A Retrospective on the Classical Gold Standard, 1821-1931 9780226066929

Citation preview

A Retrospective on the Classical Gold Standard, 1821-1931

A Conference Report National Bureau of Economic Research

A Retrospective on the Classical Gold Standard,

1821-1931 Edited by

Michael D. Bordo Anna J. Schwartz

The University of Chicago Press

Chicago and London

MICHAEL D. BORDO is professor of economics at the University of South Carolina and research associate of the National Bureau of Economic Research. ANNA J. SCHWARTZ, research associate of the National Bureau of Economic Research, is coauthor, with Milton Friedman, of Monetary Trends in the United States and the United Kingdom, 1867-1975.

The University of Chicago Press, Chicago 60637 The University of Chicago Press, Ltd., London © 1984 by The National Bureau of Economic Research All rights reserved. Published 1984 Printed in the United States of America 90 89 88 87 86 85 84 5 4 3 2 1

Library of Congress Cataloging in Publication Data Main entry under title:

A Retrospective on the classical gold standard 1821-1931. (A Conference report / National Bureau of Economic Research) Papers from a conference sponsored by the National Bureau of Economic Research, held in March, 1982. Bibliography: p. Includes index. 1. Gold standard-History-Congresses. I. Bordo, Michael D. II. Schwartz, Anna Jacobson. III. National Bureau of Economic Research. IV. Series: Conference report (National Bureau of Economic Research) HG297.R44 1984 332.4'222 84-2440 ISBN 0-226-06590-1

National Bureau of Economic Research Officers Franklin A. Lindsay, Chairman Richard Rosett, Vice Chairman Eli Shapiro, President David G. Hartman, Executive Director and Corporate Secretary

Charles A. Walworth, Treasurer Sam Parker, Director of Finance and Administration

Directors at Large Moses Abramovitz George T. Conklin, Jr. Jean A. Crockett Morton Ehrlich Edward L. Ginzton David L. Grove Walter W. Heller Saul B. Klaman

Franklin A. Lindsay Roy E. Moor Geoffrey H. Moore Michael H. Moskow James J. O'Leary Peter G. Peterson Robert V. Roosa Richard N. Rosett

Bert Seidman Eli Shapiro Stephen Stamas Lazare Teper Donald S. Wasserman Marina v. N. Whitman

Directors by University Appointment Charles H. Berry, Princeton James Duesenberry, Harvard Marcus Alexis, Northwestern J. C. LaForce, California, Los Angeles Paul McCracken, Michigan Ann F. Friedlaender, Massachusetts Institute of Technology

James L. Pierce, California, Berkeley Nathan Rosenberg, Stanford James Simler, Minnesota James Tobin, Yale William S. Vickrey, Columbia John Vernon, Duke Burton A. Weisbrod, Wisconsin Arnold Zellner, Chicago

Directors by Appointment of Other Organizations Carl F. Christ, American Economic Association Robert S. Hamada, American Finance Association Gilbert Heebner, National Association of Business Economists Robert C. Holland, Committee for Economic Development Stephan F. K.aliski, Canadian Economics Association Douglass C. North, Economic History Association

Rudolph A. Oswald, American Federation of Labor and Congress of Industrial Organizations G. Edward Schuh, American Agricultural Economics Association Albert Sommers, The Conference Board Dudley Wallace, American Statistical Association Charles A. Walworth, American Institute of Certified Public Accountants

Directors Emeriti Arthur F. Burns Emilio G. Collado Solomon Fabricant Frank Fetter

Thomas D. Flynn Gottfried Haberler George B. Roberts Murray Shields

Boris Shishkin Willard L. Thorp Theodore O. Yntema

Since this volume is a record of conference proceedings, it has been exempted from the rules governing critical review of manuscripts by the Board of Directors of the National Bureau (resolution adopted 8 June 1948, as revised 21 November 1949 and 20 April 1968).

Contents

Preface Introduction Anna J. Schwartz I.

2. The Success of Purchasing-Power Parity: Historical Evidence and Its Implications for Macroeconomics Donald N. McCloskey and J. Richard Zecher Comments: Robert E. Lipsey, Milton Friedman General Discussion

23

121

TECHNICAL PROCEDURES: RULES OF THE GAME Chairman: Robert A. Mundell 3. The Bank of England and the Rules of the Game under the International Gold Standard: New Evidence John Dutton Comment: Donald E. Moggridge Reply

vii

1

THE GOLD STANDARD AS INTERPRETED IN TRADITIONAL AND REVISIONIST WORKS Chairman: Moses Abramovitz 1. The Gold Standard: The Traditional Approach Michael D. Bordo Comment: C. Knick Harley General Discussion

II.

xi

173

viii

Contents

4. Bank of England Operations, 1893-1913 John Pippenger Comment: Charles A. E. Goodhart General Discussion of Dutton and Pippenger Papers 5. The Gold Standard and the Bank of England in the Crisis of 1847 Rudiger Dornbusch and Jacob A. Frenkel Comment: J. R. T. Hughes General Discussion III.

233

INTERNATIONAL EXPERIENCE IN THE OPERATION OF THE GOLD STANDARD Chairman: Karl Brunner 6. Canada and the Interwar Gold Standard, 1920-35: Monetary Policy without a Central Bank Ronald A. Shearer and Carolyn Clark Comment: Charles Freedman General Discussion 7. Operations of the German Central Bank and the Rules of the Game, 1879-1913 Paul McGouldrick Comment.' Heywood Fleisig General Discussion 8. Swedish Experience under the Classical Gold Standard, 1873-1914 Lars Jonung Comment: Peter H. Lindert 9. Italy in the Gold Standard Period, 1861-1914 Michele Fratianni and Franco Spinelli Comment.' Richard E. Sylla General Discussion of Jonung and Fratianni-Spinelli Papers

IV.

203

INTERNATIONAL LINKAGES UNDER THE GOLD STANDARD Chairman: Allan H. Meltzer

277

311

361

405

ix

Contents

10. The Gold Standard and the Transmission of Business Cycles, 1833-1932 Wallace E. Huffman and James R. Lothian Comment: Michael Connolly 11. Real Output and the Gold Standard Years, 1830-1913 Stephen T. Easton Comment: Geoffrey E. Wood General Discussion of Huffman-Lothian and Easton Papers 12. Canada without a Central Bank: Operation of the Price-Specie-Flow Mechanism, 1872-1913 Georg Rich Comment: Peter Temin General Discussion

v.

455

513

547

THE GOLD STANDARD AS A STABILIZER OF COMMODITY PRICES

Chairman: Richard H. Timberlake, Jr. 13. War, Prices, and Interest Rates: A Martial Solution to Gibson's Paradox Daniel K. Benjamin and Levis A. Kochin Comment: Phillip Cagan General Discussion

587

14. Some Evidence on the Real Price of Gold, Its Costs of Production, and Commodity Prices 613 Hugh Rockoff Comment: Robert J. Barro Reply General Discussion 15. The Image of the Gold Standard Leland B . Yeager

651

Participants

671

Au~&fuda

~5

Subject Index

000

Preface

The conference in March 1982 at which the papers in this book were presented brought together some fifty scholars with an interest in economic history and international monetary relations to examine the current state of knowledge of the gold standard as it operated during the 110 years until 1931, when the classical gold standard may be said to have ended. Five sessions were held at which fourteen papers were presented. Comments on each paper were prepared in advance of the conference by a designated discussant. Two comments are included with the paper by Donald McCloskey and J. Richard Zecher: one by Robert Lipsey, the assigned discussant, the other by Milton Friedman, who had no specific paper assignment at the first session but who responded to issues the authors' paper raised. A brief summary of the discussion by conference participants, prepared by Barry Eichengreen, follows the papers and the prepared comments. The replies by two authors to the comments on their papers are also included. In addition to the formal papers, at the final dinner meeting Leland Yeager provided an overview of the conference proceedings. His address is the fifteenth paper in this book. We are grateful to the Earhart Foundation, the Alex C. Walker Educational and Charitable Foundation, and others for their support of the conference. All the participants express their thanks to Kirsten Foss for her efficient management of the conference arrangements. Michael D. Bordo Anna J. Schwartz

xi

Introduction Anna J. Schwartz

Britain's abandonment of the gold standard in 1931 signaled the end of the gold standard era. The new technical economic insights and improvements in statistical and computational techniques that have developed in the half century since then constitute part of the motivation for organizing a conference on the gold standard at this juncture-they should make it possible to mine new nuggets of knowledge from the historical evidence. In addition, recent experience with floating exchange rates and unstable domestic monetary policies has made that historical evidence seem highly relevant to today's problems. Much professional attention is once again focused on the merits of fixed exchange rates and constraints on domestic monetary autonomy. We date the start of the gold standard era in 1821, when Britain resumed specie payments at the parity that had pevailed before the Napoleonic Wars, indeed, from 1717 on. By the end of the era, 110 years later, the gold standard had been transmuted. In the pre-World War I period, it evolved as a system in which countries redeemed their domestic currencies in gold and in which the offsetting of gold flows by monetary authorities to maintain existing monetary conditions, though it occurred from time to time, was not regarded as proper conduct. England-the world's largest trading nation, the center of the world's commodities markets and of the world's gold market, and the world's leading creditor nation-played a central role differing from that of other countries. Though gold was the key reserve, many countries in addition held foreign exchange (largely though not exclusively in sterling), in itself an indication of confidence in the stability of exchange rates. By 1931, gold coins did not circulate in most countries on the gold standard, and paper currency could be redeemed for gold only under severe limits. Sterilization of gold flows was accepted as a desirable way to limit the internal 1

2

Anna J. Schwartz

monetary effects of gold flows. New York rivaled London as a second reserve center. The resultant destabilizing shifts of foreign-exchange reserves from the center experiencing gold losses to the other center led to a loss of confidence in the stability of exchange rates (Dam 1982, pp. 24-60). The studies undertaken for this conference were designed to deepen our understanding of the functioning of the historical gold standard. A major by-product is to affect our response to such current economic questions as the recent resurgence of interest in a gold standard as a solution to problems of inflation, high interest rates, and low productivity. The history of the gold standard may be examined from many perspectives. The conference concentrated on five: 1. What were the main themes of the traditional literature on the gold standard, spanning several centuries, compared to the themes stressed in the analysis associated with the post-World War II monetary approach to the balance of payments? 2. Did operating procedures of the Bank of England before 1914 conform to theoretical notions of the ideal functioning of the gold standard? 3. What was the experience of a sample of four countries, with and without a formal central bank (Canada, Germany, Italy, and Sweden), in the gold standard era? This evidence was designed to supplement more extensive knowledge of the operation of the gold standard in the United States and Great Britain. 4. What links integrated the international monetary system under the gold standard? 5. Did the gold standard stabilize commodity prices? The conference was enlivened by the expression of strong conflicting views on fundamental issues. One such issue was the significance of purchasing-power parity and interest-rate parity. Do independent monetary policies have any role under fixed exchange rates if purchasingpower and interest-rate parity are important? Another issue was the role of central banks under the gold standard. Did these institutions, or less formal ones that performed similar functions, impede or assist the operation of the gold standard, or were they irrelevant, so that the operation of the gold standard was automatic? The pre-World War I period was characterized by economic growth and expanding world trade. Did adherence to fixed exchange rates under the gold standard playa major role in producing growth or was it largely irrelevant to growth? The importance of accounting for the links across the Atlantic under the pre-World War I gold standard was stressed by some participants, whereas at least one participant questioned the validity of the concept of

3

Introduction

an Atlantic economy. The reality of trend movements in commodity prices and long-term interest rates under the gold standard was another subject of debate at the conference. Was it visual spurious regression that produced the appearance of trends or did economic agents apprehend the trends as actualities? The conference did not settle these issues, but future investigators will need to confront them in dealing with the international monetary system. The studies that were prepared for the conference relied on various modes of analysis. Several were based on historical nonquantitative evidence. Some adapted the National Bureau business-cycle analysis. Others used regression analysis. Several studies presented Granger-Sims tests and analyses of ARIMA techniques. One general question that arises with high-powered statistical tests applied to pre-World War I data is the validity of the results, given the questionable reliability of some of the underlying data. Economists may nevertheless welcome the findings as a starting point for further refinement of hypotheses and a spur to efforts to improve the data sets. Section 1 of this introduction summarizes the five sessions of the conference, highlighting unresolved issues. Contemporaries regarded the gold standard as a qualified success yet later observers gave a less favorable account of the era. Section 2 attempts to account for the change in views. The implications of the findings of the conference studies and of Leland Yeager's suggestive talk at a dinner session form the basis for the speculations in the concluding section 3 on the prospects for reinstating a gold standard. 0.1

0.1.1

Summary of Issues Examined

The Gold Standard as Interpreted in Traditional and Revisionist Works

Bordo surveyed six major themes developed since' the eighteenth century in the traditional approach to the gold standard: 1. Gold represented an ideal monetary standard, both domestically and internationally, because of its unique qualities as a standard of value and a medium of exchange. The evils of a depreciated money under an inconvertible fiduciary money were contrasted with the price stability that, according to the commodity theory of money, automatic operation of a gold standard yielded over the long run. 2. The essence of the gold standard was the maintenance of a fixed price of a national money in terms of gold, linking the price levels of all countries. The price-specie-flow mechanism ensured that any disturbance away from the natural distribution of gold determined by a nation's

4

Anna J. Schwartz

real income and money-holding habits would lead to an equilibrating process through arbitrage in the gold market. Gold flows, by changing a nation's money supply, would then also change its price level. 3. The law of one price ensured that through arbitrage and individual trade in commodities, prices for similar goods, taking account of transportation costs and trade impediments, would be similar. 4. Capital flows played a role in the gold standard balance-ofpayments adjustment mechanism, supporting the price-specie-flow mechanism. A decline or rise in the domestic money stock would lead to a rise or fall in short-term interest rates affecting the movement of funds from abroad. Long-term capital flows were a source of disturbance to the balance of payments but also a balancing item in the balance of payments. By raising the price level in the capital-importing country and lowering it in the exporting country, and hence producing a current-account surplus in the latter and a current-account deficit in the former, the transfer of capital resulted in a transfer of real resources. 5. The role of central banks in the adjustment mechanism was initially examined in the context of the gold standard as an automatic monetary rule, but later views shifted to the gold standard as managed by central banks to facilitate adjustment to internal and external gold flows, and finally to discussion of the extent to which central banks in fact followed the rules of the game. 6. Schemes to reform the gold standard included its management by the central bank to shield the domestic money supply from external shocks; the separation of the medium-of-exchange function from the standard of value, by adoption of a tabular standard, bimetallism, symmetallism, the compensated dollar, or a commodity standard; and finally, the creation of some form of a supernational central bank. Harley expressed disappointment at Bordo's lack of attention to the two issues he regarded as central to a better understanding of the gold standard: transfers and reparations and the "Atlantic economy." McCloskey and Zecher took issue with the traditional view that under the gold standard, prices (interest rates) in one country could be out of line with prices (interest rates) in the rest of the world for considerable periods of time, inducing first a gold flow, than a change in the country's money supply, followed by adjustment in its price level (interest rates) to bring it (them) into line. In their view, prices (interest rates) internationally never diverge (except for impediments to trade and capital movements, transportation and transactions costs) because rational economic behavior of consumers and producers (investors) links prices (interest rates) through arbitrage. Lipsey criticized the McCloskey-Zecher identification of purchasingpower parity with the law of one price as impossible to refute and empty

5

Introduction

as a theory. Using aggregate indexes of wholesale prices in different countries to test the existence of arbitrage, as the authors do, moreover, was questionable given the differences in the. construction and composition of those series. On the other hand, if all they meant by purchasingpower parity was that foreign influences on prices cannot be ignored, the proposition was unarguable. The evidence McCloskey and Zecher cite to support the view that specie flows did not activate an adjustment process is that contrary to a claim by Friedman and Schwartz (1963, p. 99), it did not so work in 1879. McCloskey and Zecher assert that U.S. price movements that year possibly anticipated gold inflows, as if arbitrage were at work, but certainly did not lag the gold inflows. Friedman's response is that the comparison McCloskey and Zecher make between price rises and inflows of gold is the wrong one. Gold flows are a proxy for the quantity of money. Comparing prices and changes in the quantity of money directly fully supports the price-specie-flow mechanism. The initial gold flows, to which McCloskey and Zecher refer, had a direct effect on the composition, rather than the rate of growth, of high-powered money. The evidence McCloskey and Zecher cite to supporting purchasingpower parity achieved by arbitrage is the rise in wholesale prices in the United States after the trough in March 1933. They name the depreciation of the dollar rather than the National Recovery Act, to which Friedman and Schwartz alluded, as the proximate cause of the domestic price rise. According to them, domestic monetary growth cannot explain movements in prices beyond what are explained by purchasing-power parity. With respect to the rise in wholesale prices after the trough in March 1933, Friedman denies that the discussion in A Monetary History was an attempt to assess the relative contribution of several sources of price enhancement. The depreciation of the dollar, the growth in the domestic money stock, and New Deal measures, including the Agricultural Adjustment Act, the Guffey Coal Act, and still others that affected wages and prices, all played their part. Yeager's conclusion that acceptance of the general validity of purchasing-power parity and interest-rate parity need not exclude the general validity of the quantity theory of money for the analysis of domestic price movements is a judicious statement. Under a gold standard, centrifugal forces had some play, but ultimately centripetal forces triumphed. The monetary approach to the balance of payments asserts that there is one world, not a collection of separate national entities. One can easily accommodate a one-world view with degrees of autonomy over limited periods for individual economies. Under fixed exchange rates, depending on the degree of autonomy, individual economies could exercise control

6

Anna J. Schwartz

over their prices and interest rates in order to extend the period for adjustment, short of cutting loose from the restraints imposed by the gold standard. Mundell pointed out that the difference between these two views was reminiscent of an earlier discussion in 1937 between D. H. Robertson and Jacob Viner on the international adjustment mechanism. Abramovitz expressed concern that findings reported in the long-swing literature had not been integrated into the discussion of the gold standard. In early writings on the gold standard, gold flows served as the principal adjustment mechanism. Subsequently, as Bordo, Dornbusch and Frenkel, and Rich note, the role of short-term capital flows in the adjustment mechanism came to be recognized. The expansion of possible modes of adjustment to disequilibrium in a country's balance of payments due to domestic or external shocks does not damage the validity of the specie-flow mechanism. It simply indicates that adjustment became possible with a broader range of technical means. In sum, the issue of whether the dominant adjustment mechanism that links economies under fixed exchange rates is purchasing-power parity and interest-rate parity or specie flows was not resolved at the conference. 0.1.2 Technical Procedures: Rules of the Game The papers by Dutton and Pippenger give conflicting interpretations of the evidence on the Bank of England's pre-1914 performance. According to Dutton, the Bank may have violated two versions of the so-called rules of the game: one, the traditional view, requiring central banks to reinforce or not counteract the effects of gold flows on domestic money supplies; the other, a view propounded by Michaely (1971), requiring central banks to refrain from countercyclical operations, limiting their objective to maintenance of convertibility. According to Pippenger, on the other hand, the Bank was sensitive to threats to convertibility posed by international capital flows or by increases in domestic income and did not accommodate changes in the level of domestic incomes. Moggridge noted that Dutton's paper does not determine how much of the offsetting by the Bank of England, in violation of the no-offsetting rule, was "an automatic reflection of the discount market being forced into the Bank and how much reflected deliberate policy" (p. 197) and whether the Bank became more inclined to violate some rules over time. Goodhart agreed that the Bank's open-market operations imparted a procyclical impulse to the monetary base, but attributed it not to profit motives, as Pippenger does, nor to the "needs of trade," as Dutton does, but rather to the Bank's concern to protect its share of the London money market that was threatened by the growth of the London clearing banks in the 1890s.

7

Introduction

The tests of the rules may not have captured the effects of short-term and long-term capital transactions and hence may have provided an inadequate basis for judgments of the operation of the system. An interest-rate variable in a regression will not necessarily reflect the impetus for capital flows. If Brinley Thomas (1973) is correct, British capital exports and American capital imports were linked more closely to domestic- and foreign-investment activity than to interest rates (Bordo's paper in this volume, app. E). Transfers on private capital account must have had a disturbing influence on the lending country of greater significance than the financing of existing commodity trade. In the London money market, in addition, issues of foreign governments created balances for the debtor that required monetary management by the Bank of England to counteract gold movements induced by international capital transactions. The importance of those transactions is indicated by the fact that before 1914, Britain invested abroad at an annual rate of 4 percent of its national income and about 30 percent of its annual savings. Working with a small gold reserve, the Bank of England nevertheless avoided catastrophe in the wake of Britain's huge capital exports and general international interests. How did it achieve that result? One suggestion is that a strategic element was involved-all the players were aware of the consequence of collectively seeking to convert sterling into gold. Perhaps the public-good aspect of the pre-World War I managed gold standard was more important than under the Bretton Woods system. Another possibility is that the Bank of England resorted to some devices that do not appear in regressions testing its performance (on the devices, see Sayers 1936; Bloomfield 1963): 1. The use of variations in the price of gold is well known. The Bank acted on the gold points by varying its price for bars and foreign coins, refusing to sell bars or giving free advances on gold imports. Bank rate did not invariably move in line with the use of gold devices. 2. Open-market operations were conducted in such a way as to shield the domestic economy to the extent possible from actions designed to accommodate international capital movements. What seems to be a violation of the rules may in fact be the adaptation of policy to Britain's international role. 3. To discourage certain loans, typically originating in foreign transactions, the Bank manipulated the rate on advances or rediscounting, setting a higher rate than the official one. 4. England held vast short-term claims on foreign debtors that responded to Bank moves. The responsiveness of short-term capital tended to equalize open-market rates in different gold standard countries. 5. By extending or restricting short-term international loans, the Bank of England exercised a degree of control over the distribution of newly mined gold.

8

Anna J. Schwartz

Accordingly, when the Bank of England intervened in the money market, it may not have been violating the rules of the game. Dornbusch and Frenkel provide another view of the nineteenthcentury gold standard in their study of the operation of the Bank in a single year, 1847, when two crises occurred in the spring and the fall-one due to an external drain, the other to both an internal and external drain. They criticize the Bank's performance in both crises and emphasize the role of international capital flows during the adjustment process. For Dornbusch and Frenkel, suspension of the Act of 1844 limiting the Bank's fiduciary issue was required for the restoration of confidence but represented collapse of the rigid gold standard rules. Hence, for them, the gold standard provided a stable financial framework facilitating financial intermediation only when there was a lender-of-Iast-resort willing to discount freely during crises. Hughes commented that the Dornbusch-Frenkel focus on the monetary liabilities of the Bank of England ignored highly volatile movements of the domestic money supply held by the public that was a multiple of the Bank's note issues. The financial crisis in his view originated in the financial system outside the Bank, and it was the "cascading deluge upon the banks and discount houses" of the private issues "that made the Bank Act of 1844 an iron lid that had to be removed by the Treasury letter" (p. 266). 0.1.3

International Experience in the Operation of the Gold Standard

The papers on the experience of a sample of four countries under the gold standard tend to dismiss the importance of monetary actions in accord with the theoretically appropriate "rules of the game" as the explanation for the operation of the fixed-exchange-rate system. Neither Sweden nor Germany before 1914 apparently observed the rules of the game, according to lonung and McGouldrick. Fratianni and Spinelli report that Italy did not even formally adhere to a gold standard for most of the period 1861-1914. In Shearer and Clark's account, Canada, without a central bank, fortuitously returned to the gold standard after World War I and shortly thereafter abandoned it. The foregoing papers provide evidence that supplements the voluminous literature on Great Britain and the United States (for example, Ford 1962 on the asymmetry of gold standard experience as between Great Britain and the peripheral countries to which it exported capital; Morgenstern 1959 on interactions among the money markets in Great Britain, Germany, France, and the United States; and other studies cited in Bordo, app. E). Freedman noted that Shearer and Clark do not provide a generalequilibrium framework for their analysis of Canadian monetary history

9

Introduction

between the wars, but direct attention to such elements as "the relationship between the price of gold and the relative costs of borrowing in New York and borrowing under the Finance Act, the use of gold devices, and the response by the authorities to the gold flows at the end of the 1920s and in the early 1930s" (p. 307). Fleisig also objected to the absence of a model of international economic interaction underlying McGouldrick's and other papers at the conference. In his view McGouldrick does not explain Germany's gold standard experience in light of current models of the international economy that dismiss both specie flow and monetary theory of the balance of payments as inappropriate. Lindert expressed doubt that Sweden's successful performance was related to the gold standard, arguing that economic growth there was attributable to the high level of human capital and abundant natural resources that would have attracted capital inflows also under flexible exchange rates. Accordingly, he suggested that Sweden remained on the gold standard because the country grew rapidly, not vice versa. Sylla found unacceptable the implication that money was a luxury good in Italy-a conclusion Fratianni and Spinelli reach from the moneydemand function they estimate. He was also skeptical that "country risk" was the explanation for deviations from purchasing-power parity that the authors calculate. Instead, he suggested that Italian prices in fact were world prices, adjusted to take account of falling internationaltransactions costs. How does one reconcile the finding that the gold standard as a fixedexchange-rate system performed well, although country after country seems not to have observed the rules required to remain on the gold standard? One possibility is that central banks violated the rules only to a limited extent. It is clear that cumulated deficits and surpluses in balance of payments occurred only when capital flows sustained them. The system did not break down as a result of such flows. Brunner suggested in discussion from the floor that in the context of well-established expectations that the gold standard would be maintained, whether required adjustment occurred primarily in the shares of traded and nontraded goods, in long- or short-term capital movements, or in substantial changes in relatives prices depended on the character of the shocks affecting individual economies-nominal or real, transitory or permanent. It is easy for present-day observers to assume that central banks before 1914 were guided by macrostabilization goals, as in the post-World War II setting, but there is no firm basis for such an assumption. Whether formal central banks existed, as in Europe, or commercial banks or the Treasury exercised central-bank functions, their behavior was successful in maintaining the standard. Was it all waste motion, because the rules of

10

Anna J. Schwartz

the game were unimportant, as the monetary-approach-to-the-balanceof-payments theorists insist? This issue the conference did not settle. Clearly, discretionary actions were taken. Whether they served to speed adjustment was again left an open question. 0.1.4

International Linkages under the Gold Standard

The papers on linkages under the international gold standard approach the question from a number of perspectives. Huffman and Lothian examine interrelations between the United States and the United Kingdom for the century from 1833 to 1932. They conclude from a historical analysis that cyclical fluctuations were transmitted from one country to the other either by gold movements or by panic-induced changes in the money multiplier. From a battery of Granger-Sims autoregressive tests, they conclude that real income in both countries was influenced by both domestic and other-country variables. In addition, they point to weak links between U.S. and U.K. price-level movements as evidence against the monetary approach to the balance of payments. Connolly, in commenting on the paper, found the evidence against the monetary approach questionable since the Granger-Sims tests the authors rely on omit contemporaneous variables that might reveal effective price arbitrage. Easton, on the other hand, using data for eight countries for 18791914, finds no evidence in bivariate relations that either real or nominal income in one country provided any information on those variables in another. Geoffrey Wood argued that Easton's finding confounds demand and supply shocks under a fixed-exchange-rate system. Demand shocks produce positive correlations, supply shocks, negative correlations between income movements across countries. By examining the period as a whole, rather than individual episodes, Easton's approach ensures that such relationships as existed would not be found. As Wood pointed out, a predictable monetary system imposes no particular systematic behavior pattern on the real economy. Thomas was troubled by the failure to test properly for the Atlantic-economy relationships that he had investigated, not only in Easton's work but also in several other papers that had been presented. Meltzer's point in discussion from the floor bore on the importance of the institutional framework under the gold standard that the papers seemed to neglect. Not only was there a predictable monetary system, but decisions in all markets could be made with firm expectations that, for example, price controls or other arbitrary measures would not be imposed. Rich finds that the price-specie-flow mechanism operated over the long run but not over the short run in pre-World War I Canada, concluding that the failure of the mechanism in the short run was an important cause

11

Introduction

of cyclical instability under the gold standard. Thus a balance-ofpayments deficit that led to a gold outflow, a decline in the monetary base, and a rise in interest rates would induce banks to reduce their reserve ratios and expand the money supply, impeding cyclical adjustment. As evidence, Rich cites lack of correspondence between Canadian and U.S. interest rates and the lower variance of Canadian-relative-toU.S. interest rates. Temin asked why U.S. and Canadian interest rates did not move together, as one would expect if asset markets were unified. Possible explanations that he offered were data problems or noncompetitive behavior by Canadian banks. The studies of international linkages only scratch the surface of the subject. Examination of effects of transmission on national product accounts unavoidably involves the use of questionable data. Annual interpolations of benchmark figures, on which most pre-World War I nominal- and real-income data are based, raise doubts about the reliability of the estimates and the statistical significance of tests utilizing those data. Apart from the reliability of the data, we still need to learn how transmission occurred. The studies do not provide a systematic investigation of the role of transmission instruments. The instruments could have been gold flows, commodity-trade flows, capital flows, interest rates, or monetary flows. Did transmission occur through nominal linkages? If so, how was the division of nominal changes between prices. and output determined in the country that was the recipient of the transmission? Did the division vary from country to country, or were there trend influences that operated jointly on all gold standard countries, such as the secular price movements from the mid-1870s to 1896 and from 1897 to 1913? 0.1.5

The Stability of Price-Level Trends under the Gold Standard

Plots of wholesale prices before 1914 seem to be characterized by well-defined, alternately declining and rising trends. The traditional explanation has focused on the commodity theory of money. A decline in the trend of the price level reflected a more rapid growth of world real output and hence in the demand for monetary gold than the growth in the world's monetary gold stock could accommodate. The movement in the price level induced a shift from nonmonetary to monetary uses of gold and ultimately led to increased gold production. A rise in the trend of the price level reflected more rapid growth in the world's monetary gold stock than in the demand for monetary gold, inducing a shift from monetary to nonmonetary uses of gold and ultimately to decreased gold production. The widely accepted view of persistent trends in the price level under

12

Anna J. Schwartz

the gold standard was subjected to strong attack by Benjamin and Kochin, who argue that U.K. prices display the characteristics of a random walk once systematic movements are eliminated by the ARIMA technique. Similarly, the yield on consols after elimination of systematic movements is essentially a random walk. They dismiss Irving Fisher's finding that a distributed lag on past inflation rates was positively correlated with current interest rates, since such a method of forecasting would be rational only if inflation rates were positively correlated serially. They report that such was not the case and conclude that the effects of war expenditures explain comovements of the price level and interest rates. Once the influence of wars is accounted for, virtually no evidence remains of a linkage between the change in the price level and the change in long-~erm interest rates. Cagan's comments on the paper provide grounds for skepticism with regard to the authors' conclusions. Under a gold standard, prices were not in fact completely stable from year to year, despite the assurance of long-run price stability that a commodity standard is said to provide. Contemporaries might not immediately have recognized that a shift of the trend in prices, say, from falling to rising, had occurred. Such recognition would gradually develop, even though year-to-year changes would be regarded by them as random. Corresponding movements in bond yields would not then be accidental but a gradual response to price trends. Rockoff examines the response of gold production to movements in the relative price of gold before 1933. Gold output responded to market incentives in line with the commodity theory of money, which posits an inverse relationship between the general price level and the opportunity cost of producing gold. The incentives affected the timing of gold discoveries and advances in mining techniques, and influenced the willingness of producers to adopt existing capital-intensive methods of production. Imposition or relaxation of governmental environmental regulations could thwart or support market incentives. Rockoff also compares the means and standard deviations of annual growth rates of the world's monetary gold stock from 1839 to 1929 with those of the U.S. monetary base from 1949 to 1979. The gold standard regime gives more stable results. As Barro notes in his comments, the monetary-gold-stock results do not carryover to broader monetary aggregates and the price level, in a comparison with the post-World War II regime. However, the more favorable results for the post-World War II period may reflect alterations in banking institutions rather than a 'shift from the gold standard. Had these alterations been implemented during the gold standard era, the year-to-year stability of broader aggregates and prices might have surpassed the results for the post-World War II era.

13

Introduction

An existing (Bordo 1981) study compares real-output stability under the gold standard in the United States and Great Britain with the corresponding measure under the managed currency systems that superseded the gold standard. A desirable addition to the conference would have been similar studies for other countries.

0.2

Changing Professional Assessments of the Gold Standard

Looking back over world experience with monetary systems in the nineteenth century and its sequel in the twentieth century, one is struck by fluctuations in the esteem with which economists have regarded a metallic standard. After the widely known examples of paper-money inflation that occurred in the closing decades of the eighteenth century and the first decade of the nineteenth century (in the U.S. colonies, in France during the Revolution, and in Britain during the Napoleonic era), the superiority of a metallic standard seemed self-evident in theory and in practice. In both established and newly created nation-states in the nineteenth century, the evolution of monetary systems usually proceeded with the displacement of silver as the monometallic standard or the consort of gold in a bimetallic standard. The norm evolved as free and unlimited coinage of gold with subsidiary coins of silver, nickel, and bronze or copper, and government fiduciary issues and bank notes freely convertible into gold. In less-developed countries, convertibility was provided by foreign-exchange reserves linked to gold. A paper standard, by contrast, came to represent fiscal imprudence and economic backwardness. Even before the end of the nineteenth century, however, popular and professional criticism of the gold standard arose. What occasioned the criticism was the secular price rise associated with the midcentury gold discoveries and the long, secular price decline that got under way in the 1870s under an expanding international gold standard. The first challenge to the virtue of the gold standard was that it did not assure price stability. In his pamphlet, "A Serious Fall in the Value of Gold Ascertained, and Its Social Effects Set Forth" ([1863] 1884), William Stanley Jevons estimated that between 1848 and 1860 the value of gold had fallen 9 percent. In 1875 he questioned the use of metallic standards of value, in view of the extreme changes in their values, and urged as a reform a tabular standard of value ([1875] 1884). Alfred Marshall ([1887] 1925) discussed "the evils of a fluctuating standard of value" (p. 189), and concluded that "the precious metals cannot afford a good standard of value" (p. 192). He dismissed bimetallism as flawed and proposed as a remedy for the fluctuating standard of value either symmetallism or a tabular standard. With the reversal of the secular price movement after 1896, concern

14

Anna J. Schwartz

shifted to the inflationary,fluctuation of the standard. The remedy that Irving Fisher (1913) proposed was the compensated dollar. The gold standard ceased to function internationally during World War I, and the question of its merits or demerits was temporarily set aside. In the aftermath, inflation in the victorious countries and hyperinflation in the vanquished, as governments financed wartime and postwar expenditures by depreciating their currencies, again revived the attraction of the gold standard. Widely reintroduced in the years 1925 to 1929 (although attenuated by the cessation of gold-coin circulation and the limitation of convertibility to bullion bars or sales of foreign exchange), the gold standard collapsed shortly thereafter, destroyed by the economic holocaust of 1929-33. This time the main professional attack was directed to fixed exchange rates through which the gold standard works, although attention also focused on specific problems that were identified as hampering the operation of the post-World War I gold standard (the maldistribution of gold, the inadequacy of world gold output, and the poorly aligned exchangerate structure that had been restored). Fixed exchange rates required the internal economy to adjust to the balance of payments. Only by cutting loose from the gold standard were countries able to escape the deflationary pressure imposed on them by the fixed-exchange-rate system. Internal adjustment to declining world prices was no longer acceptable domestic economic policy, and growing rigidity of prices and costs allegedly placed an intolerable burden of adjustment on the economy. Moreover, far from correcting externally arising disturbances, the gold standard fostered them by transmitting maladjustments from one country to another. An additional problem under the gold standard, according to its critics, was that capital movements, short-term ones in particular, did not provide a corrective mechanism but instead aggravated the underlying situation that generated the capital flows (see Bordo, appendix E, this volume). The flows, in effect, were uncontrollable. Raising the discount rate had not stopped capital flight but had intensified it; the rising rate was interpreted as a signal that further flight would lead to devaluation. At the same time, the discount-rate rise had served to heighten deflationary pressures on the domestic economy. On the other hand, a discount-rate rise that was expected to curb internal expansion instead attracted capital from abroad and promoted further expansion. Capital movements, under fixed exchange rates, induced by interest-rate changes, operated primarily on reserves and foreign exchange of the central bank but did not immediately induce changes in the current account. Fundamental adjustment, moreover, was deterred when long-term capital exports were offset by short-term capital imports. Alternatively, when long-term capital exports ceased, the capital-importing countries confronted fixed-

15

Introduction

interest charges with deflationary impact on their economies, with reflex influence on the capital-exporting, interest-receiving countries. The gold standard thus was charged with having contributed to the instability of the world economic system after 1929. Professional approbation of a paper standard that gained ground in the 1930s was tempered by the belief that unrestrained, it would encourage beggar-thy-neighbor policies. The Bretton Woods arrangements embodied the interpretation of the views and experience of the 1930s-pegged exchange rates were essential to prevent chaos in international financial and trade transactions, but national economies should be free to restrict capital flows and to resort to the expedient of devaluation in order to be relieved of the necessity to deflate when in current-account deficit. The objectionable feature of pegged rates in forcing governments to implement monetary changes that conflicted with the goals of full employment or price stability would be removed while preserving the desirable feature of providing stable conditions in foreign exchange to promote international trade. Convertibility of many European currencies was first achieved under the Bretton Woods system in 1958. For only a few years thereafter can the system be said to have functioned fairly effectively. From the mid-1960s on, it was characterized by repeated foreign-exchange crises as market participants anticipated that existing par values were unsustainable and shifted funds from a weak currency to a strong currency, exacerbating the external position for both currencies. Since the collapse of the Bretton Woods arrangements, efforts to rehabilitate the gold standard have proceeded along two lines. One, inspired by the professional development of the monetary approach to the balance of payments, argues in favor of fixed exchange rates as a way to attain the benefits of risk-pooling and the integration of commodity and factor markets on a worldwide basis. The other line is drawn from the collapse of the Bretton Woods arrangements. The lesson, on this view, is not that only a floating-rate system can accommodate inflationary policies in the reserve-center country and conflicting policies in the nonreserve countries. Rather, the lesson is that the floating-rate system has permitted enormous growth of inconvertible paper-money issues that produced unprecedented peacetime inflation rates and extraordinary levels of interest rates. Consequently, it is argued, it is essential to establish a stable international money based on gold. The advocates offer varying prescriptions. One would rely on the changing market price of gold as an indicator to the monetary authorities of the appropriate rate of increase or decrease in the growth of the money supply, with no commitment on their part to buy or sell gold or to peg its price. Although the price of gold would thus playa part in the monetary system, it would lack crucial elements of gold standards known in the

16

Anna J. Schwartz

past. Another prescription includes stabilizing the dollar price of gold, issuance of gold coins with a face value equal to the stabilized gold parity, restoration of convertibility by linking change in money bases-in the United States to gold purchases and sales in a private gold market, and in non-reserve-center countries to changes in their holdings of gold and foreign exchange-and, finally, multilateral surveillance of country balance-of-payments problems. A more radical prescription would eliminate government-issued money. The government's role would be limited to defining a monetary unit as a specific weight of gold. Private issuers would then be free to issue claims denominated in the officially defined unit. This section suggests that support for a resurrected metallic standard of whatever form would in time dissolve, as it has for all earlier standards. What the odds are for success in the restoration of a role for gold is the subject of the concluding section. 0.3

Prospects for Reinstating the Gold Standard

The conference studies deal with historical evidence-obviously necessary to our understanding of the gold standard as it once existed. That evidence also directs our attention to the possibility that the factors that permitted the gold standard to flourish are now obsolete. What were those factors? Can we now re-create them? We can distinguish at least seven objective factors that promoted the existence of an international metallic standard: 1. the essentially fixed price of gold over the century the conference studies covered 2. a link between domestic money supply and the gold reserve 3. relative stability in conditions of gold production 4. equilibrium in mint pars among gold standard countries 5. coordination of economic policies among countries adhering to the standard 6. limited role of government in economic and social affairs 7. relative absence of political upheavals exemplified by war and revolution and the role of London as the hub of the international monetary system These objective factors were stabilizing forces that made the gold standard a stable standard of value. Whether the stability was an inherent feature of the gold standard or simply the consequence of underlying stability of other institutions is the issue. In addition to the objective factors, mention must also be made of the weight of the psychological belief in the unquestioned and unquestionable obligation to adhere to the gold standard and to the specific fixed price of gold. Flouting the gold standard risked the opprobrium of one's

17

Introduction

own countrymen and of the rest of the world. Political leaders did not regard the gold standard as a policy instrument subject to manipulation in the pursuit of other goals. The hold of the gold standard as the guarantor of the domestic value of a currency and of stable international financial dealings was sacrosanct. Can we count on the stability of the objective factors in contemporary economic circumstances? 0.3.1

The Price of Gold

A fundamental problem confronting the reinstatement of the gold standard is the choice of the dollar price at which to resume. The very conception of trying to determine the correct price somehow violates the mystique of the standard. The price then becomes a political decision, the opposite of the freedom of the standard from political influence that untlerlay its mystique. For the purpose of this analysis, assume the following solution: let the inflation rate of the general price level be reduced to zero; the price of gold at that time would be the correct price at which to resume. Once a price for gold is determined, the principal central banks, it has been suggested, should proceed to peg it. To prevent the gold price from rising, sales of gold from existing stocks could be used. To support the price, countries could use their own currencies-with possible inflationary consequences. Assuming the price were "correct," the pegging operation might be successful. Arranging the responsibility for intervention in the gold market could be managed along the lines of the Gold Pool of 1961, provided exchange rates did not vary. If they did, since an exchange-rate change is a gold-price change in at least one country, speculation in gold markets would be encouraged. The pegging operation would then become more troublesome. 0.3.2 Linking the Domestic Money Supply and Gold Reserves A pegged price of gold is not a sufficient condition for a reinstatement of the gold standard. Some link between the domestic money supply and a country's gold reserves is essential. Would it be feasible to restore convertibility of paper currency into gold for domestic and foreign holders? Countries would be required to yield the discretion they currently exercise in determining the level and growth rate of their domestic money supplies and to accept the effects on money supply that changing gold reserves would dictate. Would they be willing to accept so severe a restriction on their internal monetary policies? 0.3.3

Stability of Gold Output

If a correct price of gold were achieved for resumption, the stability of the price level under the gold standard thenceforth would then depend on

18

Anna J. Schwartz

the adequacy of gold output to provide for monetary and nonmonetary demands for gold. An adequate supply of gold is essential for adequate monetary growth. The forecasts of gold output over the rest of the century in the market economies with known gold reserves are not optimistic. Whether the forecasts might be belied by discovery of new mines or mining processes and whether the inadequacy of the flow supply might be offset by changing patterns of industrial demand for gold or shifts from investment stocks still leaves the reinstatement of the standard as a measure that risks imposing long-run deflation on the economy. The fact that the bulk of current world gold output is produced by South Africa and the Soviet Union adds an element of possible instability in future gold output for political reasons. Rockoff suggests another difference between the past and the putative future performance of the gold standard related to the gold-mining industry, namely, that the public is unlikely to tolerate long and uncertain lags in the response of the gold supply to the changing demand for money. This difference possibly could be classified as psychological, but, if accurate, it clearly impinges on an objective factor. 0.3.4 Fixing Multilateral Exchange Rates Once a correct fixed price of gold were chosen, each gold standard country would adopt par rates of exchange for its currency relative to other currencies. As Yeager remarks, the mint pars under the classical gold standard expressed an equilibrium that had gradually evolved among national price levels. This time, par rates of exchange would be arbitrarily chosen. The mistakes in choice of exchange rates when Euro'pean countries resumed in the decade of the 1920s and again under the Bretton Woods arrangements are not reassuring. 0.3.5

Coordination of-National Economic Policies

The gold standard can survive in a world in which countries allow gold to move freely; gold does not accumulate in any country and gold does not drain away from any country without being allowed to exercise an expansionary or contractionary effect, respectively, on the level of prices; and major disequilibria in price levels and financial conditions among countries are not endured. The forces that caused the breakdown of the Bretton Woods system were unleashed by actions of countries with a persistent deficit or surplus in their balances of payments. Those actions were taken to delay or resist changes in prices and costs expressed in national currencies. Under fixed exchange rates, convergence of national economic policies is essential for the system to be viable. The European Monetary System presupposes such behavior. Yet since 1979 when the system was established, member countries have repeatedly preferred to alter the relation between national price and cost levels by exchange-rate

19

Introduction

changes. This is not a good augury for restoration of an international gold standard. 0.3.6

Role of Government

Under the classical gold standard, governments in peacetime did not undertake expenditures that were financed by the printing press. In some gold standard countries, government was not divorced from business and social insurance was accepted policy. Basically, however, government participation in economic activity was restrained by concern to preserve the integrity of the national currency and to maintain its domestic and external value. These concerns receded after 1929 as governments extended their. activities to finance stabilization policies in response to interest groups wielding political influence. The question then arises whether in the future governments will reverse their course, returning to a more limited role, as in the pre-World War I era. Of course, a limited role of the state is not in itself a guarantee of a viable international monetary system, since in earlier eras international monetary affairs were often in disarray, even with limited states (Dam 1982, p. 38). 0.3.7

Civil and International Peace and London's Predominance

The gold standard collapsed when countries were engulfed by war or revolution. The relative political stability of the pre-1914 era therefore contributed to the maintenance of the standard. The significance of this factor is underscored by the prewar examples of capital flows that were steered by governments for national political and strategic reasons. French investors responded to official regulation and pressure by buying Russian government loans for railroad construction-of military value, in the eyes of both governments. Germany's foreign investments also were directed to achieve national-security goals. The ensuing war destroyed not only the gold standard but also the investments. The gold standard flourished before World War I possibly because of the special position of sterling and London. That position was threatened even before the war when Paris and Berlin became important rivals of London. Thereafter, London's predominance was never reestablished. Under the Bretton Woods system, the special position was that of the dollar and the United States. As the position of the U.S. dollar crumbled, the system collapsed. Is an important aspect of the successful operation of a gold-centered monetary system an unshakable confidence that a dominant reserve-currency would always be converted into gold on demand? Which currency would be the candidate for such a role in a future gold standard? This brief survey suggests that the objective factors that served to promote the international gold standard in the past are no longer favorable to such an institution. And, as noted, the psychological factor of

20

Anna J. Schwartz

reverence for the standard has all but vanished except among a minority of faithful believers. Like Miniver Cheevy, they probably were born too late.

References Bloomfield, Arthur I. 1963. Short-term capital movements under the pre-1914 gold standard. Princeton Studies in International Finance, no. 11. Princeton: Princeton University Press. Bordo, Michael. 1981. The classical gold standard: Some lessons for today. Federal Reserve Bank of St. Louis Review 63: 1-17. Dam, Kenneth, W. 1982. The rules of the game. Chicago: University of Chicago Press. Fisher, Irving. 1913. A compensated dollar. Quarterly Journal of Economics 27 (Feb.): 213-35, 385-97. Ford, Alec G. 1962. The gold standard, 1880-1914: Britain and Argentina. Oxford: Clarendon Press. Friedman, Milton, and Anna J. Schwartz. 1963. A monetary history ofthe United States, 1867-1960. Princeton: Princeton University Press. Jevons, William Stanley. [1863] 1884. A serious fall in the value of gold ... set forth. Reprint. In Investigations in currency and finance, ed. H. S. Foxwell. London: Macmillan. - - . [1875] 1884. An ideally perfect system of currency. Reprint. In Investigations in currency and finance. See Jevons [1863] 1884. Marshall, Alfred. [1887] 1925. Memorials of Alfred Marshall. Ed. A. C. Pigou. Reprint. London: Macmillan. Michaely, Michael. 1971. The responsiveness of demand policies to balance ofpayments: Postwar patterns. Studies in International Relations, no. 5. New York: National Bureau of Economic Research. Morgenstern, Oskar. 1959. Internationalfinancial transactions and business cycles. Princeton: Princeton University Press. Sayers, Richard S. 1936. Bank of England operations, 1890-1914. London: P. S. King and Son. Thomas, Brinley. 1973. Migration and economic growth. 2d ed. Cambridge: Cambridge University Press.

1

The Gold Standard: The Traditional Approach Michael D. Bordo

1.1

Introduction

What was the traditional approach to the gold standard? In this paper, I try to provide an answer to the question by examining the works of major writers on the subject since the eighteenth century. 1 The choice of writers and works surveyed is based on my judgment that the works encompassed a significant share of the content of the traditional approach and that the writers played a significant role in the history of economic thought. Six major themes formed the traditional approach, and five major schools of thought may be identified. 1.2 Major Themes in the Literature The first theme, which runs from Cantillon to present-day writers, was that gold (the precious metals) was an ideal monetary standard, domestically and internationally, because of its unique qualities both as a standard of value and a medium of exchange. A stable price level in the long run that an automatically operated gold standard produced, in line with the commodity theory of money, was invariably contrasted to the evils of inconvertible fiduciary money. At the hands of even well-meaning policymakers the latter would inevitably lead to depreciation of the value of money. However, most writers, following Adam Smith, emphasized the social saving from using fiduciary money instead of a commodity money Michael David Bordo is professor of economics at the University of South Carolina and a research associate of the National Bureau of Economic Research. For helpful comments and suggestions the author would like to thank Michael Connolly, John McDermott, Peter Lindert, Anna Schwartz, and Larry White; he would also like to thank Fernando Santos and Glen Vogt for able research assistance.

23

24

Michael D. Bordo

and hence were concerned with the properties of a convertible (or mixed) standard to ensure price stability. The second theme was the price-specie-flow mechanism. The essence of the gold standard was the maintenance of a fixed mint price of national money in terms of gold (achieved by specifying the weight of a nation's coinage in terms of gold). That rule ensured uniformity of the price of gold across nations (and regions) through the process of arbitrage in gold. 2 Moreover, each country's price level was determined by its stock of monetary gold, which in turn was determined (naturally distributed) by the nation's real income and money-holding habits. Consequently, the price levels of all countries were linked together under the gold standard by the fixed definition of the monetary unit in terms of gold. Any disturbance away from the natural distribution of precious metals affecting one nation's (region's) price level, and hence the market price of gold, would inevitably lead to an equilibrating process through arbitrage in the gold market. Gold flows, by changing the nation's (region's) money supply, would then also change its level of prices. For example, a gold discovery in one country would lead to an increase in its money supply, an increase in its price level, a~d a fall in the domestic market price of gold. The divergence between the domestic and world gold prices would quickly lead to a gold outflow, a contraction in the domestic money supply, and a fall of the domestic price level. An alternative way of viewing the same mechanism was to focus on the balance of trade-the rise in the domestic price level would raise prices of domestic goods and exports relative to prices of imports, leading to a balance-oftrade deficit, a gold outflow, and a contraction of the money supply. Thus the price-specie-flow mechanism was the means by which arbitrage in one commodity-gold-between nations and regions, served to keep overall national (regional) price levels in line and to maintain balance-of-payments equilibrium. Within this context, different authors stressed the pattern of adjustment of particular classes of commodities. Thus Mill focused on the behavior of the prices of tradable goods relative to those of domestic (nontradable) goods. Others focused on the secondary role of changes in the exchange rate. To the extent that gold prices between nations could differ, reflecting transportation and other costs of transferring gold (the difference between the upper and lower bounds referred to as the gold points), changes in exchange rates (the domestic relative to the foreign price of gold) would also serve to equilibrate the balance of payments without requiring a gold flow. In addition, a number of writers focused on the role of real income in the adjustment mechanism-ehanges in the quantity of money consequent upon gold flows would affect total expenditure and income in addition to, or in some cases instead of, affecting prices.

25

The Gold Standard: The Traditional Approach

The third theme, which is intimately connected to the second, was the "law of one price"-the notion that arbitrage in individual traded commodities would ensure similar prices in a common currency for similar goods, taking account of transportation costs and trade impediments. Along these lines, a distinction was made between domestic (nontraded) goods whose prices are determined primarily by domestic forces and traded goods whose prices are determined by the world mark.et. One question is how to reconcile the law of one price with the pricespecie-flow mechanism, since the latter stressed primarily consequences of arbitrage in gold, while the former stressed arbitrage in all traded commodities. For the classical economists, it was assumed that arbitrage in gold was more effective than in other commodities because of gold's special properties; moreover, since gold served as the money supply (or as the monetary base), alterations in its quantity would impinge on all prices. Ultimately, which goods serve as vehicles for arbitrage is an empirical question. The answer depends on the total costs of arbitrage, including information costs. In the eighteenth and early nineteenth centuries, gold was the commodity with the lowest arbitrage costs, hence gold flows rapidly kept gold prices in line and other goods prices followed. Later in the nineteenth century, with improvements in communications technology and the development of international securities and commodity markets, arbitrage in securities and traded commodities reduced the role for gold flows in the adjustment mechanism. The fourth theme was the role of capital flows in the gold standard balance-of-payments adjustment mechanism. The original conception of the price-specie-flow adjustment mechanism was that it operated through flows of goods and money, but by the middle of the nineteenth century, emphasis was also placed on the role of short-term capital flows as part of the equilibrating mechanism. According to the traditional approach, a decline (rise) in the domestic money stock led to a rise (fall) in short-term interest rates and consequently attracted funds from abroad. Thus in the example of a gold discovery, the increased money supply would reduce domestic interest rates relative to interest rates in other countries, producing both a short-term capital and gold outflow, thereby reducing the amount of adjustment required through changes in the domestic price level. As the nineteenth century wore on and world capital markets became more integrated, emphasis on the role of capital mobility increased to the point where it was regarded as the dominant adjustment mechanism. In addition to short-term capital flows, the role of long-term capital flows was noted as a source of disturbance to the balance of payments. Thus one element of the traditional approach was the role of long-term lending by mature countries, such as England and France, to developing

26

Michael D. Bordo

nations, such as the United States, Canada, and Argentina. Capital flows from the Old to the New World were also accompanied by gold flows, raising the price level in the capital-importing country and lowering it in the exporting country. The resultant change in relative price levels produced a current-account surplus in the capital-exporting country and a deficit in the importing country. Thus the transfer of capital resulted in a transfer of real resources. 3 The process could continue for many years, with developing (developed) countries running a persistent balance-ofpayments deficit (surplus) on current account financed by long-term capital inflows (outflows). The fifth theme, which focuses primarily on the performance of the Bank of England, was the role of central banks in helping or hindering the adjustment mechanism. This theme was a reflection of the British flavor of the gold standard literature and the key role played by the Bank of England in the analysis of the gold standard. Several aspects of the central-bank theme may be noted. One was the debate over rules versus discretion. In the early part of the nineteenth century, emphasis was placed on the advantage of combining the automatic-monetary-rule aspect of the gold standard with the benefits of low-resource-cost fiduciary money. That approach culminated in the Bank Charter Act of 1844 and the separation of the Bank of England into the Issue Department, based on a gold standard rule, and the Banking Department, based on commercial-banking principles. Second, several money-market crises and threats to convertibility in the succeeding quar.ter century led to attention in the literature to the Bank's disregard of domestic-money-market conditions in its operation as a private profit-maximizing institution following a gold standard rule. Thus the Bank, in keeping with its private role, would maintain as Iowa gold reserve as possible while using its Bank-rate weapon to protect its reserve from gold outflows. Bagehot's statement of the "responsibility doctrine" and a prescription for effective central-bank management, referred to as Bagehot's rule, emerged from the scrutiny of the Bank's behavior. A later development was the discussion of the inherent conflict between internal and external price stability under a fixed exchange rate such as the gold standard. In addition, the gold standard came to be regarded as primarily managed by central banks' use of changes in the discotmt rate to facilitate adjustment to both internal and external gold drains. Among the issues stressed were: how Bank rate was made "effective," in the sense of inducing corresponding changes in market interest rates; the use of other policy tools to protect the gold reserves; the channels by which changes in Bank rate would affect the required adjustment in the balance of payments-by inducing short-term capital flows or by changing domestic price levels, economic activity, and the terms of trade.

27

The Gold Standard: The Traditional Approach

Finally, discussion turned on the extent to which central banks followed the "rules of the game," that is, used their policy tools to speed up the adjustment mechanism to an external shock. According to the rules, the central bank of a country experiencing a gold outflow (inflow) should engage in policies to contract (expand) the domestic money supply. The sixth and final theme in the traditional approach was the advocacy of a number of proposals for reform. Many writers suggested schemes for reform of the gold standard both at the national and international levels. At the national level, a persistent theme ranging from Thornton ([1802] 1978) to Keynes ([1923] 1971) was the importance of managing the gold standard so as to reduce the conflict between external and internal stability, i.e., for the central bank to intervene and shield the domestic money supply from external shocks. Related to this theme were schemes to protect the monetary gold stock from internal currency drains, e.g., Ricardo's gold-bullion standard. Finally, schemes were designed to separate the medium-of-exchange function of gold from the store-of-value function. All these proposals attempted to rectify an important defect of the gold standard-basing a nation's money supply on one commodity subject to changing demand and supply conditions. Schemes along these lines included creation of a tabular standard, bimetallism, symmetallism, and Fisher's (1920) compensated dollar. At the international level, proposals designed to provide world price stability included schemes such as bimetallism, symmetallism, and the basing of international money on a wide commodity basket; and also, to ensure international harmony of price-level movements, they favored the creation of some form of supernational central bank. 1.3

Schools of Thought

On the basis both of common views and chronology, the five schools of thought on the gold standard are the classical school, the neoclassical school, the Harvard school, the interwar critics, and the post-World War II reinterpreters. A brief summary of the views of the leading exponents of each school follows. Detailed documentation of these views is provided in five appendixes, one for each school. 1.3.1

Classical School

Eight economists-Cantillon, Hume, Ricardo, Thornton, Mill, Cairnes, Goschen, and Bagehot-eonstituted the classical school. From the writings of these men we can distill the essence of the traditional approach. Cantillon developed the law of one price and aspects of the international adjustment mechanism. Hume is famous for the pricespecie-flow mechanism. Ricardo developed the natural distribution of precious metals and made contributions to issues related to the monetary

28

Michael D. Bordo

standard and monetary reform. Mill, perhaps the key writer of the school, covered virtually all the major themes of the traditional view, and Cairnes tested some of the theoretical implications. Finally, Goschen focused on the role of short-term capital flows, while Bagehot outlined the principles of central-bank management under the gold standard. 1.3.2 Neoclassical School Marshall, Fisher, and Wicksell of the neoclassical school extended and perfected the mechanisms analyzed by the classical school. They, however, explored some of the detrimental effects, both for individual nations and for the world, of adhering to the gold standard, and consequently the need for reform. 1.3.3 Harvard School F. W. Taussig and his students (Viner, Graham, White, Williams, and Beach) attempted to formulate and test a more comprehensive version of the traditional balance-of-payments adjustment mechanism to the external disturbance of long-term capital movements by incorporating gold flows, changes in relative price levels, short-term capital flows, and changes in discount rates. The evidence for the United States, Great Britain, France, Canada, and Argentina produced by this massive research project was largely inconclusive, and in many respects cast doubt on the traditional emphasis on relative price-level changes as the heart of the adjustment mechanism. J. W. Angell, a critic of Taussig, integrated the law of one price in the relative price-specie-flow adjustment mechanism. Despite its critical approach, his work is classified as part of the Harvard-school studies. 1.3.4 Interwar Critics After World War I, a number of writers considered the case for and against a return to the gold standard as it existed pre-World War I. Brown and Smit, accepting in the main the stylized facts of the gold standard as succinctly portrayed by the Cunliffe report (United Kingdom, Parliament [1918] 1979), assessed the gold standard as having been successful before World War I because it was a managed standardmanaged by London-and then documented the special institutional characteristics of the sterling standard. Keynes and Viner discussed the inherent policy conflict between adherence to the gold standard and domestic economic activity, and addressed a plea for more international cooperation. Whale cast doubt on the stylized facts of how the gold standard worked, suggesting that perhaps the traditional approach was incorrect.

29

The Gold Standard: The Traditional Approach

1.3.5

Post-World War II Reinterpreters

In the post-World War II period, scholars reexamined the operation of the classical gold standard on the basis of new evidence and new theoretical and statistical tools. The issues they stressed included the balance-ofpayments adjustment mechanism, capital flows, and rules of the game. The balance-of-payments adjustment mechanism under the gold standard was reexamined from a Keynesian perspective by Ford, from a modern quantity-theory perspective by Friedman and Schwartz, and from the perspective of the monetary approach to the balance of payments by Williamson, Triffin, and McCloskey and Zecher. The role of capital flows was reexamined by Morgenstern and Bloomfield. Finally the operation of central banks under the gold standard with respect to rules of the game was reconsidered by Sayers, Bloomfield, and Lindert. 1.4 A Retrospective The development of the literature on the traditional approach can be viewed from a number of perspectives. I briefly sketch out the elements of two of them: the first, that the interpretation of the gold standard by each school reflected the policy concerns of the time; the second, that the evolution of the interpretation of the gold standard has many of the characteristics of a Kuhnian scientific revolution. According to the first perspective, the development of the traditional approach by the classical economists was strongly influenced by the concern over finding the ideal monetary standard consistent with the classical principles of free enterprise and free trade. This concern thus explains the emphasis on the automatic qualities of the gold standard both as a national and an international standard, the operation of the commodity theory of money that would ensure long-run world price stability (in a stationary world), and the price-specie-flow mechanism that would ensure the natural distribution of precious metals and uniformity of price structures across the world. Behind this smoothly functioning monetary veil, real resources would be efficiently allocated to their best uses by the forces of competition between individuals and enterprises across the world. The introduction of the real-world problems of friction in the balance-of-payments adjustment mechanism and the possible conflict, at least in the short run, between the constraint of the gold standard and internal economic stability led to the development of rules of proper central-bank management of the gold standard. The neoclassical economists, writing at a time when the gold standard was the prevailing standard, accepted its rationale, but concerned themselves with removing one of its major shortcomings-specifically, the tendency for the world price level to exhibit alternating swings of defla-

30

Michael D. Bordo

tion and inflation, reflecting major shocks to the demand for and supply of gold. The Harvard economists, like the neoclassicists, writing about the heyday of the gold standard, sought a better understanding of the mechanism by which one of the most important structural changes in modern economic history took place-the transfer of real resources associated with massive lending by the mature countries of Western Europe to the developing countries of the New World. The interwar critics, writing after the collapse of the gold standard, yet strongly influenced by its heritage, were concerned with the possibility of restoring the old system. Much of their work reflected skepticism on this score because for them "special circumstances" in the prewar period made the system work: the unique interrelationship between the London gold, securities, and commodities markets that created a "sterling standard"; the commitment by major participants to maintain convertibility as their key policy goal; and relatively free trade and factor mobility. These special circumstances no longer existed. Other interwar critics focused on the negative aspects of the gold standard: the tendency for short-run price instability, the asymmetry between the adjustment mechanism of central and peripheral countries, the conflict between external and internal stability, and the tendency for economic fluctuations to be transmitted internationally by the gold standard. Hence these interwar critics doubted the wisdom of returning to the standard's iron discipline. However, in elaborating proposals for a better system, the consensus favored maintaining a fixed-exchange-rate system based on gold, with expanded national discretionary management and the establishment of a supernational central bank. Finally, in the postwar period, operating in an institutional environment far enough removed from the events before 1914, scholars of the gold standard could objectively ask how the gold standard in its many aspects worked. Armed with new theoretical and statistical tools and new compilations of data, the consensus of this work has been that the international gold standard did function smoothly in the sense of ensuring international price harmony, in allowing the international transfer of resources, and in maintaining balance-of-payments equilibria for most countries over long periods of time, but that many elements of the story-particularly the operation of the price-specie-flow mechanism and the importance of the rules of the game-were subject to doubt. According to the second perspective, the development of the gold standard literature reflected a Kuhnian scientific revolution (Kuhn 1970). Along this line, we start with the development of the classical-gold standard paradigm by the classical economists, culminating in the magnum opus of J. S. Mill. The paradigm was further extended and perfected by the neoclassical economists, especially Irving Fisher. However, anom-

31

The Gold Standard: The Traditional Approach

alies begin to appear by the end of the nineteenth century: the pricespecie-flow mechanism emphasizing the adjustment of relative price levels could not explain the actual adjustment process to international lending in a number of countries; in some cases the mechanism could be detected, in others the adjustment of price levels between countries seemed to be too rapid for the theory; and it appeared that many countries did not follow the rules of the game but engaged in extensive sterilization activities. However, the reaction to these anomalies by Taussig and others was to incorporate them into the theory as special cases. The assault on the classical paradigm began in the interwar period with the grave doubts raised by Keynes, Williams, Cassel, and others, but it was probably the 1937 article by Whale, challenging the whole classical interpretation of how the gold standard worked, that started the revolution. The further revelation of evidence inconsistent with the classical story in the postwar period added ammunition to the case presented by Williamson, Triffin, and finally McCloskey and Zecher. The last authors completely upended the classical paradigm and argued passionately that all aspects of the gold standard could be explained by the newly developed monetary approach. The scientific revolution was complete. In conclusion, we can ask: Is this the end of the gold standard story? McCloskey and Zecher, in tying together much of the unfavorable evidence against the traditional approach and then reinterpreting the facts to be consistent with the implications of the monetary approach to the balance of payments, make a strong case for a successful conclusion, except that the evidence they marshal in favor of their approach, based largely on correlation tests of commodity arbitrage, is neither extensive nor conclusive enough to end the story.

Appendix A

The Classical Economists

In this appendix the writings of eight key economists who first formulated the tenets of the traditional approach to the gold standard are summarized: 4 Cantillon, Hume, Ricardo, Thornton, Mill, Cairnes, Goschen, and Bagehot. Richard Cantillon Richard Cantillon ([1931] 1964), writing in 1755, was one of the first writers to analyze the working of a money economy. Operating within a crude quantity-theory-of-money framework ,5 Cantillon regarded the quantity of money as consisting entirely of specie-gold and silver coins. Gold and silver emerged as money commodities as a result of the evolution of natural market forces-they best satisfied the properties of

32

Michael D. Bordo

money, viz., they are "of small volume, equal goodness, easily transported, divisible without loss, convenient to keep, beautiful and brilliant in the articles made of them and durable almost to eternity" (Cantillon [1931] 1964, p. 111). Moreover, the choice between gold and silver (as well as the desired ratio in a bimetallic system) is determined by market forces-on the demand side by tastes and income, on the supply side by relative scarcity (pp. 97 and 277). In the long run, the world's monetary specie stock as well as its exchange value is determined by the foregone cost in terms of the land and labor required to extract precious metals. 6 In the short run, the two key sources of a nation's money supply are its balance-of-payments surplus and the presence of domestic gold and silver mines (bk. 2, chaps. 6, 7). Perhaps Cantillon's most important contribution was setting out a dynamic version of the quantity theory of money or what has often been referred to as monetary-disequilibrium theory. Cantillon carefully analyzed the dynamic process by which the quantity of money affected economic activity and the price level. An important element of his analysis was the international repercussions in the specie-standard world of fixed exchange rates of domestically induced monetary change: If more money continues to be drawn from the Mines all prices will owing to this abundance rise to such a point that . . . there will be a considerable profit in buying them [goods] from the foreigner who makes them much more cheaply. This will naturally induce several people to import many manufactured articles made in foreign countries, where they will be found very cheap: this will gradually ruin the Mechanics and Manufacturers of the State. (P. 165). That is, domestic inflation, by raising the prices of domestically-produced goods relative to foreign-produced goods (changing the terms of trade), will generate a balance-of-trade deficit. This deficit will induce a specie outflow, a reduction in the domestic money stock, and a reduction in domestic output and prices-the price-specie-flow mechanism. In addition to the terms-of-trade effect, the balance of payments will adjust by a direct-expenditure effect. According to this mechanism, an excess supply of money will cause domestic expenditures to exceed income; some of this expenditure will be made directly on foreignproduced goods (whose prices are determined abroad), leading to a specie outflow: It is usual in States which have acquired a considerable abundance of money to draw many things from neighbouring countries where money is rare and consequently everything is cheap: but as money must be sent for this the balance of trade will become smaller. (P. 169) Cantillon's final main contribution to the traditional view7 was a clear statement of the law of one price-eommodity arbitrage will ensure that

33

The Gold Standard: The Traditional Approach

the prices of similar traded goods will be the same across countries and across regions within countries, allowing for the influence of tariffs and transportation costs. The difference of prices in the Capital and in the Provinces must pay for the costs and risks of transport, otherwise cash will be sent to the Capital to pay the balance and this will go on till the prices in the Capital and the Provinces come to the level of these costs and risks. (P. 151) Moreover, he clearly distinguished between traded and nontraded goods on the basis of trade impediments and transportation costs. Thus the prices of traded goods that are determined abroad will be largely unaffected by domestic monetary conditions, whereas the prices of nontraded goods will respond fully, viz. In England it is always permitted to bring in corn from foreign countries, but not cattle. For this reason however great the increase of hard money may be in England the price of corn can only be raised above the price in other countries where money is scarce by the costs and risks of importing corn from these foreign countries. It is not the same with the price of Cattle, which will necessarily be proportioned to the quantity of money offered for Meat in proportion to the quantity of Meat and the number of Cattle bred there. (P. 179) This suggests that the distinction between the terms of trade and the direct-expenditure mechanisms rests on the distinction between tradedand nontraded-goods prices. The excess-money-induced expenditure falls on both traded and nontraded goods. To the extent the expenditure affects nontraded goods, their prices rise, inducing the substitution of traded goods. To the extent it affects traded goods whose prices are determined abroad, it leads to a direct specie outflow. Presumably, the effect on nontraded-goods prices will be short-lived-until substitution and the decline in the domestic money supply consequent upon the specie outflow have caused the relative prices of traded and nontraded goods to return to their initial equilibrium. David Hume In his essay, "Of the Balance of Trade" ([1752] 1955), Hume is generally believed to have originated the theory of the traditional balance-ofpayments adjustment mechanism of an international specie standard (see Viner 1937, pp. 291-92). According to Hume, a domestic monetary disturbance such as a sudden decrease in the specie stock will lead to a proportional decline in all prices and wages, a consequent decline in the prices of exports relative to the prices of imports, a balance-of-payments surplus, a specie inflow, and an increase in the domestic stock of specie.

34

Michael D. Bordo

Suppose four-fifths of all the money in Great Britain to be annihilated in one night, ... what would be the consequence? Must not the price of all labour and commodities sink in proportion . . . ? What nation could then dispute with us in any foreign market, or pretend to navigate or to sell manufactures at the same price, which to us would afford sufficient profit? In how little time, therefore, must this bring back the money which we had lost, and raise us to the level of all the neighboring nations? Where, after we have arrived, we immediately lose the advantage of the cheapness of labour and commodities; and the farther flowing in of money is stopped by our fulness and repletion. (Hume [1752] 1955, pp. 62-63) Thus, the domestic specie stock in a country (or province within a country) under a specie standard will be automatically regulated by its balance of payments. Moreover, this mechanism will ensure that each nation's (province's) price level will be consistent with adherence to the specie standard. In addition, variations in the exchange rate within the gold points will act as an additional factor to correct balance-of-payments disequilibria. There is another cause, though more limited in its operation, which checks the wrong balance of trade, to every particular nation to which the kingdom trades. When we import more goods than we export, the exchange turns against us, and this becomes a new encouragement to export; as much as the charge of carriage and insurance of the money which becomes due would amount to. For the exchange can never rise but a little higher than that sum. (P. 64n) Hume also discussed the law of one price, viz. any man who travels over Europe at this day, may see, by the prices of commodities, that money ... has brought itself nearly to a level; and that the difference between one kingdom and another is not greater in this respect, than it is often between different provinces of the same kingdom.... The only circumstances which can obstruct the exactness of these proportions [between money and real economic activity] is the expense of transporting the commodities from one place to another. (P. 66) Some writers have argued that there appeared to be an inconsistency between the law of one price, which suggests rapid adjustment of commodity prices through arbitrage, and the price-specie-flow mechanism which suggests noticeable time lags. 8 However, as is made most clear by Ricardo, the price-specie-flow mechanism is a reflection of arbitrage in the gold market, which, because of its special properties, is more rapid than arbitrage in other markets. Other prices are kept in line through the influence of changes in the quantity of gold as money. In accordance with this interpretation, Hume may have regarded the law of one price as a

35

The Gold Standard: The Traditional Approach

long-run equilibrium condition in all markets with the price-specie-flow mechanism as the means to achieve that result. David Ricardo Gold as the Standard

Ricardo ([1811] 1951, [1816] 1951), in the classical tradition, viewed the world quantity of specie as determined in the long run by cost of production (Ricardo [1811] 1951, p. 52). The quantity of precious metals used as money in each country depended "first, on its value;-secondly, on the amount or value of the payments to be made;-and, thirdly, on the degree of economy practiced in effecting those payments" (Ricardo [1816] 1951 ,po 55). Each country's share of the world specie stock and hence the natural distribution of precious metals is determined by its share of world real income and factors determining velocity (or the demand for money). The precious metals employed for circulating the commodities of the world, ... have been divided into certain proportions among the different civilized nations of the earth, according to the state of their commerce and wealth, and therefore according to the number and frequency of the payments which they had to perform. While so divided they preserved everywhere the same value, and as each country had an equal necessity for the quantity actually in use, there could be no temptation offered to either for their importation or exportation. ([1811] 1951, p. 52) The choice of gold and silver as monetary standard was "the comparative steadiness in the value of the precious metals, for periods of some duration" ([1816] 1951, p. 55). In the choice between gold and silver, gold has in its favor "its greater value under a smaller bulk" which "qualifies it for the standard in an opulent country," but coupled with the disadvantage that it is subject to "greater variations of value during periods of war or extensive commercial discredit." Silver he viewed as "much more steady in its value, in consequence of its demand and supply being more regular." The only objection to its use as a standard "is its bulk, which renders it unfit for the large payments required in a wealthy country" (p. 63). However, using both precious metals as the standard has the disadvantage that prices expressed in terms of gold and silver will vary with changing demand and supply conditions for each commodity. To avoid this instability, Ricardo suggested the substitution of paper money and "by the judicious management of the quantity, a degree of uniformity ... is secured" (pp. 57-58).9 In addition, the substitution of bank notes for specie "enables us to turn the precious metals (which, though a very

36

Michael D. Bordo

necessary part of our capital, yield no revenue) into a capital which will yield one" ([1811] 1951, p. 55). However, "the issuers of paper money should regulate their issues solely by the price of bullion" ([1816] 1951, p. 64). Indeed, for Ricardo, the key advantage of the gold standard was that adherence to the standard acted as a check against the overissue of paper money-it provided discipline (p. 78). Balance-oJ-Payments Adjustment Mechanism

Beginning with the natural distribution of precious metals, Ricardo demonstrated how this distribution would be neutral with respect to monetary changes. Any movement away from the natural distribution would be corrected by the price-specie-flow mechanism. Thus if a gold mine were discovered in one country the currency of that country would be lowered in value in consequence of the increased quantity of the precious metals brought into circulation, and would therefore no longer be of the same value as that of other countries. Gold and silver, whether in coin or in bullion, obeying the law which regulates all other commodities, would immediately become articles of exportation; they would leave the country where they were cheap, for those countries where they were dear, and would continue to do so, as long as the mine should prove productive, and till the proportion existing between capital and money in each country before the discovery of the mine, were again established, and gold and silver restored every where to one value. In return for the gold exported, commodities would be imported; and though what is usually termed the Balance of Trade would be against the country exporting money or bullion, it would be evidence that she was carrying on a most advantageous trade, exporting that which was no way useful to her, for commodities which might be employed in the extension of her manufactures, and the increase of her wealth. (Ricardo [1811] 1951, p. 54)10 Thus gold as a commodity flows to the market with the highest price and thereby maintains price uniformity between nations. As long as different countries (regions within countries) fixed the prices of their currencies in terms of gold (specified a gold weight of their coins), then arbitrage allowing for transportation costs would always keep gold prices in line. 11 This principle, referred to as the law of one price, would hold for all traded commodities, and hence in logic there was no reason why commodity arbitrage would not occur for all commodities. However, for Ricardo and other classical economists, arbitrage took place primarily in gold because of its special properties as money and because it involved the lowest arbitrage costs. Consequently since all other commodity prices were set in terms of gold-the numeraire of the system-gold flows would then keep all prices in line for countries (regions) on a gold standard. However, some prices would react more quickly to the mechanism than

37

The Gold Standard: The Traditional Approach

others, specifically prices of tradable goods, and this quick reaction probably explains the later emphasis in the literature on the role of changes in sectional prices and the terms of trade (see appendix B on Marshall). Ricardo clearly distinguished between the adjustment mechanism under the gold standard and under irredeemable paper money. An issue of convertible paper currency, e.g., Bank of England notes, will displace, through the balance of payments, a corresponding amount of specie (p. 67). However, as long as convertibility in terms of gold is maintained, the domestic price level will not be affected. 12 Once all specie is displaced and convertibility suspended, however, domestic prices will rise, and a depreciated exchange rate will be the indicator of overissue (pp. 58-59, 63-64, 72-78). In the Bullion report ([1810] 1978, pp. ccxvii-ccxxi), a distinction is made between the real exchange rate-determined by the ratio of the mint prices of gold between two gold standard countries-and the market exchange rate or the computed par. The market exchange rate includes both the influence of real factors causing a divergence from par within the gold points and the depreciation of the exchange rate (premium on the price of gold) due to a rise in the price level. Thus for Ricardo, the increase in irredeemable paper following the suspension of payments in 1797 was responsible for both a rise in all commodity prices in England with no corresponding rise in prices abroad and the depreciation of the pound (pp. ccxiv-ccxv). Proposals for Monetary Reform

As mentioned above, Ricardo viewed a properly regulated convertible paper currency as superior to a precious-metals standard. However, he believed that convertibility into gold was necessary to avoid the temptation of overissue (Ricardo [1816] 1951, p. 69). And in the Bullion report ([1810] 1978, p. ccxlvi) a strong case is made in favor of a gold standard rule and against discretionary monetary policy: The most detailed knowledge of the actual trade of the country, combined with the profound science in all the principles of money and circulation, would not enable any man or set of men to adjust, and keep always adjusted, the right proportion of circulating medium in a country to the wants of trade. When the currency consists entirely of the precious metals, or of paper convertible at will into the precious metals, the natural process of commerce, by establishing exchanges among all the different countries of the world, adjusts, in every particular country, the proportion of circulating medium to its actual occasions, according to that supply of the precious metals which the mines furnish to the general market of the world. The proportion, which is thus adjusted and maintained by the natural operation of commerce,

38

Michael D. Bordo

cannot be adjusted by any human wisdom or skill. If the natural system of currency and circulation be abandoned, and a discretionary issue of paper money substituted in its stead, it is vain to think that any rules can be devised for the exact exercise of such a discretion. As a remedy for defects of a purely metallic standard with no discretion for central bankers, he proposed a convertible banknote issue backed by bullion (Ricardo [1816] 1951, p. 66). Free export and import of bullion would be permissible. Under this scheme the costs of frequent conversions of coin into bullion would be eliminated, but the risks of attempted conver~ion of banknotes into specie in a money-market panic would not be. 13 Henry Thornton The Gold Standard

Like Ricardo, Thornton ([1802] 1978, p. 21a) viewed convertibility as a key feature of the gold standard. 14 Also, like Ricardo, he viewed the substitution of paper money for specie up to the point of convertibility as a social saving. However, he extended the analysis to consider the effects of a domestic issue of bank notes on the world price level. First, since the issue of paper money would displace specie in the domestic circulation, specie would be exported abroad, leading to an increase in the world money stock and a rise in world prices. The country displacing specie would thereby raise its capital stock (Thornton [1802] 1978, pp. 269-70). Second, to the extent the use of paper money reduced the demand for gold as money, and hence the price of bullion, this development would cause "those mines which have not yielded any rent, to be no longer worked; and the supply of gold ... to be in consequence, somewhat reduced." The process would continue until all mines will be unable to defray the charge of extracting the ore, except those which now yield the very highest rent. At this point the fall will necessarily stop . . . gold and silver must continue to bear that price, or nearly that price, ... at which they are now exchangeable for commodities. (P. 266) Balance-oj-Payments Adjustment Mechanism

Thornton clearly elucidated the price-specie-flow mechanism. The primary mechanism of adjustment following an increase in the domestic money supply (an increase in Bank of England note issue) is the effect on prices at home relative to those abroad. It is obvious, that in proportion as goods are rendered dear in Great Britain, the foreigner becomes unwilling to buy them, the commodities of other countries which come into competition with our's obtaining a

39

The Gold Standard: The Traditional Approach

preference in the foreign market; and, therefore, that in consequence of a diminution of orders from abroad, our exports will be diminished . . . . But not only will our exports lessen ... ; our imports also will increase; for the high British price of goods will tempt foreign commodities to come in nearly in the same degree in which it will discourage British articles from going out. (Thornton [1802] 1978, p. 198) The resultant deficit in the balance of payments will however be offset to a certain extent by changes in the exchange rate within the gold points. However, to the extent an unfavorable balance persists and the exchange rate falls to the specie-export point, this will lead to a specie outflow until trade balance is restored (pp. 145-47). Finally, Thornton discussed an alternative adjustment mechanismthe direct-expenditure-income mechanism: There is in the mass of the people . . . a disposition to adapt their individual expenditure to their income. Importations conducted with a view to the consumption of the country into which the articles are imported . . . are limited by the ability of the individuals of that country to pay for them out of their income.... If, therefore, ... the value of the annual income of the inhabitants of a country is diminished, either new economy on the one hand, or new exertions of individual industry on the other, fail not, after a certain time, in some measure, to restore the balance. And this equality between private expenditures and private incomes tends ultimately to produce equality between the commercial exports and imports. (Pp. 142-43) Law of One Price

According to Thornton, different prices within Great Britain for identical goods cannot exist as long as country bank notes are convertible into Bank of England notes. A very considerable advance in the price of commodities bought and sold in one quarter of this kingdom, while there was no such rise in any other, was not supposable; because the holders of the circulating medium current in the spot in which goods were . . . rendered dear, would exchange it for the circulating medium of the part in which they were assumed to be cheap, and would then buy the commodities of the latter place, and transport them to the former, for the sake of the profit on the transaction. Moreover, the law of one price can be extended from one kingdom to the whole world as long as currencies are convertible. We may ... extend our views, and conceive of Europe, and even of the world, as forming one great kingdom, over the whole of which goods pass and repass . . . nearly in the same manner in which they spread themselves through this single country.

40

Michael D. Bordo

However, prices can differ between countries within the limits of the gold points: But British paper is not exchangeable for the circulating medium of the continent, unless a discount ... be allowed. Of this fluctuating discount ... the variations in the course of exchange are the measure. (Thornton [1802] 1978, pp. 260-61) Finally, he argued that under a specie standard, one country alone can affect world (traded) goods prices only to the extent that it has monopoly power in their production. Great Britain may have this power in the short run, but in the long run the existence of substitutes will diminish the power. Policy Considerations

Thornton was one of the first to recognize the possibility of a conflict between external and internal policy goals. In the case of an unfavorable balance of trade caused by an exogenous event such as a harvest failure, the central bank could respond to the resulting gold outflow by reducing the money supply, but taking such a course of action might depress domestic activity. Hence, it would be prudent to maintain an adequate gold reserve to permit the bank to increase its loans while losing gold. IS John Stuart Mill Perhaps the clearest statement of the traditional approach to the gold standard is in J. S. Mill's Principles of Political Economy ([1865] 1961). Much of the subsequent literature is either a refinement of Mill or attempts to verify his theory. In discussing Mill, I focus on three topics: (a) gold as a commodity money; (b) the natural distribution of precious metals and the adjustment mechanism; (c) the distinction between real and nominal disturbances. Gold as a Commodity Money

Mill carefully analyzed the economics of commodity money, according to which market forces ensure a determinate money stock and price level. In the longrun, according to Mill, the exchange value of gold-what it will purchase in terms of other goods and services or the inverse of the price level-will be equal to its cost of production-"the cost in labor and expense, at the least productive sources of supply which the then existing demand makes it necessary to work" (Mill [1865] 1961, p. 502). The conformity will be maintained by deviations in gold output in response to variations in the exchange value of gold relative to its cost of production. However, because the existing stock of gold is large relative to additions to the stock, it takes a long time for full adjustment to take place.

41

The Gold Standard: The Traditional Approach

And hence the effect of all changes in the conditions of production of the precious metals are at first, and continue to be for many years, questions of quantity only, with little reference to cost of production. (P. 503) Thus, in the short run, the price level is determined by the relationship between the demand for and supply of money, and only in the long run is it determined by cost of production. Mill then compared a bimetallic standard to a single metallic standard-"There is an obvious convenience in making use of the more costly metal for larger payments and the cheaper one for smaller" (p. 507), but the arrangement only works if the ratio of the two metals is consistent with their relative costs of production. If relative values change, e.g., the value of gold rises relative to silver, this change will cause replacement of gold by silver coins and the melting of gold coins. Mill therefore preferred a limping standard. The advantage without the disadvantages of a double standard, seems to be best obtained by those nations with whom one only of the two metals is a legal tender, but the other is also coined (the more costly metal), and allowed to pass for whatever value the market assigns to it. (P. 509) Finally, Mill, like his predecessors, viewed the substitution of paper money for specie up to the point of convertibility as "a national gain," but beyond that "a form of robbery" (p. 551). Moreover, the social saving from the issue of paper money is transmitted to the rest of the world. The specie displaced through the balance of payments will initially lead to a rise in the world price level, but ultimately to a reduction in gold output. The world price level will then return to normal. 16 Natural Distribution of Precious Metals and Balance-oJ-Payments Adjustment Mechanism

In the long run, each country in the world will have that quantity of money to effect exchange consistent with keeping its value in terms of its cost of production, hence the natural distribution of precious metals across countries is determined by real forces. 17 Mill then compared the international adjustment mechanism under a barter system with that under a money system. Starting from a state of stable equilibrium where the value of exports equals the value of imports "the process by which things are brought back to this state when they happen to deviate from it, is, at least outwardly, not the same in a barter system and in a money system." Under barter, a country which wants more imports than its exports will pay for, must offer its exports at a cheaper rate, as the sole means of creating a

42

Michael D. Bordo

demand for them sufficient to reestablish the equilibrium. When money is used, the country ... takes the additional imports at the same price as before, and as she exports no equivalent, the balance of payments turn against her; the exchange becomes unfavourable, and the difference has to be paid in money. This is in appearance a very distinct operation from the former. (Mill [1865] 1961, pp. 619-20) However, this difference is only apparent; in both cases prices must adjust to restore equilibrium. In the case of a money economy: When. . . the state of prices is such that the equation of international demand cannot establish itself, the country requiring more imports than can be paid for by exports; it is a sign that the country has more of the precious metals ... than can permanently circulate, and must necessarily part with some of them before the balance can be restored. The currency is accordingly contracted: prices fall, and among the rest, the prices of exportable articles; for which, accordingly, there arises, in foreign countries, a greater demand: while imported commodities have possibly risen in price, from the influx of money into foreign countries, and at all events have not participated in the general fall. (Pp. 620-21) Thus, through the price-specie-flow mechanism the same results will be achieved as under barter, with the only difference that relative prices adjust as a consequence of changes in the quantity of money induced by specie flows rather than adjust directly. "In international, as in ordinary domestic interchanges, money is to commerce only what oil is to machinery, or railways to locomotion-a contrivance to diminish friction" (p. 622). Mill made a clear distinction between temporary and permanent disturbances to the balance of payments. When a disturbance is temporary, most of the adjustment takes place through variations in the exchange rate, within the gold points. Thus the deficit will be "soon liquidated in commodities, and the account adjusted by means of bills, without the transmission of any bullion" (pp. 617-18). In the case of a permanent disturbance to the balance of payments, the adjustment must be made by "the subtraction of actual money from the circulation of one of the countries" (p. 618).

Distinction Between Real and Nominal Disturbances Since the natural distribution of precious metals is determined by real forces, changes in that distribution will only follow from a change in real forces. Thus Mill made a clear distinction between the effects of a real disturbance, such as a remittance from one country to another, and a purely nominal disturbance, such as the discovery of a hoard of treasure. In the first case, he starts from a state of equilibrium, after the first remittance is

43

The Gold Standard: The Traditional Approach

made in money. This lowers prices in the remitting country, and raises them in the receiving. The natural effect is that more commodities are exported than before, and fewer imported, and that ... a balance of money will be constantly due from the receiving to the paying country. When the debt thus annually due to the tributary country becomes equal to the annual tribute. . . no further transmission of money takes place; the equilibrium of exports and imports will no longer exist, but that of payments will; the exchange will be at par, the two debts will be set off against one another, and the tribute or remittance will be virtually paid in goods. In addition, the terms of trade will turn against the paying country and in favor of the receiving country. "The paying country will give a higher price for all that it buys from the receiving country, while the latter. . . obtains the exportable produce of the tributary country at a lower price" (Mill [1865] 1961, pp. 627-28) In contrast, a disturbance in the money market changes the world price level with no real effects. Thus the discovery of a hoard of treasure in a country with a purely metallic currency will raise prices there, check exports and encourage imports, leading to a balance-of-payments deficit and diffusion of the new stock of money over the commercial world; consequently the country's price level will revert to its previous level. John E. Cairnes Cairnes in his essays on gold (first published in 1858-60, reprinted in 1873) used the monetary history of the Australian colonies following the gold discoveries of 1851 to test some of the principal conclusions of classical monetary and trade theory. 18 In particular, he tested the ability of the quantity theory of money to predict the comparative static effects of the gold discoveries on the money supplies and price levels of the major countries of the world, and the ability of the Hume-RicardoThornton-Mill adjustment mechanism to predict the distribution of precious metals. 19 Starting with the long-run cost-of-production theory of money, Cairnes argued that "the rate of gold earnings [is] ... the circumstance which, in the final resort, regulates the value of the metal and sets the limit beyond which depreciation cannot permanently pass" (Cairnes [1873] 1965, p. 41); and since gold earnings in Australia increased by 50 percent, i.e., the cost of gold fell by one-half, he expected the price level in Australia to double. The inflation process would spread across the world until either prices doubled or the cost of producing gold rose. Finally, as prices in the rest of the world rose, gold would become a less profitable commodity to produce and export until, in the limit, when the price of Australian imports increased by the amount of the fall in the cost of gold, Australia

44

Michael D. Bordo

would cease to have a comparative advantage in gold and would divert resources back to agriculture (p. 48). In 1872, twelve years after writing his essays on gold, Cairnes found, as he had predicted, that (a) exhaustion of the mines and the resulting rising costs of production and (b) rises in the price of imports led to a considerable shift of resources out of gold production and back into agriculture. Next Cairnes examined the factors that determine the pace of price adjustment across countries. He argued that the rise in prices would be most rapid in countries with the most advanced banking systems-the more developed the system of banking and credit, the smaller the amount of new gold required to effect a given rise in prices. Furthermore, since Australia and California conducted most of their trade with the United States and England, most of the new gold would tend to go first to these countries, and from there it would spread to the continent of Europe and to Asia (pp. 67-68).20 In general, Cairnes found the evidence agreeable to his predictionsprices increased most in Australia, followed by price increases in descending order of magnitude in Great Britain and the United States, the Continent, and finally Asia. Moreover, in a postscript, Cairnes reported that (a) world gold production doubled by 1868 and (b) most of the gold ended up in France and India, although much of it passed through England (pp. 160-65). George J. Goschen The clearest statement of the classical position on short-term capital flows in the balance-of-payments adjustment mechanism is in Goschen ([1892] 1978).

Changes in interest rates and short-term capital flows facilitate the balance-of-payments adjustment mechanism since money will be dear and scarce in the country which owes much to foreign creditors, and plentiful in that which has exported much; and, high interest will be attracting money to that quarter whence specie is flowing out in payment of foreign debts. [an] adverse balance of trade will. . . render the bills on the country which is most in debt difficult of sale, and tend to compel it to export specie; whereas the high rate of interest, which is generally contemporaneous with a drain ... of specie, will revive a demand for bills on this same country, and enhance their value in other quarters, for there will be a general desire to procure the means of remitting capital to that market where it commands the highest value. (Goschen [1892] 1978, p. 127)

Thus,

45

The Gold Standard: The Traditional Approach

where a considerable efflux of specie is taking place, the rate of interest will rise in the natural course of things. The abstraction caused by the bullion shipments will of itself tend to raise that rate. (P. 132) Moreover, a country can finance a temporary balance-of-payments deficit by borrowing abroad. Finally, arbitrage in the securities market will ensure that interest rates for a similar class of bills will be equal between financial centers, account being taken of transportation costs and exchange risks. If at any time the rate of interest here falls below that which rules on the continent, it is inevitable that the whole mass of these bills will at once be sent to London, and be discounted there at the cheaper rate, so that the proceeds may be remitted in gold to the continent to be invested there in local securities at the supposed higher rate. (P. 138) Walter Bagehot and the Responsibility of the Bank of England The pure gold standard in England-when the money supply consisted in large part of specie, and variations in its amount were determined mainly by the balance of payments-became a managed gold standard in the course of the nineteenth century-when the money supply consisted primarily of convertible notes and deposits, and variations in its amount were determined by operations of the Bank of England conforming to the external constraint of the gold standard. The evolution to a managed gold standard evoked considerable debate in the ecopomic literature. 21 In the decades after the restoration of convertibility in 1821, British monetary history was punctuated by a series of monetary crises in 1825, 1836, 1838, 1847, 1859, and 1866 (see Viner [1937] 1975, pp. 218-20). These crises occurred when the necessary contraction of the money stock consequent upon a specie outflow (an external drain) coincided with a demand by deposit and note holders for specie currency (an internal drain). In such a situation it was difficult for the Bank to maintain convertibility of its notes without resort to special measures. The Bank Charter Act of 1844 was an attempt to rectify the situation by dividing the Bank of England into an Issue Department and a Banking Department. The former was charged with the responsibility of maintaining the gold standard link by following the "currency principle": The note circulation should fluctuate one-for-one with changes in the Bank's holdings of gold. 22 The latter was to follow the principles of a profit-maximizing banking enterprise, accepting deposits and making discounts. The arrangement was criticized on two grounds: (1) the currency principle ignored the role of deposits as an increasingly important component of the money stock;23 and (2) the Banking Department in operating on a sound commercial-banking basis could not act responsibly as a central bank. In that role, it had to maintain a gold reserve large enough

46

Michael D. Bordo

to protect the rest of the banking system from the effects of both internal and external specie drains. 24 This criticism culminated in the 1860s with the formulation by Walter Bagehot, the influential editor of the Economist, of the "responsibility doctrine" and the establishment of guidelines for a central bank under a gold standard. In Lombard Street ([1873] 1969), Bagehot clearly set out the conflict between the private concern of the Banking Department to reduce the holding of specie reserves to minimize foregone interest costs and its public concern to hold larger reserves (p. 38). Bagehot argued for the "responsibility doctrine"-that the Bank of England must fulfill its special obligations as a bankers' bank and as holder of the nation's specie reserves. In consequence, he established clear guidelines for central-bank behavior in times of crisis. First, in the case of a purely external drain, the Bank of England requires the steady use of an effectual instrument. That instrument is the elevation of the rate of interest. If the interest of money is raised, it is proved by experience that money does come to Lombard Street, and theory shows it ought to come. (P. 46) The rise in Bank rate will initially lead to a short-term capital inflow and a gold inflow. And there is also a slower mercantile operation. The rise in the rate of discount acts immediately on the trade of this country. Prices fall here; in consequence imports are diminished, exports are increased, and, therefore, there is more likelihood of a balance in bullion coming to this country after the rise in the rate than there was before. (P. 47) Second, the best way for the Bank. . . to deal with a drain arising from internal discredit is to lend freely. . . . A panic. . . is a species of neuralgia, and according to the rules of science you must not starve it. The holders of the cash reserve must be ready not only to keep it for their own liabilities, but to advance it most freely for the liabilities of others. (Pp. 48, 51) In brief, the central bank has a responsibility to act as lender-of-Iastresort. Finally, in the case of both an internal and an external drain, the central bank should follow what has come to be known as Bagehot's rule. We must look first to the foreign drain, and raise the rate of interest as high as may be necessary. Unless you can stop the foreign export, you cannot allay the domestic alarm. The Bank will get poorer and poorer, and its poverty will protect or renew the apprehension. And at the rate of interest so raised, the holders ... of the final Bank reserve must

47

The Gold Standard: The Traditional Approach

lend freely. Very large loans at very high rates are the best remedy for the worst malady of the money market when a foreign drain is added to a domestic drain. (P. 56) Bagehot was aware of the limitations to the Bank's control mechanism. The Bank could only temporarily affect market interest rates. The initial effect of an increase in central-bank lending is to lower interest rates, but it also leads to an "increase of trade and increase of prices." The rise of prices and trade leads to an increase in the demand for loanable funds and also to an increase of imports and a decrease in exports. The resultant balance-of-trade deficit leads to a specie outflow and a reduction in reserves, and hence the rate of interest must be raised (pp. 12-14). In conclusion, Bagehot proposed several remedies for reform. Of key importance, "there should be a clear understanding between the Bank and the public that, since the Bank holds our ultimate banking reserve, they will recognize and act on the obligations which this implies" (p. 70); and the Bank should hold an adequate reserve to be determined by "experience" (p. 304).25 Following the publication of Lombard Street, the Bank of England's special position as both lender-of-Iast-resort and holder of the nation's reserve was recognized, but increasing stress was laid on the Bank's vulnerability in view of London's growing international liabilities (see Sayers 1957). Though Bagehot and other writers urged the Bank to meet the problem by maintaining a larger specie reserve (an approach taken by other countries), the Bank's solution was to alter its discount rate whenever its international reserves were affected, thereby primarily influencing short-term capital movements. The Bank also supplemented the use of Bank rate with other tools of monetary policy, especially open-market operations, and in addition learned to protect its reserves by using special techniques referred to as "gold devices" (see Viner [1937] 1975, p. 277; Sayers 1936, chap. 3). Thus the ultimate answer to Bagehot's concern was the use "of a powerful bank rate weapon with a 'thin film of gold' " (Sayers 1951, p. 116).

Appendix B

The Neoclassical Economists: Extension of the Traditional Approach

In this appendix, attention will center on the works of three neoclassical economists-Alfred Marshall, Irving Fisher, and Knut Wicksell-who wrote extensively on the gold standard and whose contribution can be viewed as a refinement or extension of the traditional approach.

48

Michael D. Bordo

Alfred Marshall Marshall's contribution (1923, 1926) to the traditional approach can be classified under two headings: the monetary standard and proposals for reform and the balance-of-payments adjustment mechanism.

The Monetary Standard and Proposals for Reform For Marshall, the primary purpose of a monetary standard is to ensure price stability: Violent fluctuations of prices are less distasteful to the heads of business enterprises than a gradual fall of prices. But I believe they are far more injurious both physically and morally to the community at large. (Marshall 1926, p. 20) Nevertheless, a gold or silver standard may not be the best mechanism to ensure price stability, at least in the long run. Gold and silver, separately or conjointly, can set good standards of general purchasing power, in regard to obligations and business transactions, which do not range over more than a few years; but obligations which range over long periods, call for standards that are not dependent on the hazards of mining. (Marshall 1923, p. 52) Marshall admitted that in past years, technical advances in mining precious metals had kept pace with technical advances elsewhere, so the real cost of producing the precious metals had remained constant over long periods of time, which was no accident; but he still believed that as the arts of life progress. . . man must demand a constantly increasing precision from the instruments which he uses, and from money among others: and he is beginning to doubt whether either gold or silver, or even gold and silver combined, give him a sufficiently stable standard of value for the ever widening range of space and time over which his undertakings and contracts extend . . . [indeed] gold and silver have had a less stable value, during the history of the world, than has accrued to those staple grains, which have supplied the chief means of supporting life to the great mass of the people in every age. (Pp. 53-54) Over shorter periods of time, however, gold and silver represent good monetary standards because changes in the stock are small relative to the existing stock (1926, p. 177). In addition, following the classical tradition, Marshall regarded gold as an inefficient form of money tying up scarce resources. Civilization had advanced sufficiently for an expanded role for convertible paper (p. 137). Finally, following Mill, he opposed bimetallic schemes based on a fixed ratio of gold to silver on the ground that changes in relative costs of production would lead to continuous shifts towards the lower-cost metal,

49

The Gold Standard: The Traditional Approach

thus producing more instability than reliance on one metal alone (1923, p.63). To solve the shortcomings of reliance on precious metals as a monetary standard, Marshall proposed two alternative schemes-symmetallism and a tabular standard. Under the symmetallic scheme, currency would be exchangeable for a combination of gold and silver bullion bars in fixed proportions. A gold bar of 100 grammes, together with a silver bar, say, 20 times as heavy, would be exchangeable . . . for an amount of the currency which would be calculated and fixed once for all when the scheme was introduced.... Anyone who wanted to buy or sell gold or silver alone in exchange for currency could get what he wanted by exchanging gold for silver, or silver for gold, at the market rate. Government fixing its own rates from day to day, so as to keep its reserve of the two metals in about the right proportion, might safely undertake this exchange itself, and then anyone could buy or sell either gold or silver for currency in one operation. (1926, p. 29) The scheme would provide a better monetary standard than Ricardo's gold-bullion-reserve scheme, because it causes the value of legal tender money to vary with the mean of the values of both of these metals ... and because it would be convenient both to those countries which now chiefly use gold and to those which now chiefly use silver. (P. 28). In contrast, the tabular scheme, similar to that of Fisher (see p. 53 below) would separate the standard of value from the medium of exchange. Under this scheme, long-term contracts would be tied to "an official index number, representing average movements of the prices of important commodities" (1923, p. 36). In addition, he proposed regulation of an inconvertible currency so that "the value of a unit of it is maintained at a fixed level," based on an index number (p. 50). Finally, Marshall stressed the need for an international currency or else for the international harmonization of monetary policies: There is a real, though very slow moving, tendency for national interests to overrule provincial interests, and international interests to overrule national, and I think the time will come at which it will be thought as unreasonable for any country to regulate its currency without reference to other countries as it will be to have signalling codes at sea which took no account of signalling codes at sea of other countries. (1926, p. 135) Balance-oJ-Payments Adjustment Mechanism

Following Ricardo, Marshall argued that gold flows reflect arbitrage in a widely traded commodity (gold)

50

Michael D. Bordo

[so] as to bring gold prices at the seaboards of the two countries to equality, allowance being made for carriage. If they are higher in A than in B, there will be a small temporary bounty on exportation from B to A corresponding to this difference, which must always be small. Bills drawn in B on A will multiply, and, specie point being reached, gold will go from A to B till prices in B are as high as in A. (Marshall, 1926, p. 170) However, he questioned why gold has to be the commodity used to settle payments imbalances. Thus in the case of a balance-of-payments deficit the value of gold, as a means of purchasing foreign commodities by being exported in exchange for them, will rise so much that it will be profitably exported for the purpose: ... But under these conditions merchants are likely to look around them, and see whether there is not some other thing which the country does not produce herself, and therefore does not habitually export; but which could under the circumstances be marketed profitably abroad. (1923, p. 153) For example, the balance of payments could be settled by the exportation of lead once "lead point" has been reached. The choice of which commodity is used depends partly on its portability and partly on the extent of the market which it finds in either country. The power of gold for this purpose is therefore of primary importance between two countries which have a gold currency, for gold has in each a practically unlimited market. (1926, p. 172).26 Marshall then discussed the internal adjustment mechanism to an external disturbance such as a gold outflow. The gold outflow would lower the Bank of England's reserves, the Bank would respond by raising its discount rate, and "the result would be a check to speculative investments, a diminished demand for commodities, and a fall of prices" (p. 158).27 At the same time, the rise in the discount rate, if not matched by an equal rise in the discount rate abroad, would attract gold from abroad (p. 160). Finally, he discussed the role of a force that would weaken the traditional adjustment mechanism-the international integration of markets for securities and commodities: The growing tendency of intercommunication has shown itself in the discount market more than in any other; fluctuations in the price of wheat are being held in check by the growing internationality of the wheat market; but the discount market is becoming international more rapidly even than the wheat market. (Pp. 127-28)

51

The Gold Standard: The Traditional Approach

Thus in Marshall's view, more of the adjustment to balance-of-payments disequilibrium takes place through capital flows than through the arbitrage of traded goods, with less of the burden placed on gold flows. Irving Fisher I focus on three important aspects of Fisher's (1920, [1922] 1965) treatment of the gold standard: his exposition of the operation of a commodity money standard, his discussion of the international adjustment mechanism, and his criticism of the gold standard and advocacy of a "compensated dollar. "28 The Commodity Theory of Money Fisher's exposition of the working of a commodity money standard is perhap the most lucid extension of Mill's theory. First Fisher demonstrated how, under a gold standard with unrestricted coining and melting, the price of gold bullion would conform to the price of coin, allowing for seigniorage. Thus, in the case of unrestricted coinage, if the price of gold bullion exceeds that of coin, gold users such as jewellers will melt coin into bullion; in the opposite situation, bullion owners will take bullion to the mint and have it coined. The effect of melting the coin will decrease the stock of gold coins relative to bullion, reducing the value of gold as bullion relative to gold as money, thus lowering the price level and restoring equality between bullion and money. In the opposite case, the effect of minting bullion into coin will restore equilibrium (Fisher [1922] 1965, p. 97). Next Fisher demonstrated how the world gold stock and the world monetary gold stock are influenced by the production and consumption of gold: As the stock of bullion and the stock of money influence each other, so the total stock of both is influenced by production and consumption. The production of gold consists of the output of the mines which constantly tends to add to the existing stocks both of bullion and coin. The consumption of gold consists of the use of bullion in the arts. (P.99) He then made the analogy to a reservoir, "production would be the inflow from the mines, and consumption the outflow to the arts, by destruction and loss" (p. 99). Gold production is regulated by the relationship between "the estimated marginal cost of production" and the purchasing power or exchange value of gold. He assumed that gold mining is an increasing-cost industry. Thus, gold production will always tend toward an equilibrium in which the marginal cost of production will . . . be equal to the value of the product.... If [the] purchasing power of gold is above the cost of

52

Michael D. Bordo

production in any particular mine, it will pay to work that mine.... Thus the production of gold increases or decreases with an increase or decrease in the purchasing power of gold. (Pp. 101-3). And the purchasing power of gold in turn would vary inversely with the prices of other goods. The consumption of gold "in the arts"-nonmonetary uses of gold-is related to "consumption for monetary purposes"-monetary uses of gold-by a comparison of the purchasing power of gold with the "marginal utility of what is consumed." Thus consumption of gold-the diversion of gold from monetary to nonmonetary uses-will be "stimulated by a fall in the value (purchasing power) of gold, while the production of gold is decreased" (p. 104). The two forces of production and consumption, operating in opposing directions, regulate the monetary gold stock and hence the price level. In the case of increased gold production due to the discovery of new mines, the increase in production will lead to a filling up of "the currency reservoir," and "a decrease in the purchasing power of money. This process will be checked finally by the increase in consumption. And when production and consumption become equal, an equilibrium will be established" (pp. 108-9).

International Adjustment Mechanism Fisher effectively argued that for a small open economy that is part of an international gold standard, as is the case for one state within the United States, the money supply is not an independent variable, but is determined by the need for the domestic price level to conform to foreign price levels. However, he preserved the classical quantity-theory notion of causality between the quantity of money and the price level (see Girton and Roper 1978). The price level in an outside community is an influence outside the equation of exchange of that community, and operates by affecting its money in circulation and not by directly affecting its price level. The price level outside of New York City, for instance, affects the price level in New York City only via changes in the money in New York City. Within New York City it is the money which influences the price level, and not the price level which influences the money. The price level is effect and not cause. (Fisher [1922] 1965, p. 172) Following the tradition of Ricardo and Marshall, Fisher argued that the force of arbitrage would tend to produce equality in the prices of traded commodities, but to the extent that international (and interlocal) trade does not bring about uniformity of price levels/9 "it will ... produce an

53

The Gold Standard: The Traditional Approach

adjustment of these levels toward uniformity by regulating ... the distribution of money" (p. 93). Thus gold flows because it is the most efficient international medium of exchange and because it affects the prices of all commodities. In ordinary intercourse between nations . . . there will always be a large number of commodities thus acting as outlets and inlets. And since the quantity of money itself affects prices for all sorts of commodities, the regulative effect of international trade applies, not simply to the commodities which enter into that trade, but to all others as well. (P. 93) Criticisms of the Gold Standard and Proposals for Reform

Like Marshall, Fisher criticized the basing of the monetary standard on a single commodity-gold or, for that matter, silver-because of instability in supply and demand conditions of the money metal. The commercial world has become more and more committed to the gold standard through a series of historical events having little if any connection with the fitness of that or any other metal to serve as a stable standard ... so far as the question of monetary stability is concerned, ... we have hit upon the gold standard by accident. (Fisher [1922] 1965, p. 323) Instead, in a comparison of the purchasing power of gold with that of a number of other commodities he concluded that "in terms of general purchasing power, gold is no more stable than eggs and considerably less stable than carpets" (Fisher 1920, p. 41).30 In addition, Fisher, a number of years later, criticized the gold standard because it allows price-level movements and business fluctuations to be transmitted from one country to another. 31 As a remedy for the inherent instability in the purchasing power of money under the gold standard, Fisher offered his scheme for a compensated dollar: Issue gold certificates backed by gold bullion, but vary the weight of the gold backing per dollar so as to maintain a constant purchasing power of money by tying the weight to an index number-"to mark up or down the weight of the dollar (that is, to mark down or up the price of gold bullion) in exact proportion to the deviations above or below par of the index number of prices" ([1922] 1965, p. 498). This would allow us to "keep the metal gold for the good attributes it has-portability, durability, divisibility, salability-but correct its instability, so that one dollar of it will always buy approximately [a] composite basketful of goods." It would "retain gold as a good medium [of exchange] and yet ... make it into a good standard" (1920, pp. 88-89).

54

Michael D. Bordo

Knut Wicksell Wicksell's views ([1898] 1965) on the gold standard will be discussed under three headings: the commodity theory of money, the international adjustment mechanism, and proposals for reform. Commodity Theory of Money

In contrast to Fisher, for Wicksell the stabilizing features of the commodity theory were too slow to be of consequence for price-level movements except in the very long run. Either the underlying mechanism was weak or, by the end of the nineteenth century, institutional developments had neutralized or obscured it. Thus the newly extracted gold-passes, for the most part, not into circulation, but into the stocks of cash of monetary institutions; and gold for industrial uses is mainly taken either out of these stocks or directly out of imported stocks of uncoined metal. In neither case can it be supposed that there is any direct effect on prices. (Wicksell [1898] 1965, p.31) The use of gold purely as a monetary base was utopian. In such a system the value of money would be directly exposed to the effects of every fortuitous incident on the side of the production of the precious metal and every caprice on the side of its consumption. It would undergo the same violent fluctuations as do the values of most other commodities. (P. 35) International Adjustment Mechanism

Wicksell was skeptical of the Hume price-specie-flow mechanismdisturbances affecting one country's price level relative to another's would affect the terms of trade, and the balance of payments would then be corrected by gold flows. Although the explanation was fundamentally correct, he expressed reservations: It is ... clear that international equilibrium of prices is usually reached far more rapidly and far more directly. The increase in the supply of foreign goods and the diminution in the demand for exports must themselves exert, directly and indirectly, a pressure on domestic prices which is quite independent of any simultaneous movement of precious metals. (Wicksell [1898] 1965, pp. 157-58) For Wicksell, changes in real income must be brought into the adjustment mechanism. Proposals for Reform

Like Fisher, Wicksell was concerned with price stabilization, both nationally and internationally. He favored the use of gold certificates

55

The Gold Standard: The Traditional Approach

backed by a reserve of gold bullion, with each central bank maintaining convertibility and agreeing to accept other central banks' notes and clearing them through an international clearing house (Wicksell [1898] 1965, pp. 186-87). In addition, the central bank of each nation would stabilize its internal price level by keeping the market rate of interest in line with the "natural rate of interest" following simple criteria: 32 So long as prices remain unaltered the bank's rate of interest is to remain unaltered. If prices rise, the rate of interest is to be raised; and if prices fall, the rate of interest is to be lowered; and the rate of interest is henceforth to be maintained at its new level until a further movement of prices calls for a further change in one direction or the other. To achieve international price stability, central banks would need to manipulate their gold stocks cooperatively to keep interest rates in line between nations (p. 192).

Appendix C

The Harvard Neoclassical School

F. W. Taussig of Harvard and his students-Jacob Viner, F. D. Graham, J. H. Williams, and H. D. White-and W. A. Beach (a student of Allyn Young and J. H. Williams) formulated and tested some of the main tenets of the classical Ricardo-Mill theory of the international adjustment mechanism. Taussig's reformulation of the theory and the evidence for Great Britain, the United States, France, Canada, and Argentina in the pre-World War I era are summarized here. In addition the writings of J. W. Angell, a contemporary critic of the Harvard School and a precursor of the modern monetary approach to the balance of payments are examined. F. W. Taussig In an article (1917) and a book ([1927] 1966), Taussig clearly reformulated the traditional approach in a manner suitable for empirical verification and then summarized some of the evidence. For Taussig, international borrowing was the most important disturbance to the pre-World War I international economy. He analyzed the effects of the transfer of capital on the balance of payments, money supplies, and price levels of both lender and borrower. Several of his studies of British and U. S. experience, beginning with the earliest period he covered, will be reviewed here. In an examination of the British balance of payments in the period 1853 to 1879, Taussig found that the adjustment of the merchandise balance of trade to changes in invisibles-both payments and shipping earnings-

56

Michael D. Bordo

and to capital exports was much more rapid and smooth than classical theory would lead one to expect: No signs of disturbance are to be observed such as the theoretic analysis previses; and some recurring phenomena are of a kind not contemplated by theory at all. Most noticeable ... is the circumstance that periods of active lending have been characterized by rising prices rather than by falling prices, and that the export of goods apparently has taken place, not in connection with a cheapening of goods in the lending country, but in spite of the fact that its goods have seemed to be dearer at times of great capital export. (Taussig [1927] 1966, p. 239) In addition, he found specie movements to be small relative to merchandise movements, a fact to be explained by the sensitivity of the British money supply. Moreover, he found it difficult to separate specie flows consistent with the theory from the more steady series of gold flows into Britain following the gold discoveries in the 1840s and 1850s. The general lack of conclusive evidence sympathetic to the classical mechanism for the period 1853-79 is reversed for the period 1880-1914. That period is conveniently divided into two parts: 1880-1900 and 19011914. The first period was characterized by a deficit on merchandise account financed by shipping earnings and income from abroad with no unusual capital exports. The second was dominated by massive capital exports, which were quickly reflected in a decline in the merchandise deficit. According to Taussig, the change in circumstances offered a good test of the theory. Had the trends before 1900 continued, Britain would have continued to enjoy an improvement in her terms of trade, but the terms of trade turned around after 1900. That phenomenon, he believed, was consistent with classical theory: That the gross barter terms of trade should vary as they do, becoming more favorable until 1900, thereafter less favorable, is indeed easily in accord with theory. They will naturally fluctuate in the same direction as the balance of payments. . . . More significant. . . is the fact that the net barter terms of trade move in the same direction, ... [w]e have argued that when a country has payments to receive for other items than merchandise, the direct and simple exchange of goods for goods is also affected, and is affected to the country's greater advantage. (P. 256)33 The improvement in the terms of trade raises domestic real income because money incomes rise, while the price of imported commodities falls. Taussig found the evidence that money wages rose until 1900, after which they turned down, consistent with his theory.34 However, a continuing puzzle for Taussig was how to separate the influence of equilibrating gold flows from the effects of the steady inflow

57

The Gold Standard: The Traditional Approach

of gold into Britain, the world's principal gold market. Moreover, the puzzle was complicated by the fact that commodity exports and imports [respond] with surprising promptness to the balance of international payments as a whole. The promptness is surprising because each constituent transaction . . . is purely in terms of money. . . . Yet the recorded transactions between countries show surprisingly little transfer of the only "money" that moves from one to the other-gold. It is the goods that move, and they seem to move at once; almost as if there were an automatic connection between these financial operations and the commodity exports or imports. That the flow of goods should ensue in time, perhaps even at an early date, is of course to be expected.... What is puzzling is the rapidity, almost simultaneity, of the commodity movements. The presumable intermediate stage of gold flow and price changes is hard to discern, and certainly is extremely short. (P. 261) He also examined a case under the gold standard of U.S. borrowing long-term capital from Great Britain. 35 To the extent that not all of the proceeds of the loan are spent on British goods, the increase in remittances from London to New York will cause a demand for New York exchange in London. New York exchange will rise in London, sterling exchange will fall in New York. . . . The fluctuations in foreign exchange will necessarily be confined within the gold points. (Taussig, 1917, p. 394) The next step is a gold flow from London to New York once the specie-export point is reached. Elsewhere, Taussig argued that in the case of temporary disturbances, gold rarely flows. Much of the adjustment is taken up by movements in the exchange rate within the gold points that speculators promoted. In addition, in the prewar era, sterling bills acted as a substitute for gold in important money-market centers and were transferred in lieu of gold flows. Finally short-term capital movements acted as a substitute for gold flows. 36 "In the last resort, when all expedients for adjusting and equalizing the payments between nations have been utilized and exhausted, specie will flow in payment of balances" ([1927] 1966, pp. 220-21). The gold inflow into the United States and the gold outflow from Great Britain then raises and lowers the money supplies of the two countries respectively (1917, p. 394). However, the effects of gold flows on domestic money supplies depend on institutional arrangements in each country.37 In the pre-World War I era of fractional reserve banking and convertible fiduciary money the inflow or outflow of specie. . . primarily affects the discount policy of the banks. Their discount policy in turn affects the volume of accommodation which they offer to the borrowing public, and this in turn affects the volume of notes and deposits. ([1927] 1966, p. 201)

58

Michael D. Bordo

The next step is a "train of consequences familiar to the reader of Ricardo and Mill. Prices will fall in Great Britain and will rise in the United States" (1917, p. 394). Within the United States, prices of domestic goods and exports rise relative to prices of imports, and opposite movements of sectional prices occur in Great Britain. 38 Finally, because the terms of trade have turned in favor of the United States, her citizens are better off: "Their money incomes have risen, the prices of imported commodities have fallen; as buyers of imported commodities they gain" (p. 395). The opposite takes place in Great Britain. In the case of the United States after 1879 when it was a net importer of British capital and from 1900 to 1914 when it financed a merchandise surplus with immigrant remittances and other invisibles, Taussig expected to observe gold movements and terms-of-trade effects opposite to those he had observed in the British case. As in the British case, Taussig ([1927] 1966, p. 299) found it difficult to separate the effects of equilibrating gold flows from primarily domestic gold production and consumption. However, unlike the British case, the terms of trade did not behave according to theory. It was difficult to discern a marked trend, though the net barter terms of trade were slightly less favorable before 1900 than afterwards. This, he stated is an unexpected result.... On general principles we should look for terms of trade more unfavorable in the second stage. The excess in the money volume of exports meant ... that the United States, in meeting the divers additional charges for immigrants' remittances ... sent out goods having a greater money value than the goods she bought.... The case shows an outcome different from that in Great Britain and Canada during the same period. For these countries, the actual course of events proves to be in accord with theoretical prevision. For the United States it does not. (P. 303) Taussig attempted to account for the poor results for the United States compared to those for Great Britain and Canada by reference to disturbing causes not present in the other countries. Among these were the tariff that made the terms of trade more favorable than otherwise and an exogenous increase in demand by foreigners for U.S. manufactured goods. Taussig then analyzed effects of a capital transfer under flexible exchange rates, the example relevant for the period 1862-78 when the United States had inconvertible paper money and Great Britain was on the gold standard (1917, p. 386). The initial effects of U. S. borrowing in London is to reduce the specie premium in New York (i.e., the U.S.-dollar price of gold falls). This reduction leads to a fall in the paper prices of U.S. exports (as well as domestic goods) and a rise in the paper prices of U.S. imports. As a

59

The Gold Standard: The Traditional Approach

consequence, export industries in the United States are discouraged, imports are encouraged. The opposite effects take place in Great Britain. Thus the total volume of commodities bought and sold in the United States increases, as exports are shifted to domestic consumption. Moreover, real income rises in the United States because holding the money supply constant, nominal income remains unchanged but prices have fallen. U.S. residents gain not only as purchasers of imports (they do under a gold standard as well), but also as purchasers of domestic commodities. In contrast to the gold standard case, the U.S. terms of trade improve not because U.S. residents have larger money incomes and lower prices, but because they have the same money incomes and lower prices. The opposite results occur in Great Britain. Taussig's analysis may be contrasted with that in Friedman and Schwartz (1963, pp. 84-85), of the effect on the gold premium of U.S. Treasury gold purchases abroad before resumption. The Evidence Taussig's students presented evidence for the classical adjustment mechanism based on detailed examination of the monetary history of a number of countries under the gold standard and under inconvertible paper money. Initially the case studies for Great Britain, the United States, Canada, and France under the gold standard and then for the United States and Argentina under inconvertible paper will be reported.

W. Beach In contrast to Taussig, Beach (1935) presented evidence for Great Britain for the period 1881-1913 unfavorable to the classical price-specieflow model. Beach argued that if differences in the levels of commodity prices were to dominate the adjustment mechanism, then gold would be expected to flow because the price levels of one country do not rise or fall in accordance with the levels in other countries. The movement of gold forces all countries to keep the same pace through the various phases of the business cycle. (Beach 1935, p. 170) Thus one would expect gold to flow out of a country in the upswing of the business cycle. Also, according to the classical model, one would expect capital exports to move procyclically so that other things equal, an upswing in a creditor country would induce a capital flow to a debtor country, accompanied by a gold outflow, unless offset by increased purchases of the creditor's exports. Yet Beach found that there was a tendency for gold imports to increase during the prosperity stages of business cycles and for exports to grow during depression, both in Great Britain and the United States.

60

Michael D. Bordo

Beach's explanation of this anomaly relied on several pieces of evidence. First, he observed a high correlation between business conditions and (long-term) capital exports from Great Britain. The volume of new loans, he suggested, was determined primarily by business conditions (pp. 171-73). Second, in the upswing of the business cycle in Great Britain (and the United States), there was an increased internal demand for gold currency. Third, short-term balances (loans) were sensitive to changes in discount rates, more so than domestic business and prices. Hence in the business-cycle upswing, the internal currency drain put pressure on the reserves of the banking system and the Bank of England, which led to a rise in the discount rate, a short-term capital inflow, and, ceteris paribus, a gold inflow. 39 Thus cyclical fluctuations in the movement of specie might easily be controlled by the movements of these balances. This explanation for the cyclical movements of gold found for England and the United States seems more adequate than the explanation based upon price level differences. (P. 180) Jacob Viner Viner (1924) tested the classical balance-of-payments adjustment mechanism for Canada during the period 1900-1913. 40 During that period, the great disturbing factor was the inflow of foreign capital into Canada. The adjustment to be explained was the "process whereby the Canadian borrowings, negotiated in terms of money, entered Canada in the form of goods and not in gold" (Viner 1924, p. 145). The mechanism to be tested was that of J. S. Mill. The capital inflow initially would raise the price of foreign exchange to the gold export point. This would then be followed by a gold flow from the lending to the borrowing country. Prices would rise in the borrowing country and fall in the lending country, leading to a change in imports and exports, with the borrowing country experiencing an unfavorable balance of trade and the lending country, a favorable balance. Once the unfavorable balance of the borrowing country equaled the rate of borrowing, the exchanges would return to parity, gold movements would cease, and relative prices in the two countries would stabilize at their new levels. First, examining the effects of changes in the exchange rate within the gold points, Viner argued that since transportation and insurance costs were very low between Montreal and New York, the gold points were so narrow that changes in the exchange rate were not likely to have much effect on the balance of trade (p. 155). Second, Viner found evidence that gold flows were highest in the years when Canadian borrowing was most in excess of Canadian loans to others

61

The Gold Standard: The Traditional Approach

(p. 160). At this point, Viner digressed on the role of gold in the Canadian financial system. Before World War I, gold did not circulate as hand-tohand currency but did act as a reserve asset for the chartered banks. (Canada did not have a central bank in this period.)41 The Canadian monetary base consisted of a fixed issue of government fiduciary notesDominion notes and gold reserves, largely maintained as "outside reserves" on call in New York or London, or as balances with commercial banks in those centers. According to Viner, changes in outside reserves in New York acted in a manner similar to gold flows in the classical balanceof-payments adjustment mechanism. Thus the transfer of foreign capital from London to Canada usually passed through the New York money market, raising outside reserves of Canadian banks. Then on the basis of the increased outside reserves, Canadian banks would increase their deposits (pp. 177-79). Third, the increase in the Canadian money supply would produce a rise in the Canadian price level (relative to the rest of the world) in accordance with the classical theory42 and a rise in sectional price levels. The initial effect would be on the prices of domestic goods and not on the prices of imports, which for a small open economy such as Canada are determined abroad. Some substitution away from domestic goods towards imports would follow as would also a rise in the price of exports (to the extent Canada had monopoly power in their production), leading to a reduction in exports.43 The evidence generally confirms these predictions: between 1900 and 1913 indexes of the prices of imports increased least, followed by the prices of exports, while domestic prices increased the most. Moreover, a beginning-of-period weighted index of the price of exports declined relative to an unweighted index, suggesting that commodities shifted from the export to either the domestic or import category. In addition, the ratio of domestic-goods prices to the wholesale price index increased more in Canada than in the United States. 44 Finally a decline in exports and a rise in imports completed the case for Viner in favor of the classical adjustment mechanism. 45 This theory has been verified inductively for Canada during the period 1900 to 1913. [Moreover,] a corollary of this reasoning [is] that during a period of international borrowings the terms of international exchange shift in favor of the borrowing country and against the lending country. . . . Adequate inductive verification of this proposition is supplied by the demonstration already made that export prices rose relative to import prices. (P. 295) Harry D. White White (1933) examined the French evidence for the classical pricespecie-flow explanation of the effects of capital exports. Like Britain,

62

Michael D. Bordo

France in the four decades before World War I lent large sums abroad, and as in the British case, for contemporary economists, the income from foreign investment permitted a persistent merchandise-trade deficit. In a careful reconstruction of the French balance-of-payments accounts, White demonstrated that the income from foreign investment over the 1880-1913 period was no greater than the total export of capital for the same period. Moreover he found, contrary to the official figures, that in twelve out of the thirty-four years surveyed, France had a surplus on merchandise account; and the years when the French accounts showed a deficit on merchandise trade it was paid for not by revenue from French foreign investment but by foreign-tourist expenditures in France (White 1933, p. 301). White then attempted to determine, as Taussig did for Great Britain, the adjustment mechanism by which the net capital export affected the balance of trade: Tho the totals of capital exports and net revenue from foreign investment over the period as a whole were not far apart, for anyone year during most of the period they differed considerably, thereby raising the question of the mechanism for adjustment of the disequilibrium caused by changes in the volume of capital exports. (P. 302) First, he found that in France's trade with gold standard countries, exchange-rate movements had a negligible effect on merchandise movements, but in the case of a number of countries not on the gold standard, from which France obtained one quarter of her imports, fluctuations in the exchange rate made a significant difference to French importers. Second, he concluded that in the French case, only a small portion of the sums loaned abroad were spent on French exports. Third, the movements of sectional price changes and of the volume of merchandise trade revealed a relationship in accordance with the classical sequence. Relative increases in import prices were accompanied by a decline in physical quantities and vice versa. The changes continued until there was a rough approximation between changes in the values of merchandise balances and capital exports. 46 Fourth, he could find no evidence of the linkage of gold flows to the reserves of the Bank of France and thence to the domestic money supply. According to White, the Bank of France would raise the discount rate only to offset a large gold outflow, but not to influence domestic economic activity. Moreover France's large gold reserves relative to other major gold standard countries enabled it to follow such a passive discount-rate policy. Fifth, White argued that it was possible that there existed a direct link between gold flows and price-level movements because such a large fraction of the French money supply consisted of gold currency, but

63

The Gold Standard: The Traditional Approach

no clear evidence on . . . the relationship of the quantity of money in circulation to prices is revealed by the comparison between the fluctuations in the quantity of specie and notes outside the Bank of France and the movement of prices. Moreover, even if such a relationship were revealed, the absence of correlation between the annual movements of capital and specie renders dubious the interpretation that changes in sectional price levels were induced by capital movements. (P. 304) Finally, the mixed evidence leads him to conclude that the specie-flow-specie mechanism is doubtless one of the forces, but there seems to be no justification for assuming that it is the sole or even the dominant means of adjustment. ... The neoclassical theory is not the complete explanation. The theory fails in that it explains what happens only under certain given conditions seldom found. It expounds a sequence of changes which undisturbed would in time bring about adjustment, but which seldom, if ever, operates unchecked by the frictions and rapid changes characteristic of modern economic conditions. By ignoring some of these forces and minimizing others, the neoclassical exposition exaggerates the effectiveness of gold flows and sectional price changes as a means for establishing equilibrium in international accounts. (P. 306)

Frank D. Graham Graham tested Taussig's (1917) theory of the adjustment mechanism under depreciated paper. In the period 1862 to 1878, when the United States was on an inconvertible paper standard-the greenback standard-while her principal trading partner, Great Britain, was on a gold standard, the premium on gold was a close proxy for the dollar-pound exchange rate. 47 Graham analyzed the effects of British capital flows-the major source of disturbance during this period to the balance of payments, and hence the exchange rate. The period can conveniently be divided into two episodes: 1863 to 1873, a period of heavy and continuous borrowing from London, and 1874 to 1878, a period when the borrowing dropped off. According to Taussig's theory, one would expect the period of heavy borrowing to be associated with a deficit in the balance of trade and the period of cessation with a reversal in the balance of trade. Graham's evidence showed a large annual excess of commodity imports over exports in the period of heavy borrowing, the opposite in the period of cessation (Graham 1922, pp. 231-34). Also, according to Taussig's theory, one would expect, ceteris paribus, the period of heavy borrowing to be associated with a decline in the exchange rate (the premium on gold), while the period of cessation to be associated with a rise in the exchange rate. Graham found the evidence corroborated this prediction. In comparing the price of gold with an index

64

Michael D. Bordo

number of general commodity prices, he found the quarterly average price of gold to be lower than the general price level from April 1865 through June 1876, while from July 1876 to the end of the period it was higher, with the price of gold rising relative to the overall price level after 1874. Next, according to Taussig, one would observe the following effects on sectional prices: in the first period, low paper prices of exports, gradually declining paper prices of imports, and relatively high paper prices of domestic commodities; in the second period, a gradual reversal towards higher paper prices of exports, gradually rising paper prices of imports, and relatively low prices of domestic goods. A comparison of the arithmetic means of the three different groups of commodities between the two periods provided evidence consistent with the theory. Finally, one would observe different effects on money wages (as a measure of relative prosperity) in the two periods: in the first period declining wages in the export industries relative to the domestic-goods industry, as resources are diverted from it in the face of falling prices; the opposite movement in the second period. Again, using Mitchell's wage data and classifying U.S. industries into domestic and export industries, Graham found the evidence consistent with the theory (pp. 267-70). At the same time as these effects occurred in the United States, opposite ones would occur in Great Britain, although the fluctuations in the U.S. dollar would have only limited effects on British prices since Great Britain was on a gold basis and movements of the dollar would only affect a portion of her trade. 48 Again Graham found the evidence consistent with the theory.

John H. Williams Williams (1920) tested for Argentina Taussig's theory of the adjustment mechanism under depreciated paper but unlike Graham found the evidence too inconclusive to be of more than limited support to the theory. First, according to Taussig's theory, a rising premium on gold would stimulate exports. "It does so by virtue of the fact that export prices rise more rapidly than costs, creating an extra profit or bounty for the producing and exporting classes" (Williams 1920, p. 233). Williams presented evidence indicating a rise in the price of exports concomitant with a rising premium on gold in the period 1885 to 1891. However, according to the theory, the value of exports ought to rise-but it did not. This result Williams attributed to the presence of other forces such as "the character of the Argentine exports . . . agricultural and grazing products . . . [which are] extremely susceptible to vagaries of climate" and the fact that "though [the] quantity of exports increased, the greater quantity was sold for a lower gold price per unit" (p. 234).

65

The Gold Standard: The Traditional Approach

Second, according to Taussig's theory, a rising gold premium should reduce the value of imports, which the evidence confirmed. We find first of all that the diminution of value of imports asserted by theory ... did occur: that, in fact, the diminution was very marked . . . . On comparing the course of imports with the gold premium, ... there was in every year an inverse relation between imports and the premium on gold. (P. 253) In sum, the evidence marshalled by the Harvard school in favor of the classical adjustment mechanism is mixed. Overall price levels adjust in accordance with the theory, but sectional price adjustment in accordance with the theory is limited. Little support was found for the role of gold flows, the money supply, and discount rates in the mechanism. A common finding was that the commodity trade balance adjusted rapidly to the external disturbance of capital flows, more rapidly than would be expected by a theory postulating links from price-level differences to gold flows, to changes in money supplies, to changes in sectional price levels, and then adjustment of the commodity trade balance. Explanations given by the school for the rapidity of adjustment-with no supporting evidence-included the growing integration of goods and securities markets, the role of income effects, and the sensitivity of the money-supply process. J. W. Angell

Angell ([1925] 1965), a student of Taussig's, had a distinctly different interpretation of the international adjustment mechanism under the preWorld War I gold standard. Angell focused on the relationship between individual commodity prices and national price levels. First, he argued that the classical division of the overall price level into export, import, and domestic-goods prices was "extremely misleading and may lead to erroneous conclusions." The distinction he preferred was between international or traded commodities and domestic or nontraded commodities, with the dividing line between the two types of commodities to be determined empirically: Any movable article whatsoever may enter international trade.... The first requisite for movement is that the money prices receivable, translated through a common measure, shall be higher in one country than in the other; the second, that the difference shall at least cover costs of transportation of all sorts, including tariff[s] (Angell [1925] 1965, pp. 375-76) He cited evidence for equality in the world prices of traded goods (staple commodities) at one extreme, and at the other extreme, no international competition and hence no reason for price equalization of nontraded

66

Michael D. Bordo

goods, with an in-between third category-partially traded goods. Arbitrage would ensure equality for traded-goods prices but arbitrage may not take place for many commodities because of: (a) "lack of accurate information, in each market," (b) "lack of sufficient initiative and enterprise, on the part of the manufacturers and dealers, to take advantage of the discrepancy," (c) "selling in a new market requires the prior erection of trade connections," (d) monopoly power (pp. 379-80). Second, Angell discussed evidence showing long-run similarity of national price-level movements between countries on the same metallic standard. He defined national price structures as a series of solar systems, which maintain a fairly constant relationship in their movements through space. But the component parts of the system-individual prices-are in a state of ceaseless change relative to the elements in both their own and other systems. (P. 390) However, he argued that the similarity of movement of national price levels coupled with the tendency to price equalization of internationally traded goods suggests that domestic (nontraded-goods) prices conform to the pattern set by traded-goods prices. Further, since the process of substitution between domestic and traded goods is a relatively weak one, the key mechanism that keeps prices levels in line is the classical price specie flow analysis.... No other type of explanation can adequately account for the known facts. Prices in different countries do not move together, over long periods of time, by sheer accident. There must evidently be some connecting link between them. But the influence exerted by international prices alone, on the various national price structures, is not great enough to provide this link. It is necessary to discover some condition or element that is capable of affecting the totality of prices indiscriminately, and fairly rapidly. This element is found in the mechanism by which the balance of international payments is kept in equilibrium. (P. 393) Angell offered his own version of the price-specie-flow mechanism. He downplayed the role of actual gold flows in the mechanism, arguing that neither the magnitudes nor the directions of the international flow of gold are adequate to explain those close and comparatively rapid adjustments of payment disequilibria, and of price relationships, which were witnessed before the war. (P. 400) Moreover, the character of modern banking [will not] permit the assumption of any very high degree of intimacy to the connection between a country's metallic stock and its price level. Finally, gold is among the least sensitive of the media of international payments, one of the slowest to move. (Pp. 400-401)

67

The Gold Standard: The Traditional Approach

His own version of the mechanism relies on the importance of foreign bills of exchange which act as a substitute for gold. Temporary disequilibria will be offset by movements in the exchange rate. If that proves inadequate, gold will flow, prompting a change in discount rates. Finally, if the disturbance is more than temporary, then an alteration will take place in the underlying conditions that govern the general course of international exchange itself. The volume of purchasing power in circulation in the creditor country will be built up, in consequence of the increase in the banks' holdings of bills. In the other country it will be reduced, through the decline there in bill holdings. These changes, in turn, will operate upon the corresponding general price levels. The latter effect is often, though not always, strengthened by alterations in the discount rates. The movements in prices will then influence the commodity balance of trade. . . and will continue to do so until the change in the commodity trade has become great enough to offset and correct the original disturbance in the balance of international payments. (P. 413)

Appendix D

Interwar Critics

With the outbreak of war in 1914 and Britain's suspension of convertibility, the classical gold standard expired. The gold-exchange standard existed briefly from 1925 to 1931, after which the problem of international monetary reform and the creation of a more viable international monetary system took center stage. The traditional approach to the gold standard was subjected to extensive reinterpretation and criticism, much of it derived from concern over the monetary instability of the interwar period. The reinterpretation of the gold standard began with the Cunliffe report ([1918] 1979) which succinctly restated the stylized facts of the operation of the pre-1914 gold standard and appealed for a return to the old order. Another view (Brown, Smit) of the prewar gold standard stressed that it was successful because it was a managed standard-managed by London. Followers of th~s institutional approach then documented all the many respects in which the structure of the British money, gold and commodity markets"pax Britannica"-and the astute management of the Bank of England made the system work. The key implication of the approach was that a successful gold standard could be restored if a similar institutional milieu could be re-created. A related approach (Cassel) argued that the gold standard worked well for England, and possibly for several other major countries, but not for the rest of the world. Moreover, the fact that it worked so well for England was largely an accident of history (Viner).

68

Michael D. Bordo

A recurrent theme in the interwar literature was the inherent policy conflict between internal and external stability under the gold standard fixed-exchange-rate system. According to this view (Keynes), the prewar gold standard worked because nations were willing to subsume domestic economic objectives to the maintenance of gold convertibility; but in the postwar period, a return to the harsh discipline of the prewar gold standard would be disastrous. This approach proposed the creation of a supernational monetary agency or similar means to ensure international harmonization of economic policy. The final theme in the literature was the expression of doubt about the stylized facts of how the gold standard worked. Returning to the anomalies between fact and theory discussed by the Harvard school, the critics (Whale) argued that perhaps the traditional approach itself was incorrect. The views of writers identified with each of the foregoing themes are summarized below. The Stylized Facts In the interwar period, a persistent thread in the literature was a view of the prewar gold standard as the ideal monetary standard. A classic statement of this view appeared in the Cunliffe report, but it was repeated by others including R. G. Hawtrey, T. E. Gregory, and the Interim Report of the Gold Delegation of the Financial Committee of the League of Nations ([1931] 1979). The Cunliffe report ([1918] 1979) to Parliament succinctly presented what seemed to be the salient features of the operation of the prewar gold standard in Great Britain and a series of proposals for a quick return to gold. 49 As the report documented, the money-supply process before the war was based on the Bank Charter Act of 1844. Apart from a fixed fiduciary issue, hand-to-hand currency consisted entirely of gold and subsidiary coins or of gold certificates. Gold was freely coined and there were no restrictions on the import or export of gold, so changes in the monetary base, aside from movements of gold to and from the arts, were determined by inflows from abroad and outflows. In addition, upon this base of a fixed fiduciary issue and gold and gold-backed currency rested an extensive system of checkable bank deposits, so that the pre-World War I British money supply consisted mainly of deposits. Second, the report described the operation of the balance-of-payments adjustment mechanism. A disturbance to the balance of payments led to a gold flow and a corresponding change in the money supply. Thus, e.g., when the balance of trade was unfavorable and the exchanges adverse, it became profitable to export gold, and the would-be exporter bought the gold from the Bank of England. The Banking Department in turn

69

The Gold Standard: The Traditional Approach

obtained gold from the Issue Department in exchange for notes from its reserve, with the result that liabilities to depositors and the reserve were reduced by an equal amount, and the ratio of reserves to liabilities declined. 50 The next step was a rise in the discount rate: If the process was repeated sufficiently often to reduce the ratio in a degree considered dangerous, the Bank raised its rate of discount. The raising of the discount rate had the immediate effect of retaining money here which would otherwise have been remitted abroad and of attracting remittances from abroad to take advantage of the higher rate, thus checking the outflow of gold and even reversing the stream. (Cunliffe report [1918] 1979, par. 4) Thus raising Bank rate by inducing a short-term capital inflow would be sufficient to stem a temporary balance-of-payments deficit. However, in the case of a permanent disturbance, the discount-rate rise would additionally reduce domestic credit. This description of the operation of Bank rate to facilitate the adjustment mechanism has often been referred to as the rules of the game. 51 According to Nurkse: Whenever gold flowed in, the central bank was expected to increase the national currency supply not only through the purchase of that gold but also through the acquisition of additional domestic assets; and, similarly, when gold flowed out, the central bank was supposed to contract its domestic assets also .... The chief methods to be used for changing the volume of domestic central bank assets in accordance with this principle were changes in the discount rate, designed to make borrowing from the central bank either more or less attractive, and purchases or sales of securities in the open market on the central bank's own initiative. (Nurkse [1944] 1978, pp. 66-67) Moreover, the gold standard also provided an automatic mechanism to offset an internal disturbance: When. . . credit at home threatened to become duly expanded, the old currency system tended to restrain the expansion and to prevent the consequent rise in domestic prices which ultimately causes such a drain. The expansion of credit, by forcing up prices, involves an increased demand for legal tender currency both from the banks in order to maintain their normal proportion of cash to liabilities and from the general public for the payment of wages and for retail transactions. In this case also the demand for such currency fell upon the reserve of the Bank of England, and the Bank was thereupon obliged to raise its rate of discount in order to prevent the fall in the proportion of that reserve to its liabilities. The same chain of consequences ... described [above] followed and speculative trade activity was similarly restrained. (Cunliffe report [1918] 1979, par. 6)

70

Michael D. Bordo

Thus, there was therefore an automatic machinery by which the volume of purchasing power in this country was continuously adjusted to world prices of commodities in general. ... Under these arrangements ... [the] country was provided with a complete and effective gold standard. The essence of such a standard is that notes must always stand at absolute parity with gold coins of equivalent face value, and that both notes and gold coins stand at absolute parity with gold bullion. When these conditions are fulfilled, the foreign exchange rates with all countries possessing an effective gold standard are maintained at or within the gold specie points. (Pars. 6-7) The committee recommended a restoration of the gold standard "without delay" (par. 15), to be achieved by the cessation of government borrowings and the reduction of Bank of England note issue. In addition, the committee recommended allowing the free external movement of gold and the use of Bank rate to check outflows and inflows. However, following Ricardo's "Proposal for a Secure Currency," it recommended against the use of gold coins for domestic circulation and in favor of use of all the nation's gold to be held by the Bank of England as backing for the nation's monetary base-a gold-bullion standard. Most of the proposals were adopted when Britain returned to gold in 1925. Hawtrey (1935) summarized and expanded upon the stylized facts of the Cunliffe report. He documented the domestic aspects of a gold standard. A gold-coin standard with free coinage and free export and import serves as a device to provide a limit on the supply of money, but the authorities must maintain the quality of the coin. Thus, the essence of the gold standard is that the price of gold, the value of gold in monetary units, is fixed by law and this determines the wealth value of the monetary unit itself. The use of gold coin ... provides a fairly close approximation to this ideal. (Hawtrey 1935, p. 20) A central bank acting as a bankers' bank by holding a pool of gold reserves can reduce the foregone interest cost on commercial banks' holdings of gold coins to maintain convertibility of their liabilities. However, since commercial banks keep their reserves with a central bank, it must accept the responsibility of acting as a lender-of-Iast-resort. For the banks and the public do not trouble themselves about the interchangeability of gold and credit. That is the affair of the Central Bank alone. Anyone can sell the Central Bank as much gold as he likes and can procure from it as much gold as he chooses to pay for. The Central Bank is in the gold market as both buyer and seller in unlimited quantities at a fixed price. (P. 24) The central bank has the power to control the domestic money supply by using its principal tool-its discount rate. By raising the rate, the

71

The Gold Standard: The Traditional Approach

central bank can check lending by "improvident banks" and by lowering the rate it can stimulate lending. Hawtrey then described the mechanism by which a change in Bank rate operates. The power of the central bank over the wealth value of the monetary unit ultimately depends on the deterrent effect of a high Bank rate upon the borrowing operations of the customers of the banks. Bank rate is essentially a short term rate of interest.... It is the borrowing of money for the purchase of goods that is likely to respond most promptly to a restriction or relaxation of credit, because a trader who wishes to reduce his indebtedness in respect of goods held in stock can readily do so by postponing or reducing his purchases. When traders are tending generally to do this, the effect is immediately felt by the producers of the goods in decreased orders.... . . . The installation of capital is usually financed by the raising of funds from the long-term investment market, but short-term borrowing is often resorted to in anticipation of the raising of funds from that source or for the purchase and holding of securities. If Bank rate is raised, the holding of capital assets with money temporarily borrowed is discouraged. But the effect on productive activity will be relatively slow, for the installation of capital is a prolonged process, and any such project is likely to be preceded by a long preliminary period of preparation. (Pp. 25-26) Thus, changes in Bank rate have their primary impact on the holding of inventories. Hawtrey also discussed international aspects of a gold standard. The commitment by a number of countries to fix the prices of their currencies in terms of gold establishes a fixed-exchange-rate system. The prices in anyone currency of gold in different places cannot differ by more than the cost of transporting gold between the different places-arbitrage in the gold market will ensure that outcome. To maintain convertibility of the currency-the primary responsibility of a central bank-an adequate gold reserve is essential. The threat of a loss of gold is more serious than a possible gain, since in the former case the country may be forced to leave the gold standard before the necessary adjustment can take place; in the latter case, although the central bank may temporarily lose control of the market, the stimulus to lending will eventually lead to a reversal of a gold inflow. Finally, Hawtrey, like Keynes (see below, p. 79) saw an analogy between the operation of a clearing system between banks within one country and a clearing system of central banks under the gold standard. In his analysis of the balance-of-payments adjustment mechanism following a monetary disturbance, Hawtrey incorporated elements of both the classical adjustment mechanism and the role of income changes. An increase in bank lending in an open economy by stimulating demand will

72

Michael D. Bordo

ultimately lead to an increase in production and real income, assuming less than full employment, but will initially lead to an increase in sales and reduction in inventories. Merchants and dealers will therefore seek to replenish their stocks. Those dealing in home-produced goods will do so partly by ordering fresh supplies from producers and partly by diverting to the home market goods that might have been sold abroad, while those dealing in foreign-produced goods will order fresh supplies from abroad. The diversion of production from exports and the increase in imports will create an adverse balance of payments and a gold outflow. The process continues until a new higher equilibrium level of income is reached with a lower than initial balance-of-payments deficit. Once full employment is reached, the increased spending will affect prices. However, the prices of traded goods cannot be fully raised. These, which may conveniently be called "foreign trade products" comprise not only actual imports and exports but all importable and exportable goods. The prices of foreign trade products are governed by prices in world markets and are fixed in gold. The demand for them will expand as the consumers' income expands, and as the demand expands the loss of gold grows greater and greater. (Hawtrey 1935, p. 41) In addition, some of the increased income will be diverted to the purchase of foreign securities which will worsen the current-account balance. The process can be arrested and the deficit reduced by an increase in the discount rate-quickly offsetting the capital inflow and ultimately reducing the rise in income. However, "the contraction of the consumers' income is the only substantial corrective" to the balance-of-payments deficit (p. 43). For Gregory ([1932] 1979), like Hawtrey, the essence of the gold standard is convertibility of national currency into a fixed weight of gold. One of the great advantages of the gold standard therefore is that . . . it eliminates fluctuating rates of exchange . . . [that] international trade and investment can be conducted without any fear that the sums risked in a particular trade or investment transaction will not be recovered . . . owing to changes in the relative exchange values of different moneys at the date of payment. (Gregory [1932] 1979, p. 9)52 In addition, he related the development of the international gold standard in the second half of the nineteenth century to the growth of international trade and investment in that period, and stressed the role played by gold movements in the establishment of the conditions necessary to secure equilibrium in the international balance of payments of the various countries upon the gold standard.... What the international gold standard does. . . is to force prices and incomes in different trading areas into such a relationship that the

73

The Gold Standard: The Traditional Approach

balance of payments can be adjusted without gold flows in either direction.... The international gold standard creates, not a common price level but an integrated price-and-income structure. (Pp. 11, 14-15) The gold standard can operate successfully in the context of modern banking systems and central banks (provided sterilization activity is not undertaken) and of tariffs, capital flows, and transfers. The Role of London An important theme in the literature of the gold standard, developed in the interwar period, was that the gold standard was successful primarily because it was a sterling standard. 53 Perhaps the most succinct statement of the position is in Smit (1934), although a similar viewpoint is expressed in Brown (1940). According to Smit, by 1914, all the leading money and trade centers in the world were interconnected in a triangular fashion through London, although smaller patterns, directly centered around Paris, Berlin and New York, were woven into the main pattern of the picture. (Smit 1934, p. 53) In the prewar world, sterling balances instead of gold were increasingly used by foreign financial institutions to settle international payments: The most important key to the world's foreign exchange markets lay in the sterling balances of foreign bankers kept in London.... London acted as one bank for a customer-neighborhood of bankers that comprised not only the small British island but the whole world. . . . The pound, which was internationally wanted for settling commercial and capital indebtedness with the British Isles, became more and more, as a consequence of the world-wide demand that it commanded, a conventional credit counter for settling indebtedness among all countries. (P. 54) The primary financial interconnection between countries centered on the London discount market because much of the world's foreign trade was financed by sterling bills. 54 The Bank of England by its discount rate and open-market operations was able to exercise considerable influence over this market. Thus the extraordinary effectiveness of the English official bank rate in influencing foreign exchange rates and international gold movements before the war cannot be explained unless one sees the integration that had taken place in the world credit structure (P. 55) In addition, the spread of the gold standard and integration of the international credit system were closely intertwined as "the legal guarantees of the gold standard limited the risk factor of foreign exchange

74

Michael D. Bordo

fluctuations and inspired confidence" and "London possessed the world's central gold market.... The possession of sterling balances was the surest means of getting gold when wanted" (p. 56). Finally, Smit argued that at the same time as the world moved towards a sterling standard, there was growing internationalization of commodity prices centered in British commodity exchanges. A key consequence of international integration of both commodity and money markets was that the traditional explanation of how the prewar gold standard operated placed too much emphasis on "the existence of different national monetary systems, and the quantitative relations between the separate national money and credit systems and domestic price levels" (p. 55). Even before World War I, Keynes ([1913] 1971) had noted the unique role of London in the operation of the gold standard. By World War I, England had developed a sound currency and, aided by the effective use of Bank rate, required only a small gold reserve to maintain convertibility in the face of external shocks. However, according to Keynes, most other countries were not as successful in staying on the gold standard. One key difference between Britain and other countries was that she was a net creditor in the international short-loan market, whereas most other countries were debtors. In the former case, which is that of Great Britain, it is a question of reducing the amount lent; in the latter case, it is a question of increasing the amount borrowed. A machinery which is adapted for action of the first kind may be ill-suited for action of the second. Partly as a consequence of this, partly as a consequence of the peculiar organization of the London money market, the "bank rate" policy for regulating the outflow of gold has been admirably successful in this country, and yet cannot stand elsewhere unaided by other devices. (Keynes [1913] 1971, p. 13) Most other countries in adopting the gold standard used gold as a medium of exchange but were unable to use the discount rate as an effective method to preserve the standard because, in addition to not being net international lenders, they had not established the elaborate financial network of the London money market. Consequently other countries used other mechanisms to supplement their inadequacies: they held large gold reserves "so that a substantial drain ... may be faced with equanimity"; they partially suspended payment in gold; and they kept "foreign credits and bills ... which can be drawn upon when necessary"(p. 14). Most countries (especially less-developed ones) tended to rely on the last method because it economized on the foregone interest cost of holding gold reserves. 55 Thus the gold-exchange standard evolved out of the discovery that, so long as gold is available for payments of international indebtedness at an approximately constant rate in terms

75

The Gold Standard: The Traditional Approach

of the national currency, it is a matter of comparative indifference whether it actually forms the national currency. . . . The gold exchange standard may be said to exist when gold does not circulate in a country to an appreciable extent, when the local currency is not necessarily redeemable in gold, but when the government or central bank makes arrangements for the provision of foreign remittances in gold at a fixed maximum rate in terms of the local currency, the reserves necessary to provide those remittances being kept to a considerable extent abroad. (P. 21) Cassel (1935) went even further than Keynes, arguing that the gold standard was an international standard neither in the pre- nor the postWorld War I periods. It was only a British standard. He argued that the automatic balance-of-payments adjustment mechanism of price levels adjusting to gold flows never in fact worked that way-that central banks for the sake of security maintained larger reserves than legally required, and therefore exports and imports of gold did not necessarily influence the domestic money supply or the price level. The gold supply of a country exercised such an influence only via the policy of the central bank and its regulation of the market by means of its rate of discount and its open market operations. Thus the currency necessarily became a "managed currency" whose value depended entirely on the policy of the central bank. (Cassel 1935, p. 3) In addition, capital flows hindered the automatic functioning of the international gold standard: A country normally exporting capital could compensate for a loss of gold simply by a reduction of its lending; and a country normally importing capital could compensate for a loss of gold by borrowing more. Thus it was possible to prevent gold imports or exports from having any influence on the price level of the country. (Pp. 3-4) Finally, the international gold standard did not guarantee a natural distribution of gold: Creditor countries were in a position to accumulate, if they chose to do so, disproportionate gold stocks without using them for any other purpose than for exercising political influence or merely for satisfying a national pride in the possession of gold. Debtor countries could provide gold reserves by increasing their foreign indebtedness.... The size of these reserves had very little to do with the balance of trade of the country. Nor did gold imports and exports have any distinct relation to changes in the balance of trade. (P. 4) That the international gold standard functioned so well before World War I can only be explained by the basic position that the pound sterling held in this system. Indeed, the pre-war gold standard system may not

76

Michael D. Bordo

inadequately be described as a sterling bloc held together by London's position as the world's financial clearing center and by the service of the pound sterling as a generally recognized means for international payments. (Pp. 4-5). Next, in Cassell's view, the international gold standard was fundamentally defective because it was based on the tacit assumption that the purchasing power of gold would be stable and hence that maintenance of a fixed gold parity guaranteed stability in the purchasing power of a country's currency. According to Cassel, the gold standard . . . [suffered] from an inherent and irreparable instability. This instability results partly from the instability of the value of gold itself, and partly from the insecurity of the redemption in gold of gold-standard currencies. (P. 6) On the first score, he presented evidence of considerable variability in the value of gold for the period 1850 to 1910, based on the Sauerbeck wholesale price index. This evidence he explained by "deviations of the actual gold supply from the normal gold supply"56 and "variations in the monetary demand for gold. "57 On the second score, he alleged that only Britain had a completely convertible currency in the prewar period; other countries usually put barriers in the way of the large gold exports and "eagerly watched their gold reserves." Indeed, he argued that by the end of the period most countries kept large gold reserves as a matter of national pride and consequently the key aim of policy was to protect the gold reserve rather than use the gold reserve to protect convertibility. Thus when World War I broke out, the redeemability of currency was immediately suspended in order to safeguard the gold reserve. Like Cassel, Viner (1932) argued that the prewar gold standard would. . . have been found impracticable and would have been generally abandoned ... [if not for] the development of a deliberate and centralized mechanism of control of gold movements, using central bank discount policy and credit control as its chief instruments. (Viner 1932, p. 9) Moreover, the Bank of England pioneered in the development of the technique of central-bank control. England became the manager of the international gold standard. However, the evolution of the Bank of England's effective management of the gold standard emerged as a by-product of the Bank's learning by a process of trial and error to protect its slim gold reserves. The Bank of England in the nineteenth century was primarily a profit-seeking institution and hence tried to minimize its non-interest-bearing gold holdings. However, by the close of the century, the Bank gradually began to accept responsibility for maintenance of an English gold standard: 58

77

The Gold Standard: The Traditional Approach

The Bank of England, at first as the sole issuer of paper money and the most important deposit bank, later under pressure of public opinion and in self-defense against the irresponsibility of the other English banks, partially accepted the role of a central bank with some degree of special responsibility for the mode of operation of the English gold standard and especially for the protection of the convertibility of the English paper currency. (Pp. 12-13) As a consequence, the Bank learned to makes its discount rate effective to protect its gold reserve in the face of an external drain and to hold adequate reserves to meet the exigencies of both external and internal drains. 59 Moreover, according to Viner, in the nineteenth century fluctuations in the exchange rates within the gold points and short-term capital flows aided the adjustment mechanism and reduced the size of gold flows necessary to offset a disturbance to the balance of payments. Finally, Viner made a case for the continuation of the gold standard, despite the fact that it did not produce a stable price level, because a system of inconvertible paper currencies linked by flexible exchange rates would be far worse. 60 We know too little . . . of the possibilities of stabilization to take immediately any major steps in that direction. The hostility of central bankers and the menace of political control are genuine and important factors in the situation. The gold standard is a wretched standard, but it may conceivably be the best available to us. Its past record, bad as it is, is not necessarily conclusive in this respect, as the only alternatives which have actually been tried have, on the whole, had an incomparably worse record. (P. 37) The Conflict between Internal and External Goals A major theme of the interwar period was the potential conflict between internal price stability and a fixed exchange rate under the gold standard. By fixing the price of gold in terms of domestic currency, movements in internal price levels (and real income) would be determined by external-price-Ievel (and real-income) movements. The prewar gold standard period, characterized by both price stability (in a long-run sense) and fixed exchange rates, was an accident of history, never to be repeated. At the same time, flexible exchange rates and abandonment of the gold standard rule were not embraced because of the risk of unstable exchange rates and the fear of the consequences of discretionary policy. That theme is echoed in the works of Keynes, Cassel, Viner, Nurkse, and others. To remedy the conflict between external and internal goals, various schemes were proposed to promote international harmonization of price-level movements under a managed gold standard. The views of major writers of the interwar period are surveyed briefly below.

78

Michael D. Bordo

For Keynes ([1923] 1971) the policy options facing Great Britain in the immediate postwar period were to go back to the gold standard at the prewar parity, which would involve deflation, or else to fix parity after devaluing the pound. The choice between devaluation and deflation was part of a more general dilemma-the choice between price stability and exchange-rate stability. Keynes then asked, "In the light of our answers to the first two questions, is a gold standard, however imperfect in theory, the best available method for attaining our ends in practice?" ([1923] 1971, p. 117). Because of its adverse affects on income distribution, Keynes rejected deflation. When internal and external price stability were incompatible, he chose internal price stability. Under the prewar gold standard, the choice was made in favor of fixed exchange rates and the subservience of the internal price level to external considerations. "We submitted, partly because we did not dare trust ourselves to a less automatic . . . policy, and partly because the price fluctuations experienced were in fact moderate" (p. 126). But the circumstances of the pre-1914 era were partly accidental, and it should not be presumed they would ever be repeated again. The special conditions Keynes cited for the past good performance of the gold standard were first, that progress in the discovery of gold mines roughly kept pace with progress in other directions-a correspondence which was not altogether a matter of chance, because the progress of that period, since it was characterized by the gradual opening up and exploitation of the world's surface, not unnaturally brought to light pari passu the remoter deposits of gold. But this stage of history is now almost at an end. A quarter of a century has passed by since the discovery of an important deposit. Material progress is more dependent now on the growth of scientific and technical knowledge, of which the application to gold mining may be intermittent. (P. 133) Second, the independent influences coming from the demand for gold in the arts and for hoarding purposes in Asia had a steadying influence. Third, central banks allowed their gold reserves to vary slightly, absorbing much of the additional gold produced after major discoveries and reducing some of their accumulated gold when it was relatively scarce. They thus minimized the effects on price levels. Given the special circumstances that made the gold standard successful before World War I, Keynes argued that the standard would be unlikely to work as well in the postwar period. Even if all countries adopted the gold standard-an important condition for it to be successful-the prewar system of balance-of-payments adjustment was "too slow and insensitive in its mode of operation" to handle the "large [and] sudden

79

The Gold Standard: The Traditional Approach

divergences between the price levels of different countries as have occurred lately" (pp. 128-29). Moreover, though short-term capital flows in response to interest-rate differentials helped speed up the adjustment mechanism in the prewar period, especially when the' disturbance was temporary, in the case of permanent disturbances, the adjustment "might obscure the real seriousness of the situation, and enable a country to live beyond its resources for a considerable time at the risk of ultimate default" (p. 130). This problem would be more serious in the postwar period. The case for flexible exchange rates and managed fiduciary money was that balance-of-payments adjustment to external shocks would be much more rapid under flexible than under fixed rates despite the risk of instability.61 Thus Keynes came out strongly against restoration of the classical gold standard by the United Kingdom. In truth, the gold standard is already a barbarous relic. All of us, from the Governor of the Bank of England downwards, are now primarily interested in preserving the stability of business, prices, and employment, and are not likely, when the choice is forced on us, deliberately to sacrifice these to outworn dogma, which had its value once, of £3 17s. 10V2d. per ounce. Advocates of the ancient standard do not observe how remote it now is from the spirit and the requirements of the age. A regulated nonmetallic standard has slipped in unnoticed. It exists. (P. 138) In contrast to his earlier focus on the policy dilemma facing one country alone, Keynes ([1930] 1971) concentrated on the international monetary system as a whole. The interrelationship between central banks in an international fixed-exchange-rate system such as the gold standard was ':lnalogous, he noted, to the relationship between commercial banks and the central bank within a national economy.62 Under the pre-World War I gold standard, commercial banks operated in step within one country and central banks operated in step internationally except that reserve ratios did adapt somewhat to relative scarcity or abundance of gold. Behavior of the long-run price level depended on whether new gold available for reserves was increasing faster or slower than trade of the gold standard countries, which in turn depended on the rate of discoveries and technological improvements in gold mines, the use of gold as currency, the number of countries joining the gold standard, and the growth of real per capita income. Some important differences in the relationship between central banks under the gold standard and between member banks and the central bank in a national economy, however, Keynes observed, were that central banks tended to have more variable reserve ratios;63 that more of its own money returns to the central bank than is the case for a commercial bank;

80

Michael D. Bordo

that there is a higher degree of competition for short-term capital between central banks through varying discount rates than is the case between commercial banks. 64 Thus in an international system, a central bank can pursue an independent policy-oriented primarily to domestic considerations-only within narrow limits and for short periods, with the degree of independence determined by its relative size. Thus, under a fully operative gold standard, "credit cycles have an international character" because member central banks must follow the average behavior. This implies a "real divergence of interest; and we must not expect of central banks a degree of international disinterestedness far in advance of national sentiment and of the behaviour of other organs of national government" ([1930], 1971, p. 257). The dilemma between internal-balance and external-balance considerations for one country on the gold standard is thus apparent. 65 Keynes argued that in a world of perfect capital mobility, the domestic interest rate must correspond to world interest rates: If any country tried to maintain a higher rate than its neighbours, gold would flow towards it until either it gave way or the international system broke down by its having absorbed all the gold in the world. And if it tried to maintain a lower rate, gold would flow out until either it gave way or had to leave the international system through having lost all its gold. Thus the degree of its power of independent action would have no relation to its local needs. (P. 271) The problem arises for a country if its foreign balance is inelastic, and if, at the same time, it is unable to absorb the whole of its savings in new investment at the world rate of interest. It will also tend to happen even where the foreign balance is elastic, if its money costs of production are sticky.... This, then, is the dilemma of an international monetary system-to preserve the advantages of the stability of the local currencies of the various members of the system in terms of the international standard, and to preserve at the same time an adequate local autonomy for each member over its domestic rate of interest and its volume of foreign lending. (Pp. 27172)66 As a solution to the problem, Keynes advocated a number of policies to increase the discretion of national monetary authorities while still remaining on a gold standard. One set of policies to protect a country's domestic stability in the face of "inconvenient fluctuations in the rate of foreign lending" is to maintain a large enough level of reserves: either in gold reserves at home or by holding "liquid balances in foreign centers," by arranging overdraft facilities with other central banks or by "borrowing and lending arrangements between central banks and a supernational

81

The Gold Standard: The Traditional Approach

bank" (p. 278). A second policy is to manipulate the gold points-to create an artificial spread between the official buying and selling price of gold. This can be done by direct authority or by the central bank manipulating the forward rate of exchange. 67 Finally, he suggested direct controls over capital movements. Keynes's ideal solution to the problem of combining an ideal international standard and internal equilibrium was a gold standard managed by a supernational bank. 68 The objectives of a supernational monetary agency would be to ensure long-run stability and to smooth short-run cycles around the long-run trend. One way to achieve long-run price stability would be to adopt a commodity standard based on an international aggregate of commodities. "The long-period trend in the value of gold should be so managed as to conform to a somewhat crude international tabular standard" (p. 351). To solve the problem of short-run disturbances within individual countries, Keynes advocated giving individual central banks more discretion within the fixed-exchange-rate system. The supernational bank would be established by all the world's central banks and would act as a lender-of-Iast-resort to them alone. It would hold as assets gold, securities, and advances to central banks, and its liabilities would be deposits by the central banks. These deposits, called supernational bank money (S.B.M.), would be fully convertible into gold and would serve along with gold as reserves for the member banks. The supernational central bank would then use the normal tools of monetary policy-bank rate and open-market operations-to "maintain . . . the stability of the value of gold (or S.B.M.) in terms of a tabular standard based on the principal articles of international commerce" and to avoid "general profit inflations and deflations of an international character" (p. 360). The interwar criticism of the traditional approach to the gold standard culminated in a provocative and path-breaking article by Whale (1937). He challenged both the price-specie-flow adjustment mechanism and the operation of the rules of the game. Whale referred to four pieces of puzzling evidence: (1) Taussig's finding that the adjustment of national price levels to disturbances occurred much more rapidly than the theory postulated; (2) Beach's finding that gold flows to Great Britain moved procyclically contrary to the classical prediction, and that they were more closely related to interestrate differentials than to price-level differences; (3) the finding that many prewar central banks did not follow the rules of the game, e.g., the central banks of France and Belgium rarely changed their discount rates yet remained on the gold standard; (4) the finding that price levels between regions with varying levels of economic activity moved synchronously, suggesting a linkage through arbitrage rather than adjustment with a lag to specie flows, as in the traditional theory. 69 "Might not the

82

Michael D. Bordo

national price level be similarly determined by the world system of prices?" (Whale 1937, p. 22). On the basis of this evidence, Whale suggested an alternative hypothesis to the classical mechanism: Rather than the demand for money in an open economy with a fixed exchange rate adjusting to the supply of money (and specie flows) as the classical theory predicts, specie flows and the money supply are determined by the demand for money, which in turn is determined by real income and the price level. Thus, according to Whale, an increase in real economic activity, for a given price level (determined by the world price level), would increase the demand for money, causing a balance-of-payments surplus and a gold inflow. Similarly an increase in the domestic money supply would lead to a balance-of-payments deficit, a gold outflow, and a decline in the reserves of the banking system. If the markets of the country are closely linked with foreign markets, the decline of bank reserves should lead to "an almost immediate correction" of the money supply (p. 27). Interest rates also playa different yet still important role in the alternative mechanism: What is contended is that. . . the raising of interest rates did not have the effect of producing a relative reduction of prices in certain countries. High rates in London led rather to a world fall in prices, partly because of the sympathetic movement of rates elsewhere, partly because of the effect on British entrepot trade and British long-term foreign investment. (P. 27) Two important implications follow. First, the classical transfer mechanism of Mill and Taussig must be reinterpreted. According to Whale, a transfer of capital involves a redistribution of spending power. However, rather than this process involving a change in the direction of demand and in the terms of trade, "the redistribution of spending power itself, ... apart from any change in the direction of demand and the terms of trade, may require a redistribution of money ... effected by a movement of gold" (pp. 28-29). Second, "since gold movements ... and discount rate adjustments are displaced from their central position in the process of international price adjustment, the question of 'observing the rules of the game' ... loses much of its importance" (p. 31).

Appendix E

Post-World War II Reinterpreters of the Gold Standard

In the period since World War II, economists have reexamined and reinterpreted the operation of the classical gold standard on the basis of new evidence and new theoretical and statistical tools. The principal

83

The Gold Standard: The Traditional Approach

areas of research are the adjustment mechanism, the role of capital flows, the managed gold standard, and the rules of the game. In the reconsideration of the international adjustment mechanism, several approaches can be distinguished. The Keynesian open-economymultiplier approach (Ford) explains most of the adjustment to the transfer of capital before World War I in terms of changes in economic activity rather than relative price levels. Extension of the classical price-specieflow mechanism (Friedman and Schwartz) focuses on relative price levels and interest rates in the adjustment process. The monetary approach to the balance of payments (Triffin, Williamson, and McCloskey and Zecher) integrates elements of both Keynesian and classical mechanisms and views gold flows by themselves, rather than the effects they have on price levels, incomes, and interest rates, as the equilibrating mechanism in the adjustment of the balance of payments. A major reexamination of the role of capital flows in the transmission mechanism (Morgenstern) raised serious doubts about the classical theory, though the analysis in turn was criticized (Borts). Evidence favorable. to the traditional approach was also presented (Bloomfield). Evidence on the managed gold standard amplified the view that London managed the prewar gold standard (Lindert, Sayers, Goodhart). Other studies showed that the rules of the game were largely violated before 1914 (Bloomfield) and that they were inconsequential (McCloskey and Zecher). The Adjustment Mechanism Ford (1962) downplayed the role of price-level and monetary change in the explanation of the adjustment of the balance of payments under the prewar gold standard. He stressed three themes: (1) the key element in the adjustment mechanism for Great Britain was the change in real income, working through an open-economy-multiplier process; (2) the important link between Great Britain and periphery nations via lending and exports worked primarily through changes in income; (3) there was an asymmetry between the gold standard experience of Great Britain and Argentina (an example of a periphery country). For Great Britain the gold standard mitigated the adjustment to external disturbances, for Argentina the gold standard aggravated it. These divergent experiences reflected the operation of fundamentally different gold standard financial institutions and the presence and absence of "other favorable circumstances." Ford was highly critical of the stylized facts of the traditional approach summarized in the Cunliffe report. His objections were that the report omitted the crucial role of income effects; placed weight on a link between changes in interest rates and economic activity that he believed to be tenuous; treated relative prices as the primary mechanism of adjustment in the balance of trade, based on the doubtful assumption of

84

Michael D. Bordo

elastic demands for imports and exports; inappropriately ignored the feedback effects of changes in income of the major trading nation-Great Britain-on other countries; and made no mention of the important role of foreign lending. Instead, Ford emphasized the key role of changes in income in the balance-of-payments adjustment mechanism.70 A disturbance to the balance of payments such as the exogenous decline in exports would reduce much of the initial balance-of-payments deficit by a decline in real income, working via the multiplier, and would reduce the demand for imports without any change in price levels. 71 The extent to which the balance of payments was equilibrated, without requiring a gold outflow and changes in the money supply, depended on the relative sizes of the marginal propensities to import and to save. 72 In addition, feedback effects of changes in other countries' income due to the fall in British demand for their products would (depending on the size of the other country) also facilitate the adjustment. The second adjustment mechanism was change in the money supply. To the extent the balance of payments was not equilibrated by income change, the resultant gold outflow would reduce the domestic money supply, which would raise interest rates and lower domestic economic activity, imports, and the balance-of-payments deficit. The rise in interest rates would also induce a short-term capital inflow, providing a temporary cushion for the balance of payments. A rise in the central bank's discount rate would further stimulate the reduction in domestic activity and encourage a short-term capital inflow. However, Ford did not regard the Bank of England's playing by the rules of the game as the key element in the process, and he downplayed the accommodating role of centralbank policy in other countries. 73 Finally, Ford presented evidence he regarded as consistent with the key role of income in the adjustment process for Great Britain. Comparing deviations from a nine-year moving average of exports, imports, and income from 1870 to 1914, he found that the parallelism between movements of deviations from nine year moving averages of exports, imports and national income is marked, with some slight tendency for imports to lag behind exports, ... these movements provide powerful evidence both for the exports-incomeimport automatic adjustment mechanism and for the view that in most British booms and slumps variations in export values were a vital factor. (Ford 1962, p. 61)74 According to Ford, British loans to developing nations such as Argentina, Australia, and Canada were primarily transferred through changes in real income, in contrast to the classical approach which emphasized changes in sectoral prices. A British loan to a country like Argentina

85

The Gold Standard: The Traditional Approach

would lead to an increase in imports from Britain and an increase in debt service. When the investment projects in the borrowing country matured, production of exportables would increase, some of which would be imported by Britain. The resultant rise in Argentine incomes would then generate the revenues necessary to service the foreign debt as well as increase demand for British goods. The arrangement worked well in the long run, but in the short run problems could arise for the borrowing country. If the amount lent abroad temporarily exceeded Britain's current-account surplus, a gold outflow from Britain would occur, and the Bank of England would raise Bank rate. As a result, the borrowing country would experience a gold outflow. Important investment projects might be halted before completion leading to a balance-of-payments crisis if the borrower were unable to meet the debt-service obligations. Ford examined in detail the cyclical adjustment mechanism in Argentina under the gold standard (and under inconvertible paper). The typical cycle can be described as follows: A rise in the price of Argentina's primary staple exports, generated by an increase in foreign demand, would produce a current-account surplus, a rise in economic activity, and a gold inflow. The gold inflow would lead to a rise in the domestic money supply. Some of the increased expenditure would go to the production of additional exports, some to the purchase of imports, and some into the nontraded sector-primarily for the purchase of land. A land boom would develop, followed by further foreign investment and further expansion. Ultimately, the boom would be choked off, either by a reduction in foreign economic activity that reduced the demand for Argentinian exports, or by a rise in the discount rate by the Bank of England to offset a gold outflow. Furthermore, were speculation in land to get out of hand, there would ultimately be a collapse in land prices and a domestic liquidity crisis associated with bank failures. Foreign lending would be discouraged. If the internal drain were accompanied by an external drain, then domestic gold reserves would be insufficient to withstand the onslaught and the country would suspend convertibility. This was the sequence of events in 1885 and 1913. Ford argued that the gold standard experience of periphery countries such as Argentina was considerably less favorable than that of center countries such as Britain. For two reasons the cycle in British economic activity was dampened by the gold standard whereas the cycle in Argentina was aggravated. First, Britain's position as the center of the world's money market meant that in the face of an external drain, a rise in Bank rate would draw on short-term capital and sterling balances in London without loss of gold, while Argentina lacked the cushioning financial institutions. 75 Second, it was easier for a creditor nation to obtain immediate relief from external pressure by reduced foreign lending, "whereas

86

Michael D. Bordo

. . . in a period of stringency it was difficult or even impossible for a debtor country (with a past history of depreciation) to offset gold exports by increasing its borrowing abroad" (p. 182). In their monumental study of U.S. monetary history, Friedman and Schwartz (1963) treated the gold standard in a traditional way in their analysis of the role of monetary forces and of the price-specie-flow mechanism to explain balance-of-payments adjustment under the gold standard; in their application of the commodity theory of money to explain secular price movements; and finally, in their discussion of the role of central-bank management under the gold standard. They described the role of the money supply under an international specie standard that the United States adhered to from 1879 to 1933, as follows: The amount of money in anyone country must be whatever is necessary to maintain international balance with other countries on the same standard, and the amount of high-powered money will alter through imports and exports of specie in order to produce this result . . . the amount of high-powered money is a dependent rather than an independent variable, and is not subject to governmental determination. (Friedman and Schwartz 1963, pp. 51-52) Moreover, for a country which is an economically minor part of the gold standard ... the major channel of influence is from fixed rates of exchange with other currencies through the balance of payments to the stock of money, thence to the level of internal prices that is consistent with these exchange rates.... [However] the links have much play in them, so that domestic policies can produce sizable short-term deviations in the stock of money from the level dictated by external influences. (Pp. 89-90) Under a flexible-exchange-rate regime, by contrast, such as the greenback period from 1862 to 1879, "the amount of high-powered money is determined by governmental action" (p. 51). Adjustment to both external and internal disturbances was facilitated by the classical relative-price-Ievel adjustment mechanism and capital flows. The events of the period 1879-82 are analyzed in these terms. Good harvests in 188Q-81 led to an increase in U.S. exports. The resulting increase in demand for dollars implied a relatively higher U.S. price level consistent with balance-of-payments equilibrium. 76 Pending the rise in prices, a gold inflow ensued that led to a rise in the money stock and the price level. At the same time, a gold outflow from Great Britain led to a monetary contraction and a decline in the price level. As a consequence the Bank of England raised Bank rate, reversing the gold flow to the United States.

87

The Gold Standard: The Traditional Approach

The gold inflow was a passive reaction which temporarily filled the gap in payments. In its absence, there would have had to be an appreciation of the dollar relative to other currencies-a solution ruled out by the fixed exchange rate under the specie standard-or a more rapid rise in internal U.S. prices. At the same time, the gold inflow provided the basis and stimulus for an expansion in the stock of money and thereby a rise in internal prices at home and downward pressure on the stock of money and prices abroad sufficient to bring an end to the necessity for large gold inflows. It would be hard to find a much neater example in history of the classical gold standard mechanism in operation. (P. 99) Other episodes are treated similarly: prices and incomes, aided by capital flows, adjust to maintain external balance. 77 The brunt of the adjustment was sustained in some cases by changes in the money stock, in others by changes in velocity or real output. Economically, these were the channels whereby a necessary adjustment was worked out. They were not the forces determining what adjustment was necessary.... The discipline of the balance of payments under the gold standard enforced that adjustment and determined its size. (P. 101 )78 Friedman and Schwartz applied the classical commodity theory of money to explain secular price-level movements: Under a specie standard confined to a single country, or for the world as a whole under an international standard, the existing amount of specie is determined by the available physical stock plus the relative demand for monetary and other uses; and changes in the amount of specie, by relative costs of production of specie and other goods and services. (P. 52) They explained the secular deflationary episode of 1879-96 by [a] combination of events, including a slowing of the rate of increase of the world's stock of gold, the adoption of the gold standard by a widening circle of countries, and a rapid increase in aggregate economic output, ... despite the rapid extension of commercial banking and of other devices for erecting an ever larger stock of money on a given gold base. (P. 91) The subsequent turnaround in prices and secular-inflation episode is explained by fresh discoveries of gold in South Africa, Alaska, and Colorado combined with the development of improved methods of mining and refining, especially the introduction of the cyanide process. These occurred during a period when there were few further important extensions of the gold standard yet a continued development of devices for "economizing" gold. (P. 91)

88

Michael D. Bordo

Moreover, the period of secular deflation "was an important factor in stimulating the search for gold and for economical techniques for extracting gold from low-grade ore" (p. 188). Finally, in their discussion of the sterilization of gold inflows in the 1920s by the Federal Reserve system, Friedman and Schwartz explain how violations of the rules of the game weakened the adjustment mechanism of the gold standard. The sterilization of gold could be justified as a means of insulating internal monetary conditions from external changes. Its international effect, however, was to render the maintenance of the international gold standard more difficult. Suppose all countries linked by a gold standard were to sterilize gold flows. Gold flows would then set in train no forces tending to bring them to a halt or to reverse them. The system could last only as long as the flows resulted from purely temporary imbalances of sufficiently small magnitude to right themselves before draining the countries losing gold of their reserves. The effect would be to insulate the countries from minor adjustments at the cost of letting them accumulate into major ones. (P. 283) Friedman and Schwartz attributed the Federal Reserve system's failure to stem the banking crisis of 1931 to its failure to follow the classical medicine prescribed by Bagehot for central-bank operations in the face of both an external and an internal drain-to lend freely but at a high discount rate (p. 395). Emphasis on the "stylized facts" of the gold standard on intercountry adjustment via specie flows that produced relative price-level adjustments, aided by capital flows and by central banks following the rules of the game, failed, according to Triffin (1964), "to bring out the broader forces influencing the overall pace of monetary expansion on which individual countries were forced to align themselves" (p. 2). Evidence damaging to the traditional story included: "enormous degree of parallelism-rather than divergent movements-between export and import fluctuations for anyone country, and in the general trend of foreign-trade movements for the various trading countries. . . from 1880 to 1960" (p. 3); and "overall parallelism-rather than divergence-of price movements, expressed in the same unit of measurement, between the various trading countries maintaining a minimum degree of freedom of trade and exchange in their international transactions" (p. 4); downward wage rigidity among countries that maintained exchange stability; Bloomfield's (1959) evidence of the failure of most central banks to play by the rules of the game; the ineffectiveness of changes in discount rates in many countries to stem capital flows or change relative prices; and the important role of long-term capital flows in maintaining enduring balance-of-payments disequilibrium without relative price adjustment.

89

The Gold Standard: The Traditional Approach

Triffin accordingly painted a different picture of how the gold standard worked. The most important aspect of the gold standard was exchange-rate stability maintained by "pressures for international harmonization of the pace of monetary and credit expansion [between central banks] similar . . . to those which. . . limit divergent rates of expansion among private banks within each national monetary area" (p. 11), enforced by the constraint of convertibility into gold of fiduciary money. 79 Given stable exchange rates, then national export prices remained strongly bound together among all competing countries, by the mere existence of an international market not broken down by any large or frequent changes in trade or exchange restrictions. . . . National price and wage levels also remained closely together internationally, even in the face of divergent rates of monetary and credit expansion, as import and export competition constituted a powerful brake on the emergence of any large disparity between internal and external price and cost levels. (P. 10) As a consequence, monetary expansion generating inflationary pressures could not be contained within the domestic market, but spilled out directly . .. into balance of payments deficits rather than into uncontrolled rises of internal prices, costs, and wage levels. These deficits led, in turn, to corresponding monetary transfers from the domestic banking system to foreign banks, weakening the cash position of domestic banks and their ability to pursue expansionary credit policies. (Pp. 10-11) Central banks could only temporarily slow down the adjustment process by engaging in offsetting open-market operations or using other tools of monetary policy, because ultimately their international reserve ratios would decline. Thus, in the gold standard world of fixed exchange rates, price levels were closely linked together and the balance-of-payments deficit (surplus) reflected both money-market disequilibrium and the method by which it was eliminated. In an approach similar to that of Triffin, Williamson (1961, 1963) reinterpreted U.S. experience under the gold standard. He urged analysis of the balance of payments in a general-equilibrium context, with the long-swing cycle in the growth of real output determining specie and capital flows. 80 Increased real growth would lead to both an excess demand for goods (a balance-of-trade deficit), an excess supply of bonds (a capital inflow), and an excess demand for real balances. The excess demand for money would be satisfied by a specie inflow, with little change in the price level, accompanied by a long swing in capital inflows (Williamson 1961, p. 379). The external balance was both a cause as well as a

90

Michael D. Bordo

reflection of the long swing, since in the 1830s it was British demand for U.S. cotton that was the key source of the long swing in output, which in turn induced British investment in railroads and canals. 81 In an important article applying the recently developed monetary approach to the balance of payments to the operation of the classical gold standard from 1880 to 1914, McCloskey and Zecher (1976) extended the challenge to the traditional approach intimated by Marshall in the 1880s, endorsed by Whale in the 1930s, and repeated by Triffin and Williamson. The monetary approach states that for an open economy with fixed exchange rates, the national stock of money, rather than prices, adjusts to changes in the public's demand for money (see Frenkel 1971; Johnson 1976; Mundell 1971). Contrary to the Hume price-specie-flow mechanism-which postulates significant lags in the adjustment of price-because of instant arbitrage, according to the monetary theory, no lags are observed in the adjustment of world prices. In the most rigid version of the theory, an increase in the demand for money ca~not reduce prices because prices of internationally traded goods are determined in world markets and kept comparable in different countries by international arbitrage, and prices of domestic goods and services are kept in line with prices of internationally traded goods by domestic arbitrage. The reduction in the public's demand for goods and securities leads to reduced imports and expanded exports on the goods side and to higher interest rates and capital imports on the securities side. The current account or the capital account or both move into surplus. To prevent appreciation of the currency, the monetary authority buys foreign exchange from its nationals, paying out newly created high-powered money. The increase in high-powered money leads to a multiple expansion of the domestic quantity of money which continues until the public's demand is satisfied. In open economies on a fixed exchange rate, a once-for-all increase in the quantity of money in one country and a decrease in another would produce a balance-of-payments deficit in the first and surplus in the second and lead to a flow of money to the second until equilibrium was reestablished. If a monetary authority in one country alone increased high-powered money, that would be equivalent to an increase in the world money supply. That country would experience a temporary balance-of-payments deficit until the world money supply was redistributed in proportion to the size of the country of issue, and the world price level would rise accordingly. In the long run, domestic monetary policy in a small country has a negligible influence on international prices, although in the short run the monetary authority can affect its price and income level by open-market sales (purchases) equal to its balance-of-payments surplus (deficit) that will maintain the national money stock below (above) its equilibrium value. The closer the Ijnks among world commodity markets, the higher the degree of capital mobility, the less scope for

91

The Gold Standard: The Traditional Approach

independent monetary policy in the short run. The greater the elasticity of substitution between traded and nontraded goods, the less successful will such policy be. In the long run, however, independent monetary policy is inconsistent with fixed exchange rates. McCloskey and Zecher tested the key assumptions of commodity arbitrage by examining correlations among price changes between countries and between regions within countries. For traded goods such as wheat, they found synchronous correlations equally high between regions as between nations, unlike the case of nontraded goods such as labor services and bricks. For overall price indexes, they found a significant correlation between the U.S. and U.K. wholesale price indexes, less so for GNP deflators, and even less for consumer price indexes. The higher correlation for the wholesale price than for the other index undoubtedly reflects the larger share of traded goods in the former. 82 They conclude: What has been established here is that there is a reasonable case . . . for the postulate of integrated commodity markets between the British and American economies in the late nineteenth century, vindicating the monetary theory. There appears to be little reason to treat these two countries on the gold standard differently in their monetary transactions from any two regions within each country. (McCloskey and Zecher 1976, pp. 379-80) They also cite less conclusive evidence in favor of capital-market arbitrage. They tested their model by comparing gold flows-predicted by a simple demand for money function less the money supply produced by domestic credit expansion-with actual gold flows and found a close relationship. In their view, we have established at least a prima facie case for viewing the world of the nineteenth century gold standard as a world of unified markets, in which flows of gold represented the routine satisfaction of demands for money. (P. 385) Capital Movements Both short-term and long-term capital movements play an important role in the traditional approach to the gold standard. Short-term capital flows were to act as an equilibrating mechanism, to economize on gold flows and to reduce the burden of adjustment by changes in relative price levels. Private short-term capital movements, assumed to be highly responsive to interest-rate differentials (induced by changes in the discount rate), would act to equate interest rates (adjusted for exchange risk) in different money markets. Morgenstern examined the evidence in favor of the "solidarity hypothesis," that arbitrage would ensure uniformity of interest-rate differentials to exchange risk (measured by the difference of

92

Michael D. Bordo

the exchange rate from the gold points), and found it to be inconsistent with the principles of the gold standard. Borts, in criticizing both his methodology and data, put Morgenstern's conclusions into serious doubt. Bloomfield found evidence that private short-term capital flows responded to interest differentials and expected exchange-rate changes as predicted. Long-term capital flows, a key disturbing force in the traditional approach, were believed to have been well accommodated by the gold standard balance-of-payments adjustment mechanism. They allowed both developing debtor nations and mature creditor nations to run persistent balance-of-payments disequilibria without requiring adjustment in the balance of trade. Bloomfield's study supported the integral role of long-term capital flows in the development of the "Atlantic economy." Morgenstern (1959) examined short-term-interest-rate data of different maturities, exchange rates, and the gold points for four key gold standard countries: Great Britain, France, Germany, and the United States in the periods 187~1914 and 1925-38, subjecting the "assumption of international solidarity of money markets" to three tests. 83 The first test compared derived exchange rates, based on cross rates in third markets, with the actual exchange-rate series. If arbitrage were effective, differentials between the series would vanish. Morgenstern found evidence of differentials persisting in both periods, but more so after World War I, with the greatest deviations occurring in periods of crisis or disturbance such as the 1890 Baring crisis. The second test identified deviations of exchange rates beyond the median gold import and export points. Violations persisted for long periods, with greater frequency of deviations beyond the gold export point than beyond the gold import point. The· third test compared market-interest-rate differentials with the "absolute maximum permissible" differential determined by the distance between the gold points. Violation of the principle occurred in both prewar and postwar periods. Morgenstern concluded that we ought to view the period of the classical gold standard as inadequately described by the typical mechanism at least in one respect: the interaction between two and more money markets via exchange rates and interest rates is not nearly as precise and rigid as postulated (Morgenstern 1959, p. 569) He explained these results by "friction" and central-bank intervention, suggesting the replacement of the interpretation of the gold standard as a mechanism by the notion that central banks and other market participants engage in game strategy in a struggle for gold. Borts (1964) criticized both Morgenstern's methodology and his data. Borts attributed the results of the first test comparing cross with own exchange rates to the

93

The Gold Standard: The Traditional Approach

methods of making exchange quotations. Morgenstern's data were either exchange quotations in the form of single prices at which brokers cleared the market or the average of bid-ask spreads. For Borts the bid-ask spreads were likely measures of transactions cost; hence "what appears to be an opportunity for arbitrage profit could be the difference between the clearing price and the prevailing practice of giving quotations" (Borts 1964, p. 225). Borts found the results of the second test comparing the spot exchange rate with the gold points ambiguous. In deriving the gold points, Morgenstern assumed circumstances favoring his results since he took into account neither different methods of covering exchange risk nor frequent operations by the monetary authorities on the gold points. These ambiguities could be sorted out, according to Borts, by directly examining gold movements "in an effort to confirm the position of the exchanges with regard to the gold points" (p. 227).84 Finally, Morgenstern's third test comparing interest differentials to the maximum exchange risk on uncovered funds was faulty, in Borts's view, because (1) "[he converted] the percent movement in the exchange rate into a percent difference on one-year paper. The exchange risk. . . has no time dimension and exists no matter what the maturity of the paper held" (p. 227); (2) he did not account for possible forward cover. 85 Bloomfield examined the role of both short-term (1963) and long-term (1968) capital flows under the pre-1914 gold standard. He suggested that private short-term capital movements served to equilibrate the balance of payments in the short run by tending to reduce the size of gold flows or acting as substitutes for changes in official exchange holdings (Bloomfield 1963, p. 44). To test the latter hypothesis, he compared the signs of the first differences of annual changes in the stock of net foreign short-term assets of commercial banks with annual changes in the stock of centralbank gold and foreign-exchange reserves of the Scandinavian countries and found the postulated negative relationship (p. 58).86 Long-term capital flows in the classical gold standard period in the form of portfolio investment came mainly from Britain and France, followed by Germany; the bulk of the funds went to the developing countries of the new world to finance the development of infrastructure and production and exportation of primary products. Following the work of Williamson (1964), Cairncross (1953), O'Leary and Lewis (1955), and Thomas (1973), Bloomfield (1968, pp. 18-34) found evidence of a longswing cycle in long-term capital movements. Consistent with their theories, he found for debtor countries positive correlations between capital imports and indicators of domestic investment such as domestic building, and between capital imports and net immigration; for creditor countries a negative correlation between capital exports and domestic investment, and a positive one between capital exports and net emigration. 87 These

94

Michael D. Bordo

results, plus the evidence that similar long-swing movements in the United States and Canada in turning points were inversely related to swings in British building and economic activity in general, gave support to Thomas's (1973) thesis of an Atlantic economy. Finally, contrary to theory, interest rates did not explain movements in annual data of British capital exports and U.S. capital imports. In the multiple regressions that Bloomfield estimated, domestic and foreign investment activity were the significant regressors. The Managed Gold Standard The mainstream view of the classical gold standard that emerged from the interwar period (appendix D) was that it evolved into a sterling standard. The concentration of world capital, commodity, and gold markets in London made the pound sterling an attractive reserve asset in addition to gold and made it easier for the Bank of England to control its gold-reserve ratio by altering Bank rate, in the process affecting the policies of other central banks and influencing economic conditions both at home and abroad. 88 In the post-World War II period, the degree of management of the prewar standard was further explored. Key currencies other than sterling were shown to have been important in the pre-World War I period, though sterling's role was still predominant, and the Bank of England's use of the Bank-rate weapon was deemed to be less effective than traditionally believed (Lindert 1969). The ways in which the Bank of England managed the gold standard were described (Sayers 1936, 1957; Goodhart 1972). Following Bloomfield (1963), Lindert (1969, pp. 13-27) found that holdings of several major currencies~the pound, the franc, and the mark-represented an important and growing fraction of the international reserves held by many countries in the period 1900-1913. 89 As expected, London was the primary reserve center for the world but francs and marks were popular on the Continent. According to Lindert, key currencies were held for the interest income they earned; they involved lower transaction and transportation costs than gold, and maintenance of balances in a foreign currency such as sterling often gave easier access to credit in the London money market. Perhaps the key reason these currencies were held was that their good brand name guaranteed with certainty that they could be converted into gold on demand. In the case of the pound, the location in London of the international money market and the world's gold market likely enhanced its brand name. Because their currencies were widely held, the central-reserve countries could run larger balance-of-payments deficits than otherwise and

95

The Gold Standard: The Traditional Approach

could maintain them longer. However each central bank was sensitive to its gold-reserve ratio and when it declined the bank would react by raising its discount rate. 90 Lindert found that when Great Britain raised Bank rate, other central banks on the Continent would respond, but London had stronger "pulling" power and could always attract short-term funds from the Continent. 91 A flow of short-term funds proceeded from peripheral European nations, running balance-of-payments surpluses to Paris and Berlin, and then to London. The asymmetry in discount-rate drawing power may be explained by the fact that for the center countries, short-term foreign assets tended to be less liquid than short-term liabilities; "since tighter monetary policy tends to stimulate shifts toward liquid assets, banks would react by seeking greater key-currency balances at the expense of bills on lesser centers" (Lindert 1969, p. 78). Thus, the indisputable position of London as the dominant financial center during the prewar years meant that "other countries, had, therefore, to adjust their conditions to hers."92 After the publication of Lombard Street (Bagehot [1873]1969), the Bank of England began to take seriously its responsibilities for both maintaining convertibility and preventing domestic monetary instability, doing so not by increasing its gold reserves, as Bagehot suggested, but by altering its discount rate whenever its gold reserves were threatened (Sayers 1951, pp. 109-10). Sayers described the prewar techniques used to make Bank rate "effective" in the sense of linking it tightly to short-term market rates. The methods included open-market operations, eligibility requirements, and the switch to a penalty rate in 1878. 93 In addition, under special circumstances, when it feared the internal repercussions of raising Bank rate, the Bank would protect its gold reserve by using "gold devices"-direct operations in the gold market. It is generally agreed however that the Bank achieved full control over its reserves after 1890 (Presnell 1968). According to the traditional approach, a rise in Bank rate would equilibrate the balance of payments via two principal channels: by inducing a short-term capital inflow (reducing an outflow) and by checking domestic economic activity, the domestic price level, and the price of imports. Lindert (1969, pp. 43-44), Bloomfield (1959, p. 42), and Goodhart (1972, chap. 15) evaluated the evidence for the domestic channel as indicating at best a weak and protracted adjustment with the case for the link via capital flows sacrosanct. The case made against the domestic channel was twofold: the limited response of the domestic money supply to changes in the Bank of England's gold reserve, and the limited response of domestic economic activity to changes in the interest rate. With respect to the money supply, Goodhart (1972, p. 208) was unable

96

Michael D. Bordo

to detect any close, positive association between the cash base (reserves) of the commercial banks, represented by bankers' balances at the Bank of England, and the Bank's gold reserves. According to him, the direction of causation was the reverse of the traditional one. An increase in economic activity would lead to an increase in commercial-bank lending and deposits, and the increase in bank reserves required to maintain stable reserve ratios would be supplied by the Bank of England at the expense of its other discounts, thus producing both a lower gold-reserve ratio at the Bank of England (the proportion) and a higher discount rate. The rise in the discount rate would then lead to a gold inflow restoring the Bank's proportion. 94 Goodhart concluded (1972, p. 219): Indeed, on this view, the great years of the gold standard (189~1914) were remarkable, not because the system enforced discipline and fundamental international equilibrium on this country by causing variations in the money supply, but because the system allowed for the development of such large-scale, stabilising and equilibrating, shortterm, international capital flows, that autonomous domestic expansion was rarely disrupted by monetary or balance of payments disturbances. With respect to domestic economic activity, Tinbergen (1950, p. 133) found that the influence of interest rates on the course of investment activitywhich is the chief influence interest rates exert, according to our results-is only moderate. A rise in interest rates depresses investment activity, but only to a modest extent. and Pesmazoglu (1951, p. 61) that variations ... in the long-term rate of interest did not have an important influence on fluctuations of British home investment between 1870 and 1913. 95 The Rules of the Game According to the traditional approach, the key objective of monetary policy was to maintain convertibility into gold and to use monetary policy, specifically the discount rate, to facilitate internal adjustment to external disequilibria. However, Bloomfield (1959, pp. 25-26) found that while central banks were primarily concerned with maintaining convertibility, their policy actions were discretionary, not automatic: Not only did central banking authorities ... not consistently follow any simple or single rule or criterion of policy, or focus exclusively on considerations of convertibility, but they were constantly called upon to exercise, and did exercise, their judgment on such matters as whether or not to act in any given situation and, if so, at what point of

97

The Gold Standard: The Traditional Approach

time to act, the kind and extent of action to take, and the instrument or instruments of policy to use. . . . It does indicate that discretionary judgment and action were an integral part of central banking before 1914, even if monetary management was not oriented toward stability of economic activity and prices in the broader modern sense. In a test of the extent to which central banks under the pre-1914 gold standard played by the rules of the game, Bloomfield interpreted the rules as meaning' 'that central banks were supposed to reinforce the effect of these flows [gold flows] on commercial bank reserves, not merely not to neutralize them. This implied ... that central banks were supposed to lower their discount rates in the face of persisting gains of gold. . . and to raise them when there were persisting losses." Such a policy would have the effect of "increasing central bank holdings of domestic earning assets when holdings of external reserves rose, and of reducing domestic assets when reserves fell" (Bloomfield 1959, p. 47). Following Nurkse's approach in his examination of central-bank behavior in the 1929-38 period, Bloomfield compared year-to-year changes in international and domestic assets for twelve central banks in the 1880-1914 period and found that, in the case of every central bank, the changes in the two classes of assets were more often than not in the opposite direction (wit!1 the Bank of England coming close to being the exception to the rule). Thus, Far from responding invariably in a mechanical way, and in accord with some simple or unique rule, to movements of gold ... , central banks were constantly called upon to exercise, and did exercise, discretion and judgment in a wide variety of ways. Clearly the pre-1914 gold standard system was a managed and not a quasi-automatic one from the viewpoint of the leading individual countries. (Bloomfield 1959, p.60) Based on their reinterpretation of the classical gold standard according to the monetary approach to the balance of payments, McCloskey and Zecher (1976) denied that the Bank of England could have "acted as conductor of the international orchestra" as in Keynes's ([1930] 1971) description, but "was no more than the second violinist, not to say the triangle player, in the world's orchestra" (McCloskey and Zecher 1976, pp. 358-59). The monetary theory holds that the world's economy is unified by arbitrage and that the world's price level is determined by the world's money supply. Then the Bank of England's potential influence on prices (and perhaps through prices on interest rates) depended simply on its power to accumulate or disburse gold and other reserves available to support the world's supply of money.

98

Michael D. Bordo

. . . Only by decreasing the securities and increasing the gold it held . . . could the Bank exert a net effect on the world reserves . . . Had the Bank in 1913 sold off all the securities held in its banking department it would have decreased world reserves by only 0.6 percent; had it sold off all the gold in its issue department, it would have increased world reserves by only 0.5 percent. (McCloskey and Zecher 1976, p. 359) Finally, according to McCloskey and Zecher, the central banks of the world ignored the rules of the game-stipulating that a deficit in the balance of payments be accompanied by contractionary monetary policy, a surplus by expansionary policy-because the rules were "inconsequential." According to the monetary theory "neither gold flows nor domestic deflation have effects on prevailing prices, interest rates, and incomes" (p. 361) since the central bank of a country adhering to the gold standard could only control the composition of the monetary base as between international reserves and domestic credit, not its total amount.

Notes 1. Some of the material covered is drawn from, and may overlap, earlier surveys by Viner (1937), Fetter (1965), and McCloskey and Zecher (1976). 2. The meaning of the price of gold is its relative or real price or the purchasing power of gold. This meaning is not explicitly stated by all writers but presumably it is what they intended. It is the fixed mint price of gold in terms of national currency divided by some commodity price index. 3. For some writers changes in incomes rather than changes in relative price levels produced the adjustments. 4. Some would argue that Adam Smith ([1776] 1976) should be included in this list. Smith subscribed to most of the views expressed by the other classical economists. He viewed the world specie stock and its exchange value in the long run as determined by the richness of gold and silver mines. He also stressed the social saving of using paper money for specie up to the point of convertibility. However, he did not discuss the price-specie-ftow mechanism and his belief in the real-bills doctrine (see Mints 1945) has resulted in the downgrading of his contribution to the traditional view. Recently, however, Girton and Roper (1978) and Laidler (1981) have argued that Smith may have been correct after all, if interpreted according to the recent monetary approach to the balance of payments. According to these authors, Smith viewed a country such as contemporary Scotland as a small open economy on a fixed exchange rate with an exogenously determined price level. Under such circumstances, the quantity of money would adjust to the demand for money-a result consistent with both the real-bills approach and the absence of any change in the terms of trade. 5. This discussion draws heavily on Bordo 1983. 6. Cantillon [1931] 1964, bk. 2, chap. 16. Indeed the intrinsic value of the precious metals and hence long-run supply is determined by the cost of production of the least productive mine (p. 101), Le., by marginal cost. Temporary variations of the exchange

99

The Gold Standard: The Traditional Approach

value of money from its intrinsic value can be caused by changes in the demand for precious metals (for nonmonetary uses), but in the long run exchange value will equal intrinsic value (p. 97). 7. Other aspects of the traditional view mentioned in Cantillon's Essai include a discussion on capital mobility (pp. 191-93), the gold points (pp. 253,255,257,261), the use of bills of exchange to settle international balances (pp. 229,245,247), and the operation of the forward exchange market (p. 259). 8. Viner (1937, pp. 316,319) argued that the distinction between the law of one price as pertaining to the equality of prices stated in a common currency of identical traded goods, allowance being made for transport costs, and the price-specie-flow mechanism, which involved changes in the relative prices of import and export goods (the terms of trade), was held by all the classical writers. Samuelson (1971) viewed the "inconsistency" as an error of interpretation by Viner and others. According to him, when prices in each country are measured relative to wages, the equality of identical-traded-goods prices will hold after a gold discovery initially affects the money supply and prices in one country. 9. Also paper has the advantage that it is flexible and can be supplied quickly in periods of crisis (Ricardo [1816] 1951, p. 58). 10. In addition the issue of bank notes will produce the same effect as a gold discovery (Ricardo [1811] 1951, p. 55). 11. Ricardo also described the force of arbitrage in maintaining equality between prices (the value of money) in the country and in London ([1811] 1951, p. 87). 12. Because it was legal to export bullion, but illegal to export coin, the price of bullion would initially rise above the mint price. Eventually, however, people would evade the prohibition and melt coin into bullion (Ricardo [1811] 1951, p. 64n). 13. According to Sayers (1953), Ricardo perceived that the essential condition of a gold standard is not gold coinage, but convertibility into gold for international transactions. 14. Chronologically, Thornton preceded Ricardo by several years, but it is convenient to present his views following those of Ricardo. 15. The size of the gold reserve should be determined by the degree of confidence "between independent countries" and the "largeness of the balance between the independent places" (Thornton [1802] 1978, pp. 155-56). 16. There may, however, be distribution effects between gold mining and other countries since the reduction in the use of gold as money will reduce the price levels of mining countries relative to those of the rest of the world. In addition, there will be first-round effects depending on how the money is issued-by private bankers or the government. Money issued by bankers would lead to a fall in the interest rate, a capital outflow, and a gold outflow, with no effect on the price level. If issued by government or by private manufacturers, the initial effect would be on domestic prices, leading to a current-account deficit and a gold outflow. 17. Here Mill ([1865] 1961, p. 625) cited Ricardo (Principles, 3rd ed., p. 143). "Gold and silver having been chosen for the general medium of circulation, they are, by the competition of commerce, distributed in such proportions amongst the different countries of the world as to accommodate themselves to the natural traffic which would take place if no such metals existed, and the trade between countries were purely a trade of barter." 18. This section is based on Bordo 1975. Note that Jevons ([1884] 1964) considered the same issue as Cairnes and in the course of his investigation constructed a price index to measure the extent of depreciation of the value of gold caused by the discoveries. 19. Cairnes also tested Ricardo's theory of comparative advantage in predicting Australia's switch from being a net exporter to a net importer of agricultural products, and the Cantillon transmission mechanism to predict the dispersion of price changes between different commodity groupings. 20. Cairnes argued that England and the United States, because of their efficient banking and credit systems, should gain relative to France and the rest of the Continent

100

Michael D. Bordo

since they would require less gold to finance the necessary price rise. The gold flowing out of England into France would serve to displace silver, since France was on a silver standard. The displacement of silver would for a time act as a parachute preventing French prices from rising until gold completely replaced silver; at the same time the released silver would flow eastward and into the silver currencies of Asia, augmenting gold flowing there directly. At this point in an argument similar to that of Jevons ([1884] 1964), Cairnes pointed out that the "parachute effect" would not be as important as Chevalier (1859) maintained in preventing a rise in French prices, since gold and silver were substitutes, so that as gold currency substituted for silver, the price of silver would tend to fall along with the price of gold. 21. See Viner 1937 and Fetter 1965 for a complete history of the debates. 22. That was the recommendation of the currency school which argued that a mixed currency--one consisting of both specie and notes-should be made to operate as if it were a pure specie standard. See Fetter 1965, p. 130. 23. This was the banking school's position. For them convertibility into gold and free competition in banking were sufficient to maintain an adequate money supply consistent with both internal and external balance. See Viner 1937, pp. 222-24. 24. See Viner 1937, pp. 264-70, and White (1981) who has reformulated the currencybanking schools debate as turning on the question whether to centralize the right of note issue in a single institution or to allow competition ("free banking"). 25. In addition, Bagehot was in favor of publication of the accounts of the Banking Department of the Bank of England and employing more professionals and fewer amateurs in the government of the Bank ([1873]) 1969, pp. 302, 72). 26. Also Marshall stated that "a person who had to bring home the returns of any sales in a country had to elect what commodity he would bring, and the question whether he should bring lead or tin was governed ... by exactly the same conditions as whether he should bring lead or gold. If after allowing for expenses of carriage you get a little more by bringing home the lead and selling it than by bringing home the tin, he would choose the lead; if he would get a little more by bringing home gold and selling it, he would bring home the gold" (1926, p. 121). 27. However this is only a temporary effect; in the long run changes in gold have no effect on the rate of interest, which is determined by "the average profitableness of different business" (Marshall 1926, p. 130). 28. The views on the gold standard of Fisher's contemporary, J. Laurence Laughlin (1903), according to Girton and Roper (1978), anticipated the monetary approach to the balance of payments. A critic of the traditional approach, Laughlin disagreed with the Hume price-specie-flow mechanism which postulated lengthy lags until relative-price-Ievel differences led to corrective gold flows. He argued (in a manner similar to Angell, see p. 66) that commodity arbitrage tended to keep price levels of gold standard countries always in line. In addition, in the tradition of Adam Smith, he reversed the causation of money and prices of the classical quantity theory. According to Laughlin, for an open economy, the supply of money adjusted through the balance of payments to the demand for money, determined in turn by the "needs of trade." Thus gold did not flow to equilibrate price levels but to satisfy an excess demand for (supply of) money. 29. Fisher cites a number of reasons why prices may not be equal. "Distance, ignorance as to where the best markets are to be found, tariffs, and costs of transport help to maintain price differences.... Practically, a commodity will not be exported at a price which would not at least be equal to the price in the country of origin, plus the freight" ([1922] 1965, p.92). 30. Fisher also considered the case of bimetallism and rejected it on grounds similar to those noted by other classical writers-it tends to degenerate into monometallism whenever the market ratio of gold to silver diverges from the official ratio ([1922] 1965, pp. 123,325). He also argued against an irredeemable paper standard because of the inevitable tendency of governments to overissue (p. 131), against Marshall's symmetallism scheme because it

101

The Gold Standard: The Traditional Approach

bases the standard on too narrow a base of commodities (p. 328), and against a tabular standard: how to express money to conform to that standard was a problem (p. 335). 31. On the basis of a statistical investigation in 1933 of price-level movements of twenty-seven countries, Fisher found that the price levels of gold standard countries tended to move together, those of silver standard countries moved together, and the average price level of each group varied with changes in the relative prices of gold and silver. Moreover, he found evidence for a short-run tradeoff between price-level changes and changes in trade and employment within countries. Finally, evidence that countries not on the gold standard during the Great Depression, e.g., Spain and China, avoided the deflation suffered by gold standard countries and the concomitant contraction in output and employment, convinced him that "depressions travel internationally ... the infection is carried chiefly via the monetary standard" (1935, pp. 15-16). 32. See lonung 1979 for a discussion of Wicksell's theory of price-level movements. Basically Wicksell argued that price levels will rise cumulatively if the market rate of interest, determined in the loan market, diverged from the natural rate of interest, determined by the forces of thrift and productivity. If the market rate were below the natural rate, prices would rise cumulatively, the price rise only being arrested by a gold outflow that would reduce the banking system's reserves, causing banks to raise their loan rate. When the market rate of interest was above the natural rate, a cumulative deflation would occur. Wicksell explained periods of secular inflation and deflation in the nineteenth century using this approach. In contrast to Wicksell, the Swedish economist Cassel applied classical doctrine, explaining episodes of world inflation and deflation by the growth of the world's gold supply relative to the growth in demand for gold, the former influenced primarily by the production of new gold, the latter by the growth of real income. 33. Several years later A. G. Silverman (1931) tested the classical theory that capital exports, ceteris paribus, would lead to a rise in the price of imports relative to the price of exports. Using British data over the period 188~1913, he compared "year to year percentwith year to year absolute age changes ... for the ratio of import to export prices differences for Hobson's indirect estimates of capital exports expressed in terms of its average deviation" and found that over the whole period "an annual increase or decrease in capital exports is more often than not accompanied by an opposite change in the ratio of import to export prices." Thus he concluded that "the orthodox analysis ... does not seem to be borne out. For most of the period under consideration 'net barter terms of trade' in their yearly variations become more favorable with an increase in capital exports, and vice versa" (p. 124). His explanation for this result was that "an increase in British demand for foreign securities was apparently offset by an increased foreign demand for English goods" (p. 124). 34. In Canada, the primary recipient of the new capital, the opposite took place after 190{}-an improvement in the terms of trade and rising money wages. See the discussion on Viner, pp. 6~61. 35. According to the theory, if the proceeds of the loan were spent in the lending country then the price-specie-flow mechanism would not operate (Taussig [1927] 1966, p. 230). 36. "Still another equalizing factor is the movement of securities that have an international market. They are sold between the great financial centers in a way that replaces or lessens the transmission of gold.... In any given financial center, a tight money market and a high discount rate tend to lower the prices of . . . international securities among them.... An inflow of gold, which might be expected to take place toward the country of tight money, is replaced by an outward movement of securities" (Taussig [1927] 1966, pp. 218-19). 37. In a description of the monetary system of Great Britain, the United States, Canada, and France, Taussig demonstrated how different the response mechanism to gold flows can be, ranging from the sluggish response of the French system with its high specie-money ratio to the rapid response of the British monetary system with its low gold-reserve ratio and

102

Michael D. Bordo

loaned-up banking system. The Canadian system, with gold reserves held abroad in New York and London, gave the impression that deposits and notes increased before the gold inflow (Taussig [1927] 1966, pp. 201-7). Thus "in all countries using deposits and checks freely, the looseness of the connection between bank reserves and bank deposits leads ... to a chronological order different from that assumed in the Ricardian reasoning. An inflow of specie may follow, not precede, an enlargement of the circulating medium and a rise in prices. So it may be, at least, for a short time, even for a period of many months. Indeed, if there be further forces at work than those merely monetary, it may remain so for years" (pp. 207-8). 38. In his analysis of the massive capital inflows to Canada, Taussig stated, "If the world level of prices had remained unchanged, we should have expected in Canada a fall in the prices of imported goods, and a rise in the prices of domestic goods. Exported goods in the long run would have shown a movement similar to that of the domestic, but with a lag which would for some time keep their prices either on the same low level as the imported, or in a position intermediate between that of the imported and the domestic articles" ([1927] 1966, p. 228). 39. Beach 1935, p. 180. In an appendix, he presented evidence for the United States, supportive of this explanation, that agrees with an earlier study by A. P. Andrew (1907). 40. Viner's contribution is summarized in Taussig [1927] 1966, chap. 19. For the literature critical of Viner, see Meier 1953 and Dick 1981. 41. Until 1935, although the largest commercial bank, the Bank of Montreal, performed many of the functions of a central bank. See Rich 1978. 42. Viner disputed Laughlin's view that all price levels are tied together via arbitrage, citing large differences in the price of gold among countries (Viner 1924, p. 206). He distinguished between traded goods, whose prices are closely linked internationally, and domestic goods, whose prices are only affected indirectly. 43. However, prices of most Canadian exports were determined internationally, hence the rise in the price of international goods produced in Canada by the increased price of domestic goods and services would cause a decrease in exports. 44. Also an index of the price of services rose relative to the overall price index, as did an index of money wages relative to those in the United States and Great Britain, confirming the relative price adjustment mechanism (Viner 1924, pp. 241, 248). 45. Much of the reduced exports came from the diversion of raw materials to domestic use, while a large share of the increased imports consisted of capital goods, largely from the United States-both forces conducive to economic development. The fact that most of the proceeds of the loan were not spent in the lending country, Great Britain, is given as further verification of the classical mechanism, which otherwise would not come into play. See Taussig [1927] 1966, pp. 230 and 259, where he states, "It was to be expected that Canada, getting a growing excess of imports over exports in terms of money, should also get more imported commodities in proportion to her commodities exported. But for the verification of theory it is particularly significant that the net barter terms also become more favorable. The Canadians not only got more of physical goods in proportion to the goods they sent out, but they got, on better terms, those imported goods which may be regarded as coming in payment for their own exported goods, and which had no relation to the borrowings." 46. White 1933, p. 303. Nevertheless, he concluded that "the influence of sectional price changes as a force in the adjustment does not in the case of France appear to have played so prominent a role as is presupposed by the neo-classical doctrine .... Shifts in demand schedules were doubtless a more effective medium. No substantiation of this view could be found in the French trade statistics, but actual substantiation would in any case be impossible because fluctuations in prices as a causal factor in merchandise movements could not be excluded; it would be impossible to determine what proportion of the changes in the volume of merchandise imports and exports was due to changes in demand schedules and what proportion to changes in sectional prices."

103

The Gold Standard: The Traditional Approach

47. See Kindahl 1961 and Friedman and Schwartz 1963 for further discussion. 48. According to Graham, in the first period, exporters to the United States will obtain for a time practically the same paper prices as before the lending, and these translated into gold will be higher than before. At the same time import prices in Great Britain will rise because the gold obtained by American sellers for their products when translated into paper yields less than before the depreciation of gold. Unless the British buyers can import from some other country, the U. S. sellers will gradually be able to raise their prices. The ultimate effect will be a rise in the British price level which will appear as a relative increase in the price of domestic goods. The opposite forces will occur in the second period (Graham 1922, pp. 259-60). 49. According to Nurkse ([1944] 1978), the general picture of the adjustment process and the role of central-bank policy presented in the Cunliffe report "was one which during much of that [interwar] period dominated men's ideas both as to the actual working of the gold standard before 1914 and as to the way the gold standard should be made to work after its restoration" (p. 67). Thus the prewar gold standard was portrayed in similar terms in both the Interim Report of the Gold Delegation of the Financial Committee (League of Nations [1931] 1979) and the Macmillan report (1931). 50. The report ignored the role of fluctuations in the exchange rate within the gold points and the temporary sterilization of gold flows. 51. The rules were never formally spelled out. According to the Macmillan report (1931), "the management of an international standard is an art and not a science, and no one would suggest that it is possible to draw up a formal code of action admitting of no exceptions and qualifications, and adherence to which is obligatory, on peril of wrecking the whole structure" (par. 47). 52. Moreover Gregory doubted the ability of forward exchange markets to cover exchange risk because he felt that "just when the relative values of currencies are most uncertain and when, therefore, the advantages to be derived from the organization of a forward exchange market would be greatest, the difficulties of organizing it ... are greatest also" ([1932] 1979, p. 10). 53. This theme also appeared in the prewar literature in Keynes [1913] 1971. See the discussion below. It has played an important role in the postwar explanation of the classical gold standard's success. See, e.g., Triffin 1964, Lindert 1969, D. Williams 1968, and Palyi 1972. 54. Behind this elaborate network of financial flows was the real process of the transfer of capital to developing countries from the developed countries and the real flow of goods through the current account. According to Williams (1947, p. 155), "England's creditor position in the nineteenth century had developed gradually, along with the development of a world economy involving the division of productive effort between the older industrialized areas and the younger agricultural areas and the flow of accumulated savings from the former to the latter. The same circumstances which assigned to England the leading role in capital export made London the international money market and the Bank of England the administrator of the gold standard." 55. Important pre-1914 European examples were Austria-Hungary and Russia. In Asia, India and the Philippines represented the classic successful examples of the operation of the gold-exchange standard. 56. Calculated on the basis of the underlying trend growth rate of real income. See Cassell in the Interim Report . .. of the Financial Committee (League of Nations [1931] 1979). 57. Among the key sources of variation in monetary demand in the pre-World War I era were "the large accumulation of gold in the United States in preparation for the introduction of the gold standard" in 1879 (Cassell 1935, p. 11), and the competition by central banks to strengthen their gold reserves: "the orthodox use of gold reserves for ironing out temporary deficits in the balance of trade fell into the background and often lost all

104

Michael D. Bordo

importance in comparison with the gold movements caused by competition for gold and ultimately traceable to the artificial position given to gold in the world's monetary system" (p. 13). 58. Viner (1937) described how the conflict between private and public interests of the Bank was resolved in the period 1844 to 1870. Also see the discussion above on Bagehot (p. 46). 59. Not however without courting disaster on numerous occasions (Viner 1932, p. 16); also, Viner 1937, pp. 259-74. 60. Viner's (1937) chronicle of the history of nineteenth-century debates over the monetary standard centered on the role of discretionary management under a gold standard rule. "Although most present day writers seem to believe either that the non-automatic character of the modern gold standard is a discovery of the postwar period or that it was only in the postwar period that the gold standard lost its automatic character, currency controversy during the entire nineteenth century concerned itself largely with the problems resulting from the discretionary or management elements in the prevailing currency systems. The bullion controversy ... turned largely on the difference in the mode of operation in the international mechanism of a managed paper standard currency, on the one hand, and of a convertible paper currency, on the other, with the latter treated generally, but not universally, as if it were automatic. Later, the adherents of both the currency and the banking schools distinguished carefully between the way in which a supposedly automatic 'purely metallic' currency (which, in addition to specie, included bank deposits but not bank notes) would operate and the way in which the Bank of England was actually operating a 'mixed' currency (which, in addition to specie and bank deposits, included bank notes)" (pp. 388-89). 61. "This means that relative prices can be knocked about by the most fleeting influence of politics and of sentiment, and by the periodic pressure of seasonal trades. But it also means that the postwar method [of flexible exchange rates] is a most rapid and powerful corrective of real disequilibria in the balance of international payments arising from whatever causes, and a wonderful preventive in the way of countries which are inclined to spend abroad beyond their resources" (Keynes [1923] 1971, p. 130). 62. The analogy holds perfectly if we assume that each central bank has a rigid goldreserve ratio so that the aggregate quantity of central-bank money is determined by the aggregate gold reserves of the central banks; that no gold is used as currency; and hence that variations in the world monetary gold stock are determined by the difference between the amount of new gold mined and the amount consumed in the arts (Keynes [1930] 1971, p. 250). 63. This reflects two factors: a central bank cannot turn to a "lender of last resort" if its reserves are deficient; a central bank does not maximize profits and thus may keep higher reserves than otherwise (Keynes [1930] 1971, p. 252). 64. In the British banking system there is little interest-rate competition (Keynes [1930] 1971, p. 254). 65. J. H. Williams, "Monetary Stability and the Gold Standard" (1932) in Williams 1947 covered much the same ground. 66. Keynes argued that this dilemma did not present itself to Great Britain before 1914 because "the influence of London on credit conditions throughout the world was so predominant that the Bank of England could almost have claimed to be the conductor of the international orchestra. By modifying the terms on which she was prepared to lend, aided by her own readiness to vary the volume of her gold reserves and the unreadiness of other central banks to vary the volume of theirs, she could to a large extent determine the credit conditions prevailing elsewhere" (Keynes [1930] 1971, p. 274). However since World War I, the decline of Great Britain's influence on world credit conditions meant that she now faced the dilemma.

105

The Gold Standard: The Traditional Approach

67. See Sayers 1936 and p. 95 for a discussion of how this was frequently done on a de facto basis by the Bank of England in the 1890-1914 period. 68. As in the Tract, Keynes considered the case for a managed money standard with flexible rates but ultimately rejected it because the uncertainty associated with exchangerate fluctuations would impede long-term capital mobility. "If we ... desire that there should be a high degree of mobility for international lending, both for long and for short periods, then this is, admittedly, a strong argument for a fixed rate of exchange and a rigid international standard" ([1930] 1971, p. 299). Also, see Williams (1947), who recommended maintenance of the gold standard as a restraint, setting the "limits to which monetary variation can be carried" but widening the role of discretionary monetary policy (pp. 187-88). Ultimately "the logical end of the evolution of credit management, and the only real hope of solution of the conflict between external and internal stability, would be closer cooperation of central banks looking toward some form or degree of supernational management" (p. 190). 69. Whale cited evidence of rapid price adjustment between Lancashire and the rest of England, and between England and Scotland. 70. Ford argued that accounting for real-income changes would also explain Taussig's puzzle, "that periods of active lending have been characterized by rising prices rather than falling prices and that the export of goods apparently has taken place, not in conjunction with a cheapening of goods in the lending country, but in spite of the fact that its goods have seemed dearer at times of great capital export" (Taussig [1927] 1966, p. 219). 71. Ford does not entirely dismiss the role of price changes, but doubts that the elasticities are high enough or prices flexible enough to carry the full burden (1962, p. 12; 1977, p. 17). 72. Ford estimated the marginal propensity to import at about 0.3 and the marginal propensity to save at 0.1 to 0.2. This produces an open-economy multiplier of2 to 2.5 which is not sufficient to equilibrate the balance of payments (1962, p. 54). 73. Following Keynes ([1923] 1971) and Bloomfield (1959), Ford argued that the rules of the game were followed only by creditor countries, and not even by all of them (1962, p. 16). Britain's playing by the rules was facilitated by the location in London of the world's principal capital and gold markets (pp. 11-12). 74. This evidence is contrasted to Ford's finding, unfavorable to the classical relativeprice mechanism, that "the cyclical behavior of the net barter terms of trade shows no such consistent pattern" (1962, p. 76). 75. Favorable circumstances for Britain other than the institutional environment stressed by Ford were: confidence in the convertibility of sterling, the alternating pattern of trends in home and foreign investment, and the sensitivity of British exports to British overseas lending (1962, p. 190). 76. "The price level in the U.S. relative to that in Britain rose from 89.1 to 91.1" (Friedman and Schwartz 1963, p. 98). 77. However one conspicuous example, discussed by Friedman and Schwartz, when the classical approach could not fully explain the adjustment mechanism, was the period 1896-1901. 78. A study by Macesich (1960), using a similar approach, demonstrated that the monetary instability in the period 1834-45 was not caused primarily by the Bank war and Jacksonian policy, as traditionally believed, but rather was produced by external events. Given that the United States was part of the international specie standard, the author argued, internal prices had to adjust to external prices, and how they did so did not matter. Macesich isolated the different determinants of monetary change and found that changes in the ratio of the public's holdings of deposits-plus-notes to specie and the ratio of the bank's liabilities to specie explained most of the change in the money supply reflecting uncertainty engendered by the Bank war. The approach of Temin (1969), based on different data

106

Michael D. Bordo

sources, was similar to Macesich's, but attached greater importance to changes in highpowered money, and less to the ratios, in explaining monetary movements. 79. Following Keynes ([1923] 1971, [1930] 1971) and Cassel (1935), -Triffin doubted the strength of the long-term equilibrating forces of the commodity theory of money and termed the gold discoveries of the nineteenth century favorable accidents temporarily reversing a tendency to secular deflation. Instead, "the reconciliation of high rates of economic growth with exchange-rate and gold-price stability was made possible by the rapid growth and proper management of bank money, and could hardly have been achieved under the purely, or predominantly, metallic systems of money creation characteristic of the previous centuries" (Triffin 1964, p. 15). Triffin (1960) foresaw that a gold-exchange standard based on key currencies, such as that which dominated the interwar and the post-World War II periods, would ultimately fail because of a growing threat to the convertibility of the key currencies as their use as international reserves increased. 80. See similar approaches by Abramovitz (1973) and Thomas (1973). 81. In the postbellum period it was the foreign demand for wheat that was the key source of the long swing. Williamson's interpretation of the 1830s differs markedly from that of Macesich and Temin (n. 78 above) who each stressed the role of external monetary forces as the key disturbing factor. 82. However, the evidence to date on commodity arbitrage, based on more recent evidence, is far from conclusive, with the majority of studies casting doubt on its effectiveness for other than internationally traded commodities (see Kravis and Lipsey 1978). 83. "When two (or more) countries are on the gold standard then there exist definite limits for the absolute differences between their short-term interest rates. The actual differences at a given moment depend on the absolute stand of the exchange rates at the same moment, which in turn can vary only between the gold points of the currencies.... When the interest rates of two countries conform, then we say that their money markets are in a state of solidarity; when the differentials do not conform with the respective absolute positions of the exchange rates, i.e., when they exceed the respective "permissible limits," then we say that they violate that solidarity ... the principles of the gold standard" (Morgenstern 1959, pp. 166-68). 84. Morgenstern explicitly rejected the available data on gold movements because his study (1955) found them to be unreliable. In that study, using official data of imports and exports of gold coin and bullion for the United States, Great Britain, Germany, France, and Canada and a sample of four years, 1900, 1907, 1928, and 1935, official gold exports from one country did not square with gold imports for another. 85. "Without examining the future exchange prices, the author cannot make a case that the maximum permissible interest rate differentials were in fact violated. For the market will respond to the best opportunity. If New York interest went to a 1% premium over London, Morgenstern would say the market was not operating perfectly. Yet with a premium on spot exchange, it would be perfectly consistent with the operations of a competitive market. The question of interest differentials which exceed the maximum exchange risk then involves the interest parities of forward exchange rates. Morgenstern did not examine this at all" (Borts 1964, p. 227). 86. However, a similar test for Canada revealed no correlation (Bloomfield 1963, p. 65). 87. For Canada and Sweden, merchandise exports in real terms also showed long swings that tended to lead those in other variables. 88. Scammel (1965) echoed this view by stating "it is, in the writer's view, arguable that the gold standard was in fact quasi-organizational, being operated by a team of central bankers cooperating under the leadership of the Bank of England on behalf of the world business community" (p. 34). 89. Foreign-exchange reserves accounted for 19 percent of total world reserves in 1913. Japan, Russia, and India held the largest fraction of their reserves in sterling.

107

The Gold Standard: The Traditional Approach

90. Of four effects of a rise in discount rate to correct the balance of payments, Lindert ruled out the effect on aggregate demand as involving too lengthy a lag to account "for the remarkable smoothness and rapidity with which exchange rates, international gold flows, and gold reserves of central banks seem to have been altered." Moreover he found the lag between changes in Bank rate and import prices also to be too long to be relevant, and the evidence on the effect on new issues of long-term foreign securities to be unclear. Hence only the effect on short-term funds operated as the key channel of influence of Bank rate (1969, pp. 43-47). 91. As a test of the relative "pulling" power of the different currencies, Lindert (1969, p. 50) regressed each exchange rate on the other two exchange rates, on its own discount rate, and the second center's rate. He found a hierarchy of dominance running from London to Paris to Berlin. 92. Quotation from the Macmillan report (1931, p. 125) in Lindert 1969, p. 49. The fact that London and the other centers could maintain "deficits without tears" in the pre-World War I period ultimately led to a weakening of the balance-of-payments adjustment mechanism because the longer the process continued, the more difficult it became "to undertake the contractionary measures that would have been required to restore payments 'equilibrium' " (Lindert 1969, p. 79). Also see Triffin 1960. 93. Additional methods used were the outright sale or purchase of securities, selling consols spot and buying for the account, borrowing in the market, borrowing from clearing banks, borrowing from special depositors, and moral suasion. After 1878, the Bank lent to its own customers at the market rate of interest, while at the same time it charged discount houses a penalty rate above the market rate. "The position of the penal rate was ordinarily a matter of daily concern and therefore influential over the market rate itself" (Sayers 1951, p. 115). 94. Goodhart's analysis follows closely that of Whale (1937). Both Goodhart's and Whale's results can be reinterpreted as consistent with the monetary approach to the balance of payments. According to that approach, a rise in economic activity in an open economy such as Great Britain would generate an excess demand for money that would be satisfied in part by a gold inflow. Indeed Mills and Wood (1978), applying the Granger-Sims causality test to the pre-World War I U.K. money supply and national-income data, found that income caused money, evidence that they considered sympathetic to the monetary approach. 95. Also see A. G. Ford (1977, p. 42), who cites similar evidence on investment activity.

References Abramovitz, Moses. 1973. The monetary side of long swings in U.S. economic growth. Memorandum no. 146, Stanford University Center for Research on Economic Growth. Mimeo. Andrew, A. P. 1907. The Treasury and the banks under Secretary Shaw. Quarterly Journal of Economics 21 (Aug.): 519-68. Angell, J. W. [1925] 1965. The theory of international prices. Reprint. New York: Augustus M. Kelley. Bagehot, w. [1873] 1969. Lombard Street. Reprint of the 1915 edition. New York: Arno Press.

108

Michael D. Bordo

Beach, W. 1935. British international gold movements and banking policy, 1881-1913. Cambridge: Harvard University Press. Bloomfield, Arthur I. 1959. Monetary policy under the international gold standard. New York: Federal Reserve Bank of New York. - - - . 1963. Short-term capital movements under the pre-1914 gold standard. Princeton Studies in International Finance, no. 11. Princeton: Princeton University Press. - - - . 1968. Patterns of fluctuation in international investment before 1914. Princeton Studies in International Finance, no. 21. Princeton: Princeton University Press. Bordo, M. D. 1975. John E. Cairnes on the effects of the Australian gold discoveries, 1851-73: An early application of the methodology of positive economics. History of Political Economy 7 (no. 3): 337-59. - - - . 1983. Some aspects of the monetary economics of Richard Cantillon. Journal of Monetary Economics 12 (Aug.): 235-58. Borts, G. H. 1964. Review of International financial transactions and business cycles, by O. Morgenstern. Journal ofthe American Statistical Association 59 (Mar.): 223-28. Brown, William A. 1940. The international gold standard reinterpreted, 1914-1934. New York: National Bureau of Economic Research. Bullion report. 1810. See Report . .. on the high price of bullion [1810] 1978. Cairncross, Alec K. 1953. Home and foreign investment, 1870-1913. Cambridge: Cambridge University Press. Cairnes, J. E. [1873] 1965. Essays in political economy: Theoretical and applied. Reprint. New York: Augustus M. Kelley. Cantillon, R. [1931] 1964. Essai sur la nature du commerce en general. Ed. H. Higgs. Reprint. New York: Augustus M. Kelley. (First published 1755.) Cassel, Gustav. 1935. The downfall of the gold standard. Oxford: Clarendon Press. Chevalier, M. 1859. On the probable fall in the value of gold! The commercial and social consequences which may ensue, and the measures which it invites. Translated from the French by Richard Cobden. 3d ed. Manchester: A. Ireland. Cunliffe report. [1918] 1979. See United Kingdom, Parliament [1918] 1979. Dick, Trevor. 1981. Canadian balance of payments, 1896-1913: Mechanisms of adjustment. Mimeo. Fetter, Frank W. 1965. The development of British monetary orthodoxy, 1717-1875. Cambridge: Harvard University Press. - - - . 1953. The Bullion Report re-examined. In Papers in English monetary history, ed. T. S. Ashton and R. S. Sayers. Oxford: Clarendon Press.

109

The Gold Standard: The Traditional Approach

Fisher, I. 1920. Stabilizing the dollar. New York: Macmillan. - - . [1922] 1965. The purchasing power of money. Reprint. New York: Augustus M. Kelley. - - - . 1935. Are booms and depressions transmitted internationally through monetary standards? Bulletin of the International Statistical Institute 28 (no. 1): 1-29. Ford, A. G. 1962. The gold standard, 1880-1914: Britain and Argentina. Oxford: Clarendon Press. - - - . 1977. International financial policy and the gold standard, 18701914. Mimeo. Frenkel, J. 1971. A theory of money, trade, and the balance of payments in a model of accumulation. Journal ofInternational Economics (May): 159-87. Friedman, M., and A. J. Schwartz. 1963. A monetary history of the United States, 1867-1960. Princeton: Princeton University Press. Girton, L., and D. Roper. 1978. J. Laurence Laughlin and the quantity theory of money. Journal of Political Economy 86 (Aug.): 599-625. Goodhart, C. A. E. 1972. The business ofbanking, 1891-1914. London: Weidenfield and Nicolson. Goschen, G. J. [1892] 1978. The theory of the foreign exchanges. Reprint. New York: Arno Press. Graham, F. D. 1922. International trade under depreciated paper: The United States, 1862-79. Quarterly Journal of Economics 36 (Feb.): 220-73. Gregory, T. E. [1932] 1979. The gold standard and its future. Reprint. New York: Arno Press. Hawtrey, R. G. 1935. The gold standard in theory and practice. 5th ed. London: Longmans Green. Hume, D. [1752] 1955. Of the balance of trade. Reprint. In Writings on economics, ed. E. Rotwein. Madison: University of Wisconsin Press. Jevons, W. S. [1884] 1964. Investigations in currency and finance. Reprint. New York: Augustus M. Kelley. Johnson, H. G. 1976. The monetary approach to balance of payments theory. In The monetary approach to the balance of payments, ed. J. Frenkel and H. G. Johnson. Toronto: University of Toronto Press. Jonung, Lars. 1979. Knut Wicksell and Gustav Cassel on secular movements in prices. Journal ofMoney, Credit, and Banking (May): 165-81. Keynes, J. M. [1913] 1971. Indian currency and finance. Vol. 1 of The collected writings ofJohn Maynard Keynes. Reprint. London: Macmillan and New York: Cambridge University Press for the Royal Economic Society. - - . [1923] 1971. A tract on monetary reform. Vol. 4 of The collected writings. See Keynes [1913] 1971.

110

Michael D. Bordo

- - . [1930] 1971. The applied theory of money: A treatise on money. Vol. 6 of The collected writings. See Keynes [1913] 1971. Kindahl, J. K. 1961. Economic factors in specie resumption. Journal of Political Economy 69 (Feb.): 30-48. Kravis, Irving B., and Robert E. Lipsey. 1978. Price behavior in the light of balance of payments theories. Journal of International Economics 8 (May): 193-246. Kuhn, T. 1970. The structure of scientific revolutions. 2d ed. Chicago: University of Chicago Press. Laidler, D. 1981. Adam Smith as a monetary economist. Canadian Journal of Economics 14 (May): 185-200. Laughlin, J. L. 1903. The principles of money. New York: Scribner's. League of Nations. [1931] 1979. Reports of the gold delegation: Interim report of the gold delegation of the financial committee. Reprint. New York: Arno Press. Lindert, Peter H. 1969. Key currencies and gold, 1900-1913. Princeton Studies in International Finance, no. 24. Princeton: Princeton University Press. Macesich, G. 1960. Sources of monetary disturbances in the U.S., 183445. Journal of Economic History 20 (Sept.): 407-34. Macmillan report. 1931. See United Kingdom, Parliament 1931. Marshall, Alfred. 1923. Money, credit, and commerce. London: Macmillan. - - - . 1926. Official papers. London: Macmillan. McCloskey, D. N., and J. R. Zecher. 1976. How the gold standard worked, 1880-1913. In The monetary approach to the balance of payments. See Johnson 1976. Meier, Gerald M. 1953. Economic development and the transfer mechanism: Canada, 1895-1913. Canadian Journal of Economics and Political Science 19 (Feb.): 1-19. Mill, J. S. [1865] 1961. Principles of political economy. Reprint. New York: Augustus M. Kelley. Mills, T. C., and C. E. Wood. 1978. Money-income relationships and the exchange rate regime. Federal Reserve Bank ofSt. Louis Review 60 (Aug.): 22-27. Mints, L. 1945. A history of banking theory in Great Britain and the United States. Chicago: University of Chicago Press. Morgenstern,Oskar. 1955. The validity of international gold movement statistics. Special Papers in International Economics, no. 2. Princeton: Princeton University Press. - - - . 1959. International financial transactions and business cycles. Princeton. Princeton University Press. Mundell, R. 1971. Monetary theory. Pacific Palisades, Calif.: Goodyear.

111

The Gold Standard: The Traditional Approach

Nurkse, R. [1944] 1978. International currency experience. Reprint. New York: Arno Press. (First published by the League of Nations.) O'Leary, P. J., and W. Arthur Lewis. 1955. Secular swings in production and trade, 1870-1914. The Manchester School 23 (May): 118-25. Palyi, Malchior. 1972. The twilight of gold, 1914-1936: Myths and realities. Chicago: Henry Regnery. Pesmazoglu, J. S. 1951. A note on the cyclical fluctuations of British home investment, 1870-1913. Oxford Economic Papers 3 (Feb.): 3961. Presnell, L. S. 1968. Gold reserves, banking reserves, and the Baring crisis of 1890. In Essays in money and banking in honour ofR. s. Sayers, ed. c. R. Whittlesey and J. S. C. Wilson. Oxford: Clarendon Press. Report from the select committee on the high price ofbullion. [1810] 1978. New York: Arno Press. Ricardo, D. [1811] 1951. High price of bullion: A proof of the depreciation of bank notes. In The works and correspondence of David Ricardo, ed. Piero Sraffa, 3. Cambridge: Cambridge University Press. - - - . [1816]. 1951. Proposals for an economical and secure currency; with observations on the profits of the Bank of England as they regard the public and the proprietors of Bank stock. In The works and correspondence of David Ricardo. See Ricardo [1811] 1951. Rich, G. 1978. The cross of gold: Money and the Canadian business cycle, 1867-1913. Mimeo. Samuelson, P. 1971. An exact Hume-Ricardo-Marshall model of international trade. Journal of International Economics 1 (Feb.): 1-11. Sayers, R. S. 1936. Bank of England operations, 1890-1914. London: P. S. King and Son. - - - . 1957. Central banking after Bagehot. Oxford: Clarendon Press. - - - . 1951. The development of central banking after Bagehot. Economic History Review, 2d ser., 4 (no. 1): 109-16. - - - . 1953. Ricardo's views on monetary questions. In Papers in English monetary history. See Fetter 1953. Scammel, W. M. 1965. The working of the gold standard. Yorkshire Bulletin of Economic and Social Research 17 (May): 32-45. Silverman, A. G. 1931. Some international trade factors for Great Britain, 1880-1913. Review of Economics and Statistics 13 (Aug.): 114-24. Smit, J. C. 1934. The pre-war gold standard. Proceedings ofAcademy of Political Science 13 (Apr.): 53-61. Smith, A. [1776] 1976. An inquiry into the nature and causes ofthe wealth of nations. Reprint. Chicago: University of Chicago Press. Taussig, F. W. 1917. International trade under depreciated paper, a contribution to theory. Quarterly Journal of Economics 21 (May): 380-403.

112

Michael D. Bordo

Taussig, F. W. [1927] 1966. International trade. New York: Augustus M. Kelley. Temin, P. 1969. The Jacksonian economy. New York: W. W. Norton. Thomas, B. 1973. Migration and economic growth. 2d edt Cambridge: Cambridge University Press. Thornton, H. [1802] 1978. An inquiry into the nature and effects of the paper credit of Great Britain. Fairfield, N.J.: Augustus M. Kelley. Tinbergen, J. 1950. Business cycles in the United Kingdom, 1870-1914. 2d edt Amsterdam: North-Holland. Triffin, Robert. 1960. Gold and the dollar crisis. New Haven: Yale University Press. - - - . 1964. The evolution ofthe international monetary system: Historical reappraisal and future perspectives. Princeton Studies in International Finance, no. 12. Princeton: Princeton University Press. United Kingdom. Parliament. [1918] 1979. First interim report of the committee on currency and foreign exchanges after the war. Cmnd. 9182. Reprint. New York: Arno Press. - - - . 1931. Report of the committee on finance and industry. (Macmillan report). Cmnd. 3897. London: HSMO. Viner, Jacob. 1924. Canada's balance of international indebtedness, 1900-1913. Cambridge: Harvard University Press. - - - . [1937] 1975. Studies in the theory ofinternational trade. Reprint. New York: Augustus M. Kelley. - - - . 1932. International aspects of the gold standard. In Gold and monetary stabilization, edt Q. Wright. Chicago: University of Chicago Press. Whale, P. Barrett. 1937. The working of the pre-war gold standard. Economica 4 (Feb.): 18-32. White, H. D. 1933. The French international accounts, 1880-1913. Cambridge: Harvard University Press. White, L. H. 1981. Free banking in Britain: Theory, experience, and debate, 1900-1945. Ph.D. diss., University of California at Los Angeles. Wicksell, Knut. [1898] 1965. Interest and prices. Reprint. New York: Augustus M. Kelley. Williams, D. 1968. The evolution of the sterling system. In Essays in money and banking in honour of R. S. Sayers. See Presnell 1968. Williams, John. 1920. Argentine international trade under inconvertible paper money, 1800-1913. Cambridge: Harvard University Press. - - - . 1947. Gold and monetary stabilization. In Postwar monetary plans and other essays. New York: Knopf. Williamson, Jeffrey G. 1964. American growth and the balance of payments, 1820-1913. Chapel Hill: University of North Carolina Press.

113

The Gold Standard: The Traditional Approach

- - - . 1961. International trade and u.s. economic development 1827-1843. Journal of Economic History 21 (Sept.): 372-83. - - - . 1963. Real growth, monetary disturbances, and the transfer process: The U.S., 1879-1900. Southern Economic Journal 29 (Jan.): 167-80.

Comment

C. Knick Harley

Michael Bordo has certainly presented an extensive review of some two centuries of thought on the workings of a currency either consisting of specie or based on notes freely convertible into specie. I have learned a great deal about the evolution of ideas from the 160 pages and 446 footnotes in the version of the paper I have worked with. Certainly I do not have the qualifications, the time, nor the inclination to discus individual economists or even individual schools of thought in comparable detail. Rather I would like to highlight some major themes that run through the literature. In particular, I would like to try to shift the emphasis of the discussion toward the last fifty or a hundred years and stress what Viner called "the international mechanism" partially in its theoretical context but especially within the context of the process of foreign lending in the late nineteenth century. Two issues have always dominated discussion of the gold standard. The first considers the determinants of the value of the standard in terms of other commodities and the stability of that value over time. The second considers the nature of international equilibrium and the relationships among the prices of various commodities in various locations or countries that equilibrium requires. Central to this discussion is the comparison of equilibria under differing underlying conditions and also investigation of dynamics of adjustment when underlying conditions alter. I intend to add nothing to the discussion of the value of the standard, but will state the obvious: The fluctuations of the value of the standard over the last two-thirds of a century make the concerns of Marshall, Fisher, and their contemporaries seem rather trivial. To my mind the most interesting literature relating to the gold standard deals with what can broadly be called, after Viner, "the international mechanism." Unfortunately, I am disappointed by Michael Bordo's treatment of the literature that has grown up over the past half century. In particular I am surprised that so little attention has been paid to Viner's C. Knick Harley is professor of economics at the University of Western Ontario, London, Ontario, Canada.

114

Michael D. Bordo

Studies in the Theory ofInternational Trade (six footnote references, most apparently as an afterthought) which has always seemed to me to have been the most thorough and considered statement of the traditional position as it emerged between the wars. Along the same lines I miss any discussion of the theoretical literature on transfers and reparations (Keynes [1929] 1949; Ohlin [1929] 1949; Samuelson 1952,1954). Finally, more should be said about the literature on the "Atlantic economy" that has appeared since the Second World War. I would, therefore, like to use the few minutes at my disposal to present what I see to have been the issues raised by these strands of literature and thus highlight what I believe to be major issues that should remain in our research agenda. The literature on the "international mechanism" contained two strands to which I have already referred. First, there is the real theory of comparative-static comparison of international equilibria. Second, there is discussion of what we would now call macroeconomic processes involved in moving between equilibria. Now these two issues were often not clearly separated in the literature and Ohlin ([1929] 1949, p. 179, n. 3) was moved to remark (and was echoed by Samuelson), "Professor Taussig seems to me to present two different and incompatible theories: (1) the barter theory of Mill; (2) a theory of the monetary and price mechanism." Ohlin, of course, proceeded to explain the real (barter) theory in a manner that was quickly recognized to be correct. At the same time he left the "theory of the monetary and price mechanism" to "Professor Viner, with whom I am in substantial agreement." Ohlin demonstrated that the older supposition, which underlay much of the work by Taussig and his followers, that the real transfer implied a shift in terms of trade against the paying country, was incorrect. The comparative statics of the transfer problem under a considerable range of conditions is now firmly understood and is presented with his characteristic lucidity by Samuelson (1952, 1954). Even if errors in identifying the conditions of new equilibrium are avoided, issues of the (macroeconomic) adjustment to the new equilibrium in a complex monetary economy remain. First, the transfer mechanism involves a reduction of "total buying power" in the sending country and an increase in the receiving country. In addition new equilibrium will normally require relative price adjustments, although as Ohlin demonstrated, the adjustment in the relative prices of traded goods is smaller and less predictable than earlier writers imagined. Furthermore, in a world where transportation costs result in a considerable portion of national product being untraded, resources wil have to be redirected toward (or away) from traded goods in the sending (receiving) economy, presumably through the mechanism of relative price changes. The final equilibrium will require a lower money stock in the sending country and a higher stock in the receiver. Discussion of the monetary aspects of the

115

The Gold Standard: The Traditional Approach

adjustment mechanism consists of attempts to discover the processes that are involved in these adjustments. Furthermore, most investigators seem to believe that alteration of relative prices is not frictionless and can be expected to affect both employment and the general price level during a period of adjustment. Thus Keynes ([1929] 1949, p. 167) characterized those who see no such adjustment prlblems as "applying the theory of liquids to what is, if not a solid, at least a sticky mass with strong internal resistance." Discussion of the operation of the gold standard in the short run must now proceed among the other battles in the arena of short-term macroeconomic adjustment where heroes of flexibility confront gladiators of rigidity. Historical fact-as well perhaps as historical myth-maintains scholars' interest in the late-nineteenth-century gold standard. The international economy seems to have experienced less short-run instability than either theoretical models or twentieth-century experience would suggest. Britain, in particular, seems to have succeeded in maintaining enviable short-run stability despite recurring real shocks in the form of large periodic fluctuations in foreign investments that reflected the cycles in expansion in areas of recent settlement. For example, in 1903 foreign investment was about 2.5 percent of British GNP, but by 1913 it had grown to some 7 percent. Now the remarkable feature of these cycles to Taussig and his students when they discovered them and to subsequent investigators who believe in "sticky mass" theories of short-term adjustment was both the near absence of fluctuations in unemployment and inflation and the absence of monetary crises despite Britain's fractional reserve banking system. What happened factually is clear enough. First, the balance of payments adjusted rapidly, primarily through fluctuations in exports, so that an increase in foreign lending from about £50 million in 1903 to well over £200 million in 1913 was transferred abroad without significant gold flows and thus without placing any noticeable strain on the Bank of England's gold holdings that were always well under £40 million. Thus loans were transferred without monetary strain. Second, as Cairncross (1953) demonstrated, fluctuations in employment and the domestic price level were largely avoided because domestic investment, particularly building, declined as exports rose in response to foreign investment and then displaced exports as the latter declined when foreign investment declined. Now it is obvious to anyone who examines the evidence that the international system could not have worked in the manner postulated by the simplest version of the gold standard adjustment mechanism. That is to say, the increase in lending could not possibly have been initially transferred in gold and the real transfer then been effected by gradual price adjustments. In the first place there was insufficient gold in Britain, and in the second the adjustment seems too rapid.

116

Michael D. Bordo

The historical literature, it seems to me, is best seen as trying to understand this adjustment. One strand has paid attention to short-run capital flows and the use of British balances as a substitute for gold as the mechanism that allowed fluctuating capital flows in the face of very moderate gold movements. That explanation, however, does not seem to offer any strong mechanism acting in Britain to effect the real transfer, nor does it help to explain how sterling in its role as a reserve currency avoided the problems of sterling since the First World War or of the dollar since 1960. A second major strand in the literature has maintained that the capital movements were not exogenous shocks but part of a wider process of growth in the Atlantic economy as a whole. The working out of that process of growth involved simultaneously changes in the composition of production in the borrowing and lending country necessary to effect the real transfer and the capital flow in financial terms. In this tradition I would obviously include Brinley Thomas's and Jeffrey Williamson's work. I would also draw attention to a very interesting theoretical article by George Borts (1964). Others like A. G. Ford argue that while long-run forces in the expanding international economy aided adjustment, it is probably somewhat misleading to focus exclusively on Britain's stability. Their argument would run that there were inflationary periods alternating with financial crises and deflation, but the international market for short-term funds insulated Britain and forced the instability onto the periphery. I remain agnostic about the mechanism that smoothly allocated resources alternately between building houses in Oldham and Oklahoma over at least a half century. Until we understand that mechanism, we must take care in drawing conclusions about the classical gold standard. Perhaps, as H. J. Habakkuk (1962) has argued, there was not a mechanism but rather chance. It seems more likely to me that there were elements of the systematic relationships suggested in the literature, but these may well have been peculiar to their own historical situation. It certainly seems that the international transfer of capital within the late nineteenth century has been and should remain central to our study and understanding of the workings of the gold standard. More central, I would argue, than Michael Bordo's summary allows. Finally, I cannot concur with Bordo's vision of McCloskey and Zecher's (1976) article as a Copernican revolution in our thinking about the gold standard. Certainly they have made some useful arguments about what sorts of disequilibria are possible. But they have not considered what is to my mind the major issue of the adjustment mechanism in the late-nineteenth-century gold standard that I have just discussed at length. Finally, I have been unable to resist reading a few sentences of Viner (1937, pp. 316--17) to them:

117

The Gold Standard: The Traditional Approach

The classical school and its important followers all held the same views on this point: after allowance for transportation costs, the market prices of identical transportable commodities must everywhere be equal or tend to be equal when expressed in or converted to a common currency. When, therefore, critics of the classical theory have taken it to task on the grounds that it explained the adjustment of international . balances by the influence on the course of trade of divergent market prices in different markets of identical transportable commodities. . . they have misinterpreted the classical doctrine. References Borts, George H. 1964. A theory of long-run international capital movements. Journal 0/ Political Economy 72 (Aug.): 341-59. Habakkuk, H. J. 1962. Fluctuations in house-building in Britain and the United States in the nineteenth century. Journal 0/ Economic History 22 (June): 198-239. Keynes, J. M. [1929] 1949. The German transfer problem. Reprint. In Readings in the theory o/international trade, ed. H. S. Ellis and L. S. Metzler. Philadelphia: Blakiston. (First published in Economic Journal39 [Mar.]: 1-7.) McCloskey, D. N., and j. R. Zecher. 1976. How the gold standard worked, 1880-1913. In The monetary approach to the balance o/payments, ed. J. Frenkel and H. G. Johnson. Toronto: University of Toronto Press. Ohlin, B. [1929] 1949. Transfer difficulties, real and imagined. Reprint. In Readings in the theory o/international trade, ed. H. S. Ellis and L. S. Metzler. Philadelphia: Blakiston. (First published as The reparation problem: A discussion, in Economic Journal 39 [June]: 172-78.) Samuelson, Paul A. 1952. The transfer problem and transport costs: The terms of trade when impediments are absent. Economic Journal 62 (June): 278-304. - - - . 1954. The transfer problem and transport costs II: Analysis of effects of trade impediments. Economic Journal 64 (June): 264--89. Viner, Jacob. [1937] 1975. Studies in the theory 0/ international trade. Reprint. New York: August M. Kelley.

General Discussion ABRAMOVITZ emphasized and endorsed the view that a central mystery concerning the adjustment mechanism under the international gold standard lies in reconciling the relative freedom enjoyed by Great Britain from the long swings in· aggregate economic activity that characterized

118

Michael D. Bordo

the United States with the magnitude of international capital movements in the nineteenth century. FRIEDMAN elaborated upon the issue raised by Abramovitz by referring to the findings of a study that Anna Schwartz and he recently completed on monetary trends in the United States and the United Kingdom. One of their most surprising findings was the independence of phase-cycle-tophase-cycle movements in real income in Great Britain from events in the United States. Friedman and Schwartz had been unable to find any systematic factors explaining real-income fluctuations in Britain. Friedman suggested that this phenomenon was precisely the one cited by Abramovitz. MCCLOSKEY responded to a point raised by Harley. He noted in several of the papers at the conference a definition of equilibrium in the international affairs of nations that is both artificial and a potential source of confusion. According to that definition, a country's international affairs are in equilibrium only when the balance of payments is zero or, in a more extreme version, when the current-account balance is zero. In fact, countries can maintain surpluses or deficits in their balance of payments for periods of decades or longer. This is evident in the recent experience of Japan, for example. Similarly, countries sometimes go for centuries with surpluses or deficits in their trade balances. Thus, a zero trade balance or zero current-account balance may not be a useful definition of equilibrium. ZECHER, commenting on David Hume's methodology, pointed out that though Hume may have believed in the law of one price as always in effect, yet he might still ask what would happen if prices did differ between countries or regions. For exan:tple, he used this method in explaining the law of gravity by starting with a situation where water did not find its own level. KOCHIN noted Bordo's exclusion of a notable authority on the gold standard, Adam Smith. He noted also another issue omitted from Bordo's paper, namely, Hume's recognition that one of the purposes of the gold standard is to possess a treasure against the contingency of war. FRENKEL, commenting on Harley's remarks, was skeptical that one should require a discussion of the gold standard to bring to the forefront the terms of the trade adjustment consequent on the transfer. Frenkel suggested that as far as the gold standard is concerned, the terms-of-trade effect is probably not a central issue, since much of the adjustment mechanism operates through the income-expenditure mechanism a la Ohlin. Frenkel also asked discussants to define precisely what they meant when referring to the law of one price. In his own discussion of the law of one price, Hume had stated merely that any man who traveled in Europe in his day observed that differences in the prices of commodities between

119

The Gold Standard: The Traditional Approach

one kingdom and another are no greater than such differences between provinces in the same kingdom. DORNBUSCH expressed surprise at Frenkel's statement that the transfer issue is not central to the analysis of the gold standard. It is central, he suggested, if we think that relative price changes are part of the gold standard adjustment mechanism. WHITE offered a supplement to Bordo's discussion of the classical period in which Bordo considers Ricardo and Mill but omits the currency-banking school controversy. White suggested that Bordo's account of the Act of 1844 relies too heavily on the arguments of its advocates. It may be misleading to say that the currency school wished to combine the automaticity of the gold standard with the capital savings of a fiduciary currency, since in fact the Act of 1844 was not required in order to accomplish that goal. Simply permitting the existence of an unregulated financial system with fractional reserve banking would have been sufficient to achieve that goal. White argued that the currency school's principal complaint was that the gold standard was insufficiently automatic; they sought to eliminate what they saw as a slippage between exports of gold and reductions of the quantity of money. White went on to argue that the so-called banking school was made up of two distinct sets of opponents of the Act of 1844. Bordo indicates that the opponents of the act favored free competition, but in fact certain members of the banking school, such as Thomas Tooke, were quite hostile to free competition in banking and sympathetic with the Bank of England's monopoly. At the same time, there were other opponents of the Bank Act, best referred to as the free banking school, who advocated free competition in banking. The free banking school had a monetary theory of the trade cycle and blamed the Bank of England for the economy's cyclical instability. Tooke, on the other hand, had a nonmonetary theory and placed little of the blame on the Bank of England. BORDO concurred with Frenkel's view that the transfer problem was not central to the gold standard story. He accepted White's interpretation of the existence of divisions within the banking school.

2

The Success of Purchasing Power Parity: Historical Evidence and Its Implications for Macroeconomics Donald N. McCloskey and J. Richard Zecher

2.1

Two Views of the International Economy and Why They Matter

A model of the economic life of nations that emphasizes the mutual interdependence of the nations is easy to believe in a period of relative tranquility, such as the heyday of the gold standard, 1880-1914, or of the gold-dollar standard, 1945-71. In less tranquil periods, such as the thirty years of war and depression from 1914 to 1945 or the decade just past, one might suppose that the history is less favorable to the model. We propose to show that this supposition is misleading, and that interdependence was strong. The strength of interdependence depends on the strength of purchasing-power parity. And purchasing-power parity is stronger than it looks. Purchasing-power parity has recently been much in the scholarly news. Some of the new interest in an old idea is attributable to the recent turbulence of international finances, giving practical reasons for wanting Donald N. McCloskey is chairman of the department of economics and professor of history at the University of Iowa, Iowa City. J. Richard Zecher is senior vice-president and chief economist of the Chase Manhattan Bank, New York, New York. The authors are grateful to Francisco Comprido, Daniel Vesper, and Benjamin Russo for valuable research assistance. Part of section 2.4 draws on a paper in preparation by McCloskey and John Lewis, "The Anomalous Price Rise of 1933-34." Earlier versions have been presented at the Macroeconomics Seminar at the University of Iowa, the Economic History Seminar at Northwestern University, the Southern Economics Meetings in New Orleans, a seminar at the University of California at Davis, a conference on the gold standard at the University of Southern California, and seminars at Ohio State University and the Chase Manhattan Economics Group. The participants in these various sessions were tough but open-minded in their skepticism, for which thanks especially to Fischer Black, Eric Gustafson, Steven Kohlagen, Peter Lindert, Robert E. Lipsey, Joseph Vinso, Bob Slighton, Bill Dewald, Sykes Wilford, Blu Putnam, and Dayle Nattress.

121

122

Donald N. McCloskey and J. Richard Zecher

to know when exchange rates or prices are in equilibrium. But some interest is a result of autonomous intellectual change by itself. The chief of these changes was the monetary approach to the balance of payments developed in the early 1970s by Robert Mundell and his students, with involvement by Harry Johnson, Ronald McKinnon, Jeffrey Williamson, and various others. The monetary approach can do without the law of one price. 1 One can approach the balance of payments as a monetary phenomenon-that is to say, not primarily the real phenomenon that the elasticities approach believes it to be-without committing oneself to any particular view of the working of international arbitrage. The monetary approach, after all, merely notes that the balance of payments is by definition a balance on monetary account (just as the current balance is a balance on commodity account and the capital account is a balance on an account of future claims), then makes the innocent-sounding suggestion that its explanation might focus on the excess supply of and demand for money. It is not essential to adopt purchasing-power parity to believe that there may be merit in this view. Nonetheless, purchasing-power parity was in fact commonly invoked in the early theorizing in the monetary approach. Those present at the creation in Chicago in the late 1960s and early 1970s felt they were merely appropriating for use in the study of monetary affairs an assumption that was a necessary commonplace in the study of real affairs. It is hard to see how the real theory of international trade could have gotten far without postulating enough rationality on the part of economic actors to arbitrage away for each commodity any price differences outside the gold pointsthe term "gold points" defined to include all the risks and other costs of transportation. And, to go further, if the gold points were as wide for many commodities as is often implied in criticisms of purchasing-power parity, it is hard to see what usefulness there could have been in the standard propositions in the real theory, such as a tendency to factor price equalization or a tendency to satisfy the Heckscher-Ohlin theory of exports or indeed any tendency to equilibrium. The pioneers of the monetary approach felt they were simply bringing to international finance the intellectual habits formed in the study of the real theory, especially the intellectual habit of supposing that people exploit opportunities for profit. It is no accident that the second generation of leaders in bringing rigor to the study of international trade-the Mundells and Johnsons, following on the first generation of Samuelsons and Meadeswere the inventors of the monetary approach and were inveterate users of the assumption of purchasing-power parity.l We believe that the doctrine of purchasing-power parity, brought into theorizing about the monetary approach by the back door, should be the guest of honor. It is a more radical proposition than the one that the

123

The Success of Purchasing-Power Parity

nation's liquidity balance (or, still more generally, the asset balance) probably has much to do with the amount of money it imports or exports. It challenges in fact the whole way of doing macroeconomics. The usual way of doing macroeconomics might be called the Martian approach. A national economy, usually the United States, is taken in its relations with the rest of the world to be similar to Mars. The price level on Mars, obviously, is determined by Martian demand and supply curves (whether for money or for aggregate goods is not important); likewise, the interest rate. An occasional spaceship might land from Earth bearing gold or Federal Reserve notes, thereby driving up the price level on Mars (or increasing speculative balances in the presence of a liquidity trapagain, the rest of one's economic ideology is irrelevant to the point at issue). The arrival of the spaceship might even be occasioned by events on Mars, but only in a very long run, since it is a long way from Earth to Mars. Mars is a closed economy that has to adjust to the money supply or aggregate demand or expectations that Martians have, period. The Martian approach characterizes 90 percent of the articles and books on macroeconomics, written mostly by Americans. The theoreticians among them can always argue that it is unimportant to them whether or not any actual economy matches their models, for they are concerned with higher things. So much the worse for theory, one might say. A floor or two down the ivory tower the more empirically concerned theorists can argue, quite correctly, that their models might well apply to the whole world even if they are inappropriate for one part of the world. It is strange then to include institutions (such as a central bank with a national policy) in the models that have no worldwide equivalents, one might say. Empiricists, living with the computer down in the basement, can and do argue wearily that they are working on a still larger model (with 10,001 sectors) and will perhaps be able to fit the international sector into one of these. The rest of the world is to them merely another, rather small, sector of the American economy, similar in importance, say, to office equipment. They view an appeal to include the international milieu as a tiresome request to further complicate an already complex model by inserting an office-equipment sector. At the most they are willing to consider project LINK, with its plan to cure the maladies of misspecified national models by putting all the models into the same hospital. It is notable where the other 10 percent of the books and articles on macroeconomics originate. They originate from small open economies. Cassel was Swedish. His countryman, Knut Wicksell (1918), had no difficulty believing purchasing-power parity, debating with the American Taussig (1918) on the matter.3 The assumption of thoroughgoing arbitrage between regions and countries pervades the work of Heckscher and Ohlin. Canadians were the heirs to the Swedes in producing dispro-

124

Donald N. McCloskey and J. Richard Zecher

portionate numbers of excellent economists. They too, living with the great bear of the United States as the Swedes lived with the great bear of the German Empire, have found it difficult to think in Martian terms about their homeland, and even about their new American home. Johnson, McKinnon, and Mundell were raised in Canada, as were many of their students. Martian thinkers are accustomed to dismissing such cases with the remark, "precisely: they came from small economies; the United States is large." The remark is irrelevant, reflecting a common notion that recent developments in balance-of-payments theory depend somehow on the assumption that we are dealing with small economies. The psychological disposition to recognize the existence of an international milieu may be smaller in a big country, but the milieu is still there. The United States may be so big, to be sure, that it can significantly alter the world's price level or interest rate (at least so the finance ministers of other countries believe). But the American price level and interest rate are no less the world's on this account. That General Motors is big does not put it in a different market for automobiles than British Leyland or Simca. America's money supply may well in some periods act as the world's high-powered money, with multiple effects (although the usual accounts of monetarism do not talk this way); America's policy in some periods may well affect expectations abroad (although the usual accounts of rational expectations do not talk this way). Adopting such arguments would constitute a radical break with Martianism. And in any case it would entail recognizing that prices and interest rates were world, not national, phenomena, which is quite another way of doing macroeconomics. The other way of doing macroeconomics may be called the Iowa City approach. No one doubts that Iowa City has virtually no control over its price level and its interest rates. In very short order an attempt by Iowa City bankers to raise interest rates on loans to twice the market rate would empty the loan offices of the banks. In rather longer order (a month, say), an attempt by Iowa City grocers to raise prices to twice the market would empty the grocery stores. In still longer order (a couple of years, say), an attempt by house owners to raise rents above the cost and value of housing determined by the substitutability of housing for other goods in production and consumption would empty the city. Likewise, no one would believe there was a useful sense in which Iowa City could have a monetary policy. It could impose tariffs or price controls, to be sure. But the more usual and subtle instruments of monetary policy would be blunt in the hands of the First National Bank. If Iowa City had its own money supply (and under the free-banking legislation before the Civil War it, like many American cities, in fact did), increasing the money supply would have no effect on the Iowa City price

125

The Success of Purchasing-Power Parity

level or interest rate, so long as the exchange rate was fixed. No economist would place any credence in a model of the economy of Iowa City that allowed Iowa City's prices and interest rates to be determined wholly or even largely within Iowa City by the forces of aggregate demand and supply. The choice is one between a world in which purchasing-power parity does and does not work well enough to be a good description. Is the price level of the United States (when the exchange rate is fixed) substantially or importantly free to move independent of the price level of the rest of the world? Is the United States (or the United Kingdom or whatever) more like Mars or is it like Iowa City? The choice between the Martian and the Iowa City approaches is an empirical one. The United States is not literally either Mars or Iowa City. The question is which approach is closer to the truth, or if you prefer, which mix of the two is true. In particular, it is not enough to remark blandly that both approaches apply to some degree and then proceed to use one or the other to buttress some conclusion on policy or history. For much of the period 1880 to the present the major economic powers were on literally or virtually fixed exchange rates, and it is to such a case that the argument applies most easily. But it is not true, as some think, that a regime of flexible exchange rates completely unhinges an economy from the world market. With a correction for the exchange rate, purchasing-power parity might still apply (though one would expect uninsurable exchange risk to make the gold points wider). And if purchasing-power parity does apply, then the central bank can have only a neutral effect on the economy. The bank would be free to push the general price level up or down (and could just as well make the exchange rate the policy instrument as the money supply), but could not alter relative prices, pegged by world markets. Relative prices-for instance price of investment goods relative to consumer goods-are commonly objects of monetary policy. The common objectives are unattainable if purchasingpower parity works well. And, to repeat, if governments bind themselves to a fixed exchange rate, they cannot even have a neutral influence on prices. Another red herring sometimes drawn across the trail should be avoided as well: purchasing-power parity is assured in the very long run by the price-specie-flow mechanism. Therefore, the argument goes, the monetary approach, which assumes that markets operate very quickly among nations, is merely another way of expressing conventional monetarism, which assumes that markets do not operate quickly among nations.4 The argument is misleading. The price-specie-flow mechanism is a disequilibrium model. It is essentially that two economies that for some reason develop a divergence in their purchasing-power parities will

126

Donald N. McCloskey and J. Richard Zecher

generate flows of gold that will realign the parities. By contrast, the monetary approach, subspecies purchasing-power parity, is an equilibrium model. It claims, nonetheless, by virtue of an alleged quickness with which price divergences among countries are arbitraged away, to be relevant to a much shorter run than could reasonably be supposed for the other. That in one respect (namely, purchasing-power parity) the models happen to have the same outcome in the long run should not be allowed to obscure that the two exhibit radically different behavior in most other ways. In particular, monetary policy does work in the price-specie-flow model (at least in some short run and at least if the model does not belie itself by introducing a rapid price-specie flow) but does not work in the purchasing-power-parity model except by way of influences on the world money supply. The monetary approach takes much from monetarism but, in the end, differs importantly from the monetarist aproach to national monetary policy. We should point out that the historical record contains little evidence that the price-specie-flow mechanism actually happened. Economists, accustomed to thinking of the Facts of History, may be surprised; it will not surprise historians, hardened to the ubiquity of the Myths of History. The intellectual status of the mechanism is similar to the kinked demand curve of oligopolists: it does not work empirically and is unreasonable besides (for instance, it would provide opportunities for speculative profit). In an earlier paper (1976, p. 367) we reviewed the empirical anomalies in the price-specie-flow mechanism. For instance, we argued that Milton Friedman and Anna Schwartz misapplied the mechanism to an episode in American history. The United States went back on the gold standard in January 1879 at the pre-Civil War parity. The American price level was too low for the parity, allegedly setting the mechanism in motion. Over the next three years, Friedman and Schwartz argued from annual figures, gold flowed in and the price level rose just as Hume would have had it. They conclude (1963, p. 99) that "it would be hard to find a much neater example in history of the classical gold-standard mechanism in operation." On the contrary, however, we believe it seems much more like an example of purchasing-power parity and the monetary approach than of the Humean mechanism. In the monthly statistics (Friedman and Schwartz confined themselves to annual data), there is no tendency for price rises to follow inflows of gold, as they should in the price-specie-flow mechanism; if anything, there is a slight tendency for price rises to precede inflows of gold, as they would if arbitrage were shortcutting the mechanism and leaving Americans with higher prices directly and a higher demand for gold. Whether or not the episode is a good example of the monetary theory, it is a poor example of the price-specie-flow mechanism .5

127

The Success of Purchasing-Power Parity

The price-specie-flow mechanism, therefore, is not a good way to harmonize closed-economy monetarism with the fact that we live in a world of many economies. What should be clear by now is that if purchasing-power parity is found to be a useful characterization of the world, then closed-economy theorizing and empirical work in macroeconomics should be changed to allow for the direct effects of international price arbitrage. Whether monetarists or Keynesians or rational expectationists, economists should begin thinking and measuring in global terms. 2.2

The Root Definition of Purchasing-Power Parity

"Come, come," the representative Martian will say, "don't waste my time-we know that prices diverge. Purchasing-power parity fails." That prices are not identical everywhere is not an important failure of purchasing-power parity. A minor reply is that prices can be different in level but related in their changes, a distinction made in the usual statistical tests of parity. The main reply is that purchasing-power parity is a consequence of rationality in arbitraging. If all the opportunities for riskless (or insured) arbitrage among countries that are profitable at existing interest rates and other costs of arbitrage have taken place, then the price level of the world may be said to have exhausted its ability to determine the price level of one country. Now it may have exhausted it, yet be trivial. To take the single commodity case for illustration (and only for illustration, it being a major theme below that commodity-by-commodity arguments do not suffice), the gold points might be so wide that even though they are not violated they are not useful as describing a constraint on the economy. Wheat and boomerangs in Rumania and Tasmania in 1682 (or in 1982?) may have offered no opportunities for profit by arbitrage ex ante, yet their prices were surely free to move within wide limits independent of each other. Our hypothesis is that in the modern world among the main trading nations the forces of rationality in arbitrage were powerful enough to fix anyone general price level and interest rate ceteris paribus, in terms of the others. The usual statistical question is whether or not the result is a unit elasticity of one price level with respect to another. We shall see in a moment whether the statistical question is the right one. But the price level in one country could be determined by the world in the sense that it was fixed ceteris paribus by the action of arbitrageurs-even if the elasticity were not unity. For this reason arguments that ratios of purchasingpower parities tend to drift are irrelevant.6 They do drift, just as demand curves drift. To say that the law of demand fails because in an uncontrolled experiment the observed price does not correlate well inversely

128

Donald N. McCloskey and J. Richard Zecher

with the quantities would be an error. So too here. The arbitrageurs could be making the American price level exogenous to, say, American monetary policy even if changing technologies of traded as against nontraded goods, changing compositions of market baskets, and changing errors in the underlying statistics caused the American price level to be poorly correlated with other price levels. It is easy to construct examples that illustrate the point. Imagine that electrical equipment produced in the United States sells for 20 percent more than that produced in Germany, after translating the prices at the exchange rate, because of differences in the energy efficiency of the equipment. Imagine further that deviations as little as plus or minus one percentage point from the 20-percent differential would create opportunities for profitable arbitrage, exploited instantly. Now suppose the equilibrium price ratio rises (that is, the ratio dividing profitable from unprofitable arbitrage) suddenly to 30 percent from 20 percent, because of changes in energy prices (worldwide). Suppose finally that the same plus-or-minus-one-percentage-point band exists at the 30-percent differential as at the old 20-percent differential. The economist stumbling on such data might conclude that parity fails: the drift from a 20-percent to a 30-percent differential would be interpreted by him as indicating the poorness of correlation between prices in one place and in another. Yet in the sense relevant here of ceteris paribus the prices in Germany and the United States are mutually linked (plus or minus one percentage point) just as strongly at 30 percent as at 20 percent. A monetary policy in the United States that had as one intended result a rise in electrical-equipment prices in the United States relative to (exchange-adjusted) German prices would fail. We wish, then, to appropriate for purchasing-power parity the prestige of the postulate of rationality. It surpasses belief that many opportunities to make easy money buying low and selling high persist long enough to be observed in economic data. Yet much of the opposition to purchasingpower parity seems to believe that it is so. When specialists in finance such as Richard Roll (1979) think about international markets, they assume with hardly a comment that all opportunities for arbitrage are exhausted in a matter of weeks. Roll remarks that "in the monetary approach ... prices and exchange rates tend toward equilibrium ... in the very long run, say a year or more" (italics added; p. 135).1 His criticisms of the monetary approach come from a novel direction: instead of criticizing the approach for supposing that the long run is as short as a year, he criticizes it for supposing that the long run is as long as a year. The view from finance is refreshing and highly relevant. We consider purchasing-power parity to be a proposition similar in more ways than one to the efficient-markets hypothesis, to be demonstrated on similar grounds, namely, on the ground of the shared belief of economists in

129

The Success of Purchasing-Power Parity

rationality and on grounds of whatever evidence can be adduced to confront the belief. The belief is that there must be nothing systematic or predictable in the future of prices in one country relative to present prices in another (except, indeed, instantaneous equality), or else there is money to be made in exploiting the prediction. If differences in price levels take years to be eliminated by trade, then trade right now-in one commodity or in the CPI-is profitable. The result, we believe, is a "theorem" linking the pursuit of profit and the exogeneity of the general price level. If opportunities for arbitrage are exploited (allowing fully for the cost of transport and information), then the price level of one country is fixed by the rest of the world, even in the very short run. The argument is not a test by itself, but merely a theorem, a higher-order proposition about the relation between equilibrium and the exogeneity of prices. The theorem is a curious product, to be sure, for it is not merely a logical proposition. It requires the world to be arranged in a certain way to be true. We arrive again at an empirical question. Three points of logic nonetheless may make the fundamental theorem of the Iowa City approach more palatable. First, arbitrage does not need to occur commodity by commodity. All prices in an economy are connected to each other. Prices of bricks in New York and London are held together not only by thp. direct forces of arbitrage in the market for bricks itself, weak as they are, but by the indirect forces of arbitrage in related markets-the markets for brick-making labor, say, or the market for lumber for which bricks are a substitute. That nontraded goods exist is sometimes thought to be a rebuttal to purchasing-power parity. Not so, at least if the nontraded goods are provided in markets sensitive to costs (military bases, for instance, may be an exception). It is also thought to be a rebuttal to note that the law of one price need not hold for such-andsuch commodity if that commodity's market is obstructed. Not so, at least if the commodity in question is related in production and consumption to other goods. Commodity-by-commodity thinking can be misleading. The simplest form of the argument is the Walrasian point that one absolute price (the numeraire) serves to set all other prices in the economy, given resources, technology, and tastes. At the extreme, then, if Mars were connected to Earth by the market in chewing gum alone, the two price levels would nonetheless be fixed in relation to each other. How much tighter, one might suppose, would be the relation between two economies connected by the prices of thousands of goods and services. A second point of logic, seldom recognized, is that arbitrage across space down to the extent of the transport-cost wedge is reinforced, for storable commodities (housing, wheat, automobiles, cement), by arbitrage across time down to the wedge of storage costs. Suppose the price of cement rises in the United States because demand has risen. One might say that cement will not immediately flow from Canada or Spain to the

130

Donald N. McCloskey and J. Richard Zecher

United States, that it takes time to reorganize the direction of transport. But if the cement price is expected eventually to take up its usual relation with foreign prices, then the present relation is constrained-it cannot deviate from the long-run relation by more than the cost of "transporting" cement from the long into the short run. The argument applies even to nontradable goods: if the forces of general equilibrium would eventually bring even housing prices, say, into a rough parity, then the storability of housing will enforce the parity earlier. A third point is that a properly measured price index would be an index of characteristics, not named goods and services. The imperfections of the usual price indexes should not be used as evidence for the failure of purchasing-power parity, whether the imperfection is a sheer error in reporting or, as here, an error of concept. It is commonly argued that goods and services, especially manufactured goods, are not perfect substitutes across countries, that the category "vacuum cleaners" contains Panasonic (Matsushita Electric) model Me-881 and Sears Power-Mate model 20 A 2099 with different characteristics, and therefore that departures from the gold points are rational. The argument is that competition between the two models of vacuum cleaners is not perfect. The response must be, for one thing, that the degree of monopoly in international trade is beside the point. Matsushita Electric might well be discriminating in the prices it sets for vacuum cleaners in Japan and in the United States, yet it would still be true that there was a stable relationship between the Japanese and the American price determined, say, by relative elasticities of demand. For another thing, the characteristics making up the good may well have perfect markets. Vacuuming power, ease of use, reliability of service, and the like are separately measurable, at least in the consumer's mind, and each is perfectly substitutable across brand names. The bundle of characteristics called a Panasonic vacuum cleaner may not be exactly duplicated in any other vacuum cleaner, but the price of vacuuming power may be set on a competitive market. Another way of making the point is to think of the prices of named goods and services as being composed, speaking statistically, of many factors in principal components, shared with many other goods. This is a statistical way of stating the general equilibrium point: what must be arbitraged is a relatively small number of characteristics, not each of millions of named goods and services one by one. The degree of identity in products across countries is no more relevant than is the degree of identity of other measures of the composition of consumption. 2.3

Tests of the Efficacy of Arbitrage

Thus armed against irrelevant doubt, we turn to the empirical questions. One can cast light on the degree of thoroughness of equilibrium (or

131

The Success of Purchasing-Power Parity

"arbitrage") by measuring it directly. The measurement is extremely difficult, essentially because a measurement entails second-guessing people in the business. If the right amount of resources is being used for arbitrage, then prices in one place are not free to move independent of prices elsewhere. To deny purchasing-power parity in this behavioral sense is to deny that the right amount of resources is being used for arbitrage. Often unrecognized by critics of purchasing-power parity is that their conclusion that it has failed usually implies an ability to make money. Anyone who knew that purchasing-power parity was true in some long run but not yet true in the present would have a rosy financial future. The divergences from purchasing-power parity detected in the literature are so gross and the statistics purporting to show the divergences so easy to collect that the opportunities for profit are large. Go thee and prosper. The only direct test of the rationality of arbitrageurs, then, is literal second-guessing. That is, one assembles the facts on prices and transport costs, being careful to allow for such subtleties as the cost-of-exchange risk between the time the arbitrage opportunity arises and the time it is exploited, and does the calculation that the arbitrageuse presumably did, or should have done. She herself would use list prices only after ascertaining that they reflected prices at which she could actually transact, allowing for delivery lags, credit terms, and risk of default. If she missed a profitable opportunity, one can either doubt the completeness of one's calculations or doubt her rationality, depending on the strength of one's devotion to the working hypothesis of rationality. The test is very difficult to perform, though it has the compensating merit of being most relevant to the question at issue. We urge others to attempt it.8 Another, more practical test is to examine whether the shipments of goods implied by the supposed opportunities for arbitrage in fact occurred. Enthusiasts of purchasing-power parity would argue that the mere threat of such flows suffices to bring prices speedily back to unprofitable levels of divergence. Doubters of purchasing-power parity would argue that only the flows themselves suffice, and these only over a considerable period. Taking the doubters' view, then, the flows themselves should be observable. A country that exhibits divergences from purchasing-power parity convincing to the doubters should also exhibit lowered exports and increased imports, a trade deficit: a place with "high" prices would have a hard time selling and an easy time buying. The reasoning is, of course, the first step in the elasticities approach to the balance of payments. In other words, the balance of trade should become more negative for a country that was above its trend of purchasing-power parity. Nothing of the sort shows through in the U.S. statistics. One would expect excessive U.S. inflation to hurt U.S. exports. That is, one would expect a rise in the deviation from purchasing-power parity, expressed as

132

Donald N. McCloskey and J. Richard Zecher

a rise in the difference between the U.S. price index and the U.K. or Canadian price index in U.S. dollars, to cause a fall in the trade balance. The equation for the United Kingdom meets the expectation: the sign on d (Pus - ePUK ) is indeed negative (standard errors in parentheses): Change in U.S. trade] = 5.3 - 2.8[d(Pus - ePUK )]. [ balance with U.K. (19.3) (2.5) The Durbin-Watson is 1.59, in the indeterminant range, but of course much better than the result of running the levels instead of the differences. The slope coefficient, however, is insignificant at conventional levels (notice that the insignificance of the constant indicates that there is no linear trend to worry about). It is unclear what insignificance might mean with observations that are not a sample but the universe of the relevant U.S. and British variables in the period. The R 2 of the equation is a mere .02. It is clear what that means. The equation for Canada, which in view of its contiguity with the United States should prove a better test, is worse: Change in U.S. trade] = 3.5 + 0.61[d(Pus - ePCan )]. [ balance with Canada (6.1) (0.73) The sign is perverse, the R 2 only .01, and again the coefficients are insignificant .9 One could certainly raise the R 2 ,s here and perhaps correct the sign by embarking on a search through all possible specifications of lags and functional forms and periods. For instance, from 1880 to 1912 the United States deflates faster (down to 1896, the usual turning point for price series in the period), then inflates faster than the United Kingdom and Canada. True to expectations, the trade balance with the United Kingdom (though not with Canada) exhibits the same U, but inverted. U.K. clamor about German and U.S. competition becomes great in the 1890s and diminishes in the Edwardian boom of exports and investment abroad that followed. But such evidence would have to contend with a markedly parallel movement of net U.S. inflation and net U.S. exports-the reverse of expectations-relative to both the United Kingdom and Canada during the 1920s and 1930s. In other words, the apparent deviations from purchasing-power parity appear not to represent unexploited opportunities for arbitrage. Another way to say it is that the deviations were only apparent. The "deviations" could be, for instance, the result of peculiarities in the price indexesU.S. price indexes are always more volatile than the U.K. ones in the late nineteenth century, which may well say more about how the indexes were constructed than about the underlying real character of price formation. The rise and fall of net exports to the United Kingdom, then, could be a result of matters wholly internal to the countries, such as building booms

133

The Success of Purchasing-Power Parity

in the face of industrial maturity. Alternatively, the "deviations" could be once-for-all (or for-the-duration) changes in the cost of moving goods from one place to another, as it appears are the violent leaps up and down of parity during the Great War. Changes in tariffs or in other legal costs of movement can permit a change in the relative position of two price levels without in any way belying the assertion that the price levels are not free to move where they will. Such a tariff inflation appears to have taken place in Germany during the 1880s (although we have not as yet been able to find out how powerful the explanation is). While prices fell elsewhere, led by agricultural prices responding to competition from Russia and the United States, prices in Germany held up, as Bismarck forged a tariff politics of rye and iron. Further tests of the hypothesis that profitable opportunities for arbitrage arise when measured prices diverge might proceed commodity by commodity, but the aggregate results at least are unpromising for the hypothesis. 2.4 The Irrelevance of Price Regressions Unadorned The representative Martian will by now be close to apoplexy. "Innumerable regressions of one country's prices on another have been performed recently, and half of them show that purchasing-power parity fails. Surely such tests are conclusive." No, they are not. There are two points here. The first, the more fundamental, is that the regressions are not useful tests of our hypothesis, which has to do with the rationality of arbitrageurs, not with how closely one price correlates with another. Our hypothesis says that prices are linked and therefore insensitive to internal forces such as monetary policy (as we shall show presently), not that the prices are linked by some linear relationship having such-and-such a slope. The second point is that even the hypothesis of linear relationship, which unlike ours is not based on a foundation of individual rationality, has been tested inadequately. The tests take the form of regressing, say, the U.S. GNP deflator on the U.K. GNP deflator multiplied by the exchange rate of dollars for pounds.

Pus = a + J3[(e)(PUK )] + error term. If the slope coefficient J3 is significantly different from 1.0 at conventional levels, then purchasing-power parity is said to fail. A number of criticisms can be and have been leveled at such equations, although none is the main point here. It can be argued, for instance, that allowing for simultaneity bias brings the J3 coefficient closer to 1.0 (Krugman 1978). It can be argued that the equation is misspecified, not allowing properly for lags or for secular trends. It can be argued that "significance" is beside the point for a nonsample. It can be argued that the prices, especially the

134

Donald N. McCloskey and J. Richard Zecher

foreign prices on the right-hand side, are measured in error, as they are for instance if the United Kingdom is taken to stand for the world in the calculation of prices, biasing the coefficient towards zero. It can be argued that the slope coefficient could differ from 1.0 if such a difference were implied by the Walrasian general equilibrium. It can be argued that the whole procedure is misleading, because any coefficient different from zero bespeaks a relationship between domestic and foreign prices dangerous to ignore in explaining domestic prices. It can be argued that to judge the hypothesis of parity a failure, and to abandon it, is to flee to evils that we know not of. The main point, however, is that the "failure" of purchasing-power parity in such an equation is not measured against a standard. How close does the slope have to be to the ideal of 1.00 to say that purchasing-power parity succeeds? The literature is silent. The standard used is the irrelevant one of statistical significance. A sample size of a million yielding a tight estimate that the slope was .9999, significantly different from 1.00000, could be produced as evidence that purchasing-power parity had "failed", at least if the logic of the usual method were to be followed consistently. Common sense, presumably, would rescue the scholar from asserting that an estimate of .9999 with a standard error of .0000001 was significantly different from unity in a significant meaning of significance. Such logic also could be applied to findings of slopes of .90 or 1.20, but usually it is not. lD The point is not that levels of significance are arbitrary. Of course they are. The point is that it is not known whether the range picked out by the level of significance affirms or denies the hypothesis. Nor is the point that econometric tests are to be disdained. Quite the contrary. The point is that the econometric tests have not followed their own rhetoric of hypothesis testing. For one thing, as we have said, the errors that tests of significance deal with are errors of sampling, but in many cases there is no sampling involved: we have the entire universe of observations of the general price level in the United States and the United Kingdom 18801940. For another, nowhere in the literature of tests of purchasing-power parity does there appear a loss function. We do not know how much it will cost in policy wrecked or analysis misapplied or reputation ruined if purchasing-power parity is said to be true when by the measure of the slope coefficient it is only 85-percent true. That is, the argument (due to Neyman and Pearson 1933) that undergirds modern econometrics has been set aside here as elsewhere in favor of a merely statistical standard, and an irrelevant one related to sampling error at that. We are told how improbable it is that a slope coefficient of .90 came from a distribution centered on 1.00 in view of the one kind of error we claim we know about (unbiased sampling error, with finite variance), but we are not told

135

The Success of Purchasing-Power Parity

whether it matters to the truth of purchasing-power parity where such limits of confidence are placed. Silence on the matter is not confined to the literature of purchasingpower parity. Most texts on econometrics do not mention that the goodness or badness of a hypothesis is not ascertainable on merely statistical grounds. Statisticians themselves are more self-conscious, although the transition from principle to practice is sometimes awkward. A practical difficulty in the way of using the Neyman and Pearson theory in pure form, say Mood and Graybill (1963, p. 278), is that the loss function is not known at all or else it is not known accurately enough to warrant its use. If the loss function is not known, it seems that a decision function that in some sense minimizes the error probabilities will be a reasonable procedure. Such a procedure might be reasonable for a general statistician who makes no claim to know what is a good or bad approximation to truth in fields outside statistics itself. The procedure is not reasonable for a specialist in international trade or macroeconomics. If the loss function is not known it should be discovered. Finding the loss function amounts to finding out how close the slope has to be to 1.00 in order for it to be reckoned close enough. Every student of the matter is more or less aware of the need for some standard against which to judge the closeness, a standard beyond the probability of being misled by sampling errors (if there really are any) into gullibility or skeptjcism, but no one has provided them. In a superb paper that has by its sheer weight and subtlety turned the tide of battle recently against purchasing-power parity, for example, Kravis and Lipsey (1978), in reporting some calculations of parity relative to base years and some correlation coefficients (both of which "ought" to be 1.00), remark: Each analyst will have to decide in the light of his own purposes whether the PPP relationships fall close enough to 1.00 to satisfy the theories. As a matter of general judgment, we express our opinion that the results do not support the notion of a tightly integrated international price structure. (Pp. 214ff.) The only guidance they provide to evaluating their "general judgment" is a footnote (p. 214) reporting that in the general judgment of Houthakker, Haberler, and Johnson deviations from parity of anything under 10 or 20 percent are acceptable to the hypothesis. ll It happens, incidentally, that the bulk of the Kravis and Lipsey evidence passes rather than fails such a test. But accepting or rejecting one unargued standard of truth by comparing it with another unargued standard of truth does not advance the art of argument in economics very much.

136

Donald N. McCloskey and J. Richard Zecher

To be fair, Kravis and Lipsey are in fact unusually sensitive to the case for having some standard, more sensitive than are economists working the field with more powerful statistical tools. So frequently does their paper make the point that it must be accounted one of the major ones made. Repeatedly, for instance, they draw a distinction between the statistical and the economic significance of their results: "Indeed, even high coefficients of correlation [between domestic and foreign prices] may conceal shifts in relationships that are economically important" (p. 204); "The difference [of slope coefficients from 1.00] may not be large enough to be picked up by a statistical test yet be economically significant" (p. 236); "even a high correlation does not preclude what may be economically significant variations between the two indexes" (p. 242); and so forth, passim. The intellectual sword is sharp. Remarkably, however, they use the sword only against, not in favor of, purchasing-power parity, and never turn it on themselves. After stating repeatedly that they do 'not have a standard by which to judge the hypothesis, they nonetheless conclude: We think it unlikely that the high degree of national and international commodity abitrage that many versions of the monetarist theory of the balance of payments contemplate is typical of the real world. This is not to deny that the price structures of the advanced industrial countries are linked together, but it is to suggest that the links are loose rather than rigid. (Italics added; p. 243.) Every italicized word involves a comparison against some standard of what constitutes likelihood or highness or typicality or linkage or rigidity, yet no standard is proposed. Indeed, the lone page (p. 204) in their fifty that addresses the issue, entitled "Criterion of Similarity of Price Movements," is devoted to dismissing the lone attempt in the literature to offer a criterion (namely, the Genberg-Zecher criterion: if the correlations of prices among different countries are as good or bad as they are inside one country, then the different countries act as one) and to remarking again that the merely statistical standard used elsewhere in the paper is in fact irrelevant. The irrelevance of the merely statistical standard of fit does not bedevil only that half of the empirical literature that finds purchasing-power parity to be wrong. Towards the end of a fine article favorable to purchasing-power parity, Krugman writes (1978 p. 405): There are several ways in which we might try to evaluate PPP as a theory. We can ask how much it explains [that is, R-square]; we can ask how large the deviations from PPP are in some absolute sense; and we can ask whether the deviations from PPP are in some sense systematic. The defensive usage "in some absolute sense" and "in some sense" betrays his unease, which is fully justified. There is no "absolute sense" in

137

The Success of Purchasing-Power Parity

which a description is good or bad. The sense must be comparative to a standard. Similarly, Jacob Frenkel (1978 p. 175) says "if the market is efficient and if the forward exchange rate is an unbiased forecast of the future spot exchange rate, the constant term [in a regression of the spot rate today on the future rate for today quoted yesterday] ... should not differ significantly from zero, the slope coefficient should not differ significantly from unity" (italics added). In a footnote on the next page, speaking of the standard errors of the estimates for such an equation in the 1920s, Frenkel argues that "while these results indicate that markets were efficient and that on average forward rates were unbiased forecasts of future spot rates, the 2-8 percent errors were significant" (italics added).12 What he appears to mean is that he judges a 2- to 8-percent error to be large in some economic sense. In any event, what his results imply about their subject is unclear-purchasing-power parity, because significance in statistics, however useful it is as an input into economic significance, is not the same thing as economic significance. 2.5

The Search for Standards

Results typical of the conventional tests of purchasing-power parity are easy to replicate. Regressions for 1880 to 1940 (running through the First World War, as a more extreme test of the argument) of the U.S. GNP deflator (to avoid the usual criticism of the domination of wholesale prices by traded goods) against Canadian or U.K. prices adjusted by official exchange rates are (all variables in logarithms and standard errors in parentheses): Price in U.S. = 0.83 (0.19) R 2 = .87,D.W. Price in U.S. = 0.26 (0.14) R 2 = .94,D.W.

+ 0.87 [Exchange-adjUsted] (0.04)

Canadian price

.

= 0.29 + 0.93 [Exchange-adjUsted] (0.03)

U.K. price

.

= 0.46

Additional regressions adjusting for autocorrelation and for the simultaneity of prices and exchange rates (during the floating-rate period) yielded ~'s for the United Kingdom that ranged from .91 to 1.02, and for Canada from .35 to 1.00. Most of these regressions have in common the result that at conventional levels of significance, the slope coefficients would be said to be different from 1.00, and about half of the trends would exhibit "drift." In other words, by the usual standards, these regression results would lead to the conclusion that purchasing-power parity "fails." But the regressions also have something else in common-a uniformly

138

Donald N. McCloskey and J. Richard Zecher

5.25 A. Natural Log of the U.S. Implicit Price Deflator 5.00 4.75 4.50

4.25

1889

1899

1909

1919

1929

1939

1919

1929

1939

5.20 B. Purchasing Power Parity Prediction of the U.S. Implicit Price Deflator Using Canadian Prices 5.00 4.80 4.60 4.40

1889

c.

1909

Standardized Prediction Errors, 1880 to 1940

1889

Fig. 2.1

1899

1899

1909

1920

1919

1929

1939

u. S. prices predicted by Canadian prices.

"high" correlation as measured by R 2. Consider the plots in figures 2.1 and 2.2 of the actual U.S. price, the price estimated from foreign prices, and their difference, for Canada and the United Kingdom. Although the period includes the Great War and the Great Depression, the foreign prices do predict the gross outlines of the U.S. price. Such eyeballing is another way of saying that the RZ's are high, as they are. A fuller treatment would make the sensitivity analysis explicit, introducing the cost of being wrong in adopting a model in which a was zero and f3 was 1.00. A second standard is suggested by the bottom graphs, which plot the forecasting error in units of standard deviation. The period 1880-1914 is a standard for relative tranquility. This is the standard, a relative one over

time. If one is willing to think of 1880-1914 as tranquil (one may not, of course), then it lends meaning to the "success" or "failure" of purchasing-power parity. Comparing 1880-1914 with 1921-40 reveals no difference in the average deviation from purchasing-power parity. The comparison involving the United Kingdom, indeed, can even include the Great War and its immediate aftermath with no change in result. The turbulence of the 1920s and 1930s, which is said to have loosened the economic ties among nations, appears not to have done so. The price relations are equally close, which suggests that the economic behavior causing prices to move in parallel was uniform. It is no surprise that turbulence would offer high rewards to arbitrageurs and that arbitrageurs would take them.

140

Donald N. McCloskey and J. Richard Zecher

A third standard (our favorite) is the Genberg-Zecher criterion (see section 2.4 above). Our paper, and papers by Genberg (1976, 1978), expanded on this theme in the early 1970s.13 It will suffice for the present to recall the conclusion of our work, namely, that the matrix of the United States, the United Kingdom, Germany, and Sweden (with special attention to the United Kingdom) had correlations of prices in the 1880-1914 period similar to those between different parts of the United States. Genberg performed similar tests with similar results for recent times. The few attempts since to undermine the conclusion leave us unmoved. Arthur Gandolfi and James Lothian (1982) have written an interesting paper on the subject, but demand that the hypothesis pass a test of a lower correlation between very close states of the United States than between far nations of the world. A common reaction to the standard is, "Well, suppose we do not even accept the premise that the United States is a unified market?" If one does not accept such a premise, then of course one will refrain from speaking of "the" U.S. price level. Nor will one talk of "the" U.S. GNP. If the United States itself is a mere collection of wholly local markets, there is no more use for the talk than for talk about macroeconomic variables for a federation of Fiji-Botswana-Iceland-Saskatchewan, that is, for a random assortment of places. Fourth and finally, purchasing-power parity can be tested against the standard of explaining major events or illuminating puzzles in history. We have mentioned the event of the United States returning to gold in 1879 at the wrong parity. Consider another episode, puzzling to all who have noticed it, the inflation of 1933-34. It was a spectacular outlier of the Phillips curve, an example of stagflation forty years before that hideous neologism was coined. Unemployment was very high, yet prices rose, whatever notion of "the" price level one uses: a 14 percent rise in wholesale prices would have caused alarm even in 1981; a 3 percent rise in consumer (retail) prices would have caused alarm at least in 1960 (table 2.1). Why did it happen? While recognizing the indirect force of the world price level, Friedman and Schwartz (1963, pp. 498-99), argued that: Another [factor] was almost surely the explicit measures to raise prices and wages undertaken with government encouragement and assistance, notably, NIRA [the National Industrial Recovery Act, leading to the National Recovery Administration, or NRA], the Guffey Coal Act, the agricultural price-support program, and the National Labor Relations Act. ... We have grave doubts that autonomous changes in wages and prices played an important role [after World War II. But] there seems to us a much stronger case for a wage-price or price-wage spiral interpretation of 1933-37.14

141

The Success of Purchasing-Power Parity

Table 2.1

U.8. Price Indexes 1932-36 (1933 = 100)

Definition of Prices 1. GNP deflator, 1929 weights 2. GNP deflator, 1954 weights 3. Personal consumption, 1954 weights 4. Consumer price index 5. Retail food 6. Wholesale prices, 1926 base 7. Average annual earnings per full-time employee, all industries

1931

1932

1933

1934

1935

1936

113

101

100

106

105

109

113

102

100

106

107

108

117

104

100

106

108

109

118 124

106 103

100 100

103 112

106 119

107 120

100

114

121

123

100

104

108

113

111

122

98.3

107

Source: U.S. Bureau of the Census 1960. Row 1, series F 1/2; row2, F67, 87; row 3, F68, 88; row 4, E 113; row 5, E 114; row 6, E 13; row 7, D 696.

This is an unusual line of reasoning for such crusaders against mixing up the determination of relative and absolute prices. It also seems to us to square poorly with the evidence. The chief factual difficulties with the notion that the official cartels sanctioned by the NRA codes caused a rise in the general price level is that most of the NRA codes were not enacted until after the price rise. Ante hoc ergo non propter hoc. Look at the plot of wholesale prices of 1933 in figure 2.3 (retail prices, including such nontradables as housing, show a similar pattern). Most of the rise occurs in May, June, and July of 1933, but the NIRA was not even passed until June. A law passed, furthermore, is not a law enforced. However eager most businessmen must have been to cooperate with a government intent on forming monopolies, the formation took time. As the Bureau of Labor Statistics described it: The monthly load of code approvals reached its peak in the period from October 1933 to March 1934; thereafter there was a rapid decrease. Many of the large employing industries were codified in the latter part of 1933 [mentioning cotton textiles, petroleum, bituminous coal, retail trade, fabricated metal products, retail food]. The National Recovery Administration estimated in a report issued in February 1934 that codification of American industry under the industrial self-government program contemplated by the act was 90 percent completed.I5

Fig. 2.3

Exchange Rate

Jan

I

Feb

£

g>

~

j

4-

:-...'

~/~

~

Scale)

April

~

f?

a:

c(

~

j

g

. -."... .....1 ~

~ March

c:

.:.t.

(5 J:

:2

~

~

.!Cl

~ ~c:

~

~

j

51o

(3

....

May

' '

June

J,...... ~ •J......

~g

::J -,..JU

~~~ OO

oc&i

~~ ~

a:~ o:!2

§15

~~

j~

c:-o

~~

o &.~

"

~/

July

,

~

a::

.~

"0

-64

Dec ~ - 6~ '0

~ -66

~ -67

&

8- 68

t-!e9 ~

-70

~ u..

~

-60

- 6:

~-

62

Nov

1934

Range _ 72

..

_----..{

,,--~

~ - 63

Oct

"

,.. ]

Wholesale Price Index, BLS Handbook of Labor Statistics 1936 Edition, pp, 684-85. Averages for Weeks Ending Friday (Same Friday as Exchange Rates).

sept

l

"

Friday Spot Exchange Rates in London from Paul Einzig, The Theory of Forward Exchange (1937).

Sources:

Aug

\ .. ~I••_ ......' "

f'....·........

~I ::J'

T"""

l

Close weekly relationship between the exchange rate and the wholesale index, U.S.A., 1933. Source: Friday spot exchange rates in London, from Einzig 1937; wholesale price index, BLS Handbook of Labor Statistics, 1936 Edition, pp. 684-85, averages for weeks ending Friday (same Friday as exchange rates).

$3.00 -

...............J......-.......

$3.50-

$4.00-

$4.50 -

$5.00-

($ pert)

143

The Success of Purchasing-Power Parity

By September 1933, apparently before the approval of most NRA codes-and, judging from the late coming of compulsion, before the effective approval of agricultural codes-three-quarters of the total rise in wholesale prices and more of the total rise in retail food prices from March 1933 to the average of 1934 was complete. On the face of it, at least, the NRA is a poor candidate for a cause of the price rise. It came too late. What came in time was the depreciation of the dollar, a conscious policy of the Roosevelt administration from the beginning, although not usually believed to have taken effect until the fall of 1933. There was certainly no contemporaneous price rise abroad to explain the 28-percent rise in American wholesale prices (and in retail food prices) between April 1933 and the high point in September 1934. In fact, in twenty-five countries the average rise was only 2.2 percent, with the American rise far and away the largest. The close link between exchange rates and U.S. wholesale prices is clear in the weekly series graphed in figure 2.3 above. Note especially the two sharp jumps around 20 April and 2 July in response to explicit announcements by Roosevelt of the intent of his administration to devalue the dollar. Wholesale prices move up simultaneously. It would appear, in short, that the economic history of 1933 cannot be understood with a model closed to direct arbitrage. The inflation was no gradual working out of price-specie flow; less was it an inflation of aggregate demand. It happened quickly, well before most other New Deal policies (and in particular the NRA) could take effect, and it happened about when and to the extent that the dollar was devalued. By the standard of success in explaining major events, parity here works. 2.6

Purchasing-Power Parity and Monetary Policy

In the style of the doubts expressed above about price-specie flow, the success of parity can be judged by the failure of the alternatives. A common view in much of purchasing-power-parity literature is that while international trade places limits on what exchange-adjusted domestic prices can be, there is nevertheless considerable flexibility for prices to move up and down. It is argued in particular that a country can raise its price level relative to the exchange-adjusted price abroad by expanding the domestic money supply. If the activities of traders and arbitrageurs fix prices within very narrow ranges, however, such policies would not work. Consider, then, the relationship between monetary policy and the "errors" in purchasing-power-parity forecasts. The model is monetarist, postulating that excessively rapid money growth will put upward pressure on prices, leading to domestic prices that are systematically over the purchasing-power-parity predictions. This is a

144

Donald N. McCloskey and J. Richard Zecher

test, then, to what degree a country, the United States in this case, can through its monetary policies affect its price level (Pus) relative to prices (adjusted for exchange rates) in the rest of the world (Pus). In general equilibrium terms, the hypothesis concerns a state of disequilibrium in the goods markets that is matched by an offsetting disequilibrium in the domestic money market. The regression model below represents the goods-markets disequilibrium by the difference between actual U.S. prices and the purchasing-power-parity prediction of U.S. prices using Canadian and U.K. prices and exchange rates. Disequilibrium in the U.S. money market is the difference between growth in money supply a (M S ) and money demand a (M D ). Two measures of money supply are used: (1) M2, which reflects the Martian view that the United States could control its total money supply over the 1880-1940 period, and (2) domestic credit, omitting the effects of specie flows on money supply, which reflects the view that for most of this period the United States could only affect domestic credit, not total money supply. Growth in money demand is represented by the sum of growth in real income and in prices, on the assumptions that money demand is unit elastic with respect to both of these variables: i.e., a M D = a y + a P, where y is real income and P is the price level. Thus the regression equation becomes:

a (Pus -

-"

Pus)

= 0:

+ ~a (M S - M D )t

+ ~a (M S

-

MD)t_l.

The regression results are reported in tables 2.2 and 2.3. Of the thirty-six estimated coefficients relating excess money growth to changes in the purchasing-power-parity forecast error, twenty-two are negative and sixteen are positive. Two of these coefficients are significant at the 5 percent level; both are negative and both are for regressions using M2 as the measure of money. Only two of the twenty-four regressions have R 2 s above .08. There is little support here for the notion that the errors in purchasing-power parity are related to domestic monetary conditions. The alternative to the international determination of prices appears to work poorly. 2.7

Conclusion

The argument and evidence presented here make a pronouncement of the failure of purchasing-power parity impossible, and without a pronouncement of failure much of modern macroeconomics is badly damaged. The failure of purchasing-power parity must be large indeed to leave the Martian models and thei'r empirical implementations unscathed, whether these are Keynesian, monetarist, or rationally expecta-

145

The Success of Purchasing-Power Parity

tionist. Under either the cleanly fixed exchange rates that typify the historical periods used to test the models or under the dirty float that typifies the years in which the conclusions thus tested have been used for policy, it is hard to believe that foreign prices or interest rates did not matter. Yet the silence of most American macroeconomists on the role of

Table 2.2

Calculated Forecast Errors Implicit Deflator Canadian

U.K.

CPI Canadian

U.K.

WPI Canadian

U.K.

Purchasing-Power-Parity Forecast Errors For the U.S. as a Function of Measures of Excess Money Growth, 1880-1940

Constant

Current Excess Money Growth

0.018 (0.038) 0.019 (0.037)

0.109 (0.166) 0.125 (0.166)

0.087 (0.098) -0.089 (0.098)

0.040 (0.429) -0.080 (0.431)

0.026 (0.060) 0.028 (0.060)

0.077 (0.263) 0.112 (0.262)

-0.117 (0.117) -0.117 (0.118)

0.154 (0.512) -0.167 (0.519)

0.489 (0.096) -0.488 (0.097)

-1.114 (0.421) -1.126 (0.427)

-0.404 (0.139) -0.402 (0.140)

-0.343 (0.610) -0.304 (0.615)

Lagged Excess Money Growth

0.163 (0.166)

0.417 (0.431)

0.360 (0.262)

0.132 (0.519)

-0.123 (0.427)

-0.406 (0.615)

R2

D.W.

0.007

1.905

0.023

1.893

0.000

2.160

0.016

2.108

0.001

2.208

0.033

2.327

0.001

2.481

0.003

2.484

0.106

1.609

0.107

1.621

0.005

2.100

0.013

2.124

Sources: Canada: Deflator, Firestone 1958; CPI, Urquhart and Buckley 1965, tables J165,

139; WPI, ibid., table J34. U.K.: Deflator, Feinstein 1971; CPI, McCloskey and Zecher 1976; Mitchell 1975 , series 12; WPI, Mitchell 1975 series 11; Board of Trade series spliced at 1919-20 and 1929-30. U.S.: Deflator, U.S. Bureau of the Census 1975, series F5; CPI, ibid., series E 135; WPI, ibid., series E 23; money supply and gold flows, Friedman and Schwartz 1963, pp. 704-7; real GNP, McCloskey and Zecher 1976; U.S. Bureau of the Census 1975, series F 3. Note: Values in parentheses are the standard errors of the coefficients.

146

Donald N. McCloskey and J. Richard Zecher

Table 2.3

Calculated Forecast Errors Implicit Deflator Canadian

U.K.

Purchasing-Power-Parity Forecast Errors For the U.S. as a Function of Measures of Excess Domestic-Credit Growth, 1880-1940

Constant

Current Excess DomesticCredit Growth

0.004 (1.175) -0.410 (1.223)

-0.002 (0.010) -0.004 (0.010)

-0.674 (1.069) -1.014 (1.116)

-0.008 (0.009) -0.010 (0.009)

0.146 (0.756) 0.012 (0.795)

-0.003 (0.006) -0.003 (0.006)

-0.094 (1.270) -0.233 (1.338)

-0.010 (0.010) -0.010 (0.011)

-0.405 (0.609) -0.088 (0.264)

-0.006 (0.005) -0.004 (0.005)

-0.012 (0.835) 0.098 (0.880)

-0.008 (0.007) -0.008 (0.007)

Lagged Excess DomesticCredit Growth

0.144 (0.124)

0.119 (0.113)

R2

D.W.

0.001

1.719

0.026

1.726

0.015

2.564

0.035

2.632

0.003

1.912

0.010

1.926

0.016

1.694

0.019

1.726

0.023

2.307

-0.076

2.246

0.026

1.937

0.029

1.898

CPI Canadian

U.K.

WPI Canadian

U.K.

0.047 (0.080)

0.048 (0.135)

-0.110 (0.063)

-0.038 (0.089)

Sources: See table 2.2. Note: Values in parentheses are the standard errors of the coefficients.

the rest of the world in their models implies such a belief. It is hard to believe that American prices and interest rates are not at all constrained directly by the forces of arbitrage. Yet the journals are filled with work embodying this belief. The failure of arbitrage necessary to validate a Martian model must be gross, not a matter of the fourth digit of accuracy but of the first. To the first digit of accuracy, and even to the second, the hypothesis of parity succeeds. One wonders what would happen to estimates of wage and price equations, or of the effects of domestic monetary policy on prices and interest rates, or of optimal economic policy under

147

The Success of Purchasing-Power Parity

rational expectations, if they were each asked to embody the international milieu of the U.S. economy to the second digit of accuracy. The hypothesis of parity survives the test for the reasons usual in economic arguments. Economists are embarrassed to assert in print that they possess the economic equivalent of a perpetual-motion machine, and the gross violations of rationality that opponents of purchasingpower parity believe they see entail such a machine. General equilibrium makes the hypothesis still more robust; even nontraded goods are substitutes in consumption and production with traded goods; a few characteristics are tradable in markets even when named goods are not; and the tendency for parity to hold in the long run gives opportunities for speculative profits if it does not hold in the short. The standards by which parity "fails" empirically are unclear, as many opponents of the hypothesis readily admit. The literature contains no articulation of standards. The introduction of standards casts into doubt all the recent attacks on parity. By standards that make intellectual sense the hypothesis succeeds. It succeeds in explaining the U.S. price level from 1880 to 1940 to a standard of accuracy demanded of such explanations. 16 It succeeds in explaining the price level in turbulent periods by the standard of tranquil periods. It succeeds in explaining the difference in prices among countries by the standard of the difference in prices among places in a single country. One of its competitors, closedeconomy monetarism, fails to explain the residual deviations from parity by the standard of statistical fit. Another of its competitors, the elasticity approach, fails to explain the balance of trade by the same standard, although again we express our doubt that much can be inferred from the uncontrolled experiments in curve fitting that characterize the literature and that we have dutifully followed here. And by the standard of good storytelling that underlies all economics, applied to episodes from their beginning to their end, the hypothesis of parity explains what happened under the gold standard. Purchasing-power parity is not a failure. On the contrary, by the standards we have examined, it is a great success. And at the least, speaking to the most skeptical reader, it is not so great a failure that macroeconomics can go on ignoring the rest of the world.

Notes 1. See R. Dornbusch and D. Jaffee's (1978) introduction to the special issue of the Journal of International Economics on purchasing-power parity. They remark that the Kravis and Lipsey paper in the issue leaves purchasing-power parity "rather in a shambles" (p. 159). 2. Samuelson, the chief theoretician of the real theory, however, is an implacable foe of purchasing-power parity.

148

Donald N. McCloskey and J. Richard Zecher

3. Gottfried Haberler once described Cassel's work as "one part Wicksell and nine parts water." One can also read the General Theory as taking prices to be given (not constant), because given by international factors such as purchasing-power parity. 4. Friedman and Schwartz (1982, p. 318) assume no quick operation of price-specie flow. Were it as quick as in the monetary approach, many of their other conclusions would be wrong, especially the effect of domestic money on prices. 5. Prices might have risen in anticipation of gold flows, though such rationality runs counter to the usual price-specie-flow argument. 6. See Darby 1982. Kravis and Lipsey (1978) make a similar point, arguing that the drift of purchasing power from parity over periods of a decade or so shows that parity is false. 7. Roll says further that "In an efficient market, something so easy to detect and so intoxicating to the arbitrager as a relative price difference in two locations would presumably display an immeasurably short half-life" (p. 136). 8. One of us (McCloskey 1981, chaps. 4, 6) has in fact done similar work with the notion that English businessmen failed to adopt profitable novelties in the late nineteenth century. The task is to see whether the second-guesser could have done better, recognizing the limitations of resources they faced (including a limitation on prescience). It is worth knowing for the present context that the second-guessing showed that the businessmen knew what they were doing. 9. The source for the trade statistics is U.S. Bureau of the Census 1960 based on declarations to American customs. The period covered is 1880-1940. 10. A good-or bad---example is the paper by J. D. Richardson (1978). Richardson regresses Canadian on U.S. prices (multiplied by the exchange rate) for a number of industries and concludes: "it is notable that the 'law of one price' fails uniformly. The hypothesis of perfect commodity arbitrage is rejected with 95 percent confidence for every commodity group" (p. 347). 11. To which may be added an authority overlooked by Kravis and Lipsey, Leland B. Yeager (1958). Yeager reckoned that if twenty-six out of thirty-five countries were within 25 percent of their 1937 parity by 1957, the hypothesis is confirmed. Richard Caves and Ronald Jones (1973) may be added to the affirmation. They remark on Yeager's results that "this performance seems rather good-just how good it is hard to say" (p. 338). The rueful remark, "just how good is hard to say," illustrates well how urgent it is to develop some standard. The state of play, largely favorable to purchasing-power parity, is well described in a comprehensive review by Lawrence H. Officer (1976). 12. In another article, Frenkel (1981), concludes that the "collapse" of PPP during the 1970s was due to a fundamental difference between exchange rates and prices. "Exchange rates reflect expectations about future circumstances while prices reflect more present and past circumstances" (p. 162). In this view, arbitrageurs in commodity markets look backwards in time while arbitrageurs in financial markets look forwards. For a different view, see Nattress and Zecher 1982, where a theory of the arbitrageur is developed. 13. Genberg writes (1976, p. 302): "it is evident ... that the differences between OECD countries are no greater on the average than those between cities within the United States. Thus, if we believe that the whole of the U.S. can be treated as a single market in a macroeconomic context, then the area composed of the above countries can be treated likewise." 14. Friedman and Schwartz speak of 1933-37 because of their commitment to analysis by cycles, 1933-37 being an upswing. Notice that by all measures except the consumer price index most of the price rise had occurred before 1935. The discussion of indirect influence of devaluation occurs at any rate on pp. 465ff, at the beginning of which a direct influence on "most farm products and raw materials exported by the United States" is mentioned. 15. BLS Handbook of Labor Statistics, 1937, pp. 512ff. The report referred to was "Report on the Operations of the National Recovery Act," p. 7. 16. Another related standard that we shall explore in later work is that of the relative

149

The Success of Purchasing-Power Parity

convergence of prices over time. Spooner's magnificent graphs (Braudel and Spooner 1967, pp. 470-71) show ranges of wheat prices expressed in silver in Europe and its offshoots of 6.66 to 1 around 1400, falling steadily to 1.88 to 1 around 1750. The divergences of the late nineteenth century, not to speak of the twentieth, look trivial beside these. Correspondingly, fixation on the "failure" of the unity of world markets for the period 1880-1939 looks odd indeed.

References Braudel, F. P., and Frank C. Spooner. 1967. Prices in Europe from 1450 to 1750. In The Cambridge Economic History of Europe, ed. E. E. Rich and C. H. Wilson, vol. 4. Cambridge: Cambridge University Press. Caves, Richard, and Ronald Jones. 1973. World trade and payments. Boston: Little Brown. Darby, Michael R. 1982. Does purchasing power parity work? In Proceedings of Fifth West Coast Academic/Federal Reserve Economic Research Seminar, December 1981. San Francisco: Federal Reserve Bank of San Francisco. Dornbusch, Rudiger, and Dwight Jaffee. 1978. Purchasing power parity and exchange rate problems: Introduction. Journal of International Economics 8 (May): 157-61. Einzig, Paul. 1937. The theory offorward exchange. London: Macmillan. Feinstein, Charles H. 1971. National income, expenditure, and output of the United Kingdom, 1855-1965. Cambridge: Cambridge University Press. Firestone, O. J. 1958. Canada's economic development, 1867-1913. Income and Wealth series VII. London: Bowes and Bowes. Frenkel, Jacob A. 1978. Purchasing power parity: Doctrinal perspective and evidence from the 1920s. Journal of International Economics 8 (May): 169-91. - - - . 1981. The collapse of purchasing power parities during the 1970s. European Economic Review 16 (May): 145-65. Friedman, Milton, and Anna J. Schwartz. 1963. A monetary history ofthe United States, 1867-1960. Princeton: Princeton University Press. - - - . 1982. Monetary trends in the United States and the United Kingdom, 1867-1975. Chicago: University of Chicago Press. Gandolfi, Arthur E., and James R. Lothian. 1982. A comparison of price movements among countries and states. Citibank. Typescript. Genberg, A. Hans. 1976. Aspects of the monetary approach to balance-of-payments theory: An empirical study of Sweden. In The monetary approach to the balance of payments, ed. J. A. Frenkel and H. G. Johnson. Toronto: University of Toronto Press.

150

Donald N. McCloskey and J. Richard Zecher

- - - . 1978. Purchasing power parity under fixed and flexible exchange rates. Journal of International Economics 8 (May): 247-76. Kravis, Irving B., and Robert E. Lipsey. 1978. Price behavior in the light of balance of payments theories. Journal of International Economics 8 (May): 193-246. Krugman, Paul R. 1978. Purchasing power parity and exchange rates: Another look at the evidence. Journal of International Economics 8 (Aug.): 397-407. McCloskey, Donald N. 1981. Enterprise and trade in Victorian Britain: Essays in historical economics. London: Allen and Unwin. McCloskey, Donald N., and J. Richard Zecher. 1976. How the gold standard worked, 1880-1913. In The monetary approach to the balance of payments. See Genberg 1976. Mitchell. B. R. 1975. European historical statistics. New York: Columbia University Press. Mood, Alexander M., and Franklin A. Graybill. 1963. Introduction to the theory of statistics. 2d ed. New York: McGraw-Hill. Nattress, Dayle, and J. Richard Zecher. 1982. The theory of the arbitrageur and purchasing-power parity. Chase Manhattan Bank. Typescript. Neyman, Jerzy, and E. S. Pearson. 1933. On the problem of the most efficient tests of statistical hypothesis. Royal Society of London. Philosophical Transactions 231 (series A): 289-337. Officer, Lawrence H. 1976. The purchasing-power-parity theory of exchange rates: A review article. IMF Staff Papers 23 (Mar.): 1-60. Richardson, J. David. 1978. Some empirical evidence on commodity arbitrage and the law of one price. Journal of International Economics 8 (May): 341-51. Roll, Richard. 1979. Violations of purchasing power parity and their implications for efficient international commodity markets. In International finance and trade, ed. M. Sarnat and G. Szego. Cambridge: Ballinger. Taussig, Frank W. 1918. International freights and prices. Quarterly Journal of Economics 32 (Feb.): 410-14. U.S. Bureau of the Census. 1960. Historical statistics ofthe United States, colonial times to 1957. Washington, D.C.: Government Printing Office. - - - . 1975. Historical statistics of the United States, colonial times to 1970. Part 1. Washington, D.C.: Government Printing Office. Urquhart, M. C., and K. A. H. Buckley. 1965. Historical statistics of Canada. Toronto: Macmillan of Canada. Wicksell, Knut. 1918. International freights and prices. Quarterly Journal of Economics 32 (Feb.): 404-10. Yeager, Leland B. 1958. A rehabilitation of purchasing-power parity. Journal of Political Economy 66 (Dec.): 516-30.

151

The Success of Purchasing-Power Parity

Comment

Robert E. Lipsey

How to define purchasing-power parity and how to test its "success" is the central issue of McCloskey and Zecher's paper. The authors come to two conclusions about purchasing-power parity, both favorable. One conclusion is derived from their reasoning and one from their tests. The first conclusion is that purchasing-power parity works instantaneously and perfectly, immediately eliminating any genuine differences in prices or divergences in relative price movements. The second is that the theory works sufficiently well that foreign influences on prices cannot be ignored. I shall argue that the first conclusion, as stated, depends on a definition of purchasing-power parity that makes it a tautology, not susceptible to proof or disproof, and that the second conclusion is valid but would not be disputed by many "critics" of the theory. I shall argue in addition that the strongest test they propose has been performed, with results that are unfavorable to the theory by their own criterion. The Definition of Purchasing-Power Parity In discussing the meaning of the concept, the authors say that they view purchasing-power parity as resulting from arbitrage in commodity markets. To me that means they identify the concept with the operation of the law of one price rather than with vaguer notions of aggregate country price levels or price changes. For example, "purchasing-power parity is a consequence of rationality in arbitraging. If all the opportunities for riskless (or insured) arbitrage among countries that are profitable at existing interest rates and other costs of arbitrage have taken place, then the price level of the world may be said to have exhausted its ability to determine the price level of one country" (p. 127), and "If opportunities for arbitrage are exploited (allowing fully for the cost of transport and information), then the price level of one country is fixed by the rest of the world, even in the very short run" (p. 129). Two things can be said about this definition or "theorem" or "higherorder proposition" as the authors refer to the latter version. One is that it is not what most people mean by purchasing-power parity. A second is that, stated this way, it is only a definition, not a theory, and is not susceptible to testing. The authors make no effort to apply it empirically. If one includes and takes seriously all the qualifying phrases about costs of arbitrage over space and time, most of which are not measured and probably cannot be measured, it is impossible to show that purchasingpower parity has been violated. The price of product x in country a plus all costs of arbitrage (including information costs, advantages of continuity of supplier relationships, costs of adaptation of existing machinery and Robert E. Lipsey is professor of economics at Queens College of the City University of New York and research associate of the National Bureau of Economic Research.

152

Donald N. McCloskey and J. Richard Zecher

work habits, etc.) equals the price of product x in country b. That statement is an identity; it tells us nothing about how the world works. By this standard, prices on Mars and Venus satisfy purchasing-power parity even though there is no communication between them. The information and arbitrage costs are so high that any prices or price changes satisfy the authors' criterion: no one can make money by arbitrage between the two planets. Critics of purchasing-power parity do not deny this tautology. What they argue is that adjustment costs are large in some cases, that it may take a long time to overcome them, that prices can move relative to each other in the meantime, and that consumers and producers react to these price differences and relative price movements. The authors illustrate their point about arbitrage with examples of changing price differences for electrical equipment, cement, and vacuum cleaners. It is clear from these examples that they picture the mechanism often called the law of one price, referring to prices of individual commodities, as enfor.cing purchasing-power parity. They do not give vivid examples of arbitrageurs buying the U.S. CPI or WPI or GNP deflator and selling that of the U.K. Of course, if the law of one price operated exactly and instantaneously, that is, if prices of carefully defined individual products were identical everywhere or moved identically in different countries, the levels of prices in general and their movements would be similar or would move similarly. However, aggregate price levels and price changes would not be the same. Given the differences among countries in the composition of consumption and production and what are probably even greater differences in the way that aggregate price measures are constructed, one must be careful in moving from one kind of statement to the other. The authors take little note of this point and speak about the movements of vaguely defined and badly measured price aggregates in different countries and the responses to them as if they were useful for testing the opportunities for, or existence of, arbitrage. I do not think they are. Testing Purchasing-Power Parity The first test the authors propose is that a critic of purchasing-power parity show that he has gotten rich on arbitrage profits. "It surpasses belief that many opportunities to make easy money buying low and selling high persist long enough to be observed in economic data" (p. 128). "Often unrecognized by critics of purchasing-power parity is that their conclusion that it has failed usually implies an ability to make money. Anyone who knew that purchasing-power parity was true in some long run but not yet true in the present would have a rosy financial future. The divergences from purchasing-power parity detected in the literature are so gross and the statistics purporting to show the divergences so easy

153

The Success of Purchasing-Power Parity

to collect that the opportunities for profit are large. Go thee and prosper" (p. 131). "Economists are embarrassed to assert in print that they possess the economic equivalent of a perpetual-motion machine, and the gross violations of rationality that opponents of purchasing-power parity believe they see entail such a machine" (p; 147). The number of times the point appears shows that the authors take it seriously, but I have never seen a criticism of purchasing-power parity that implied irrationality on the part of purchasers. There is strong evidence that in the capital goods and other complex products that form a large part of the trade of developed countries, price differences and divergences in price movements exist. They are not arbitraged away immediately, but they do eventually bring about shifts in trade that tend to remove them (an indication that they are not simply illusory or due to differences in specifications). The reasons are implicit in some of the authors' own discussion: information is costly and the risks of purchasing unknown types of machinery and dealing with unfamiliar suppliers are high and uninsurable. A fall in the price of a Japanese machine might at first produce no shift in purchases because buyers were unaware of the change, uncertain about its permanence, or skeptical about the quality of the machine or the availability of spare parts. There would be no violation of the law of one price by the authors' definition, because the price difference was insufficient to offset information or risk or insurance costs. After the lower price had been in effect for a year, information would become more widely and cheaply available, risk and insurance costs would decrease, and some buyers would switch. After another year more information would be available and still more buyers would switch. At each point, the price in Japan plus information, risk, and insurance would have been equal to the price in the United States. Therefore this sequence of events, which would suggest a violation of the law of one price to most observers, would be in conformity with it by the authors' definition. In view of the impracticality of calculating in retrospect all the costs of arbitrage, the authors go on to suggest a "more practical" test. That test is "whether the shipments of goods implied by the supposed opportunities for arbitrage in fact occurred.... A country that exhibits divergences from purchasing-power parity convincing to the doubters should also exhibit lower exports and increased imports ... a place with 'high' prices would have a hard time selling and an easy time buying" (p. 131). One might expect that the authors would perform their "practical" test on the data for commodities defined as narrowly as possible to observe the action or inaction of arbitrageurs. Instead, they perform their tests of responses to deviations from purchasing-power parity between the United States and the United Kingdom and the United States and Canada using aggregate price indexes. If they seriously believe that profit

154

Donald N. McCloskey and J. Richard Zecher

seeking by arbitrageurs is the force at work, and they wish to observe that action, why measure purchasing-power parity from price aggregates? The arbitrageur cannot easily buy the V.S. wholesale price index or GNP deflator. A more appropriate test of the workings of arbitrage, and one the authors themselves suggest, would be to compare price and trade changes for identical or related goods such as the electrical equipment they refer to. As they say, "Further tests of the hypothesis that profitable opportunities for arbitrage arise when measured prices diverge might proceed commodity by commodity." In fact, Irving Kravis and I, using price indexes that were constructed so that the same goods with the same weights were represented in two countries' price measures did perform many versions of the "practical" test they suggest, although not with the intention of testing purchasingpower parity. That is, we investigated whether price changes and differences in price levels did lead to shifts in trade. We found strong evidence that they did and that these shifts took years before they were completed. We explained the reasons for these lags in our book on price competitiveness (Kravis and Lipsey 1971) and in a series of later articles. l The reasons we gave for the price and price-change divergences we found did not imply that there were overlooked opportunities for aboveaverage profit, given the costs of information, costs of adjustment, and uncertainties about the permanence of price changes. It was true that the first V.S. buyers of foreign electrical generating equipment and foreign steel paid less than their competitors who hesitated. We cannot say what the rational policy was for a buyer of generating equipment given the possible uncertainties at the time. The first buyers were public systems which may have faced less danger from mistakes than private utilities. The private utilities may have had little incentive to lower costs, given the way their prices were regulated and the lack of incentives for managers to break cozy relationships with domestic suppliers. The important point is that whatever the reason, gaps in prices persisted for a long time and produced not sudden but gradual shifts in trade; but they did produce the shifts that McCloskey and Zecher imply would refute purchasing-power parity. In looking back at these past episodes in which price differences gradually gave rise to shifts in trade, we do not know whether the first V. S. buyers of foreign electrical generators and transformers or the first V.S. buyers of foreign steel reaped exceptional profits for a time. Even if they did, the profits, as usually measured, may have done no more than compensate for the risks, given the uncertainties about the quality of foreign products, the reaction of the V.S. government and regulatory agencies, the commitment of foreign suppliers to technical assistance, service, and continuity of supply, and many other factors. Thus these events may not represent a violation of purchasing-power parity by their

155

The Success of Purchasing-Power Parity

definition (it would be difficult to do that), but fail their test of conformance with the theory by the "practical" test. If, as stated in the text (p. 127), the hypothesis has to do with the rationality of arbitrageurs, the tests they offer are irrelevant. I know of no writer skeptical of purchasing-power parity who has stated or implied or assumed that the reason for deviations from it is that traders ignore profit opportunities. They believe there have been deviations from purchasingpower parity because it was not profitable to exploit all opportunities or to exploit them immediately, given the costs and uncertainties involved. One does not add to understanding of the international economy by assuming away all these deviations from purchasing-power parity. The paper includes several of the authors' own tests, but these are tests of what I called at the beginning of my comments their second conclusion, or the second version of purchasing-power parity. That second, or weak, version is that "it is hard to believe that foreign prices . . . did not matter" or "that American prices are not at all constrained directly by the forces of arbitrage" (p. 146). They test this version by regressions of the U.S. GNP deflator against aggregate Canadian and British price indexes adjusted by exchange rates, and by plots of the U.S. price indexes predicted by Canadian and British prices and of the prediction errors from the equations. The correlations are "high," they conclude, and "the foreign prices do predict the gross outlines of the American price." Furthermore, they report, the equations predict as well in the turbulent period 1921-40 as in the "tranquil" period 1880-1914. Aside from the point that these equations represent the "uncontrolled experiments in curve fitting" the authors are so critical of in others, the equations are unconvincing in other ways. For example, for the first twenty years of the chart the Canadian price predicts nothing of U.S. price movements. We are then told that foreign prices did no worse as predictors of U.S. prices in turbulent times than in "tranquil" times. But for half of the "tranquil" period foreign price did not predict at all-not a very exacting standard for judging the predictive power of Canada's prices to estimate later U.S. prices. A more serious objection to these tests is that there is no consideration of the possibility that the high correlation stemmed from the effects of common factors, such as World War I and the Great Depression on U.S., Canadian, and British prices at the same time, rather than from the dependence of one country's prices on another's. Judging the Results of Tests of Purchasing-Power Parity A substantial part of the paper is devoted to standards by which to judge the results of tests of purchasing-power parity. The authors take many of their colleagues to task for using vague and ill-defined standards. I cannot see that they escape the same problem. Their paper is filled with the same undefined terms they deplore in others. One reason for their

156

Donald N. McCloskey and J. Richard Zecher

vagueness and that of everyone else is that no one standard can be used for all purposes. The precision required by a speculator might be greater than that needed by a company making a long-term investment or by a government or international agency trying to set an exchange rate, and all of these standards may be far above that required for satisfaction by devotees of purchasing-power parity. The authors' judgments of the results reported by others are clouded by the fact that they seem to conceive of only two possible conclusions. Either the theory is a "success" and is graded A + , or it is a failure and is graded F. They seem to be determined to come out with a grade of A + and to think that all analyses that find deviations from purchasing-power parity imply a grade of F. In fact, most of the studies they cite seem to imply judgments of B or C rather than total success or failure. In summary, my reaction to the two tests of purchasing-power parity proposed by the authors is that one test cannot be failed and that the other test, conducted with the right type of data, usually is failed. Fortunately, there is another theme to the paper, although it is obscured by the extravagant claims made for purchasing-power parity. It is the important and reasonable one that "it is hard to believe that foreign prices or interest rates did not matter. It is hard to believe that American prices and interest rates are not at all constrained directly by the forces of arbitrage," and "it [purchasing-power parity] is not so great a failure that macroeconomics can go on ignoring the rest of the world." If that is the point the authors really want to make, even many skeptics about purchasing-power parity could agree. Note 1. Some of these were referred to in the article (Kravis and Lipsey 1978) quoted by McCloskey and Zecher; we might add to the list Kravis and Lipsey 1982. Of course, similar studies by many others of price elasticities in trade have yielded similar results.

References Kravis, Irving B., and Robert E. Lipsey. 1971. Price competitiveness in world trade. New York: National Bureau of Economic Research. - - - . 1978. Price behavior in the light of balance of payments theories. Journal of International Economics 8 (May): 193-246. - - - . 1982. Prices and market shares in the international machinery trade. Review of Economics and Statistics 64 (Feb.): 110-16.

157

The Success of Purchasing-Power Parity

Comment

Milton Friedman

I believe that a fundamental confusion runs through the McCloskeyZecher treatment, both in this paper and in their earlier paper-a confusion between two very different propositions. One proposition is whether the quantity of money in a country is an endogenous or an exogenous variable-an important and very interesting question. There is no doubt that in a world of fixed exchange rates and stable barriers to trade, the quantity of money in each country separately is ultimately an endogenous variable. That proposition is perfectly clear and everybody has accepted it for a long time. It's the proposition that Keynes presented so well in the appendix to his Tract on Monetary Reform ([1923] 1971), that there are three things of which a country can choose any two: stable internal prices, stable exchange rates, and free trade. You can't have all three; you can have only two. That proposition is critical. It should be noted, however, that while the quantity of money is ultimately an endogenous variable, there can be and is much leeway in the short run, before the external forces overwhelm the independent internal effects. And we have repeatedly been surprised in our studies by how much leeway there is and for how long-frequently a number of years. There's a very different proposition that is easily confused with the endogeneity or exogeneity of the quantity of money, namely, if money is endogenous, there is no causal relation between money and prices. That is a whole different proposition. Whatever may determine the quantity of money within a country, that quantity of money may still largely determine-or at least, be the conduit through which other forces determineprices within a country. The confusion between these two wholly different propositions is apparent in the statement by McCloskey and Zecher that "if purchasing-power parity is found to be a useful characterization of the world"-and they should have added "and fixed exchange rates characterize the world"-"then closed-economy theorizing and empirical work in macroeconomics should be changed to allow for the direct effects of international price arbitrage. Whether monetarists or Keynesians or rational expectationists, economists should begin thinking and measuring in global terms" (p. 127). Economists have consistently thought and measured in global terms in examining the determinants of the quantity of money in a country, both for periods of fixed exchange rates and of dirty floating-and McCloskey and Zecher cite no examples to the contrary. In respect of the second proposition, the money supply Milton Friedman, Nobel laureate, is senior research fellow at the Hoover Institution, Stanford University, and Paul Snowden Russell Distinguished Service Emeritus professor of economics at the University of Chicago.

158

Donald N. McCloskey and J. Richard Zecher

may be endogenous after a sufficient interval, yet not in the short run. And whether endogenous or exogenous, the domestic money supply is absorbed primarily within the domestic economy; it is highly fruitful to examine the process whereby that occurs and the relation within a country between changes in the money supply and changes in other variables. There is no such sharp dichotomy between "closed economy" and "global" thinking and measuring as the straw man set up by McCloskey and Zecher. Some findings from Anna Schwartz's and my book Monetary Trends are relevant to this subject. We calculated correlations between the United States and the United Kingdom in the rates of change of various variables between cycle phases for almost a century. We pointed out, and this is strictly in accord with the McCloskey-Zecher view, that the correlation between prices in Britain and the United States is closer than the correlation between any other two magnitudes. It is closer than the correlation between money supplies in the two countries, closer than the correlation between income in the two countries, real or nominal. We stressed that that result is strictly consistent with what they call our Martian view of the economy, and indeed dema·nded by it. Because, we said, in a world of fixed exchange rates, the money supplies in the different countries have to accommodate themselves in such a way as to be consistent with equality of prices and goods among countries. And therefore, the relation between the quantity of money in the two countries would be expected to be less close than between prices because changes in the demand curve for money within an individual country must be accommodated by changes in money rather than in prices (Friedman and Schwartz 1982, pp. 310-15). Let me turn to a couple of other points. First, McCloskey and Zecher assert that "the turbulence of the 1920s and 1930s, which is said to have loosened the economic ties among nations, appears not to have done so" (p. 139). One comparison in our book supports a very different conclusion. We started with the Kravis, Heston, and Summers estimate for the purchasing-power-parity exchange rate between the United States and the United Kingdom in 1970. We used price indexes in the two countries to extrapolate the purchasing-power exchange rate annually from 1870 to 1970. We then calculated the ratio of the purchasing-power-parity exchange rate to the market exchange rate. The resulting chart is fascinating (Friedman and Schwartz 1982, chart 6.5, p. 291). Before about 1932, the ratio of the purchasing-power-parity exchange rate to the market exchange rate varied within a range of plus or minus 10 percent. All of us would say that is a fairly close relationship to purchasing-power parity. After 1931, the range is between minus onequarter and plus one-third-an enormously wider range. There is no

159

The Success of Purchasing-Power Parity

doubt that a comment by Jacob Frenkel is right: Hume traveling in the 1950s would have found more deviations from purchasing-power parity than he would have found in the 1780s. Exchange controls, tariffs, other impediments to trade were far more important in dividing the world than improvements in communication and other technologies were in uniting it. I don't see that notion is any contradiction to the purchasing-powerparity theory of exchange rates. It is simply a consequence of the fact that there has been an enormous increase in barriers to trade among countries since 1931. I want now to discuss two particular episodes for which McCloskey and Zecher take Anna Schwartz and me to task for our analysis in A Monetary History (1963). The first example is resumption in 1879. They quote our statement that "it would be hard to find a much neater example in history of the classical gold-standard mechanism in operation" (p. 99). Their look at that episode on the basis of monthly data is interesting and most welcome, but on closer examination it does not, contrary to their claims, contradict our interpretation of the episode. McCloskey and Zecher compare price rises to inflows of gold, concluding, "In the monthly statistics ... there is no tendency for price rises to follow inflows of gold . . . ; if anything, there is a slight tendency for price rises to precede inflows of gold, as they would if arbitrage were shortcutting the mechanism" (p. 126). Their comparison is the wrong one for determining whether prices were reacting to arbitrage rather than reflecting changes in the quantity of money. For that purpose the relevant comparison is with the quantity of money. Gold flows are relevant only as a proxy for the quantity of money. If we compare price rises with changes in the quantity of money directly, a very different picture emerges than McCloskey and Zecher draw (see table C2.1). Our basic estimates of the quantity of money for this period are for semiannual dates, February and August. Resumption took effect on 1 January 1879. From August 1878 to February 1879, the money supply declined a trifle, continuing a decline that had begun in 1875 in final preparation for resumption. From February 1879 to August 1879, the money supply rose sharply, according to our estimates, by 15 percent. The Warren-Pearson monthly wholesale price index fell in the first half of 1879, reflecting the earlier decline in the money stock. It started its sharp rise in September 1879, or at least seven months later than the money supply.1 As to gold, the total stock of gold, as well as gold held by the Treasury, had been rising since 1877 as part of the preparation for resumption. But it had been rising at the expense of other components of high-powered money, which actually fell slightly. However, the decline in the money stock before 1879 had been due primarily to a decline in the depositcurrency ratio and the deposit-reserve ratio. After successful resumption,

160

Donald N. McCloskey and J. Richard Zecher

both ratios rose, which enabled the stock of money to rise despite no initial increase in gold flows. The large step-up in gold inflows in the fall of 1879, to which McCloskey and Zecher call attention, was mostly absorbed in raising the fraction of high-powered money in the form of gold rather than in speeding up monetary growth. Table C2.1 shows these developments in more detail. On the basis of this reexamination, I am inclined to repeat the statement in our Monetary History, "It would be hard to find a much neater example in history of the classical gold standard mechanism in operation" (italics added), at least on a sophisticated interpretation of both the gold standard mechanism and the historical data. At any rate, the gold standard mechanism, as I understand it, has always incorporated a variety of channels of adjustment, subject to different lags. Any result one wants can be gotten, depending on the relative speed of adjustment of the various channels. It is an important scientific question to try to identify and isolate these relative speeds of adjustment. I believe that McCloskey and Zecher make a real contribution in examining aspects of that issue. The second episode is the behavior of prices after the u.s. went off gold in 1933. The figure 2.3 shows a close parallelism between the weekly movements of wholesale prices and the exchange rate. However, the different scales used for the price index and the exchange rate in the chart give a misleading impression? For example, from April 1933 to July 1933, wholesale prices rose less than a sixth, the exchange rate by nearly a half, yet the total impression from their chart is that prices rose more sharply. Logarithmic scales would give a more accurate picture and make clear Table C2.1

Aug. Feb. Aug. Feb. Aug.

1878 1879 1879 1880 1880

Relations between U.S. Prices, Money, High-Powered Money, and Gold, 1878-80 Wholesale Prices (P) (1)

Money Stock (M) (2)

Highpowered Money (HPM) (3)

90 88 86 105 97

1.57 1.55 1.78 1.94 2.05

0.767 0.752 0.815 0.897 0.972

Gold Stock (G)

MIHPM

(4)

(5)

GIHPM (6)

0.182 0.198 0.219 0.302 0.378

2.05 2.06 2.18 2.16 2.11

0.24 0.26 0.27 0.34 0.34

Sources: Col. (1) U.S. Bureau of the Census 1949, app. 24, p. 344; col. (2) Friedman and Schwartz 1970, p. 5; col. (3) Friedman and Schwartz 1963, p. 799; col. (4) ibid., notes to table A-I, p. 723 and table A-3, p. 765. Notes: Col. (1) Warren-Pearson index of wholesale prices (191~14 = 100); col. (2) currency held by the public plus adjusted deposits of commercial banks (billions of dollars); col. (3) in billions of dollars; col. (4) in billions of dollars.

161

The Success of Purchasing-Power Parity

how much narrower the relative movement in the wholesale price index was than in the exchange rate. In any event, as McCloskey and Zecher note, we pointed out in A Monetary History that there was a direct effect of devaluation on prices. However, the existence of a direct effect on wholesale prices is not incompatible with the existence of many other prices, as Moe Abramovitz has remarked, such as non-tradable-goods prices, that did not respond immediately or responded to different forces. An index of rents paid plotted against the exchange rate would not give the same result. An index of wages would not give the same result. It may be worth quoting what we actually said on the issue, especially in view of the McCloskey-Zecher comment on a quotation from the Monetary History that "this is an unusual line of reasoning for such crusaders against mixing up the determination of relative and absolute prices" (p. 141).3 Here is what we said: "The aim of the gold policy was to raise the price level of commodities, particularly farm products and raw materials.... Most farm products and raw materials exported by the United States had a world market in which this country ... was seldom dominant.... Hence, the decline in the foreign exchange value of the dollar meant a roughly proportional rise in the dollar price of such commodities, which is, of course, what did happen to the dollar prices of cotton, petroleum products, leaf tobacco, wheat and similar items" (Friedman and Schwartz 1963, pp. 465-66). Thirty-odd pages later, after noting that the rise in the implicit price index from 1933 to 1937 was of roughly the same order of magnitude as in 1879-82 and 1896- or 1897-99, but in wholesale prices decidedly larger: "What accounts for the greater rise in wholesale prices in 1933-37, despite a probably higher fraction of the labor force unemployed and of physical capacity unutilized than in the two earlier expansions? One factor, already mentioned, was devaluation with its differential effect on wholesale prices" (p. 498). This was followed by the passage McCloskey and Zecher quote in which we referred to "the implicit measures to raise prices and wages undertaken with government encouragement and assistance" (p. 498). Contrary to the impression McCloskey and Zecher give, we did not try to assess the relative importance of various factors in explaining the rise in prices from 1933 to 1934---the period to which they limit their chart and discussion. On the contrary, we explicitly cited these measures as helping to explain the "rise in wholesale prices in 1933-37." The wholesale price index continued to rise after its initial sharp rise in 1933 and did not reach its peak until mid-1937 when it was 47 percent above its low point in February 1933 and 28 percent above its level in July 1933. Hence there was ample time for the factors we referred to to play their part after the enactment of the legislation we listed.

162

Donald N. McCloskey and J. Richard Zecher

Note finally that McCloskey and Zecher have faced up in this paper to the problem of floating or flexible exchange rates to only a very minimal extent. This paper is written primarily for a world of fixed exchange rates, and indeed, fixed exchange rates with nonchangeable barriers to trade. To be applicable to the current world, those elements must be added. Notes 1. This paragraph and the next two were added after the conference in revising my comments for publication. 2. This sentence and the next two were added in revising these comments for publication. 3. This paragraph and the next three were added in revising these comments for publication.

References Friedman, Milton, and Anna J. Schwartz. 1963. A monetary history ofthe United States, 1867-1960. Princeton: Princeton University Press. - - - . 1970. Monetary statistics of the United States. New York: National Bureau of Economic Research. - - - . 1982. Monetary trends in the United States and the United Kingdom, 1867-1975. Chicago: University of Chicago Press. Keynes, J. M. [1923] 1971. The applied theory of money: A treatise on money. Vol. 4 of The collected writings of John Maynard Keynes. London: Macmillan and New York: Cambridge University Press for the Royal Economic. Society.

General Discussion ABRAMOVITZ suggested another way of posing the question that lies at the heart of the McCloskey-Zecher paper: How can tradable-goods prices remain equal to one another or move in similar ways in different countries without destabilizing the gold standard, in the face of differing national and sectoral rates of productivity growth? Abramovitz pointed out that McCloskey and Zecher offer one possible adjustment mechanism. In contrast, the traditional specie-flow mechanism, involving changes in prices and nominal wages in different countries, offers a rather different mechanism. MCCAULEY asked McCloskey and Zecher to justify the leap from purchasing-power parity, however defined, to the assertion that monetary policy cannot alter prices. This assertion appears in weak form (p. 146)-that American prices are not at all constrained by the forces of

163

The Success of Purchasing-Power Parity

arbitrage-and in rather stronger form (p. 128)-that American price levels are exogenous to American monetary policy. Surely as the share of world output produced by a single country approaches unity, an individual country's monetary policy becomes capable of raising prices. McCauley argued that these questions imply technical objections to McCloskey and Zecher's analysis. While McCloskey and Zecher analyze the residuals from their purchasing-power-parity equation, their procedure presumes that U. S. monetary policy is incapable of influencing price levels in the rest of the world. ZECHER responded that the Zecher-McCloskey paper effectively incorporates McCauley's point. In response to Friedman, Zecher disagreed with his statement that McCloskey and Zecher fail to take into account the difference between periods of flexible and fixed exchange rates. Under flexible exchange rates, McCloskey and Zecher's assertion is not that a country cannot affect its own price level or rate of inflation, but rather that relative prices, or the deviation from purchasing-power parity, is constrained by arbitrage in commodity and other markets. Thus, their analysis is capable of dealing with flexible-exchange-rate periods. FRIEDMAN restated the central point of his argument: suppose a country's money supply is endogenous, determined by the outside world. One can still examine the relationship between the quantity of money in that country (call it Illinois) and the price level and nominal income in Illinois. The change in the quantity of money, however produced, has effects internal to Illinois. MCCLOSKEY responded that all monetarists share a belief in a stable demand for money. But to go from the presumption that money demand is stable to the assertion that money supply in Illinois determines prices and interest rates in Illinois is a large jump. FRENKEL raised the question of what exactly McCloskey and Zecher mean when they speak of purchasing-power parity? He suggested that McCloskey and Zecher essentially mean the law of one price. Purchasingpower parity· is enforced by the mechanism of commodity and asset arbitrage, converting the whole discussion of purchasing-power parity into a discussion of financial flows and profit opportunities. Frenkel suggested posing a very different question, which was the original question underlying the development of purchasing-power parity: How can one determine an appropriate exchange rate for the period following a serious market dislocation? How much information can be obtained from aggregate price indexes? One issue is which "aggregate" to look at. Needless to say, this question is based on the presumption that aggregates provide useful information for determining equilibrium exchange rates; changes in relative prices call this view into question. The crucial question, therefore, is under what conditions it is likely that aggregate price indexes will provide useful information about equilib-

164

Donald N. McCloskey and J. Richard Zecher

rium exchange rates? It may be very important to know whether the shocks to the system originate from the real side or from the monetary side. In the wake of monetary shocks, Frenkel argued, it is advisable to focus on the aggregate that best represents monetary conditions. This view suggests looking at the price indexes that cover the broader domain of goods and services, which is what Cassel had in mind. In his view, purchasing-power parity was not a theory of individual prices but of price indexes because it was intended as a measure of the monetary conditions. His view is fundamentally different from the view that foreign exchange rates have nothing to do with aggregate price levels but only with individual commodity prices. Frenkel pointed out that the original view of purchasing-power parity refers to the ratio of rates of inflation of purchasing power, where inflation is to be understood as inflation of the quantity of money. These concepts are completely divorced from individual commodity prices per see Frenkel made a number of points concerning McCloskey and Zecher's econometric results. One interesting exercise in their paper asks whether deviations from purchasing-power parity have real effects? McCloskey and Zecher choose to concentrate on the trade balance; they ask if there is a visible, statistically significant relationship between apparent deviations from purchasing-power parity and the trade balance. They find no such evidence. Frenkel hesitated to infer from these results any particular conclusion about purchasing-power parity, since it is not clear that changes in relative prices should always have a particular trade-balance effect. If one thinks of the current account as the difference between income and spending, then there is a determinate theory that links changes in the relative prices of commodities to the aggregate saving ratio and hence to the current account. There is a determinate link between the terms of trade and the current account-the so-called Laursen-Metzler effect. However, under a variety of plausible circumstances, the LaursenMetzler effect might not hold. ZECHER responded to Frenkel by emphasizing the importance of questioning the extent to which national markets are integrated. The problem with many recent empirical studies, he suggested, is that after concluding that purchasing-power parity fails, subsequent theorizing simply neglects the rest of the world. It is important to attempt to define criteria that permit one to label markets as more or less integrated. PIPPENGER pointed out that several economists have analyzed the residuals from purchasing-power parity calculations. He himself had examined deviations from purchasing-power parity using annual data, going back in some cases to the 1870s. The evidence indicates that deviations follow a random walk. There appears to be no tendency for

165

The Success of Purchasing-Power Parity

relative prices to return to any normal long-run level. This random-walk property holds even for the regional price indexes within the United States, which is indeed curious. Of course, this result is inconsistent with the continuous maintenance of purchasing-power parity, because the result permits exchange rates to drift in any direction. It also raises some interesting questions about the proper way of interpreting time-series data. One possible interpretation is that purchasing-power parity simply fails to hold. Another, suggested by Richard Roll, is that the randomwalk property is evidence of efficient commodity markets. This latter interpretation would suggest that Zecher and McCloskey are right, but for a different reason. What we may be seeing is simply the fact that price indexes for different countries are made up of different commodity bundles. Many problems must be sorted out before we can distinguish between the traditional view and the Zecher-McCloskey view. BRUNNER argued that one may wish to distinguish a shorter run, perhaps up to one-and-a-half or two years, over which the money stock is exogenous, and a longer run over which it is endogenous. Support for this distinction can be found in the history of the Italian monetary affairs in the 1960s. At that time there were a number of one-and-a-half- or two-year periods when macroeconomic accelerations were fueled essentially by the domestic-credit component of the monetary base. Such credit creation was able to alter the money supply for one-and-a-half or two years with accompanying adjustments in prices. Only thereafter was the balance of payments affected. Such lagged responses are quite consistent with some of the adjustment mechanisms sketched here. DORNBUSCH suggested that McCloskey and Zecher had provided insufficient room in their framework for the considerations emphasized by Abramovitz. When an economy is growing and the composition of activity is changing over time, simple tests of purchasing-power parity will be biased. For example, anyone who tests the purchasing-power-parity hypothesis for the last twenty years will find that real exchange rates in manufacturing, of the United States as well as of any other industrialized country, are well explained by differentials in sectoral growth rates but not by national rates of price inflation. Therefore, serious tests of purchasing-power parity must incorporate time trends or other variables designed to account for differentials in sectoral growth rates and other real changes. Dornbusch also drew attention to the financial research of the last three years, which demonstrates that one cannot reject the hypothesis that the stock market devia~es for long periods from market fundamentals. On purely statistical grounds, even fifty years of Dow-Jones data are incapable of rejecting, at a .99, a .95 or even a .90 level of confidence, the hypothesis that the stock market is driven for long periods by fads and fashions.

166

Donald N. McCloskey and J. Richard Zecher

MCGOULDRICK asked whether empirical tests of purchasing-power parity should include in the price indexes not only the prices of goods but the prices of securities. MCCLOSKEY agreed that in principle the capital market should be considered in studies of purchasing-power parity. He pointed out that Lipsey takes the McCloskey-Zecher analysis of purchasing-power parity as an empirical test and argues that it is not in fact properly interpreted as a statistical test of a hypothesis. Rather, the analysis is better thought of as a way of looking at particular episodes which might persuade people of the plausibility of a particular view of how markets function. McCloskey suggested that Friedman was in substantial agreement with the authors' main point. Friedman concedes that under fixed exchange rates the money supply of Illinois does not determine prices and interest rates in Illinois. That was McCloskey and Zecher's main point. FRIEDMAN referred back to some of the work in his recent book written with Anna Schwartz. Friedman and Schwartz discovered an appreciable difference in the relation between interest rates in the United States and United Kingdom before 1896 and after 1896. Interest rates in the United States before 1896 are much higher relative to the British interest rates than after 1896. In other words, before-1896 interest rates act as if there was widespread anticipation of a depreciation of the U.S. dollar. After 1896, they act as if there was widespread expectation of an appreciation of the U. S. dollar. This behavior bears on the question of whether there can be significant deviations in prices and interest rates in various countries over substantial periods of time. FRATIANNI suggested parallels with the Italian experience. In Italy, he argued, systematic deviations from purchasing-power parity are matched by deviations from interest-rate parity. Also, the way in which governments finance budget deficits should be accounted for in regressions. GREGORY reported an experiment conducted by himself and colleagues (Baltagi and Sailors) at the University of Houston. They estimated a three-equation model, pooling thirty-four cross-section time-series observations for seven countries (France, Germany, Japan, Russia, Sweden, the United Kingdom, and the United States), to investigate the working of the gold standard. The model included a money-demand equation, a balance-of-payments equation, and a money-supply equation, with all variables expressed as first differences. The pooled regression results suggest that the classical gold standard was a fairly simple system, not the complex one that Bloomfield reported. According to the model, if supply shocks were to set off domestic inflation, domestic inflation would then cause a worsening of the external balance, and a worsening of the external balance would cause the domestic money supply to drop. While the pooled model yielded these statistically significant and plausible results, the individual-country time-series regressions

167

The Success of Purchasing-Power Parity

yielded generally insignificant coefficients. Gregory suggested that pooled data might serve as an avenue for expanding the empirical data set in exploring the working of the classical gold standard. THOMAS raised a major question with respect to the model that Gregory described. An outstanding characteristic of the period 1880-1913 was the unique role of Great Britain in the Atlantic economy, as major capital exporter and center of what was virtually a sterling standard; her status was different in kind from that of the borrowing countries of the periphery. In 1880 the United Kingdom was responsible for 41 percent of world exports of manufactured goods as against 3 percent for the United States; even as late as 1889 the U.K. share of world exports of capital goods was 44 percent as compared with the U.S. proportion of 23 percent. A unique feature of the growth process was the fact that the long swings in capital formation, productivity, and real income in the center country, Great Britain, were inverse to the corresponding long swings in the borrowing countries of the periphery. The voluminous evidence confirming the validity of these inverse long swings was summarized by Arthur Bloomfield in his well-known Patterns of Fluctuation in International Investment before 1914 (1968). No account of the working of the international gold standard can afford to neglect these special features of the pre-1913 period. Five of the seven countries on which the pooled model was testedGermany, France, Sweden, Russia, and Japan-were not Atlanticoriented either in trade or foreign lending, so that only two-Great Britain and the United States-reflected the special characteristics of the pre-1913 international economy, namely, the center-periphery interaction. However, the way the model was specified made it impossible to pick up this interaction. The pooled results have drowned the peculiarity of Britain's interaction with the United States in general averages for the seven countries. The working of the international gold standard between the "regions" of the nineteenth-century Atlantic economy has a close resemblance to the working of the internal gold standard between the regions of the United States. Under this internal standard the ease of adjustment was greatly facilitated by the existence of two fundamental conditions-free interregional migration of labor and the transfer of Treasury funds into weak regions. Among the most important reasons why the international gold standard worked fairly smoothly were, first, the high degree of international mobility of labor and, second, the fact that Britain, the dominant creditor, with a high propensity to import, was always putting money back into international circulation, either through a substantial upswing in imports or through a substantial upswing in foreign lending. Thomas also commented on the McCloskey and Zecher paper. Proceeding from the analogy between the international and the national

168

Donald N. McCloskey and J. Richard Zecher

economy, McCloskey and Zecher emphasize what they regard as the necessary corollaries-unified prices of products, assets, and labor across national boundaries. Hence the monetary theory, which is an equilibrium model; but the theory overlooks some awkward facts. One has to recognize the nature of the growth process in the pre-1913 Atlantic economy, particularly the inverse relation between investment upsurges in the center country and in the overseas country of new settlement. Export capacity in a given phase of the long swing was a function of the infrastructure investment in that country in the previous swing. There was a long-run symbiotic relationship, but it necessarily entailed opposite movements at the center and the periphery, and serious disequilibria when the peaks of the long swing were reached and the Bank of England had to protect its reserve. This process would occur particularly when under-effected transfer was experienced. The U.S. trade balance determined the gold flow, and the gold flow determined the rate of growth of the money supply. There is no basis for the notion that investment upswings, by generating excess demands, attracted net capital inflows that more than offset the unfavorable trade balance, thereby inducing gold inflows. Gold inflow, and as a consequence the money supply, rose most rapidly in the phases of the long swing when U.S. exports were surging upwards and infrastructure expenditure and imports were in a downswing. Simultaneously Britain was having an upswing in home investment, her exports as a proportion of imports were falling, and there was an external flow of gold from the Bank of England. When it was the turn of the United States to have its upswing in investment, her trade balance deteriorated, gold flowed out, and the rate of growth of the money stock fell. See what happened during the 1890s. Between June 1892 and June 1896 there was an absolute fall in the U.S. money stock, the first such decline since the 1870s, whereas the Bank of England's reserve increased spectacularly from £15 million to no less than £49 million. When the United States was struggling desperately to stay on the gold standard, Britain was enjoying such a surfeit of liquidity that the market rate of discount was below 1 percent. In the second half of the 1890s the reverse happened as a result of the massive upsurge in United States exports in relation to imports coinciding with the opposite in Britain. The Bank of England reserve as a proportion of liabilities fell almost as fast as it had risen, while the money stock of the United States went up by 52 percent. The Old Lady of Threadneedle Street was not managing the international gold standard: she was just minding her own business and doing it on an investment in gold stocks inexcusably small in relation to her responsibilities. Her status as central bank of the center country endowed her with clout. McCloskey and Zecher are scornful of Keynes's description of the Old Lady as "conductor of the international orchestra"; they

169

The Success of Purchasing-Power Parity

regard her as "no more than the second violinist, not to say the triangle player, in the world orchestra" ("How the Gold Standard Worked, 1880-1913," in The Monetary Approach to the Balance of Payments, ed. J. Frenkel and H. G. Johnson. [London: Allen and Unwin, 1976], pp. 358-59). As a superb understatement that must constitute something of a record. EICHENGREEN elaborated upon one of Brinley Thomas's points. The purpose of pooling national time series and of attempting international comparisons is to extract as much information as possible from historical data. This approach is predicated upon the assumption that the structural relationships under consideration are identical across countries. In the case of monetary relations under the classical gold standard, there is considerable historical evidence of the existence of important structural asymmetries that would call into question the validity of this assumption. Eichengreen drew attention to the work of Triffin and others that pointed to the unique degree of market power exercised by the Bank of England under the classical gold standard and to asymmetries in the impact that changes in the monetary conditions in different countries had on the balance of payments of the countries participating in the system. For example, changes in monetary conditions in Britain appear to have had a much more powerful impact on short-term capital flows than did comparable changes in monetary conditions abroad. As Triffin suggests, the Bank of England had an ability to influence international gold flows unrivaled by other central banks. Asymmetries of this sort are not taken into account in Gregory's analysis. ABRAMOVITZ pointed out that there is no obvious connection between long swings, such as fifteen-to-twenty-year Kuznets cycles, and international gold movements. This is not surprising, since many factors can substitute for the actual movements of gold: the growth of high-powered money from domestic sources, changes in high-powered money multipliers, and changes in the income velocity of money. What, then, produced the long swings in high-powered money that parelleled so closely the long swings in the growth rate of real output? Abramovitz suggested focusing on the growth rate of the sum of exports and capital imports. That sum traces long swings that parallel the long swings in the growth rate of nominal and real incomes in the United States. This line of inquiry suggests a further question: are exogenous changes in the growth rate of the sum of exports plus capital imports driving the growth rate of the money supply and of nominal income? Or is the growth rate of nominal income determined by independent changes in real output and world prices to which the money supply of a country must adjust? The latter is recognizably the view of the monetary approach to the balance of payments, which is arguably the right view when one has in

170

Donald N. McCloskey and J. Richard Zecher

mind very long periods of time. In shorter periods, however, and even in the long swings that run across ordinary business cycles, the answer is much less clear. In the long swings of income in the United States, there appears to have been. an interaction between the sum of commodity exports and capital imports which together constitute the positive elements underlying the balance of payments and the real-income changes with which they are associated. Independent movements of commodity exports or capital imports stimulated change in nominal income and in real income as well. The changes in real income, in turn, generated further increases in capital imports. Within limits, there was a selfsustaining cumulative process. Abramovitz noted that the central point, raised by Brinley Thomas, concerns the inverse pattern of long swings in Britain and the United States. Abramovitz maintained that there were no inverse swings in real output or in nominal income in the aggregate in the United States and United Kingdom. In point of fact, the long upswings in the United States were matched by surges of capital export from Britain and capital import into the United States. Similarly, surges of exports from Britain were matched by surges of imports into the United States, and declines in home investment in Britain were matched by rises in home investment on the other side of the Atlantic. In Britain fluctuations in commodity and capital export offset one another and left the British economy growing smoothly over the business cycle, in contrast to the United States where capital imports and exports were not matched so closely bf surges and declines in commodity imports and exports. These observations lead to a further question: Why did the long swings come to an end? Why did they result in the United States in serious depressions that culminated each of these episodes, while in Great Britain there were no comparable breakdowns that would have produced, had they occurred, the appearance of long swings in aggregate output as in the United States? One reason was the difference between the banking systems of the two countries. At intervals, the United States suffered banking and financial panics far more violent than those to which Britain was subject. A more severe impact on money and real income was felt in the United States. The two countries, however, differed in other respects as well, and the matter deserves a lot more study.

3

The Bank of England and the Rules of the Game under the International Gold Standard: New Evidence John Dutton

3.1

Introduction

This paper is an investigation of the Bank of England's actions under the gold standard in the decades immediately preceding World War I. In particular, I use monthly data on Bank reserves, domestic activity, price changes, and gold inflows from the late 1880s to mid-1914 to determine which variables the Bank reacted to. The years from 1870 or 1880 until 1914 are frequently regarded as a halcyon period in international monetary relations. Several financial crises occurred (in 189O-the Baring crisis-and in 1893 and 1907), but throughout the period the major central banks were able to maintain gold convertibility of their currencies. Had war not intervened in 1914, the system might have operated reasonably well for decades longer. At the center of the system was London (Lindert 1969 describes its importance), and at the center of the London financial community was the Bank of England. Actually, in 1914 the Bank was still a private profit-making institution; however, it had for a century and more increasingly taken on quasi-official functions. After Walter Bagehot's publication of Lombard Street in 1873, the Bank's position as central bank had become widely recognized and accepted. Because of London's key position in the international financial community, the Bank at that time was clearly the foremost central bank of the world-a role that has special interest for anyone studying the international financial system. John Dutton is assistant professor of economics at North Carolina State University, Raleigh, North Carolina. The author wishes to thank Michael David Bordo, Paul McGouldrick, John Pippenger, Anna Schwartz, Richard Sylla, and the numbers of a North Carolina State University International Trade Workshop for helpful comments and David Dickey for econometrics advice. They are absolved of all responsibility for the final result.

173

174

John Dutton

A salient feature of that international financial system is its apparent stability. Bloomfield (1959) points out that exchange rates among countries on gold were essentially unchanging; devaluations were a rarity. Implicitly, balance-of-payments problems were not the disrupting influence they came to be under the Bretton Woods system. Admittedly some economies may have experienced greater instability with respect to certain measures of prices and output than after World War II (see Bordo 1981 for statistics for the United States and the United Kingdom); nonetheless, the international aspects of the system were stable. Analysts of the pre-1914 era have stressed two explanations of international stability. In the short run, if trade imbalances developed (caused perhaps by exogenous shocks), capital would flow in the opposite direction. This capital counterflow would take place because of the effects of gold flows on money supplies and interest rates. In the United Kingdom, according to the descriptive literature (e.g., Clapham 1970; Sayers 1976), such interest-rate effects were produced by central-bank authorities. Of course, changed flows of capital in response to interest-rate changes were primarily stock-adjustment processes and therefore inherently temporary. For this reason capital flows are not a satisfactory explanation of long-term flow adjustment. Once any capital stock adjustment had taken place and capital stopped moving, the outward gold flows caused by the original trade imbalance would reappear. A second explanation of long-term adjustment was the Hume pricespecie-flow mechanism. According to that explanation, internationalpayments imbalances resulted in money-supply changes in the countries involved. These in turn caused prices to change (upward in the country receiving gold, downward in the country losing gold), and these price changes altered the international flow of goods in such a way as to eliminate the initial trade imbalance. Although it does not appear in analyses of international adjustment before World War I, after the war the phrase "rules of the gold standard game" was applied to a prescription for a specific central-bank role in the adjustment process. According to the "rules," central banks had the important role of facilitating international-payments adjustment, either by reinforcing the effects of payments imbalances on the domestic economy so as to speed the adjustment process, or at least by not hindering those effects. According to one definition of the rules articulated by Bloomfield (1959 pp. 47-48): A discount rate and credit policy ... was supposed . . . to have the effect of increasing central·bank holdings of domestic income-earning assets when holdings of external reserves rose, and of reducing domestic assets when reserves fell. In this way, the effect of changes in central bank reserve holdings on the domestic credit base would be magnified by central bank action.

175

The Bank of England and the Rules of the Game

Bloomfield mentions that Nurkse (1944) applied this definition to the period between the world wars and found that during that period central banks in general did not follow this form of the rules. Bloomfield himself applied the definition to annual data in the 1880-1914 period. He computed the number of times international and domestic interest-earning assets of each central bank moved together, indicating rules compliance, and moved in opposite directions, indicating rules violation. He found for all the observations covered a compliance rate of 34 percent and a violation rate of 60 percent, with the remainder being cases of no change. The comparable figures for the Bank of England in Bloomfield's study were 47 percent and 50 percent. Pippenger's (1974) study uses a definition of the rules somewhat like Bloomfield's. He regresses Bank of England monetary liabilities on the Bank's stock of gold during the 1890-1908 period and finds a positive and significant relationship when data for periods of a quarter and longer are used, though not when the data are monthly. Pippenger's conclusion is that the Bank responded to gold flows by changing the money supply in the medium to long run but not in the short run. Such long-run behavior, if the response were sufficient, would be in keeping with the rules. Pippenger's coefficient for annual data indicates a bit less than a one-forone transmission. Bloomfield notes in his description of the rules that the definition quoted above is by no means the only one. Another possible definition, which he does not explicitly mention, is that central banks refrain from countercyclical domestic policy. Under the rules, central banks were obligated to maintain the convertibility of their currencies. Given the limited nature of their policy tools, that obligation would have left them little capacity to pursue domestic stabilization. In addition, pursuit of the stabilization goal might well have interrupted the process of long-term adjustment to eliminate payments imbalances. Qualitative discussions provide mixed answers to the question of the incidence of domestic countercyclical policy. Bloomfield (1959, p. 24) himself says: "the view . . . of central banking policy as a means of facilitating the achievement and maintenance of reasonable stability in the level of economic activity and of prices was scarcely thought about before 1914, and certainly not accepted, as a formal objective of policy," but then he goes on to note increasing awareness and sensitivity on the part of central banks to the domestic effects of their actions. Sayers (1976) also notes some sensitivity of the Bank of England to the effects of its policy on business conditions. Apparently, critics of the effects of Bank policy on the domestic economy made their views known (see Dornbusch and Frenkel, this volume). Mints (1945, pp. 188-89) cites several writers of the pre-1914 period who decried those effects. Palgrave, writing in 1903, complains: "Great instability in the rate of dis-

176

John Dutton

count is a very prejudicial thing to the interests of commerce," and favors minimizing the transmission of disturbances from abroad to the domestic economy ([1903] 1968, p. viii). Whether the Bank actually pursued countercyclical policies or not, certainly the possibility of doing so must have been evident to its directors. One problem in testing for Bank countercyclical policy is a possible strong negative relationship between reserves and domestic activity. Ford (1962, p. 21) points out that at cyclical peaks high import demand might have coincided with high capital exports, with consequent demands on central-bank reserves. If low reserves led to high Bank discount rates, then the Bank might appear to be following a countercyclical policy without actually doing so. Ford apparently believes the Bank-ratedomestic-activity relationship to be such an indirect one, for he states, "the Bank did not pursue a conscious contra-cyclical policy in its use of Bank Rate" (1962, p. 34). Goodhart (1972, p. 199) presents evidence to support the negative reserve-activity relationship. A regression he reports of the balance-of-trade surplus on an activity variable for the 1890s and early 1900s yielded a significant negative coefficient. A possible test for separate effects of reserves, domestic activity, and inflation on Bank policy is to include all three in the equations estimated. Dutton (1978) reports such a regression on Bank rate using annual data for 1862-1913. Despite the presence of reserves as a control in the equation, domestic activity, as measured by the rate of unemployment, had a significant positive effect on Bank rate. Another problem in the interpretation of the effects of domestic activity involves the Bank's profits. If high activity led to high market interest rates, the Bank might have raised its own discount rate as a way of enhancing its profits. Such profit-motivated behavior might appear to be countercyclical policy. It would be hard to discriminate between profitmotivated and purely countercyclical Bank-rate policy. Most of the descriptive literature takes for granted that profit was secondary to other Bank policies. In this paper I assume that it was. Actually, if Bank policy was countercyclical, whether for profitability reasons or otherwise, the Bank would have broken the rules. The need to discriminate between profitability and stabilization as motives may not be important in testing for violations of the rules. In the present study, I use monthly data available for several series from the late 1880s to 1914 to find out how well Bank actions fit the rules of the game. In doing so, I test for both countercyclical policy on the part of the Bank (ignoring profit motives) and for relationships between Bank holdings of non-interest-bearing reserves and interest-bearing domestic assets. I expect to find, if the Bank followed the rules, that it did not react to domestic activity and inflation in its conduct of policy. I also expect to find, if the Bank followed the rules, that reserve changes and Bank

177

The Bank of England and the Rules of the Game

interest-bearing assets moved together, indicating that the Bank did not sterilize the effects of reserve changes on the money supply. 3.2 3.2.1

Specific Variables Considered Policy Tools

The Bank of England's chief instrument of monetary control was its discount rate, known as "Bank rate." According to Sayers (1976, p. 23), "the accepted doctrine both inside and outside the Bank was that its most important action was the fixing of Bank Rate." Of course, for Bank rate to have an effect, it had to influence market rates of interest. Accordingly, the Bank had to ensure that capital markets in London were dependent on it for at least part of their funds. In fact, many commercial concerns borrowed regularly from the Bank at Bank rate, or at rates closely related to Bank rate. Consequently, during much of the period, that rate constituted the opportunity cost of funds at the margin for market participants. A great deal of the energies of those governing the Bank was expended in making sure at certain key times that Bank rate did constitute that opportunity cost. "Making Bank rate effective" generally involved the use of other more limited tools of the Bank to back up Bank rate. Altering the Bank's holdings of interest-earning assets-the tool used frequently to "make Bank rate effective"-ean be thought of as an early form of open-market operations. To force market participants to borrow at Bank rate, the Bank would in one way or another gain command of additional funds in the market. It did so at times by selling securities, at other times by borrowing against its securities holdings, and at still other times by selling consols spot and buying them forward. All of these procedures removed funds from the market and eventually forced customers to borrow from the Bank. How common these operations were is difficult to say. Clapham notes their use as far back as the midnineteenth century and earlier; he goes on to say that "in the seventeen years from 1873 to 1890 there are only four in which no market borrowing is done, and in several the borrowings are repeated and complex" ([1944] 1970, pp. 295-98). Sayers (1976, pp. 37-43), on the other hand, stresses the relative infrequency of these operations before 1890, though he describes them as used with increasing frequency after that time (see also Morgan 1965 for a description of the practices involved). A third instrument, in addition to Bank rate and altering the Bank's portfolio, was use of the so-called gold devices. The Bank was required by law to purchase gold with notes and to redeem notes with gold sovereigns on certain set terms of exchange. However, the Bank was free to alter those terms somewhat in favor of persons bringing gold for

178

John Dutton

exchange. It was also free to force its customers to follow the law's conditions more or less exactly in presenting or seeking gold. Thus the Bank could draw more gold by making it more advantageous for customers to bring gold and more costly for them to remove it from the Bank. These practices-lumped together under the name "gold devices"-had the effect of increasing the spread between the gold points, the rates of exchange at which gold could profitably be imported or exported. Sayers (1953, 1976) describes these gold devices in some detail. Their use was in a sense a violation of the rules, since the devices interfered with the free convertibility of gold at clearly specified rates of exchange. They were usually used to retain or attract gold without resorting to extreme Bankrate changes that would otherwise be necessary. In this paper I ignore the gold devices largely because I have no data series on their use. The descriptive literature portrays them as a relatively minor part of Bank policy. The policy tool I focus on in the empirical work is Bank rate, since all descriptions feature it as the principal embodiment of policy. Bank-rate information is available in weekly form from several sources, including, for the period 1888 to 1909, a very useful U.S. National Monetary Commission statistical volume (1910). Following Goodhart (1972), I use for each month the Bank rate in effect at the beginning of the month. I also use a form of the second policy tool, the Bank's prototype form of open-market operations. The U.S. National Monetary Commission volume contains information on Bank holdings of government securities and other domestic interest-bearing assets. As a securities-holdings variable, I use the reported holdings of all such assets at the beginning of each month. 3.2.2 Target Variables By all accounts, the Bank of England devoted a large share of its attention to ensuring the adequacy of the liquid reserves in its Banking Department. Those reserves formed the base of the credit structure in the United Kingdom. As a result of its lending, the Banking Department's liabilities were generally two or three times its reserves. Some of those liabilities were deposits of the government or of customers of the sort a commercial bank might have. Others, however, were "bankers' balances," and as such constituted a large part of the reserves of the commercial-banking system. Thus any unusual demand for money anywhere in the U.K. economy or from abroad was likely to end up quickly as a demand on the Bank. Only the Banking Department's reserves were available in normal times to meet that demand. Generally the gold reserves in the other half of the Bank, the Issue Department, were considerably larger than Banking Department reserves; however, that

179

The Bank of England and the Rules of the Game

gold was by law (the Bank Charter Act of 1844) only to be given up in return for Bank of England notes. Gold held for backing notes not in the hands of the Bank was unavailable for meeting a money demand. Of course that gold was physically within the country and therefore available for emergency use in meeting a demand from abroad. In fact, the 1844 law governing Bank behavior with respect to gold was suspended by a special letter from the cabinet on rare crises occasions. (See Dornbusch and Frenkel, this volu~e, on the 1847 crises in which domestic liquidity demands supervened on foreign gold demands.) The Bank and government, however, were highly reluctant to suspend the law; for essentially all purposes, then, the Bank's available reserve was that of the Banking Department. For the purposes of this paper, therefore, I use Banking Department reserves as a variable indicating the Bank's ability to meet its liabilities, including its obligation to provide gold in return for its bank notes. I also use another widely watched indicator of reserve position, the "proportion. " The proportion was the ratio of reserves to liabilities of the Banking Department. It was of course closely related to the level of reserves. Either of these variables, if they act as expected, should demonstrate a significant inverse relationship to Bank rate in the empirical work. As reserves and/or the proportion dropped, Bank rate should have risen. In some equations a reserve-change variable will be used with Bank securities-holdings changes. In that case, if the Bank adapted its interestbearing asset-holdings to reinforce the effects of changes in reserve holdings, in accordance with the rules, then the measured relationship should be positive. In the description of the rules above, I mentioned the rules interpretation that would prohibit a countercyclical reaction of the Bank to domestic activity and prices. The empirical work described below tests for such reactions. On an annual basis, estimates of gross domestic product are available for years quite a bit earlier than those covered in this study. However, on a monthly basis they do not exist. I therefore employ proxies for domestic activity. Goodhart (1972) collected and used railway freight receipts, which were available by month back to the early 1890s. I have borrowed that series for this work. As an alternative I also use an unemployment-rate series originally collected from trade unions. Neither series is a perfect proxy for domestic activity. The railway-receipt series, of course, is unduly representative of certain sectors of the economy. Since it is a revenue series, it reflects not only quantity movements but price movements. The unemployment series applies only to a subset of workers, those in trade unions. In addition that subset altered as the number of trade unions reporting increased. Despite these limitations I employ the two series to test for Bank reactions. If the Bank followed the rules, then Bank rate should not respond to them. The rate of inflation, like domestic activity, should not have affected

180

John Dutton

Bank rate if the Bank followed the rules. I have computed a monthly rate of inflation based on a crude wholesale price index collected by Augustus Sauerbeck. Another variable I use, gold inflows, will serve as an indicator of internationally induced money movements. In some ways a gold inflow is like a positive movement in reserves. Therefore, I enter it in some versions of the securities-holdings equation to see if the Bank responded to gold flows rather than to changes in its own reserves. Unfortunately the data on gold flows are poor. Goodhart (1972) presents two monthly series, one collected by the Bank and the other by the English Board of Trade, both purportedly measuring the same thing. The two series are quite different, indicating that one or the other or both are defective. I have chosen to include gold flows, despite the limitations of the data. I use the Bank of England series; presumably the Bank would use that series, if it used any, to govern its behavior. 3.3 Modeling and Estimation The question posed is: What variables did the Bank of England react to? To answer it I treat the Bank's policy tools as dependent variables and regress them on several variables to which the Bank might have responded. Of course, any response is founded on a belief by the Bank that its tools in turn would affect these variables. The process of feedback of target variables to the actions of a policy authority is discussed by Sargent (1979, chap. 15). As he notes, proponents of the recently developed rational-expectations hypothesis tend to discount the effectiveness of systematically applied policy tools in influencing economic events. The argument is that individuals and firms, if they come fully to anticipate policy reactions, will take compensatory actions that will obviate the effects of policy. If these analysts are right, then feedback processes, being very regular and predictable, will be ineffectual in altering economic events. As Sargent points out, however, a sufficient condition for feedback policy responses to work is greater and more timely information availability to the policy authority. If such a condition holds, then individual agents will be unable fully to anticipate and compensate for policy actions; those policy actions as a result will have real effects on the economy. In this paper, I assume that the Bank of England at least believed its policy actions, including those determined by feedback rules, to be effective. The appearance of any significant coefficients in the feedback equation would bear out that assumption. For purposes of testing, the Bank is assumed to react each month to the values of target variables for that month, or more strictly speaking, to forecasts of target variables for that month. The feedback process is

181

The Bank of England and the Rules of the Game

modeled in linear form, i.e., the policy-tool variable is treated as a linear function of forecasts of the target variables. The forecasts, rather than the actual variables themselves, are used for two reasons. First, it is unlikely that the Bank would have full information about the most recent values of its targets at the time of undertaking policy action. Second, the forecasts serve to eliminate problems of simultaneous-equation bias. Two-way causality is implicit in the formulation of the model; policy actions influence targets, and targets, through the feedback mechanism, influence policy actions. The second avenue of causality is the one I test for. One way to reduce the chances of detecting the first direction of causality and mistaking it for the second is to use regressors that are predetermined at the time policy action is taken. In this case, I use forecasts of the target values based on information from previous months. The ideal forecast series, of course, would be those actually used by Bank officials. However, such series were likely never made completely explicit, much less written down. Proxies for the forecast series must be sought. One approach to forecasting used a great deal of late is that of Box and Jenkins (1976). To predict a variable, they use the past history of the variable, including both past values of the variable itself and past differences between the variable and its predicted value. The equations fitted for forecasting are of the form: Zt =

1

Zt-l

- a2 at -2 -

+ 2 Z t-2 + ... + p Zt-p + at ... -

a1 a t -l

aqa t - q ,

where Zt is the variable at time t (or in some cases changes in the variable), Zt _ i is the variable at time t - i, at is the prediction error at time t, at _ i is the prediction error at time t - i, and i and ai are parameters. The Zt-i terms form the autoregressive part of the process and the at-i terms are denominated the moving average part. Usually p and q are low numbers ranging from zero to two. If differencing is used, the process is said to be integrated. The acronym ARIMA, for "autoregressive integrated moving average" is commonly used with these models. In much of the empirical work that follows, I use ARIMA models to obtain forecasts for the target variables. Several objections to this method of generating forecasts may be raised. First, the Bank likely used information other than a variable's past history in generating forecasts for that variable. It probably used other variables, for example. Second, even if the Bank used only a variable's past history in generating its forecast, it might have used a different forecasting process than the equation estimated here. The forecasting equation estimated for a variable in this paper is based on data covering the whole period under consideration. At any given time, of course, the

182

John Dutton

Bank could have used only data preceding that time. If the pattern a variable followed were stable over the whole period covered, as well as over a suitably long preceding period, then the Bank in principle could have used the forecasting equation used here. If, on the other hand, that pattern were changing, then the equation I estimate here based on data up to 1914 could not, even in principle, have been available to the Bank before 1914. These considerations all imply that the forecast series are at best quite imperfect proxies for the Bank of England's forecasts. The problem can be viewed as one of errors in the variables. In such a case, the observed coefficients will be biased toward zero, making detection of any existing feedback less likely. Earlier I mentioned the use of predetermined forecasts as a way of eliminating simultaneous-equation bias. Unfortunately, use of those forecasts does not eliminate another "back-door" route by which causality running in the "wrong" direction could be picked up by the equations. If the policy-tool variable is autocorrelated, i.e., if its value at time t is correlated with its value at time t - i, then the possibility exists that any results from regressing the policy tool at time t directly on the forecasts could simply reflect relationships of both with the policy tool at time t - i. This problem, however, would not exist if the dependent variable at time t were not correlated with its own previous values. One way to meet this condition is to subtract from the dependent variable the part of it that can be predicted from its own past. For this purpose, I use an ARIMA model fitted to each dependent variable to generate predicted values, which are then subtracted from the actual values. The residuals can be thought of as the innovations that occurred each period in the policy variable. The innovations are uncorrelated with each other and therefore will not evidence any spurious back-door relationships with the predetermined target forecasts. This method should yield results reflecting causality going only in the direction from target variables to Bank policy. The equations fitted with this techinque are of the form at = X; ~ + Et , where at is the time t residual in the ARIMA equation for the policy tool, Xt is a vector of forecasts of the target variables, ~ is a vector of coefficients, and Et is the error at time t. Despite alterations of the Z and X series, the errors in this equation may still exhibit autocorrelation. If they do, the standard errors of the coefficients may be biased. To remove that autocorrelation and resulting bias, a generalized least-squares procedure may be used (see Theil 1971, p. 253, for a general description of the procedure) . The monthly data for the empirical work come from several sources, all detailed in a data appendix. Most of the series coyer the period from the late 1880s to mid-1914; the specific period for each series is listed in table 3.1. Most of the series exhibited marked seasonality. Some also exhibited distinctive time trends. To remove those influences the data were regres-

-

1

+ = in; - = out.

9.65 13.98 15.22 13.11 12.83 13.82 14.07 12.99 14.92 12.13 10.82 11.77 .039/mo.

JAN. 1888JUNE 1914

JAN. 1888JUNE 1914

4.15 3.54 3.31 3.22 3.07 3.06 2.81 2.92 3.13 3.58 4.12 4.12

Bank of England Reserves (million £)

Bank Rate (percent per annum)

Monthly Means and Trends

Sources: See Appendix.

January February March April May June July August September October November December Trend

Period

Table 3.1

-

37.3 50.0 48.3 44.3 47.0 48.6 44.5 48.0 51.9 45.6 45.9 48.1

4.96 4.54 4.48 4.01 3.99 3.94 4.07 4.40 4.51 4.57 4.39 4.90

442.5 509.3 520.3 471.1 492.9 442.1 445.7 442.4 523.8 565.8 558.4 468.9 1.73/mo. -

JAN. 1888JUNE 1914

JAN. 1887JUNE 1914

JAN. 1893JUNE 1914

Bank of England Proportion (percent)

Unemployment Rate (percent)

U.K. Railway Receipts (thousand £)

52,721 44,399 47,779 48,560 45,905 45,080 49,441 45,212 43,671 46,775 44,848 43,961 60.4/mo.

JAN. 1888DEC. 1909

Bank of England Securities (thousand £)

0.11 0.46 0.93 0.83 1.36 1.46 0.70 1.02 -1.28 -2.27 0.08 -0.03

JULY 1891JUNE 1914

Gold Flows l (million £)

184

John Dutton

sed on monthly dummy variables and trend terms. The residuals from those equations were used in subsequent steps. These residuals should be trendless and free of seasonal effects. Table 3.1 also presents the monthly means and estimated-trend coefficients. These statistics, though a secondary feature of this paper, are of interest in themselves. Note the unusually low average reserves and high average interest-earning assets of the Bank during January-the most usual month for issuing financial reports. Those averages apparently reflect the "window-dressing" in which English banks of the period engaged. The practice involved boosting gold and Bank of England note holdings just before public statements of financial condition were to be issued. The higher reserve holdings made the financial conditions of the banks appear more favorable to depositors and others. Table 3.2 presents the ARIMA equations used to generate the forecasts used as independent variables and the innovations used as dependent variables in the final equations. The Z's represent the detrended series with monthly means removed. L is a lag operator; i.e., .80L Zt = .80Zt - 1 and . 16L5 Zt = .16Zt - 5 . The a's are residuals that are assumed to be "white noise." A variable is a white-noise series if the value of the variable at time t is independent of its value in any other period. If an equation in table 3.2 has "captured" all the explanatory power of the variable's history in explaining its value in the present, then the residual series in that equation will be white noise. The Q-statistic probabilities on the right-hand side of the table are meant to answer the question: What is the probability of obtaining these residuals if they are from a white noise series? The Q-statistics are computed using autocorrelation coefficients of the residual series up through lags 6, 12, 18, and 24. The probabilities reported are all well above .1, indicating that we cannot reject the hypothesis that they are white noise. This fact demonstrates that the equations of the table are good fits and do capture most of the explanatory power present in each variable's history. One is led to ask why the particular lag structures of table 3.2 show up in the data. The low-order lags are what one might expect, and they show up fairly uniformly. The twelfth-order lags are also to be expected; they represent a bit of seasonality left in the data despite removal of monthly means. The other scattered lag terms of the Bank-rate and -reserves equations are harder to explain. I have been unable to divine any economic explanation of their presence but have simply accepted them. Before proceeding to report the results of the outlined procedures, I note an additional, similar mode of estimating the feedback equations described above. The use of the forecasting equations of table 3.2 effectively makes the policy variables into functions of lags of the target variables, with specific lag structures imposed from prior information. It seems reasonable also to estimate those equations without imposing lag

5

6

9

Zt(l- .80L - .16L + .16L - .09L ) = at (21.76) (2.74) (2.64) (2.28) Zt(l- .85L - .19L5 + .11L7 + .14Lll - . 16L12 ) = at (25.19) (4.06) (2.32) (2.39) (2.83) Zt(1-1.21L + .24L2 ) = at (1- .64L) (5.46) (8.78) (1.94) Zt(l- .91L)(1- .24L12 ) = at(l- .69L) (3.64) (18.52) (8.34) Zt(l- .72L - .17L2 - .07L3 ) = at (13.08) (2.51) (1.30) Zt(l- .24L) = at (4.70) Zt(l- L)(l- .20L12 ) = at(l- .57L) (11.01) (3.28) Zt = at(l + .21L - .30L2 ) (3.60) (5.18)

Fitted Equation

a

Forecasting Equations and Filters

.142 .343 .675 .243

.214 .671 .436 .154 .487 .416

.132 .232 .628 .503 .292 .642

.465

.233

.945

.991

.956

.727 .

.512

.424

6

Q = Statisticb Probabilities for Lags 12 18

.584

.394

.234

.849

.165

.330

.989

.732

24

.16Zt - 6 . Beginning year of data indicated; end was June 1914 except where indicated. Asymptotic t-statistics are in parentheses. bryhe probabilities of obtaining the estimated a's if they are from a white-noise series.

az indicates the variables before filtering; a indicates the variables after filtering and ideally is white noise. L is a lag operator; for example, .16L6 Zt =

Bank rate (1888) Reserves (1888) Proportion (1888) Railway receipts (1893) Unemployment (1887) Inflation rate (1885) Securities holdings (1888-1909) Gold inflow (1891)

Table 3.2

186

John Dutton

structures, i.e., to enter the lagged target variables themselves (with trend and monthly means removed) rather than the functions of those variables reported in table 3.2. Results of such an estimate procedure are reported below. It follows in some respects a procedure outlined in Nerlove, Grether, and Carvalho (1979, chap. 11). 3.4 Empirical Results

Table 3.3 reports coefficients for forecast variables regressed on the transformed Bank-rate variable. As expected from the descriptive literature on Bank policy, both the reserve level and the proportion showed up as strong influences on Bank rate. The relationship in each case was consistently negative, indicating that Bank rate was raised in response to low values of those variables and vice versa. Apparently the reserve level and the proportion are highly correlated; as a result, when both appear together in an equation, the effect of neither can be measured with precision. When they appear separately, however, each evidences statistically significant coefficients. Domestic economic activity was entered in some equations in the form of railway freight receipts and in others in the form of the unemployment rate. The former had consistently positive coefficients and the latter consistently negative ones. Both types indicate that Bank rate rose when activity was high and dropped when it was low. The level of statistical significance of the coefficients varied somewhat. The coefficient of unemployment was always high relative to its standard error, but the coefficient of railway receipts dropped when reserves or the proportion were included in the equation. Possibly railway receipts were more closely related to imports than was unemployment; if that were the case, it would be easier to estimate precisely a separate effect of unemployment than it would to estimate an effect for activity as represented by railway receipts. The rate of inflation also had significant coefficients in several of the equations. Its positive sign apparently indicates that Bank rate rose in response to high 'predicted rates of inflation and vice versa. Table 3.4 reports results when the change in Bank holdings of domestic interest-bearing assets is the dependent variable. The relationship between that variable and changes in reserves is of particular interest because of the emphasis Bloomfield (1959) and Nurkse (1944) placed on it. The negative sign on reserve changes (measured here as predicted reserves less actual reserves in the previous month, both adjusted for trend and seasonality) indicates that Bank securities holdings generally increased when reserves were predicted to fall, and vice versa. An increase in those security holdings, other things equal, would lead to an increase in Bank of England notes and gold held by commercial banks and the public, and to an increase in the domestic money supply. The

187

The Bank of England and the Rules of the Game

increase would tend to counteract any decrease caused by the withdrawal of reserves to the foreign sector. Thus the Bank at least in part seems to have sterilized the effects of international money flows on the domestic supply of money. As an alternative approach to the Bank's reactions, I estimate similar equations using gold inflows rather than reserve changes. The results, reported in equations 6 through 9, are similar to those described above, though the significance level of the coefficients is substantially lower. Gold inflows, which would normally suggest money-supply increases, led the Bank to respond by selling off securities, thus countering some or all of the change in the supply of money. In none of the equations of table 3.4 was either of the activity variables significant. The rate of price change, however, appears to be nearly significant. Its positive coefficient indicates that expected increases in prices led to higher Bank holdings of interest-earning assets. The response, if correctly measured, is a procyclical one; that is, higher inflation appears to cause the Bank to expand the British supply of money. This procyclical response contrasts with the apparently countercyclical response reported in table 3.3. The reserve level was also included in several equations. It had no significant effect in the reserve-change equations, though its coefficients were always positive. When gold inflows were substituted for reserve changes, the reserve level almost assumed statistical significance. Its positive coefficient indicates that the higher the reserves, the greater the inflow of interest-bearing securities into the Bank's portfolio, other things equal. Table 3.5 contains results of stepwise regressions paralleling the equations of table 3.3. These regressions are estimated by allowing independent variables to enter the equations in order of statistical significance. Only variables significant at the 0.15 level were included in the equations of table 3.5. The independent variables in these equations are lagged values of the detrended, deseasonalized series. Entering lagged values directly avoids imposing a lag structure of the sort imposed in using the ARIMA forecasting equations. The dependent variables are the same "innovations" in Bank rate used for table 3.3. Because of the loose specification of the equations, table 3.5 is harder to interpret than table 3.3. However, it appears that reserves and/or the proportion had the most significant overall effects on Bank rate. In each case, the sum of the effects is negative. The fact that much of the effect comes from lags of low order lends credibility to the results. The other variable that seems significant is inflation. Its overall effect is positive, as it was also in the table 3.3 equations. Neither of the activity variables appears to be important, though lags 2 and 12 of the railway-receipts variable enter equation 3. Possibly the specification of the equation is too loose for domestic activity to appear significant in the presence of re-

Nb

258

258

306

306

258

258

318

318

306

1. (1893)

2. (1893)

3. (1889)

4. (1889)

5. (1893)

6. (1893)

7. (1888)

8. (1888)

9. (1889)

-0.75 (0.64)

-1.47 (2.76)

-1.54 (2.82)

-1.72 (2.49)

-2.07 (2.74)

Reserves x 10- 2

-1.18 (1.38)

-0.99 (2.16)

-1.15 (2.29)

-1.57 (2.98)

-1.80 (3.20)

Proportion X 10- 2

Bank-rate Equations, 1888-1914

Equation a

Table 3.3

1.01 (0.62)

0.52 (0.30)

2.05 (1.28)

1.62 (0.95)

Rlwy Rcpts X 10- 3

-2.97 (2.12)

-3.07 (1.98)

-3.19 (2.70)

-3.35 (2.79)

Unemp X 10- 2

1.99 (2.34)

1.27 (1.31)

1.72 (1.92)

1.14 (1.16)

Inflation

Lag 3, - .10(1.76) Lag 12, .12(2.11)

Lag 2, .10(1.80), Lag 3, - .10(1.84) Lag 12, .10(1.80)

Lag 3, - .10(1.87) Lag 12, .10(1.74)1.74)

Lag 2, .11(1.72)

None

Lag 2, .12(2.16) Lag 12, .13(2.32)

Lag 2, .10(1.76) Lag 12, .13(2.29)

None

Lag 3, - .11(1.76)

Autoregressive Termsc

306

306

318

258

318

318

258

318

11. (1889)

12. (1889)

13. (1888)

14. (1893)

15. (1888)

16. (1888)

17. (1893)

18. (1888)

-1.60 (3.08)

-0.42 (0.51)

-1.02 (0.97)

-2.12 (3.28)

-1.55 (1.80)

-0.02 (0.02)

3.33 (2.31)

2.99 (1.92)

1.02 (0.61)

-3.90 (2.98)

-4.15 (2.84)

-2.87 (2.15)

Lag 2, .11(1.98), Lag3, - .09(1.69) Lag 12, .11(1.94)

Lag 2, .11(1.82)

1.71 (1.76) 2.14 (2.52)

Lag 3, - .11(2.07) Lag 12, .11(1.93)

Lag 3, - .10(1.72) Lag 12, .10(1.87)

None

Lag 3, - .11(2.07)

Lag 3, - .11(1.99) Lag 12, .13(2.25)

Lag 2, .12(2.16) Lag 12, .12(2.19)

Lag 2, .11(1.71)

1.92 (2.19)

1.70 (1.89)

1.27 (1.30)

Note: t-statistics are in parentheses. 8Date is beginning year of data; all data extend through June 1914. bN = number of observations. clndicated are lag number of autoregressive error term, estimated autocorrelation coefficient used in data transformation, and t-statistic of parameter.

258

10. (1893)

252

252

204

252

252

222

222

204

222

1. (1889)

2. (1889)

3. (1893)

4. (1889)

5. (1889)

6. (1891)

7. (1891)

8. (1893)

9. (1891)

-6.81 (2.08) -6.70 (2.06)

(l.31)

-6.61 (2.53) -5.53 (1.76) -8.96

Reserve

6.35 (1.95) 5.30 (1.33) 6.48 (1.80)

2.20 (0.61) 0.79 (0.17) 1.86 (0.48) 1.80 (0.47)

FCx 10

-4.93 (1.49) -3.83 (1.15) -4.94 (1.42) -4.52 (1.36)

Gold Inflow 2 X 10

FC

Rlwy Rcpts

-6.03 (0.61)

-3.21 (0.33)

FC

1.22 (0.14)

-3.11 (0.40)

Unemp x 10

Note: t-statistics are in parentheses. Fe

= forecast. aBeginning year in parentheses; final year is 1909. bN = number of observations. CFigures are lag number of autoregressive term, parameter estimate, and t-statistic (in parentheses).

Nb

Reserve Forecast - Reserves ( -1) X 102

Securities-Holdings Equations, 1888-1909

Equatibn a

Table 3.4

9.70 (1.62) 9.65 (1.73)

10.08 (1.70) 9.67 (1.91) 9.84 (1.96)

Inflation 3 X 10

None

None

Lag 7, .13(1.96)

Lag 7, .12(1.73)

None

None

None

Lag 7, .11(1.82)

Lag 7, .11(1.78)

Autoregressive Termsc

191

The Bank of England and the Rules of the Game

Table 3.5

Stepwise Regressions on Bank Rate

EQUATION Explanatory Variables Entered Variables a Significant

2

Reserves, Lags 1-5 Rlwy Rcpts, Lags 1-12 Inflation, Lags 1-12 Reserves Lag 1 2 4 5 -

.0865 .0578 .0411 .0210

3

Proportion Lags 1-12 Rlwy Rcpts, Lags 1-12 Inflation, Lags 1-12

Reserves Lag 1 2 4 7 8 -

(6.08) (3.56) (2.57) (1.50)

at 0.15 Proportion Lag 1 4 6 11 -

Level

R2

Reserves Lags 1-12 Proportion Lags 1-12 Rlwy Rcpts, Lags 1-12 Inflation, Lags 1-12

.0218 .0127 .0163 .0157

(3.75) (1.89) (2.35) (2.92)

.0989 .0609 .0289 .0391 .0482

(7.04) (3.83) (2.38) (2.49) (3.28)

Proportion Lag 6 - .0234 (3.90) 10 .0134 (1.87) 11 .0125 (1.82)

Railway Receipts None Significant

Railway Receipts None Significant

Railway Receipts Lag 2 .0012 (1.57) 12 - .0014 (1.93)

Inflation Lag 1 7 11

Inflation Lag 1 .372 (1.75) 5 - .471 (2.17) 7 .417 (1.91) 11 .468 (2.17) 0.19

Inflation Lag 1 .356 (1.76) 5 - .507 (2.40) 6 .431 (2.01) 11 .344 (1.69) 0.32

0.24

.448 (2.19) .591 (2.91) .543 (2.64)

at-statistics in parentheses following coefficients. Intercept included but not shown.

serves and/or the proportion. Of course, if either appeared significant despite the loose structure, then the hypothesis that the Bank engaged in countercyclical policy would have been strengthened. 3.5

Interpretation

The question addressed in the paper is: Did the Bank of England follow the rules of the gold standard game? The answer, insofar as it can be given, is a soft-spoken no. The rules, at least in spirit, required that the Bank not react countercyclically to domestic activity or price changes. The results of the equations of table 3.3 indicate that the Bank did react countercyclically. The reaction is present even controlling for the effects of reserves on Bank policy, suggesting that the domestic-activity and price-change variables are not simply acting as proxies for reserves in the equations. Those variables appear to have had direct effects on Bank policy independent of their indirect effects via reserves. This finding

192

John Dutton

conflicts with the beliefs of several previous analysts (e.g., Ford 1962; Bloomfield 1959) that the Bank did not engage in countercyclical policy of any sort. On the other hand, the finding is supported by much descriptive literature, which frequently mentions the Bank's sensitivity to the effects of its policies on the domestic economy. Table 3.3 suggests that the effects of reserves and the proportion were strongest. They showed up consistently and generally had the highest levels of statistical significance. This suggests that whatever domestic cyclical variables affected Bank behavior, they were outweighed by the need to maintain convertibility of the currency. The conclusion is also supported by the less constrained equations of table 3.5; there reserves and the proportion show up as the most significant variables and with low-order lag structures of the type that seem most reasonable. Thus it is necessary to speak softly of Bank rule-breaking; the results of the Bankrate equations indicate a strong preoccupation with convertibility. Table 3.4 also supports a negative answer to the question: Did the Bank follow the rules? Bank-reserve decreases seem to have led to increases in Bank holdings of interest-earning assets. Instead of amplifying the effects of reserve changes on the money supply, the Bank seems to have sterilized them. The sterilization mayor may not have been intentional. Bloomfield (1959) described Bank policy with respect to its securities holdings as somewhat passive; demands for additional liquidity were met by passively acceding to requests for additional discounting. Whether passive or active in the process, the Bank apparently acted as a buffer between reserve movements and money-supply changes. The rules would demand that it be an amplifier. The Bank at least to some extent seems to have violated the rules, yet the international monetary system was stable. Were the Bank's violations too minor to be important? Or were the rules themselves unimportant for the adjustment process? Must we rely on the unsatisfying attribution of stability to blind luck, the confluence of fortuitous circumstances?

Appendix

Data Sources

Sources for the monthly data used are Goodhart (1972, appendixes VA and VB), the U.S. National Monetary Commission statistics volume on the United Kingdom, Germany, and France (1910), and some National Bureau of Economic Research data sheets kindly supplied me by Anna Schwartz. The latter include price data based on a 1928 Journal of the Royal Statistical Society article ("Wholesale Prices of Commodities") attributed to "the Editor of the 'Statist' ," and unemployment data taken from the British Abstract of Labour Statistics. The unemployment data

193

The Bank of England and the Rules of the Game

are based on returns collected by the Board of Trade and the Ministry of Labour from trade unions paying unemployment benefits. The price data are based on price indexes for forty-five commodities computed by Augustus Sauerbeck; the overall index is a simple arithmetic average of the individual ones. As such, it suffers from obvious weaknesses. However, I have been unable to locate an alternative monthly index. The sources of the data used are listed below. Bank rate, Banking Department reserves, and the proportion: January 1888-June 1891, U.S. National Monetary Commission 1910; July 1891June 1914, Goodhart 1972. Railway receipts ("average weekly gross goods receipts of major British Rlys"): January 1893-June 1914, Goodhart 1972. The March and April 1912 figures were substantially lowered by a coal strike; because the effect on domestic activity was likely much less drastic, I have substituted higher numbers (1182 and 1161) for those two months. Unemployment rates: January 1887-June 1914, NBER data sheets. Price index: January 1885-June 1914, "Wholesale Prices" 1928. Securities holdings: January 1888-December 1909, U.S. National Monetary Commission 1910. This series consists of the sum of government securities in both Bank departments, plus "other securities" in the Banking Department. Gold inflows: July 1891-June 1914, Goodhart 1972. The series used is that attributed to the Bank of England.

References Bagehot, Walter. 1873. Lombard Street: A description of the money market. London: John Murray. Beach, W. Edwards. 1935. British international gold movements and banking policy, 1881-1913. Cambridge: Harvard University Press. Bloomfield, Arthur I. 1959. Monetary policy under the international gold standard, 1880-1914. New York: Federal Reserve Bank of New York. Bordo, Michael David. 1981. The classical gold standard: Some lessons for today. Federal Reserve Bank of St. Louis Review 63: 2-17. Box, George E. P., and Gwilym M. Jenkins. 1976. Time series analysis: Forecasting and control. San Francisco: Holden-Day, Inc. Clapham, Sir John. [1944] 1970. The Bank of England: A history. Vol. 2, 1797-1914. Reprint. Cambridge: Cambridge University Press.

194

John Dutton

Dutton, John. 1978. Effective protection, taxes on foreign investment, and the operation of the gold standard. Ph.D. diss., Duke University. Fetter, F. W. 1965. Development of British monetary orthodoxy, 17971875. Cambridge: Harvard University Press. Ford, A. G. 1962. The gold standard, 1880-1914: Britain and Argentina. Oxford: Clarendon Press. Goodhart, C. A. E. 1972. The business of banking, 1891-1914. London: Weidenfeld. and Nicolson. Great Britain. Ministry of Labour. Statistics Branch. 1894-1926. Abstract of Labour Statistics, annual issues. Lindert, Peter H. 1969. Key currencies and gold, 1900-1913. Princeton Studies in International Finance, no. 24. Princeton: Princeton University Press. McCloskey, Donald N., and J. Richard Zecher. 1976. How the gold standard worked, 1880-1913. In The monetary approach to the balance of payments, ed. J. A. Frenkel and H. G. Johnson. London: George Allen & Unwin. Mints, Lloyd W. 1945. A history of banking theory in Great Britain and the United States. Chicago: University of Chicago Press. Morgan, E. Victor. 1965. The theory and practice of central banking, 1797-1913. London: Frank Cass & Co. Nerlove, Marc, David M. Grether, and Jose L. Carvalho. 1979. Analysis of economic time series: A synthesis. New York: Academic Press. Nurkse, Ragnar. 1944. International currency experience. Princeton: League of Nations. Palgrave, R. H. Inglis. [1903] 1968. Bank rate and the money market in England, France, Germany, Holland, and Belgium, 1844-1900. Reprint. New York: Greenwood Press. Pippenger, John. 1974. Bank of England operations, 1890-1908. Mimeo. Sargent, Thomas J. 1979. Macroeconomic theory. New York: Academic Press. Sayers, R. S. 1976. The Bank ofEngland, 1891-1944. Vol. 1. Cambridge: Cambridge University Press. - - . 1953. The bank in the gold market, 1890-1914. In Papers in English monetary history, ed. T. S. Ashton and R. S. Sayers. Oxford: Clarendon Press. Theil, Henri. 1971. Principles ofeconometrics. New York: John Wiley & Sons. Triffin, Robert. 1964. The,evolution ofthe international monetary system: Historical reappraisal and future perspectives. Princeton Studies in International Finance, no. 12. Princeton: Princeton University Press. U.S. National Monetary Commission. 1910. Statistics for Great Britain,

195

The Bank of England and the Rules of the Game

Germany, and France. Washington, D.C.: Government Printing Office. Wholesale prices of commodities in 1927. 1928. Journal of the Royal Statistical Society 91 (pt. 3): 403.

Comment

Donald E. Moggridge

The phrase "rules of the game" only came into the economist's vocabulary as the interwar gold standard neared its end. So far as I can ascertain, the term was first used by Sir Robert Kindersley, a director of the Bank of England, on February 1930 in the course of his evidence to the Committee on Finance and Industry (United Kingdom 1931b, question 1595). The phrase attracted Keynes's attention and found its way into his "private evidence" to the committee two weeks later (Keynes 1981, p. 42) and subsequently into the committee's report (United Kingdom, Committee on Finance and Industry 1931a, pars. 46-47). Between Kindersley's coining of the phrase and the present the exact meaning of the "rules" has varied. For Keynes, they meant that "you so conduct your affairs that you tend neither to gain nor to lose large quantities of gold" (Keynes 1981, p. 42). The Macmillan committee was even more general when it argued: It is difficult to define in precise terms what is implied by the "rules of the game". The management of an international standard is an art and not a science, and no one would suggest that it is possible to draw up a formal code of action, admitting of no exceptions and qualifications, adherence to which is obligatory on peril of wrecking the whole structure. Much must necessarily be left to time and circumstance. (United Kingdom, Committee on Finance and Industry 1931a, par. 47) Nevertheless, when economists have come to assess the possible reasons for the success or failure of fixed-exchange-rate regimes, and occasionally even more "flexible" ones (Chisholm 1979), they have often specified a set of rules conducive to the stability of the regime and tested for the relevant authorities' adherence or nonadherence thereto. The most famous of these exercises are those of Nurkse (1944) and Bloomfield (1959) for the interwar and pre-1914 gold standard periods. Both men tested a relatively activist rule that internationally equilibrating behavior would move central banks' foreign and domestic assets in the same direction as they reinforced the impact of reserve movements on financial Donald E. Moggridge is professor of economics at Scarborough College, University of Toronto, West Hill, Ontario.

196

John Dutton

markets. In both cases, using annual data, they found adherence to such a rule was the exception rather than the norm. Alternative suggestions for rules have been provided by Bloomfield (1959, 1968) and Michaely (1968). The former suggested that behavior would be equilibrating if central banks did not offset the effects of reserve changes or if their discount rate moved inversely with their reserve holdings or reserve ratios. The latter suggested that suitable behavior would see the money supply varying directly and the central-bank discount rate moving inversely with reserve changes. Bloomfield only chose to test his discountrate rule during the pre-1914 period while Michaely's rule was applied to the Bretton Woods period. Doubtless one could also present another rule involving relative rates of change of a suitable monetary aggregate and apply it to the pre-1914 period, but as yet I know of no such exercise. John Dutton's paper brings to the discussion of the observance or nonobservance of the rules of the game by the pre-war Bank of England a new twist to the rules and a new test of observance of the NurkseBloomfield reinforcement rule-and incidentally a check on the more passive no-offsetting and discount-rate rules. Dutton's new twist states that given equilibrating central-bank behavior under classical gold standard conditions, one would not observe central banks pursuing countercyclical policies. One can see the sense of such a rule in the abstract, for it would mean that the banks involved would avoid meeting potentially destabilizing dilemma cases and it would incidentally economize on the need for international reserves. However, I wonder whether it is an appropriate rule for the pre-1914 international economy given that Bloomfield (1959, p. 38), Morgenstern (1959, chap. 2) and Triffin (1964, chap. 1) have all noted the strong parallelism in movements of economic activity during the period. In such circumstances, nonadherence to the rule might still be consistent with the successful operation of the standard. As well as providing a possible new rule, Dutton's paper tests the Bank of England's observance of various rules in a new form. Rather than simply comparing the authorities' actual behavior to the rules, he proposes a more complex model in which the Bank reacts to forecasts of its possible target variables (the Banking Department reserve, the proportion, domestic activity, prices and gold movements) by altering the level of its discount rate and its domestic assets. The rationale for this more involved procedure is twofold. First the Bank might not have full information as to the most recent values of its target variables at the time of making policy adjustments, and second, the procedure eliminates some problems of simultaneous-equation bias. Leaving the second reason to one side, I cannot fully see the strength of Dutton's first procedural rationalization. It is true, given Dutton's-and Charles Goodhart's (1972)-problem of finding a good monthly index of

197

The Bank of England and the Rules of the Game

activity, that the Bank was unlikely to have had full information on that score. The same would almost certainly be the case as regards the price level. But why Dutton should think that the Bank did not have full information about its own reserve, its proportion, and gold movements strikes me as odd, given the information the governors received each working day at the daily "books" meeting (Sayers 1976, 1: p. 31). Thus it would seem to me that the justification for Dutton's technique must depend more on the usefulness of the new rule he wishes to test and on its statistical characteristics than on its being representative of the details of Bank behavior. This is particularly the case when the forecasting model assumes that the Bank had information on each variable's behavior over the entire period and used the same processes consistently. Both of these assumptions seem to me suspect, for the seasonal variability of certain matters such as internal drains changed over the period (Sayers 1976, 1: p. 32) and it is clear from the narrative material available that the Bank's procedures and techniques, as well as the balance among the latter, were changing markedly over the twenty-five years before 1914. Thus it would seem to me that one cannot really be certain exactly what Dutton's procedures are capturing at the end of the day. What appears to be going on is that the Bank reacted most markedly to the traditional stimuli, changes in the level of its reserves and the proportion, and in the expected direction. One would expect these reactions; they would be consistent with one Bloomfield rule and part of the Michaely rule of the game. There is as well the confirmation of the inverse relationship between changes in the Bank's reserves and its holding of securities that one might expect given Bloomfield's evidence on annual data that the Bank followed the reinforcing rule of the game just less than half the time between 1880 and 1914 (Bloomfield 1959, p. 50). What we do not know is whether this offsetting was partial as Bloomfield suggested (1959, p. 50) or complete-whether the Bank was merely inclined to lean against the wind or stand resolutely against it. The former might be within the spirit of a possible rule, given its Bank-rate reaction, while the latter would represent a violation of the rules in almost any common formulation. Nor do we know, although the Bank did, how much offsetting was an automatic reflection of the discount market being forced into the Bank and how much reflected deliberate policy. Perhaps some day the Bank or some private scholars will extract the necessary information from the "books" and thus help remove another puzzle. Finally we have the suggestion that the Bank responded countercyclically to the activity and price variables, something that Bloomfield (1959, p. 33) regarded as incidental and Ford (1962, p. 34) believed was not a conscious policy at all, although narrative accounts of the Bank's behavior have understood it as a subsidiary but growing preoccupation. Whether this behavior reflected a continuing but changing Bank concern or the long-standing

198

John Dutton

suggestion that given the power of Bank rate the money-supply process in Britain was to some extent endogenous (Ford 1962, p. 36; Goodhart 1972, chap. 15) is not pursued in this paper, concerned as it is with the rules of the game. Thus the paper leaves us in the position of confirming the Bank's adherence to some rules and suggesting violations of others. Perhaps further work by Dutton will clarify the extent of these violations, especially of the no-offsetting rule, and indicate whether the Bank became more inclined to violate some rules over time-perhaps because it did become more concerned about levels of economic activity. For the present we can thank Dutton for the questions he has raised and hope that discussion and more work will help us come to find answers. References Bloomfield, Arthur I. 1959. Monetary policy under the international gold standard, 1880-1914. New York: Federal Reserve Bank of New York. - - - . 1968. Rules of the game of international adjustment In Essays in Money and Banking in Honour of R. S. Sayers, ed. c. R. Whittlesey and J. S. C. Wilson. Oxford: Clarendon Press. Chisholm, Derek. 1979. Canadian monetary policy, 1914-1934: The enduring glitter of the gold standard. Ph.D. diss., Cambridge University. Ford, A. G. 1962. The gold standard, 1880-1914: Britain and Argentina. Oxford: Clarendon Press. Goodhart, C. A. E. 1972. The business of banking, 1891-1914. London: Weidenfeld and Nicolson. Keynes, J. M. 1981. Activities, 1929-1931: Rethinking employment and unemployment policies. Vol. 20 of The Collected Writings of John Maynard Keynes, ed. Donald Moggridge. London: Macmillan for the Royal Economic Society. Michaely, Michael. 1968. Balance-of-payments adjustment policies: Japan, Germany, and the Netherlands. Occasional paper 106. New York: National Bureau of Economic Research. Morgenstern, Oskar. 1959. International financial transactions and business cycles. Princeton: Princeton University Press. Nurkse, Ragnar. 1944. International currency experience: Lessons of the inter-war period. Princeton: League of Nations. Sayers, R. S. 1976. The Bank of England, 1891-1944. Cambridge: Cambridge University Press. Triffin, Robert. 1964. The evolution ofthe international monetary system: Historical reappraisal and future perspectives. Princeton Studies in International Finance, no. 12. Princeton: Princeton University Press.

199

The Bank of England and the Rules of the Game

United Kingdom. Committee on Finance and Industry. 1931a. Report. London: His Majesty's Stationery Office. - - - . 1931b. Minutes of evidence. London: His Majesty's Stationery Office.

Reply

John Dutton

I would like to respond to two points in Professor Moggridge's "Comment. " First, he indicates concern that the behavior of the Bank of England varied substantially over the twenty-seven years studied. He is also bothered by the possibility of changes in seasonal and other patterns in the target variables over the period and by the effects those changes might have on the forecasting equations. I share those concerns and so have attempted to allay them. Tables C3.1 and C3.2 contain linear feedback equations, similar to those of tables 3.3 and 3.4, but for two shorter periods. Some changes in the coefficients appear. However, the earlier findings for the whole period remain intact in the results for the two subperiods. Bank rate in each subperiod reacted negatively to reserves, positively to economic activity, and positively to inflation. The reserve variable is clearly significant in each case (although its coefficient is much greater for the second period). The activity variables give much the same results as those for the whole period. Unemployment has a significant, or nearly significant, negative effect; railway receipts show a weak but consistently positive effect. Inflation has statistically significant positive coefficients for the first period but not the second. Likewise, in both periods, securities holdings of the Bank tend to change in the opposite direction from predicted reserve changes. Securities appear to have decreased about one-half to two-thirds as much as the change in reserves. This decrease is consistent with results for the whole period, indicating that the processes governing the Bank probably did not change dramatically between the two periods. Moggridge mentions concern that seasonal patterns of the variables changed over the twenty-seven years of the study. Figures C3.1 and C3.2 contain plots of monthly means of several variables for periods ending in 1900 and beginning in 1901. The patterns for the two periods seem remarkably similar. Gold inflows evidence the largest change (as well as large standard deviations for the means). Unemployment patterns during the first quarter of the two subperiods differ somewhat. The overall impression, however, is of substantial likeness between the overall seasonal patterns. Professor Moggridge rightly worries about stability of the forecasting

200

John Dutton Bank-rate Equations

Table C3.1

Equation a

Reserves x 10- 2

Nb

Rlwy Rcpts X 10- 3

Unemp X 10- 2

Inflation

Autoregressive Termsc

First Period (through 1900)

none

1. (1893)

96

-1.34 (1.82)

4.01 (1.01)

2. (1893)

96

-1.10 (1.47)

4.87 (1.23)

3. (1888)

156

-1.45 (1.95)

-3.49 (1.83)

4. (1888)

156

-1.28 (1.75)

-3.12 (1.65)

5. (1888)

156

-1.81 (2.23)

2.40 (1.59)

none lag 7, .14(1.73) lag 10, - .14(1.80)

2.18 (2.03)

lag 7, .15(1.87) lag 10, - .14(1.80) lag 10, - .16(2.08)

Second Period (through 1914)

lag 1, .13(1.68)

6. (1901)

162

-18.42 (6.19)

1.70 (1.05)

7. (1901)

162

-18.46 (6.18)

1.73 (1.07)

8. (1901)

162

-18.37 (6.37)

-3.79 (2.53)

9. (1901)

162

-18.43 (6.39)

-3.84 (2.56)

10. (1901)

162

-18.50 (5.92)

0.25 (0.39)

lag 1, .13(1.70) lag 1, .15(1.99)

0.34 (0.55)

lag 1, .18(2.02) none

Note: t-statistics are in parentheses. aDate in parentheses is beginning year. b N = number of observations. cIndicated are lag number of autoregressive error term, estimated autocorrelation coefficient used in data transformation, and t-statistic of parameter.

Table C3.2

Securities-Holdings Equations a

Period

N

1888-1900

155

1901-1909

108

Reserve Forecast - Reserves (-1) x 102 -7.39 (1.40) -5.25 (1.84)

Note: t-statistics are in parentheses. aThe errors in this equation showed no statistically significant autocorrelation. Securities changes measured in thousand-pound units, reserves in million-pound units. Coefficient x 10- 1 indicates portion of reserves changes offset.

201

The Bank of England and the Rules of the Game ~.j

11 ions of

ounds

Percent

15 10

Balk of

Eng I Old

Brnking

~oortrrent

reserves

55

/

//'\

50

\

\

'.-/

\ \,...- -"

45 40

.,r-\

Prnk rate \

/ /

4.0 3.5

/

~

.......

/

"""----- -_--....,

///

/

'-'"

3 .0

2.5

65

60

55 50

Fig. C3.1

Balk of

Eng IOld

securi ties holdings

Bank of England financial variables, means, by months, 18881900 and 1901-14. Earlier and later period values are indicated by broken and solid lines, respectively. Source: See section 3.6.

process over the whole period in question. The ARIMA forecasting equations are at best only rough proxies for the Bank's forecasts. One would expect the coefficients on these rough proxies to be biased toward zero. That significant coefficients are obtained for the period as a whole and for the two subperiods seems to indicate that the ARIMA forecasts are doing their proxying job reasonably well. The second of Moggridge's concerns to which I should like to respond is the validity of the no-countercyclical-activity rule of the game used in the paper. The rule seems reasonable to me because the Bank, with limited policy tools, would have needed to concentrate its efforts on the single major goal of maintaining convertibility. Efforts at countercyclical policy would likely have interfered with pursuit of that goal. They might

202

John Dutton ~'j

11 ions of

Percent

ounds

.6

,""-

-~

Railway receipts

,...----...

''----......... ' - - - - - - . . / /

//

,

//

,

",

5.0

"

Lharployrrent rate

4.5

4.0

1.5

0.5

--- -----

(?DId infICl'ls

-0.5 -1. 5

-2.5 .JAN

Fig. C3.2

F B

~~AR

APr

~~AT

JUN

JlTL

AUG

SEP

OCT

NOV

DEC

U.K. railway receipts, unemployment rate, and gold flows, means, by months, 1888-1900 and 1901-14. Source: See fig. C3.1.

also have had destabilizing effects. There is some reason to believe that gold inflows, Banking Department reserves, and domestic activity moved together (see Pippenger, this volume). If they did, then countercyclical Bank-rate policy would have tended to reinforce those cyclical movements of gold and reserves, rather than tempering them. Moggridge also suggests in his comment that the apparent countercyclical policy evidenced in my equations might have reflected a common response of Bank rate and economic activity to some third variable. Such a possibility cannot be completely dismissed. Possible candidate variables are reserves and the money supply. However, if reserves and the money supply were positively related to domestic activity, as seems likely, and Bank rate negatively related to those two, then any indirect effect of activity via reserves on Bank rate would be negative. The coefficients in the paper consistently indicate a positive relationship. Other scenarios for explaining the empirical results are of course possible. It is plausible, however, that the results could signify countercyclical actions on the part of the Bank. In any case, Professor Moggridge is certainly correct in calling for more work to reduce the extent of uncertainty about the Bank's policies.

Bank of England Operations, 1893-1913

4

John Pippenger

The Bank of England did not publish figures for bankers' deposits until 1967. The first economist to use that information was Goodhart (1972). This study builds on and reexamines the work of Goodhart, whose conclusions conflict with the conventional wisdom about the Bank and the gold standard. Section 4.1 reviews some of Goodhart's results, section 4.2 examines the long-run operations of the Bank, section 4.3 analyzes short-run behavior, and the final section 4.4 presents the conclusions. An appendix provides spectral estimates of key variables examined in this study. 4.1

Goodhart's Results

Goodhart (1972) analyzed the operations of the Bank of England and British commercial banks and their roles in the functioning of the gold standard from 1891 to 1914. His conclusions about the role of the Bank in the operation of the gold standard challenge the conventional wisdom. The strongest link in the causal chain of the classical analysis of the working of the gold standard mechanism is generally considered to be that connecting changes in the reserve base of the commercial banks with fluctuations in the (gold) reserve, or liquidity, position of the central bank. Yet in this study of the working of the system in the UK this is the link which shatters . . . . there is no simple direct relationship between the variations in the levels of bankers' balances at the Bank and in the level of the reserve in the Bank. (Goodhart 1972, p. 209) This conclusion rests primarily on two regressions. In the first, monthly data on bankers' balances at the head office of the Bank of England are John Pippenger is professor of economics at the University of California, Santa Barbara.

203

204

John Pippenger

regressed against time, reserves in the Banking Department, and seasonal factors. There is no link between bankers' balances and reserves.

= 7913.5 + 37.74 time

Bankers' balances

(967.4) (1.97) - 1.29 reserves + seasonals. (34.53) With seasonals R2

= 0.64, D.W. = 1.07. Without seasonals R2 = 0.56.

The numbers in parentheses are the standard errors. The second equation is in logs and adds railway freight receipts as a proxy for nominal income. Log bankers' balances = 4.843 + 0.0006 time (1.142) (0.0003) + 0.707 log freight receipts (0.167) + 0.092 log reserves. (0.40)

R2

= 0.679, D.W. = 0.96. Without seasonals R = 0.579.

With Seasonals

2

Now a positive relation between bankers' balances and reserves emerges, but the estimated response to income is several times larger than the response to reserves. Goodhart also estimates two other relationships that are relevant for the operations of the Bank of England. One attempts to explain the ratio of reserves in the Banking Department to total liabilities of that department----otherwise known as the proportion.

= 7.39 + 0.0006 time

Log proportion

(0.78) (0.0002) - 0.79 log freight receipts (0.11) + 0.53 log reserves + seasonals. (0.03)

R2

= 0.765, D.W. = 0.90. Without seasonals R2 = 0.584. With seasonals

The standard errors are in parentheses. Goodhart (1972, p. 206) interprets this result as follows: "It suggests that the Bank must have regularly accommodated, to some large extent, variations in the demand for cash

205

Bank of England Operations, 1893-1913

caused by changes in the level of domestic activity by varying its holdings of other assets, independently of the level of gold reserves." The final relationship attempts to explain Bank rate in terms of trend and the liquidity position of the Bank of England, first using the proportion and then the reserves as a measure of liquidity:

= 4.48 + 0.0012 time

Bank rate

(0.89) (0.0004) - 1.09 log proportion (0.22)

+ 0.746 Bank rate (t - 1) + seasonals. (0.037)

"R 2

= 0.795. Without seasonals "R = 0.736. Bank rate = 2.49 + 0.0021 time With seasonals

2

(0.55) (0.0005) - 0.714 log reserves (0.165) + 0.756 Bank rate (t -1) + seasonals. (0.038) With seasonals

"R 2

Without seasonals

= 0.791.

"R

2

=

0.749.

The numbers in parentheses are the standard errors. The results show the expected inverse relation between Bank rate and the liquidity position of the Banking Department. The next two sections reexamine the operations of the Bank of England, employing as much as possible the data used by Goodhart.2 The first section concentrates on the long-run and the second looks at the short-run behavior of the Bank.

4.2

Long-Run Operations

This section concentrates on long-run behavior by using annual averages of monthly data.3 The next section, in order to emphasize short-run operations, uses monthly changes. Sayers (1976, p. 8) points out that the governor of the Bank of England had three primary objectives. He had a statutory duty to maintain the convertibility of the note into gold coin; he had a political duty to look after the financial needs of government; and he had a commercial duty to maintain an income for

206

John Pippenger

the stockholders. Whenever possible, he was running all three horses at once, but if there was a conflict, he knew which he had to put first. He would think of his primary duty as the maintenance of the gold standard. Although a variety of special situations probably influenced the short-run operations of the Bank, the duties cited by Sayers, particularly the statutory and commercial duties, appear to dominate long-run behavior of the Bank. 4.2.1

Bankers' Deposits and Reserves

Goodhart's most challenging discovery is the weak relationship between reserves in the Banking Department and bankers' deposits at the Bank of England. His results threaten a crucial link in the conventional interpretation of the gold standard. Consider a very simple model of Bank-portfolio behavior in which desired reserves R depend on deposits and interest rates. (1)

R

=

O.

Again, we concentrate on the short term and therefore leave relative prices and output as exogenous to the model. 5.3.3

Formal Dynamics

Equations (4'), (5), and (6) represent a dynamic model of the interaction between the Banking Department's credit policy and the balance of payments. Substituting equation (4') in equations (5) and (6) yields the following pair of ·equations: (7) (8)

;= G(r, B); GrO. B = H(r, B); H r > 0, H B < O.

where the signs of the partial derivatives follow from the previous assumptions. It is readily verified that the system shown in figure 5.4 must be stable.

242

Rudiger Dornbusch and Jacob A. Frenkel

r

In figure 5.4 the = 0 schedule shows the locus of reserve-deposit ratios and levels of bullion at which the Banking Department is in equilibrium with respect to its liquidity position. Therefore, along that schedule the reserve-deposit ratio is neither rising nor falling. At points above the schedule, the high reserve-deposit ratio implies a low realmoney supply and thus high interest rates. The preferred reserve-deposit ratio is low, and therefore above the r = 0 schedule the reserve-deposit ratio is being lowered. Conversely, below the = 0 schedule, the Banking Department seeks to become more liquid because interest rates are low, and ther~fore the reserve-deposit ratio is raised. Along the B = 0 schedule the balance of payments is in equilibrium. Points below and to the right of the schedule correspond to high money supplies, low interest rates, capital outflows, and therefore deficits and falling bullion. By contrast, points to the left of the sch~dule involve high interest rates and growing levels of bullion. Along the B = 0 schedule the interest rate is compatible with external balance. The interest rate is higher above and to the left of the schedule and lower below and to the right of the schedule. The relative slopes of the two schedules are implied by the previously assumed restrictions. As the arrows indicate, the dynamic model of the financial sector must be stable and the approach to equilibrium cannot be oscillatory. From any initial reserve-deposit ratio and stock of bullion, the adjustment process leads to the steady state at point A where the Bank's liquidity position is in equilibrium and external payments are balanced. The response of the Bank's reserve-deposit ratio to the rate of interest is reflected in the slope of the r = 0 schedule. The less responsive the

r

Reserve - Deposi t Ratio

8=0

r= 0 r

o

~----_-----&_----- Bullion

o Fig. 5.4

Financial model.

B

243

The Bank of England in the Crisis of 1847

Bank, the flatter the schedule; in the extreme, when the Bank is entirely unresponsive, the desired reserve-deposit ratio is constant and the = 0 schedule is horizontal. As the reserve-deposit ratio becomes more responsive, th~ schedule steepens until, in the limit, its slope coincides with that of the B = 0 schedule. The responsiveness of the reserve-deposit ratio, of course, determines the extent to which interest rates move in the adjustment process. If the reserve-deposit ratio declines in response to high interest rates, then a shortage of bullion will in part be offset by increased lending on the part of the Bank and interest rates therefore will tend to be lower, the balance of payments will be smaller, and the rate of adjustment will be slower. Conversely, if the reserve-deposit ratio is unresponsive, a shortage of bullion implies a sharper reduction in the money stock, high equilibrium interest rates, larger capital flows, and faster adjustment.

r

5.3.4 The Adjustment Process The traditional model of the price-specie-flow mechanism, originating with David Hume, emphasizes the impact of relative prices on the trade balance and hence on the balance of payments and the international flow of bullion. A deterioration of the external balance due to increased aggregate spending or an adverse development of net exports will lead to bullion export, monetary deflation, declining spending, and price deflation. Both the decline in spending and deflation work to restore external balance. The model we have sketched here, on the contrary, places emphasis on capital flows and banking policy as the main factors in the adjustment process. The two views of the adjustment mechanism are of course complementary, although they may well correspond to different adjustment periods. In the short run, banking policy and capital flows are likely to be the main factors determining bullion flows, since, in the short run, prices and trade flows do not adjust to the full, possible extent. The role of capital flows in the adjustment process was recognized by contemporaries. John Stuart Mill, in particular, noted: It is a fact now beginning to be recognised, that the passage of the precious metals from country to country is determined much more than was formerly supposed, by the state of the loan market in different countries, and much less by the state of prices. (Mill I8?1 , bk. 3, chap. 8, section 4) In addition to the difference between the balance of trade and the capital account in facilitating adjustment, there is another aspect of the adjustment process that deserves emphasis. The traditional representation of the gold standard takes it to be an automatic, nondiscretionary

244

Rudiger Dornbusch and Jacob A. Frenkel

adjustment. Bullion flows are matched one-for-one by changes in the amount of currency outstanding. This is, of course, not the case once the reactions of the Banking Department are taken into account. Changes in the reserve-deposit ratio of the Banking Department affect the money stock independently of the existing stock of bullion. The question then arises whether during the 1847 episode the Banking Department's credit policy might in fact have amounted to partial or even complete sterilization of bullion flows. The possibility of credit expansion by the Bank and of loss of note reserves to finance the export of bullion is suggested by the data which reveal a high correlation between weekly changes in bullion and in note reserves. Consider an autonomous, transitory improvement in the trade balance which leads to an inflow of bullion and therefore to a monetary expansion. The monetary expansion lowers the interest rate, and, with a constant reserve-deposit ratio (or a flat; = 0 schedule in figure 5.4), the lower interest rate leads to capital outflows and thereby to restoration of the initial equilibrium. Now if, on the contrary, the reserve-deposit ratio rises due to the Banking Department response to the reduced profitability of loans, then the rise in the reserve-deposit ratio dampens the decline in interest rates and therefore slo.ws down the adjustment process. The Banking Department's reaction to the interest rate will only slow the speed of adjustment but will not eliminate the adjustment process. Thus our model is also capable of incorporating a partial sterilization policy with an effect of dampening interest-rate movements and reducing the speed of adjustment. Changes in the reserve-deposit ratio enter consideration in another respect. If the Bank, perhaps in response to a loss of confidence on the part of the public, decides to raise the reserve-deposit ratio, then this raise, of course, leads to a reduction in the supply of money and credit. Interest rates rise and that state persists until bullion inflows accommodate the desired increase in reserves. The model suggests therefore that changes in the Bank's reserve preferences may be an important source of macroeconomic disturbance. The possibility of internal inconvertibility turns out to be an important issue in the 1847 crisis. Internal inconvertibility would arise if the Banking Department should become sufficiently illiquid not to be able to redeem its deposit liabilities in notes. Thus there is a clear distinction between external or gold convertibility and internal or note convertibility. Note convertibility involves the Banking Department's reservedeposit ratio. If the reserve-deposit ratio falls too low, the public loses confidence and reacts by raising the currency-deposit ratio. While our model embodies this reaction of the public, the reaction is for the moment not allowed to exercise a dominating influence.5

245

The Bank of England in the Crisis of 1847

5.3.5

Some Evidence

Before discussing in detail the various crises that occurred during 1847, we look at some evidence that is consistent with the general analytical framework outlined in this section. The dynamic model was summarized by equations (7) and (8). Changes in the reserve-deposit ratio depend negatively on the level of that ratio and positively on the stock of bullion, while changes in the stock of bullion depend positively on the reserve-deposit ratio and negatively on the stock of bullion. In table 5.4 we report regressions of the changes in the reserve-deposit ratio and bullion on the previous-week levels of these variables. The coefficients have the predicted sign and are statistically significant. We view these estimates as providing support for the analytical framework that was developed in this section, and we turn next to a more detailed analysis of the crises of 1847. 5.4 Financial Markets and the Balance of Payments in 1847

5.4.1

The April Crisis

The harvest failure of 1846-47 depleted the bullion in the Bank in the fall of 1846 and more so in early 1847. Table 5.5 shows the development of bullion, note reserves, the reserve-deposit ratio, the discount rate, and the stock of notes in the hands of the public during the first half of 1847. The table brings out forcefully the magnitude of this depletion. Indeed, over the period 2 January to 17 April 1847, bullion fell by about 40 percent and note reserves in the Banking Department declined by about 70 percent. The extraordinary decline in the reserve-deposit ratio from 46 percent to 19.6 percent implies that the Bank sterilized substantially the effect of gold outflows. The decline in the reserve-deposit ratio occurred The Dynamic Model, 1847 Weekly Data (standard errors in parentheses)

Table 5.4 Dependent Variable Constant -0.119 (0.099)

~'t

0.337(107 ) (0.099)107

~Bt

't

't-1

-0.625 (0.151) 0.262(107 ) (0.118)107

Bt -

1

0.301(10- 7 ) (0.122)10- 7 -0.413 (0.116)

R2

D.W.

P

.36

1.85

.75

.66

2.00

.85

Notes: and B t denote, respectively, reserve-deposit ratio and bullion in the Bank of England; ~'t and ~Bt denote the weekly change in these variables. R 2 denotes the coefficient of determination and p the first-order autocorrelation coefficient.

246

Rudiger Dornbusch and Jacob A. Frenkel

along with an increasing discount rate. The table reports the weighted average discount rate applied by the Bank. From a level of 3 percent at the beginning of the year, the rate was gradually raised toward 5 percent in early April 1847. These developments suggest that part of the effects of the external drain on the money supply were sterilized. Whether sterilization was a conscious policy, or whether it was a banking response to increasing interest rates and credit tightness, is open to question. But it is certainly interesting to note that F. T. Baring, the ex-chancellor of the exchequer, argued that the possibility of sterilization was a major defect of the Bank Act of 1844: I believe, if we look back, we shall find that the operation of the deposits and the question of reserve was not sufficiently considered, either by those who were favourable or those who were opposed to the bill. I cannot find in the evidence before the committee of 1840 more than a few sentences leading me to suppose that danger arising from such a cause was contemplated or referred to; yet this was a most important consideration; for it was by the reserve, the bank was enabled to do what was contrary to the spirit of the bill when gold was running out, not to reduce their circulation by a single pound. I do not think that the system works satisfactorily in this respect; and in fact, the point did not receive anything like a sufficient consideration. Perhaps it was impossible before the bill was in practical operation to see how the reserve of notes would operate; but it certainly never entered into the contemplation of anyone then considering the subject that £7,000,000 in gold should run off, yet that the notes in the hands of the public would rather increase than diminish. (MacLeod 1896, pp. 141-42) The relative constancy of notes in the hands of the public, to which Baring refers, is shown in table 5.5. Through late April 1847, notes were practically unchanging while bullion declined by nearly one-third. In this period the Bank of England expansion "financed" the export of bullion.

Table 5.5

2 January 6 March 3 April 17 April 1 May 5 June

The April 1847 Crisis (million £)

Note Reserves

Bullion in Issue Department

ReserveDeposit Ratio (%)

Discount Rate (%)

Notes in Public Hands

8.23 5.71 3.70 2.56 2.74 5.09

14.26 10.99 9.55 8.80 8.51 9.43

46.0 36.0 23.9 19.6 23.6 32.0

3.10 4.14 4.25 5.25 5.25 5.20

20.0 19.3 19.9 20.2 19.8 18.3

Sources: See table 5.2 and Appendix.

247

The Bank of England in the Crisis of 1847

Bullion losses did not exert their full contractionary effect on money and credit because the reserve-deposit ratio was declining.6 In March and April things became troublesome. The ongoing decline of bullion tightened credit-market conditions, and the failure of the Bank to change its accommodating stance in the face of a deteriorating balance sheet evoked concern about a sudden reversal of policies that would leave the public without notes and without loans. In April, therefore, the ongoing drain of bullion was reinforced dramatically by the seasonal payment of the dividend, which meant a significant run-down of public deposits. During the week of 17 April reserves fell to a level of only £2.56 million; the reserve-deposit ratio fell to less than 20 percent. Consol prices fell in March-April by 4.5 percent and short-term interest rates skyrocketed as the Bank moved vigorously to restore its liquidity position by reduced discounts, consol sales, and high discount rates. Figure 5.5 shows the series for private and public deposits. The figure makes it clear that whatever influence the bullion drain had on the liquidity position of the Bank, the sharp public-deposit withdrawal could not but accentuate the problem. Table 5.6 shows that the public-sectordeposit withdrawal led only partially to a loss of note reserves and that the money stock (currency held by the public plus private deposits at the Bank of England) did not change substantially. The table confirms that the Bank of England managed to face the runoff by selling securities. Figure 5.6 shows the weekly series of the reserve-deposit ratio during the year. The effects of the extraordinary loss of reserves (by 17 April) and the reaction of the financial markets and the Bank of England have been described by MacLeod (1896, p. 142):

12.0,..........-------------------.....,

9.5

, ~~

,

, 4.5

,J

"II ~

\

u , = ,

\ \ Q. ,

, L--

I 1847

Fig. 5.5

9 1847

I I I ,

I \.~,'

\ 0,

(A2)

M t - M t-

(A3)

1";*- Yi-l

1

=

C

< 0;

a(Mi - M t - 1 ) + ut; 0 < a

= ~(1";- 1";~1);

0