The Origins, History, and Future of the Federal Reserve : A Return to Jekyll Island 9781107333185, 9781107013728

Citation preview

more information - www.cambridge.org/9781107013728

THE ORIGINS, HISTORY, AND FUTURE OF THE FEDERAL RESERVE

This book contains essays presented at a conference held in November 2010 to mark the centenary of the famous 1910 Jekyll Island meeting of leading American financiers and the U.S. Treasury. The 1910 meeting resulted in the Aldrich Plan, a precursor to the Federal Reserve Act that was enacted by Congress in 1913. The 2010 conference, sponsored by the Federal Reserve Bank of Atlanta and Rutgers University, featured assessments of the Fed’s near-100-year track record by prominent economic historians and macroeconomists. The final chapter of the book records a panel discussion of Fed policy making by current and former senior Federal Reserve officials. Michael D. Bordo is professor of economics and director of the Center for Monetary and Financial History at Rutgers University, New Jersey. He is a research associate of the National Bureau of Economic Research, Cambridge, Massachusetts. He holds a PhD from the University of Chicago. He has published many articles in leading journals in monetary economics and economic history. Recent publications include A Retrospective on the Bretton Woods International Monetary System (1993, with Barry Eichengreen), The Defining Moment: The Great Depression and the American Economy in the Twentieth Century (1998, with Claudia Goldin and Eugene White), Essays on the Gold Standard and Related Regimes (Cambridge University Press, 1999, paperback 2005), and Globalization in Historical Perspective (2003, with Alan Taylor and Jeffrey Williamson). He is also a founding and managing editor of the Cambridge University Press Studies in Macroeconomic History series. William Roberds is a research economist and senior policy adviser with the Federal Reserve Bank of Atlanta. His areas of responsibility include basic research and policy analysis. Dr. Roberds’s research focuses primarily on the fields of payment systems, macroeconomics, and econometrics. His research has been published in leading economic journals including the Journal of Monetary Economics, the International Economic Review, and the Journal of Money, Credit, and Banking, as well as in Federal Reserve System publications. Dr. Roberds joined the bank in July 1987. Previously he was an assistant professor at Brown University (1982–4) and an economist at the Federal Reserve Bank of Minneapolis (1984–7). He received his PhD from Carnegie Mellon.

Studies in Macroeconomic History Series Editor: Michael D. Bordo, Rutgers University Editors: Marc Flandreau, Institut d’Etudes Politiques de Paris Chris Meissner, University of California, Davis Franc¸ois Velde, Federal Reserve Bank of Chicago David C. Wheelock, Federal Reserve Bank of St. Louis The titles in this series investigate themes of interest to economists and economic historians in the rapidly developing field of macroeconomic history. The four areas covered include the application of monetary and finance theory, international economics, and quantitative methods to historical problems; the historical application of growth and development theory and theories of business fluctuations; the history of domestic and international monetary, financial, and other macroeconomic institutions; and the history of international monetary and financial systems. The series amalgamates the former Cambridge University Press series Studies in Monetary and Financial History and Studies in Quantitative Economic History. Other books in the series: Howard Bodenhorn, A History of Banking in Antebellum America [9780521662857, 9780521669993] Michael D. Bordo, The Gold Standard and Related Regimes [9780521550062, 9780521022941] Michael D. Bordo and Forrest Capie (Eds.), Monetary Regimes in Transition [9780521419062] Michael D. Bordo and Roberto Cort´es-Conde (Eds.), Transferring Wealth and Power from the Old to the New World [9780521773058, 9780511664793] Claudio Borio, Gianni Toniolo, and Piet Clement (Eds.), Past and Future of Central Bank Cooperation [9780521877794, 9780511510779] Richard Burdekin and Pierre Siklos (Eds.), Deflation: Current and Historical Perspectives [9780521837996, 9780511607004] Trevor J. O. Dick and John E. Floyd, Canada and the Gold Standard [9780521404082, 9780521617062] (Continued after index)

The Origins, History, and Future of the Federal Reserve A Return to Jekyll Island Edited by MICHAEL D. BORDO Rutgers University, New Jersey

WILLIAM ROBERDS Federal Reserve Bank of Atlanta

cambridge university press Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, S˜ao Paulo, Delhi, Mexico City Cambridge University Press 32 Avenue of the Americas, New York, NY 10013-2473, USA www.cambridge.org Information on this title: www.cambridge.org/9781107013728  C The Federal Reserve Bank of Atlanta 2013

This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 2013 Printed in the United States of America A catalog record for this publication is available from the British Library. Library of Congress Cataloging in Publication Data The origins, history, and future of the Federal Reserve : a return to Jekyll Island / [edited by] Michael D. Bordo, Rutgers University, New Jersey; William Roberds, Federal Reserve Bank of Atlanta. pages cm. – (Studies in macroeconomic history) Includes bibliographical references and index. ISBN 978-1-107-01372-8 1. Board of Governors of the Federal Reserve System (U.S.) 2. Federal reserve banks – History. 3. Banks and banking, Central – United States – History. 4. Monetary policy – United States – History. I. Bordo, Michael D., editor of compilation. II. Roberds, William, editor of compilation. HG2563.O75 2013 332.1 10973–dc23 2012025995 ISBN 978-1-107-01372-8 Hardback Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party Internet Web sites referred to in this publication and does not guarantee that any content on such Web sites is, or will remain, accurate or appropriate.

Contents

Conference Speaker Bios

page ix

Introduction

1

Michael D. Bordo and William Roberds

1

“To Establish a More Effective Supervision of Banking”: How the Birth of the Fed Altered Bank Supervision Eugene N. White Comment Warren Weber

2

55

The Promise and Performance of the Federal Reserve as Lender of Last Resort 1914–1933 Michael D. Bordo and David C. Wheelock Comment Ellis W. Tallman

3

59 99

Where It All Began: Lending of Last Resort at the Bank of England Monitoring During the Overend-Gurney Panic of 1866 Marc Flandreau and Stefano Ugolini Comment Barry Eichengreen

4

7

113 162

Volatile Times and Persistent Conceptual Errors: U.S. Monetary Policy 1914–1951 Charles W. Calomiris Comment Allan H. Meltzer

166 219

vii

viii 5

Contents Government Policy, Credit Markets, and Economic Activity

226

Lawrence J. Christiano and Daisuke Ikeda

6

7

Policy Debates at the Federal Open Market Committee: 1993–2002 Marvin Goodfriend

332

Two Models of Land Overvaluation and Their Implications

374

Narayana R. Kocherlakota

8

Panel Discussion: November 6, 2010 Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

Index

399

423

Conference Speaker Bios

Ben S. Bernanke began a second term as chairman of the Board of Governors of the Federal Reserve System on February 1, 2010. He also serves as chairman of the Federal Open Market Committee. Before his appointment as chairman, Bernanke was chairman of the President’s Council of Economic Advisers. Previously Bernanke had served the Federal Reserve System in several roles: as a member of the Board of Governors of the Federal Reserve System from 2002 to 2005; as a visiting scholar at the Federal Reserve Banks of Philadelphia, Boston, and New York; and as a member of the Federal Reserve Bank of New York’s academic advisory panel. He has also taught at Princeton University, Stanford University, New York University, and the Massachusetts Institute of Technology. Bernanke has published articles on a variety of economic issues and is the author of several books. He served as the director of the Monetary Economics Program of the National Bureau of Economic Research (NBER) and as a member of the NBER’s business-cycle dating committee. He received a BA in economics from Harvard University and a PhD in economics from the Massachusetts Institute of Technology. Charles W. Calomiris is the Henry Kaufman Professor of Financial Institutions at the Columbia University Graduate School of Business and a professor at Columbia’s School of International and Public Affairs. He is a member of the Shadow Financial Regulatory Committee, the Shadow Open Market Committee, and the Financial Economists Roundtable and a research associate of the National Bureau of Economic Research. He is also a member of the Task Force on Property Rights at the Hoover Institution and the Pew Trusts Task Force on Financial Reform. Calomiris codirected the Project on Financial Deregulation at the American Enterprise Institute ix

x

Speaker Bios

for more than a decade and was a senior fellow at the Council on Foreign Relations. He has received research grants from the National Science Foundation and other foundations. He has worked for numerous government organizations, central banks, multilateral organizations, and corporations. Calomiris served on the International Financial Institution Advisory Commission, a congressional commission to advise the U.S. government on the reform of the International Monetary Fund, the World Bank, regional development banks, and the World Trade Organization. Calomiris is the author of numerous books and academic articles in the areas of banking, corporate finance, financial history, monetary economics, and economic development. In 2010–11 he was a Houblon-Norman Senior Fellow at the Bank of England and a Podlich Fellow at Claremont McKenna College. Currently, he teaches master’s and PhD-level courses on advanced corporate finance, emerging financial markets, and the history of financial crises. He received a BA in economics from Yale University and a PhD in economics from Stanford University. Lawrence J. Christiano is the Alfred W. Chase Chair in Business Institutions at Northwestern University, where he joined the faculty in 1992. From 1985 to 1992 he was first an economist, then a research officer, and finally the director of the Institute for Empirical Macroeconomics at the Federal Reserve Bank of Minneapolis. Prior to joining the Minneapolis Fed, he was a visiting assistant professor at Carnegie Mellon University and an assistant professor at the University of Chicago. Christiano is currently a visiting scholar at the Federal Reserve Bank of Cleveland and an adviser at the Federal Reserve Bank of Atlanta. He is a research associate with the National Bureau of Economic Research and a fellow of the Econometric Society. He has been a visiting scholar at the International Monetary Fund and the European Central Bank and has served as a consultant at the Federal Reserve Board and the Federal Reserve Bank of Chicago. In addition, he has taught short courses on monetary economics at numerous universities in Europe and the Middle East. The author of many articles and papers, Christiano is an associate editor of the Journal of Money, Credit, and Banking and a former associate editor of eight other academic journals, including the International Economic Review and the American Economic Review. Christiano earned a BA in history and economics and an MA in economics from the University of Minnesota, an MSc in econometrics and mathematical economics from the London School of Economics, and a PhD in economics from Columbia University.

Speaker Bios

xi

E. Gerald Corrigan is a managing director at Goldman, Sachs & Company, a position he has held since 1994. Among his other activities at Goldman Sachs, he is chairman of Goldman Sachs Bank USA, cochair of the Firmwide Risk Management Committee, vice chair of the Firmwide Business Practices Committee, and a member of the Firmwide Commitments Committee. Since joining Goldman Sachs in 1994, Corrigan has served as chair or cochair of a number of firmwide and industrywide groups dealing with a range of issues with major implications for financial market efficiency and stability. In addition, he provides a wide range of strategic advice to the firm and its clients. Corrigan ended a twenty-five-year career with the Federal Reserve System when he stepped down from his position as president and chief executive officer of the Federal Reserve Bank of New York in 1993. He had been chief executive officer of the New York Fed and vice chairman of the Federal Open Market Committee since 1984. He has also served as president of the Federal Reserve Bank of Minneapolis and special assistant to Fed Chairman Paul Volcker. Corrigan is the chairman, a trustee, or a member of a number of nonprofit organizations. He earned a bachelor of social science degree in economics from Fairfield University and MA and PhD degrees in economics from Fordham University. Barry Eichengreen is the George C. Pardee and Helen N. Pardee Professor of Economics and professor of political science at the University of California, Berkeley, where he has taught since 1987. He is a research associate of the National Bureau of Economic Research in Cambridge, Massachusetts, and a research fellow of the Centre for Economic Policy Research in London. In 1997–98 he was senior policy adviser at the International Monetary Fund. He is a fellow of the American Academy of Arts and Sciences. Eichengreen is the convener of the Bellagio Group of academics and economic officials and chair of the academic advisory committee of the Peterson Institute of International Economics. He has held Guggenheim and Fulbright Fellowships and has been a fellow of the Center for Advanced Study in the Behavioral Sciences in Palo Alto, California, and the Institute for Advanced Study in Berlin. He is a regular monthly columnist for Project Syndicate. He has written or edited numerous books. The most recent are Emerging Giants: China and India in the World Economy (coedited with Poonam Gupta and Ranjiv Kumar, 2010), Labor in the Era of Globalization (coedited with Clair Brown and Michael Reich, 2009), and Fostering Monetary & Financial Cooperation in East Asia (coedited with Duck-Koo Chung, 2009). Eichengreen was awarded the Economic History Association’s Jonathan R. T. Hughes

xii

Speaker Bios

Prize for Excellence in Teaching in 2002 and the University of California at Berkeley Social Science Division’s Distinguished Teaching Award in 2004. He received a doctor honoris causa from the American University in Paris and the 2010 Schumpeter Prize from the International Schumpeter Society. He is the 2010–11 president of the Economic History Association. He received an AB degree from the University of California, Santa Cruz, and MA and MPhil degrees (economics), an MA in history, and a PhD in economics from Yale University. Marc Flandreau is a professor of international history and politics at the Graduate Institute for International Studies and Development, Geneva. He is a specialist in the history of the international monetary system and has written extensively on the history of money, exchange rates, monetary policy, and financial crises. In the past, he has taught or held visiting positions in Sciences Po Paris, Stanford, Berkeley, and the London School of Economics. He has held consulting positions at the International Monetary Fund, the Bank of France, the Bank for International Settlements, and Lehman Brothers, France. Since 1994 he has been a research fellow at the Centre for Economic Policy Research, London. His books include The Glitter of Gold: France and the Emergence of the International Gold Standard, 1848–1878 (Oxford University Press, 2003) and Money Doctors: The Experience of International Financial Advising, 1850–2000 (Routledge, 2004). A former graduate from Ecole Normale Sup´erieure in Paris and Fulbright Scholar at the University of California, Berkeley, Flandreau holds a PhD in economics from the Ecole des Hautes Etudes, Paris, and the London School of Economics. Marvin Goodfriend is a professor of economics and chairman of the Gailliot Center for Public Policy at the Tepper School of Business at Carnegie Mellon University. He was senior vice president and policy adviser at the Federal Reserve Bank of Richmond from 1993 to 2005, where he regularly attended meetings of the Federal Open Market Committee. In 1984–85 he served as a senior staff economist for the President’s Council of Economic Advisers at the White House. He was a visiting professor at the Graduate School of Business of the University of Chicago from September 1988 to June 1990. He has been a visiting scholar at the Board of Governors of the Federal Reserve System, the European Central Bank, the Institute for International Economic Studies at the University of Stockholm, the International Monetary Fund, the Swiss National Bank, and the Federal Reserve Banks of Atlanta, Cleveland, Kansas City, and New York. He served on external

Speaker Bios

xiii

review panels to evaluate research and policy advice at the European Central Bank, Norges Bank, the Swedish Riksbank, and the Swiss National Bank. Goodfriend is coeditor of the Carnegie-Rochester Conference Series on Public Policy and has served on editorial boards of the Journal of Money, Credit, and Banking, the International Journal of Central Banking, and the Journal of Monetary Economics. He is a member of the Economic Advisory Panel of the Federal Reserve Bank of New York and a research associate of the National Bureau of Economic Research. He holds a BS in mathematics from Union College and a PhD in economics from Brown University. Alan Greenspan currently heads Greenspan Associates, a consulting firm in Washington, DC, and is the author of The Age of Turbulence. He served from 1987 to 2006 as chairman of the Board of Governors of the Federal Reserve System. He also served as chairman of the Federal Open Market Committee, the Federal Reserve System’s principal monetary policy-making body. He originally took office as chairman and to fill an unexpired term in 1987 and was reappointed to a full fourteen-year board term in 1992. He was designated Fed chairman by Presidents Reagan, Bush, Clinton, and Bush. From 1954 to 1974 and from 1977 to 1987, Greenspan was chairman and president of Townsend-Greenspan & Company Inc., an economic consulting firm in New York City. From 1974 to 1977, he served as chairman of the President’s Council of Economic Advisers under President Ford. From 1981 to 1983 he was chairman of the National Commission on Social Security Reform. He was appointed a member of President Reagan’s Economic Policy Advisory Board, the President’s Foreign Intelligence Advisory Board, the Commission on Financial Structure and Regulation, the Commission on an All-Volunteer Armed Force, and the Task Force on Economic Growth. Before his appointment to the Federal Reserve Board, Greenspan served as a director of numerous corporations, including J.P. Morgan & Company Inc., Mobil Corporation, Aluminum Company of America (Alcoa), General Foods Corporation, and Capital Cities/ABC Inc. He was a term member of the board of trustees of the Rand Corporation, a member of the board of overseers of the Hoover Institution (at Stanford University), and vice chairman and trustee of the Economic Club of New York. Greenspan has served as chairman of the Conference of Business Economists, president and fellow of the National Association of Business Economists, and fellow of the American Statistical Association. He has received honorary degrees from Harvard, Yale, Pennsylvania, Notre Dame, Leuven (Belgium), and Edinburgh universities. He received the Legion of Honor (Commander) from France, became an honorary Knight Commander of the British

xiv

Speaker Bios

Empire, and received the Medal of Freedom, the United States’ highest civil award. He received a BS in economics summa cum laude, an MA, and a PhD from New York University. Daisuke Ikeda is Deputy Director and Economist at the Bank of Japan. He joined the bank in 2004. He holds a BA from Kobe University in 2002, an MA from the University of Tokyo in 2004, and a PhD from Northwestern University in 2011. Narayana R. Kocherlakota became the twelfth president of the Federal Reserve Bank of Minneapolis in 2009. As president, Kocherlakota serves on the Federal Open Market Committee, the policy-making arm of the Federal Reserve System. Previously he was a professor of economics at the University of Minnesota, where he chaired the economics department, and a consultant to the Federal Reserve Bank of Minneapolis. He was also a research associate at the National Bureau of Economic Research. From 1996 to 1998, he was a research staff member at the Minneapolis Fed. Kocherlakota has published more than thirty articles in academic journals, including Econometrica, the Journal of Political Economy, the Journal of Economic Theory, the Journal of Monetary Economics, and the Journal of Money, Credit, and Banking. His work includes theoretical and empirical contributions to macroeconomics, monetary economics, financial economics, and public finance. His book The New Dynamic Public Finance (Princeton University Press) describes an approach to tax design pioneered by him and others over the past decade. He earned an AB in mathematics from Princeton and a PhD in economics from the University of Chicago. Dennis P. Lockhart is president and chief executive officer of the Federal Reserve Bank of Atlanta. In this role he is responsible for all of the bank’s activities, including monetary policy, bank supervision and regulation, and payment services. He also chairs the bank’s management committee and serves on the Federal Reserve’s chief monetary policy body, the Federal Open Market Committee. Prior to joining the Atlanta Fed in March 2007, Lockhart served on the faculty of Georgetown University’s Walsh School of Foreign Service, teaching in the master’s program and chairing the program’s concentrations in international business–government relations and global commerce and finance. He was also an adjunct professor at Johns Hopkins University’s Nitze School of Advanced International Studies. Prior to his academic career, Lockhart was a managing partner at the private equity firm Zephyr Management LP. He also worked at Heller Financial, where he

Speaker Bios

xv

served as executive vice president and director of the parent company and as president of Heller International Group. Previously Lockhart held various positions, both domestic and international, with Citicorp/Citibank. In addition to his professional activities, Lockhart was a member of the boards of directors of several companies and was chairman of the Small Enterprise Assistance Funds, a not-for-profit operator of emerging markets venture capital funds. He holds a BA in political science and economics from Stanford University and an MA in international economics and American foreign policy from the Johns Hopkins University School of Advanced International Studies. Allan H. Meltzer is the Allan H. Meltzer University Professor of Political Economy and Public Policy at Carnegie Mellon University and a visiting scholar at the American Enterprise Institute in Washington. He has been a visiting professor at Harvard, the University of Chicago, the University of Rochester, the Yugoslav Institute for Economic Research, the Austrian Institute for Advanced Study, the Getulio Vargas Foundation in Rio de Janeiro, and the City University, London. He has served as a consultant for several congressional committees, the President’s Council of Economic Advisers, the U.S. Treasury Department, the Board of Governors of the Federal Reserve System, the World Bank, foreign governments, and central banks. He has been a member of the President’s Economic Policy Advisory Board and the President’s Council of Advisers and an honorary adviser to the Institute for Monetary and Economic Studies of the Bank of Japan. In 1999–2000 he served as chairman of the International Financial Institution Advisory Commission, known as the Meltzer Commission, which proposed major reforms of the International Monetary Fund and development banks. Meltzer has published more than 300 papers on economic theory and policy in numerous journals and is the author of several books, the most recent being A History of the Federal Reserve (University of Chicago Press, two volumes, 2003 and 2010). He is a former coeditor of the Carnegie-Rochester Conference Series on Public Policy, the Journal of Economic Literature, and the Journal of Finance. From 1973 to 1999, Meltzer was chairman of the Shadow Open Market Committee. He is a past president of the Western Economic Association and a fellow of the National Association of Business Economists. He is a distinguished fellow of the American Economic Association. In 2003 he received the Irving Kristol Award of the American Enterprise Institute and the Adam Smith Award from the National Association for Business Economics. He received an AB and an MA from Duke University and a PhD from the University of California, Los Angeles.

xvi

Speaker Bios

Raghuram Rajan is the Eric J. Gleacher Distinguished Service Professor of Finance at the University of Chicago’s Graduate School of Business. He is also currently an economic adviser to the prime minister of India. Prior to resuming teaching in 2007, Rajan was the economic counselor and director of research at the International Monetary Fund. He chaired the Indian government’s Committee on Financial Sector Reforms, which submitted its report in September 2008. Rajan’s research interests are banking, corporate finance, and economic development, especially the role finance plays in it. His papers have been published in all the top economic and finance journals, and he has served on the editorial boards of the American Economic Review and the Journal of Finance. His books include Saving Capitalism from the Capitalists (with Luigi Zingales) and the just-published Fault Lines: How Hidden Fractures Still Threaten the World Economy. He is a senior adviser to Booz and Company and BDT Capital and is on the academic advisory board of Moodys and the international advisory board of Bank Itau-Unibanco. He is a director of the Chicago Council on Global Affairs and a member of the Comptroller General of the United States’ Advisory Council. He is the president-elect of the American Finance Association and a member of the American Academy of Arts and Sciences. In January 2003, the American Finance Association awarded Rajan the inaugural Fischer Black Prize, given every two years to the financial economist under the age of 40 who has made the most significant contribution to the theory and practice of finance. He received a BTech degree from the Indian Institute of Technology, an MBA from the Indian Institute of Management, and a PhD from the Massachusetts Institute of Technology. Ellis W. Tallman is the Danforth-Lewis Professor of Economics at Oberlin College. He has published scholarly articles in the fields of macroeconomics, economic forecasting, and historical episodes of financial crises. His articles have appeared in scholarly journals such as the Journal of Monetary Economics, the Journal of Business and Economic Statistics, and the Journal of Economic History. Prior to joining Oberlin College, Tallman was a vice president and team leader for the macrogroup in the research department at the Federal Reserve Bank of Atlanta. From January 1996 to December 1997, he was a visiting senior research economist at the Reserve Bank of Australia, where he engaged in policy support and economic research for the Australian central bank. He is currently a visiting scholar at the research department of the Federal Reserve Bank of Cleveland. He received a BA from Indiana University, Bloomington, and an MA and PhD from the University of Rochester.

Speaker Bios

xvii

Stefano Ugolini is Assistant Professor of economics at the University of Toulouse 1 (Institute of Political Studies and LEREPS). A former graduate from Scuola Normale Superiore di Pisa, he holds a PhD from Sciences Po Paris and a post-doc from the Graduate Institute in Geneva; he has also served as a consultant to the Bank of Norway in the framework of Norges Bank’s Bicentenary Project. Ugolini’s research focuses on the history of central banking and monetary policy and the evolution of financial markets and international monetary systems, as well as financial development. His recent publications include “Bagehot for Beginners: The Making of Lending of Last Resort Operations in the Mid-Nineteenth Century” (with Vincent Bignon and Marc Flandreau), Economic History Review, 65:2, 2012; “The Origins of Foreign Exchange Policy: The National Bank of Belgium and the Quest for Monetary Independence in the 1850s,” European Review of Economic History, 16:1, 2012; and “Foreign Exchange Reserve Management in the Nineteenth Century: The National Bank of Belgium in the 1850s,” in ¨ Anders Ogren and Lars F. Øksendal (eds.), The Gold Standard Peripheries: Monetary Policy, Adjustment and Flexibility in a Global Setting (Palgrave Macmillan, 2012). Paul A. Volcker worked in the United States Federal Government for almost 30 years, culminating in two terms as Chairman of the Board of Governors of the Federal Reserve System from 1979 to 1987, a critical period in bringing a high level of inflation to an end. In earlier stages of his career, Mr. Volcker served as Undersecretary of the Treasury for Monetary Affairs during the early 1970s, a period of historic change in international monetary arrangements. He was subsequently President of the Federal Reserve Bank of New York, and in earlier years was an official of The Chase Manhattan Bank. Mr. Volcker retired as Chairman of Wolfensohn & Co. upon the merger of that firm with Bankers Trust. From 1996 to 1999, Mr. Volcker headed The Independent Committee of Eminent Persons, formed by Swiss and Jewish organizations to investigate deposit accounts and other assets in Swiss banks of victims of Nazi persecution and to arrange for their disposition. From 2000 to 2005 Mr. Volcker served as Chairman of the Board of Trustees of the newly formed International Accounting Standards Committee overseeing a renewed effort to develop consistent, high-quality accounting standards acceptable in all countries. Upon leaving public service in 1987, and again in 2003, he headed private, non-partisan Commissions on the Public Service, each recommending a sweeping overhaul of the organization and personnel practices of the United States Federal Government. In 2004, Mr. Volcker was asked by UN Secretary General Kofi Annan to chair the Independent

xviii

Speaker Bios

Inquiry into the United Nations Oil-for-Food Program, resulting in identification of substantial corruption and malfeasance. In 2007, Mr. Volcker was asked by the President of the World Bank to chair a panel of experts to review the operations of the Department of Institutional Integrity. That effort has culminated in broad reform of the Bank’s anti-corruption effort. In November 2008, President Elect Obama chose Mr. Volcker to head the President’s Economic Recovery Advisory Board, which ended in February 2011. Pursuing his many continuing interests in public policy, Mr. Volcker, among his extensive non-profit activities, has been former Chairman of the Trilateral Commission and former Chairman of the Trustees of the Group of 30. Educated at Princeton, Harvard, and the London School of Economics, Mr. Volcker is a recipient of honorary doctorates from each of his “alma maters,” as well as a number of other American and foreign universities. Warren Weber recently retired as a senior research officer in the research department at the Federal Reserve Bank of Minneapolis. He was also an adjunct economics professor at the University of Minnesota and a research associate at the Center for the Advanced Study in Economic Efficiency. Before joining the Minneapolis Fed, he taught at Virginia Polytechnic Institute and State University, Tulane University, and Duke University. He has published papers in the American Economic Review, the Journal of Political Economy, and the Journal of Economic History, among others. His research agenda focuses on monetary theory, the history of banking in the United States, and medieval monetary systems. He earned his PhD from Carnegie Mellon University. David C. Wheelock is Vice President and Deputy Director of the Research Division of the Federal Reserve Bank of St. Louis, where he has been employed since 1993. He serves as an advisor to the bank president on monetary and financial policy issues and conducts policy-oriented research. Before joining the Federal Reserve Bank of St. Louis, Dr. Wheelock was a member of the faculty of the Department of Economics at the University of Texas at Austin. Dr. Wheelock has written numerous articles on banking and monetary policy issues for professional journals and Federal Reserve publications. He is the author of The Strategy and Consistency of Federal Reserve Monetary Policy, 1924–1933 (Cambridge University Press, 1991). Dr. Wheelock received his BS degree from Iowa State University, and his MS and PhD degrees from the University of Illinois at Urbana-Champaign. Eugene N. White is a professor of economics at Rutgers University and a research associate at the National Bureau of Economic Research. He

Speaker Bios

xix

is a former editor of Explorations in Economic History. He has been a visiting professor at the Paris School of Economics, Erasmus University, and New York University and a visiting scholar at the Federal Reserve Banks of Atlanta and Philadelphia. His current research focuses on the history of housing market booms from the 1920s to the present, the evolution of U.S. bank supervision and regulation, the Banque de France and financial crises in the nineteenth century, and the microstructure of securities markets in Europe and America. His most recent book is Conflicts of Interest in the Financial Services Industry: What Should We Do About Them? (with Andrew Crockett, Trevor Harris, and Frederic Mishkin, London, 2003). He received an AB in history magna cum laude from Harvard University, a BA in history and economics from Oxford University, and an MA and PhD in economics from the University of Illinois, Urbana. Other participants in the 2010 Jekyll Island conference James Bullard, President, Federal Reserve Bank of St. Louis Charles L. Evans, President, Federal Reserve Bank of Chicago Richard W. Fisher, President, Federal Reserve Bank of Dallas Anil K. Kashyap, Professor of Economics and Finance, University of Chicago Robert G. King, Professor of Economics, Boston University Jeffrey Lacker, President, Federal Reserve Bank of Richmond Bennett McCallum, Professor of Economics, Carnegie Mellon University Athanasios Orphanides, Governor, Central Bank of Cyprus Sandra Pianalto, President, Federal Reserve Bank of Cleveland Charles I. Plosser, President, Federal Reserve Bank of Philadelphia Eric S. Rosengren, President, Federal Reserve Bank of Boston Thomas J. Sargent, Professor of Economics and Business, New York University

Introduction Michael D. Bordo and William Roberds

This volume contains the proceedings of a recent conference organized by the Federal Reserve Bank of Atlanta and Rutgers University. The conference was held on November 5 and 6, 2010, to mark the centenary of the famous 1910 Jekyll Island meeting. That meeting led to the drafting of legislation – the Aldrich Plan – for a central bank for the United States. The Aldrich Plan would serve as a template for the Federal Reserve Act, which was eventually passed into law. The 2010 conference was held at the same location as the 1910 meeting, the Jekyll Island Club (which is now a hotel operating under the same name). The purpose of the conference was to offer a forum for examination of the Federal Reserve’s near-100-year track record, in the context of the vision of its founders. The methodology of the papers included both economic history and macroeconomics. Presentations at the conference included eight academic papers and a concluding panel discussion by current and past Federal Reserve officials. THE SIGNIFICANCE OF THE 1910 MEETING

Following the panic of 1907, Congress passed the Aldrich-Vreeland Act of 1908. This act established a study group, the National Monetary Commission, chaired by Republican Senator Nelson Aldrich of Rhode Island. The commission was charged with finding a way to reform the nation’s monetary system and proposing possible designs for a central bank. Progress toward those goals began with the commission taking trips to major European banking centers and holding hearings in the United States. However, by 1910 the eighteen-member commission could not agree on a plan. Aldrich then took matters into his own hands, meeting secretly with a group of bankers on Jekyll to formulate a plan. The group included Henry 1

2

Michael D. Bordo and William Roberds

Davison, Frank Vanderlip, and Paul Warburg; also present were Aldrich’s secretary, Arthur Shelton, and A. Piatt Andrew, a U.S. Treasury official. Warburg’s attendance was critical because of his knowledge of European banking practices. Aldrich was well aware that his meeting with bankers outside of the commission proceedings would generate controversy; for this reason, the group met in private at a remote location. Over the years, the clandestine nature of the meeting has often been criticized as allowing undue Wall Street influence over the founding of the U.S. central bank. However, the meeting itself was just one step in the process that led to the creation of the Federal Reserve, and many details of Aldrich’s original design were changed in the legislation that was eventually passed. What emerged from the Jekyll meeting was the so-called Aldrich Plan, which was presented to the National Monetary Commission as a legislative blueprint. Although not realized in its original form, the plan was an effective catalyst for debate about the role of the government and the banking industry in the central bank’s governance. The plan, and the bill that followed in 1912, proposed that a National Reserve Association would function as the U.S. central bank. The association would consist of a federation of regional bank associations, each presided over by boards of directors elected by local banks. This diffuse structure represented a departure from the centralized governance of European central banks. Essentially a private sector organization, Aldrich’s proposed central bank would promote the public goal of financial and macroeconomic stability. The proposals that came out of the Jekyll meeting remained the core of the subsequent steps in the creation of the Federal Reserve. Aldrich put forward his central banking bill to the U.S. Senate in January 1912, during the last year of the Taft administration. The bill ran into significant opposition, and the Senate did not act on it. After the election of Woodrow Wilson in 1912, Democratic Congressman Carter Glass of Virginia assumed leadership of monetary reform. Glass had served for ten years as a minority member of the House Banking and Currency Committee and had led an investigation into concentration of power on Wall Street. Working in coordination with Glass was Oklahoma Senator Robert L. Owen. Their proposal for a central bank, the Glass-Owen Bill of 1913, kept the key financial and macroeconomic stabilization mechanisms of the Aldrich Plan, but it differed in terms of the proposed institution’s structure and governance. The Glass-Owen Bill provided for eight to twelve districts, with a reserve bank in each district – a departure from the Aldrich Bill, which called for fifteen districts, with a central bank branch in each. However, the most salient difference was the governance of the central bank. The debate

Introduction

3

about a U.S. central bank had been mired in disagreements about who would control the system – bankers or appointed government officials. The Glass-Owen Bill achieved a compromise by allowing limited banking interest representation. Of the nine-member reserve bank boards of directors, only three could be bankers. Three other directors engaged in commerce, industry, or agriculture were to be elected by bankers, and the remaining three directors were to be named by the Federal Reserve Board in Washington, DC. The bill also identified the Federal Reserve Board as the controlling agency. The board was made up of two ex officio members – the secretary of the treasury and the comptroller of the currency – and five other members appointed by the president and confirmed by the Senate. The reconciled bill was passed as the Federal Reserve Act and signed into law by President Wilson in 1913, and the twelve reserve banks opened about a year later. THE 2010 CONFERENCE

The conference program was divided into three sessions of scholarly papers and a concluding panel discussion. The three papers in the first session focused on the origins of the Federal Reserve and its initial impact on the United States’ financial landscape. The first paper, by Eugene White, documents the changes that took place in bank supervisory practices following the founding of the Fed. White argues that national bank regulation during the National Banking Era was characterized by a “light touch.” Direct supervision was minimal, but double liability of shareholders provided strong incentives for prompt closure of insolvent banks. This regime changed with the arrival of the Federal Reserve, which was more concerned about ensuring the liquidity of the banking system than about controlling moral hazard at the supervisory level. The payback for this regulatory laxity came with the deflationary shocks of 1929–33, which ultimately resulted in the closure of over half of the nation’s banks. New Deal reforms of banking regulation attempted to strengthen bank supervision by introducing a regime of more intensive, discretionbased supervision. The effectiveness of this reform was limited, according to White, because New Deal legislation also eliminated shareholders’ double liability, thereby reducing incentives for prompt closure of insolvent banks. The second paper in the session, by Michael Bordo and David Wheelock, examines the Fed’s early performance as a lender of last resort. The authors argue that the Fed’s inability to counteract the banking panics of the 1930s stemmed from a failure to provide a monetary environment conducive to effective policy intervention. Most critically, the Federal Reserve Act did

4

Michael D. Bordo and William Roberds

not succeed in its goal of creating a deep and liquid market in bankers’ acceptances, a key objective of Paul Warburg and other framers of the Aldrich Plan. Bordo and Wheelock contend that if such a market had existed, it would have allowed the Fed to more easily extend liquidity to the banking system during crisis periods. In addition, the act did nothing to reform the unit structure of the U.S. banking industry, which made it especially vulnerable to systemic shocks. These two major shortcomings were compounded by additional policy mistakes. From this analysis, Bordo and Wheelock conclude that in order to be effective, a lender of last resort’s policy tools must be a good match for the prevailing financial environment. The topic of policy effectiveness is also explored in the third paper of this session, contributed by Marc Flandreau and Stefano Ugolini. Their paper explores the role of the Bank of England during the Overend-Gurney crisis of 1866. This episode was known to the framers of the Aldrich Plan and would have contributed to their favorable view of the capabilities of a lender of last resort. Using previously unexplored data from the Bank of England’s archives, the authors document the Bank’s interventions in the London bill market. These data show that the support extended by the Bank was of a surprisingly “cross-border” nature, that is, much of it went to prop up the market in international acceptances. The data also indicate that the main recipients of the Bank’s emergency liquidity infusions were not deposit banks, but nonbank financial market participants known as bill brokers – the “shadow bankers” of the time. On the whole, the authors suggest, the archival data portray a more active role for the Bank of England than has been indicated in earlier accounts. Hence, according to their analysis, the framers of the Aldrich Plan may well have underestimated the financial firepower necessary for the Fed to succeed as a lender of last resort. The second session of the conference examined the Fed’s historical performance, from its founding in 1913 to the present day. The first paper of the session, contributed by Charles Calomiris, analyzes the key policy concept underlying the early decades (until the 1951 Fed–Treasury accord) of the Fed’s monetary policy decisions, the real bills doctrine. The founders of the Fed believed that adherence to this doctrine would lessen the likelihood of credit contractions as were experienced during the National Banking Era. In practice, however, Calomiris contends that adherence to real bills, which viewed the main role of monetary policy as accommodating shifts in demand for trade credit, only served to amplify cyclical fluctuations in money and liquidity. A major problem with implementation of the real bills doctrine lay with Fed policymakers’ inability to distinguish between demand and supply shocks. During the Depression

Introduction

5

this shortcoming was compounded by policymakers’ confusion of real and nominal interest rates. Moreover, the Fed was slow to learn from its mistakes and abandon real bills, according to Calomiris. One factor behind this reticence may have been the Fed’s success in smoothing seasonal fluctuations in interest rates, apparently validating the real bills approach. Another may have been the sheer size of external forces during this period: World War I, the Roaring Twenties, the Great Depression, World War II, and the Korean War. The final paper of the second session, by Lawrence Christiano and Daisuke Ikeda, offers theoretical analyses of policy interventions in credit markets. With an eye to the unconventional policies that were deployed during the 2007–8 crisis, Christiano and Ikeda consider the efficacy of four types of interventions: (1) reductions in the costs of funds to financial firms, (2) equity injections into financial firms, (3) loans to financial and nonfinancial firms, and (4) transfers of net worth to financial firms. Their paper evaluates these policies by constructing four different theoretical models of credit subject to financial frictions, and considering the effects of the different policies in each model environment. Christiano and Ikeda find reductions in funding costs and net worth transfers to be the most robust policies, in the sense that these policies reduce market dysfunction in all of the models considered. Loans and equity injections, by contrast, are shown to be effective in some environments but not in others. The last two papers of the conference focused on the future direction of Federal Reserve. The first paper, by Marvin Goodfriend, offers a detailed look at the decentralized nature of Fed policymaking. The paper recounts the contributions of J. Alfred Broaddus, former president of the Federal Reserve Bank of Richmond, to Federal Open Market Committee (FOMC) meetings over the period 1993–2004. Goodfriend discusses Broaddus’s positions on seven different major policy debates that came before the FOMC during this time. Many of these positions were motivated by new insights from academic research. In every case considered, Broaddus’s judgment concerning the appropriate direction of policy diverged from the views of many other committee members, including the chairman’s. In some instances, Broaddus’s views were eventually incorporated into Fed policy, in others not. Taken as a whole, Goodfriend argues, this microlevel analysis confirms the ongoing value of the Fed’s decentralized structure. The second paper of the session was contributed by Narayana Kocherlakota. His paper considers the issue of how policy should interact with asset bubbles. A bubble is said to occur when buyers are willing to pay more for an asset than they would if they were required to hold that same asset without

6

Michael D. Bordo and William Roberds

reselling it. The paper proposes two theoretical models of bubbles in land prices. The first model illustrates how land price bubbles can be exacerbated through ineffective bank regulation. The second model employs the theory of rational bubbles to illustrate the negative growth consequences of a bubble collapse, which stem from both a dampening of credit markets and a redistribution of wealth. Kocherlakota notes that in the latter model, the ill effects of a bubble collapse can always be undone through (fiscal) policy interventions, but that in practice there will be limitations on the ability of governments to effect such interventions. The conference concluded with a panel discussion by Federal Reserve policymakers. Members of the panel included Ben Bernanke (chairman of the Board of Governors since 2006), Alan Greenspan (chairman, 1987– 2006), and Gerald Corrigan (former president of the Federal Reserve Banks of New York and Minneapolis). There was also a brief video presentation by Paul Volcker (chairman, 1979–87). The discussion began with a request from the moderator (Raghuram Rajan) to the panelists for their recollections of a particularly challenging moment during their tenure as Fed policymakers. These were the beginning of the Volcker disinflation (Corrigan), the 1987 stock market crash (Greenspan), and the 2008 Lehman crisis (Bernanke). The conversation then moved on to a wide range of topics, with an emphasis on the interactions between Federal Reserve policy and financial markets.

ONE

“To Establish a More Effective Supervision of Banking” How the Birth of the Fed Altered Bank Supervision Eugene N. White An Act to provide for the establishment of Federal reservaue banks, to furnish an elastic currency, to afford means of rediscounting commercial paper, to establish a more effective supervision of banking in the United States, and for other purposes. – The title of the Federal Reserve Act of 1913 [emphasis added]

Although the formation and development of the Federal Reserve have been intensively studied, histories of the Fed, from Milton Friedman and Anna J. Schwartz’s Monetary History (1963) to Allan H. Meltzer’s History of the Federal Reserve (2003–2009), have focused almost exclusively on monetary policy.1 The quiet work of bank supervision has seemed somewhat pedestrian compared with the high-profile determination of interest rates, monetary aggregates, and exchange rates. Yet, the founders of the Fed regarded bank supervision as a key element in their new policy regime, placing it in the title of the Federal Reserve Act. Unfortunately, in spite of this headlining, the effective approach to bank supervision that had developed under the National Banking System was slowly undermined in the Fed’s first decade and a half. Although the National Banking System had frequent banking panics that increased the severity of recessions, its regulation and supervision successfully limited losses to depositors. Supervision before 1913 had a relatively light hand and may be characterized as a regime that aimed at reinforcing market discipline with prompt closure of insolvent institutions. Its panics were primarily driven by liquidity problems rather than by solvency issues, even if a few bank failures inaugurated a panic. The underlying causes of these frequent crises were the prohibition on branch banking, which created a fragmented system of undiversified unit banks, and the absence of a central bank. 7

8

Eugene N. White

The situation in which reformers found themselves after the crisis of 2008 would not have been unfamiliar to the fathers of the Federal Reserve System. After a half century, the banking regime created during the Civil War was outdated. Ignited by the collapse of fast-growing new institutions outside the federal “safety net,” the panic of 1907 led to a demand for an overhaul of the regulatory system. In response, the Federal Reserve Act created a lender of last resort and sought to subject state-chartered banks and trust companies that had operated with weaker regulations to federal rules. When the Fed opened for operation in 1914, a conflict immediately arose with the Comptroller of the Currency over which agency would control federal supervision. Competition between these agencies slowly weakened oversight of banking. In addition, the opening of the discount window provided a new option for troubled member banks. Previously, when a bank saw declining profitability and the possibility of failure, its directors might close the bank in advance of insolvency, fearful of the penalty of double liability that would fall on them and the other shareholders should the bank fail. These voluntary liquidations kept the number of actual insolvencies relatively low. Under the new regime, a troubled member bank now could obtain loans from the discount window, enabling it to continue operation in the hope of recovery. Although the new Federal Reserve regime of bank supervision weakened the effectiveness of the pre-1913 system and raised the costs of bank failures, it left most of its principal elements in place. This old regime was toppled only when the Great Depression led to a loss of faith in markets. Under the New Deal, supervision designed to reinforce market discipline was replaced by discretionary supervision with forbearance, setting the stage for the crises of the late twentieth century.2

I. The Problems of the National Banking System, 1864–1913 I.A. Origins The panic of 1907 and subsequent recession of 1907–8 highlighted the need to redesign the National Banking System. This aging regulatory regime was the product of an earlier effort to tackle the key issues of how to ensure financial stability. Following the expiration of the Second Bank of the United States’ charter in 1836, the federal government abandoned any role in the regulation of the banking system. The states then experimented with a variety of regimes, but the predominant one was “free banking.” This type of banking system had two key characteristics: free entry or very low

“To Establish a More Effective Supervision of Banking”

9

barriers to entry and the issue of banknotes by each individual bank that were guaranteed by the purchase of an equivalent amount of state bonds valued at par.3 After some initial difficulties, these state free banking systems expanded, providing increased intermediation and a relatively safe currency. However, the unanticipated shock of the Civil War undermined many state systems. As the values of state bonds fell, the backing or insurance behind banknotes declined, producing a wave of failures.4 In 1861, there were 1,601 banks, dropping to 1,492 a year later.5 Congress responded to this crisis and the desire to build a deeper market for federal bonds by creating the National Banking System. Patterned on the states’ antebellum systems, the National Currency Act of 1863 and the National Bank Act of 1864 established a federal free banking system. Although setting high standards for reserve requirements, minimum capital, and lending, the objective was to create a nationwide federal system generating a large number of new “national” banks and absorbing the state banks.6 The banknotes issued by national banks were given a much better backing than state banknotes had secured – U.S. government bonds – that ultimately provided a uniform, safe currency for the public. The Bank Act of 1864 also created the Office of the Comptroller of the Currency (OCC) to examine the national banks and ensure that they complied with the regulations. The agency’s name emphasized the initial primacy of protecting the security of national banknotes. In terms of the key design issues for a supervisory regime, the National Banking System was remarkably simple. There was no central bank, and high-powered money was largely determined by the balance of payments under the gold standard.7 Hence there was no potential conflict price stability and financial stability because of the monetary policy conducted by the central bank. The absence of a central bank also made bank supervision independent. If Congress had succeeded in winding down the state banking systems, with their separate state regulatory agencies, there would have been only one nationwide regulator of the banking system, the OCC. Although the OCC was formally a bureau of the Treasury Department, it was granted a considerable degree of independence. The head of the agency, the comptroller, was appointed by the president on the nomination of the secretary of the Treasury, with the advice and consent of the Senate for a term of five years, freeing the comptroller for the immediate pressures of the electoral cycle. Although most pre-1914 comptrollers did not complete a full term, some were renewed.8 The comptroller was given a salary of $5,000, but he was required to post a substantial bond of $100,000 for the faithful discharge of his duties. He was also given a deputy comptroller with half the

10

Eugene N. White 14 12 10

Thousands

State Banks 8 6 4 National Banks 2

Trust Companies

0

1863 1867 1871 1875 1879 1883 1887 1891 1895 1899 1903 1907 1911

Figure 1.1. The number of banks by charter type. Sources: Carter et al. (2006, Series Cj203, Cj212, Cj149); Bell (1928).

salary and half the bond and the right to hire the necessary clerks and examiners. The agency reported regularly to Congress, providing highly detailed annual reports. Concerned about the safety of the bank-issued currency, the examiners valued assets according to the market and were charged with prompt closure in the event that they discovered a bank to be insolvent. Yet, state banks did not all join the National Banking System, as they recoiled at the prospect of meeting the new, tougher, national bank regulations. In 1865, Congress responded to this resistance by imposing a 10 percent tax on state banknotes, a vital means for funding their loans. Although state banks did not completely disappear, national banks became the dominant financial institutions. This victory is seen in Figure 1.1, which reports the number of banking institutions by charter type. By 1870, there were 1,612 national banks and a mere 325 state banks, with national banks holding 88 percent of all bank assets.9 For the next two decades, national banks maintained their preeminence across the country. Although a few states permitted some limited type of branching, all national banks were unit or single-office banks. Because of the National Banking Act’s requirement that a national bank’s “usual business shall be transacted at an office or banking house located in the place specified in its organization certificate,” the second comptroller of the currency ruled in 1866 that branches were forbidden.10

“To Establish a More Effective Supervision of Banking”

11

100 Trust Companies 80 State Banks 60

40 National Banks 20

0 1886

1890

1894

1898

1902

1906

1910

Figure 1.2. Shares of banking assets. Sources: Carter et al. (2006, Series Cj204, Cj213, and Cj150); White (1983, pp. 12–13).

I.B. The Reemergence of State Banks The strict regulatory regime imposed on national banks created incentives for innovations that would slowly undermine the National Banking System. Given that states retained their authority to charter financial institutions, they played a crucial role in these developments, which would have profound consequences for the stability of the financial system. Over the course of the next several decades, individual states, seeking to expand the number of financial institutions within their borders, revised their bank legislation. They engaged in “competition in laxity” by lowering their minimum capital, reserve, and lending requirements, which resulted in a “regulatory arbitrage” with institutions searching for the least onerous regulations.11 National banks found two new competitors, typically one type on the agricultural frontier – state banks – and the other in the growing urban centers – trust companies. State-chartered banking systems began to revive in the third quarter of the nineteenth century when states revised their banking codes and set regulations that were considerably weaker than those governing national banks. Most importantly, the states substantially lowered the minimum capital needed to open a bank.12 Combined with the general prohibition of branch banking and economic growth, these regulations encouraged a rapid

12

Eugene N. White 100 90 80 SB Deposits/Liabilities

Percent

70 60 NB Deposits/Liabilities 50 40 30 20 10

SB Notes/Liabilities

NB Notes/Liabilities

0 1834 1840 1846 1852 1858 1864 1870 1876 1882 1888 1894 1900 1906 1912

Figure 1.3. Sources of funding for state banks and national banks. Source: Carter et al. (2006, Series Cj159, Cj160, Cj155, Cj170, Cj156, Cj204, Cj227, Cj209, Cj209, Cj210, Cj235).

increase in the number of very small banks throughout the country, particularly in the Midwest and West. State banks eventually outnumbered the national banks by the late 1890s and then experienced an explosive growth in the early twentieth century, garnering an increased share of assets, as seen in Figures 1.1 and 1.2. They were able to expand because they developed deposit banking more quickly than national banks could. Figure 1.3 shows that before the Civil War, banknotes and deposits played roughly an equal role in funding state bank lending, with capital representing the other major share. By the 1850s, deposit banking was gradually moving ahead of the use of banknotes. The national banks seemed to follow these trends; but the state banks, forced by the 1865 tax to abandon note issue, focused on deposit creation, reaching a share of deposits to liabilities that national banks matched only in 1900. The constraints on national banknote issue, notably the declining number of outstanding U.S. government bonds, kept banknotes’ shares of funding very modest.13 The result was that the national bank regulations which “insured” banknotes protected a smaller and smaller share of bank customers as “uninsured” deposits soared, opening the potential for greater losses in the event of a failure. The rapid growth of single-office banks led to the emergence of two serious weaknesses in the American banking system that did not appear in Canadian or European banking systems where branching was permitted

“To Establish a More Effective Supervision of Banking”

13

and entry often limited by high initial minimum capital requirements: (1) Many small American banks’ loan and deposit bases were relatively undiversified so they were more vulnerable to shocks and prone to failure and (2) an elaborate correspondent banking system was necessary to link these banks with national money markets and facilitate the clearing of checks. The correspondent system was further encouraged by high federal and state reserve requirements that permitted banks to hold their part of their reserves outside of their own vault and in accounts at major city banks where they earned interest. These large pools of liquid funds, concentrated primarily in New York, were invested by city banks in call loans to the stock market. The financial system was thus integrated but potentially fragile. Panicked country banks could withdraw their deposited funds from money centers, stressing city banks and financial markets. This fragmentation of the banking system and the correspondent banking system thus played key roles in nineteenth-century panics, including that of 1907.

I.C. The Rise of the Trust Companies The second class of institutions that challenged the National Banking System was the trust companies, permitted to combine traditional banking with trust operations, which were banned to national banks. These institutions opened under state charters with regulations that were often weaker than the regulations governing state banks.14 They prospered in the financial centers, where they became tough competitors of national banks.15 Trust companies grew at an astonishing rate in New York State. In 1897, national banks had $915 million in assets, state banks $297 million, and trust companies $396 million in New York.16 However, only a decade later, in 1907, while national banks had grown to $1.8 billion and state banks to $541 million in assets, trust companies had expanded to $1.4 billion in assets. The effects of the new entrants on the banking system are visible in Figures 1.1 and 1.3. The number of trust companies rose quickly in the last decade of the nineteenth and the first decade of the twentieth centuries. As they were larger urban institutions, they did not challenge national and state banks in sheer numbers, but they did garner a larger share of all bank assets. Although their share was just over 20 percent nationally, they were more important in financial centers. Furthermore, they operated outside the federal regulatory system, sometimes with only modest state oversight. The fragmented system of unit banks, tied together by correspondent balances, could turn a banking panic by the public into a banker’s panic. In

14

Eugene N. White

spite of this fragility, there was no central bank. Central banks were formed in most developed countries by the end of the nineteenth century; but the political economy of the United States prevented the establishment of one. The problems created by the absence of a central bank were highlighted by the discussion about the “inelasticity” of national banknotes or what today would be termed the inability to expand high-powered money or currency to the financial system in response to crises. National banknotes were tied to the outstanding stock of government bonds, and they did not increase when there was a flight to cash during seasonal stringencies, common in this period, and financial panics. The private market provided only a partial solution to the absence of a central bank by its temporary issue of clearinghouse loan certificates in crises. These instruments, backed by the assets of individual banks, gave some additional liquidity, especially to the clearing operations between banks. However, to participate, banks had to be members of the clearinghouses. In these institutions, banks needed to trust one another completely because of the large balances that they built up against one another before settlement. To alleviate concern, they monitored each other more closely and frequently than did the official federal and state regulators. Thus, they had particularly good information on the solvency of their members. During panics, the creation of high-denomination clearinghouse loan certificates allowed banks to replace coin and banknotes in their clearinghouse exchanges, enabling them to pay out more cash to their nervous depositors. Their success eventually led to the issuance of small-denomination notes that circulated with the public briefly in times of crisis.17 Clearinghouse loan certificates helped to alleviate earlier panics, but the panic of 1907 reflected the most recent evolution of the financial system: the emergence of the trust companies as a “shadow” banking system, competing with but not fully accepted by their more heavily regulated rivals, the banks. As they did not conform to the same regulations as the banks, they could have only corresponding membership in a clearinghouse, as the banks were dead set against allowing these rivals to gain direct access to its services. The panic of 1907 exploded when the Knickerbocker Trust Company, a major New York financial institution, approached the New York Clearing House for assistance. When it was denied, the news prompted a run on all trust companies that eventually spread to the banks. The intervention neither by J.P. Morgan nor by the U.S. Treasury proved sufficient to halt this panic, which was contained only by the declaration of a suspension of payments by banks.18

“To Establish a More Effective Supervision of Banking”

15

I.D. Panics The obvious concern of reformers in the latter years of the National Banking era was the frequency and severity of banking panics, which occurred at both regional and national levels. Two studies commissioned by the National Monetary Commission, established after the panic of 1907, examined these crises. O. M. W. Sprague’s volume focused on the crises of 1890, 1893, and 1907, but Edwin W. Kemmerer’s close reading of the Commercial and Financial Chronicle uncovered eight major and twenty-one minor panics.19 Subsequent scholarship has identified five major banking panics between 1864 and the founding of the Fed: 1873, 1884, 1890, 1893, and 1907.20 Some of these panics did not degenerate into a full-blown scramble for liquidity, and 1884 and 1890 are sometimes classified as “incipient panics” or moments of “financial distress.” It is widely agreed that these panics had severe consequences for the economy. Pointing to six recessions that hit the United States between 1895 and 1912, Allan H. Meltzer concluded that “financial panics, interest rates temporarily at an annual rate of 100 percent or more, financial failures, and bankruptcies were much too frequent.”21 He emphasized that other countries did not suffer the financial trauma that afflicted the U.S. economy because they had lenders of last resort. Christina Romer determined that the pre–Fed era, compared with the post–World War II era, had greater volatility in the gross national product (GNP), industrial production, commodity output, and the unemployment rate.22 Although the length of recessions was about the same, expansions were considerably longer in the post–World War II period. In general, she found that the frequency and severity recessions were greater before the founding of the Fed. For the years 1890–1908, Jeffrey Miron reported that panic years had substantially lower real GNP growth than nonpanic years and that the elimination of major panics after the founding of the Fed shortened the length of recessions from 17.5 to 14.25 months.23 Most recently, Andrew J. Jalil confirmed that panics independently and significantly diminished the growth of output and contributed to deflation. More important, recessions with major banking panics were more severe and longer than recessions without panics.24 Although most scholars would agree that the severity of the 2007–9 recession was amplified by the panic of 2008, confirming the pre-1914 pattern, the pre–Fed panics were different. In contrast to the most recent financial collapses, including the banking and savings and loan disasters of the

16

Eugene N. White

1980s and the crisis of 2008, the panics of the National Banking Era were primarily liquidity events rather than solvency events. The scramble for liquidity contributed to the severity of economic downturns, but the panics were not driven by widespread bank insolvencies, nor did they create insolvencies that ultimately led to substantial losses to shareholders, depositors, and other stakeholders. As for liquidity events, the banking industry was subject to massive withdrawals of deposited funds by the public and corresponding banks, often leading to a suspension of payments. However, unlike contemporary events, the solvency of the entire banking industry or even a significant fraction was not ultimately in question. In this pre-1914 world, bank regulation and supervision played a crucial role in setting the incentives that limited solvency problems.

II. Supervision, 1864–1913 II.A. Disclosure Of the three basic components of bank supervision – disclosure, examination, and disciplinary action – disclosure was emphasized under the National Banking System. The emphasis on disclosure reflected the belief that market discipline was the best means to ensure the soundness of banks. Because there was detailed proprietary information that could not be revealed to the public, examinations also played a key role in ensuring bank solvency; but, apart from fines for late disclosure of information, the only disciplinary actions a comptroller could take were the revocation of a bank’s charter and the declaration of a suspension with appointment of receivers to liquidate the bank. The OCC was established to ensure compliance with the new federal regulations.25 Initially, national banks were required to provide a detailed quarterly report and a very limited monthly statement. The fixed dates and absence of auditing permitted banks to engage in “window dressing.” In 1869, Congress responded to comptrollers’ complaints and instituted “call” reports of condition to improve disclosure. National banks were required to provide the comptroller with five call reports per year with information on their balance sheets; and three of these were set on dates randomly chosen by the comptroller to limit opportunities for manipulation of banks’ books. Every day’s delay in delivery of the call report was subject to a $100 fine.26 These reports appear to have provided significant information, as they were valued by the financial industry. Banks in reserve cities requested the comptroller to publish more frequent information from the call reports so

“To Establish a More Effective Supervision of Banking”

17

that they could better follow the changing condition of their corresponding banks in the countryside.27

II.B. Examination and Discipline Although providing information on the general condition of a bank, the call reports were aggregate by nature, and examination was thus a vital component of supervision for detecting fraud and ensuring compliance with regulations by delving into details of bank operations that would be deemed proprietary and inappropriate for public disclosure. The initial purpose behind the examinations was to ensure that banks would be able to redeem their banknotes upon presentation. Examinations were conducted from the “bottom up,” in which examiners scrutinized the cash, assets, and accounts of the bank to ensure that they complied with the letter of the law.28 Although comptrollers sent examiners instructions for examination, they emphasized that there could be no “cast-iron rules covering minute details.”29 The Comptroller of the Currency was charged with performing a minimum of two examinations per year for all national banks. These examinations were to be unannounced so the bank officials had no opportunity to hide any problems. Examiners were appointed by the comptroller with the approval of the secretary of the Treasury. Their compensation was set by statute, as were the assessments levied on banks to fund their payment. Out of these fees, examiners had to pay for expenses and their assistants. Beyond these very basic tenets, no limits were set on the number of examiners who could be appointed; and there were no fixed terms of office, oath of office, bond, or geographical districts, leaving the comptroller with considerable discretion for deployment of his examiners.30 However, the comptroller’s ability to increase disclosure by printing more frequent call reports or managing examiners by expanding his staff were limited by the fact that his office in Washington, DC, was funded by congressional appropriation and increases were rarely forthcoming. National bank examiners were paid a fee for examining a bank, based on the bank’s capital. Before the 1875 amendment to the National Bank Act, an examiner was paid five dollars for each day of examination and two dollars for every twenty-five miles he traveled by the examined bank.31 As large city banks were bigger and more complex, compensation for examiners of banks in reserve cities was determined by the secretary of the Treasury on the recommendation of the comptroller. A schedule of assessments for banks was set to fund these fees. For banks outside the established reserve cities, the

18

Eugene N. White

examination fee ranged from twenty dollars for a bank with capital under $100,000 to seventy-five dollars for a bank with a capital over $600,000. Although examinations were supposed to be unannounced, the surprise was often compromised by the predictability of an examiner’s travel plans, given his efforts to minimize cost, enabling banks to prepare for a visit if forewarned. Comptrollers regularly complained about the incentive effects of these legislated fees. In hearings before the House Committee on Banking and Currency, Representative Joseph W. Babcock reported that cashiers in a chain bank that had been visited by an examiner would telegraph to the other member banks, “the examiner here to-day; lookout, he is coming.”32 In the 1901 Annual Report, comptroller William Ridgely recommended the payment of fixed salaries to examiners instead of fees.33 Ridgely reiterated his complaint in his report of 1906.34 However, Congress remained immune to these pleas throughout the National Banking era. The third and last element of supervision, disciplinary action, was limited to the revocation a bank’s charter. Consequently, prudential supervision was circumscribed, with the comptroller relying on moral suasion to induce the directors and officers of a bank to correct violations.35 Although many comptrollers emphasized the importance of discussing the principles of good management with bank officials and requesting correction of problems, the Office of the Comptroller did not accept responsibility for a bank’s mistakes. Comptroller Knox wrote, It is scarcely to be expected, if a robber or a forger is placed in control of all its assets, that a national bank can be saved from disaster by the occasional visits of an examiner.36

II.C. The Operation of the Office of the Comptroller of the Currency Supervision by the OCC was thus intended to reinforce market discipline, primarily through disclosure requirements and surprise examinations, by which assets were marked to market and prompt closure was enforced for insolvent institutions. A review of the extant records of the OCC reveals a remarkably light supervisory hand, as one might expect in such a regime. Table 1.1 presents some basic numbers about national banks and the OCC from 1884, when the first budgetary data were reported, to 1913.37 The second to fourth columns of Table 1.1 report the number of national banks and their nominal and real assets.38 The more rapid growth after 1900 reflects effects of the Gold Standard Act of 1900, which lowered the minimum capital requirement, permitting entry of considerably smaller institutions. At

“To Establish a More Effective Supervision of Banking”

19

Table 1.1. The Operation of the OCC 1884–1913 Number of Real Assets Assessments for Real National Assets ($ (1914 $ Examinations Assessments Number of Banks per Year Banks Millions) Millions) (Current $) (1914 $) Examiners Examiner 1 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913

2 2,625 2,689 2,809 3,014 3,120 3,239 3,484 3,652 3,759 3,807 3,770 3,715 3,689 3,610 3,581 3,582 3,731 4,163 4,532 4,935 5,330 5,664 6,047 6,422 6,817 6,886 7,138 7,270 7,366 7,467

3 2,283 2,422 2,475 2,637 2,731 2,938 3,062 3,113 3,494 3,213 3,422 3,471 3,354 3,563 3,978 4,709 4,944 5,674 6,007 6,285 6,653 7,325 7,781 8,472 8,710 9,365 9,892 10,378 10,857 11,032

4 2,908 3,134 3,282 3,461 3,578 3,946 4,142 4,166 4,625 4,246 4,689 4,799 4,596 4,882 5,391 6,324 6,487 7,266 7,524 7,554 7,820 8,582 8,838 9,095 9,405 10,113 10,410 10,968 11,035 11,137

5 99,642 107,782 107,273 110,220 121,778 130,726 136,773 138,969 161,984 162,445 251,967 238,252 237,804 222,859 225,445 244,904 259,165 277,816 307,297 324,599 346,895 388,307 396,766 425,158 429,398 510,928 524,039 492,269 526,170 556,210

6 126,941 139,460 142,266 144,627 159,512 175,560 185,016 185,959 214,441 214,669 345,236 329,479 325,830 305,354 305,521 328,896 340,068 355,758 384,897 390,148 407,725 454,962 450,678 456,446 463,675 551,713 551,464 520,239 534,795 561,493

7

8

30

108

75 74 76 83 91 100 114

60 67 70 68 66 64 60

113a

64

a

Estimate based on the average salary of an examiner and total salaries. Source: U.S. Comptroller of the Currency (1884–1913).

the same time, the size of large-city institutions was expanding, so that even with the growth of small banks the average size of a bank was 20 percent larger in real terms after 1900 compared with that of the previous decade. Columns 5 and 6 show the total nominal and real assessments levied on the banks to cover the salaries and expenses of examiners. Unfortunately, the number of examiners in column 7 is much rarer information; but taken with

20

Eugene N. White

the data on revenues to pay their compensation, the pattern shows a very modestly sized agency growing with its expanding industry. In 1889, there were only 30 examiners for 3,239 national banks. The number of examiners grew, increasing from 75 in 1902 to 114 in 1908. Yet, this remained a small staff, as the number of national banks increased from 6,007 in 1902 to 8,710 in 1908. Examiners had considerable independence in conducting their examinations and hiring and paying their assistants, though the number of these individuals was apparently never reported to the comptroller.39 Nevertheless, given the fixed compensation it is unlikely that the number of these assistants grew relative to the number of examiners over time, and thus the number of examiners and assessments represents a reasonable proxy for the effort expended in examination. The independence of examiners seems to have left some comptrollers uncomfortable, and there were frequent requests to Congress to amend the National Bank Act to gain control over the appointment of assistants, set fixed salaries for both examiners and assistants, and require them to post bonds while taking an oath of office.40 However, there was no response from Capitol Hill. Even though the number of banks per examiner fell from over 100 in 1889, it remains to modern eyes surprisingly high. For the period 1902–11, it varied between 60 and 70 banks per examiner. At two required visits a year, each examiner would have to be performing 120 to 140 examinations per year. Given the time needed for travel between banks in the rural parts of the country and given the size and complexity of large-city banks, these reports could not have been exhaustive. Furthermore, bank assets appear to have been growing faster than the number of examiners. These assets rose from $80 million per examiner in 1902 to $91 million in 1911, suggesting that visits to a bank required more and more work. In spite of this workload, national bank examiners seem to have been relatively well compensated. The average compensation for an examiner for 1902–11 was approximately $4,307. Even if his expenses consumed up to half of this sum, his pay would appear to be good in comparison with that of the deputy comptroller, who was paid a salary of $2,500, the clerks in the OCC’s Washington, DC, office who received $900 to $1,800, or perhaps even the comptroller who earned $5,000.41 At the same time, the average cost of an examination remained relatively low for a bank, rising from about forty to seventy dollars per year during this fifty-year period. Relatively little was written about the actual work of national bank examiners, given the confidential nature of their activities. One very knowledgeable banker, James B. Forgan, president of the First National Bank of

“To Establish a More Effective Supervision of Banking”

21

Chicago, believed that examiners had good intelligence about the state of a bank: A competent examiner – and there are many such now in the government employ – while he cannot pass judgment on all the loans in a bank, can, after a careful examination, or series of examinations, form a wonderfully correct judgment as to the general character of its assets and as to whether its management is good or bad, conservative or reckless, honest or dishonest.42

Yet, examiners were not always this well received. The comptroller reported that although “some bank officers look upon the work of the examiners as unnecessary and inquisitorial,” the “benefits derived from the visits of a competent examiner are, as a rule, fully appreciated by the managers of the associations.”43 The comptroller found that this “friendly attitude” was more common among banks “possessing ample capital and transacting an extended business” in which the managers would not be able to personally supervise all the details of the business. Therefore, an independent knowledgeable outside review was sometimes appreciated in these circumstances. Examinations, except when fraud or insolvency was discovered, remained advisory. Forgan summarized the value of an examination: Examinations, as they are now conducted, have a most beneficial influence on bank management, especially by way of restraint. The correspondence carried on by the Comptroller, based on the examiners’ reports, does an inestimable lot of good in the way of forcing bank officers to comply with the law and in compelling them to face and provide for known losses as they occur. Supervision by examination does not, however, carry with it control of management and cannot, therefore, be held responsible for either errors of judgment or lapses of integrity. Examination is always an event after the act, having no control over a bank’s initiative, which rests exclusively with the executive officers and directors, and depends entirely on their business ability, judgment, and honest of purpose.44

Well aware that the examination process gave him only an advisory role, the comptroller appears to have become much more anxious after the panic of 1907. Attempting to pressure bank directors to become more attentive to the problems of national banks, the comptroller ordered examiners to ask directors a set of fixed questions. When Philadelphia’s national bank examiner, Frank L. Norris, confronted the directors of the Philadelphia National Bank in 1908, they were angry and resentful. Questions included “How many of the Directors know the conditions of the Bank in all its details?” “How many know nothing at all about the condition of the Bank?” “Have the Directors full knowledge of the habits and general standing of the Bank’s employees?” The president of the bank answered them perfunctorily

22

Eugene N. White

and protested the right of the comptroller to interrogate its directors and officers.45 The comptroller’s frustration appears to have been increasing on the eve of the passage of the Federal Reserve Act, even though there was no apparent increase in the number of insolvencies or their costs. The 1912 Annual Report reported an effort by the comptroller to more forcefully engage the directors of national banks in the process of examination.46 The comptroller reported that for years the agency had been urging directors to create examining committees of their own to look into the affairs of their banks at stated intervals in order to supplement the work of the national bank examiners.47 The comptroller complained that directors did not understand the nature of examinations and were unfamiliar with “the proper methods of verifying many assets and liabilities of the bank.”48 To remedy the situation, the comptroller sent a circular to all national banks on July 9, 1912, directing them to ensure that the directors would be able to provide national bank examiners with the best possible assistance at the next examination. In spite of these efforts by the comptroller, the system of supervision set up by the National Bank Act remained largely unchanged until the Federal Reserve Act.

II.D. The Supervision of State Banks Like regulation, supervision of state-chartered banks was generally lighter than the supervision experienced by national banks. In some states, the creation of a state bank commission or state banking board occurred only after the panic of 1907. In general, the federal system of supervision came to be regarded as best practice, and the slow drift to adopting rules similar to those for the Comptroller of the Currency sped up after 1907. While state banks were generally subject to weaker supervision than national banks, trust companies largely escaped state supervision until after 1907. In New York, for example, the superintendent of banking did not have the power to take possession of a failing trust company until 1907, although he had been given this power for state banks in 1892.49 For most of the National Banking era, state banks were required to provide one or two reports per year on their condition to a designated state official on specified dates. Toward the end of the period, states revised their laws on disclosure and followed the OCC’s practice of requiring reports on days that were not known in advance by banks.50 Yet, even in 1910, the number of required reports was generally fewer than the comptroller’s five. Only nine states required five annual reports, twenty-two required four, nine required

“To Establish a More Effective Supervision of Banking”

23

two, four required three, and one required one report. In all cases, state law required that the reports be published in local newspapers. After the panic of 1907, more states came to regard the comptroller’s rules as best practice and several began making “call reports” on the same days as the OCC.51 Regular examinations began to become the norm for state banks in the latter years of the nineteenth century. Before 1887, only New York, Indiana, Minnesota, California, and Iowa required regular examinations, though five additional states could make examinations if they believed that a bank was in trouble.52 By 1910, forty-one states and territories authorized regular examinations of state banks, though 20 called for just one annual examination. Some of these states set up regular examinations only after the panic of 1907. For example, California had no state supervision by examination until 1909, when responding to public concern and a plan by the clearinghouses to institute their own examinations, the legislature empowered the state supervisor to conduct them.53 In almost all states, examiners were paid a fixed salary, in contrast to the fees for national bank examiners that so troubled the comptrollers. Most of the funds to compensate state examiners came from fees assessed on the banks. Like those of the comptroller, state bank supervisors’ powers over their chartered banks was circumscribed and generally limited to taking action only when a bank was determined to be insolvent. Although the comptroller gained no additional authority immediately after the panic of 1907, state legislatures granted more powers and discretion to state bank supervisors. By 1920, fifteen states gave bank supervisors the authority to “direct the discontinuance of unsafe and unauthorized practices.”54 A post-panic New York special commission was blunt: Under existing law, he [the superintendent of banks] may criticize objectionable practices when they come to his knowledge, and report continued delinquencies to the attorney-general. His criticism is hence in large measure academic and may be given scant consideration by delinquents. . . . Were he clothed with the powers to “direct the discontinuance of unsafe practices,” no institution would dare continue the same after having been admonished by him.55

Interestingly, no ability to fine or otherwise discipline banks, short of closure, was considered. Even in this respect, state bank supervisors had less power than the Comptroller of the Currency. A comptroller had the power to appoint a receiver for a national bank, but a state bank examiner had to apply to the courts to appoint receivers for state banks and trust companies. Only after the panic of 1907 did some states begin to follow the national bank practice.56

24

Eugene N. White

The minimal data that are available for the state banking authorities suggest that the resources available for state supervisors were more modest than those at the disposal of the comptroller. The OCC’s 1911 Annual Report provided a brief survey of the state systems of examination. According to this report, there were a total of 224 state bank examiners for all states compared with the 113 national bank examiners. However, there were far more state banks, numbering 17,913,57 implying that there were 80 state banks per state bank examiner, significantly more than the 64 banks per national bank examiner. This higher workload was not accompanied by higher compensation. State bank examiners were paid an average of $2,300 compared with the average national bank examiner income of $4,356.58 Overall, although national bank supervision may have exercised a light hand, state surveillance of state-chartered banks and trust companies was minimal in the third quarter of the nineteenth century and then slowly began to move toward the OCC’s model of supervision.

III. Consequences and Costs of Supervision, 1864–1913 III.A. The Role of Double Liability To evaluate the effectiveness of supervision in the National Banking era, there is one additional important feature of bank regulation – double liability – that needs to be discussed but that has hitherto received relatively little attention.59 Concerned about the incentives for shareholders, Congress imposed double liability on the shareholders of national banks in the National Banking Act. In discussion of the act, Senator John Sherman emphasized that the purpose of the rule was not only to provide greater protection from loss in the event of a failure but also to provide the appropriate incentives to shareholders. He argued that in addition to giving security to creditors, double liability “tends to prevent the stockholders and directors of a bank from engaging in hazardous operations.”60 Under the national banks’ double liability rule, if a bank failed, shareholders at the time of failure could be forced to pay an assessment up to the par value of the stock in order to compensate depositors and other creditors. This regulation provided a strong incentive to owners to check the risk-taking activities of bank management. If a bank were faring poorly, the directors of a bank had the statutory right to vote to voluntarily liquidate the bank while it was still solvent, enabling them to protect shareholders from assessments. Consequently, to understand the operation of bank

“To Establish a More Effective Supervision of Banking”

25

supervision, one needs to look at both insolvencies and voluntary liquidations – phenomena that have been overlooked. Like many other dimensions of bank regulation, the assignment of liability to shareholders of state banks evolved over the National Banking Era. In 1870, twelve states imposed single liability, eighteen states double, and the remainder had no law or statutes that were ambiguous.61 By 1900, only five states did not fix shareholder liability, eleven states elected to impose single liability, and thirty-two states chose double liability. However, there was a notable difference between the states’ assignments of liability and the National Banking Act. Although the comptroller had the right to impose an assessment on shareholders to pay out depositors, creditors of failed state banks, in most states, had to pursue shareholders in court, making collection of funds more costly and difficult. Furthermore, this state liability was enforceable only after the assets of the bank had been exhausted. In contrast, a national bank receiver could impose an assessment and begin distribution to depositors before a bank was finally closed.62 Given this generally weaker liability regime for state banks, it would be expected that the constraints on risk taking would be reduced and it would be less likely that a troubled bank would voluntarily liquidate before it became insolvent.

III.B. Voluntary Liquidations and Failures A comparison of national bank and state bank voluntary liquidations would be very instructive, but unfortunately there are no data on state bank voluntary liquidations. Consequently, analysis must be limited to national banks. Figures 1.4 and 1.5 display the number of national bank voluntary liquidations and insolvencies and the percentage of these relative to the total number of national banks. For the National Banking era, over four times the number of banks, 2,373, were voluntarily liquidated compared with 501 that were closed as insolvent. The total capital of voluntarily liquidated banks was $432.8 million and for insolvent banks it was $89.1 million.63 To the twenty-first-century eye, the surprising feature of these charts is the number of banks that were placed in voluntary liquidation compared with the number that were insolvent. It suggests that shareholders and directors were quite cautious. When directors voluntarily liquidated a bank, they appear to have been successful in picking the right moment, as few of these banks were subsequently found to be insolvent.64 When a bank was past saving, it was quickly closed by the authorities, revealing a readiness to close an insolvent bank, which stands in stark contrast to today, when closed banks often show substantial losses. Occasionally, the comptroller

26

Eugene N. White 160 140 120 Voluntary Liquidations

Number

100 80 60 40 20

Insolvencies

0 1864 1868 1872 1876 1880 1884 1888 1892 1896 1900 1904 1908 1912

Figure 1.4. The number of national bank voluntary liquidations and insolvencies 1864– 1913. Source: U.S. Comptroller of the Currency (various years).

erred and suspended a bank that was later found to be technically solvent and some of these were re-opened. Figures 1.4 and 1.5 also suggest that the worst years for national bank failures were the 1890s, when the number and percentages of insolvencies 4.00 3.50 3.00 Voluntary Liquidations Percent

2.50 2.00 1.50 1.00 0.50

Insolvencies

0.00 1864 1868 1872 1876 1880 1884 1888 1892 1896 1900 1904 1908 1912

Figure 1.5. The percentage of national bank voluntary liquidations and insolvencies 1864–1913. Source: U.S. Comptroller of the Currency (various years).

“To Establish a More Effective Supervision of Banking”

27

180 160 140 State Banks Number

120 100 80 Trust Companies 60 40 20

National Banks

0 1864 1868 1872 1876 1880 1884 1888 1892 1896 1900 1904 1908 1912

Figure 1.6. The number of bank insolvencies 1864–1913. Sources: Barnett (1911, pp. 187, 190, 193, and 195); U.S. Comptroller of the Currency, (1913, p. 104).

peaked. By the first decade of the twentieth century, insolvencies had sunk to the low levels preceding the nineties, whereas voluntary liquidations were considerably higher. Whether it was the chastising experience of the 1890s or the increased vigilance of examiners, thanks to the growth in their number seen in Table 1.1, the national banking system appears to have more carefully protected the interested of its depositors. What losses were experienced by insolvent banks? Figures 1.6 and 1.7 show the number and percentage of national banks, state banks, and trust companies declared to be insolvent in each year between 1864 and 1913. Unfortunately, there are limited data for failures of state banks and trust companies before the 1890s; however, failures of all types of institutions do appear to have been more frequent in the last decade of the nineteenth century than in the first decade of the twentieth, in spite of the severity of the panic of 1907. For the years for which there are comparable data, the percentage of state banks failing was greater than the percentage of national banks and the percentage of trust companies failing higher than both. There is a notable spike in trust company failures in 1907, reflecting their fate during the panic of 1907. These higher failure rates by state banks and especially trust companies suggest that they may have been taking greater risks. Because of the regulations that promoted smaller, less diversified state banks, they might be expected to have a higher failure rate and a higher loss rate. They might have

28

Eugene N. White 7 6

Percent

5 4 3 Trust Companies 2 State Banks 1

National Banks

0 1864 1868 1872 1876 1880 1884 1888 1892 1896 1900 1904 1908 1912

Figure 1.7. The percentage of bank insolvencies 1864–1913. Sources: Barnett (1911, pp. 187, 190, 193, and 195); U.S. Comptroller of the Currency (1913, p. 104); Carter (2006, Series Cj203 and Cj212).

protected themselves by increasing their capital relative to their assets to offset this danger. Figure 1.8 graphs the capital-to-asset ratio for state banks and national banks from 1834 until the creation of the Federal Reserve. This longer time frame provides a necessary perspective on the risk exposure of these institutions. Unfortunately, trust companies did not provide reports in most states that would permit an aggregate national capital-to-asset ratio to be measured for this group of intermediaries. Figure 1.8 captures the well-known, long downward trend of the capitalto-asset ratio over the course of the nineteenth century.65 This key ratio hovered around 40 percent for state banks in the late antebellum period, though it was higher in the 1840s. Visually, the national bank capital-toasset ratio appears to continue the trend for pre–Civil War state banks, ultimately falling to around 20 percent. The surprise in Figure 1.8 is the much lower capital-to-asset ratio for state banks in the late nineteenth century, suggesting that they were taking greater risks than national banks and perhaps gaining higher returns. This assessment probably understates their risk taking as their loan portfolios were not as constrained as national banks in most states, permitting them, for example, to lend extensively on real estate, enabling a greater maturity mismatch. If local banking markets had been fully competitive, one would not expect to see the persistently lower capital-to-asset ratio and higher failure rates for state-chartered institutions.

“To Establish a More Effective Supervision of Banking”

29

70 National Banking System Founded

60

Percent

50

Federal Reserve Founded

40 National Banks 30 State Banks 20 10 0 1834 1842 1850 1858 1866 1874 1882 1890 1898 1906 1914 1922

Figure 1.8. Capital- to-asset ratios for national banks and state banks 1834–1929. Source: Carter et al. (2006, Series Cj150, Cj 157, Cj159, Cj175, Cj204, Cj211, Cj213, and Cj236).

However, local banking markets, especially on the frontier, had strong elements of local monopolies. The lower minimum capital requirements of state banks permitted the creation of very small institutions in small towns where a national bank could not set up shop.66

III.C. The Costs of Bank Failures The record of costs from the closure of national banks looks quite favorable relative to the experience of the Great Depression, the savings and loans crisis, or the 2008 financial collapse. Between 1865 and 1913, 540 banks were declared to be insolvent. This number is higher than the 501 insolvencies that were compared with voluntary liquidations in Figure 1.5 because 39 were restored to solvency. These 540 national banks had an initial capital of $86.8 million or roughly, on average, $160,000 each. The total assets of these banks on the date that were closed were $360 million, of which 35.9 percent were estimated to be good, 31.5 percent to be doubtful, and 18.9 percent to be worthless, with an additional 13.5 percent recovered since suspension. There were $28.6 million of offsets for these banks. The receivers recouped $183.9 million from the sales of assets and $22.5 million in assessments on shareholders or 89.5 and 10.5 percent, respectively, of the total collections. Tracking down and enforcing the assessments were

30

Eugene N. White

clearly difficult and costly. Receivers assessed shareholders $46.4 million but recovered only 48.6 percent of the total. The costs of closing these banks down were modest, with $5.4 million spent on legal expenses and $9.5 million spent on receivers’ salaries or 4.1 percent of assets and 7.8 percent of the proven claims. Payments to depositors and other creditors totaled $146.9 million out of $191.0 million of proven claims for a payout ratio of 76.9 percent. Total losses thus amounted to $44 million for this fifty-year period. The sum of $44 million represents a very modest cost, no matter how it is measured. In 2009 dollars, $44 million is approximately $1 billion, which would seem a relatively small sum for the cumulative losses. On average, losses were $880,000 per year. This was a tiny fraction of the deposits of national banks, which even in 1870 stood at $706 million in 1870, growing to $1,085 million in 1880 and $1,978 million in 1890.67 In terms of the gross domestic product (GDP), $880,000 and $44 million were 0.01 and 0.6 percent, respectively, of the GDP in 1870 and 0.006 and 0.3 percent, respectively, of the GDP in 1890. These total losses are on the same order of magnitude as the losses that Hugh Rockoff calculated for the free banking era from 1838 to 1860, a nominal $1.9 million or 0.01 percent of the GDP.68 However, they pale next to the losses experienced during the Great Depression. Friedman and Schwartz calculated that $2.5 billion were lost by depositors and shareholders during the years 1929–33, representing 2.4 percent of the contemporary GDP or $39 billion in 2009 dollars.69 One estimate of the costs of the savings and loan and bank failures of the early 1980s was $126 billion, or 3.4 percent of the contemporary GDP and $200 billion in 2009 dollars.70 One estimate of the losses from the 2008–9 collapse was $1.7 trillion or 11.6 percent of the 2008 GDP.71 One might argue that these aggregate numbers hid the pain of high losses in individual years, but that does not seem to be the case. Figure 1.9 reports the number of national bank failures in each year between 1865 and 1913 and the payout to depositors and other creditors. The decade of the 1890s, punctuated by panics and four recessions, stands out as the worst experience, but national banks then returned to the previous pattern of failures and payouts. Yet, even in the most extreme years of 1891, 1894, and 1896 when the payout ratios fell to 38.7, 42.1, and 46.7 percent, losses totaled $4.2, $2.2, and $3.6 million. Given that national bank deposits stood at $1,974, $1,939, and $2,141 million, losses were no more than 0.3 percent of all deposits. Unfortunately, the record of losses from the state bank insolvencies has not been well preserved. As previously discussed, state bank supervisors were

“To Establish a More Effective Supervision of Banking”

31 120

70 Percentage Payout

60

100

50 40 60 30

Percent

Number

80

40 20 Number of Failures 20

10

0

0 1865

1870

1875

1880

1885

1890

1895

1900

1905

1910

Figure 1.9. National bank insolvencies and payout ratios 1865–1913. Source: U.S. Comptroller of the Currency (1918). (Note: The payout ratio is calculated as the payments divided by the proven claims. If offsets are included, the recoveries by depositors would be higher.)

generally not given control over failed banks, and consequently few statistics on failures were collected by these officials. Barnett’s (1911) study for the National Monetary Commission found only four partial studies. Inference from these is difficult because no attempt was made to separate state, private, and savings banks, partly because definitions of these institutions varied from state to state. In 1879, the comptroller investigated the failures of state, private, and savings banks for the three previous years and found that 210 banks failed, with an average payout ratio of 66 percent.72 Trust companies were omitted from this and subsequent studies. The comptroller’s own records show that there were thirty-three insolvent national banks from 1876 to 1878, with payout ratios of 75.0, 98.5, and 91.3 percent for each year.73 In subsequent studies in 1895 and 1896, the comptroller found that 1,234 of state and private institutions had failed since 1863 and had paid out less than 50 percent of claims. For national banks over the same span of years, 330 failed, paying out an average of 67.7 percent of proven claims. In 1899, the comptroller examined state, private, and savings banks that had failed between 1893 and 1899 and determined the payout ratio to be 56.19 percent. Over this same interval, national banks paid out 74.5 percent. Barnett found one additional state study. In 1909, after reviewing the receivers’ reports, the secretary of the Nebraska State Banking Board

32

Eugene N. White

compiled some statistics for 1901–9. During this period, the average deposits of state banks totaled $50 million. The total deposits in failed banks were $451,557. These banks ultimately had $187,955 in unpaid claims, implying a recovery rate of 58 percent.74 Assessing this information is difficult, though it would appear that the stricter liability rules and supervision of national banks led to better payout ratios. The only econometric study of the effects of double liability was conducted by Richard Grossman, who found that double liability state banks were associated with lower risk taking and lower failures rates than single liability state banks.75 However, his work compares the effects of only state regulations; the effects of stricter national bank rules, in which the comptroller, not the courts, enforced the laws, may have had a more profound effect on the safety of banks in the national system. Nevertheless, the state systems did not result in much larger losses to depositors. Even though failures were more numerous, state banks were significantly smaller than national banks, so that total losses were probably around the same order of magnitude as those experienced by depositors of national banks. The addition of these losses to the totals for national banks would not alter the picture of very modest losses during the years 1863–1913 compared with subsequent crises.

III.D. The National Banking System: An Assessment The National Banking era receives a mixed review in most historical accounts. Although there was an expansion of banking services and an integration of money and capital markets, it is viewed as inherently flawed because of the large number of banking panics compared with the contemporary experience of other nations.76 However, it is important to understand the defects that contributed to these crises. First and foremost, the prohibition of branch banking created a system of thousands of unit banks, many of which had undiversified deposit bases and loan portfolios. To clear and collect checks and find better use for their seasonally fluctuating balances, the country and small-city banks relied on their correspondents in large cities. Federal and state reserve requirements that permitted these banks to keep more than half their reserve on deposit in interest-earning accounts with these same big-city correspondent banks added further to these sizable “bankers’ balances.” Investing heavily in the most liquid market of the day–call and time loans to the stock market – tied the fate of Main Street to Wall Street and vice versa. Many experts realized this key weakness of the system and its origin – the general prohibition on branching. However, calls for reducing the barriers to branching were met with fierce opposition

“To Establish a More Effective Supervision of Banking”

33

from many unit banks that feared being driven out of business by branches of large banks. In spite of the weakness of individual banks, regulation and supervision set the incentives and level of monitoring that ensured that losses from bank failures were very modest, if not minimal. The second deficiency of the National Banking era was more widely acknowledged: the absence of a central bank. Contemporary experts and most historians believe that a lender of last resort could have squelched most panics by providing credit to liquidity-constrained banks in the midst of a crisis. The identification of this problem points to the key fact that these panics were liquidity rather than solvency events. Although a few insolvent institutions might provoke initial bank runs, there were no system-wide failures or losses. Which of these two problems was preeminent? Many countries, such as the United Kingdom, France, and Germany, had both central banks and widespread branching. However, the Canadian experience offers a useful comparison. With a similar distribution of economic activity and seasonally fluctuating demands for credit, Canada had an economic structure similar to that of the United States. The critical difference was that Canada permitted nationwide branching but had no central bank until 1935.77 In contrast to the United States, Canada did not suffer from frequent, numerous bank runs and panics, pointing to a well-diversified integrated banking system as essential to limiting panics. This comparison suggests that the first priority of Congress should have been to reform the nation’s banking structure. The political economy of banking, however, made this impossible, and hence reformers focused on the establishment of the Federal Reserve.

IV. The Advent of the Fed IV.A. The Incomplete Reform of Regulation and Supervision Although the primary focus of the Federal Reserve Act of 1913 was the creation of a central bank that would be palatable to the banks and the public, the legislation did not attempt the needed restructuring of the banking system. The reforms of 1913 were not as bold as the reforms undertaken in 1863–4, leaving a good portion of the financial system under control of the states. During the Civil War, Congress was willing to take a stick to the financial system – a tax on state banknotes – to induce banks to join the National Banking System; but the act of 1913 and subsequent amendments offered mostly carrots to bring banks into the Federal Reserve System. The problem with the establishment of the Federal Reserve from the point of

34

Eugene N. White

view of regulation and supervision is that it set a precedent for patching up some of the problems and failing to fundamentally reform the system. The inability to do more directly contributed to the next great financial crisis of 1929–33. By setting up a central bank, the Federal Reserve Act created the potential for a conflict between the objectives of price stability and financial stability. The failure of the early leaders of the Fed to quickly recognize the implications of this conflict contributed to the waves of bank failures in the early 1920s and the Great Depression. In addition, by giving the Federal Reserve authority to supervise state member banks, the act did not make a decisive decision about whether supervision should be located inside or outside the central bank. Whereas before 1914 there had been one federal banking agency that squared off with the numerous state agencies, the Federal Reserve Act was the first step in the multiplication of federal regulatory agencies that have inherently conflicted interests.78 Concerned about effectively exercising their mandates, state and federal agencies in the 1920s gradually weakened regulation and supervision to induce banks to change their charter or member status, permitting “regulatory arbitrage.” Compounding these problems was a gradual reduction in transparency because the Federal Reserve banks were not official government agencies like the OCC and were not similarly subject to Congressional oversight. The overall philosophy of bank supervision that aimed at reinforcing market discipline continued to inform the activities of the OCC, but the Fed’s additional concerns about its discount operations slowly began to undermine it.

IV.B. Conflict between the Fed and the OCC: Examinations The Federal Reserve Act created two conflicts between the Federal Reserve Board and the OCC. First, although membership in the Fed of existing national banks was mandatory, the Federal Reserve Board had the right to decide on the membership application of any state-chartered banks. By granting a national charter to a bank, a comptroller could create a new member bank without the approval of the Fed. Second, the Federal Reserve Act did not contain any provision that required the comptroller to provide the Federal Reserve Board or a Federal Reserve bank with copies of national bank examinations.79 In his memoir of the Fed’s early years, The Formative Period of the Federal Reserve System (1925), Governor W. P. G. Harding reported that during his eight-year tenure, the comptroller granted charters for several new national banks against the recommendations of Federal Reserve banks and even granted national charters to state banks

“To Establish a More Effective Supervision of Banking”

35

that had been refused membership, thus ensuring that they automatically acquired it without the consent of the board. As the OCC was directed to examine national banks twice a year, the board deemed it unnecessary to have regular examinations, assuming that the OCC would furnish the Federal Reserve banks with copies of its examiners’ reports. However, the comptroller provided only limited reports, reserving the right to decide what information the Fed would receive. The Comptroller of the Currency in office at the time of the creation of the Fed, John Skelton Williams, fought to maintain the prerogatives of the OCC and had icy if not hostile relations with the Fed. When Williams’ five-year term as comptroller expired on February 2, 1919, President Wilson nominated him for another five-year term. But, several Senators blocked his nomination and no action was taken when Congress adjourned on March 4, 1919. Under the law, Williams could continue as comptroller until his successor was appointed. When the Sixty-Sixth Congress convened, the chairman of the Senate Committee on Banking and Currency refused to make a favorable report and no action was taken, even though there appeared to be a majority of senators who would have voted for his appointment. By the end of 1920, this appeared to be a permanent impasse and Williams decided to retire from office early in March 1921. During this time he became increasingly hostile to the Board. He attacked the amount of credit available to New York City member banks in contrast to those in the interior and argued for lowering interest rates.80 In these circumstances and given the escalating number of banks failures, it is surprising how little discussion in the literature there is about examination and supervision. W. Randolph Burgess (1927) mentions the supervisory activities of the Federal Reserve banks only twice in his authoritative book The Reserve Banks and the Money Market. In his analysis of the lending activities of these banks, he emphasizes that the safety of both rediscounts and advances is ensured by the quality of the collateral and the shortness of the credit. Losses to the Federal Reserve banks were thus “negligible.” Member banks were required to file statements of the customers who presented paper to the banks in excess of $5,000. To verify the quality of the collateral, the New York Fed maintained a file on 50,000 to 60,000 individuals and businesses. Examination gets short shrift: “They [the Federal Reserve Banks] maintain staffs of bank examiners who from time to time examine member banks and still more frequently collaborate with federal and state authorities in such examinations.”81 In Burgess’s view a key function of examination was to provide information to the Federal Reserve bank officers in charge of lending to members

36

Eugene N. White

to “prevent too constant or too large use of borrowing facilities.” In New York, this gave the Fed a list of banks whose conditions were not satisfactory so that loans to them could be scrutinized. Those banks that made use of Fed credit for an unreasonably long period were then subjected to special inquiries to determine the “necessity for the borrowing.”82 If borrowing was in excess of its capital and surplus, the bank was automatically placed on the list for inquiry.83 Burgess emphasized the importance of discretion in handling each of these cases. The example he gives is very telling because it is a very different from the comptroller’s view of the nature of supervision: Take as an example the perplexing problem of lending to a bank in the farming area of the Middle West in recent years. The First National Bank of Crestland is loaded with doubtful farm paper, much of it representing sometime equities in real estate. They bring all their good paper to the Federal Reserve Bank to rediscount. Shall the Reserve Bank take it and lend them the money? If the Reserve Bank refuses, failure may follow. If it makes the loan, it assumes the responsibilities of continuing in operation a bank probably insolvent. If failure should then come the depositors might find much of the good assets rediscounted at the Reserve Bank and unavailable to pay depositors. The Reserve Bank must consider not only the safety of its loan, but the interests of the depositors. Can the bank be saved by a loan? If not, will the depositors be better off under an immediate liquidation, or a later liquidation, when the bank may have dissipated many of its best assets? These are some of the questions the Reserve Bank has to face. The answer depends on a careful scrutiny of each bank, in constant cooperation with state and national supervisory authorities. (Burgess, 1927, pp. 236–7)

On this matter, the Fed ran into a conflict with the OCC, which believed that it should not disclose the information collected during examinations to the Fed, except where it might have direct bearing on its discount operations. What Burgess recognized was the discount window had created moral hazard for banks that borrowed. Following Walter Bagehot’s Lombard Street, a nineteenth-century European central bank might not have worried about the total condition of a borrowing bank, just whether it had good collateral. But the Fed apparently assumed responsibility for the depositors and creditors of the bank. This shift in the approach to supervision was abetted by the struggle between the Fed and the OCC for control of supervision. This contest arose because the Fed tried to appease state member banks who objected being subject to the same, more stringent regulations as national banks. Rather than coerce them to adhere to these regulations and risk their departure from the system, the Fed slowly weakened supervision for all national and state member banks. Just as state banks did not jump to become national banks after the National Bank Act of 1864, so too were few state-chartered banks

“To Establish a More Effective Supervision of Banking”

37

convinced of the benefits of Fed membership that they willingly submitted to its generally more rigorous regulations. The Federal Reserve Act had tried to bring in most institutions into the fold by a variety of changes, lowering reserve requirements and differentiating between demand and time deposits and giving trust powers to national banks.84 However, few state banks responded; and by 1917, only 53 of the over 19,000 state banks had taken out membership. Amendments to the act of 1913 made membership slightly more attractive to larger state institutions, and membership rose to 513 in 1918 and 1,042 in 1919, cresting at 1,648 in 1922. Although these state banks accounted for half of all state banks’ assets, there were still 19,000 outside the regulatory purview of any federal agency. At the same time, the number of national banks under the aegis of the comptroller slowly rose from 7,518 in 1914 to a peak of 8,236 in 1923. Needless to say, the Federal Reserve was very sensitive to any slippage in the number of its members. The most visible example of the effects of competition between agencies followed from the transfer of authority to request call reports from the OCC to the Fed and the resultant decline in the number of call reports. The Federal Reserve Act gave the Board of Governors the power to demand reports and examine member banks, but initially the OCC carried out examinations of state member banks in addition to national banks. When, in 1915, comptroller Williams asked for a sixth report and more detailed information, he provoked a flood of complaints.85 As a consequence of this uproar and the inequality between the requirements imposed on national and state member banks, the 1917 amendment to the Federal Reserve Act ordered state member banks to make their reports of condition to their Federal Reserve bank, setting the minimum number of call reports at three – not the five required of national banks. Furthermore, the power to set call dates was transferred to the Board. In 1916, the surprise call year-end call reports were abandoned.86 This regime shift was not completed until after Williams left office in late 1921. In 1922, the number of reports fell back to 5 – the number that had been requested continuously since 1870.87 Then, in 1923, it dropped to 4 and remained at that level for 1924 and 1925. There is no comment in the Annual Reports of the Comptroller of the Currency or the Federal Reserve Board or in the Federal Reserve Bulletin; but in 1926, there were only three call reports – one for April 12, June 30, and December 3. In 1927, the Board called for four reports, a number it adhered to in subsequent years. What is missing for 1926 is the report that was traditionally called for in October, a shocking omission as this is the most fateful month of the year for financial

38

Eugene N. White

crises. Why was it omitted? Although the answer may lie somewhere in the archives of the Federal Reserve, there are two possible reasons. The Fed could have been under pressure from the national banks to reduce their reporting and put them on a par with state member banks. If so, this is an example of the “competition in laxity” between state and federal regulators that had led to a reduction of capital and reserve requirements in the late nineteenth and early twentieth centuries and continues to bedevil contemporary American regulators.88 Alternatively, the Fed may have been alarmed by the condition of some banks, given that real estate values had begun to decline in the third quarter of 1926. The decision to skip the October call might have been made to give banks time to raise capital or make other adjustments. If this were the case, it represented a significant shift toward discretionary supervision. The record of bank examination reveals no such obvious deterioration, and in fact, the Federal Reserve Act adopted some of the recommendations that pre-1913 comptrollers had long requested. The payment of a fixed fee for each bank examined had caused the examinations to be less than a surprise, as examiners’ movements became more predictable as they sought to minimize travel costs. The new Federal Reserve regime eliminated this incentive and put examiners on a salary and paid their expenses. In addition, they were provided with paid assistants.89 The OCC also gained an increased ability to monitor its examiners, reorganizing its operations by Federal Reserve District and appointing a chief national bank examiner with responsibility for all examiners in the district.90 Although the comptroller was initially responsible for examining state member banks, the 1917 Amendment transferred this power to the Federal Reserve banks who organized their own examination departments.91

IV.C. Examination in the Early Fed Years The resources available to the comptroller under this new regime are shown in Table 1.2. The increase in resources devoted to examination is marked but probably reflects the need to manage the growing number of insolvencies. In 1915, the number of examiners was little different, perhaps a bit lower than in 1913. However, beginning in 1916, there is a steady rise, peaking in 1923 at 234 examiners, nearly double the number before the foundation of the Fed. As the number of national banks changed little, the number of banks per examiner declined to the mid-1930s. This change suggests that examiners had more time to spend per bank; however, it is unclear how

“To Establish a More Effective Supervision of Banking”

39

Table 1.2. The OCC and Bank Supervision 1914–1929 Number Number Real Nominal Real of of Banks per Nominal Assessments Examination Expenses Year Banks Examiners Examiner Assessments 1914 $ Expenses 1914 $ 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929

7,518 7,597 7,571 7,599 7,699 7,779 8,024 8,169 8,225 8,184 8,049 8,054 7,912 7,765 7,635 7,408

103 124 131 137 149 153 195 214 234 226 221 221 227 219 204

74 61 58 56 52 52 42 38 35 36 36 36 34 35 36

520,607 536,300 577,763 849,816 994,626 1,050,977 1,184,027 1,769,395 2,159,510 2,145,392 2,293,545

520,607 519,540 496,943 592,561 595,335 613,758 607,110 1,064,212 1,374,402 1,327,930 1,437,571

1,181,449 1,363,870 7,808,273 3,839,805 2,145,391 2,168,731 2,091,059 2,295,122 2,291,408 2,308,250 2,409,858

689,951 699,324 4,696,326 2,443,811 1,327,929 1,359,339 1,287,871 1,407,725 1,439,871 1,439,512 1,497,857

Source: U.S. Comptroller of the Currency (1914–1929).

much of this increase was associated with the higher number of bank failures in the early 1920s. There are two series representing the resources that are not completely compatible, nor do they overlap. The first is the traditional revenue from the fees or assessments from examinations, and the second is the reported examination expenses, which would include salaries and travel expenses of the examiners. Nevertheless, they show the same pattern, doubling in real terms from the years around the founding of the Fed to the mid-1920s, but again the published record is silent as to whether this represents any change in supervision activity or the management of failing banks. The Federal Reserve banks initially relied on the comptroller’s examiners and state supervisors examiners, only slowly creating their own examination staffs. The Federal Reserve Board created its own staff of examiners whose numbers are shown in Table 1.3. Conscious that state-chartered banks had no desire to be examined by two regulatory authorities, the Fed was given discretion to use the reports of state authorities. As a result, the relatively large number of banks per examiner does not imply that the Fed made less effort than the OCC to examine its banks.

40

Eugene N. White

Table 1.3. State-Chartered Member Banks and Federal Reserve Board Examiners 1915–1929 Year

Number of State Member Banks

Number of Examiners

Banks per Examiner

1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929

17 34 53 513 1042 1374 1595 1648 1620 1570 1472 1403 1309 1244 1177

6 5 11 14 20 18 18 22 22 21 21 18 17 17 18

3 7 5 37 52 76 89 75 74 75 70 78 77 73 65

Note: For 1918–1929, the number of examiners includes assistant examiners. Source: Board of Governors (1943, p. 22); Board of Governors of the Federal Reserve System, Annual Report (1914–1929).

V. Price Stability Versus Financial Stability V.A. The Unexpected Deflation Shock Against this backdrop of interagency conflict, the conflict between the twin goals of price stability and financial stability quickly emerged. When inflation began to pick up speed in 1920, the Federal Reserve responded by quickly raising interest rates.92 This action, often deemed to be an excessive response, produced a sharp recession and a significant deflation. There was no historical precedent that would have prepared the financial sector for this action. During the Civil War, the price level had more than doubled but the deflation that permitted the United States to join the gold standard at its prewar parity was gradual, lasting from 1866 to 1879. After World War I when the price level nearly doubled between 1914 and 1920, it tumbled 22 percent between 1920 and 1922.93 Banks that had been riding the wartime and postwar booms were hit hardest, with the small banks in agricultural areas suffering the most as commodity prices plummeted. Figure 1.10 shows the percentages of state and national bank failures and the inflation rates between 1865 and 1929. The unprecedented deflation was accompanied

“To Establish a More Effective Supervision of Banking” 30

4.5

25

World War I

4

20

3.5

15

3

10 2.5

Inflation 5

2 0 1.5

–5

State Banks 1

Inflation–Percent

Percent of Bank Failures

41

–10

National Banks

0.5

–15 –20

0 1866

1872

1878

1884

1890

1896 1902

1908

1914

1920

1926

Figure 1.10. Bank failures and inflation 1865–1929. Sources: U.S. Comptroller of the Currency (1914, p. 104); Board of Governors of the Federal Reserve System (1943, p. 283); Carter et al. (2006, Series Ca13).

by a lagged increase in bank failures, as its effects gradually brought a rise in farm foreclosures.94 During the years 1921–9, 766 national banks and 4,645 state banks failed. The number of national bank insolvencies was far greater than the 501 insolvencies of 1865–1913. Worse yet, the payout ratio for national banks dropped to 40 percent for a total loss of $565 million ($6.9 billion in 2009 dollars) or 0.8 percent of the 1925 GDP.95 The banking system might have recovered from this onetime shock, as it had recovered from the 1890s, with modest numbers of failures and relatively high payouts. However, the second great unanticipated deflationary shock of the Great Depression prevented a complete recovery. In addition, this onetime shock muddied the water for identifying other more subtle changes in policy.

V.B. Seasonal Interest Rate and Discount Policies Beyond this deflationary event, two permanent features of the Federal Reserve changed the incentives for banks to take risks: its seasonal interest rate policy and its discount policy. In the Federal Reserve Act, the Fed was charged with furnishing “an elastic currency.” For the Fed it was a central obligation to eliminate the seasonal strain in financial markets. As its first Annual Report emphasized, “its duty is not to await emergencies but by

42

Eugene N. White 14 Time Loans

January 1914

12 Commercial Paper

Percent

10 8 6 4 2 0 1890 1893 1896 1899 1902 1905 1908 1911 1914 1917 1920 1923 1926 1929 1932

Figure 1.11. Interest rates 1890–1934. Source: Board of Governors of the Federal Reserve System (1943, pp. 448–451).

anticipation, to do what it can to prevent them.”96 Miron (1986) documented that the Federal Reserve promptly carried out policies that reduced the seasonality of interest rates. Because panics occurred in periods when seasonal increases in loan demand and decreases in deposit demand strained the financial system, accommodating credit to seasonal shocks reduced the potential of a crisis. Comparing 1890–1908 and 1919–28, Miron found the standard deviation of the seasonal for call loans fell from 130 to 46 basis points, with the amplitude dropping from 600 to 230 basis points. The reduction of seasonality in interest rates lowered the stress on the financial system, leading Miron to conclude that it had eliminated banking panics during the period 1915–29. Most striking was the absence of a panic during the severe recession of 1920–1. Both the timing in the decline of seasonality and the role of the Fed have been challenged, but Miron’s basic results have been upheld.97 This seasonal interest rate policy is visible in Figure 1.11, which displays the time rate for stock market loans and the commercial paper rate. Before the establishment of the Fed, businessmen and their bankers had to be wary of the sharp seasonal fluctuations, with the sometimes accompanying financial panic. Once fears had been dampened, banks could more readily take risks, knowing that they would not be subject to the sharp interest rate spikes in the past. In this environment, a quick deflationary shock such as

“To Establish a More Effective Supervision of Banking”

43

that of 1920–1 or a longer one, such as that of 1929–33, would take a heavier toll on exposed financial institutions. In addition, a bank that found itself in trouble could, if it were a member of a Federal Reserve bank, borrow at the discount window. Banks in trouble now had a new option. Whereas before they could voluntarily liquidate or continue with the risk of ultimately becoming insolvent, at the discount window, they could borrow using some of their good assets and try to recoup their losses. The danger for depositors and other creditors was that such a bank would pay out less if it failed because it had pledged more good assets at the Fed’s window. This was the problem that Burgess identified, though he thought that the “careful scrutiny” of the regulators would be sufficient to prevent this outcome. Unfortunately, contrary to Burgess’s hopeful expectations, many banks became long-term borrowers at the discount window. In 1925, the Federal Reserve Board collected data on the indebtedness of member banks to their Federal Reserve banks. This information revealed that contrary to the expectations of the fathers of the Fed, the discount window turned out to have been more than a facility to increase short-term liquidity. On August 31, 1925, 593 banks had been borrowing for a year or more.98 Out of this total, 239 had been borrowing continuously since 1920. These banks appear to have been severely troubled institutions, and the Fed estimated that 80 percent of the 259 national member banks that had failed since 1920 had been “habitual borrowers.” This problem was confirmed in later years. Of the 457 banks that had been borrowing for more than a year in 1926, 41 suspended operations in 1927 and 24 liquidated voluntarily or merged.99

V.C. The Growing Cost of Bank Failures How did this regime affect the closing of banks and the losses to depositors and creditors? First, there was no change in the liability of national bank shareholders who were still subject to double liability. However, there was one significant change engineered by the act of November 7, 1918. Before this act, if two national banks wanted to merge, one had to be liquidated to permit the other to purchase its assets and assume its liabilities. Out of the numerous voluntary liquidations, very few banks chose this option. After 1918, the two banks could consolidate under either charter, subject to approval of the comptroller of the currency. This legal change gave national banks, which were previously prohibited from acquiring branches, a crude means to acquire them in states where branching was permitted.100 Whereas liquidations before 1918 had rarely been used as a means to consolidate,

44

Eugene N. White Table 1.4. Number of National Bank Closures 1921–1929

Year

Suspensions

Voluntary Liquidations

Consolidations and Absorptions

1921 1922 1923 1924 1925 1926 1927 1928 1929 Total

52 49 90 122 118 123 91 57 64 766

12 8 20 28 14 12 9 7 3 113

74 107 96 96 81 129 160 136 225 1104

Source: Board of Governors of the Federal Reserve, Banking and Monetary Statistics (1943, p. 52).

now banks that were weak might be tempted to combine with a stronger bank. This change in the menu of choices makes the comparison of bank closures before and after 1913 difficult. Table 1.4 provides data on national bank closures for the years 1921– 9. The number of annual suspensions, which approximates the number of insolvent banks, rose considerably during the 1920s, largely because of the post–World War I agricultural crisis. Voluntary liquidations were less frequent, and consolidations and absorptions were on the rise. Yet, even if one treats consolidations and absorptions as voluntary liquidations, as they might have been before 1914, then the sum of the two, 113 and 1104, is still not twice the number of suspensions. In the National Banking era, voluntary liquidations outnumbered insolvencies by a factor of 4. White knights may have been found for some troubled banks, but fewer banks were taking the cautious path of voluntary liquidation. Again the temporary deflationary shock of 1920–1 makes it difficult to evaluate the change in the payout ratio for insolvent banks from the permanent innovations by the Fed. Although this ratio averaged around 40 percent for the 1920s, the deflationary shock should have abated considerably by the end of the decade. However, the ratio reached only 50 percent in 1929, which is suggestive, though not conclusive, evidence that banks were responding to the new incentives, taking increased risks relative to the years before 1913.101 The changes in bank supervision witnessed in the first fifteen years of the Federal Reserve had subtle effects on bank behavior and outcomes. A small but significant minority of banks became dependent on the discount

“To Establish a More Effective Supervision of Banking”

45

window, voluntary liquidations were down, suspensions increased, and payouts declined. These changes did not destabilize the system that had arisen during the National Banking era, and aggregate losses remained very modest. In the absence of the Great Depression, Burgess’ optimism might have eventually proved warranted if one is willing to make some strong assumptions. If competition in laxity and regulatory arbitrage had been brought under control and supervision had reduced the number of borrowers at the discount window, bank failures and payouts might have returned to the lower levels of the pre-1913 era. If one doubts the ability to make these changes effective, then the American banking system was stuck with a more costly supervisory regime.

VI. Apr`es Le Deluge These conjectures seem “academic” because of the change that swept over supervision with the Great Depression. The assumption of the New Dealers was that regime of regulation and supervision inherited from the National Banking System had failed. Yet this regime had kept losses to depositors to a minimum for the fifty years before 1914, in spite of a fragmented banking structure and the absence of a central bank. It even appears to have held up reasonably well after the creation of the Fed and the large unanticipated deflationary shock of 1920–1. Nevertheless, because of a grand misdiagnosis, the New Deal swept aside this successful regime and imposed a radically different one that sharply increased moral hazard and risk taking. The Great Depression devastated the banking system in the years 1929– 33. The unexpected series of deflationary shocks led prices to fall 23 percent and contributed to the 39 percent drop in the GDP.102 In July 1929, commercial banks, which numbered 24,504, held $49 billion in deposits. By the time the bank holiday ended, only 11,878 banks with $23 billion in deposits were deemed strong enough to open immediately. The losses from insolvent banks, half to depositors and half to shareholders, reached an unprecedented $2.5 billion or 2.4 percent of the GDP.103 As is well known, the defects of the banking system and even the characters of bankers were held to be primary causes of the Depression, and the New Deal responded with a major reform effort to alter regulation and supervision. Although the influence of specific lobbies within the banking industry, notably unit bankers and independent investment bankers, have been identified in designing and lobbying for new regulation, the general argument was that the competitive market had failed and needed to be

46

Eugene N. White

subjected to a thorough regulation. The competitive market was replaced by a loosely organized government cartel with controls on many dimensions of entry and pricing. This change was based on a failure to recognize that it was large deflationary shocks that had undermined the banking system and the rest of the economy, not unbridled competition. The massive number of bank failures presented supervision with huge new challenges. In the past, it was thought that bank examiners could use market prices to judge the solvency of an institution. The increased volatility of price expectations accompanying the jagged downward path of the economy made the valuation of assets extremely difficult. However, instead of identifying the mistakes in monetary policy as the primary culprit, the reformers argued that markets failed to accurately value assets. Examiners were now instructed to value assets according to the “intrinsic value” they would have when the economy recovered. Supervision abandoned efforts to reinforce market discipline and instead was given discretion to make independent judgments, permitting forbearance in closing insolvent institutions that might recover later The most high-profile New Deal change, the creation of the Federal Deposit Insurance Corporation (FDIC), was accompanied by a change that has all but been forgotten – the abandonment of double liability. Shareholders under the threat of high assessments pleaded with legislatures to change the law, whereas the general public saw little benefit in retaining this rule.104 Additionally, it was assumed that because depositors were protected by FDIC insurance, they would no longer need the protection afforded by double liability. The extra incentive for shareholders to more carefully monitor directors was erased, leaving supervision with a heavier monitoring burden. The visible effect of double liability, voluntary liquidations, disappeared. Banks were not closed before they failed, or even shortly after they became insolvent. The increased difficulty of monitoring them and a growing moral hazard, with perhaps forbearance, meant that they would be closed only when they were gravely insolvent. The proliferation of New Deal and post–New Deal federal regulatory agencies created more competition in laxity, with opportunities for regulatory arbitrage and even regulatory capture. The sheer number of agencies involved in bank supervision, the OCC, the Fed, the FDIC, the Securities and Exchange Commission (SEC), the Federal Home Loan Bank Board, the Federal Home Loan banks, the Federal Savings and Loan Insurance Corporation, and the National Credit Union Administration made it more difficult for Congress to provide effective oversight.

“To Establish a More Effective Supervision of Banking”

47

The incentive effects of these vast changes were, however, hidden for decades. The Great Depression had eliminated all but the stronger institutions and induced them to abandon loans for U.S. government bonds. To serve its enormous financing needs during World War II, the federal government pushed banks to enlarge even further their bond portfolios.105 By the end of the war, banks had become extremely safe institutions, and then it would take decades to unwind from their bond-saturated positions and permit the full effects of the New Deal incentives to operate. This adjustment was finished by the late 1970s, when inflationary shocks created a perfect storm that caused the New Deal regulatory system to collapse. References Alesina, A., & Stella, A. (2010). The politics of monetary policy (Working Paper No. 15856). Cambridge, MA: National Bureau of Economic Research. Alston, L. J., Grove, W. A., & Wheelock, D. (1994). Why do banks fail? Evidence from the 1920s. Explorations in Economic History, 31, 409–431. Barnett, G. E. (1911). State banks and trust companies since the passage of the NationalBank Act. Washington, DC: National Monetary Commission. Barsky, R. N., Mankiw, G., Miron, J., & Weil, D. (1988). The worldwide change in the behavior of interest rates and prices in 1914. European Economic Review, 32, 1123– 1154. Bell, J. F. (1928). The growth and development of banking activities of trust companies. Unpublished doctoral dissertation, University of Illinois, Urbana. Board of Governors of the Federal Reserve System. (1914–1929). Annual report. Washington, DC: Author. Board of Governors of the Federal Reserve System (1943). Banking and monetary statistics 1914–1941. Washington, DC: Author. Bodenhorn, H. A more perfect union: Regional interest rates in the United States, 1880–1960. In M. D. Bordo and R. Sylla (Eds.), Anglo-American financial systems (pp. 415–454). New York: Irwin. Bordo, M. D., & Redish, A. (1987). Why did the Bank of Canada emerge in 1935? Journal of Economic History, 47, 405–417. Bordo, Michael D., Redish, A., & Rockoff, H. (1994). The U.S. Banking System from Northern Exposure. Journal of Economic History, 54, 325–341. Burgess, W. R. (1927). The reserve banks and the money market. New York: Harper & Brothers. Cagan, P., & Schwartz, A. (1991). The National Bank Note puzzle reinterpreted. Journal of Money, Credit, and Banking, 23, 293–307. Calomiris, C. W., & White, E. N. (1994). The origins of deposit insurance. In C. Goldin & G. D. Libecap (Eds.), The regulated economy: A historical approach to political economy (pp. 145–188). Chicago: Chicago University Press. Cannon, J. G. (1910). Clearing Houses. Washington, DC: National Monetary Commission.

48

Eugene N. White

Caporale, T., & McKiernan, B. (1998). Interest rate uncertainly and the founding of the Federal Reserve. Journal of Economic History, 58, 1110–1117. Carter, S. B., Gartner, S. S., Haines, M. R., Olmstead, A. L., Sutch, R., & Wright, G. (2006). Historical statistics of the United States (Millennium ed.). Cambridge: Cambridge University Press. Champ, B., Wallace, N., & Weber, W. (1992). Resolving the National Bank Note paradox. Federal Reserve Bank of Minneapolis Quarterly Review, 16, 13–21. Clark, T. (1986). Interest rate seasonals and the Federal Reserve. Journal of Political Economy, 94(1), 76–125. Dunbar, C. F. (1929). The theory and history of banking (5th ed., rev.). New York: Knickerbocker. Economopoulos, A. J. (1988). Illinois free banking experience. Journal of Money Credit and Banking, 20, 249–264. Fisher, R. P. H., & Wohar, M. (1990). The adjustment of expectations to a change in regime: Comment. American Economic Review, 80, 968–976. Friedman, M., & J. Schwartz, A. J. (1963). A monetary history of the United States, 1863–1960. Princeton, NJ: Princeton University Press. Grossman, R. S. (1993). The macroeconomic consequences of bank failures under the National Banking System. Explorations in Economic History, 30, 294–320. Grossman, R. S. (2001). Double liability and bank risk taking. Journal of Money, Credit and Banking, 33(2), 143–159. Grossman, R. S. (2007). Fear and greed: The evolution of double liability in American banking, 1865–1930. Explorations in Economic History, 44, 59–80. Grossman, R. S. (2010). Unsettled account: The evolution of banking in the industrialized world since 1800. Princeton, NJ: Princeton University Press. Harding, W. P. G. (1925). The formative period of the Federal Reserve System during the world crisis. London: Constable. James, J. A. (1976). The conundrum of the low issue of National Bank Notes. Journal of Political Economy, 84, 359–367. James, J. A. (1978). Money and capital markets in postbellum America. Princeton, NJ: Princeton University Press. Jalil, A. (2009). A new history of banking panics in the United States, 1825–1929: Construction and implications. Unpublished paper. University of California, Berkeley. Kaufman, G. G. (1992). Bank capital: Past, present and future. Journal of Financial Services Research, 5, 385–402. Kemmerer, E. W. (1910). Seasonal variations in the relative demand for money and capital in the United States. Washington, DC: National Monetary Commission. Kirn, B. A. (1945). Financial reports of American commercial banks. Washington, DC: Catholic University of America Press. Kool, C. J. M. (1995). War finance and interest rate targeting: Regime changes in 1914– 1918. Explorations in Economic History, 32, 365–382. Kumbhakar, S. C., & Wheelock, D. C. (1994). The slack banker dances: Deposit insurance and risk-taking in the banking collapse of the 1920s. Explorations in Economic History, 31, 357–375. Meltzer, A. H. (2003). A history of the Federal Reserve (Vol. 1). Chicago: Chicago University Press.

“To Establish a More Effective Supervision of Banking”

49

Macey, J. R., & Miller, G. P. (1992). Double liability of bank shareholders: History and implications. Wake Forest Law Review, 27(Spring), 31–62. Mankiw, N. G., Miron, J. A., & Weil, D. N. (1987). The adjustment of expectations to a change in regime: A study of the founding of the Federal Reserve. American Economic Review. 77, 358–374. Miron, J. (1986). Financial panics, the seasonality of the nominal interest rate, and the founding of the Fed. American Economic Review, 76(1), 125–140. Mishkin, F. S. (2001). Prudential supervision: Why is it important and what are the issues? In F. S. Mishkin (Ed.), Prudential supervision: what works and what doesn’t (pp. 1–29). Chicago: Chicago University Press. Moen, J., & Tallman, E. W. (1992). The panic of 1907: The role of the trust companies. Journal of Economic History, 52, 611–630. Neal, L. (1971). Trust companies and financial innovation, 1897–1914. Business History Review, 45(1), 35–51. Robertson, R. (1968). The comptroller and bank supervision: A historical appraisal. Washington, DC: Office of the Comptroller of the Currency. Rockoff, H. (1974). The free banking era: A reexamination. Journal of Money Credit and Banking, 6(2), 141–167. Rockoff, H. (1975). Varieties of banking and regional development in the United States, 1840–1860. Journal of Economic History, 35, 160–181. Rolnick, A. J., & Weber, W. E. (1983). New evidence on the free banking era. American Economic Review. 73, 1080–1091. Rolnick, A. J., & Weber, W. E. (1984). The causes of free bank failures: A detailed examination. Journal of Monetary Economics, 14(3), 267–291. Romer, C. D. (1990). Changes in business cycles: Evidence and explanations. Journal of Economic Perspectives, 13(2), 23–44. Romer, C. D. (1993). The nation in depression. Journal of Economic Perspectives, 7(2), 19–39. Schwartz, A. J. (1992). The misuse of the Fed’s discount window. Federal Reserve Bank of St. Louis Review, 74(5), 58–69. Sprague, O. M. W. (1910). History of crises under the National Banking System. Washington, DC: National Monetary Commission. U.S. Comptroller of the Currency (1864–1929). Annual Report. Washington, DC: U.S. Government Printing Office. U.S. House of Representatives. (1894). Hearings before the Committee on Banking and Currency (53rd Congress, First and Second Sessions, 1893–1894). Washington, DC: U.S. Government Printing Office. Wainwright, N. B. (1953). History of the Philadelphia National Bank: A century and a half of Philadelphia banking, 1803–1953. Philadelphia: Philadelphia National Bank. White, E. N. (1983). The regulation and reform of the American banking system, 1900– 1929. Princeton, NJ: Princeton University Press. White, E. N. (1985). The merger movement in banking, 1919–1933. Journal of Economic History, 45, 285–291. White, E. N. (1992). The comptroller and the transformation of American banking. Washington, DC: Office of the Comptroller of the Currency.

50

Eugene N. White

White, E. N. (2009). Lessons from the history of bank examination and supervision in the United States, 1863–2008. In A. Gigliobianco & G. Toniolo, G. (Eds.), Financial market regulation in the wake of financial crises: The historical experience. Rome: Banca d’Italia. White, L. J. (1991). The S&L debacle: Public policy lessons for bank and thrift regulation. New York: Oxford University Press. White, L. J. (2009). The role of capital and leverage in the financial markets debacle of 2007–2008. Mercatus on Policy, No. 37, 1–4.

Notes 1 Even contemporary authorities remained relatively mute on the subject. Charles Dunbar’s classic banking text, The Theory and History of Banking (1st ed., 1891, 5th ed., 1929), makes only the briefest mention of examination and supervision: “The Board exercises many supervisory functions over the reserve banks which are similar to those which have long been exercised by the Comptroller of the Currency over national banks. Examination of reserve banks is under its direction” (Dunbar. 1929, p. 311). 2 For a discussion of the various U.S. supervisory regimes and regulatory issues, see Mishkin (2001), White (2009), and Alesina and Stella (2010). 3 See Rockoff (1974), Rockoff (1975), Rolnick and Weber (1983). 4 Economoupoulos (1988), Rolnick and Weber (1984). 5 Carter et al. (2006, Series Cj203). 6 For details, see James (1976) and White (1983). 7 The United States went off the bimetallic standard in 1861. Resumption took place in 1879, with the nation moving to a de facto gold standard. 8 The longest serving pre-1914 comptroller was John Jay Knox, April 25, 1872– April 30, 1884. 9 Carter et al. (2006, Series Cj203 and Cj204). 10 In an era in which communications were slow, this decision may have made sense from an examiner’s point of view, as it would be difficult to examine a bank and its branches simultaneously. 11 The success of the growing state banking systems ultimately forced Congress to reduce the lowest of the tiered minimum capital requirements. In 1864, this requirement set $50,000 as the minimum capital for towns with a population of fewer than 6,000. The Gold Standard Act of 1900 set a new lower bound of $25,000 for the minimum capital required for banks starting up in towns with a population of fewer than 3,000. White (1983). 12 For details of this competition in laxity, see White (1983, Chapter 1). 13 There is a large literature on the “conundrum” of low national banknote issue. For example, see James (1976), Cagan and Schwartz (1991), and Champ, Wallace, and Weber (1992). 14 Barnett (1911). 15 Neal (1971). 16 See Barnett (1911, p. 235) for more details. 17 Cannon (1910); Friedman and Schwartz (1963). 18 Friedman and Schwartz (1963); Moen and Tallman (1992).

“To Establish a More Effective Supervision of Banking”

51

19 Sprague (1910); Kemmerer (1910). 20 See Jalil (2009) for a discussion of the differences between the standard sources on the classification of banking panics. 21 Meltzer (2003, p. 9). 22 Romer (1990) compares the years 1886–1916 with 1948–97. 23 Miron (1986). Additionally, for the National Banking era, Grossman (1993) found that a “small bank failure” shock could lead to a 2 percent decline in real GNP and a large bank failure shock could produce a 20 percent decline in GNP. 24 For the years 1825–1914, the average decline in output for recessions with panics was 7.9 percent, with the average time from peak to trough 1.4 years and troughto-pre-downturn peak 1.7 years, contrasting 4.8 percent, 1.2 years, and 1 year for recessions without panics. If only the post-Civil War era is considered, the severity and length of recessions with panics is even greater (Jalil, 2009, p. 34). 25 Regular examinations first appeared in the United States before the Civil War in six states, New York, Vermont, Indiana, Michigan, Ohio, and Iowa (Robertson, 1968, pp. 25–6). 26 Banks were required to report the payment of a dividend within ten days or face a similar penalty (Robertson, 1968, pp. 79–81). 27 In the 1889 Annual Report, the comptroller asked for an additional appropriation of $10,000 to fulfill to a request from the American Bankers Association to publish not only call report data for each national bank on the date nearest the first of October but also for the date nearest the first of April (U.S. Comptroller of the Currency, 1889, pp. 53–4). 28 Officers and directors must be “complying with the requirements of the law and [confirm] whether they are in any way violating any of its provisions” (U.S. Comptroller of the Currency, 1881, pp. 35–6. 29 U.S. Comptroller of the Currency (1891, p. 26). 30 U.S. Comptroller of the Currency (1889, p. 54). 31 Robertson (1968, pp. 76–9). 32 U.S. House of Representatives (1894, p. 197). 33 U.S. Comptroller of the Currency (1901, p. XVIII). 34 “This is a matter of very great importance in the interest of good administration, and the effective supervision of national banks. Every Comptroller of the Currency has agreed in this opinion, and has recommended that this change be made. The examination and supervision of national banks will never be what it should be, until this recommendation is carried out” (U.S. Comptroller, 1906, p. 65). 35 Robertson (1968, p. 71, note 13). 36 U.S. Comptroller of the Currency, (1881, p. 38). 37 There may have been a very important private complement to the limited examinations under the OCC in this period. The 1907 U.S. Comptroller of the Currency’s Annual Report mentions the examination by the clearinghouses and their improving quality. See Cannon (1910, pp. 137–147). 38 The GDP deflator is used to measure the real values (Carter et al. 2006, Series Ca13). 39 The U.S. Comptroller of the Currency’s Annual Report (1889, p. 55) provides a description of some aspects of examination and refers to “the examiner, assisted by competent assistants.” 40 U.S. Comptroller of the Currency (1889, p. 57).

52 41 42 43 44 45 46

47 48 49 50 51 52 53 54 55 56

57 58 59

60 61 62

63 64 65 66 67 68 69 70

Eugene N. White U.S. Comptroller of the Currency (1889, pp. 85–6). Cannon (1910), p. 138. U.S. Comptroller (1889, pp. 55–6). Cannon (1910, p. 138). Wainwright (1953, p. 165). The concern about the quality of management appears to have been shared the clearinghouses. Before the panic of 1907, only the Chicago Clearing House had appointed an examiner (in 1906) to examine its member banks on a regular basis. Afterward, the clearinghouses in Minneapolis-St. Paul, St. Louis, Los Angeles, Kansas City, and Philadelphia employed an examiner (Cannon, 1910, pp. 137–47). U.S. Comptroller (1912, pp. 84–85). U.S. Comptroller (1912, p. 84). Barnett (1911, p. 174). Barnett (1911, p. 145). Barnett (1911, p. 146). Barnett (1911, p. 149). Cannon (1910, p. 146). Barnett (1911, p. 162). Barnett (1911, p. 163). Barnett (1911, pp. 167–8). The 1907 New York Special Commission found that the New York court-appointed receivers in 1907 were more costly than those appointed by the OCC, absorbing 13 percent compared with 8.9 percent for the comptrollerappointed receivers for the period 1865–1906 (Barnett, 1911, p. 169). Carter et al. (2006, Series Cj159). It should be noted that it is unclear in the report if the state examiners had to pay for their expenses out of their compensation. Grossman (2001, 2007, 2010) is the only exception. He provides a careful survey of liability rules but probably understates their effects as his analysis focuses on the differences in liability among the state systems, which had weaker enforcement than national banks. Macey and Miller (1992). Grossman (2007). Barnett (1911, pp. 78–85). Seven states did provide an additional guarantee for depositors after 1907 in the form of deposit insurance schemes for their statechartered banks. However, these short-lived guarantee systems did not play a major role in protecting the banking system. See White (1983) and Kumbhakar and Wheelock (1994). Grossman (2007) finds that the seven states that adopted deposit insurance had double liability in effect or legislated it at the same time U.S. Comptroller of the Currency (1913, p. 104). A few of the liquidating banks were consolidated with other banks. Kaufman (1992). Bodenhorn (1995); James (1978). Carter et al. (2006, Series Cj2090. Rockoff (1975). Friedman and Schwartz (1963, p. 351). White (1991).

“To Establish a More Effective Supervision of Banking”

53

71 White (2009). 72 Unfortunately, these types of banks cannot be separately analyzed (Barnett, 1911, p. 183). 73 U.S. Comptroller of the Currency (1920, Table 1.38, pp. 80–122). 74 Cited in Barnett (1911, p. 182). 75 Grossman (2001). 76 On the integration of financial markets, see Bodenhorn (1995). 77 Bordo and Redish (1987); Bordo, Redish, and Rockoff (1994). 78 While the Federal Reserve Act was pending in Congress, there was discussion about whether to have the OCC moved from the jurisdiction of the Treasury to the Federal Reserve Board or to simply abolish the OCC and transfer its operations to the Board. Such suggestions were rejected and the Comptroller was made a member of the Reserve Bank Organization Committee and a member ex officio of the Federal Reserve Board. The Comptroller was granted a salary of $7,000 for his membership on the Board in order to raise his total salary to that of other board members (Harding, 1925, p. 6). 79 Harding (1925, pp. 6–7). 80 Harding (1925, pp. 201–2, 209). 81 Burgess (1927, p. 29). 82 Burgess (1927, pp. 234–5). 83 Under the national banking system banks were not permitted to borrow in excess of their capital, but the Federal Reserve Act allowed them to do so. 84 White (1983, Chapter 3). 85 See Kirn (1945). 86 The next time a surprise year-end call was issued was in 1962. See White (1992, p. 18). 87 For the random distribution of the call dates from 1869 to 1913, see the U.S. Comptroller of the Currency (1913, p. 215). 88 See White (1983). 89 U.S. Comptroller of the Currency (1919). 90 Kirn (1945, pp. 164–8). 91 State banks complained about bearing the cost of state examinations in addition to those of the Fed, though the board could accept state examinations in lieu of additional federal ones (Kirn, 1945, p. 164). 92 Friedman and Schwartz (1963, pp. 231–9); Meltzer (2003, pp. 109–19). 93 The price level is measured by the GDP deflator (Carter et al., 2006, Series Ca13). 94 Alston, Grove, and Wheelock (1994). 95 Calomiris and White (1994, pp. 170–2). 96 U.S. Comptroller, Annual Report (1914, p. 17). 97 See Clark (1986), Mankiw, Miron, and Weil (1987), Barsky, Mankiw, Miron, and Weil (1988), Fisher and Wohar (1990), Kool (1995), and Carporale and McKiernan (1998). 98 On June 30, 1925, there were 9,538 member banks. The 593 borrowing banks thus represented 6.2 percent of the total (Board of Governors of the Federal Reserve System, 1943, pp. 22–3). 99 Schwartz (1992, p. 58) and Burgess (1927, p. 236) give figures for 1923 and 1925. 100 White (1985, p. 288).

54 101 102 103 104 105

Eugene N. White Calomiris and White (1994, p. 171). Romer (1993, p. 22). Friedman and Schwartz (1963, p. 351) and the author’s calculation. Macey and Miller (1992). Friedman and Schwartz (1963, Chapter 10).

Comments to “To Establish a More Effective Supervision of Banking”: How the Birth of the Fed Altered Bank Warren Weber1

This interesting chapter is well worth reading for two reasons. First, it contains a wealth of historical information on the National Banking System (NBS) and on the early years of the Federal Reserve System. Second, it says some extremely important things about the importance of supervisory incentives and the ability of supervisors to take disciplinary actions. The chapter describes and evaluates the changes in banking supervision and regulation that occurred with the establishment of the Federal Reserve System in 1913. It has two main theses: 1. The establishment of the Fed did not improve the supervision and regulation of banks as compared with the NBS, at least in the short run. 2. The problems experienced by the NBS, which stemmed from its basic structure, would not have been cured by better supervision and regulation. The chapter proceeds by discussing the supervisory and regulatory environment under the NBS. It then discusses the supervisory and regulatory changes that occurred with establishment of the Fed. It concludes by discussing why, at least in the short run, these changes did not necessarily improve the safety and soundness outcomes of the banking system.

I. Supervision and Regulation Under the National Banking System The supervisor of national banks was the comptroller of the currency. National banks were required to submit five call reports per year. The timing of two of these reports was random. In addition, national banks were subject to two examinations per year. The main disciplinary action that could be taken against a bank was to revoke its charter. 55

56

Warren Weber

In White’s view, the National Banking System is seen as having performed poorly, primarily because of the numerous banking and financial crises that occurred between the time it began and when the Federal Reserve System was established. White argues that these crises arose because national banks were prohibited from branching, national bank notes were an inelastic currency that could not expand and contract with the needs of trade, and there was no entity that could act as a central bank in times of financial strain. A further problem, not mentioned by White, was the structure of reserve requirements on deposits, along with the fact that banks could pay interest on “bankers’ balances,” which meant that the country’s specie reserve piled up in a few central reserve city banks. White does not argue that the NBS was not without flaws. He views the deficiencies just mentioned to be serious ones. However, his main point about the NBS is that its deficiencies were not in the areas of supervision and regulation.

II. Changes with the Establishment of the Fed White identifies two changes in the structure of supervision and regulation of banks that occurred with the establishment of the Fed. In his view, these changes were not good ones, as least as they were implemented. The first of these was that the Fed was added as a regulator of state chartered banks that were members of the Federal Reserve System. In White’s view, this led to “competition in [supervisory] laxity” between potential regulators, namely the Office of the Comptroller, state regulators, and the Fed. In being more lax in its regulation, White argues, the Fed’s purpose was to provide an incentive for more state banks to become members of the Federal Reserve System. I think that White should have built a stronger case for a weakening of supervision. He cites as evidence a decrease in the number of call reports required of national banks. I think it would have been useful to see if state bank data showed an increase in the rate of insolvencies and the losses to bank creditors after establishment of the Fed. The second change in supervision and regulation was the establishment of the discount window. This change created moral hazard problems. The discount window gave banks an incentive to make risky loans, because the Fed would be stuck with part of any losses. In addition, the regulators running the discount window allowed banks to borrow for too long. Here, I think White builds a strong case. He states that 593 banks, out of the approximately 9,500 in existence, borrowed from the discount window for more than a year. Further, for the entire period of 1920–5, 239 banks

Comments

57

borrowed continuously. And 80 percent of the failures that occurred over the 1920–5 period were banks that were “habitual borrowers” from the discount window. What I would have found fascinating would have been some documentary evidence of what Fed officials were thinking when they permitted banks to have credit through the discount window for such extended periods. It would also have been interesting to see if the borrowing patterns differed by reserve district, and if they did, whether any reasons were given for the differences.

III. Some Thoughts This chapter is about incentives, supervision, and disciplinary action, and it uses a historical event, the establishment of the Fed, to show the importance incentives in designing banking supervision and regulation. White’s effective use of history led me to ask whether there were other historical examples that illustrate the same point. I came up with two, both of which suggest considering regulatory schemes with mutualization of losses in which potential loss-sharers have supervisory and regulatory powers as a way to get agents to have more “skin in the supervision/discipline game.”

Example 1: The Suffolk Banking System The Suffolk Banking System was a clearing system for the notes of New England banks run from Boston by the Suffolk Bank. It existed from 1825 to 1858. Note that clearing under this system was done very much like the Fed clears checks today. Banks that were members of the system had accounts with the Suffolk Bank, and these accounts were debited and credited as banks brought in the notes of other member banks. The mutualization of losses under this system occurred because during the note clearing process, the Suffolk Bank made overdraft loans to banks that had low or negative balances with it. If a bank failed and was unable to pay off these overdrafts, Suffolk would lose. The amount of overdrafts that Suffolk had on its books was close to the amount of its capital at times. Because of the potential to suffer large losses, Suffolk “supervised” and “regulated” the banks that were members of the Suffolk Banking System. An example is a letter sent by the president of the Suffolk Bank to the president of the Bank of Woodstock in Vermont: It appears evident . . . that too large a portion of your loan is in accommodation paper, which cannot be relied upon at maturity to meet your liabilities. . . . [W]e

58

Warren Weber

hope you will take measures to change the character of your loan, and render it more available in case of need.

In other words, clean up your portfolio or risk getting kicked out of the system. It also appears that the Suffolk Bank’s supervision was quite effective in preventing bank failures. Banks in New England had low failure rates compared with those of banks in other parts of the country, and there were times when banks in New England did not suspend specie payments on their notes when banks in other parts of the country did.

Example 2: The State Bank of Indiana Despite the name, the State Bank of Indiana was in actuality a system of independent, mostly privately owned banks that were called branches. The business decisions of each branch were made by the directors of that branch, and the profits of a branch accrued solely to the stockholders of that branch. The mutualization of losses under this system occurred because the branches mutually guaranteed “all debts, notes, and engagements of each other.” And each branch had a large amount of skin-in-the-game. A rough calculation shows that the amount of exposure that a branch had to the failure of another branch would have amounted to approximately 20 percent of its capital. Supervision under the State Bank of Indiana system was done by a state board that consisted of some members appointed by the state legislature and one director from each branch. This board had broad powers. It could limit a branch’s ratio of loans and discounts to capital, limit the amount of dividends a branch could pay, and close a branch. No branch of the State Bank of Indiana ever failed.

IV. Conclusion This paper is well worth reading for two reasons. One, it has a wealth of historical information. Two, it convincingly points out that when thinking about bank regulation, it is important to think about supervisory incentives and the ability of supervisors and regulators to take disciplinary actions. Note 1 The views expressed herein are those of the author and not necessarily those of the Federal Reserve Bank of Minneapolis or the Federal Reserve System.

TWO

The Promise and Performance of the Federal Reserve as Lender of Last Resort 1914–1933 Michael D. Bordo and David C. Wheelock

INTRODUCTION

The actions by the Federal Reserve to defuse the financial crisis of 2007– 8 renewed long-standing debates about how central banks should act as lenders of last resort. The Fed’s defenders contend that the central bank’s response to the crisis was effective and consistent with the long-accepted principles of Bagehot (1873) (e.g., Madigan, 2009). Critics, however, argue that the Fed’s actions did little to alleviate financial strains (Taylor and Williams, 2009), contributed to instability (Meltzer, 2009), and may have helped sow the seeds of future crises by protecting creditors of large financial firms (Buiter, 2009; Poole, 2009).1 Clearly, there remain many unsettled questions about how central banks should carry out their responsibilities as lenders of last resort. This chapter examines the origins and early performance of the Federal Reserve as lender of last resort. We believe that a look back at the successes and failures of central banks in the past can inform current discussions about how central banks should act as lenders of last resort. Here we consider why the Fed’s performance as lender of last resort, especially during the Great Depression, failed to live up to the promises of those who designed the System. The Fed was established to overcome the problems of the National Banking era. Those problems, which included seasonal money market stringency and recurrent banking panics, had brought calls for reform by the 1870s. Following the panic of 1907, Congress enacted the Aldrich-Vreeland Act of 1908, which established the National Monetary Commission as well as a temporary mechanism for increasing the supply of currency during banking panics. The studies of the National Monetary Commission identified defects of the U.S. banking system and drew lessons from the performance of banking systems in other countries. One study, in particular, argued that 59

60

Michael D. Bordo and David C. Wheelock

the panic of 1907 and earlier crises revealed the superiority of the European “discount system” and the vital role played by central banks in maintaining financial stability. The study’s author, Paul Warburg (Warburg, 1910a), convinced Nelson Aldrich, the powerful chairman of the Senate Banking Committee, of the efficacy of the European system, and Aldrich became the principal champion of a central banking system for the United States. Aldrich convened the now famous Jekyll Island meeting of November 1910, where he met with a small group of leading bankers, including Warburg, to determine how to organize and operate the proposed central bank. The bill Aldrich submitted to Congress in 1912 was the product of that meeting. It included many features that reflected Warburg’s desire to emulate the European systems. Aldrich proposed a National Reserve Association that would oversee the operations of a system of local and regional reserve associations and set a discount rate at which the local branches would rediscount notes and bills of exchange for member banks (Wicker, 2005). Congress, which then was controlled by Democrats, rejected the Aldrich bill. However, the Federal Reserve Act of 1913 resembled the Aldrich bill in many respects, including the provisions concerning the rediscounting of commercial paper and bills of exchange for member banks, which were fundamental to how the central bank would serve as lender of last resort to the banking system. Neither the Aldrich bill nor the Federal Reserve Act dealt explicitly with financial crises nor prescribed how the Fed should respond to banking panics. The authors believed that their proposed reforms would prevent banking panics from occurring in the first place. Indeed, the United States had no banking panics during the first 15 years of the Fed’s existence, despite the occurrence of several shocks, including a world war, a short, but severe, postwar recession, and the failure of several thousand mostly small, rural banks during the 1920s. Banking panics returned with a vengeance during the 1930s, however, and the Fed’s failure to prevent or counteract panics was, many believe, a principal cause of the Great Depression (e.g., Friedman & Schwartz, 1963; Bernanke, 1983). Numerous explanations for the Fed’s failures during the Depression have been suggested, including defects in the system’s structure and leadership, the Fed’s devotion to the gold standard, and policymakers’ misreading of monetary conditions. Although each of these likely contributed to the Fed’s highly deflationary monetary policy, we believe that there is more to the story, especially as to why the Fed failed to prevent or offset serious banking panics. We trace the Fed’s failure to act as an effective lender of last resort during the Great Depression to defects of the Federal Reserve Act and, more broadly,

The Promise and Performance of the Federal Reserve as Lender

61

of the U.S. banking system. In particular, the act failed to re-create the money market conditions and other institutions that enabled the Bank of England and other European central banks to function effectively as lenders of last resort. In addition, the act created a system that depended critically on the competence of the individuals running the system – a point that Friedman and Schwartz (1963) emphasize – rather than a set of rules or principles to guide lender of last resort policy. Finally, and perhaps at least as important, the Federal Reserve Act failed to replace the crisis-prone U.S. unit banking system with a more stable, concentrated branch banking system, such as those of the United Kingdom and Canada. The following section briefly describes the defects of the National Banking System identified by the National Monetary Commission and others and the European discount system that advocates saw as the appropriate reform. Next we discuss how the Federal Reserve System was intended to overcome the flaws of the National Banking System by creating a discount mechanism to supply bank reserves and currency as needed to support economic activity and avoid panics. We then examine how well this mechanism performed and consider why the Fed failed to serve effectively as lender of last resort during the Great Depression. The Fed’s failures led to numerous reforms in the mid-1930s, including expansion of the Fed’s lending authority and changes in the system’s structure, as well as changes that made the U.S. banking system less prone to banking panics. Finally, we consider lessons about the design of lender-of-last-resort policies that might be drawn from the Fed’s early history. BANKING REFORM

Defects of the National Banking System The recurrent instability of the National Banking era was the principal motivation for the reform movement that led to the Federal Reserve Act. A related impetus was the desire to enhance the international role of the dollar and to have a central bank to manage the gold standard (Broz, 1997). The United States experienced numerous bank failures, banking panics, and persistent seasonal stringency in the money market throughout the nineteenth century, which reflected two fundamental problems: (1) unit banking and (2) the absence of an effective lender of last resort. Unit banking resulted from legal restrictions imposed by the federal government on interstate branch banking and by most states on branching within state borders. Unit banking made it difficult for banks to pool risks

62

Michael D. Bordo and David C. Wheelock

and exposed them to local and regional shocks. Further, it hampered their ability to grow in size and scope to satisfy the credit and payments requirements of their major business customers, especially toward the end of the nineteenth century, when large industrial firms with national operations were becoming more prevalent (Calomiris, 1993, 1995). Unit banking was largely a U.S. phenomenon. Over time, the banking systems of Canada and most European countries became increasingly dominated by small numbers of large banks with nationwide branches (Bordo, Redish, & Rockoff, 1996; Grossman, 2010). Branching restrictions and the absence of a central bank reflected deepseated populist fears about the concentration of financial power. Alexander Hamilton, the first U.S. secretary of the treasury, established a prototypical central bank in 1791. The federally chartered First Bank of the United States was inspired by the experience of the Bank of England. The First Bank acted as the government’s fiscal agent, provided a uniform national currency, and promoted economic development. Its large capitalization and nationwide scope (it had branches in every state) allowed it to evolve quickly into a bankers’ bank for the nation’s nascent state-chartered banks. In 1811, the bank’s twenty-year charter was not renewed amid allegations of corruption and populist and states-rights opposition to its power (Timberlake, 1993). A second federal bank was chartered in 1816, with a similar charter as the First Bank and a similar fate. Under the direction of its president, Nicholas Biddle, the Second Bank was even more successful than the First Bank in providing a uniform national currency by effectively policing the note issues of the state banks and performing many central banking functions (Knodell, 2003). The Second Bank encouraged and backstopped the development of a liquid market in bills of exchange (two-name paper) and on occasion provided liquidity to correspondent banks in times of stringency, actions similar to those undertaken by the Bank of England at this time (Broz, 1997). It also conducted exchange market policy to manage the gold standard (Bordo, Humpage, & Schwartz, 2007). Andrew Jackson vetoed the bill to recharter the Second Bank in 1836. The demise of the Second Bank left the chartering and regulation of banks entirely to the states, the majority of which enacted free banking laws in the 1830s and 1840s. These laws significantly reduced entry barriers into banking and permitted banks to issue notes based on the collateral of eligible bonds. The free banking era of 1836–63 was characterized by a multiplicity of banknotes circulating at varying rates of discount, reflecting the soundness of the banks and the distance from the issuer (Gorton, 1996), frequent bank failures, in some states fraud (“wildcat banking”), and several serious panics

The Promise and Performance of the Federal Reserve as Lender

63

(1837, 1839, 1857) (Rockoff, 1974; Temin, 1969).2 The National Banking Acts of 1863 and 1865 were intended to overcome perceived flaws of the free banking era, as well as to create a demand for U.S. government debt. Under the National Banking System, a uniform currency emerged in the form of national banknotes backed by U.S. government bonds. National banks had higher reserve and capital requirements than the banks chartered by most states and were supervised by the U.S. comptroller of the currency. State banknotes were taxed out of existence, but state banks continued to thrive under more lax state regulation as deposit-taking institutions (White, 1983). The National Banking System prevailed for fifty years. Although it created a uniform national currency, the system itself had several serious defects that reformers viewed as responsible for a series of banking panics in 1873, 1884, 1890, 1893, and 1907 that were as severe as the major antebellum panics. Reformers identified three problems in the U.S. banking system: (1) an “inelastic” currency stock; (2) seasonal stringency in the money market; and (3) an “inverted pyramid” of banking system reserves.

Inelastic Currency The framers of the National Banking System sought to avoid the periodic suspensions of convertibility of banknotes into specie or other forms of high-powered money that plagued the free banking era by requiring national banks to pledge U.S. government bonds as security for their currency liabilities. The remedy, however, resulted in an inflexible currency stock that did not vary rapidly or sufficiently to meet normal seasonal variations in economic activity, let alone extraordinary demands. Under the National Banking System, the stock of national banknotes could expand only through an increase in the volume or value of U.S. government bonds held by national banks, which were unlikely during the short period of a crisis. Through local clearinghouses, banks developed ways to conserve reserves and expand high-powered money by issuing emergency currency in the form of clearinghouse certificates (Timberlake, 1984; Gorton, 1985). Beginning with the panic of 1857, the New York Clearing House issued loan certificates to its member banks based on the discounted value of the collateral they posted in proportion to each bank’s share of the total assets of the clearinghouse. These certificates served as substitutes for reserves, which allowed member banks to pay out cash that otherwise would have been tied up in interbank settlements. In the panic of 1873, the New York Clearing House pooled the reserves of the member banks. Later, in the

64

Michael D. Bordo and David C. Wheelock

panics of 1893 and 1907, clearinghouse currency was issued in exchange for loan certificates. This arrangement provided depositors with insurance against the failure of individual banks and discouraged runs. The issuance of clearinghouse certificates was sufficient to allay crises in 1884 and 1890 but not other crises of the era. The U.S. Treasury also intervened on occasion during panics to add reserves to the banking systems by depositing tax and customs receipts in New York banks and by other methods. These interventions were often too little and too late (Schwartz, 1986). Banking panics in 1873, 1893, and 1907 ended only after the suspension of convertibility of deposits into currency.

Seasonal Stringency Banking reformers noted that financial crises tended to occur at times of the year when the demands for currency and credit peaked. Seasonal fluctuations in credit demand produced seasonal swings in interest rates and in the ratio of reserves to deposits in the banking system (Miron, 1986). Capital inflows relieved the seasonal pressures to some extent (Goodhart, 1969). However, if an unusual gold outflow or a shock to the domestic financial market, such as major bank failure or stock market crash, occurred at a seasonal peak in currency and credit demand, a banking panic was more likely to occur (Kemmerer, 1910; Sprague, 1910).

Inverted Pyramid of Reserves Reform advocates pointed to the distribution of reserves across the banking system as a third problem with the National Banking System. Under the National Banking System, national banks were required to maintain minimum levels of reserves against their deposit liabilities, which were intended to prevent excessive deposit expansion and to protect banks in the event of runs (Bordo, Rappoport, & Schwartz, 1992, pp. 211–13). National banks located in small cities and rural areas (i.e., “country” national banks) were subject to a 15 percent reserve requirement, of which three-fifths could be held as balances with correspondent banks in reserve cities (cities with populations greater than 50,000) or in central reserve cities (New York, Chicago, and St. Louis). The remaining two-fifths of required reserves were to be held in lawful money (U.S. notes, specie, gold, and clearinghouse certificates). National banks in reserve cities were required to hold 25 percent of their deposits in reserves, half of which had to be held in lawful money, the other half as balances on deposit in central reserve-city national banks.

The Promise and Performance of the Federal Reserve as Lender

65

Central reserve-city national banks were required to hold 25 percent of their deposits in lawful money. Country and reserve-city banks tended to hold the maximum allowable amount of reserves in the form of deposit balances in central reserve cities. The reserve structure of the National Banking System was described as an inverted pyramid because many of the nation’s bank reserves were held in the form of correspondent balances in a handful of banks in the central reserve cities, especially New York City (Myers, 1931; James, 1978). Most of the reserves held as correspondent balances in New York City were invested in the call loan market. Call loans were demand loans secured by equities traded on the New York Stock Exchange and by U.S. government and other bonds. Most call loans were made to brokers who would then consign the securities serving as collateral to the banks. Commercial banks considered call loans their most liquid investment. The New York City national banks dominated the call loan market – close to 75 percent of bankers balances in New York were held in call loans (Myers, 1931, p. 290). In addition, country and reserve-city national banks, state commercial banks, savings banks, and trust companies all invested directly in the call loan market (using their central reserve-city correspondents as intermediaries) whenever the call loan rate rose significantly above the 2 percent rate normally paid on correspondent balances. Thus an inverse relationship existed between the call loan rate and correspondent balances in New York City and a direct one between the call loan rate and country bank excess reserves invested directly in the call loan market (Myers, 1931, p. 290; James, 1978, p. 304). The inverted pyramid of reserves and the intimate connection between the correspondent balance system and the call loan market were widely regarded as key elements in the financial crises of the National Banking era (Sprague, 1910). All of the major banking panics were marked by withdrawals of correspondent balances by the country and reserve-city banks from the New York banks. The decline in correspondent balances in turn put pressure on the call loan market, causing call loan rates to rise sharply and stock prices to fall. The decline in the reserves of New York City banks could be so severe as to precipitate a panic, which could be stopped only by a suspension of convertibility of deposits into currency or by the issuance of clearinghouse loan certificates. A related critique of the National Banking System was that banks viewed the reserves they held to meet the minimum requirements as unavailable in a crisis. Banks rarely were willing to let their actual reserve holdings fall below the minimum because of the possible legal sanctions that might be

66

Michael D. Bordo and David C. Wheelock

imposed; hence, reserves were not an effective line of defense in the event of a bank run (Cagan, 1963; West, 1977, pp. 30–1).

The Reform Movement The first wave of reform proposals followed the 1873 panic and focused on the pyramiding of banking system reserves. Several proposals called for prohibiting the payment of interest on interbank deposits as a way of discouraging banks from holding reserves in the form of correspondent balances, but none resulted in legislation (West, 1977, Chapter 2). A second wave of reform proposals followed the panic of 1893. They focused on making the currency stock more elastic by replacing the nation’s “bond-backed” currency, that is, national banknotes backed by government bonds, with an “asset-backed” currency tied to banks’ holdings of commercial paper. The American Bankers Association’s Baltimore Plan of 1894 was one of the first to call for a currency backed by commercial paper. A similar plan was recommended by J. Laurence Laughlin at the Indianapolis Monetary Convention in 1897. Laughlin, a professor of economics at the University of Chicago, was a strong proponent of the real bills doctrine. He argued that basing the currency stock on self-liquidating short-term commercial paper (real bills) used to finance commerce, industry, and agriculture would always prevent overissue or underissue. Laughlin’s views became more influential as momentum for reform built up in the next decade (Broz, 1997; Mehrling, 2002).3 The Aldrich-Vreeland Act of 1908 was an important step in the reform movement. In addition to creating the National Monetary Commission, the Aldrich-Vreeland Act institutionalized the emergency currency provisions developed by major clearinghouses to alleviate banking panics. The act permitted groups of ten or more national banks to form currency associations to issue emergency currency in the event of a crisis equal to as much as 75 percent of the value of commercial paper deposited with the association. The act also permitted individual banks to issue notes if authorized by the secretary of the treasury. Currency issued under the Aldrich-Vreeland Act was subject to tax to ensure its speedy retirement after an emergency had passed. Aggregate circulation under the act was limited to $500 million (Friedman & Schwartz, 1963, Chapter 3). The Aldrich-Vreeland Act was a temporary measure, and it was used only once, to stem a crisis in 1914 at the outbreak of World War I (Wicker, 2005; Silber, 2007). Nelson Aldrich, chairman of the Senate Committee on Banking and Currency, remained at the center of the banking reform movement. As

The Promise and Performance of the Federal Reserve as Lender

67

noted previously, he was persuaded of the efficacy of the European-style discount and central banking system by Paul Warburg, a partner in the firm of Kuhn Loeb. Warburg had been a successful banker in Germany before immigrating to the United States in 1902. Based on his experience and knowledge of the operations of the German Reichsbank, the Bank of England, and the Banque de France, Warburg made the case for a Europeanstyle central bank for the United States. He argued that in the advanced countries of Europe the presence of a discount market and a central bank that provided liquidity to back up the market and serve as lender of last resort in times of stringency prevented the type of financial instability experienced in the United States (Warburg, 1910a). Warburg believed that a market for bills of exchange (two-name bills), as exemplified by the market for bankers acceptances, would be more liquid than the existing U.S. commercial paper market (which was based on single-name promissory notes). Acceptances were short-term instruments in which the IOU issued by, for example, a merchant to one of his suppliers, would be guaranteed (accepted) by a bank. The bank’s reputation would allow the bill to be traded in an open market and hence provide liquidity. Two types of acceptances were used in Europe: trade acceptances used to finance domestic trade (inland bills in England), and bankers’ acceptances used to finance international trade. The latter type of bill was not legal in the United States during the National Banking era and the former had declined in use after the Civil War.4 Warburg argued that the U.S. money market would be more liquid if banks were permitted to issue bankers acceptances. In addition, he noted that the creation of a U.S. acceptance market would break the monopoly that sterling bills (bankers acceptances drawn on British merchant banks) had over U.S. international commerce and help the dollar become an international currency (Broz, 1997). Warburg believed that recreating as closely as possible the money market environment of England, France, and Germany was a crucial step in bringing stability to the U.S. banking system. The European financial system in the late nineteenth century, especially the most highly developed one in England, was both sophisticated and complex. The Bank of England took many years to evolve into an effective lender of last resort and money market maker that Aldrich, Warburg, and the other New York bankers so admired. The bank was chartered in 1694 as a joint stock bank of issue and served as the government’s fiscal agent. The bank’s charter required that its liabilities be convertible into gold at the official parity. Over time, the bank’s strong capitalization and position as the government’s bank enabled it to become a bankers’ bank. However, it did not always act as a true lender of last resort

68

Michael D. Bordo and David C. Wheelock

in times of panic because of its responsibility to its shareholders (Goodhart, 1987). Only after the Overend Gurney crisis of 1866 did the bank accept Bagehot’s Responsibility Doctrine and agree to subsume its private interest to the public good in times of crisis (Bordo, 1990). The English financial system had become very sophisticated by the middle of the nineteenth century. It consisted of merchant banks that financed international trade, bill brokers that dealt in bills of exchange, discount houses that evolved from bill brokers and purchased and rediscounted bills, and commercial banks. The market for bills of exchange (acceptances) was both deep and liquid (Bignon, Flandreau, & Ugolini, 2011). Discount houses, such as Overend Gurney, had gained considerable prominence in the English market by the mid-nineteenth century. The discount houses acted as intermediaries between commercial banks and the Bank of England. When in need of liquid funds, the commercial banks would turn to the discount houses to rediscount their paper, and the discount houses in turn would go to the Bank of England for accommodation. The discount houses would pass their bills to the bank, which would judge the quality of the paper offered as collateral and return cash if the collateral was deemed acceptable. According to Capie (2002, p. 311), the bank lent anonymously to the market: The mechanism can be envisaged as the central bank having a discount window made of frosted glass and raised just a few inches. Representatives of institutions could appear at the window and push through the paper they wanted discounted. The central banker would return the appropriate amount of cash, reflecting the going rate of interest. The central banker does not know, nor does he care, who is on the other side of the window. He simply discounts good quality paper or lends on the basis of good collateral. In this way, institutions holding good quality assets will have no difficulty in obtaining the funds they need. Institutions with poor quality are likely to suffer. In times of panic the interest rate would rise.

Thus, the Bank of England did not as a rule lend to individual banks, but to the market. The bank’s discount rate, “bank rate,” served as an anchor to the financial system. In times of crisis the bank followed Bagehot’s strictures: (1) to lend freely in the face of an internal drain (a domestic liquidity crisis) and to discount all sound collateral; (2) to charge a high rate in the face of an external drain (an outflow of gold reserves); and (3) to lend freely at a high rate when faced with both an internal and an external drain. Bagehot is commonly believed to have said “lend freely at a penalty rate” to discourage moral hazard. However, according to Goodhart (1987), Bagehot used the term “high” and not “penalty.” Moreover, according to Bignon et al.

The Promise and Performance of the Federal Reserve as Lender

69

(2009), there is considerable confusion in the subsequent literature over the term penalty and that by the 1850s the bank rarely discounted paper that would be subject to moral hazard. In the years following Bagehot’s Lombard Street (1873), the bank never faced another banking panic (Schwartz, 1986; Capie, 2002). Warburg was most familiar with the German system. Germany also had a well-developed discount market and a highly concentrated banking system. As Warburg (1910a) describes, the German banks discounted directly with the Reichsbank (Wicker, 2005).5 Germany faced only one minor banking crisis in 1901 (Bordo & Eichengreen, 2002). The institutional frameworks of the European banking systems that Warburg admired were very different from those of the United States in the National Banking era. The European banking systems were relatively concentrated, dominated by large banks with nationwide branches, and had active markets in bankers acceptances, which could be rediscounted with a central bank. By contrast, the United States had thousands of small unit banks, no acceptance market, and no central bank. Would the new Federal Reserve regime adapt U.S. institutions to match the performance of the pre-1914 European central banks? Unit banking remained in the United States after the Fed was established, as did the “dual” banking system, in which some banks were chartered and supervised by national authorities and others by state authorities. With its system of semi-autonomous regional reserve banks, the Federal Reserve System was made to fit the structure of the U.S. banking system. Warburg pushed for the development of a bankers acceptance market in the United States. The Federal Reserve Act permitted national banks to issue bankers acceptances and authorized the Federal Reserve banks to purchase acceptances in the open market. The U.S. acceptance market never became large, however, and fell off sharply during the Depression. Thus, a key element of the original Warburg plan was never realized.

The Warburg-Aldrich Plan Warburg (1910b) first proposed the creation of a central bank with 20 regional branches controlled by bankers but regulated, to some extent, by government officials. His proposed United Reserve Bank would rediscount bills of exchange for its member banks, thereby providing liquidity to the market and establishing a lender of last resort that, following Bagehot’s strictures, would lend freely in a banking panic:

70

Michael D. Bordo and David C. Wheelock

The relationship between the central bank and the discount market is a most important one. While in normal times only a small proportion of the business is done by the central bank, the existence of this bank is all important to the whole financial structure, because even if a bank makes it a rule not to rediscount with the central bank and in its general business keeps independent of this institution, the fact remains that in case of need it can nevertheless rediscount with the central bank every legitimate bill, both bankers or mercantile acceptance, so that every legitimate bill represents a quick asset, on the realization of which every bank or banker can rely. Consequently no investor, bank, banker, private capitalist or financial institution will ever hesitate to buy good bills. Furthermore, there will not be in critical times any rush to sell good bills, as everybody in these countries knows that there is no better and safer investment, because for no other investment is there an equally reliable market. (Warburg, 1910a, p. 14)

Echoing Bagehot (1873), Warburg (1910a, p. 37) explained how the central bank should respond to crises: Thus certain periodic and normal demands for cash, as well as a domestic drain caused by distrust, must be met by paying out freely. A foreign drain, on the other hand, must generally be met by an energetic increase of the rate, while a drain both domestic and foreign must be treated by various combinations of both methods. (Warburg, 1910a, p. 37)

Under Warburg’s plan, the discount rate would be the key instrument of policy, but it would be supplemented by open-market operations to help make the discount rate effective, that is, to ensure that changes in the discount rate could always determine the behavior of market rates (Sayers, 1957). As in Europe, adherence to the official gold parity would anchor the price level. The United Reserve Bank would issue currency backed by gold and bills of exchange, and manage the gold standard, that is, intervene in the exchange market and manipulate the gold points, in accordance with the “rules of the game” as the European central banks did (Bordo et al., 2007). Warburg argued that a discount market would replace the call loan market as the principal source for liquidity for U.S. banks. This in turn would eliminate, in his opinion, a key source of financial instability – the link between the stock market and the banking system. The Aldrich bill drafted at Jekyll Island was similar in many ways to the Warburg plan. The Aldrich bill called for the establishment of a National Reserve Association, headquartered in Washington, DC. The association’s branches would be located throughout the United States and serve member commercial banks. The association would issue asset-backed currency and rediscount eligible paper consisting of short-term commercial and

The Promise and Performance of the Federal Reserve as Lender

71

agricultural loans for its members at a discount rate set by the National Association’s board of directors. The discount rate would be uniform throughout the nation. The association also would be permitted to conduct open market operations and exchange market intervention. THE FEDERAL RESERVE ACT

Congress rejected the Aldrich bill. Popular distrust of Wall Street power and of bankers in general (“the money trust”) killed it. Moreover, when the Democrats took control of Congress in the election of 1912, the Aldrich bill was considered anathema. Yet, under the direction of Carter Glass, the new chairman of the Senate Banking and Currency Committee, and with the aid of his advisor H. Parker Willis (a student of J. Laurence Laughlin and strong advocate of the real bills doctrine), a bill was put forward in early 1913 for what became the Federal Reserve Act. The act almost completely replicated the key monetary and international policy provisions of the Warburg plan and the Aldrich bill (Wicker, 2005). The initial Glass bill was later modified to include important revisions requested by President Woodrow Wilson to increase government oversight of the system, and the bill became known as the Glass-Owen bill (Robert Owen was chairman of the House Committee on Banking and Currency). The Federal Reserve System differed markedly from Aldrich’s proposed National Reserve Association in terms of structure and governance. Rather than a central organization with many branches, the Federal Reserve System consisted of twelve semi-autonomous regional reserve banks and the Federal Reserve Board, which had a general oversight role. Whereas the Federal Reserve Board was made up of five members appointed by the president and chaired by the secretary of the treasury, the reserve banks were owned by their member banks and managed by officers appointed by local boards of directors.6 A key difference between the Federal Reserve Act and the Aldrich plan was that the individual Federal Reserve banks set their own discount rates (subject to review by the Federal Reserve Board) and each bank was required to maintain a minimum reserve in the form of gold and eligible paper against its note and deposit liabilities.

Lender of Last Resort Provisions of the Federal Reserve Act The preamble to the Federal Reserve Act states that it is an “Act to provide for the establishment of Federal Reserve banks, to furnish an elastic

72

Michael D. Bordo and David C. Wheelock

currency, to afford means of rediscounting commercial paper, to establish a more effective supervision of banking in the United States, and for other purposes.” The act does not contain explicit instructions for how the Fed should respond in the event of a banking panic; in other words, how it should serve as lender of last resort. Apparently, the authors of the act believed that they had created a foolproof mechanism that would prevent panics from occurring in the first place.7 The Federal Reserve Act also did not address sources or forms of financial instability outside the banking system. For example, although trust companies had been at the epicenter of the panic of 1907, the act did not permit Federal Reserve membership for most trust companies, as well as mutual savings banks, building and loan companies, or other nonbank financial institutions. Membership was required only of national banks. State-chartered banks were permitted to become Fed members if they agreed to the same minimum capital, reserve, and other requirements imposed on national banks. Only member banks were given access to Fed services, including the discount window, although the act left open the possibility that discount window loans could be extended to nonmember banks in special circumstances with the approval of the Federal Reserve Board. Relatively few state-chartered banks joined the Federal Reserve System. By June 1915, only 17 state banks had become Fed members. Membership grew slowly to a peak of 1,620 state-chartered banks (compared with 19,345 nonmember banks) in June 1923. Member banks tended to be larger than nonmembers, however, and by June 1915 member banks held nearly half the total deposits of all U.S. commercial banks ($8.9 billion vs. 9.1 billion held by nonmember commercial banks). Still, even by June 1923, member banks held less than two-thirds of total U.S. commercial bank deposits ($27.1 billion vs. 10.6 billion in nonmember banks).8 The Fed’s member banks were required to maintain reserve balances with the Federal Reserve Banks, which the founders expected would reduce the concentration of correspondent balances held in the central money markets and invested in stock market call loans and thereby lessen the transmission of instability from the stock market to the banking system. Reformers argued that the flow of surplus funds to the central money markets, principally New York City, and invested in stock market loans had contributed to market instability and reduced the supply of credit available for commercial and agricultural borrowers, especially outside the principal financial centers. To address the problem of an inelastic currency (defined broadly to include both currency and bank reserves), the Federal Reserve Act permitted

The Promise and Performance of the Federal Reserve as Lender

73

member banks to rediscount eligible paper with Federal Reserve banks in exchange for currency (Federal Reserve notes) or reserve deposits. Federal Reserve notes were asset backed in the sense that the reserve banks were required to hold reserves in the form of eligible commercial paper or gold equal to their outstanding note issues (plus an additional 40 percent gold reserve). The stocks of Federal Reserve notes and member bank reserve deposits were elastic in that their volumes would vary with the amount of eligible paper that member banks rediscounted with the reserve banks, which in turn varied with fluctuations in the demands for currency and credit. The Federal Reserve Act imposed two checks on the amount of notes and bank reserve deposits the Fed could issue. First, each reserve bank was required to maintain gold reserves equal to at least 40 percent of its outstanding notes and 35 percent of its deposit liabilities. Federal Reserve banks were further required to hold eligible paper equal to 100 percent of their outstanding note issues.9 The Federal Reserve Act did not specify the criteria that reserve banks were to use in setting their discount rates, but clearly it was expected that a reserve bank would increase its discount rate as necessary to maintain adequate gold reserves. The Federal Reserve Act also limited the types and maturities of loans and securities that member banks could rediscount with the reserve banks, which served as a second brake on Federal Reserve credit. Glass and Willis were strong proponents of the real bills doctrine who believed that Federal Reserve credit should be extended only by the rediscounting of short-term, self-liquidating commercial and agricultural loans. The act permitted rediscounting of “notes, drafts, and bills of exchange arising out of actual commercial transactions,” but forbid rediscounting of loans or securities “covering merely investments or issued or drawn for the purpose of carrying or trading in stocks, bonds or other investment securities, except bonds and notes of the Government of the United States.” Further, the Federal Reserve Act specified that only those loans with a term to maturity of 90 days or fewer (180 days for agricultural loans) were eligible for rediscount. In setting a maximum term, Congress cited the experience of other countries. For example, Senator John F. Shafroth stated When we look around in the history of the world we find that . . . in England the paper [that is eligible for rediscounting] must run only 28 days, . . . in France it runs but 26 days, . . . in Germany it does not exceed 90 days, and that there is no bank in the world which discounts paper in excess of 90 days . . . [Does] it not become us, in the interest of caution, to say that until it is demonstrated the other way we had better adhere to 90-day paper? (Quoted in Hackley, 1973, p. 14)

74

Michael D. Bordo and David C. Wheelock

The provisions of the Federal Reserve Act defining eligible paper were similar to those in the Aldrich bill. However, the Aldrich bill would have permitted rediscounting of any direct obligation of the borrowing bank if approved by the secretary of the treasury and the bank’s local association which, as Wicker (2005, p. 86) notes, might have been particularly helpful during a banking panic. The Glass-Steagall Act of 1932 amended the Federal Reserve Act to permit the Fed to extend loans on the basis of any satisfactory collateral in emergency situations (Hackley, 1973, p. 100). The Federal Reserve Act granted access to the Fed’s discount window only to member banks. Further, the act specified that “no member bank shall act as the medium or agent of a nonmember bank in applying for or receiving discounts from a Federal Reserve Bank . . . except by permission of the Federal Reserve Board.” The provision granting the board authority to permit exceptions was not in either the House or Senate versions of the act, but was added in conference committee (Hackley, 1973, p. 119). During World War I, the board authorized the reserve banks to discount for nonmembers, with the endorsement of a member bank, notes secured by U.S. government securities if the proceeds were to be used for holding government securities (Hackley, 1973, pp. 118–19). Then in 1921, the board authorized the reserve banks to discount for member banks any eligible paper acquired from nonmember banks, but that authority was rescinded in 1923 (Hackley, 1973, p. 119). Thereafter, Federal Reserve credit was extended to nonmember banks only in exceptional circumstances with board approval. One such occurrence helped to end a local banking panic in Florida in 1929 (Carlson, Mitchener, & Richardson, 2011). However, in general, nonmember banks were shut out from Federal Reserve loans, which was especially problematic during the Great Depression when banking panics arose among nonmember banks. Fundamentally, the provisions of the Federal Reserve Act pertaining to the discount window were those advocated by Warburg and contained in the Aldrich bill.10 Warburg’s views were also reflected in sections of the act that permitted member banks to offer bankers acceptances based on international trade and that authorized the Federal Reserve Banks to rediscount or purchase acceptances in the open market. Similar to the discount rates they set for the rediscount of eligible paper, rates of discount (“bill buying rates”) were set by reserve banks on acceptances they offered to purchase in the open market. The Fed’s acceptance buying facility was closer in form to the Bank of England’s discount facility than to the Fed’s discount window. Typically, the reserve banks would purchase all of the eligible acceptances offered to them at their set bill buying rates. If the U.S. acceptance

The Promise and Performance of the Federal Reserve as Lender

75

market had developed to the extent it had in England (or in Germany) in the nineteenth century, it is conceivable that the Fed would have been a more effective lender of last resort during the Great Depression. However, as discussed in the next section, the U.S. acceptance market remained small and fell off sharply during the Depression. Although the Federal Reserve purchased a significant share of outstanding acceptances, those purchases generally were not a significant source of Federal Reserve credit during the Depression. THE FED’S PERFORMANCE AS LENDER OF LAST RESORT TO 1933

In addition to creating the Federal Reserve System, the Federal Reserve Act also extended for one year a provision of the Aldrich-Vreeland Act of 1908 that permitted commercial banks to form associations to issue emergency currency backed by commercial paper or certain long-term bonds in the event of a banking panic. Some $375 million of emergency currency was issued under the terms of the act when bank runs occurred at the start of World War I in August 1914. The response is widely credited with stemming the panic and ensuring continued loan growth (e.g., Sprague, 1915; Friedman & Schwartz, 1963; Wicker, 2005, pp. 44–9). The Federal Reserve banks began to operate in November 1914 but were minor players in the money market until the United States entered World War I in 1917 (West, 1977, pp. 181–6). The Fed then helped to finance the war effort by offering preferential discount rates on loans (“advances”) to member banks secured by government bonds.11 The reserve banks also purchased over $1 billion of government securities in the open market. In deference to the Treasury Department, the Fed maintained preferential discount rates on loans secured by government securities for some months after the war ended. The Fed regained its independence and began to raise its discount rate (and end preferential rates) in December 1919. The continued expansion of Federal Reserve credit after the war significantly reduced the reserve banks’ gold reserve ratios, from a postwar peak of 50.6 percent in June 1919 to a low of 40.6 percent in March 1920.12 The falling reserve ratio was a principal motivation for raising the discount rate, though the Fed also sought to control inflation and to limit stock market speculation (Wicker, 1966, pp. 37–45). The reserve banks maintained their discount rates at high levels even after inflation had ceased to be a threat and economic activity had begun to decline. Both wholesale prices and industrial production fell sharply in 1920, but the Fed did not reduce its discount rate until May 1921, when

76

Michael D. Bordo and David C. Wheelock

officials were confident that the reserve banks could maintain their reserve ratios (Chandler, 1958, p. 186). The Fed was heavily criticized for both postwar inflation and the subsequent deflation and decline in economic activity. Perhaps partly in response to this criticism, there appears to have been a marked change in Fed policy after 1921 that is evident in the system’s use of open-market operations in government securities. By 1923, the reserve banks had formed a committee to coordinate their open-market operations, and a Special System Investment Account was established at the Federal Reserve Bank of New York in December 1923 to handle the committee’s operations. Friedman and Schwartz (1963, Chapter 6) refer to the period from 1921 to 1929 as the “High Tide” of the Federal Reserve System, during which the Fed, under the leadership of New York Fed Governor Benjamin Strong, pursued a well-conceived monetary policy that promoted economic stability and low inflation. Although other studies have been more critical of Strong and his policies (e.g., Meltzer, 2003), economic activity and the price level were remarkably stable during this period. Moreover, despite the failures of several hundred, mostly small, rural banks, there were no episodes of banking distress or panic in the major financial centers. It seemed that the Federal Reserve Act had indeed solved the problems that had produced recurrent banking panics during the nineteenth and early twentieth centuries. The elimination of seasonal tightness in credit markets and interest rates was a principal goal of the Fed’s founders, who had noted that banking panics tended to occur at times of the year when the demands for currency and credit were at seasonal peaks. The Fed seems to have accomplished this objective as seasonal variations in money market interest rates dropped sharply after the Fed’s founding (Friedman & Schwartz, 1963, pp. 292–3; Miron, 1986). Federal Reserve credit exhibited a distinct seasonal pattern throughout the 1920s, reflecting increases in both discount window borrowing and Fed purchases of bankers acceptances at times of the year when currency and credit demands reached seasonal peaks and decreases in discount window borrowing and in the Fed’s acceptance holdings when seasonal demands ebbed (see Figure 2.1).13 Seasonal accommodation was largely automatic, as the Fed’s founders had intended. At relatively fixed discount rates on loans to member banks and on purchases of acceptances in the open market, the reserve banks made more discount window loans and purchased more bankers acceptances at times of the year when demands for credit and currency were high. Unfortunately, the Great Depression demonstrated that accommodating seasonal variations in money and credit demand was not sufficient to eliminate the problem of banking panics.14

The Promise and Performance of the Federal Reserve as Lender

77

2,000 1,800 1,600

$ Millions

1,400

Federal Reserve Credit

1,200 1,000 800

Discount Window Credit

600 400 200 0 1922

Bankers Acceptances 1923

1924

1925

1926

1927

1928

1929

Figure 2.1. Federal Reserve credit, monthly, 1922–1929. Source: Banking and Monetary Statistics, 1914–41. Board of Governors of the Federal Reserve System.

The Great Depression The Fed’s policy failures during the Great Depression are legendary. The Fed – specifically, the Federal Reserve Bank of New York – reacted swiftly to the October 1929 stock market crash by lowering its discount rate and lending heavily to banks, and by purchasing government securities in the open market. However, the Fed largely ignored the banking panics and failures of 1930–33, and did little to arrest large declines in the price level and output. The Fed clearly failed to serve effectively as lender of last resort.

Fed Policy from the Stock Market Crash to Bank Holiday Figure 2.2 shows the level and composition of Federal Reserve credit during 1929–34, providing one measure of the Fed’s response to the major financial crises of the Great Depression. In the figure, we plot the sum of Federal Reserve discount window loans and holdings of bankers acceptances (labeled Federal Reserve loans), as well as Federal Reserve holdings of U.S. government securities and other sources of Federal Reserve credit (mainly float). Following the stock market crash, the Federal Reserve Bank of New York used open-market purchases and liberal discount window lending to inject reserves into the banking system, which enabled New York City banks to absorb a large amount of loans made by securities brokers and dealers. The New York Fed’s actions were “timely and effective” in containing the

78

Michael D. Bordo and David C. Wheelock 4,000 3,500

$ Millions

3,000

Final Banking Panic Other Fed Credit

2,500 2,000 1,500

U.K. Off Gold Standard Stock Market Crash Fed U.S. Govt. Securities Portfolio First Banking Panic

1,000 500 Federal Reserve Loans 0 1929

1930

1931

1932

1933

1934

Figure 2.2. Federal Reserve credit, 1929–34. Source: Banking and Monetary Statistics, 1914–41. Board of Governors of the Federal Reserve System.

crisis and preventing widespread panic in money markets and among bank depositors (Friedman & Schwartz, 1963, p. 339). The Federal Reserve Board reluctantly approved the New York Fed’s actions ex post, but many members expressed displeasure that the New York Fed had acted independently. The New York Fed pressed for additional easing in early 1930. However, the Federal Reserve Board rejected several requests for discount rate cuts and additional open-market purchases. As Figure 2.2 shows, total Federal Reserve credit fell by about one-third during the first half of 1930, mainly because of declines in discount window loans and Fed purchases of bankers’ acceptances. As Figure 2.3 shows, the monetary base and broader measures of the money stock mirrored Federal Reserve credit outstanding – increasing sharply after the stock market crash, but then falling with the decline in Fed credit during 1930. Friedman and Schwartz (1963) contend that the decline in the money stock was the main cause of the subsequent decline in economic activity.15 The stock market crash was the first in a series of financial shocks during the Great Depression. Friedman and Schwartz (1963) identify major banking panics in the fourth quarter of 1930, early 1931, fourth quarter of 1931, and in February–March 1933. As Figure 2.2 shows, total Federal Reserve credit surged briefly following the stock market crash and during the banking panics of October–December 1930, September–December 1931 (which followed the United Kingdom’s decision to leave the gold standard), and January–March 1933. On each occasion, the increase in Federal Reserve credit (and its impact on the monetary base) was quickly reversed. Moreover,

The Promise and Performance of the Federal Reserve as Lender 18,000

79 60,000

16,000 50,000 Money Stock (right axis)

14,000

40,000

10,000 Monetary Base

30,000

8,000 6,000

20,000

4,000

panic

2,000 crash 0 1929

$ Millions

$ Millions

12,000

Federal Reserve Credit

gold

10,000

panic 0

1930

1931

1932

1933

1934

1935

1936

1937

1938

Figure 2.3. Federal Reserve credit and the monetary aggregates. Source: Federal Reserve Credit (Banking and Monetary Statistics, 1914–41), St. Louis Adjusted Monetary Base (FRED), Money Stock (Friedman and Schwartz, 1963, Appendix A, Table A-1).

as Figure 2.3 shows, when Federal Reserve credit finally began to grow in 1932, it only temporarily halted the decline in the broader money stock. Why did the Fed permit Federal Reserve credit to contract after each financial shock of 1929–33? Meltzer (2003) argues that Fed officials misinterpreted the signals coming from money market interest rates and discount window borrowing. Following the policy guidelines that Federal Reserve Bank of New York Governor Benjamin Strong and his colleagues had developed during the 1920s, policymakers inferred that low levels of interest rates and borrowing meant that monetary conditions were exceptionally easy and that there was no benefit, and possibly some risk, from putting more liquidity into the banking system.16 Strong explained the use of the level of discount window borrowing as a guide to policy as follows: Should we go into a business recession while the member banks were continuing to borrow directly 500 or 600 million dollars . . . we should consider taking steps to relieve some of the pressure which this borrowing induces by purchasing government securities and thus enabling member banks to reduce their indebtedness. . . . As a guide to the timing and extent of any purchases which might appear desirable, one of our best guides would be the amount of borrowing by member banks in the principal centers. . . . Our experience has shown that when New York City banks are borrowing in the neighborhood of 100 million dollars or more, there is then some real pressure for reducing loans, and money rates tend to be markedly higher than the discount rate. On the other hand, when borrowings of these banks

80

Michael D. Bordo and David C. Wheelock

are negligible . . . the money situation tends to be less elastic and if gold imports take place, there is liable to be some credit inflation, with money rates dropping below our discount rate. When [New York City] member banks are owing us about 50 million dollars or less the situation appears to be comfortable, with no marked pressure for liquidation.17

Discount window borrowing declined sharply, from $500 million for all Federal Reserve member banks in January 1930 ($39 million for New York City banks) to $231 million in April 1930 ($17 million for New York City banks), and $226 million in July 1930 ($0 for New York City banks). Fed officials interpreted the decline as indicating that monetary conditions were extremely easy and that additional stimulus was not required. For example, the governor of the Federal Reserve Bank of San Francisco argued in June 1930 that “with credit cheap and redundant, we do not believe that business recovery will be accelerated by making credit cheaper and more redundant.”18 Indeed, some officials described monetary conditions as too easy and argued for a tighter policy. For example, in January 1930, the governor of the Federal Reserve Bank of Minneapolis wrote that “I cannot see the desirability of further ease of credit. It seems to me money is getting almost ‘sloppy.’”19 Several Fed officials believed that Federal Reserve credit should be withdrawn whenever economic activity slows. For example, the governor of the Federal Reserve Bank of Philadelphia stated, “We have been putting out credit in a period of depression, when it was not wanted and could not be used, and will have to withdraw credit when it is wanted and can be used.”20 Federal Reserve credit increased temporarily in late 1930, as shown in Figure 2.2. Federal Reserve credit normally rose in the autumn when loan demand and interest rates tended to rise. However, in December 1930, the Federal Reserve Bank of New York also purchased $175 million of U.S. government securities and bankers acceptances to ease financial market strains that appeared following the failure of the Bank of United States. Numerous banks failed throughout the United States between October and December 1930. Most were small banks that were not members of the Federal Reserve System and thus were unable to borrow at the Fed’s discount window.21 Friedman and Schwartz (1963) note that Fed officials felt no particular responsibility for nonmember banks. However, they argue that the Fed made a critical error in not saving the Bank of United States, which was a midsized New York City bank and a member of the Federal Reserve System. Although the Federal Reserve Bank of New York participated in discussions about a possible merger to save the Bank of United States, those talks broke down when neither the New York Fed nor the New York Clearing House banks would guarantee $20 million of the Bank of United States assets

The Promise and Performance of the Federal Reserve as Lender

81

(Meltzer, 2003, pp. 323–4). Instead, the Fed and clearinghouse elected to let the Bank of United States fail and focus on containing the resulting fallout. Federal Reserve credit outstanding declined sharply in early January 1931 as money market strains eased. However, in 1931, the economic contraction deepened, deflation took hold, and interest rates and discount window borrowing declined still further. Another wave of bank failures occurred in the first half of 1931, but again most of the banks that failed were nonmember banks located outside of New York City and other financial centers. For the first time, banks began to hold excess reserves, that is, reserves in excess of legal requirements. Fed officials took this as further evidence of exceptionally easy monetary conditions and considered engaging in open-market sales to “soak up” excess liquidity (Meltzer, 2003, p. 328). The next major financial shock occurred in late September 1931. After running through most of its gold and foreign exchange reserves, the United Kingdom abandoned the gold standard on September 21, 1931, and allowed the pound to float freely. Speculation that the United States would soon also leave the gold standard caused large withdrawals of gold and currency from U.S. banks. The Federal Reserve responded by increasing its discount and acceptance buying rates in an attempt to halt and then reverse the gold outflow and to demonstrate the system’s resolve to maintain the gold standard. Federal Reserve officials interpreted their response to the gold outflow as being consistent with Bagehot’s rule to lend freely at a high interest rate (Meltzer, 2003, p. 348). The Fed did not make significant open-market purchases to offset the withdrawal of gold and currency from banks, however, which exacerbated the decline in the monetary aggregates (see Figure 2.3). Moreover, when gold began to flow back into the banking system, Federal Reserve credit outstanding fell by more than the gold inflow, which resulted in a net decline in total bank reserves. Fed officials apparently were hesitant to make open-market purchases because they saw a “disinclination on the part of member banks to use Federal Reserve credit for the purpose of extending credit to their customers.”22 Besides doubting that open-market purchases would serve any useful purpose, at least some Fed officials were concerned that large open-market purchases would threaten the system’s gold reserves. Although researchers subsequently have concluded that the Fed did have sufficient gold reserves to make significant open-market purchases (e.g., Friedman & Schwartz, 1963; Bordo, Choudhri, & Schwartz, 2002), Fed officials may have been concerned that large purchases would have touched off a resumption of gold outflows (Wicker, 1966). In any event, the excuse became moot when Congress enacted legislation in February 1932 that enabled the Fed to use U.S. government securities as collateral for Federal Reserve notes.23 Under

82

Michael D. Bordo and David C. Wheelock

pressure from Congress, the Fed then purchased approximately $1 billion of government securities between February and August 1932. Discount window loans totaled $848 million when the Fed began to purchase government securities in February 1932, and, hence, monetary conditions were tight according to the Fed’s traditional policy guide. Discount window borrowing declined and banks began to build up substantial excess reserves as the Fed continued its purchases. Several Fed officials interpreted the increase in excess reserves as indicating that the Fed’s purchases had little benefit. The Fed ended its purchases when the level of discount window borrowing had fallen to what it had been before Britain left the gold standard. Nonetheless, the purchases caused Federal Reserve credit to rise substantially (see Figure 2.2), which for a few months arrested the decline in the money stock (see Figure 2.3). The final and most severe banking crisis of the Depression began in February 1933. Banking panics, marked by heavy withdrawals of currency and gold reserves, swept across the United States. As it had done in response to gold outflows in 1931, the Fed reacted by increasing its discount and acceptance buying rates. Federal Reserve credit increased sharply in March 1933, as discount window loans rose from $253 million on February 8 to $1.4 billion on March 8, and the Fed purchased some $400 million of bankers acceptances. The Fed also purchased $100 million of government securities, but this was far too little to offset the decline in bank reserves caused by currency and gold withdrawals. In response to a request from the secretary of the treasury for larger purchases of government securities, Federal Reserve Governor Eugene Meyer replied that a rise in bond yields was a “necessary readjustment in a market which has been too high” and that “Purchases of Government securities at the present time would be inconsistent from a monetary standpoint.”24 Instead of supplying additional liquidity to the banking system, the Federal Reserve Board voted to suspend the Fed’s gold reserve requirement and to recommend that President Hoover declare a national bank holiday. This action and many others were subsequently taken by President Roosevelt on his first day in office on March 5, 1933. WHY DID THE FED FAIL TO ACT AS LENDER OF LAST RESORT DURING THE DEPRESSION?

Many studies have considered why the Federal Reserve failed to act effectively as lender of last resort during the Great Depression. Friedman and Schwartz (1963) emphasize the Fed’s decentralized structure and lack of strong leadership. Without a forceful leader, they argue, the system was

The Promise and Performance of the Federal Reserve as Lender

83

paralyzed by in-fighting, petty jealousies, and sharp differences of opinion. Other studies downplay the significance of the Fed’s structure, contending that the policies pursued by the Fed during the Depression were fundamentally consistent with those of the 1920s (e.g., Wicker, 1966; Wheelock, 1991; Meltzer, 2003). Those studies contend that Fed officials misinterpreted the behavior of nominal interest rates and the level of borrowing from the Fed’s discount window. Officials interpreted low nominal interest rates and little borrowing at the discount window as evidence that monetary conditions were exceptionally easy and that there was little more the Fed could or should do to promote economic recovery. Still other studies focus on the role of the gold standard (Temin, 1989; Eichengreen, 1992). The Federal Reserve Act affirmed the fundamental role that the gold standard played in the U.S. monetary system. Federal Reserve banks were required to maintain gold reserves to back their note and deposit liabilities. Although the act permitted the Federal Reserve Board to suspend the system’s gold reserve requirement, Fed officials were deeply committed to maintaining the gold standard and were extremely reluctant to take any action that would threaten or even that might be perceived as threatening the gold standard. We believe that the Fed’s decentralized structure, misreading of monetary conditions, and commitment to the gold standard all contributed to the Fed’s highly deflationary monetary policy of 1929–33 and limited response to banking panics. However, we also believe that there is more to the story, especially as to why the Fed failed to prevent or offset serious banking panics during the Depression. In particular, we argue that the Federal Reserve Act failed to re-create the features of the British banking system that made the Bank of England an effective lender of last resort during the late nineteenth and early twentieth centuries; or, similarly, the features of the German banking system that made the Reichsbank an effective lender of last resort that Warburg admired. Further, we argue that the restrictions on discount window lending imposed by the Federal Reserve Act were both too limiting and left too much to the discretion of policymakers in administering the discount window to make it an effective mechanism for responding to banking panics.

The Discount Window – A Flawed Lender of Last Resort Mechanism The authors of the Federal Reserve Act intended the discount window to be the primary means by which the Fed would “furnish an elastic currency.”

84

Michael D. Bordo and David C. Wheelock

The authors sought to provide a mechanism that would ensure ample supplies of currency and bank reserves to support commercial and agricultural activity, but not be a source of funds for speculation or long-term investment. Accordingly, the types of loans and securities that were eligible for rediscounting with Federal Reserve Banks were restricted to shortterm commercial and agricultural paper (and U.S. government securities). During the Depression, many banks apparently lacked paper that was acceptable for rediscounting with Federal Reserve Banks (or that could be used as collateral for advances from the Fed) under the Federal Reserve Act. Although the total amount of eligible paper (government securities and private commercial paper and acceptances) far exceeded the amount of discount loans made by the Fed, holdings of eligible paper varied widely across banks of different sizes and across Federal Reserve districts (Chandler, 1971, pp. 227–33). According to Chandler (1971, p. 232), “the narrow definition of eligible assets, symbolizing a persistence of commercial loan ideas, contributed to bank closings and to liquidation of credit by banks that succeeded in remaining open.”25 A second problem with the discount mechanism as designed and implemented under the Federal Reserve Act was that member banks were apparently quite reluctant to borrow from the Fed in the event of a crisis. In part, this reluctance reflected the Fed’s administration of the discount window. Throughout the 1920s, Fed officials had tried to discourage banks from continuous borrowing and sought to instill the idea that banks are hesitant to borrow from the Fed and do so reluctantly when confronted with a short-term liquidity need (see Meltzer, 2003, pp. 161–5, for discussion). A number of Fed officials, especially members of the Federal Reserve Board, were concerned that banks were borrowing from the Fed to finance loans for the purchase of stocks. In February 1929, the Federal Reserve Board directed the reserve banks to crack down on banks that were borrowing from the Fed and simultaneously making speculative loans: “A member bank is not within its reasonable claims for rediscount facilities at its Federal Reserve Bank when it borrows either for the purpose of making speculative loans or for the purpose of maintaining speculative loans” (quoted by Chandler, 1971, pp. 56–7). The message clearly was that the reserve banks should administer the discount window more tightly and that member banks should think twice before coming to the Fed for a loan. The reluctance to borrow at the discount window took on another dimension during the Depression, when bank depositors became concerned about the condition of the banks in which they held funds. According to Friedman and Schwartz (1963, pp. 318–19), “The aversion to borrowing by banks,

The Promise and Performance of the Federal Reserve as Lender

85

which the Reserve System had tried to strengthen during the twenties, was still greater at a time when depositors were fearful for the safety of every bank and were scrutinizing balance sheets with great care to see which banks were likely to be the next to go.” Wheelock (1990) estimates a model of the demand for discount window loans as a (nonlinear) function of the spread between the discount rate and a market interest rate, the change in nonborrowed reserves, and a measure of economic activity, and finds evidence that the demand for discount window loans shifted downward during the Depression. This suggests greater reluctance (or inability) of banks to borrow at the discount window during the Depression than during the 1920s, and that the discount mechanism was not well suited for ameliorating a banking panic. Further, it suggests that the level of discount window borrowing was an especially poor indicator of monetary conditions during the Depression.

Bankers Acceptance Purchases as an Alternative to the Discount Window A second mechanism for supplying currency or bank reserves in the event of a crisis – Federal Reserve purchases of bankers acceptances – also failed to prevent or counteract banking panics during the Depression. The Fed did make large purchases of bankers acceptances during banking panics in the fall of 1931 and in March 1933, but the purchases were not large enough to offset the effects of currency and gold withdrawals from the banking system. The 1931 panic was triggered by the United Kingdom’s departure from the gold standard on September 21. Speculation that the United States would soon follow Britain off the gold standard precipitated large withdrawals from U.S. banks. The Fed responded to the crisis by raising its discount rates on discount window loans and acceptance purchases in an effort to discourage gold outflows and to shore up confidence in the Fed’s resolve to stay on the gold standard. Despite the higher cost of borrowing from the Fed, both discount window borrowing and sales of acceptances to the Fed rose sharply. Discount window loans increased from $263 million on September 16 to $698 million on October 21 and peaked at $1,024 million on December 30. The Fed’s holdings of acceptances rose from $218 million on September 16 to a peak of $769 million on October 21. Despite the increases in discount loans and Fed purchases of acceptances, the total reserves of Fed member banks fell by $142 million between September 16 and October 21 and continued to fall until the end of February 1932. The Fed allowed its acceptance portfolio to dwindle. After peaking at $769 million

86

Michael D. Bordo and David C. Wheelock

on October 21, the Fed’s acceptance holdings declined to $327 million on December 30 and continued to fall steadily to under $100 million by mid-March 1932. The banking crisis of 1933 resulted in a similar short-lived increase in both discount window loans and Fed purchases of bankers acceptances. Discount loans increased sharply from $327 million on February 27 to $1,414 million on March 15, and the Fed’s holdings of acceptances rose from $180 million to $417 million. These increases, however, were insufficient to replace all of the reserves the banking system lost to outflows of currency and gold. Although the Fed’s purchases of bankers acceptances provided some support to the banking system during the panics, the acceptance market was small and highly concentrated in New York City, which limited the usefulness of Fed purchases in a crisis. The size of the acceptance market fell sharply from about $1.6 billion at the end of 1929 to around $700 million by mid-1932 (Balabanis, 1935). For comparison, as of September 16, 1931, the reserves of Federal Reserve member banks totaled $2.4 billion. Moreover, the market was concentrated in New York City, and most banks outside of New York City held few, if any, acceptances they could sell to the Fed.26 The Fed purchased approximately 80 percent of the outstanding acceptances in October 1931 and 50 percent in March 1933 (Balabanis, 1935). Conceivably, the Fed could have made it more attractive for banks to sell acceptances to the reserve banks by lowering their bill buying rates, but it seems doubtful that they could have purchased enough acceptances to prevent declines in bank reserves, especially in 1931 and early 1932.

The Fed’s Decentralized Structure The authors of the Federal Reserve Act emphasized that they were not creating a U.S. central bank, but rather a federal system of reserve banks that would respond to and support the banking and currency needs of their individual districts. Accordingly, each reserve bank had the discretion to set its own discount rate and to administer its discount window, and each was required to satisfy a reserve requirement. The purpose of this structure was to reduce the concentration of the banking system’s reserves in the central money markets, especially New York City, and to limit the power of New York and Washington over the nation’s banks and economy. The Fed’s decentralized structure, however, proved unwieldy, especially in responding to financial crises. Friedman and Schwartz (1963) contend that the Fed’s failures during the Great Depression stemmed from a lack of effective leadership, which enabled parochial interests and petty jealousies

The Promise and Performance of the Federal Reserve as Lender

87

to hamstring policy. The individual reserve banks acted competitively, rather than cooperatively, at critical points during the Depression. For example, in March 1933, the Federal Reserve Bank of Chicago refused a request from the New York Fed to exchange gold for U.S. government securities when gold outflows threatened to push the New York Bank’s reserve ratio below its legal minimum (Meltzer, 2003, p. 387). Although the Federal Reserve Board eventually required the Chicago Fed and other reserve banks to lend to New York, the episode illustrates how the system’s structure hampered its response to crises. The Banking Acts of 1933 and 1935 addressed this by substantially increasing the authority of the Federal Reserve Board over both discount and open-market policy. Although the authors of the Federal Reserve Act sought to impose a rule that would guarantee adequate supplies of currency and bank reserves to support economic activity and prevent banking panics, in fact, the act left considerable discretion to the individual reserve banks and the Federal Reserve Board for implementing policy. Some of the reserve banks were more liberal in determining and valuing acceptable discount window collateral than others (Chandler, 1971, p. 233). In addition, some of the reserve banks moved more aggressively than others to supply currency to banks threatened by a panic. According to Richardson and Troost (2009), the Federal Reserve Bank of Atlanta responded to local panics by moving large quantities of cash to affected regions, extending emergency loans to member banks, and helping member banks extend loans to nonmember banks. Comparing the performance of the Atlanta and St. Louis Federal Reserve Banks during a 1930 banking panic that straddled both districts, Richardson and Troost (2009) conclude that the Atlanta Fed’s more aggressive response to the panic kept bank failure rates lower and commercial lending and economic activity higher in the Atlanta district than in the St. Louis district. Further evidence that officials of the Federal Reserve Bank of Atlanta were aggressive in responding to banking distress is reported by Carlson, Mitchener, and Richardson (2011), who investigated a banking panic that arose in Florida in 1929. The study reports that the Atlanta Fed shipped large amounts of currency to the afflicted region to shore up the confidence of bank depositors and that the reserve bank secured permission from the Federal Reserve Board to lend to nonmember institutions. The study concludes that the bank’s swift action stopped the panic and held down the number of bank failures. A third example of discretion on the part of a reserve bank was the Federal Reserve Bank of New York’s aggressive response to the 1929 stock market

88

Michael D. Bordo and David C. Wheelock

crash. Following the crash, the New York Fed extended discount window loans liberally to member banks so that they could take on stock exchange loans held by brokers. The bank also purchased more than $100 million of government securities in the open market. The actions by the Federal Reserve Banks of Atlanta and New York suggest that the Federal Reserve had the tools and the power to respond effectively to financial crises. However, an effective response required leaders who were willing to improvise and test the limits of the Federal Reserve Act. The act did not provide an automatic, foolproof mechanism for dealing with crises, as the founders had hoped. Instead, effective lender-of-last-resort action depended a great deal on the discretion of individual policymakers. CONCLUSION

The lender-of-last-resort provisions of the Federal Reserve Act mimicked features of the Aldrich bill of 1912, which in turn originated with the Warburg plan of 1910. In particular, Warburg envisioned that the central bank would rediscount bankers acceptances to backstop a highly liquid and deep money market like those of England and other advanced European countries and to provide lender-of-last-resort facilities in the event of a financial crisis. The Fed navigated its first 15 years without any banking panics. It succeeded in smoothing out seasonal fluctuations in short-term interest rates which had contributed to instability in the National Banking era. The Federal Reserve failed to prevent a series of banking panics in the early 1930s, however, which worsened the Great Depression. An extensive literature has posited several explanations for the Fed’s failure. These include (1) flaws in the system’s structure that impeded coordination between the Federal Reserve Board and the reserve banks, especially after the death of New York Fed Governor Benjamin Strong in 1928 (Chandler, 1958; Friedman & Schwartz, 1963); (2) devotion to the gold standard, which kept the Fed from following expansionary policies to offset banking panics (Temin, 1989; Eichengreen, 1992); and (3) adherence to a flawed policy framework that relied on nominal interest rates and the level of discount window borrowing as policy guides (Wheelock, 1991; Meltzer, 2003). We argue that a fourth important factor was the failure of the Federal Reserve Act to provide a discount mechanism and money market environment of the sort that had enabled the Bank of England and other European central banks to function effectively as lenders of last resort. This was manifest in three flaws: (1) the reluctance of member banks to turn to the

The Promise and Performance of the Federal Reserve as Lender

89

discount window in times of stress (the “stigma” problem); (2) the Fed’s limited membership and the fact that except in extreme circumstances, only member banks had access to the discount window; and (3) the restrictive eligibility requirements on collateral posted for borrowing at the Fed’s discount window. Furthermore, the Federal Reserve Act did nothing to reform the inherently unstable unit banking system. A better alternative might have been to allow nationwide branch banking and consolidation of the banking industry as in Canada and the United Kingdom, which over time proved to be a more stable market structure with less need for a lender of last resort (Bordo, Redish, & Rockoff 2012). The Federal Reserve Act sought to create a U.S. bankers acceptance market and offered support to that market by authorizing the reserve banks to purchase bankers acceptances in the open market. The Fed’s acceptance purchase facility was similar to the Bank of England’s discount window in that the Fed purchased all of the eligible acceptances offered to it, much as the Bank of England rediscounted all good-quality bills offered by dealers under its “frosted glass” discount window. The acceptance facility had at least two characteristics that seem good lender-of-last-resort practice: (1) Lending was to the market, rather than to individual institutions or classes of institutions, against a standard financial instrument; and (2) the facility entailed little scope for hesitation or discretion, except in setting the discount (“bill buying”) rate. Bankers acceptances never became the core instrument of the U.S. money market, however, and the acceptance market fell off sharply during the Depression. Thus the United States never developed the money market conditions that had enabled European central banks to be effective lenders of last resort before World War I.27 In response to the disaster of the early 1930s, the Banking Acts of 1933 and 1935 made significant changes in the structure and authority of the Federal Reserve System. The acts concentrated policymaking authority within the Board of Governors, expanded the Fed’s ability to lend on the basis of any sound collateral, and authorized the Fed to lend to nonbank financial institutions in a crisis (the infamous Section 13(3)). In addition, the banking system was subject to major reforms, including the introduction of federal deposit insurance, the forced separation of commercial and investment banking (Glass-Steagall Act), the regulation of deposit interest rates (Regulation Q), and strict limits on market entry. These reforms were intended to enhance the Fed’s ability to respond to crises while making the banking system less prone to instability. What are the lessons of the Fed’s early experience as the lender of last resort? The Fed’s early history provides much information about what

90

Michael D. Bordo and David C. Wheelock

does not work. In particular, it shows that a lender-of-last-resort system that works well in one environment may not work in another environment. A lender-of-last-resort structure should match the financial environment. Paul Warburg sought to emulate the European central bank mechanism and discount market. For political economy reasons (especially the deep-ingrained fear of concentration of financial power), U.S. banking institutions were not fully adapted to the European system (e.g., the United States retained a dual state and federal banking system, maintained unit banking, and the single-name commercial paper market remained dominant after the Fed was founded). The Federal Reserve Act overcame some of the flaws of the National Banking System (e.g., the inelastic currency) that promoted instability, but not all of them. Perhaps the Fed’s lenderof-last-resort mechanism would have performed better with a Canadian– European-style branch banking system coupled with a deep acceptance market. Amendments to the Federal Reserve Act in the 1930s addressed many of the technical flaws that had caused the Fed to be an ineffective lender of last resort in the Depression. In addition, other reforms promoted stability of the banking system (e.g., deposit insurance, the Glass-Steagall separation of commercial and investment banking). But these created a banking system that was slow to innovate and lost business to less-regulated financial institutions and markets (the “shadow” banks). The reforms of the 1930s focused on protecting bank depositors and preventing runs by depositors, and hence they proved only partly helpful during the crisis of 2007–8. As with the original Federal Reserve Act, the 1930’s reforms did not contemplate how to protect the banking system from instability coming from outside the banking system (e.g., runs on investment banks). The Section 13(3) lending programs created by the Fed in 2007–8 were, for the most part, helpful in alleviating the crisis, but required considerable discretion and judgment on the part of Fed officials – there was no playbook to follow, so policymakers had to invent on the fly. Moreover, the Fed seemed to have had no way to save the financial system without resorting to bailouts (e.g., AIG), and these actions may cause other problems (e.g., bailouts lead to moral hazard and compromise the Fed’s independence). A key lesson of both the Fed’s early experience and the crisis of 2007–8 is that the tools of a lender of last resort must match the financial environment. A lender-of-last-resort mechanism must adapt to be effective. It remains to be seen whether the reforms of 2010 have got it right. We now will have a Financial Stability Council charged with keeping track of key risks to the

The Promise and Performance of the Federal Reserve as Lender

91

financial system and regulating systemically important firms and markets. We now will have an expedited mechanism for winding down troubled large nonbank financial institutions. We also have prevented the Fed from making Section 13(3) loans to individual firms (no more bailouts?). Will it all work to protect the financial system from the next crisis? Only time will tell. APPENDIX

Legislated Changes to the Federal Reserve’s Lending Authority28 December 1913: The Federal Reserve Act is signed into law. The act authorizes Federal Reserve Banks to rediscount eligible paper for member commercial banks. September 1916: An amendment to the Federal Reserve Act adds Section 13(8), which authorizes Federal Reserve Banks to make advances to member banks on their own notes secured by paper eligible for discount or purchase or by the “deposit or pledge of bonds or notes of the United States.” February 1932: The Glass-Steagall Act amends the Federal Reserve Act by permitting U.S. government securities to serve as collateral for Federal Reserve notes. The act also adds Section 10(a), which permits Federal Reserve banks to lend to groups of five or more member banks, and Section 10(b), which permits Federal Reserve banks to lend to any member bank with capital not exceeding $5 million on the basis of any satisfactory assets, whether or not technically eligible for rediscount, in exceptional and exigent circumstances. The provision stipulates that any such loans require the approval of at least five members of the Federal Reserve Board and shall bear an interest rate of not less than 1 percent above the regular discount rate. Sections 10(a) and 10(b) are set to expire in March 1933. July 1932: The Emergency Relief and Construction Act adds Section 13(3) to the Federal Reserve Act, permitting the reserve banks to lend in unusual and exigent circumstances and with the authority of the Federal Reserve Board, to any individual, partnership, or corporation unable to secure adequate credit accommodations from other banking institutions. Section 13(3) loans must be secured by paper “of the kinds and maturities eligible for discount for member banks under other provisions” of the Federal Reserve Act and carry the endorsement of the borrower or a thirdparty surety. March 1933: The Emergency Banking Act extends Section 10(b) until March 1934 and eliminates the provisions that restrict Section 10(b) loans to banks with capital not exceeding $5 million and requires the approval of

92

Michael D. Bordo and David C. Wheelock

at least five members of the Federal Reserve Board. The act also adds Section 13(13) to the Federal Reserve Act, authorizing Federal Reserve advances to any individual, partnership, or corporation for periods of not more than 90 days on notes secured by direct obligations of the United States or issued or fully guaranteed by U.S. agencies. A subsequent amendment to the Emergency Banking Act authorizes for a period of one-year Federal Reserve loans to nonmember banks in limited circumstances on the same terms as advances to member banks under Section 10(b). This authority expired in March 1934 and was not renewed. June 1934: The Federal Reserve Act is amended by adding Section 13(b), which authorizes Federal Reserve banks to make working capital loans to businesses. Section 13(b) was repealed under the Small Business Investment Company Act of 1958 (Hackley, 1973, pp. 135, 145). August 1935: The Banking Act of 1935 makes permanent Sections 10(b) and 13(3) of the Federal Reserve Act. The act removes the requirement that discount loans under Section 10(b) be made only under “exceptional and exigent circumstances.” It also eliminates a provision of Section 13(3) requiring that loans to individuals, partnerships, or corporations be secured by both collateral eligible for discount by member banks and the endorsement of the borrower or a third-party surety. March 1980: The Monetary Control Act of 1980 subjects most depository institutions to reserve requirements and provides that any depository institution that holds transactions accounts or nonpersonal time deposits subject to Federal Reserve requirements shall be entitled to the same discount and borrowing privileges as Federal Reserve member banks. December 1991: The Federal Deposit Insurance Corporation Improvement Act of 1991 amends Section 13(3) of the Federal Reserve Act by deleting the requirement that Federal Reserve loans to individuals, partnerships, or corporations be secured by collateral “of the kinds and maturities made eligible for discount for member banks.” The act also subjects the Federal Reserve to a potential liability to the FDIC on discount window loans to undercapitalized or critically undercapitalized depository institutions. July 2010: The Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010 amends the Federal Reserve Act to permit only “broad-based” liquidity facilities under Section 13(3) and requires approval of the secretary of the treasury for any facilities established under Section 13(3). Further, the Federal Reserve must disclose the identities of borrowers under any 13(3) facilities within one year after the termination of the facility. The legislation also requires the Federal Reserve to disclose the identities of all discount window borrowers after a two-year delay.

The Promise and Performance of the Federal Reserve as Lender

93

References Bagehot, W. (1873). Lombard Street: A description of the money market. London: H. S. King. Balabanis, H. P. (1935). The American discount market. Chicago: University of Chicago Press. Bernanke, B. S. (1983). Nonmonetary effects of the financial crisis in propagation of the Great Depression. American Economic Review, 73 (3), 257–276. Bignon, V., Flandreau, M., & Ugolini, S. (2011). Bagehot for beginners: The making of lending of last resort operations in the mid 19th century Economic History Review, 65(2), 580–608. Board of Governors of the Federal Reserve System. (1943). Banking and monetary statistics, 1914–1941. Washington, DC: U.S. Government Printing Office. Bordo, M. D. (1990). The lender of last resort: Alternative views and historical experience. Federal Reserve Bank of Richmond. Economic Review, 76 (1), 18−29. Bordo, M. D., & Eichengreen, B. (2002). Crises now and then: What lessons from the last era of financial globalization. In P. Mizen (Ed.), Monetary history, exchange rates and financial markets: Essays in honor of Charles Goodhart (Vol. 2, pp. 52–91). London: Elgar. Bordo, M. D., Ehsan, C., & Schwartz, A. J. (2002). Was expansionary monetary policy feasible during the Great Contraction? An examination of the gold standard constraint. Explorations in Economic History, 39 (1), 1−28. Bordo, M. D., Humpage, O., & Schwartz, A. J. (2007). The historical origins of U.S. exchange market intervention policy. International Journal of Finance and Economics, 12 (2), 109–132. Bordo, M. D., Rappoport, P., & Schwartz, A. J. (1992). Money versus credit rationing: Evidence for the National Banking Era, 1880–1914. In C. Goldin & H. Rockoff (Eds.), Strategic factors in nineteenth century American economic history, A volume to honor Robert W. Fogel (pp. 189–224). Chicago: University of Chicago Press. Bordo, M. D., Redish, A., & Rockoff, H. (1996). A comparison of the stability and efficiency of the Canadian and American banking systems, 1870–1925” Financial History Review, 3(1), 49–68. Bordo, M. D., Redish, A., & Rockoff, H. (2012). Why didn’t Canada have a banking crisis in 2008 (or in 1930, or 1907, or 1893)? Economic History Review (in Press). Broz, J. L. (1997). The international origins of the Federal Reserve System. Ithaca, NY: Cornell University Press. Buiter, W. (2009). The Fed’s moral hazard maximising strategy. FT.com/maverecon, accessed March 6. Available at http://blogs.ft.com/maverecon/2009/03/the-fedsmoral-hazard-maximising-strategy/. Cagan, P. (1963). The first fifty years of the Federal Reserve. In Deane Carson (Ed.), Banking and monetary studies. Homewood, IL: Irwin. Calomiris, C. W. (1993). Regulation, industrial structure, and instability in U.S. banking: An historical perspective. In M. Klausner & L. J. White (Eds.), Structural change in banking. Homewood, IL: Business One-Irwin. Calomiris, C. W. (1995). The costs of rejecting universal banking: American finance in the German mirror. In N. Lamoreaux & D. Raff (Eds.), The coordination of activity within and between firms. Chicago: University of Chicago Press.

94

Michael D. Bordo and David C. Wheelock

Capie, F. (2002). The emergence of the Bank of England as a mature central bank. In D. Winch & P. O’Brien (Eds.), The political economy of British historical experience 1688–1914 (pp. 295–315). London: Oxford University Press. Carlson, M., Mitchener, K. J., & Richardson, G. (2011). Arresting banking panics: Federal Reserve liquidity provision and the forgotten panic of 1929. Journal of Political Economy, 119 (5), 889–924. Chandler, L. V. (1971). American monetary policy 1928–1941. New York: Harper and Row. Chandler, L. V. (1958). Benjamin Strong, central banker. Washington, DC: Brookings Institution. Eichengreen, B. (1992). Golden fetters: The gold standard and the Great Depression, 1919– 1939. New York: Oxford University Press. Friedman, M., & Schwartz, A. J. (1963). A monetary history of the United States 1867–1960. Princeton, NJ: Princeton University Press. Goodfriend, M. (2009). Central banking in the credit turmoil: An assessment of Federal Reserve practice. Prepared for the Bank of Japan 2009 International Conference, Financial System and Monetary Policy Implementation, Tokyo. Goodhart, C. (1969). The New York money market and the finance of trade: 1900–1913. Cambridge, MA: Harvard University Press. Goodhart, C. (1987). The evolution of central banks. Cambridge, MA: MIT Press. Gorton, G. (1996). Reputation formation in early bank note markets. Journal of Political Economy, 104 (2), 346–397. Gorton, G. (1985). Clearing houses and the origin of central banking in the United States. Journal of Economic History, 45, 277–283. Grossman, R. S. (2010). Unsettled account: The evolution of banking in the industrialized world since 1800. Princeton, NJ: Princeton University Press. Hackley, H. H. (1973). Lending functions of the Federal Reserve banks: A history. Washington, DC: Board of Governors of the Federal Reserve System. Holden, J. M. (1955). The history of negotiable instruments in English law. London: University of London/Athlone. James, J. (1978). Money and capital markets in postbellum America. Princeton, NJ: Princeton University Press. James, J. (1995). The rise and fall of the commercial paper market, 1900–1929. In M. D. Bordo & R. Sylla (Eds.), Anglo-American financial systems: Institutions and markets in the twentieth century. New York: Irwin. Kemmerer, E. W. (1910). Seasonal variations in the relative demand for money and capital in the United States (National Monetary Commission, Senate Document No. 588, 61th Congress, 2nd Session). Washington, DC: U.S. Government Printing Office. Knodell, J. (2003). Profit and duty in the Second Bank of the United States exchange operations. Financial History Review, 10, 5–30. Madigan, B. F. (2009). Bagehot’s dictum in practice: Formulating and implementing policies to combat financial crisis. Remarks at the Federal Reserve Bank of Kansas City’s Annual Economic Symposium, Jackson Hole, Wyoming, August 21, 2009. Meltzer, A. H. (2003). A history of the Federal Reserve (Vol. 1: 1913–1951). Chicago: University of Chicago Press.

The Promise and Performance of the Federal Reserve as Lender

95

Meltzer, A. H. (2009). Regulatory reform and the Federal Reserve. Testimony before the Subcommittee on Monetary Policy, House Committee on Financial Services, July 9, 2009. Mehrling, P. (2002). Economists and the Fed beginnings. Journal of Economic Perspectives, 16, 207–218. Miron, J. A. (1986). Financial panics, the seasonality of the nominal interest rate, and the founding of the Fed. American Economic Review, 76 (1), 125–140. Myers, M. (1931). The New York money market: Origins and development. New York: Columbia University Press. Parker, R. E. (2007). The economics of the Great Depression: A twenty-first century look back at the economics of the interwar era. Cheltenham, UK: Elgar. Poole, W. (2009). The Bernanke question. Cato Institute Commentary, July 28. Available at www.cato.org/pub-display.php?pub id=10388. Richardson, G., & Troost, W. (2009). Monetary intervention mitigated banking panics during the Great Depression: Quasi-experimental evidence from a Federal Reserve district border, 1929–1933. Journal of Political Economy, 117, 1031– 1073. Rockoff, H. (1974). The free banking era: A reexamination. Journal of Money, Credit and Banking, 6, 141–167. Rolnick, A., & Weber, W. (1983). New evidence on the free banking era. American Economic Review, 73, 1080–1091. Sayers, R. (1957). Central Banking after Bagehot. Oxford, UK: Clarendon. Schwartz, A. J. (1986). Real and pseudo-financial crises. In F. Capie & G. E. Wood (Eds.), Financial Crises and the World Banking System. London: Macmillan. Silber, W. (2007). When Washington shut down Wall Street: The great financial crisis of 1914 and the origins of America’s monetary supremacy. Princeton, NJ: Princeton University Press. Sprague, O. M. W. (1910). History of crises under the National Banking System (U.S. National Monetary Commission, Senate Document No. 538, 61th Congress, 2nd Session). Washington, DC: U.S. Government Printing Office. Sprague, O. M. W. (1915). The crisis of 1914 in the United States. American Economic Review, 5, 499–533. Taylor, J. B., & Williams, J. C. (2009). A black swan in the money market. American Economic Journal: Macroeconomics, 1 (1), 58–83. Temin, P. (1969). The Jacksonian economy. New York: Norton. Temin, P. (1989). Lessons from the Great Depression. Cambridge, MA: MIT Press. Timberlake, R. (1984). The central banking role of clearing house associations. Journal of Money, Credit and Banking, 16 (1), 1–15. Timberlake, R. (1993). Monetary policy in the United States: An intellectual and institutional history. Chicago: University of Chicago Press. Warburg, P. M. (1910a). The discount system in Europe (National Monetary Commission. U.S. Senate Document No. 402). Washington, DC: U.S. Government Printing Office. Warburg, P. M. (1910b). The United Reserve Bank Plan. In P. Warburg (Ed.), The Federal Reserve (Vol. 1, Chapter III). New York: MacMillan. West, R. C. (1977). Banking reform and the Federal Reserve, 1863–1927. Ithaca, NY: Cornell University Press.

96

Michael D. Bordo and David C. Wheelock

Wheelock, D. C. (1990). Member bank borrowing and the Fed’s contractionary monetary policy during the Great Depression. Journal of Money, Credit, and Banking, 22(4), 409– 426. Wheelock, D. C. (1991). The strategy and consistency of Federal Reserve monetary policy, 1924–1933. New York: Cambridge University Press. Wheelock, D. C. (1992). Seasonal accommodation and the financial crises of the Great Depression: Did the Fed ‘Furnish an elastic currency?’ Federal Reserve Bank of St. Louis Review, 74(6), 3–18. Wheelock, D. C. (2010). Lessons learned? Comparing the Federal Reserve’s responses to the crises of 1929–1933 and 2007–2009. Federal Reserve Bank of St. Louis Review, 92 (2), 89–107. White, E. N. (1983). The regulation and reform of the American Banking System. Princeton, NJ: Princeton University Press. Wicker, E. (1966). Federal Reserve monetary policy 1917–1933. New York: Random House. Wicker, E. (2005). The great debate on banking reform: Nelson Aldrich and the origins of the Fed. Columbus, OH: Ohio State University Press.

Notes 1 Information about the Federal Reserve’s response to the financial crisis of 2007– 8 is available on the Federal Reserve Board of Governors website: (http://www .federalreserve.gov/monetarypolicy/bst crisisresponse.htm). For a summary, see Wheelock (2010). 2 A revisionist literature has downplayed the defects of free banking. Losses to note holders were minimal, wildcatting occurred in only a few states, and information about the quality of individual state banknote issues was widely available (Rockoff, 1974; Rolnick & Weber, 1983; Gorton, 1996). 3 Other proposals to improve banking stability included reserve pooling, issuing emergency currency as done by the clearinghouses, and adopting Canadian-style nationwide bank branching (West, 1977, Chapter 2). 4 The two-name bill was a common financial instrument in the United States before the Civil War, and Biddle’s Second Bank of the United States helped make it the kind of liquid market that Warburg favored. The two-name bill declined markedly after the Civil War, however, because of the currency instability during the Greenback period, which made it difficult to make longer-term contacts. It was replaced by singlename promissory notes and the commercial paper market developed considerably by the end of the nineteenth century (James, 1995). The commercial paper market, however, was not as liquid as the two-name bill markets in Europe. 5 The French system was similar to the German system (Broz, 1997). 6 Federal Reserve bank boards of directors consist of nine directors, three of whom, including the chairman and vice chairman, are appointed by the Board of Governors, and six (three bankers and three others) are elected by the Reserve Bank’s member banks. The member banks are required to purchase stock in their local reserve bank. 7 Senator Claude A. Swanson stated optimistically that the Federal Reserve Act made “impossible another panic in this country” (quoted in Hackley, 1973, p. 10). 8 These data are from Board of Governors of the Federal Reserve System (1943, pp. 16–17).

The Promise and Performance of the Federal Reserve as Lender

97

9 An amendment to the Federal Reserve on June 21, 1917, reduced the amount of eligible paper that the Reserve Banks were required to hold from 100 percent to 60 percent of their outstanding note issues (the 40 percent gold reserve requirement against their note issues was retained). 10 Wicker (2005) argues that Warburg was not a proponent of the real bills doctrine and that the Federal Reserve Act was “real bills neutral” because it did not require that the paper eligible for rediscounting with the Fed be “self-liquidating.” Meltzer (2003, p. 70), however, contends that Warburg held the view that “the Federal Reserve could prevent wide swings in interest rates without risking inflation if it purchased real bills.” 11 An amendment to the Federal Reserve Act in 1916 permitted member banks to borrow directly from Federal Reserve Banks using eligible paper as collateral, which gave banks an alternative to rediscounting eligible paper to obtain Federal Reserve notes or reserve deposits. 12 This is an aggregate of the ratios of total reserves to deposits and Federal Reserve notes for all twelve reserve banks (Board of Governors of the Federal Reserve System, 1943, p. 346). 13 Figure 2.1 plots total Federal Reserve credit and credit extended through the discount window and by Federal Reserve purchases of bankers acceptances. The data shown are monthly averages of daily figures (Board of Governors of the Federal Reserve System, 1943, pp. 369–71). 14 Miron (1986) contends that the Fed was less accommodative of seasonal demands during the Depression, which could explain the increased incidence of financial crises. However, Wheelock (1992) finds that any changes in the seasonal patterns of interest rates and Federal Reserve credit after 1929 were not statistically significant. 15 Friedman and Schwartz’s (1963) monetary explanation of the Great Depression is widely, but not universally, accepted among economists. See Parker (2007) for a survey of alternative views on the causes of the Great Depression. 16 According to Meltzer (2003), the Fed’s monetary policy in the 1930s and even in the 1950s and later was based on the guidelines developed by Strong and his colleagues W. Randolph Burgess and Winfield Riefler to guide open-market policy in the 1920s (the so-called Riefler-Burgess Doctrine). 17 Presentation to the Federal Reserve Governors’ Conference, March 1926 (quoted by Chandler, 1958, pp. 239–40). 18 Quoted by Chandler (1971, p. 118). 19 Quoted by Chandler (1971, p. 143). 20 Minutes of the Open Market Policy Conference, September 25, 1930 (quoted by Chandler 1971, p. 137). See Chandler (1971) and Meltzer (2003) for information about the policy views expressed by different Federal Reserve officials during the Depression. 21 Before the Monetary Control Act of 1980, only Federal Reserve member banks had access to the Fed’s discount window. 22 Federal Reserve Bank of New York Governor George Harrison (quoted by Meltzer (2003, p. 350). 23 The Federal Reserve Act required each Reserve Bank to maintain gold reserves equal to 40 percent of its note issue and reserves in the form of gold or other eligible

98

24 25

26

27

28

Michael D. Bordo and David C. Wheelock securities (which did not include U.S. Treasury securities) equal to the remaining 60 percent. See Friedman and Schwartz (1963, pp. 399–406). Quoted by Meltzer (2003, p. 385). The Federal Reserve Act was amended in February 1932 to add Section 10(b), which permitted Federal Reserve Banks to lend to any member bank with capital not exceeding $5 million on the basis of any satisfactory assets, whether or not technically eligible for rediscount, in exceptional and exigent circumstances. Section 10(b) loans required the approval of at least five members of the Federal Reserve Board and bore an interest rate of not less than 1 percent above the regular discount rate. Section 10(b) was set to expire in March 1933, but was extended for one year by the Emergency Banking Act of 1933. Subsequent amendments made permanent changes to the Federal Reserve Act that permitted discount window loans on any satisfactory assets of the borrowing bank. The appendix lists major statutory changes affecting Federal Reserve discount window lending. Friedman and Schwartz (1963, p. 404n) report that only New York City banks held bankers acceptances. Paul Warburg viewed the failure of large, active acceptances markets to develop outside of New York City as the system’s principal failure (cited by Meltzer, 2003, p. 119n). The UK bill market also declined in the 1920s and 1930s. According to Holden (1955, p. 302), declining international trade and changing trade patterns reduced the demand for bills to finance international trade during the Great Depression, whereas the greater stability of the Brutish banking system, which resulted from a merger wave that increased concentration, “made it no longer necessary for bankers to guard against the consequences of a panic or run upon the banks by discounting bills.” A compilation of textual changes in the Federal Reserve Act is available from the Board of Governors of the Federal Reserve System on the following website: (http://fedweb.frb.gov/fedweb/board/legal/lawlib/textual-FRA.htm).

Comments on “The Promise and Performance of the Federal Reserve as a Lender of Last Resort: 1914–33” Ellis W. Tallman

The chapter by Michael Bordo and David Wheelock titled, “The Promise and Performance of the Federal Reserve as a Lender of Last Resort: 1914–1933” examines the failure of the Federal Reserve System to provide liquidity to the banking system effectively to stem the banking crises of the Great Depression. Work on the Great Depression has received heightened attention since the recent financial crisis, and this investigation takes on a daunting task because there is an extensive literature on the question of why the Fed failed as a liquidity provider throughout 1929–33. And the Federal Reserve System failed in a number of ways in addition to its lender-of-last-resort function. The main thesis of the chapter is that the structure of the banking system in the United States was the underlying problem and that the structure of the Federal Reserve System, as designed by the Federal Reserve Act, left the Fed incapable of serving effectively as a lender of last resort. More specifically, the chapter argues that the design of the Federal Reserve System as an institution in the Federal Reserve Act was not capable of compensating for the structural shortcomings in the United States’ banking system. Bordo and Wheelock propose this explanation as a complement to existing explanations for why the Federal Reserve System failed. There are few predictions that conflict with existing explanations, which makes the present chapter uncontroversial. However, it is not clear whether the paper contributes significantly to the understanding of the reasons that underlie the Fed’s performance during the Great Depression. In that regard, the paper revisits three conventional explanations for why the Federal Reserve System failed so dramatically as a lender of last resort to the U.S. banking system. The existing literature has, over the span of decades, centered on three main explanations for how (and why) the Fed failed as a lender of last resort: 99

100

Ellis W. Tallman

1. The flawed design of the Fed; its decentralized structure was unwieldy and awkward for coordinating a coherent and effective lender of last resort policy (Friedman & Schwartz, 1963). 2. The Federal Reserve System was excessively devoted to the gold standard. The Federal Reserve System’s adherence to the gold standard was especially harmful after the British departed from the gold standard in 1931. That devotion prevented expansionary monetary policy during the Great Depression (Eichengreen, 1992). 3. The Federal Reserve System followed a flawed policy framework that relied on a nominal interest rate and the level of aggregate bank borrowing from the discount window as indicators of the tightness (or looseness) of policy (Wicker, 1966; Wheelock, 1992; Meltzer, 2003). These explanations suggest that the Fed was unable or unwilling to respond effectively to banking panics and forestall the collapse of intermediation that took hold during the Great Depression. In such circumstances as presented by the banking panics, policies of a lender of last resort can be close to policies that are perceived as monetary policy, for example, adequate provision of monetary base for aggregate liquidity purposes. The Fed failed on both counts. In the first case, the Fed did not respond consistently to member bank demands for liquidity through inconsistent discount window policies across districts and across time. In the aggregate case, the Fed apparently lacked understanding about the policy actions that it could take to forestall or at least combat the sequence of financial crises from 1929 to 1933. The main theme of the Bordo–Wheelock paper is effectively consistent with any (and all) of these explanations, which is what makes the chapter’s thesis so challenging to assess separately. The authors offer a number of subtle insights that indicate the potential for additional knowledge arising from their thesis in an attempt to add explanatory power to existing explanations. But the burden on the authors is to produce compelling evidence to support the relevance of those insights, especially evidence to distinguish their arguments from the existing explanations. To highlight the shortcomings of the preexisting banking structure in the United States, the chapter devotes more than half its text to describing the history and evolution as well as to explaining the flaws of the banking system in the United States – the National Banking Era (1863–1914). The synopsis provides a conventionally accepted view of the central shortcomings of the banking system that existed just prior to the Federal Reserve Act. The key points are as follows:

Comments

101

1. the pyramid structure of bank reserves, 2. the lack of a nationwide market for liquidity, 3. the absence of a source of short-term increases in a final payment media. There were vast differences of opinion – not about the flaws, but about the possible solutions to them. The extensive historical description provides more than sufficient background to support the main idea in the chapter – that the Warburg central bank proposal and the goal of a nationwide discount market were motivated by the examples provided by European banking systems and their experiences. As models for the central bank and nationwide market for rediscounts in the United States, Bordo and Wheelock emphasize the reasonable hypothesis that the U.S. banking structure was never comparable to those European systems. The U.S. banking system remained “dual” with national and state-based chartering, with limitations or prohibitions on branching (both within state in some cases and across states), and unit banking by design. The chapter highlights these functional and structural differences between the “model” European systems and the actual U.S. banking system. In essence, the authors argue that the Warburg plan was inspired by banking systems that were not comparable to the U.S. banking system and that the central bank and nationwide discount market would not be effective in overcoming the structural flaws. This idea is noteworthy. The problems in the U.S. banking structure were known; however, it was widely recognized that the predominant political sentiments would prevent progress toward relaxing branching restrictions, for example, or toward eliminating dual chartering. And, although the Warburg plan influenced the design of the Federal Reserve Act, there were some important differences that arose from the political process. In several important cases, the authors would find support and agreement among contemporary critics that the Federal Reserve Act would not repair the flaws of the U.S. banking system. However, there was also recognition among the reformers that a central bank of the sort in Europe would be neither effective nor possible in the United States. J. Laurence Laughlin (1912) wrote: Although European conditions differ so much from ours that their institutions cannot be bodily transferred to our country, we have much to learn from them. Direct copying would be folly; we must take only that which aids us in a natural evolution out of our past. The central bank of Europe is not suited to our conditions; it is contrary to the spirit of our institutions; and, in any event, it would be politically impossible. (Laughlin 1912, p. 15)

102

Ellis W. Tallman

Contemporary policymakers and bankers highlighted both the flaws in the banking system and the weaknesses in the Federal Reserve Act. As noted in the chapter, Warburg (1930) was adamant about establishing a national market for rediscounting bank assets and was trying to establish such a market in bankers’ acceptances. Laughlin (1912) also complained that there was no nationwide market for liquidity, and the collaboration of Laughlin with Warburg was important for the passage of a Federal Reserve Act. THE SEPARATION OF THE MONEY MARKET FROM THE CAPITAL MARKET

The Federal Reserve Act was the legislative outcome of a reform movement catalyzed by the panic of 1907. Implicit in the reform movement was an attempt to separate the call loan money market from the payment system, which was believed would remove “speculation” from the issue of productive credit. During the National Banking Era, the large, money center banks held a large proportion of their loans in the form of call loans to brokers on the stock market, and the extension of credit in this form largely arose from the “pyramid” structure of bank reserves in the National Banking System. The collateral of the call loans was stock market issues or bond issues. Experiences during the panic of 1907 and the earlier panics of the National Banking Era demonstrated how call loans became illiquid and noncallable (despite their name) during panics. The illiquidity of the call loan market and the funding of the market by the large, money center banks became a lightning rod for banking reform. J. Laurence Laughlin and Paul Warburg had divergent views on some elements of monetary reform, but they both agreed that the call loan market should not serve as a key storage center of short-term liquidity for banks. Both economists aimed their reform proposal at separating the banking and payments system – banks and “productive” credit – from the capital markets by reducing the proportion of bank assets funneled to the call loan market. Both wanted to encourage the development of an alternative, national market for liquidity by using other assets. The funds that were used for call loans were typically from correspondent banking balances of interior banks that were held as legal reserves with New York City national banks. The reformers believed that the Federal Reserve System, with its inherent objective of holding bank reserves, would be a key factor in reducing such balances at New York City national banks. Then, as a result, it was perceived that a lower level of correspondent banker balances at New York City national banks would reduce the proportion of call loans in the asset

Comments

103

portfolios of New York national banks. The chapter stresses these functional ideas of breaking the link between bank reserves and the call loan market, but it would enhance the chapter to extend the analysis to the overarching theme of separating capital markets and the payments system. Further, the views of Laughlin and Warburg reflect some of the biases of the “real bills doctrine” that advocated credit extension by banks only to self-liquidating, short-term loans. Some of the collateral distinctions in the Federal Reserve Act associated with “real bills” incorporate these views by limiting the range of acceptable collateral for discount window credit. As discussed in the next section, the collateral distinctions reflect an aversion to having securities – even U.S. Treasury securities in the original Federal Reserve Act – as collateral for Federal Reserve discount window loans. The chapter describes the real bills elements of the Federal Reserve Act and associates the prescriptions as arising from the views of H. Parker Willis and Carter Glass. A number of the monetary reformers at that time accepted to some degree the real bills doctrine. But the resulting limitations on the eligible collateral of banks for discount with the Federal Reserve Banks was a lasting legacy with negative ramifications for the provision of liquidity during the Great Depression. Warburg (1930 and Goodhart (1969) note that call loan money market arose in the United States because of the daily clearing requirement of the New York Stock Exchange. Warburg’s goal of establishing a national bankers’ acceptance market for rediscounting was motivated to provide an alternative to the call money market for bank liquid assets. The lack of the development of the bankers’ acceptance market was well documented by Bordo and Wheelock. What was less apparent in the chapter and what offers an opportunity for analysis in further work is how the call loan market expanded substantially despite the restrictions on collateral implicit in the Federal Reserve Act and the anticipated reduction in correspondent balances among New York City banks arising from the creation of the Fed and the migration of reserves to the Fed. Further, the separation of investment banking and commercial banking in 1933 legislation may reflect a similar perspective as Warburg and Laughlin expressed toward the call loan market and banking assets. That is, the legislation was another attempt to separate fundamentally the banking and payments system from the capital markets. THE MAIN FLAWS OF THE FEDERAL RESERVE SYSTEM

The chapter highlights three main flaws of the Federal Reserve System design:

104

Ellis W. Tallman

1. the reluctance of member banks to borrow at the discount window, 2. the Federal Reserve System’s limited membership of banks, and 3. the restricted eligibility of Federal Reserve discount window collateral. Each of these flaws could have been overcome with small adjustments to the provisions of the Federal Reserve System, as subsequently discussed. Adjustments to the provisions had precedents that were successful during World War I and immediately following, and the chapter highlights the perceived effectiveness of these adjustments on the performance of the Fed during those periods. The most notable example, discussed in the chapter and in the next section as well, was the ability for member banks to borrow for nonmember correspondent banks. If the Fed had implemented similar actions during the Great Depression, its performance as a lender of last resort may have been more successful. It is a discussion of what the chapter views as key structural flaws in the design of the Federal Reserve System and the analysis of the subsequent Federal Reserve policy failures on which the chapter must expand in order to provide evidence for its thesis. RELUCTANCE TO BORROW

The chapter suggests that reluctance to borrow from the discount window was a structural flaw of the Federal Reserve System. Yet there is plenty of evidence to the contrary. Member banks had borrowed from the discount window effectively throughout the years 1914–27. In fact, immediately following World War I, the discount window lending was the largest component of the monetary expansion (see Figure 1). From this perspective, the reluctance to borrow was not a defect in the design of the Fed. Rather, the apparent reluctance on the part of member banks to borrow from the Federal Reserve System may have been the unforeseen result of a misguided economic policy. Member bank reluctance to borrow became a problem toward the late 1920s, perhaps as a result of eighteen months of “direct pressure” by the Federal Reserve to constrain banks from lending money to the stock market through the call loan market.1 The policy of direct pressure aimed at reducing bank discount window borrowing; it was perceived by some Federal Reserve policymakers that the discount window loans were effectively the source of funding for loans on call to buy stock. At the time, discount window borrowing to fund (even indirectly) call loans was viewed as speculative borrowing, which was specifically prohibited in the Federal Reserve Act. The negative externality of “direct pressure” may have led banks to adopt a reluctance to borrow from the Federal Reserve System,

Comments

105

4 3.5 Federal Reserve Credit

Billions of US $

3 Discount Window Borrowing

2.5 Government Securities Held

2 Bankers' Acceptances

1.5 1 0.5 0 Jan-1917

Jan-1921

Jan-1925

Jan-1929

Jan-1933

Figure 1. Federal Reserve Credit, 1917–35.

but the reluctance to borrow was not a structural flaw of the design of the Fed. LIMITED MEMBERSHIP OF THE FEDERAL RESERVE SYSTEM

The limited membership of the Fed was recognized as a defect from the time of its establishment, or even before. Key banking reform leaders, like Frank Vanderlip, National City Bank president, wanted to compel all intermediaries – national banks, state banks, and trusts – to become members of a reserve association system. Universal membership was a characteristic Vanderlip wanted included in the Aldrich Plan, which was the proposal that arose from the Jekyll Island meeting, whose centennial anniversary marks the date of this conference. The limited membership issue therefore is a structural flaw in the Federal Reserve Act. However, the chapter provides evidence that is effectively contradictory to the thesis – at times (1920–3, especially) the Board of Governors made adjustments to the provisions to the Federal Reserve Act that enabled the Fed to act more effectively as a lender of last resort. The best example of this idea is the provision of Federal Reserve discount window loans to nonmember banks. The Federal Reserve Act prohibited member banks from borrowing for other, nonmember banks through the discount window function. In 1921, the Board of Governors authorized the reserve banks to discount for member banks any eligible paper acquired from

106

Ellis W. Tallman

nonmember banks, which would have been extremely helpful for the nonmember banks during 1929–33. Unfortunately, that authority was rescinded in 1923. As a result, during the Great Depression, nonmember banks were unable to access Federal Reserve liquidity, except in a few isolated emergency instances. There is no disagreement with the facts of the chapter – the Federal Reserve System failed to serve as lender of last resort, and adherence to the unadjusted Federal Reserve Act provisions for discount window lending left the nonmember banks isolated from Fed discount window loans. The irony is that the motivation to enforce universal membership in a reserve association was precisely to prevent such an outcome. The isolation of the trusts from the New York Clearing House, the source of liquidity provision during the panic of 1907 put into motion the successful movement toward a central bank in the United States.2 The previous adjustments to the act and their perceived success at expanding Federal Reserve effectiveness within the period 1914–23 weakens somewhat the thesis that the structure of the banking system and the design of the Fed was the source of the failure of the Fed to act as lender of last resort during the Great Depression. Whether the Fed failed to initiate those provision adjustments because it was unable to form a consensus (decentralized Fed explanation) or because it was incapable of understanding the benefits of doing so (the knowledge gap or the flawed policy model explanation) is not clear. The point that seems clear is that there were actions that the Fed could have taken to address some of its design flaws. The adjustments to the original act that were effective during World War I and afterward may have alleviated some of its bad performance as a lender of last resort. The chapter isolates the provisions that were adjusted during the period prior to the Great Depression; the discussion could then address directly those activities taken by the Fed during those periods that allowed it to perform its role effectively. Further research could then address characteristics that the Fed did not possess during the Great Depression and how the lack of the adjustment of the act’s provisions destined the Fed to failure. RESTRICTED ELIGIBILITY OF DISCOUNT WINDOW COLLATERAL

The Federal Reserve Act imposed significant restrictions on the composition of discount window collateral, and it is certain that such restrictions hindered severely the ability of the Federal Reserve System to act as a lender of last resort. The chapter describes effectively how many of the same restrictions were elements of the original Aldrich Plan, so the flaws in the Federal

Comments

107

Reserve Act were associated with the banking reformers. It is possible that the reformers and the authors of the Federal Reserve Act lacked a full understanding of the complexity of providing liquidity to the financial market. Part of the limitation, however, was also a result of the excessive influence (or misapplication) of the real bills doctrine. The chapter hints at this explanation, but Meltzer (2003) takes this idea further and in greater detail. The real bills doctrine had a profound effect on the composition of acceptable collateral for discount window credit extensions. Federal Reserve discount window credit could be extended only on the receipt of “good collateral,” the composition of which could not be for the purposes of “covering merely investments or issued or drawn for the purposes of carrying or trading in stocks, bonds or other investment securities.” As a result, the banks had few alternatives to bankers’ acceptances, commercial paper, or other forms of short-term, private credit to use for discount window collateral. Collateral requirements were also applied differently across reserve bank districts, which also supports the criticism of the Federal Reserve’s decentralized structure as a key flaw. However, there are contrasts between the Federal Reserve Act and the Aldrich Plan that are mentioned in the chapter and are worth commenting on again because there were unique features of Aldrich Plan that may have been useful if they were included in the Federal Reserve Act. The Aldrich Plan had, for example, a provision to allow the rediscounting of any direct obligation of the borrowing bank if that action was approved by the secretary of the treasury and the local bank association of the borrowing bank. This characteristic would have been helpful in a panic, and it was not part of the Federal Reserve Act. The chapter explains accurately that the market for bankers’ acceptances, the market that Paul Warburg hoped would supplant the call loan market as a key element of banker “liquid” investments, never grew to be large enough to provide a sufficient source of bank rediscounting. The Federal Reserve System purchased substantial amounts of bankers’ acceptances, but the volume of credit was too small to fulfill the role that Warburg envisioned. That market, had it developed, may have provided a mechanism for the Fed to provide liquidity to the market through various counterparties that did not involve discount window lending to individual banks. This point, emphasized in the chapter, is one of the subtle insights that could be emphasized in subsequent research to support this chapter’s thesis. Then again, the banker acceptance financed international trade, which was becoming progressively more risky as the Depression deepened.3

108

Ellis W. Tallman

The model of providing credit through the discount window function likely arose from the private clearinghouse function of issuing “clearinghouse loan certificates” during financial crises of the National Banking Era (1863–1913). Liquidity provision is “bank specific” and requires banks to provide the issuing party (the clearinghouse or the Fed) with information on its balance sheet as well as adequate collateral for the loan. However, the Federal Reserve System had, by 1923, become cognizant of its ability to engage in open market operations to adjust the level of reserves and currency outstanding. And in the Annual Report of 1923, the description of open market policy effectively describes how open market actions would affect the demand for credit by member banks. That said, the Federal Reserve Act, left unadjusted, faced constraints arising from the strict collateral requirements for discount window lending that made the Federal Reserve System ineffective as a lender of last resort. In addition, the Federal Reserve Act also imposed restrictions that limited the Fed’s effectiveness as a provider of liquidity to the market. COLLATERAL QUALITY REQUIREMENTS ON CURRENCY BACKING IN THE FEDERAL RESERVE ACT

The Federal Reserve Act allowed the Fed to hold government securities on its balance sheet. However, there was a restriction on the asset composition of collateral that backed currency; the currency collateral required a 40 percent gold reserve and the other 60 percent of collateral, which had to be approved securities, could not be composed of government securities. Meltzer (2003, pp. 356–7) describes how the gold reserve constrained Fed open market operations from 1931 through 1932, and Bordo and Wheelock emphasize that the passage of the Glass-Steagall Act of 1932 allowed government securities to be used as collateral for Federal Reserve notes and thereby removed this barrier from Fed open market purchases of government securities. Yet Wicker (1966) and Meltzer (2003) argue that the Fed had at its disposal other methods to relax the gold reserve constraint and chose not to use them. Further, the Fed chose not to engage in expansionary policies by using open market operations during 1929 and 1930 when the gold reserve was not an effective constraint on its policies. Here again, the poor performance of the Fed as a liquidity provider seems more consistent with any of the three existing explanations than the main argument of the chapter. In essence, the restriction on asset composition of the backing of currency was perhaps a design flaw, but one that again could have been overcome with some adjustment to the provisions of the Federal Reserve

Comments

109

Act. Then, the poor performance was a result of ineffective policy, arising because a decentralized Fed could not coordinate a policy, because the Fed was too devoted to the gold standard, or because there was a knowledge lacunae. AN ASSESSMENT OF THE BORDO–WHEELOCK ARGUMENTS AND THE EXISTING THREE EXPLANATIONS

The three explanations appear interconnected and interdependent, and, when combined, they provide a coherent explanation for why the Federal Reserve System failed 1929–33. The Fed as a decentralized institution failed to act decisively to intervene in financial markets; even its successes, like the provision of liquidity following the 1929 stock market crash, was a result of a New York Fed–initiated policy, which then led to friction between the New York Fed and the Board of Governors. Coordinated and effective policy was unlikely because of the decentralized structure, but also for other reasons. The Fed held to the gold standard too long and too rigidly, and the Fed also implemented a flawed policy framework. Both these elements contributed to the ineffective, decentralized Fed because the development of a policy relied on a shared view of how things worked. That shared view did not exist. Meltzer (2003) suggests that the Fed applied a flawed policy framework and thereby misdiagnosed the problems in the economy in 1929–33. Similarly, Wicker (1966) emphasized that there was an understanding lacunae among board members, in which the participants were uncertain about how policies had worked successfully in the past and how successful past experiences in policymaking might have relevance for 1929–33. Further, Wicker suggests that the Fed lacked a clear model for when to increase the monetary base permanently and by what methods to do it. The provision of adequate aggregate liquidity, though not directly considered the lenderof-last-resort facility, relates to the lender-of-last-resort role because such liquidity provision may alleviate some of the need for lender-of-last-resort lending. Unfortunately, the Fed limited its holding of government securities in its portfolio for various reasons, which may reflect the influence of the real bills doctrine, an excessive adherence to the gold standard, or both. The gold standard and its potential constraint on Fed policies arise in various elements of the other explanations. As one of the main explanations for the failure of the Fed as a lender of last resort, a decentralized Fed may have failed to resolve the disagreements among board members and governors about how the gold reserve constraint may bind, or not, during the

110

Ellis W. Tallman

instances – like 1930 – when monetary expansion was the best response to monetary conditions. Analogously, the understanding gap as discussed by Wicker may involve a misunderstanding of the effects of gold on the Fed policy decisions. The Bordo–Wheelock chapter provides an opportunity to assess the explanatory power of these three existing explanations for why the Federal Reserve System failed to provide sufficient liquidity to the banking system and failed to act effectively as a lender of last resort. I came away with a sincere appreciation for how comprehensively these characterizations explain how the Fed failed. Overall, the reader of the chapter should come away with the impression that these existing explanations of the failure of the Federal Reserve System to act as a lender of last resort are three layers of a cake that is more satisfying if taken together. The contribution of the Bordo–Wheelock chapter aims to be an icing. The cake is more satisfying with the icing, but the cake is the main dessert. SUMMARY

The Bordo–Wheelock chapter offers an additional explanation for why the Federal Reserve System failed as a lender of last resort from 1929 to 1933. The chapter demonstrates that several of the weaknesses in the Federal Reserve Act were also weaknesses in the Aldrich Plan and weaknesses that may have arisen from what may have been a mismatch between the institution as designed and the flawed banking system that it would have to serve. The evidence and arguments provided in the chapter provide a suggestive case for the characterization of the Federal Reserve Act as an institution that was designed ineffectively to support a banking structure that was deeply flawed. There are opportunities to make these arguments more compelling with additional evidence and a more complete analysis. Yet the essential explanatory power of the three leading hypotheses regarding the failure of the Federal Reserve System remain central to understanding why the Fed as an institution did not perform as the bank reformers had hoped. The existing arguments appear robust to scrutiny and complementary to all others. WHAT DID BEN BERNANKE LEARN FROM 1929–33 FEDERAL RESERVE POLICY FAILURES?

The recent worldwide financial crisis of 2007–9 has brought the performance of policymaking institutions to the forefront of economic and political

Comments

111

debates. During the crisis, the central metric in those debates was whether the actions taken by those institutions would avoid “another Great Depression.” With two years of perspective, the Federal Reserve System, the U.S. Treasury Department, and the Federal Deposit Insurance Corporation engaged in policies successful in that regard. The subsequent economic downturn has been called the “Great Recession.” Yet, despite the complaints of critical commentators, it could have been worse, considering the reference point. The Federal Reserve System in 2008 established liquidity provision facilities like the Primary Dealer Credit Facility and the Securities Lending Credit Facility in order to establish some way of providing credit to intermediaries that were not members of the Federal Reserve System. Further, the Federal Reserve may have relaxed collateral requirements in order to provide liquidity to institutions that were considered essential to the financial market. And the Term Auction Credit Facility, initiated in late 2007, was a direct attack on what was perceived as an aversion to borrow from the discount window on the part of member banks. Thus the recent experience suggests that the Fed as an institution, led by a former academic with expertise in the economic history of the Great Depression, has apparently learned from its history of failure. In this recent episode, the institution overcame the three key flaws in the Federal Reserve performance during 1929–33 that were emphasized in the Bordo and Wheelock chapter. A repeat of the Great Depression was avoided, and for achieving that outcome the Federal Reserve should be commended. References Eichengreen, B. (1992). Golden fetters: The gold standard and the Great Depression, 1919– 1939. New York: Oxford University Press. Friedman, M., & Schwartz, A. J. (1963). A monetary history of the United States, 1867– 1960. Princeton, NJ: Princeton University Press. Goodhart, C. (1969). The New York money market and the finance of trade: 1900–1913. Cambridge, MA: Harvard University Press. Laughlin, J. L. (1912). Banking reform. Chicago: The National Citizens’ League for the Promotion of a Sound Banking System. Laughlin. J. L. (1933). The Federal Reserve Act: Its origin and problems. New York: Macmillan. Meltzer, A. H. (2003). A history of the Federal Reserve (Vol. 1: 1913–1951). Chicago: University of Chicago Press. Moen, J. R., & Tallman, E. W. (1999). Why didn’t the United States establish a central bank until after the panic of 1907? (Working Paper 1999–16). Atlanta, GA: Federal Reserve Bank of Atlanta.

112

Ellis W. Tallman

Warburg, P. M. (1930). The Federal Reserve System: Its origin and growth (Vols. I and II). New York: Macmillan. Wheelock, D. (1991). The strategy and consistency of Federal Reserve policy, 1924–1933. New York: Cambridge University Press. Wheelock, D. (1992). Monetary Policy in the Great Depression: What the Fed did and Why. Federal Reserve Bank of St. Louis Review, March/April 1992, 74(2), 3–2. Wicker, E. R. (1966). Federal Reserve monetary policy, 1917–1933. New York: Random House.

Notes 1 See Wicker (1966, pp. 121–3). 2 See Moen and Tallman (1999). 3 Barry Eichengreen made this point at the conference.

THREE

Where It All Began: Lending of Last Resort at the Bank of England Monitoring During the Overend-Gurney Panic of 1866 Marc Flandreau and Stefano Ugolini*

During the consultations organized after 1908 by the U.S. National Monetary Commission with a view to creating the Federal Reserve System, banker Paul Warburg delivered an account of the functioning of the money market in Europe (Jacobs, 1910; Warburg, 1910). He drew a comparison with the American system. In America, he explained, the money market was based on the stock exchange. As there was no central bank standing ready to rediscount short-term commercial credit instruments (“acceptances”), the liquid portion of the money market was made up of repos to the stock exchange. Those with available short-term cash lent it to stock market dealers in exchange for securities and got their cash back or renewed their positions periodically. According to Warburg, this arrangement lacked resilience. It made the U.S. financial system vulnerable to balance-of-payment shocks. If payments abroad increased, for example, in the event the trade balance deteriorated and foreign creditors demanded settlement, cash was withdrawn. Also, because cash was with the stock exchange, it was withdrawn from there and sent speculators wrong-footed. This forced fire sales. The ensuing decline in the price of securities (i.e., the deterioration of collaterals) prompted brokers to increase their margin requirements. Lenders distributed money sparingly. The balance-of-payment shock morphed into a stock exchange crisis and then into a credit crisis. Reluctance to lend and declines in values completed the circle and led to commercial bankruptcies. By contrast, in Europe – Warburg reasoned – the existence of a large volume of bills that could be rediscounted at the central bank provided more leeway and facilitated financial stabilization. Most contemporary observers (and this included U.S. economists and policymakers involved in the debates surrounding the U.S. National Monetary Commission) were struck by the 113

114

Marc Flandreau and Stefano Ugolini

fact that the tradable bill or acceptance was the staple instrument in European money markets. Their generous availability, their liquidity, and the fact that central banks stood willing to rediscount them in crises were seen as a source of financial resilience. When liquidity requirements grew, banks could turn to the central bank and rediscount acceptances. The liquidity thus obtained enabled banks to keep supporting their customers. The central bank thus acted as a lender of last resort. The result was that Europeans could deal with crises more effectively than Americans. The mix of acceptances and central bank support was seen as Europe’s secret recipe for financial stability, and the only thing the United States needed to do was to remove the regulatory constraints that impeded the emergence of such a market. This would bring one century of financial stability. History may have eventually decided otherwise, but later scholars have generally concurred that there was widespread belief that creating a market for acceptances “`a la Europe” would provide the public good of financial stability. “Europe” was code for England. Whereas the U.S. National Monetary Commission concerned itself with studying other central banks, U.S. bankers and policymakers cast their sights on the Bank of England and the London market for acceptances, the center of world liquidity. This was natural, because this was where the best practice was defined. Franc and mark bills enjoyed some international circulation, but they were junior to sterling.1 Moreover, beyond the goal of finding a remedy to crises, one concern of the National Monetary Commission was to devise ways to shortcircuit London and save on the “tribute” that was paid annually to UK bankers in the form of acceptance commissions. Warburg and his supporters intended to defeat Europe on its own turf, and this started in London (Broz, 1997; Eichengreen & Flandreau, 2012). At that time, the central banking wisdom that prevailed in central banking was what Frank Fetter would call the “British monetary orthodoxy” (Fetter, 1965). A prominent feature of this orthodoxy was its identification (which contemporaries associated with the Bank Charter Act of 1844) of the central bank’s key lending rate (the discount rate) as the legitimate policy tool to protect the gold reserve and peg the external value of the currency, although some observers grew uneasy with the interest rate volatility this induced and made suggestions for improvements (Palgrave, 1903).2 But this said little as to how one should construct a market.

Where It All Began: Lending of Last Resort at the Bank of England

115

The U.S. National Monetary Commission produced four reports on the English banking system.3 There were two books. One was an alreadypublished book by an Austrian scholar, Eugen von Phillippovich, now translated from German (Phillippovich, 1910). It was devoted to the historical evolution of the relations between the Bank of England and the state. The other was a joint volume, with contributions of varied lengths, by a number of city experts (Withers, 1910). And there were two pamphlets. The first was by Jacobs (1910), and the other was the already mentioned contribution by Warburg (1910). The reports by Jacobs and Warburg were superlative on the beauties of the European system, but they were concise. The contribution by the city writers also lacked detail. Withers dealt with “the merchant bankers and accepting houses” in fewer than five pages, although there was laid manifestly the secret of making fire.4 A characteristic of most reports submitted to the U.S. National Monetary Commission is that they generally were abstracted from more tedious microstructure aspects.5 This omission is intriguing. It may have reflected an English antipathy for detail. But a lot of relevant information was concealed that way. We fail to understand why American counterparts were content with material that was so general it could hardly serve as the basis of a blueprint for monetary design. This conflicts with the National Monetary Commission’s mission to inspire the creation of a market and new instruments – a mission that would succeed or fail on microeconomic cleverness, not on abstract principles. This makes the historical experience of the Bank of England as it was known or ought to have been known at the time when the Federal Reserve was created an important subject. This should shed light on how the European precedent shaped the U.S. policy choices of the 1910s and improve our knowledge of issues that are still relevant today. Indeed, as we shall see, there is a fascinating precedent in the way the Bank of England found itself involved in helping a shadow banking system of nonbank, limited liability, money market institutions known as bill brokers (or perhaps, more adequately, “discount houses,” although the two words were used interchangeably) despite its initial insistence on not supporting it because it saw it as a source of speculation and financial vulnerability. When markets learned of the failure of Overend-Gurney, which the bank had refused to bailout, liquidity seized and a violent panic set it. The Bank of England was forced to concede support to the shadow banking system. The analogy with the Fed’s refusal to help Lehman in September 2008 and the events that followed is not only tempting: It is legitimate.

116

Marc Flandreau and Stefano Ugolini

In effect, the crisis of 1866 ended up being a turning point. As we argued in earlier joint research, the adoption of a “modern” policy of “lending of last resort” (materialized by generous lending against good collateral) consolidated precisely at this time.6 The result was the adoption of “Bagehotian” principles for lending of last resort. These were expounded in The Economist during the 1840s (before Bagehot’s time) and then, with increasing assertiveness, during the 1860s, by Bagehot himself. These ideas came to be organized in Bagehot’s book, Lombard Street, published in 1873. The book advocated generous liquidity support to the money market in periods of crisis. This begs for a greater research effort to provide for a better understanding of the how and why of this revolution. However, it is fair to say that little recent work has been done to understand in finer detail the microeconomics of the Bank of England’s lending of last resort. The way it selected bills, the way it protected itself against moral hazard, the way it monitored the market, and so on, are not really known – both qualitatively and quantitatively.7 We are not aware of any recent study providing an empirical exploration of the relations between the Bank of England and the London money market.8 As a result, older accounts still rule. They are of superior quality, and this per se has acted as an entry barrier. Classics include important works by historians who discussed in detail the operation of the money market and the Bank of England’s relation to it (King, 1936; Sayers, 1936). Another important work is Sayers’s (1968) account of Gilletts, a midsized bill broker. Later research by Sayers (1976), Goodhart (1988), or Capie (2002) completes the picture. An important feature of the picture that emerges from this literature is the notion that, paramount in the transformation of the Bank of England into a modern central bank, was the development of “anonymous” dealing with the market. The Bank of England is usually portrayed as having managed to ignore the identity of borrowers as long as “good collateral” would be brought in for discount. Capie (2002) reflects this view in a powerful metaphor in which he depicts the bank’s discount window as “made of frosted glass and raised just a few inches” so as to enable customers to discount anonymously. In Forrest Capie’s account, the central banker “does not know, nor does he care, who is on the other side of the window. He simply discounts good quality paper or lends on the basis of good collateral.”9 These works are still outstanding and inspiring, and they constitute our starting point. However, some recent progresses in the availability of sources open new possibilities. We exploit here two types of ledgers that provide

Where It All Began: Lending of Last Resort at the Bank of England

117

critical information on Bank of England crisis lending. First, we use the ledgers for daily discounts, which record, as they occur, the succession of liquidity provision operations the bank performed with counterparties. Second, we use the customers’ ledgers, which were the instruments through which the bank monitored its exposure to individual risks. As far as we know, neither the daily discounts ledgers nor customers’ ledgers have been exploited systematically so far. The reason is that they involve accounts of private customers of the bank, for which a full embargo used to apply, now shortened to a moving wall of one hundred years.10 Thus, although known to some previous scholars, these sources could not be used as openly as we do here.11 The tremendous value of such material is obvious. First, its very existence and the way it was organized suggests that there was more to the bank– discounter relation than the frosted window metaphor would suggest. The bank kept detailed records and monitored the position of discounters (who presented the paper) and of acceptors (who had guaranteed it). Second, this source enables us to get an extremely detailed view on how the market operated. In principle, and subject to some limitations owing to partly missing sources for some dates and periods, a more or less complete characterization of the interactions between the Bank of England and the British money market throughout the succession of crises that occurred in England before the Federal Reserve Act – namely 1847, 1857, 1866, 1878, 1890, and 1907 – is feasible. This chapter, however, falls short of fulfilling such a vast scheme and sets itself a more limited intermediary target. First, the amount of work needed to master the enormous volume of information in the bank’s ledgers precludes a systematic study of all crises in just one chapter. This explains this chapter’s focus on the Overend-Gurney panic of 1866, and if consolation is needed, it has been argued before, and we concur, that this crisis was a turning point. Second, working with the bank ledgers cannot control for the self-selection involved in presenting given financial instruments to the discount window. The view we give of the money market must by construction be partial, and the only defense is that the central bank’s perspective to the matter remains crucial and that future research ought to provide further scrutiny of our main findings. The methodology in this chapter is as follows. We provide a statistical exploration of the financial instruments the Bank of England purchased during May 1866 (the month when the so-called Overend-Gurney crisis of 1866 peaked) and compare it with a “normal” month exactly one year

118

Marc Flandreau and Stefano Ugolini

earlier (May 1865). By combining these two pictures (“normal” and “crisis time”), we seek to understand better the changeover that occurred both in the type of instrument and in the type of customer. There are four key findings. First, we unearth a rich system of monitoring by the Bank of England, which suggests that its approach to the money market was probably less anonymous than implied by earlier accounts. Second, we discover the considerable importance of the nonbank counterparties for the Bank of England’s operation during crisis periods. This can be put in relation to the importance of the present shadow banking system and the way the central bank can end up being a hostage of financial innovation.12 Third, we discover that more than two-thirds of the bills discounted at the bank’s window had originated abroad, that is, had been issued by foreign correspondents of British banks. It is a striking feature that the staple instrument for the conduct of British monetary policy and crisis lending was related to foreign trade. We also find that the surge of foreign acceptances as the staple instrument of monetary policy was a development of the period 1850–70. Last, we emphasize the importance of central bank’s supervision and prudential role in fostering the liquidity of certain instruments. We argue that “trade acceptances” were convenient instruments to supervise, and this is why they ended up as the chief support for liquidity provision. This suggests that the root of the special status of sterling was not solely Britain’s predominance international trade but Bank of England’s advance in supervision. The remainder of the chapter is organized as follows. Section I reviews our new source in relation to the operation of the money market. Section II explores the rise of the shadow banking system in England until the crisis of 1866. Section III looks at who came to secure cash in 1865 and 1866. Section IV explores what was brought in. Section V deals with the question of central bank supervision. We end with conclusions.

I. The London Money Market and Bank of England’s Ledgers Conventional descriptions of the set of instruments comprised under the heading “British money market” traditionally emphasize the role of acceptances. Acceptances were bills that one merchant or banker (the drawer) had drawn on another merchant or banker (the drawee) and that the drawee had “accepted” by putting his signature on the bill. Prestigious drawees were leaders in the acceptance business and sold their signature for a fee. Previous literature emphasized the role of “merchant banks”

Where It All Began: Lending of Last Resort at the Bank of England

119

but also of some British foreign and colonial banks as key providers of acceptances, and mentioned the presence of some private and joint-stock banks, although their importance is said to have started only much later (Jenks, 1927; Chapman, 1984). From that point, conventional accounts suggest, flowed a kind of “circuit” whereby the acceptances (initially supplied by correspondents of leading British merchant banks and then certified by those very merchant banks) were purchased through the agency of bill brokers by large commercial banks, for reserve purposes. As time passed, bill brokers started to get invested in these acceptances for their own account, using resources they collected through “call loans” (essentially, time deposits) from the commercial banks. The result was that they gradually evolved into money market funds. In case of crisis, commercial banks would secure liquidity by going to the Bank of England, rediscounting the acceptances they held and getting cash in return. In addition, they would call back their deposits with the bill brokers who, in order to meet the cash requirements of the commercial banks, would in turn have to unwind their own balance sheets by discounting bills with the Bank of England.13 Although this description will subsequently receive qualification, it has a heuristic value as a starting point. It helps to understand that the Bank of England’s discount window was a bit like the pond in the savannah – the place where the wild beasts of the money market came to water. Data pertaining to what was happening at the discount window have rich informational value. In this chapter we exploit information on the London ledgers of the Bank of England in order to provide a detailed picture of the bank’s lending-of-last-resort activities. There were two different ways in which the Bank of England provided cash to customers. The first one, called “discounts,” consisted of an outright purchase of acceptances. The second one, known as “advances,” amounted to a modern repo operation: The Bank took in bills or bundles of bills (“parcels”), but the counterparty was understood to repurchase the security from the bank at a given date. Advances were secured by the security given in repo, to which a haircut was added. As decades passed, the number of securities eligible for advances increased,14 but for the period under study, advances could be made on acceptances, on domestic sovereign stock such as consols (for “consolidated,” as British government bonds were known), or on Indian government bonds.15 Bank of England discounts, on the other hand, were exclusively based on acceptances and were secured by the signature of both the discounter and the acceptor.

120

Marc Flandreau and Stefano Ugolini

Not just anybody could come to the Bank of England’s window. The bank had a list of eligible discounters. In London (on which we focus here), discounters could be any kind of firm involved in “trading” (i.e., commerce or industry), merchant banks, commercial banks, and bill brokers. To become eligible, one had to be recommended by some authority, and the so-called “rating books” bear mention of the authority that had provided recommendation (often a senior merchant bank or a bank director).16 There were 438 discounters in 1865, 503 in 1866.17 The bank kept a record of discounters approaching it for cash procurement regardless of whether it agreed to the loan (in the overwhelming majority of cases it did), and then, when it agreed to the operation, several entries were created according to a very meticulous system, bearing witness of a careful monitoring of risks and exposure by the Bank of England. First, there was a Bank of England’s window’s journal. Day after day the so-called “daily discounts” ledger entered individual discounters’ applications as they came. Table 3.1, reproducing the entry for May 3, 1866, shows (see the figure accompanying Table 3.1 for a picture of original) the information recorded in this ledger. It included the number of bills brought in for discount by individual discounters, the rate, the name of the counterparty benefiting from the discount or advance, the amounts discounted or advanced, the number and amount of bills rejected if relevant, and a “remarks” column that could be used to give reasons for rejecting applications. In the example displayed, reasons for rejecting a bill included “sighting altered” (suggesting a poor-looking bill, perhaps a forgery) and “beyond 95 days” (usually the bank restricted its discounting to bills with less than three months to run).18 Consistent with the evidence in the “rating books,” the bank also controlled its exposure on a per-customer basis. Every operation (discount or advance) was reported in individual accounts in ledgers that were organized by institutional types – as we subsequently explain. Because the discounting of bills was secured by both the acceptor’s and the discounter’s signatures, two entries were created each time a bill was taken in: one in the discounter’s account and the other in the acceptor’s. Ledgers were manifestly used to monitor “at a glance” the position of customers: they did show, for each entity, the outstanding amount of credit guaranteed either as a discounter or as an acceptor. Table 3.2 shows the entry for one random merchant bank and date (Smith Fleming & Co., May 11–14, 1866). As can be seen, for each liquidity provision event the ledger documents several characteristics: the place where the drawer of the bill (if the event involved a bill) was located (first

121

7 7 7 7.5 7 7 7 7 7 7 7.5 7 7 7 7 7 7 7 7 7 7 7 7

Hooper R & Sons Jenkinson W Hart J & Co Lloyd & Attree Nelson T & Son Lovering & Minton Rattray W & Co Sieveking Droops & Co Morgan Bros Bailey Pegg & Co Stephenson Clarke Rutherford Drury & Co Tamvass Mierulachi & Co Naylor Benzon & Co Hooper R & Sons Smith Fleming & Co The Colonial Co Ltd Bell A & Co Gaury J & Co Hart J & Co Gibbs A & Sons Byass & Son Pawson JF & Co

Source: Bank of England Archive, Daily Discounts, C28/26. Note: Figures given in italic in the last row are totals.

202

10 2 6 13 5 1 6 11 19 1 6 5 6 3 17 23 40 15 3 2 8

For whom Discounted, or To whom Advanced

151,787

3,258 221 6,865 300 562 500 1,350 5,643 5,698 1,000 2,537 864 4,329 2,973 1,623 44,007 31,962 6,972 4,832 951 25,332

£

2 7 2 6 10 1 9 11 5 9

13 6 12 7 18 12 7 6 13 2 3 12 4

5

4

19

11

3 10

11 5

d

12 12

s

13

2 1

8

1

1

N◦ of Bills Rejected

6,078

745 131

3,501

500

1,200

£

6

7 8

10

s

4

9 3

4

d

Amount Rejected

Rate Per Cent

Amount of Bills Brought in for Discount

N◦ of Bills Brought in for Discount

Amount Advanced

£ 3,000.00

Amount of Advances going off

Amount Discounted

£ 92,000.00

Amount of Bills Discounted going off

Thursday, May 3, 1866

8,000 9,600 17,600

£

Amount Advanced

Table 3.1. An excerpt from the “Daily Discounts” ledgers (May 3, 1866). (Also see accompanying figure).

Beyond 95 Days

Beyond 95 Days

Sighting Altered

Remarks

£ 17,600.00

£ 145,708.00

122

Marc Flandreau and Stefano Ugolini

An excerpt from the “Daily Discounts” ledgers, May 3, 1866. Source: Bank of England Archive, Daily Discounts, 1866, C28/26.

column), his name (second column), the date of the liquidity provision event (third column), a reference to the other page in which the same operation was recorded (fourth column), the name of the “other signature involved” (i.e., the name of the acceptor if Smith Fleming & Co. was the discounter, or the name of the discounter if Smith Fleming was the acceptor) (fifth column), the maturity (sixth column), and the amount of credit granted by each operation (registered in the relevant column in the last group).19 As can be seen, the bank could constantly monitor customers. If need be, it could at any time measure its exposure to any single entity. As said, the bank did recognize differences among customers, and this motivated the use of different ledgers. The “discounters” ledger (as the bank

123

Finlay & Co

Nicol W & Co Nicol W & Co Stewart G & Co

Bombay Bombay Colombo

Angers R

Bombay

Karachi

Barthold E & Co Oriental Bank

Montevideo Bombay

Robinson & Co Robinson & Co Robinson & Co North Western Bank Bank of Hindustan Jardine M & Co Oriental Bank

Nicol W & Co

Bombay

Bombay Bombay Bombay Liverpool Calcutta Shanghai Bombay

Drawer

Whence Drawn

1866 May 11 May 11 May 11

Date of Discount

1487 – –

–

– – – – – – –

–

– 214 1674 –

Folio Brought Forward National Discount Co LOAN Melly Fargo & Co Union Bank of London London Joint Stock Bank North Western Bank North Western Bank North Western Bank Barclay & Co National Bank Matheson & Co Union Bank of London Union Bank of London Nicol D & Co Nicol D & Co Arbuthnot L & Co

Acceptors or Discounters

Smith, Fleming & Co.

Aug 11 Aug 11 Aug 16

Aug 10

June 29 June 29 June 29 July 9 July 11 Aug 1 Aug 10

June 9

July 9 Aug 9 May 24 June 9

Due

5,000 5,000 1,000 131,606

1,000

2,000 2,000 2,000 1,000 2,500 2,500 2,500

920

9

14

15

s

With

859 5,000

98,326

£

7

6

1

d

18,469

15

15

s

Upon

13,469 5,000

£

Discounts

Table 3.2. Excerpt from Smith, Fleming & Co.’s entry in Bank of England’s “Discounters” ledgers

10

10

d

F. 1632

8,000

8,000

£

Upon

(continued)

32,000 196,000

164,000

£

With

Advances

124

Bell JG & Co

Manchester

May 14

May 14

May 12

Date of Discount – – – – – 211 220 220 220 222 224 232 237 – 232 229 –

Folio

Source: Bank of England Archive, Discounters’ Ledgers, C22/34. Note: Figures given in italic are totals.

Nicol W & Co

Brown & Co Nicol W & Co Nicol W & Co Hemming S & Co Dewham GR & Co

Sheffield Bombay Bombay London Manchester

Bombay

Drawer

Whence Drawn Alexanders Cunliffes & Co Alexanders Cunliffes & Co Alexanders Cunliffes & Co Alexanders Cunliffes & Co National Discount Co Smith R & Co Discount Corporation Ltd Discount Corporation Ltd Discount Corporation Ltd Frith Sands & Co Cunliffe R & Co Harwood Knight & Co Discount Corporation Ltd National Discount Co Harwood Knight & Co Alliance Bank Barclay & Co

Acceptors or Discounters

Smith, Fleming & Co.

May 23 May 26 May 26 May 26 July 18 Aug 9 May 25 May 25 May 25 June 11 June 8 May 28 May 26 July 9 May 28 June 11 July 27 May 17

Due

Table 3.2 (continued)

302 131,304

£

4 4

s

With

9 10

d

2,000 42,060

1

2

2

6

4 17

1

10

d

3

s

Upon

5,000 40,060

622 5,000 5,000 855 5,112

£

Discounts

With £

F. 1632

2,000 435 44,368

1,800 2,007 5,000 5,726 2,000 4,800 6,000 6,600

£

Upon

Advances

Where It All Began: Lending of Last Resort at the Bank of England

125

called it) included predominantly merchant banks and trading houses, such as Smith Fleming & Co. It was a mixed bag by nature: Merchant bankers being heavily involved in international commodity trade, the line separating “traders” and “merchants” was thin. In the beginning, all customers of the bank were included in this one ledger but, as specialization increased in the money market, some classes of customers were given special ledgers. “Bill brokers” had their own. Under this item were found a number (though not all of them) of money market funds variously known as “bill brokers,” “discount houses,” “discount brokers,” or “credit companies.”20 Yet another group, the “bankers,” initially included in the bill brokers ledgers, came to have a ledger of its own (in 1864).21 These were joint-stock, such as the London Joint-Stock Bank, or private, such as Glyn Mills Currie & Co. They could have the bulk of their business in London (such as the ones just mentioned), in the rest of the United Kingdom (such as the Royal Bank of Liverpool), in colonies (such as the Union Bank of Australia), or abroad (such as the Imperial Ottoman Bank).22 Finally, there were discounted bills that had not been drawn on customers of the bank (probably because the guarantees offered by the discounters were considered sufficient). These were recorded in the so-called “upon ledgers.” Taken together, these ledgers covered all the material that was taken to the bank.

II. The Shadow Banking System and the Crisis of 1866 As noted earlier, Withers (1910) called the bill brokers “ancillary.” They are the Cinderella of the reports to the National Monetary Commission, perhaps because, by the late nineteenth century, they had managed to become such a perfectly integrated part of the money market machinery that they could go unnoticed. Yet their importance never escaped the attention of the best connoisseurs of the London money market (King, 1935, 1936; Sayers, 1968). King (1935, 1936) strongly emphasized the role of bill brokers in promoting the market for acceptances in the first half of the nineteenth century. These were started as private finance companies, with unlimited liability, and essentially matched the supply and demand of bills. As funds do, they looked for safe instruments with higher returns (for instance, because they came from initially segmented markets) and leveraged (King, 1935). According to King (1936), bill brokers flourished after the 1825 crisis, when complete illiquidity of the interbank acceptance market caused by rampant credit rationing pushed many London-based

126

Marc Flandreau and Stefano Ugolini

commercial banks to bankruptcy.23 Survivor banks would have pledged never to experience such a situation any more and stopped holding their liquid resources in bills. This created the opportunity for bill brokers to transform into money market funds: They gradually came to attract on-call deposits from commercial banks, which they reinvested in the acceptance market.24 To put it differently, bill brokers would have emerged because of their willingness to bear the risk of balance-sheet mismatches – which commercial banks were no longer ready to do. In the 1840s, two private firms, Overend-Gurney and Alexanders, stood as leaders of this industry. They had very large credibility and are generally described as having captured an increasing market share (see King, 1935, for rough midcentury estimates). The resulting leverage, which was increased over time, boosted returns (King, 1936, gives the ratio of capital to deposits as 1:10–15 in 1847 and rising). Partners had the reputation of having amassed “fabulous fortunes.”25 In a first stage, the development of bill brokers is said to have been supported by the Bank of England. The growth of the money market in London occurred when the Bank of England permitted certain chosen bill brokers to open discount accounts.26 When the brokers sought liquidity, they could go to Threadneedle Street and found the bank ready to discount the bills of exchange they held.27 Several authors have described the relation between the Bank of England and the bill brokers as symbiotic. There were constant exchanges between leaders of the industry and the bank. Another theme we find in the literature is the role of bill brokers as a transmission mechanism for monetary policy. The bill brokers had large, leveraged inventories and tended to suffer when the bank rate rose, because liquidation then occurred at a loss. According to Sayers (1968), a sudden hike in the Bank of England’s rate could easily wipe out one year’s profits. As a result, when a bank rate increase was in sight, bill brokers covered themselves by pushing interest rates in the open market, thus making the bank rate effective ahead of actual changes.28 Because of the deep structural changes dictated by Peel’s Act, however, an adversarial relation developed after 1844, and further deteriorated with the 1857 crisis. As the subsequent Parliamentary Committee pointed out, the bank found that, during the crisis, about 36% of London advances had been made to bill brokers “partly upon securities which, under other circumstances, the Bank would have been unwilling to accept.”29 This large number contrasted with the smaller figures that were observed in normal times, when bill brokers tried to minimize their refinancing and reserves at

Where It All Began: Lending of Last Resort at the Bank of England

127

the Bank of England.30 The bank decided that the brokers were free riding on the bank’s window. This concern, according to Bagehot (1873) and a number of other contemporary and subsequent writers, was amplified by the directors’ preoccupation with profitability. The bank would have suffered from brokers’ competition and was thus less and less willing to help them out in difficult times. Assurance that they would be supported by the bank, it was argued, made them even more aggressive in normal times. As a result, in March 1858, the bank inaugurated a new rule that banned bill brokers from discounts and, practically, advances too.31 Support in crisis times was not excluded, but the bank would see. This led to an era of conflicting relations, and King argued that the “sixties therefore, were marked by a pronounced lack of co-operation between the Bank and the bill market.”32 Verbal threats and retaliatory moves followed. The bank was said to be discriminating against bill brokers. The heart of the confrontation was with the leading discount house (Overend-Gurney). In 1860, in an act of defiance, Overend-Gurney suddenly withdrew from their account at the bank “no less than £1,650,000 all in £1,000 notes.” The bank had no other solution than to raise brutally the interest rates, causing chaos in the money market.33 De facto, this put an end to the relationship between the bank and Overend-Gurney: Despite still showing up in the list of agreed discounters, the bill broker ceased to rely on the bank’s rediscounting facilities in the following years.34 The events of 1866 were a direct consequence of this situation. The full story of the money market during that period remains to be written. An issue that emerges clearly from earlier accounts is that the bank was preoccupied with what we would call today supervisory and prudential issues. The decade from the mid-1850s saw the expansion of international trade and the increased role of British capital in funding it. The liquidity of the London market gave it a competitive edge for both imports and exports. Reflecting the initial illiquidity of a number of trading niches, merchant banks moving into trade acceptances secured large commissions, which they could get without immobilizing any resource, provided that there were ready buyers for the bills.35 Continental merchant bankers moved to London to avail themselves of these enormous benefits, and joint-stock banks with an international orientation were created in the 1860s.36 Attracted by the fortunes of the early leaders in the field, and pushed by the resulting supply of bills, which looked for holders, money market vehicles were created in the form of joint-stock discount companies that took

128

Marc Flandreau and Stefano Ugolini

advantage of the new limited liability law. Several companies were created such as the London Discount, the National Discount, the Joint Stock Discount, or later the Discount Corporation, the Consolidated Discount Company, the Mercantile Discount Company, and the Financial Discount Company. Just like modern money market funds, they were supposed to invest in blue chip bills but often ended up attempting to boost returns by taking more risk on board – in the familiar way: They invested in illiquid securities. It was also said that their limited liability setup made them less vigilant than their private predecessors. There were suggestions of “questionable operations” and adventures in “paper that [ . . . ] should not have [been] touched.”37 Allegations were made that weakening in investing standards facilitated dubious forms of origination. These included “finance” or “accommodation bills” whereby a firm asked a correspondent to draw on itself without real counterpart transaction, the employment of agents to push the bills into discount houses, the creation of a circulation of fictitious credit among networks of suppliers, or the mortgaging of bills with long maturities (which amounted to securing the bill not by a real security, but by another one). Overend-Gurney had the misfortune to buck the trend. Whereas it was previously known as a standard of prudence concerned with “setting its face” against questionable practices, it developed after 1857 into something that looked more like a financial conglomerate. Successive failures of companies whose bills it had subscribed led it to end up with industrial assets, which it tried to run for its own account. At one point, Overend-Gurney owned two miniature fleets that had belonged to Anglo-Greek merchants. The firm was also heavily invested in railway shares and other industrial securities, thus essentially becoming a universal bank. In what shareholders later described as a last-ditch attempt to hide its collapse (but judges, and King, 1936, disagreed), the firm finally transformed itself into a limited liability company. The stock market collapse of late 1865 and early 1866 battered the company’s balance sheet. Failure of a number of customers forced Overend-Gurney into further losses. The Bank of England was approached but the “Governor took the view that the Bank could not assist one concern unless it was prepared to also assist the many others which were known to be in similar plight.”38 This was decided after a confidential report was commissioned to investigate whether assistance by the bank or a consortium of London commercial banks was merited. Desperate calls to other bankers were unsuccessful and at 3:30 p.m., May 10, 1866, Overend Gurney & Co. suspended payment.

Where It All Began: Lending of Last Resort at the Bank of England

129

The result (May 10 and 11) was the “wildest panic”: Contemporaries compared the event to an “earthquake,” and King (1936) writes that it is “impossible to describe the terror and anxiety which took possession of men’s minds for the remainder of that and the whole of the succeeding day.”39 Markets seized, all transactions were suspended. The financial system ground to a halt, and the only thing people wanted was Bank of England notes or bullion. Several banks and discount houses stopped payments or came close to it. Meanwhile, the bank met all “legitimate” demands and lent over £4 million in one day while its reserve fell by close to £3 million. Then the governors sought from the treasury the permission to infringe on the Bank Act (suspend convertibility), obtained it, raised the bank rate further, and stood ready to provide massive relief. As in previous crises, “suspension” of the Bank Charter Act was the signal for the panic to subside.

III. Credit in Ebb and Flow: Who Came? The evidence we constructed in this chapter is destined to provide help in understanding better the lending policy of the Bank of England during the crisis of 1866. In this section we begin our foray by documenting the profile and needs of those who came to the Bank of England to get cash. We work with the daily discount ledgers and identify both volumes and the identity of those who came to secure cash. Because the bank worried about customers’ types, such information is available in the ledgers. Figures 3.1(a) and (b) show daily (nominal) amounts sought for in either discounts or advances, as well as the amounts rejected for each category, in May 1865 and 1866, respectively (each business day is represented as a bar). As can be seen, lending literally exploded on May 11 as soon as the Bank of England’s discount window reopened. As a result, the crisis month (May 1866) was characterized by much larger amounts of cash supplied compared with the normal month (May 1865). The share of rejected bills was also reduced in May 1866 compared with its 1865 counterpart. This is suggestive of an extensive role of the Bank of England in supporting the market. Last, we see that discounts predominated during both periods, but the relative share of advances increased markedly during the crisis and neared half of the amounts provided in the peak of the crisis. The implication is that in crisis times the range of instruments supplied and accepted by the bank was broadened as people desperately sought to provide adequate collateral in exchange for cash.

(b)

Figure 3.1. Total amounts discounted, advanced, and rejected by the Bank of England. (a) 1865, (b) 1866. Source: Bank of England Archive, Daily Discounts, C 28/25–26.

31/05/1866

30/05/1866

29/05/1866

28/05/1866

26/05/1866

25/05/1866

Advances Accorded 31/05/1865

30/05/1865

29/05/1865

27/05/1865

26/05/1865

25/05/1865

23/05/1865 24/05/1865

22/05/1865

20/05/1865

19/05/1865

18/05/1865

17/05/1865

16/05/1865

Advances Accorded

23/05/1866 24/05/1866

22/05/1866

21/05/1866

19/05/1866

18/05/1866

Bills Rejected

17/05/1866

15/05/1865

12/05/1865 13/05/1865

11/05/1865

10/05/1865

09/05/1865

Bills Rejected

16/05/1866

15/05/1866

12/05/1866 14/05/1866

11/05/1866

Bills Discounted

10/05/1866

08/05/1865

06/05/1865

04/05/1865 05/05/1865

03/05/1865

02/05/1865

01/05/1865

Thousands

Bills Discounted

09/05/1866

08/05/1866

07/05/1866

04/05/1866 05/05/1866

03/05/1866

02/05/1866

01/05/1866

Thousands

130 Marc Flandreau and Stefano Ugolini May 1865

400 Advances Rejected

200

(a)

May 1866

1,800 Advances Rejected

1,600

1,400

1,200

1,000

800

600

400

200

Where It All Began: Lending of Last Resort at the Bank of England

131

The next figures decompose the amounts distributed in discounts (Figures 3.2(a) and (b)) and advances (Figures 3.3(a) and 3(b)) according to the “institutional” categories previously identified. We separate amounts received by bill brokers, bankers, and “ordinary” discounters (which mix together merchant bank and other “trading houses”). Let’s begin with discounts (Figures 3.2(a) and (b)). The crisis saw a dramatic transformation in the identity of those who came in. Although bill brokers and bankers were virtually absent in 1865, they became very important customers during the crisis. A similar pattern is observed for advances: Again, banks and bill brokers represent a large share of the amounts advanced during the crisis (Figures 3.3(a) and (b)).40 The reasons for the changeover in the position of bill brokers and banks are of course natural in view of what we said earlier: Banks faced the risk of a run of depositors and sought to increase their cash holdings. This was secured at a lower rate on the interbank market, but in the case of a panic this market froze and the only options they had were either going to the bank or withdrawing their on-call deposits from the bill brokers. Bill brokers, who managed their portfolio of bills with resources from the banking sector, had to meet the banks’ cash withdrawals. The Bank of England then became the natural counterparty in a vanishing market. There was just nowhere else to go, explaining why the bank had to support the market and why, in such instances, it could always expect to benefit from a suspension of the Bank Act – as this was the only way to backstop the market. Figure 3.3(b) suggests why the bank’s anti-bill-brokers rhetoric started in 1858 was immediately put to rest when the crisis hit: The share of bill brokers in advances made in London during the crisis of 1866 is of the same order of magnitude as the one that had been observed during the crisis of 1857. The Bank of England may have had normal times’ customers, loyalties, and preferences. But in a crisis it was just impossible to escape the responsibilities laid on its shoulders by the community (and encapsulated in “suspensions” of the Bank Act, granted by the treasury). Although it could continue to tender to its regular customers in difficult times (and we see that discounts and advances to merchant banks and trading houses increased as well), it was also bound to enlarge the scale and scope of its liquidity provision operation. To deepen our foray, we now take a look at the characteristics of the population of customers who came to the bank’s window in 1865 and 1866 to get discounts or advances. This we do by collecting data from the daily discounts ledgers. The results are organized, not on a discount event basis, but on a discounter basis: This means that in the case in which a discounter came in several times during that month, we use the total of all discounts

01/05/1866 02/05/1866 03/05/1866 04/05/1866 05/05/1866 07/05/1866 08/05/1866 09/05/1866 10/05/1866 11/05/1866 12/05/1866 14/05/1866 15/05/1866 16/05/1866 17/05/1866 18/05/1866 19/05/1866 21/05/1866 22/05/1866 23/05/1866 24/05/1866 25/05/1866 26/05/1866 28/05/1866 29/05/1866 30/05/1866 31/05/1866

Thousands

01/05/1865 02/05/1865 03/05/1865 04/05/1865 05/05/1865 06/05/1865 08/05/1865 09/05/1865 10/05/1865 11/05/1865 12/05/1865 13/05/1865 15/05/1865 16/05/1865 17/05/1865 18/05/1865 19/05/1865 20/05/1865 22/05/1865 23/05/1865 24/05/1865 25/05/1865 26/05/1865 27/05/1865 29/05/1865 30/05/1865 31/05/1865

Thousands

132 Marc Flandreau and Stefano Ugolini

Commercial Banks Discounts - May 1865

400

Bill Brokers

Merchant Banks and Trading Houses

200

-

(a)

Discounts - May 1866

1,800 Commercial Banks Merchant Banks and Trading Houses

1,600

1,400

1,200

1,000

800

600

400

200

-

(b)

Figure 3.2. Total amounts discounted by the Bank of England, per type of customer. (a) 1865, (b) 1866. Source: Authors, from database.

Thousands

Merchant Banks and Trading Houses

Advances - May 1865

Commercial Banks

(a)

133

Merchant Banks and Trading Houses

Advances - May 1866

Commercial Banks

Where It All Began: Lending of Last Resort at the Bank of England

400

Bill Brokers

01/05/1865 02/05/1865 03/05/1865 04/05/1865 05/05/1865 06/05/1865 08/05/1865 09/05/1865 10/05/1865 11/05/1865 12/05/1865 13/05/1865 15/05/1865 16/05/1865 17/05/1865 18/05/1865 19/05/1865 20/05/1865 22/05/1865 23/05/1865 24/05/1865 25/05/1865 26/05/1865 27/05/1865 29/05/1865 30/05/1865 31/05/1865

200

-

1,800

1,600

1,400

1,200

1,000

800

600

400

01/05/1866 02/05/1866 03/05/1866 04/05/1866 05/05/1866 07/05/1866 08/05/1866 09/05/1866 10/05/1866 11/05/1866 12/05/1866 14/05/1866 15/05/1866 16/05/1866 17/05/1866 18/05/1866 19/05/1866 21/05/1866 22/05/1866 23/05/1866 24/05/1866 25/05/1866 26/05/1866 28/05/1866 29/05/1866 30/05/1866 31/05/1866

200

-

(b)

Figure 3.3. Total amounts advanced by the Bank of England, per type of customer. (a) 1865, (b) 1866. Source: Authors, from database.

Thousands

134

Marc Flandreau and Stefano Ugolini Discounts 100 90

Sums Discounted (%)

80 70 60 50

Cumulative Discounts 1865

40 30 20

Cumulative Discounts 1866

10 0 0

10

20 30 40 50 60 70 80 Number of Discounters (%)

90 100

(a) Advances 100 90

Sums Advanced (%)

80 70 60 Cumulative Advances 1865

50 40 30 20

Cumulative Advances 1866

10 0 0

10

20 30 40 50 60 70 80 Number of Borrowers (%)

90 100

(b)

Figure 3.4. Pareto curves for bank’s counterparties: cumulative proportion of operations by x% borrowers. (a) Discounts, (b) Advances. Source: Authors, from database.

Where It All Began: Lending of Last Resort at the Bank of England

135

Table 3.3a. Descriptive Statistics for the Population of Borrowers (Discounts) May 1865

May 1866

Number

269

Number

372

% of Bank’s Customers

61%

% of Bank’s Customers

74%

30.00 100,974.62 7,999.22 3,690.00 2,151,791.03

Min Max Mean Median Total

43.81 692,520.76 27,584.59 5,820.18 10,261,467.88

Min Max Mean Median Total

made with that customer for that month. As can be seen in Table 3.3a, there were 269 customers who came to the bank in May 1865 to get discounts, and 372 in May 1866 (representing respectively 61 percent and 74 percent of the bank’s eligible discounters for the respective years).41 Customers asked for widely varied amounts of cash, ranging from £30 to more than £100,000 in 1865, and from £43.81 to £692,520 in 1866. Reflecting this increase in maximum amounts required, the mean also shot up from about £8,000 to about £27,000 and the median also rose (from about £3,700 to about £5,800). Similar features are observed for the advances, which are reported in Table 3.3b. A nice way to capture what was going on is to construct “Pareto curves” of the demand for discounts and advances during the two periods. This is done in Figures 3.4(a) and (b). Whereas in 1866 the 20% largest discounters received 80% of the cash dispensed by the Bank of England (the Pareto rule!), in 1865 the proportion had been only 65% (Figure 3.4(a)). A similar Table 3.3b. Descriptive Statistics for the Population of Borrowers (Advances) May 1865

May 1866

Number

25

Number

69

% of Bank’s Customers

6%

% of Bank’s Customers

14%

1,200.00 138,000.00 20,084.00 10,400.00 502,100.00

Min Max Mean Median Total

800.00 750,000.00 74,611.59 20,000.00 5,148,200.00

Min Max Mean Median Total

Source: Authors, from database.

May 1865: Top 50 Discounters Commercial Banks

Merchant Banks and Trading Houses

Bischoffsheim & Goldschmidt Cavan Lubbock & Co Frith Sands & Co Allard J Ebbw-Vale Co Lemme JL & Co Raphael & Sons Finlay Campbell & Co Pinto Perez Ashley & Co Samuel Montagu & Co Wiggin J & Co Zarifi Brothers & Co Lane Hankey & Co Dent Palmer & Co Suse & Sibeth

0

100

200

300

400

500

600

700

Thousands (a) May 1866: Top 50 Discounters Bill Brokers

Commercial Banks

Merchant Banks and Trading Houses

Alexanders Cunliffes & Co The National Discount Co Barclay & Co Harwood Knight & Allen Oriental Bank Corporation The City Bank London & County Bank Brightwen Gillett & Co Drake Kleinwort & Cohen Smith Fleming & Co The Colonial Co Ltd Oppenheim S & Sons Raphael & Sons Huth F & Co Frith Sands & Co

0

100

200

300

400 Thousands (b)

500

600

700

Figure 3.5. Top discounters at the Bank of England, per type of customer. (a) 1865, (b) 1866. Source: Authors, from database.

May 1865: Top 50 Advances Merchant Banks and Trading Houses Smith Fleming & Co Frith Sands & Co Wallace Bros Ebbw-Vale Company Robert Smith & Co Elias & Co Yonle F Dadalhai Naoroji & Co Gledstanes & Co Framjee & Co Boutcher Mortimore & Co Elworthy FJ John Crossley & Sons Ltd Arthur & Co Morgan Gellibrand & Co

0

100

200

300

400

500

600

700

800

Thousands

(a) May 1866: Top 50 Advances Bill Brokers

Commercial Banks

Merchant Banks and Trading Houses

Agra & Masterman Bank London & Westminster Bank Bank of London Harwood Kinght & Allen Alliance Bank The Discount Corporation Ltd Lawes R & Co Frith Sands & Co Barclay & Co Smith Fleming & Co Barnett & Co Sheppards Pelly & Co Bank of Hindustan Ebbw-Vale Company Cunliffe R Son & Co

0

100

200

300

400

500

600

700

800

Thousands (b)

Figure 3.6. Top advances at the Bank of England, per type of customer. (a) 1865, (b) 1866. Source: Authors, from database.

138

Marc Flandreau and Stefano Ugolini

pattern is observed for advances: The 20% largest receivers of advances secured more than 75% of totals in 1866 but only about 60% in 1865 (Figure 3.4(b)). Another way to put it is to note that the top three discounters received 13% of the total in 1865, but 18% in 1866. Respective numbers for the top ten are 30% and 36%. In other words, the distribution of funds was more unequal during crises. This is consistent with the view that there was more “commercial lending” in normal times, whereas during crises generous lending to the needy predominated. Last, we also note that this “concentration” of lending should not obfuscate the fact that lending remained quite scattered: The number of institutions receiving significant amounts was not modest (there were more recipients in 1866 than in 1865, we found). In other words, central bank lending in crises was both extensive (more aggregate lending to all) and intensive (more relative lending to some). Of course, in view of the previous finding that the crisis was also characterized by the emergence of certain customers, it is tempting to argue that the reason for the increase in inequality was the arrival at the bank’s window of cash-hungry financial intermediaries who sought to secure large amounts of refinancing. To explore this, we delve further in the data and take a look at the identity of the top discounters and recipients of advances. Figures 3.5(a) and (b) and 3.6(a) and (b) show “market shares” (shares in total amounts during the relevant months) of the top-fifty institutions receiving the biggest amounts of discounts and advances in May 1865 and May 1866, respectively. As can be seen, the evidence fully confirms the impression from earlier figures. The increase in the share of bill brokers and commercial banks during the crisis, as well as the rise of Gini coefficients, does reflect the arrival of a limited number of customers who asked for (and received) generous credit. In 1865, the top-three discounters belonged to the “merchant banks and trading houses” category.42 In 1866, however, the top three, for much bigger amounts, were two leading bill brokers (private Alexanders Cunliffes & Co. and joint-stock National Discount Co.) and one private commercial bank (Barclay & Co.). A very similar phenomenon occurs for advances. There again, merchant banks and trading houses dominated in 1865, whereas bill brokers and commercial banks led the way in 1866.43 We conclude by emphasizing that the increase in the concentration of discounts during crises was due to the sudden arrival of big requests from institutions that were not regular customers of the bank – financial intermediaries facing liquidity shocks. In crisis mode, Bank of England lending did continue to service the London traders and merchants.44 But their requests were dwarfed by the support granted to London “financial Gibraltars” – both

Where It All Began: Lending of Last Resort at the Bank of England

139

banks and shadow banks, which the bank was prepared to shore up in what it saw as “urgent circumstances.”

IV. Discounting in Ebb and Flow: What Did They Bring in? Next, we study what discounters did bring in. This means opening the black box of the money market and getting an understanding of the types of instruments that were allowed to flow freely from the market to the bank. Data limitation imposes our narrowing down the focus of our study to the geography of bills discounted. We thus exclude securities pledged as collateral for advances.45 Given the statistical importance of discounts, this should nonetheless provide relevant information. Addressing this question breaks new ground. We are not aware of any related previous attempt to uncover the nature of the material traded in the London money market, apart from the discussion by Sayers (1968) of the portfolio of a junior bill broker (Gilletts) at two benchmark dates. Most available evidence we are aware of is qualitative.46 Ideally, one would want to get some idea of the “risks” associated with the categories of paper that were taken by the bank. However, for lack of an independent, “rating-like” assessment of the bills, we are bound to circumnavigate the issue a bit and find indirect ways to approach the contours of eligible instruments. Three questions guide our discussion. First, we are interested in knowing the respective proportion of domestic versus foreign bills taken by the bank. The reason is that the expansion in discount houses was related to the increase in international trade: It would be interesting to know the extent to which the bank did support this trend in the market.47 Second, we are interested in the identity of the acceptors, and in particular in knowing whether the crisis led to distortion in the type of paper that was brought in. Earlier accounts, such as that of Chapman (1984), suggest that the market for acceptances was very concentrated, reflecting the quality of a limited number of signatures.48 One interesting issue would be to determine whether the Bank of England delegated to prestigious acceptors the responsibility for screening the bills (in which case it would tend to concentrate its discounts on a few high-prestige signatures), or whether instead it sought to diversify its exposure (in which case we would expect the bank to buy bills endorsed by many different acceptors). Third, we are also interested in knowing more about the geography of sterling acceptances. In particular, we would like to test whether it reflected British trade patterns. To the extent that acceptances were predominantly created through the infrastructure of trade finance, we expect Bank of England’s material to have reflected

140

Marc Flandreau and Stefano Ugolini

underlying opportunities (trade shares), provided that the bank did support the new tendencies in an undiscriminating way. As discussed in Section I, the location of the drawer and thus the geographical origin of the bill was documented in the customers’ ledgers: In Smith Fleming & Co.’s account shown in Table 3.2, we see that discounted bills had been drawn from places such as Bombay, Shanghai, and Montevideo. As a result, it is possible to reconstruct most of the geographical origins of the bills. Provided that a proper sample is constructed, we should in principle be able to provide answers to the questions we raised. In what follows, we exploit information for two separate samples, corresponding to the portfolio of bills discounted by the “top discounters” and “top acceptors.”49

IV.A. The Inland/Foreign Split We begin with the inland/foreign split of Bank of England’s discounts. That is, we document the share of the value of bills drawn from abroad in the value of total discounts by the Bank of England. Using both the top discounters and top acceptors, we found that the share of foreign bills was huge in 1865 and 1866. The percentages of foreign bills were 85 percent in 1865 and 63 percent in 1866. With the top acceptors sample, the proportions were 89 percent and 86 percent, respectively. Beyond the difference across the samples subsequently discussed, the evidence provides strong supportive evidence for the foreign orientation of the prime material traded in the London money market (and thus willingly taken in by the bank). For comparison, some relevant archival material is available for the early nineteenth century. We also found a number of totals computed by the Bank of England itself and reported in the last pages of the Annual Volumes for the Daily Discounts (1854, 1855, 1856, and 1859). Table 3.4 summarizes the evidence and bears witness of a drastic progression in the share of foreign bills compared with that at the beginning of the century and accelerating during the late 1850s and early 1860s. Does it mean that the Bank of England modified its behavior or instead was the composition of originated sterling bills being modified? Figure 3.7 looks at this by comparing the share of foreign bills within the bank’s discounts with the share of foreign bills within the whole British acceptance market during the 1850s.50 As can be seen, the two lines are parallel, with the bank’s portfolio ahead of the entire domestic market in terms of exposure toward international bills – a natural outcome, given the Bank of England’s relationship with London-based, more internationally oriented intermediaries. The suggestion therefore is that the evolution of the material taken by the Bank of England reflected

Where It All Began: Lending of Last Resort at the Bank of England

141

Table 3.4. Share of Foreign Bills in Bank of England’s Discounts, Selected Dates

Date

Method

1800

“Amounts Discounted to Agents with Specialized Activities” (see Clapham 1944, I, pp. 205–6) “Amounts Discounted to Agents with Specialized Activities” (see Clapham 1944, I, pp. 205–6) Bank of England Ledgers Bank of England Ledgers Bank of England Ledgers Bank of England Ledgers Estimated from sample of “top discounters” Estimated from sample of “top discounters”

1810

1854 1855 1856 1859 1865 1866

Percent Foreign Bills

Source

32.5%

M6/1

35.5%

M6/1

35.5% 43.8% 45.6% 51.5% 85.0%

C28/14 C28/15 C28/16 C28/19 authors

63.0%

authors

Source: Authors, from Bank of England Archive.

global trends (although data for the 1860s would be useful too). To conclude, the data emphasize that the development of lending-of-last-resort operations was intrinsically related to the growth of trade finance. And if stories that the growth of discount companies was motivated by the concern with taking advantage of expanding trade finance are to be taken at face value, we are bound to conclude that the Bank of England bucked the trend. The second interesting feature from our data is the fact that during the crisis of 1866, the relative share of foreign bills in the discounters’ sample declined (but their total increased a lot). Top acceptors in the London market were specialists in foreign bills unlikely to change their specialization in the event of a crisis. Therefore, the relative decline in foreign bills in the discounters’ sample during the crisis month (very relative, as it nonetheless remained a hefty 65 percent) does reflect the scramble for cash and the use of domestic bills that did not normally reach the central bank. This is reflected by the rise in the number of acceptors with a greater domestic orientation. Using our top discounters sample, we found 369 identifiable acceptors in May 1865, but 1,055 in May 1866: This increase is much more substantial than the increase in the number of discounters (see Table 3.3a).51 Figure 3.8 looks at the domestic and foreign decomposition of the material turned in by bill brokers, commercial banks, and merchant banks and trading

142

Marc Flandreau and Stefano Ugolini Percentage of Foreign Bills within Total Bank of England Discounts

Percentage of Foreign Bills within Total Annual Stamp Yields in Britain

60 50

Percent

40 30 20 10 0 1854

1855

1856

1857

1858

1859

Figure 3.7. Share of foreign bills in Bank of England’s discounts during the 1850s. Source: Bank of England Archive C28/14–19, and Hughes (1960, p. 299).

Foreign

Inland

6 5

In millions

4 3 2 1

Bill Brokers (May 1865)

Commercial Merchant Banks Bill Brokers and Trading Banks (May (May 1866) Houses (May 1865) 1865)

Commercial Merchant Banks Banks (May and Trading 1866) Houses (May 1866)

Figure 3.8. Breakdown of inland vs. foreign bills (estimates are taken from the “top discounters” sample, and then normalized for the true total of discounts provided by the daily discounts ledgers). Source: Authors, from database.

Where It All Began: Lending of Last Resort at the Bank of England

143

Table 3.5a. Rankings of top 25 acceptors from the top acceptors sample, May 1865 May 1865 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

London Joint Stock Bank 166,862.66 7.75% Union Bank of London 84,419.34 3.92% London & County Bank 69,317.37 3.22% City of Glasgow Bank 52,555.49 2.44% Imperial Ottoman Bank 42,580.81 1.98% Fr¨uhling & Goschen 42,560.03 1.98% The City Bank 39,170.67 1.82% Drake Kleinwort & Cohen 29,261.21 1.36% Bank of London 26,359.61 1.23% Agra & Masterman’s Bank 24,504.00 1.14% Baring Brothers & Co 21,635.55 1.01% Finlay Campbell & Co 19,216.32 0.89% F Huth & Co 19,029.89 0.88% The National Bank 15,793.46 0.73% Finlay Hodgson & Co 14,456.01 0.67% NM Rothschild & Sons 12,853.00 0.60% Union Bank of Australia 12,498.68 0.58% Dadalhai Naoroji & Co 12,000.00 0.56% Glyn Mills Currie & Co 11,956.26 0.56% Merchant Banking Co of London 11,264.87 0.52% Oriental Bank Corporation 11,139.60 0.52% Moses Brothers 10,200.00 0.47% Colonial Bank 10,179.34 0.47% Alliance Bank 9,101.34 0.42% JH Schroder & Co 8,421.57 0.39% TOTAL 777,337.08 36.13%

houses in 1865 and 1866. As can be seen, the increase in domestic material occurred across the board. Therefore, the increase in domestic paper had not so much to do with changes in the identity of discounters, but with the fact that customers brought to the bank proportionately more of the domestic instruments.

IV.B. Key Acceptors Tables 3.5(a) and (b) document, for each period, the ranking and the market share of the biggest acceptors in the top acceptors sample. As can be seen, both ranks and market shares are rather stable. Because this corresponds more closely to the biggest acceptors in either period of both, the suggestion

144

Marc Flandreau and Stefano Ugolini Table 3.5b. Rankings of top 25 acceptors from the top acceptors sample, May 1866 May 1866 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

London Joint Stock Bank 637,028.01 Union Bank of London 474,520.92 The National Bank 321,824.83 Fr¨uhling & Goschen 279,321.03 Agra & Masterman’s Bank 191,511.83 The City Bank 188,088.95 North Western Bank 175,129.64 London & County Bank 150,793.66 Baring Brothers & Co 147,425.16 Royal Bank of Liverpool 146,905.89 Drake Kleinwort & Cohen 144,033.20 F Huth & Co 125,467.88 Finlay Hodgson & Co 123,896.58 City of Glasgow Bank 96,051.60 JS Morgan & Co 95,764.03 Bank of Liverpool 85,577.62 Ebbw-Vale Company Limited 80,771.80 Smith Fleming & Co 80,741.91 Consolidated Bank 80,253.50 R & J Henderson 77,485.63 Oriental Bank Corporation 77,025.64 Finlay Campbell & Co 75,030.05 Merchant Banking Co of London 72,484.53 Dickinson W & Co 62,141.31 Glyn Mills Currie & Co 61,882.74 TOTAL 4,051,157.91

6.21% 4.62% 3.14% 2.72% 1.87% 1.83% 1.71% 1.47% 1.44% 1.43% 1.40% 1.22% 1.21% 0.94% 0.93% 0.83% 0.79% 0.79% 0.78% 0.76% 0.75% 0.73% 0.71% 0.61% 0.60% 39.50%

Source: Authors, from database.

is that, unlike discounters, the ranking of acceptors (i.e., the composition of what was brought, in terms of accepting houses) did not change much notwithstanding the big increase in their total number. One major finding that emerges is that merchant banks hardly represent the only, let alone the main, source of acceptances.52 Contrary to what has been often emphasized, several commercial banks were a prominent source of acceptances in this early period. This seems to be in blatant conflict with accounts that have emphasized the undisputed role of merchant banks in this market. Of course, it could be that there is a major selection bias in the Bank of England material. It could also be that the rise of the supremacy of merchant banks was a later phenomenon, although scholars usually argue exactly the

Where It All Began: Lending of Last Resort at the Bank of England Top 3

145

Top 10

40 35

Percent

30 25 20 15 10 5 0 Discounters 1865

Discounters 1866

Acceptors 1865

Acceptors 1866

Figure 3.9. Market shares of top discounters and top acceptors. Source: Authors, from database.

opposite.53 Another interpretation is that the market for acceptances was much more diverse and scattered than has been recognized so far and that the Bank of England, rather than delegating to a few prestigious houses the responsibility of screening the bills, preferred to diversify its exposure. We come back to this issue in Section V. Next, we ask whether and how things changed with the crisis. We saw that rankings remained stable at the top, but this could go along with many entries at the bottom.54 Figure 3.9 documents the market share of the top-ten and top-three acceptors in 1865 and 1866 and compares it with rankings for discounters. We see that although top discounters had a greater share of totals during the crisis, the share of top acceptors remained stable, emphasizing that the great change was with the amounts brought rather than with the nature of the paper presented. Another interesting thing that can be explored with the help of our data is the way the paper brought to the bank was structured. We do this by looking at the material drawn on a number of leading acceptors. The reason for focusing on leading acceptors is tied to the fact that these will have endorsed substantial values, making inferences more meaningful. There are several ways in which acceptances per acceptor can be organized. We look here at the geographical makeup in order to assess whether acceptors were diversified geographically.55 Results for ten leading acceptors are shown in Figures 3.10(a) and (b).56 Different acceptors had different geographic biases. One sees, for instance, the importance of drafts from the United

146

Marc Flandreau and Stefano Ugolini May 1865 Undetermined

Sub-Saharian Africa

Canada

Rest of England and Wales

Australia and New Zealand

Continental Latin America

Mediterranean

London

Far East

Caribbean

Atlantic and Baltic Europe

India

United States

Scotland and Ireland

200

Thousands

150

100

50

Suse & Sibeth (MB)

Raphael & Sons (MB)

NM Rothschild & Sons (MB)

Fruhling & Goschen (MB)

Drake Kleinwort & Cohen (MB)

Baring Brothers & Co (MB)

Oriental Bank Corporation (CB)

London Joint-Stock Bank (CB)

Colonial Bank (CB)

Barclay & Co (CB)

-

(a) May 1866 Undetermined

Sub-Saharian Africa

Canada

Rest of England and Wales

Australia and New Zealand

Continental Latin America

Mediterranean

London

Far East

Caribbean

Atlantic and Baltic Europe

India

United States

Scotland and Ireland

400

350

300

Thousands

250

200

150

100

50

Suse & Sibeth (MB)

Raphael & Sons (MB)

(MB)

NM Rothschild & Sons

(MB)

Fruhling & Goschen

Cohen (MB)

Drake Kleinwort &

(MB)

Baring Brothers & Co

Oriental Bank

Corporation (CB)

Bank (CB)

London Joint-Stock

Colonial Bank (CB)

Barclay & Co (CB)

-

(b)

Figure 3.10. Geography of drawers for a selection of leading acceptors. (a) 1865, (b) 1866. Source: Authors, from database.

Where It All Began: Lending of Last Resort at the Bank of England

147

States and the Caribbean in Rothschild’s acceptances, or the importance of the United States for Barings, or that of the Mediterranean for Fruhling & Goschen. Therefore, the conventional view that bankers on whom drafts were drawn operated in niches, for which they had information advantages, is fully supported by the data. A big finding from Figure 3.10 is that foreign bills were by no means the exclusive hunting ground of merchant banks. The vast majority of paper on the London Joint-Stock Bank or on the Colonial Bank, for instance, was drawn from abroad. Paper on Barings, on the other hand, included a nontrivial amount of domestic drafts. Of course, banks with clear domestic orientation, such as Barclay & Co., show a much greater share of domestic drafts – as one would expect. In fact, the real split is with the political arrangements prevailing in the foreign zone(s) under consideration: Commercial banks showed mostly drafts from the British Empire (India, the Caribbean, Australia, Hong Kong, and Singapore), whereas merchant banks were drawn by correspondents in non-Empire locations (the United States, Latin America, Continental Europe).57

IV.C. Geography of Acceptances To conclude, we now take a look at the aggregate geography of acceptances. Focusing on the top discounters sample, we document the spatial breakdown for bills drawn abroad and compare it with data on the structure of British trade at about the same date. The assumption, implicit in this exercise, is that there is a sort of “gravity theory” for London bills, which would have brought the imprint of British trade. In other words, as trade finance provided the infrastructure for the development of the money market, regions heavily related to Britain as far as trade is concerned ought to have also drawn more bills on London bankers, and this should be reflected in the composition of the portfolio we study. This is consistent with the notion that the Bank of England did not set its face against the development of the London money market along lines heavily influenced by international trade. To construct the relevant data, we extracted information on the drawing place for all bills covered by the top discounters sample. This was performed for both May 1865 and May 1866. Then we extracted from the Statistical abstract for the principal and other foreign countries (Board of Trade, 1874) and the Statistical abstract for the several colonial and other possessions of the United Kingdom (Board of Trade, 1876) material to organize a cross-sectional analysis of the geographical composition of British trade for the year 1865. Both databases

148

Marc Flandreau and Stefano Ugolini Exports

Imports

Origin of Discounts 1865

Origin of Discounts 1866

45 40 35

Percent

30 25 20 15 10 5 0 British Empire

Northern Europe

Latin America, Caribbean and Atlantic

USA

Other

Figure 3.11. Breakdown of British trade (1865) and of the geography of drawers of the bills included in the “top discounters” sample (May 1865 and May 1866). Source: Authors, from database and Board of Trade (1874, 1876).

were then aggregated in the following way. We identified five main geographic areas that corresponded to broad regions with trade relevance for Britain: British Empire, Northern European markets (Holland, Scandinavia, northern Germany, the Baltic, and Russia), Latin America and the Caribbean, the United States of America, and, finally, “Other,” which covers the rest of Europe, non-Empire Asia and Africa, and the Middle East. The correspondence between trade and finance is shown in Figure 3.11. Although there is no formal criterion to judge the “fit” of the two distributions, the eye impression is indeed one that is consistent with the assumption that trade patterns were closely associated with financial patterns. This is a striking result: Our exploration of the portfolio of acceptances bought by the Bank of England yields a map that is not dissimilar to that implied by the statistics of the Board of Trade. We conclude that this vindicates the view of an association between British trade supremacy and

Where It All Began: Lending of Last Resort at the Bank of England

149

the rise of sterling as an international currency supported by the National Monetary Commission. We also add a twist to the story: It is probably no coincidence that the dramatic increase in international trade of the mid-nineteenth century immediately preceded the rise of modern central banking. During the 1850s and 1860s, both old (merchant banks) and new institutions (joint-stock banks) took advantage of the expansion of global trade and of London financial know-how to boost, with the help of a booming shadow banking system, the size and liquidity of the sterling money market. To perfect this market, the help of the Bank of England was needed, and, given the importance finance had for British political supremacy, it is perhaps not surprising that the Bank of England, despite its vituperation against the bill brokers, found itself doing what was needed when it was needed.

V. Lending of Last Resort and Supervision The pending question is that of determining whether the Bank of England’s policy created moral hazard. As indicated, it has often been argued in the literature that the salient ingredient of nineteenth-century Bank of England lending (and of its transformation into a modern central bank) reached its perfect stage when it managed (in the age of Bagehot) to ignore the identity of borrowers as long as good collateral would be brought in for discount (Goodhart, 1988; Capie, 2002). To some extent, this view is vindicated by our finding that the 1866 crisis changed the identity of those who came to secure liquidity more dramatically than it changed observable characteristics of the paper that was turned in. On the other hand, our evidence brings some substantial qualification to this conventional picture. Had Bank of England lending been really anonymous, one would expect its counters to have been fully open to the public: In principle, anybody bringing in good collateral should have been entitled to borrow. Yet this was not the case. People eligible to borrow from the bank were part of an exclusive club whose membership was subject to a number of conditions; what is more, as Bignon et al. (2012) find, “membership” shrank dramatically throughout the nineteenth century. This did reflect in part consolidation in the banking industry, but not exclusively. In spite of the expansion of London as an international financial center, the Bank of England was restricting direct access to its discount window. Another issue is that of determining what “good collateral” actually was. In the case of advances, the definition of good collateral might have been rather plain, consisting of those exchange-traded securities considered as

150

Marc Flandreau and Stefano Ugolini

eligible by the bank, although “haircuts” were probably applied (we suspect they were based on the “character” of the customer and not only on the collateral).58 Discounts revealed a similar pattern. Although “normally” generated in the course of actual transactions and for the finance of “physical” commodity shipping, the bill of exchange was, from a legal point of view, more similar to a promissory note with multiple guarantees rather than to a physical asset-backed security. In the case of default of the acceptor, the holder of the bill (e.g., the Bank of England) had the right to ask for its payment from its last endorser (viz., the person from whom the bill had been discounted or purchased), who in turn could get even with previous endorsers; in no case, however, did the holder have the right to seize directly the collateral that might have been mentioned on the bill (Seyd, 1868, pp. 81–3). The commodities explicitly mentioned on the bill did not represent a material lien and may have acted as a “psychological” guarantee that the people involved in the transaction (either the acceptor or the drawer, who was legally equivalent to the first endorser) were not engaged in phony operations and would therefore be able to meet their engagements at maturity; in the case of default of all the people involved, the discounter had the right to seize a proportional share of the debtors’ assets, not the commodities concerned by the given transaction. This means that the “value” of a bill of exchange was the names written of it (acceptors and discounters) – hardly an anonymous feature! Differently said, the collateral that backed a bill discounted by the Bank of England was not a real commodity or transaction but the character of the participants to the origination of the bill and the belief that the bank had leverage over this character (perhaps because access to the discount window created value). Withers (1920) summarized this process, through which the so-called finance bill had come into being,59 by saying that the acceptor had “grown from a merchant into an accepting house.”60 This is, according to him, through that process that finance bills (which did not provide any motive for the credit operation) had become the staple instrument of the London money market. As one can see from London money market bulletins published by The Economist, prime finance bills (i.e., endorsed by first-class financial intermediaries) were constantly discounted in the open market at a lower interest rate than trade bills (endorsed by firms specialized in trade rather than in banking). The implication of all this is that lending could not be anonymous, even if the Bank of England had really wished it to be so. Even if its discount window had been “made of frosted glass,” by looking at the paper pushed through the chink, the bank would have immediately known who was on

Where It All Began: Lending of Last Resort at the Bank of England

151

the other side of the window. When it examined the paper, it immediately saw names (information provided by the bill itself), and it is on the basis of these names that it made its decision. This, we argue, is the reason why the bank thought it wise to put in place the monitoring system described in Section I, which allowed it to track its own exposure toward individual acceptors and discounters, but also quite certainly to track with accuracy whether certain market participants were overextending credit. It is likely that the Bank of England could act on this information. This conclusion is of utmost importance to understanding how and why the new policies adopted by the Bank of England in the 1860s did not succumb to moral hazard. The system that emerges from the preceding analysis can be described as a cascade of embedded monitoring and screening operations. The acceptors were in charge of screening the drawers, whom the Bank of England never saw, because they were often foreign residents. The large number of top accepting houses, we found, rarely or never discounted at the bank although the bank held big amounts of bills accepted by them. This suggests that they acted as competitive delegated certifiers (akin to competitive rating agencies) working for the market and committed to honor their pledges and make ends meet. Obviously, their credit would have been badly damaged if the Bank of England had started refusing their paper, and this would have driven them out of the profitable acceptance market.61 In effect, the discount window was the place where the bank supervisory role was really enforced through tight monitoring of discounters. Monitoring of discounters took the form of guarantees taken on discounters’ assets, reputation (discounters had to be “introduced”), word of mouth, or the request to make deposits. The guarantees taken by the bank protected it against losses, of which, in practice, discounters bore all risks. If acceptors would not pay, discounters would be immediately asked to refund the bank – and if unable to do so, they would be expelled from the club. This implies that the discounters were also effectively charged with the responsibility to screen acceptors, just as acceptors were expected to screen drawers. It also implies that the enforcement of this system worked because discounters feared being deprived of membership to the club. Exclusion from the discount window in effect forced valuable information to be revealed and limited moral hazard. To put it bluntly, the Bank of England did limit moral hazard by not lending “anonymously.” This was permitted by the combination of an accurate, individualized screening procedure and individualized penalties that rested on the bank’s crisis lending monopoly and on its credible exclusion threats.

152

Marc Flandreau and Stefano Ugolini

VI. Conclusions This chapter has brought fresh light to the relation among central banking, crisis lending, the shadow banking system, and the making of sterling as an international currency, providing emphasis on the relation between the Bank of England and the international money market. We used so-farunexplored Bank of England’s ledgers to provide a picture of liquidity provision during both stress periods (May 1866, when the Overend-Gurney panic reached its apex) and normal times (May 1865, or one year earlier). Important findings can be organized into three groups. First, regarding liquidity support, we have observed the emergence of new borrowers when crisis hit – and among them, the considerable importance of bill brokers: Although brokers were not coming to the bank in normal times, they drew about one-third of total cash during the crisis. As a result of this scramble for cash, there was a spectacular increase in amounts discounted in crisis periods. Second, regarding the instruments that were brought in: While there was a big modification in the volumes discounted and in the identity of discounters, the material that was discounted did not change much during the crisis compared to normal time. Only “more of the same” paper was brought in, suggesting a mutual understanding between discounters and authorities as to what would constitute eligible paper in any situation. An interesting finding is also that the list of what constituted eligible paper was long: There were very many different houses involved in the origination of bills and whose signature the Bank of England looked upon favorably. In emergencies, the bank recognized that there were many houses, both merchant and commercial institutions, all with geographical niches, that were being drawn upon by foreign correspondents and whose signature the bank recognized (we saw the large predominance of foreign bills and the overlap between the geographical reach of British trade and the geographical composition of acceptances). Lack of a predominant source of acceptances suggests that at this date, leading accepting firms never controlled more than a fraction of the market: to a large extent, therefore, the Bank of England enforced the contestability of the market for acceptances. Although several of these findings were anticipated in earlier literature, some actually run counter to modern wisdom (or go on side tracks). For instance, we found abundant evidence that the Bank of England was minding the shop: Its elaborate system of reporting enabled it to control its exposure to both discounters and acceptors and provided it, at the same time, with intimate knowledge of the state of the market. A new finding of this study is the extent to which this monitoring of the market was personalized

Where It All Began: Lending of Last Resort at the Bank of England

153

rather than anonymous as suggested by previous research. The Bank of England recognized a good bill not by looking at it but by checking its exposure to the signatures on it. Our study also underscores the extent to which the London money market was very international at an early time. There was an extremely tight correspondence between the patterns of the London money market and trade finance, implying continuity between what was “domestic” (monetary policy) and what was “international” (the instruments used to conduct monetary policy). Next, an important feature that emerges from the evidence here is the enormous importance of specialized operators of the London money market, namely the bill brokers. Whereas British writers commissioned by the National Monetary Commission saw them (as already stated) as “chiefly ancillary to banks,” we found that their refinancing was a central facet of lending of last resort, at least in 1866. Why was it so? We speculate that, given the scattered nature of the products traded in the London money market, there was a serious need for experts able to sort bills and construct portfolios, and thus screen, secure, and eventually make ends meet. These specialized dealers did not compete against commercial banks, and as a result they were wholly interested in the success of the market they operated. They provided the Bank of England with a convenient instrument to support the market at arm’s length, provided that they accepted to “play by the rules” – unlike what Overend-Gurney attempted to do. Conversely, to keep ensuring the liquidity of bills, given the risk of mismatch, the bill brokers obviously needed the Bank of England just as badly as the bank needed their knowledge to screen risks. Thus, in summary, although international trade provided one necessary condition for a successful market for acceptances to emerge (the National Monetary Commission’s insight), other microstructural features were needed, and they included the bill brokers and the Bank of England’s lending-of-last-resort operations – together with the monopoly on crisis lending and the enforcement power over money market participants these operations implied. That this was not emphasized too strongly by British experts when asked by American policymakers should not surprise us exceedingly. There, after all, laid the secret of making fire. ARCHIVAL SOURCES

Bank of England Archive (London), Cashiers’ Department: C22/27–34 (Discounters’ Ledgers) C23/3 (Drawing Office Discounters’ Ledgers)

154

Marc Flandreau and Stefano Ugolini

C24/1 (Bankers’ Ledgers) C25/3 (Brokers’ Ledgers) C28/15 and 25–26 (Daily Discounts) C30/3 (Discount Office Accounts: Analyses and Summaries) References Aldrich, N. W. (ed.) (1910). Interviews on the banking and currency systems of England, Scotland, France, Germany, Switzerland, and Italy (Sixty-First Congress, Second Session, Senate Document No. 405). Washington, DC: U.S. Government Printing Office. Bagehot, W. (1873). Lombard Street: A description of the money market. London: King. Bankers’ Almanac (1866). [Banking almanac: Containing a complete banking directory of the United Kingdom and the British colonies, the principal banks of the world, and a banker’s guide to the principal insurance]. London: Thomas Skinner. Bignon, V., Flandreau, M., & Ugolini, S. (2012). Bagehot for beginners: The making of lending of last resort operations in the mid-19th century. Economic History Review, 65(2), 580–608. Board of Trade (1874). Statistical abstract for the principal and other foreign countries in each year from 1860 to 1872. London: Her Majesty’s Stationer’s Office. Board of Trade (1876). Statistical abstract for several colonial and other possessions of the United Kingdom in each year from 1860 to 1874. London: Her Majesty’s Stationer’s Office. Broz, J. L. (1997). International origins of the Federal Reserve System. Ithaca, NY: Cornell University Press. Capie, F. (2002). The emergence of the Bank of England as a mature central bank. In D. Winch and P. K. O’Brien (Eds.), The political economy of British historical experience, 1688–1914 (pp. 295–315). Oxford, UK: Oxford University Press. Chapman, S. (1984). The rise of merchant banking. London: Routledge. Clapham, J. H. (1944). The Bank of England: A history. Cambridge, UK: Cambridge University Press. Clare, G. (1891). A money market primer, and key to the Exchanges. London: Effingham Wilson. Eichengreen, B. J., & Flandreau, M. (2012). The Federal Reserve, the Bank of England, and the rise of the dollar as an international currency, 1914–1939. Open Economies Review, 23(1), 57–87. Fetter, F. W. (1965). Development of British monetary orthodoxy, 1797–1875. Cambridge, MA: Harvard University Press. Flandreau, M. (2004). The glitter of gold: France, bimetallism and the emergence of the international gold standard, 1848–1873. Oxford, UK: Oxford University Press. Flandreau, M., & Gallice, F. (2005). Paris, London and the international money market: Lessons from Paribas, 1885–1913. In Y. Cassis and E. Bussi`ere (Eds.), Paris and London as international financial centers (pp. 78–106). Oxford, UK: Oxford University Press.

Where It All Began: Lending of Last Resort at the Bank of England

155

Flandreau, M. & Jobst, C. (2005). The ties that divide: A network analysis of the international monetary system, 1890–1910. Journal of Economic History, 65(4), 977– 1007. Goodhart, C. A. E. (1972). The business of banking, 1891–1914. London: Weidenfeld & Nicolson. Goodhart, C. A. E. (1988). The evolution of central banks. Cambridge, MA: MIT Press. Gorton, G. (2010). Questions and answers about the financial crisis: Paper prepared for the U.S. Financial Crisis Inquiry Commission. Mimeo. New Haven, CT: Yale University. Hidy, R. W. (1949). The House of Baring in American trade and finance: English merchant bankers at work, 1763–1861. Cambridge, MA: Harvard University Press. Hughes, J. R. T. (1960). Fluctuations in trade, industry and finance: A study of British economic development, 1850–1860. Oxford, UK: Clarendon. Jacobs, L. M. (1910). Bank acceptances (Sixty-First Congress, Second Session, Senate Document No. 569). Washington, DC: U.S. Government Printing Office. Jenks, L. H. (1927). The migration of British capital to 1875. New York: Knopf. King, W. T. C. (1935). The extent of the London discount market in the middle of the nineteenth century. Economica, 2, 321–326. King, W. T. C. (1936). History of the London discount market. London: Routledge. Nishimura, S. (1971). The decline of inland bills of exchange in the London money market, 1855–1913. Cambridge, UK: Cambridge University Press. Okazaki, T. (2007). Micro-aspects of monetary policy: Lender of last resort and selection of banks in prewar Japan. Explorations in Economic History, 44(4), 657–679. Palgrave, R. H. I. (1903). The bank rate and the money market in England, France, Germany, Holland, and Belgium, 1844–1900. London: Murray. Perkins, E. J. (1975). Financing Anglo-American trade: The House of Brown, 1800–1880. Cambridge, MA: Harvard University Press. Philippovich, E. von (1910). History of the Bank of England and its financial services to the state (Sixty-First Congress, Second Session, Senate Document No. 591). Washington, DC: U.S. Government Printing Office. Roberts, R. (1992). Schroders: Merchants and bankers. London: Macmillan. Rozenraad, C. (1900). The international money market. Journal of the Royal Statistical Society, 63(1), 1–40. Sayers, R. S. (1936). Bank of England operations, 1890–1914. London: King. Sayers, R. S. (1968). Gilletts in the London money market, 1867–1967. Oxford, UK: Clarendon. Sayers, R. S. (1976). The Bank of England, 1891–1944. Cambridge, UK: Cambridge University Press. Seyd, E. (1868). Bullion and foreign exchanges, theoretically and practically considered. London: Effingham Wilson. Tamaki, N. (1974). The merchant bankers in the early 1830s. Keio Business Review, 13, 59–70. Tullio, G., & Wolters, J. (2003a). The objectives of British monetary policy during the classical gold standard, 1876–1913: An econometric analysis of domestic and foreign determinants of bank rate (Discussion Paper No. 13). Berlin: Freie Universit¨at Berlin, Fachbereich Wirtschaftswissenschaften. Tullio, G., & Wolters, J. (2003b). The objectives of French monetary policy during the classical gold standard, 1876–1913: An econometric analysis of the determinants of

156

Marc Flandreau and Stefano Ugolini

the Banque de France’s official discount rate (Discussion Paper No. 12). Berlin: Freie Universit¨at Berlin, Fachbereich Wirtschaftswissenschaften. Tullio, G., & Wolters, J. (2003c). The objectives of German monetary policy during the classical gold standard, 1876–1913: An econometric analysis of the determinants of the Reichsbank’s official discount rate (Discussion Paper No. 14). Berlin: Freie Universit¨at Berlin, Fachbereich Wirtschaftswissenschaften. Tullio, G., & Wolters, J. (2007). Monetary policy in Austria-Hungary, 1876–1913: An econometric analysis of the determinants of the central bank’s discount rate and the liquidity ratio. Open Economies Review, 18(5), 521–537. Ugolini, S. (2011). An ‘atypical’ case? The first emergence of Brussels as an international financial centre, 1830–1860. In L. Quennou¨elle-Corre and Y. Cassis (Eds.), Financial centres and international capital flows in the nineteenth and twentieth centuries (pp. 47–70). Oxford, UK: Oxford University Press. Warburg, P. M. (1910). The discount system in Europe (Sixty-First Congress, Second Session, Senate Document No. 402). Washington, DC: U.S. Government Printing Office. Withers, H. (1910). The English banking system. In R. H. I. Palgrave, E. Sykes, and R. M. Holland, The English banking system (Sixty-First Congress, Second Session, Senate Document No. 492, pp. 3–148). Washington, DC: U.S. Government Printing Office. Withers, H. (1920). The meaning of money. New York: Dutton.

Notes * Marc Flandreau is Professor at the Graduate Institute of International and Development Studies, Geneva (Switzerland). Stefano Ugolini is Assistant Professor at Sciences Po Toulouse and LEREPS – University of Toulouse 1 Capitole (France). Many thanks to Will Roberds and Mike Bordo for their comments on an early draft. We also thank our discussant Barry Eichengreen for detailed comments and Charlie Calomiris for valuable feedback. We are immensely grateful to the archivists of the Bank of England (Sarah Millard, Jeanette Sherry, Lara Webb, and Ben White) for their help and patience with our requirements. Extremely dedicated research assistance from Pierre Turgeon is also gratefully acknowledged. We thank David Altig, James Nason, Will Roberds, and Michael Bordo for their invitation to contribute. Ugolini acknowledges financial support from the Bank of Norway. 1 See Lindert (1967); Flandreau and Gallice (2005); Flandreau and Jobst (2006). 2 R. H. Inglis Palgrave was an editor of The Economist between 1877 and 1883, reflecting the continued influence of this journal in setting the tune of proper policy making. 3 We leave aside the statistical volumes that combined information on various countries, such as Aldrich (1910). 4 Withers (1910, p. 56) has an intriguing digression on the Bank of England being the “final arbiter” of the market when the credit of certain houses came under suspicion. The brief discussion suggested a command of deep and complex interactions between prudential regulation and market making, which modern policymakers have (re?)discovered in the course of the subprime crisis. 5 For more detailed discussions of the international dimensions of the London market, see Clare (1891) and Rozenraad (1900).

Where It All Began: Lending of Last Resort at the Bank of England

157

6 Bignon, Flandreau, and Ugolini (2012). 7 This is in contrast with the situation for some other central banks of lesser international importance such as the Bank of Japan, for which recent econometric work is available (Okazaki, 2007). 8 Except for some investigations on the determinants of interest rate setting by the Bank of England such as that by Tullio and Wolters (2003a). For counterpart studies on the central banks of France, Germany, and Austria-Hungary, see Tullio and Wolters (2003b, 2003c, and 2007, respectively). 9 Capie (2002, pp. 310–11). 10 This rule has an effect on study of the 1907 crisis. Because some ledgers contain material covering the period after 1910 (and thus still embargoed), the bank remains reluctant to communicate them today. 11 An exception is Sayers (1968), who was shown by Bank of England archivists the entries for Gilletts in the bank’s “Brokers Ledger.” Having been commissioned a history by Gilletts themselves, he had most probably been provided the needed clearance, thus releasing the bank from its confidentiality duties toward customers. As a result, Sayers was able to document the episodes when Gilletts sought Bank of England support (as rarely as possible). Sayers (1968, pp. 55 ff) also notes that his Bank of England Operations “had not the benefit of access to the Bank’s records, but fits tolerably well with Gilletts’ transactions at the Bank, inspection of which has now graciously been allowed by the Bank.” 12 On the shadow banking system, see Gorton (2010). 13 Withers’s report for the National Monetary Commission (Withers, 1910, pp. 53–5) is thus characteristic when it emphasizes that the business of acceptance is “largely in the hands of the leaders among the old merchant firms, whose acceptance of a bill stamps it at once as a readily negotiable instrument” and states later on (Withers, 1910, p. 61) that “the discount houses in London carry on a business that is chiefly ancillary to that of the banks.” 14 Although the practice of doing advances on securities goes back to the eighteenth century (Clapham, 1944), the statistical separation between “discounts” and “advances” is not available for early times. C30/3 gives the breakdown from only 1853 onward. 15 Bagehot (1873) complained that railway bonds ought to have been included as well. At the time of the National Monetary Commission, these included all securities traded on the London Stock Exchange except those relating to mining companies (highly speculative; see Aldrich, 1910, p. 20). 16 These are found in ledgers from the discount office archive, bearing the title “Rating books, showing each discounter’s credit limit.” These handwritten “rating books” were updated when needed and bore many corrections until a wholly new rating book was issued and in turn updated, corrected, etc. 17 Bank of England Archive C30/3. The source does not tell whether the number corresponds to eligible discounters or those of the eligible discounters who sought discounts from the bank, although we suspect the former to be the case. 18 The “daily discounts” ledgers also contain convenient monthly and yearly recapitulations, with some useful totals, such as the aggregate value of applications received, rejected, a breakdown of advances and discounts, and, occasionally, some additional evidence such as the breakdown between bills drawn by domestic and foreign entities, and known as “inland” and “foreign” bills, respectively.

158

Marc Flandreau and Stefano Ugolini

19 That is, in the “With” column if Smith Fleming & Co. was the borrower, and “Upon” if it was the acceptor, controlling for whether the operation was a discount or advance. “Advances Upon” corresponded to a repo contracted by another customer on bills accepted by Smith. 20 Adding up the entries in the Bank of England ledgers, we find fifty-seven “bill brokers,” a subset of which operated only in 1865–6. The Bankers’ Almanac of 1866 does distinguish between “recognized discount brokers” (i.e., private houses such as Alexanders Cunliffes & Co) and “principal discount and credit companies” (i.e., joint-stock structures that operated as money market funds). All of the Almanac’s “recognized discount brokers” are listed in the bank’s ledgers. But the Bank of England’s bill brokers ledgers also have a large list of private bill brokers not found in the Bankers’ Almanac of that year. Conversely, only seven of the nineteen “principal discount and credit companies” in the Almanac are in the bank’s ledgers (implying that they did not have a discount account). 21 Before that date they were included in the bill brokers ledgers. 22 For completeness we should mention the “Drawing Office discounters,” which included a variety of merchant banks and other trading houses that used to be customers on a regular basis and were thus allowed to have a current account opened at the bank. Within these commercial customers we find bullion dealers and refining houses, industrial concerns such as shipbrokers, brewers, and linen factors, but also the Crown Agents for the Colonies. For practical purposes, in this study, we decided to aggregate this category to the “merchant banks and trading houses.” 23 See Bignon et al. (2012) for evidence on the surge of defaults after the crisis of 1825 and a discussion of pre-Bagehotian credit rationing technique. 24 King (1936, pp. 62–70). 25 Bankers’ Magazine about Samuel Gurney, quoted in King (1936, p. 217). 26 See King (1936, pp. 68–9 and 89–90) as well as Clapham (1944, Vol. II, p. 142). 27 Focusing on the 1830s, Tamaki (1974) describes a system whereby bills drawn by U.S. correspondents on Barings could be invested in by Gurney & Co, who could in that case get refinanced at the Bank of England. 28 This was even recognized by Withers (1910, p. 63). 29 See King (1936, p. 200), who relies on the report of the Select Committee on the Operation of the Bank Act (1858). 30 They earned money from leverage and the difference between the lending rate and the rate at which they secured funds, so any balance at the Bank of England or rediscount there was a loss of money. 31 King (1936). Only a lifeline to advances routinely made during “shuttings” (i.e., when dividends on British debts were paid) was maintained. The text of the March 1858 decision in the Bank of England Archive, which Charles Calomiris kindly communicated to us read, “that habitual advances by Discount or Loan to Bill Brokers, Discount Companies and Money Dealers being calculated to lead them to rely on the assistance of the Bank of England for their security in time of pressure; Advances to Bill Brokers, Discount Companies and Money Dealers shall be confined to Loans made at the period of the Quarterly Advances [shuttings] or to Loans made under special and urgent circumstances which shall be communicated by the Governors at the earliest opportunity to the Court for its approval.” 32 King (1936, p. 216).

Where It All Began: Lending of Last Resort at the Bank of England

159

33 This was abundantly discussed in the contemporary press. See the famous discussion in Bagehot (1873). See King (1936, p. 213) for details. 34 Bank of England Archive, C25/3 and C24/1. 35 For instance, Roberts (1992, pp. 527–37) reports figures suggesting that fees on acceptances (which involved no immobilization of capital provided that bills could be readily sold) were as large as 1.5% in the early 1860s. This large number is consistent with figures for commissions, which are said to have varied from 0.5 percent to 1.5 percent, and shows how specialization in a new market could put a merchant bank close to the upper bracket. 36 See King (1936, pp. 176–7). 37 King (1936, p. 228). 38 King (1936, p. 242). 39 King (1936, p. 243). 40 In aggregate numbers for the entire two months, bill brokers and bankers represent 0 percent and 2 percent, respectively, of the total discounts and advances in 1865, but 21 percent and 33 percent in 1866. 41 The year 1866 saw an upsurge in the number of customers, possibly because the crisis led to an increase in applications. We computed these numbers under the already-mentioned assumption that data in C30/3 relate to the number of eligible discounters, not to the number of applicants. 42 Bischoffsheim & Goldschmidt, Cavan Lubbock & Co., Frith Sands & Co.: These merchant banks had connections with Continental Europe, Canada, and India, respectively. 43 An interesting feature is the greater lead by banks compared with that of bill brokers in advances. This seems to confirm the above-mentioned preference of commercial banks for not discounting bills directly on the market. It may be that, as banks were competing against the Bank of England for commercial credit, they were reluctant to reveal private information by discounting their own paper. In any case, this issue should receive more attention in the future, especially because it has been argued that (albeit for a longer period) the Bank of England really discriminated against advances. Wither (1910, pp. 6–7) argues that at the time of his writing, a customer “taking advances on securities [ . . . ] usually pays one-half of 1 per cent above Bank [discount] rate” that applied to bills. 44 The large presence of Bischoffsheim & Goldschmidt at a time of active international bullion arbitrage reflects the use by some merchant banks of central bank facilities to conduct their operations (Flandreau, 2004). Ugolini (2011) discusses Bischoffsheim’s emergence as leading international merchant bankers. 45 As shown by Figure 3.2, advances are not systematically documented by our sources. This is because some material was made of bundles of bills called “parcels” that were not “unpacked” in statistics (except concerning the identity of acceptors), and securities other than bills are not documented. 46 Such as the discussion in Bagehot (1873) that during the 1825 crises “anything” had been brought to the bank. 47 This would also have an independent value for discussion of the aggregate supply of inland and foreign bills. This matter is not well known. King (1936, p. 271) argues without quoting numbers that “the decline [of the inland bill at the expenses of foreign bills] may be said to have begun shortly after the 1857 crisis, although it

160

48

49

50 51 52 53 54 55 56

57

58

Marc Flandreau and Stefano Ugolini was not until the seventies that it became at all marked.” Later scholars have argued that the growth in the use of foreign bills as opposed to inland bills was a later phenomenon, tied to the “amalgamation movement” in British banking of the late nineteenth century (Nishimura, 1971). There is disagreement, however, as to when this concentration occurred. King (1936, pp. 280–1) suggests that this occurred only after 1870. He argues that other banks (private and joint-stock) were also active in the market of acceptances, but less so – and only later in the century. Chapman (1984, pp. 39–41) believes in a fairly early concentration of the market for acceptances (as early as in the 1830s). A conventional view in previous work is that when the amalgamation movement occurred in the 1890s, the giant clearing banks also became large suppliers of acceptances – although here again it is usually said that merchant banks’ material still reigned supreme. Because covering the whole material discounted by the bank was not practical, we constructed two “samples” that were used to address, either simultaneously or separately, different sets of questions. The first sample uses information obtained from the material brought in by the “top discounters” (for both May 1865 and May 1866) as emerging from the “daily discounts” ledgers. Because we collected all the material for daily discounts, we do know the league tables for top discounters, and this is how the sample was constructed. However, given the modification in the identity of those who came to the Bank of England, this means that the ranking of “top discounters” changes a lot between the two periods. The second sample was captured through a field technique that amounted to collecting the accounts that had, on an eyeball test, “substantial length” (viz., entries with several pages). We called it “top acceptors” for simplicity, but it should be emphasized that we do not know the actual ranking of “top acceptors.” Acceptors were not sorted out by the bank in any way that would permit collection of this information. An interesting test of consistency between the two samples is to match estimated acceptors market shares according to their “random drawing” from information about discounters, with information on their exact market share from acceptors’ data. This enabled us to draw charts that suggest broad consistency. Computed from stamp duty statistics from Hughes (1960, p. 299). Note that this latter number is vastly superior to that for discounters with access to the Bank of England. This result is fully robust for considering the “top discounters” sample instead. See preceding discussion and also Withers (1910, p. 56). The two portfolios of “top discounters” for each period suggest a drastic increase in the number of acceptors. Most accounts suggest they were not (Chapman, 1984). The selection was mostly heuristic, and we do not seek to make any general inference. There are four commercial banks and six merchant banks. Bill brokers do not appear as they usually did not accept bills. This is consistent with Chapman’s (1984, pp. 111–15) claim that the finance of Eurasian trade was left open to newly founded commercial banks by the collapse of agency houses after the 1847 crisis. Unfortunately, we lack information concerning the size and determinants of the haircuts. Because advances were closely monitored by the bank on a per-customer basis (as shown in Table 3.2), we cannot exclude that haircuts were charged according

Where It All Began: Lending of Last Resort at the Bank of England

161

to the borrower’s own situation and rating. As a result, even purely collateralized lending by the bank was probably not anonymous. 59 “It is this quality, inherent in a genuine bill, that gave rise to the saying that banking is the easiest possible business to conduct, when once the banker has grasped the difference between a bill of exchange and a mortgage. We have seen that the genuine bill of exchange is easily negotiable before maturity, and on maturity is cash by the sale of the goods on which it is based. . . . But, as a matter of practical fact, a very large number of the bills drawn are not of this genuine character, and the use of this admirable and efficient instrument of credit has been so extended, that the distinction between it and a mortgage on real property is nowadays sometimes in favour of the latter, which has at any rate something behind it. . . . More often the bill takes a form like this: ‘At ninety days after sight of this First of Exchange (Second Unpaid) pay to the order of Messrs. Jones Two Thousand Pounds Sterling, Value received, and charge the same to a/c as advised’. Experts in credit may be able to hazard a shrewd guess from the appearance of a bill, as to what is behind it. But the phrase ‘Value received’ covers a multitude of mystery, and the difference between a genuine produce bill and a piece of finance paper is often difficult to detect” (Withers 1920, pp. 46–53). 60 “When John Smith [a fictive acceptor] is described as having grown from a merchant into an accepting house, he is supposed to have passed through a process which has been fairly common experience. Like many other merchant houses, he has given up the actual handling and selling of merchandise, though retaining the title of merchant, which is highly honored in the City, and is confining his attention to the profits which he can more easily earn, if his name be good enough, by placing his acceptance at the disposal of borrowers who want to draw on him. The arrangement that he has made with Watt’s [a fictive drawer] banker, and with many other dealers in bills of exchange in other parts of the world, enables them to draw on one another at any time, whether there be produce passing or no, and brings into being the instrument known as a finance bill. By this operation he and they create credit instruments which can be discounted and turned into cash, on the security of their names which are on the bills” (Withers, 1920, pp. 48–9, our italics). 61 There were a number of practical issues that facilitated control of discounters, too. As time passed, the practice of discounting parcels of bills became widespread. Parcels were presented by one discounter but might be made of bills accepted by myriads of different people: In such cases, verification costs became prohibitive and threatened to clog the workings of the bank.

Comments on “Where It All Began: Lending of Last Resort and Bank of England Monitoring During the Overend, Gurney Panic of 1866”1 Barry Eichengreen

Marc Flandreau and Stefano Ugolini have written a fascinating chapter that contains a wealth of new information, archival evidence, and historical analysis. Like other Flandreau and Ugolini papers, it brings to mind a set of Russian dolls. Open one up and you will find another inside. Open that one, and – well, you get the idea. We find here a number of interesting papers hiding within one another. All constitute important additions to the literature and provide valuable new insights. But understanding what lies inside can be a challenge. The first paper – the first doll we encounter – is concerned with Paul Warburg, the National Monetary Commission, and the road to Jekyll Island. This story is familiar. It is the tale of how Warburg saw the existence in Europe and corresponding absence in the United States of a market in bankers acceptances and finance bills, along with a central bank to rediscount them, as a stabilizing (in the United States, destabilizing) influence. Flandreau and Ugolini’s telling of this tale is compelling, although I might quibble with their emphasis on London and the Bank of England as the paradigmatic case. Warburg’s 1910 pamphlet, The Discount System in Europe, in fact, makes as many references to Germany and France as to Britain. This is not surprising: Warburg’s most intimate knowledge was of the German market, because that was where the family firm of M.M. Warburg was headquartered. But I would not challenge the basic story. The second paper – the second doll we encounter – is on the origins of the Bank of England’s lender-of-last-resort function. Flandreau and Ugolini trace this to the lessons the Bank of England drew from the Overend, Gurney crisis of 1866. Overend, Gurney, by their account, was the turning point when the bank assumed the mantle of lender of last resort. This argument, too, follows in a long-standing tradition. The classic statement of the lenderof-last-resort function by Bagehot, in Lombard Street (1873), dates precisely 162

Comments

163

from the aftermath of Overend, Gurney.2 Flandreau and Ugolini argue, likewise, that modern central banking practice was consolidated in Britain and elsewhere in Europe with this episode. But others (viz., Wood, 2003) would dispute this. They characterize the bank as still reluctant to act as a lender of last resort owing to moral hazard concerns until the next crisis and turning point in 1890. They document the bank’s reluctance to build additional reserves that might be used to carry out this function and attribute the improvement in financial stability after 1866 to other factors such as the increased concentration and diversification of the banking system. The third paper – the third doll we encounter – is concerned with the growth of the London market in trade bills. Flandreau and Ugolini show that bills discounted at the Bank of England were at mid-century primarily overseas-trade-related paper. This suggests that the development of a liquid financial market and international-financial-center status are both related to commercial preeminence; in other words, it suggests that finance follows trade. Widely branched British banks could provide credit in the form of loans for inland trade, but another mechanism, namely the provision of trade bills, was needed to provide credit for overseas transactions. Flandreau and Ugolini then establish that this market in trade bills did not grow up spontaneously but rested on the market-making services of the Bank of England. It depended on the existence of a specialized set of bill brokers to screen and rate trade-related paper. And this second prerequisite, in particular, grew up only gradually over time. This last point has obvious implications for the efforts of American policymakers to cultivate the development of an analogous market in the United States, not all of which they drew. The fourth paper – the fourth doll we encounter – analyzes the Bank of England’s actions in the Overend-Gurney crisis itself. It asks this question: What did the bank discount? Who approached the bank for emergency liquidity? How developed was the lender-of-last-resort function? Flandreau and Ugolini’s evaluation is broadly favorable to the bank. Again, not everyone would agree. As Collins (1988) writes, “In the weeks that followed the crisis, several firms complained bitterly that the Bank had not extended adequate assistance. Thus, while the Bank of England had made advances to the market on a scale unequaled in the post-Charter period, its failure to assist Overend, Gurney and its reluctance to extend assistance to other members of the financial community suggest that it did not fulfill its role as lender of last resort.” More generally, one wants to ask this: Was the Bank of England right to let Overend, Gurney fail? Was Overend, Gurney illiquid or insolvent? And even if it was insolvent, did the bank provision adequately

164

Barry Eichengreen

for the corollateral damage caused by its failure? After all, more than 200 other companies and banks failed in the aftermath of Overend, Gurney. This was a serious financial crisis. My impression is that Flandreau and Ugolini would reject the view that 1866 was the Bank of England’s Lehman Brothers moment. Scrutinizing its ledgers, they find evidence of the Bank of England lending to nonbank counterparties. But why, then, were there 200 failures? Did the bank lend only to select nonbank counterparties? Did it lend only against select financial instruments, notwithstanding the fact that the range of instruments discounted by the bank expanded dramatically during the crisis? Finally, and to bring us to the topic of this conference, there are the implications for the United States, the Fed, and the dollar. Flandreau and Ugolini observe that Bank of England intervention in the market required a high-quality, homogeneous instrument to “repo.” Such an instrument did not exist in the first half of the nineteenth century but developed with the growth of British trade, which produced trade acceptances collateralized by goods in transit, carried on British ships, and discounted by British bankers. After 1913, the problem for the new U.S. central bank was the absence of an equally deep and liquid market in trade acceptances. To be sure, the volume of acceptances rose sharply in the 1920s, when the Fed was active in the market, but fell equally sharply in the 1930s when the Fed withdrew (Ferderer, 2003). The reader wants more analysis of why the market failed to sustain itself. Was the key prerequisite for its development the organic growth of a class of brokers, something that can occur only over a long period of time and that the Fed’s indiscriminate purchases of acceptances in the 1920s tended to suppress? Did this mean that when trade acceptances became riskier in the 1930s and the Fed grew more reluctant to discount them, there were no private brokers to screen and sort them? Whether in fact this was the binding constraint is not clear. Contemporary experts paid considerable attention to the growth of a handful of specialized “discounters” or “bill brokers” in New York in the 1920s, which maintained inventories of acceptances and acted as market makers. Could the reason that the market declined in the 1930s be instead that acceptances became risky with the German standstill and the spread of exchange control? It was problematic that acceptances were heavy claims on foreigners – claims on foreigners that became risky with, inter alia, the German standstill – because ability to enforce contracts across borders was limited. The Bank of England had not faced an analogous problem before 1913 because most of the trade acceptances it discounted were generated in connection with

Comments

165

trade with members of the British Empire, where contract enforcement was not an issue. Could it be, then, that the development of modern central banking in Great Britain had as much to do with the fact of the British Empire as it did with the keen insights of Walter Bagehot? References Bagehot, W. (1873). Lombard Street: A description of the money market. London: Henry King. Collins, M. (1988). Money and banking in the UK: A history.London: Croom Helm. Ferderer, P. (2003). Institutional innovation and the creation of liquid financial markets: The case of bankers’ acceptances, 1914–1934. Journal of Economic History, 63, 666–694. Warburg, P. (1910). The discount system in Europe. Washington, DC: U.S. Government Printing Office. Wood, J. (2003). Bagehot’s lender of last resort: A hollow hallowed tradition. Independent Review, 7.

Notes 1 Prepared for the Federal Reserve Bank of Atlanta’s conference on the history of the Federal Reserve System, Jekyll Island, Georgia, November 2010. 2 Bagehot had, of course, already proposed his “lend freely at a penalty/high rate” principle in The Economist, which he edited, in 1866.

FOUR

Volatile Times and Persistent Conceptual Errors U.S. Monetary Policy 1914–1951 Charles W. Calomiris*

“If stupidity got us into this mess, then why can’t it get us out?” – Will Rogers1

I. Introduction This chapter reviews the history of the early (1914–51) period of “monetary policy” under the Federal Reserve System (FRS), defined as policies designed to control the overall supply of liquidity in the financial system, as distinct from lender-of-last-resort policies directed toward the liquidity needs of particular financial institutions (which is treated by Bordo & Wheelock, 2010, in another chapter of this volume). The history of monetary policy generally focuses on four key sets of questions: What did the monetary authority do, why did it behave the way it did, what effects did its policies have, and what should it have done differently? In addressing these questions, the monetary history of 1914–51 can be usefully broken down into three subperiods – World War I, the interwar period, and World War II and its immediate aftermath. The most interesting monetary policy questions relate to the interwar period, during which wartime constraints on monetary policy were absent. The monetary policy of the interwar period has been the subject of voluminous research, and substantial controversy for decades, both with respect to the intentions and constraints that guided monetary policy during this period and with respect to the effects of monetary policy on the economy. Such studies include those by Chandler (1958, 1971), Friedman and Schwartz (1963), Wicker (1966, 1980, 1982, 1996), Brunner and Meltzer (1964, 1968a), Frost (1971), Meltzer (1976, 2003), Temin (1976), Bernanke (1983, 1995), Miron (1986), Wheelock (1990, 1991, 1992), Eichengreen (1992), Romer (1992), Calomiris and Wheelock (1998), Bordo, Choudri, and Schwartz (2002), Calomiris and 166

Volatile Times and Persistent Conceptual Errors

167

Mason (2003a, 2003b), Hanes (2006), Hsieh and Romer (2006), Bernstein, Hughson, and Weidenmier (2010), Fishback (2010), and Calomiris, Mason, and Wheelock (2011). It would not be possible in this chapter to do justice to these and other contributions to this vast literature, or to describe fully, much less lay to rest, the controversies contained in them. Happily, reviewing that literature in detail is also unnecessary, given the recent comprehensive and authoritative contribution by Meltzer (2003), whose first volume of the history of the Fed not only provides the definitive summary of the actions and intentions of policymakers during this period, but also summarizes, and arguably often resolves, the academic controversies that have surrounded those actions ever since.2 My review focuses primarily on a few of the most important and controversial issues surrounding the intentions and consequences of monetary policy during the interwar period, and it does so from the perspective of a larger, historical question: How do monetary policymakers and their critics learn about the proper approach to monetary policy? The obvious part of the answer to that question is that learning must happen over time, largely as a result of trial and error and subsequent analysis of that trial-and-error policy process. The less obvious part of the answer is that the learning process is not uniform across times and places and depends on particular historical circumstances, especially the initial conditions that define one’s priors, and the specific sequence of shocks that one experiences and from which lessons are supposed to be derived. Historical circumstances were not very favorable for clear and speedy learning about U.S. monetary policy from 1914 to 1951, and that explains why so little seems to have been learned during these years. Indeed, we are still debating some of the fundamental questions about what went wrong with monetary policy during this period, what drove those errors, and what effects they had, as my review will show. The remainder of this chapter is organized as follows: Section II begins with a brief background on the founding of the Fed, its mission, and structure. Section III reviews the perceived failure of the Fed during the Depression and argues that the Fed’s failure and the protracted delay in learning from that failure reflected the initial conditions in which it was founded – which included the unique circumstances of the U.S. banking system, the guiding conceptual framework of the real bills doctrine, and the dramatic shocks and rapidly changing structure of the economy in its early years. Section IV considers five key questions in the history of monetary policy during the interwar period: (1) To what extent did the Fed cause the stock market crash of 1929, or alternatively, was the stock market boom

168

Charles W. Calomiris

of 1928–9 a source of instability that warranted Fed intervention to prick a ballooning bubble? (2) To what extent did the gold standard limit the Fed’s ability to prevent monetary contraction during the Depression? (3) To what extent were the four banking panics identified by Friedman and Schwartz times of unique strain that should have awakened the Fed to the need to prevent a contraction of the money supply? (4) Was the economy in a liquidity trap in the early 1930s, or alternatively, would increased open market operations have prevented the Depression? (5) To what extent were the reserve requirement increases of 1936–7 responsible for the recession of 1937–8? Reviewing the changing answers to these questions that have evolved over time illustrates why it was so challenging for Fed officials, and even for subsequent observers of monetary policy, to properly gauge the role that monetary policy played in the Depression and to identify the particular sources of policy errors. These questions, one can argue, do have answers, but that does not make them easy questions to answer, and the difficulty of answering such questions helps explain the protracted process of learning about monetary policy that is the central theme of this chapter. Section V reviews the literature on the Fed’s role in smoothing seasonal fluctuations during the interwar period. Section VI describes the role of the Fed during World War II and the reestablishment of monetary independence from the treasury in the Accord of 1951. Section VII concludes.

II. Intentions of the Fed’s Founders The original mission of the Fed, as revealed by its structure, by the debates that gave rise to it, and by the statements of its leaders, was quite different from its current focus: “[s]table growth was not part of the Federal Reserve’s formal mandate in the early years. Most of the System’s leadership would have denied any responsibility for economic activity or employment” (Meltzer, 2003, p. 9). The primary concerns that gave rise to the Fed were the so-called “inelasticity” of the supply of money and credit and the peculiar U.S. propensity for banking panics. After 1866, the United States was the only economy in the world that continued to suffer from banking panics: Major panics, defined as events in which the New York City Clearing House banks acted cooperatively to deal with liquidity risk, occurred in 1873, 1884, 1890, 1893, 1896, and 1907 (Calomiris & Gorton, 1991). The peculiar propensity for panics in the United States reflected its unique “unit” banking structure: Unlike banks in other countries, banks in the United States historically (with the exception of the antebellum South, where bank branching was widespread, and a few states in the postbellum

Volatile Times and Persistent Conceptual Errors

169

North, where branching was permitted on a limited basis) were constrained to operate in one location, and in the few states that permitted branching, branches were generally not allowed throughout the state, and in no case could banks operate branches across state lines. This unit banking structure made the U.S. financial system uniquely risk prone in its response to real shocks; limitations on branching prevented interregional diversification of loan risks ex ante and hampered the coordination of the banking system’s response to shocks ex post (Calomiris, 2010). Other countries had grown out of banking panics, strictly defined, after 1866 (Bordo, 1985), but in the United States, panics continued.3 Banking panics in the United States were clearly not random events. They occurred at cyclical peaks of economic activity and at seasonal peaks of credit demand in the fall harvesting or spring planting seasons (when the banking system was at its maximum leverage). Calomiris and Gorton (1991) show that, from a cyclical perspective, the panics of the national banking era were quite predictable based on a simple dual-threshold criterion: A banking panic occurred in a particular quarter, if and only if, during the preceding quarter both the liabilities of failed businesses rose by 50 percent or more (seasonally adjusted), and stock market returns fell by 8 percent or more. In the minds of the founders of the Fed, the problems of inelasticity and banking panics were closely related. The Fed was founded in the belief that (1) an elastic supply of currency (in the form of reserves supplied by the Fed) would result in an elastic supply of bank credit, and that (2) this would reduce the incidence of banking panics by reducing the exposure of the banking system to liquidity risk. Spikes in the demand for liquidity (at seasonal or cyclical frequency), if not accommodated by increases in currency, produce momentary scrambles for liquidity that would raise interest rates, result in distress sales of assets, and potentially lead, in extreme circumstances, to banking panics. The Fed’s job was to accommodate the swings in the demand for liquidity, at seasonal and cyclical frequencies, which would also prevent disruptive interruptions to the availability of credit. To help fix ideas, consider the following simple model, illustrated in Table 4.1 and Figure 4.1. Table 4.1 summarizes the balance sheet of a typical bank. Its assets consist of loans and reserves (vault cash plus balances at correspondent banks, and after the founding of the Fed, at Federal Reserve banks); its liabilities consist of deposits, with the difference between assets and liabilities comprising the bank’s equity capital, or net worth. Assume that banks target a low level of default risk on their deposits (for details, see Calomiris & Wilson, 2004), which they maintain through a combination of

170

Charles W. Calomiris Table 4.1. A Representative Bank’s Balance Sheet Assets

Liabilities and Net Worth

80 Loans 20 Liquid Reserves 100 Total Assets

70 Deposits 30 Equity Capital 100 Total Liabilities and Net Worth

a sufficiently high ratio of reserves to loans and a sufficiently high ratio of equity to deposits. In this simple framework, an adverse economic shock causes loan losses, which reduce the ratio of equity to assets. Banks adjust to this shock to restore the low risk of default on deposits by some combination of accumulating capital (which is hard to do at high frequency) and raising their ratio of cash to loans by not renewing some maturing loans (which is easier to do at high frequency). Banks charge for bearing the credit and liquidity risk associated with expanding loans. The short-term loan supply function (Figure 4.1, loan supply without Fed) is upward sloping because short-term increases in lending (holding constant capital and liquid assets) raises the exposure of the bank to both credit risk and liquidity risk. After the founding of the Fed, the new loan supply function is flatter (Figure 4.1, loan supply with Fed). The reason is that, in the presence of the Fed, banks have a new means of dealing with liquidity risk other than recapitalization or loan liquidation; that third option is borrowing from the Fed against some of their loans.

Real Loan Interest Rate

Loan Supply (without Fed)

Loan Supply (with Fed)

High Loan Demand Average Loan Demand Low Loan Demand Real Quantity of Loans

Figure 4.1. The loan market with and without the Fed.

Volatile Times and Persistent Conceptual Errors

171

Banks still have to adjust their lending and capital over time to restore their low default risk, as needed, and the Fed does not bear significant default risk through discount window lending, but banks do not have to make adjustments suddenly out of fears of illiquidity (during a moment of high seasonal leverage in the banking system). That means that, in the presence of the Fed, the banking system will avoid magnifying loan loss risk by creating a scramble for liquidating assets at a time of high leverage. In the presence of the Fed, it is also true that variation in loan demand, seasonally or cyclically, should result in greater variation in the quantity of lending and lower variation in interest rates, as shown in Figure 4.1, which is drawn here assuming that the primary source of shock in the system is variation in loan demand (an assumption that we subsequently evaluate in more detail). Although this model was not articulated in any explicit form by the Fed’s founders, it is a reasonable representation of their view that the founding of the Fed would produce greater elasticity of reserves, which would result in greater elasticity of credit, which in turn would reduce the propensity for banking panics. This model also can explain why the United States was uniquely prone to banking panics. Nationwide branching would have produced greater diversification of loan risk (via the law of large numbers), which would have reduced the effect of loan portfolio shocks on banks’ risk profiles. Furthermore, branching banks could coordinate better to borrow from each other in response to shocks because they were geographically coincident (see Calomiris & Schweikart, 1991, and Calomiris, 2000). Some of the adherents to this view of banking system liquidity risk and the role of the Fed in mitigating it also lamented a closely related structural characteristic of the U.S. banking system that contributed to its liquidity risk: namely the “pyramiding” of reserves in New York City (the tendency for banks in rural areas to park their funds at other banks during the summer and winter seasonal lulls in credit demand). New York City banks reinvested those reserves in the call loan market, resulting in potential linkages between Wall Street securities risks and credit and liquidity risk to the nation’s bank deposits. The banking panics of 1857, 1873, and 1907, and arguably others, had their origins in securities market problems concentrated initially in New York. According to those most concerned about this practice (especially Carter Glass, who chaired the House banking committee that passed the Federal Reserve Act and later oversaw the reform of banking and Fed practices as a senator in the 1930s), the Fed would bring an end to this destabilizing tendency by replacing the interbank deposit market based in New York with a decentralized system of deposits by member banks in each Federal Reserve District at their reserve banks.

172

Charles W. Calomiris

As envisioned by Glass and other Fed founders, the Federal Reserve banks would be a repository of excess reserves during times of low demand for their member banks and a source of additional reserves (via either lending to members or buying assets from them) during high-demand periods. To prevent accommodation of destabilizing increases in demand for reserves (e.g., the fueling of undesirable speculation in stock markets or real estate), the Federal Reserve banks would focus their activities only on purchasing or lending against “real bills,” defined as commercial loans related to the financing of trade. At the Fed’s founding, real bills were expected to be the key asset that would pass between member banks and their reserve banks. After an initial struggle for control over the FRS between the reserve banks and the board in Washington (Meltzer, 2003, pp. 75–82), the twelve Federal Reserve banks succeeded in maintaining a decentralized system until 1935. The reserve banks coordinated their policy decisions through a committee, headed by the New York Fed, but each was free to opt out of any coordinated intervention decided by that committee. That arrangement preserved the decentralized structure originally envisioned by the Fed’s founders.4 The original decentralized structure of the Fed was designed to facilitate a close relationship between the reserve banks in each district and their local members, who collectively owned them, and was designed to prevent any capture of the system by Wall Street speculators or Washington politicians. The early Federal Reserve System’s operational structure and behavior reflected its founders’ intentions regarding the creation of an elastic supply of reserves, adherence to the real bills doctrine, and the decentralization of decision making, albeit imperfectly. The Fed failed to achieve some founders’ goals. Most obviously, banking system reserve pyramiding in New York City continued after the founding of the Fed, owing to a central flaw in the design of the Fed: Reserve banks, unlike New York City banks, did not pay interest on interbank deposits; that gave peripheral banks a strong incentive to deposit their reserves in correspondent banks rather than Federal Reserve banks (White, 1983). Also, contrary to its founders’ vision, the politicization of the system by Washington began almost immediately after the Fed’s founding. Under the pressures of World War I’s financial challenges, the Fed became an important partner in assisting the U.S. government to market its debts. In 1917, reserve requirements were reduced to permit expanded credit to finance the war (Meltzer, 2003, p. 79, footnote 31). And collateral rules for Federal Reserve note issues were relaxed in 1917: The total amount of collateral was reduced, and perhaps more important, promissory notes of

Volatile Times and Persistent Conceptual Errors

173

member banks secured by government bonds could be used as collateral for the notes (Meltzer, 2003, p. 89). At the end of World War I, in the interest of boosting demand for outstanding treasury debts, the Fed also reduced its discount rate for loans collateralized by treasury securities. This had longterm effects. The discount rate reduction led the Fed to abandon its “penalty” rate policy for targeting the discount rate, which had been one of its core founding principles (Meltzer, 2003, pp. 73, 86).5 This change subverted the Fed founders’ intent that the Fed would use a penalty discount rate as its primary tool of managing the cyclical and seasonal availability of credit in the money market. More broadly, the World War I precedent of making the Fed subservient to the interests of marketing treasury debt not only produced the short-term inflationary binge of 1917–20 (Meltzer, 2003, pp. 90–107), it also set the stage for subsequent changes that eventually made the Federal Reserve a fiscal instrument of the U.S. Treasury. Those changes include the 1932 Glass-Steagall Act, a temporary measure later made permanent, which permitted the use of treasury securities as collateral for Federal Reserve note issues (Meltzer, 2003, pp. 358, 417–18), and the various changes in treasury monetary powers (discussed in the next section) after 1933, which gave the treasury effective control over monetary policy, which it used to target the yields on government debts until the Treasury-Fed Accord of 1951.

III. Initial Conditions As the Fed began operations in 1914, consider what it had to go on. First, there was no established view of the proper approach to monetary policy, of the precise mission of the Fed, or of how the Fed might accomplish its illdefined mission. If the point of having a Fed was to increase the elasticity of credit, as distinct from the elasticity of the narrowly defined currency supply (which was the clear purpose of the Aldrich-Vreeland Act of 1908), how would the Fed do so? Was the Fed supposed to focus mainly on seasonal or cyclical elasticity? What sorts of market indicators were likely to give the most reliable signals of loose or tight conditions in the money and credit markets? What were the most effective tools to use to react to those signals, and how much reaction was called for under what circumstances? How would the Fed reconcile its seasonal or cyclical policy objectives with its adherence to the gold standard? Most of these questions had been the subjects of bitter debates leading up to the founding of the Fed, and the debates were not resolved prior to the Fed’s founding. There was no monetary authority track record from the past to guide the new central bank. Furthermore, potentially relevant data were absent,

174

Charles W. Calomiris

and even some of the simplest conceptual ideas that would be taken for granted today about how to define interesting concepts in order to collect data about them had yet to evolve (e.g., monetary aggregates, according to Friedman & Schwartz, 1963, p. 628, are first mentioned by the Federal Reserve in 1948 and are not discussed seriously as a potential criterion for targeting monetary policy until 1952).6 The founding of the Fed was also motivated by problems that were recognized at the time as unique to the United States’ financial system; despite the many volumes of excellent studies of other countries’ experiences produced by the National Monetary Commission in 1910, the unique structure of the U.S. financial system (a geographically fragmented unit banking system serving the needs of a vast continent) substantially limited the ability of U.S. policymakers to learn from, say, the Bank of England or any other preexisting central banks. Thus, the goals, relevant monetary concepts, data, and understanding of policy instruments of monetary policy had to evolve “in real time” alongside policy decisions. U.S. monetary policymakers had no choice but to improvise in the meantime. Improvising meant relying on their preexisting beliefs, and based on those beliefs, arguing with one another about the appropriate goals and indicators on which to focus, and on the appropriate reactions to market indicators. The dominant monetary policy doctrine at the Fed from the beginning was the “real bills doctrine” – a view of monetary policy that saw trade credit as the desirable focus of commercial bank lending and that viewed the main role of monetary policy as accommodating shifts in the demand for trade credit. Because trade credit was perceived as closely linked to the real needs of business, and in that respect, distinct from the speculative whims of stock market or real estate speculation, changes in the demand for trade credit were perceived as a more reliable indicator of legitimate shifts in credit demand that were worth accommodating. By 1923, the Fed had developed a coherent toolkit to implement the real bills doctrine, which was described in its tenth annual report and which Meltzer (2003, pp. 161–5) refers to as the Riefler-Burgess doctrine. According to that view, the Fed should respond to shifts in the domestic demand for credit by easing through a combination of open market purchases and discount rate reductions (typically, implemented together, with open market operations leading subsequent discount window changes) when borrowed reserves were high and nominal interest rates were high. If it followed the real bills doctrine, the Fed believed that increases in the money supply would be “self-liquidating” and therefore would pose no risk of fueling speculative bubbles or inflation, nor any threat to the

Volatile Times and Persistent Conceptual Errors

175

maintenance of the gold standard, so long as the Fed also responded to international shocks in the gold market by tightening credit conditions in response to outflows of gold and loosening in response to gold inflows (the so-called “rules of the game” of the gold standard).7 The real bills doctrine was an appealing ideology because it offered policymakers a rule that promised to provide short-term flexibility (or “elasticity” of currency supply) without disrupting the long-term discipline of adherence to gold or helping to perpetuate asset pricing bubbles.8 Wicker (1966), Brunner and Meltzer (1964, 1968a), Wheelock (1990, 1991, 1992), and Meltzer (2003, Chapter 5) show that – with a few exceptions – the Fed consistently adhered to that real bills approach in the 1920s and the 1930s. In particular, contrary to Fisher’s (1935)9 and Friedman and Schwartz’s (1963) view that Benjamin Strong’s death caused the Fed to adopt a new and unwise monetary policy targeting regime after 1928, subsequent econometric work by Wheelock (1900, 1991, 1992) has shown that the same policy reaction function describes Fed actions consistently in the 1920s and 1930s. An analysis of Fed officials’ statements by Meltzer (2003) confirms that Fed officials intended to do so. Under Strong’s leadership, the Fed had followed that reaction function, and Strong’s less powerful but supposedly enlightened successor at the New York Fed, George Harrison, tended to advocate expansion or contraction in a manner fully consistent with real bills thinking (Meltzer, 2003, p. 274). The supposed exceptions to real bills targeting – Strong’s successful advocacy of expansionary policy in response to international events in 1924 and 1927 – were not really exceptions, because both occurred at times of high borrowed reserves (when real bills targeting agreed that expansionary policy was called for). Furthermore, as Meltzer (2003, p. 274) shows, the monetary expansion in 1927 was viewed in 1928 by most of the Fed leadership as having been too expansionary because it had occurred without commensurate growth in real bills; Meltzer argues that even if Strong had lived, he likely would have overseen a tightening of policy in reaction to what was widely perceived in 1928 as an error. Finally, Meltzer (2003, pp. 289–323, 409–10) shows that there is little to the view that there were deep divisions within the Fed or paralysis in the aftermath of Strong’s untimely death; most of the disagreements that occurred about policy in 1929 were procedural rather than substantive. The key error of seeing monetary policy as appropriately “easy” despite the worsening contraction in 1930–2 reflected a common misperception by virtually all Fed officials, which reflected their adherence to the Riefler-Burgess doctrine and their failure to distinguish between nominal and real interest rates.

176

Charles W. Calomiris

In the wake of the failure of the Bank of United States in December 1930, and at other times of high bank failure, there was little impetus within the Fed to offset the resulting declines in the supplies of money and credit. As banking system risk increased and banks responded to the risk with higher reserve demand, Fed officials, including Harrison, followed the RieflerBurgess doctrine and interpreted rising excess reserves and low borrowings as indicating a lack of demand for money (Meltzer, 2003, pp. 326–34). As banking failures increased in 1930 and subsequently, Fed officials were aware of failures, but did not see monetary policy as an instrument to be used to prevent bank or commercial failures, which were part of the natural process of maintaining market discipline. Although Fed officials differed from time to time on the precise levels of purchases or sales of securities, they generally agreed about the direction of policy. Real bills thinking, as depicted in Figure 4.1, underlay the primary errors of the Fed in the 1930s: (1) ignoring supply-side shocks in the loan market (which also appeared in the 1930s as shocks to the demand for reserves) and (2) failing to distinguish between nominal and real interest rates. With respect to the first of these errors, the Fed failed to consider the shift in the demand for reserves by banks in response to recession-induced losses. As banks experienced losses in their loan portfolios and equity capital-toassets ratio resulting from economic decline, and facing the discipline of the deposit market, they had to either restore lost capital (with either new offerings or increased retentions) or reduce asset risk by increasing the ratio of cash assets relative to loans (Calomiris & Wilson, 2004, model that choice). Virtually no banks chose to raise new capital during the 1930s. They chose instead to substantially reduce the ratio of loans to cash assets and also to cut dividends to boost capital. This inward shift in loan supply is depicted in Figure 4.2. As banks were cutting lending, the effect on real interest rates in the loan market was positive. Of course, the recession also caused declines in loan demand, which somewhat offset that effect. As Meltzer (2003) repeatedly emphasizes, however, the Fed did not focus on the real interest rate, which was rising during the early 1930s, in response to increasing deflation; instead, it took the low nominal interest rates in the market as an indication of declining interest rates, which it saw as an indication of a plentiful supply of loans. The result was a sharp contraction in money and credit, which was caused by a contraction in loan supply, which substantially aggravated the macroeconomic consequences of deflation (Bernanke, 1983; Calomiris & Mason, 2003b; Calomiris & Wilson, 2004). According to Calomiris and Mason (2003b), after controlling for the effects of the general decline in the

Volatile Times and Persistent Conceptual Errors

177

Loan Supply (1932) Real Loan Interest Rate

Loan Supply (1931) Loan Supply (1929)

Loan Demand (1929) Loan Demand (1931) Loan Demand (1932)

Real Quantity of Loans

Figure 4.2. Loan supply contraction and rising real loan rates during the Depression.

supply of money nationally, a contraction in loan supply of 1 percent (in the state-level cross section) resulted in an incremental decline of roughly half a percent in income during the early 1930s. Thus, the combination of adherence to the real bills and Riefler-Burgess doctrines and interest rate money illusion were fundamental conceptual errors that permitted the Fed to oversee the sharp contraction in the supplies of money and credit during the Depression. The real bills doctrine and the focus on borrowed reserves and nominal interest rates as reliable signals of market tightness and looseness to which the Fed should respond have no adherents today.10 It is now widely recognized that following the Riefler-Burgess framework will tend to aggravate business cycles by expanding the money supply at times of exogenous expansion in aggregate demand. It is also understood today that low nominal interest rates and low levels of borrowed reserves are not reliable indicators of loose money. But that does not mean that Fed officials were stupid. They engaged, unavoidably, in a process of ongoing trial-and-error experimentation, beginning with some simple rules of thumb that seemed to offer a means of balancing short-term flexibility with long-term discipline. Over time, retrospectively, the Fed would have to draw inferences about past actions and their consequences and incorporate lessons to guide policy more scientifically. The Fed would have to learn that the real bills doctrine created a procyclical bias in monetary policy (because it was accommodating aggregate demand shocks with increases in money supply). It would have to learn that decreases in borrowed reserves (at a time when banks were

178

Charles W. Calomiris

scrambling for cash because of system-wide increases in banking risk) could signal tightening credit conditions (as in the early 1930s) or that low nominal interest rates (again, in the early 1930s) could be consistent with high borrowing costs and tight credit conditions (in a deflationary environment). One might have expected that this process would take years.

Why Learning Was So Slow That expectation would have turned out to be wrong. Despite the huge costs of these conceptual errors in the 1930s, learning did not take years, but rather, several decades (i.e., these lessons were generally, if not universally, understood only by the 1960s). To economists, reading this chapter in 2012, perhaps the most remarkable single fact to note about monetary policy at the end of the interwar period is that its architects were, for the most part, quite pleased with themselves. Far from learning about the errors of their ways during the interwar period, Fed officials congratulated themselves on having adhered to appropriate principles, and to the extent that they were self-critical, it was because they thought that they had been too expansionary (Meltzer, 2003, 410–13). In part, Fed officials absolved themselves of blame because they believed that monetary policy was of limited use in combating recessions. Monetary policy could exert proper influence only through its effect on bank lending, but Fed officials (who saw low borrowed reserves, high excess reserves, and low nominal interest rates as indicators of loose credit) believed that the economy was in a liquidity trap in the early 1930s. As one official put it, monetary policy did not stop the Depression because “you must have borrowers who are willing and able to borrow” (Meltzer, 2003, p. 478). The Fed had not expanded its open market operations because it did not see the point of doing so; more liquidity would not have made a difference. As we subsequently see, that view was mistaken and reflected two errors: (1) the Riefler-Burgess doctrine’s approach to measuring the tightness of credit markets and (2) interpreting rising reserve ratios in the 1930s as indicative of a passive willingness to accumulate reserves by banks, rather than as an active scramble for liquidity. With some significant exceptions, central bankers from outside the United States and independent monetary experts – including the famed banking crisis historian Oliver Sprague, who authored (Sprague, 1910) one of the most influential reviews of the pre-Fed experience with banking panics as part of the work of the National Monetary Commission – shared the view that the Depression had not been caused by monetary policy

Volatile Times and Persistent Conceptual Errors

179

errors (Meltzer, 2003, pp. 277–82).11 Congressional leaders and President Roosevelt (who asked the preeminent advocate of the real bills doctrine in Congress, Senator Carter Glass, to craft regulatory reforms in 1933) seem to have agreed, because the regulatory changes affecting monetary policy during the 1930s were designed in large part to assist Fed monetary policy by making the real bills doctrine work better, based on the view that bank lending for non–real bills purposes had been a contributor to excessive financial expansion (Meltzer, 2003, pp. 429–34; Calomiris, 2010). This lack of learning had consequences; once the Fed had restored itself to a position of substantial independence from the U.S. Treasury after the 1951 Accord, it followed a policy reaction function that was remarkably similar to what it had done in the 1920s and early 1930s (Friedman & Schwartz, 1963, pp. 614–32; Brunner & Meltzer 1964, 1968a), and consequently, the Fed made similar “procyclical” errors to those it had made before. What explains this lack of learning during the interwar period? It is difficult to answer such a question convincingly, but one thing is for sure: The period 1914–47 was an unusually unstable three decades in U.S. history. The sequence of large and unique shocks, along with the changing economic structure of the economy, was not conducive to learning about monetary policy; the possibility of meaningful central bank learning about the dynamic structures of the financial and real sectors and their interlinkages requires a minimum of stability in the basic economic processes that form the backdrop for monetary policy. As Wicker (1966) emphasized, the uniqueness of the events of the interwar period made it difficult for policymakers to react to them properly, especially given the inadequacies and procyclical bias of the policy doctrines the Fed started with. Not only was the economy buffeted by large and unique shocks; its structure was also changing over time. No sooner had the Fed been founded, but World War I began. The Fed had been constructed to operate under the international adherence to the gold standard, but World War I saw a collapse of that gold standard. Domestically, World War I saw unprecedented increase in government control of the economy (Rockoff, 2008), making it difficult for the nascent central bank to learn about the normal dynamic structure of the financial system and its interactions with the real economy. The 1920s were also an era of significant structural change and large shocks. The 1920s saw a dramatic mixture of divergent regional and sectoral trends within the United States. Some agricultural areas, for example, those specializing in grains and cotton, were hard hit after World War I by declines in commodity prices. This produced the worst rural mortgage foreclosure outcomes and the highest rural bank failure rates up to that point since the

180

Charles W. Calomiris

1830s. Indeed, the high bank failures of the early 1930s are best seen as a continuation of the rural bank failures that had begun in the 1920s (White, 1984; Calomiris, 1990, 1992; Alston, Grove, & Wheelock, 1994). At the same time, however, other sectors of the economy thrived. Not only did much of industry expand, but its structure was changing dramatically. Industrial production was undergoing something of a revolution. New industries and firms based on new products and new production methods were ushering in a new era of dynamic technological progress, which was reflected in the the rapid appreciation of the stock market in the late 1920s (Nicholas, 2007; Kabiri, 2009). The number of patents was unusually high in the 1920s, but more important, of the 19,948 patents granted to firms between 1920 and 1929, an unusually high percentage of them (21%) are cited in patents granted between 1976 and 2002 (Nicholas, 2007). Field (2003) shows that the progress of the 1920s was so powerful that it continued into the 1930s, despite the inhospitable macroeconomic environment. The period 1929–41 saw the “fastest rate of multifactor productivity growth over the last century and a half, and probably two centuries” (p. 1406), making these years “the most technologically progressive of any comparable period in U.S. economic history” (p. 1399). Mowery and Rosenberg (2000) find that the employment of research scientists and engineers grew by 72.9 percent from 1929 to 1933. Bresnahan and Raff (1991, 1992) examine technological change in the automobile industry and argue that the Depression hastened the “shake out” of relatively backward firms and the consolidation of market share in technological leaders. Regional integration through increased communication and transport linkages was accelerating at a rapid rate (automobiles, radios, and motion pictures became widespread, and aviation began to spread). In 1919, there were roughly 7 million automobiles in the United States; by 1929, there were more than 23 million. Electricity became widespread in the 1920s, even in rural areas. Annual radio sales grew from $60 million in 1922 to $843 million in 1929. Integration of communication and transportation affected the nature of consumers’ behavior in the 1920s, too, in ways that may have altered aggregate consumption dynamics and financial markets. Electrical appliance use became popular, as did consumer credit for durables purchases. Canned and other prepackaged food purchases grew dramatically. National consumer fads and dramatic news coverage of controversial events (e.g., the Scopes trial in 1925), dramatic increases in attendance at major sporting events, the publication of national magazines with large circulations, and the birth of a national advertising industry all reflected fundamental

Volatile Times and Persistent Conceptual Errors

181

changes in the 1920s (Allen, 1931). Brokerage houses developed the first true network for retail stock investing in the mid-1920s (Ferderer, 2007), and consumers’ investments in stock became important to an unprecedented degree. Stock prices, especially for high-growth firms and for the large New York banks, grew dramatically during the 1920s and prompted a growing controversy over whether a bubble had begun in these markets and whether and how the central bank should respond to it (more on this in the following discussion). Banking underwent dramatic changes in the 1920s, too. Rural bank failures prompted widespread bank consolidation and a relaxation of branching restrictions. Many states relaxed their branching laws in the 1920s, and the McFadden Act of 1927 allowed national banks to branch in states that permitted state-chartered banks to branch (Calomiris, 2000, Chapter 1). From 1920 to 1930, the number of banks operating branches and the number of branches increased from 530 branching banks with 1,281 branches to 751 branching banks with 3,522 branches (Calomiris, 2000, p. 57). As banks grew in scale, they also grew in scope. U.S. banks entered asset management (trust activities) aggressively during this period. They also participated as underwriters in the growth of the securities markets. And money center banks expanded their operations abroad. The first era of true universal banking in the United States began in the 1920s. Construction underwent changes, too. Cities saw the spread of skyscrapers of ever-expanding ambition. The first real estate boom and bust in Florida, amid an awakening appreciation of the unique opportunities for development there, occurred from roughly 1920 to 1926. The population of Miami (only one of the many new resort destinations) rose from 30,000 in 1920 to 75,000 by 1925. A national housing market boom and bust, ending in 1929, was of similar magnitude to the 2000–7 boom and bust, although the price decline did not have the same impact on the financial sector, owing to the low levels of mortgage leverage in the 1920s (White, 2009; Nicholas & Scherbina, 2010). The 1930s, of course, brought even more volatility and shifts in economic and political circumstances than the 1920s. Government policy was a source of substantial volatility, and government grew substantially in scale and scope (Rockoff, 1998; Wallis and Oates, 1998). Many of those shifts affected the structure of the economy, and some of them took the form of institutional changes in the structure of the monetary system and the organizational structure of the Fed. New Deal industrial policies initially had important microeconomic consequences through the National Industrial Recovery Act’s (NIRA’s) attempts

182

Charles W. Calomiris

to fix prices and wages. Rising unionization and collective bargaining, culminating in the Wagner Act of 1935, altered the behavior of wages, and the macroeconomic consequences of large-scale strikes (like those in the automobile industry in 1937) posed new questions for how monetary policy should react to such shocks. With respect to banking, failure rates in 1930–3 were far higher than those of the 1920s, and because failures were concentrated in smaller banks, they produced a significant change in the size distribution of surviving banks. The 1920s were a turbulent decade for agricultural states. The 5,712 banks that failed during the years 1921–9 had total deposits equal to $1.6 billion at the time of their failure, constituting 3.1 percent of average total deposits in the banking system from 1921–9. Losses to depositors for the period 1921–9 amounted to $565 million, which was 1 percent of average deposits during the period 1921–9 and 0.6 percent of the average annual gross national product (GNP).12 By way of comparison, during the period 1873–1913, no year had seen losses to depositors in excess of 0.1 percent of the GNP. Bank failures accelerated in the early 1930s. The number of banks fell 39 percent from 24,633 in December 1929 to 15,015 in December 1933. The 9,096 banks that failed during the years 1930–3 tended to be small banks. Failed banks, as defined by the Federal Reserve (1943), represented 37 percent of the banks in existence at the end of 1929, but the deposits of those failed banks (at their dates of failure) were only 14 percent of the average level of bank deposits over the years 1930–3, and losses borne by depositors in failed banks were roughly $1.3 billion, representing 2.7 percent of the average amount of deposits in the banking system for the years 1930–3, and 2 percent of average annual GNP for 1930–3. Consolidation was the twin of bank fragility during the 1920s and 1930s. Thousands of banks were absorbed during the 1920s and early 1930s, accounting for a large share of bank assets, and this process was facilitated by regulatory reforms that expanded permissible bank branching (White, 1985; Calomiris, 2000, Chapter 1). The 1933 Banking Act, among other things, brought an end to the 1920s merger wave in banking by explicitly limiting consolidation and by subsidizing small risky banks, who were the primary beneficiaries of the new Federal Deposit Insurance created under the act. The act also set demand deposit interest rates at zero, as a means of trying to shrink the interbank deposit market, and restructured the banking system by separating commercial and investment banking. The Reconstruction Finance Corporation, which had begun its lending operations to banks and other firms in 1932, was reformed in 1933 to permit preferred stock investments in

Volatile Times and Persistent Conceptual Errors

183

banks and other firms, which had important consequences for the banking system (Mason, 2001; Calomiris & Mason, 2004). All of these changes had potentially large and uncertain effects on the structure of the financial system and the riskiness of financial institutions. The macroeconomic rules of the game were also in flux. The international return to the gold standard that had been happening during the 1920s came apart in 1931, as Britain and many other countries, but not the United States, left the gold standard. The Fed was enjoined to maintain a commitment to maintaining the gold standard (prior to 1933), and also facilitating the domestic needs of trade. These sometimes conflicted. Furthermore, effective interventions in international currency markets by central banks can require coordinated actions, which further complicated the Fed’s responsibilities. In 1933, President Roosevelt left the gold standard and ushered in a new era of gold price targeting, which had dramatic consequences, in particular, by encouraging large gold inflows, which were the main component responsible for expanding the monetary base from 1933 to 1936. Major fiscal policy changes, especially the large tax increases in 1936 and 1937, also contributed greatly to macroeconomic volatility (Romer, 1992; Calomiris & Hubbard, 1995; Fishback, 2010). The post-1933 period also changed the structure of monetary policy. Most important, it saw substantial increases in the monetary powers of the U.S. Treasury. This occurred via its control of the exchange rate, the control over the sterilization of gold inflows, the Exchange Stabilization Fund, established in 1934, and silver-related authority, all of which gave the treasury the ability to grow the money supply. Because the treasury wanted to grow the money supply, and because the Fed’s balance sheet was too small to stop that growth, even if it had tried, the Fed had no effective tools to use to oppose treasury expansion (Calomiris & Wheelock, 1998). The period of Fed policy from 1933 to 1940 is best captured by Meltzer’s (2003) chapter title, “in the back seat.” The back seat also was reupholstered. The Banking Act of 1935 changed the structure of the Federal Reserve System and its powers, as subsequently discussed further. No sooner had the Fed begun to adjust to those new rules and structure than mobilization for World War II began, with its sectoral shifts, price controls, and rationing. Monetary policy did not have to adapt to those changes, because it practically ceased to exist, as the Fed became an instrument for pegging low, constant nominal interest rates on national debt, in support of the treasury’s war financing effort. In many respects, the America that we have come to know (a technologically advanced, highly mobile society, with nationally integrated financial,

184

Charles W. Calomiris

product, and factor markets, and eventually, nationwide banks) had its beginnings during the first three decades of the founding of the Fed, but a full understanding of those shifts and their implications for monetary policy could not have been known at the time. Even a central bank equipped with the tools and experience of a 2010 central banker would have found this a challenging environment in which to direct monetary policy. For example, on a forward-looking basis, it would have been difficult to say what rate of monetary growth, or level of interest rates, was likely to produce price stability in this economy. Furthermore, unstable times like World War I, the 1920s, and the 1930s, made it hard to identify the contribution of mistaken Fed doctrines and practices (e.g., “real bills” thinking) or the deleterious effects of the treasury’s monetary policies (e.g., the December 1936 decision to sterilize gold flows, which helped to precipitate the recession of 1937–8, as subsequently discussed). After all, it is difficult to identify monetary policy errors when the effects of other shocks are conflated with monetary policy in their effects on output and employment. In light of those shocks, perhaps it is not surprising that the Fed found it difficult to identify accurately the negative role that it had played in the economy or to derive lessons from those errors about more appropriate targets and instruments of monetary policy. Indeed, monetary historians and economists are still arguing over the nature of monetary policy errors during the interwar period and are trying to disentangle the effects of monetary policy on the economy from a myriad of other influences. Friedman and Schwartz (1963) saw the Depression of 1929–33 and the recession of 1937–8 largely as consequences of Fed errors that could and should have been avoided. The Friedman-Schwartz narrative of the Great Depression has not always been embraced by others. Temin (1976) argued that monetary policy had largely played a passive role during the Depression. Eichengreen (1992) argued that the Fed was constrained prior to 1932 in its ability to expand the money supply by the gold standard and the rules of the Fed’s charter governing its holdings of gold, and others (Bordo et al. 2002; Meltzer, 2003; Hsieh & Romer, 2006) have disputed Eichengreen’s contention. Wicker (1966, 1980, 1996) and Calomiris and Mason (2003a) contested the Friedman and Schwartz view that a national banking panic had occurred in 1930 and early 1931; those alleged panics were a crucial part of the Friedman-Schwartz argument that the Fed had failed to bolster the money supply at a time when it should have foreseen a panicinduced contraction in the money multiplier. Frost (1971), Calomiris and Wheelock (1998), Hanes (2006), and Calomiris et al. (2011) dispute the

Volatile Times and Persistent Conceptual Errors

185

Friedman-Schwartz account of the role of reserve requirement changes in 1936–7 in precipitating the recession of 1937–8. Overall, recent research has reinforced many aspects of the FriedmanSchwartz view that failures of monetary policy were central to the economic troubles of the 1930s, but researchers disagree about the relative importance of various monetary policy shocks and errors Friedman and Schwartz identify, about the extent to which the Fed, as opposed to the treasury, was responsible for monetary policy errors in 1936–7, and about the ideological origins of avoidable mistakes in monetary policy. I review these arguments in more detail in Section IV, which offers a synthesis of the literature on some of these key issues, but I mention them here as a reminder: It has not been easy for monetary economists and historians to identify clearly the nature of monetary policy errors of the interwar period and their effects on the economy, even decades after they allegedly occurred. If it has taken economic historians decades to identify the Fed’s contribution to the macroeconomic performance of the Great Depression, perhaps it is not surprising that the Fed did not learn immediately from its alleged mistakes. Those controversies about the cyclical effects of monetary policy during the interwar period contrast sharply with the literature on the Fed’s great success story of the interwar period: the reduction of seasonal volatility in the financial system (reviewed in Section V). Unlike the literature on the Fed’s contribution to cyclical volatility, the seasonal volatility literature has been remarkably uniform in its conclusions. From its founding, a key part of the Fed’s mission (stated clearly by its founders) was to increase the seasonal elasticity of the supply of reserves, to facilitate the seasonal smoothing of the supply of credit. The Fed was given the tools to do so, understood that mission, and seems to have accomplished it reasonably well (Miron, 1986, Bernstein, Hughson, & Weidenmier, 2010). It is interesting to ask why seasonal smoothing was so much easier than cyclical smoothing, and we shall return to that topic in Section V.

IV. Identifying Policy Errors Is Challenging, Even With the Benefit of Hindsight Here I consider five key questions in the history of monetary policy during the interwar period listed in the Introduction. These questions relate to some of the most egregious alleged errors of the Fed. Our understanding of these errors has evolved over time, and continues to evolve, which is the overarching point of this review. Academics continue to reasonably disagree about the extent to which the Fed can be faulted for the stock market crash

186

Charles W. Calomiris

of 1929, the failure to sufficiently expand the supply of money in 1930–3, the extent to which the banking crises of the 1930s were “panics” (which relates to the question of how egregious the Fed’s failure was to respond to them), the extent to which monetary policy was limited by a liquidity trap in the 1930s, and the extent to which the reserve requirement increases imposed by the Fed in 1936–7 helped precipitate the recession of 1937– 8. The durability of these academic debates itself shows how difficult it was for the Fed to avoid errors in real time, especially given that it was hobbled by conceptual failings (money illusion and real bills thinking). The slow progress toward achieving a consensus on these difficult issues helps explain the protracted process of learning about monetary policy after the Depression.

The Stock Market Boom and Bust of 1928–9 From the beginning of 1927 to October 1929, U.S. stocks more than doubled in value (Board of Governors, 1943). That impressive average performance masked important differences across types of stocks. High-growth stocks grew much faster than average. These consisted of new technology firms like General Motors, RCA, and General Electric, and companies in high-growth sectors, like electricity distributers and money center banks. New York City banks were experiencing a boom in revenues as the result of new services and a growing customer base (trust management and securities underwriting) and the establishment of global networks. As previously discussed, the United States was experiencing an unparalleled technology boom in the 1920s, one in which many new product and process technologies were developed that would have lasting significance (Field, 2003). Nicholas (2007) shows that the number of patents with lasting applications and large numbers of subsequent citations were particularly high during the 1920s. He also shows that the number of patents (weighted by their future citations) is a powerful explanatory variable for the cross section of stock appreciation during the 1927–9 boom. The implied value of patents rose over the last years of the 1920s, reflecting the rising perceived real options associated with new technology. For some sophisticated observers – notably Charles Dice (1929) and Irving Fisher (1930) – productivity growth and other positive long-run trends justified the boom in stock prices. Kabiri (2009) analyzes the valuation models that were used by professional investors during the 1920s. He finds that stock prices are consistent with those models. In essence, the high prices of the 1928–9 boom reflected an expectation of a continuation of the

Volatile Times and Persistent Conceptual Errors

187

revenue growth of the recent past into the near future. Those expectations also explain the rising leverage that was tolerated by bond market investors during the 1920s (Calomiris, 1993). Stock valuations appear not to have been driven by amateur investors; rather, they reflected the consensus among professional investors that growth in earnings for high-growth firms would persist into the future. Calomiris (2012) examines the composition of stockholders of Citibank (then First National City Bank) from 1925 to 1931. Large stockholders, especially sophisticated investors who were also bank insiders with very large preexisting stakes, increased their holdings over time; virtually no insiders sold during the boom, and all held on through November 1931. But for other observers, the rapid rise in prices was seen as an unsustainable and destabilizing frenzy. Some recent academic studies have also agreed with that perspective, emphasizing the new presence of retail investors in the market, the importance of buying on credit, the high markups on closed end mutual funds, and the ex ante perceptions of increasing risk as stock prices rose implicit in securities loan interest rates (DeLong & Shleifer, 1991; White, 1990: Rappaport & White, 1993, 1994). For real bills adherents, like Senator Glass and many Fed officials, the major concern was keeping the banking system from becoming entangled in financing the stock market boom. The Fed’s actions began as an initiative to prevent bank involvement in the call loan market, which real bills adherents believed would fuel the bubble and leave the money center banks weakened. These beliefs led the Fed to impose limits on member bank lending to the securities market, beginning in 1928 (White, 1990; Rappaport & White, 1993, 1994). This policy seems to have had little effect either on the amount of credit available for the stock market or on stock prices. Real bills advocates pressed for additional policies to deflate what they saw as a bubble. Ultimately, partly in reaction to the continuing surge in stock prices, the Fed pursued contractionary monetary policy in 1929, which precipitated the recession. The recession dashed expectations of near-term growth and resulted in a severe drop in stock prices. However, in 1930, in expectation of a normal economic recovery, stock prices recovered much of the ground they had lost, until the deepening recession caused them to fall once again. Whether the Fed was prescient about the risks of the stock market or, alternatively, the source of an unnecessary collapse of the stock market depends on how much weight one attaches to the likelihood of an optimistic growth scenario in the absence of Fed action. For those who equate run-ups in securities prices with bubbles, the Fed was obviously justified, but for

188

Charles W. Calomiris

those who entertain the possibility that stock market booms may reflect fundamentals, there is a lot of evidence consistent with that point of view from the 1920s. Unsurprisingly, the lesson Carter Glass learned was that he had been right all along about the market. That point of view underlay his aggressive and successful actions to limit connections between commercial banking and the securities markets in 1933. The 1933 act gave the Fed powerful tools to limit bank involvement in securities lending (Meltzer, 2003, p. 434). It also separated investment banking and commercial banking and limited interest payments on checking accounts (a convenient means of undermining the pyramiding of reserves by eliminating the interest payments on interbank deposits). Subsequent academic research has seen these as unnecessary and unwise actions, but it took many decades for these regulations to be undone (see Calomiris, 2010, for a review).

Did the Gold Standard Constrain Monetary Expansion During the Depression? Some of the thorniest issues of interpretation about the Fed’s actions and intentions during the Great Depression relate to the question of whether the maintenance of the gold standard prevented monetary expansion, especially during the period October 1931–January 1932. Eichengreen (1992) is the primary advocate in favor of the proposition that a lack of “free gold” constrained policy, although support for that point of view can also be found in Temin (1989) and Bernanke (1995). According to Eichengreen (1992), only a coordination of actions by the world’s central banks could have freed the Fed from the concerns that inhibited its monetary expansion. Friedman and Schwartz (1963) had taken the opposite view, arguing that the Fed could have expanded the money supply without any risk to its ability to maintain the gold standard. When the British left the gold standard in September 1931, the United States began to experience a significant drain of gold reserves. By law, the Fed had to maintain the gold standard (the convertibility of its notes into gold), and it also faced statutory requirements regarding the minimum gold reserves it had to maintain against outstanding Federal Reserve notes.13 Eichengreen (1992) corroborates the statements of Fed officials (see also Meltzer, 2003, pp. 349–56), who had expressed concern about these risks, both during the 1931–2 period, and subsequently. Eichengreen argues that Fed officials were rightly concerned that an expansion of the money supply in the presence of a reserve outflow would have put the Fed at risk of running

Volatile Times and Persistent Conceptual Errors

189

out of “free gold” (gold in excess of its 40 percent reserve requirement against notes) and could have prompted a run on the dollar. Wicker (1966) and Meltzer (2003, pp. 274–6, 347–57, 404–7) review the claims that have been made on both sides of this issue and document the thinking of the Fed decision makers at the time. They conclude that the Fed officials often felt constrained by the gold standard, but they side with Friedman and Schwartz on the question of whether it would have been possible for the Fed to have expanded the money supply during October 1931–January 1932 without being concerned that doing so would have been difficult or risky. Bordo et al. (2002) argue that the United States was a large country with ample gold reserves and that it had substantial latitude to expand the money supply without departing from the gold standard. They argue that the United States was not constrained from using expansionary policy to offset banking panics, deflation, and declining economic activity. They perform simulations, based on a model of a large open economy, which indicate that expansionary open market operations by the Federal Reserve at what they regard as two critical junctures (October 1930 to February 1931 and September 1931 through January 1932) would have been successful in avoiding economic contraction without endangering convertibility. That is also the point of view of Hsieh and Romer (2006), who perform both statistical tests of the effects of monetary expansion on exchange rate devaluation risk, and review market perceptions of that risk, as indicated by market prices, as well as statements by market participants. Hsieh and Romer (2006) show that the expansion of the money supply that occurred after the relaxation of reserve requirements (in February 1932, under the 1932 Glass-Steagall Act) created no risk of devaluation, and they argue that this shows that monetary expansion would not have put the gold standard at risk in late 1931. There are two separate questions that are both worth addressing, and recognizing them as separate is useful. (1) Did Fed officials hold back in monetary expansion during October 1931–January 1932 that they otherwise would have wanted to undertake because they believed that they were constrained by the gold standard? (2) Irrespective of the beliefs of Fed officials, would it have been possible for the Fed to expand substantially more without threatening the gold standard? With respect to the intentions and beliefs of Fed officials, it is pretty clear that they were concerned about the gold drain after the British departure from gold and that this played an important role in delaying expansion during October 1931–January 1932: “Did the free gold problem delay open market purchases? The answer is certainly yes” (Meltzer 2003, p. 357).

190

Charles W. Calomiris

But Meltzer also shows that Harrison delayed monetary expansion during this period for several reasons, and that several other governors opposed purchases for reasons unrelated to gold. One of those reasons was a realbills-doctrine version of the “liquidity trap” (subsequently described); some thought that monetary expansion would have little effect for that reason (Meltzer 2003, p. 350). Also, many Fed officials believed that the speculative excesses that had produced the Depression had to run their course and should not be combated with expansion of credit (Meltzer, 2003, p. 405). Contrary to Eichengreen (1992), Meltzer argues that the failure of international coordination was not the binding constraint on Fed actions. International coordination was feasible when the Fed saw it as desirable (Meltzer, 2003, p. 344). Meltzer quotes Oliver Sprague, who said in a speech in May 1931 that the lack of monetary expansion by the central banks was not the result of “any difficulty in securing agreement among the three banks (France, the UK, the United States), but because none of them harbored the belief that [monetary expansion] was the appropriate remedy” (Meltzer, 2003, p. 278). Sprague himself agreed with that “liquidationist” position. Furthermore, the willingness of the Fed to expand after February 1932 (an aggressive but brief expansion that unfortunately came to a halt in June 1932) reflected not only the passage of the 1932 Glass-Steagall Act and the calming of international markets, but also the signals being provided by the real bills indicators: The action was consistent with the Riefler-Burgess framework. Member bank borrowing and short-term rates had not declined. Borrowing was well above the $500 million range considered high in an ordinary recession and was almost back to the 1929 peak. A program to reduce the volume of borrowing by undertaking purchases was consistent with the dominant view that credit markets could be eased by forcing a reduction in the System’s portfolio of real bills (Meltzer, 2003, p. 359).

The fact that the Fed chose to expand the money supply after the passage of the February 1932 Glass-Steagall Act also provides only weak support for the view that the gold reserve was the key limit that prevented policymakers from acting earlier. The 1932 act did more than just relax the gold reserve requirement; it allowed member bank borrowing at 1 percent above the discount rate against previously ineligible commercial paper, and it permitted groups of banks (e.g., clearinghouses) to borrow collectively from the Fed on the credit of the group (Meltzer, 2003, p. 358). In summary, Fed officials clearly worried about the gold drain of late 1931, and it clearly had a negative effect on their willingness to expand the money supply. After the February 1932 Act, they were willing to expand aggressively. It is clear that free gold and concerns about a potential run on

Volatile Times and Persistent Conceptual Errors

191

the dollar played an important role in Fed thinking, but not clear precisely how much weight to attach to that factor. Finally, it is important to remember that the period in question during which free gold concerns mattered as a constraint in the minds of Fed officials was brief (less than six months), and the subsequent period of expansion of the money supply after the passage of the February 1932 act was also brief (February–June 1932). Free gold was a small issue in the grand scheme of the errors of the Fed during the Depression and has nothing to say about the failures of Fed policy prior to September 1931 or from June 1932 to March 1933.14 With respect to the second question – whether, in fact, the Fed was constrained by its gold reserve requirement on notes and by the risk of an attack on the dollar – there are two aspects to that question (the importance of the physical reserve requirement on notes and the potential market reaction), and it is important to consider them separately. The primary risk relating to the gold standard did not revolve around the 40 percent gold reserve requirement per se. The possibility that the Fed would be unable to support the dollar because it would keep its gold locked in a vault (as backing for its notes) was remote. Gold inflows in 1930 and early 1931 had been large (raising the gold stock at the Fed 15 percent above its August 1929 value), implying substantial free gold (Meltzer, 2003, p. 275). Even after the large $750 million outflow of gold in September and October 1931, there was about $2 billion in surplus gold and gold certificates that the Fed could access, if necessary (Meltzer, 2003, p. 276). Harrison himself dismissed the importance of free gold, per se (Meltzer, 2003, p. 345). Furthermore, as Meltzer (2003, p. 356) points out, the Fed enjoyed substantial latitude with respect to its requirement. It could fall below the 40 percent requirement with little cost. A shortfall of reserves would have entailed a small tax payment by the Fed, and even that could have been mitigated by canceling Federal Reserve notes that were held by the Fed itself, an action that would have freed substantial gold – Meltzer (2003, p. 356). And, of course, it was common for governments to suspend gold reserve requirements on central banks during crises; the British had suspended the Peel Act of 1844 during all three of Britain’s post-1844 banking crises (in 1847, 1857, and 1866). The more difficult question is whether aggressive monetary expansion in October 1931–January 1932 might have produced an attack on the dollar. This seems to have been more central to the concerns of policymakers. Charles Hamlin, a Federal Reserve Board member, who expressed confidence that free gold, per se, was not a concern, was nonetheless concerned about the escalating and uncertain risks that the Federal Reserve faced in the foreign exchange markets in late 1931:

192

Charles W. Calomiris

The experience of recent weeks brings home to Federal Reserve officials their heavy responsibility, the necessity for keeping their powder dry, so that in these troublous times they may remain the rock that can withstand all storms and upon which world confidence may once more be reconstructed (Meltzer, 2003, p. 276).

High interest rates and tight monetary policy were seen as confidence builders, not just by Fed officials (like Harrison), but also by their counterparts abroad. Governor Clement Moret of the Bank of France encouraged Governor Harrison to raise rates in October 1931 for precisely that reason (Meltzer, 2003, pp. 344–5). Hsieh and Romer (2006) argue that the lack of any market reaction (increased devaluation risk) to the monetary expansion of 1932 offers evidence against the view that an expansion of the money supply in late 1931 would have produced a run on the dollar. However, one could argue that they overstate the usefulness of their evidence. The market environment of late 1931 was quite different from the one of early 1932. Monetary expansion in an environment that is fearful of collapse and looking for signals of commitment from monetary authorities likely would have punished a move to expand the money supply much more than the market reacted to similar measures in 1932. That is not to say that Hsieh and Romer are wrong in their conclusion, just that it is very hard to have confidence about the counterfactual, as applied to October 1931–January 1932. In summary, the beliefs and actions of Fed officials regarding the gold standard were not obviously wrong in late 1931. Fed officials did not regard the physical gold requirement as much of a barrier to expansion. They were, however, justifiably worried in late 1931 about a potential run on the dollar and also were loath to loosen out of a misguided philosophical belief in the benefits of allowing the speculative excesses of the Depression to run their course. Worries that led the Fed to project dollar strength in late 1931 receded by early 1932, partly because of several changes introduced in the 1932 Glass-Steagall Act, and partly because of changes in the market’s perception of devaluation risk. The reason the Fed was eager to pursue a path of expansion, however, from February to June 1932 was that doing so was fully consistent with the policy rules of the Riefler-Burgess doctrine.

Were the Friedman-Schwartz “Panics” Really Panics? Friedman and Schwartz (1963) argued that from an early date (December 1930) many bank failures resulted from unwarranted “panic” that occurred in several waves and that failing banks were in large measure illiquid rather than insolvent. Friedman and Schwartz’s emphasis on contagion posited

Volatile Times and Persistent Conceptual Errors

193

that bank failures, early as well as late in the Depression, mainly reflected a problem of illiquidity rather than insolvency. Illiquid but solvent financial institutions, in their view, failed as the result of withdrawal demands by depositors, particularly during sudden moments of panic. In contrast, an insolvent institution fails to repay depositors as the result of fundamental losses in asset value, rather than the suddenness of depositor withdrawals. If this account of the banking distress of December 1930–March 1933 is correct, then the failure of the Fed was egregious indeed. In the FriedmanSchwartz view, the Fed not only followed a misguided targeting rule, it failed to recognize and try to address widespread banking panics – the very problem that it was created to prevent. And it failed to do so as early as the end of 1930 and the first half of 1931. That timing is important, because it implies that much of the worst of the Depression might have been avoided if early banking system liquidity risks had been addressed. Friedman and Schwartz attach great importance to the banking crisis of late 1930, which they attribute to a “contagion of fear” after the failure of a large New York bank, the Bank of United States, which they regard as itself a victim of panic. They also identify two other banking crises in 1931 – from March to August 1931 and from Britain’s departure from the gold standard (September 21, 1931) through the end of the year. The fourth and final banking crisis they identify occurred at the end of 1932 and the beginning of 1933, culminating in the nationwide suspension of banks in March. The 1933 crisis and suspension was the beginning of the end of the Depression, but the 1930 and 1931 crises (because they did not result in suspension) were, in Friedman and Schwartz’s judgment, important sources of shock to the real economy that turned a recession in 1929 into the Great Depression of 1929–1933. The Friedman and Schwartz argument that these episodes of banking distress reflected liquidity rather than insolvency problems is based on the suddenness of banking distress and the absence of collapses in relevant macroeconomic time series prior to those banking crises (see Charts 27–30 in Friedman and Schwartz, 1963, p. 309). But there are reasons to question Friedman and Schwartz’s view of the exogenous origins of the banking crises of the Depression. As Temin (1976) and others have noted, the bank failures during the Depression mainly marked a continuation of the severe banking distress that had gripped agricultural regions throughout the 1920s. Of the nearly 15,000 bank disappearances between 1920 and 1933, roughly half predate 1930. And massive numbers of bank failures occurred during the Depression era outside the crisis windows identified by Friedman and Schwartz (notably,

194

Charles W. Calomiris

in 1932). Wicker (1996, p. 1) estimates that “[b]etween 1930 and 1932 of the more than 5,000 banks that closed only 38 percent suspended during the first three banking crisis episodes.” Recent studies of the condition of the Bank of United States indicate that it too may have been insolvent, not just illiquid, in December 1930 (Lucia, 1985; Wicker, 1996). Banks that considered merging with it determined at the last minute not to do so (Meltzer, 2003, pp. 323–4). So there is some prima facie evidence that the banking distress of the Depression era was more than a problem of panic-inspired depositor flight. Friedman and Schwartz omitted important aggregate measures of the state of the economy relevant for bank solvency, for example, commercial distress and construction activity. Second, aggregation of fundamentals masks important sectoral, local, and regional shocks that buffeted banks with particular credit or market risks. The empirical relevance of these factors has been demonstrated in the work of Wicker (1980, 1996) and Calomiris and Mason (1997, 2003a). Using a narrative approach similar to that of Friedman and Schwartz, but relying on data disaggregated at the level of Federal Reserve districts, Wicker (1996) argues that it is incorrect to identify the banking crisis of 1930 and the first banking crisis of 1931 as national panics comparable to those of the pre-Fed era. According to Wicker, the proper way to understand the process of banking failure during the Depression is to disaggregate, both by region and by bank, because heterogeneity was very important in determining the incidence of bank failures.15 Microeconomic studies of banking distress provide useful evidence on the reactions of individual banks to economic distress. White (1984) shows that bank failures in 1930 were a continuation of the agricultural distress of the 1920s and are traceable to fundamental disturbances in agricultural markets. Declines in railroad bonds were also significant in some cases (Meltzer, 2003, p. 346). Calomiris and Mason (1997) study the Chicago banking panic of June 1932 (a locally isolated phenomenon). They find that the panic resulted in a temporary contraction of deposits that affected both solvent and insolvent banks. Fundamentals, however, determined which banks survived. Apparently, no solvent banks failed during that panic. Banks that failed during the panic were observably weaker ex ante, judging from their balance sheet and income statements, and from the default risk premia they paid on their debts. Furthermore, the rate of deposit contraction was not identical across banks; deposits declined more in failing banks than in surviving banks.

Volatile Times and Persistent Conceptual Errors

195

Calomiris and Wilson (2004) study the behavior of New York City banks during the interwar period, and, in particular, they analyze the contraction of their lending during the 1930s. They find that banking distress was an informed market response to observable weaknesses in particular banks, traceable to ex ante bank characteristics. It resulted in bank balance sheet contraction, but this varied greatly across banks; banks with higher default risk were disciplined more by the market (i.e., experienced greater deposit withdrawals), which encouraged them to target a low risk of default. Calomiris and Mason (2003a) construct a survival duration model of Fed member banks throughout the country from 1929 to 1933. This model combines aggregate data at the national, state, and county levels with bankspecific data on balance sheets and income statements to identify the key contributors to bank failure risk and to gauge the relative importance of fundamentals and panics as explanations of bank failure. Calomiris and Mason find that a fundamentals-based model can explain most of the failure experience of banks in the United States prior to 1933.16 They identify a significant, but small, national panic effect around September of 1931 and some isolated regional effects that may have been panics, but prior to 1933, panics were not important contributors to bank failures nationally. The fact that a consistent model of fundamentals can explain the vast majority of bank failures prior to 1933 has important implications. First, the influence of banking panics as an independent source of shock to the economy was not important early in the Depression. Only in 1933, at the trough of the Depression, did failure risk become importantly delinked from local, regional, and national economic conditions and from fundamentals relating to individual bank structure and performance. Second, the timing of this observed rise in risk unrelated to indicators of credit risk is itself interesting. In late 1932 and early 1933, currency risk became increasingly important; depositors had reason to fear that President Roosevelt would leave the gold standard, which gave them a special reason to want to convert their deposits into (high-valued) dollars before devaluation of the dollar (Wigmore, 1987). As part of their bank-level analysis of survival duration, Calomiris and Mason (2003a) also consider whether, outside the windows of “panics” identified by Friedman and Schwartz, the occurrence of bank failures in close proximity to a bank affects the probability of survival of the bank, after taking into account the various fundamental determinants of failure. Calomiris and Mason consider this measure of “contagious failure” an upper bound, because in part it measures unobserved cross-sectional heterogeneity common to banks located in the same area, in addition to true contagion. They

196

Charles W. Calomiris

find small, but statistically significant, effects associated with this measure. The omission of this variable from the analysis raises forecasted survival duration by an average of 0.2 percent. They also consider other regional dummy variables associated with Wicker’s (1996) instances of identified regional panics and again find effects on bank failure risk that are small in national importance. The large number of bank failures in the United States during the Great Depression, a phenomenon that was largely confined to small banks, primarily reflected the combination of extremely large fundamental macroeconomic shocks and the vulnerable nature of the country’s unit banking system. Panic was not a significant contributor to banking distress on a nationwide basis until near the trough of the Depression, at the end of 1932. For these reasons, the Great Depression bank failure experience has more in common with the bank failures of the 1920s than with the panics of the pre–World War I era. It is probably not correct to argue, then, that the Fed failed to detect avoidable national liquidity crises and prevent waves of bank failures in 1930 and 1931. The Fed properly did not see its role as bailing out failed banks, and thus it is not surprising that it allowed failed banks to be closed. That is not to say that there were no bank liquidity problems or panic episodes early in the Depression; there were various local panics at different times during the period 1929–32 (Florida in 1929, various parts of the South in 1930, Chicago in mid-1932, to name a few), and some were associated with significant bank distress. Furthermore, it would be wrong to presume that the Fed did all it could do address those local problems. Although there are inherent limits to what the Fed could accomplish with liquidity assistance through collateralized lending, there is evidence suggesting that more could have been done.17 For example, the Atlanta Fed was more active than the St. Louis Fed and seems to have been able to prevent liquidity problems from spreading, in Florida in 1929 and in the areas of Mississippi under its control in 1930, whereas the St. Louis Fed’s conservatism may have led to unnecessary bank failures (Richardson & Troost, 2006; Carlson, Mitchener, & Richardson, 2010).18 Most important, even if the early phases of the Depression lacked nationwide banking panics, that does not mean that the banking distress of 1930–1 should have been ignored by the Fed. The Fed should have recognized that the contraction in deposits and increase in the banks’ demands for reserves were causing the money multiplier to shrink. The Fed should have responded to this contraction in the supplies of money and credit with expansionary open market operations and discount window lending. In that sense,

Volatile Times and Persistent Conceptual Errors

197

Friedman and Schwartz are correct to criticize the Fed for its failure to respond. And if the Fed had responded aggressively to bank failures and regional panics early in the Depression, it likely would have prevented the economic and banking collapse that came later. That failure of the Fed did not reflect its inability to perceive financial panic, but rather, its adherence to the Riefler-Burgess doctrine, which prevented the Fed from detecting or responding to a contraction in banks’ supply of loans.

Was the Economy Stuck in a Liquidity Trap in the 1930s? The concept of the “liquidity trap,” while not consistently defined, has been proposed in various forms to explain the persistence of the Great Depression in the United States as the result of the impotence of monetary policy to promote a recovery. Roughly speaking, there are three versions of the liquidity trap concept: (1) the na¨ıve view of the liquidity trap, based on an assumption of static expectations and a model of the economy, (2) a more sophisticated view, and (3) the Fed’s own view of the liquidity trap during the Depression, which was based on a particular version of the real bills doctrine.19 The na¨ıve view is the one most commonly espoused: When short-term interest rates are at zero, any further expansion of the money supply has no effect on interest rates, as money demand passively accommodates all increases in money supply. Thus, expansion in money supply has no effect on economic activity. According to the na¨ıve view, when short-term interest rates are at zero (implying perfect substitutability between money and treasury bills), open market operations are a useless tool. The sophisticated version of the liquidity trap recognizes that there are other assets in the economy, including longer-term treasury bonds and foreign exchange (or gold), and that expectations of inflation are not constant, but will adapt to changes in policy on a forward-looking basis. Under these assumptions, the liquidity trap is a possibility, but only if the monetary authority acts unwisely. If the short-term nominal interest rate is zero and the long-term interest rate is positive and above its nominal floor, then a purchase of longer-term bonds can provide monetary stimulus, even before taking into account changes in expected inflation. Furthermore, the central bank can purchase other assets, including foreign exchange, thereby depressing the foreign exchange rate and boosting the demand for exports. Such expansionary policies, especially if done aggressively, would also lead to expected rises in the price level. An expected increase in inflation will lower real interest rates, implying further potency to monetary policy.

198

Charles W. Calomiris

If the central bank is unwilling to purchase foreign exchange, long-term bonds or other assets, and if the central bank is not willing to send a signal to the market of continuing expansion in the money supply, then it may be stuck in a liquidity trap. But if the central bank is willing to do these things, then monetary policy can have potent effects, even when short-term nominal interest rates are at their minimum value of zero. In theory, monetary policy never need be stuck in a liquidity trap, if central banks are willing to expand the supply of money on an ongoing basis (so-called “quantitative easing”) and purchase assets other than just short-term treasury bills. Clearly, during the 1930s, despite episodes when nominal short-term interest rates were at their zero floor, the Fed and the treasury, through various mechanisms, always had the means at their disposal to expand the money supply, increase inflation, and buy a wide range of assets. Monetary policy therefore could have been very effective during the Great Depression. Indeed, Hanes (2006) shows that expansions in the monetary base (e.g., via gold inflows) were effective during the 1930s, in particular, through their effects on long-term interest rates. Basile, Landon-Lane, and Rockoff (2010) show that there was substantial ability for the Fed to affect private sector interest rates, too. The liquidity trap did not constrain the effectiveness of monetary policy. If the Fed and the treasury had expanded the supply of high-powered money through open market purchases at any time during the period, doing so would have reduced long-term interest rates, expanded money and credit, and resulted in increased economic activity and higher prices. Nevertheless, Fed officials believed that policy was stuck in a liquidity trap. Their view of that liquidity trap differed from those of economists, and reflected real bills thinking and a misunderstanding of reserve demand. As Meltzer (2003, p. 336) explains, “Harrison favored delaying further purchases [in 1931], at first because the international monetary system had deteriorated and he believed the timing was poor, later because the banks held excess reserves. Although he fully discussed the rising rate of failure and insolvency among New York banks, he never mentioned the relation between rising excess reserves and rising failure rates. He believed that open market purchases would be useful only if the banks acquiring reserves used them to acquire low-quality bonds.” In other words, in mid-1931, and afterward, the reason the Fed chose not to expand the money supply was that it believed monetary expansion would not result in an expansion of credit. This was an implication of the RieflerBurgess doctrine: Because loan demand was weak, banks were accumulating excess reserves. Adding more money to the system would just result in

Volatile Times and Persistent Conceptual Errors

199

greater holdings of bank reserves, not more lending, and thus would have little effect. The Fed failed to see that the expansion in excess reserves reflected banks’ increased demand for reserves in the wake of large losses of capital and increased liquidity risk (Calomiris & Wilson, 2004; Van Horn 2008; Calomiris et al., 2011). The Fed should have interpreted the increase in reserves as indicative of a decline in loan supply (Calomiris & Mason, 2003b), as depicted in Figure 4.2, but instead they saw it as a decline in loan demand and believed (because of their adherence to the real bills doctrine) that they should accommodate that decline. Harrison was not alone in that view. Marriner Eccles, in his 1935 testimony before the House Committee on Banking and Currency, agreed with Congressman Goldsborough, who coined the phrase: “one cannot push on a string” (Meltzer, 2003, p. 478). Eccles argued that monetary policy was not capable of restoring economic activity, hence Eccles’ desire for fiscal activism; for monetary policy to work, “you must have borrowers who are willing and able to borrow” (Meltzer, 2003). The liquidity trap did not constrain monetary policy during the 1930s. False beliefs about the liquidity trap, resulting from the flawed conceptual framework used to define Fed objectives and interpret economic data, did constrain policy. Those false beliefs about the liquidity trap were central in explaining why the Fed failed to expand the money supply during the Depression.

Did the Reserve Requirement Increases of 1936–7 Cause the Recession of 1937–8? The reorganization of the Federal Reserve System in 1935 was presented as a means of centralizing decision making over monetary policy at the Board and giving greater authority to the Federal Reserve System. This reflected the rising influence of Marriner Eccles within the Roosevelt team and the declining influence of Carter Glass (the architect of the decentralized approach to monetary policy). In the name of coordination, Eccles was especially desirous of diminishing the power of the New York Fed. A modified version of Eccles’ plan was adopted, which reorganized the Federal Open Market Committee (FOMC) to put control over monetary policy in the hands of presidential appointees for the new Board of Governors. The Fed also was granted more authority over the setting of reserve requirements. The Banking Act of 1935 eliminated the so-called Thomas Amendment of 1933, which had required presidential approval of changes in bank reserve requirements.

200

Charles W. Calomiris

On the surface, the Banking Act of 1935 seems like it should have strengthened the power of the Fed to control the money supply. But appearances can be deceiving, especially when the intent is deception. As already noted, after 1933, the treasury occupied the driver’s seat of monetary policy, and the Fed was in the “back seat.” The Fed made no change to the size of its open market portfolio from the end of 1933 until 1937. Although monetary policy was highly expansionary from 1933 to the end of 1936, the primary contributor to monetary expansion during that period was gold inflows, which were controlled by the treasury. Under the Gold Reserve Act of January 30, 1934, the treasury set the exchange rate, managed the Exchange Stabilization Fund (ESF), and decided on gold sterilization policy. The Silver Purchase Act of June 19, 1934, added further treasury authority over the creation of money. The power of the treasury over monetary affairs was not accidental. There was little the Fed could do to oppose the treasury’s aggressive monetary expansion. That fact reflected Secretary Henry Morgenthau’s intentions. Morgenthau’s reflections on the limits to Fed power after the Banking Act of 1935 in his diary are particularly telling: “[The Treasury’s] power has been the Stabilization Fund plus the many other funds that I have at my disposal and this power has kept the open-market committee in line and afraid of me” (Blum, 1959, p. 352). And the appearance of centralized Fed power, alongside the fact of Fed impotence, particularly suited his intentions: “I prophesy that . . . with the seven members of the Federal Reserve Board and the five governors of the Federal Reserve Banks forming an open market committee, that one group will be fighting the other and that consequently they will not be able to do anything constructive, and that therefore if the financial situation should go sour the chances are that the public will blame them rather than the Treasury” (Blum, 1959, p. 352). The power of the treasury relative to the Fed was asymmetric; the Fed was not able to challenge the treasury’s desire to expand the money supply for the simple reason that the treasury’s capacity to expand high-powered money was greater than the Fed’s capacity to contract it, which was limited by the size of the Fed’s balance sheet (Calomiris & Wheelock, 1998). But the Fed could have forced an increase in the money supply (via open market purchases) even if the treasury had tried to oppose that increase with a contraction, because, in that case, the Fed’s power to expand could not have been checked by the treasury. In effect, the reason the Fed was in the backseat after 1933 was that the treasury wished to expand as much as or more than the Fed would have liked. In December 1936, fears of increasing inflation risk voiced by Chairman Eccles, as well as by President Roosevelt (Meltzer, 2003, pp. 502–7, 516), led

Volatile Times and Persistent Conceptual Errors

201

the treasury to begin sterilizing gold inflows, which ended the long period of persistent growth of the monetary base. A brief power struggle between the Fed and the treasury over who would control sterilization policy was resolved in 1936, when the treasury won that battle, as Secretary Morgenthau predicted he would (Blum, 1959, pp. 360–7). Although the Fed lacked the balance sheet clout to play chicken with the treasury credibly over the control of the monetary base, as already noted, the Fed was given new power under the 1935 act to control reserve requirements. In 1936 and 1937, in a series of actions, the Fed raised the reserve requirements on deposits substantially. The treasury was sometimes displeased by those actions and was sometimes successful (in March 1937) in pressuring the Fed to expand the monetary base in order to preempt threatened use of the Exchange Stabilization Fund by Morgenthau to accomplish that same end. On March 15, 1937, under the threat that the treasury would use the Exchange Stabilization Fund to increase the money supply (described in the minutes of the March 13, 1937, FOMC Executive Committee meeting), the FOMC responded to the treasury’s concerns over a very slight increase in treasury yields by agreeing to stabilize yields with open market operations, if necessary. As recorded by Morgenthau in his diary, he and President Roosevelt understood that they controlled the money supply, and they were not shy about reminding the Fed of the fact: The President suggested that I should say to the Federal Reserve: “Now Congress gave you the job of managing the money market and that is your responsibility. You muffed it. You haven’t done it. You have not maintained an orderly market, and this thing is getting steadily worse. . . . Now I, Henry Morgenthau, Jr., talking for the U.S Government, serve notice on the Federal Reserve Board that I ask you to do what Congress has given you the power to do, namely, to increase your portfolio. If you don’t do it, the Treasury will step in. . . . We are putting you on notice.” (Blum, 1959, pp. 373–4, quoted in Calomiris & Wheelock, 1998)

This passage shows that the treasury’s definition of the Fed’s job during this period was to peg the interest rates on treasury securities. That was the “orderly market” that Morgenthau was referring to. This passage also shows that, in effect, even the power to change reserve requirements did not give the Fed monetary control. Reserve requirement changes could be offset by an expansion of the base, and the treasury had enough power to insist on such an expansion, or if necessary, to produce one. In fact, the reserve requirement increases of 1936–7 were not intended as a means of contracting the money supply, nor did Fed officials believe that the increases in reserve requirements had done so. Rather, the Fed desired to increase the reserve requirement so that required reserves would

202

Charles W. Calomiris

be closer in amount to total reserves, to facilitate future monetary policy actions, and thus to forestall long-term inflation risk. This view was part and parcel of adherence to the real bills doctrine, which saw excess reserves as a redundant slack in the system that created an obstacle to the Fed’s ability to use its targets predictably (e.g., borrowed reserves).20 Friedman and Schwartz (1963) recognized that this was the Fed’s intent, but they argued that, in fact, the reserve requirement changes had a major contractionary effect on the money supply, over and above the effect of the sterilization of gold inflows (pp. 459, 510–11, 520–2, 544). More recent scholarship has questioned the importance of the reserve requirement increases for producing the recession of 1937–8. Meltzer (2003, p. 503) notes that the reserve requirement change of mid-1936 “had no perceptible effect on the economy in 1936,” although he is more accepting of the Friedman and Schwartz view for the 1937 reserve requirement increases. Hanes (2006) and Calomiris et al. (2011) argue that there is no evidence of an effect from reserve requirement changes on the economy, and that one must look elsewhere (gold inflow sterilization, contractionary fiscal policy) to explain the recession of 1937–8. Hanes (2006) constructs a model of the effects of monetary policy on long-term interest rates. He shows that treasury bond yields did not respond to the reserve requirement changes. He concludes that policies affecting the supply of reserves – especially gold inflows before sterilization – had significant effects on interest rates, but reserve requirement changes did not. Calomiris et al. (2011) provide the microfoundations that explain Hanes’s (2006) results. Reserve holdings by member banks, and other banks, reflected risk management (see also Calomiris & Wilson, 2004; Van Horn, 2008). Given the lost capital of banks in the 1930s and the liquidity risk banks had experienced, bankers accumulated large reserves, and almost all banks held substantially more reserves than what the Fed required. The Fed reserve requirement increases, although substantial, were still not even close to binding on bank preferences for the most part, because banks wanted to hold substantially more reserves than the Fed ever required them to hold. To test this theory, Calomiris et al. (2011) employ microeconomic data on individual bank balance sheets and income statements for 1934 and 1935, and other data, to construct a model that predicts bank reserve ratios in 1935. The coefficients from this model are then used to forecast the ratios of reserves to total assets (or the ratio of reserves to total deposits) for fifteen mutually exclusive categories of banks: twelve district-level reserve city bank aggregates, two central reserve city aggregates (New York City and Chicago),

Volatile Times and Persistent Conceptual Errors

203

and a non-reserve city bank aggregate, using various alternative definitions of reserves (including or excluding various categories of liquid assets in the definition of reserves). Calomiris et al. (2011) then test to see whether the actual reserve ratios in 1936 and 1937 for these fifteen bank aggregates were higher than the counterfactual forecasts based on the 1935 estimates. They find no evidence that actual reserve ratios were higher than those forecast based on 1935 estimates. In other words, there is no evidence that reserve requirement changes in 1936–7 had any effect on the reserve ratios chosen by banks.

Summary Even with the benefit of seven decades of hindsight, some of the most important facts about the nature of the shocks of the interwar period, and some of the most important alleged sources of errors of monetary policy identified by academics during those seven decades, are not so obvious. Should the Fed have tried to discourage the run-up in the stock market, or did its attempt to do so create a shock rather than respond to one? The evidence seems at least as compelling in favor of the latter alternative. Was the Fed wrong to be concerned about a run on the dollar in October 1931, or was the reaction to gold outflows and perceived risk prudent? Again, it is hard to dismiss out of hand the possibility that the Fed was right to delay monetary expansion until international markets had calmed a bit. (Of course, that possibility in no way absolves the Fed from its disastrous policies before September 1931 or after June 1932.) Were the alleged Friedman-Schwartz panics of late 1930 and early 1931 national events that should have grabbed the Fed’s attention, or were they local affairs and largely a continuation of the patterns of local bank failures that had occurred for a decade in agricultural areas? The evidence seems to be fairly strong in favor of the latter. Was there a liquidity trap that rendered monetary policy impotent during the Depression? There are various ways to define the liquidity trap, and none seem applicable to the Depression. Nevertheless, at some points Fed officials thought that there was no point in expanding the money supply because they were observing rises in excess reserves (which indicated no need for additional liquidity, according to the Riefler-Burgess framework). Were the reserve requirement changes of 1936 and 1937 key sources of the recession of 1937–8, as Friedman and Schwartz (1963) argued, or were they largely neutral in their effects, as Fed officials had believed? Again, the evidence seems to be fairly strong in favor of the latter.

204

Charles W. Calomiris

Even if one disagrees with the conclusions reached here – and there is room for reasonable disagreement – it would be hard to argue that opposite conclusions could be proven with existing data and analysis even as of 2010. That should give those who are puzzled by the slow learning curve of the Fed during the interwar period some pause. If some of the smartest policy analysts in the world, thinking at leisure decades after the Depression was over, were not able to get key aspects of the story about the Depression right in the 1960s, the 1990s, or the 2000s, then perhaps it is not so surprising that the Fed took so long to reject the Riefler-Burgess framework.

V. The Fed’s Success Story: Stabilizing Seasonal Swings in Interest Rates There is substantial evidence (Miron, 1986; Richardson & Troost, 2006; Bernstein et al., 2010) that the founding of the Fed reduced seasonal volatility of interest rates, increased seasonal variation in lending, and reduced liquidity risk in the banking system, which in turn reduced the propensity for bank panics. This must be regarded as the Fed’s “success story” of the interwar period. After reviewing the evidence of that success, I consider why the Fed was able to stabilize the financial system at the seasonal frequency so much better than it was able to manage monetary policy over the business cycle. Miron (1986) showed that the Fed’s founding was associated with reduced seasonal variability of interest rates and increased seasonal variability of lending. Why, exactly, did Fed lending practices make the loan supply function more elastic? Miron’s (1986) findings can be explained by a variant of the deposit risk targeting model in Calomiris and Wilson (2004), as discussed in Section II. In that model, the riskiness of deposits is a function of bank asset risk and bank leverage. Because total bank capital and total cash (gold and currency) assets in the economy do not vary much from month to month, a seasonal increase in bank lending (especially to finance crop harvesting and transport in the fall, which Davis, Hanes, & Rhode, 2007, show was largely driven by the cotton cycle) implies a commensurate increase in bank asset risk and in bank leverage, which unambiguously means an increase in the riskiness of deposits (the actuarialy fair default risk premium). This is a source of seasonal variation in the risk of deposit withdrawals, since market discipline makes the risk of withdrawal in the deposit market sensitive to increases in default risk (i.e., some depositors are intolerant of risk and will withdraw when risk increases). A bank that increases its lending, ceteris

Volatile Times and Persistent Conceptual Errors

205

paribus, faces increased deposit withdrawal risk, particularly if an adverse cyclical shock hits during a seasonal lending spike. As noted before, all six of the major national banking era panics happened at cyclical peaks; they were clearly responses to adverse economic shocks to banks’ balance sheets (Calomiris & Gorton, 1991). Furthermore, these panics all occurred either during the spring planting season or the fall harvest, at times when lending (and bank liquidity risk) was at a seasonal peak. Liquidity risk in the banking system peaked in the fall and spring. Seasonal liquidity risk, combined with cyclical changes in default risk, can explain why the panics of the national banking era happened at cyclical peaks (defined by GDP, the stock market, and business failures) in months of highest liquidity risk. From the perspective of this model, the founding of the Fed provided a means of reducing liquidity risk to banks by giving them a source of liquidity, if needed, to stem deposit withdrawals (making them less vulnerable to withdrawal risk at times when seasonal lending peaks coincided with cyclical downturns). The founding of the Fed thus flattened the bank loan supply function, making loans vary more over the seasonal cycle, and interest rates vary less. This seasonal flattening of the loan supply curve is identical to the pattern already illustrated in Figure 4.1. Bernstein et al. (2010) provide additional evidence consistent with that interpretation of the Fed’s effect on seasonal liquidity risk. They compare the standard deviations of stock returns and short-term interest rates over time in the months of September and October (the two months of the year when markets were most vulnerable to a crash because of financial stringency from the harvest season) with the rest of the year before and after the establishment of the Fed. Stock volatility in those two months fell more than 40 percent and interest rate volatility more than 70 percent after the founding of the Fed. They also show that this result is driven by years in which business cycles peaked. In other words, the main risk that the Fed eliminated was associated with combined cyclical peaks in economic activity and seasonal peaks in lending. What explains the Fed’s ability to successfully manage seasonal fluctuations? Most important, a largely passive policy of freely discounting bills for short periods, supplemented as need be with short-term season-specific open market operations, should work reasonably well for accommodating seasonal shifts in demand, so long as there are few seasonal shifts in supply (which seems likely). That is, the seasonal variation of the loan market is probably captured reasonably accurately by Figure 4.1. It also should have been fairly easy for the Fed to learn that its approach to seasonal smoothing

206

Charles W. Calomiris

of interest rates and seasonal accommodation of loans was working well, because the Fed could observe the increased variation in loans over the season and the decreased variation in interest rates. Thus, in contrast to the cyclical frequency – in which relying on the Riefler-Burgess framework was disastrous – that framework worked pretty well for managing seasonal fluctuations in the market. That suggests another possibility. It is possible that seasonal success slowed the Fed’s learning process about cyclical failure. Fed officials may have been encouraged by the observable success of the Riefler-Burgess thinking, as applied to the seasonal cycle, and may have wrongly extrapolated that evidence as proof that their policy approach was also useful for managing cyclical variation. Given how hard it was to learn at the cyclical frequency (owing to few observations of cycles, multiple and diverse shocks affecting cycles, and various structural shifts of the economy), the Fed might have given greater weight to its seasonal success story in reinforcing its confidence in the Riefler-Burgess framework.

VI. Toward the Accord The late 1930s brought World War II, which was associated with continuing fundamental changes in economic structure – mobilization for war, high government spending, sectoral production shifts, price controls, rationing, and continuing improvements in technology, transportation, and communication. World War II saw large increases in government debt, which increased by roughly one quarter of GDP. Fed policy was essentially dormant during the war; the Fed assumed a largely passive role in support of the treasury’s war financing efforts by placing a ceiling of 0.375 percent on the interest rates for treasury bills and 2.5 percent for treasury bonds (Meltzer, 2003, p. 580). After the war, the treasury continued to control Fed policy, in support of treasury yields, reflecting treasury and Fed concerns of postwar recession and deflation (Meltzer, 2003, Chapter 7). The size of Fed holdings of marketable debt grew dramatically during the war and its aftermath. From December 1941 to December 1950, Fed holdings rose from $2.3 billion to $20.8 billion (Meltzer, 2003, p. 720). After some seventeen years of Fed subservience to the treasury, in 1951 the Fed and the treasury agreed that the Fed’s independence would be restored, at least in part. This change seems to have reflected two key influences: (1) Rising inflation risk led the Fed to seek and obtain support in Congress for its independence and for the desirability of shifting its focus from bond yield

Volatile Times and Persistent Conceptual Errors

207

pegging to controlling inflation (Meltzer, 2003, p. 723). (2) The increased size of the Fed’s balance sheet increased its clout (Calomiris & Wheelock, 1998). Recall that in 1935 Secretary Morgenthau was confident of his ability to force the Fed to do his bidding precisely because his ability to expand (through his various monetary powers) exceeded the Fed’s ability to contract. World War II increased the scale of the Fed, substantially enhancing its power to shrink the money supply. This fact gave the Fed a new strategic advantage that it could use to defend itself from treasury coercion. The 1951 Treasury-Fed Accord reestablished substantially greater Fed independence, but the Fed returned to its old ways, adopting once again the Riefler-Burgess framework to guide its policies (Meltzer, 2003, p. 723).

VII. Conclusion The Fed was established to bring stability to the U.S. financial system through the creation of an elastic supply of liquidity. Exactly what that meant and how it would be accomplished was not clear, and given the unique microeconomic (unit) structure of the American banking system, the Fed would not be able to rely much on precedents from abroad to guide it. Perhaps ironically, the policy tool created to reduce the probability of banking panics like those of 1873, 1893, and 1907 (which, in comparison with the Great Depression, were short-lived events involving few bank failures) became the source of the worst economic disaster and banking system collapse in American history. The most remarkable aspect of the period 1914–51 was its volatility, economically, politically, and financially. The first four decades of the Fed’s existence coincided, practically without interruption, with World War I, the roaring twenties, the Great Depression, World War II, and the Korean War. These episodes saw dramatic restructuring of the U.S. economy, diverse and severe shocks to the economy, and fundamental changes in government powers and Fed structure. The Fed made numerous errors in cyclical policy, especially during the period 1929–33, when its actions precipitated the Great Depression. Although there were many contributing sources of error, the key source of the Fed’s errors was adherence to the real bills doctrine, implemented through the Riefler-Burgess framework, which failed to distinguish between demand and supply shocks in the loan market or between real and nominal interest rates. This led the Fed to misinterpret signals about the economy, effectively to mistake declining loan supply for declining loan demand during the Depression.

208

Charles W. Calomiris

Interestingly, the Fed did not learn from its mistakes. Fed officials were pleased with their performance during the Depression because they believed that they had followed sound principles, rather than being distracted by ad hoc goals, which they believed would have made matters worse. The volatility of the period probably helps explain why learning was so slow. Diverse shocks and structural change made it harder for the Fed to distinguish supply and demand shocks or to identify the effects of its own actions as a source of economic troubles. A review of five major controversies surrounding Fed actions during the Depression (whether the Fed’s intervention to cool the stock market in 1928–9 was warranted, whether the gold standard was a constraint on monetary expansion from October 1931 to January 1932, whether banking distress of the early Depression years reflected national panics or fundamentals, whether monetary policy was impotent because of a “liquidity trap” during the Depression, and whether reserve requirement increases were a cause of the recession of 1937–8) shows that even decades after the end of the Depression, many of the issues at the heart of Fed decision making in the 1930s remain controversial. From that standpoint, the slow pace of Fed learning about its flawed policy framework is not surprising. The Fed’s greatest success during the period was the smoothing of cyclical variations in interest rates and liquidity risk, an achievement that contributed to financial stability and was very much at the heart of the founding of the institution. Ironically, that success may itself have slowed the process of learning about cyclical policy errors, because the demand-side-dominated mindset of the Fed about the loan market, embodied in the Riefler-Burgess framework, worked well for purposes of seasonal smoothing; seasonal success may have boosted the Fed’s confidence in its cyclical miscalculations. The structure of the Fed changed dramatically over time. Its founders envisioned a decentralized structure, which would avoid concentration of power, ensure a connection to the local banks’ needs, and avoid politicization of Fed actions. The desire for decentralization, however, conflicted somewhat with the need for a national monetary policy. In its early years, the Fed balanced its decentralized structure and its national mission reasonably well. In 1935, in reaction to perceived problems of insufficient coordination, power in the Fed was centralized. However, real power resided at the treasury until the 1951 Treasury-Fed Accord. These structural shifts are a reminder that the Fed is a creature of government, subject to political influence. The recognition of that fact led the Fed to become an active political force in Washington (a process that could be said to have begun in earnest with the lobbying for its own independence in 1951).

Volatile Times and Persistent Conceptual Errors

209

References Ahamed, L. (2010). Lords of finance: 1929, The Great Depression, and the bankers who broke the world. New York: Windmill Books. Allen, F. L. (1931). Only yesterday: An informal history of the 1920s. New York: Harper & Row. Alston, L. J., Grove, W. A., & Wheelock, D. C. (1994). Why do banks fail? Evidence from the 1920s. Explorations in Economic History, 30, 409–431. Basile, P. F., Landon-Lane, J., & Rockoff, H. (2010). Money and interest rates in the United States during the Great Depression (NBER Working Paper No. 16204). Cambridge: National Bureau of Economic Research. Bernanke, B. S. (1983). Nonmonetary effects of the financial crisis in the propagation of the Great Depression. American Economic Review 73, 257–276. Bernanke, B. S. (1995). The macroeconomics of the Great Depression: A comparative approach. Journal of Money, Credit and Banking, 27, 1–28. Bernanke, B. S., & James, H. (1991). The gold standard, deflation, and financial crisis in the Great Depression: An international comparison. In R. Glenn Hubbard (Ed.), Financial markets and financial crises (pp. 33–68). Chicago: University of Chicago Press. Bernstein, A., Hughson, E., & Weidenmier, M. D. (2010). Identifying the effects of a lender of last resort on financial markets: Lessons from the founding of the Fed. Journal of Financial Economics, 98(1), 40–53. Blum, J. M. (1959). From the Morgenthau diaries: Years of crisis, 1928–1938. Boston: Houghton Mifflin. Board of Governors of the Federal Reserve System. (1943 [1976]). Banking and monetary statistics: 1914–1941. Washington, DC: Board of Governors. Bordo, M. (1985). The impact and international transmission of financial crises: Some historical evidence, 1870–1933. Revista di Storia Economica, 2(2), 41–78. Bordo, M. D., Choudri, E. U., & Schwartz, A. J. (2002). Was expansionary monetary policy feasible during the Great Contraction? An examination of the gold standard constraint. Explorations in Economic History, 39(1), 1–28. Bordo, M. D., & Eichengreen, B. (2003). Crises now and then: What lessons from the last era of financial globalization? In Monetary history, exchange rates and financial markets: Essays in honour of Charles Goodhart, Paul Mizen (Ed.), (Vol. 2, pp. 53–84), London: Edward Elgar. Bordo, M. D., & Landon-Lane, J. (2010). The lessons from the banking Panics in the United States in the 1930s for the financial crisis of 2007–2008 (NBER Working Paper No. 16365). Cambridge: National Bureau of Economic Research. Bordo, M. D., & Wheelock, D. C. (2010). The promise and performance of the Federal Reserve as lender of last resort 1914–1933. In this volume. Bresnahan, T. F., & Raff, D. M. G. (1991). Intra-industry heterogeneity and the Great Depression: The American motor vehicle industry, 1929–1935. Journal of Economic History, 51, 317–332. Bresnahan, T. F., & Raff, D. M. G. (1992). Technological heterogeneity, adjustment costs, and the dynamics of plant shut-down behavior: The American automobile industry in the time of the Great Depression (Working Paper No. 92–097). Cambridge, MA: Harvard Business School.

210

Charles W. Calomiris

Brunner, K., & Meltzer, A. H. (1964). The Federal Reserve’s attachment to the free reserves concept (Subcommittee on Domestic Finance, Committee on Banking and Currency, United States House of Representatives, Eighty-Eighth Congress, Second Session). Washington, DC: U.S. Government Printing Office. Brunner, K., & Meltzer, A. H. (1968a). What did we learn from the monetary experience of the United States in the Great Depression? Canadian Journal of Economics, 1, 334–348. Brunner, K., & Meltzer, A. H. (1968b). Liquidity traps for money, bank credit and interest rates. Journal of Political Economy, 76, 1–37. Calomiris, C. W. (1989). Deposit insurance: Lessons from the record. Economic Perspectives, May/June, 10–30. Calomiris, C. W. (1990). Is deposit insurance necessary? A historical perspective.” Journal of Economic History, 50, 283–95. Calomiris, C. W. (1992). Do vulnerable economies need deposit insurance? Lessons from U.S. Agriculture in the 1920s. In Philip L. Brock (Ed.), If Texas were Chile: A primer on bank regulation (pp. 237–349, 450–458). San Francisco: The Sequoia Institute. Calomiris, C. W. (1993, Spring). Financial factors in the Great Depression. Journal of Economic Perspectives, 7(2), 61–85. Calomiris, C. W. (2000). U.S. bank deregulation in historical perspective. New York: Cambridge University Press. Calomiris, C. W. (2009). Banking crises and the rules of the game (Working Paper). New York: Columbia Business School. Calomiris, C. W. (2010). The political lessons of Depression-era banking reform. Oxford Review of Economic Policy, 26, 540–560. Calomiris, C. W. (2012). New York City banks’ policies affecting ownership structure, capital raising, and listing in 1927–1931 (Working Paper). New York: Columbia University. Calomiris, C. W., & Gorton, G. (1991). The origins of banking panics: Models, facts, and bank regulation. In R. G. Hubbard (Ed.), Financial markets and financial crises (pp. 107–173). Chicago: University of Chicago Press. Calomiris, C. W., & Hubbard, R. G. (1989). Price flexibility, credit availability, and economic fluctuations: Evidence from the U.S., 1894–1909. Quarterly Journal of Economics, 104(3), 429–452. Calomiris, C. W., & Hubbard, R. G. (1995). Internal finance and investment: Evidence from the undistributed profits tax of 1936–1937. Journal of Business, 68(4), 443–482. Calomiris, C. W., & Kahn, C. M. (1991). The role of demandable debt in structuring Optimal banking arrangements. American Economic Review, 81, 497–513. Calomiris, C. W., & Mason, J. R. (1997). Contagion and bank failures during the Great Depression: The June 1932 Chicago Banking Panic. American Economic Review, 87, 863–883. Calomiris, C. W., & Mason, J. R. (2003a). Fundamentals, panics and bank distress during the Depression. American Economic Review, 93, 1615–1647. Calomiris, C. W., & Mason, J. R. (2003b). Consequences of bank distress during the Great Depression. American Economic Review, 93, 937–947. Calomiris, C. W., & Mason, J. R. (2004). How to restructure failed banking systems: Lessons from the U.S. in the 1930s and Japan in the 1990s. In T. Ito & A. Krueger

Volatile Times and Persistent Conceptual Errors

211

(Eds.), Governance, regulation, and privatization in the Asia-Pacific region (pp. 375– 420). Chicago: University of Chicago Press. Calomiris, C. W., Mason, J. R., & Wheelock, D. C. (2011). Did the doubling of reserve requirements cause the recession of 1937–1938? A microeconomic approach.” NBER Working Paper No. 16688, January. Calomiris, C. W., & Powell, A. (2001). Can emerging market bank regulators establish credible discipline: The case of Argentina, 1992–99. In F. S. Mishkin, (Ed.), Prudential supervision: What works and what doesn’t (pp. 147–196). Chicago: University of Chicago Press. Calomiris, C. W., & Schweikart, L. (1991). The panic of 1857: Origins, transmission, and containment. Journal of Economic History, 51, 807–834. Calomiris, C. W., & Wheelock, D. C. (1998). Was the Great Depression a watershed for American monetary policy? In M. D. Bordo, C. Goldin, & E. N. White (Eds.), The defining moment: The Great Depression and the American economy in the twentieth century (pp. 23–66). Chicago: University of Chicago Press. Calomiris, C. W., & White, E. N. (1994). The origins of Federal Deposit Insurance. In C. Goldin & G. Libecap (Eds.), The regulated economy: A historical approach to political economy (pp. 145–188). Chicago: University of Chicago Press. Calomiris, C. W., & Wilson, B. (2004). Bank capital and portfolio management: The 1930s ‘Capital Crunch’ and scramble to shed risk. Journal of Business, 77, 421– 455. Carlson, M., Mitchener, K. J., & Richardson, G. (2010). The Florida banking panic of 1929: Real shocks, runs and intervention (Working Paper). Irvine: University of California. Chandler, L. V. (1958). Benjamin strong, central banker. Washington, DC: Brookings Institution. Chandler, L. V. (1971). American monetary policy 1928–1941. New York: Harper & Row. Currie, L. (1934). The failure of monetary policy to rrevent the Depression of 1929–32. Journal of Political Economy, 42(April), 145–177. Davis, J., Hanes, C., & Rhode, P. (2007). Harvests and business cycles in nineteenth century America (Working Paper). SUNY, Binghamton. DeLong, J. B., & Shleifer, A. (1991). The stock market bubble of 1929: Evidence from closed-end mutual funds. Journal of Economic History, 51. 675–700. Dice, C. A. (1929). New Levels in the Stock Market. New York: McGraw-Hill. Demirguc-Kunt, A., Kane, E., & Laeven, L. (Eds.). (2009). Deposit insurance around the world. Cambridge, MA: MIT Press. Eichengreen, B. (1992). Golden Fetters: The Gold Standard and the Great Depression, 1919–1939. Oxford: Oxford University Press. Ferderer, P. (2007). The rise of the American over-the-counters markets, 1876–1924 (Working Paper). St. Paul, MN: Macalaster College. Field, A. J. (2003). The most technologically progressive decade of the century. American Economic Review, 93, 1399–1413. Fishback, P. V. (2010). U.S. monetary and fiscal policy in the 1930s (NBER Working Paper No. 16477). Cambridge: National Bureau of Economic Research. Fisher, I. (1930). The stock market crash – And after. New York: Macmillan. Fisher, I. (1933). The Debt-deflation theory of Great Depressions. Econometrica, 1, 337–357.

212

Charles W. Calomiris

Fisher, I. (1935). Testimony on the Banking Act of 1935 (U.S. House of Representatives Committee on Banking and Currency, Seventy-Fourth Congress, First Session). Washington, DC: U.S. Government Printing Office. Flood, M. (1991). The great deposit insurance debate. Federal Reserve Bank of St. Louis Review, 74(July-August), 51–77. Friedman, M., & Schwartz, A. J. (1963). A monetary history of the United States, 1867– 1960. Princeton, NJ: Princeton University Press. Frost, P. A. (1971). Banks’ demand for excess reserves. Journal of Political Economy, 79 (July), 805-25. Gorton, G. (1985). Clearing houses and the origin of central banking in the United States. Journal of Economic History, 45, 277–283. Grossman, R. S. (1994). The shoe that didn’t drop: Explaining banking stability during the Great Depression. Journal of Economic History, 54, 654–682. Hanes, C. (2006). The liquidity trap and U.S. interest rates in the 1930s. Journal of Money, Credit, and Banking, 38 (1), 165–194. Hsieh, C.-T., & Romer, C. (2006). Was the Federal Reserve constrained by the gold standard during the Great Depression? Evidence from the 1932 Open Market Purchase Program. Journal of Economic History, 66(March), 140–176. Kabiri, A. (2009). Wall St. 1929: Modern valuation models, money market arbitrage and liquidity black holes. Doctoral dissertation, Cass Business School, City University, London. Lucia, J. L. (1985). The failure of the Bank of United States: A reappraisal. Explorations in Economic History, 22, 402–416. Mason, J. R. (2001). Do lender of last resort policies matter? The effects of Reconstruction Finance Corporation assistance to banks During the Great Depression. Journal of Financial Services Research, 20 (September), 77–95. McCallum, B. T. (2001). Inflation targeting and the liquidity trap (NBER Working Paper No. 8225). Cambridge: National Bureau of Economic Research. Meltzer, A.H. (1976). Monetary and other explanations of the start of the Great Depression. Journal of Monetary Economics, 2 (November), 255–471. Meltzer, A. H. (2003). A history of the Federal Reserve, 1913–1951. Chicago: University of Chicago Press. Mints, L. (1945). A History of Banking Theory. Chicago: University of Chicago Press. Miron, J. (1986). Financial panics, the seasonality of the nominal interest rate, and the founding of the Fed. American Economic Review, 76, 125–140. Mowery, D., & Rosenberg, N. (2000). Twentieth century technological change. In S. Engerman & R. Gallman (Eds.), The Cambridge Economic History of the United States (pp. 693–742). New York: Cambridge University Press. Nicholas, T. (2007). Stock market swings and the value of innovation, 1908–1929. In N. R. Lamoreaux & K. L. Sokoloff (Eds.), Financing Innovation in the United States, 1870 to the Present (pp. 217–245). Cambridge: MIT Press. Nicholas, T., & Scherbina, A. (2010). Real estate prices during the Roaring Twenties and the Great Depression (Working Paper). Cambridge, MA: Harvard Business School. Orphanides, A. (2004). Monetary policy in deflation: The liquidity trap in history and practice. North American Journal of Economics and Finance, 15, 101–124. Orphanides, A., & Wieland, V. (2000). Efficient monetary policy design near price stability. Journal of Japanese and International Economics, 14, 327–365.

Volatile Times and Persistent Conceptual Errors

213

Rappaport, P., & White, E. (1993). Was there a bubble in the 1929 Stock Market? Journal of Economic History, 53, 549–574. Rappaport, P., & White, E. (1994). Was the Crash of 1929 expected? American Economic Review, 84, 271–81. Richardson, G., & Troost, W. (2006). Monetary intervention mitigated banking panics during the Great Depression? Quasi-experimental evidence from the Federal Reserve District Border in Mississippi, 1929 to 1933 (NBER Working Paper No. 12591). Cambridge: National Bureau of Economic Research. Rockoff, H. (2008). US Economy in World War I. EH.Net Encyclopedia. Robert Whaples (Ed.). Retrieved from http://eh.net/encyclopedia/article/Rockoff.WWI Romer, C. D. (1992). What ended the Great Depression? Journal of Economic History, 52, 757–784. Sprague, O. M. W. (1910). History of crises under the National Banking System. Washington, DC: U.S. National Monetary Commission. Temin, P. (1976). Did monetary forces cause the Great Depression? New York: Norton. Temin, P. (1989). Lessons from the Great Depression. Cambridge, MA: MIT Press. U.S. Department of Commerce (1970). Historical Statistics of the United States. Washington, DC: Author. Van Horn, P. (2008). Excess reserves during the Great Contraction: Evidence from the Central Money Market of New York City, 1929 to 1932 (Working Paper). Dearborn, MI: University of Michigan. Wallis, J. J., & Oates, W. E. (1998). The impact of the New Deal on American federalism. In M. D. Bordo, E. Goldin, & E. N. White (Eds.), The defining moment: The Great Depression and the American economy in the twentieth century. Chicago: University of Chicago Press. Wheelock, D. C. (1990). Member bank borrowing and the Fed’s contractionary monetary policy during the Great Depression. Journal of Money, Credit and Banking, 22, 409–426. Wheelock, D. C. (1991). The strategy and consistency of Federal Reserve monetary policy, 1924–1933. Cambridge, MA: Cambridge University Press. Wheelock, D. C. (1992). Monetary policy in the Great Depression: What the Fed Did and Why. Federal Reserve Bank of St. Louis Review, 74, 3–28. White, E. N. (1983). The regulation and reform of the National Banking System, 1900– 1929. Princeton, NJ: Princeton University Press. White, E. N. (1984). A reinterpretation of the banking crisis of 1930. Journal of Economic History, 44, 119–138. White, E. N. (1985). The merger movement in banking, 1919–1933. Journal of Economic History, 45, 285–291. White, E. N. (1990). The stock market boom and crash of 1929 revisited. Journal of Economic Perspectives, 4(Spring), 67–83. White, E.N. (2009). Lessons from the Great American real estate boom and bust of the 1920s. NBER Working Paper 15573. Cambridge: National Bureau of Economic Research. Wicker, E. (1966). Federal Reserve monetary policy, 1917–1933. New York: Random House. Wicker, E. (1980). A reconsideration of the causes of the banking panic of 1930. Journal of Economic History, 40, 571–583.

214

Charles W. Calomiris

Wicker, E. (1982). Interest rate and expenditure effects of the banking panic of 1930. Explorations in Economic History, 19, 432-445. Wicker, E. (1996). The Banking Panics of the Great Depression. Cambridge University Press. Wigmore, B. A. (1987). Was the Bank Holiday of 1933 a run on the dollar rather than the banks? Journal of Economic History, 47, 739–756.

Notes * This paper was presented November 3, 2010, at a conference sponsored by the Atlanta Fed at Jekyll Island, Georgia. It will appear in a 100th anniversary volume devoted to the history of the Federal Reserve System. I thank my discussant, Allan Meltzer, and Michael Bordo and David Wheelock, for helpful comments on earlier drafts. 1 Cited in Ahamed (2010), p. 347. 2 For similar reasons, I do not review the literature on the transmission mechanism of monetary policy during the interwar period. This literature, especially the contributions of Fisher (1933), Bernanke (1983), Temin (1989), and Bernanke and James (1991), are reviewed in Calomiris (1993). See also Calomiris and Hubbard (1989) and Calomiris and Mason (2003b). 3 This is not to say that banking systems outside the United States avoided problems during recessions. Australia (in 1893), Argentina (in 1890), Italy (in 1893), and Norway (in 1900) suffered severe bank solvency crises, defined as episodes in which the negative net worth of failed banks exceeded 1 percent of GDP (see Calomiris, 2009). Furthermore, banking system shrinkage and distress during recessions, manifested in significant deposit and loan contraction and loan losses, gave rise to credit crunches even when it was not associated with a banking crisis. For a cross-country analysis of the macroeconomic effects of banking system credit contraction, see Bordo and Eichengreen (2003), who study the effects of historical banking distress (broadly defined) on business cycle severity. These episodes, however, are generally not properly regarded as true banking crises, as defined in Calomiris (2009) – that is, either episodes of widespread and sudden withdrawals of deposits that were due to a panic or episodes in which losses from bank insolvencies were large as a fraction of GDP. 4 To limit concentration of power, the Fed was divided into twelve quasi-autonomous districts. But the dollar was a national currency, the banking system was connected through the interbank reserve holdings and the interregional clearing of checks, and the securities markets and real economy were becoming increasingly integrated nationally. The increasing integration of the national economy favored greater coordination of policy. At the same time, sharply divergent regional shocks suggested benefits from preserving district autonomy. The severity of the Great Depression and the diagnosis that decentralization had contributed to insufficient open market operations in 1931 because of internal dissent (Meltzer, 2003, pp. 470–86), brought an end to the era of district autonomy in 1935. 5 There have been moments when the discount rate was above the market rate, but that was not part of a consistent policy rule implemented to achieve that outcome. 6 For a brief review of the development of theories of central banking, see Meltzer (2003, Chapter 2). This review correctly points out that there were early precedents

Volatile Times and Persistent Conceptual Errors

7

8

9

10

11

12

13

215

(notably the early nineteenth-century work of Henry Thornton) to what we would now regard as the “correct” broad view of the role of central banks in stabilizing the economy by targeting the overall supply of liquidity, but these views were not in the mainstream of thinking, inside or outside the United States at the time of the founding of the Fed. The theory of central banking, as it evolved in the nineteenth and early twentieth centuries, focused on two key themes: (1) appropriate mechanisms for ensuring that central banks helped to maintain the long-run commitment to a specie standard and (2) constructing an appropriate relationship between the central bank and other financial intermediaries to limit liquidity risk and avoid panics. When the two sets of signals conflicted with one another – especially when gold flows were signaling the need for an expansion of the money supply, as in 1930, when gold flowed in and real bills declined – the Fed favored the real bills doctrine. In 1930, that meant monetary contraction, contrary to the expansion that adherence to the gold standard’s rules of the games would have implied (Meltzer, 2003, pp. 401–2). As one Fed official put it: “Probably the most important effect of the Federal Reserve Act was to set up the machinery necessary to provide elastic currency; elastic in that it would be based on self-liquidating credit instruments arising out of the production and distribution of commodities. An obligation of the United States does not represent a transaction of this character . . . to the extent such obligations back the currency such currency is fiat currency” (statement by John U. Calkins at the Federal Reserve Governors Conference, May 1922, cited in Meltzer, 2003, p. 70). This view was widely shared by some of the most prominent monetary economists of the time, including A. Piatt Andrew, H. Parker Willis, J. Laurence Laughlin, and Horace White. For more details, see Mints (1945). Wheelock (1991, p. 2) cites Fisher’s 1935 House Testimony: “ . . . this depression was almost wholly preventable, and . . . it would have been prevented if Governor Strong had lived.” The notion that ensuring an elastic supply of trade credit is the central goal of monetary policy is not advocated by any economist or by the central bank of any country today, to my knowledge. There still are many people who advocate the use of nominal interest rates as a monetary policy tool, but economists and central bankers today recognize that real, not nominal interest rates, capture the cost of credit and that borrowed reserves and nominal interest rates do not offer unambiguous signals about the tightness or looseness of credit markets. The most notable exception was Lauchlin Currie (1934), who rejected the real bills doctrine and enjoined the Fed to control the quantity of money, arguing that “there exists no valid theoretical justification for the Commercial Loan Theory of Banking” and he described the Fed’s role during the Depression not as a failure of activist policy, but a policy that failed because it “was one of almost complete passivity and quiescence” (cited in Meltzer, 2003, p. 474). Deposits and failures are from the Federal Reserve Board’s data in Banking and Monetary Statistics: 1914–1941 (1943), using suspensions as the measures of failures. Nominal GNP is from the U.S. Department of Commerce’s Historical Statistics of the United States, Vol. I (1970). The Fed was required by law to maintain a 40 percent gold reserve against its notes, with the remaining 60 percent in eligible private sector paper. If reserve banks held

216

Charles W. Calomiris

less than the required 40 percent gold reserve, they were subject to a tax (e.g., if the reserve ratio fell to a percentage between 32.5 and 40, the tax rate was 1.5% (Meltzer, 2003, p. 356)). 14 The possibility of devaluation becomes a policy issue again at the end of 1932, but this has to do with the risk of Roosevelt’s leaving the dollar. For conflicting perspectives on that issue, see Wigmore (1987) and Hsieh and Romer (2006). 15 Once one disaggregates, Wicker argues, it becomes apparent that at least the first two of the three banking crises of 1930–1 identified by Friedman and Schwartz were largely regional affairs. Wicker (1980, 1996) argues that the failures of November 1930 reflected regional shocks and the specific risk exposures of a small subset of banks, linked to Nashville-based Caldwell & Co., the largest investment bank in the South at the time of its failure. Temin (1989, p. 50) reaches a similar conclusion. He argues that the “panic” of 1930 was not really a panic and that the failure of Caldwell & Co. and the Bank of United States reflected fundamental weakness in those institutions. Wicker’s analysis of the third banking crisis (beginning September 1931) also shows that bank suspensions were concentrated in a very few locales, although he regards the nationwide increase in the tendency to convert deposits into cash as evidence of a possible nationwide banking crisis in September and October 1931. Wicker agrees with Friedman and Schwartz that the final banking crisis (of 1933), which resulted in universal suspension of bank operations, was nationwide in scope. The banking crisis that culminated in the bank holidays of February–March 1933 resulted in the suspension of at least some bank operations (bank “holidays”) for nearly all banks in the country by March 6. From the regionally disaggregated perspective of Wicker’s findings, the inability to explain the timing of bank failures by using aggregate time series data (which underlay the Friedman-Schwartz view that banking failures were an unwarranted and autonomous source of shock) would not be surprising even if bank failures were entirely due to fundamental insolvency. Failures of banks were local phenomena in 1930 and 1931, and so may have had little to do with national shocks to income, the price level, interest rates, and asset prices. The unique industrial organization of the American banking industry plays a central role in both the Wicker view of the process of bank failure during the Depression and in the ability to detect that process empirically. Because banks in the United States were smaller, regionally isolated institutions, large region-specific shocks might produce a sudden wave of bank failures in specific regions even though no evidence of a shock was visible in aggregate macroeconomic time series (see the cross-country evidence in Bernanke & James, 1991, and Grossman, 1994). The regional isolation of banks in the United States, because of prohibitions on nationwide branching or even statewide branching in most states, also makes it possible to identify regional shocks empirically through their observed effects on banks located exclusively in particular regions. 16 Bordo and Landon-Lane (2010) argue that examiners’ reports of failed banks, which sometimes attribute failure to bank illiquidity in the 1930 and early 1931 period, therefore provide evidence of nationwide banking panics. That inference is not warranted, in my view, for two reasons. First, individual bank illiquidity of a failed bank, as inferred by an examiner, is not clearly distinguishable from expected insolvency. After all, suspended banks would have reopened (rather than remained closed) if

Volatile Times and Persistent Conceptual Errors

217

they did not ultimately become insolvent. A bank whose depositors lose confidence in it, causing insufficiency of reserves, which is also unable to open subsequently, may be deemed to have been “illiquid” by examiners, but may have been made illiquid precisely because it was deemed as insolvent by its depositors (Calomiris & Kahn, 1991). Second, even if some banks were driven to failure by their illiquidity (which clearly was the case in some instances) that does not imply the existence of a nationwide panic. Wicker and Calomiris and Mason (2003a) recognize that there were some regional panics during that period, but show that they were not significant at the national level. Calomiris and Mason’s (2003a) finding that identifiable fundamental shocks explain individual bank failures as well during the alleged panic episodes of 1930 and early 1931 as during other times in Depression, when Friedman and Schwartz did not allege the existence of a national panic, shows that there was not much additional aggregate importance of nonfundamental factors in explaining bank failures during the early alleged panics. 17 Discount window lending helps preserve only banks that are suffering from illiquidity, which was not the primary problem underlying large depositor withdrawals. Indeed, in 1932, President Hoover created the Reconstruction Finance Corporation (RFC) to enlarge the potential availability of liquidity, but this additional source of liquidity assistance made no difference in helping borrowing banks avoid failure (Mason, 2001). As commentators at the time noted, because collateralized RFC and Fed loans were senior to deposits, and because deposit withdrawals from weak banks reflected real concerns about bank insolvency, loans from the Fed and the RFC to banks experiencing withdrawals did not help much, and actually could harm banks, because those senior loans from the Fed and the RFC reduced the amount of highquality assets available to back deposits, which actually increased the riskiness of deposits and created new incentives for deposit withdrawals. In 1933, however, once the RFC was permitted to purchase banks’ preferred stock (which was junior to deposits), RFC assistance to troubled banks was effective in reducing the risk of failure (Mason, 2001). 18 Richardson and Troost (2006) show that, despite the limited ability of Fed discount window lending to absorb credit risk, Fed provision of liquidity to member banks mitigated bank failure risk associated with illiquidity somewhat in 1930 and could have played a greater role in stemming illiquidity-induced failures if the Fed had been more willing to relax lending standards to member banks. They study the failure propensities of Mississippi banks. The Federal Reserve Act of 1913 divided Mississippi between the 6th (Atlanta) and 8th (St. Louis) Federal Reserve Districts. The Atlanta Fed championed a more activist role in providing loans to member banks experiencing troubles, whereas the St. Louis Fed rigidly adhered to the real bills doctrine and eschewed the extension of credit to troubled banks. Mississippi banks in the 6th District failed at lower rates than in the 8th District, particularly during the banking panic in the fall of 1930, suggesting that more aggressive discount window lending reduced failure rates during periods of panic. Carlson, Mitchener, and Richardson (2010) show that timely and effective liquidity provision by the Atlanta Fed during the Florida agricultural crisis of 1929 (caused by an insect infestation) was also helpful. 19 For a more detailed treatment of the liquidity trap literature and the three categories of ideas described here, see Brunner and Meltzer (1968b), Orphanides and Wieland

218

Charles W. Calomiris

(2000), Orphanides (2004), McCallum (2001), Meltzer (2003, pp. 336–7, 478), Hanes (2006), and Basile et al. (2010). 20 See Meltzer (2003, pp. 495–9). Fed officials, particularly Eccles, sometimes made statements that seemed to recognize the possibility that the rising reserve holdings of banks may reflect a shift in demand, but this was an aberration from the typical view, and in any case, Eccles, in pushing for the increases in reserve requirements, does not seem to have acted in a manner consistent with that belief (Meltzer, 2003, pp. 514–15). Other concerns – especially about inflation risk that was due to gold inflows – were also relevant (Meltzer, 2003, pp. 504–5).

Comments on “Volatile Times and Persistent Conceptual Errors” Allan H. Meltzer Prepared for the Federal Reserve Bank of Atlanta Conference on Jekyll Island, November 2010

As usual, Charles Calomiris has written a very comprehensive and valuable account of the Federal Reserve’s failings and mistakes from its beginning to the 1951 Accord with the U.S. Treasury. Much of the chapter is based on his earlier work and my history. Together they give a comprehensive history of monetary and regulatory policy in these years. It seems safe to predict that the chapter will end up on many reading lists. Instead of going over the well-known details of financing the two world wars, Calomiris concentrates on five major, disputed issues about the interwar period, issues that he calls the Federal Reserve’s “egregious alleged errors.” The errors he chose are (1) allowing the stock market crash in 1929; (2) failure to expand the supply of money in 1930–3; (3) the designation “banking panics” to describe the several waves of bank failures in the early 1930s; (4) the claim that a liquidity trap prevented the Federal Reserve from undertaking a more expansive monetary policy in the mid-1930s; and (5) the claim that doubling reserve requirements in 1937 caused the recession of 1937–8. Some similar issues remain prominent about current policy decisions. I will add a few items to the list of errors. THE 1929 STOCK MARKET CRASH

The proper response to asset market exuberance remains in dispute. In 1928–9 and again in the dot-com price explosion of the late 1990s, the board was reluctant to raise interest rates enough in an effort to stop or slow the increase. Their reasoning was expressed by Alan Greenspan’s comment that the Federal Reserve could not prevent the rise but would work to prevent deflationary effects on the financial system when the boom collapsed. This reasoning seems to say that small increases in interest rates cannot stop asset prices from rising at a 20 percent or greater annual rate, and large increases 219

220

Allan H. Meltzer

in interest rates will likely cause a recession. That seems to me to be a correct interpretation of the board’s reluctance to raise interest rates in 1928 and until August 1929. That was not the only argument the board used and, as Calomiris points out, they were misled by their mistaken reliance on the real bills doctrine and later by the failure to distinguish real and nominal interest rates. In several papers and a book, Karl Brunner and I always treated credit and money as related but distinct. Monetary transmission works by changing relative prices, especially the price of assets relative to the price of output. Households and business borrow when monetary policy lowers interest rates and raises prices of existing assets. Borrowing to purchase newly produced housing or newly produced capital has a major role in the business cycle. The 1928–9 stock market recorded the substantial expansion in the United States that lasted until August 1929. Expansion ended after a world recession started. This period gave rise to the post hoc fallacy made famous by Ken Galbraith. One actual Federal Reserve Board error was prohibition of efforts by the New York Fed to continue lending to the market right after the crash. That was a major error, unfortunately not the last. Economic recovery after the more than 20 percent fall in the Dow Jones average in 1987 should put to rest the idea that stock price declines caused the Depression. As Calomiris notes, stock prices recovered in the early 1930s. And in 1988, the U.S. economy expanded and equity prices rose. Economic policy, especially differences in monetary policy, explains the difference. FAILURE TO EXPAND MONEY IN 1930–3

The 1913 Federal Reserve Act made the real bills doctrine law. At least by the mid-1920s Benjamin Strong recognized the doctrine’s principal flaw: Following real bills rules did not prevent expansion of money and credit. The reason is that limiting member bank borrowing to real bills restricts what banks discount but does not restrict their loans, at least not much. As Calomiris points out, Carter Glass – an author of the act and a secretary of the treasury at the end of the Wilson administration – never understood the point. Members of the Federal Reserve Board and some reserve bank governors made the same error repeatedly. Governor Norris of the Philadelphia Reserve Bank expressed the mistaken view of several when he said that open market purchases in the Depression would put out money when there was no demand for it and would require them to withdraw it when demand increased. That was the real bills error. The “free gold” excuse for policy does not die. Here are some facts. For the first two years of the Depression, the gold stock rose. The Federal

Comments

221

Reserve did not use the gold inflow to expand. Also, it observed the currency withdrawals but allowed bank reserves to decline in response instead of expanding money. As I point out in the first volume of my History, they ignored the work of Thornton, Bagehot, Fisher, and Wicksell, although they cite Bagehot in their discussions. Calomiris and I agree fully. I think the gold issue should be considered settled. The claim that “free gold” prevented expansion is not true except possibly for a few months after Britain devalued. Governor Harrison said that free gold was not the problem at the time. BANKING PANICS

Calomiris’s point is that most of the bank failures were local, not national, panics. November 1932–March 1933 is an exception. He bases his conclusion on the extensive research that he and others have done. I learned a great deal from his work. I raise this question. Didn’t the fact that the Federal Reserve failed to prevent the spread of failures, locally to be sure, raise doubts and concerns nationally? The reason President Hoover proposed the National Monetary Commission (and that later became the Reconstruction Finance Corporation) was that examiners closed banks when their capital was impaired by the decline in prices of (mostly) railroad bonds. The regulators then sold the bonds, further lowering their prices. That spread the problem from local to regional and perhaps national. Many nonmember banks failed. The regulators did nothing. I suggest this had an impact on other nonmember bank lending. The Bank of the United States was a New York City bank engaged in lending to the garment trade. It was permitted to fail in part because it was owned by “outsiders” and was not a member of the clearing-house. Permitting failure in New York must have raised concerns at other banks in major cities. We know that the money center banks wanted a guarantee that they could discount advances made to the National Monetary Commission in an emergency. The Federal Reserve would not give the guarantee, so President Hoover’s private market solution did not prevent growing failures. A government agency soon after replaced the commission. LIQUIDITY TRAPS

A liquidity trap is a plaything for monetary theorists. Calomiris cites my 1968 paper showing that where there are many different assets, a zero interest rate on short-term treasury bills does not eliminate the central bank’s

222

Allan H. Meltzer

ability to influence money and the economy. That model has asset prices, and expected real returns on real capital in the money and goods demand equations in addition to interest rates. The central bank has to purchase something other than treasury bills. Several central banks have recently purchased longer-term assets. The Federal Reserve lowered mortgage rates by purchasing mortgages and long-term treasuries. No liquidity trap there. As I write, the market has lowered long-term rates in anticipation of additional Fed purchases of notes and bonds. Portfolio substitution lowered other interest rates and increased equity prices. I agree with Calomiris. Chairman Eccles was wrong in the 1930s, and many economists have repeated the error many times. The zero interest rate on treasury bills in 1938 did not prevent recovery. The near–zero (3/8 percent) treasury bill rate was a negative real rate during World War II. No liquidity trap prevented wartime expansion of money and output. The liquidity trap requires strong assumptions about perfect substitution of all asset returns at the so-called zero bound. That’s not a good assumption about short-term adjustment. I believe that Keynes’ conjecture reflected an error by Ralph Hawtrey. Hawtrey observed that the Bank of England did not lower the bank rate below 2 percent during the deflation at the end of the 19th century. Hawtrey failed to distinguish real and nominal interest rates. Consequently, he concluded that borrowing declined at the 2 percent floor instead of recognizing that borrowing declined because real rates rose with deflation. RESERVE REQUIREMENT INCREASES IN 1936–7

I agree in part with Calomiris that the doubling of reserve requirements in 1937 was not the main cause of the 1937–8 recession. Over the president’s objections, in election year 1936, Congress voted to pay in cash the bonus promised to World War I veterans later. Veterans spent much of the transfer in 1936 setting off a boom in automobile and durable purchases. The transfers ended in 1937. Also, in December 1936, the treasury began to sterilize the gold inflow that had been the source of virtually all the increase in the monetary base and money after the 1934 devaluation. The increase in reserve requirement ratios removed $3.l billion of reserves in nine months. The reduction was approximately 28 percent of bank reserves in June 1936, before the policy changed. Banks did not soon restore the reserves that the policy removed. And unlike most changes in reserve requirements in Federal Reserve history, the Fed did not hold interest rates constant to let banks restore the lost reserves (A. H. Meltzer (2003). A history

Comments

223

of the Federal Reserve, 1913–1951 (p. 518). Chicago: University of Chicago Press). The induced interest change was relatively small, however. OTHER COMMENTS

I would be an unusual discussant if I did not disagree with some of the author’s comments. First, I think it is wrong to say that the use of nominal interest rates to indicate monetary ease and restraint has “no adherents today.” The Fed has always used nominal interest rates as an indicator of policy thrust, and the elegant Woodford model dismisses money, ignores credit, and focuses attention on interest rates. Second, I do not agree that periods like the 1920s or the 1930s were sufficiently turbulent that it was difficult to identify the error in the real bills doctrine. As previously noted, Governor Strong understood why the real bills doctrine was not a proper guide in the 1920s. And Lloyd Mints’s book thoroughly demolished the claim. Third, the 1951 Federal Reserve–the Treasury Accord did not fully restore Federal Reserve independence. The Fed became freer to raise interest rates, but it retained some responsibility for debt management. That is a main reason why the Fed continued “even keel” operations until well into the 1970s. These operations supported treasury market sales by supplying reserves to keep interest rates from rising during and shortly after the marketing and, at times, by purchasing a substantial part of the offering. Finally, I have two comments not directly related to the chapter, two comments on an important point in the chapter, and I point out some current errors. Calomiris says that the design of the Federal Reserve was flawed. Indeed, it was. Woodrow Wilson’s compromise gave the reserve banks semiautonomous status subject to supervision by a politically appointed board in Washington. That set off conflict over who was in charge – the politically appointed board or the reserve banks. With each crisis, including the current one, the board has increased its power relative to the reserve banks. The board has always been considered more political than the reserve banks. That is certainly true at present when many of the presidents would prefer a less-aggressive policy than the board has chosen. Independence is a related issue because the board has often been more politically responsive. Independence began under the gold standard. That limited the extent of permissible monetary expansions and contractions. When the gold standard ended, the restriction disappeared. The Federal Reserve was able to use its discretion to produce the Great Depression, several recessions, and the Great Inflation of the 1970s. When it feared

224

Allan H. Meltzer

deflation in the middle of the last decade, it ended the policy that appears to have followed a Taylor rule approximately. Ending that quasi-rule-like behavior helped to finance the housing surge. Another big mistake for discretionary policy took the form of actions during the recent crisis that sacrificed independence to help the treasury and placate the Congress. I believe the Federal Reserve will not restore its independence without adopting and following rule-like policy. And given the large errors made by discretion, rule-like behavior that restricts action has considerable appeal. Data collection and economic research are among the Fed’s impressive achievements that started early in Fed history and have continued. Both the board staff and several of the banks have produced important research on the role of money, rational expectations, and banking. I also commend the recent return to interest in credit markets and away from models with a single interest rate set by policy in which money and credit growth are of no interest. Earlier work by James Tobin, Karl Brunner, and me always used models that start from the consolidated balance sheet of the private sector. There are three assets – money, bonds, and real capital. Monetary analysis develops an equilibrium in which relative prices, prices, and output are such that the three stocks are willingly held. In several papers, Ben Bernanke and his coauthors analyze a model in which there are bank loans. I believe they must assume that bonds and real capital are a single asset. This does not strike me as a good idea or as appropriate at a time when we have an unprecedented debt and preventing inflation is likely to be a major problem after a while. Further, I do not see much evidence that the distribution of loans and bonds between banks and other lenders has an important effect on the outcome. But I do think that the future will be greatly concerned about the interest rate and exchange rate at which the market will willingly hold the government debt and money. Alas, errors did not end in 1951. Two of the most persistent errors are the excessive attention to near-term events over which the Federal Reserve has little control and the persistent neglect of longer-term consequences. One can read many, many pages of minutes and transcripts without ever seeing a discussion of longer-term consequences. This is disturbing as well as surprising because several significant achievements focused on mediumto long-term results. Two examples are the Volcker disinflation and the apparent effort to follow the Taylor rule from the middle 1980s to about 2004. Another example is the successful response to the productivity increase in the 1990s. Concentration on near-term events overresponds to transitory disturbances. The Volcker-Reagan disinflation would not have succeeded without

Comments

225

attention to the permanent response. This is one of the best examples of the Federal Reserve at its best. In each of these examples, the Federal Open Market Committee relaxed its attention to model forecasts based on the Phillips curve. There are many examples of excessive attention to forecasts of near-term events. Two wellknown examples are the response to a possible predicted deflation in 2003 and currently (2010). The first of these resulted in negative real interest rates that helped finance the housing boom. This mistake would seem to have been avoided or mitigated by continuing to follow the Taylor rule. The current attention to deflation increases the longer-term problem of reducing excess reserves while avoiding major inflation.

FIVE

Government Policy, Credit Markets, and Economic Activity* Lawrence J. Christiano and Daisuke Ikeda

1. Introduction 1.1. Preliminary Observations The recession that began in late 2007 poses new challenges for macroeconomic modeling. Asset values collapsed, initially in housing and then in equity [see Figure 5.1(a)]. In late 2008, interest rate spreads suddenly jumped to levels not seen in over seventy years [see Figure 5.1(b)].1 There was widespread concern among policymakers that financial markets had become dysfunctional because of a deterioration in financial firm balance sheets associated with the fall in asset values.2 These concerns were reinforced by the dramatic fall in investment in late 2008 [see Figure 5.1(c)], which suggested that a serious breakdown in the intermediation sector might have occurred. The U.S. Treasury and Federal Reserve (Fed) reacted forcefully. The Fed’s actions had the effect of reducing the cost of funds to financial institutions. For example, the Federal Funds rate was driven to zero [see Figure 5.1(d)] and the interest rate on the three-month commercial paper of financial firms also fell sharply. In addition, the Fed took a variety of unconventional actions by acquiring various kinds of financial claims on financial and nonfinancial institutions. Standard macroeconomic models are silent on the rationale and on the effects of the Fed’s unconventional monetary policies. Still, there is casual evidence that suggests the Fed’s unconventional monetary policy helped.3 The Fed began to purchase financial assets in late 2008, and financial firm commercial paper spreads dissipated quickly thereafter. In March 2009 the Fed expanded its asset purchase program enormously and corporate bond spreads also began to come down [Figure 5.1(b)]. Soon, aggregate output began to recover and the National Bureau of Economic 226

Government Policy, Credit Markets, and Economic Activity

227

S&P/Case-Shiller 10-city Home Price Index S&P 500 Index

Annual percent return

1 .9

log

.8 .7 .6 .5 .4

3

2

1

2007

2008

2009 month

2010

2007

2008

(a)

2009 Month

2010

(b)

index of industrial production real gross private domestic investment

5 Annual, Percent Rate

1

Log

.9 .8 .7

2007

2008

2009 Month

(c)

2010

4 3 2 1 2007

September, 2008

2008

2009 Month

2010

(d)

Figure 5.1. (a) Real equity and housing prices. (b) Spreads, BAA over AAA rated bonds. (c) Production and investment. (d) Federal funds rate.

Research declared an end to the recession in June 2009 [Figure 5.1(c)]. Of course, it is difficult to say what part of the recovery (if any) was due to the Fed’s policies, what part was due to the tax and spending actions in the American Recovery and Reinvestment Act of 2009, and what part simply reflects the internal dynamics of the business cycle. Many observers suppose that the Fed’s policies had at least some effect. These observations raise challenging questions for macroeconomics: r What are the mechanisms whereby a deterioration in financial firm

balance sheets causes a drop in financial intermediation and a jump in interest rate spreads?

228

Lawrence J. Christiano and Daisuke Ikeda

r How do reductions in interest rate costs for financial firms and large-

scale government asset purchases correct these financial market dysfunctions? What are the effects of these actions on economic efficiency? The answers to these questions are important for determining which asset market program should be undertaken and on what scale. Traditional macroeconomic models used in policy analysis in central banks have little to say about these questions. Although our analysis is primarily motivated by events in the United States since 2007, the questions we ask have a renewed urgency because of recent events in Europe. There is a concern that a collapse in the market value of sovereign debt may, by damaging the balance sheets of financial firms, plunge that continent into a severe recession. Models are required that can be used to think about the mechanisms by which such a scenario could unfold. We survey the answers to questions raised in the preceding two bullets from the perspective of four standard models borrowed from the banking literature and inserted into a general equilibrium environment. In each case, we drastically simplify the model environment so that we can focus sharply on the main ideas. Accordingly, the kind of details that are required for ensuring that models fit quarterly time series data well are left out. For example, the models have only two periods, most shocks are left out of the analysis, and we abstract from such things as labor effort, capital utilization, habit persistence, nominal variables, money, price, and wage-setting frictions. We also abstract from the distortionary effects of seigniorage and the other mechanisms by which governments and central banks acquire the purchasing power to finance their acquisition of private assets. We abstract from these complications by assuming that revenues are raised with nondistorting, lump-sum taxes. Finally, we make assumptions that allow us to abstract from the effects of changes in the distribution of income in the population. For the reasons described in Section 2, it is important to relax this assumption in more general analyses of unconventional monetary policy. Ultimately, the questions just raised must be addressed in fully specified dynamic, stochastic general equilibrium (DSGE) models. Only then can we say with confidence which of the financial frictions subsequently discussed is quantitatively important. Similarly, we require a DSGE model if we are to quantify the magnitude of the required policy interventions. Work on the task of integrating financial frictions into DSGE models is well under way.4 Our hope is that this chapter may be useful in this enterprise by providing a bird’s-eye view of the qualitative properties of the different models, in terms of their implications for the questions just raised.

Government Policy, Credit Markets, and Economic Activity

229

Our survey does not examine all models of financial frictions. For example, we do not review models that can be used to think about the effects of government asset purchases on a liquidity shortage (see, e.g., Moore, 2009; Kiyotaki & Moore, 2008).5 Instead, we review models that are in the spirit of Mankiw (1986), Bernanke, Gertler, and Gilchrist (1999) (BGG), Gertler and Karadi (2009), and Gertler and Kiyotaki (2011) (GK2 ). We review four models. The first two feature moral hazard problems, and the third features adverse selection. The fourth model features asymmetric information and monitoring costs. The latter model resembles that of BGG closely, although we follow Nowobilski (2011) by assuming that the financial frictions apply to financial rather than to nonfinancial firms. Our models capture in different ways the hypothesis that a drop in bank net worth caused the rise in interest rate spreads and the fall in investment and intermediation that occurred in 2007 and 2008.6 In our first two models, these effects involve the operation of fundamental “nonlinearities.” In particular, in these models there is a threshold level of bank net worth, such that when net worth falls below it, the equations that characterize equilibrium change. The second two models involve nonlinearities in the sense that the equations characterizing equilibrium are not linear. However, they are not characterized by the more fundamental type of nonlinearity found in the first two models. Where possible, we use our four models to investigate the consequences for economic efficiency of the following tax-financed government interventions: (i) reductions in the cost of funds to financial firms, (ii) equity injections into financial firms, (iii) loans to financial and nonfinancial firms, and (iv) transfers of net worth to financial firms. Regarding (ii), we define an equity injection as a tax-financed transfer of funds to a bank in which all the resulting profits are repaid to the government. In the case of (iii), we define a government bank loan as a tax-financed commitment of funds that must be repaid on the same terms as those received by ordinary depositors. All the models suggest that (i) helps to alleviate the dysfunctions triggered by a fall in net worth, though the precise mechanisms through which this happens vary. There is less agreement among the models in the case of (ii) and (iii). Whether these policies work depend on the details of the financial frictions. All the models suggest that (iv) helps. This is perhaps not surprising, because (iv) in effect undoes what we assume to be the cause of the trouble. Still, this aspect of our analysis is best viewed as incomplete, for at least two reasons. First, our models are silent on why markets cannot achieve the

230

Lawrence J. Christiano and Daisuke Ikeda

transfer of net worth to financial firms. Within the context of the models, there is no fundamental reason why it is that when funds are transferred to banks they must go in the form of credit and not net worth. We simply assume that the quantity of net worth in financial firms is fixed exogenously. We think there is some empirical basis for the assumption that bank net worth is hard to adjust quickly in response to a crisis, but whatever factors account for this observation should be incorporated into a full evaluation of (iv). Second, policy (iv) entails a redistribution of wealth and income among the population. Our models abstract from the effects of wealth redistribution. Some policies are best analyzed in only a subset of our models. Examples of such policies include leverage restrictions on banks, as well as a policy of bailing out the creditors of banks experiencing losses on their portfolios. We study the first of these policies in only two of our models, the ones in Sections 4 and 6. We study creditor bailouts in Section 4.

1.2. Overview of the Model Analysis 1.2.1. Moral Hazard I: “Running Away” Model We first describe a simplified version of the analysis in GK2 , which focuses on a particular moral hazard problem in the financial sector.7 This problem stems from the fact that bankers have the ability to abscond with a fraction of the assets they have under management. A repeated version of the oneperiod model that we study provides a crude articulation of the post2007 events. Before 2007, interest rate spreads were at their normal level (actually, zero according to the model), and the financial system functioned smoothly in that the first-best allocations were supported in equilibrium. Then, with the collapse in banking net worth, interest rate spreads jumped and financial markets became dysfunctional, in the sense that the volume of intermediation and investment fell below their first-best levels. According to the model, banks respond to the decline in their own net worth by restricting the amount of deposits that they issue. Banks do so out of a fear that if they tried to maintain the level of deposits in the face of the decline in their net worth, depositors would lose confidence and take their money elsewhere.8 Depositors would do so in the (correct) anticipation that a higher level of bank leverage would cause bankers to abscond with bank assets. From this perspective, a sharp cut in the cost of funds to banks calms the fears of depositors by raising bank profits and providing bankers with an incentive to continue doing business normally.

Government Policy, Credit Markets, and Economic Activity

231

In the case of direct equity injections and loans, we follow GK2 in assuming that the government can prevent banks from absconding with government funds.9 Under these circumstances, it is perhaps not surprising that government equity injections and loans, (ii) and (iii), are effective. With the government taking over a part of the economy’s intermediation activity, the amount of intermediation handled by the banking system is reduced to levels that can be handled efficiently with the reduced level of banking net worth. Of course, if the nature of the financial market frictions is not something that can be avoided by using the government in this way, then one suspects that (ii) and (iii) are less likely to be helpful. This is the message of our second model. 1.2.2. Moral Hazard II: Unobserved Banker Effort Our second model captures moral hazard in banking in a different way. We suppose that bankers must exert a privately observed and costly effort to identify good investment projects. The problem here is not that bankers may abscond with funds. Instead, it is that bankers may exert too little effort to make sure that assets under management are invested wisely. Bankers must be given an incentive to exert the efficient amount of effort. One way to accomplish this is for bank deposit rates to be independent of the performance of bank portfolios, so that bankers receive the full marginal return from exerting extra effort. But bankers must have sufficient net worth of their own if the independence property of deposit rates is to be feasible. This is because we assume that bankers cannot hold a perfectly diversified portfolio of assets. As a result, bankers – even those that exert high effort – occasionally experience a low return on their assets. For deposit rates to be independent of the performance of banker portfolios, bankers with poorly performing portfolios must have sufficient net worth to pay the return on their deposits. We show that when bankers have a sufficiently high level of net worth, then bank deposit rates are independent of the performance of bank portfolios and equilibrium supports the efficient allocations. Financial markets become dysfunctional when the banks whose assets perform poorly have too little net worth to cover their losses. Depositors in such banks must in effect share in the losses by receiving a low return. To be compensated for low returns from banks with poor assets, depositors require a relatively high return from banks with good assets. But when deposit rates are linked to the performance of bank assets in this way, bankers have less incentive to exert effort. Reduced effort by bankers pushes down the average return on bank assets and hence deposit rates for savers. With lower deposit

232

Lawrence J. Christiano and Daisuke Ikeda

rates, household deposits – and hence investment – are reduced below their efficient levels. Consider the implications for policy. Interest rate subsidies, policy (i), help by reducing the cost of funds to banks. This policy reduces banks’ liabilities in the bad state and so increases the likelihood that deposit rates can be decoupled from bank asset performance. This result is of more general interest, because it conflicts with the widespread view that interest rate subsidies to banks cause them to undertake excessive risk. In our environment, an interest rate subsidy increases bankers’ incentive to undertake effort, leading to a rise in the mean return on their portfolios and a corresponding reduction in variance. Interest rate subsidies have this effect by raising the marginal return on banker effort. Government equity injections and loans, policies (ii) and (iii), have no effect in the model. Although the proof of this finding involves details, the result is perhaps not surprising. The government equity injections and bank loans that we consider do not offer any special opportunity to avoid financial frictions in the way that our first model of moral hazard does. It is not obvious (at least to us) what unique advantage the government has in performing intermediation when that activity involves a costly and hidden effort. Our hidden-effort model illustrates the general principle that the sources of moral hazard matter for whether a particular government asset purchase program is effective. Our hidden-action model is well suited to studying the effects of leverage restrictions and bailouts of creditors to banks with poorly performing assets. We have previously noted that when net worth is low, it may not be possible for deposit rates to be decoupled from the performance of bank assets. Obviously, if the quantity of deposits were sufficiently low, then deposit rates could be fixed and independent of bank asset performance even if net worth is low. We show that when binding leverage restrictions are placed on banks when net worth is low, social welfare is increased. 1.2.3. Adverse Selection Our third model focuses on adverse selection as a source of financial market frictions.10 In our model, the portfolios of some banks are relatively risky in that these banks have a high probability of not being able to repay their creditors. Banks have access to credit markets. However, because bank creditors cannot assess a given bank’s riskiness, all banks must pay the same interest rate for credit.11 This interest rate must be high enough to take into account the bankers with high-risk portfolios that are likely not to

Government Policy, Credit Markets, and Economic Activity

233

repay. As is the case in adverse-selection models, under these circumstances “good” bankers – those who could potentially acquire low-risk assets – find it optimal not to borrow at all. This reflects the fact that good bankers repay creditors with high probability, so that their expected profits from borrowing and acquiring securities are low. When the net worth of bankers drops, the adverse-selection effect driving out good bankers becomes stronger. Because the rise in the interest rate spread on credit to banks drives away potentially good bankers, the quality of the assets on the balance sheet of banks seeking credit deteriorates. The result is a decline in the overall return on bank assets and thereby a fall in the equilibrium return on household saving. The reduction in saving in turn causes a fall in investment. We show that this fall in investment corresponds to an increase in the gap between the equilibrium level of investment and investment in the firstbest equilibrium. In this sense, the decline in banker net worth makes the banking system more dysfunctional. For these reasons the adverse-selection model formalizes a perspective on the financial events since 2007 that is similar to the one captured by the models in the previous two sections. We consider the policy implications of the adverse-selection model. A taxfinanced transfer of net worth to bankers improves equilibrium outcomes. This is not surprising, because an increase in banker net worth reduces banks’ dependence on external finance and hence reduces the adverseselection distortions. Government policies that have the effect of subsidizing the cost of funds to bankers also improve equilibrium outcomes. The reason is that they raise the return on saving received by households and have the effect of reducing the gap between the equilibrium interest rate and the social return on loans. 1.2.4. Asymmetric Information and Monitoring Costs Our fourth model of financial frictions focuses on asymmetric information and costly monitoring as the source of financial frictions. At this time, the costly state verification model is perhaps the most widely used model of financial frictions in macroeconomics.12 In the model, bankers combine their own net worth with loans to acquire the securities of firms with projects that are subject to idiosyncratic risk. We assume that a bank can purchase the securities of at most one firm, so that the asset side of bank balance sheets is risky. There are no financial frictions between a bank and the firm whose securities it purchases. The realization of uncertainty in a firm’s project is observed by its bank, but can be observed by bank creditors only by their paying a monitoring cost. We assume that

234

Lawrence J. Christiano and Daisuke Ikeda

creditors offer banks a “standard debt contract.” The contract specifies a loan amount and an interest rate. The bank repays the loan with interest, if it can. If the securities of a bank are bad because the issuing firm has an adverse idiosyncratic shock then the bank declares bankruptcy, is monitored by its creditor, and loses everything.13 Our characterization of the 2007–8 crisis follows the line explored with our other models by supposing that the crisis was triggered by a fall in bank net worth. In addition, our model also allows us to consider the idea that an increase in the cross-sectional dispersion of idiosyncratic shocks played a role.14 Our environment is sufficiently simple that we obtain an analytic characterization of the inefficiency of equilibrium. We show that in the model the marginal social return on credit to banks exceeds the average return, and it is the latter that is communicated to bank creditors by the market. Lending to banks is inefficiently low in the equilibrium because a planner prefers that the credit decision be made based on the marginal return on loans. The problem is exacerbated when the net worth of banks is low. Not surprisingly, we find that a policy of subsidizing bank interest rate costs improves welfare. Also, the optimal subsidy is higher when bank net worth is low. In addition, we study the effects of direct government loans to banks, but find that this has no impact on the equilibrium. The rest of the chapter is organized as follows. Section 2 describes what we call the Barro-Wallace irrelevance proposition, which sets out a basic challenge that any model of government asset purchases must address. The following two sections describe the two models of moral hazard. Section 5 studies the model of adverse selection. Section 6 studies the model with asymmetric information and costly monitoring. The final section presents concluding remarks.

2. The Barro-Wallace Irrelevance Proposition Any analysis of unconventional policy must confront a basic question. If the government acquires privately issued assets by levying taxes (either in the present or in the future), then the ownership of the asset passes from private agents to the government, which later reduces households’ tax obligations as the asset bears fruit. The question any analysis of asset purchases by the government has to answer is why it makes a difference whether private agents hold assets themselves or the government holds them on taxpayers’ behalf. In the simplest economic settings, households’ intertemporal consumption opportunities are not affected by government asset purchases, so that such purchases are irrelevant for allocations and prices. We refer

Government Policy, Credit Markets, and Economic Activity

235

to this irrelevance result as the Barro-Wallace irrelevance proposition, because it is closely related to the Ricardian equivalence result emphasized by Barro (1974) and extended by Wallace (1981) to open market operations.15 Any analysis in which government asset purchases have real effects must explain what assumptions have been made to defeat the BarroWallace irrelevance result. One way to defeat Barro-Wallace irrelevance builds on heterogeneity in the population. For example, suppose that a subset of the population has a special desire to hold a certain asset (e.g., thirty-year treasury bonds). If the government engages in a tax-financed purchase of that bond, then in effect the bond is transferred from the subset of the population that holds it initially to all taxpayers. Such a redistribution of assets among heterogeneous agents may change prices and allocations. This type of logic may be useful for interpreting the recent substantial changes that have occurred in the Federal Reserve’s balance sheet.16 We do not pursue this line of analysis further here. There are other ways in which tax-financed purchases of private securities may have real effects. In the examples we explore, this can happen by changing the market rate of interest.

3. Moral Hazard I: “Running-Away” Model17 We construct a two-period model. In the first period, households make deposits in banks. Bankers combine these deposits with their own net worth and provide funds to firms. In the second period, households purchase the goods produced by firms by using income generated by bank profits and interest payments on bank deposits. The source of moral hazard is that bankers have an option to default by absconding with an exogenously fixed fraction of their total assets, leaving the rest to depositors. When a sufficiently large fraction of a bank’s assets is purchased with bankers’ own net worth, then a bank simply hurts itself by defaulting, and it chooses not to do so. We show that, when the net worth of banks is sufficiently large that the option to default is not relevant, then the equilibrium allocations correspond to the first-best efficient allocations. We refer to this scenario as a “normal time.” When banks’ net worth is sufficiently low, banks restrict the supply of deposits. Banks do this because they know that if they planned a higher level of deposits, depositors would rationally lose confidence and take their deposits elsewhere. With the supply of deposits reduced in this way, and no change in demand, the market-clearing interest rate on deposits is low. Because the return on bank assets is fixed by assumption, the result

236

Lawrence J. Christiano and Daisuke Ikeda

is an increase in banks’ interest rate spreads.18 We refer to the situation in which bank net worth is so low that the banking system is dysfunctional and conducts too little intermediation as a “crisis time.” Thus, the model articulates one view about what happened in the past few years: “a fall in housing prices and other assets caused a fall in bank net worth and initiated a crisis. The banking system became dysfunctional as interest rate spreads increased and intermediation and economic activity was reduced.” In contemplating such a scenario, we imagine a version of our two-period model, repeated many times. Government policy can push the economy out of crisis and back to normal by undoing the underlying cause of the problem. One way the government can do this is by purchasing bank assets. In the Gertler-Karadi and Gertler-Kiyotaki analysis, it is assumed that the government has the ability to prevent banks from absconding with bank assets financed by equity or deposit liabilities to the government. We show that sufficiently large government purchases of bank assets can restore the banking system to normal. In particular, government asset purchases cause interest rate spreads to disappear and total intermediation to return to its first-best level. Interest rate spreads disappear because government-financed purchases of assets induce a fall in household demand for deposits. If the government purchases are executed on a large-enough scale, the fall in the demand for deposits is sufficient to push the deposit interest rate back up to the efficient level where it equals banks’ return on their funds. The logic of the BarroWallace irrelevance result does not hold in a crisis time because tax-financed government purchases of bank assets have an impact on the interest rate. Another policy that can resolve a crisis is one in which the government provides tax-financed loans to firms. Under this policy the government returns the proceeds of its investment in firms to households in the form of lower taxes in the second period. Households understand that this government policy is a substitute for their bank deposits, and so they reduce the supply of deposits. With the supply and demand for bank deposits both reduced, the deposit interest rate rises back up and the interest rate spread is wiped out. Total intermediation returns to its normal level because, though household deposits are relatively low, this is matched by a corresponding increase in government provision of funds. In this way, tax-financed loans to nonfinancial business can resolve a crisis. Finally, we show that a policy of subsidizing banks’ cost of funds can push the economy out of a crisis. Such a policy works by increasing banks’ profits during a crisis and so reducing their temptation to abscond with bank assets. Understanding that their depositors are aware of this, banks expand their deposits back to the first-best level.

Government Policy, Credit Markets, and Economic Activity

237

We first describe the model. We then formally establish the properties of government policy just reviewed.

3.1. Model There are many identical households, each with a unit measure of members. Some members are “bankers” and others are “workers.” There is perfect insurance inside households, so that all household members consume the same amount c in period 1 and C in period 2. In period 1, workers are endowed with y goods and the representative household makes a deposit d in a bank subject to the period 1 budget constraint: c + d ≤ y. The representative household’s period 2 budget constraint is: C ≤ Rd + π. Here, R represents the gross return on deposits and π denotes the profits brought home by bankers. The household treats π as lump sum transfers. The intertemporal budget constraint is constructed by using period 1 and period 2 budget constraints in the usual way: c+

C π ≤y+ . R R

(3.1)

The representative household chooses c and C to maximize u(c) + βu(C),

u(x) =

x 1−γ , 1−γ

γ > 0,

(3.2)

subject to (3.1). The solution to the household problem is c=

y+ 1+

π R 1

(βR) γ R

,

d = y − c, C = Rd + π.

(3.3)

We can see the basic logic of the Barro-Wallace irrelevance proposition from (3.1). Suppose the government raises taxes T in period 1, uses the proceeds to purchase T deposits, and gives households a tax cut RT in period 2. Periods 1 and 2 budget constraints are replaced by c + d ≤ y − T , C ≤ Rd + π + RT.

(3.4)

Using these two equations to substitute out for d + T , we obtain (3.1) and T is irrelevant for the determination of c and C. Deposits are determined residually by d = y − T. If the government increases T , then d drops by

238

Lawrence J. Christiano and Daisuke Ikeda

the same amount. Of course, if we change the environment in some way, then the Barro-Wallace irrelevance proposition may no longer be true. This could happen, for example, if T affected R. To investigate this, we need to flesh out the rest of the model. Bankers in period 1 are endowed with N goods. They accept deposits from households and purchase securities s from firms. Firms issue securities in order to finance the capital they use to produce consumption goods in period 2. Intermediation is crucial in this economy. If firms receive no resources from banks in period 1, then there can be no production, and therefore no consumption, in period 2. We first consider the benchmark case in which there are no financial frictions and the banking sector helps the economy achieve the first-best allocations. We suppose that the gross rate of return on privately issued securities is technologically fixed at Rk . Bankers combine their own net worth N with the deposits received d to purchase securities s from firms. Firms use the proceeds from s to purchase an equal quantity of period 1 goods that they turn into capital. The quantity of goods produced by firms in period 2 using this capital is sRk . Goods-producing firms make no profits, so sRk is the revenue they pass back to the banks. Banks pay Rd on household deposits in period 2. Bankers solve the following problem: π = max[sRk − Rd], d

(3.5)

where s = N + d and N is the banker’s state. An equilibrium is defined as follows: Benchmark equilibrium: R, c, C, d, π such that (i) (ii) (iii) (iv)

the household and firm problems are solved, the bank problem, (3.5), is solved, markets for goods and deposits clear, and d, c, C > 0.

Condition (iv) indicates that we consider only interior equilibria, both here and elsewhere in the chapter. A property of a benchmark equilibrium is R = Rk . To see this, suppose it were not so. If R > Rk the bank would set d = 0 and if R < Rk the bank would set d = ∞, neither of which is consistent with the equilibria that we study. Thus, in the benchmark case the interest rate faced by households in equilibrium coincides with the actual rate of return on capital. It is therefore not surprising that the first-best allocations are achieved in this version of the model. That is, the allocations

Government Policy, Credit Markets, and Economic Activity

239

in the efficient, benchmark equilibrium coincide with the allocations that solve the following planning problem19 : max u(c) + βu(C )

c, C, k

subject to c + k ≤ y + N , C ≤ Rk k.

(3.6)

The interest rate spread in this economy is defined as Rk − R. In the benchmark equilibrium the interest rate spread is zero. This makes sense, because there are no costs associated with intermediation and there is no default. We summarize this result as follows: Proposition 3.1: A benchmark equilibrium has the following properties: (i) the interest rate spread, Rk − R, is zero, (ii) d takes on its first-best value. To gain intuition, it is useful to define the demand for d by banks and the supply of d by households. The supply of d is obtained by solving (3.1) for d, after substituting out for bank profits from (3.5). This provides an upward sloping curve in a diagram with R on the vertical axis and d on the horizontal. Also, a drop in bank net worth, N, induces a less than onefor-one increase in the supply of d, for consumption smoothing reasons. The bank demand for d is simply a horizontal line at R = Rk . Representing the demand and supply for d in this way provides an immediate graphical illustration of the observation that in equilibrium, R = Rk . Note that in this economy, the Barro-Wallace irrelevance proposition is satisfied. Tax-financed government purchases of private assets have no impact on consumption or total intermediation, d + T. We now introduce the moral hazard problem studied by Gertler-Karadi and Gertler-Kiyotaki. A bank has two options: “default” and “not default.” Not defaulting means that a bank simply does what it does in the benchmark version of the model. In this case, the bank earns profits π = Rk (N + d ) − Rd.

(3.7)

The option to default means that the banker can take an exogenously fixed fraction θ of the assets and leave whatever is left for the depositors. A defaulting bank receives θRk (N + d ), and its depositors receive (1 − θ )Rk (N + d ). The bank chooses the no-default option if and only if doing so increases its profits: (N + d )Rk − Rd ≥ θ (N + d )Rk .

(3.8)

240

Lawrence J. Christiano and Daisuke Ikeda

By rearranging terms, we see that (3.8) is equivalent to (1 − θ )(N + d )Rk ≥ Rd.

(3.9)

That is, a bank chooses the no-default option if and only if doing so reduces what depositors receive. Each bank takes the interest rate on deposits as given, and sets its own level of deposits d. Banks are required to post their intended values of d at the start of the period, so that households can assess whether a bank will default. We consider symmetric equilibria in which no bank chooses to default and the d posted by banks satisfy (3.8). In such an equilibrium, an individual bank has no incentive to choose a level of deposits that violates (3.8) because depositors would in this case prefer to take their deposits to another bank, where they obtain a higher return [see (3.9)]. In this setting, the banker solves the following problem: π = max[Rk (N + d ) − Rd], subject to (3.8). d

(3.10)

Our formal definition of equilibrium in the case in which the banker has a default option is as follows: Financial equilibrium: R, c, C, d, π such that (i) (ii) (iii) (iv)

the household and firm problems are solved, the bank problem, (3.10), is solved, markets for goods and deposits clear, and c, C, d > 0.

It is useful to represent the equilibrium in a demand and supply diagram. As before, a bank’s demand for deposits is the mapping from each possible R into a value of d that solves (3.10). For R > Rk a bank maximizes profits by setting d = 0. For R ≤ (1 − θ) Rk , (3.8) does not constrain d and the bank would choose d = ∞. For R = Rk a bank makes no profits on d. Combining this fact with a bank’s incentive not to violate (3.8) implies that for R = Rk a bank is indifferent over values of d such that 0 ≤ d ≤ N(1 − θ)/θ. Finally, for (1 − θ)Rk < R < Rk a bank wishes to hold the largest amount of deposits that is consistent with (3.8). This implies d = Rk N(1 − θ)/(R − (1 − θ)Rk ). Thus, in the presence of the financial friction, a bank’s demand for d is no longer simply a horizontal line at R = Rk , with the demand for d infinite for R < Rk and the demand for d zero for R > Rk . The demand for d is now a horizontal line at R = Rk , extending over the interval 0 ≤ d ≤ N(1 − θ)/θ.

Government Policy, Credit Markets, and Economic Activity

R

241

Household supply of d

Financial constraint non-binding Rk

Bank demand for d

(1-θ) Rk

Financial constraint binding

N(1-θ)/θ

d

Figure 5.2. Interior, no default equilibrium.

For (1 − θ)Rk < R < Rk a bank’s demand for deposits is finite and declining in R and it is infinite for R ≤ (1 − θ)Rk . See Figure 5.2 for an illustration. This analysis implies that if a bank’s net worth N falls, then the quantity of deposits it demands shifts left. The shift is substantial for the plausible case, θ < 1/2 (see Figure 5.2). The supply of deposits is unaffected by the financial frictions. Equilibrium can again be represented as the intersection of the household’s upwardsloped supply of d with the generally downward-sloping demand curve for banks just described. Now, of course, it is not necessary for equilibrium to imply R = Rk . This will occur only if the financial constraint is nonbinding, but R < Rk if the constraint is binding (see Figure 5.2). The constraint will be binding if N is sufficiently low. We summarize our results in the following proposition: Proposition 3.2: When (3.8) is nonbinding, the financial equilibrium allocations are first-best and the interest rate spread is zero. When (3.8) binds, then the equilibrium values of d and R are below their first-best levels and the interest rate spread is positive. A sequentially repeated version of this model economy provides a rough characterization of events before and after 2007. Suppose that N was large enough in the early period, so that the economy was operating at its efficient level and no part of actual spreads was due to the type of default considerations addressed here. Then, in late 2007 the net worth of banks

242

Lawrence J. Christiano and Daisuke Ikeda

suddenly began to fall as a consequence of the collapse in housing prices. When the participation constraint began to bind, spreads opened up. The volume of intermediation – and the investment it supported – then collapsed. The preceding scenario can be visualized using a diagram like the one in Figure 5.2. Suppose that the pre-2007 economy corresponded to the left of the two equilibria depicted there. With a drop in N, the demand for d by banks shifts left by a relatively large amount (we assume θ < 1) and the supply of d by households shifts right by a relatively small amount (recall the results for supply derived earlier). With these shifts, the economy can end up in the constrained region where R < Rk and intermediation d is smaller.

3.2. Implications for Policy We now consider the effects of four kinds of tax-financed unconventional monetary policies: injections of equity into banks, deposits in banks, direct loans to firms, and subsidies to banks’ cost of funds. In each case, the policy is financed by lump sum taxes T in the first period. In the case of the asset purchase policies, the government transfers the proceeds back to households in the form of a second-period tax reduction. 3.2.1. Equity Injections into Banks In the case of an equity injection, the government transfers T to each bank. The government requires the banks to repay the earnings Rk T on the assets financed by the equity. The government transfers the Rk T back to households in period 2 in the form of a tax reduction. We assume that, unlike the household, the government has the power to prevent the bank from absconding with any part of the assets financed by T. Thus, for a bank that receives an equity injection of T, the incentive to default is still the object on the right of the inequality in (3.8). An equity injection also has no impact on a bank’s profits: (N + T + d )Rk − Rd − Rk T = (N + d )Rk − Rd. Thus, for a given level of deposits d, an equity injection has no effect on a bank’s decision to default. However, the government’s equity injection does affect the representative household’s choice of d.

Government Policy, Credit Markets, and Economic Activity

243

To understand how the representative household responds to the tax implications of an equity injection, a suitable adjustment of (3.3) implies that c=

π R

y−T + 1+

+

Rk T R

1

(βR) γ R

.

Note that T does not directly cancel in the numerator because the rate of interest enjoyed by the government when it does an equity injection is different from the household’s rate of return on deposits when (3.8) binds and Rk = R. To understand the general equilibrium impact of T on c, it is necessary to substitute out for π (3.7): c=

y−T +

Rk (N +d )−Rd R

1+

+

Rk T R

1

(βR) γ R

.

The household’s period 1 budget constraint implies d = y − T − c. Using this to substitute out for d in the preceding expression and rearranging, we obtain c=

Rk 1 (βR) γ

(N + y), +

Rk

(3.11)

d = y − c − T. Interestingly, the general equilibrium effect of T on consumption is nil, despite the difference between the government’s and the household’s interest rate. From the latter expression, we see that a rise in T has no impact on c and so it has a one-for-one negative impact on d. If (3.8) is nonbinding, then the equity injection is irrelevant. There is no impact on total intermediation d + T , and the interest rate spread remains unchanged at zero. Now suppose that (3.8) is binding. Given R, the marginal fall in d with a rise in T reduces the right-hand side of (3.9) by R and reduces the lefthand side of (3.9) by (1 − θ)Rk , thus making the incentive constraint less binding. This is because R > (1 − θ)Rk ; otherwise, (3.8) never binds. With T large enough, the incentive constraint ceases to bind altogether, and an analogous argument to the one leading up to proposition 3.2 establishes that the interest rate spread is eliminated, R = Rk , and total intermediation T + d achieves its first-best level. To see what level of T achieves the first-best, let d ∗ denote the level of deposits in a benchmark equilibrium [we can find d ∗ by solving (3.6) and

244

Lawrence J. Christiano and Daisuke Ikeda

setting d ∗ = k − N ]. Our assumption that (3.8) is strictly binding implies that N Rk < θ (N + d ∗ ), so that d ∗ is not part of a financial equilibrium. Set T to the value T ∗ that solves N Rk = θ (N + d ∗ − T ∗ ).

(3.12)

We summarize the preceding results in the form of a proposition: Proposition 3.3: When (3.8) is nonbinding, tax-financed equity injections have no impact on total intermediation d + T and on the interest rate spread Rk − R. When (3.8) binds, tax-financed equity injections reduce the interest rate spread and increase total intermediation. A sufficiently large injection restores spreads and total intermediation to their first-best level. We can express the equations of the model in words as follows. When N falls enough, the supply of deposits by banks decreases because the incentive constraint binds on the banks. This creates an interest rate spread by reducing the deposit rate (recall, the return on assets is fixed in this model). A tax-financed government purchase of assets causes the demand for deposits by households to decrease, pushing the deposit rate back up and reducing the interest rate spread. The decrease in deposits is somewhat offset by the rise in the deposit rate and this is why d + T increases with the government intervention. The intervention is welfare improving because it pushes the economy back up to the first-best allocations. 3.2.2. Government Deposits in Banks and Loans to Firms Suppose the government makes tax-financed deposits T in banks in period 1. In period 2 it returns the proceeds to households in the form of a tax cut in the amount RT. It is easy to verify that c and d are determined according to (3.11) in this case. As a result, total deposits d + T are invariant to T for a given R. If we assume that banks can as easily default on the government as on households, then total deposits d + T enter the incentive constraint and the tax-financed deposits are irrelevant. However, suppose that the government can prevent banks from defaulting on any part of the government’s deposits. In that case, the profits earned by banks on government deposits,

Government Policy, Credit Markets, and Economic Activity

245

(Rk − R)T , are not counted in the incentive constraint, (3.8). With only household deposits in the incentive constraint, the analysis is identical to the analysis of equity injections. Now consider the case in which the government makes tax-financed loans directly to firms. This case is formally identical to the case of tax-financed equity injections. For a given R, d + T is invariant to T. However, because only d enters the incentive constraint, (3.8), the reduction in d that occurs with a rise in T relaxes the incentive constraint in case it is binding. This results in an increase in R and hence a rise in total intermediation. We summarize these results in the following proposition: Proposition 3.4: If the government can prevent bank defaults on its own bank deposits, then the effects of tax-financed government deposits in banks resemble the effects of equity injections summarized in proposition 3.3. Direct government loans to firms have the same effects as those of equity injections. 3.2.3. Interest Rate Subsidies and Net Worth Transfers to Banks We now consider a policy in which the government subsidizes the interest rate that banks pay on deposits. Suppose that the equilibrium is such that the incentive constraint, (3.8), is binding. As in the previous subsection, this implies that the first-best level of deposits [i.e., the one that solves (3.6) with “deposits” identified with k − N ] violates (3.8): (N + d ∗ )Rk − Rd ∗ < θ (N + d ∗ )Rk ,

(3.13)

when the deposit rate R is at its efficient level Rk . Let τ > 0 be the solution to (N + d ∗ )Rk − Rk (1 − τ )d ∗ = θ (N + d ∗ )Rk .

(3.14)

Note that there exists a unique value of τ > 0 that solves this equation because the left-hand side is increasing in τ and the left-hand side exceeds the right-hand side when τ = 1. To finance the transfer τ Rk d ∗ to banks the government levies taxes, T = τ Rk d ∗ , on households in the second period. We now verify that this policy, together with d = d ∗ , R = Rk , and c, C at their first-best levels c ∗ , C ∗ , satisfies all the equilibrium conditions. Bank profits in the second period are π = (N + d ∗ )Rk − Rk (1 − τ ),

d ∗ = (N + d ∗ )Rk − Rk d ∗ + Rk τ d ∗ .

246

Lawrence J. Christiano and Daisuke Ikeda

Total household income is Rd + π − T = (N + d ∗ )Rk . The latter result and the assumption that c ∗ , C ∗ solve (3.6) imply that the household problem is solved. The fact that the incentive constraint is satisfied implies that the bank problem, (3.10), is solved. We summarize these findings as follows: Proposition 3.5: Suppose (3.8) binds in equilibrium so that deposits are strictly below their first-best level in a financial equilibrium. Then, a subsidy to bank deposit liabilities at the rate defined by (3.14) ensures that the first-best allocations are supported as a financial equilibrium. Next, we consider the case in which taxes are levied on households in the first period and the proceeds are given to bankers as a supplement to their net worth. The net worth transfer is financed by taxes on households in period 1. Suppose the equilibrium is such that the incentive constraint, (3.8), is binding. This implies that the first-best level of deposits d ∗ violates (3.8) and that (3.13) is satisfied with R at its efficient level Rk . Let T denote the tax-financed transfer of net worth to bankers. The pretax level of banker net worth is N and after taxes it is N + T . We conjecture, and then verify, as for T sufficiently large, that the financial equilibrium has the property that deposits equal d ∗ − T , the incentive constraint is nonbinding, and c, C coincide with their first-best values. Let T be the solution to (N + d ∗ )Rk − Rk (d ∗ − T ) = θ (N + d ∗ )Rk .

(3.15)

Note that N + d ∗ is unaffected under the tax policy and the conjecture about the equilibrium. A unique T > 0 that solves (3.15) is guaranteed to exist because the left-hand side is monotonically increasing in T and the left-hand side is assumed to be smaller than the right-hand side when T = 0. To understand how the representative household responds to the taxfinanced equity injection, a suitable adjustment of (3.3) implies that c=

y−T + 1+

π R 1

(βR) γ R

.

Under our conjecture, R = Rk and π is given by the expression on the lefthand side of the equality in (3.15). Substituting, we obtain (3.11), the level

Government Policy, Credit Markets, and Economic Activity

247

of consumption in the first-best equilibrium. This verifies our conjecture about the period 1 level of consumption. It is straightforward to verify that the first-best level of period 2 consumption satisfies the period 2 household budget constraint. We summarize our findings as follows: Proposition 3.6: Suppose (3.8) binds in equilibrium so that deposits are strictly below their first-best level in a financial equilibrium. Then, a taxfinanced transfer of net worth to bankers at a level defined in (3.15) ensures that the first-best allocations are supported as a financial equilibrium.

4. Moral Hazard II: Unobserved Banker Effort The basic framework of the model used here is similar to the one in the previous section. The difference lies in the source of moral hazard. We assume that bankers, to make a high return for their depositors, must exert an unobserved and costly effort. As in the case of the model in the previous section, the model used here can articulate the idea that the banking system supported efficient allocations prior to 2007, but then became dysfunctional as a consequence of a fall in bank net worth. As in the previous section, the fall in net worth pushes the economy against a nonlinearity, which causes an increase in interest rate spreads and a fall in intermediation and in the activities that intermediation supports. Despite the similarities, there are some important differences between the models in terms of their implications for policy. For example, the model used here implies that equity injections into banks during a crisis have no impact on equilibrium allocations. The model of the previous section implies that injections of bank equity can move the economy to the efficient allocations. In addition, we use the model of this section to study a broader range of policy interventions. We consider the effects of government bailouts of the creditors of banks whose assets perform poorly. The model is also useful for thinking about the benefits of imposing leverage restrictions on banks. The following section provides an intuitive summary of the analysis. After that comes the formal presentation.

4.1. Overview There are two periods. There are a large number of households. Each household has many bankers and workers. Bankers are endowed in the first period

248

Lawrence J. Christiano and Daisuke Ikeda

with their own net worth, and they combine this with deposits to acquire securities from firms.20 There are a large number of firms, each having access to one investment project. The investment project available to some firms is a good one in that it has a high (fixed) gross rate of return. If these firms invest one unit of goods in period 1, they are able to produce Rg goods in period 2. The investment project available to other firms is bad, and we denote the gross rate of return on these investment projects by Rb , where Rb < Rg . The rates of return Rg and Rb are exogenous and technologically determined. Empirically, we observe that some banks enjoy higher profits than others, and we interpret this as reflecting that banks cannot hold a fully diversified portfolio of assets. This could be because there are many different types of investment projects – differentiated according to industry, geographic location, etc. – and there are gains to specializing in the identification of good projects of a particular type. In the model, these observations are captured by the assumption that banks can purchase the securities of at most one firm. Similarly, a firm can issue securities to at most one bank. Production for a firm is costless, and the rate of return on bank securities is identical to the rate of return on the underlying investment.21 The task of a banker is to exert an unobserved and costly effort e to identify a firm with a good project. The ex post rate of return on the banker’s securities is observed, but this does not reveal the banker’s effort. This is because e affects only the probability p(e) that a banker identifies a good firm. We define the efficient levels of effort and of intermediation as those that occur in competitive markets in the special case that the efforts exerted by bankers are fully observed. For a banker to have the incentive to exert the efficient level of effort when effort is not observed requires that he or she receive a reward that is linked in the right way to the performance of the securities. Let Rbd and Rgd denote the interest rate on bank deposits when the bank’s securities pay Rb and Rg , respectively. We show that a banker sets effort to the efficient level when Rbd = Rgd , that is, when the cost of funds is independent of the performance of the securities. We characterize the situation in which Rbd = Rgd as one in which the banker’s creditors (i.e., the depositors) do not share in the losses when a banker’s securities do not perform well. The banker exerts the efficient level of effort when Rbd = Rgd because the banker fully internalizes the marginal benefit of increased effort. For the arrangement Rbd = Rgd to be feasible, it is necessary that the banker have a sufficiently large amount of net worth. Otherwise, the banker would not have enough funds to pay depositors in the probability 1 − p(e) event that the banker’s loan turns out to be bad.22 When net worth is too low in

Government Policy, Credit Markets, and Economic Activity

249

this sense, then a bank’s depositors share in the loss that occurs when their bank’s securities generate a bad return. In this case, depositors must receive a relatively high return, Rgd > Rbd , in the good state as compensation. But, with this cross-state pattern in deposit rates, the banker does not fully capture the marginal product of increased effort. Thus, when banker net worth is not sufficiently high to permit an uncontingent deposit rate, the banker’s incentive to exert effort is reduced. This reduced effort has a consequence that relatively more low-quality projects are funded. As a result, the overall rate of return on deposits falls and so the quantity of deposits falls too. With the fall in deposits, intermediation and investment are reduced. We now briefly discuss the concept of the “interest rate spread.” We can loosely think of the bad state as a bankruptcy state, a state that occurs with relatively low probability. For the purpose of defining the interest rate spread, we think of the “interest rate” paid by a bank on its source of funds as the rate it pays, Rgd , when the good state is realized. This notion of the interest rate is similar to that of the face value of a bond, which specifies what the holder receives as long as nothing goes wrong with the issuing firm. Households are the ultimate source of funds for banks, and they receive an interest rate R that is risk free. This is so because the representative household is perfectly diversified across banks (he or she accomplishes this by using a mutual fund) and so he or she receives the average rate of return across all banks. With these considerations in mind, we define the interest rate spread as follows: Rgd − R.

(4.1)

When bank net worth is sufficiently high, then Rgd − R = 0, so that the interest rate spread is zero. When net worth falls enough, then Rbd must be low in the bad state and thus Rgd must be relatively high in the good state. As a result, the interest rate spread is positive when bank net worth is low. In sum, when bank net worth is high (we refer to this as normal times), then the interest rate spread is zero and effort and deposits are at their efficient levels. When bank net worth is low (a crisis), then there is a positive interest rate spread and deposits are below their efficient levels. In this sense, the financial system is dysfunctional when net worth is sufficiently low. From this perspective, the model implications are qualitatively similar to those of the model in the previous section. Still, the economics of the two models differ. For example, in the model considered here, the interest rate spread compensates for the low returns paid by banks with bad investments. In principle, one could perform an empirical study to measure the bank losses that are reflected in the

250

Lawrence J. Christiano and Daisuke Ikeda

high-risk spread. In the model of Section 3, the interest rate spread reflects a fear of out-of-equilibrium misbehavior by banks. As such, the fear is about something that does not actually happen. The two models also differ in terms of their implications for policy. In the model of the previous section, equity injections have no effect in normal times and they improve the efficiency of the economy in a crisis. In the model here, equity injections in normal times are counterproductive because they reduce bankers’ incentives to exert effort. The intuition is simple. We treat an equity injection as a “loan” from the government that must be repaid according to the actual return that the bank receives as a consequence of the government loan. The direct impact of this sort of loan on the bank is nil because it generates zero net cash flow regardless of whether the bank identifies a good or bad firm. However, there is a general equilibrium effect that matters. From the point of view of the household, an equity injection corresponds to a tax hike in the first period, followed by a tax reduction in the second period. Because this pattern of taxes satisfies part of the household’s desire to save, the household responds by reducing his or her own deposits. With fewer deposits, the banker has less incentive to exert effort. With less effort, the average quality of bank securities falls. This produces a fall in the risk-free interest rate paid to households and causes them to save less. The net effect is that intermediation falls below its ideal level. It turns out that in a crisis, an equity injection has no effect in the model. This is because in a crisis there is an additional positive effect from equity investments that cancels the negative effects in normal times that were discussed in the previous paragraph. Recall, the definition of a crisis time is that net worth is too low to permit a state-non-contingent interest rate on deposits. When household deposits with banks are reduced in response to an equity injection, it becomes possible to reduce the degree of state contingency in deposit rates. This is because, with lower deposits, the amount of money owed by banks in the bad state is smaller and more likely to be manageable with bank net worth. The reduced state contingency in deposit rates improves the incentive of banks to exert effort. This positive effect exactly cancels the negative effects that occur in a normal time. We also investigate other policies. For example, we study the effects of placing tax-financed government deposits in banks during a crisis. Such a policy has no effect because, consistent with the Barro-Wallace proposition, households respond by reducing their deposits by the same amount. Subsidizing banks’ cost of funds in a crisis is helpful, because this policy improves the likelihood that a bank can cover losses with its own net worth. Bailing out the creditors of banks whose loans perform badly is also welfare

Government Policy, Credit Markets, and Economic Activity

251

increasing in a crisis. Finally, we find that leverage restrictions improve welfare in a crisis. The reason for this is that by forcing banks to reduce their level of deposits, a leverage restriction increases the likelihood that a bank can cover its losses with its own net worth, thus increasing its incentive to exert effort. This is welfare improving in a crisis, when banker effort is below its efficient level, absent government intervention. The following subsections present the formal description of the model and the results, respectively.

4.2. Model There are many identical households, each composed of many workers and bankers. The workers receive an endowment y in period 1, and the households allocate the endowment between period 1 consumption c and period 1 deposits in mutual funds d. All quantity variables are expressed in per-household-member terms. The gross rate of return on deposits is risk free and is denoted by R. The preferences of the representative household are as in the previous example, in (3.2). Optimality of the deposit decision is associated with the usual intertemporal Euler equation. This Euler equation and the first-period budget constraint are given by u (c) = βRu (C ),

(4.2)

c + d = y.

(4.3)

In the second period, households receive Rd and profits from their bankers π. In the interior equilibria that we study, the second-period budget constraint is satisfied as a strict equality: C = Rd + π. We impose the following restriction on the curvature parameter in the utility function [see (3.2)]: 0 < γ < 1.

(4.4)

The upper bound on γ ensures that the equilibrium response of d to R is positive, which we view as the interesting case. Bankers receive an endowment N in the first period. They combine N with deposits received from mutual funds and buy securities that finance the investment of a firm. Firms are perfectly competitive and costless to operate, so the bank receives the entire return on its firm’s investment project.

252

Lawrence J. Christiano and Daisuke Ikeda

The probability p(e) that the firm whose securities the bank buys are good is specified as follows: p(e) = a + be,

b > 0,

(4.5)

so that p (e) = b, p (e) = 0. We consider only model parameter values that imply 0 < p(e) < 1 in equilibrium. The mean m(e) and variance V (e) of a bank’s asset are given by m(e) = p(e)Rg + (1 − p(e))Rb , V (e) = p(e)(1 − p(e))(Rg − Rb )2 ,

(4.6)

respectively. Note that V  (e) = (1 − 2p(e))(Rg − Rb )2 b. This expression is negative for p(e) > 1/2. In our analysis, we assume that p(e) satisfies this condition. Thus, when bankers increase effort, the mean of the return on their securities increases and the variance decreases. Our primary interest is in the scenario with “financial frictions,” in which the mutual fund does not observe the effort e made by the banker. To this end, it is of interest to first discuss the observable-effort version of the model in which e is observed by the mutual fund. Throughout, we assume that e is observed by the banker’s own household. Absent this assumption, a banker would always set e = 0 because e is costly to the banker and because a banker’s consumption while at home is independent of the return on the banker’s portfolio.

4.3. Observable-Effort Benchmark A loan contract between a banker and a mutual fund is characterized by four numbers (d, e, Rgd , Rbd ). Here, Rgd , Rbd denote the gross returns on d paid by bankers whose firms turn out to be good and bad, respectively. All four elements of the contract are assumed to be directly verifiable to the mutual fund in the observable-effort version of the model. Throughout, we assume that sufficient sanctions exist so that verifiable deviations from a contract never occur. The representative competitive mutual fund itself takes deposits d from households and commits to paying households a gross rate of return R. The mutual fund is competitive in that it treats R as exogenous. Because the representative mutual fund is perfectly diversified, its revenues from deposits d are p(e)Rgd d + [1 − p(e)]Rbd d. The mutual fund must repay

Government Policy, Credit Markets, and Economic Activity

253

Rd to depositors, so that the profits of the mutual fund are p(e)Rgd d + [1 − p(e)]Rbd d − Rd. Because mutual funds are competitive, profits must be zero23 : p(e)Rgd d + [1 − p(e)]Rbd d = Rd.

(4.7)

We assume the banker’s only source of funds for repaying the mutual fund is the earnings on his or her investment. In each state of nature the banker must earn enough to pay his or her obligation to the mutual fund in that state of nature: Rg (N + d ) − Rgd d ≥ 0, Rb (N + d ) − Rbd d ≥ 0.

(4.8)

In practice, these constraints will either never bind or they will bind only in the bad state of nature. Thus, an additional restriction on the menu of contracts (d, e, Rgd , Rbd ) available to a bank is Rb (N + d ) − Rbd d ≥ 0. The problem of the banker is to select a contract (d, e, Rgd , Rbd ) from the menu defined by (4.7) and (4.8). A banker’s ex ante reward from a loan contract is      1 λ p(e) Rg (N + d ) − Rgd d + (1 − p(e)) Rb (N + d ) − Rbd d − e 2 , 2 (4.9) where e 2 /2 is the banker’s utility cost of expending effort and λ denotes the marginal value of consumption for the household of the banker. In addition, d denotes the deposits issued by the banker and is distinct from the deposit decision of the banker’s household. As part of the terms of the banker’s arrangement with the household, the banker is required to seek a contract that maximizes (4.9). Throughout the analysis, we assume that the banker’s household observes all the variables in (4.9) and that the household has the means to compel the banker to do what the household requires. The Lagrangian representation of the banker’s problem is      1 max λ p(e) Rg (N +d )−Rgd d +(1− p(e)) Rb (N +d )−Rbd d − e 2 d d 2 e,d,Rg ,Rb    d  d d b + μ p(e)Rg d + (1 − p(e))Rb d − Rd + ν Rb d − R (N + d ) , (4.10) where μ is the Lagrange multiplier on (4.7) and ν ≤ 0 is the Lagrange multiplier on (4.8).

254

Lawrence J. Christiano and Daisuke Ikeda

An interior equilibrium for this economy is as follows: Observable-effort equilibrium: c, C, e, d, R, λ, Rgd , Rbd such that (i) (ii) (iii) (iv) (v)

the household maximization problem is solved, mutual funds earn zero profits, the banker problem, (4.10), is solved, markets clear, c, C, d, e > 0.

We now study the properties of this equilibrium. The first-order conditions associated with the banker problem in equilibrium are     e : λp (e) Rg − Rb (N + d ) − Rgd − Rbd d   − e + μp (e) Rgd − Rbd d = 0,      d : λ p(e) Rg − Rgd + (1 − p(e)) Rb − Rbd     + μ p(e)Rgd + (1 − p(e))Rbd − R + ν Rbd − Rb = 0, Rgd : −λp(e)d + μp(e)d = 0, Rbd : −λ(1 − p(e))d + μ(1 − p(e))d + νd = 0, μ : p(e)Rgd d + (1 − p(e))Rbd d = Rd,   ν : ν Rbd d − Rb (N + d ) = 0, ν ≤ 0,

Rbd d − Rb (N + d ) ≤ 0,

where x indicates the first-order condition with respect to the variable x. Adding the Rgd and Rbd equations, we obtain μ = λ − ν.

(4.11)

Substituting (4.11) back into the Rgd equation, we find ν = 0, so that the cash constraint is nonbinding. Substituting the latter two results back into the system of equations, they reduce to e : e = λp (e)(Rg − Rb )(N + d ),

(4.12)

d : R = p(e)R + (1 − p(e))R ,

(4.13)

μ : R = p(e)Rgd + (1 − p(e))Rbd .

(4.14)

g

b

Note from (4.12) that in setting effort e, the banker looks only at the sum N + d and not at how this sum breaks down into the component reflecting

Government Policy, Credit Markets, and Economic Activity

255

the banker’s own resources N and the component reflecting the resources d supplied by the mutual fund. By committing to care for d as if these were the banker’s own funds, the banker is able to obtain better contract terms from the mutual fund. The banker is able to commit to the level of effort in (4.12) because e is observable to the mutual fund, and throughout the analysis we assume that all actions that are verifiable are enforceable. The profits π brought home by the bankers in the representative household in period 2 are     π = p(e) Rg (N + d ) − Rgd d + (1 − p(e)) Rb (N + d ) − Rbd d = RN , (4.15) using the zero profit condition of mutual funds. Thus, the representative household’s second-period budget constraint is C = R(N + d ).

(4.16)

The five equilibrium conditions, (4.12), (4.13), (4.3), (4.2), and (4.16), can be used to determine values for c, C, e, d, R. Also, λ = βu (C ).

(4.17)

Although the observable effort version of the model uniquely determines variables like c, C, d, and R, it does not uniquely determine the values of the state contingent return on deposits, Rgd , Rbd . These are restricted only by (4.14) and (4.8). For example, there is an equilibrium in which deposits have the following state contingent pattern: Rgd = Rg , Rbd = Rb . There may also be an equilibrium in which deposit rates are not state contingent, so that Rgd = Rbd = R. However, for the latter to be an equilibrium requires that N be sufficiently large. The equilibrium values of c, C, e, d, and λ are the same across all state-contingent returns on deposits that are consistent with (4.14) and (4.8).

4.4. Unobservable Effort We now suppose that the banker’s effort e is not observed by the mutual fund. Thus, whatever d, Rgd , or Rbd is specified in the contract, a banker always chooses e to maximize:      1 λ p(e) Rg (N + d ) − Rgd d + (1 − p(e)) Rb (N + d ) − Rbd d − e 2 . 2

256

Lawrence J. Christiano and Daisuke Ikeda

The first-order condition necessary for optimality is     e : e = λp (e) (Rg − Rb )(N + d ) − Rgd − Rbd d .

(4.18)

Note that Rgd > Rbd reduces the banker’s incentive to exert effort. This is because in this case the banker receives a smaller portion of the marginal increase in expected profits caused by a marginal increase in effort. Understanding that e will be selected according to (4.18), a mutual fund will offer only contracts (d, e, Rgd , Rbd ) that satisfy not just (4.8), but also (4.18). In light of the previous observations, the Lagrangian representation of the banker’s problem is   g  d  max λ p(e) R (N + d ) − Rg d e,d,Rgd ,Rbd

 1  + (1 − p(e)) Rb (N + d ) − Rbd d − e 2 2   d d + μ p(e)Rg d + (1 − p(e))Rb d − Rd      + η e − λp (e) (Rg − Rb )(N + d ) − Rgd − Rbd d   + ν Rbd d − Rb (N + d ) ,

(4.19)

where η is the Lagrange multiplier on (4.18). The equilibrium concept used here is as follows: Unobservable-effort equilibrium: c, C, e, d, R, λ, Rgd , Rbd such that (i) (ii) (iii) (iv) (v)

the household maximization problem is solved, mutual funds earn zero profits, the banker problem, (4.19), is solved, markets clear, and c, C, d, e > 0.

To understand the properties of this equilibrium, consider the first-order necessary conditions associated with the banker problem, (4.19):       e : λp (e) (Rg − Rb )(N + d ) − Rgd − Rbd d − e + μp (e) Rgd − Rbd d      + η 1 − λp (e) (Rg − Rb )(N + d ) − Rgd − Rbd d = 0,     d : 0 = λp(e) Rg − Rgd + λ(1 − p(e)) Rb − Rbd   + μ p(e)Rgd + (1 − p(e))Rbd − R

Government Policy, Credit Markets, and Economic Activity

257

     − ηλp (e) (Rg − Rb ) − Rgd − Rbd + ν Rbd − Rb , Rgd : −λp(e) + μp(e) + ηλp (e) = 0, Rbd : −λ(1 − p(e)) + μ(1 − p(e)) − ηλp (e) + ν = 0,

(4.20)

μ : R = p(e)Rgd + (1 − p(e))Rbd ,     η : e = λp (e) (Rg − Rb )(N + d ) − Rgd − Rbd d ,     ν : ν Rbd d − Rb (N + d ) = 0, ν ≤ 0, Rbd d − Rb (N + d ) ≤ 0. We refer to these equations – and their subsequent counterparts – by their names to the left of the colon. Add the Rgd and Rbd equations to obtain (4.11). After using (4.11) to substitute out for μ in (4.20), making use of (4.5), and rearranging, we obtain24   e : (λ − ν)b Rgd − Rbd d + η = 0, d : R = p(e)Rg + (1 − p(e))Rb , Rgd : ν p(e) = ηλb,

(4.21)

μ : R = p(e)Rgd + (1 − p(e))Rbd ,     η : e = λb (Rg − Rb )(N + d ) − Rgd − Rbd d ,   d   Rb d − Rb (N + d ) ≤ 0. ν : ν Rbd d − Rb (N + d ) = 0, ν ≤ 0, We distinguish two cases. Equilibrium in a normal time corresponds to the case in which N is sufficiently large that the cash constraint is nonbinding, so that ν = 0. Equilibrium in a crisis time corresponds to the case in which ν < 0. We first consider the properties of equilibrium in a normal time. Substituting ν = 0 into the Rgd equation, we deduce that in an interior equilibrium with d, λ > 0, the multiplier on the incentive constraint η is zero. With η = 0 and the fact that λ − ν > 0, the e and μ equations imply that Rgd = Rbd = R.

(4.22)

It then follows from the η equation that e : e = λb(Rg − Rb )(N + d ).

(4.23)

Equations (4.23) and the μ equation in (4.21), together with the three household equilibrium conditions (4.3), (4.2), and (4.16), represent five

258

Lawrence J. Christiano and Daisuke Ikeda

conditions. These conditions can be used to determine the following five variables: c, C, e, d, R. A notable feature of the equilibrium in a normal time is that the incentive constraint, (4.18), is nonbinding and the allocations are efficient in the sense that they coincide with the allocations in the version of the model in which effort is observable. The level of effort exerted by the banker in the unobservable effort equilibrium coincides with what it is in the observable effort equilibrium because the loan contract transfers the full marginal product of effort to the banker. This is accomplished by making the rate of interest on banker deposits not state contingent [see (4.22)]. The interest rate spread in this equilibrium [see (4.1)] is zero in a normal time. We state these results as a proposition: Proposition 4.1: When the cash constraint, (4.8), does not bind (i.e., ν = 0), then the allocations in the unobserved effort equilibrium coincide with those in the observed effort equilibrium and the interest rate spread is zero. We now turn to the case in which the cash constraint is binding, so that ν < 0 and ν : Rbd d = Rb (N + d ).

(4.24)

In this case, the observed and unobserved effort equilibria diverge, because the cash constraint never binds in the observed effort equilibrium. The Rgd equation in (4.21) implies η < 0, so that according to the e equation in (4.21), Rgd > Rbd

(4.25)

in an interior equilibrium with d > 0. It follows from the η equation in (4.21) that e < λb(Rg − Rb )(N + d ). That is, the banker in a crisis equilibrium exerts less effort, for a given N + d, than he or she does in the observed effort equilibrium. The reason is that with (4.25), the banker does not capture the full marginal return from effort. With reduced effort, equation d in (4.21) shows that equilibrium R

Government Policy, Credit Markets, and Economic Activity

259

is smaller. Given (4.4), the household equilibrium conditions, (4.3), (4.2), and (4.16), imply a lower d, reinforcing the low e. We summarize these findings in the following proposition: Proposition 4.2: When the cash constraint, (4.8), binds (i.e., ν < 0), then e, d, and R in the unobserved effort equilibrium are lower than they are in the observed effort equilibrium, and the interest rate spread is positive.

4.5. Implications for Policy In this section, we consider the impact of government deposits and equity injections into banks, and show that these are not helpful in a crisis. We then show that bank deposit rate subsidies and transfers of net worth to banks can solve the crisis completely by eliminating the interest rate spread and moving allocations to their efficient levels. Finally, we study the effects of bailing out the creditors of banks with poor-performing securities and the effects of leverage restrictions. 4.5.1. Government Deposits Into Mutual Funds Consider the case in which the government raises taxes T and deposits the proceeds in the mutual fund. The household’s period 1 budget constraint is given by c + d˜ = y,

(4.26)

where d˜ denotes d + T and d denotes deposits placed by households in the mutual fund. The intertemporal condition, (4.2), is unaffected by the change. The household’s second-period budget constraint is unaffected ˜ Similarly, the equilibrium by the change, except that d is replaced by d. conditions associated with the banker problem, (4.19), are unchanged, with ˜ In particular, if the government deposits the exception that d is replaced by d. taxpayer money into the mutual funds, taxpayers reduce their deposits by the same amount and there is no change. From the point of view of households in the economy, it is the same whether deposits are held in their capacity as taxpayers or directly in their own name. That is, the policy considered in this section does not overcome the Barro-Wallace irrelevance proposition. This conclusion makes use of the assumption we use throughout our analysis, that an equilibrium is interior. In the present context, this implies that T is

260

Lawrence J. Christiano and Daisuke Ikeda

not so large that the constraint d ≥ 0 is nonbinding. We summarize these results in the form of a proposition: Proposition 4.3: In an interior equilibrium, the level of tax-financed government deposits are irrelevant for the equilibrium levels of c, C, R, d + T , and e. 4.5.2. Equity Injections Into Banks In this section we adopt the same interpretation of equity injections as in Subsection 3.2.1. That is, the government raises taxes T and hands these over to the banks in period 1. The government requires that the banks repay the earnings they actually make on these funds in period 2. Under this policy, the expected profits of the bank are   p(e) Rg (N + T + d ) − Rgd d − Rg T   +(1 − p(e)) Rb (N + T + d ) − Rbd d − Rb T . Note that taxes enter revenues symmetrically with deposits and the bank’s own net worth. On the cost side, equity injections require that banks pay the government the actual rate of return on its securities. Thus, equity injections have no direct impact on bank profits, because they enter revenues and costs in exactly the same way. For the same reason, equity injections also do not change the banker’s cash requirement in the bad state. That is, the bank requirement that revenues be no smaller than costs is, in the presence of equity injections, Rb (N + d + T ) ≥ Rbd d + Rb T , so that T cancels from both sides and thus coincides with (4.8). We conclude that the banker’s problem, (4.19), is completely unaltered by the presence of equity injections. Now consider the household problem. The period 1 budget constraint is c + d ≤ y − T,

(4.27)

reflecting that equity injections T are financed with taxes on households. The government transfers the revenues from equity injections back to households in period 2. In this way, the period 2 household budget constraint is     C = Rd + p(e) Rg (N + d ) − Rgd d + (1 − p(e)) Rb (N + d ) − Rbd d   + p(e)Rg + (1 − p(e))Rb T = R(N + d + T ).

(4.28)

Government Policy, Credit Markets, and Economic Activity

261

The last term on the right-hand reflects that the government’s distribution of equity among banks is completely diversified. The intertemporal Euler equation, (4.2), is unchanged. From the household problem we see that an increase in T induces an equal reduction in d for a given value of R. Note that although T does not enter the banker’s problem, d does. Thus, it is possible that T has an indirect effect on the equilibrium. Consider first the case in which the cash constraint in the bad state is not binding, ν = 0. In this case, the problem solved by the banker’s contract is given by (4.19) with ν = 0, so that (4.22) and (4.23) are satisfied. In this case, increased equity injections for a given interest rate R reduce deposits and so reduce the banker’s incentives to exert effort e [see (4.23)]. This in turn produces a fall in R, so that d falls some more. Thus, d + T falls with a tax-financed equity injection in a normal time when the cash constraint is not binding. The intuition for this result is described in Section 4.1. We summarize this result in the form of a proposition: Proposition 4.4: If the cash constraint, (4.8), is not binding, then an equity injection produces a fall in effort e, the interest rate R, and total intermediation d + T. In a crisis time when the cash constraint in the bad state is binding, the fall in d + T that occurs with an equity injection is offset by a second effect. The two cancel, and so equity injections are irrelevant in a crisis. The second effect occurs because a fall in deposits d loosens the cash constraint, (4.8), in the bad state. This relaxation of (4.8) requires an increase in the rate of return on deposits for banks in the bad state. The reduction in the state contingency of deposit rates enhances bankers’ incentives to exert effort. As a result, the fraction of good projects that are identified is increased, so that the risk-free rate rises, leading to a rise in deposits. Formally, we have the following proposition. Proposition 4.5: If the cash constraint is binding, then an equity injection has no impact on consumption c, C, the interest rate R, and the volume of intermediation d + T. See Appendix A for a proof of this proposition. We summarize our two propositions as follows. In a normal time, when the cash constraint is not binding, equity injections are counterproductive, as they lead to a reduction in effort by bankers. In crisis times, an equity

262

Lawrence J. Christiano and Daisuke Ikeda

injection satisfies the Barro-Wallace irrelevance result, so that they have no impact on c, C, R, or d + T as long as the cash constraint remains binding. Once equity injections reach a sufficient scale, then the cash constraint ceases to bind and proposition 4.4 is relevant. That is, equity injections that are large enough to render the cash constraint nonbinding are counterproductive in that they reduce effort. 4.5.3. Interest Rate Subsidies and Net Worth Transfers to Banks As we have emphasized, the heart of the problem in a crisis is that banks with poor-performing securities do not have enough resources to fully absorb their losses. Equilibrium in this case requires that deposit rates covary positively with the return on the bank portfolio. But this positive covariance leads to welfare-reducing allocations by reducing banks’ incentive to exert effort. State noncontingency in the banks’ deposit rate is crucial if they are to have enough incentive to exert the efficient level of effort. This reasoning suggests two policies that can help solve the problem. First, by reducing the costs of their deposits, a tax-financed subsidy on banks’ cost of funds makes it possible for banks to cover their losses in bad states and for bank deposit rates to be state noncontingent. Second, a tax-financed transfer of equity to banks also allows them to cover their losses in bad states with state-non-contingent deposit rates. Consider first the case of interest rate subsidies. Suppose we have the allocations and returns in the observable effort equilibrium. The assumption that we are in a crisis implies that if R = Rbd = Rgd , where R solves (4.13), then the cash constraint, (4.8), is violated: Rd > Rb (N + d ).

(4.29)

(1 − τ )Rd = Rb (N + d ).

(4.30)

Let τ solve A value of τ > 0 is guaranteed to exist because the left-hand side of this expression is monotonically decreasing in τ and it is zero when τ = 1. All the equilibrium conditions associated with the banker problem [see (4.21)] are satisfied, with ν = 0. As a result, the banker exerts the level of effort that occurs in the observed effort equilibrium [see (4.23)]. The key thing is that state noncontingency of deposit rates causes the banker to exert effort as though the deposits belonged to the banker. The fact that the level of deposit rates is lower across the realized returns of its securities is irrelevant to the effort exerted by the banker.

Government Policy, Credit Markets, and Economic Activity

263

It remains only to verify that the household decisions in the observableeffort equilibrium also solve their problem in the unobservable-effort equilibrium with an interest rate subsidy. The households’ period 1 budget constraint, (4.3), is unaffected. The household’s intertemporal Euler equation, (4.2), is also not affected. The only household equilibrium condition that requires attention is the second-period budget constraint, because the tax subsidy to banks is financed by period 2 taxes, T = τ Rd, C = Rd + π − T. Bank profits π are higher under the interest rate subsidy than they are in the observable effort equilibrium. However, they are higher by exactly T. So, the assumption that the period 2 household budget constraint is satisfied in an observable-effort equilibrium implies that the allocations in that equilibrium also satisfy the preceding budget constraint with taxes. We summarize these findings as follows: Proposition 4.6: Suppose the cash constraint in an unobservable-effort equilibrium is binding. The interest rate subsidy, (4.30), financed by a period 2 tax on households, causes the allocations in the unobservable-effort equilibrium to coincide with those in the observable-effort equilibrium. The interest subsidy policy is of wider interest because it allows us to address a common view that interest rate subsidies to banks lead them to undertake excessive risk. In the environment here, an interest rate subsidy in a crisis induces bankers to exert greater effort e. This leads to a rise in the mean return on assets and a fall in their variance. Thus, this environment does not rationalize the common view about the impact of interest subsidies on risk taking by banks. We now turn to tax-financed transfers of net worth to banks. Suppose again that the cash constraint is binding in the unobservable effort equilibrium. The government raises taxes T in period 1 and transfers the proceeds to banks. If the transfer is sufficiently large, then the cash constraint in the unobservable effort equilibrium ceases to bind. To establish this result, suppose we have the allocations in the observable-effort equilibrium in hand. The assumption that we are in a crisis implies that if R = Rbd = Rgd , where R solves (4.13), then the cash constraint, (4.8), is violated, as in (4.29). We first consider the response of the observable-effort equilibrium to T > 0. Banks’ pretax net worth is N and after taxes their net worth is N + T. We conjecture, and then verify, that with T > 0, deposits decline

264

Lawrence J. Christiano and Daisuke Ikeda

one-for-one in the observable effort equilibrium and period 1 and period 2 consumption allocations do not change. Suppose that T satisfies R(d − T ) = Rb (N + d ).

(4.31)

The value of T that satisfies this equation exists and is unique because the left-hand side is monotonically decreasing and continuous in T and it is zero when T = d. According to (4.31), the cash constraint is (marginally) nonbinding. It is easily verified that the household period 1 budget constraint and Euler equations in the observable-effort equilibrium are satisfied [see (4.3) and (4.2)]. It is also easily verified that households’ second period income is invariant to T .25 Finally, the bank equilibrium conditions, (4.12), (4.13), (4.14), are easily seen to be satisfied. We conclude that we have an observable-effort equilibrium. Because in addition the cash constraint is satisfied, it follows that we have an unobserved-effort equilibrium too. We summarize our finding as follows: Proposition 4.7: Suppose the cash constraint in an unobservable-effort equilibrium is binding. The net worth subsidy, (4.31), financed by a period 1 tax on households, causes the allocations in the unobservable-effort equilibrium to coincide with those in the observable effort equilibrium. 4.5.4. Creditor Bailouts In this subsection we explore another policy that can increase welfare in a crisis. This policy subsidizes bank creditors (i.e., the mutual funds) when their portfolios perform poorly (i.e., when banks earn Rb ). This policy is helpful because it goes to the heart of the problem. The problem when net worth is too low is that creditors must share in the losses when bank portfolios perform poorly. Under these circumstances, creditors require Rgd to be high to compensate them for the losses associated with the low Rbd . This increase in Rgd − Rbd causes bankers to reduce effort below the efficient level [recall (4.18)]. By subsidizing creditors in the bad state, Rgd − Rbd is reduced and effort moves back in the direction of its efficient level. We explore the quantitative magnitude of these effects in this section. Let Rbd denote, as before, the bank’s payment in the bad state. The amount the mutual fund actually receives is (1 + τ )Rbd . We assume that the bailout τ Rbd is financed by a lump-sum tax on households in the second period. The zero-profit condition of the mutual fund is p(e)Rgd d + (1 − p(e))(1 + τ )Rbd d = Rd.

Government Policy, Credit Markets, and Economic Activity

265

With this change, the equilibrium loan contract is the (e, d, Rgd , Rbd ) that solves the following analog of (4.19): 

    1 p(e) Rg (N +d )−Rgd d +(1− p(e)) Rb (N +d )−Rbd d − e 2 2   + μ p(e)Rgd d + (1 − p(e))(1 + τ )Rbd d − Rd      + η e − λp (e) (Rg − Rb )(N + d ) − Rgd − Rbd d   (4.32) + ν Rbd d − Rb (N + d ) .

 max λ e, d, Rgd , Rbd

Note that τ enters only the zero-profit condition of mutual funds. Because τ does not explicitly enter the banks’ own profits, the incentive constraint on bank effort is not affected. For a detailed characterization of the loan contract and an algorithm for computing the equilibrium, see Section A.3 in Appendix A. We compute the socially optimal value of τ in a numerical example. The social welfare function aggregates the utility of everyone in the household: 1 u(c) + βu(C) − e 2 . 2 We do the computations for a crisis situation, one in which N is sufficiently low that ν = 0. We construct an example by first selecting an equilibrium with τ = 0 in which the cash constraint is nonbinding, that is, ν = 0. We then reduce N sufficiently so that the cash constraint is binding and we then compute equilibria for a range of values of τ. We must assign values to the following parameters: β, γ , Rg , Rb , a, b, y, N , where a and b are the parameters of p(e) [see (4.5)], and γ , β are parameters that govern household utility [see (3.2)]. We set β = 0.97, γ = 0.9, a = 0.5, and N = 1, and we set the other four parameters, Rg , Rb , b, and y to achieve R = 1/β and the following three calibration targets: p(e) = 0.99, V (e) = 0.0036,

d = 0.26, y

where V (e) denotes the variance, across banks, of returns [see (4.6)]. The equilibrium associated with this parameterization is characterized by a nonbinding cash constraint. In this equilibrium, Rbd = Rgd = R when τ = 0. We verified numerically that τ = 0 corresponds to a local maximum of the social welfare function.

266

Lawrence J. Christiano and Daisuke Ikeda Social welfare: u(c)+β u(C) - e2/2

Saving, d

20.3501

.775 .77

20.35

.765 20.35

.76 .755

20.3499

.75 20.3499

.745 .74

20.3498 0

.5

1

1.5

τ

0

2

.5

.486

1 .98 .96 .94 .92 .9 .88 .86 .84

.485 .484 .483 .482 .481 .48 .479 .5

1

τ

1.5

2

1.5

2

Rdg and Rdb

Effort, e

0

1

τ

1.5

2

Rd g

Rd

b

0

.5

1

τ

Figure 5.3. Hidden effort model properties, various τ .

We reduced the value of N to 0.70, in which case the cash constraint is binding. Figure 5.3 displays features of the equilibrium for values of τ ∈ (0, 2). The optimal value of τ is roughly 0.7282. Note that equilibrium effort e is increasing in τ. As indicated in the introduction to this section, this result reflects that Rgd − Rbd is falling in τ. The rise in equilibrium effort gives rise to an increase in the return R generated by the financial system and hence produces a rise in deposits d. 4.5.5. Leverage Restrictions In normal times, a binding leverage restriction on banks reduces welfare because the equilibrium is efficient. However, bankers make inefficiently low efforts in a crisis because their cost of funds is positively correlated with the performance of their assets. This correlation reflects that bankers’ net worth is too low for them to insulate creditors from losses when banks experience a low return Rb . Obviously, if banks had a sufficiently low level of deposits when net worth was low, then bankers’ net worth would be sufficient to cover losses. This raises the possibility that leverage restrictions may be welfare improving when net worth is low. However, recall that bank incentives are a

Government Policy, Credit Markets, and Economic Activity

267

function not only of Rgd − Rbd , but also of the level of deposits d [see (4.18)]. So it is not so obvious, ex ante, that leverage restrictions are desirable in a crisis. For this reason, we investigate the desirability of leverage restrictions in a numerical example. In the example, we use the same parameter values as the ones in the previous subsection. We find that leverage restrictions indeed are desirable in crisis times. We suppose that the government imposes a restriction on the equilibrium contract, which prohibits banks from exceeding a specified level of leverage L: N +d ≤ L. N This leads to the following alternative formulation of the problem solved by the equilibrium contract:   g   b  d d  max λ p(e) R (N + d ) − Rg d + (1 − p(e)) R (N + d ) − Rb d e, d, Rgd , Rbd

1 − e2 2  + μ p(e)Rgd d + (1 − p(e))(1 + τ )Rbd d − Rd       + η e − λp (e) Rg − Rb (N + d ) − Rgd − Rbd d     + ν Rbd d − Rb (N + d ) + δ LN − (N + d ) , where δ ≥ 0 is the multiplier on the leverage constraint. We assume that the last two constraints are binding, so that δ > 0, ν < 0. The nine panels in Figure 5.4 display selected characteristics of the equilibrium for a range of values of L and for two values of the bailout rate τ. The two values of τ are τ = 0 and τ = 0.7282, which are its optimal values when there are no leverage restrictions. When τ = 0 and 0.7282, leverages in the absence of leverage restrictions are 2.0453 and 2.0684, respectively. The highest value of L reported in Figure 5.4 is 2.0453. Consider the case τ = 0 first. Note from Figure 5.4(b) that social welfare initially rises as L is reduced from L = 2.0453. The optimal value of L is 1.9980. This represents a 2.3 percent cut in leverage, which translates into a reasonably substantial cut of roughly 5 percent in deposits d. Consistent with the intuition previously provided, the reduction in L reduces the state contingency in banks’ costs of credit Rgd − Rbd [see Figure 5.4(i)]. According to Figures 5.4(e) and (h), the fall in Rgd − Rbd results in higher effort e, despite the lower level of deposits. As a result, the leverage restriction produces an increase in the cross-sectional average return on bank portfolios [Figure 5.4(d)], as well as a fall in the cross-sectional variance V (e) in (4.6). To be

268

Lawrence J. Christiano and Daisuke Ikeda τ=Ramsey optimal tax (a) Bankers utility

τ=0 (c) Interest rate, R

(b) Social welfare

.43 .42

20.351

1

20.3505

.98

20.35

.41

.96

20.3495

.4 .39

20.349

.38

20.3485 1.92 1.94 1.96 1.98 2 Leverage

2.02 2.04

.94 .92 1.92 1.94 1.96 1.98 2 Leverage

(d) Average return on banker assets

1.92 1.94 1.96 1.98 2 Leverage

2.02 2.04

2.02 2.04

(f) Consumption in period 1

(e) Deposit, d 1.58

1.025

.72 1.56

1.02 .7

1.015

1.54 .68

1.01

1.52

1.005

.66 1.5

1 1.92 1.94 1.96 1.98 2 Leverage

2.02 2.04

1.92 1.94 1.96 1.98 2 Leverage

2.02 2.04

1.92 1.94 1.96 1.98 2 Leverage

2.02 2.04 d

d

(i) Difference between Rg and Rb

(h) Effort, e

(g) Consumption in period 2

.15

1.44

.53

1.43

.52

1.42

.1

.51

1.41

.5

.05

1.4 .49 1.39

0

.48 1.92 1.94 1.96 1.98 2 Leverage

2.02 2.04

1.92 1.94 1.96 1.98 2 Leverage

2.02 2.04

1.92 1.94 1.96 1.98 2 Leverage

2.02 2.04

Figure 5.4. Hidden effort model properties for various restrictions and bailout rates, τ .

consistent with clearing in the market for deposits, the deposit rate R must fall as the leverage restriction becomes more binding [see Figure 5.4(c)]. With the fall in the deposit rate and the increase in the average return on assets, there is a rise in the net profits earned by banks on deposits. In the absence of government intervention, competition drives these profits to zero.26 In effect, the government reduction in L causes the banking sector to behave as a monopsonist. Restricting L raises banker utility, (4.9), according to Figure 5.4(a). We now turn to the case τ = 0.7282. The results suggest that bailouts are, to some extent, a substitute for leverage restrictions. To see this, note that when τ is positive, then the optimal level of leverage is raised. This property of the model starkly contradicts conventional wisdom, which maintains that leverage restrictions are required to undo the bad side effects of bailout commitments. The conventional wisdom on leverage can perhaps be paraphrased as follows: “bailouts reduce the incentive for creditors to play a socially important role in monitoring bankers, and this leads bankers to choose overly risky

Government Policy, Credit Markets, and Economic Activity

269

portfolios.” In our model, creditors have no ability to monitor bankers and so this monitoring channel is not present. However, from comparing the unobservable- and observed-effort versions of our model, we can conjecture what would happen if we modified our model so that creditors could decide, at a cost, whether and how much to monitor banks. Recall that in the observed effort version of our model studied in Section 4.3 creditors perfectly monitor the activities of the banker. In that model, even if net worth is so low that creditors must share in banker losses, the effort level of bankers is efficient. Thus, suppose net worth is low, so that the observableand unobservable-effort equilibria differ. Suppose further that the economy is repeated twice, with the observable-effort equilibrium occurring at a first date and the unobservable-effort equilibrium occurring at the second date. Loosely, one can interpret this two-date economy (each date has two subperiods) as one in which creditors monitor in the first date but do not monitor in the second date. In this model, the effort level of bankers is inefficiently low in the second date (recall proposition 4.2) and the crosssectional variance of their portfolios increases as a result (see (4.6)). This reasoning suggests to us that our model would be consistent with the conventional wisdom if creditor monitoring of bankers were endogenized. Of course, an important empirical question is whether in fact creditors do have the ability to monitor banks apart from observing the performance of banker securities. The other results corresponding to the case τ = 0.7282 are consistent with the idea that leverage complements bailouts in this model. For example, at every level of L, banker effort is higher with τ > 0 than with τ = 0. Finally, note that the response of d [Figure 5.4(e)] and c [Figure 5.4(f)] are invariant to τ. This is because net worth and y are fixed in the figure and in this case leverage immediately determines d and c.

5. Adverse Selection We consider an environment that is similar to the one in the previous sections, except that the friction now is adverse selection. As in the previous section, we capture the notion that banks do not hold a diversified portfolio of assets with the assumption that each bank can acquire the securities of at most one firm. Each firm in the economy has access to an investment project that requires a fixed input of resources to operate. Firms have no resources of their own and must rely on funding from a bank. Moreover, a firm can have a relationship with at most one bank.27 A firm earns no rent from its investment project and turns over all revenues to the bank that holds its securities. Because a bank’s own net worth is not sufficient

270

Lawrence J. Christiano and Daisuke Ikeda

to finance the investment project of the firm whose securities it purchases, banks must also obtain deposits. Banks obtain deposits from mutual funds. Mutual funds are competitive and each one is perfectly diversified across banks. Finally, mutual funds are completely financed by risk-free deposits from households. Because firm investment projects differ according to how risky they are, it follows that the asset side of the balance sheets of different banks differ in terms of their risk. We assume that the mutual funds that lend to banks cannot differentiate the high- and low-risk banks. To compensate for losses from deposits in the riskier banks the interest rate spread – the difference between the rate paid by banks to mutual funds and the rate paid by mutual funds on their risk-free deposits – must be positive.28 The distortions associated with the interest rate spread imply that intermediation and investment are below their efficient levels. A drop in bank net worth aggravates the distortions because banks become more dependent on external finance. We insert the banks and mutual funds into the type of general equilibrium environment considered in the previous sections of this chapter. When bank net worth falls, interest rate spreads jump and intermediation and investment drop. In this way the environment rationalizes the type of observations that motivated this chapter. Consistent with the analysis of Mankiw (1986) and Bernanke and Gertler (1990), who consider a similar environment, we find that a subsidy to banks’ cost of funds can ameliorate the problem.29 Indeed, a suitable choice of the interest rate subsidy can make the market allocations coincide with the firstbest efficient allocations. This is so, even though the subsidy policy does not require observing the riskiness of individual bank portfolios whereas our efficient allocations are those chosen by a benevolent planner who does observe those risks. A subsidy to banks, by reducing their dependence on external finance, can also improve allocations. We consider government deposits in banks, but these do not overcome the Barro-Wallace irrelevance result. That is, they have no effect on the allocations. Finally, we show that a tax-financed transfer of net worth to banks moves the allocations closer to first-best. The gains from doing this are greater in a crisis, because a decline in bank net worth increases the gap between equilibrium and first-best efficient allocations.

5.1. Model The economy is populated by many large and identical households. The representative household has a unit measure of members composed of workers

Government Policy, Credit Markets, and Economic Activity

271

and bankers. The measure of bankers is e < 1.30 All these agents receive perfect consumption insurance from households. Workers and bankers receive endowments of y and N , respectively, at the start of the first period. Here, y > 0 is measured in household per capita terms. We find it convenient to measure N < 1 in banker per capita terms. Thus, in household per capita terms the quantity of banker net worth is eN. The conditions that characterize household optimization are as in the other parts of this chapter but are reproduced here for convenience: c + d = y, c

−γ

= βRC

−γ

,

(5.1) γ > 0,

C = Rd + eπ.

(5.2) (5.3)

Here, c denotes first-period household consumption, d denotes deposits, and C denotes second-period consumption. These three variables are measured in household per capita terms. The object π denotes earnings, in banker per capita terms, brought home in period 2 by bankers. Finally, R denotes the gross rate of interest on household deposits in mutual funds. We now discuss the problems of firms, bankers, and mutual funds.31 Each banker meets a firm with an investment project characterized by two parameters, θ > 0 and p ∈ [0, 1], which are drawn independently from the cumulative distribution function (cdf), F (θ, p). These parameters are subsequently explained. The banker must then select between one of two options. The net worth can be deposited in a mutual fund and earn RN. Alternatively, the net worth can be combined with loans obtained from a mutual fund and securities can be purchased from the firm with which the banker is paired. From here on, we simplify the language by pretending that the investment project operated by a bank’s firm is operated directly by the bank. A banker’s realized values of θ and p are known only to the bank and to the household to which the bank belongs. In particular, the mutual fund from which the banker obtains funds does not observe θ and p. The distribution F is known to all. All investment projects are indivisible and require an investment of one good in period 1. We explain the reason for our assumption that there is an upper bound on the scale of investment in the discussion of Proposition 5.3, which appears in Subsection 5.2.1. In period 2, the investment project yields θ goods with probability p and zero goods with probability 1 − p. Our analysis is greatly simplified by placing the following restriction on F : θ p = θ,

(5.4)

272

Lawrence J. Christiano and Daisuke Ikeda

where θ is a nonrandom parameter known to all. Thus, each banker’s investment project generates the same expected return, but differs in terms of riskiness. We characterize F by specifying a distribution for p and then setting θ = θ /p. We assume that p is drawn from a uniform distribution with support [0,1]. Because N < 1, a banker must obtain a loan from a mutual fund in order to operate any investment technology. We suppose that the mutual fund can observe only whether the banker’s project succeeds or fails. In case the project succeeds, the mutual fund cannot tell ex post what that project’s value of θ was. As a result, the payment made by the banker to the mutual fund can be contingent only on whether the banker’s project is successful. We denote the interest rate paid by the banker in the event that the project succeeds by r. Because the banker has no resources in the event that the project fails, the interest rate in that event must be zero. Bankers who choose not to activate their investment projects earn RN with certainty by depositing their net worth in mutual funds. For a banker who decides to operate a project with probability p earns θ − r(1 − N ) and with probability 1 − p earns nothing. It is in the household’s interest that each of his or her bankers makes the project activation decision with the objective of maximizing expected earnings. The law of large numbers and the fact that there are many bankers in each household guarantee that if each banker behaves in this way, the total resources brought home by all bankers in a family is maximized. Households are assumed to be able to compel bankers to maximize expected returns and not divert any profits because of the assumption that households observe everything (including θ and p) about their bankers. Thus, a given banker invests his or her net worth N in a project and borrows 1 − N if and only if the realized value of p satisfies θ − pr(1 − N ) ≥ RN.

(5.5)

The values of p that satisfy (5.5) are as follows: 0 ≤ p ≤ p(r ),

p(r ) ≡

θ − RN . r(1 − N )

(5.6)

The object p(r ) in (5.6) summarizes several interesting features of the equilibrium. For example, note that when r increases, bankers with higher values of p decide not to activate their investment project [i.e., p(r ) is decreasing in r]. The reason is that under our assumptions expected investment income is fixed at θ whereas bankers with high-p projects are more likely to experience success and pay r to their mutual fund. As a result, the

Government Policy, Credit Markets, and Economic Activity

273

expected return on investment is lower for bankers with less risky investment projects, that is, those with high p. Also, p(r ) is the fraction of bankers who invest:

p(r ) d p = p(r ). (5.7) 0

Here, we have used the implication of the uniform distribution that the density of bankers with each p ∈ [0, 1] is unity. Similarly, because the quantity of goods used in each investment project is unity, p(r ) also corresponds to the total quantity of goods invested by all bankers in a household. The average value of p among the bankers who invest is denoted as (r ), where  p(r ) pd p 1 = p(r ). (5.8) (r ) = 0 p(r ) 2 Expression (5.8) reflects that, among the bankers who operate their investment technologies, the density of bankers with p ∈ [0, p(r )] is 1/p(r ). Finally, we subsequently show that p(r ) is inversely proportional to the interest rate spread, the difference between the interest rate r paid by bankers with successful projects and the risk-free rate R. We restrict our attention to model parameterizations that imply the efficient allocations (see Subsection 5.2.1) and the equilibrium allocations are interior. This means the usual nonnegativity constraints on quantities and also 0 < p(r ) < 1. We now turn to the mutual funds. Because each mutual fund is fully diversified across bankers, its revenues are nonrandom. Because we also assume that mutual funds are competitive, it follows that their profits must be zero. A mutual fund’s average earnings per unit of loan is (r )r. The cost of a unit of deposits for a mutual fund is R, so that the each mutual fund’s zero-profit condition is (r )r = R.

(5.9)

Much of the economics of the model is summarized in (5.9). For example, multiplying (5.8) by r and using (5.9), we obtain a simple expression for the interest rate spread: interest rate spread =

2 r = . R p(r )

(5.10)

According to this expression, the interest rate spread is at least 2 and can be much higher. The intuition for (5.10) is simple. Suppose all bankers activated their investment project, so that p = 1. In this case, the average

274

Lawrence J. Christiano and Daisuke Ikeda

probability of success is 1/2 [see (5.8)]. With half the bankers unable to pay, the ones that do pay must pay 2R if the mutual fund is to be able to pay R to its depositors.32 Interestingly, (5.9) determines the equilibrium rate of interest R in the model. To see this, substitute (5.6) into (5.8) to obtain (r )r =

1 θ − RN = R, 2 1−N

(5.11)

where the second equality reflects (5.9). Evidently, R is determined exclusively by variables specific to the loan market and not by, for example, households’ intertemporal preferences. That the zero-profit condition is compatible with only one R is a striking result, though well known in the literature on adverse selection. To understand the result requires understanding why mutual fund revenues per loan are independent of the interest rate r that they charge on a loan. Note that a higher r implies that mutual funds earn more revenues from bankers who borrow and repay their loan. However, this positive impact on revenues is canceled by an adverse-selection effect. Recall that when a bank raises r, bankers with a high probability of repaying their loan decide to become inactive.33 As a result, the average probability that a banker repays the loan falls [see (5.6) and (5.7)]. In principle, this need not be a problem because the lower-probability bankers also enjoy a better outcome when they are successful. However, this is little comfort to the mutual funds in the model, because they must charge the same interest rate r to all borrowers. A fixed interest rate on loans prevents mutual funds from sharing in the huge payoffs experienced when low-p bankers are successful. This is why a mutual fund’s revenues are independent of r. Adverse selection also explains why the revenue function (r )r is decreasing in R. As R increases, high-p bankers switch to being inactive, and this reduces the average p among borrowers from mutual funds, reducing mutual fund revenues per unit of loan extended. Because R is determined by the zero-profit condition, in equilibrium the quantity of saving by households adjusts passively to the R that is implied by (5.11). If, for example, the supply of saving were perfectly elastic at an interest rate that is different from the one that solves (5.11), then a small perturbation in the variables (such as N ) that determine (r )r would have an enormous impact on intermediation. In a one-sector model such as ours, the notion that the supply of saving is highly inelastic seems implausible. However, in a multisector (or open-economy) version of the model, the situation would be different. Thus, suppose that the zero-profit condition in (5.9) pertained to mutual funds specializing in the supply of

Government Policy, Credit Markets, and Economic Activity

275

funds to banks in a particular sector that is small enough that it takes the economy-wide deposit rate R, as given. In this case, the supply of funds to the particular sector could be expected to be perfectly elastic at the interest rate R. If a decrease in net worth N among the bankers of the given sector drives revenues per loan [i.e., the object to the right of the first equality in (5.11)] down, then a fall in N could cause intermediation in that sector to collapse entirely. We do not explore the multisector version of our model more here, though this would clearly be of interest. Clearing in financial markets requires that the quantity of investment e p(r ) equal the quantity of household deposits d plus the quantity of net worth eN in the hands of bankers: e p(r ) = d + eN.

(5.12)

We now obtain a simplified expression for period 2 household income. Averaging earnings over all bankers, we obtain

p(r )

1   π= θ − pr(1 − N ) d p + N Rd p p(r )

0

= p(r )[θ − (r )r(1 − N )] + (1 − p(r ))N R.

(5.13)

Adding eπ to household earnings on deposits yields the equilibrium expression for total household income in the second period: Rd + e p(r )[θ − (r )r(1 − N )] + e(1 − p(r ))N R = e p(r )θ. The expression after the equality is obtained after substituting out for R and d using (5.9) and (5.12). The object e p(r )θ represents the total period 2 output from bankers’ investment projects in household per capita terms. Replacing total household income with its equilibrium value of e p(r )θ in (5.3), we obtain the household’s second-period budget constraint in equilibrium: C = e p(r )θ .

(5.14)

Consistent with Walras’s law, (5.14) is also the second-period resource constraint. We have the following definition of equilibrium: Adverse-selection equilibrium: c, C, d, r, R, π, p(r ) such that (i) c, C, d solve the household problem given R,π, (ii) mutual funds earn zero profits, and (iii) bankers maximize expected revenues.

276

Lawrence J. Christiano and Daisuke Ikeda

An equilibrium is straightforward to compute for this economy. The five equilibrium conditions, (5.1), (5.2), (5.9), (5.12), and (5.14), as well as the definitions of π and p(r ) in (5.13) [with (r ) defined in (5.8)] and (5.6), respectively, are sufficient to determine the seven equilibrium objects. Substitute (5.6) into (5.8) and solve the resulting expression for R: R=

θ . 2−N

(5.15)

Combine (5.14) and (5.2) and use (5.6): −

c = (βR)

1 γ

θ − RN θe. r(1 − N )

(5.16)

Use the latter expression and (5.12) to substitute out for c and d in (5.1). Solving the resulting expression for r, we obtain −1

(βR) γ θ + 1 r = 2eR , y + eN

(5.17)

with the understanding that R is determined by (5.15). With r in hand, c can be computed from (5.16), C from (5.2), d from (5.1), p from (5.6), and π from (5.13) and (5.8). In this way, all the equilibrium variables can be computed uniquely as long as the model parameters are such that an interior equilibrium, p(r ) < 1 and c, C > 0, exists.34 The ratio of equations (5.15) and (5.17) provides a convenient expression for the interest rate spread r/R: −1

(βR) γ θ + 1 r = 2e . R y + eN

(5.18)

According to (5.15), R falls with a decrease in N. According to (5.18), this fact alone drives the spread up. The total effect of a decrease in N on the interest rate spread also involves the denominator in (5.18), and this drives the interest rate spread up too. We summarize these results as follows: Proposition 5.1: When an interior adverse selection equilibrium exists, it is unique and characterized by (5.15), (5.16), (5.17), and the observations thereafter. The interest rate spread, given by (5.18), rises with a reduction in N.

Government Policy, Credit Markets, and Economic Activity

277

5.2. Implications for Policy Subsection 5.2.1 discusses a planner problem for our model economy. In addition, we use this subsection to explain why we assume the existence of an upper bound on the scale of bankers’ projects. In the subsequent subsections, we show that two types of subsidy schemes improve the equilibrium allocations: a tax-financed transfer of net worth to bankers and a tax subsidy to mutual funds. Tax-financed government deposits with the mutual funds do not overcome the Barro-Wallace proposition. They have no effect because they do not affect the equilibrium interest rate on bank deposits. Households respond to the increase in taxes by reducing their deposits one-for-one with the increase in taxes and government deposits. 5.2.1. Efficient Allocations We consider the allocations selected by a benevolent planner who observes a banker’s p. We use these allocations as a benchmark from which to evaluate the adverse-selection equilibrium and various policy interventions studied in the subsequent subsections. Although here we assume the planner observes each bank’s p, the policy interventions studied in subsequent subsections do not require that policymakers observe p. The planner faces the period 1 resource constraint, c + d ≤ y.

(5.19)

To describe the planner’s decisions about which and how many projects to activate and to derive the planner’s period 2 resource constraint, we find it useful to describe the model environment using a particular figure. Figure 5.5 arranges all the agents in the economy in the unit square. Each point in the square corresponds to a particular household (vertical dimension) and member of household (horizontal dimension). There is a unit measure of households and a unit measure of members of any given household. We suppose that the box is constructed in period 1, just after each banker has drawn his or her value of p. A horizontal line inside the box highlights one particular household. The points on the line to the left of e correspond to the bankers. The points to the right of e correspond to the workers. The bankers are ordered according to their value of p, from p = 0 to p = 1 passing from left to right. For any particular p ∈ [0, 1], the banker with that investment project is indicated by the point pe on the horizontal axis. Each point on a vertical line through pe corresponds to the bankers

278

Lawrence J. Christiano and Daisuke Ikeda

Household Active bankers

Workers

1 e

eN

d

Line pertains to one particular household

Inactive bankers

ep

0

1

Household members

Figure 5.5. Agents in the adverse selection model.

with the given p in the cross section of households. Each of those bankers has the value of θ that is given by (5.4). The planner must decide how many bankers in the interval 0 to e to activate. If the planner elects to activate a banker with a particular p, it instructs all the bankers in the cross section of households with that p to activate their project. The planner is indifferent about which projects (i.e., which p’s) to activate. Each project is the same to the planner because each has the same mean productivity θ, and bankers suffer no cost to activate their project. As a result, there is no loss of generality in simply assuming that the planner selects bankers with p’s extending from p = 0 to p = p for some p ≤ 1. This corresponds to the mass of bankers in the interval 0 to e p in the figure. Consider a mass of bankers on an arbitrary interval inside [0, e]. The resource cost of activating these bankers in the cross section of households is the area of the rectangle with base inside the unit square. The latter area is just itself. This reflects the assumption that there is a unit mass of households and that each project costs one unit of resources to activate. The available net worth N per banker is sufficient to operate the bankers corresponding to the interval 0 to eN . Because these resources have no alternative use, the planner applies them. Activating additional bankers is costly to the planner because this requires suppressing consumption in period 1. Suppose the planner considers activating an additional mass d of bankers. This corresponds to the bankers extending from the point eN to the

Government Policy, Credit Markets, and Economic Activity

279

point eN + d in the figure. Activating these bankers requires d resources. So, if the planner wishes to activate a measure e p of bankers, then d + eN resources are needed, subject to e p ≤ d + eN.

(5.20)

When the planner activates bankers from 0 to e p, the total amount of goods available in period 2 is e p θ . Thus the second-period resource constraint for the planner is C ≤ ep θ.

(5.21)

The planner’s problem is to solve max u(c) + βu(C ),

c, C, p, d

subject to 0 ≤ p ≤ 1, (5.19), (5.20), (5.21), and c, C ≥ 0. The unique interior solution is characterized by the first-order conditions evaluated at equality. Solving these, we obtain c=

ep =

y + eN θ, 1 βθ γ + θ

(5.22)



y + eN 1 + θ (βθ )

−

1 γ

,

(5.23)

1

C = c(βθ ) γ ,

(5.24)

with d given by solving (5.19) with equality. It is convenient to compare these allocations with the allocations in the adverse-selection equilibrium. Substituting (5.17) into (5.16) and using (5.15), we obtain that first-period consumption in the equilibrium is c=

y + eN

θ.

1

(βR) γ + θ Using (5.6) and making use of (5.15) and (5.17), we find that total resource use in the adverse-selection equilibrium is c=

y + eN 1

1 + θ (βR) γ

θ.

(5.25)

280

Lawrence J. Christiano and Daisuke Ikeda

According to (5.2), second-period consumption in equilibrium is 1

C = c(βR) γ . Evidently, the sole factor preventing the equilibrium from replicating the planner’s allocations is that the interest rate R is too low. In the adverseselection equilibrium, R is θ /(2 − N ), but the social rate of return on investment is θ. With the market sending the wrong signal to households about the return on investment, saving and investment are too low. The more severe the problem, the smaller N is. The ratio of investment in equilibrium to its first-best level is given by dividing (5.25) by (5.23): 1 + θ (θβ )

−

1 γ

−

1 γ

1 + θ (Rβ )

.

From this expression, we see that equilibrium investment falls relatively more than the first-best level of investment when N decreases [here we have used the relation between R and N in (5.15)]. We summarize the preceding results in the form of a proposition: Proposition 5.2: The equilibrium household deposit rate R is less than the social return on investment θ, and R falls with a reduction in N. Equilibrium investment falls relatively more than the first-best level of investment with a reduction in N. In our adverse-selection model we suppose that there is an upper bound on the resources that bankers can invest in their projects. In Appendix B we consider a version of the model that does not impose an upper bound on the scale of banker projects. For that version of the model we find the following proposition: Proposition 5.3: Suppose banker projects have constant returns and can be operated at any scale. If there is an equilibrium, then (i) only bankers with the lowest value of p operate their projects, (ii) the aggregate profits of these bankers is zero, and (iii) the allocations in equilibrium coincide with the first-best efficient allocations and R = θ . To help ensure the existence of an equilibrium under the assumption of proposition 5.3, we modify the distribution of p slightly by supposing that it

Government Policy, Credit Markets, and Economic Activity

281

has positive mass at the lower bound of its support and that the lower bound is a (very) small positive number. For our purposes, property (i) renders the equilibrium of this version of our model uninteresting. Also, we suspect that in reality investment projects have diminishing returns to scale. Although the diminishing returns to scale implicit in our upper bound assumption for investment projects is extreme, we find this assumption more interesting than the constant returns to scale alternative used in proposition 5.3. Intermediate scenarios are presumably also of interest, but we do not examine these here. 5.2.2. Interest Rate Subsidies According to the analysis in previous subsections, the problem with the adverse-selection equilibrium is that the deposit rate R is too low. In addition, when net worth drops, the problem is aggravated as R falls even more and investment, relative to first-best, drops (see proposition 5.2). Consistent with the empirical phenomenon we seek to understand, the interest rate spread also rises with a drop in N (see proposition 5.1). The inefficiently low deposit rate R reflects that mutual funds do not recover the full return made possible by their loans. This suggests two direct ways to repair the market mechanism: subsidize mutual fund earnings or, equivalently, their cost of funds. We consider the latter here. We show that an appropriate interest rate subsidy can make the allocations in the adverse-selection equilibrium coincide with the first-best efficient allocations. Significantly, implementation of this policy does not require that the government observe any characteristics of the banks that borrow from mutual funds. Denoting the pretax cost of funds to the mutual fund by R, the aftersubsidy cost under our policy is R/(1 + ν ) < R. We suppose that this subsidy is financed by a lump-sum tax on households in period 2 in the amount

R T = (1 − N )e p(r ) R − . 1+ν

Here, the terms in front of the square bracket represent the total amount of loans made by the mutual funds to the active banks. The mutual funds finance these loans with deposits taken from inactive bankers and households. The amount of the subsidy is R − R/(1 + ν) per unit of loans made. The household’s second-period budget constraint, (5.3), is replaced by C = Rd + eπ − T.

282

Lawrence J. Christiano and Daisuke Ikeda

Repeating the substitutions leading up to (5.14), taking account of the modified second-period budget constraint of the household, we find that (5.14) continues to hold. The impact of the tax subsidy on the equilibrium value of R is determined by studying the appropriately modified mutual fund zero-profit condition, (5.9): (r )r =

R . 1+ν

(5.26)

Substituting (5.6) and (5.8) into the latter expression and solving for R, we obtain R=

θ (1 + ν ) . 2 − N + Nν

Evidently, achieving R = θ requires ν = 1. Recall that the seven conditions determining c, C, d, r, R, π, and p(r ) are (5.1), (5.2), (5.9), (5.12), (5.14), (5.13), and (5.6). We have verified that (5.14) continues to hold. Apart from (5.14), the other conditions are obviously unaffected by T. The only equilibrium condition that must be adjusted is the mutual fund zero-profit condition, (5.9), which we replace by (5.26). With ν = 1, the interest rate that solves (5.26) is the efficient one, R = θ. Given that the other equations are unaffected, it follows from the discussion in the previous subsection that the efficient allocations are supported by the subsidy policy. We summarize this result in a proposition: Proposition 5.4: With an interest cost subsidy, ν = 1, the allocations in the adverse-selection equilibrium are efficient. 5.2.3. Tax-Financed Transfers to Bankers Here, we consider a policy of raising lump-sum taxes T on households in period 1 and transferring T /e to each banker. By setting T = e(1 − N ),

(5.27)

the transfer ensures that each banker has enough funds to fully finance an investment project. With this policy the financial frictions are completely circumvented. To see this, note that the value of R that satisfies the mutual fund zero-profit condition, (5.9), is still the one in (5.15), except that N is replaced by the posttransfer level of banker’s net worth. Because that is unity under the tax-transfer scheme, we have that R = θ , its value in the

Government Policy, Credit Markets, and Economic Activity

283

efficient equilibrium. It is also straightforward to verify that c, C, and p satisfy (5.22), (5.23), and (5.24). We have assumed that model parameters imply that p ≤ 1. When p < 1, then d < 0. That is, in this case some of the net worth transferred to bankers is recycled back to households through the loan market. We summarize these results as follows: Proposition 5.5: Under the tax-transfer scheme in (tax), the allocations in the adverse-selection equilibrium are efficient. A problem with the tax-transfer scheme considered in this section is that it presses hard on a feature of the model in which we have little confidence. In particular, we assume that it is known how much net worth each banker has and how much he or she needs for his or her investment project. In practice, these assumptions may be difficult to assess. In addition, our environment abstracts from any problems associated with distributing wealth between bankers and other agents. These would have to be considered in case such a policy were actually implemented. 5.2.4. Tax-Financed Government Deposits in Banks Suppose the government levies a lump-sum tax T on households in period 1, deposits the proceeds in a bank, and then transfers the earnings RT on the deposits to households in period 2. It is easy to see that this has no impact on allocations, as long as the policy does not drive the economy into a boundary. The equations that characterize an interior equilibrium are (5.1), (5.2), (5.9), (5.12), (5.14), (5.13), (5.6), and (5.8) (see the discussion after the definition of equilibrium in Section 5.1). The new policy simply requires replacing d with d + T in (5.1), (5.9), and (5.12). If T is increased, then d falls by the same amount. Total investment in period 1, consumption in the two periods, and the interest rate R do not change. Although a drop in net worth makes the banking system more dysfunctional, a policy of tax-financed loans to banks has no impact.

6. Asymmetric Information and Monitoring Costs This section presents a version of the model in the previous sections, in which the financial friction on the liability side of banks’ balance sheets stems from an asymmetric information problem similar to the one studied in BGG.35 As in the previous sections, we assume the trigger for the crisis is a fall

284

Lawrence J. Christiano and Daisuke Ikeda

in the net worth of banks. Our environment is well suited to contemplating the effects of an increase in a particular type of microeconomic uncertainty, and so we consider this as an additional trigger for the crisis. Like the model in the previous section (though unlike our first two models), the model analyzed here does not display the nonlinearity of a sharp dichotomy between normal and crisis times. This is because equilibrium conditions of the model do not include equations that are satisfied as strict equalities for some values of the state and strict inequalities for other values of the state. As a result, we simply define a normal time as one in which banking net worth is high and a crisis time as one in which net worth is substantially lower. We now provide a sketch of the model, which most closely resembles the setup in the previous section. There are two periods. A large number of workers and bankers with identical utility functions live in a representative, competitive household. Workers have an exogenous endowment y of income in the first period, and bankers possess N units of net worth. All variables are expressed in household per capita units. Households allocate y to firstperiod household consumption and to deposits in identical, competitive mutual funds. Each banker combines his or her net worth with a loan from a mutual fund to acquire the securities of one firm. The firm operates a technology perturbed by an idiosyncratic technology shock to build capital in period 1 and uses the capital in period 2 to produce goods. Because we assume markets are competitive and because firms contribute nothing themselves to building capital and selling goods, they enjoy zero earnings after paying the return on the security sold to their bank. Because a bank can invest in the securities of at most one firm, the asset side of a bank’s balance sheet is risky. As in the previous section, we simplify the exposition by pretending that a firm’s activity building capital and selling goods is undertaken directly by its bank. A financial friction arises because the bank observes the realization of the idiosyncratic technology shock on the asset side of its balance sheet, whereas its mutual fund can observe the technology shock only by paying a monitoring cost. The loan received by banks from their mutual fund comes in the form of a standard debt contract. The contract specifies a loan amount and an interest rate. Banks with a sufficiently low realization of their idiosyncratic productivity shock are not able to repay their loan, and these banks must transfer whatever assets they have to their mutual fund after being monitored. The measure of microeconomic uncertainty referred to in the opening paragraph pertains to the variance across banks in their idiosyncratic technology shock. We consider a characterization of

Government Policy, Credit Markets, and Economic Activity

285

the financial crisis that involves just a drop in bank net worth as one that also involves an increase in the cross-sectional variance of the technology shock. Household consumption in the second period is financed out of income on period 1 deposits as well as bank profits. The model implies that a fall in bank net worth causes interest rate spreads – the cost of funds to banks minus the risk-free rate – to rise and investment to fall. We consider various policy responses. We find that a policy that subsidizes the cost of funds to banks is welfare improving in both normal and crisis times. Moreover, the optimal value of the subsidy is greater in crisis times, so that the model suggests a more aggressive policy stance then. We also find that, absent government intervention, bank leverage is too low. In this sense, the model does not include the kind of features that rationalize the current interest in imposing leverage restrictions on economic agents. The reason for our “underborrowing” result is that the marginal return on loans by mutual funds to banks is higher than the average return, whereas the economics of the model implies that bank creditors focus on the average return. Next, we show that the Barro-Wallace irrelevance result applies for loans made by the government to banks. We also show that a sufficiently large tax-financed transfer of net worth to bankers allows the economy to support the first-best equilibrium outcomes. Because the model abstracts from the income distribution consequences of this type of policy, we think of the result only as illustrating the logic of the model. The latter result is reported in proposition C.4 in Appendix C.2.6. Although the model framework into which we insert the BGG-type financial frictions in this section has the virtue of consistency with the other models in the paper, it has one potential drawback. In our framework, the price of capital is always unity, whereas in the literature (see, e.g., BGG or Christiano, Motto, & Rostagno, 2003, 2010, 2011), the price of capital is endogenous. Moreover, the recent literature on pecuniary externalities (see Bianchi, 2011; Korinek, 2011; Lorenzoni, 2008; Mendoza, 2010) raises the possibility that there is overborrowing when banker net worth is in part a function of the market price of an asset like capital (see, e.g., Bianchi, 2011; Mendoza, 2010). These findings motivate us to investigate the robustness of our underborrowing result to endogenizing the price of capital the way it is done in BGG. We do so by introducing curvature into the technology for converting consumption goods into capital. Because capital is the only source of net worth for bankers, the change introduces the type of pecuniary externality studied in the literature. We display numerical results that suggest that our underborrowing result in fact is robust to this change.

286

Lawrence J. Christiano and Daisuke Ikeda

6.1. Banks and Mutual Funds The typical banker takes a net worth N and approaches a mutual fund for a loan B. The bank combines its net worth and the loan to produce output in period 2, using the following production function: ω(N + B)r k ,

(6.1)

where r k is a fixed parameter of technology. Also, ω is an idiosyncratic productivity shock,

∞ dF (ω) = 1, (6.2) ω ∼ F, 0

where F is the cdf of ω. Before the realization of the bank’s productivity shock, the bank and its mutual fund have the same information about ω. Both know the shock will be drawn from F. After the realization of the shock, the bank and the mutual fund are asymmetrically informed. The bank observes the realization of its ω, but the bank’s mutual fund can observe the shock only by paying a monitoring cost. Townsend (1979) showed that under these circumstances a “standard debt contract” works well. Under such a contract, the bank pays the mutual fund an amount ZB in period 2 if it is able to do so. Given F , some banks experience such a low ω that they are not able to repay ZB. Under a standard debt contract, those banks are “bankrupt.” The mutual fund associated with such a bank verifies bankruptcy by undertaking costly monitoring, μω(N + B)r k , where μ > 0 is a parameter. A bankrupt bank transfers everything it has, ω(N + B)r k , to its mutual fund. The cutoff level of productivity ω that separates the bankrupt and nonbankrupt banks is defined by ω(N + B)r k = ZB.

(6.3)

According to (6.3), ω=

ZB = (N + B)r k

Z NB

(N +B) k r N

=

Z L−1 , rk L

where L denotes the leverage of the banker, L≡

N +B . N

(6.4)

Government Policy, Credit Markets, and Economic Activity

287

Evidently, as L → ∞, ω converges to a constant, that is, the value of ω that a bank with no net worth needs to be able to pay back its debt. From the perspective of period 1, an individual bank’s expected profits π in period 2 are given by  ∞ 

∞   k k π= ω(N + B)r − ZB dF (ω) = N Lr [ω − ω]dF (ω) ω

ω

(6.5) using (6.3). Expression (6.5) is written in compact notation as follows:

∞ k π = N Lr (6.6) [ω − ω] dF (ω) = N Lr k (1 − (ω)), ω

where



ω

(ω) ≡ G(ω) + ω [1 − F (ω)] , G(ω) ≡

ωdF (ω).

(6.7)

0

The bank maximizes (6.6) and remits all its profits to its household. The banker does so in exchange for perfect consumption insurance. It is in the interest of the household that each of its banks maximize expected returns because, by the law of large numbers, this implies that bankers as a group maximize the total resources brought home to the household in period 2. Households can observe everything about their own banks (including ω), and we make the (standard) assumption that what is observable is enforceable. That is, the household has the means to make sure that each of its bankers actually maximizes (6.6) and does not, for example, divert any proceeds toward private consumption. From (6.6) and the observation about ω for large L, we see that the banker’s objective is unbounded above in L for any given Z and r k . As a result, we cannot use the classic Marshallian demand and supply paradigm in which the banker takes Z as given and chooses a loan amount B. To describe our market arrangement, we must first discuss the situation of the banks’ creditors, mutual funds. There are a large number of identical and competitive mutual funds, each of which makes loans to banks and takes deposits from households. Because mutual funds are diversified, there is no risk on the asset side of their balance sheet. Although a mutual fund does not know whether any particular bank will fully repay its loan, the mutual fund knows exactly how much it will receive from its banks as a group. Because there is no risk on the asset side of the balance sheet, it is feasible for mutual funds to commit in period 1 to paying households a fixed and certain gross rate of interest R

288

Lawrence J. Christiano and Daisuke Ikeda

on their deposits in period 2. Because mutual funds are competitive, they take R as given. A mutual fund that makes size B loans to each of a large number of banks earns the following per bank:

ω

[1 − F (ω)] ZB + (1 − μ)

ωdF (ω)r k (N + B).

0

Here, the term before the plus sign corresponds to the revenues received from banks that are not bankrupt, that is, those with ω ≥ ω. The term after the plus sign indicates receipts, net of monitoring costs, received from banks that cannot fully repay their loan. Because the cost of funds is RB, the mutual fund’s zero-profit condition is [using (6.3)]

[1 − F (ω)] ω(N + B)r + (1 − μ) k

ω

ωdF (ω)r k (N + B) = RB,

0

or [(ω) − μG(ω)]

r k (N + B) = R. B

(6.8)

An important feature of this environment is that the interest rate paid to households is proportional to the average return, r k (N + B)/B, on loans and not, say, to the marginal return. We subsequently discuss the policy implications of this feature, evident from (6.8). Mutual funds are indifferent between loan contracts, as long as they satisfy the preceding zero-profit condition. Expression (6.8) motivates the market arrangement that we adopt. Let the combinations of ω and B that satisfy (6.8) define a “menu” of loan contracts that is available to banks for a given R.36 It is convenient to express this menu in terms of L and ω. Rewriting (6.8), we obtain L=

1 1−

rk R

[(ω) − μG(ω)]

.

(6.9)

Banks take N , R, and r k as given and select the contract (L, ω) from this menu, which maximizes expected profits, (6.6). Using (6.9) to substitute out for L in the banker’s objective, the problem reduces to one of choosing ω to maximize N rk

1 − (ω) 1−

rk R

[(ω) − μG(ω)]

.

Government Policy, Credit Markets, and Economic Activity

The first-order necessary condition for optimization problem is   rk 1 − F (ω) − μωF  (ω) 1 − F (ω) R = , k 1 − (ω) 1 − r [(ω) − μG(ω)]

289

(6.10)

R

which can be solved for ω given R. Given the solution for ω, L solves (6.9) and Z solves (6.3). A notable feature of the contract is that L and Z are independent of net worth N. That is, if banks had different levels of net worth, the theory as stated predicts that each bank in the cross section receives a loan contract specifying the same leverage and rate of interest. This feature of the model reflects the assumption that all banks draw ω from the same distribution, F. In a more realistic setting, different banks would draw from different F ’s and they would receive different debt contracts. 37

6.2. Households and Government In period 1 the household budget constraint is c + B ≤ y,

(6.11)

where c, B, and y denote consumption, deposits in mutual funds, and an endowment of output y, respectively. Expression (6.11) is also the period 1 resource constraint. The second-period budget constraint of the household is C ≤ (1 + τ )RB + π − T , where C denotes period 2 consumption, T denotes lump-sum taxes, τ denotes a subsidy on household saving, and π denotes the profits brought home by bankers.38 Households maximize utility, u(c) + βu(C ),

(6.12)

subject to their periods 1 and 2 budget constraints. In practice, we assume that u(c) =

c 1−α , 1−α

α > 0.

The government’s budget constraint is T = τ RB.

(6.13)

290

Lawrence J. Christiano and Daisuke Ikeda

The second-period resource constraint is obtained from the household budget constraint by substituting out for π from (6.6), RB from (6.8), and T from (6.13), to obtain the second-period resource constraint: C ≤ r k (N + B) [1 − μG(ω)] .

(6.14)

According to (6.14), period 2 consumption is no greater than total output, net of the output used up in monitoring by mutual funds.

6.3. Equilibrium We define an equilibrium as follows: Definition 6.1: For given τ , a private sector equilibrium is (C, c, R, ω, B, T ) such that (i) (ii) (iii) (iv) (v)

the household problem is solved (see Section 6.2), the problem of the bank is solved (see Section 6.1), mutual fund profits are zero (see Section 6.1), the government budget constraint is satisfied (see Section 6.2), and the first- and second-period resource constraints are satisfied.

For convenience, we collect the equations that characterize a private sector equilibrium here: Equation number Equation

Economic description 1 τ ]R) α

(6.10)

C = c (β[1 + C = r k [N + B][1 − μG(ω)] c+B =y rk 1−F (ω)−μωF  (ω)] 1−F (ω) R[ = 1−(ω) rk

(6.9)

N +B N

(6.14) (6.11)

1−

=

1−

R [(ω)−μG(ω)] 1

rk R [(ω)−μG(ω)]

household first-order condition period 2 resource constraint period 1 resource constraint contract efficiency condition mutual fund zero-profit condition (6.15)

These equations represent five equations in five private sector equilibrium objects: C, c, R, ω, B.

(6.16)

The equilibrium value of T can be backed out by imposing either the household or government budget constraint. Note too that the rate of interest on banks Z is determined from (6.3).

Government Policy, Credit Markets, and Economic Activity

291

6.4. Implications for Policy The first part of this section shows that the equilibrium in our economy is characterized by too little borrowing, so that subsidizing the cost of funds to banks is welfare improving. We then show that a policy of extending loans directly to banks fails to overcome the Barro-Wallace irrelevance result and so has no impact. 6.4.1. Subsidizing the Cost of Funds to Banks and Leverage Restrictions Interest rate subsidies are desirable from a welfare point of view because they correct a particular inefficiency in the model economy. Households make their deposit decision by treating R as the marginal return on a deposit. However, the structure of financial markets is such that R corresponds to the average, not the marginal, return on loans to banks. Not surprisingly, a planner prefers that household deposit decisions be made based on the marginal return. We show that in our environment, the marginal return on loans is higher than the average return, so that the market signal received by the households, R, does not provide them with an appropriately strong incentive to save. An interest rate subsidy corrects this problem. In the second subsection we turn to quantitative simulations. In view of the results in the previous paragraph, it is perhaps not surprising that restrictions on leverage in the model reduce welfare. We find that the market inefficiency in the model – the excess of the marginal return on loans over the average return – increases in a crisis time. As a result, the interest rate subsidy ought to be expanded substantially then. In addition, the efficient policy expands leverage by a greater percent in a crisis time than in normal times. Qualitative Analysis. A private sector equilibrium is defined conditionally on a particular value of τ. If we treat τ as an undetermined variable, then the system is underdetermined: There are many equilibria, one for each possible value of τ. The Ramsey equilibrium is defined as the best of these equilibria in terms of social welfare, (6.12). That is, the Ramsey equilibrium solves max

c, C, B, ω, R, τ

u(c) + βu(C),

subject to (6.14), (6.11), (6.10), (6.9), and the household intertemporal first-order condition. The latter is nonbinding as it can simply be used

292

Lawrence J. Christiano and Daisuke Ikeda

to define τ without placing any limitation on the maximization problem. Making use of the latter observation, and substituting out for c and C using (6.14), (6.11), we can find the Ramsey equilibrium allocations by solving max u(y − B) + βu(r k [N + B][1 − μG(ω)]), B, ω

(6.17)

subject to (6.10), (6.9). It is convenient to further simplify the problem by solving (6.9) for r k /R and using the result to substitute out for r k /R in (6.10):   L(B) − 1 1 − F (ω) = 1 − F (ω) − μωF  (ω) , (6.18) 1 − (ω) [(ω) − μG(ω)] where L(B) denotes the private sector equilibrium level of leverage as a function of B: N +B . (6.19) L(B) = N Equation (6.18) defines a mapping from B to ω. We denote this mapping by ω(B).39 When ω(B) is substituted into (6.17), the Ramsey problem reduces to40 max u(y − B) + βu(r k [N + B][1 − μG(ω(B))]), B

The first-order necessary condition for an (interior) optimum is    u (c) = r k 1 − μ G(ω(B)) + (N + B)G  (ω(B))ω (B) ≡ Rm , βu (C ) (6.20) say. The term Rm denotes the marginal social return on loans. Once the Ramsey problem is solved, the remaining objects in Ramsey equilibrium can be found as follows. The cost of funds to mutual funds, R, is obtained by computing the average return on loans: R=

r k L(B) [(ω) − μG(ω)]. L(B) − 1

Here, we have used (6.8) and (6.19). The saving subsidy τ is computed so that the household’s saving decision is based on the marginal return on loans Rm and not on the average return R: Rm = R(1 + τ ).

(6.21)

Two features of (6.20) deserve emphasis. First, r k is the return to loans in the first-best version of our economy. The first-best allocations are those

Government Policy, Credit Markets, and Economic Activity

293

that solve the problem max u(c) + βu(C )

c, C, B

subject to c + B ≤ y, C = r k [N + B]. This is the problem in which allocations are chosen by a planner who can observe each bank’s productivity shock ω. In this problem, there are no monitoring costs. The object in braces in (6.20) represents a wedge introduced by the presence of asymmetric information. A second interesting feature of (6.20) is that the solution to the Ramsey problem implies that τ > 0. This is because the intertemporal marginal rate of substitution in consumption is equated with the marginal return to loans in the Ramsey problem Rm , whereas in the private sector equilibrium with τ = 0 it is equated to the average return on loans R [recall the discussion surrounding (6.8)].41 Proposition C.1 in Section C.1 of Appendix C establishes that under a certain regularity condition, Rm > R, so that marginal return on loans exceeds the corresponding average return. Thus, by (6.21), τ > 0. To define the regularity condition, let the hazard rate be denoted as follows: h(ω) ≡

F  (ω) . 1 − F (ω)

(6.22)

The regularity condition is ωh(ω) increasing in ω.

(6.23)

BGG study (6.23) and argue that it is satisfied when F corresponds to the log-normal distribution. Although we assume μ > 0 in our model, it is interesting to consider the limiting case, μ → 0. In this case, R = r k according to (6.10). But (6.20) implies that Rm = r k when μ = 0, so we conclude that in this case the average and marginal returns coincide. That is, if μ = 0 then τ = 0 in the Ramsey equilibrium. We summarize the preceding results in the form of a proposition: Proposition 6.2: Suppose μ > 0 and condition (6.23) holds. Then, the interest rate subsidy τ is positive in a Ramsey equilibrium. This reflects that (i) the household bases his or her saving decision on the after-tax average return on deposits, (1 + τ )R, whereas the Ramsey planner wants the household to base his or her saving decision on the corresponding marginal return Rm and (ii) Rm > R. Suppose μ = 0. Then R = Rm ; τ = 0 in a Ramsey equilibrium; and the allocations in a private sector equilibrium with τ = 0 are first-best.

294

Lawrence J. Christiano and Daisuke Ikeda Table 5.1. Parameters of the asymmetric information model

β σ N

discount factor standard deviation banker’s endowment

0.97

α

0.37

μ

relative risk aversion monitoring cost

1

rk

0.2

y

return on capital household’s endowment

1.04 3.11

1

Quantitative Analysis. We construct several numerical examples to illustrate the observations in the previous subsection. We suppose that the cumulative distribution of ω, denoted by F , is log-normal. This distribution has two parameters, E log ω and σ 2 ≡ Var(log ω). The assumption Eω = 1 implies that E log ω = −σ 2 /2 so σ is the only free parameter in F. The baseline values of our model parameters are displayed in Table 5.1. These parameter values were chosen in part to ensure a bankruptcy rate of 4 percent, that is, F (ω) = 0.04, a leverage ratio L = 2, and R = 1.01 when τ = 0. The value of F (ω) that we use is about one percentage point higher than what appears in the literature (see Christiano et al., 2010, for a review). That literature uses models that are specified at a quarterly frequency. Given our setting of β, we are tempted to interpret the period in the model as one year, in which case our specification of F (ω) is somewhat low relative to that in the literature. However, given that the model has only two periods, it is unclear how to compare the time dimension of the model with the quarterly time dimension in empirical models. In light of these considerations, we decided to use a relatively conventional value for F (ω) in our calibration. Our value of μ is also within the range used in the literature. We select values for σ , r k , and ω so that (6.8), (6.10), and the calibrated value of F (ω) are satisfied. The resulting value of σ , reported in Table 5.1, is within the range used in the literature. As in other sections of the chapter, we capture the onset of the crisis with an exogenous drop in N.42 The quantitative properties of the model are displayed in Table 5.2. Panel A in that table displays the properties of the model under the benchmark parameterization. This is our characterization of the economy in a normal time. Panels B and C display two representations of our model economy in a crisis time. Panel B corresponds to the case in which N is reduced. Our second representation of a crisis is that the drop in N is accompanied by a 20 percent rise in σ. We are interested in this

295

0.32 0.98 1.24 0.321 4.03 2.003 0.678 0.666 0.643

0 1.00 1.23 0.321 4.00 2.000 0.679 0.665 0.642

Ramsey

0.671 0.676 0.650

– – – – –

–

First best

0.604 0.564 0.556

−5.45 7.83 0.161 21.27 3.466

−5.36 7.68 0.161 20.94 3.434 0.609 0.559 0.551

1.84

Ramsey

0

τ =0

0.589 0.594 0.571

– – – – –

–

First best

Panel B: Drop in N

0.612 0.552 0.549

−6.99 12.32 0.161 28.95 3.418

0

τ =0

0.606 0.557 0.554

−7.08 12.53 0.161 29.35 3.452

1.95

Ramsey

0.589 0.594 0.571

– – – – –

–

First best

Panel C: Drop in N and rise in σ

Note: (i) The columns headed “Panel A: Baseline” correspond to the baseline parameterization reported in the text. The columns headed “Panel B: Drop in N ” correspond to the baseline parameterization with replaced by N = 1/2. The columns headed “Panel C: Drop in N and rise in σ ” correspond to the baseline parameterization with N = 1/2 and σ replaced the product of its value in the baseline parameterization and 1.2. (ii) The column headed “τ = 0” reports a private sector equilibrium with zero interest rate subsidy. The column headed “Ramsey” reports the Ramsey equilibrium. The column headed “First best” reports the first best allocation.

Interest rate subsidy, 100τ Financial variables (R − 1)100 risk-free rate (Z/R − 1)100 spread N/y net worth bankruptcy rate F (ω) 100 L leverage ratio Real variables c/y time 1 consumption C/y time 2 consumption (B + N)/y investment

τ =0

Panel A: Baseline

Table 5.2. Properties of the asymmetric information model

296

Lawrence J. Christiano and Daisuke Ikeda

representation as a way to capture casual evidence that there was a general increase in “uncertainty” during the crisis. The results for this case are reported in Panel C. In each panel, the column labeled τ = 0 displays the private sector equilibrium with τ = 0. The column “Ramsey” indicates the equilibrium with the Ramsey-optimal τ. Finally, the column marked first-best indicates the first-best allocations. Note from Panel A that in normal times the Ramsey interest rate subsidy τ is 0.3 percent. Thus, consistent with proposition 6.2, in normal times the marginal return on investment exceeds the average return. Because equilibrium saving increases in the subsidy, saving and investment are both higher in the Ramsey equilibrium than they are with τ = 0. The increased supply of saving reduces the equilibrium interest rate R, so that τ is effectively a subsidy to banks’ cost of funds. The interest rate spread is slightly higher in the Ramsey equilibrium than it is in the private sector equilibrium with τ = 0. This reflects that loans to banks are larger in the Ramsey equilibrium, so that monitoring costs associated with bankruptcies are larger. We now turn to Panel B. With the drop in N the economy is poorer and so we expect substantial effects, even in the first-best allocations. According to Panel B, the first-best level of consumption in the first and second periods drops. There is a rise in household saving in response to the shock, but the rise is smaller than the fall in N so that investment drops. In terms of the response in the private sector equilibrium with τ = 0, note that there is a substantial jump in the interest rate spread, from 1.23 percent to 7.83 percent. This jump is associated with a very large rise in bankruptcies, from 4 percent to 21 percent. In addition, consumption in both periods and investment all drop by large amounts. In terms of interest rates and quantities, the drop in N generates a response qualitatively similar to what we found in the previous models. In terms of policy, the optimal interest rate subsidy rises more than fivefold in response to the drop in N. The intervention reduces the cost of funds to banks (R falls an additional nine basis points in the Ramsey equilibrium, compared with the τ = 0 equilibrium). Thus this model shares the implication of the other models in this chapter that indicate that a crisis triggered by a fall in net worth justifies a policy of (increased) interest rate subsidies to banks. For the results in Panel C, we set N = 1/2 and σ = 1.2 × 0.37, where 0.37 is the value of σ in the baseline parameterization (see Table 5.1). Of course, this change in our experiment has no impact on the first-best allocations.

Government Policy, Credit Markets, and Economic Activity

297

The jump in σ causes the interest rate spread to jump by nearly 5 percentage points in both the τ = 0 and Ramsey equilibria. In addition, investment is reduced, though only by a small amount. Note that the increase in σ produces a rise in period 1 consumption. This happens because the increase in σ exacerbates the financial frictions and induces substitution away from activities (investment) that involve finance and toward activities (period 1 consumption) that do not. Christiano et al. (2011), who incorporate the financial frictions described here into a dynamic model, find that fluctuations in σ account for a substantial portion of business cycle fluctuations. This is because other features of their dynamic general equilibrium model reverse our model’s prediction that consumption and investment comove negatively in response to disturbances in σ. Turning to the implications for policy, note in Panel C that the σ shock magnifies the rise in the Ramsey tax subsidy rate. Thus, an increase in uncertainty calls for a greater subsidy to banks’ cost of funds and, hence, more leverage. In sum, the numerical analysis shows that, at a qualitative level, the model of this section rationalizes the view that a drop in net worth reduces investment, raises interest rate spreads, and warrants interest rate subsidies to banks. .

6.4.2. Government Loans and Net Worth Transfers to Banks We consider a government policy that raises a lump-sum tax T on households in the first period. It then lends T to banks using the same technology available to mutual funds and transfers the proceeds to households in the second period in the form of a lump-sum tax rebate. We show that this policy has no impact on equilibrium allocations because it simply displaces, one-for-one, private saving. That is, the Barro-Wallace irrelevance result holds for government purchases of the financial assets of financial businesses. Tax-financed transfers of net worth to banks do help. Indeed, if they are carried out on the right scale, then banks do not require credit and the first-best allocations are supported. This result, which is not surprising, is proved in Appendix C, Section C.2.6. We suppose that the government and mutual funds offer the same menu of loan contracts to banks. Because banks are all identical, each chooses the same loan contract, regardless of whether they borrow from mutual funds or from the government. That is, each receives the same leverage and interest rate, characterized by (6.9) and (6.10). Denote the fraction of banks that receive their loans from mutual funds by ν, whereas the complementary

298

Lawrence J. Christiano and Daisuke Ikeda

fraction receives their loans from the government. Thus, the net worth of banks receiving their loans from mutual funds and the government is νN and (1 − ν )N , respectively.43 In the first period, the representative household deposits b with the banking system. The household’s first-period budget constraint is c + b + T ≤ y.

(6.24)

Expression (6.24) is also the period 1 resource constraint. The household’s second-period budget constraint is C ≤ Rb + government lump sum taxes + banker earnings. By the zero-profit condition of private mutual funds, household deposits generate the following return in equilibrium: Rb = r k (νN + b) [(ω) − μG(ω)] . The banks financed by mutual funds return the following profits to the households: r k (νN + b) [1 − (ω)] . Thus, household income from deposits plus the profits generated by the bankers financed by those deposits is Rb + r k (νN + b) [1 − (ω)] = r k (νN + b) [1 − μG(ω)] . Similarly, the households receive a tax rebate from the government plus profits from the bankers financed by the government, in the amount r k ((1 − ν)N + T )[1 − μG(ω)]. Then, total household income in the second period is r k [(νN + b) + ((1 − ν )N + T )] [1 − μG(ω)] = r k (N + b + T )[1 − μG(ω)]. Combining the household’s budget constraint with the equilibrium expressions for banker profits and deposit interest, we obtain the second-period resource constraint: C ≤ r k (N + b + T ) [1 − μG(ω)] .

(6.25)

The necessary and sufficient conditions for an (interior) household optimum are that the first- and second-period budget constraints [i.e., (6.24) and (6.25)] hold with equality and that the intertemporal marginal rate of substitution in consumption be equated to the household deposit rate R.

Government Policy, Credit Markets, and Economic Activity

299

These three conditions coincide with the first three equations in (6.15) if we identify B with b + T and we set τ = 0. Thus, the equilibrium allocations of the model in this subsection are determined recursively. The five equations in (6.15) with τ = 0 determine the five equilibrium objects in (6.16). Private lending b is then determined by b = B − T. If the government increases loans T to bankers, then household loans to bankers via mutual funds are reduced by the same amount. Of course, this assumes (as we do throughout this chapter) that we consider only interior equilibria. For example, if B < T then the nonnegativity constraint b ≥ 0 would be binding and we expect the tax policy to have real effects.44 We state this result in the form of a proposition: Proposition 6.3: Suppose the government has the same technology for making loans to banks as do mutual funds. Then in an interior equilibrium, taxfinanced government loans to bankers have no impact on rates of return, asset prices, consumption, and investment. Of course, this proposition is not true if there are differences between the loans provided by the government and those provided by mutual funds. We suspect that interesting asymmetries would involve the government’s having a less-efficient technology for making loans than mutual funds have. If so, then we conjecture that social welfare would be reduced with an increase in T. Thus the environment of this section, in contrast to the one in Section 3, appears to provide little rationale for government purchases of securities issued by financial businesses.

6.5. Pecuniary Externalities: A Robustness Check Recent literature emphasizes the importance of pecuniary externalities that occur when expenditures are constrained by the market value of agents’ assets and the market price of assets is endogenous. Our setting incorporates collateral constraints, but our assumptions about technology imply that asset prices are fixed. Here, we report calculations that suggest the conclusions of our analysis are robust to endogenizing asset prices. In particular, we show that the underborrowing property of the model is robust to endogenizing asset prices, as is the result that it is desirable to subsidize the interest rate costs of banks. We consider a sequence of economies, starting with our baseline specification in which the technology for converting banker resources, N + B, into productive capital is linear. In the baseline specification, the price of capital is unity. When there is curvature in the technology for producing

300

Lawrence J. Christiano and Daisuke Ikeda

capital, the price of capital becomes endogenous and in principle this introduces a pecuniary externality. We established in proposition 6.2 that in our baseline linear case the equilibrium is characterized by underborrowing by banks and mutual funds and undersaving by households. We find that when curvature in the production of capital is increased the optimal value of the subsidy τ at first increases, so that underborrowing becomes even more severe. For yet higher levels of curvature, the optimal value of the subsidy converges slowly to zero from above, so that the underborrowing result is attenuated. On net, our numerical results indicate that our underborrowing result is robust to all but the very highest levels of curvature. And even then, we never find overborrowing. To explain our results, we now indicate the key features of the modified model, which we adapt from BGG. In particular, in the version of the model with endogenous capital prices, bankers are endowed with a quantity of capital k at the beginning of the period. As in BGG, bankers sell this capital to capital producers for a price, Pk . This price defines banker net worth, N = Pk k. Capital producers operate a technology that combines k with investment goods to produce and sell new capital K at a price PK . The technology operated by capital producers is  γ I K =k+ k, 0 ≤ γ ≤ 1. k Banks go to mutual funds with their net worth and obtain a standard debt contract with loan amount B. They purchase K by using this loan and their net worth, subject to PK K ≤ B + N. The banker operates a production technology analogous to the one in (6.1) and (6.2): ωKr k , where ω is distributed as in (6.2) and r k is a fixed parameter. With our modification, the prices of old and new capital, Pk and PK , respectively, are endogenous. The rate of return on capital Rk is also endogenous, with Rk ≡

rk . PK

Now, suppose a mutual fund deviates from the standard debt contract and makes an additional loan to one particular bank. This has two effects that may involve pecuniary externalities. First, when the banker uses the loan to purchase new capital, its price PK is driven up by a small amount. This

Government Policy, Credit Markets, and Economic Activity

301

0.02 0.018 0.016

tax rate, τ

0.014 0.012 0.01 0.008 0.006 0.004 0.002 0

0.1

0.2

0.3

0.4

0.5 γ

0.6

0.7

0.8

0.9

1

Figure 5.6. Optimal subsidy rate, in presence of pecuniary externality.

encourages capital production and leads to a rise in Pk , thus raising the net worth of other bankers and raising the value of their net worth. As a result, the other bankers are able to borrow more too. The second effect of extending a loan to a banker is that the things the banker buys (in this case K ) become more expensive (i.e., PK increases). Note that the two effects work against each other. Depending on the relative strengths of the two effects, various results could occur in principle. The details of the modified model economy are presented in Appendix C. We now report our quantitative experiment. We solved for the Ramsey optimal subsidy rate τ for 0.01 ≤ γ ≤ 1, holding all other parameters fixed at their baseline level (see Table 5.1). Figure 5.6 displays the computed values of τ. Note that as γ decreases, the inefficiency of the economy initially increases because τ increases. However, for γ below roughly 0.85, the inefficiency of equilibrium decreases monotonically with additional reductions in γ . For γ near 0.3, the inefficiency is virtually completely eliminated because the Ramsey optimal τ is close to zero. We interpret the finding τ ≥ 0 as indicating the robustness of our underborrowing result and of our result concerning the desirability of subsidizing banks’ cost of funds.

7. Concluding Remarks In the past decade, DSGE models have been constructed that have proved useful for analyzing questions of interest to policy makers.45 In recent years,

302

Lawrence J. Christiano and Daisuke Ikeda

the Federal Reserve has undertaken various actions – a large-scale asset purchase program, reductions in banks’ cost of funds – with the objective of correcting dysfunctions in credit markets. The DSGE models developed to place structure on the policy discussions before 2007 are silent on the rationale, design, and appropriate scale of recent policy actions. DSGE models must integrate the right sort of financial frictions to be useful, given the new policy questions. This chapter surveys four candidate models and summarizes some of their implications for recent policy actions.

APPENDIX A: NOTES ON THE UNOBSERVED EFFORT MODEL A.1. Computational Strategy for Solving the Baseline Model Computing an equilibrium when the cash constraint, (4.8), is not binding (i.e., ν = 0) is straightforward. Here, we describe an algorithm for computing an equilibrium when the cash constraint is binding, so that ν < 0. There are ten equilibrium conditions. This includes the six conditions associated with the banks, (4.21), the three conditions associated with household optimization, (4.3), (4.2), and (4.16) and the definition of λ, the marginal utility of second-period consumption, (4.17). The ten variables to be solved for are λ, c, C, Rgd , Rbd , R, e, d, ν, η. When ν < 0, the e and ν equilibrium conditions associated with the banks can be simplified and we do so first. That ν = 0 implies that the ν equation in (4.21) can be replaced by ν : Rbd d = Rb (N + d ). Note that the μ equation implies that Rgd − Rbd =

R − Rbd . p(e)

Use this expression to substitute out for Rgd − Rbd in the η equation:  R − Rbd d . e = λb (R − R )(N + d ) − p(e) 

g

b

(A.1)

Government Policy, Credit Markets, and Economic Activity

303

Note that the d equation implies that Rg − Rb =

R − Rb . p(e)

Use this expression to substitute out for p(e) in (A.1) and use the ν equation to substitute out for Rbd in (A.1) to obtain   R − Rb (N d+d ) g b g b d . e = λb (R − R )(N + d ) − (R − R ) R − Rb Factor (Rg − Rb ), collect terms in d, and rearrange, to obtain the adjustment to the η equation that is possible when ν = 0: η : e = λb(Rg − Rb )

R N. R − Rb

(A.2)

The equilibrium conditions associated with the banks, with the appropriate adjustments that are possible when ν = 0, are   e : (λ − ν)b Rgd − Rbd d + η = 0, d : R = p(e)Rg + (1 − p(e))Rb , Rgd : ν p(e) = ηλb, μ : R = p(e)Rgd + (1 − p(e))Rbd , R η : e = λb(Rg − Rb ) N, R − Rb ν : Rbd d = Rb (N + d ).

(A.3)

To solve this system, fix R > Rb . Solve for λ, c, C, and d by using (4.17) and the household equations, (4.3), (4.2), and (4.16). Compute Rbd using the ν equation. Compute e using the η equation. Compute Rgd using the μ equation. Adjust the value of R until the d equation is satisfied. To investigate the possibility of multiple equilibria, we considered values of R on a fine grid over a wide range of values, and it appeared that there is only one value of R that satisfies the d equation. Finally, the e and Rgd equations can be used to solve linearly for ν and η. For example, with some algebra we find   b2 Rgd − Rbd λ2 d  . ν = 2 d b Rg − Rbd λd − p(e) This completes the discussion of computing the equilibrium.

304

Lawrence J. Christiano and Daisuke Ikeda

A.2. Proof of Proposition 4.5 In this appendix, we prove a slightly more precise statement of proposition 4.5 in Subsection 4.5.2. Let B ≡ d + T. Then we obtain the following proposition. Proposition A.1: Let T and T  denote two different levels of tax-financed equity finance. r Equilibrium is characterized by the following invariance property. If

c, C, e, R, λ, B satisfy the equilibrium conditions under T , then c, C, e, R, λ, B also satisfy the equilibrium conditions under T  as long as ν < 0 in both cases. r In addition, (i) while ν < 0, ν is monotone increasing in T and (ii) there is a T large enough, say T ∗ , such T ∗ < B and ν = 0 for B > T ≥ T ∗ . It is worth stressing that the invariance property applies only to c, C, e, R, λ, and B and not to all the endogenous variables of the model. These include, in addition, ν, η, Rgd , and Rbd .

(A.4)

The equilibrium values of the variables in (A.4) do change with T , when ν < 0. We now prove the preceding proposition. The four equilibrium conditions associated with the household are given by (4.17), (4.27), (4.28), and (4.2): C = R(N + d + T ), y = c + d + T, c

−γ

= βRC −γ ,

λ = βu (C ). Note that if c, C, R, λ, and B solve these equations for one value of T, then the same c, C, R, λ, and B solve these equations for another value of T. That is, the household equilibrium conditions satisfy the invariance property. Now consider the equilibrium conditions associated with the banks. Recall from Subsection 4.5.2 that only private deposits d (i.e., not T ) enter the bank equilibrium conditions. The equilibrium conditions for the case

Government Policy, Credit Markets, and Economic Activity

305

ν = 0 are stated in (A.3). For convenience, we rewrite the equations here, replacing d with B − T :   e : (λ − ν)b Rgd − Rbd (B − T ) + η = 0, d : R = p(e)Rg + (1 − p(e))Rb , Rgd : ν p(e) = ηλb,

(A.5)

μ : R = p(e)Rgd + (1 − p(e))Rbd , R η : e = λb(Rg − Rb ) N, R − Rb ν : Rbd (B − T ) = Rb (N + B − T ).

The d and η equations clearly satisfy the invariance property. It remains to verify that the four equations, e, Rgd , μ, and ν, do so too. For given B, λ, e, and R, these four equations represent four linear equations in the four unknowns in (A.4). We now verify that these equations have a unique solution. With this result, our invariance property is established. Using the μ equation to substitute out for Rgd − Rbd in the e equation and the Rgd equation to substitute out for η, the remaining two equations are e : (λ − ν )b

R − Rbd p(e) (B − T ) + ν = 0, p(e) λb

ν : Rbd (B − T ) = Rb (N + B − T ). Use the second of these equations to substitute out for Rbd in the e equation: e : (λ − ν )b

p(e) R(B − T ) − (N + B − T )Rb +ν = 0. p(e) λb

Solving for ν, we obtain −ν =



λ2

.

2

p(e) b R(B−T )−(N +B−T )Rb

(A.6)

− λb

Given the value of ν < 0 in (A.6), we can now uniquely solve for η, Rgd , and Rbd in (A.4). The invariance property is established. Note that the invariance property applies only to the variables in proposition A.1.

306

Lawrence J. Christiano and Daisuke Ikeda

We now turn to properties (i) and (ii) in the proposition. According to (A.6), our assumption that the cash constraint is binding in the bad state, ν < 0, implies that R(B − T ) − (N + B − T )Rb > 0. Also, note that R > Rb according to the d equation in (A.5). As a result, for fixed B,R, the preceding expression is linear and decreasing in T. Let T ∗ denote the value of T , where the preceding expression passes through zero. Note that T ∗ < B, so that deposits B − T ∗ are positive. We can see from (A.6) that as T → T ∗ , − ν → 0. That is, the cash constraint is marginally nonbinding for T = T ∗ . For T > T ∗ the cash constraint is nonbinding and ν = 0. This completes the proof of the proposition.

A.3. Solving the Version of the Model with Bailouts and Leverage We describe algorithms for solving the versions of the model studied in Subsections 4.5.4 and 4.5.5 in the main text. We begin with the version of the model in Subsection 4.5.4. The first-order conditions associated with the problem in (4.32) are a suitable adjustment on (4.20):      e : λp(e) Rg − Rb (N + d ) − Rgd − Rbd d   − e + μp(e) Rgd − (1 + τ )Rbd d       + η 1 − λp(e) Rg − Rb (N + d ) − Rgd − Rbd d = 0,     d : 0 = λp(e) Rg − Rgd + λ(1 − p(e)) Rb − Rbd   + μ p(e)Rgd + (1 − p(e))(1 + τ )Rbd − R       (A.7) − ηλp (e) Rg − Rb − Rgd − Rbd + ν Rbd − Rb , Rgd : −λp(e) + μp(e) + ηλp (e) = 0, Rbd : −λ(1 − p(e)) + μ(1 − p(e))(1 + τ ) − ηλp (e) + ν = 0, μ : R = p(e)Rgd + (1 − p(e))(1 + τ )Rbd ,      η : e = λp (e) Rg − Rb (N + d ) − Rgd − Rbd d , ν : Rbd d − Rb (N + d ) = 0. Here, we assume ν = 0, so that the cash constraint is binding.

Government Policy, Credit Markets, and Economic Activity

307

We eliminate the multipliers μ and η and two equations from this system. We add the Rgd equation to the Rbd equation and solve for μ: λ−ν . 1 + (1 − p(e))τ

μ= From the Rgd equation,

λ−ν ηλb = λ − 1 + (1 − p(e))τ = ν˜ p(e),

p(e)

where ν˜ ≡

ν + (1 − p(e))τ λ . 1 + (1 − p(e))τ

(A.8)

We substitute the expressions for μ and ηλb, as well as the μ equation, into the d equation:     d : 0 = λp(e) Rg − Rgd + λ(1 − p(e)) Rb − Rbd       − ν˜ p(e) Rg − Rb − Rgd − Rbd + ν Rbd − Rb . or, after rearranging, we obtain      ˜ p(e) Rg − Rgd + (1 − p(e)) Rb − Rbd d : 0 = (λ − ν)   ˜ Rbd − Rb . +(ν − ν) Note that when τ = 0, then ν˜ = ν, and this equation, together with the μ equation in (4.21), reduces to d in (4.21).46 We must adjust the equilibrium conditions of the household to accommodate the taxes required to finance the bank bailouts. The profits π brought home by the bankers in the representative household in period 2 are     π = p(e) Rg (N + d ) − Rgd d + (1 − p(e)) Rb (N + d ) − Rbd d . The representative household’s second-period budget constraint is C = Rd + π − T ,

308

Lawrence J. Christiano and Daisuke Ikeda

where T denotes the lump-sum taxes required to finance τ : T = (1 − p(e))τ Rbd d. Substituting out for T and π in the second-period budget constraint, we obtain     C = Rd + p(e) Rg (N + d ) − Rgd d + (1 − p(e)) Rb (N + d ) − Rbd d −(1 − p(e))τ Rbd d = Rd + p(e)Rg (N + d ) + (1 − p(e))Rb (N + d )   − p(e)Rgd d + (1 − p(e))Rbd d − (1 − p(e))τ Rbd d = Rd + p(e)Rg (N + d ) + (1 − p(e))Rb (N + d )   − p(e)Rgd d + (1 + τ )(1 − p(e))Rbd d − (1 − p(e))τ Rbd d −(1 − p(e))τ Rbd d = Rd + p(e)Rg (N + d ) + (1 − p(e))Rb (N + d ) − Rd, where the fourth equality makes use of the zero-profit condition of mutual funds. Then,   C = p(e)Rg + (1 − p(e))Rb (N + d ).

(A.9)

From the household intertemporal Euler equation, c −γ = βRC −γ , we have 1

C = c(βR) γ . Substituting d out by using the first- and second-period budget constraints, we obtain c+

p(e)Rg

C = N + y. + (1 − p(e))Rb

We use the intertemporal first-order condition to substitute out for C, so that the equilibrium conditions for the household are c=

N +y 1+

1

(βR) γ p(e)Rg +(1−p(e))Rb

, d = y − c.

Government Policy, Credit Markets, and Economic Activity

309

We now collect the equilibrium conditions. We replace the d equation in (A.7), and make use of the expressions for η and μ to obtain the following system:   λ−ν ν˜ b Rgd − (1 + τ )Rbd d + p(e) = 0, 1 + (1 − p(e))τ λb      ˜ p(e) Rg − Rgd + (1 − p(e)) Rb − Rbd d : 0 = (λ − ν)   + (ν − ν˜ ) Rbd − Rb , e:

μ : R = p(e)Rgd + (1 − p(e))(1 + τ )Rbd ,     η : e = λb (Rg − Rb )(N + d ) − Rgd − Rbd d ,

(A.10)

ν : Rbd d − Rb (N + d ) = 0, N +y c= , 1 γ 1 + p(e)Rg (βR) +(1−p(e))Rb d = y − c, c

−γ

= βRC −γ ,

where the last three equations are the equilibrium conditions associated with the household. It is understood that λ is determined according to (4.17) and ν˜ according to (A.8). In addition, the functional form for p(e) has been used. The expressions in (A.10) represents eight equations in eight unknowns: ν, e, Rgd , Rbd , d, c, C, R. These equations reduce to (4.21) when τ = 0 in the case ν = 0. We solve the preceding equations by solving one equation in R. We fix a value for R. We now define a mapping from e into itself. We fix a value for e. We use the last three equations in (A.10) to solve for c, C, and d. Then, we use the ν equation to solve for Rbd :  Rbd = Rb

N +d d

 .

We use the zero-profit condition for mutual funds to solve for Rgd : Rgd =

R − (1 − p(e))(1 + τ )Rbd . p(e)

310

Lawrence J. Christiano and Daisuke Ikeda

Finally, we use the η equation to solve for e:     η : e = λb (Rg − Rb )(N + d ) − Rgd − Rbd d . We adjust the value of e until a fixed point is obtained. We now have d, c, C, R, Rbd , Rgd , and e in hand. It remains to determine a value for ν. Substituting out for ν˜ in the e equation and multiplying by 1 + (1 − p(e))τ , we obtain    1  ν + (1 − p(e))τ λ p(e) = 0. (λ − ν)b Rgd − (1 + τ )Rbd d + λb Note that    1  ν + (1 − p(e))τ λ p(e) (λ − ν )b Rgd − (1 + τ )Rbd d + λb     d d = λb Rg − (1 + τ )Rb d − νb Rgd − (1 + τ )Rbd d 1 1 + ν p(e) + (1 − p(e))p(e)τ b λb

  d 1 d =ν p(e) − b Rg − (1 + τ )Rb d λb   1 + λb Rgd − (1 + τ )Rbd d + (1 − p(e))p(e)τ. b Then,   λb Rgd − (1 + τ )Rbd d + 1b (1 − p(e))p(e)τ   . ν= 1 b Rgd − (1 + τ )Rbd d − λb p(e) Finally, we adjust the value of R until the d equation is satisfied. We now turn to the model considered in Subsection 4.5.5. Relative to the solution to (4.32) that we just analyzed, the solution to (4.33) involves only replacing the zero in the d equation in (A.10) with δ and adding LN = (N + d ) as an additional equation. We now have nine equations in nine unknowns. We adapt the strategy just described to solve this model. We solve two equations in two unknowns, R and e. We fix values for R and e and solve for d, c, C, Rbd , ν, and Rgd in the same way as before. These calculations do not involve the d equation. Then, we solve for δ using the modified d equation:      ˜ p(e) Rg − Rgd + (1 − p(e)) Rb − Rbd δ = (λ − ν)   ˜ Rbd − Rb . + (ν − ν)

Government Policy, Credit Markets, and Economic Activity

311

Finally, we adjust R and e until the η equation in (A.10) and LN = (N + d ) are satisfied. In effect, this solution strategy replaces the d equation with the leverage constraint as an equality.

APPENDIX B: PROOF OF PROPOSITION 5.3 FOR THE ADVERSE-SELECTION MODEL In this appendix we prove proposition 5.3, which underlies the reason for the assumption in Section 5 that there is an upper bound on the scale of investment projects. Proposition 5.3 addresses what happens if we instead adopt an assumption at the opposite extreme, that investment projects have constant returns to scale without any upper bound. We show that if there is an equilibrium in this version of the model, then it must be that all borrowing is done by the bankers with the lowest value of p and the aggregate profits of those bankers are zero. Thus in this equilibrium the household receives the whole marginal product of his or her saving, and the first-best efficient allocations are supported. For equilibrium to be well defined in this case, we obviously require that there be positive mass on the lower bound of the support for p and that that lower bound be positive. Thus, we suppose that p ∈ [ε, 1], where ε is a very small, positive number. Suppose a banker with a particular p∗ ∈ [ε, 1] chooses to activate a project; let B p∗ denote the amount borrowed by that banker. Such a banker’s profits must be no less than what the banker can earn by simply depositing his or her net worth in the bank and not borrowing anything.47 That is, the analog of (5.5) must hold: (θ − p∗ r )B p∗ + N (θ − R) ≥ 0.

(B.1)

For each p ≤ p∗ there exists a B p such that (B.1) also holds, so that bankers with p ≤ p∗ also choose to activate their projects. In equilibrium, it must be that θ ≤ pr,

p ≤ p∗ ,

(B.2)

for otherwise there is no choice of B p that optimizes expected banker profits. For an interior equilibrium in which there is a positive supply of deposits to mutual funds, it must be that B p > 0 for some p. Bankers with lower probabilities of success also borrow positive amounts, and we conclude that

312

Lawrence J. Christiano and Daisuke Ikeda

in an interior equilibrium there exists a pu ∈ [ε, 1] such that bankers with ε ≤ p ≤ pu choose B p > 0. For p in this interval, it must be that θ ≥ pr,

p ≤ pu ,

(B.3)

for otherwise the supposition B p > 0 is contradicted. Let p+ ≡ min(pu , p∗ ). Combining (B.2) and (B.3), we conclude that θ = pr,

for ε ≤ p ≤ p+ .

But this expression can hold only if p+ = ε. We conclude that in an interior equilibrium only the bankers with the lowest probability of success operate their projects. Each of these bankers pays r = θ /ε in interest and earns zero profits ex ante. Because mutual funds are competitive and so make zero profits, R = θ . That is, in equilibrium, households receive the actual social marginal return on loans. As a result, the allocations in equilibrium coincide with the first-best efficient allocations. This establishes proposition 5.3.

APPENDIX C: NOTES ON THE ASYMMETRIC INFORMATION MODEL C.1. Proof That the Marginal Return Exceeds the Average Return on Loans For convenience, we repeat the expression for the marginal return on loans, the object to the right of the equality in (6.20), here:    Rm ≡ r k 1 − μ G(ω(B)) + (N + B)G  (ω(B)) ω (B) . (C.1) We wish to establish Rm > R. Here, R is the equilibrium cost of funds to mutual funds, which we showed is also equal to the average return on loans to bankers [see (6.8)]: r k (N + B) . (C.2) B Thus, we establish that (under a regularity condition subsequently stated) the marginal return on loans exceeds the corresponding average return. In (C.2) and (C.1), ω(B) is defined by (6.18) and (6.19). We reproduce these expressions here [after substituting out for L in (6.18) by using (6.19)] for convenience: N   (ω)   (ω) − μG  (ω) = . (C.3) (ω) − μG(ω) B 1 − (ω) R = [(ω(B)) − μG(ω(B))]

Government Policy, Credit Markets, and Economic Activity

313

The mapping ω(B) characterizes how ω changes with a change in B, given the zero-profit condition of mutual funds, (6.9), and the efficiency condition characterizing the solution to the banker contracting problem, (6.10). For our analysis, we require the derivative of ω(B) with respect to B: ω (B) =

(1 − )(  − μG  ) ,   (  − μG  )(N + B) +   ( − μG)N − (1 − )(  − μG  )B (C.4)

where we have omitted the argument ω in , G,   , and G  for notational simplicity. The loan contract between banks and mutual funds solves the following optimization problem: maxr k (N + B)[1 − (ω)],

(C.5)

r k (N + B) [(ω) − μG(ω)] − RB = 0.

(C.6)

ω,B

subject to

Letting λ ≥ 0 denote the Lagrange multiplier associated with constraint (C.6), we find that the first-order conditions of the problem are   (C.7) r k [1 − (ω)] + λ r k [(ω) − μG(ω)] − R = 0, λ=

  (ω) .   (ω) − μG  (ω)

(C.8)

It is easy to verify that   (ω) = 1 − F (ω) > 0,

G  (ω) = ω f (ω) > 0,

(C.9)

where f (ω) ≡ dF (ω)/dω. Conditions (C.8), (C.9), and λ ≥ 0 imply48 λ > 1.

(C.10)

Solving (C.7) for λ and multiplying the numerator and denominator of (C.8) by r k (N + B)ω (B), we obtain r k [1 − (ω)] , R − r k [(ω) − μG(ω)]

(C.11)

r k (N + B)  (ω)ω (B) , r k (N + B)[  (ω) − μG  (ω)]ω (B)

(C.12)

λ= λ=

where ω (B) is defined in (C.4).

314

Lawrence J. Christiano and Daisuke Ikeda

Combining (C.11) and (C.12), we obtain the following expression for λ: λ=

r k [1 − ] − r k (N + B)  ω (B) , R − r k [ − μG] − r k (N + B)[  − μG  ]ω (B)

(C.13)

where we suppress the argument ω when doing so does not risk confusion. The numerator of (C.13) consists of the numerator of (C.11) minus the numerator of (C.12). Similarly, the denominator of (C.13) consists of the denominator of (C.11) minus the denominator of (C.12).49 Now we rewrite the marginal return (C.1), making use of the expression for λ in (C.13): Rm = r k [(ω) − μG(ω) + 1 − (ω)] − r k (N + B)[  (ω) + μG  (ω) −   (ω)]ω (B),   = r k [1 − (ω)] − r k (N + B)  (ω)ω (B)  − R − r k [(ω) − μG(ω)] − r k (N + B)    ×   (ω) − μG  (ω) ω (B) + R,  = R + (λ − 1) R − r k [(ω) − μG(ω)] − r k (N + B)    (C.14) ×   (ω) − μG  (ω) ω (B) , where R denotes the average return, (C.2). From (C.10), λ > 1. Hence, if the object in braces in (C.14) is positive, then Rm > R. The object in braces in (C.14) is the denominator of λ in (C.13). Because λ > 0, we know that the denominator is positive if the numerator of λ is positive. Using the expression for ω (B), (C.4), we rewrite the numerator as follows: r k [1−]−r k (N +B)  ω (I ) r k (N +B)  (ω)(1−)(  −μG  ) = r k [1−]−     ( −μG )(N +B)+  (−μG)N −(1−)(  −μG  )B   1 k = r [1−] 1− .   −μG  )B 1+  (−μG)N−(1−)(    (N+B) (ω)( −μG ) This object is positive if   ( − μG)N − (1 − )(  − μG  )B (N + B)  (  − μG  )

(C.15)

is positive. A sufficient condition for (C.15) to be positive is that ωh(ω) is increasing in ω, where h(ω) denotes the hazard rate [see (6.22)]. According to BGG, this implies that (i)   − μG  > 0 in equilibrium, and (ii)   G  −   G  > 0 for all ω.50 Condition (i) implies that the denominator of (C.15)

Government Policy, Credit Markets, and Economic Activity

315

is positive. The following result shows that (ii) implies that the numerator of (C.15) is positive:   ( − μG)N − (1 − )(  − μG  )B (1 − )(  − μG  ) B − (1 − )(  − μG  )B =   ( − μG)   ( − μG) 1−  B =   (  − μG  ) −   (  − μG  )   1− = μ   G  − G    B > 0.  Here, the first equality uses (C.3) to substitute out for N. This completes the proof of the proposition that the marginal return on B exceeds the corresponding average return. We state this proposition formally as follows: Proposition C.1: Suppose that ωh(ω) is increasing in ω. Then Rm > R.

C.2. Model with Curvature in the Production of Capital Here, we introduce the modifications to the model that cause the price of capital to be endogenous and possibly be the source of a pecuniary externality. We introduce a representative, competitive firm that produces capital. Rather than building the banker’s own capital (as we assume in the main text), the banker uses the net worth and mutual fund loan to purchase capital from a representative capital producer. The following subsection describes the problem of that firm. We then derive all the model equilibrium conditions. The model is virtually identical to the one in the main text. Still, there are some differences in notation, and so we spell out all the details at the risk of some overlap. After that, we describe the algorithm used to compute the equilibrium. This algorithm is the basis for the calculations discussed in the text. Finally, the last subsection describes the proposition that establishes that a sufficiently large net worth transfer to bankers supports the first-best allocations. C.2.1. Capital Producers At the start of period 1, each banker is endowed with an equal quantity k of capital goods. Each banker sells his or her capital to capital producers and receives N = Pk k, where Pk is the market price of capital in terms of the period 1 numeraire good, consumption. Here, N denotes a banker’s net worth after selling his or her capital. In the main text (apart from Section

316

Lawrence J. Christiano and Daisuke Ikeda

6.5), Pk = 1 always, and so N = k there. The fact that N is a function of the market price Pk creates the potential for a pecuniary externality in this model. A perfectly competitive, representative capital producer operates the following production function:   I K =k+ f k, k where

   γ I I f = , k k

0 < γ < 1.

Here, I denotes a quantity of investment goods, measured in units of the period 1 numeraire good and K denotes the quantity of new capital produced by the capital producer. Profits of the capital producer are given by PK K − I − Pk k.

(C.16)

Here, PK denotes the market price of new capital, in units of the period 1 numeraire good. The representative capital producer takes prices PK and Pk , as given. In an interior equilibrium, profit maximization leads to the following first-order condition for k:     I I  I PK 1 + f − f = Pk . k k k The first-order condition for I is PK f 

  I = 1. k

Combining the two first-order conditions with the production function implies the capital producer’s profits, (C.16), are zero. C.2.2. Banks and Mutual Funds The typical bank takes its net worth N and approaches a mutual fund for a loan B. It combines its net worth and the loan to purchase new capital: PK K = N + B. In period 2, the banker uses his or her capital K to produce ωKr k

Government Policy, Credit Markets, and Economic Activity

317

goods, where r k is a fixed parameter of technology. In addition, ω is an idiosyncratic productivity shock,

∞ ω ∼ F, dF (ω) = 1, 0

where F is the cdf of a log-normal distribution. A representative mutual fund offers the banker a standard debt contract in period 1, before the realization of ω. Under the contract, the bank pays the mutual fund an amount ZB in period 2 if it is able to do so. Bankers whose ω is too low to pay ZB in full are “bankrupt.” Mutual funds verify this by monitoring those bankers, at a cost of μωKr k goods, where μ > 0 is a parameter. Bankrupt banks must transfer everything they have to their mutual fund. The cutoff level of productivity, ω, that separates the bankrupt and nonbankrupt banks is defined by ωKr k = ZB.

(C.17)

From the perspective of period 1, an individual bank’s expected profits in period 2 are given by

∞

∞     k ωKr − ZB dF (ω) = ωKr k − ωKr k dF (ω) ω ω  ∞  k (C.18) = NR L [ω − ω] dF (ω) , ω

using (C.17). In (C.18), L denotes leverage [see (6.4)] and Rk denotes the rate of return on capital: Rk ≡

rk . PK

Expression (C.18) is written in compact notation as follows:

∞ N Rk L [ω − ω] dF (ω) = N LRk (1 − (ω)), ω

where



ω

(ω) ≡ G(ω) + ω [1 − F (ω)] , G(ω) ≡ 0

ωdF (ω).

(C.19)

(C.20)

318

Lawrence J. Christiano and Daisuke Ikeda

A bank agrees to maximize (C.19) and remit all its profits to its household in period 2 in exchange for perfect consumption insurance. With bankers maximizing (C.19), the household ensures that his or her bankers as a group maximize the total resources available to the household in period 2. Households can observe everything about their own member bankers (including ω). We now turn to the mutual funds. There are a large number of competitive mutual funds, each of which makes loans to a diversified group of bankers and takes deposits from households. Because there is no risk on the asset side of the balance sheet, it is feasible for mutual funds to commit in period 1 to paying households a fixed and certain gross rate of interest R on their deposits in period 2. Because mutual funds are competitive, they take R as given. A mutual fund that makes size B loans to each of a large number of banks earns the following per bank:

ω [1 − F (ω)]ZB + (1 − μ) ωdF (ω)r k K. 0

Here, the term before the plus sign indicates the revenues from bankers that are not bankrupt, that is, those with ω ≥ ω. The term after the plus sign indicates receipts, net of monitoring costs, from bankers that cannot pay interest, Z. Because the cost of funds is RB, the mutual funds’ zero-profit condition is [using (C.17)]

ω k ωdF (ω)r k K = RB, [1 − F (ω)]ωKr + (1 − μ) 0

or [(ω) − μG(ω)]

Rk PK K = R. B

(C.21)

As in the main text, the interest rate paid to households is proportional to the average return Rk PK K/B on loans and not, say, to the marginal return. Let the combinations of ω and B that satisfy (C.21) define a menu of loan contracts that is available to bankers for given PK and Rk .51 It is convenient to express this menu in terms of L and ω. Rewriting (C.21), we obtain L=

1 1−

Rk R

[(ω) − μG(ω)]

.

(C.22)

Bankers take N and Rk as given and select the contract (L,ω) from this menu, which maximizes expected profits, (C.19). Using (C.22) to substitute

Government Policy, Credit Markets, and Economic Activity

319

out for L in the banker’s objective, the problem reduces to one of choosing ω to maximize 1 − (ω) N Rk . k 1 − RR [(ω) − μG(ω)] The first-order necessary condition for optimization is   Rk 1 − F (ω) − μωF  (ω) 1 − F (ω) R = , k 1 − (ω) 1 − R [(ω) − μG(ω)] R

which can be solved for ω given R and R. Given the solution for ω, L, and Z, we solve 1 L L= , , Z = Rk ω Rk L−1 1 − R [(ω) − μG(ω)] k

respectively. As in the text, L and Z are independent of net worth N. C.2.3. Households and Government In period 1, the household budget constraint is c + B ≤ y,

(C.23)

where c, B, and y denote consumption, bank deposits, and an endowment of output y, respectively. In the second period, deposits generate an after-tax return, (1 + τ )RB = (1 + τ )r k K [(ω) − μG(ω)] , where τ denotes a subsidy on household saving and (C.21) has been used. Total profits brought home by a household’s bankers are denoted by π : π = Kr k (1 − (ω)) . Taking into account that mutual funds have zero profits, the second-period budget constraint is C ≤ (1 + τ )RB + π − T ,

(C.24)

where T denotes lump-sum taxes and C denotes second-period consumption. Substituting, we obtain C ≤ (1 + τ )r k K [(ω) − μG(ω)] + Kr k (1 − (ω)) − T. The government’s budget constraint is T = τ RB = τ r k K [(ω) − μG(ω)] .

(C.25)

320

Lawrence J. Christiano and Daisuke Ikeda

If we combine the government’s budget constraint with the household’s second-period budget constraint, we obtain the second-period resource constraint: C ≤ r k K [1 − μG(ω)] ,

(C.26)

which is Walras’s law in our environment. That is, period 2 consumption is no greater than total output, net of the output used up in monitoring by banks. The representative household maximizes u(c) + βu(C), subject to (C.23) and (C.24). The first-order necessary and sufficient conditions corresponding to this problem are C π −T u (c) = (1 + τ )R, c + =y+ .  βu (C ) (1 + τ )R (1 + τ )R In practice, we assume that u(c) =

c 1−α . 1−α

C.2.4. Equilibrium We define an equilibrium as follows: Definition C.2: A private sector equilibrium is a (C, c, R, K , ω, B, Pk , PK , I, T ) such that (i) (ii) (iii) (iv) (v) (vi)

the household problem is solved for given τ , the problem of the bank is solved, mutual fund profits are zero, the problem of the capital producer is solved (see Subsection C.2.1), the government budget constraint is satisfied, and the first- and second-period resource constraints are satisfied.

For convenience, we collect the equations that characterize a private sector equilibrium here. We divide these equilibrium conditions into a household block, and a bank/mutual fund/capital producer block. The first block is Equation number (1) (2) (3)

Household Economic description 1 C = c (β[1 + τ ]R) α household first-order condition k C = r K [1 − μG(ω)] period 2 resource constraint c+I =y period 1 resource constraint

Government Policy, Credit Markets, and Economic Activity

321

Expression (3) is the household’s first-period budget constraint, with B replaced with I. This replacement is possible for the following reason. Total bank assets PK K can be written as follows: PK K = N + B = Pk k + B. Here, the first equality is the banker’s expenditure constraint and the second equality uses the definition N = Pk k. Zero profits for the capital producers (see Subsection C.2.1) implies that PK K = I + Pk k, and the fact that B=I

(C.27)

follows. The impact of the household and government budget constraints is completely captured by the period 1 and period 2 resource constraints and expression (1), and so the budget constraints are not included among the equilibrium conditions associated with the household. The set of equilibrium conditions associated with banks, mutual funds, and capital producers is as follows: Equation number (4) (5)

Efficiency conditions for firms Rk = r k /PK Rk 1−F (ω)−μωF  (ω)] 1−F (ω) R[ = 1−(ω) Rk 1−

(6)

PK K Pk k

(7)

Pk =

(8)

1 = PK f 

(9)

=

1−

I  k

K = 1+ f

contract efficiency condition

R [(ω)−μG(ω)] 1

Rk [(ω)−μG(ω)] R   PK 1 + f kI − f 



Economic description of condition rate of return on capital

 I  k

k

mutual fund zero-profit condition I  I  efficiency condition of k k capital producers optimality of I choice by capital producers capital accumulation technology

Equations (1)–(9) represent nine equations in nine private sector equilibrium objects: C, c, R, Rk , K , ω, Pk , PK , I. The equilibrium value of T can be backed out by imposing either the household or government budget constraint. Note too that the equilibrium rate of interest on banks Z is determined from (6.3) and the facts, (C.27), N = Pk k.

322

Lawrence J. Christiano and Daisuke Ikeda

It is of interest to show that the equilibrium allocations are first-best when μ = 0. In this case, Eq. (5) can be satisfied only with R = Rk and ω disappears from that equation. Combining Eqs. (6), (7), and (9), we obtain  γ 1 + kI 1  γ . = 1 − (ω) 1 + (1 − γ ) kI We can think of this equation as defining ω and, hence, Z (i.e., the spread). But this variable does not enter the other equations, and so it plays no role in determining the quantity allocations. The other equations are (1), (2), and (3). Expressing these with μ = 0 and using (8) and (9), we obtain  α  γ −1 , (1) β1 Cc = (1 + τ )r k γ kI   I γ   k (2) C = r k 1 + k , (3)

c + I = y,

where a prime indicates that the equation has been adjusted using (8) or (9). The preceding system represents three equations in three unknowns, c, C, and I. We substitute out I by using (3): (1) (2)

 γ −1 = (1 + τ )r k γ y−c k     γ y−c . C = rkk 1 + k 1 β

 C α c

In sum, the system can be solved as follows. First, solve (1) and (2) for C and c. Then, I = y − c and PK can be computed from (8). Finally, (4) and (5) imply that R = Rk =

rk . PK

We define the first-best problem as the problem for a planner who observes the bankers’ ω realizations. Such a planner obviously does not have to pay monitoring costs. The problem of this planner is max u(c) + βu(C ),   y−c γ . C = rkk 1 + k Expressing this in Lagrangian form, we have     y−c γ k −C . max u(c) + βu(C ) + λ r k 1 + k

(C.28)

Government Policy, Credit Markets, and Economic Activity

The first-order conditions are





u (c) = λr γ k

y−c k

323

γ −1 ,

βu (C ) = λ, so that the necessary and sufficient conditions for optimization reduce to marginal return on investment

  γ −1 y − c , rkγ k γ  y−c . C = rkk 1 + k

u (c) = βu (C )





Evidently, these equations coincide with (1) and (2) when τ = 0. We conclude that the equilibrium supports the first-best consumption and investment allocations. We summarize our findings as follows: Proposition C.3: Suppose 0 < γ ≤ 1. When monitoring costs are zero, then the equilibrium consumption and investment allocations coincide with the solution to the first-best problem.

C.2.5. Computation of Private Sector Equilibrium Here we describe an algorithm for computing the private sector equilibrium conditional on an arbitrary value of τ. Unlike in the previous subsection, in this subsection we take the model parameters as given and compute values for the nine model endogenous variables that satisfy the nine model equations, (1)–(9) in section C.2.4. Our strategy is to find R˜ ≡ RPK such ˜ = 0. To define the mapping g from R˜ into the real line, we fix a that g(R) ˜ We solve for ω by using (5) and value for R. rk Rk rk = = . R RPK R˜ Next, we compute the object on the right-hand side of (6) and call the result X. Then, (6) is written as    γ 1 + f kI 1 + kI PK K     =  γ = X. = Pk k 1 + f kI − f  kI kI 1 + (1 − γ ) kI Rewriting this expression, we obtain  γ I X −1 = k 1 − X (1 − γ )

324

Lawrence J. Christiano and Daisuke Ikeda

and solve for I/k. Then, we solve (9) for K/k, (8) for PK , and (7) for Pk . Finally, R=

R˜ . PK

We combine (1), (2), and (3): 1



c(β[1 + τ ]R) α = r k K [1 − μG(ω)],

1  y − I (β [1 + τ ] R) α = r k K [1 − μG (ω)] ,

1 I y − (β[1 + τ ]R) α = r k Kk [1 − μG(ω)]. k k

We solve the latter for y/k: 1

I (β[1 + τ ]R) α + r k Kk [1 − μG(ω)] y = k . 1 k (β[1 + τ ]R) α

˜ and C(R). ˜ We solve (2) and (3) for c,C. In this way, we define mappings c(R) Then, we let 1

˜ ≡ C(R) ˜ − c(R)(β[1 ˜ g(R) + τ ]R) α . ˜ = 0, that is, (1) is satisfied. We adjust R˜ until g(R) C.2.6. Net Worth Transfers to Bankers Here, we consider a policy in which the government raises taxes in the first period and then simply gives the banks the proceeds as a subsidy. With this policy, the government in effect can cause the economy to fully circumvent the financial frictions and move to the first-best allocations. Suppose the government raises T > 0 in the first period and transfers the proceeds as a lump-sum gift to banks. Let I ∗ denote the first-best level of investment [see (C.28)]. The optimal policy is to set T = I ∗ . In this case, the banks have enough net worth that they do not need to borrow from mutual funds. In this way there are no monitoring costs. It is easy to verify that with one exception the equilibrium conditions of the model considered here coincide with Eqs. (1)–(9) in section equilibrium. The only change is that the object on the left-hand side of the equality in (6) is replaced by PK K . Pk k + T

Government Policy, Credit Markets, and Economic Activity

325

This is the leverage ratio of the bankers, and the preceding expression reflects that, with the policy considered here, bankers’ net worth consists of the value of the capital they are endowed with, plus the tax transfer received from the government. If we set T = I, then leverage is unity by the zeroprofit condition on capital producers, (C.16). But if the left-hand side of (6) is unity, the right-hand side must be too. This can be accomplished by ω = 0, because in this case (ω) = F (ω) = G(ω) = 0 [see (6.7)]. Equation (5) then requires Rk /R = 1. The remaining equations coincide with the equations of the first-best equilibrium, and so these can be satisfied by setting T = I ∗ . We summarize these results as follows: Proposition C.4: A policy of lump-sum tax-financed transfers of net worth to bankers can support the first-best allocations. References Ajello, A. (2010). Financial intermediation, investment dynamics and business cycle fluctuations. Unpublished doctoral dissertation, Northwestern University, Evanston, IL. Akerlof, G. A. (1970). The market for “lemons”: Quality uncertainty and the market mechanism. Quarterly Journal of Economics, 84(3), 488–500. Arellano, C., Bai, Y., & Kehoe, P. (2010). Financial markets and fluctuations in uncertainty (Research Department Staff Report). Minneapolis, MN: Federal Reserve Bank of Minneapolis. Barro, R. J. (1974). Are government bonds net wealth? Journal of Political Economy, 82, 1095–1117. Bernanke, B., & Gertler, M. (1990). Financial fragility and economic performance. Quarterly Journal of Economics, 105(1), 87–114. Bernanke, B., Gertler, M., & Gilchrist, S. (1999). The financial accelerator in a Quantitative business cycle framework. In J. B. Taylor & M. Woodford (Eds.), Handbook of macroeconomics (pp. 1341–1393). Amsterdam, New York, and Oxford, UK: Elsevier Science, North-Holland. Bianchi, J. (2011). Overborrowing and systemic externalities in the business cycle. American Economic Review, 101, 3400–3426. Bigio, S. (2011). Endogenous liquidity and the business cycle. New York: New York University. Bloom, N. (2009). The impact of uncertainty shocks. Econometrica, 77(3), 623–685. Bloom, N., Floetotto, M., & Jaimovich, N. (2010). Really uncertain business cycles. Unpublished manuscript, Durham, NC: Duke University. Carlstrom, C. T., & Fuerst, T. (1997). Agency costs, net worth and business fluctuations: A computable general equilibrium analysis. American Economic Review, 87, 893–910. Chari, V. V., Shourideh, A., & Zetlin-Jones, A. (2010). Analyzing policies to repair credit markets. Minneapolis, MN: Federal Reserve Bank of Minneapolis. Christiano, L. J., Motto, R., & Rostagno, M. (2003). The Great Depression and the Friedman-Schwartz hypothesis. Journal of Money, Credit and Banking, 35(6), 1119– 1198.

326

Lawrence J. Christiano and Daisuke Ikeda

Christiano, L. J., Motto, R., & Rostagno, M. (2010). Financial factors in business cycles (Working Paper No. 1192). Frankfurt, Germany: European Central Bank. Christiano, L. J., Motto, R., & Rostagno, M. (2011). Risk shocks. Evanston, IL: Department of Economics, Northwestern University. Christiano, L. J., Trabant, M., & Walentin, K. (2011). DSGE models for monetary policy analysis. In B. M. Friedman & M. Woodford (Eds.), Handbook of monetary economics (Vol. 3A). Amsterdam, The Netherlands: North-Holland. Curdia, V., & Woodford, M. (2009). Credit frictions and optimal monetary policy (BIS Working Papers No. 278). Basel, Switzerland: Bank for International Settlement. Del Negro, M., Eggertsson, G., Ferrero, A., & Kiyotaki, N. (2010). The Great Escape? A quantitative evaluation of the Fed’s non-standard policies. New York: Federal Reserve Bank of New York. Dib, A. (2010). Capital requirement and financial frictions in banking: Macroeconomic implication. Ottawa: Bank of Canada. Eggertsson, G., & Woodford, M. (2003). The zero interest-rate bound and optimal monetary policy. Brookings Papers on Economic Activity, 2003(1), 139–211. Eisfeldt, A. (2004). Endogenous liquidity in asset markets. Journal of Finance, 59(1), 1–30. Fisher, J. D. M. (1999). Credit market imperfections and the heterogeneous response of firms to monetary shocks. Journal of Money, Credit and Banking, 31(2), 187–211. Fishman, M., & Parker, J. (2010). Valuation and the volatility of financing and investment. Evanston, IL: Northwestern University. Fuerst, T. (1994). The availability doctrine. Journal of Monetary Economics, 34, 429– 443. Gagnon, J., Raskin, M., Remache, J., & Sack, B. (2010). Large-scale asset purchases by the Federal Reserve: Did they work? New York: Federal Reserve Bank of New York. Gertler, M., & Karadi, P. (2009). A model of unconventional monetary policy. New York: New York University. Gertler, M., & Kiyotaki, N. (2011). Financial intermediation and credit policy in business cycle analysis. In B. M. Friedman & M. Woodford (Eds.), Handbook of monetary economics (Vol. 3A). Amsterdam, The Netherlands: North-Holland. Hirakata, N., Sudo, N., & Ueda, K. (2009a). Chained credit contracts and financial accelerators (Discussion Paper Series E-30). Tokyo: Bank of Japan. Hirakata, N., Sudo, N., & Ueda, K. (2009b). Capital injection, monetary policy, and financial accelerators. Tokyo: Bank of Japan. Hirakata, N., Sudo, N., & Ueda, K. (2010). Do banking shocks matter for the U.S. economy? Tokyo: Bank of Japan. Holmstrom, B., & Tirole, J. (1997). Financial intermediation, loanable funds, and the real sector. Quarterly Journal of Economics, 112, 663–691. House, C. L. (2006). Adverse selection and the financial accelerator. Journal of Monetary Economics, 53, 1117–1134. Ikeda, D. (2011a). Adverse selection, uncertainty shocks and business cycles. Unpublished doctoral dissertation, Northwestern University, Evanston, IL. Ikeda, D. (2011b). Adverse selection, uncertainty shocks and monetary policy. Unpublished doctoral dissertation, Northwestern University, Evanston, IL. Jermann, U., & Quadrini, V. (2010). Macroeconomic effects of financial shocks. Los Angeles, CA: University of Southern California Press.

Government Policy, Credit Markets, and Economic Activity

327

Kiyotaki, N., & Moore, J. (2008). Liquidity, business cycles, and monetary policy. Princeton, NJ: Princeton University Press. Korinek, A. (2011). Systemic risk-taking: Amplification effects, externalities, and regulatory responses. College Park, MD: University of Maryland. Krishnamurthy, A., & Vissing-Jorgensen, A. (2010). The effect of purchasing long-term securities on interest rates. Evanston, IL: Kellogg School of Management, Northwestern University. Kurlat, P. (2010). Lemons, market shutdowns and learning. Stanford, CA: Stanford University. Liu, Z., Wang, P., & Zha, T. (2010). Do credit constraints amplify economic fluctuations? (Working Paper No. 2010-1). Atlanta, GA: Federal Reserve Bank of Atlanta. Lorenzoni, G. (2008). Inefficient credit booms. Review of Economic Studies, 75, 809–833. Mankiw, N. G. (1986). The allocation of credit and financial collapse. Quarterly Journal of Economics, 101(3), 455–470. Meh, C., & Moran, K. (2010). The role of bank capital in the propagation of shocks. Journal of Economic Dynamics and Control, 34, 555–576. Mendoza, E. G. (2010). Sudden stops, financial crises and leverage. American Economic Review, 100, 1941–1966. Moore, J. (2009). Contagious illiquidity. Plenary lecture given at the XIV Annual Meeting of the Latin American and Caribbean Economic Association, and the XXV Latin American Meeting of the Econometric Society, Buenos Aires, October 1–3. Nowobilski, A. (2011). Liquidity risk and the macroeconomic costs of financial crises. Evanston, IL: Department of Economics, Northwestern University. Shimer, R. (2010). Wage rigidities and jobless recoveries. Chicago: University of Chicago. Stiglitz, J. E., & Weiss, A. (1981). Credit rationing in markets with imperfect information. American Economic Review, 71, 393–410. Stiglitz, J. E., & Weiss, A. (1992). Asymmetric information in credit markets and its implications for macro-economics (Oxford Economic Papers, New Series, Vol. 44, No. 4, Special Issue on Financial Markets, Institutions and Policy, October, pp. 694– 724). New York: Oxford University Press. Taylor, J. B., & Williams, J. C. (2009). A black swan in the money market. American Economic Journal: Macroeconomics, 1, 58–83. Townsend, R. (1979). Optimal contracts and competitive markets with costly state verification. Journal of Economic Theory, 21, 265–293. Ueda, K. (2009). Banking globalization and international business cycles. Tokyo: Bank of Japan. Wallace, N. (1981). A Modigliani-Miller theorem for open-market operations. American Economic Review, 71(3), 267–274. Williamson, S. (1987). Financial intermediation, business failures and real business cycles. Journal of Political Economy, 95, 1196–1216. Zeng, Z. (in press). A theory of the non-neutrality of money with banking frictions and bank recapitalization. Economic Theory.

Notes * Prepared for the conference, “A Return to Jekyll Island: the Origins, History, and Future of the Federal Reserve,” sponsored by the Federal Reserve Bank of Atlanta

328

1

2

3

4

5 6 7 8

9 10

11

12

Lawrence J. Christiano and Daisuke Ikeda and Rutgers University, November 5–6, 2010, Jekyll Island, GA. We are grateful to Dave Altig, Toni Braun, Marty Eichenbaum, and Tao Zha for discussions. We are especially grateful for detailed comments and suggestions from Andrea Ajello, Lance Kent, and Toan Phan. We examined monthly data on the interest rate on BAA and AAA rated bonds taken from the St. Louis Federal Reserve bank website. In December 2008 the interest rate spread on BAA over AAA bonds peaked at 3.38 percent, at an annual rate. This is a higher spread than was observed in every month since 1933. The extent to which balance sheets became impaired was hard to assess because there did not exist clear market values for many of the financial assets in the balance sheets of financial institutions. See, for example, Gagnon, Raskin, Remache, and Sack (2010). For a less sanguine perspective on the effectiveness of the Fed’s policy, see Krishnamurthy and VissingJorgensen (2010) and Taylor and Williams (2009). There is now a large literature devoted to constructing quantitative dynamic, stochastic general equilibrium models for evaluating the consequences of government asset purchase policies. For a partial list of this work, see Ajello (2010), Bernanke, Gertler, and Gilchrist (1999), Carlstrom and Fuerst (1997), Christiano, Motto, and Rostagno (2003, 2010, 2011), Curdia and Woodford (2009), Del Negro, Eggertsson, Ferrero, and Kiyotaki (2010), Dib (2010), Fisher (1999), Gertler and Karadi (2009), Gertler and Kiyotaki (2011), Hirakata, Sudo, and Ueda (2009a, 2009b, 2010), Liu, Wang, and Zha (2010), Meh and Moran (2010), Nowobilski (2011), Ueda (2009), and Zeng (in press). The Moore and Kiyotaki and Moore ideas are pursued quantitatively in Ajello (2010) and Del Negro et al. (2010). By a “bank” we mean any institution that intermediates between borrowers and lenders. For other analyses in this spirit, see Holmstrom and Tirole (1997) and Meh and Moran (2010). The model (and others in this chapter) assumes that banks cannot increase their net worth. The model offers no explanation for this. The assumption does appear to be roughly consistent with observations. In a private communication, James McAndrews shared the results of his research with Tobias Adrian. That work shows that bond issuance by financial firms declined sharply in the recent crisis, whereas equity issuance hardly rose. In addition, we assume that – unlike the bankers in the model – government employees do not have the opportunity to abscond with tax revenues. There are several analyses of the recent credit crisis that focus on adverse selection in credit markets. See, for example, Chari, Shourideh, and Zetlin-Jones (2010), Fishman and Parker (2010), House (2006), Ikeda (2011a, 2011b), and Kurlat (2010). In addition, see also the argument in Shimer, http://gregmankiw.blogspot.com/2008/ 09/case-against-paulson-plan.html. Eisfeldt (2004) presents a theoretical analysis that blends adverse-selection and liquidity problems. In particular, equilibrium in the market for credit to banks is a pooling equilibrium. One reason why equilibrium involves pooling is that the environment has the property that the quantity of funds that banks must borrow is fixed. For a prominent example, see BGG. Another example is given by the Christiano et al. (2003, 2010, 2011) analysis of the U.S. Great Depression and the past three decades

Government Policy, Credit Markets, and Economic Activity

13

14

15 16 17 18 19 20 21

22 23

24

329

of U.S. and Euro Area business cycles. An earlier DSGE model application of the costly state verification and costly monitoring idea can be found in the influential contribution by Carlstrom and Fuerst (1997), as well as in Fuerst (1994). For another contribution of this idea in a DSGE model, see Jonas Fisher’s 1994 Northwestern University doctoral dissertation, published in Fisher (1999). Finally, see Williamson (1987). Herein lies a sharp distinction between the model analyzed here and the one in BGG. In BGG, the asymmetric information and monitoring costs lie on the asset side of the bank balance, that is, between the bank and the firm to which it supplies funds. In addition, the bank is perfectly diversified across firms so that in BGG, banks are perfectly safe. Other modifications of the BGG model that introduce risk in banking include, for example, Hirakata et al. (2009a, 2009b, 2010), Nowobilski (2011), Ueda (2009), and Zeng (in press). To our knowledge, the first papers to consider the economic effects of variations in microeconomic uncertainty are Williamson (1987) and Christiano et al. (2003). More recently, this type of shock has also been considered in Arellano, Bai, and Kehoe (2010), Bigio (2011), Bloom (2009), Bloom, Floetotto, and Jaimovich (2010), Christiano et al. (2010, 2011), Ikeda (2011a, 2011b), Jermann and Quadrini (2010), and Kurlat (2010). What we are calling the Barro-Wallace irrelevance proposition is applied to government purchases of long-term debt in Eggertsson and Woodford (2003). The logic in the text may also provide the foundation for a theory of the effectiveness of sterilized interventions in the foreign exchange markets. This section is based on the joint work with Tao Zha. So profits per unit of bank deposits rise when banker net worth is low. However, total bank profits may be low because of the lower net worth of the banks. We assume the environment is such that c < y. Whether the “securities” take the form of loans or equity is the same in our model. Given that a bank undertakes a costly search to find a good firm, it would be interesting to explore an alternative model formulation in which the firm and bank that find each other engage in bilateral negotiations. Bankers do not have access to funds other than their own and those provided by depositors. If instead profits were positive, mutual funds would set d = ∞, but this exceeds the resources of households who make the deposits. If a positive value of d produced negative profits, then profit-maximizing mutual funds would earn zero profits by setting d = 0. But this would be less than the positive amount of deposits supplied by households in the interior equilibria that we study. The d equation in (4.21) is a simplified version of the d equation in (4.20), obtained as follows. Substitute from (4.11) and the Rgd and μ equations in (4.20) into the d equation in (4.20) to obtain, after some algebra,   0 = (λ − ν ) p(e)Rg + (1 − p(e))Rb − R .

The result follows from the observation that (λ − ν ) is strictly positive because ν ≤ 0 and λ > 0 in an interior equilibrium. The e equation in (4.21) is a simplified version of the e equation in (4.20), after making use of (4.11) and the ν equation in (4.20). 25 This requires performing substitutions similar to those in (A.9).

330

Lawrence J. Christiano and Daisuke Ikeda

26 See the d and μ equations in (4.21). 27 Although unmodeled, this might reflect that banks must expend a fixed cost to evaluate a firm’s project. 28 Our model has the property that the only equilibrium is a pooling equilibrium, that is, one in which banks with different risks receive the same loan contract from mutual funds. For additional discussion of pooling and separating equilibria under adverse selection in credit markets, see Stiglitz and Weiss (1992) and the references they cite. 29 Our model is most closely related to the one in Mankiw (1986). 30 The letter e is not to be confused with effort in the previous section. 31 Our model is an adaptation of the model in Mankiw (1986), especially the example on p. 463. 32 Clearly, an empirically plausible version of our model would require a density function for p that places greater mass on higher p. 33 This phenomenon is captured by a quote from Adam Smith’s Wealth of Nations, cited in Stiglitz and Weiss (1992). According to Stiglitz and Weiss, Adam Smith wrote that if the interest rate was fixed too high, “ . . . the greater part of the money which was to be lent, would be lent to prodigals and profectors . . . Sober people, who will give for the use of money no more than a part of what they are likely to make by the use of it, would not venture into the competition.” 34 A sufficient condition is that, in addition to N < 1, the parameters satisfy

1 1 (y + eN )(θβ ) γ ≤ e (θβ ) γ + θ . 35 For other asymmetric information and monitoring cost models applied specifically to banking, see Hirakata et al. (2009a, 2009b, 2010), Nowobilski (2011), and Zeng (in press). 36 Note that a collection (ω, B) is equivalent to a collection (Z, B) by ω = ZB/[(N + B)r k ] and the fact that N and r k are exogenous to the banks at the time the contract is undertaken. 37 Here, we have used   (ω) = 1 − F (ω) and G  (ω) = ωF  (ω). See BGG for a formal discussion of the fact that (6.10) uniquely characterizes the solution to the bank optimization problem. A sufficient condition is the regularity condition subsequently defined in (6.23). 38 As noted previously, by the law of large numbers, the expected profits of individual banks, what we defined in (6.6) as π , also corresponds to aggregate profits (in per capita terms) for all the banks in the household. 39 In the numerical examples we have studied, we have found that when there exists a value of ω that solves (6.18) for a given B, that value of ω is unique. 40 We generally ignore imposing the usual nonnegativity constraints such as B ≥ 0, y − B ≥ 0, to minimize clutter and in the hope that this does not generate confusion. Throughout, we assume that model parameters are set so that nonnegativity constraints are nonbinding. 41 In his analysis of costly state verification problems, Fisher (1999) also emphasizes that the quantity of lending is determined by the average, not the marginal, return on a loan. 42 See Gertler and Kiyotaki (2011) and Shimer (2010) for other studies that model the shock that triggers the recent financial crisis as a drop in wealth.

Government Policy, Credit Markets, and Economic Activity

331

43 Recall that all variables are in household per capita terms. 44 We suspect that the environment of this section is not an interesting one for considering equilibria in which the constraint, b ≥ 0 is binding. 45 See Christiano et al. (2011) for a review. 46 Recall that λ − ν > 0 because λ, − ν > 0. 47 We implicitly ignore another option for the bank: deposit N in the mutual fund and then borrow from the mutual fund and invest in the project. A property of equilibrium is that R ≤ θ , so that no bank could increase profits by choosing this option. 48 To see that λ ≥ 0, suppose on the contrary that λ < 0. Note that (0) = G(0) = 0, so that the solution to the Lagrangian representation of the problem solved by the loan contract is B = ∞ and ω = 0. This does not solve the problem of maximizing (C.5) subject to (C.6) because (C.6) is violated. 49 We have used the following result. Suppose C A λ= = . B D Then λ = (A + C )/(B + D). 50 Condition (i) reflects a combination of the result at the top of p. 1382 in BGG, as well as the observation at the top of p. 1385. Condition (ii) is established on p. 1382 in BGG. 51 Note that a collection (ω, B) is equivalent to a collection (Z, B) by ω = ZB/[(N + B)Rk ] and the assumption that banks view N and Rk as parametric.

SIX

Policy Debates at the Federal Open Market Committee: 1993–2002 Marvin Goodfriend1

I. Introduction The consideration of a variety of points of view and the absorption of new thinking over time are two of the Federal Open Market Committee’s (FOMC) primary responsibilities. The range of perspectives brought to the committee by its individual members is one the committee’s great strengths. At each meeting, nineteen participants – twelve voting members plus the seven nonvoting reserve bank presidents – have ample opportunity to present their respective views on U.S. and global economic and financial conditions, monetary policy, and other matters that may be brought before the committee. Because individual members hold widely varying views on these matters, committee discussions are typically substantive and robust. Whereas the general public naturally and appropriately focuses primarily on the Fed chairman, other FOMC participants, including the reserve bank presidents, are well qualified to make meaningful contributions to policy deliberations and do so. This chapter is a case study of how one FOMC participant, J. Alfred Broaddus, Jr., president of the Federal Reserve Bank of Richmond from 1993 to mid-2004, brought the Richmond Fed’s perspective to bear on several important policy questions concerning the committee.2 Broaddus sought consistently to address questions of concern to the FOMC with the analytical tools of modern monetary and financial economics. He worked closely with economists in the research department at the Richmond Fed to draft positions for every policy question considered by the committee. The story is told through Broaddus’s policy statements – and on one occasion a presentation by the author, who attended FOMC meetings as Broaddus’s policy advisor. It also presents the reactions to these statements of Alan Greenspan, who was Fed chairman throughout the period, and other 332

Policy Debates at the Federal Open Market Committee: 1993–2002

333

meeting participants. This chapter covers seven policy debates that were of particular concern to the Richmond Fed: transparency and communication, preemptive interest rate policy, lending to Mexico, foreign exchange intervention, inflation targeting, Fed asset acquisition, and monetary policy at the zero interest bound.3 The account is based on statements, presentations, and debate preserved in lightly edited transcripts of Federal Open Market Committee meetings.4 Among other things, the debate illustrates nicely the interplay between theoretical and practical policy matters and shows how advances in the theory and practice of central banking were mutually reinforcing during the period. Although the seven issues considered in the paper arose at different times during the period, in retrospect they all illustrate a common principle: Fed (and other central bank) policies have lasting effectiveness only if the policies are credible to the public, that is, the public is confident that the Fed’s actions are free of political influence or manipulation and seek consistently to advance attainment of the Fed’s central mandates of maintaining price stability and promoting maximum sustainable economic growth. Broaddus’s positions on issues such as the loans to Mexico, foreign exchange market intervention, and Fed asset acquisition, in particular, were all motivated by the need to protect the Fed’s independence from political interference, both in fact and in perception.

II. Transparency and Communication The story begins with the FOMC’s historic decision in February 1994 to announce its federal funds rate policy actions without delay for the first time, thereby taking responsibility fully and publicly for the course of shortterm interest rates. The decision was taken after months of inquiry into Fed secrecy by Henry Gonzales, chairman of the House Banking Committee. The demand for Fed transparency had been building since the Volcker Fed grabbed the headlines and took responsibility for inflation in the early 1980s. Academic work helped prepare the way. John Taylor characterized interest rate policy as the “Taylor Rule” in a 1993 paper.5 With Broaddus’s support and encouragement, the author argued against central bank secrecy in a 1996 paper and documented the role of interest rates in the conduct of monetary policy in a 1991 paper.6 It is important to distinguish between two aspects of the FOMC’s information policy. One involves the disclosure of the process by which the FOMC arrives at its decisions. Another involves communication undertaken

334

Marvin Goodfriend

to accompany and enhance the effectiveness of its policy actions. The FOMC reformulated both aspects of its information policy in the early 1990s. In late 1992, Gonzalez began a campaign to improve the transparency of the FOMC’s disclosure practices. A primary concern of Gonzalez was the discontinuance of the Memoranda of Discussion (MOD) of FOMC meetings. The MOD, which dated back to the establishment of the modern FOMC in 1936, were edited narratives of points made by named speakers at FOMC meetings. In 1970, the FOMC began to release the MOD with a five-year lag, eventually making the entire MOD record public. However, the FOMC discontinued the preparation of the MOD in March 1976 in response to a Freedom of Information Act (FOIA) request.7 A letter from Gonzales to Greenspan on October 8, 1992, expressed regret at the discontinuance of the MOD in 1976 and requested the Federal Reserve Board’s opinion on releasing a videotape of FOMC meetings with a two-month lag. Gonzales also called inadequate the lack of released FOMC minutes and complained that the FOMC then released its current policy stance only after the subsequent FOMC meeting. The FOMC formed a subcommittee on disclosure policy in February 1993 to address these concerns. At its March 1993 meeting the committee decided to combine two and then currently released documents into one that would be released on the same schedule but more visibly on the Friday after the subsequent FOMC meeting. These highly edited, not for attribution, reports of policy discussions and actions at FOMC meetings have evolved today into minutes released roughly three weeks after each FOMC meeting. In September 1993, Gonzales invited the entire Federal Reserve Board and all twelve reserve bank presidents to testify on a bill that he had introduced. Among other things, the bill would have required a transcript and videotape of each FOMC meeting to be made public within 60 days and that any policy action be made public within a week. The invitation asked about notes or records that FOMC participants made of FOMC meetings. On October 5 and 15, two FOMC conference calls were held in preparation for the congressional testimony. At the October 5 meeting, Greenspan informed the committee that “raw materials still existed for drafting the MOD going back a number of years.”8 The existence of a nearly complete set of verbatim written transcripts of past FOMC meetings came out subsequently in Greenspan’s congressional testimony and in the media. Greenspan’s concern was to avoid “premature, detailed disclosure of our deliberations” that would compromise “openness and free exchange of views so essential to monetary policy.”9 Increasingly the Fed appeared in an unfavorable light with regard to its disclosure policy.10

Policy Debates at the Federal Open Market Committee: 1993–2002

335

As the FOMC reconsidered the disclosure policy for its deliberative process, events forced its hand on the disclosure of its policy actions. When the FOMC considered taking preemptive interest rate policy actions against inflation at its February 1994 meeting, Greenspan asked that its interest rate actions be made fully transparent: . . . This really gets to the issue that when we move in this particular context, which of course will be the first time we have moved since September 1992, We are going to have to make our action very visible. . . . I am particularly concerned that if we choose to move tomorrow, we make certain that there is no ambiguity about our move. . . . In the circumstances a draining [of reserves] action will be unambiguous to the professionals. But I’m not sure that more widespread recognition will come out very quickly; it will sort of dribble out. Under ordinary circumstances, that is not only fine but desirable. One of the things that we have argued . . . is that there is a distinction between a discount rate and a federal funds rate action in the sense that we don’t want an announcement effect ordinarily on the funds rate. It gives us a much more calibrated instrument.[11 ] But a federal funds change in this particular instance is a discount rate change, as far as the Federal Reserve System is concerned. I am very strongly inclined to make it clear that we are doing this but to find a way to do it that does not set a precedent. . . . we are going to have to deal with this issue in conjunction with the question of transcripts and tapes and all of the other disclosure issues. . . . So I’m caught in this particular situation where I would feel very uncomfortable if when we make a move . . . we do not make it very clear that we are moving . . . I would very much like to have the permission of the Committee to announce that we’re doing it and to state that the announcement is an extraordinary event.

This was a dramatic moment for those in the room like the author who were aware of the long-standing reluctance of the Fed to be fully clear about its interest rate policy, and for those who thought more openness was necessary and beneficial. Greenspan did not present his proposal as revolutionary. But by announcing its federal funds rate action explicitly, the Fed would take full public responsibility for the level of short-term interest rates. And by announcing its federal funds rate action immediately, the Fed would make its federal funds rate action front-page news for the first time and likely forever. Despite Greenspan’s pretense to the contrary, the step would be revolutionary. There would be no turning back. The market would become evermore demanding of FOMC interest rate policy concerns and intentions. In time, the FOMC would talk explicitly in terms of its intended federal funds rate target and sharpen the forward guidance on its rate intentions. The FOMC would become increasingly ambitious in managing interest rate expectations. The following statement was released immediately following the decision to raise the federal funds rate by twenty-five basis points:

336

Marvin Goodfriend

Chairman Greenspan announced today that the Federal Open Market Committee decided to increase slightly the degree of pressure on reserve positions. This action is expected to be associated with a small increase in short-term money market interest rates. The decision was taken to move toward a less accommodative stance in monetary policy in order to sustain and enhance the economic expansion. Chairman Greenspan decided to announce this action immediately so as to avoid any misunderstanding of the Committee’s purposes given the fact that this is the first firming of reserve market conditions by the Committee since early 1989.

The statement was transitional in nature, continuing to speak of “pressure on reserve positions” as it spoke of the FOMC’s interest rate intentions. In spite of implying that the explicitness about its action was exceptional, the FOMC has announced interest rate policy actions explicitly and immediately ever since. The FOMC was inclined to be more open about its policy actions for three practical reasons. First, more timely announcements of policy actions would not impair the FOMC’s deliberative process, which it was most anxious to protect. Second, to do otherwise would be to invite leaks of its intended policy stance. Third, continued delayed announcement of its interest rate policy stance would feed increasingly unfavorable opinion about Fed secrecy building in Congress and the media. More fundamental, by 1994 the ground had been prepared for enhanced transparency of the Fed’s interest rate policy actions. The Fed was more inclined to talk openly in terms of interest rate policy because academics had begun to do so a few years earlier and an academic literature had developed indicating that communication could enhance the effectiveness of interest rate policy. The Fed was positioned to be more transparent because under Volcker and Greenspan it had put in place a strategy based on maintaining price stability within which the FOMC could talk systematically and productively about policy. The days of go–stop policy – when interest rates were lowered to stimulate employment and increased only when unemployment had to rise to stabilize inflation – were over.12 On February 2, 1995, without any explicit guarantee from Congress to protect its deliberations from immediate release, the FOMC announced its intention to produce lightly edited transcripts of its meetings and release them with a five-year lag. Congress and the Fed have yet to reach an explicit agreement on delayed release. Nevertheless, a number of factors work to sustain that equilibrium. Most important is the fact that the FOMC has announced its interest rate policy actions immediately and explicitly since February 1994, and preemptive interest rate actions against inflation in 1994 worked to extend the economic expansion. It is worth noting that this

Policy Debates at the Federal Open Market Committee: 1993–2002

337

chapter could not have been written if the FOMC had not decided to restart the disclosure of its deliberative process in February 1995.

III. Preemptive Interest Rate Policy in 1994 The campaign of interest rate policy actions initiated at the February 1994 FOMC meeting to preempt rising inflation was only the second time that the committee had ever raised interest rates without a sustained prior increase of inflation. The first time was in 1983–4, when preemptive interest rate actions by the Volcker Fed succeeded in holding the line on inflation at 4 percent without pushing unemployment higher. But the Fed had allowed inflation to rise in the late 1980s after the 1987 stock market correction. The question was this: Could the Fed preempt rising inflation again without pushing unemployment higher? The committee was largely supportive of preemptive action. The problem was how to prepare markets for what would be the first preemptive interest rate policy actions of the Greenspan Fed. In pursuing preemptive policy in 1994, the committee wished to “hold the line” on inflation at 3 percent and anchor inflation expectations firmly. Given its new disclosure policy, the FOMC would take full responsibility for raising the federal funds rate sharply to 5 percent or 6 percent during the year from 3 percent. The magnitude, timing, and communication of interest rate policy would be particularly challenging for the FOMC in 1994.13 At issue in early February 1994 was whether to raise the federal funds rate by twenty-five or fifty basis points. Greenspan argued against doing 50 basis points: . . . I am very sympathetic with the view that we’ve got to move and that we’re going to have an extended period of moves, assuming the changes that are going on now continue in the direction of strength. I would be very concerned if this Committee went 50 basis points now because I don’t think the markets expect it. You want to hit a market when it needs to be hit; there is no significant evidence at this stage of imbalances that require the type of action that a number of us have discussed. Were we to go 50 basis points with the announcement effect and the shock effect, I am telling you that these markets will not hold still . . . I think there will be a time; and if the staff’s forecast is right, we can get to 150 basis points pretty easily. We can do it with a couple of 1/2 point jumps later when the markets are in the position to know what we’re doing and there’s continuity.14

Greenspan’s concerns about the market reaction to the FOMC’s twenty-five basis point federal funds rate action were justified. The FOMC’s twenty-five basis point action provoked an unusually large move in long bond rates. The

338

Marvin Goodfriend

thirty-year U.S. Treasury rate rose from a low of 5.8 percent in October 1993 to around 6.75 percent by the end of February.15 At the February 28 FOMC conference call, Greenspan pointed out that “[i]f you look at the pattern of long-term rates from December through the current period and you plot the German, French, American, and I suspect a couple of other rates, they fit remarkably closely. And it’s hard to argue that inflation expectations are really beginning to pick up in Europe.” He added that “My own guess at the moment is that this is more real than inflation expectations.”16 At the March 22nd FOMC meeting Greenspan added, Prior to our move on February 4th, the market had drifted into a state of somnambulance at low risk premiums, and there were steady upward price pressures. While we all recognized at the time that the stock market was a little dicey and we were worried about the mutual funds, I don’t think we were aware of the apparent underlying speculative elements involved in the markets on a worldwide basis that I think our February move unearthed. . . . We have seen some increase in yield spreads but they are still quite low by any historic standard, which suggests to me that the adjustment process in the capital markets, in the portfolios of pension funds, mutual funds, and individual households still has a long way to go.17

Interestingly, there was little mention of what had provoked the larger-thanexpected international increase in long-term bond rates in the aftermath of the FOMC’s February 25 basis point action.18 At the March 22 FOMC meeting, the author heard for the first time Greenspan observe that the economy was behaving differently than it had in the last few decades: . . . we have an economy which doesn’t look like anything that we have experienced in the last 30 years. In the last 30 years, when we saw a tightening of slack in the system, we also saw a pickup in credit demands and we had inflation. That, however, has not been the universal experience in this country. If we go back to the 1920s, there were several periods of significant tightness in markets but no significant credit expansion and no inflation; that was also the experience even in the 1950s. I raise this issue because we are pretty far along in this business cycle. So why is inflation not showing its head a little more?19

The author wondered whether the Fed’s commitment to price stability under Volcker and Greenspan was succeeding in anchoring inflation and inflation expectations firmly as under the “gold standard” of the 1920s and 1950s. Greenspan moved on to tactical questions. He was “hard pressed, he said, to remember when the outlook itself looked as unequivocally expansionary as it does today.”20 He reiterated the need to restore [interest rate] policy neutrality and acknowledged in effect that doing so meant to move in a way

Policy Debates at the Federal Open Market Committee: 1993–2002

339

somewhat divorced from data, catching up, so to speak, so that markets had trouble predicting the size and timing of the FOMC’s interest rate policy actions.21 Greenspan put the tactical problem this way: “ . . . so the question is, having very consciously and purposely tried to break the bubble and upset the markets in order to sort of break the cocoon of capital gains speculation, we are now in a position . . . to try to restore some degree of confidence in the System.22 Greenspan recommended that the Fed again move only twenty-five basis points at the March meeting; and the Fed moved another twenty-five basis points at an April conference call, still sensitive to upsetting markets. At the May 17 FOMC meeting Greenspan reached the conclusion that the FOMC’s three successive twenty-five basis point federal funds rate actions – in February, March, and April – had re-created a healthy degree of uncertainty in markets and readjusted speculative security holdings from weak hands into firmer hands, as he put it. He concluded that “we have the capability I would say at this stage to move more strongly than we usually do without the risk of cracking the system.”23 Given that the committee was in agreement on the need to take decisive action on rates and disagreed only on how much the financial system could take before its “tensile strength” broke, the committee raised the federal funds rate fifty basis points at its May meeting. The FOMC moved the funds rate up another fifty basis points at its August 16 meeting, saying that it expected the August action to be “sufficient at least for a time to meet the objective of sustained, noninflationary growth.” When the committee declined to raise rates again at its September meeting, Broaddus dissented, arguing that he saw no compelling reason to delay another move, in part because rising inflation expectations reflected in a rising long-term bond rate would put the Fed in a situation in which all the choices were bad.24 At the November 15 FOMC meeting following the midterm elections, Greenspan acknowledged . . . we are behind the curve . . . I think that creating a mild surprise would be of significant value . . . I suspect that while the majority think we are going to do 50, the vast majority will think that that is not enough and they will immediately price an additional 50 or more basis points in the December forward contracts. In my judgment we would be risking – a low probability risk but a potentially very large outcome if it were to happen – a run on the dollar, a run on the bond market, and a significant decline in stock prices. . . . So I think that we have to be very careful at this stage and be certain that we are ahead of general expectations. I think we can do that with 75 basis points.25

340

Marvin Goodfriend

The thirty-year Treasury bond rate peaked in early November 1994 at around 8.2 percent. Presumably the speculative factors that concerned Greenspan had been worked out. Yet the bond rate remained elevated. Apparently, much of the run-up in bond rates during 1994 ultimately reflected fundamentals such as inflation expectations, which were stabilized after the FOMC moved rates up aggressively by seventy-five basis points in November. In December, Greenspan observed again how the recovery appeared to behave differently from the historic norm: I think the interesting question is why wages are not responding to what is a very rapidly tightening labor market. After speaking to some labor leaders and others who talk to their members and have a sense of this, I get the impression that longterm job insecurities are quite pervasive especially with respect to the portability of health insurance and pensions that make workers more cautious about changing jobs. . . . This is crucial because so long as that is the case and productivity is positive, unit costs are very well contained. Any endeavor to move final prices up in that environment induces competitors to come in and try to steal a firm’s market share, which erodes the firm’s pricing capability. So long as we have some evident flexibility in the system, then prices cannot readily move. . . . we are having difficulty getting banks to notice that interest rates are up. The interest rates on automobile installment paper are really lagging. Everybody is trying to protect market share, and this whole thing just doesn’t seem to be coming together. But it will. It always does. And the question is essentially pretty much when.26

Greenspan observed further, I think it is really worth recognizing that there is something quite different about the timing of this recovery. Ordinarily, a recovery has a much higher rate of growth in the early stages and slows in the later stages. Probably what is happening here is that we really didn’t have the classic movement to a cyclical recovery until well into the cycle, and we are probably now at effectively the earlier stages in a geriatric sense as distinct from the calendar . . . I wonder to what extent we can attribute all of this to monetary policy and monetary policy lags . . . the fact that inflation is relatively low may – despite all the discussions we have had about the inadequacy of the evidence – be contributing to improved productivity. If that is the case, we will get some greater growth in potential.27

The FOMC concluded its series of interest rate increases by raising the federal funds rate another fifty basis points to 6 percent at its January 31 – February 1, 1995 meeting. In recommending this action, Greenspan said that . . . an argument can be made to stay where we are at this particular time. That argument would have considerable force were it not for the fact that the markets expect a 50 basis point rise in the context of an exchange market for the dollar that has not been all that impressive. We know that to the extent that we choose to

Policy Debates at the Federal Open Market Committee: 1993–2002

341

go against market expectations, we create a degree of volatility; indeed, that is the purpose of going against the market. But there are times when doing so is probably unwise. And were we to hold still at this point we would in my view be taking unnecessary and undue risks.28

In the end, Fed interest rate policy in 1994–5 preempted rising inflation without pushing up unemployment. Talk of the death of inflation soon followed. Inflation expectations were anchored and the recovery extended. From the perspective of modern monetary theory, one can understand that outcome as the result of a demonstrated commitment by the Greenspan Fed to achieve price stability in the aftermath of the 1990–1 recession. The Greenspan Fed worked the inflation rate down from around 6 percent in 1990 to 3 percent in 1994, even allowing the unemployment rate to peak temporarily at 7.8 percent in June 1992 during the so-called “jobless recovery.” In 1994, it raised interest rates preemptively against inflation in the open against great political and market skepticism. By demonstrating its commitment to price stability, the Greenspan Fed credibly anchored expected and actual inflation. Interestingly, Greenspan did not then talk in terms of the Fed’s own credibility for low inflation. Yet, his observations about the changing nature of the business cycle were what monetary theory would predict as the central bank acquires such credibility.

IV. Lending to Mexico: 1994–5 At the March 1994 FOMC meeting the committee considered a request by Mexico for a substantial, permanent increase in its swap lines with the Federal Reserve and the U.S. Treasury, then $700 million and $300 million, respectively. The Fed and the Treasury ultimately decided to increase the combined swap lines to $3 billion dollars each. The increase in the swap line was linked to the recent passage of NAFTA and the planned shift in Mexico to an independent central bank. As initially contemplated, the Fed would rely on provisions of the U.S. Treasury to establish conditions under which Mexico could activate the swap line to ensure its repayment. The request triggered a lively debate within the FOMC. Broaddus set the terms: . . . The 1951 Treasury/Federal Reserve Accord established the principle that the Fed needs to be meaningfully independent within the government in order to conduct our monetary policy effectively. . . . We also know that this independence is granted only grudgingly and it’s under constant scrutiny in the Congress and elsewhere.

342

Marvin Goodfriend

Consequently, I think we ought to be very reluctant to take any action that might have even the appearance of abusing this independence lest we lose some or all of it. . . . As I understand it, the [swap] line was initially established back in 1967 to help deal with balance of payments issues and specifically to help support the then-fixed exchange rate between the peso and the dollar. That rationale no longer exists; in fact, the materials . . . distributed make it explicit that this facility would not be used for this purpose. So, it seems clear to me that any loan to Mexico in the current circumstances in essence would be a fiscal action of the U.S. government. And fiscal actions – expenditures of the government – are supposed to be authorized by Congress and Congress is supposed to authorize the funds. So whatever the general merits may be of making loans to Mexico, I don’t think we should be involved without explicit Congressional authorization.29

William McDonough, president of the Federal Reserve Bank of New York, disagreed: . . . one of the functions of the Federal Reserve is to seek monetary stability in a broader framework than just the American economy itself because of the obvious inter-linkages of world markets. . . . [Mexico] also is a country, being on our border, in which serious financial instability would have a very definite possibility of spreading across the border and creating problems in our own markets. . . . So I don’t share the view that it is something that is inconsistent with the role of the central bank and therefore would demand approved funding from Congress. I think it’s just as much a part of the responsibilities of the Federal Reserve as the swap line with Germany is.30

After considerable debate, Greenspan observed . . . this is a much more profound issue about the nature of the Federal Reserve System than I think we realize. . . . I think one can make a very strong argument for the central bank as a narrow institution that basically maintains strictly central banking operations. In that narrow sense, it’s not obvious to me that we would be involved in currency intervention. We would restrict ourselves strictly to domestic monetary policy. What this question is really all about is the stance of the central bank in a broader context, in other words a role in which we do get involved in issues which are in many aspects at the Treasury’s lead.31

The peso collapsed against the dollar on the foreign exchange market in December 1994. In January 1995, the Administration asked Congress to authorize a package of $40 billion dollars of U.S. government guarantees to back Mexican government securities. After Congress refused to authorize the package, the Administration asked the Fed to participate in a program with the U.S. Treasury’s Exchange Stabilization Fund (ESF) to provide financial support for Mexico. The U.S. Treasury asked the Fed’s participation because

Policy Debates at the Federal Open Market Committee: 1993–2002

343

liquid dollar assets on the ESF’s balance sheet were then only about $5 billion. At its January 31–February 1, 1995 meeting the FOMC agreed to raise from $3 billion to $20 billion its willingness to “warehouse,” deutsche marks and yen holdings for the ESF.32 The FOMC also agreed to show its support for the package by increasing its swap line from the $4.5 billion voted in December 1994 to $6 billion in return for a “take-out” in the form of a commitment from the Treasury that the ESF would take over any Federal Reserve obligation that was outstanding for more than 12 months. The administration’s request precipitated a spirited debate at the January 31 meeting that focused at first on the Treasury’s promised take-out. Tom Melzer, president of the Federal Reserve Bank of St. Louis, asked “[w]hat ability do the Treasury or the ESF have to take us out of an obligation if funds are not appropriated by Congress?” Greenspan later admitted “[t]here is a question here of whether or not the amount the United States Treasury gives us has to be appropriated funds, which I think is really where our examination of the issue has to be. In examining the take-out, we ought to make certain that we talk to them with respect to the question of what happens if they do not get the appropriated funds.”33 Ted Truman, director of the Division of International Finance at the Federal Reserve Board, pointed out that the ESF does not have appropriated funds. Greenspan asked “[a]re we going to be getting a take-out from the Exchange Stabilization Fund? Truman replied “I think that is what is in the program.”34 Truman observed that in the 1960s the Treasury floated Roosa bonds to obtain foreign currencies and used some of those currencies to take the Fed out in similar circumstances. The Roosa bonds were issued under the Treasury’s debt-management portfolio in a way that did not involve appropriated funds. The debate resumed on the morning of February 1 with Governor Larry Lindsey pointing out that the International Monetary Fund (IMF) released a statement the day before that the Fed, along with the IMF, would monitor developments closely in Mexico. But the Fed had not yet agreed to the package or the conditions that Mexico would be expected to fulfill to get the financial support. Truman answered: Our good friends at the Treasury apparently felt that the statement was needed for two reasons. . . . The first was to add to the credibility, if that’s the right word, of this revised proposal on Capitol Hill by continuing to assure certain members of Congress that we would be involved in the process. Secondly, they felt that the process would be somewhat less formal than would have been the case under the legislative approach, and therefore they apparently wanted to signal in the IMF’s press release that we – we the United States and we the Federal Reserve in particular –

344

Marvin Goodfriend

would be involved in the normal monitoring, if I can put it that way, and that the IMF would do the managing.35

Later on, the debate returned to the committee’s concern with the promised take-out. Finally, Greenspan assured the Committee Yes, it is a take-out. Precisely how it will be constructed is something which our General Counsel and their General Counsel will work on. . . . Basically, the agreement is that the Federal Reserve has zero credit risk and zero market risk; that is the principle. . . .

Greenspan proceeded to outline what he saw as the real problem: The real risks relate to the issue that Larry Lindsey is raising and implicitly what Ted is raising. The problem is the fact that we have managed to build up a high degree of credibility. The difficulty is that the Federal Reserve has now become the honest broker . . . as crucial players in the American government, I frankly don’t know how to get around accepting this responsibility when we are being asked by the Congress and the Administration to somehow oversee this operation, which is a very fuzzy deal. . . . 36

McDonough put the problem succinctly: . . . [e]ssentially, we are in a situation where our involvement has to be maintained and where we have to do the best we can to ensure that the terms for our involvement are done in the best possible way. This means some very tough conditionality.37

Melzer remained concerned that “ . . . [i]n effect, one could argue that we would be participating in an effort to subvert the will of the public, if you will.”38 Greenspan doubted that, arguing “the Republicans up on the Hill look at the Federal Reserve as the good guy . . . they are willing to be supportive of the President and the Administration only if we are involved. That is where our problem lies.” Greenspan later added “from the point of view of the country as a whole, it would have been irresponsible on our part not to be involved. When your neighbor is lying in the street, screaming, you can just walk over him and keep walking and not get involved, I grant you that. I do not think we have that choice.” 39 Broaddus responded: . . . I understand fully the very difficult position that you and the System have been put in. I take that in consideration. But I still just have to make a brief comment in general support of Tom Melzer’s position and Larry Lindsey’s position . . . as I see it, this action – the whole package – is by any reasonable definition in substance a fiscal action, not a monetary policy action. It is therefore the province of the Congress. The Congress did not have the will to take what I think we all agree was the appropriate action, so we are being left holding the bag. I guess I just see it as a

Policy Debates at the Federal Open Market Committee: 1993–2002

345

raid on our independence, and I regret it. I agree with you that the risk to us, while I think it is substantial, is probably remote, although I’m not sure I think it is as remote as you do.40

Greenspan replied, “I’m not sure I think it is as remote as I think it is either!” and the debate drew to a close shortly thereafter.41 The ESF never warehoused foreign currencies with the Fed during the Mexican peso crisis. The warehousing authorization reverted to $5 billion in 1996 and has remained there ever since. The last time the ESF warehoused foreign exchange with the Fed was in 1992. Mexico drew around $1.5 billion on its swap lines with the Fed during the peso crisis and paid it back in full by January 1996.42

V. Foreign Exchange Operations The Fed recognizes the U.S. Treasury’s preeminence in foreign exchange operations and works closely with the Treasury in conducting them. The Fed intervened in foreign exchange markets at the Treasury’s request on a number of occasions between 1993 and 2002. The interventions precipitated lively debates within the FOMC that altered the way the committee regarded exchange rate policy. The story is one of steadily diminished enthusiasm for foreign exchange intervention in the FOMC.43 Broaddus’s participation in the debate on foreign exchange intervention began following the interventions of April 29 and May 4, 1994, when the Fed and the Treasury’s ESF purchased dollars against German marks and Japanese yen, with resources evenly split between the two institutions. The May 4 intervention in support of the dollar was coordinated internationally. The interventions were triggered after the dollar declined steadily from mid-April onward and foreign exchange markets on April 29 were said to become “disturbed.” According to Peter Fisher, manager of the System Foreign Exchange Desk at the Federal Reserve Bank of New York, . . . on May 4th, market participants – quite skeptical as to whether any form of international cooperation existed – were impressed to see 19 central banks put an exclamation point behind Secretary Bentsen’s statement that movements in exchange markets had gone beyond what is justified by economic fundamentals.44

The committee questioned Fisher at its May meeting about the details of the foreign exchange operation, but the intervention appeared to calm markets, and not much more was discussed. In mid-June 1994, the dollar’s decline accelerated as foreign exchange markets wondered whether the Fed could preempt rising inflation without a recession, and economic prospects appeared to strengthen in Germany

346

Marvin Goodfriend

and Japan. At the Treasury’s request the Fed and sixteen other central banks intervened on June 24, 1994, in a high-profile effort to support the dollar. According to Fisher, Treasury officials felt under pressure to stop the acceleration of negative views of the Administration’s policies, increasingly identified with the dollar, and also to follow their words in support of the dollar with some kind of action. The operation did not generate any appreciation of the dollar, however, and was widely portrayed in the media as a failure.45 The failed foreign exchange intervention precipitated a lively questioning by the committee at the July FOMC meeting. Governor Lindsey noted that “[r]eading the transcript from the last meeting . . . the reason given for the previous intervention in late April and early May was that it was to underline a change in policy.” Lindsey asked Fisher, “Was there any change in policy that this intervention underlined?” Fisher replied, “No, and I think that may have been one of its problems.”46 Lindsey later emphasized that the main reason for intervening in foreign exchange markets is to underline policy already in place. Greenspan responded: . . . actually there is another that is important, and I think we have intervened for that reason on occasion. It is to break a psychology that is building up irrationally in the market. And if intervention can accomplish that, maybe it can have a constructive effect. . . . The only way it can do something is to catch the market short and break the back of a cumulative psychological downtrend. We have accomplished that on occasion when we tried it and sometimes it has failed.

Lindsey asked Greenspan whether he viewed the June 24 intervention as a success or a failure. Greenspan replied that “for the last attempt we did not have the benefit of a net short position in the market. We knew that at the time, but we were at a point where there were no easy solutions.”47 Broaddus framed the issue more broadly: . . . I just have to say that this intervention . . . really bothered me a lot. It got a lot of attention even in the Richmond papers. The headline on the front page the next day was about the failure of this operation. What concerns me is that our involvement in such a conspicuously unsuccessful operation is bound to raise a question in the public’s mind about our general effectiveness as an institution, about our ability to do what we set out to do. In that sense, it could tend over time to undermine or at least weaken the credibility of our longer-term monetary policy objectives . . . I think this shows pretty conclusively that sterilized intervention operations only have lasting effects if the markets believe they are going to be backed up strongly by basic monetary policy actions. Clearly, on June 24th that expectation or that conviction was not there. Given this experience and others like it in the past, Mr. Chairman, I would respectfully recommend that to the best of our ability we avoid getting involved in these operations unless there is pretty general agreement

Policy Debates at the Federal Open Market Committee: 1993–2002

347

within this Committee in a particular instance that we will back up the intervention with whatever monetary policy actions are necessary.48

Greenspan replied, You know it depends really on whether we expect markets to be wholly efficient and not run periodically into some significant abnormalities that an intervention could rebalance. This is the key question. . . . If we are put in a position where we opt out of coordinating with the Treasury, I think that would do the financial system more damage. In all these actions it has been we who have fended off recommended interventions which we thought were superfluous and potentially counterproductive. We have only gone forward when we thought there was at least a reasonable shot. Our choice is not, in my judgment, to withdraw from these discussions because the damage it could do to the financial system if we are perceived to be at loggerheads with the Treasury in this sort of arrangement, I think, would be very substantial. If you are asking me whether I personally disagree with the underlying philosophy that you were outlining, the answer is no, I do not; I do agree. The question is what to do about it.49

At the November 1994 FOMC meeting, Broaddus dissented against renewal of existing swap lines because they were used primarily to facilitate exchange market intervention.50 He pointed out that . . . Because sterilized intervention cannot have sustained effects in the absence of conforming monetary policy actions, Federal Reserve participation in foreign exchange markets risks one of two undesirable outcomes. First, the independence of monetary policy is jeopardized if the System adjusts its policy actions to support short-term foreign exchange objectives set by the Treasury. Alternatively, the credibility of monetary policy is damaged if the System does not follow interventions with compatible policy actions, the interventions consequently fail to achieve their objectives, and the System is associated in the mind of the public with the failed operations.51

The next major Fed intervention with the US Treasury occurred on June 17, 1998, to strengthen the yen against the dollar after the yen/dollar rate reached 146. The intervention precipitated a lively debate at the June 30–July 1 FOMC meeting. Again Broaddus registered his disapproval: . . . I see a risk in resisting the depreciation of the yen, if that begins to occur again . . . Deflationary forces in Japan have now pushed that very important economy into recession. . . . I hope that as we go forward we will keep in mind that any international effort to resist further yen depreciation carries some important risks of its own. The reason is that it has potential consequences for both Japanese and U.S. monetary policy. Fundamentally, to prevent further yen depreciation if underlying economic forces renew downward pressure on that currency, either the Bank of Japan must

348

Marvin Goodfriend

pursue a tighter monetary policy or we must pursue a more accommodative monetary policy or some combination of the two. The former, of course, risks an even deeper recession in Japan. The latter risks creating stronger inflationary pressures in the U.S. economy at a time when, at least in my view, rising inflation is already a risk. . . . Indeed, a case could be made that we really need just the opposite, a more expansionary monetary policy in Japan and a less expansionary one here.52

Broaddus drew the following response from Greenspan: Let me just say that I do not think anyone at the Treasury would seriously disagree with the way you framed the issue. There was a great reluctance at the Treasury to intervene, and the decision was touch-and-go for a considerable period of time. I would say that the chances of repeating such intervention verge on the remote or even less than that. I believe there is a real understanding that Japan’s . . . effort to stabilize the yen in April clearly demonstrated that intervention per se does not work. . . . The only reason that intervention seemed to work in the latest episode had nothing to do with the size of the intervention. It was the result of what somehow was perceived as a signal that either we were going to ease monetary policy or the Japanese were going to tighten. Clearly, neither policy option is even remotely on the agenda at this stage.53

A few minutes later Fisher reported, [t]he intervention in which we participated on June 17 certainly took the market by surprise. We had a much bigger price effect than I and, I think, a number of my colleagues anticipated. Some large speculative players seemed to have been extending their long-dollar/short-yen positions, and as the yen strengthened from 146, where we entered the market, down to 142, they had to close those positions. That’s why we had such a “pop” in the market. The market is wary of further intervention. Obviously, when people sustain significant losses on a mark-to-market basis and some of them close their positions that gets the market’s attention.54

The Fed decided to eliminate all of its standing swap lines when the euro came into existence in December 1999, except for lines with its NAFTA partners, Canada and Mexico. Foreign exchange intervention came to be seen in the Fed as ineffective at best and counterproductive at worst given the paramount importance that domestic price stability had assumed as a priority for monetary policy. Nevertheless, the Fed intervened again on the foreign exchange market with the Treasury on September 22, 2000, in conjunction with the G-7 monetary authorities to weaken the dollar against the euro, then trading below 90 cents/euro. The intervention was generally regarded as a success in that it appeared to push the dollar/euro rate up somewhat. William Poole, president of the Federal Reserve Bank of St. Louis, expressed his disapproval

Policy Debates at the Federal Open Market Committee: 1993–2002

349

of the operation at the October FOMC meeting because sterilized interventions themselves are ineffective and because the Fed should not support interventions of questionable value with its prestige and resources.55 Greenspan followed: Between June of 1998 and today – or I should say last week – we probably had 20 or so different requests, at various levels, for intervention. We turned them all down. And indeed, in this latest case we were not happy with the notion of intervening. What occurred essentially was that the Treasury, feeling under very considerable pressure from the rest of the G-7, concluded that in the spirit of international comity we had very little choice but to accommodate the Europeans. We, the Federal Reserve, did have the choice of saying we would not participate. The presumption that we could do that without it being known is not believable. . . . Now that would create a schism within the United States government. And the question we have to evaluate is whether our concerns about the inefficacy and potentially counterproductive effects of sterilized intervention are far more significant than the schism that would be created with the Treasury. In my judgment the answer is no. . . . We do have to maintain a relationship with them. We are independent, to be sure, but not from the Congress. . . . [W]hen we were talking about it in advance . . . the thing we ha[d] to be concerned about is the probability that this intervention [might] fail. And a failure of the intervention as we go into a G-7 meeting would put extraordinary political pressure on everybody to say something . . . Would I have preferred that we didn’t intervene? I certainly would. We did not shut the door completely. . . . And I would say that this Treasury has been a lot less likely to cave in to foreign pressures than previous ones.56

At the January 29–30, 2002 FOMC meeting, Greenspan summed up the evolution in the thinking on foreign exchange intervention that had occurred since the early 1990s: . . . let me note that I have recently had conversations with the Secretary of the Treasury in which he reiterated the Treasury’s position with regard to foreign currency intervention. It is about as close to ours as you can get. The general view at Treasury is that the history of intervention shows clearly that it has not been effective. And except under extraordinary circumstances, which would be less economic and perhaps more political in an international sense, there is no inclination on their part to do any intervention. Consequently, I think the discussions we’ve had with pervious Treasury officials when we’ve had activist Treasury Departments are not relevant in this case.57

VI. The Inflation Targeting Debates After the Republicans took control of the House of Representatives in November 1994 and expressed interest in inflation targeting, Chairman

350

Marvin Goodfriend

Greenspan asked the committee to consider the issue. Broaddus had earlier expressed support for inflation targeting and was invited to present the pro case. Governor Janet Yellen agreed to present the con case. The FOMC held its first formal debate on inflation targeting at the January 31–February 1, 1995 meeting. Broaddus began: . . . The basic argument . . . for implementing an operationally meaningful explicit inflation objective is that it would allow us over time to foster a better economic performance. This would occur . . . because we would be moving away from the almost purely discretionary approach to policy we have followed historically, with its focus on reacting to emerging short-term economic developments, toward an approach where the central focus would be on precommitment to a permanent low inflation objective that would be clear and feasible. . . . I think it is fair to say that it is supported by much, if not most, of the important research done in monetary economics over the course of the last twenty years . . . [E]xperience over the years under our current approach to policy . . . suggests that periodic inflation scares in the financial markets and the damage inflation has done to the economy naturally make a lot of people think there has got to be a better way. The main objection to such an approach . . . is that a short-term trade-off is said to exist between real activity and inflation . . . [A]s I see it at least, there is nothing incompatible between a credible long-term inflation objective on the one hand and having the flexibility to cushion the economy against supply shocks as long as the public understands and is confident that the longer-term commitment remains in place while we are dealing with the short-term problem. Indeed, far from reducing our flexibility, it seems to me that a credible long-term objective arguably would increase our flexibility in dealing with such shocks because we would not be worried about losing credibility in that situation. . . . I would recommend that the Committee commit itself firmly and publicly to the objectives contained in the Neal amendment . . . in the way that it defines price stability and also importantly with respect to the 5-year horizon.58 . . . There are some very strong advantages to proceeding in the way that I just suggested. For one thing, we are already on record in favor of the Neal amendment . . . [D]oing what I have suggested would not prevent the Fed from taking the kinds of policy actions that we take today to stabilize employment and output. What it would do, and this probably is the most important thing I am saying today, is to discipline us to justify our short-term actions designed to stabilize output and employment against our commitment to protect the purchasing power of our currency.59

Yellen followed: I am strongly opposed to the adoption of formal multi-year inflation targets. . . . I am taking this proposal to be . . . that the inflation rate should be the sole objective of policy for current and future years with no weight being placed on achieving

Policy Debates at the Federal Open Market Committee: 1993–2002

351

competing ultimate goals for real variables. I am going to speak against that proposal, and I note that it is a somewhat stronger proposal than I heard Al just support. . . . I began by asking myself the question, what is it that the public cares about? The answer seems straightforward to me. It is not just high and variable inflation . . . [t]he public also cares about fluctuations in output and employment. . . . Then I ask myself, what is it that the Fed can accomplish? I conclude that the actions of this Committee affect not just the level and variability of inflation but also at a minimum the variability of output and employment. . . . The moral I draw is that the Fed should pursue multiple goals. . . . Fortunately, the goals of price stability and output stability are often in harmony, but when the goals conflict and it comes to calling for tough trade-offs, to me, a wise and humane policy is occasionally to let inflation rise even when inflation is running above target. . . . Let me turn to the issue of credibility. A key argument in favor of inflation targeting – Al made this point – is that it would raise the FOMC’s credibility and result in a lower sacrifice ratio. Clearly, if we could achieve this, it would be a very worthwhile benefit. The problem in my view is that it is not achievable. I look at countries that have adopted and carried through inflation targeting programs, I consider the results discouraging. . . . The second point concerning credibility is that I do not think inflation targets would raise credibility for the simple reason that they would not be credible. . . . We could talk a little about dynamic inconsistency, but for the sake of time I think I will pass that up. . . . If we testify, it seems to me that we should point out that the benefits of price stability are elusive and that the costs of additional output instability with such a plan could easily outweigh the benefits of greater inflation stability.60

After a subsequent airing of the various points of view, Greenspan commented: Let me say what I think the purpose of this discussion is. To go to inflation targeting without a Congressional statute is probably unwise. We do not have a Neal bill, but there clearly is going to be a Connie Mack bill that will be very close to the Neal bill, and we are going to be asked to comment on it. The basic purpose of this discussion is to get our first cut as to where this Committee stands for purposes of testifying on that legislation. My own judgment is that if we do not announce any specific inflation targets, our policy can actually be similar to what Al Broaddus was suggesting. If we do announce explicit inflation targets, they become in effect a statutory obligation for this Committee to adhere to; and I am not sure by any reading of the Humphrey-Hawkins statute that inflation targeting is consistent with it.

Immediately following Greenspan, Tom Melzer, president of the Federal Reserve Bank of St. Louis, reiterated a key point:

352

Marvin Goodfriend

. . . My view is that monetary policy only affects prices in the long run. I have a hard time justifying setting objectives with respect to things that we can’t influence in the long run. So, I would very much like to set objectives that are consistent and that we can influence in the long run. . . . . . . The other thing I would say is that it is not clear to me, and this again is not the time to discuss it, that the present legislation under which we operate would absolutely preclude some sort of inflation targeting regime. I think there would be some advantage to go along the course that Al described where as soon as we can reach some sort of consensus, assuming that there is some consensus in this direction, we could move ahead to take some actions on our own.61

After a while, Greenspan summarized the proceedings: . . . Now we understand why this Committee has had difficulty confronting this issue. It is because we are as split down the middle as we could possibly get. . . . This really raises some very interesting questions as to where we are. My own impression is that even if we now locked into law a fixed inflation rate – say 2 percent or 1 percent – and the Congress voted for it with a large majority, in the first recession everyone would be arguing to go in a different direction.62

The committee resumed its discussion of inflation targeting at the July 2–3, 1996, meeting. Yellen made the opening statement, emphasizing the costs of bringing inflation down from the then current rate of around 3 percent to zero. Yellen concluded her opening presentation: . . . As I total things up, it appears to me that a reduction of inflation from 3 percent, which I take as roughly our current level, to 2 percent, very likely, but not surely, yields net benefits. . . . To my mind, to go below 2 percent measured inflation as currently calculated requires highly optimistic assumptions about tax benefits and the sacrifice ratio. . . . 63

Broaddus followed with his opening statement: . . . [L]istening to Janet Yellen has served to convince me that if we are really going to make progress, we need to prioritize some of these issues. . . . This may involve some risks, but I would assert that there are, or at least there may be, some points of agreement. I think most of us would accept the view that at a minimum we want to hold the line on inflation – that is, to preserve the gains we have made over the last 15 years or so in bringing the trend inflation rate down and then to bring the rate down at least somewhat further over a period of time. Moreover, I think many of us would regard the line to be held as an underlying rate of something like 3 percent on the core consumer price index (CPI), although we can debate which measure it should be.64

Robert Parry, resident of the Federal Reserve Bank of San Francisco, followed soon after:

Policy Debates at the Federal Open Market Committee: 1993–2002

353

Mr. Chairman, is there a way to focus on what was agreed to between them as an interim step? It looked as though both had the same view about the desirability of not allowing inflation to go higher. There also was a very explicit agreement that inflation should begin to move lower, and I think I heard Janet say something like a full percentage point lower.65

Yellen immediately replied, “I agree, yes.” Parry then asked, “Why not set that out as an objective, and then we can have another meeting when we reach it. Broaddus replied, “I would second that. I think that is a great idea.” Parry observed, “That would mean more progress than we have made in 11 years.” Cathy Minehan, president of the Federal Reserve Bank of Boston, said soon after “ . . . I am in complete agreement with the two things on which I think they agreed. That is, we should at a minimum hold the line on inflation where it is and go somewhat further if we can do so.”66 Tom Hoenig, president of the Federal Reserve Bank of Kansas City, added “ . . . In my view that requires that we take the legislation that is in place now and pursue stable prices seriously, I think the mandate is there. In implementing that mandate, I would agree with Bob Parry and Al Broaddus that we ought to start somewhere. I would accept 2 percent inflation as the interim goal if we can agree on a reasonable timeframe in which we would move systematically toward that goal.” Governor Larry Meyer noted, . . . But without deciding on exactly what the path is, we have an agreement that we want to hold the line at about 3 percent on core CPI and that provisionally we want to set a very explicit target for ourselves. We seem to be headed for agreement on 2 percent inflation. We still have to debate how to get there, but this is a lot of progress. . . . 67

A few minutes later Greenspan asked, “Can I switch the subject? Since we have now all agreed on 2 percent, my question is, what 2 percent?” And then Greenspan proceeded to wonder which measure of inflation would be best.68 But Parry interjected: It seems to me that when Janet and Al were talking, they had implicit in their minds something like a CPI or core CPI. They were able to generate some consensus or agreement about the desirability of reducing the rate of CPI inflation from its current level of about 3 percent down to 2 percent. That is the critical point.69

Minehan immediately added, “Hold the line where we are.” Meyer asked, “What do you think the personal consumption expenditure (PCE) deflator is now? Greenspan answered “It is 2 percent.” And Meyer answered, “So we are there.” Parry then said, “That’s fine. Then we will start with 2 percent

354

Marvin Goodfriend

inflation and go to 1 percent. But Meyer asked, “Maybe 2 percent inflation as measured by the PCE is where you want to be?”70 This was a dramatic moment for those in the room, like the author, who favored an explicit inflation objective. The committee implicitly had reached a working consensus for holding the line on current inflation measured with either the core CPI or PCE. But just as suddenly the committee veered away from that consensus. Instead of prioritizing the issues in order to make progress, as Broaddus had suggested, and focusing on the consensus that had been achieved to “hold the line where we are,” as Minehan put it, the discussion dissolved again into a debate on how low the committee would ultimately want to go on inflation. Greenspan adjourned the debate shortly thereafter without any formal acknowledgment of the progress that had been made. The author was disheartened by the turn of events. Another small but important step toward announcing an explicit inflation objective was taken a couple of years later at the February 2–4, 1998 FOMC meeting. The committee was setting M2 money growth ranges required for the Humphrey-Hawkins hearings before congress. Broaddus proposed an inflation target once more: . . . A couple of recent developments make me want to put that issue back on the table. . . . First, we are very close to price stability now . . . with measured CPI at around 2 percent with an upward bias of around 1 percent. That has a couple of implications. One is that we no longer need to get hung up on the topic of the transition costs of moving back to price stability. We are already there essentially. I think that is a powerful argument for trying to lock in price stability by announcing a somewhat more specific inflation objective. Also, with inflation as low as it is currently, the public has become more aware of and more concerned about deflation. For obvious and understandable reasons, there has been much more public comment about deflation recently than in the past. In effect, one might say we recently have gone through a mini-deflation scare. In the current low inflation environment, it seems to me that our longer-term policy strategy now needs to address deflation concerns as well as inflation concerns. For that reason, I believe we should consider stating explicitly a lower bound for our longer-term inflation objective. It could be zero on an accurate price index or maybe 1 percent or so on the actual measured CPI. . . . If we did that, I think it would help to clarify our strategy for situations that have begun to receive some attention very recently but that, understandably, had not previously received a lot of attention for four or five decades. In my view, that could help us avoid getting into the kind of situation the Japanese have found themselves in from time to time in recent years. Of course, if we announce a lower bound on inflation, then it would make sense for us to announce simultaneously an explicit upper bound as well. It might be in the neighborhood of 3 percent on the measured CPI.71

Policy Debates at the Federal Open Market Committee: 1993–2002

355

Later in the meeting, Broaddus again urged the adoption of an explicit lower bound on the committee’s tolerance range for inflation and drew this response from Meyer: Could I add a point? It is not just our ability to communicate to the public it is our own internal deliberations that are at stake here. We might ask where we are heading. Do we think we are where we want to be or do we think we want inflation to move down 1/2 percentage point, 1 percentage point, 11/2 percentage points? The vague definition [of price stability] was perfectly adequate when inflation was 10 percent, and it worked when inflation was 5 percent, but now I think there is a real question – maybe among us and certainly among the public at large. There is an issue here.72

Greenspan replied “that is a nice problem to have” but immediately asked for a vote on policy and nothing more was said on the matter. Yet, the issue that Meyer raised was important: If the FOMC needed a precise explicit working definition of its inflation objective for internal purposes, which it surely did, then shouldn’t the Congress and the public be informed of that working definition? In effect, the approach to price stability revived the issue of Fed secrecy – this time not about policy deliberations, or interest rate policy actions, but about the long-run objective of monetary policy.

VII. Asset Acquisition Policy The emergence of large federal budget surpluses in 2000 and 2001 led to a substantial paying down of federal government debt and the possibility that the stock of U.S. Treasury debt could be reduced substantially in subsequent years. Fed assets at the time, accumulated in providing currency and bank reserves to the economy, consisted almost entirely of roughly $500 billion of Treasury securities. At its March 2000 meeting, the FOMC authorized a subcommittee led by Peter Fisher, manager of the System Open Market Account at the Federal Reserve Bank of New York, and Don Kohn, director of the Division of Monetary Affairs at the Federal Reserve Board and secretary of the FOMC, to consider a variety of options to study what assets it should acquire in place of Treasuries should they be retired. The subcommittee presented its findings at the January 2001 FOMC meeting. Don Kohn introduced the report, highlighting a number of points: I think the first important point to highlight . . . is that the issue cannot be put off for much longer. Under a wide variety of assumptions about the growth of the economy and the political process, Treasury debt will be repaid over coming years, . . . . Meanwhile, the Treasury market will become increasingly illiquid, ultimately for RP as well as outright transactions, especially considering that many

356

Marvin Goodfriend

of these securities are held by investors who will be loath to give them up even at elevated price premiums. A second point apparent in all the papers is that there are no easy, obvious solutions to the problem of what assets to hold under this circumstance. All options seemed to have significant drawbacks. Some people have proposed continuing to rely on Treasury securities, even as the debt is paid down, by acquiring them through special arrangements with the Treasury or the Social Security Trust Fund. While the System would be able to continue to hold risk-free government assets, such plans themselves do raise a number of questions. They would transfer the problem of possibly accumulating private assets to another part of the government that may not be as well equipped to deal with it. They would leave the Federal Reserve with a portfolio of illiquid assets as the Treasury market disappears, and they would make the central bank dependent on agreements with the rest of the government for its assets. Of course, the alternative of taking on private obligations raises other issues, including those involved with potential effects on private credit allocation and the management of risk and liquidity in the System’s portfolio. . . . A key tradeoff would be between minimizing the effects of System portfolio choices on relative asset prices on the one hand, and minimizing risk and maximizing liquidity on the other. A broadly diversified portfolio, which included credit to financial intermediaries holding nonmarketable assets, would have the greatest chance of exerting as little influence as possible on private credit decisions. With such a portfolio, the System would have a low profile in each market and it would not be favoring one type of asset over another. But the System would be acquiring riskier and less liquid assets. . . . At the other end of the spectrum, if the Committee chose to concentrate operations in a small subset of markets that promised the least credit risk and the greatest liquidity – for example those for GSE securities or A1/P1 commercial paper – it would increase the odds on eventually affecting relative asset prices.73

Broaddus believed that the subcommittee report gave insufficient attention to staying with Treasuries, and he opened the committee discussion of the report by making the case for a “Treasuries-only” asset acquisition policy. Given the extensive holdings of non-Treasury securities on the Fed’s balance sheet in 2010 and the potential for the Fed to acquire more in the future, Broaddus’s case for Treasuries only is particularly relevant today. Hence, Broaddus’s lengthy statement from the January 2001 FOMC meeting is presented in its entirety followed by the exchange with Greenspan and other FOMC participants that helped to clarify Broaddus’s recommendation. Broaddus began: Mr. Chairman. I want to make a pitch for trying to arrange with the Treasury a way for us to stay with investing in Treasury securities only. In my view this is a really important issue that goes to the heart of our institutional position in the government and also to our ability to conduct monetary policy effectively over the longer run.

Policy Debates at the Federal Open Market Committee: 1993–2002

357

I would begin my remarks by suggesting that we think about how fortunate we have been over the years to be able to pursue a “Treasuries only” policy – or at least approximately Treasuries only – for so long. As the Beebe/Cumming study recognizes, a Treasuries only policy alone among the alternatives that are being considered and suggested on this issue satisfies all four principles that are laid out in that paper with respect to how we should guide our portfolio selection. Such a policy would allow us to maintain instrument independence, minimize credit allocation and distortions to relative prices, maintain essential liquidity and credit quality, and provide appropriate transparency and accountability. I would underscore the benefits of Treasuries only as follows: Monetary policy basically determines the quantity of the monetary base and as a by-product the aggregate volume of Federal Reserve credit that we will extend. The beauty of Treasuries only, as I see it, is that it has allowed the government as a whole to implement monetary policy by essentially buying back interest-bearing government debt and replacing it with the liability of the central bank. Consequently, and this is the key point, neither the Fed nor the government as a whole for that matter has had to invest in any private assets to conduct monetary policy or to make the potentially very difficult choices among private assets that might have to be made if we consider these other alternatives. Now, of course, we face a situation where the outstanding stock of Treasury debt may disappear. I think this presents the Fed with a huge problem because all of the alternative approaches available to us – that is, the other assets that are being considered – will involve us to one degree or another in decisions about allocating credit across particular sectors of the private economy. Some of you may recall that at the March [2000] meeting last year I argued that credit allocation would inevitably embroil us over time in politically charged decisions that could undermine our independence and the effectiveness of monetary policy. And I urge that before we go down the path of these other alternatives we at least consider the possibility of persuading the fiscal authorities to continue to issue sufficient government debt to allow us to stay with the Treasuries only approach. I believe we still ought to do that and I ask you just to consider it. A Treasuries only option is sketched out briefly on page 16 of Chris Cumming’s and Jack Beebe’s study. Let me briefly summarize it. The idea is that even if continued surpluses were to permit the Treasury to stop issuing debt, the Treasury would continue to issue debt for the Fed to buy in order to replace maturing debt already on our balance sheet and to provide for secular growth in the monetary base. Note here, and I think this is an important point, that this debt would be costless to the Treasury since we would be remitting to the Treasury the interest on the debt we would buy. The question of short-term cyclical needs for increases in the monetary base would still remain, and it might be that we would need to satisfy those needs by purchasing liquid, low-risk private assets in the form, say, of RPs. But since the acquisition of private assets in that case would be self-reversing and relatively limited in size, it would involve the Fed only minimally in credit allocation. I don’t think it would raise the kinds of issues and concerns that a more fundamental change would.

358

Marvin Goodfriend

Now, I know that when you first hear this proposal it seems eminently dismissible. [Laughter.] A lot of questions come up and a lot of objections can be raised, and Don has already cited some of those. But let me address a couple of them. The first question is: Isn’t this proposal just a way for the Fed to shift the burden of investing in private assets, if we have to do that in this new world, from itself to the Treasury? Well, this proposal would respect the integrity of the fiscal policymaking process by leaving all fiscal decisions to the fiscal authorities, Congress and the Treasury, which would protect the Fed’s independence. The key point here, though, is that the government wouldn’t have to accumulate assets with the revenue it would get from selling securities to the Fed. It would simply be the revenue that the government gets from the seigniorage tax – that is, from the act of creating money – and I think that’s an important point. The government could use this revenue to permanently reduce other taxes or to increase expenditures. That covers one question. A second question, closely related to the first, is whether the government as a whole shouldn’t take advantage of at least relative political independence of the Federal Reserve to let us acquire the assets and make the choices among these private assets. Presumably, we would be subject to less potential political interference than other parts of the government. This question I think is more likely to be asked by people who feel our independence is secure rather than by people like me who think it is inherently fragile. In my view the answer to this question is the same as the answer to the first question. It is not necessary for the government to acquire private assets permanently in order to conduct monetary policy. I doubt that many people around this table would think it’s a good idea, just on the face of it, for the government to buy and hold private assets. If not, then I believe we should be wary of letting the Fed be the instrument for doing that. And that’s one of the reasons why I think we need to adopt the Treasuries only proposal seriously. But what about some of the alternative approaches like expanded use of the discount window discussed by Craig Hakkio and Rick Lang in their paper, or the expanded use of RPs not only for short-term liquidity purposes but to meet the secular need for increases in the base? Let me make a few comments about each of those. With respect to discount window loans, at first blush that appears to be an attractive alternative. We have the authority without seeking new legislation to expand our use of the window in implementing monetary policy. And in principle we could increase our discount window lending from the relatively small amount that’s on our books now to several hundred billion dollars.[74 ] Presumably, as I think Craig’s and Rick’s paper recognizes, we would need to restrict our lending to banks with CAMEL ratings of 1 and 2. Also, we probably would want to limit our lending to any particular bank to a prudent fraction of that bank’s capital and we would want to back our loans with good collateral. I think we could start out down this road successfully. But let’s recognize that this would be a profound change in the way we do things. It would make the Fed a major, continuous creditor to hundreds of depository institutions instead of an infrequent lender to particular institutions. What worries me under this proposed regime is, what to do if a bank to which we have extended substantial credit gets into serious trouble? In my view that would put us in a very difficult situation.

Policy Debates at the Federal Open Market Committee: 1993–2002

359

Presumably, we would want our portfolio to be public knowledge. I think transparency is essential in establishing accountability for our portfolio. It is hard for me to see that we would not be forced to make our portfolio transparent if we went in this direction. But in that situation if we pulled a loan because of the deterioration in a bank’s condition, that action would signal publicly that the bank has significant difficulties. Currently we don’t publicize CAMEL ratings, so this would be a fairly radical change in our supervisory approach to safety and soundness. Hypothetically, even if we didn’t make the composition of our loans public, it seems inevitable that if we pulled a loan to a sizable institution, the markets would quickly detect it. My real worry is that in such a situation we would be unwilling to pull the loan. Worse, troubled banks would tend to replace lost uninsured funding with discount window loans, which is what has happened historically. It has happened in today’s world. But since our loans would be backed by much of a bank’s good collateral, this would greatly increase the exposure of the FDIC – and potentially taxpayers – to losses when a bank ultimately fails. In sum, if we greatly expand our discount window lending, we will put ourselves even more in the middle of contentious issues surrounding the potential resolution of the problems of a troubled bank. We’ve had that happen historically. If we go to this kind of approach, I think we will have it in spades. The kinds of difficulties we could encounter in this regime would be bad enough in the case of an individual troubled bank, but they could be quite damaging if we faced any kind of general banking crisis. I think it could threaten our independence and our ability to conduct monetary policy independently. So I believe that dealing with our portfolio problem by expanding discount window lending would be a mistake. Let me turn now to RPs and then I will be finished. I appreciate your patience. Expanding the use of RPs would not raise some of the issues that expanding discount window lending would raise. RPs are self-liquidating, which would allow us to exit problem bank situations more quietly if they arise. And we could do RPs on a wide variety of assets with appropriate haircuts. So at first blush it looks as if RPs might be the way to go. But expanded use of RPs to support the secular growth in the monetary base is distinctly different from the use of RPs to deal with the short-run problems that I mentioned earlier. To use them in a long-term way would still be problematic, I think. First, while RPs would raise fewer obvious credit allocation issues than some of the other alternatives that are being considered, over time there is a good chance that political pressures on the System would adjust to this change. And we could find ourselves dealing with political problems in credit allocation issues with respect to RPs as well as with some of the other alternatives that are on the table here. Beyond this, though, there is one other less obvious but I think very important problem with the RP alternative – namely, that precisely because of the desirable properties of RPs that I just listed, they pay a relatively low return. Remember that in this situation the return would be the government’s revenue for money creation. So if we went to RPs because of their nice properties from the Fed’s standpoint, essentially we would be limiting the government’s revenue from money creation. In essence, we would be using a large part of this revenue to buy liquidity services and to protect ourselves – the Fed, that is – from credit and price risk, thereby denying the rest of

360

Marvin Goodfriend

the government the use of these funds for whatever other purposes it wanted to use them. This last point, in my view, is the answer to one other objection mentioned in the study to the Treasuries only proposal. The concern was that because the Treasury would be doing us a favor in some sense by allowing us to continue with the Treasuries only approach, they would demand some sort of quid pro quo. Now, if it were understood that perhaps arguably the most feasible alternative to Treasuries only, namely the RP alternative, would be costly to the government, then it would be in the narrow budgetary interest of the fiscal authorities to prefer that we stay with the Treasuries only approach. So in that instance a quid pro quo wouldn’t be necessary. Having said that, I recognize that arguing this point and getting it across to others elsewhere in the government would be challenging. But in my mind it’s a valid point and I think we should try to make it. Well, that is basically my argument and I appreciate your patience. Let me list quickly the four main points I’ve tried to make. First, there is no need for the Fed or the government to acquire private assets, except maybe temporarily, to implement monetary policy. Second, I believe it is feasible for the Fed to follow a Treasuries only policy with the cooperation of the Treasury even if the Treasury has no other reason to issue debt. Third, it wouldn’t cost the government anything to provide debt for the Fed to buy. Finally, with respect to the RP alternative, the government would forgo revenue if the Fed held a portfolio of very safe and liquid but lowyielding private RPs. So from that perspective it would be in the interest of the fiscal authorities to cooperate with the Fed in a Treasuries only approach. I know that pushing this proposal is a hard idea to get used to. But looking at the disadvantages and problems associated with the other alternatives, I find the argument for at least trying to do that compelling. And I hope we will consider doing it. Thank you.75

Broaddus’s remarks immediately prompted a series of comments and questions from Greenspan that helped clarify the key part of the Broaddus’s proposal. The exchange is reproduced below exactly as it appears in the Transcript: CHAIRMAN GREENSPAN. In your scheme, what does the Treasury do with our payments to them for their debt? MR. BROADDUS. They, of course, would be paying interest to us and we simply would be turning around and paying it back to them. CHAIRMAN GREENSPAN. The issue basically is that they have to invest the proceeds from our purchases in something else. MR. BROADDUS. As I see it, they first sell their securities in the market. We buy them in the market; they are not selling directly to us. So that would be the form in which they would take the funds. CHAIRMAN GREENSPAN. It doesn’t matter how it’s done.

Policy Debates at the Federal Open Market Committee: 1993–2002

361

MR. BROADDUS. They take in revenue and it could be used for whatever purposes they want. CHAIRMAN GREENSPAN. But the issue, in the context we’re talking about, is that if the debt to the public is down to zero, they have to accumulate private assets. MR. GOODFRIEND. May I answer that please? MR. BROADDUS. Sure! MR. GOODFRIEND. I’m sorry to interrupt; I know this is unusual. The revenue for money creation could be regarded as basically the result, on a secular basis, of the growing demand for real currency balances that the public wants to hold. If one regards the revenue from money creation, which is sometimes called seigniorage, as a tax flow just like any other tax that the government receives on a yearly basis– CHAIRMAN GREENSPAN. Money is fungible; I understand that. It’s not that I object to what President Broaddus is saying, I’m just asking a question. There are two regimes, one in which we accumulate private sector assets and one in which we don’t. In the regime that you’re suggesting there is double entry bookkeeping. My question is: What appears on the asset side of the U.S. government’s balance sheet? Since by hypothesis we are stipulating that there is zero debt to the public, it means the Treasury can’t pay off debt. Therefore, if they hold a liability, they must hold an asset. What is the asset? MR. BROADDUS. Well, the asset in the short run is probably some private asset. But over time adjustments could be made that would take that off the books. CHAIRMAN GREENSPAN. The only way to do that is to run a government deficit. MR. MEYER. They can basically rebate it as a tax refund immediately or spend it. MR. GOODFRIEND. That’s right. MR. BROADDUS. Those are the two alternatives. MR. MEYER. The government would not accumulate private assets; the funds would just flow right through. MR. BROADDUS. That’s right. MR. MEYER. The government never acquires debt. It just rebates it right back to the public instantaneously. CHAIRMAN GREENSPAN. Yes, but then that alters the view that we’re in surplus.

362

Marvin Goodfriend

SEVERAL. Right. MR. MEYER. That’s right; there is no surplus. CHAIRMAN GREENSPAN. One could argue that it is better from our point of view to have the credit allocation process be in the hands of the Federal government than in the hands of the Federal Reserve. But it is difficult to get around the fact that there is an allocation process going on in the consolidated monetary authority system, given the accounting process. What is it about that statement that’s not true? MR. GOODFRIEND. The main issue is that we can provide the public– CHAIRMAN GREENSPAN. Let me stipulate something very important, which is that the government balance in terms of deficit/surplus is the same in the two regimes. The only difference is who is holding which assets. If you consolidate the Federal Reserve into the system, then there is a unique solution. The only issue occurs when you disassociate the Federal Reserve from the authorities. If you’re assuming a unified budget, obviously, that does create a change in the balance sheet. I’m only saying that in this context if we don’t purchase, or through RPs acquire, private instruments, somebody else has to. There’s no way of getting around that. MR. GOODFRIEND. That’s certainly true given your assumptions about the rest of the government’s fiscal position. President Broaddus’s point is that no one in the government needs to acquire private assets to implement monetary policy. CHAIRMAN GREENSPAN. Nobody disagrees with you on that. MR. GOODFRIEND. Okay. Then if one of our goals is to minimize private assets acquired by the government, we could make that understood by the rest of the government, in which case they would do with the money what Governor Meyer is saying– CHAIRMAN GREENSPAN. Meaning, lower their surpluses and refund taxes. MR. MEYER. Think of it as a “money rain” every day!

VIII. Monetary Policy at the Zero Interest Bound At its January 2002 meeting, in the wake of the 9/11 attack on the World Trade Center, the FOMC invited three Fed economists who had presented papers at a 1999 Federal Reserve System conference on monetary policy in a low-inflation environment to lead a discussion of monetary policy at or near the zero interest bound.76 Dave Reifschnieder and John Williams

Policy Debates at the Federal Open Market Committee: 1993–2002

363

focused on interest rate policy near the zero bound. The author was asked to present his work on monetary policy at the zero bound. Because monetary policy at the zero bound is an important policy question in 2010, much of the author’s January 2002 FOMC briefing is reproduced below together with the committee’s comments and questions. Greenspan introduced the session, saying “We now move on to what hopefully is going to be a rather interesting conversation and I trust an academic one, but we never know about these things.” The author began his presentation: I will spend most of my time explaining how what I think is the best option – expanding the monetary base – could work to stimulate the economy at the zero bound . . . . Usually, open market operations are constrained to accommodate the demand for the monetary base at the opportunity cost spread between the intended federal funds rate and zero.[77 ] There is a need to defend an interest rate spread when the federal funds rate is positive, and as a result, the monetary base is not an independent instrument available for policy in those normal circumstances. But once the federal funds rate is zero, there is no need to defend an interest rate spread, and policymakers are free to expand the monetary base further to stimulate the economy. Central banks can pursue what I call quantitative monetary policy – as distinguished from interest rate policy – at the zero bound. To appreciate the power of quantitative policy at the zero bound, we need to distinguish between narrow and broad liquidity services. In models of the demand for money, narrow liquidity services are provided by the medium of exchange, which allows banks and the public to economize on transactions costs or so-called shopping time costs. For example, people hold currency to minimize trips to the ATM; they hold checkable deposits to avoid sales of nonmonetary assets in order to replenish money balances and make payments; and banks hold reserves to save on transactions costs in the federal funds market. When short-term nominal interest rates are at zero, narrow liquidity services of the kind I just described are no longer scarce because there is no opportunity cost of holding currency or bank reserves, and the channel of monetary policy transmission that we ordinarily use is exhausted. Broad liquidity services are not exhausted, however, and they provide what I will argue is the leverage for quantitative monetary policy to stimulate the economy further at the zero bound. Let me first define what I mean by broad liquidity. Broad liquidity is a service yield provided by assets according to how easily they can be turned into cash, either by their sale or by serving as collateral for external financing. Broad liquidity services are valued because they minimize the exposure of households and firms to what I call the external finance premium. That premium is a consequence of imperfect information, costly enforcement, and costly monitoring of loan contracts that create a wedge between the cost of funds raised externally and those generated internally. The existence of an external finance premium gives rise to a demand for broadly liquid assets that over time has been referred to in the

364

Marvin Goodfriend

profession as “precautionary savings,” “a liquid buffer stock,” or “self-insurance.” Broad monetary instruments include bank deposits, money market mutual fund shares, and short-term government securities. These could be used to meet spending needs in excess of current income – in other words, to protect households and firms from having to go to banks or credit markets to borrow and thus pay the external finance premium. Quantitative monetary policy at the zero bound must expand broad liquidity in order to be stimulative because stimulus from narrow liquidity is no longer available. Open market purchases of short-term government securities, the usual vehicle for an expansive monetary policy, would create monetary base by withdrawing from the system an equal value of short-term government securities. At zero interest, the liquidity services provided by base money and short-term securities would be roughly equivalent. So, if the central bank operates in a “business as usual” mode that would not increase broad liquidity in the economy. Other methods of operations must be employed. There are three avenues that could be pursued to increase broad liquidity by expanding the Federal Reserve’s balance sheet. One would be to buy from the public relatively illiquid assets such as long-term government bonds, which would provide the public with base money that would then be deposited in banks.[78 ] The banking system would expand, deposits would grow, and the banks would hold reserves against these additional deposits. It’s important to understand that even if banks do not use these reserves to expand lending – and thus there is no secondary expansion of bank deposits – the Fed has increased broad monetary liquidity in the economy as a result of its purchase of relatively illiquid long-term bonds on the open market. Alternatively, the Fed could buy assets other than long-term government bonds – anything one can imagine that is relatively illiquid – and in that way also increase broad liquidity. The third way would be to monetize a government budget deficit. That would involve the government’s issuing new short-term securities, say, which the Fed would buy. We would thereby be providing the government with monetary base, which in turn it would transfer to the public through a tax cut or in some other way. The funds placed in the hands of the public would be deposited in the banking system and deposits would increase, as would broad liquidity. I now discuss the channels of monetary transmission by which quantitative policy might be expected to work. The two components of the transmission mechanism are the portfolio rebalancing channel and the credit channel. I will discuss those two channels sequentially, though you will see in a minute that to a large extent they are intertwined. By expanding broad liquidity in the economy the Federal Reserve reduces what I’m going to call the “marginal implicit broad liquidity services yield on monetary assets.” The implicit marginal liquidity yield falls because a greater abundance of broad liquidity reduces the exposure of households and firms to the external finance premium. The portfolio rebalancing channel operates this way: After an injection of broad liquidity that drives down its implicit yield, people will feel compelled to hold assets that are less liquid but have a higher explicit rate of return. Portfolio balance would require a similar fall in the explicit yield on nonmonetary assets. Equilibrium prices of nonmonetary assets would be bid up to restore the required return differential.

Policy Debates at the Federal Open Market Committee: 1993–2002

365

Higher asset prices raise desired consumption out of current income. And higher asset prices relative to their cost of production would revive investment. The increased investment would raise employment, and higher utilization rates and profits would raise asset prices further. That is the essence of the portfolio rebalancing channel. Let me talk for a minute about the credit channel. Because asset prices are higher, collateral values would be higher, net worth would be higher, and bank capital would be higher. As a result of the higher valuations available to back loans, the external finance premium would come down. Credit spreads would narrow, bank lending would revive, and spending would rise as the cost of borrowing against future income prospects falls. Those developments would occur along with the portfolio rebalancing channel, but I call them the credit channel because they are distinct in that they operate in the credit markets. I want to note two implementation problems that I think would be hard to overcome. Even if the transmission mechanism outlined above works well otherwise, these two problems would hinder making the broad liquidity channels operative. Injections of monetary base can provide an impulse to get the recovery going, but for the recovery to be self-sustaining the public must be confident that base money will be expanded by as much and for as long as needed. That is, monetary policy must be supportive until the economy expands enough to support asset prices on its own. To acquire such credibility, the central bank must overcome the perception that it is excessively concerned about the inflationary risk of potentially very high growth in the monetary base. The monetary authority must be prepared to overshoot – perhaps by a wide margin – base money that will be demanded at stable prices after the economy recovers. There is a second related, but distinct, implementation problem that would make gaining credibility for quantitative policy difficult. Ordinarily, relatively small changes in bank reserves suffice to support interest rate policy. We hardly have to move the System’s balance sheet at all to support even large changes in the intended federal funds rate. At the zero bound, however, policy will have to exert its effect through broad liquidity rather than very narrow reserves liquidity. What we’re talking about is operating on a monetary aggregate like M3 – which is roughly around 8 trillion dollars – plus the stock of short-term Treasury securities, which involves another $1 to $2 trillion. The order of magnitude of that aggregate is about that of GDP. That will require large-scale injections of monetary base, substantially increasing the size of the Federal Reserve’s balance sheet, I believe, in order to have the desired effect through the two broad liquidity channels of monetary transmission. If all this is true, the Federal Reserve will need more fiscal support for quantitative policy at the zero bound than we usually are granted by the fiscal authorities. For one, there might not be enough long-term bonds to buy in order to expand the monetary base. Of course, we could buy other assets. But either way the Federal Reserve would be exposed to capital losses that might leave it with insufficient assets to reverse the huge expansion of its balance sheet that is being contemplated. In other words, to be willing to use quantitative monetary policy at the zero bound, the

366

Marvin Goodfriend

central bank must be able to inject large quantities of base money into the economy and be confident that it will have the assets to drain this money after the economy has recovered. In particular, we’d need to be able to drain money that threatens to become inflationary, or we would be reluctant to embark on this process in the first place.[79 ] The fiscal authorities could come into the process in a number of ways. They could promise to transfer to us enough assets – in effect to recapitalize the central bank if necessary – to allow us to drain the amount of base money that needs to be withdrawn from the economy. Alternatively, the fiscal authorities could agree to run a budget deficit at the central bank’s request to help us inject broad liquidity into the economy. The central bank could monetize short-term debt issued to finance the deficit and then withdraw excess base money later by selling that debt to the public. Quantitative monetary policy actions at the zero bound could result in a significant increase in government debt in the hands of the public when this process is over. The fiscal authorities might be unwilling to allow the public debt to expand at the discretion of the central bank. Yet the central bank might be unwilling to pursue an aggressive expansion of its balance sheet without a commitment from the fiscal authorities to support monetary policy fully. A prearranged agreement could enable quantitative policy to work credibly, flexibly, and effectively – at least the way this story goes.[80 ]

The author’s presentation on monetary policy at the zero interest bound provoked a number of comments and questions. Greenspan made the initial comments and asked the first question: I’m a little curious to get from the three of you a sense of how robust you believe the results of your conclusions are. For one, you’re operating generally outside the scope for which the data are fitted in your models, and of necessity there’s a linearity that is implicit in the structure of your models. Very serious questions arise as to whether in fact, as you approach some of these bounds, linearity is the appropriate presumption regarding how economies function. Very specifically, one issue that Marvin raises is that . . . [t]here may in fact be a discontinuity at some point, in which case we may build up a significant amount of inflationary tinder in a deflationary environment, nothing happens for a time, but then the tinder ignites in a way that induces a very dramatic reversal from deflation to inflation. That creates all sorts of instabilities. . . . To a certain extent, Marvin’s results are almost an accounting system. I believe you’re merely describing the mechanisms by which we can change the monetary base, Marvin, and those are essentially the result of our institutional structure. We could go ahead as a central bank and just print money and buy assets – we could buy baseball teams for all we need – and we can generate as much currency as we want. I’m curious to get your impression not of what the standard deviation of your simulations off the existing structure is but what the standard deviation of your models is from reality.81 [Laughter]

Goodfriend replied “In a way, my discussion had an advantage in that I didn’t have a model.” [Laughter.] Greenspan replied, “You have double-entry

Policy Debates at the Federal Open Market Committee: 1993–2002

367

bookkeeping, which is better than some of the accounting practices we’ve heard about recently! [Laughter.] Goodfriend continued: I will say that the discontinuity – though I didn’t use that word – worried me. There were a number of possible discontinuities that worried me, such as banking system distress or fiscal interventions – things that don’t usually happen but that occur at the zero bound. That’s why I was focusing on mechanisms that could get money into and out of the system. I was thinking about how we would position ourselves against this kind of unleashing of inflation – that’s the term I used – rather than “punting” on where that glass ceiling would be. I wanted to think about designing a system that would be robust against it. That is more or less what I was doing.”82

Parry asked Goodfriend to relate this to Japan. Goodfriend answered: Yes, Japan has gone so far down the road that, you’re right, it’s very hard to imagine that it could dig itself out with quantitative policy. But I’ll say this . . . [I]f the Bank of Japan were going to engage in quantitative policy – if it were willing to expand the size of its balance sheet by as much as I believe would be necessary to be effective – I believe it would need to get some prior commitment from the fiscal authorities that it would be able to pull the money out later. As you know, the Bank of Japan was made independent from the Ministry of Finance just two or three years ago. I think it’s hard for them to get back together and cooperate when they’ve spent a good deal of the last fifteen years trying to get a divorce. So that’s one problem. As a central banker, I have some sympathy with the BOJ situation on that score. The next issue is the banking situation. Much of the problem in Japan is not a central banking problem; it’s a fiscal bank regulation problem. If I were a central banker in Japan, I would be willing to pursue quantitative policy aggressively if I could get two commitments from the Ministry of Finance: one, for the fiscal authorities to support monetary policy along the lines of my earlier discussion and, two, for the fiscal authorities to deal with the problems in the banking system. To get to your last point, it’s true that as the deflation gathers speed, it’s harder to get out of this jam. But the mechanism I was talking about – the margin upon which quantitative policy would operate – would still be there. The implicit shadow broad liquidity services yield would be still there. In fact, it may be bigger in Japan than it has been for a long time because the value of having liquidity is much higher now than previously. This is the case, in turn, because the external finance premium is elevated as a consequence of the fact that asset valuations are depressed and loans are more costly to monitor and enforce. Therefore, I believe that there would be a substantial shadow margin for Japanese monetary policy to exploit if the BOJ were to do quantitative policy in the way I was suggesting.83

A few minutes later, Poole made this observation: I want to note first of all, without going into any detail about the lessons we may or may not have learned from Japan that the relevance of the Japanese experience for so-called quantitative policy is clear because it hasn’t been tried. Japanese money growth, as I understand it, never exceeded 4 percent in the decade, and for most

368

Marvin Goodfriend

of the 1990s it was a good bit less than 4 percent. And in the 1980s it was more like 8 percent or close to it. So what Japan did was to allow money growth to sag as the country went into a period of ongoing recession and the Bank of Japan never attempted to use quantitative policy to get out. So we don’t know from the Japanese experience whether that is going to work or not. In fact, the Japanese experience would be consistent with the traditional Chicago view, which says that the economy can fall into a depression if money growth is allowed to be too low for too long.84

Jerry Jordan, President of the Federal Reserve Bank of Cleveland, observed that quantitative policy on the scale contemplated would depreciate a nation’s foreign exchange rate very significantly. He suggested, in effect, that Japan’s unwillingness to allow the yen to depreciate sufficiently would have blocked the pursuit of quantitative policy of the scale necessary to be truly effective against deflation. Broaddus made the last comment and asked the last question: The key element that works in Marvin’s mechanism is the broad liquidity premium and yield and the ability to move that. So my question is this: Is it necessarily the case that the mechanism can’t work incrementally at the zero bound? In other words, because of the nature of the mechanism that is at work, it seems to me that we can’t rule out the possibility that we wouldn’t have to pump in a lot of money and then run the risk of a big resurgence in inflation. Is that right?85

Goodfriend answered: Yes, I think that’s true. If this policy were known to be effective and all the pieces were in place to make it work, it’s probably true that you could get by with a much smaller increase in the monetary base. That’s because once people saw that it was going to work, they would be more optimistic about the future. As with all these models, credibility is everything. If the policy is really credible, asset prices would begin to rise, and then the economy would start to recover. I think that would be true.86

IX. Conclusion The foregoing account of FOMC deliberations makes clear that the committee depends heavily on the chairperson to maximize its potential. The chairperson must encourage a diversity of views and then forge a consensus course of action from that diversity. Pressed by events, the committee must act decisively; yet it must explore uncharted contingencies and new ideas in an unhurried way.

Policy Debates at the Federal Open Market Committee: 1993–2002

369

Under pressure from Congress, Chairman Greenspan led the committee decisively in 1994 to improve its transparency and communication policies and Greenspan steered preemptive interest rate policy successfully. Greenspan encouraged unhurried consideration of problems less pressing. Treasuries did not disappear, but in the process of studying alternative assets to buy, the committee conceived of credit programs that proved useful during the recent turmoil. The committee also revived its appreciation for Treasuries only, which should help to motivate the normalization of the Fed’s balance sheet following the credit turmoil. Greenspan allowed ample time for the consideration of inflation targeting, although he chose not to expedite its adoption. He invited the committee to consider monetary policy at the zero interest bound, though most then thought it was a purely academic exercise. Greenspan’s ambivalence on relations with the Treasury enriched the debates on lending to Mexico and intervening in foreign exchange markets. Ultimately, Greenspan worked to end the Fed’s foreign exchange interventions. Above all, Chairman Greenspan created a culture of equality in the committee based on a respect for economic analysis. Economic reasoning is a great equalizer. By encouraging debate in terms of economic reasoning, Greenspan enabled the Richmond Fed to maximize its contribution to the FOMC. References Berry, J. (1993, December 12). The Fed feels the heat. The Washington Post, pp. H1, H5. Board of Governors of the Federal Reserve System. (1994). Minutes of the Federal Open Market Committee, Meeting of November. www.federalreserve.gov/monetarypolicy/ fomchistorical1994.htm Bordo, M. D., Humpage, O. F. & Schwartz, A. J. (2010). U.S. foreign-exchange intervention during the Volcker-Greenspan era (Working Paper 16345). Cambridge, MA: National Bureau of Economic Research. Broaddus, J. A., & Goodfriend, M. (1996). Foreign exchange operations and the Federal Reserve. Federal Reserve Bank of Richmond Economic Quarterly (Winter), 1–20. Broaddus, J. A., & Goodfriend, M. (2001). What assets should the Federal Reserve buy? Federal Reserve Bank of Richmond Economic Quarterly (Winter), 7–22. Campbell, J. Y. (1995). Some lessons from the yield curve. Journal of Economic Perspectives, 9(Summer), 129–152. Federal Open Market Committee. (1993–2002). Transcripts. Board of Governors of the Federal Reserve System. www.federalreserve.gov/monetarypolicy/fomc.htm Federal Reserve Bank of Kansas City. (2006). The Greenspan era: Lessons for the future. Jackson Hole, WY: Author. Goodfriend, M. (1986). Monetary mystique: Secrecy and central banking. Journal of Monetary Economics, 17(1), 63–92.

370

Marvin Goodfriend

Goodfriend, M. (1991). Interest rates and the conduct of monetary policy. In A. Meltzer & C. I. Plosser (Eds.), Carnegie Rochester Conference Series on Public Policy (Vol. 34, pp. 7–30). North Holland: Amsterdam. Goodfriend, M. (2005). Inflation Targeting in the United States? In B. Bernanke & M. Woodford (Eds.), The inflation targeting debate (pp. 311–337). Chicago: University of Chicago Press. Goodfriend, M. (2011). Central Banking in the credit turmoil: An assessment of Federal Reserve practice. Journal of Monetary Economics, 58(January), 1–12. Greenspan, A. (1990). Statement before the U.S. Congress, House of Representatives, Subcommittee on Domestic Monetary Policy of the Committee on Banking, Finance, and Urban Affairs. Hearing, 101 Congress, First Session. In Zero Inflation. Washington, DC: U.S. Government Printing Office. Greenspan, A. (1994a). Testimony by Alan Greenspan, Chairman of the Board of Governors of the Federal Reserve System before the Joint Economic Committee, United States Congress, January 31. Greenspan, A. (1994b). Hearing before the Subcommittee on Economic Growth and Credit Formation of the Committee on Banking, Finance and Urban Affairs, House of Representatives, 103 Congress, Second Session, Monetary Policy Report to Congress, February 22. In Conduct of Monetary Policy. Washington, DC: U.S. Government Printing Office. Hetzel, R. (2008). The monetary policy of the Federal Reserve: A history. New York: Cambridge University Press. Mallaby, S. (2010). More money than God: Hedge funds and the making of a new elite. New York: Penguin Press. Meltzer, A. H. (2009). A history of the Federal Reserve (Vol. 2). Chicago: University of Chicago Press. Monetary policy in a low inflation environment. (2000, November). Journal of Money, Credit, and Banking, Part 2. Taylor, J. (1993). Discretion versus policy rules in practice. In A. Meltzer & C. I. Plosser (Eds.), Carnegie Rochester Conference Series on Public Policy (Vol. 39, pp. 195–214). North Holland: Amsterdam.

Notes 1 The Friends of Allan Meltzer Professor of Economics, Tepper School of Business. Email: [email protected]. The paper was prepared for the Federal Reserve Bank of Atlanta–Rutgers University Conference, “A return to Jekyll Island: The origins, history, and future of the Federal Reserve,” November 5–6, 2010. Conversations with Al Broaddus greatly improved the paper. The research was supported by the Gailliot Center for Public Policy at the Tepper School, Carnegie Mellon University. 2 Obviously many other participants made significant contributions to FOMC discussions in the period considered here and throughout the FOMC’s history. The focus on Broaddus’s contributions reflects the author’s extensive collaboration with Broaddus in formulating and articulating the Richmond Fed’s policy positions. 3 The last two of these issues are directly relevant to policy issues the Fed is confronting currently.

Policy Debates at the Federal Open Market Committee: 1993–2002

371

4 Federal Reserve Bank of Kansas City (2006), Hetzel (2008), and Meltzer (2009) cover many of these issues and contain useful background material. 5 Taylor (1993). 6 Goodfriend (1986, 1991). 7 Goodfriend (1986). 8 Transcripts, October 5, 1993, p. 2. 9 Transcripts, November 16, 1993, p. 6. 10 Berry (1993, pp. H1, H5). 11 Goodfriend (1991, pp. 19–22). 12 See Goodfriend (2005, pp. 314–5) and Hetzel (2008, pp. 280–310). 13 Greenspan prepared markets for higher short-term interest rates in testimony before the Joint Economic Committee a few days before the February 1994 FOMC meeting (Greenspan, 1994a). He explained preemptive interest rate policy in the February 1994 Monetary Policy Report to Congress (Greenspan, 1994b). 14 Transcripts, February 3–4, p. 55. 15 Campbell (1995) constructs and reports a set of forward rate curves extracted from the corresponding U.S. spot yield curves at different dates in 1994. Interestingly, judging by the behavior of the forward rate curves, the bulk of the Fed’s policy impulses were delivered in three major steps – the first percentage point increase by early January, an additional percentage point in early February, and a third by early May. An announcement in mid-May constituted an impulse for easier policy and so forth. Most of the seven federal funds rate actions in 1994–5 did not constitute impulses in longer-term rates at the time they were implemented. 16 Transcripts, February 28, 1994, p. 9. 17 Transcripts, March 22, 1994, pp. 40–1. 18 Mallaby (2010), pp. 172–92, tells the story of the transmission of the Fed’s interest rate actions to the world bond markets from the point of view of hedge funds. 19 Transcripts, March 22, 1994, p. 42. 20 Transcripts, March 22, 1994, p. 41. 21 Transcripts, March 22, 1994, p. 43–4. 22 Transcripts, March 22, 1994, p. 44. 23 Transcripts, May 17, 1994, p. 32. 24 Transcripts, September 27, 1994, p. 42. 25 Transcripts, November 15, 1994, p. 36. 26 Transcripts, December 20, 1994, pp. 31–2. 27 Transcripts, December 20, 1994, p. 32. 28 Transcripts, January 31–February 1, 1995, p. 107–8. 29 Transcripts, March 22, 1994, pp. 4–5. 30 Transcripts, March 22, 1994, p. 5. 31 Transcripts, March 22, 1994, p. 12. 32 The FOMC has an agreement to “warehouse” foreign currencies for the U.S. Treasury and the Exchange Stabilization Fund (ESF). This is an arrangement under which the FOMC agrees to exchange, at the request of the Treasury, U.S. dollars for foreign currencies held by the Treasury or ESF over a limited period of time. The purpose of the warehousing facility is to supplement U.S. dollar resources of the Treasury and ESF for financing purchases of foreign currencies and related international operations.

372 33 34 35 36 37 38 39 40 41 42 43

44 45 46 47 48 49 50 51 52 53 54 55 56 57 58

59 60 61 62 63 64 65 66 67 68 69 70 71 72 73

Marvin Goodfriend Transcripts, January 31–February 1, 1995, p. 63. Transcripts, January 31–February 1, 1995, p. 64. Transcripts, January 31–February 1, 1995, p. 118. Transcripts, January 31–February 1, 1995, p. 131. Transcripts, January 31–February 1, 1995, p. 133. Transcripts, January 31–February 1, 1995, p. 136. Transcripts, January 31–February 1, 1995, pp. 136–7. Transcripts, January 31–February 1, 1995, p. 141. Transcripts, January 31–February 1, 1995, p. 142. Bordo, Humpage, and Schwartz (2010), p. 40. See Bordo et al. (2010) for a discussion of the Fed’s participation in foreign exchange operations with the Treasury in cooperation with financial authorities abroad in the 1980s. Transcripts, Presentation Materials, FOMC Notes, Peter Fisher, May 17, 1994, p 4. Transcripts, Presentation Materials, FOMC Notes, Peter Fisher, July 5–6, 1994, pp. 1–10. Transcripts, July 5–6, 1994, p. 2. Transcripts, July 5–6, 1994, p. 4. Transcripts, June 5–6, 1994, p. 5. Transcripts, June 5–6, 1994, p. 6. The work by Broaddus and Goodfriend (1996) is an extended critique of the Fed’s involvement in foreign exchange operations along the lines of Broaddus’s dissent. Board of Governors of the Federal Reserve System (1994). Transcripts, June 30–July 1, 1998, p. 6. Transcripts, June 30–July 1, 1998, p. 7. Transcripts, June 30–July 1, 1998, p. 10. Transcripts, October 3, 2000, pp. 11–14. Transcripts, October 3, 2000, p. 15. Transcripts, January 29–30, 2002, p. 6. The Neal Amendment defines price stability as a situation in which expectations of future inflation do not play a significant role in economic decision making. See Greenspan (1990). Transcripts, January 31–February 1, 1995, pp. 39–41. Transcripts, January 31–February 1, 1995, pp. 42–45. Transcripts, January 31–February 1, 1995, pp. 47–48. Transcripts, January 31–February 1, 1995, p. 58. Transcripts, July 2–3, 1996, p. 45. Transcripts, July 2–3, 1996, p. 47. Transcripts, July 2–3, 1996, p. 50. Transcripts, July 2–3, 1996, pp. 50–1. Transcripts, July 2–3, 1996, p. 58. Transcripts, July 2–3, 1996, p. 63. Transcripts, July 2–3, 1996, p. 64. Transcripts, July 2–3, 1996, p. 64. Transcripts, February 2–3, 1998, pp. 87–8. Transcripts, February 2–3, 1998, pp. 97–8. Transcripts, January 30–31, 2001, pp. 3–4.

Policy Debates at the Federal Open Market Committee: 1993–2002

373

74 The asset acquisition policy of the Eurosystem relies heavily on lending to banks. 75 Transcripts, January 30–31, 2001, pp. 7–14. Broaddus and Goodfriend (2001) contains an extended defense of Treasuries-only along the lines of Broaddus’s FOMC statement. 76 Monetary policy in a low-inflation environment (2000). 77 The Fed did not have authority to pay interest on reserves until 2008. 78 Alternatively, the funds could be used to repay bank loans – offsetting a tightening of bank credit, reducing the finance premium on bank loans, or shifting bank credit to borrowers more in need. 79 Since authorized to pay interest on reserves in 2008, the Fed can raise interest rates against inflation by paying interest on reserves without first selling assets or draining reserves (Goodfriend, 2011). 80 To be fully flexible against either inflation or deflation at the zero interest bound, the Fed should enlarge its surplus capital with the help of the fiscal authorities to guarantee its financial independence to raise interest rates against inflation by paying interest on reserves, whatever the size of its balance sheet (Goodfriend, 2011). 81 Transcripts, January 29–30, 2002, pp. 22–4. 82 Transcripts, January 29–30, 2002, p. 26. 83 Transcripts, January 29–30, 2002, pp. 27–8. 84 Transcripts, January 29–30, 2002, pp. 32–3. 85 Transcript, January 29–30, 2002, p. 42. 86 Transcript, January 29–30, 2002, p. 42.

SEVEN

Two Models of Land Overvaluation and Their Implications* Narayana R. Kocherlakota

I. Introduction From early 1996 through mid-2006, the price of residential land (henceforth, land) nearly tripled in the United States. It has since fallen by over 60 percent. As of the fourth quarter of 2009, the price of land was not even 10 percent above its price in 2000.1 These observations suggest to me, as they have to others, that in the mid-2000s, the price of land was overvalued (relative to the expected discounted value of future land rents). In this chapter, I construct two distinct models of this overvaluation. In the first, the land overvaluation is due to government guarantees of bank debt. I examine the implications of this model for the regulation of banks and other financial institutions. In the second, the land overvaluation is due to an asset price bubble. I examine the implications of this model for fiscal policy.2 In the remainder of the Introduction, I briefly lay out the arguments and conclusions described with greater detail and more technically in the two parts of the chapter that follow. The first part of the chapter constructs a simple model in which bank creditors (depositors and debt holders) receive implicit and explicit guarantees from the government.3 The key to the model is that the markets for land, mortgage origination, and bank debt are all competitive. This assumption implies that the benefits from debt guarantees end-up flowing to the owners of scarce factors in the economy. In the model, the only such scarce factor is land, and so the government guarantees affect the economy by increasing the value of land. I demonstrate that these guarantees can lead land to be overvalued. It is generally believed that the benefits of debt guarantees accrue to bank equity holders. However, they only do so if their bank enjoys market power in the mortgage origination process. The rents from government guarantees flow to the owners of scarce factors in the economy. In my model, the only scarce factor is land. 374

Two Models of Land Overvaluation and Their Implications

375

Interestingly, this overvaluation occurs even if bank debt holders are the only people in the economy who are actually aware of the guarantees. All other actors (including bank equity holders) simply respond to natural competitive pressures. Thus, landowners may have no idea why their buyers are willing to pay so much. Indeed, the buyers themselves have no idea why they are paying so much. All they know is that the banks are offering low mortgage rates. The banks are willing to offer these low mortgage rates because they are able to borrow at a low rate from debtors who believe themselves to be insured against loss. I show that regulations like capital requirements and leverage restrictions can eliminate this overvaluation. However, to be effective, bank regulation must react to shifts in the probability distribution of future land values. In particular, a given capital requirement or leverage cap might successfully eliminate all pricing distortions associated with debt guarantees for a given probability distribution of future land values. However, suppose that the distribution changes so that its left tail becomes longer. Then, the debt guarantees will lead to land being overvalued. Intuitively, the debt guarantees imply that taxpayers make transfers to bank creditors when they suffer losses. To forestall losses to taxpayers, good regulation should require that bank equity holders or landowners hold as much of land’s risk as possible. Such regulations must adjust to the underlying distribution of land values, and especially to changes in its left tail. The second part of the chapter models land overvaluation as being due to a rational bubble. What do I mean by a bubble? I mean that the asset’s price is higher than the price that a buyer would be willing to pay if he were required to never resell it. I refer to this “buy-and-hold-forever” price as the asset’s fundamental. So, with a bubble, assets are overvalued relative to fundamentals. What do I mean by rational? I mean that the overvaluation is common knowledge among all people in the model economy, and so risk-adjusted returns are always equalized across assets. At this point, the term “rational bubbles” may sound like an oxymoron. The key to this apparent contradiction is that returns are related to the expected rate of growth of prices and not to the level of the price. Two prices might grow at the same rate, but have consistently different levels. Two prices might have the same expectation, but have different variances around their means. Hence, even if risk-adjusted returns are equalized across all assets, some asset prices may be elevated relative to fundamentals, and some asset prices might fluctuate more than fundamentals.4 Beginning with Samuelson (1958), there have been many years of research into rational bubbles.5 In this literature, rational bubbles emerge as a

376

Narayana R. Kocherlakota

response to a particular kind of friction in loan markets. In private credit markets, the real rate of interest adjusts so as to equalize the demand for loans by borrowers with the supply of loans by lenders. However, in many circumstances, borrowers face credit constraints, in that they are unable to fully capitalize their future income. In those settings, the demand for loans is restricted. The loan market can clear only if the interest rate is low enough so as to reduce the supply of loans correspondingly. If the credit constraints are sufficiently severe (so that they drive down the market interest rate sufficiently), equilibrium asset prices may exhibit bubbles. These bubbles have a benefit, in that they provide savers with a vehicle to store wealth for future consumption purposes. To understand how this works, suppose a parcel of land has a fundamental value of $200,000. Without a bubble, person A can buy the land, using a downpayment of $40,000 and a nonrecourse mortgage from bank B for the remaining $160,000. But suppose the land has a bubble of $100,000, with the same risk-adjusted return as the fundamental. Now person A can buy the land using a down payment of $60,000 and a mortgage from bank B for $240,000. Person A pays more – but gets the same expected return. The bubble in the land price does not necessarily give either A or B a higher risk-adjusted return on their portfolios. However, it does allow A to save an additional $20,000 in the form of land and the owners of B to save an additional $80,000 in the form of land. Thus, bubbles enhance the ability of savers to save. At the same time, though, bubbles are intrinsically unstable because they rely on a fragile chain of mutual understanding. (I’m willing to pay too much for land today because someone else will pay too much and so on.) Changes in people’s beliefs in the viability of this chain can have big effects on the size of the bubble and on asset prices themselves. I use this framework to discuss useful fiscal policy responses to the emergence of bubbles. My basic argument is that, while they have some benefits, bubbles introduce instability into the economy. It may be possible for the government to avoid this possibility by using appropriate debt management, but only at some risk and cost. I then discuss optimal policy responses to the collapse of bubbles. A bubble collapse introduces two distinct problems into the economy: r The real interest rate has fallen (below the growth rate of the econ-

omy), because the supply of savings has contracted. This problem is necessarily highly persistent.

Two Models of Land Overvaluation and Their Implications

377

r The distribution of wealth has shifted, as those who owned the bubbly

asset lose, and those who did not have the bubbly asset would gain. This problem may be (relatively) transitory. A government can fix the first problem by committing to borrow at a higher fixed real interest rate that is lower than the growth rate in the economy. Because the real interest rate is lower than the growth rate of the economy, these loans are essentially a free lunch. The government raises funds in the first period and then rolls over the debt perpetually. Fixing the second problem, wealth redistribution, is more challenging. The government needs to be able to target the proceeds of the debt issue to the right people in the economy. It may well be difficult to figure out, from an economic standpoint, who should receive a transfer and who should not. My discussion of optimal policy responses to the emergence and collapse of bubbles is contingent on the validity of the rational bubble framework. I close with some admittedly speculative suggestions about how to use the policy prescriptions in this chapter as a benchmark for a more general class of models. There has been a great deal of discussion about how monetary policy should respond to bubbles. I will have nothing to say on this issue. In the United States, monetary policy generally takes the form of open market operations. In an open market operation, the Federal Reserve exchanges some government liabilities (reserves) for other government liabilities (treasuries or securities issued by government-sponsored enterprises) of equal market value. In this way, the Federal Reserve can influence the composition of outstanding government liabilities, but not the total value of outstanding government liabilities. (The latter is shaped by Congress.) As we will see, in the rational bubble framework, the time path of the total value of government liabilities matters a great deal for economic outcomes. However, the composition of those liabilities is, at a minimum, less essential. Hence, I abstract from the latter consideration entirely – and, in doing so, I abstract from monetary policy.6

II. Debt Guarantees and Land Overvaluation In this section, I consider the impact of bank debt guarantees on land prices. By “bank debt guarantees,” I mean to include guarantees for creditors of so-called too-big-to-fail institutions and guarantees of depositors. My discussion in the text is verbal. The interested reader is referred to Appendix A for mathematical details.

378

Narayana R. Kocherlakota

II.A. Model and Equilibrium I begin by describing a simple abstract two-period model. The key feature of the model is that bank creditors – both depositors and debt holders – have debt guarantees provided by the government. In equilibrium, these debt guarantees drive up the price of assets that are in fixed supply. In the model, there are equal numbers of four types of economic actors: landowners, land buyers, bank equity holders, and bank debt holders. (Implicitly, there is a fifth type – taxpayers – but I treat them as being outside the model.) All people are risk neutral, and they do not discount between the two periods. The owners each own a unit of land. Land pays off V units of consumption in period 2. Here, V is random, with cumulative distribution function  and probability density function φ. V has support [Vmin , Vmax ]. All land payoffs are always equal to one another, because I want to focus on aggregate risk. The owners do not desire consumption in period 2, but do desire consumption in period 1, and so are willing to sell to the buyers. The buyers buy land from the owners at an endogenous price p. The buyers are required to put down a downpayment of αp, where α is exogenous. They borrow the remaining (1 – α)p from the banks, using debt with an endogenous face value Fland . The loan is nonrecourse, except for the underlying land itself, and so the buyers get a random payoff equal to max(V − Fland ,0) from their purchase. I assume that the bank faces an exogenous restriction κ on its debt/equity ratio (this is supposed to stand in for capital requirements or, more meaningfully, for leverage caps). Bank debt holders make a loan with an endogenous face value Fbank to the bank equity holders. The equity holders face limited liability, and so they receive a payoff equal to max[min(V ,Fland ) − Fbank ,0]. The bank’s debt holders are fully guaranteed against any losses from fluctuations in V. Hence, their payoff in period 2 is simply the loan amount Fbank . Trade is competitive within the model. The price of land p, the face value of bank debt Fland , and the face value of land loans Fbank adjust so that debt holders, equity holders, and land buyers earn zero economic profit in equilibrium. Without debt guarantees, the price p equals the expected value

Two Models of Land Overvaluation and Their Implications

379

E(V). However, with debt guarantees, the price of land may be larger than E(V). Define λ = (1 − α)/(1 + 1/κ ). This is the key regulatory parameter. It ranges between 0 and 1, is decreasing in the downpayment α, and is increasing in the leverage restriction κ. When α is high, landowners hold much of the land risk. When κ is low, equity holders hold much of the land risk. But when λ is high, much of the land risk is held by guaranteed bank debt holders – and therefore by taxpayers. Given a value of λ, the land price is determined as follows7 : E (V )λ ≤ Vmin ⇒ p = E (V ). E (V )λ > Vmin ⇒ p = E{λp − V |V ≤ λp}(λp) + E (V ). If λE (V ) ≤ Vmin , then bank equity holders or land buyers hold all of the land risk. The probability of taxpayers making transfers to creditors is zero, and land is not overvalued. If λE (V ) > Vmin , then the bank and land buyers are jointly sufficiently leveraged so that bank debt is risky. There is a positive probability of taxpayers making transfers to creditors, and land is overvalued. Intuitively, the amount of the overvaluation exactly equals the expected value of the transfers from taxpayers to bank creditors.

II.B. Lessons of the Model In this section, I draw out some simple implications of the model for bank regulation. Implications for Leverage I have treated both the leverage restriction κ and the downpayment α as being exogenous variables (implicitly under the control of bank regulators). Together, these two parameters combine to form the key regulatory parameter λ, which ranges between 0 and 1. In this model, bank regulation eliminates all pricing distortions caused by government guarantees if λ is less than E(V)/Vmin . Bankers and policymakers have long debated the merits and costs of having more stringent capital requirements when banks are doing well. In this model, the cutoff value for λ does not change if the distribution of land values is scaled up or down proportionately. Instead, the requisite regulation has more to do with the left tail of the probability density of asset (here, land) values. If that tail becomes

380

Narayana R. Kocherlakota

long, then bank regulation needs to be tight. If the left tail is shorter, then bank regulation can be more relaxed. Lesson: Bank leverage restrictions should become tighter if the left tail of the probability density of future land values becomes longer. Implications for Down Payments I have treated the downpayment requirement α as being exogenous. Suppose E (V )(1 + κ −1 ) ≤ Vmin. Then, for any specification of α, λE (V ) ≤ Vmin and p = E (V ). Hence, the price of land is independent of the downpayment α, and there are no competitive pressures on its level. In contrast, if E (V )/(1 + κ −1 ) > Vmin , then the price of land is decreasing in α. Banks will compete to make α as low as is possible. Lesson: Bank leverage restrictions should become tighter when banks are actively competing with one another by lowering downpayment requirements. Implications for the Distribution of Rents and Knowledge In this model, landowners receive all of the rents from debt guarantees. Nonetheless, only bank creditors need to know about the guarantees. In equilibrium, the guarantees imply that bank debt holders make loans at a low rate, given their risk, to bank equity holders. The equity holders do not care why bank debt holders are willing to do so. The bank equity holders respond by being willing to make cheap loans, given the underlying risk, to land buyers. Again, the land buyers do not care why their loans have such a low interest rate. Finally, because their loans are nonrecourse and have such a low interest rate, the land buyers are willing to borrow a lot to finance their land purchases. In fact, they find that to buy land at all, they must be willing to borrow a lot and bid high. The landowners are, of course, delighted to get such a high price and, again, do not care why buyers are willing to pay so much. This chain of reasoning points out how competition spreads rents around in an economy. There are ex ante rents because of ex post debt guarantees. But debt markets are competitive, and so the actual recipients of debt guarantees compete away those rents. The ex ante owners of scarce factors are the ones who get the rents, without necessarily knowing their source. Lesson: Insufficiently stringent bank regulation may have relatively little impact on bank profits, but nonetheless lead scarce factors (finance talent, land) to be overvalued. Summary This simple model shows how ex post debt guarantees for bank creditors can lead to the overvaluation of land. It indicates how optimal bank regulation

Two Models of Land Overvaluation and Their Implications

381

needs to respond to changes in the shape of the probability distribution of future land values. In this discussion, I treated the probability distribution of future land values as exogenous. The distribution may change for demographic, technological, or much less fundamental reasons. Bank regulation should respond in the same way to these changes, regardless of their source.

III. Rational Bubbles in Land Prices In this section, I describe the basics of rational bubble models. I then discuss the implications of these models for optimal fiscal policy. As I mention in the introduction, I abstract completely from monetary policy. There are many kinds of models with rational bubbles: overlappinggenerations models, models with exogenous borrowing constraints, and models with endogenous borrowing constraints. My discussion attempts to treat all at once. In Appendix B, readers can find an explicit mathematical framework that illustrates the verbal points made in the text.

III.A. Rational Bubble Basics This treatment draws on the works of Tirole (1985), Kocherlakota (1992), and Santos and Woodford (1997). The discussion is designed to apply to a model with a large number of infinitely lived households or to an overlapping-generations model. In either case, I assume that all households are symmetrically informed at all dates. Definition and Equilibrium Restrictions I begin by defining a bubble. Definition 1. In a given equilibrium, an asset’s price has a bubble if the price exceeds the asset’s buy-and-hold-forever value (which is, again, the price that a buyer would be willing to pay if he or she were required to never resell it). In what follows, I assume that there are no aggregate fluctuations in the economy. This assumption means that all assets earn the same constant rate of return r. This assumption can be relaxed, but only at a nontrivial notational cost. Suppose an asset’s price has a bubble. By assumption, the asset’s expected return is constant over time and equal to r. Hence, by the definition of expected return, the asset’s price and dividend satisfy the restriction Pt = Et {Pt +1 + Dt +1 }/(1 + r ),

(1)

382

Narayana R. Kocherlakota

where Et is the conditional expectation, based on information available at date t. I can roll this first-order difference equation forward to conclude that Pt = EDVt + ηt .

(2)

Here, EDV stands for expected discounted value of future dividends (discounted at rate r) and is the buy-and-hold-forever value of the asset. The random variable ηt is the bubble; it is necessarily nonnegative and satisfies the restriction ηt = Et ηt +1 /(1 + r ).

(3)

The asset has a bubble if ηt > 0. Note that Eq. (3) implies that an asset that has a bubble currently must have had one in the past. However, the size of the bubble may fluctuate over time. Equation (3) immediately implies that there is a sample path along which the asset price grows at rate r. But the asset price cannot grow faster than the economy. Hence, if the economy grows at g, then r ≤ g. In other words: If the asset has a bubble, the equilibrium rate of return r cannot exceed the growth rate of the economy g.8 Note that this requirement is a long-run restriction. Over short periods of time, an asset’s price could grow more rapidly than the economy – but this trend cannot continue forever. There is an immediate implication from this result: At any point in time, current investors must face restrictions on their ability to borrow against the future income of the society. The discount rate r is no larger than g. Hence, the present value of the future income of the economy is infinite. If people could fully capitalize all of this future wealth, they could buy an infinite amount of goods. It follows that if some asset’s price has a bubble, then, at any date, everyone faces borrowing constraints that bind currently or at some point in the future. These borrowing constraints can take a number of forms. They could be purely exogenous limits on people’s debt positions. Or, they could be endogenous restrictions, as in Kehoe and Levine (1993). Or, they could represent the inability of parents to borrow against their children’s incomes (as in an overlapping-generations economy). For similar reasons, no currently traded asset with rate of return r can have an underlying dividend stream that grows at the same rate as the economy. It follows that growth in a bubbly economy occurs through the addition of assets, not through growth in the payoff streams of a currently available asset.

Two Models of Land Overvaluation and Their Implications

383

Low Interest Rates and Mutual Trust Suppose a household has a temporarily low income in a given year compared with its average income realization. There are two ways for that household to smooth its consumption over that temporary downward blip in income: It can have saved from the past or it can borrow from the future. In a well-functioning loan market, some households save and some borrow. But now suppose borrowers become more tightly constrained in their ability to borrow. There are many ways to think about such frictions, but at their heart, they all represent an erosion of interpersonal trust. Suppose everyone knows, with certainty, that I will have $X next year. In principle, I should be able to borrow $X. A borrowing constraint means that I cannot borrow the full $X, even though everyone knows that I will have $X. I am not trusted to fulfill my obligation – because of my personal characteristics, legal institutions, or both. More severe borrowing constraints mean that more households will try to smooth their consumption by saving and fewer by borrowing. In terms of basic economics, the demand for loans by borrowers has fallen relative to the supply of loans by lenders. The loan market can be equilibrated only by the interest rate r falling. It is when the borrowing constraints are so tight that r falls below g that bubbles can occur. To understand why bubbles occur, it is useful to think through why people save. Savers defer consumption from the current period to some future date, when it is more valuable to them. For example, in a model with incomplete markets, a saver gives up current consumption in order to have more consumption at some later date when his or her income will be low. In an overlapping-generations economy, the saver gives up consumption today for consumption when he or she is older. Borrowing constraints make it hard for savers to accomplish this deferral of consumption. The saver simply cannot trust borrowers to make repayments when needed. Bubbles allow savers to defer consumption without the aid of individual borrowers. The saver buys the bubbly asset at an elevated price today and then sells it when he or she needs consumption. The key is that he or she anticipates (with reasonable confidence) that there will be a buyer willing to buy the bubbly asset at that critical point in the future. In this sense, a bubble is a form of collective trust. Savers are unwilling to lend $X to a given borrower, because the borrower may not repay the loan. However, savers are willing to invest $X in a bubbly asset because they feel confident that they can sell the asset for an elevated price in the future. Bubbles substitute collective trust for individual trust.

384

Narayana R. Kocherlakota

In this sense, bubbles expand the range of what people can accomplish. The problem is that arrangements based on mutual trust are intrinsically unstable. In the future, buyers of the asset may have considerably more doubt in their ability to resell. This doubt can lower the future price of the asset and potentially eliminate the bubble permanently. Thus, fluctuations in the mutual trust that underlies the bubble can lead to fluctuations in a bubbly asset’s value and the ability of people in the society to effectively transfer wealth across periods.

III.B. Macroeconomics of Bubbles In this section, I turn to the consideration of optimal fiscal policy in the context of a bubble. Responding to Ongoing Bubbles I begin with the first question that arises in any discussion of bubbles: How can governments “pop” bubbles? I’ve argued that bubbles represent a form of collective trust. Person A is willing to pay a price above fundamentals for land because he or she believes that the bubble component of the price will grow at the rate of return. That belief relies on person A’s trust that someone else will be willing to pay that higher price at some point in the future. Of course, I say collective trust – others might say collective delusion. Whatever one might call it, though, I see no easy way to end it. On the one hand, it is not clear if private citizens will be greatly influenced by government statements about land or other assets being overvalued. On the other hand, as I show in the next section, the consequences of a bubble suddenly bursting can be extremely adverse. Even if a government has the ability to reduce a bubble, it would prefer to do so only gradually – deflate it rather than pop it. The government can attempt to replace a private sector land bubble with a public sector bubble in its debt. Suppose that the government gauges that the total bubble across all private sector assets is $Y. The government can try to siphon off this bubble by offering Y dollars of debt for sale, in which the debt pays real interest rate r. Because there is a bubble, the real interest rate is no larger than g, and so the government can then roll this debt over perpetually. Under this policy, there is an equilibrium in which the price of the private sector assets reverts to fundamentals and the bubble is transferred to government debt. This kind of shift is attractive in some ways if the public debt bubble is regarded as being more stable than the private sector’s bubble. However, the

Two Models of Land Overvaluation and Their Implications

385

shift will make some agents worse off, and the bubble in public debt may also be unstable.8 The Consequences of a Bursting Bubble With any bubble, there is some probability that it will fall in size – possibly greatly. This shrinkage in the size of the bubble can have significant effects on an economy. As I have argued earlier, bubbles emerge in response to significant borrowing frictions. Bubbles insulate economies against the impact of these frictions by providing savers a vehicle with which to transfer current resources into the future. A fall in the bubble’s size reduces the ability of savers to save. With that in mind, the exact impact of the bubble’s collapse depends on the role of asset accumulation in the economy. However, a bubble bursting generally has two key consequences. First, because the bubble contributed to the supply of savings in the economy, its collapse means that the supply of savings has fallen, and so the real interest rate should fall. Moreover, a burst bubble has a low or zero value today because people expect its value to be low or zero in the future. The fall in the real interest rate is therefore likely to be persistent. Second, some agents held bubbly assets and others did not. The fall in the value of the bubbly asset redistributes wealth from the bubbly asset holders to the nonholders. The persistence of this effect depends on the law of motion of the distribution of wealth in the society. In many models with borrowing constraints, the long-run distribution of wealth is invariant to shocks. In such models, this latter effect will be transitory. The fall in the real interest rate and the redistribution of wealth can impact the economy in a number of ways. I focus on the implications for labor supply9 and investment. Bubbly asset holders are less wealthy after the bubble’s collapse, and so are more willing to supply labor, whereas nonholders are less willing to do so. At the same time, the fall in the real interest rate retards labor supply for everyone. Overall, the impact of the bubble’s collapse on aggregate labor supply, and on aggregate output, depends critically on the joint distribution of labor productivities and asset holdings in the economy. If the holders of a bubbly asset have high labor productivities, then the bubble’s collapse may actually increase output as asset holders supply more labor. The fall in the real interest rate after a bubble bursts will increase the demand for investment goods. Firms and others with nonbubbly sources of resources should engage in more investment. However, a bubbly economy is an economy with significant credit constraints. When the bubble is large, it is

386

Narayana R. Kocherlakota

likely to be entrepreneurs and firms who anticipate having good investment projects who are holding the bubbly assets.10 If the bubble bursts, they lose that source of funds. Given the credit constraints that generated the bubble in the first place, they will be unable to take advantage of desirable projects. What Should Be Done After a Bubble Bursts? As I described previously, the bursting of the bubble has two consequences for an economy: a fall in the real interest rate and a redistribution of wealth. The government can potentially fix both problems. To fix the first problem, the government can issue one-period debt with real interest rate r, where r was the real interest rate before the bubble burst. It can then commit to roll that debt over forever. If credible, this commitment will serve to raise the real interest rate permanently. The fresh debt gives savers the needed vehicle for savings. Because the bubbly interest rate r was less than or equal to the growth rate g, this perpetual rollover plan is feasible without any extra taxes ever being collected. However, it does rely on the same kind of mutual trust that supported the bubble. At any date, people are willing to buy the debt today only because they believe that the government will be able to pay it off with a future sale of debt. Instability in this chain of beliefs may lead to the government having to raise taxes or default. The debt rollover plan provides the government with extra resources. As well, the increase in the interest rate generates more labor supply and more output. The government can generate even higher levels of output after the bubble’s collapse by allocating these funds in an appropriate way. For example, it can target the funds to people with relatively low labor productivities. Given the usual wealth effects on labor supply, this shift in the distribution of wealth from highly productive to less productive workers will generate higher output.11 Similarly, the government could generate higher output by targeting funds to entrepreneurs with highly productive investment projects. It may be hard for the government to identify the relevant targets in this scheme. A more feasible option might be to allocate government funds so as to re-create the distribution of wealth that existed under the bubble. By doing so, the government can also re-create the higher levels of output that existed under the bubble. This distribution of funds can work in many ways. Consider person A who has a mortgage from bank B with principal $200,000. Person A’s property was worth $300,000 at the peak of the bubble and is now worth $150,000. How should the government allocate the proceeds of its new debt issue between A and B? There are many ways to proceed, but my favorite

Two Models of Land Overvaluation and Their Implications

387

is that the government pays bank B $150,000 to write down the value of the mortgage to $50,000. I like this approach because person A keeps her property and again has $100,000 of equity in her property. In addition, B has a (presumably viable) debt worth $50,000 from A and has received $150,000 from the government. This strategy is reminiscent of many suggested foreclosure mitigation plans. However, within the context of the model, the government can recapture the bubbly distribution of wealth only by using the same strategy regardless of when A bought her property or how she financed it. For example, suppose A bought the property for $100,000 cash in 1995, its value rose to $300,000 in 2006, and after the bubble burst, the property is now worth $150,000. To restore the bubbly distribution of wealth, the government should simply give A $50,000 to replace the lost bubbly wealth.

III.C. But Is It a Bubble? I have described how a government should use debt management to respond to the emergence of, and the collapse of, a bubble in land prices. I have done so in the context of models in which everyone agrees that there is a bubble in land prices. In reality, we always face this question: How do we know if a given run-up in land prices is due to a bubble or due to fundamentals? Answering this question is always hard. However, the preceding discussion suggests that it is better to ask a different – perhaps easier – question. In the models that I consider, the desirable interventions involve the government raising interest rates and then raising revenue through a debt issue. Because r ≤ g in the bubbly economies, the government can plan on rolling the debt over perpetually. In this way, the interventions can raise resources today without requiring new taxes in the future.12 Of course, the models in this chapter are artificially deterministic. In reality, these interventions always run the risk of requiring the government to raise taxes in the future if interest rates move in the wrong direction. For example, the 10-year real yield on Treasury Inflation-Protected Securities is currently less than 1 percent. The government could issue more of these bonds. However, in 10 years, it will need to pay the principal on these bonds, and it does not know what the real interest rate (or expected growth rate) will be then. So, in the models that I use in this chapter, the government can construct a costless intervention that makes everyone better off after a bubble

388

Narayana R. Kocherlakota

bursts. In alternative models, the intervention may have costs. Nonetheless, I conjecture that there are many models in which large falls in asset prices would prompt governments to contemplate fiscal interventions. The point is that in any model with borrowing constraints, large asset price run-ups – whatever their source – provide agents with a larger supply of savings. If the asset’s price falls, the pool of savings shrinks, and the real interest rate will fall. In reaction, just as in the model in this chapter, the government can sell more of its own debt and make well-chosen transfers.13 In the models in this chapter, this intervention could be done without any costs to anyone. In a broader class of models without bubbles, the government’s intervention would necessarily expose some taxpayers – maybe in the future – to some loss of income. These potential losses mean that the government must face hard distributional questions about the optimality of intervention.

IV. Conclusions Many observers have characterized land as having been overvalued in the 2000s relative to fundamentals. In this chapter, I have explored two distinct models of that overvaluation. In the first, overvaluation emerges as a result of implicit guarantees for bank creditors (depositors and debt holders). In the second, overvaluation is a consequence of a rational bubble. I have described appropriate regulatory and fiscal responses in these models. It should go without saying that these analyses are contingent on particular models. An important topic for future research is the evaluation of the robustness of the results in this chapter to other modeling approaches.

APPENDIX A I describe a simple abstract two-period model. The key feature of the model is that bank debt holders have debt guarantees. These debt guarantees drive up the price of an asset that is in fixed supply. There are equal numbers of four types of economic actors: owners, buyers, bank equity holders, and bank debt holders. (Actually, there is a fifth type of actor – taxpayer – who exists outside the model.) All people are risk neutral, and they do not discount between the two periods. The owners each own an asset that pays off V units of consumption in period 2. Here, V is random, with cdf  and density φ over support [Vmin ,Vmax ]. The assets’ payoffs are always equal to one another, because I want to focus on aggregate risk. The

Two Models of Land Overvaluation and Their Implications

389

owners do not desire consumption in period 2, but do desire consumption in period 1, and so are willing to sell to the buyers. The buyers buy the assets from the owners at an endogenous price p. The buyers are required to put down a downpayment of αp, where α is exogenous. They borrow the remaining (1 – α)p from the banks, using debt with an endogenous face value Fland (to match up with the notation in the chapter). The loan is collateralized only by the asset itself. Hence, the buyers get a random payoff equal to max(V − Fland ,0) from their purchase. I assume that the bank faces an exogenous restriction κ on its debt/equity ratio. (This is supposed to stand in for capital requirements or, more meaningfully, for leverage caps.) Bank debt holders make a loan with an endogenous face value Fbank to the bank equity holders. The equity holders enjoy limited liability, and so they receive a payoff equal to max[min(V ,Fland ) − Fbank ,0]. The bank’s debt holders are fully guaranteed by the government against any losses from fluctuations in V. Hence, their payoff in period 2 is simply the loan amount Fbank . Trade is competitive within the model.14 It follows that the buyers’ downpayment must equal the period 1 expected value of their period 2 payoff:

αp =

Vmax

Vmin

max (V − Fland ,0)φ(V )dV.

As well, the bank owners’ payment to the buyers must equal the expected value of their period 2 payoffs:

V max (1 − α)p = Fbank + {max[min(V ,Fland ) − Fbank ,0]}φ(V)dV. Vmin

Finally, the leverage restriction implies that

V max κ {max[min(V ,Fland ) − Fbank ,0]}φ(V )dV = Fbank Vmin

so that the debt/equity ratio equals κ. (I am requiring the bank to issue as much debt as possible. However, given the existence of the bailouts, the bank will find this behavior at least weakly optimal.) We can find the endogenous asset price p, asset loan face value Fland , and bank loan face value Fbank by

390

Narayana R. Kocherlakota

solving the previous three equations (given the downpayment α and the leverage cap κ). Without debt guarantees, the risk neutrality of the people in the model implies that the asset price p = E(V). However, the presence of the debt guarantees can make the asset price higher than E(V). Proposition 1. If (1 − α)E (V )/(1 + 1/κ ) ≤ Vmin , then p = E (V ). If (1 − α)E (V )/(1 + 1/κ ) > Vmin , then p > E (V ), and p is the unique solution to the equation

p=

(1−α)p/(1+1/κ )

[(1 − α)p/(1 + 1/κ ) − V ]φ(V )dV + E (V ).

Vmin

Proof. Note that min(x,c) + max(x,c) = c + x, and so min(x,c) + max(x − c,0) = x. This simple mathematical relationship plays a key role in what follows. Consider the zero profit conditions for the banks and the asset buyers:

(1 − α)p =

Vmax

Vmin

=

Vmax

Vmin



αp =

min(V ,Fland )φ(V )dV

Vmax

+

max[min(V ,Fland ) − Fbank ,0]φ(V )dV + Fbank

Vmin Vmax

Vmin

{Fbank − min[min(Fland ,V ),Fbank ]}φ(V )dV ,

max(V − Fland ,0)φ(V )dV.

Adding these together, I get

p = E (V ) +

Vmax

Vmin

Fbank − min[min(Fland ,V ),Fbank ]φ(V )dV.

The leverage restriction says that

κ

Vmax

Vmin

{max[min(V ,Fland ) − Fbank ,0]}φ(V )dV = Fbank .

Two Models of Land Overvaluation and Their Implications

391

If Fland ≤ Fbank , then Fbank = Fland = 0. Hence Fbank < Fland . It follows that

Vmax

p = E (V ) +

Vmin

[Fbank − min(Fbank ,V )]φ(V )dV.

The leverage restriction implies that Fbank (1 + κ −1 ) = (1 − α)p, and so

p = E (V ) +

Vmax

{(1 − α)p/(1 + κ −1 )

Vmin

− min[(1 − α)p/(1 + κ −1 ),V ]}φ(V )dV

(1−α)p/(1+κ −1 ) = E (V ) + [(1 − α)p/(1 + κ −1 ) Vmin

−V ]φ(V )dV.

(4)

If I set p = E (V ), the left-hand side (LHS) is no larger than the right-hand side (RHS). If I set p = Vmax (1 + κ −1 )/(1 − α), then the LHS is larger than the RHS. Hence, there exists a solution. The derivative of the LHS of Eq. (4) with respect to p is 1. The derivative of the RHS is (1 − α)[(1 − α)p/(1 + κ −1 )]/(1 + κ −1 ) < 1. It follows that there is a unique solution to (4). If E (V )(1 − α)/(1 + κ −1 ) ≤ Vmin , then p = E (V ) satisfies this equation. If E (V )(1 − α)/(1 + κ −1 ) > Vmin , then p = E (V ) does not solve this equation, because the right-hand side is larger than the left-hand side. It follows that p > E (V ) and solves the equation in the proposition. If the downpayment requirement α is sufficiently small or the leverage cap κ is sufficiently large, then debt guarantees lead the asset’s price to be larger than its fundamental value E(V). The degree of “mispricing” is decreasing in α and increasing in κ. Here, I have treated the downpayment requirement α as exogenous. It is worth noting that if E (V )/(1 + 1/κ ) ≤ Vmin , the banks are indifferent across the choice of downpayment requirement, because p = E (V ) regardless of the downpayment requirement. However, if E (V )/(1 + 1/κ ) > Vmin , the price of the asset is a strictly decreasing function of α. Under this condition, if banks are allowed to choose their own α and Fb , they will compete the downpayment requirement down to whatever minimum regulators might allow.

392

Narayana R. Kocherlakota

In the same vein, if banks can choose their debt/equity ratio, they will compete κ up to infinity. If α = 0 and κ = ∞, then p = Vmax . The bank is fully debt financed. Its debt has a face value of Vmax , and the loan to the asset buyer also has a face value of Vmax . The asset buyer defaults almost always, and the bank defaults almost always on its obligation to its debtor. As a result, the government is forced to pay the debtor Vmax − V when the value of the asset is V.

APPENDIX B Consider an economy with equal measures of infinitely lived odd and even agents. Odd agents are productive in odd periods; they produce nt units of consumption by exerting nt units of effort in those periods. They are unproductive in even periods. Even agents are productive in even periods; they produce nt units of consumption by exerting nt units of effort in those periods. They are unproductive in odd periods. Agents have a momentary utility function ln(ct ) − nt and a discount factor equal to β, where 0 < β < 1. Suppose agents are able to borrow and lend with one another, using oneperiod bonds, subject to a borrowing constraint of the form that requires their bond holdings to be larger than – B: bt +1 ≥ −B. This means that their flow constraint looks like ct + bt +1 ≤ wt nt + bt (1 + Rt ), where wt = 1 when the agent is productive and 0 when the agent is unproductive.

B.1. Borrowing Constraints and Interest Rates For the moment, I focus on stationary equilibria in which R is constant. This means that I am assuming that the initial asset position of the odd agents equals –B(1+R) and the initial asset position of the even agents equals B. The equilibrium consumptions, effort, and interest rate satisfy cprod + B = n − B(1 + R),

Two Models of Land Overvaluation and Their Implications

393

cunprod − B = B(1 + R), βcprod (1 + R)/cunprod = 1, 1/cprod = 1. I can solve these equations to get15 (1 + R) = (B−1 β − 1)−1 , cunprod = cprod =

B(2 + R), 1,

n = 1 + B(2 + R).

(5)

It follows that R is an increasing function of B. As B falls, so does the equilibrium interest rate.

B.2. Bubbles The equilibrium interest rate is less than 0 if and only if B < β/2. I claim that, for any such B, I can construct an equilibrium with a bubble. In particular, endow each of the even agents with a unit of land that pays no dividend. Suppose that agents can trade land and one-period debt in each period. They cannot short-sell land, and they cannot borrow more than B in each period. Then, their budget sets are defined by the constraints ct + bt +1 + Lt +1 pt = wt nt + bt (1 + R) + Lt pt , Lt ≥ 0, bt ≥ −B, where Lt is the agent’s landholdings at the beginning of period t and pt is the price of land in period t. There is an equilibrium (as previously defined) in which R < 0 and land has no value. But there is also an equilibrium in which R is constant at 0 and land has constant value p: cprod = 1, cunprod = β, n = 1 + β, p = β − 2B.

394

Narayana R. Kocherlakota

The bubble in land allows agents to engage in more effective intertemporal trades. Output is higher with the bubble, the unproductive agents consume more, and the productive agents consume the same.

B.3. Crowding Out Bubbles I have described two possible equilibria in the economy with land. There are a host of others that feature stochastic fluctuations in land prices over time. A government may wish to implement policies that remove this source of instability from the economy. Suppose that the economy is in the bubbly equilibrium previously described. In a given period t, the government in period t offers β − 2B units of one-period debt for sale, with a commitment to keeping the real interest rate at R = 0 forever. (I assume that this offer was unanticipated in earlier periods.) The government uses the proceeds of the sale to buy goods from the productive agent. It repays the debt by rolling it over in perpetuity. There are many equilibria in this economy. The government’s interest rate offer is no different from that in the private market. Hence, agents may simply ignore the government’s debt offer in every period. In this case, the bubbly equilibrium is unaffected by the government’s policy. However, there is also an equilibrium in which the price of land immediately falls to zero in period t. In that period, the productive agents produce 1 + β units of consumption. The government gets β − 2B of that production, though, and the unproductive agents get only 2B units of consumption. In period (t + 1), the economy returns to the bubbly equilibrium. The government may be able to substitute a perpetual rollover in its own debt for a land bubble. However, the substitution may not work. If it does work, then owners of the private sector bubble will be made worse off.

B.4. Aftermath of a Bubble Collapse As I previously indicated, there is a continuum of stochastic equilibria in this economy. Consider, for example, that there is an equilibrium in which the price of land is positive and constant but may fall to zero with a small probability. (In terms of quantities, the aftermath of a “wholly unanticipated” bursting of the bubble is a good approximation to its bursting with a low but positive probability.) In the first period after the bubble bursts, the productive agents produce 1 + 2B units of consumption. The unproductive agents consume 2B units of consumption. Then, the agents live in the nonbubbly equilibrium [with the lower interest rate R defined by (1)] after the bubble bursts. From a macroeconomic point of

Two Models of Land Overvaluation and Their Implications

395

view, there is a gradual (two-period) fall to a permanently new level of output.16 What is an appropriate response to the bursting of the bubble? I think about three distinct policies. I consider only their ex post consequences. In all three policies, the government commits to sell an arbitrary amount of one-period debt with real return 0. This part of the policy fixes the problem that the real interest rate is too low. The government commits to collecting no new taxes. Hence, the government plans to repay the debt issue by rolling over the debt in each period. Any debt rollover plan of this kind allows the government to generate extra resources in the first period after the bubble collapses. They differ in how the government spends those extra resources.

B.5. Public Investment Consider a policy in which the government uses the resources from its debt issue to fund public projects that are additively separable from the private economy. Then, in the first period after the bubble collapses, the productive agents produce 1 + β, and there is no recession at all. Nonetheless, unproductive agents do suffer a loss relative to the bubbly equilibrium, because they consume only 2B, rather than β. The government gets the other resources (β − 2B) from its debt issue. In the second period after the bubble’s collapse, the economy reverts to the bubbly outcome (because R = 0). This longer-run effect is a consequence of any of the three policies.

B.6. Uniform Transfer Suppose next that the government spreads the proceeds of the debt issue evenly across all agents. (Somewhat unrealistically, I think about this redistribution as happening in the first period after the bubble collapses. Thus, the government is redistributing the results of the debt issue simultaneously with actually doing the debt issue.) Then, in the first period after the bubble collapses, the productive agents produce 1 + β/2 + B. They spend β − 2B on the government’s debt issue, receive (β − 2B)/2 back as a transfer, repay their debt of B units of consumption to the unproductive agents, and lend B units of consumption to the unproductive agents. Unproductive agents consume β/2 + B. There is a one-period recession. In this sense, the fiscal multiplier associated with transfers is lower than that associated with public goods expenditures. The

396

Narayana R. Kocherlakota

lower fiscal multiplier is generated by a wealth effect: giving resources to productive agents leads them to work less hard.

B.7. Directed Transfer Finally, suppose the government gives out the proceeds of the debt issue only to unproductive agents (again as the debt issue is taking place). Now, the distribution of wealth is the same as in the bubbly economy and the equilibrium interest rates are the same. The equilibrium from this policy exactly mimics the bubbly equilibrium outcome.

B.8. Intuition Intuitively, there are two effects from a bubble collapse. First, the real interest rate falls permanently because there are fewer vehicles for saving. Second, there is a redistribution of wealth from bubble owners to nonowners. The impact of the first effect is a permanent one, whereas the impact of the second effect is transient. It is relatively easy to solve the first problem, using appropriate fiscal policy. It is harder to fix the second, because it requires targeted transfers. References Abel, A., Mankiw, N. G., Summers, L. H., & Zeckhauser, R. (1989). Assessing dynamic efficiency: Theory and evidence. Review of Economic Studies, 56(1), 1–20. Aiyagari, S. R., & McGrattan, E. R. (1998). The optimum quantity of debt. Journal of Monetary Economics, 42, 447–469. Caballero, R. J. (2006). On the macroeconomics of asset shortages. In A. Beyer & L. Reichlin (Eds.),The role of money: Money and monetary policy in the twenty-first century. Frankfurt, Germany: European Central Bank. Caballero, R. J., & Krishnamurthy, A. (2006). Bubbles and capital flow volatility: Causes and risk management. Journal of Monetary Economics, 53(1), 35–53. Davis, M. A., & Heathctoe, J. (2005). Housing and the business cycle. International Economic Review, 46, 751–784. Davis, M. A., & Heathcote, J. (2007). The price and quantity of residential land in the United States. Journal of Monetary Economics, 54, 2595–2620. For data, see www .lincolninst.edu/subcenters/land-values/. Diamond, P. A. (1965). National debt in a neoclassical growth model. American Economic Review, 55, 1126–1150. Farhi, E., & Tirole, J. (2010). Bubbly liquidity (Working Paper). Cambridge, MA: Harvard University Press. Hall, R. E. (2011). The long slump (Working Paper). Stanford, CA: Stanford University Press. Hellwig, C., & Lorenzoni, G. (2009). Bubbles and self-enforcing debt. Econometrica, 77, 1137–1164.

Two Models of Land Overvaluation and Their Implications

397

Kehoe, T. J., & David K. Levine, D. K. (1993). Debt-constrained asset markets. Review of Economic Studies, 60, 865–888. Kiyotaki, N., & Moore, J. H. (2008). Liquidity, business cycles, and monetary policy (Working Paper). Princeton, NJ: Princeton University Press. Kocherlakota, N. R. (1992). Bubbles and constraints on debt accumulation. Journal of Economic Theory, 57(1), 245–256. Kraay, A., & Ventura, J. (2007). The dot-com bubble, the Bush deficits, and the U.S. current account. In R. Clarida (Ed.), Current account imbalances: Sustainability and adjustment. Chicago: University of Chicago Press. Samuelson, P. A. (1958). An exact consumption loan model of interest, with or without the social contrivance of money. Journal of Political Economy, 66, 467–482. Santos, M. S., & Woodford, M. (1997). Rational asset pricing bubbles. Econometrica, 65(1), 19–57. Scheinkman, J. A., & Weiss, L. (1986). Borrowing constraints and aggregate economic activity. Econometrica, 54(1), 23–45. Tirole, J. (1985). Asset bubbles and overlapping generations. Econometrica, 53, 1499– 1528.

Notes * I thank R. Anton Braun, Doug Clement, Robert Hall, Anil Kashyap, Futoshi Narita, and Eric Rosengren for their comments. Some of the results and ideas in this chapter are similar to those in a working paper that I wrote in 2009, “Bursting bubbles: Consequences and cures.” 1 Note that I’m talking about land, not housing. Theoretically, it is hard to motivate the existence of significant overvaluation in housing structures (they’re readily replaceable). Empirically, there is considerably less evidence of overvaluation for structures than for land. Here, I refer to data from the Lincoln Institute of Land Policy that separates the price of housing into the price of structures and the price of land. The data were originally constructed by Davis and Heathcote (2005, 2007) and are derived from the Case-Shiller housing price index. These data indicate that the price of housing structures rose by less than 100 percent in nominal terms from 1996 to 2006, and has fallen by less than 10 percent since that date. 2 The views expressed do not necessarily reflect those of others in the Federal Reserve System. Indeed, I would hesitate to say that the results in this paper represent my own view. There are many possible models of the behavior of land prices in the 2000s. My goal in this chapter is only to describe the implications of two of them. I also make no claims to originality: I am sure that much of what I say is well known to most macroeconomists and financial economists. 3 I am primarily interested in the ex ante impact of the guarantees. What matters is whether creditors expect such guarantees, not whether they actually receive them. 4 How can I say that risk-adjusted land returns were equalized to those of other assets in the 2000s, when land prices were growing so fast? An asset’s price may grow rapidly even if the asset’s risk-adjusted return is the same as other assets’ returns. However, in the equilibrium of an efficient market, an asset’s price can grow rapidly from one year to the next only if there is a nontrivial probability that it might fall. In other words, a strong believer in efficient market theory who watched land

398

5

6

7 8

9 10 11

12

13 14

15

Narayana R. Kocherlakota appreciate rapidly in the 2000s would have said that there must be some probability of a big fall in land prices. An even more sophisticated believer in efficient markets might have speculated that land’s high returns are possible only because land has a large beta. Such a true believer in efficiency would have predicted that a fall in land prices would take place simultaneously with a large stock market crash, a deep recession, or both. Along with many others, Diamond (1965), Tirole (1985), Kocherlakota (1992), Santos and Woodford (1997), and Hellwig and Lorenzoni (2009) provide theoretical treatments of rational bubbles. As I do in this paper, Caballero (2006), Caballero and Krishnamurthy (2006), Kraay and Ventura (2007), and Farhi and Tirole (2010) analyze policy interventions in the context of models in which bubbles emerge in response to financing frictions. As I previously emphasized, a bubble collapse leads to a fall in the real interest rate. In the presence of a lower bound on the nominal interest rate, the real interest rate might not be able to fall as much as is necessary to clear all markets (as in Hall, 2011). As a result, the bubble’s collapse might have even more deleterious effects. See Appendix A for the full mathematical analysis. Because of the government’s ability to tax, public sector bubbles may be more stable than private sector bubbles. See Caballero and Krishnamurthy (2006) for a full discussion. This discussion is reminiscent of the behavior of the equilibrium described in Scheinkman and Weiss (1986). This discussion is reminiscent of that in Kiyotaki and Moore (2008). These same wealth effects imply that the government could also generate higher output by spending its new resources on public goods. The welfare consequences of such an approach depend on the details of how those public goods affect private agents. Abel, Mankiw, Summers, and Zeckhauser (1989) consider dynamic efficiency in overlapping-generations economies with aggregate shocks. They show that in these models, one cannot reliably evaluate dynamic efficiency by comparing average riskfree interest rates to average growth rates. This conclusion applies to my model as well. Their solution is to use the sign of net inflows into firms. This solution is valid in their context but does not work in my model with financial constraints. (Kraay and Ventura, 2007, make a similar point.) These kinds of interventions are studied by Aiyagari and McGrattan (1998). There are a variety of ways that trade could be structured. The only requirement is that, in equilibrium, the buyers, bank equity holders, and bank lenders compete among themselves so as to maximize the price p that buyers bid for a given piece of land, given the zero economic profit constraints subsequently described. I am assuming that B is sufficiently low that 1/cunprod − β(1 + R) > 0

so that the unproductive agents have binding borrowing constraints. This is true for an interval of B that is bounded from above by β/(1 + β ). 16 In the wake of the bubble’s collapse, the labor wedge grows. Labor productivity is unaffected by the fall in the bubble. However, the amount of labor falls. Through the lens of a representative household model, it would appear that there has been a shock that reduces the household’s marginal utility of consumption.

EIGHT

Panel Discussion: November 6, 2010 Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

The conference concluded with a panel discussion. Participating in the panel were Ben Bernanke (chairman, Board of Governors of the Federal Reserve System since 2006), Alan Greenspan (chairman, 1987–2006), and Gerald Corrigan (president, Federal Reserve Bank of New York, 1984–93). In addition, there was a video presentation by Paul Volcker (chairman, 1979–87). The panel was introduced by Dennis Lockhart (president, Federal Reserve Bank of Atlanta since 2007) and moderated by Raghuram Rajan (professor of finance, University of Chicago).

Dennis Lockhart: For this last session, which I know is going to be very special, we invited Paul Volcker to attend, and unfortunately he’s out of the country. But we got the next best thing: Paul Volcker, in fact, will introduce this panel; so we’ll start with a video. Paul Volcker video: Well, I’m sorry I’m halfway around the world and unable to participate in this great restoration of Jekyll Island and the return to that place where the Federal Reserve idea took form. You know, none of us can claim personal memory of the creation of the Federal Reserve, but your panel discussions are going to encompass years of experience. Gerry Corrigan started back in the 1970s; Alan Greenspan was there for a very long tenure (it almost seemed forever); now Ben Bernanke’s there and in the middle of a great crisis, one of the most challenging periods in all of Federal Reserve history. In that group, I claim only one distinction: I walked into the Federal Reserve Bank in New York, that great fortress downtown, in 1949. I think that’s before Ben Bernanke was born. So I do have a little perspective; somehow, I sometimes think that was only yesterday when I first entered the Federal Reserve. You know, we have had a large crisis and still have the remnants of that crisis, or maybe more than the remnants. And we’re tempted to think that the Federal Reserve has never been so challenged as in these recent years, and that may be true. But it’s certainly not true that there weren’t a lot of controversy 399

400

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

and challenges right through the history of the Federal Reserve. You know, the United States experimented with central banking back in the late 1700s and early 1800s, and that was very controversial, not very successful; it became a matter of controversy in presidential elections. The central bank ended in the 1820s, and it wasn’t until 1913 when Woodrow Wilson, the great achievement of his administration, founded the Federal Reserve. It was quite a different institution; there was a lot of suspicion that it would be politicized or dominated by Wall Street or too-concentrated power. Some of those suspicions were reflected in the fact that the Federal Reserve in those days didn’t even have the authority to buy government securities. There wasn’t any idea of open market operations. There really wasn’t any concept of monetary policy in the interest of stabilizing a growing economy. Now that’s all past, but the controversy clearly has remained, the challenges have remained. The Federal Reserve has taken heroic action in recent years; it’s acted fully up to the limits of its authority, it stabilized the financial system, saved the economy. It’s been, to me anyway, an unexpected reward for those efforts. Somehow we come out of that crisis with the Federal Reserve regulatory authorities enhanced. Now I don’t want to suggest the controversy surrounding the Federal Reserve System has in any way diminished, but I do want to make just one point: I think there’s a consistent thread through the Federal Reserve’s history, and it’s a thread that I consider of enormous importance, not just for monetary policy, not for the economy, but for the country more generally. The leadership of the Federal Reserve – and it’s exemplified by Gerry Corrigan and Alan Greenspan and Ben Bernanke – have really embodied a certain confidence. It’s embodied the nonpolitical nature of the system, it’s embodied a leadership of the system through the country. The Federal Reserve is respected. And it’s respected at a time when respect and trust in all our government institutions is all too rare. It’s that respect and that trust that, at the end of the day, is vital to the acceptance of its independence and to support for its policies. It’s those intangibles – to me more important than any technical analysis or intellectual brilliance – that in times of crisis makes it possible for the Federal Reserve to step in, to act, and to act forcibly in the national interest. Respect and trust – it’s those qualities that need to be maintained, and I’m sure that’s the challenge for the system in the future. Lockhart: We’re pleased to have an eminent economist in his own right, Raghu Rajan, as the moderator of this historic session, so let me turn it over to Raghu.

Panel Discussion: November 6, 2010

401

Raghuram Rajan, moderator: Thank you very much. I think from the last day and a half of conversations, we have had a very rich set of papers and discussions, and I hope this panel is the crowning glory of this wonderful conference. This is a great opportunity to go back in time, to see how three stalwarts of the Fed dealt with many of the big issues that have confronted us over the last twenty-five years. And it would be wonderful if we could get your perspective on what the big challenges were at that moment, what the uncertainties were, how you thought about them, and how we moved forward and the system learned. So let me start by asking each one of you about a particular episode that you confronted. Let me start with Mr. Corrigan. You were there during the great fight against inflation, the “Volcker disinflation,” if you will. What was the thinking going on at that time? Did you anticipate the extent of economic disruption that would take place, and how did you stay the course against what presumably were tremendous political pressures? Gerald Corrigan: First of all, if I could, I want to join others who have congratulated Dennis and his team at the Atlanta Fed for putting together this program and say that for me and many others, just the opportunity to spend a couple of days and a couple of long nights with my former colleagues in the Federal Reserve has been a true joy. So Dennis, thank you very much. I guess I also have to acknowledge that Chairman Volcker always has a way with words, and “respect and trust,” I think, captures what all of this is all about, as the chairman made clear. Now, on the events of the very late ’70s and the early ’80s, let me try to capture as much of it in a couple of minutes as I can. Needless to say, it was discussed at great length yesterday, the inflation period of the ’70s, the emergence of Volcker’s pragmatic monetarism after he became president of the Federal Reserve Bank in New York in 1975 and then was appointed chairman by President Carter in the summer of 1979. Part of the folklore that existed after he was appointed chairman was that Paul went off to the IMF meeting, which I think was in Belgrade, in September of 1979 and that somehow or other he was persuaded at the Belgrade meeting that he had to do something very dramatic. The fact of the matter is, I think he had already made that decision long before he went to Belgrade. And I certainly know that, in the days and weeks leading up to that Saturday FOMC meeting – on, I think it was, the sixth of October if I’m not mistaken – he was laying the groundwork for what emerged at that meeting and thereafter. That Saturday FOMC meeting had something in common

402

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

with the Jekyll Island meeting in 1910, and that is the participants in the meeting were instructed to go to great lengths to arrive in Washington in a surreptitious fashion; no one was allowed to stay at the same hotel so as not to attract attention. I remember very distinctly that the heavy drapes found in the boardroom for that Saturday meeting were completely closed; it was almost pitch dark in the room! And I also remember early on (I was of course seated on the sidelines there), early on that Saturday morning I heard a great deal of commotion out on the Mall across Constitution Avenue from the Fed’s headquarters. And I managed to open the blinds and the drapes a little bit, and it turned out that the pope’s helicopter had just landed on the Mall, and I said to myself, “Well, we got a little ecclesiastical support going here!” But, clearly, that meeting, which lasted, I think, from 8 in the morning till 6:30 at night, really was quite dramatic. Because while Paul had shared with a few of the governors some of the specifics of what he had in mind, it all had to be laid out in great detail at the meeting because he understandably was quite reluctant about circulating papers in advance of the meeting under the circumstances. And it was really quite dramatic, and of course I was a thirty-eight-year-old kid; it was especially dramatic for me because I kept saying to myself, “Oh my God, you know, this is the real thing!” I think it’s probably fair to say that everyone realized the consequences, the short-run consequences, which were going to be quite dire in terms of the economy, unemployment, and all the rest of it. I don’t remember, or maybe I chose not to remember, whether there were actual forecasts of unemployment, things like that presented at the meeting, but everybody knew what was involved here. And, of course, the Volcker “pragmatic monetarism” put great emphasis on the money supply, and there was always this kind of cynical view that that was an excuse to rationalize how and why you could produce these gargantuan increases in interest rates. I don’t think it was all that clear that this was kind of a sham. I think Paul did realize, and I think the committee realized, that there was something to this proposition – that money and inflation kind of went together – and there was that inherent attraction of trying to paint the picture in those terms. But it is also true that, as things unfolded, that focus on the money supply produced its own problems. Because, as some of you will remember, in those days the Fed produced money supply statistics on Thursday afternoons; and it got to be that Thursday afternoons were just utter chaos. And to make matters worse, early on in the process, I’ll never forget this, one of the banks made a big mistake in reporting their statistics that underlie the M1 numbers, and I mean all hell broke loose; I mean there were congressional hearings,

Panel Discussion: November 6, 2010

403

and then there were the other problems with Regulation Q and all the rest of it. But nonetheless, the chairman and the committee, for that extremely difficult period, were literally always of one central mind, and that central mind was we had to get that inflation problem behind us once and for all. You know, there were very difficult moments. Homebuilders were mailing two-by-fours to the Federal Reserve headquarters. There were a whole collection of other events: President Carter had his credit control program, we had the Monetary Control Act, we had the Hunt silver market crisis; it was really an extraordinary period. But it was when I, at least as a young guy, saw the best of the Federal Reserve. And this has already been mentioned several different ways, but the collegiality of the committee in that time frame was off the charts in terms of a willingness to work with each other, listen to each other, respect each other, and stay the course. Of course, I probably don’t want to take other people’s time, but the rest of it kind of is history – there was a gradual, gradual movement away from sole emphasis on money supply to nonborrowed reserves. I think it’s unquestionably true that unemployment rates, for example, ended up higher than I think certainly most of us thought. But at the end of the day, it worked! And I still think, and I’m sure Chairmen Bernanke and Greenspan would agree with this, that it is not an accident that we in the United States, and in other countries as well, since then have had essentially thirty years of something pretty damn close to price stability. And the dividends of that, for our people and for our societies, despite the problems of the last couple of years, I think have far outweighed the short run, and you know, substantial costs that were incurred to get from point A to point B. So I’ll stop there; thank you for your attention. Moderator: Thank you. Turning now to Chairman Greenspan, your – soon after you took over the helm of the Fed – you had a baptism by fire with the Great Crash of 1987. What happened? Alan Greenspan: That’s exactly the question I asked when it happened! You know, Paul was mentioning the fact that he was showing up at the New York Fed in 1949, and I at the time had a best friend by the name of Tilford Gaines, whom I suspect Gerry remembers. And he was an assistant vice president at the New York Fed, and I met Paul at Til’s 1950 Christmas party; and he was taller then than anybody I had remembered, and his stature has remained the same as far as I’m concerned. But, you’re quite right. I showed up on the scene despite actually really quite considerable knowledge about how the Federal Reserve System worked, and obviously I was chairman of the Council of Economic Advisers for two and a half years and had very

404

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

close relationships with Arthur Burns at the time, and I knew everybody on the board, and I knew the people on the staff, so it wasn’t as though I was coming in cold. But when you’re at a new institution, having just gone through a oneperson seminar about that extraordinary staff, and every subject under the sun – 103 percent of which I did not have a clue about – but it was remarkable in the sense that the crash occurred basically as I was getting on a plane for Dallas, out of Washington. And I knew that the market had started down but I didn’t realize the size it was until I arrived at the gate in Dallas and was told that the market was down – it was 502 or something – I said, “Wow . . . it must have rallied all the way back.” And I got a look, and I realized at that point that we had just had the most remarkable collapse in equity values, percentagewise, in the history of this country. I also knew that all of our past histories had indicated that the “wealth effect” – as we have defined it now, or in a more general way, the issue of assets as a very significant determinant in how our economy functions – was critical. And I remember very distinctly when I arrived on the scene saying, “There is no way we can get out of this without a significant economic contraction.” And then, as Gerry will remember, we all got on the phone, and you had a remarkably welcoming remark as I recall, saying, “OK, buddy, it’s all on your shoulders.” I think I forgot to thank you. Thank you! Corrigan: That’s not an exaggeration either! Greenspan: Any event, what . . . the following morning I get a call from the White House operator, who says, “Senator Baker wishes to speak to you.” Howard Baker was then chief of staff for Ronald Reagan. And I pick up the phone, and I said, “Senator,” and then there was a little silence, and then out comes, “Help!” And so I got trundled off to Washington as quickly as I can. But as Gerry remembers, the conversations that we had – remember there were several different meetings we had by phone – it was generally agreed that in that condition, in that event, you just open up the taps because you had no notion of the shock that, what the liquidity was, and as a consequence of that, you had no real judgment as to, with this historic event, what the appropriate actions were, but you knew the direction. And what we did, as you may recall, is we moved the funds rate down, fairly significantly, for a short period of time. And the thing that startled me was I had expected many subsequent disruptions to the financial markets because it’s extremely rare to get a crisis of that type in which a speculative market, whether it’s Hunt silver or whether or not it’s stocks or anything else. You’re looking at a situation in which you get the first shock down, you get tendencies to

Panel Discussion: November 6, 2010

405

recover, but usually there are second and third testings of the market. And we indeed went through that. And it was touch and go with such problems as whether or not the New York Stock Exchange would shut down. I remember having conversations with John Phelan (I think was then chairman of the New York Stock Exchange), and he was telling me, “We’re going to have to shut down in an hour.” And I said, “John, don’t think of shutting down. Think of how you will reopen if you shut down.” And we were very fortunate, because as that hour approached, for some, from where I do not know, but a flood of investment funds moved in, it stabilized it, and we got past that. And it was pretty rough for the week subsequently. Now I haven’t a clue to this day whether it was the injection of massive amounts of liquidity in the short run which stabilized the system, or it was utterly irrelevant. But we had never seen that type of phenomenon before, and we had no way of judging it, and we haven’t had, really – with the exception of the most recent crisis, which is complex by so many other aspects – really essentially, what one makes of that period. But the data show, of course, is that we had a major decline; and one could describe it as a bubble because the run-up in stock prices in the summer of that year was really quite extraordinary. But granted, it was a bubble, it had no footprint on the GDP that I could find even to this day, and we ran into the same thing in the dot-com boom. And the dot-com boom had many of the characteristics of what happened in 1987; and, again, you cannot tell that there was any significant financial problem. Then we ran into 2007 and 2008, and it’s a wholly different ball game. I think that is going to create more PhDs and more work for advisers and a lot of paper from dissertations. And I hope it is to a successful end because to this day, we fully don’t understand what has gone on and how it’s going to come out, how it’s all going to come out. Well, I’ve spoken long enough, Mr. Moderator, Mr. Chairman, whatever you want to be called, Raghu. Moderator: No, Raghu’s fine. Chairman Bernanke, no surprise about what I’m going to ask you about – the panic of 2008 and the Fed’s response. It’s rumored that if you work on a book on the crisis, eventually, it will be called “Before Asia Opens.” You had many tough weekends full of activity. What was the thinking at that time? How do you see the actions with the perspective of a little bit of hindsight? And, you know, while you were doing all this was the political fallout a concern? How did it play out? Ben Bernanke: Well, let me start off by saying . . . thanking President Lockhart and Jekyll Island staff and all for putting this interesting conference on. In the last paper, someone noted the importance of sharp theoretical

406

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

models as a way of improving our thinking about monetary policy. Well, I think there’s a dialectic between theoretical models and history. And in my own academic career, I tried to look at both sides of that, and I found, in my experience of the last few years, that both modern monetary economics of the most, you know, the most sophisticated type, together with, you know, my reading and work on the Depression and other periods, that both of those were extraordinarily helpful in thinking about the crisis. I came into the Federal Reserve, having been appointed by President [George W.] Bush, thinking that crisis was an important part of my job. And, in fact, one of my very first acts was to create a staff committee of senior staff members who were supposed to be looking at financial stability issues and trying to identify problems. Now obviously they didn’t anticipate the problems, and that was no criticism of them; it was a relatively small group of people looking at a restricted set of issues. What we found out, of course, was that the crisis that began in 2007 and hit its peak in the fall of 2008 was an extraordinarily complicated phenomenon. It wasn’t simple bank runs or a stock market crash; it was an interaction of so many different, complex, and global aspects of the financial system. I recently testified before the Financial Crisis Inquiry Commission, which is charged with explaining the crisis, and I wish them great luck. But my testimony, which is available, sort of summarizes what I think are some of the key features, including leverage, which you talked about in the last session, problems with risk management, gaps in the regulatory system, and the shadow banking system, and many other complex and interacting features that led to the crisis. But I think one example which I found very instructive relates to work by Gary Gorton, who looked at the interesting analogies between the shadow banking system and the classic, or the traditional, banking system. Gary pointed out that the traditional commercial banking system didn’t have enough insured or safe liabilities to satisfy the huge demand for liquid assets, this was a point also that Marvin Goodfriend was making earlier. And so the shadow banking system arose, and what it did was essentially it created, on the one hand, all of these so-called AAA assets. And Anil’s [Kashyap] picture of how magic of the alchemy of financial engineering transformed sort of mediocre underlying assets into AAA, that the shadow banking system held those as assets and then created short-term, liquid liabilities with which to finance those assets. And so it was very much the same structure as a bank except it didn’t have the protections, the guarantees, and the oversights that a bank has. As Gary explains it, in some nice papers (and others have worked on this as well), the essence of a banking system when you have liquid assets is that you believe that the assets on the other

Panel Discussion: November 6, 2010

407

side are essentially perfectly safe, or guaranteed in some way, and once that belief is shattered then your best response is to pull out – is to run – is to pull out your short-term liquidity as quickly as possible. That’s essentially what happened in the fall of 2007 when a couple of hedge funds said they couldn’t value some underlying collateral and problems with subprime mortgages were becoming evident and the decline in house prices was putting pressure on those mortgages. It became increasingly evident that the so-called AAA securities were not AAA, that there were going to be losses, and that, of course, as in any classic bank run situation, generated a broad panic in the shadow banking system and in the money markets. So it was interesting in that this was, in some ways, an absolutely clear, predictable, and understandable phenomenon in the sense that it looked very much like a banking crisis of the 1830s. And yet because it was in such a complex and novel institutional context, one that was, had outgrown the underlying regulatory structure, our ability to detect it and to anticipate it and protect against it – well, perhaps someone more omniscient than me could have done it – but it was difficult for us to do given the . . . particularly given the sort of silo aspects of our regulatory system, where individual regulators had different responsibilities. So the crisis, as you know, began in ’07 and intensified with the Bear Stearns phenomenon, which, again, was very much tied into a run, in that case in the repo market. After Bear Stearns there was a period of some calm, but then we saw Fannie and Freddie and, of course, things picked up with Lehman Brothers and AIG, etc., etc., in the fall. Now people who have written about the crisis have focused very much on the individual institutions and the attempts of the Federal Reserve and the Treasury and others to protect or rescue them. Let me just say, first of all, that I, you know, based on historical experience and also just knowledge of markets and theory, I was persuaded that, given where we were, and given the global nature of the crisis, that we had to do all we could to avoid the collapse of these major global financial institutions. Not again, of course, as you all understand, not because of any desire to protect the predators or equity holders of those firms, but because, as we found out, with Lehman Brothers, that the collapse of one of these firms could have enormous effects on the broad financial system. Now the reason that those, in some sense, that those episodes were so dramatic was in a sense that we as a country were completely unprepared to deal with that kind of phenomenon. Given our congressional-presidential system (unlike, not a parliamentary system where the government can act relatively quickly), we didn’t have mechanisms for quick response from the government. We didn’t have tools to safely resolve firms that were not banks;

408

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

you know we have (the FDIC has) a set of tools for resolving banks. We didn’t have any analogous tools to resolve a large insurance company or a large investment bank. And so a lot of that, a lot of those episodes were ad hoc; we were using the tools that we could put together, cobble together. I think if anything good has come out of this, I hope it is that we will in the future both create a much stronger ex ante supervisory regime over systemically critical firms, which is what the Dodd-Frank bill attempts to do, so that it’s much less likely that they will come to this kind of brink situation and that, should that happen, that we have a resolution regime and a set of well-established principles for resolving these firms in a way that does minimal damage to the broader system. So that part of the whole episode, the response to individual firms, which was the thing that cost us the most sleepless nights, was really an attempt to deal with what was really a very inadequate institutional structure and political structure for dealing with this particular kind of problem. Now beyond those, though, I think beyond these dramatic events, I think that history and theory provided some very useful tools for addressing what happened in the crisis. And I’ll just talk briefly about two broad sets. First of all, you know, the Federal Reserve was set up to ensure financial stability. We know from Bagehot, we know from Thornton, how central banks are supposed to respond in a period of panic. In particular, we are supposed to lend freely against good collateral, to provide liquidity; and that’s precisely what we did. I mean I thought of this whole thing as being that Bagehot would have been happy with what we did. We set up all kinds of new mechanisms, recognizing that commercial banks were not the only run-prone institution in the current environment; we had to worry about money market mutual funds, and the repo market, and many others. And we were creative, and we worked with the Treasury and did what we could to set up provision of liquidity to those troubled firms and markets. And I think ultimately that proved successful. We ended essentially all our major interventions, our major liquidity programs, at the beginning of this year. So that part, you know, I really think of it as the classic central bank lenderof-last-resort response, again, adapted for the complexity of the financial system. I would add, in addition to that, that we also made good use of our supervisory function throughout the entire episode, as was the case in the crises faced by my colleagues here. An example I would give would be the stress tests that we led in the spring of ’09 that were very successful in providing information to investors and to the public about the state of the banks and which was essential for them to be able to raise capital and strengthen themselves in the subsequent months.

Panel Discussion: November 6, 2010

409

So that’s the liquidity/financial stability half; the other part of policy was monetary policy. And there again, I think that our approach was, you know, nothing unusual about it in the sense that we were trying to respond to a collapsing demand and a contracting economy; we cut rates very quickly. Of course, we ran into zero bound, which we had had some brush, some discussion with, in 2003. We ran into the zero bound quite literally in December of 2008, when we brought the federal funds rate down to the range of zero to twenty-five basis points, and from there on, we had to, you know, change our methods of operation. Again, Marvin talked about this very well. I think the point I’d like to leave with you is that there’s a sense out there that “quantitative easing” or asset purchases or some completely foreign new and strange kind of thing, and we have no idea what the hell’s going to happen, and you know, just, it’s just an unanticipated, unpredictable policy. Quite the contrary – this is just monetary policy. Monetary policy involves the swapping of assets – essentially, acquisition of Treasuries and swapping those for other kinds of assets. And similarly, the way federal funds rate management essentially is aimed at reducing medium- and longer-term interest rates and raising asset prices, quantitative easing has exactly the same objectives, and so it’ll work or not work in much the same way that monetary policy – ordinary, or conventional, familiar monetary policy – works. So there is not really, in my mind, as much of a discontinuity as people think. Monetary policy is monetary policy; it’s just that the specific tools that are used to create easier, more supportive financial conditions are somewhat different. But, you know, we took action in this line, as you know, earlier this week. The motivation is precisely the one that we all have learned is the appropriate motivation, which is that we have inflation; we’ve seen disinflation; we see inflation that is low, below the target, or below the preferred range of most of the members of the FOMC. And we see an economy which has a very high level of underutilization of resources and a relatively slow growth rate. So, you know, the standard considerations suggest that we should be using expansionary monetary policy, and that was the purpose of the action. Again, nothing extraordinary, just a different set of tools to achieve essentially the same kind of results. So I’ll stop there. Moderator: Thank you. Now let’s get into a little bit of conversation. President Corrigan, you have a note up there which suggests that you worry a little bit about the inflationary consequences of, I would guess, more stimulus. Let me let you explain what your concerns are at this point.

410

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

Corrigan: First of all, that note that’s out there was stated a little more precisely than you just referred to it. First of all, let me say at the outset that, as you would expect, I have asked myself hundreds of times over, what would have I done had I been in the situation that Chairman Bernanke and his colleagues have been in the past couple of years? And this is not flattery, Ben – I think I have said this to you privately – I think the best answer I could give is I would have done pretty much exactly what Chairman Bernanke and his team did except I might not have been clever enough to think of some of the things that they thought of. I just want to make sure that’s very much a part of the record. And, again, that is what central banking is all about, so there’s no great mystery to this. Now the issue that is referred to in the note that was circulated at this meeting (that is, I’m so compulsive that if I have something like this I feel like I have to write something), it is simply the following. And I went out of my way to describe this as a very low-probability risk, and I went out of my way to say (as many of the people in this room know) I kind of have a fetish about low-probability risks, because those, when something goes wrong, it can really go wrong. So let’s keep this all in perspective. The point that I raised – which I’m sure Chairman Bernanke and the FOMC itself has thought about many times – is that if you seek to nudge up the inflation rate, even with these very, very, very low rates of capacity utilization, labor market conditions, and all the rest of it, is there a risk that nudging it up to, pick a number, 2 percent or whatever you want to use, may turn out to be easier than capping it at 2 percent or whatever. That’s essentially the source of – and I use the word very carefully – uneasiness that I wanted to register. That is in no way intended to be taking exception with the course that the FOMC has set out for itself as was announced earlier this week, and it was just summarized by Chairman Bernanke this morning. But I think all of us, when we’re on the Open Market Committee or when we’re not, should think about contingencies, and that was the spirit in which I raised that issue. Bernanke: Could I comment? I’m very sympathetic, Gerry, to what you’re saying. Of course, there are contingencies in different directions, and there are both upside and downside risks. We are very committed on the FOMC to our price stability objective. I have rejected any notion that we are going to try to raise inflation to a supernormal level in order to have effects on the economy. I think that our credibility must be maintained; I think it’s critical for us to maintain inflation at an appropriate level, and we’ve provided through our projections a pretty clear sense of what the committee thinks is the appropriate long-run level for inflation.

Panel Discussion: November 6, 2010

411

Now that being said, we’ve had, at least according to many measures, we’ve had a fairly significant disinflation since the peak of the business cycle in December of 2007. That’s at least indicative of a world in which monetary policy is insufficiently stimulative. We can all debate about how much the monetary policy can do for the real side of the economy – personally, I think it can be helpful – but nominal quantities are the Federal Reserve’s responsibility. And if inflation is declining, and continuing to decline, I think at a minimum we should not be satisfied with a situation where we have both a large amount of slack on the employment side and inflation which is below our generally agreed-on preferred level and seems to be declining over time. So, in that respect, I think that’s a signal that more should be done, as I said in my speech in Boston. That was the motivation for the action taken earlier this week. But I couldn’t agree with you more. Again, we’re not in the business of trying to create inflation; our purpose is to try to provide some additional stimulus to help the economy recover and to avoid potentially additional disinflation, which I think we all agree would also be a worrisome outcome. But we are equally committed to both sides of our mandate. Moderator: Chairman Greenspan, let me get back to you on an issue which is somewhat closely linked with your tenure, which is asset prices, asset price bubbles, and the notion that there was a “Greenspan put” out there in the markets. What’s your sense of that? Was it fair or unfair? And how did you think about that episode? Greenspan: Well, first of all, it’s a slightly ambiguous concept in the sense that I get a lot of different definitions of what that means, so I will interpret it by the definition I would like to respond to. The issue is basically that the Federal Reserve tended to react far more significantly to weakness in the economy than the upside – that there was an asymmetry of response to the economic events. And if you look at the data, there is no question that monetary policy eases far more rapidly than it tightens. But that has got nothing to do with the issue of asset prices or the issue of setting up a put; that’s what the economy is doing. Because the National Bureau, for example, say, over the past fifty years, indicates that our economy was in recession 14 percent of the time. And if you look at the unemployment rate, you will find that it rises very sharply, and then eases down very generally. So what is being observed and what is being interpreted as the Greenspan put – and I appreciate the notion that this is a great invention of mine – is essentially the way one would expect a central bank to respond to the economy.

412

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

Now I don’t deny that because we endeavored increasingly over time to try to be anticipatory and preemptive as best we could on the basis of the knowledge we had – and obviously if you begin to see very major changes in equity values in the economy, and you know what the wealth effect concepts are, and you know the impact of changes in equity prices on, for example, the market value of financial intermediaries. Because clearly, the extent to which you have problems in a financial intermediary to a large extent depends on the market value of the equity of the intermediary because it is that market value which is the support for the liabilities of the institution. And, essentially, a big infusion of unrealized capital gains in a financial institution enables it to very substantially improve on its ability to lend, and, indeed, the credit rating agencies, hopefully better than they’ve been done, obviously rise for the senior debt of a financial institution if the market value of equity expands. The converse of that, when you’re seeing a sharp decline, is that we know that there are certain effects that are going to occur in the economy, and we tend to respond to them. But we’re responding to the economy, not to the markets, not to interest rates; we’re responding essentially to what our job is – namely, to stabilize the system. Now if, in effect, the Greenspan put is a notion which says you’re stabilizing the system, I say, “Well, I hope so.” That’s indeed what we’re here for, to celebrate the origination of the Federal Reserve System, which was created largely because of the extraordinary panic in 1907, which cumulated a long series of such panics. And monetary policy, if it’s a functioning issue at all, has got to largely be an issue of trying to stabilize the economy, so that I don’t really have an understanding of why that has become a pejorative term. I mean I listen to it, I know what they’re saying, and I’m saying, if I understand you correctly, what we’re doing is what we should be doing! And in that sense, it’s an issue of success, not failure. Moderator: Let me just push a little harder; maybe I’ll ask President Corrigan on this. Is there a role for the Fed to maybe lean a little bit against the tide, as it’s taking prices up, so that you get more . . . you prevent the upturn which has to be in a sense . . . ? Greenspan: Well, if I may say, I don’t think that’s what happens. In fact, going back to, say, 1994, what we observed – and Marvin Goodfriend’s sort of piece had many quotes directly out of the FOMC hearings – at that time, we were talking about bubbles in the financial markets and in the stock market, and our general view was basically that there was a speculative froth going on, and we indeed were tightening. And that three hundred basis

Panel Discussion: November 6, 2010

413

point rise in 1994 was not directly related to stock market speculation, but we were aware that there was an incipient bubble in the market emerging, and it appears directly in the FOMC discussions at the time. What happened, in fact, is that we raised the funds rate that is pushing against the issue by three hundred basis points and did stop the stock market cold for a year. As soon as we stopped raising rates, the market took off again. And we went through the same process later in the decade. I concluded from that that while we may have thought that we were trying to lean against the wind and defuse it, we in fact did precisely the opposite. Because what the market (this is now an obvious interpretation) was probably saying to itself, “Wow, the stock prices went through a three hundred basis point crunch, and nothing happened. Therefore, we have underestimated the equilibrium level of the Dow Jones Industrial Average.” And it took off from there. So this presumption that you can incrementally defuse these bubbles is working very fine in an econometric model in which we define the structure of the model in such a way to get that conclusion. All it tells me is that we’re missing a variable in our models which doesn’t capture this feedback effect, which I suspect is very real, and, in a sense, this notion that you can incrementally impact on the economy or on the markets is really something which exists only within our models. It doesn’t exist in the real world, and there’s another reason incidentally. You know, if monetary policy has long and variable lags, how do you get the feedback effect in time to know whether or not the incremental actions you’re taking are actually working? I mean, if we’re sitting there, and we’re saying, “Well, we’re going to increment the funds rate,” the problem is that the delayed effect is so long and so variable, we have no way of knowing how to increment. So this notion that is fairly general I find very puzzling, because I don’t think the evidence at all supports it. Corrigan: Let me try to quickly put this issue in perspective. And part of the perspective is the following. If we look at the thirty years between 1980 and today, by my count over that thirty-year period we have had eight financial disturbances, all of which had at least some potential elements of systemic risk associated with them. Now in the vast majority of those disturbances, including the most recent one, the overwhelming majority of losses and write-downs were directly or indirectly attributed to the credit origination process, although in this last sequence, as Ben has already mentioned, the securitization process and all of the excesses that went with that aggravated that problem. But the underlying problem was still largely on the credit origination side. I think you can make a decent argument that of those eight

414

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

episodes, the two that don’t seem to fit the credit-dominated phenomenon were the two that Chairman Greenspan mentioned – in other words, the ’87 stock market crash and the dot-com thing. Because those were the two in which it’s hard to see, as the chairman suggested, convincing evidence of any significant damage to the real economy. Now, that pattern – I’m sure there are scholars in this room that have looked at this in a more systematic way than I have – but that pattern raises a variety of questions in my mind in terms of “Well, is there something inherently unstable about the system?” And you can’t casually rule that out. Or, in the alternative, have there been at least limited cases in which the extent to which bubbles were financed through credit expansion contributed to these sources of instability? Now I don’t pretend to have all the answers to those questions, but I do think how you think about those questions and their answers bears enormously on this whole question of financial reform. In other words, if we just all put ourselves in a time capsule and said, “Well, we’ll wake up on,” today is what, “November 6th, 2015.” And will we find on November 6, 2015, that we’ve inherited a financial system that is inherently more stable, less accident prone, and even when accidents do happen, they don’t cause as much damage as we’ve just witnessed? That question – of what the system is going to look like in the future – I think personally the probabilities favor a more stable system, but I don’t think it’s baked in the cake. Greenspan: Gerry’s raising a very important point here. I think you have to ask yourself whether credit restraint on lending or capital is what you want to do. And I ask myself, what, in retrospect, would have been required to prevent the crisis of the post-Lehman Brothers default crisis? And the obvious question is that Lehman Brothers, I think, went to the wall with 3 percent tangible capital. What if it had had 20 percent capital? Or, 50 percent capital, like they used to have prior to the Civil War? The answer essentially has got to be at some point if there is adequate capital – and the word “adequate” is a crucial word here – by definition, you do not have defaults of senior debt. And you do not have contagion, with the exclusion of the wealth effect, in a system in which there are no defaults. And as far as I’m concerned, much of what we have run into here is a misjudgment as to what the tail risk of the probability distributions were (and, in effect, as I recall – I’m not sure, Ben has been far more involved in this than I, there has been a very interesting discussion going on prior to the Lehman event, or even prior to the Bear Stearns event) that the tail risk was not a normal distribution – that it was a very nonlinear distribution which had a fat tail but which we had never actually observed because we never got

Panel Discussion: November 6, 2010

415

to the point that we got to in 2008. And, in the event, if you look through the actual tracing and make it retrospectively, putting it in where the model risk distributions would appear, that the tail was not fat – it was obese. And the problem essentially is that we had an extraordinary implosion as a consequence of that. And I think the lessons that at least I’ve learned – I don’t know to what extent we’re going to get it in the regulatory system – but I think that the financial intermediary system in the United States, and I suspect elsewhere as well, was implicitly subsidized by the taxpayers. I know in our case because we had less capital through that period than in retrospect we should have. And the question is that therefore the financial system was subsidized by the taxpayer. The bill didn’t come due until after September 15, 2008; that’s just because the billing relationships took a long time. But it’s all capital that I see is the problem, and I’m not saying there are lots . . . I think there was rampant fraud in a lot of what was going on in these markets. But my general judgment is that there are two fundamental reforms that we need; one is to get adequate capital, and two, to get far higher levels of enforcement of fraud statutes. Existing ones – I’m not even talking about new ones. Things were being done which were certainly illegal and clearly criminal in certain cases, which in fraud is a fact. Fraud creates very considerable instability in competitive markets. If you cannot trust your counterparties, it won’t work, and indeed we saw that it didn’t. Bernanke: Just one comment that all of these points lead to, which is that there is a difference between an equity bubble and a credit bubble. An equity bubble, particularly if it’s an unleveraged bubble, is a loss of wealth, and unless it’s a huge bubble, like perhaps the Japanese case, it’s going to be absorbable by the system, as we saw in ’87 and a relatively mild recession in 2001. If it’s a credit bubble, which is highly leveraged and leads to loss of capital and loss of capacity to intermediate new credit, then the damage is much, much greater. Now what that suggests is that all of this talk about the Greenspan put or the Greenspan-Bernanke put or whatever, which thinks implicitly about the stock market, is more or less beside the point. I mean the real issue is what we can do to keep our financial intermediation system strong. And the position I’ve maintained throughout is that the first line of defense, at least, should be capital, supervision, liquidity, and other restrictions to make the intermediation system as strong as possible, so I agree with Chairman Greenspan about that. Corrigan: If I could, Ben and Alan, just add a footnote to that, it seems to me unambiguous that one of the lessons learned here is that going forward,

416

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

particularly for the so-called systemically important institutions, that we really have to look at capital adequacy and liquidity adequacy as a single discipline. Because it’s very clear that if you focus on just capital without a systematic integration of how you think about liquidity, the probability of train wrecks of a major proportion go up. So I think that the work, Ben, that the Basel Committee and the Fed is doing to pull together this single integrated approach to capital and liquidity adequacy is probably one of the two or three single most important things I think we have to do. Moderator: We don’t have a huge amount of time, so I want to make sure we touch on all the issues that were on the table in this conference. One of the issues that Marvin Goodfriend brought up was the difficulty for the Fed, in occasionally being the only game in town, having the flexibility, having the capacity to undertake certain actions which should in theory be the realm of other agencies of the government. The example he brought up was the Mexican loan. One could think of, for example, today, where a number of economists have argued that fiscal policy might be more effective at this point than monetary policy, but monetary policy might be the only game in town because of the state of play in Washington. What’s your sense about how much . . . how do you choose – difficult choice – between being a team player, stepping up to the plate but at the same time risking, sort of being part of the political process and perhaps losing a certain amount of independence? I think Marvin had a case study there where he described some of the content of the discussion. Has that thinking . . . how has that thinking evolved over time? Maybe first with Chairman Greenspan . . . Greenspan: Well, remember that the Federal Reserve is the fiscal agent for the United States Treasury, that there is really only one monetary authority in the United States. We, unfortunately, in our bookkeeping back in 1968 when we decided to create the unified budget, left the Fed outside of that, and it’s caused all sorts of distortions, and it’s created an illusion that the sovereign credits that are employed by the Federal Reserve, the central bank, are different from those of the Treasury. They are not. There is only one sovereign credit that’s involved in this whole process. And so I think that you have no choice but to coordinate with Treasury, and the question there raises a major problem of Federal Reserve independence. And that is one of the most difficult problems I think that the institution has – to fulfill its role as a partner in the maintenance and care of the sovereign credit of the country and the independence of the central bank from government. Those are not obviously reconcilable views, and there is a considerable tension

Panel Discussion: November 6, 2010

417

that always exists; it always will exist, and I don’t know how you make any difficulty there. I, for example, think, I’ve argued that 13(3) – everyone knows what 13(3) is by now, I hope – had to be used when we had what I think was the greatest global financial crisis ever. Not the greatest economic crisis – that was the Great Depression. But I don’t recall the history indicating the extent to which the shutdown occurred in financial markets, in the commercial paper market, in the RP market, in money market mutual funds, and trade credits. Those things never shut down previously, and the call money market, for example, was open during the Great Depression; the rates went to 20 percent, but it never shut down. It is true that there was a major contraction of credit availability because of failed banks in the ’30s, but the institutions stayed open. In 1907, there was one day in which the call money market shut down, and I think the bid rate went up to, I think, 168 percent or something like that. But I don’t recall, and I’ve reached back into history as best I can, but I’ve never yet found an instance of anything remotely close to what the Federal Reserve had to deal with here. And, I think that 13(3) is clearly useful, and indeed it was the thing that we actually amended as late as 1994. I mean, it wasn’t as though this is some obscure depression vehicle which we never thought of and we pulled out. The interesting question here, however, is that if the Fed is going to put things on its balance sheet, I think that it’s incumbent upon the United States Treasury Department to take the Fed out. Those are not monetary assets; 13(3) is, can be, and has turned out to be a nonmonetary vehicle, and it’s properly used as essentially a fiscal agent. But the Fed should not be asked to keep that stuff on its balance sheet, and in so doing, you create really serious political effects, which I think are self-evident at this stage and I think are unnecessary. The reason, of course, goes all the way back to the 1960 Unified Budget Procedure. For the Treasury to do that, they would have had to have gotten appropriated funds, where the same credits are employed by the Federal Reserve without such things. Bernanke: I think there’s a lot of intellectual confusion about what Federal Reserve independence means and what it’s for and also about cooperation and what it’s for. The clearest case for independence is in making monetary policy because we want to avoid short-term political interference in monetary policy decisions, and there’s a good case for why a longer-term perspective makes sense. And throughout the crisis, the Federal Reserve has been completely and entirely independent in all of its monetary policy decisions; there’s been no issue there whatsoever. A second area is supervision.

418

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

There are good reasons for supervisors to be independent – so that they can carry out closing a bank, for example, without interference from Congress. But that being said, the independence that a supervisor has is different from the kind that a monetary policymaker has, and when the Fed is operating in its role as a supervisor, its relationship to Congress is somewhat different. For example, we are carrying out the statutory laws and the framework that the Congress has put down. Now, in the case of a financial crisis, obviously we don’t want to compromise our independence, but I think there is both plenty of precedent and plenty of clarity that cooperation, in trying to arrest the financial crisis, is perfectly appropriate. And, in fact, you can see that in things like the systemic risk exception in the FDICIA law, which requires a collaborative decision by the Fed, the Treasury, etc., and the FDIC, that a systemic crisis exists. The Financial Stability Oversight Council it just passed is also a collaborative institution. So a collaboration, in itself, to arrest a financial crisis I think is not in any way inconsistent with monetary policy independence, which is a different function. Now that said, in carrying out financial stability operations, there are different roles. And the Federal Reserve made every effort, and I think we were pretty successful, in restricting our role throughout the process to liquidity provision. That is, we were required by law to lend against good collateral, or technically, to be secure to the satisfaction of the president of the Reserve Bank making the loan, and we’ve had no losses, and we’ve been repaid, and most of our programs are being shut down. So I would argue that, you know, as best as we could, and I think pretty effectively, we collaborated with the Treasury on managing the crisis, but we still tried to respect each other’s roles, where the Treasury injected capital and took losses and the Federal Reserve, wherever possible, lent against good collateral. The reason that we couldn’t save Lehman, which was before the Treasury was authorized to put capital into companies, was that, given that the company was insolvent, there was no way that the Fed, using its strictly liquidityproviding authorities, could avoid that. Later, of course, that would have been a different story. So even as we collaborated we tried to make sure as best we could that we were each using our own separate authorities, and there was an agreement between us and the Treasury, which tried to lay out – it was sort of a new accord, I guess you might call it – but at least tried to lay out clearly that credit risk is the Treasury’s responsibility; liquidity provision is the Federal Reserve’s responsibility. Corrigan: I guess the only thing I would add is that, human nature being human nature, I think it’s inevitable – notwithstanding the obvious accuracy

Panel Discussion: November 6, 2010

419

of what Ben has just said about cooperation with the Treasury and so on, that’s clear enough – but I do think again human nature being human nature that there is some risk coming out of this last episode (not to say things about earlier episodes) that the public and politicians may find it all too convenient to really try to convince themselves that the Fed or monetary policy or both, you know, can solve all of our economic and financial ills. I think there has to be some danger at the margin that the great successes of the past couple of years – and indeed, I would go back to the successes of the Volcker initiatives – create that perhaps a bit of a false sense of comfort. I do think we have to be careful about that, and I do think that regardless of whether you use the word independence or autonomy, or however you want to dress up the semantics of it, that that concern, I think, inevitably does raise questions about the standing and status of the central bank of the United States over the longer term. I think it can be managed, but to think that those issues won’t surface, or haven’t surfaced, I think is perhaps being a bit naive. Moderator: Well, we have about five minutes, so I’d be remiss if I didn’t ask a slightly tougher question. Allan Meltzer yesterday said that the Fed “displayed persistent interest in near-term events while persistently ignoring longer-term consequences of its behavior.” And, specifically, he suggested the attention paid to deflation in 2002–2003 was an example of this, and I would guess, by extension – and I’m putting words in his mouth – he would argue that the deflation scare of 2010 might fall in the same bucket. Is he being unfair? Greenspan: Are you asking me? I think, as I recall, Allan is a very great supporter of Milton Friedman. And let me just argue, for example, with respect to the 2002–2003 experience, that I wrote a rather extensive analysis of that period for a Brookings paper, which published in June of this year, in which I think the evidence that the “bubble” was a consequence of the endeavor of the Fed to move rates down when we were confronted with a fall in inflation rate in that period. My judgment is that at that point, and I think Ben Bernanke remembers it, I’m sure, because we both sat and worried about it, but it was clear that the probability of getting deflation at that time was less than fifty-fifty. But had it occurred, the impact would have been much too difficult to deal with. And the result is that we brought rates down. And what were the consequences? Well, first of all, all of the evidence that I can see says that housing prices are a function of long-term mortgage interest rates with a lag of a significant number of months. It worked. There was a very important relationship between the federal funds rate and long-term

420

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

rates, and therefore housing prices in earlier years. But with the emergence of a globally arbitraged bond market, the ten-year note in the United States became a function of what was going on in the global markets and essentially detached from the federal funds rate. Prior to then, that correlation between the federal funds rate and the long-term rates – for example, the ten-year Treasury note – was very high, and when you moved the federal funds rate, you would get a movement in the ten-year note, effectively saying that the term structure was stable. And the Federal Reserve staff was very much aware of the fact that we were moving incremental estimates of the term structure. But the notion that it would essentially collapse, as it did in the period following 2001, when we got this tremendous flood, as Ben pointed out, of savings moving into the global economies, the consequence of, in my judgment, the end of the Cold War, which led to a major shift toward market economies in much of the developing world, which in turn created a huge level of income, which they couldn’t spend because they didn’t have an infrastructure or a culture to do that, and that spewed over into the markets, and we had a major rise in ex ante savings relative to ex ante investment in the United States and globally. And the consequence of that is that, if you look at the data, subsequent to 2002 we were getting a M2 . . . however, I think money supply by any measure was very stable during that period. I just find it very nonconceivable that the overnight rate in that sort of environment can capitalize twenty-year to thirty-year assets. It’s got to have . . . the capitalization rate has got to be fixed at the same maturity that the life of the asset is. Now the notion of this pumping of liquidity is a very fine analogy, but the channels are never defined sufficiently. You can’t just jump and say, “It’s therefore . . . ” – and I find the word “therefore” is utterly inappropriate – and if you look at the correlations (and I suggest to those of you who are interested in refuting my argument, please go to the June publication at Brookings, it’s the Brookings Panel on Economic Activity, and if the argument is wrong, I’d very much like to hear it because I’ll change my point of view). But let me just say this: With respect to the period, Milton Friedman said, this is in January 2006, “There is no other period of comparable length in which the Federal Reserve System has performed so well. It is more than a difference of degree; it approaches a difference of kind.” Now, being that Milton Friedman was the most severe critic of the Federal Reserve up to that point, I do think that sometimes his viewpoints ought to be considered in this particular case, and I’m surprised that – I don’t know whether or not you’re quoting Allan correctly, frankly – but if he said that, and I don’t like to quote people, especially when they’re

Panel Discussion: November 6, 2010

421

here – but all I can say is that if that’s the case, Allan, both Milton and I disagree with you. Bernanke: I have great respect for Allan and particularly his magisterial work on the history of the Federal Reserve, but Chairman Greenspan reminded me of something I think will be provocative, because Allan Meltzer wrote an editorial recently, an op-ed, saying that Milton Friedman wouldn’t approve of the Federal Reserve’s behavior. I grasp the mantle of Milton Friedman. I think we are doing everything Milton Friedman would have us do. Besides the stability of the period that Chairman Greenspan just referred to, I think what Milton Friedman would say is that the Federal Reserve is responsible for the stability of nominal aggregates, including prices. And that means that, in particular with respect to inflation, that you don’t want inflation to be too high, but you also don’t want it to be too low. And as someone mentioned earlier, based on what we knew, given the data in 2003, the downside risk was greater than the upside risk. And we adjusted appropriately, and we did so in a way that did not lead to long-term inflation, and achieved quite a bit of stability of the monetary aggregates and of inflation. Likewise, in the more recent period, our objective has been, among other things, to try to keep inflation neither too high nor too low. If you look at what’s happened to key nominal aggregates, like nominal GDP growth or broad money aggregates, they’re all saying that we need to do more. Because nominal GDP has been growing very, very slowly for the last two or three years. Broad money aggregate’s been growing very, very slowly. So, even from a strictly monetarist perspective, we need to do more. So, I just disagree. We have . . . the issue in the long term for the central bank, of course, is inflation. There are many other issues that the fiscal authorities and so on have to deal with, but as I said at the very beginning, we are committed to both parts of our mandate. We are confident we have the tools to manage our balance sheet in a way that will allow us to achieve price stability in the longer term. But what we are trying to achieve now, which is entirely consistent with monetary theory and practice, is stability of inflation and stability of monetary policy within an extraordinary circumstance, an extraordinary context. So I think we’re considering both the long term and the short term. Moderator: Mr. Corrigan, you have the last word on the panel. Corrigan: My last word is simply that note that I distributed here. I talk a lot about the culture of the Federal Reserve. And I do think that, especially in these trying times, we should recognize, even for our critical friends in the academic community, that the culture of the Federal Reserve is strong,

422

Ben S. Bernanke, E. Gerald Corrigan, and Alan Greenspan

it’s been that way for a long time, and I think it will continue to be that way at the end of the day. The maintenance of that culture, including the feature of collegiality that I’ve spoken about, that I think at the end of the day is the name of the game, and to the extent that culture is preserved, I think that will just reinforce the capacity of Ben and his colleagues to cope with this incredible agenda and challenges that they’re facing right now, and the same will apply to his successors down the road. That’s my parting thought. Moderator: That’s fantastic. Greenspan: All I can say is, “Hear! Hear!” Moderator: Thank you very much; this has been a fascinating panel. Thanks again to the Federal Reserve Bank of Atlanta for getting us here.

Index

acceptance bankers, 59, 67, 69, 85–86, 89, 107, 113 defined, 118 European money markets and, 114 trade, 118 types of European, 67 Act of November 7, 1918, 43 advances, 119 Aldrich Plan, 1, 2, 107 Aldrich, Nelson, 1, 59, 60, 66 Aldrich-Vreeland Act of 1908, 1, 59, 66, 75, 173 Alexanders Cunliffes & Company (bill brokers), 126, 138 allocations, efficient, 277–281 American Bankers Association’s Baltimore Plan of 1894, 66 American Recovery and Reinvestment Act of 2009, 227 Andrew, A. Piatt, 2 arbitrage, regulatory, 7, 11, 34, 45 assets, 218, 224 equity, and 2007 housing collapse, 226 fundamental, 375 purchase of, 2008, 226 swapping (monetary policy), 409 typical bank, 169 valuing, 46 Babcock, Joseph W., 18 Bagehot, Walter, 36, 68, 70, 127, 218, 221, 408 Baker, Howard, 404 balances bankers, 32, 54, 56 correspondent, 13, 65, 66

deposit, 65 real currency, 361 reserve, 72 seasonally-fluctuating, 32 Bank Acts 1933 (Emergency Banking Act), 182 1935, 183 Bank Charter Act of 1844, 114 Bank of England, 4, 67, 114 monitoring during Overend, Gurney Panic of 1866, 153 acceptances, 143, 149 discounting, 124, 140: splits, 143 last resort lending, 151 supervision, 151 Bank of the United States, 218, 221 First, 62 Second, 62 Bank of United States, 80, 193 banking branch, 7, 11, 32, 43, 54, 56, 58, 101, 168 correspondent, 13, 32, 62, 64–65 dual chartering, 101 era of free (1836-63), 62 shadow, 4, 14, 90, 113, 115, 149, 152, 406 unit, 61, 101, 169 wildcat, 62 Banking Acts 1933 (Emergency Banking Act), 87, 89, 91 1935, 87, 89, 92, 199 banking, free, key characteristics of, 8 banknotes Civil War tax on state, 33 deposit banking and, 12 inelasticity of, 14, 54, 56

423

424 banknotes (cont.) insured, 12 national, 63, 66 pre–Civil War role of, 12 banks categories of, 202 central, 54, 56, 60, 69 absence of a U.S., 14, 33, 45 and Federal Reserve Act of 1913, 33 as lenders of last resort, 59, 60, 114 discount market and, 70, 88, 101 European, 67, 70, 101 gold reserve requirements and, 191 in the late 1700s, 400 monetary policy and, 342 qualitative monetary policy and, 363 rediscounting and, 88, 113 responsive role in a panic, 408 closures, summary of, 1921–29, 44 country national, 64 failures of 1890s, 26 1921–29, 41 1929–33, 34, 182 1930s, 81, 182 costs, 29–32, 43–45 local, 218, 221 prior to 1933, 195 national, 64 regulation of double liability and, 24 under the Federal Reserve Act, 91 under the National Banking Era, 3 reserve, 69, 71, 218, 223 single-office, weakness of, 12 state-chartered, 11–13, 72 Banque de France, 67 Barclay & Company (commercial bank), 138, 147 Barings, 147 Barnett, George E., 31 Barro, Robert J., 235, 250 Barro-Wallace Irrelevance Proposition, 234–235, 236, 237, 239, 259, 262, 270, 277, 285, 291, 297 defined, 235 Basile, Peter F., 198 Beebe, Jack, 357

Index Bentsen, Lloyd, 345 Bernanke, Ben, 6, 218, 224, 229, 270, 399, 405–409, 410–411, 415, 417–418, 421 Bernstein, Asaf, 205 Biddle, Nicholas, 62 Bignon, Vincent, 68, 149 bills Connie Mack, 351 finance or accommodation, 128 Neal, 350, 351 real, doctrine of, 103, 107, 174–178, 187, 197, 207, 218, 220, 223 pro-cyclical bias in monetary policy and, 177 trade, 144, 150, 163 Bordo, Michael, 3, 99, 189 Bresnehan, Timothy F., 180 Broaddus, J. Alfred, 5, 332, 339, 341, 344, 345, 346, 360–362 brokers, bill, 4, 113, 115, 119, 125–127, 131, 144, 152, 153, 164 Brunner, Karl, 218, 220, 224 bubble, 375–377, 405 2002-03, 419 as a form of collective trust, 383 asset, defined, 5 credit, 415 defined, 375, 381 equilibrium, 392–394 equity, 415 land price, 6 microeconomics of, 384–387 rational, 6, 381–388 defined, 375 bubble, collapse, problems of, 6, 376, 385–387, 395, 396 Burgess, W. Randolph, 35, 43 Burns, Arthur, 404 Calomiris, Charles, 4, 176, 184, 187, 194, 202, 204, 218, 219 Capie, Forrest, 68, 116 capital, equity (net worth), 169, 230–234, 235, 241, 245–246, 247, 248–249, 250, 262–264, 266, 270, 284, 296, 297–299 Carlson, Mark, 87 Carter, Jimmy, 401 certificates, clearing-house loan, 14, 63, 64, 108

Index Chapman, Stanley, 139 Chaudri, Ehsan U., 189 Christiano, Lawrence, 5 collateral currency, 108 good, defining, 103, 107, 149–150 requirements, 107 Collins, Michael, 144, 163 Commercial and Financial Chronicle, 15 Consolidated Discount Company, 128 Corrigan, Gerald, 6, 399, 401–403, 410, 413–414, 415–416, 418–419, 421–422 Crash 1929 stock market, 186–188, 218, 219–220 1987 stock market, 403, 414 credit, elastic, 173 crisis 1825, 125 1866, 125–129 1914, 66 1930s, 82–86 2007–08, 59, 90, 110 defined, 250 Hunt (Brothers) silver market, 403, 404 Cummings, Chris, 357 currency asset-backed, 66, 70 associations, 66 bond-backed, 66 elastic, 41, 173 and bank panics, 169 real bills doctrine and, 175 inelastic, 59, 63–64, 72, 168 uniform national, 62, 63 Davis, Joseph, 204 Davison, Henry, 2 debt, short-term, 366 deflation, 15, 42, 44, 45, 81, 83, 176, 218, 222, 225, 354, 419 shock of unexpected, 40–41 deposits converting, 65 equity and, 170 evaluating riskiness of, 204 household, 235, 236, 250 interest on interbank, 66, 172 on-call, 126 pre–Civil War role of, 12

425

reserve requirements on, 54, 56, 73, 201 reserves and, 64 supply and demand of, by banks, 244 uninsured, 12 Dice, Charles, 186 discount houses, 68, 115 market, central bank and, 70 policies, and the Federal Reserve, 43, 61 rates, 70, 89, 173 Federal Reserve Banks and, 71 window, 8, 36, 43, 79 and the Federal Reserve, 59, 74, 80, 83–85, 88: alternative to, 86 Bank of England and, 68 frosted glass metaphor, 68, 89, 116, 150 limited access to, 89 monitoring, 151 moral problems of, 54, 56 reluctance to borrow from, 104–105 restricted eligibility of collateral, 108 savannah pond analogy, 119 Discount Corporation, 128 Discount System in Europe, The (Warburg), 144, 162 discounts, 119 daily, 120 disinflation, 411 Volcker, 218, 224, 401 Dodd-Frank Wall Street Reform and Consumer Protection Act of 2012, 92, 408 down payments, 380 Eccles, Marriner, 199, 200, 218, 222 Eichengreen, Barry, 184, 188, 190 Emergency Relief and Construction Act, 91 equilibrium, benchmark, versus financial, 240 Euler equation, 251 examiners bank, 17, 38, 46 National Bank, 10, 17 Exchange Stabilization Fund, 183, 200, 201, 342 exchange, stock, 88, 113, 405 Federal Deposit Insurance Corporation (FDIC), 46 Federal Deposit Insurance Corporation Improvement Act of 1991, 92

426

Index

Federal Home Loan Bank Board, 46 Federal Open Market Committee (FOMC), 332–369 asset acquisition, 362 foreign exchange operations, 349 inflation targeting, 355 lending to Mexico (1994–95), 345 memorandum of Discussion discontinuance, 334 monetary policy at the zero interest bound, 368 presumptive interest rate policy of (1994), 341 transparency and communication, 337 Federal Open Market Committee (FOMC), primary responsibilities of, 332 Federal Reserve Act of 1913, 3, 37, 38, 60, 71, 86 amendments to, 37, 90 1917, 37 1933, 199 collateral quality requirements under, 108–109 failures of, during Great Depression, 60 LLR provisions of the, 71–75 purposes of, 7 regulation under the, 71–91 Section 10(a), 91 Section 10(b), 91, 92 Section 13(13), 92 Section 13(3), 89, 90, 91, 92 Section 13(b), 92 Federal Reserve Banks Atlanta, 87, 196 Boston, 353 Chicago, 87 Cleveland, 368 Kansas City, 353 Minneapolis, 80 mission of, 172 New York, 76, 342, 345, 355 Crash of 1929 and the, 77, 87 operational structure of, 172 Philadelphia, 80, 218, 220 Richmond, 5, 332 San Francisco, 80, 352 St. Louis, 87, 196, 343, 348, 351 Federal Reserve Board, 71 Federal Reserve System, 33–39 1914–51, 166–208

egregious alleged errors of (Calomiris), 218, 219 1929–33 failure of the, 109–110 lessons learned, 111 1930s, 206–207 as lender of last resort, 59, 60, 75–91, 103, 106 during the Depression, 83 to 1933, 76 board examiners and member banks (1915–29), 40 conflict with the OCC, 34–38 decentralized structure of, 88 early years of, 38–39 Great Depression and, 77 incomplete regulation and supervision and the, 34 lending authority of legislated changes to, 92 limited membership in, 105–106 main flaws of, 103–108 original mission of, 168–173, 185, 207 policy of, 1929-34, 77–82 politicization of, 172 purpose of, 59, 342 reorganization of (1935), 199 stabilizing seasonal interest rate swings, 76, 206 structural changes under the, 54, 57, 208 U.S. Treasury and the, 206, 416, 418 Federal Savings and Loan Insurance Corporation, 46 Fetter, Frank, 114 Financial Crisis Inquiry Commission, 406 Financial Discount Company, 128 Financial Stability Council, 90 Financial Stability Oversight Council, 418 First National Bank Chicago, 21 Crestland, 36 Fisher, Irving, 175, 186, 218, 221 Fisher, Peter, 345, 346, 348, 355 Flandreau, Marc, 4, 69, 149 Forgan, James B., 20 Formative Period of the Federal Reserve System (Harding), 34 Friedman, Milton, 7, 30, 61, 84, 168, 175, 184, 188, 192–197, 202, 419, 420–421 Fruhling & Goschen, 147

Index Gaines, Tilford, 403 Galbraith, Kenneth, 218, 220 Gertler, Mark, 229, 236, 239, 270 Gilchrist, Simon, 229 Gilletts (bill brokers), 139 Glass, Carter, 2, 71, 73, 103, 171, 179, 187, 188, 199, 218, 220 Glass-Owen Bill of 1913, 2, 71 Glass-Steagall Act of 1932, 74, 89, 90, 91, 108, 173, 189, 190, 192 Glyn Mills Currie & Company, 125 gold free, 188, 189, 191, 218, 220–221 standard, 40, 70, 85, 100, 109, 175, 179, 183, 184, 188–192, 195, 208, 218, 223, 338 and the Federal Reserve System, 60 and the National Banking System, 9 central bank and, 61 fundamental to U. S. monetary policy, 83 limit, 168 panics and the, 88 U.K.’s abandonment of, 78, 81, 193 Gold Reserve Act of 1934, 200 Gold Standard Act of 1900, 18 Goldsborough, T. Alan, 199 Gonzales,Henry, 333–334 Goodfriend, Marvin, 5, 361, 362, 366–367, 368, 406, 412, 416 Goodhart, Charles, 103, 116 Gorton, Gary, 406 Great Depression, 45, 60, 184, 207, 417 bank panics and the, 192–197 Federal Reserve System and, 77, 86 Greenspan put, 411, 415 defined, 412 Greenspan, Alan, 6, 218, 219, 332, 334–336, 337–341, 342, 343, 344–345, 346–347, 349–355, 360–362, 363, 366, 369, 399, 403–405, 411–413, 414–415, 416–417, 419–421 Grossman, Richard, 32 Hakkio, Craig, 358 Hamilton, Alexander, 62 Hamlin, Charles, 191 Hanes, Christopher, 198, 202, 204 Harding, W.P. G., 34 Harrison, George, 175, 176, 190, 191, 192, 198

427

Hawtrey, Ralph, 218, 222 hazards, moral adverse selection, 232–233, 269–283 asymmetric information, 233–234, 283–301 running away, 230–231, 235–236 unobserved banker effort, 231–232, 247–269 History of the Federal Reserve (Meltzer), 7 Hoenig, Tom, 353 Hoover, Herbert, 82, 218, 221 Hsieh, Chang-Tai, 189, 192 Hughson, Eric, 205 Humphrey-Hawkins statute, 351, 354 Ikeda, Daisuke, 5 Imperial Ottoman Bank, 125 Indiana, State Bank of, 54, 58 Indianapolis Monetary Convention (1897), 66 inflation, 197, 337, 421 1865-1929, 40 1980s, 333 operationally meaningful, 350 injection, equity, 245, 250 defined, 229 into banks, 260–262 mutual funds and, 259–260 insolvency, bank 1838–60, 30 1865–1913, 29 1929–33, 30 2008–09, 30 insurance, deposit, 89, 90, 182 interventions, policy, four types of, 5 investment call loans as liquid, 65, 107 public, 395 versus commercial banking, 90, 103, 182, 188 Jackson, Andrew, 62 Jacobs, Lawrence M., 115 Jalil, Andrew J., 15 Jekyll Island Club, 1910 meeting at, 1–3 Joint Stock Discount Company, 128 Jordan, Jerry, 368 Kabiri, Ali, 186 Karadi, Peter, 236, 239 Karadi, Peter, 229

428

Index

Kemmerer, Edwin W., 15 Keynes, John Maynard, 218, 222 King, Robert G., 5, 125, 127, 128 Kiyotaki, Nobuhiro, 229, 236, 239 Knickerbocker Trust Company, 14 Knox, John Jay, 18 Kocherlakota, Narayana, 5, 381 Kohn, Don, 355 Kuhn Loeb, 67 land overvaluation consequences of, 374–388 debt guarantees and, 377–381 Landon-Lane, John, 198 Lang, Rick, 358 Laughlin, J. Laurence, 66, 101, 102 leverage, 379–380 liabilities, 169 liability double, 7, 24–25, 43, 46 defined, 24 shareholder, 25 single, 25 Lindsey, Larry, 343, 346 liquidation, 36, 84, 170 voluntary, 8, 25–29 liquidity acceptance, 114 broad, 363 implicit marginal, 364 panics and, 7, 15, 16, 33 provision, 108 risks, reducing, 205 solvency and, 192 traps, 190, 197–199, 203, 208, 218, 219, 221–222 versions of, 197–199 loans call, 65, 102, 119 government bank defined, 229 restrictions on, by Federal Reserve Act, 73, 84 supply, 176 Lombard Street (Bagehot), 36, 116, 144, 162 London Discount Company, 128 London Joint-Stock Bank, 125 Mankiw, N. Gregory, 229, 270 markets

call loan, 65, 70, 417 money American, 113 London, 118–125 versus capital, 102–103 Mason, Joseph R., 176, 184, 194, 202 McDonough, William, 342, 344 Meltzer, Allan H., 7, 15, 79, 107, 174, 175, 176, 183, 189, 190, 191, 198, 202, 419, 421 Melzer, Tom, 343, 344, 351 Mercantile Discount Company, 128 Meyer, Eugene, 82 Meyer, Larry, 353, 355, 361 Minehan, Cathy, 353 Mints, Lloyd, 218, 223 Miron, Jeffrey, 15, 42, 204 Mitchener, Chris James, 87 model, two period, described, 388–392 monetarism, pragmatic, 401, 402 Monetary Control Act of 1980, 92, 403 Monetary History (Friedman and Schwartz), 7 monetary policies American, 1914-51, 166–208 unconventional creditor bailouts, 266 equity injections into banks, 244, 250 four kinds of, 242 government deposits in banks and loans to firms, 245 interest rate subsidies and new worth transfers to banks, 246 leverage restrictions, 269 observable effort benchmark, 255 unobservable effort, 259 unobserved banker effort, 251 Moret, Clement, 192 Morgan, J. P., 14 Morgenthau, Henry, 200, 201, 207 National Bank Acts 1864, 9, 36 amendment to (1875), 17 1933, 188 National Banking Acts 1863, 63 1865, 63 National Banking Era (1863–1914) call loans and, 102 clearing house loans and, 108

Index cylindrical panics in the, 205 liquidity panics in the, 16 shortcomings during, 100 National Banking System assessing the, 32–33 defects of, 61–63 depositor losses under the, 7 organization of, 9–10 problems (1864–1913), 15–16 origins of, 8–10 rise of trust companies, 14 state bank re-emergence, 13 regulation and supervision under the, 54, 56 supervision (1864–1913) bank failures and, 32 Comptroller of Currency and, 22 consequences of, 33 disclosure and, 17 double liability and, 25 state banks and, 24 voluntary liquidations and, 29 National Credit Union Administration, 46 National Currency Act of 1863, 9 National Discount Company, 128, 138 National Monetary Commission, 1, 15, 31, 59, 113, 144, 149, 153, 162, 174, 178, 218, 221 mission of, 114, 115 reports of, 115 National Reserve Association, 2, 60, 70, 71 Nebraska State Banking Board, 31 New Deal, 45–47 New York Clearing House, 14, 63, 106 Nicholas, Tom, 186 Norris, Frank L., 21, 218, 220 Nowobilski, Andrew, 229 O’Neill, James, 406 Office of the Comptroller of the Currency (OCC), 8, 9 bank supervision (1914–29), 39 examination requirements of, 17 operation (1884–1913), 19 operation of, 22 purpose of, 16 Owen, Senator Robert L., 2 panics, 218, 219 1857, 63, 171 1866, 125–129

429

1873, 63, 66, 171 1893, 64, 66 1907, 8, 14, 21, 59, 64, 102, 171, 412 1929, 196 1932, 194, 196 2008, 15, 405–409 American propensity for, 168, 171 cyclical and seasonal, 169, 205 events precipitating, 64 Federal Reserve System and, 60 Great Depression, 192–197 incipient, 15 National Banking System and, 7, 15–16, 63, 218, 221 Parry, Robert, 352, 367 Peel Act of 1844, 126, 129, 191 Phelan, John, 405 Philadelphia National Bank, 21 Phillippovich, Eugen Von, 115 Phillips curve, 218, 225 Poole, William, 348, 367 Raff, Daniel M. G., 180 Rajan, Raghuram, 6, 401 rate bill-buying, 74 Federal Funds, 226, 337–341, 409 inflation, 410 interest nominal, 5, 18, 83, 88, 100, 174, 175, 176, 177, 183, 207, 218, 220, 222, 223: short-term, 197, 363 real, 376, 385 seasonal, 41–43 spread, 249–250 subsidies, 281–282 recessions, 218, 220 1907–08, 8 1920–21, 40, 42, 60 1937–38, 199–204, 218, 222 1990–91, 341 2007–09, 15, 227 between 1895 and 1912, 15 panics and, 7 Reconstruction Finance Corporation, 182, 218, 221 rediscounting, 35, 36, 60, 68, 69, 70, 72, 84, 88, 91, 101, 103, 113, 119, 127, 144, 162 under the Aldrich Plan, 107 reform, 66–69, 71

430

Index

Regulation Q, 89, 403 Reichsbank, 67, 69, 83 Reifschnieder, David, 362 rents, 380 Reserve Banks and the Money Market, The (Burgess), 35 reserves gold, 75 reserve banks and, 73 inverted pyramid of, 64–66 pyramiding, 171 requirement, 201, 203, 218, 219, 222–223 Responsibility Doctrine (Bagehot), 68 Rhode, Paul, 204 Richardson, Gary, 87 Ridgely, William, 18 Riefler-Burgess doctrine, 174, 176, 177, 178, 190, 192, 197, 198, 203, 206, 207 Rockoff, Hugh, 30, 198 Romer, Christina, 15, 189, 192 Roosevelt, Franklin Delano, 183, 195 Roosevelt, Theodore, 179 Rothschilds, 147 Royal Bank of Liverpool, 125 RPs (repurchase agreements), 357–360 Samuelson, Paul A., 375 Santos, Manuel S., 381 Sayers, Richard S., 116, 126, 139 Schwartz, Anna J., 7, 30, 61, 84, 168, 175, 184, 188, 189, 192–197, 202 Second Bank of the United States, 8 Securities Exchange Corporation (SEC), 46 seigniorage, 361 Shafroth, Senator John F., 73 Shelton, Arthur, 2 Sherman, John, 24 shocks balance-of-payment, 113 deflationary, 3, 40–41, 45 demand for reserves, 176 inflationary, 47 seasonal, 42 supply and demand, 4, 207 supply-side, 176 Silver Purchase Act of 1934, 200 Small Business Investment Company Act of 1958, 92 Smith Fleming & Company, 140 society, American, changes to 1920s, 179–181

1930s, 181–185, 206–207 solvency, panics and, 7, 16, 33 Special System Investment Account, 76 Sprague, O. M. W., 15, 178, 190 stability, price versus financial, 40–45 stringency, seasonal, 64 Strong, Benjamin, 76, 79, 88, 175, 218, 223 Suffolk Bank, 54, 57–58 Suffolk Banking System, 54, 57 supervision bank, 7 before 1913, 7 components of, 16: discipline, 18; disclosure, 16–17; examination, 17–18 consequences of, 24–33 during Overend, Gurney Panic of 1866, 151 prudential, 18 state banks and, 22–24 under the Federal Reserve Act, 91 Taylor rule, 218, 224, 333 Taylor, John, 333 Temin, Peter, 184, 193 Thornton, Henry, 218, 221, 408 Tirole, Jean, 381 Tobin, James, 218, 224 transfers directed, 396 tax-financed, 283 uniform, 395–396 Treasury-Federal Accord of 1951, 166 Treasury-Federal Reserve Accord of 1951, 173, 207, 208, 218, 223, 341 Truman, Ted, 343 Ugolini, Stefano, 4, 68, 149 Unified Budget Procedure (1960), 417 Union Bank of Australia, 125 United Reserve Bank, 69 Vanderlip, Frank, 2, 105 Volcker, Paul, 399–400, 401–402 Wagner Act of 1935, 182 Wallace, Neil, 235, 250 Warburg, Paul, 2, 4, 59, 60, 67, 90, 102, 107, 113, 144, 162 Warburg-Aldrich Plan, 69–71, 101 Weidenmier, Marc D., 205

Index Wheelock, David, 3, 99, 175, 184, 202 White, Eugene, 3 Wicker, Elmus, 179, 184, 189, 194 Wicksell, Knut, 218, 221 Williams, John, 362 Williams, John Skelton, 37 Willis, H. Parker, 71, 73, 103 Wilson, Barry, 187, 195, 204

Wilson, Woodrow, 35, 71, 218, 223, 400 Withers, Hartley, 125, 150 Woodford model, 218, 223 Woodford, Michael, 381 Yellen, Janet, 350, 352 zero bound, 409

431

Continued from page iii Barry Eichengreen, Elusive Stability [9780521365383, 9780521448475, 9780511664397] Barry Eichengreen (Ed.), Europe’s Postwar Recovery [9780521482790, 9780521030786] Caroline Fohlin, Finance Capitalism and Germany’s Rise to Industrial Power [9780521810203, 9780511510908] Michele Fratianni and Franco Spinelli, A Monetary History of Italy [9780521443159, 9780521023450, 9780511559686] Mark Harrison (Ed.), The Economics of World War II [9780521620468, 9780521785037, 9780511523632] Robert L. Hetzel, The Monetary Policy of the Federal Reserve: A History [9780521881326] Kenneth Mour´e, Managing the Franc Poincar´e [9780521394581, 9780521522847, 9780511572630] Aldo Musacchio, Experiments in Financial Democracy: Corporate Governance and Financial Development in Brazil, 1882–1950 [9780521518895] Larry Neal, The Rise of Financial Capitalism [9780521382052, 9780521457385, 9780511665127] Lawrence H. Officer, Between the Dollar-Sterling Gold Points [9780521038218] Angela Redish, Bimetallism [9780521570916, 9780521028936] Aurel Schubert, The Credit-Anstalt Crisis of 1931 [9780521365376, 9780521030298, 9780511664632] Pierre Siklos, The Changing Face of Central Banking: Evolutionary Trends since World War II [9780521780254, 9780521034494, 9780511606427] Tobias Straumann, Fixed Ideas of Money: Small States and Exchange Rate Regimes in Twentieth Century Europe [9780521112710, 9780511750953] Norio Tamaki, Japanese Banking [9780521496766, 9780521022330, 9780511586415] Mark Toma, Competition and Monopoly in the Federal Reserve System, 1914–1951 [9780521562584, 9780521022033, 9780511559761] Gianni Toniolo (with the assistance of Piet Clement), Central Bank Cooperation at the Bank for International Settlements, 1930–1973 [9780521845519, 9780521043700] David C. Wheelock, The Strategy and Consistency of Federal Reserve Monetary Policy, 1924–1933 [9780521391559, 9780521531399, 9780511528743] Elmus Wicker, Banking Panics of the Great Depression [9780521562614, 9780521663465, 9780511591985] John H. Wood, A History of Central Banking in Great Britain and the United States [9780521850131, 9780521741316]