Unequal Gains: American Growth and Inequality since 1700 9781400880348

Table of contents :
CONTENTS
List of Illustrations
Preface
1 Persistent Debate, a New Approach, More Data, Rich Findings
2 Colonial Incomes on the Eve of the Revolution
3 When Did Colonial America Get Rich?
4 Losing the Lead: The Cost of Revolution and Independence
5 Unequal Economic Growth, 1800–1860
6 The Civil War: Growth Lost, Freedom Gained, Inequality Maintained
7 Contending Forces: American Incomes across the Late Nineteenth Century
8 The Greatest Leveling of All Time
9 Rising Inequality Once More, since the 1970s
10 Inequality and Growth: History Lessons for the Future
Appendix A A Guide to the 1774 and 1800 Income Estimates
Appendix B Salaries, Payment in Kind, and Workdays
Appendix C Estimating Slaves’ Retained Earnings, Colonial Times to 1860
Appendix D American versus British Prices, 1640–1875
Appendix E A Guide to the 1860 Income Estimates, and Some Modifi cations for 1850
Appendix F A Guide to the 1870 Income Estimates
Appendix G Farm Operators’ Incomes in 1870
Appendix H Sources and Notes to Tables and Figures in Main Text
References
Index

Citation preview

UNEQUAL GAINS

The Princeton Economic History of the Western World Joel Mokyr, Series Editor A list of titles in this series appears at the end of the book

UNEQUAL GAINS American Grow th and Inequalit y sin c e 1 7 0 0 PETER H. LINDERT AND JEFFREY G. WILLIAMSON

Princeton University Press Princeton and Oxford

Copyright © 2016 by Princeton University Press Published by Princeton University Press, 41 William Street, Princeton, New Jersey 08540 In the United Kingdom: Princeton University Press, 6 Oxford Street, Woodstock, Oxfordshire OX20 1TW press.princeton.edu Jacket design by Faceout Studio All Rights Reserved Library of Congress Cataloging-in-Publication Data Names: Lindert, Peter H., author. | Williamson, Jeffrey G., 1935– author. Title: Unequal gains : American growth and inequality since 1700 / Peter H. Lindert and Jeffrey G. Williamson. Description: Princeton : Princeton University Press, [2016] | Series: The Princeton economic history of the Western world | Includes bibliographical references and index. Identifiers: LCCN 2015044019 | ISBN 9780691170497 (hardcover : alk. paper) Subjects: LCSH: Income distribution—United States—History. | Wealth—United States—History. | Economic development—United States—History. | United States—Economic conditions Classification: LCC HC110.I5 L5624 2016 | DDC 339.20973—dc23 LC record available at http://lccn.loc.gov/2015044019 British Library Cataloging-in-Publication Data is available This book has been composed in Minion Pro and Scala Sans OT Printed on acid-free paper. ∞ Printed in the United States of America 10 9 8 7 6 5 4 3 2 1

For Lin and Nancy plus the two wonderful generations that have followed

CONTENTS

List of Illustrations Preface

1

ix xv

Persistent Debate, a New Approach, More Data, Rich Findings

1

2

Colonial Incomes on the Eve of the Revolution

13

3

When Did Colonial America Get Rich?

43

4

Losing the Lead: The Cost of Revolution and Independence

77

5

Unequal Economic Growth, 1800–1860

96

6

The Civil War: Growth Lost, Freedom Gained, Inequality Maintained

142

Contending Forces: American Incomes across the Late Nineteenth Century

166

8

The Greatest Leveling of All Time

194

9

Rising Inequality Once More, since the 1970s

219

7

10 Inequality and Growth: History Lessons for the Future Appendix A A Guide to the 1774 and 1800 Income Estimates Appendix B Salaries, Payment in Kind, and Workdays Appendix C Estimating Slaves’ Retained Earnings, Colonial Times to 1860 Appendix D American versus British Prices, 1640–1875

242 263 279 287 304

Appendix E A Guide to the 1860 Income Estimates, and Some Modifications for 1850 Appendix F A Guide to the 1870 Income Estimates Appendix G Farm Operators’ Incomes in 1870 Appendix H Sources and Notes to Tables and Figures in Main Text References Index

viii

•

CONTENTS

311 320 327 349 369 391

ILLUSTRATIONS

TABLES 2-1. Main Data Inputs for 1774 Income Estimates

18

2-2. Estimated American Personal Incomes, 1774

29

2-3. Alternative Estimates of National Income, 1774, in Current and 1840 Dollars (Millions)

30

2-4. Inequality in the American Colonies, 1774

38

2-5. Inequality in Western Europe, 1732–1808

39

3-1. Past Estimates of Eighteenth-Century Colonial Income per Capita Growth

46

3-2. Colonial Price Volatility, 1700–1776, in Long-Run Perspective

51

3-3. Trends in the Colonies’ Terms of Trade, 1700–1776

53

3-4. Indicators Guiding the Income Backcasts from 1774 to 1650

60

3-5. Predicted Wealth for Forty-Five-Year-Old Colonial New Englanders, by Time and Place, for Selected Occupations

61

3-6. Conjectural Estimates of Gross Income per Capita, 1650–1774

62

4-1. Estimated American Incomes, Eastern Seaboard, 1800

80

4-2. Alternative Estimates of National Income, 1774 and 1800, in Current and 1840 Dollars (Millions)

81

4-3. Alternative Property Incomes and Total Incomes, 1774 and 1800

83

4-4. Real Income per Capita, 1774–1840

84

4-5. Wage Gaps and Skill Premiums, 1774 and 1800

85

5-1. Estimated American Personal Incomes in 1850 (Millions of Current Dollars)

98

5-2. Estimated American Personal Incomes in 1860 (Millions of Current Dollars)

99

5-3. Real Income per Capita Growth, 1800–1860

102

5-4. West European Growth Rates, 1800–1860 (% per Annum, per Capita)

104

5-5. Regional Relative Incomes per Capita, 1800–1860

108

5-6. The Inequality of American Household Incomes, 1850

115

5-7. The Inequality of American Household Incomes, 1860

116

5-8. The Inequality of American Property Incomes, 1774–1870

122

5-9. Earnings Inequality in America, 1774–1860

122

5-10. Wage Gaps and Skill Premiums, 1800–1860

126

5-11. The Kuznets Urbanization Effect, 1800–1870, All Households

128

5-12. Was There Artisan Middle Class Hollowing Out between 1800 and 1860?

134

6-1. Estimated American Personal Incomes in 1870

145

6-2. Real Incomes per Capita, 1860–1870

148

6-3. Southern Free and White Incomes, 1860–1870

152

6-4. The Inequality of American Household Incomes, 1870

154

6-5. The Inequality of American Household Labor Earnings, 1870

157

x

•

LIST OF ILLUSTRATIONS

6-6. Labor Force as a Percentage of Population, by Race and Region, 1860–1870

162

6-7. The Progress of Black Incomes, 1860–1870

164

7-1. GDP per Capita Growth, 1870–1910: Western Europe versus United States

170

7-2. Inequality Trends in America, 1870–1929

173

7-3. The Kuznets Urbanization Effect, 1870–1910, All Households

175

7-4. Wage Gaps and Skill Premiums, 1870–1910

176

7-5. Personal Income per Capita across Regions, 1880–1910

184

7-6. Racial Differences in School Quality in the South, 1890–1950

188

7-7. Black–White Income Gaps, 1774–2010

190

8-1. Growth of the Fifteen- to Sixty-Four-Year-Old Population, 1870–1970

210

8-2. Average Years of Schooling for the Fifteen- to Sixty-Four-Year-Old Population, United States, 1930–2010

213

8-3. The School Enrollment Histories of the Population of Ages Fifteen to Sixty-Four in Selected Countries, 1900–1940

215

9-1. Growth of the Fifteen- to Sixty-Four-Year-Old Population per Decade, 1910–2010

229

9-2. Average Years of Schooling for the Population of Ages Fifteen to Sixty-Four, in Twenty-Three Countries, 1960–2010

231

10-1. American Real Income per Capita, Relative to Great Britain, 1700–2010

252

A-1. Six Examples of Calculating Household Incomes in 1774 (Incomes in Dollars at $4.44 per Pound Sterling)

267

LIST OF ILLUSTRATIONS

•

xi

A-2. Free Farm Incomes in 1774: Baseline Results and an Alternative (Dollar Values in Thousands)

273

A-3. Main Data Inputs for 1800 Income Estimates

276

C-1. Slave Household Size in 1860

290

C-2. Slave Retained Earnings by Region, for 1850 and 1860

294

C-3. Slave Consumption Components in 1800 (Mancall, Rosenbloom, and Weiss)

298

C-4. Total Consumption of Food, Fuel, Shelter, and Clothes per Slave Worker in 1800

299

C-5. Slave Earnings Retention Rates, Based on Hire Market Data for 1796–1804

301

D-1. Ratios of American to English Prices, Twenty-Two Commodities, 1640–1774

304

D-2. Ratios of American to English Prices, Thirty Commodities, 1792–1808

306

D-3. Ratios of American to English Prices, Thirty Commodities, 1840–1875

308

D-4. American versus British Costs of a Respectability Bundle of Consumer Goods, 1800–1858

310

E-1. Marginal Labor Products and Profit Residuals for Northern Farm Households, 1859–1860

312

F-1. Southern Farm Wage Rates by Race, 1879–1880

325

G-1. Marginal Products of Farm Labor and Pure Farm Profit (in Dollars) for the Non-South in 1870

329

G-2. Marginal Products of Farm Labor and Pure Farm Profit (in Dollars) for Tennessee in 1870

332

G-3. Twelve Kinds of Agricultural Census Returns for 1879–1880

337

G-4. Determinants of a Farm’s Own-Labor Income in the Ransom and Sutch Sample of Southern Farms, 1879–1880

343

xii

•

LIST OF ILLUSTRATIONS

G-5. Alternative Bridges to Tennessee: Three Different Sets of Marginal Products for a Tennessee Tenant in 1870

347

G-6. The Limited Sensitivity of 1870 Income Results to Different Income Intercepts for Tennessee Farm Operators

348

FIGURES 2-1. Household Heads by Region and Occupational Class, 1774

15

2-2. Comparing American and English Income Ranks in the Late Eighteenth Century

40

3-1. Population of British America, 1610–1780

55

3-2. Urban Share of America’s Population, 1650–1870

58

3-3. Purchasing Power per Capita in Thirteen American Colonies and Great Britain, 1650–1774

64

3-4. Workers’ Purchasing Power in England and the Colonies, 1650–1820 (Allen)

69

3-5. The Cost of Certain Necessities in England and America, 1632–1894

70

5-1. Nominal GDP per Capita, Britain and America, 1650–1870

105

5-2. Real GDP per Capita, Britain and America, 1650–1870

106

5-3. Income Inequality in America, Britain, and the Netherlands, 1732–2010

119

5-4. Top 1 Percent’s Share of Income in Four Countries in Recent Centuries

120

7-1. Blacks and Other Victims of Lynching in the United States, 1882–1964

186

7-2. Black–White Ratios of Earnings per Worker and Income per Capita, 1774–2010

191

LIST OF ILLUSTRATIONS

•

xiii

8-1. Shares of National Income Received by the Top 1 Percent, United States and Elsewhere, 1870–1970

197

8-2. Inequalities among Non-Elite Incomes and Wages, United States, 1910–2010

199

8-3. Relative Salaries of Financial Occupations, with Some Correlates, 1909–2006

201

8-4. The Relative Income per Person of Southern Regions, 1774–2010

204

8-5. Changing Concentration of Income at the Top versus Labor Supply Growth, 1920–1970

211

8-6. Education Attainment in the United States, 1930–2010

214

9-1. Share of National Income Received by the Top 1 Percent, United States and Elsewhere, 1950–2012

220

9-2. Female–Male Wage Ratios in the United States, 1820–2010

226

9-3. Female–Male Wage Ratios, Several Countries, 1967–2006

226

9-4. Changing Concentration of Income at the Top versus Labor Force Growth, 1920–1970 versus 1970–2010

230

9-5. Rates of Return for an Extra Year of School, for a Young Man in the United States, 1914–2005

233

9-6. Rise in Wage Inequality, 1977–2007, Compared with the Growth in Adults’ Schooling Attainment, in Ten Countries

234

10-1. Growth in the Working-Age Population, United States, 1774–2050

245

10-2. American Real Income per Capita, Relative to Britain, 1700–2011

253

10-3. American Income Leadership and Convergence, 1920–2011

255

10-4. Real Wages Relative to the United Kingdom, 1830–1988

256

A-1. Assembling the Parts of National Income in 1774

265

xiv

•

LIST OF ILLUSTRATIONS

PREFACE

W

e began this journey in 2009, more than six years ago. With Branko Milanovic, we were fi nishing up what we called our ancient inequality project, which had collected social tables over two millennia to explore trends in inequality in preindustrial times.1 As chapter 2 will describe, social tables started to appear in the seventeenth century as Englishmen like Sir William Petty and Gregory King began to construct them to inform policy. As we amassed these “ancient” social tables, we kept asking why smart colonialists like Benjamin Franklin hadn’t done the same for British America since they copied so much else that was English. But Franklin and his colleagues hadn’t, so we decided to fill the gap. We were familiar with this method and we had written on American inequality history more than thirty-five years ago.2 Early social tables make it possible to extend the history of American incomes back before 1914, the well-worn starting point used by previous economic historians, all the way to the colonial era. No doubt we have made errors along the way. In a review of Thomas Piketty’s now-famous Capital in the Twenty-First Century, Nate Silver had this to say: It can be tempting to assume that the information contained in a spreadsheet or a database is pure or clean or beyond reproach. But this is almost never the case. All data . . . [are] subject to human error.3

Silver’s dictum surely applies to the innumerable data and calculations that have gone into our own study. For starters, many of the original data were gathered in a way that was not meant for the kind of analysis we 1

Milanovic, Lindert, and Williamson 2011. Lindert and Williamson 1982, 1983b; Williamson and Lindert 1980a. 3 Silver 2014. 2

have attempted. Some of the original numbers are probably themselves in error. We and other scholars may have transcribed the numbers incorrectly. And our calculations are based on many assumptions— some of them rather courageous. We have, of course, presented detailed appendixes (in this book) and Internet files (at http://gpih.ucdavis.edu, hereafter cited as http://gpih) showing some of our data and the steps in our calculations. These are certainly not easy reading, and we may not have succeeded in getting every file’s calculations revised to reflect all the updates in other files. But we have tried. For reasons like these, the probability that all of our calculations are error free is zero. On a more positive note, we also think there is a low probability that any of the main findings we announce in chapter 1 are false, though the quantitative extent of their truth is probably subject to error. We will in the future—to the best of our ability—revise the Internet files whenever necessary. Crucial to the development of our estimates has been the financial assistance of the US National Science Foundation, under grants 0649062, 0922531, and 1227237. These grants allowed us to hire superb research assistants, with the greatest contributions being those of Leticia Arroyo Abad, Sun Go, Oscar Méndez Medina, Brock Smith, and Nick Zolas. Archival help was supplied by, among others, Jan Kinzer (Pennsylvania State Archives), Diana McCain (Connecticut Historical Society), Clifford C. Parker (Chester County Archives), and Marc Thomas (Maryland Historical Society). Along the way, we have accumulated many debts to other scholars who have been generous with their time, advice, criticism, and resources. In the early going, Tom Weiss was a constant adviser on the colonial and early federalist years. Gloria Main then helped us immensely by sharing her huge New England probate data file, which had been shelved away for decades. As we got deeper into the project, Allan Kulikoff shared his extensive knowledge of colonial Chesapeake, Bob Allen shared some of his unpublished annual “bare-bones” price data, and Richard Sutch shared his impressive knowledge of slavery and the post-emancipation era, and even the recent Ransom and Sutch farm sample. We accumulated many other debts to the following: Jeremy Atack, Samuel Bowles, Steve Broadberry, Trevor Burnhard, Lois Carr, Paul Clemens, Peter Coclanis, Bill Collins, Lee Craig, Paul David, John xvi

•

PREFACE

Devereux, Stanley Engerman, Joseph Ferrie, Frank Garmon Jr., Henry Gemery, Claudia Goldin, Farley Grubb, Michael Haines, Douglas Irwin, Louis Johnston, Herbert Klein, Robert Margo, Branko Milanovic, Alan Olmstead, Roger Ransom, Paul Rhode, Joshua Rosenbloom, Winnifred Rothenberg, Peter Rousseau, Carole Shammas, Billy Gordon Smith, Richard Sylla, Daniel Waldenström, Lorena Walsh, Marianne Ward, Samuel Williamson, and Gavin Wright. We have also benefited from feedback on portions of our work when presenting in various forums: All-UC Conference in Economic History (Pasadena, 2015), American Historical Association Annual Meetings (Washington, DC, 2014), Bank of England (December 2014), University of California at Berkeley, Cliometrics Conference (Clemson, 2014), Economic and Business History Annual Meetings (LaCrosse, WI, 2015), European History Economics Society Annual Meetings (Dublin, 2011; Pisa, 2015), International Institute for Social History Workshop (Amsterdam, 2012), Latin American RIDGE Program (Montevideo, 2015), London School of Economics, NBER Development of the American Economy Meetings (Cambridge, MA, 2012, 2014), Oxford University, Stanford University, University of Michigan (October 2015), University of Warwick at Venice (2011 and 2014), University of Wisconsin, and the World Economic History Congress (Stellenbosch, 2012). We apologize in advance to those who want the terms “America” and “American” reserved only for the collection of continents and islands named after Amerigo Vespucci since 1507. For their rhetorical convenience, we use the adjective “American,” and occasionally the noun “America,” in the narrowly conventional sense. Our reference here is to the thirteen British colonies on the North American mainland that with independence became the United States of America. Our coverage omits the rest of North America, and also omits Central and South America and the Caribbean—that is, the rest of “the Americas,” New World, or Western Hemisphere. The convenience of doing so stems largely from the problem of settling on a readable adjective. We anticipate that readers would want the adjective “US” to be used only sparingly. The English language has failed to come up with a reader-friendly counterpart to the Spanish adjective estadounidense, which itself requires a burdensome seven syllables. Hence, the book uses “American” frequently and also sprinkles in references to “America.” PREFACE

•

xvii

UNEQUAL GAINS

CHAPTER 1

Persistent Debate, a New Approach, More Data, Rich Findings

H

ow and when did Americans become so prosperous and so unequal? Generations of Americans have debated competing visions of what was happening to national income and how it was divided. Yet they lacked the solid evidence needed to choose between the competing visions. We still know little about the growth and (especially) the inequality of American incomes before the twentieth century. We also need to understand what has caused the dramatic movements in inequality over the last century, and its future prospects. When did America grow fast enough to make it a world leader in average living standards? There is little disagreement about how American incomes have grown since the early twentieth century and even as far back as 1870, thanks to the pioneering work of Simon Kuznets and many others. But income estimates are weak for the years before 1870—weaker than in some western European countries. To be sure, others before us have struggled admirably to reduce uncertainties about the pre-1870 history of national income.1 Yet the debate continues about income levels and growth before and during the Civil War. And even those estimates that most people agree on probably give the wrong impression about how American incomes compared with those of other countries. Our history textbooks imply that the road to prosperity was paved by the institutional wisdom of the country’s founding fathers and those who refined that wisdom over the two centuries that followed. While those institutions were well chosen and 1 For the current state of knowledge about the history of US gross domestic product (GDP) and national income, see Richard Sutch’s encyclopedic coverage in volume 3 of Historical Statistics of the United States (Carter et al. 2006, vol. 3, chap. Ca) and the sources cited there.

largely well borrowed, this book will show that America had reached world leadership in living standards long before the country’s founding fathers constructed their new republic. We will also find that the road to prosperity was far bumpier than the standard, benign tale of American economic progress implies. How unequally was income distributed between the rich, middle, and poor, and why? The steep rise in inequality since the 1970s is now unmistakable. New measures of inequality avoid the faulty official numbers that hid most of the true movements in the incomes of the richest. Since the 1960s, the official US Census Bureau estimates have badly understated top incomes and (unintentionally) hidden much of the rise in the share going to the richest 1 percent.2 Fortunately, an international research team led by Anthony Atkinson, Thomas Piketty, and Emmanuel Saez solved that twentieth-century problem. Starting from income tax returns, this team has charted the dramatic twentieth-century fall and rise of top incomes in countries around the world. Their evidence, however, is only available for the twentieth century. Since the US income tax was only introduced in 1913, there is still no history of American top income inequality for the centuries before, though economic historians have certainly offered many plausible guesses.3 Why does all this new twentieth-century American evidence matter, and why do we think it’s necessary to use new methods to mine the thinner evidence documenting the three centuries before? The answer is that two fundamental questions important for policy debate have 2 The main shortcoming, though not the only one, of the official Census Bureau estimates based on its Current Population Survey (CPS) is known as the “top-coding” problem. The Census Bureau knew it would be sensitive and difficult to ask top income people about the exact magnitude of their incomes. The official response to this difficulty has been a top-coding solution that seriously understates top incomes. As others have pointed out in Congress and the media back in the 1990s, the census values all top household incomes at the floor of that top income class. That floor was only $50,000 for 1967–1976, then $100,000 for 1977–1984, $300,000 for 1985–1992, and $1 million after 1993. The official CPS estimates imply that between 1980 and 1997, Bill Gates of Microsoft earned less than $8 million—from which he somehow accumulated a personal net worth valued at over $36 billion in 1997 (Newsweek, August 4, 1997, 49– 50). Worse yet, the published official CPS figures display even lower top-class cutoffs, frustrating any attempt to view what has happened within the top 5 percent of households. 3 For a summary of what is now known about American inequality in the twentieth century, see Atkinson, Piketty, and Saez 2011. Rough numbers on American inequality before the twentieth century were summarized more than three decades ago by the two of us (Williamson and Lindert 1980b, chaps. 2– 4), with updates by Lindert (2000).

2

•

CHAPTER 1

been left unanswered by American history. First, does modern economic growth inevitably drive up inequality? And second, does inequality favor or disfavor growth? This book speaks to those two questions by exploring American incomes since the 1600s.

A DIFFERENT APPROACH WITH NEW DATA New Evidence, Helped Greatly by Four Scholars Information about the distant past keeps growing, thanks to advances in archival recovery technology. The leading estimates of nineteenthcentury American gross domestic product (GDP) date from pioneering research in a great quantification wave from the 1960s through the late 1980s. Nineteenth-century evidence on inequality and growth did not advance so quickly, but the same quantification wave did give us new impressions about colonial American inequality. Those impressions were still limited by incomplete evidence on the distribution of wealth, property income, labor earnings, and thus total income. Since then, several new sources have become available that this book exploits—new evidence supplemented by some old sources that have been underutilized in the past. The new evidence did not appear by some official release of long-locked archives. Rather, it came from the previous efforts of several others. We are delighted to acknowledge their labors before describing our own method of extracting a new income history from the mass of information they have patiently extracted. Our American incomes history has benefited especially from the contributions of four scholars. The landmark study of American wealth around 1774 by Alice Hanson Jones had already appeared by 1980 and launches our new income history in chapter 2.4 Jackson Turner Main scoured the colonial archives and delivered much of what we know about rates of pay on the eve of the revolution.5 As far as we know, we are the first to mine systematically the numbers in his Social Structure of Revolutionary America. Gloria Lund Main, first with her late husband and then on her own, wrote widely on colonial American wealth inequality. Central 4 5

Jones 1977, 1980. J. Main 1965.

PERSISTENT DEBATE

•

3

to our chapter 3, she has just made available—in machine-readable form—their rich sample of New England probates from 1631 to 1776. Finally, Steven Ruggles, director of the University of Minnesota Population Center, leads the continuing development of the Integrated Public Use Microdata Series (IPUMS), which has revolutionized the use of past censuses. One of its many accomplishments is the set of 1  percent samples of the US population censuses from 1850 onward. Our new history of American incomes reported in chapters 5 and 6 would have taken vastly longer to research without the 1850, 1860, and 1870 IPUMS samples of individual wealth, occupation, location, and other attributes. Building Social Tables on the Income Side Armed with new evidence, this book applies a different approach to the historical estimation of what Americans have produced, earned, and consumed. National income and product accounting reminds us that we should end up with the same number for GDP by assembling its value from any of three sides—the production side, the expenditure side, or the income side. All previous American estimates for the years before 1929 have proceeded from either the production side or the expenditure side. Taking the production route, others have assembled real GDP by applying fi xed base-period weights to time series of such physical output indicators as bushels of grain harvested, pigs slaughtered, yards of cloth woven, bricks used to build houses, and service workers employed.6 These weighted output trends are then applied to some benchmark year, where the evidence is thick enough—like an early census—to build what are hoped to be solid estimates for that year. Here the leading historical extension has been the pioneering work of Robert Gallman, who provided annual estimates back to 1834.7 Paul David used his “controlled conjectures” to push aggregate output back to 1800, and more recently, Thomas Weiss and his collaborators have 6 Production-side estimates of GDP or gross national product (GNP) before 1929 include the following series in Historical Statistics of the United States (Carter et al. 2006, hereafter HSUS): the Millennial Edition series Ca9–17 back to 1790, by Richard Sutch and others; the Balke and Gordon as well as Romer series Ca208–18 for 1869–1929, which can be traced back via Kuznets to William Howard Shaw’s (1947) commodity output series by sector. 7 See the Gallman estimates for 1834–1909 in HSUS series Ca192–207, Ca219–32.

4

•

CHAPTER 1

used the same method to push the aggregates back into the colonial era.8 The second leading approach to GDP estimation before 1929 has taken the expenditure side, adding up estimates of household consumption, capital formation, government expenditures, and the difference between exports and imports. The production and expenditure approaches have helped support each other by using much the same data from federal censuses. We work instead on the income side, constructing nominal (current price) GDP from free-labor earnings, property incomes, and (up to 1860) slaves’ retained earnings (that is, slave maintenance or actual consumption).9 What are called “social tables” are built up to income aggregates from occupation and location (described at greater length in chapter 2) in the “political arithmetic” tradition spawned by such Englishmen as Sir William Petty and Gregory King in the seventeenth century.10 Development economists will recognize a similarity between our social tables and their social accounting matrices. We have built five social tables for the benchmark years 1774, 1800, 1850, 1860, and 1870— years where the data are most plentiful. No such income estimates were available for any year before 1929 until now. Our different approach leads to rewards not attainable by sticking to the production or expenditure side. One reward is the chance to confront and challenge the production-side estimates using very different data, sources, and methods. The production side and the income side should add up to the same GDP total, once one either multiplies the 8 David 1967, 2005; Weiss 1993a, 1994; Mancall and Weiss 1999; Mancall, Rosenbloom, and Weiss 2003; Rosenbloom and Weiss 2014. 9 We use the terms GDP, gross national income (GNI), and household income interchangeably in this book. This rhetorical convenience violates some accounting conventions, but it seems harmless given that the different national income and product aggregates are so close in magnitude and concept. For a discussion of why these different measures come out nearly equal today and almost exactly equal before the twentieth century, see appendix A. 10 For previous uses of this approach, see Lindert and Williamson 1982, 1983a; Milanovic, Lindert, and Williamson 2011. We are preceded by at least two early Americans who imitated Petty and King with their own calculations of what their region was worth—presumably to estimate its ability to pay taxes and fight wars. Colonial governor James Glen of South Carolina made an imaginative social table for his colony in 1751 (cited in McCusker 2006), and Samuel Blodget (1964, 99) made another social table a half century later for the United States as a whole. Both Glen and Blodget started with occupations and/or social classes in building their social tables, and in so doing, appear to have been readers of the English political arithmeticians, whose writings multiplied with the growing need to finance wars. On the rise of the quantification culture in late eighteenth-century England, see Hoppit 1996.

PERSISTENT DEBATE

•

5

production side’s real GDP estimate by a price index or divides the income side’s nominal GDP by the same price index. As we will see, some instructive tensions arise between the two kinds of estimates, exposing the need to rethink the index-number alignment of real GDP and its price deflator. An even bigger reward from using the income approach is that it exposes how income was distributed by socio-occupational class, race, and gender as well as by region and urban–rural location. Furthermore, our income approach allows us to travel deeper into the past than just 1790, 1834, or 1870. Our estimated social tables capture the distribution of national income going back to the colonial era. In addition, we can break that distribution down into component parts—skill premiums, urban–rural wage and income gaps, regional inequality, earnings inequality, property income shares, and property income distribution— thus better to understand the determinants of aggregate income inequality and its change over time. Why Not Wealth or Capital? We elect to chart a new history of American incomes rather than revisit the history of American wealth. Our choice of income, as opposed to wealth or capital, may seem surprising given that it comes in the wake of Piketty’s best-selling book Capital in the Twenty-First Century and his article with Gabriel Zucman proclaiming that “capital is back.” Piketty dwells at great length on wealth inequality and the ratio of wealth to GDP. We have explored the history of American wealth before, so why not return to it with new data to address the debate that Piketty has reignited?11 The answer is that wealth is an incomplete measure of one’s lifetime resources. It only includes nonhuman assets, missing the investments people make to augment their earnings capacity—formal education, on-the-job training, health, and migration. At this point, we need to emphasize that the inequality we should really care about is the distribution of lifetime resources, as shared within a household. It can be measured either as an inflow, by lifetime earnings 11 See Piketty 2014; Piketty and Zucman 2015. For our earlier history of American wealth inequality, see Williamson and Lindert 1980a, chap. 3.

6

•

CHAPTER 1

plus inheritance, or an outflow, by lifetime consumption plus bequest. For most people, any calculation of their lifetime income reveals the quantitative dominance of labor earnings or consumption flows, not wealth. Earnings from accumulated skills, ability, and effort account for more than half of lifetime incomes, and wealth misses this.12 More important, the modern distribution debate has failed to note the fact that inequality of lifetime income must have been reduced dramatically by rising life expectancy, as we have pointed out some years ago.13 Since the nineteenth century, the spectacular decline in infant and child mortality—and that of young mothers in childbirth—has not only improved average life expectancy but also caused a spectacular convergence in those rates across income classes. This point is never introduced into inequality debates. While we do not fully approve of the narrower focus on current income, we follow the convention in this book so that our new evidence can be compared with that of others. Granted, studying wealth does have some practical benefits. First, the study of wealth inequality is a useful prelude to the study of inheritance—an issue worth public debate. In addition, data on household wealth offer clues about income inequality in earlier centuries, when direct income measures are sparse. Yet because we have found a way to trace the long history of income levels and income inequality, we can afford to set aside a separate, and narrower, discussion of the distribution of wealth. Three Things Left Out This book omits three things that matter. It excludes Native Americans—a big part of the colonial population—since information on their living conditions is simply too limited.14 Second, our seventeenthand eighteenth-century analysis only covers the thirteen mainland British colonies, ignoring the West Indies, Canada, and all other North American settlements. Third, and most important, we see no way to 12 While a few authors have tried to estimate lifetime income inequality (e.g., Lillard 1977), it is much too large a task for this book. We must be content with a short-term proxy for life-span inequality—current income. 13 Lindert and Williamson 1985, 347. 14 See, however, the conjectured incomes per capita for Native Americans in the lower South, 1720–1800, in Mancall, Rosenbloom, and Weiss 2003, table 4.

PERSISTENT DEBATE

•

7

place any monetary value on the freedom that slaves were denied. Nor can their inhumane treatment be quantified. Only slave consumption is measured here—a much narrower concept than their well-being. The last of these qualifications deserves particular emphasis since so much of this book will deal with income inequality. While we will stress that the distribution of American incomes was strikingly “equal” or “egalitarian” before 1790 or 1800 and the start of modern economic growth, this evidence should be understood to mean “equal in income” or “income egalitarian.” A society with slavery should not be viewed as egalitarian in any broader sense. Though we use a different income-building approach than have others, our estimates should be viewed as part of a research tradition that David so aptly described as “new evidence and controlled conjectures.” Our estimates use new evidence that was not available when others were writing on this topic, and we offer them only as controlled conjectures, since they are laden with explicit assumptions about information that is still lacking from the historical record. Far from claiming closure, we present the implications of currently available evidence, awaiting revision as more and better evidence accumulates.

NEW FINDINGS Our new approach and new data yield a rich harvest of new findings. They are: • American world leadership in income per person has waxed and waned for centuries. Before the twentieth century, the period in which Americans most clearly led Britain and all of western Europe in purchasing power per capita was during colonial times—that is, when North Americans were still British. They were already ahead by the late seventeenth century. America lost that lead in the Revolutionary War and the Articles of Confederation years, gained it back by 1860, lost most of it again in the Civil War decade, gained it back once more by 1900, and briefly lost it again in the Great Depression of the 1930s. Angus Maddison’s

8

•

CHAPTER 1

claim that American income per capita did not catch up to that of Britain until the start of the twentieth century seems to be off the mark by at least two centuries. Over the whole span of over 360 years since the mid-seventeenth century, America’s income advantage over Britain has not increased and may have decreased slightly. The only historical moment in which the United States soared far ahead of the rest of the world in average income came at the end of World War II. Since then, western Europe and Japan have been growing faster than the United States in terms of incomes per person. • Demography mattered from the start. American colonists probably had the highest fertility rates in the world, and their children probably had the highest survival rates in the  world. Thus, the American colonies had much higher child dependency rates and family sizes than did Europe, and even higher than does the Third World today.15 What was true of the colonies was also true of the young republic. It follows that America’s early lead in income per capita was exceeded by its early lead in income per household or per worker. • The colonial era saw little growth per capita, because extensive growth in the poorer hinterland offset intensive growth on the richer coast. Our evidence supports the slow- or no-growth side of the colonial growth debate. This is not a “pessimist” result, however, since it is consistent with more than a century of relative prosperity based on a growing colonial supply of primary products to Atlantic markets and the rapid expansion of an interior poorly integrated with those markets. It was a dualistic economy, with the richer coastal strip producing high-value exports and undergoing intensive growth, and with an interior producing a high level of subsistence (or what colonial historians have called “subsistence-plus”) and undergoing extensive development. The interior won the colonial population race, bringing de-urbanization over the century up to independence. 15 We are talking about free whites here, but we suspect the same was true of the mainland slave population, especially inland from the southern coast, and especially by the mideighteenth century.

PERSISTENT DEBATE

•

9

• The southern reversal of fortune started early. The South’s relative income per capita fell for at least two hundred years between 1670 and 1870, starting from its being clearly the richest part of the thirteen colonies—even when slaves are counted as lowincome residents.16 The South’s relative decline had multiple causes, but prominent among them were diminishing returns to land as the frontier pushed inland, Revolutionary War losses, declining export prices after independence, Civil War losses, and its failure in the nineteenth century to provide public education, even for free whites. • Independence was costly. The Revolutionary War and the dysfunctional confederation years were costly. The large American colonial per capita income lead over Britain was lost by 1800. The causes seem clear: war damage, mortality and morbidity among young adult males, the destruction of loyalist social networks, a collapse of foreign markets for American commodity exports, hyperinflation, a dysfunctional financial system, and much more. The per capita income loss up to 1790 may have been as large as 30 percent. • The young republic was a modern economic growth leader. From 1800 (and probably even from 1790) to 1860, American per capita incomes grew much faster than in western Europe, well above Kuznets’s criterion for modern economic growth (more than 1 percent per annum over many decades). This was a period of catching up with and overtaking the average income of western Europe, including that of Britain. Fast per capita income growth and even faster population growth quickly made the American economy by 1860 one of the biggest in the Western world.17 • America’s first great rise in inequality was as big as the rise since 1970. There was a long steep rise in American inequality between 1800 and 1860, matching the widening of income gaps we have lived through 16 The South was probably not the richest of all the British American colonies in terms of free white incomes. What little we know about white wealth and indirectly about income suggests that white incomes in the British West Indies were higher than in any mainland colony. See McCusker and Menard 1985, 61, table 3.3; Higman 1996, 321–24; Burnhard 2001. But income per capita of all—white and black—was probably higher on the mainland since the percent slave was much lower than in the West Indies. 17 Maddison (2010) reports the 1860 GDP of the United Kingdom as the biggest in the West, with France at 86.3 percent and the United States at 84.8 percent of the leader. Our aggregate income revisions suggest that the US economy was probably the second largest by 1860.

10

•

CHAPTER 1

since the 1970s. The earlier rise was not dominated by a surge in the property income share, however, as has been true since the 1970s. Rather, the fi rst great rise in inequality was even more broadly based, with a widening of income gaps throughout the whole income spectrum—urban–rural income gaps, skill premiums, gaps between slave and free, North–South income gaps, earnings inequality, and even property income inequality. • The Civil War maintained income inequality overall, despite its equalizing incomes. Income inequality rose in the North during the 1860s, continuing the long upward march that started with the creation of the republic in 1789. But emancipation and defeat greatly reduced southern inequality. The widening within the North and the widening between the North and the South were sufficient to offset the massively egalitarian redistribution within the South, thus keeping income inequality from falling at the national level. • Emancipation raised black incomes by about 30 percent. The big rise in black incomes is apparent despite the predictable decline in black labor force participation. Emancipation meant that the freed slaves could now capture something close to their marginal product rather than the 40 to 50 percent of it that slavery permitted. As of 1870, African Americans of a given age and sex still worked more days per year than did their white counterparts, but many fewer than under slavery. • The Great Leveling between the 1910s and 1970s offered America a second chance to start with great equality. For almost all countries supplying the necessary data, the income share captured by the richest 1 percent fell dramatically between the 1910s and 1970s, and that of the bottom half rose, for several reasons. Wars and other macro-shocks destroyed private wealth (especially financial wealth) and shifted the political balance toward the Left. The labor force grew more slowly and automation was less rapid, improving wages for the less skilled. Rising trade barriers lowered the import of labor-intensive products and the export of skill-intensive products, favoring the less skilled in the lower-middle ranks and at the bottom. And in the United States, the financial crash of 1929–33 was followed by half a century of tight financial regulation, which held down the PERSISTENT DEBATE

•

11

incomes of those employed in the financial sector and the net returns reaped by rich investors. • The second great rise of American inequality after the 1970s was probably avoidable. The equality gained during the Great Leveling slipped away after the 1970s in North America, the United Kingdom, and Australia, while inequality hardly changed at all in continental western Europe. These countries’ new income gaps were partly due to inegalitarian policy shifts. The United States lost its lead in the quantity and quality of mass education, and its gaps in educational achievement have widened relative to other leading countries. Financial deregulation has also worked poorly in the United States since 1980. In addition, a regressive pattern in tax cuts has allowed more wealth to be inherited rather than earned. All three of these shortfalls—in basic education, financial regulation, and the taxation of heritable wealth—are potentially reversible, without any clear loss in GDP. • A booming, unregulated financial sector contributes to inequality. Unregulated financial growth amplified the two great rises of American inequality: the longer, slower rise before 1910, and the shorter, faster one since 1970. The intervening Great Leveling era saw financial crises, stiff financial regulations, a fall in incomes of those employed in finance, and a drop in top income shares. The correlation between high finance and inequality is not spurious. Skilled individuals with financial knowledge have been well rewarded during the two booms and heavily penalized during the one slump. • There is no fundamental law driving the history of income inequality. Inequality movements are driven not by any fundamental law of capitalist development but instead by episodic shifts in six basic forces: politics, demography, education policy, trade competition, finance, and labor-saving technological change. These forces appear to be exogenous with respect to inequality. If they are indeed exogenous and hard to predict, then four centuries of American inequality can hardly have been driven by some capitalist law of motion.

Now then, how did we arrive at these conclusions?

12

•

CHAPTER 1

CHAPTER 2

Colonial Incomes on the Eve of the Revolution

F

or the year 1774, on the eve of the revolution, we now have better information about incomes and wealth than for any other year before the middle of the nineteenth century. This chapter uses that information surrounding the 1774 benchmark year to reopen a host of questions about people’s relative fortunes. Was America rich or poor compared with Britain and the rest of the western European leaders? Were colonial incomes distributed more equally than in western Europe, or did slavery make the American colonies more unequal? Which were the economically leading colonies, and which were the laggards? Part of our motivation to construct new estimates of American incomes around the time of the revolution arises from an urge to compare them with some long-established views. Maddison estimated that it was not until around 1900 that the United States caught up with and surpassed the United Kingdom in real GDP per capita, although active debate has ensued about the quality of his estimates and that of others.1 Maddison’s assertion needs to be reconciled with the fact that North America attracted so many migrants from the mother country and that colonial population growth was much faster than anywhere else (as Benjamin Franklin wrote in 1751, a half century before Thomas Robert Malthus did). In the light of current research on the history of European incomes and the continued use of the Maddison world income estimates, we think the time is ripe to add new data from America to be compared with the new estimates for Europe.2 1 Maddison published many world GDP and GDP per capita estimates. Here we rely on Maddison 2007. 2 The income estimates in this chapter are reported, and defended, in much greater detail in appendixes A– C and the supplementary materials relating to our published article (Lindert and

We will find higher colonial per capita incomes in 1774 than did previous scholars, especially for the southern colonies. In 1774, average colonial incomes were considerably higher than those in England and Wales, using either exchange rates or purchasing power parity calculations, although, as we will see in chapter 4, American per capita incomes were almost certainly lower than English incomes in 1800. Chapter 4 will raise new questions about what happened during the Revolutionary War and the early years of the young republic. The 1774 estimates will also provide a much clearer view of colonial American inequality and how the incomes of different classes compared with those of their counterparts in England. We will find that American inequality in 1774 was much lower than in England. Inequality was also much lower in New England and the middle colonies than in the United States today. Before we can resume exploring these 1774 findings (page 28), we must describe how we got them. HOW THE 1774 ESTIMATES ARE DERIVED We start by counting people by occupation or social class, and then muster evidence about their average incomes.3 That is, we build national income and product accounts (NIPA) from the income side. Historians will recognize our approach as that of building what are called social tables, in the political arithmetick tradition, as mentioned earlier, spawned by such Englishmen as Petty and King in the seventeenth century. A snapshot of that society appears as figure 2-1. Here we show the relative size of social classes in the cities and countryside of each colonial region, where the classes are ranked from top to bottom by average income per household. An outstanding feature of the colonial social structure was the dominance of farmers and planters. In 1774, the American colonies were still a small farming periphery of the Atlantic Williamson 2013a) on the Journal of Economic History website. The underlying quantitative fi les are downloadable from http://gpih.ucdavis.edu, within the folder “American Incomes ca 1650– 1870.” These fi les are hereafter cited by their fi le name, followed by http://gpih. 3 A condensed step-by-step guide to the 1774 estimates appears as “Appendix A: A Guide to the 1774 and 1800 Income Estimates,” both in this book and at the gpih website. Like this chapter, it gives citations to the more detailed data fi les.

14

•

CHAPTER 2

New England

Middle Colonies

South

Urban, all Rural white collar Farmers and planters

Rural artisans

Farmers and planters

Farmers and planters

Rural artisans Rural laborers

Rural, no occupation given

Rural, no occupation given

Slaves

Slaves Figure 2-1. Household Heads by Region and Occupational Class, 1774

economy. Their urban populations were tiny. Philadelphia, the largest city, had only 42,500 residents, New York had only 32,300, and Boston only 16,000—versus about 50,000 in Leiden, a European city that ranked only fortieth in size there, and with London ranked number one with 2.3 million in 1750.4 In fact, the society sketched in figure 2-1 was even more dominated by rural farming around 1774 than it was in the late 1600s. Chapter 3 will take a closer look at this de-urbanization, or ruralization, across those colonial years. Filling in this social profile of America on the eve of the revolution has taken us to local censuses, tax lists, occupational city directories, probates, and other archival sources, all supported by the earliest national population censuses of 1790 and 1800. Fortunately for us, the recent electronic revolution has made local enumerations from the late eighteenth century much more accessible. While all records before 4

Hohenberg and Lees 1985, 227, table 7.2.

COLONIAL INCOMES IN 1774

•

15

1790 were local, aggregate regional counts can be developed by assuming that proportions from one documented locality represent those of other localities in the same region, with the same population density, urbanization level, and qualitative attributes. Counting early Americans by work status, location, and living arrangement starts from basic population totals, and then converts them into labor force estimates by applying labor participation rates by age, sex, and slave/free status. That labor force is then allocated by occupation and household headship status. Population census counts. The few local censuses from the colonial period are now collated and referenced in the colonial section of Historical Statistics of the United States.5 These offer population detail by age, sex, race, and free/slave status for seven colonies, and we clone the demography of the six missing colonies from these seven. Labor force participation rates (LFPR). Next we derive labor force participants in each demographic group by relying on LFPR estimates from the early nineteenth century. Following convention, the labor force consists of all persons producing goods and services sold in significant part (or for slaves, demanded in significant part) outside the household. To convert the 1774 population into labor force, we use the detailed labor participation rates for 1800 supplied by Weiss. It seems reasonable to assume there were no behavioral changes between 1774 and 1800 in the rates defined in the detailed, cell-specific Weiss estimates, which give separate rates for detailed categories such as urban Pennsylvania’s free white females age ten to fifteen, rural South Carolina’s male slaves over the age of ten, or small-town Connecticut’s free white males age sixteen and older, and so on. Since these categories changed in relative importance over time, however, the regional and national labor participation rates could and did change between 1774 and 1800. Recorded occupations. Constructing the social makeup of the labor force requires detailed occupation counts for different localities. We draw on newly accessible counts for years near 1774, although only for a few places, only for parts of the labor force, and only with the help of some comparison of occupational mixes over time and space. 5 They are in both the bicentennial (US Census Bureau 1976) and millennial editions (Carter et al. 2006).

16

•

CHAPTER 2

Our new reconstruction of the social structure of America on the eve of the revolution uses local tax assessment lists and occupational directories, as reported in table 2-1. Such lists allow us to create the following occupational groups for the free population: Group 1 = officials, titled, professionals Group 2 = merchants and shopkeepers Group 3 = skilled artisans in manufacturing Group 4 = skilled in the building trades Group 5 = farm operators (renters, sharecroppers, planters, owneroperators) Groups 6A and 6B = male and female menial laborers

The new data modify the occupational structure of the colonies constructed previously by others. For example, relative to the famous colony-wide probate samples taken and analyzed by Jones, our estimates shift a lot of the middle colonies’ labor force from middling farmers to less wealthy artisans and laborers, and males with no stated occupation.6 In the urban South, the 1790 directory for Charleston is used, after scaling the numbers back to the estimated Charleston 1774 population. One gets the same occupational patterns by starting with the weights for Jones’s sample drawn from four southern colonies. In either case, one must adjust for the overrepresentation of absentee landowners and slaveholders living off the plantation in town. We adjust the Jones weights, guided by some useful local censuses from three North Carolina counties in 1779–1782. These enumerate the whole population of free household heads (HHs) according to whether they held slaves or real estate, or both, or neither. We assume that the same adjustment of weights is required in Charleston as in the rest of the South. For the rural South, we carried out the same adjustment away from slaveholders and landowners, instead giving more weight to ordinary farmers. One could wish, of course, for a broader sample of the rural South than just the Jones sample from four colonies, plus our new sampling from the three North Carolina counties. There are other rural southern county assessment documents on the Internet, but only a few 6

Jones 1977, 1980.

COLONIAL INCOMES IN 1774

•

17

Table 2-1 Main Data Inputs for 1774 Income Estimates Data sources and adjustments for occupational shares A. Population and labor force

Local censuses and labor force participation rates for 1800 supplied by Thomas Weiss, expanding on his estimates in Weiss 1992.

B. Occupations of HHs and the labor force New England (CT, MA, NH, RI) Big city = Boston

Nineteen lesser cities and rural

Middle colonies (NY, NJ, PA, DE) Big city = Philadelphia, New York City Three New Jersey lesser cities Rural South (GA, MD, NC, SC, VA) Big city = Charleston*

Rural

• Boston 1780 shares in Main 1965; backed by Boston 1790 shares in Price 1974; downloaded Boston 1800 occupational directory. • Use the 1771 Massachusetts and Maine tax returns to estimate the shares of landowning farmers, non-land-owning farmers, and others with positive versus zero realty. Then for the towns, apply the nonfarm, nonbig-city occupation mix from Lancaster, PA, 1800 to lesser cities in 1774, and the Chester County, PA, rural occupation mix of nonfarmers in 1800 to the rest of New England in 1774. • Philadelphia 1772 occupations from assessment lists supplied by Billie Gordon Smith. • Lancaster Borough, PA, 1773. • Chester County, PA, 1800, nine rural townships. • Charleston 1790 directory, downloaded. Reweighted away from slaveholders and landowners, based on assessments for three North Carolina counties, 1779–1782. • Start with Alice Hanson Jones’s rural w weights from four colonies (MD, VA, NC, SC), and apply the same adjustment as for Charleston based on three North Carolina counties.

Table 2-1 (cont.) Data sources and adjustments for occupational shares C. Free-labor earnings

Secondary literature: J. T. Main, Stanley Lebergott, Carroll Wright, Bureau of Labor Statistics, T. M. Adams, Donald Adams, Winnifred Rothenberg, and others.

D. Slave retained earnings

Slave retained earnings by age and sex could be derived from any two of these three parameters: free wage rate for same occupations, expropriation rate, and the slave hire rate (Fogel and Engerman, for Queen Anne’s County, MD, 1796–1804). We used the latter two, cross-checked against the literature on slave consumption (e.g., Mancall, Rosenbloom, and Weiss 2001).

E. Property income

Applying a 6% rate of net return plus asset-specific depreciation to Alice Hanson Jones’s wealth details, reweighted using new occupational data.

* The earliest Baltimore occupational directory available is for 1799, and the earliest for Norfolk, Virginia, is for 1801. Both are posted on http://gpih.

are for dates earlier than 1798, and none of the lists we have seen record the occupations of the HHs. Unrecorded occupations. Persons with occupations recorded by tax assessment lists or city occupational directories fall short of persons in the labor force. In most cases they even fail to capture all HHs, with the exception of those three counties in rural North Carolina between 1779 and 1782, for which the listings seem to have captured all free HHs. Some labor force participants lack an occupational label despite a positive amount of assessed wealth. Some lack an occupational label, and are listed as tax exempt because they had zero or near-zero wealth. Thus, we distinguish between the following groups of HHs without recorded occupations: Groups 7 and 8 = free males and females with positive wealth but no recorded occupation

COLONIAL INCOMES IN 1774

•

19

Group 9 = free persons recorded as having zero or near-zero wealth, and no stated occupation, and others who are in the labor force but unlisted in local records

Counting households. One could avoid measuring household headship if one were only interested in measuring aggregate national income, since it depends only on who is in the labor force and their average income. We need the headship rates by occupation, however, to measure the distribution of income and thus inequality. Households are the income recipient units used here to measure income inequality, for both practical and theoretical reasons. Previous investigators have been forced to confront the simple fact that all household members use its property, such as real estate, even if only one is the owner and taxpayer. The prevailing practice is to measure income inequality among households, not among individual income earners. In order to compare apples with apples, we do the same. That’s the practical reason. The theory comes from Kuznets, who in 1976 warned against measuring inequality among individual earners and emphasized the superiority of the household focus. Caring about economic inequality means caring about how unequally people consume resources over their lifetimes. Even if data constraints force us to study annual incomes rather than life cycle incomes, Kuznets pleaded for the measurement of income per household member. The numerator should capture the incomes of all economically active household members, and the denominator should capture all adult-equivalent consumers in the household. The earliest population censuses usually did not count households, unlike the more helpful 1850, 1860, and 1870 censuses we will use in later chapters. Some assumptions must be invoked to count colonial HHs. Fortunately, historians of early America have already grappled with this issue. Following their lead, we estimate the number of HHs from population data around 1774 invoking the following assumptions:7 (1) All free white males, twenty-one-years old and up, were HHs, subject to (4) below.

7 In particular, we make heavy use of the work of Billy Gordon Smith (1981, 1984, 1990) and the late Lucy Simler (1990, 2007).

20

•

CHAPTER 2

(2) All free white widows with any indication of property ownership or occupations were HHs. (3) One-sixth of the free black population consisted of HHs.8 (4) The number of free white males, twenty-one-years old and up, who were not HHs is matched by the number of free white females, age eighteen and up, who were HHs, despite not being included in (2)  above. That is, we assume that the two errors offset each other when using the white males age twenty-one and up as HHs.

These assumptions generated total households by location—that is, by region and urban versus rural. By subtraction, we derive the number of HHs who are missed by the listed-occupation counts. The shares of heads omitted are often large when the occupational data come from the tax lists and city directories. As estimated in detail elsewhere, the colonial business directories and the tax lists may have missed more than 30 percent of all households.9 Left uncorrected, such counts would underestimate total income and inequality, if many of the unregistered were poor. In contrast, the tax lists from around 1800 are more likely to have captured something like the full population, or so it appears in our samplings from New York State property tax rolls that began around 1799. The same should have been true of the federal direct tax of 1798, which required a household enumeration subject to external audit. Still, we have found significant omissions in some tax lists and business directories from the colonial era. Which groups were most frequently omitted? The literature has advanced the plausible intuition that the omitted consisted mainly of the tax-excused poor. Yet there is also some evidence that many in the middling and rich groups may also have been omitted, or at least that their wealth was underassessed. We 8 The one-sixth assumption is supported by the somewhat-distant 1820 census—the earliest census to give an age distribution for free blacks. As of 1820, 24.3 percent of free blacks consisted of likely HHs, using the same assumptions as for free whites. We believe that the headship rate was probably lower in 1774, both because children were a higher population share of whites and slaves, and because fewer free black adults would have been able to establish separate households then. Hence, we opt for the one-sixth assumption for 1774. For an elaboration, see “Estimated Mix of Occupations 1774 by Region,” http://gpih. 9 See Excel fi le “Occs 1774 by Region,” worksheet (2), and the worksheets on our weighting in the three regional “Property 1774” fi les in the supplementary materials for our article (Lindert and Williamson 2013a) on the Journal of Economic History website.

COLONIAL INCOMES IN 1774

•

21

have used the tax assessments to help us divide the HHs of no recorded occupation into groups 7–9 listed above. Three questions remain about those who were in the labor force, according to the censuses and the Weiss estimates of LFPRs, yet who were not reported as HHs. First, how many of them were there for each place defined by region and urban/rural? Second, what kinds of occupations and earnings rates did they have? Third, in whose households did they live? Guided by the censuses, we identify the following groups in the labor force that were not HHs: Groups 10 and 11 = free white males and females age ten to fifteen Groups 12 and 13 = free white males and females age sixteen and up, but not HHs Groups 14 and 15 = free black males and females age ten to fifteen Group 16 = free black males age sixteen and up, minus free black HHs Group 17 = free black females age sixteen and up Group 18 = white indentured servants in Maryland, the only colony that labeled them separately in a census near 1774 Group 19 = slaves age ten and up (65 percent of whom were assumed not to be HHs)

Many of these groups contained laborers who were almost surely paid only unskilled wage rates, while others could have been spread over occupations of higher earnings. Our income estimates make the following assumptions within each location: Groups 10, 11, 14, and 15 (free non–household heads [non-HHs], age ten to fifteen) are allocated to groups 6A and 6B, menial or unskilled, by sex. Groups 12, 13, 16, and 17 (free non-HHs age sixteen and up) are allocated location specifically across groups 2– 6 in the same proportion as are groups 2– 6. Group 18, Maryland servants, are allocated across occupations that were relatively urban and skilled.10 Group 19, slaves age ten and up, uses other estimates of the amount of slaves’ earnings they were allowed to retain for consumption, specific to region and occupation. In the South, the rate varied between 41.4 10 The allocation follows David Galenson (1981) and Farley Grubb (1985). See Excel fi le “OwnLabor Incomes, 1774,” worksheet (3).

22

•

CHAPTER 2

percent for field hands to 52.7 percent for Charleston labor. We assume that slaves doing nonfarm work were artisans, construction workers, or unskilled (including servants).11 The nonfarm share unskilled is based on the Charleston 1848 census, while the residual is divided equally between artisans and construction workers.12

We must also decide in whose households these non-HH members of the labor force lived. The data are almost nonexistent on this issue. We make the following assumptions about the non-HH earners “imported” into the households of others: (1) For each region and urban/rural place (e.g., New England big cities or rural South), the non-HHs and their individual earnings are absorbed into the same region and place. (2) For the free population, within each group defined by region and urban/rural, we assume that the average earnings of each non-HH imported into free families is the same for all free persons of that occupation in that place. (3) Slave non-HHs are taken into slave households only, leaving household income the same as the retained earnings of all slaves. (4) This same assumption holds for the separately recorded group of Maryland servants, though the assumption is redundant here because these are one-person households.

These assumptions and the resulting allocations can certainly be challenged. But one point should be emphasized: for each place defined by region and urban/rural, the aggregate imports of non-HHs are driven by census information, labor participation rates, and household headship rates. The allocation of non-HHs to households by place cannot yet be derived from microstudies because there are still too few such studies. Labor earnings by occupation. Annual incomes can be assigned to the most ubiquitous occupations in each location, thanks to the enormous archival gleanings by many since Carroll Wright.13 Their time-consuming 11 These assumptions follow the work of Claudia Goldin (1976), Richard Wade (1964), Richard Sutch (1975a, 1975b), and others. 12 See http://gpih, Excel fi le “Own-Labor Incomes, 1774,” worksheet (4). 13 See “Wage Data c 1774,” http://gpih. The main sources are J. Main 1965; Lebergott 1964; C. Wright 1885; D. Adams 1968, 1970, 1982, 1986, 1992; T. Adams 1944; US Bureau of Labor Statistics 1929; Rothenberg 1985.

COLONIAL INCOMES IN 1774

•

23

collection of newspaper quotes and account book entries must be used with care. Some are in the depreciated local colonial currency, whereas others are in (British) pounds sterling. Fortunately, most sources, and Jack Main in particular, were careful to say which was which. Some of the earnings are annual, as for white-collar professionals and farmers, but others are monthly, weekly, or daily rates of pay, requiring assumptions about how many days or months they spent in gainful employment each year. We believe that for the days or months when a person did not hold their main stated job, they nonetheless fi lled in with other productive work, like weaving and farming at home, and some of this output was traded on the market. Thus, one could use our “full-time” estimates, which assume that daily or monthly full-time equivalent (FTE) workers performed productive work of some kind for 313 days a year (excluding only Sundays). This assumption implies, of course, that we include more nonmarket work in our income estimates than do others, who include only or mainly market work in their output estimates. The full-time assumption is especially relevant when looking backward earlier in the colonial era, when even large, market-oriented plantations in the South spent much of their labor time producing food, drink, fuel, and structures for the use of slaves, servants, overseers, and free families. As we move forward in time, all families, farms, and firms marketed larger and larger shares of their output, implying that unemployed labor was more likely to have been idle the more distant it was from 1774. Thus, as sensitivity analysis and to improve comparisons backward and forward in time, we will emphasize our 1774 “part-time” estimates that use fewer labor days per year. The alternative days worked annually assumptions that seem most plausible to us are: 313 days for those households with the head employed in the professions, commerce, and skilled manufacturing artisanal jobs, and slave households; 280 days for households with the head employed in construction trades, rural unskilled workers, and farm-operator households; and 222 days  for households headed by free, urban unskilled laborers and zero-wealth HHs of unknown occupation. Th is “sensitivity analysis” range is certainly wide enough to encompass such estimates for England: 1760 and 1771 averaged 278 days, and 1800 averaged 280

24

•

CHAPTER 2

days.14 For 1774, these part-time assumptions yield the following ratios of part-time to full-time total incomes (labor plus property): New England

Four middle colonies

South

Thirteen colonies

Free households

0.918

0.957

0.948

0.943

All households

0.919

0.958

0.954

0.948

These ratios imply that choosing between part- and full-time estimates could not change much of the gap between our income estimates and those of others for 1774 (see tables 2-3 and 2-4 below). Again, we will emphasize the part-time estimates in what follows because they are closer to the income concepts used by others.15 We enlarged the concept of labor earnings to include farm operators’ profits, plus slaves’ and indentured servants’ consumption, or what might be called the retained share of what they earned.16 As noted previously, we have called this labor income amalgam “own-labor incomes.”17 We conclude this section on labor earnings with an important reminder. The cells of our 1774 social table are fi lled with averages: we have little or no information on the variance of incomes among individuals within the cells, even though the cells are quite specific to gender, location, and occupation. Modern evidence documents considerable variance across individuals by gender, location, and occupation, driven as they are by age, health, background, and luck. Thus, the true distribution of income in 1774 may have been a bit more unequal than we estimate. The same must be said for our 1800, 1850, 1860, and 1870 income estimates reported later in this book, however, so comparisons over time are not affected much. It is also true that the western European income inequality estimates with which the American evidence is compared are similarly based on social tables with the same limitations, so comparisons across countries are not affected either. 14

For these English estimates, see Broadberry et al. 2015, 264, table 6.02. On the length of the work year in 1774, see “Wage Data 1774,” worksheet (7) on “1774 FTE Work,” http://gpih. For our handling of the work year by different occupations at all benchmark years from 1774 through 1870, see appendix B. 16 Here we rely on J. Main 1965. 17 See the Excel fi le “Own-Labor Incomes 1774.” 15

COLONIAL INCOMES IN 1774

•

25

Property income.18 Our property income estimates benefit from Jones’s masterly study of American wealth in 1774, based on her probate inventory samples and supporting documents.19 An important advantage of her data is that they identify the occupation or social status of most of those probated in her colonial sample. We have examined her data and procedures in great detail, and find no flaws.20 Jones realized that a probate-based sample ran the risk of overstating average wealth and understating wealth inequality, because probate was more likely for the deceased rich than the poor. She went to enormous lengths to adjust for this, ending with what she called w*B estimates that were meant to capture more of the poor. We have moved in the same direction, using a different procedure. Our greater weighting of the poorer households was achieved by introducing the new data on occupational structure described earlier in this chapter. As it turns out, our estimates imply an even greater probate-wealth markdown than did her w*B estimates. Wealth is not property income, nor is it total income. Jones confined her income-measurement efforts to brief conjectures about wealth– income ratios, using twentieth-century aggregate capital-output ratios borrowed from the macroeconomics literature of the 1970s. We have followed a different route in order to exploit the labor earnings data just described. Our reading of the limited evidence on colonial rates of return suggests that, on average, assets earned a net rate of return of around 6 percent per annum. True, colonial interest rates around 1774 are thinly documented. Around 1800, federal government bonds had a market yield of 6.94 percent per annum, while New England municipals yielded 6.13 percent.21 But for earlier years, Winifred Rothenberg notes that colonial law stipulated that 6 percent was the “lawful 18 For a detailed description of our estimation of property incomes from Jones’s wealth estimates for 1774, see appendix 3 in the supplemental materials to Lindert and Williamson 2013a. 19 See Jones 1977, 1980; see also her ICPSR data fi le 7329 at the Inter-University Consortium for Political and Social Research at the University of Michigan. 20 In a set of side experiments, we tried to replicate Jones’s A*-weighted estimates using her own data and procedures. In no case did we achieve exact replication, and for one regional wealth total, we were off by 4 percent. We could not fi nd the source of this discrepancy, but suspect that she had to take some shortcuts in the pre-spreadsheet era that we have not understood. Despite the discrepancy, we feel confident of both her estimates and ours. See the http://gpih Excel fi les “Property 1774” for each region. 21 Homer and Sylla 1996, 276–79, 286.

26

•

CHAPTER 2

interest,” even though “beginning in 1785, interest rates began to climb to 7, 8, and 9 percent,” reflecting postwar inflation.22 Later we will quantify the sensitivity of our income estimates to this 6 percent rate of return assumption. The gross rate of return, which is more appropriate for calculating gross national product (GNP) for comparison with other studies, equals this net 6 percent plus rates of depreciation that differed by asset. Following NIPA accounting standards, we have assumed zero depreciation on financial assets and real estate (positive depreciation offset by rapid capital gains), 5 percent for servants and slaves, 10 percent for livestock and business equipment, and zero for net changes in producers’ perishables and crops. Combining own-labor and property incomes. Here we gain from having invested so much effort in gathering occupation data. Since ownlabor and property incomes are both arranged by occupation, we can combine the two to get their total incomes. For farmers, the largest occupational group, we can even exploit some data on the regional size distribution of property income, dividing it into the top 2 percent of farmers, the next 18 percent of farmers, a middling 40 percent, and a bottom 40 percent. This disaggregation helps us judge the degree of income inequality within each region.23 Households were practically the whole economy. Our calculations offer what NIPA accountants call the total private income of the household sector. The colonial government sector’s contribution, however, consisted only of the wages and salaries of government officials and military personnel, which are already included in our occupations and own-labor earnings. There were no government corporations in 1774. Nor do we need to worry about the retained earnings of private corporations since they amounted to little by the end of the century. The same assumptions will be made for 1800 in chapter 4. When chapter 5 22 Rothenberg 1985, 790. In personal communication, Farley Grubb notes late colonial evidence that could argue for either a 5 or 6 percent rate on government borrowing. 23 How we combined different kinds of incomes, both for HHs and their dependents, and the derivation of the fi nal inequality estimates for 1774 are elaborated in “American Incomes 1774, Full-Time Assumptions,” http://gpih; “American Incomes 1774, Baseline Part-Time Assumptions,” http://gpih.

COLONIAL INCOMES IN 1774

•

27

compares 1774–1800 incomes with 1850–1870 incomes, the non– household sector will take a larger share of national income for the later dates.

PROVISIONAL CONCLUSIONS ABOUT 1774 INCOME LEVELS Our estimates of labor and property incomes shed new light on economic conditions in 1774. The levels and composition of total personal income are shown in tables 2-2 to 2-5, for the three regions used by Jones and a geographically fi xed “nation,” defined as the thirteen colonies in 1774 (the same land area as the fifteen East Coast states plus the District of Columbia in 1800). Table 2-2 can be used to calculate any of several important ratios, using the denominators in the lower half of the table and the price deflators in the notes to the table.24 Our estimates suggest that the thirteen colonies were richer and more productive in 1774 than other estimates have implied. The left half of table 2-3 underlines the contrast, focusing on the more recent and more prominent set of competing estimates. Our thirteen-colony, current price (part-time) estimate of $163.8 million is 20 percent greater than the average of the Jones and John McCusker estimates ($136.9 million). Yet our colonial income estimates differ greatly from those of Jones for only one region. There is little difference between us for New England or the middle colonies. The main source of the big difference with Jones arises in the South, for which our income estimate ($98.8 million) is almost twice that of Jones ($59.2 million). There are two gaps between Jones and ourselves to confront here: the gap in estimates for the thirteen colonies as a whole, and the gap for the South alone. While there will be errors in any measure of early incomes, including our own, we offer reasons for believing that the thirteen-colony gap and the southern colonial surprise are not due to errors we have introduced. For the thirteen colonies as a whole, the large gap is not driven by any higher estimate of wealth per household, since we rely on Jones’s own work. Supplementing her data with our new occupation weights, we get 24 Th is section draws on additional evidence reported in Lindert and Williamson 2013a; see appendix 4 of the supplementary materials, downloadable from the Journal of Economic History’s website.

28

•

CHAPTER 2

Table 2-2 Estimated American Personal Incomes, 1774 New England Middle Atlantic

South Atlantic

All thirteen colonies (fifteen states and District of Columbia)

Gross income, millions of current dollars ($4.44/£ sterling) FTEs, free own-labor income

31.09

28.85

62.81

122.75

Ditto, part time (see text)

28.07

27.22

58.09

113.38

Slave retained earnings

0.13

1.06

12.18

13.37

Gross property income

4.84

8.37

23.83

37.04

Gross total income

36.06

38.28

98.81

173.15

Ditto, with part time

33.04

36.65

94.10

163.79

Relevant denominators Free labor force

185,999

156,875

195,938

538,812

Total labor force

188,230

175,655

436,136

800,021

Free population

657,567

582,134

719,875

1,959,577

Total population

661,563

613,685

1,101,151

2,376,399

Some average incomes (current 1774 dollars) Free own-labor income per member of free labor force full-time assumptions

167.15

183.90

320.56

227.82

part-time assumptions

150.91

173.51

296.47

210.43

per slave labor force member

58.27

56.44

50.71

51.19

per capita

32.53

33.60

31.95

32.08

per capita, free

50.05

61.14

113.80

76.76

per capita, total

49.94

59.72

85.46

68.92

Slave retained earnings

Gross total part-time incomes

Table 2-3 Alternative Estimates of National Income, 1774, in Current and 1840 Dollars (Millions)

United States (original thirteen)

1774 (current $m)

1774 (1840 $m)

142.2

152.5

GDP: McCusker 2000

131.7

141.6

Gross income: Jones 1980

173.2

185.7

444.1

305.9

Gross income, FTE: Lindert and Williamson 2013a

163.8

176.0

422.8

291.3

Gross income, part time: Lindert and Williamson 2013a

508.7

350.3

GDP: McCusker 2000

515.5

355.5

GDP: Mancall and Weiss 1999

United States (all) 150.3

161.6

135–57.0

145–69.0

134.8

145.0

132.6

142.6

1800 (current $m)

1800 (1840 $m)

Source

GDP: Gallman 1972 500.1

344.9

GDP: Goldin and Lewis 1980 Narrow GDP: Weiss 1992

430.9

297.2

Berry 1988

446.3

307.8

David 2005

510.4

351.5

GDP: Mancall, Rosenbloom, and Weiss 2003

470.7

324.2

Gross income: Lindert and Williamson 2013a

448.1

308.7

Gross income, part time: Lindert and Williamson 2013a

Table 2-3 (cont.)

New England

Middle Atlantic

South Atlantic

Lower south Atlantic

1774 (current $m)

1774 (1840 $m)

1800 (current $m)

1800 (1840 $m)

Source

35.5

38.2

33.0

35.4

36.5

39.3

36.7

39.4

59.2

63.6

94.1

100.8

214.9

148.0

Gross income, part time: Lindert and Williamson 2013a

22.0

23.7

93.5

64.4

GDP: Mancall, Rosenbloom, and Weiss 2003

94.1

64.8

Gross income: Lindert and Williamson 2013a

Income: Jones 1980 95.1

65.5

Gross income, part time: Lindert and Williamson 2013a Income: Jones 1980

134.1

92.4

Gross income, part time: Lindert and Williamson 2013a Income: Jones 1980

a slightly lower net worth per wealth holder than she did. Furthermore, because we find many fewer households with wealth than her estimated number of “potential wealth holders,” our aggregate wealth estimate is only about 70 percent of her implied total wealth.25 Thus, our higher income estimate is not due to a higher wealth estimate, since ours is lower than that of Jones. The difference between them is explained by the way wealth is converted into property income. Our conversion, as described earlier, involved multiplying different asset holdings by net and gross rates of return. A reader feeling that our 25 In this passage, “wealth” means household net worth. On property incomes for 1774, see http://gpih.

COLONIAL INCOMES IN 1774

•

31

income estimates seem too large might want to challenge both our net and gross rates of return as being somewhat high. But it seems to us unlikely that our 6 percent figure overstates the net rate of return—the opportunity cost of not having lent at interest. The colonies and the early republic had a legal usury limit of 6 percent that was vigorously supported by law and custom.26 That is, the usury constraint seems to have checked a strong demand for capital, so that the 6 percent ceiling might well have been below market. Could the (illegal, market) rate of interest forgone by holders of directly productive assets have been higher—say, 8 percent? This is a distinct possibility, especially for 1800, for which the literature suggests even greater capital scarcity than for 1774. As we have noted elsewhere, shifting to the higher rate of return would raise our total income estimates further above those conjectured by others.27 One might also challenge our depreciation assumptions in deriving gross property incomes. Indeed, some might argue that depreciation should not have been included in the income estimate at all. If the reader prefers net property income estimates, ones that only include that 6 percent rate of return on wealth, then this net household income estimate would eliminate about half the gap between our gross income estimate and the $136.9 million average estimate offered by Jones and McCusker.28 That partial convergence might seem comforting, but it should not. The debate over early American economic growth has consistently used GNP (or GDP), not net national product. We should conform to the same convention for the purposes of comparing apples with apples. Thus, our favored 1774 aggregate income estimates remain the gross income figures shown in tables 2-2 and 2-3. So much for possible biases in property income. What about our own-labor income estimates for 1774, supported as they are by new occupation weights, full-time employment assumptions, and occupationspecific wage rates? Could these have exaggerated labor income for the thirteen colonies as a whole, thereby raising our aggregate income estimates above that of previous scholars? The source of the difference cannot lie with our new occupation shares, which give greater weight to 26

Homer and Sylla 1996, 271, passim. Lindert and Williamson 2013a, 746, table 5. 28 Using the full-time estimates, see Lindert and Williamson 2013a. 27

32

•

CHAPTER 2

poorer and less probated whites in the North, since this would serve, once again, to make our estimates lower than Jones’s, not higher. We also do not think the full-time employment assumption of 313 working days per year for those hiring out is inappropriate, given the widespread prevalence of home production and direct nonmarket consumption in the eighteenth and early nineteenth centuries, when subsistence farming at the frontier dominated. And as we have already noted, switching to a more conventional set of assumptions about occupations’ annual workdays has cut our estimated national income by only 4.8 percent. That is, using the part-time work years defined above reduces our 1774 national income from $173.23 million to $164.11 million, or still 20.4 percent above the Jones and McCusker average.29 What did Jones assume about rates of pay for labor, including the earnings retained by servants and slaves? In fact, she did not make any assumption at all but rather took a single leap of faith that we have already noted: by picking up some capital–output ratios quoted in the aggregate growth literature from the 1970s, she jumped from her impressive and reliable wealth estimates to total income guesswork that stands or falls on her assumed aggregate wealth–income ratio (not necessarily the same as a capital–output ratio). The macro literature offered Jones capital–output ratios ranging from 2.5 to 10 for the nineteenth and twentieth centuries. Within this wide range, she said, “I hazard that ratios of three or three and a half to one may be reasonable.” 30 Yet we find that the 1774 ratio of net worth (wealth) to national income was only 1.89. Why did southern colonists have higher incomes than northerners in 1774? The wide income per capita gap between the North and South in 1774 has some simple explanations. First, the colonial South was still 29 There is another alternative calculation that could bring our 1774 estimates still closer to those of Jones and McCusker. For a discussion of alternative ways of interpreting Jackson Turner Main’s information on farm profits around 1774, see appendix A. Still, as we say in appendix A, our present interpretation is more consistent with the information on farm profits in the mid-nineteenth century and also consistent with Main’s gleanings from colonial sources on the subject. 30 Jones 1980, 62. Gallman and Weiss have preferred her top wealth-income ratio, 3.5–1, and that is used in table 2-2’s display of her estimates. Decomposing our aggregate thirteen-colony wealth-income ratio of 1.89 into regional wealth–income ratios for 1774, we estimate the wealth–income ratio of 0.96 for New England, 1.80 for the Middle colonies, and 2.25 for the South.

COLONIAL INCOMES IN 1774

•

33

more of a frontier area, reaping the early gains from a rich and undepreciated land endowment suitable for producing tobacco, rice, and (royally subsidized) indigo. Second, the total southern population, including all age groups, worked more days per year per person—especially when one remembers to include slaves, with their high labor force participation and full-time employment. By contrast, the northern population had a more noticeable share of unskilled workers who were less regularly employed. Third, we emphasize that the high incomes of southern whites were not something that people from New England or the middle colonies could easily attain just by moving to the South. Free southerners differed in their occupations more than in their annual incomes for any given occupation. Unlike 1860 and later, in 1774 free southern men had a different occupation mix, with a much higher share propertied and a much lower share poor. On the eve of the revolution, the South was still a staple-exporting frontier with high returns on the production of tobacco, rice, and indigo. We find this contrast between the regional occupation mixes among free HHs in 1774—a contrast implicit already in figure 2-1’s societal sketch: New England

Middle colonies

Southern colonies

Farm operators

43.9

25.8

72.7

Professions, commerce, crafts

11.0

32.5

14.3

No occupation given, some wealth

16.7

28.7

11.0

Menial laborers and those with zero wealth

28.4

13.0

1.9

Southern farm operators not only had higher average incomes than other farmers but also constituted a larger share of households, while low-paying occupations took a lower share among free southerners (but not among total southerners, including the hardworking slaves). That is, what drove the income gap between regions was not pay differentials mysteriously unexploited by potential migrants but rather a 34

•

CHAPTER 2

mix of free southern occupations favoring those with higher wealth and income. An exercise in accounting for income differences between regions supports this point.31 Of the 107 percent gap between average free household income in the South ($705) and the middle colonies ($340), most would be accounted for by giving free southerners the occupational mix of the middle colonies, and only a small share would be due to differences in average rates of pay for given occupations. Other evidence supports our finding of a richer colonial South than previously thought. Jones’s wealth estimates had already shown that wealth per wealth holder was 56 percent higher in the South than the average for other colonies. That again stemmed largely from her estimates of the occupational mix, even before our revisions magnified the contrast. Both her occupational mix and ours show a remarkable lack of poor southern whites, defined as those of menial trades or having zero tax-assessable wealth. While our occupational calculations for the South have used only Jones’s estimates plus the finding that there may have been no HHs with zero wealth in three North Carolina counties, two archival data sets from Virginia in the 1780s also suggest that there were few demonstrably poor white HHs in the South. Tax returns from 1787 for rural Loudoun County at the north end of Virginia appear to have covered all white households, yet with few poor whites having zero net assets. Similarly, while a special 1782 census of the rural town of Richmond, Virginia, did reveal some white HHs with menial occupations or zero wealth, their share of white HHs was no higher than in the rural towns of Chester County, Pennsylvania, and lower than the share in rural Massachusetts. More important than these comparisons within rural regions, or between their small towns, was their rural and urban differences. The South was overwhelmingly rural, where the poverty share among the free population was lowest. While the percentage of true white paupers was not zero in the South, it was not as high as in the North.32 The suggestion that the colonial South was richer than the colonial North has evoked surprise in our seminars. Given that the data are still meager, despite our having added fresh information, we need to 31

See Lindert and Williamson 2013a, appendix 4 in the supplementary materials, table A4.1. For evidence from Virginia in the 1780s and commentary on the inequality literature for the Chesapeake, see Lindert and Williamson 2013a, appendix 4 of the supplementary materials. 32

COLONIAL INCOMES IN 1774

•

35

consider the sensitivity of our conclusions about the South to possible errors. Suppose that we, and Jones, somehow missed the southern, free rural poor—that is, the rural, tax-exempt, poor free households of 1774. Suppose that as much as 10 percent of the true southern free households were missed, and that they were as poor as the zero-wealth rural New Englanders, our poorest free male-headed households, with annual incomes of only $40 per household. That 10 percent figure is possible, given a high-side reading of the Virginia evidence we have explored elsewhere.33 Adding these possibly missed poor, at $40 per household, would lower free southern income per household from $587 to $533 (about 10 percent lower), with a corresponding 10 percent adjustment to the $392 average if slave households are included—still far above the average free household incomes of $255 for New England and $277 for the middle colonies. With or without a possible error of this sort, the southern colonies appear to have been richer, as Jones had thought using wealth data alone.

AN INCOME-EGALITARIAN COLONIAL AMERICA Until the appearance in 1965 of Main’s The Social Structure of Revolutionary America, inequality and social structure was a marginal topic in the early American literature. Things changed dramatically afterward, as happened in the economics literature with the appearance in 1955 of Kuznets’s presidential address to the American Economic Association on what came to be called the Kuznets curve hypothesis. Since then, there has been an outpouring of empirical work on American colonial wealth inequality, income inequality, and wage inequality led by Jones’s impressive work.34 Yet getting income distributions from Jones’s colonial wealth distributions alone is an insurmountable challenge, as we have seen. The rest of the colonial inequality literature is even less adequate since it relies on local observations, and we are never sure just how representative these cities, towns, and counties are. We think the problem is solved with our 1774 estimates. 33 34

36

See Lindert and Williamson 2013a, appendix 4 in the supplementary materials. For summaries, see Williamson and Lindert 1980a, chap. 2; Henretta 1991, 148–53.

•

CHAPTER 2

Incomes were more equally distributed in colonial America than in any other place that can be measured. Among all American households, slaves included, the richest 1 percent had only 8.5 percent of the total income, and the Gini coefficient was 0.441 (table 2-4). Among free households, the top 1 percent had only 7.6 percent of total incomes, and the Gini was 0.409. Compare colonial American inequality with that of the United States today, where almost 20 percent of the total income accrues to the top 1 percent, and where the Gini coefficient is about 0.5.35 That colonial America was a more egalitarian place is even more apparent when we compare the United States today with colonial New England (Gini 0.367), the Middle Atlantic (Gini 0.376), and surprisingly, the free South (Gini 0.341). It might seem puzzling that the free populations in each region could have a Gini less than that for the total (e.g., 0.367, 0.376, and 0.341—all less than 0.441), but recall that there was also great inequality between the three regions. In short, free citizens had much more equal incomes than do today’s Americans. Free American colonists also had much more equal incomes than did western Europeans at that time. The average Gini for the four northwestern European observations reported in table 2-5 is 0.57, or 0.13 higher than the American colonies and 0.20 higher than New England. Indeed, there was no documented place on the planet that had a more egalitarian distribution in the late eighteenth century.36 If people had more equal incomes in America than elsewhere, which kinds of colonists were better off than their counterparts in Europe? Figure 2-2 offers an Anglo-American comparison. On the horizontal axis, each society is ranked from its poorest to its richest, and on the vertical axis, their average group incomes are displayed in logarithms. 35

Atkinson, Piketty, and Saez 2011, 31, table 5. For a preindustrial comparison across continents and centuries, see Milanovic, Lindert, and Williamson 2011. We should repeat that these inequality calculations relate to total household income. Were it possible, one would like to compare the inequality of income per household member or income per consumer unit measured in adult-male equivalents, given that children and women consume less than grown men. Th is has not proven possible, and it might matter for a society that had large families and high child dependency rates (chapter 3). For the colonies in 1774 and the English social tables of the eighteenth century, we lack average sizes for free households by the head’s occupation, sex, region, and urban–rural residence. In the colonies, the labor force size per free household seems to have been higher for richer classes, and higher for urban than for rural. Th is would make household incomes per consumer unit even more equal in the free colonial population. 36

COLONIAL INCOMES IN 1774

•

37

Table 2-4 Inequality in the American Colonies, 1774 Region

All thirteen colonies

All thirteen colonies

New England

Middle colonies

South

South

Households

All

Free only

All

All

All

Free only

Gini coefficient

0.441

0.409

0.367

0.376

0.464

0.341

Income shares in % of total income Top 1% of HHs

8.5

7.6

4.1

6.7

8.3

6.7

Top 5%

23.1

22.7

12.0

20.9

26.6

22.4

Top 10%

32.2

31.1

20.8

28.8

35.2

32.2

Top 20%

47.8

45.3

36.3

43.3

49.7

43.6

Next 40%

39.6

39.7

52.5

40.5

38.3

35.0

Bottom 40%

12.6

14.6

11.2

16.2

12.0

21.9

Household income levels in $ (at $4.44/£ sterling) Mean

327

383

255

277

392

587

Median

279

334

309

232

321

508

Top 1% of HHs

2,774

2,900

1,039

1,862

3,243

3,910

Top 5%

1,509

1,740

614

1,156

2,089

2,635

Top 10%

1,052

1,192

531

799

1,381

1,889

Top 20%

781

868

463

600

976

1,281

Next 40%

324

380

335

281

376

638

Bottom 40%

103

143

72

113

117

190

It appears that an American colonist of any rank had a higher income than their English counterpart of the same rank until we reach the top percentile. Indeed, it turns out that even American slaves were above the bottom of the Anglo-American income ladder, although such comparisons fail to account for loss of freedom, (presumably) longer hours worked, and harsher working conditions. As one can gather from

38

•

CHAPTER 2

Table 2-5 Inequality in Western Europe, 1732–1808 Region (All households)

England and Wales

England and Wales

Holland

Netherlands

Year

1759

1802

1732

1808

Gini coefficient

0.522

0.593

0.610

0.563

Income shares in % of total income Top 1% of HHs

17.5

14.6

13.7

17.0

Top 5%

35.4

39.2

37.0

39.5

Top 10%

45.1

48.8

50.9

51.3

Top 20%

57.5

63.2

65.8

64.7

Next 40%

30.0

27.8

25.6

22.8

Bottom 40%

12.5

9.0

8.5

12.5

Household income levels Mean £ Median

£

43.4

90.6*

florins

67.8

319.3

25.0

55.0

florins

35.0

150.0

* £106.8 if we count government revenue, the king, and certain pensioners, listed separately by Colquhoun (Lindert and Williamson 1983a).

figure 2-2, colonial households as a whole had higher incomes than households in England and Wales. If one simply converted from dollars directly into sterling at the exchange rate of $4.44 per pound, colonial households averaged £78 each versus about £50 per family in England and Wales, converting either the revised version of Massie’s 1759 social table or that of Colquhoun for 1801–1803. At a first rough comparison, the American colonists in 1774 had much higher incomes—56 percent higher, as one might have predicted given that so many English continued to risk migration across the Atlantic (see chapter 3). However, we need to assess the impact of relative purchasing power on such income comparisons, since simple exchange rate conversion does not adequately account for cost-of-living differences between places. This familiar point has a number of important applications in the

COLONIAL INCOMES IN 1774

•

39

Household income (£ / year) (log scale)

20,000 10,000

Richest colonists

1,000 500

All colonists 100

Slaves

England–Wales, 1759 (in 1774 £) 3 Poorest

20th %

40th %

60th %

80th %

Richest

Income ranks Figure 2-2. Comparing American and English Income Ranks in the Late Eighteenth Century

colonial American context, and they deserve emphasis here and in later chapters. One is that the cost of a standard consumption bundle was lower in New England and the middle colonies than it was in the Lower South or England and Wales. So say some recent calculations for this era.37 Thus, nominal income comparisons understate the colonial living standard advantage: New England—with its cheap fish, corn, beans, rum, and molasses—was not so much poorer than the southern colonies, as the nominal figures in table 2-4 imply; similarly, the middle colonies—with their cheap grains (exported to England where they were expensive)—were not so much poorer than the southern colonies, as table 2-4 implies.38 Perhaps New England wasn’t so much worse off relative to southerners as our figures suggest, and perhaps workers in the Middle Atlantic were even better off compared with English workers 37

Allen, Murphy, and Schneider 2012; see also chapter 3. Mancall, Rosenbloom, and Weiss 2008b. Such adjustments should also deal with the relative cost and quality of housing (Shammas 2007). 38

40

•

CHAPTER 2

than our figures suggest.39 These “real inequality” issues will be explored further in the next chapter, but the results do not overturn the inequality contrasts shown here.40 Our new inequality evidence for 1774 also speaks to a new institutional literature that argues that economic inequality breeds political power that favors rent-seeking (or extractive) institutions and policies rather than growth-enhancing institutions and policies, while a large middle class does just the opposite. Table 2-4 offers American colonial evidence consistent with this thesis. There the “middle class”—which we take to be the middle 40 percent—claimed 52.5 percent of incomes in New England, the cradle of the revolution, but it claimed only 35 percent in the slaveholding South. We also know that the policies introduced in the early republic were pro-growth, as were those inherited from England. These are only correlations, but they are certainly consistent with the new institutional views of economic growth.41

SUMMING UP The only way to push back in time the quantitative frontiers of inequality and living standards is to adapt to the scarce data environments of the deep past. In the archaeological extreme, good use has been made of skeletal remains and DNA as our main forms of evidence. Similarly, our journey back to the late eighteenth century has had to use an eclectic array of incomplete evidence. Still, we have found one of the most underexploited tools—the social table—effective. This tool has allowed us to estimate aggregate incomes and their distribution by colonial regions.

39 Of course, an upper-class cost-of-living bundle, including the cost of music, theater, and servants, must have been much lower in London than in the American colonies. 40 For the specific contrast of consumer prices between New England and other regions, see the fi le “Massachusetts vs. England and WV,” http://gpih (where WV = West Virginia). See also Allen, Murphy, and Schneider 2012, table 3. Unfortunately, these two sources do not offer the price evidence we need for the more export-oriented South. On the more general issue of classand place-specific costs of living, see Williamson 1977; Hoff man et al. 2002. 41 On the global imprint of institutions on economic growth, see Acemoglu and Robinson 2006, 2012; Engerman and Sokoloff 2012. On the institutions developed in the early American republic, see Irwin and Sylla 2011.

COLONIAL INCOMES IN 1774

•

41

These social tables have produced a rich harvest of early American income and inequality estimates. It appears that the colonists had higher incomes in 1774 than previously thought—more than half again higher than Great Britain. In addition, we have found that free American colonists had more equal incomes than did households in England and Wales, or any other measurable part of the world. The colonists also had greater purchasing power than their English counterparts over all the income ranks except at the top 1 percent, although chapter 4 will show that their income advantage was lost in the fight for independence. Regional inequality was also an important feature of the American colonies. Our estimates suggest that per capita incomes were far higher in the South than in the North in 1774, and that poor whites were much less common there than in other colonies. It appears that the colonial South lacked the large number of poor whites that historians have counted in Boston, Philadelphia, New York, and lesser coastal towns. In short, our results suggest that mass poverty did not spread among the southern white population until the nineteenth century.42 Or so it seems to us, given the much improved, but still limited, historical evidence now available. 42 The late colonial income distribution in the South needs further research of the sort already done for the Chesapeake. See, for example, Stiverson 1977; Kulikoff 1986; Carr, Menard, and Walsh 1991; Walsh 2010. See also Gallman’s (1982) study of Perquimans County, North Carolina.

42

•

CHAPTER 2

CHAPTER 3

When Did Colonial America Get Rich?

T

he previous chapter showed that the average American colonist had by 1774 a far higher income than did the average Englishman. Colonial America was also probably the most egalitarian society in the world. The South was the richest of the colonies, and even its slaves had higher living standards than did the poorest in England. The average American colonist ate better, was taller, and lived longer than did the average Englishman. When and how did colonial America get so rich? The near-total statistical darkness about colonial living standards has prevented any consensus. There is still not enough evidence to get the answers by constructing an early colonial social table, as we did for 1774. But there is enough indirect evidence to say something about colonial per capita income growth over the seventeenth and eighteenth centuries. Early settlement conditions have long been well-known, even without economic numbers. In the beginning there was misery. Jamestown, the earliest colony to survive, barely survived. Of the ten thousand persons that the Virginia Company transported to Jamestown between 1607 and 1622, only one-fift h of them were still alive there in the latter year. The company had to keep delivering more settlers from England to replenish the population, just as fresh African slaves had to be imported to maintain the slave labor force of the high-mortality West Indies. Mistakes were made, repeatedly. The Jamestown site was badly chosen. The settlers were ill-equipped for agriculture in the new setting, partly because so many of them had imagined that they could strike it rich just by seizing land and forest resources from the natives, or trading for them. New shipments of settlers kept arriving with insufficient

rations. To feed themselves, the settlers had to trade away their European goods for maize, squash, game, and other basic foods. Nor did they have any firm understanding of how to deal with the local tribes.1 Similarly, the initial settlers on the not-so-fertile coast of Massachusetts were ill-equipped for their new agricultural environment. Several forces turned the original colonists’ misery into prosperity for their descendants. The Native American population soon died off from European disease, leaving abundant land for the Europeans, who eventually learned how to work it.2 The colonists also discovered, in 1616, that the Chesapeake was excellent for cultivating tobacco, which quickly became their main source of export income. Another adjustment was the shift from company rule to generous land allotments for homesteading, which improved incentives and thus returns on investments. Finally, the colonists began importing servants and slaves—a labor force they could control more easily than the disappearing Native Americans. The evidence to which we now turn suggests that prosperity had arrived by the late seventeenth century, especially in the South. The thirteen mainland colonies as a whole were about as prosperous in the late seventeenth century as they were on the eve of the revolution, and perhaps even more so. Second- and third-generation colonists learned how to cultivate abundant and high-quality land, and how to exploit Atlantic markets for the export of their commercial staples like fish, barrel staves, wheat, tobacco, naval stores, and rice. The narrow band of coastal settlements and their ports achieved higher per capita incomes and per capita income growth, but much slower population growth than the rapidly expanding though economically undeveloped interior. Southern commodity export prices had secular booms like that of the nineteenth century. Prices of all export staples, North and South, were even more volatile than in the nineteenth century. Since volatility is bad for growth, this must have slowed down colonial economic progress. In addition, abundant land induced early marriage, high fertility, low child mortality, 1 See Cederberg 1977, especially chap. 4; McCusker and Menard 1985, 117–23; Kulikoff 1986, 30– 44; Taylor 2001, chap. 6, including the sources cited there. 2 For a summary of the best guesses on the size of Native North American tribes and their population decline, see Ubelaker 1992.

44

•

CHAPTER 3

and the fastest population growth in the world. Child dependency rates were therefore much higher in the colonies than anywhere else in the world, and labor participation rates were accordingly much lower. Those high dependency rates may therefore have suppressed per capita income growth. But those high dependency rates implied that colonial income per worker was even further above Europe than was colonial income per capita. We turn next to the conflicting views of how fast incomes per capita in the colonies grew after the mid-seventeenth century. Most early American historians imagined fairly rapid growth up to the 1770s, implying poverty for the colonists a century earlier. A small minority has argued that growth was slow or nonexistent, implying the same prosperity in the late seventeenth century as in the 1770s. To address this disagreement, we explore indirect evidence on colonial demography, trade, staple export prices, and frontier settlement, all of which should have been central actors in the economic progress of this young settlement. We then offer direct conjectures on the movements of income per capita itself, using a “backcasting” technique and some data previously unavailable. The lack of consensus over the rate of growth across the colonial era is summarized in table 3-1. Earlier authors tended to posit high growth rates, averaging 0.61 percent per annum, which would imply a doubling of average incomes from the mid-seventeenth century to 1774. In contrast, the newer slow-growth estimates posit rates averaging 0.05 percent per annum, or roughly zero, based on “controlled conjectures” prepared by Peter Mancall and Weiss on all colonies, Mancall, Joshua Rosenbloom, and Weiss on the Lower South, and Rosenbloom and Weiss on the middle colonies. We offer some novel empirical contributions, suggesting little or no growth in average incomes between 1650 and 1774. The next section examines the movements in trade, commodity prices, demography, and frontier settlement, all of which should have affected colonial growth. We will then turn to our new direct evidence on the movement of labor and property incomes before 1774.

WHEN DID COLONIAL AMERICA GET RICH?

•

45

Table 3-1 Past Estimates of Eighteenth- Century Colonial Income per Capita Growth Period

Region

Per capita annual growth (%)

Data

Source

Fast-growth estimates 1688–1764

Lower South

0.5

Product per capita

Menard 1996, 257

1700–1770

All colonies

0.4

Wealth per capita

Jones 1980

1713–1775

All colonies

0.5

Mainly imports per capita

Egnal 1998

1713–1775

Lower South

0.9

Mainly imports per capita

Egnal 1998

1650–1770

Southern New England

0.35

Wealth per capita

Main and Main 1988

1700–1779

Southern New England

0.51

Wealth per capita

Anderson 1979

1705–1776

Chesapeake

0.4

Wealth per capita

Kulikoff 1979

1722–1762

South Carolina coastal

1.9

Wealth per capita

Coclanis 1989

1713–1775

North

0.6

Mainly imports per capita

Egnal 1998

1713–1775

Upper South

0.1

Mainly imports per capita

Egnal 1998

1700–1775

Upper South

0.5

All evidence ca. 1991

Henretta 1991, 176

Average

0.61

Slow-growth estimates 1700–1770

All colonies

0.05

GDP per capita

Mancall and Weiss 1999

1720–1770

Lower South

−0.03

GDP per capita

Mancall, Rosenbloom, and Weiss 2003

1720–1770

Middle colonies

0.13

GDP per capita

Rosenbloom and Weiss 2014

Average Note: See also appendix H.

0.05

LEADING ACTORS: LIKELY INFLUENCES ON GROWTH IN A FRONTIER ECONOMY Several kinds of forces should have buffeted fortunes in colonial America. Some arose from world trade conditions, and others from colonial demographic influences. These forces should have been dominant in a preindustrial world, in which productivity growth was slow even among the western European leaders. Overseas Trade Patterns The North American colonies made up a tiny frontier on the periphery of the Atlantic economy. Colonial incomes above subsistence depended on the prices they could get for their primary product exports, or what we call today “commodities.” While exports varied in importance across local economies, all four mainland colonial zones that later became the United States shared much the same price volatility and secular price trends. New England—then consisting of Connecticut, Massachusetts, New Hampshire, and Rhode Island, with Boston as the region’s main port— was the most diverse of the four regions, even after it started to harvest fish off the Grand Banks. By 1770, fish accounted for only 34.7 percent of the region’s exports, and the rest was a mixture of rum (4.3 percent), wood products (14.4 percent), whale products (14.1 percent), livestock (20.5 percent), and other commodities (12 percent). These New England commodities were exported everywhere in the Atlantic economy, not just to Britain. The salted fish went to Mediterranean ports, livestock to the West Indies, whaling products to England and the Continent, and wood products (mainly staves and cask heads for barrels) everywhere. Beyond such commodities, New England stood out with its high earnings on exporting “invisibles” (services), such as shipping services. Overall, its export earnings amounted to 11.1 percent of its regional product, of which nearly half consisted of invisibles.3 3 For a seminal work that captures the colonial patterns of foreign transactions for the period 1768–72, see Shepherd and Walton 1972. The regional income denominators in this and the following paragraphs are our own for 1774; see http://gpih.

WHEN DID COLONIAL AMERICA GET RICH?

•

47

Another way to summarize New England’s position in interregional trade is to note that its comparative advantage was close to that of England itself, implying that it might have served as a trade competitor against the mother country, even before the nineteenth-century rise of its manufactures.4 Indeed, Sir Josiah Child was already lamenting New England’s role in the empire in the late seventeenth century: New-England is the most prejudicial Plantation of the Kingdom of England. . . . All our American Plantations, except that of New England, produce commodities of different Natures from those of this Kingdom. . . . Whereas New-England produces generally the same we have here, viz. Corn and Cattle.5

By the eve of the revolution, the middle colonies—that is, New Jersey, New York, and Pennsylvania, with Philadelphia and (to a lesser extent) New York City as their main ports—had also emerged as significant exporters of flour, pork, wheat, and other classic farm products of the temperate zone. Yet exports of goods and services accounted for only 9.4 percent of the middle colonies’ overall income, and the commodity (goods) export share was even smaller. The Upper South—Delaware, Maryland, and Virginia, with Norfolk as the region’s only large port before Baltimore began to emerge after 1750—exported mainly tobacco, making up 60 percent of its foreign exchange earnings, with grains adding another 26.3 percent. Thus, just two products dominated the region’s export revenues (86.3 percent).6 As late as 1768–1772, these staples directly generated 13 percent of the region’s total income—a somewhat higher share than in other regions— and the dependence on foreign trade was presumably even greater in the tobacco boom of the late seventeenth century. The Lower South—the Carolinas and Georgia, with Charlestown and later Savannah their main ports—exported rice and naval stores throughout the eighteenth century, and added indigo to the list in the late 1740s. These three staples took up a large share of the Lower South’s 4 We use the words “might have” advisedly, since like European imperial powers over the next two centuries, Britain did not let America develop its manufacturing. 5 Sir Josiah Child, A New Discourse of Trade (London: T. Sowle, 1698), 166, 204– 6, quoted in Galenson 1996, 201. Beyond corn and cattle, Child should have mentioned the fisheries of both countries. Yet his point seems correct enough. 6 McCusker and Menard 1985, 132, table 6.2.

48

•

CHAPTER 3

total export revenues; in 1768–1772, 55.4 percent of the region’s foreign exchange earnings were from rice, 20.3 percent from indigo, and 5.7 percent from naval stores (pitch, tar, and turpentine). Other items included deerskins (from trade with natives), wood products, grains, and livestock, but more than 75 percent consisted of the big two—rice and indigo.7 While the Lower South’s exports were concentrated on two products, they did not constitute a particularly large share of the regional product—only 9 percent. The Lower South was thus heavily committed to production for domestic consumption. For all the attention given to its exports, it was not as trade dependent as the other colonies. As such, there seem to have been two economies present in each of the four regions: a small, coastal-based, export-staple, higher-income economy, and a large, lower-income frontier economy only poorly integrated with the rich coast. While we have used the word “subsistence” to describe levels of living on the large inland frontier, it should not be taken to imply low living standards, since a large land and forest endowment per yeoman farmer produced a higher per capita income than the English produced. Historians of early America have called it  “subsistence-plus.” Perhaps a better word to describe the stillundeveloped colonial frontier might be noncommercial. Colonial Price Volatility 1700–1776 It was just as well that the colonies depended so little on overseas trade, since the world economy around them generated wide swings in prices for traded goods. The qualitative histories of colonial America are sprinkled with commentary on economic ups and downs, booms and slumps, good times and bad. As we will see, these traditional qualitative narratives are borne out by new quantitative evidence. The Indian Wars on the borders, embargoes, European conflicts on the seas, and parliamentary decree drove some of this economic volatility.8 In agriculture, 7

McCusker and Menard 1985, 174, table 8.2. Th is volatility can certainly be documented on the high seas. Philadelphia insurance firms quoted maritime insurance as a percent of the value of the cargo carried. On the Philadelphia and London route, and without convoy, over the thirty years before the revolution, the rate in percent fell from a high of 15 (1745– 46) to a low of 2.5 (1749–55), rose to a high of 22.5 (1757), fell to a low of 6.7 (1759), rose again to a high of 15 (1762), and fi nally fell to a low of 2 or 3 (1767–71). See Egnal 1998, 184–85, appendix D. 8

WHEN DID COLONIAL AMERICA GET RICH?

•

49

rice, indigo, grain, and tobacco crops were certainly influenced by variance in climate and pests. In the colonial staple economy, however, economic volatility was driven mainly by the influence of overseas conditions on export prices. These prices were none of their doing, since the American colonies were only marginal world suppliers of fish, flour, wheat, indigo, and rice. The evidence gathered in table 3-2 agrees with these narrative accounts. Indeed, all four colonial regions recorded higher volatility in their export prices than do developing countries in the nineteenth, twentieth, and our current century. Since they were more violent, one might expect that these price storms should have brought more economic damage to the American colonies than to today’s Third World commodity exporters. True, commodity or staple prices have always been more volatile than prices of manufactures or services, and table 3-2 shows that the colonial experience fits the rule.9 On average, such primary-staple prices were more than three times as volatile as manufactured goods prices (PM), the highest ratio by far being rice (6.53) and the lowest being codfish (0.70). In short, with few exceptions, staple prices were much more volatile than manufactured goods prices in eighteenth-century colonial America, and especially so for the Lower South. And they were more volatile than the world’s commodity exporters have faced since. It is well established by economists and economic historians that volatility is bad for growth.10 We see no reason to think otherwise for colonial America. So if colonial America’s commodity export prices were more volatile, does it necessarily follow that it suffered more economic damage? Perhaps not, since foreign trade was only about a tenth of its incomes; most of its economic activity had already moved far inland by the start of the eighteenth century.

9

Jacks, O’Rourke, and Williamson 2011; Williamson 2011, chap. 10. Acemoglu, Johnson, and Thaicharoen 2003; Hnatkovska and Loayza 2005; Blattman, Hwang, and Williamson 2007; Huff 2002; Williamson 2011, 2012. 10

50

•

CHAPTER 3

Table 3-2 Colonial Price Volatility, 1700–1776, in Long-Run Perspective Period

Standard deviation log change

Relative to import prices

Cod

1700–1776

0.042

0.70

Rum

1720–1775

0.235

3.96

Pine

1720–1775

0.235

3.96

Exports (Px)

1700–1776

0.180

3.04

Imports (Pm)

1700–1776

0.059

—

Terms of trade (Px/Pm)

1700–1776

0.186

3.14

Flour

1720–1775

0.192

3.15

Wheat

1720–1775

0.209

3.42

Pork

1720–1775

0.042

0.69

Exports (Px)

1700–1776

0.189

3.10

Imports (Pm)

1700–1776

0.061

—

Terms of trade (Px/Pm)

1700–1776

0.200

3.28

Wheat

1720–1775

0.154

2.43

Corn

1720–1775

0.185

2.92

Flour

1720–1775

0.157

2.48

Tobacco

1700–1776

0.163

2.56

Exports (Px)

1700–1776

0.128

2.02

Imports (Pm)

1700–1776

0.063

—

Terms of trade (Px/Pm)

1700–1776

0.139

2.18

Rice

1720–1775

0.295

6.53

Indigo

1747–1775

0.136

3.01

New England

Middle colonies

Upper South

Lower South

(continued)

Table 3-2 (cont.) Period

Standard deviation log change

Relative to import prices

Naval stores

1720–1775

0.148

3.26

Exports (Px)

1700–1776

0.210

4.64

Imports (Pm)

1700–1776

0.045

—

Terms of trade (Px/Pm)

1700–1776

0.210

4.64

0.181

3.16

Commodities

Manufactures

Colony commodity average As compared with United States

1873–1896

0.065

0.105

United Kingdom

1820–1869

0.137

—

Nineteenth- and twentieth-century international

1860–2005

0.062

—

0.088

—

Average of the three

Secular Trends in the Terms of Trade Price volatility may have suppressed colonial growth, but were the prices of each region’s staples booming in the long run, thereby—on that account at least—fostering growth in the coastal staple districts?11 What would we expect to find? First, a quickening of GDP growth in western Europe would have put upward pressure on commodity prices, just as rapid growth in China and India do today. Second, declining transport costs in the Atlantic economy would have fostered price convergence.12 As we will see below, high imported food prices fell in England, while low exported food prices rose in the American colonies.13 Thus, 11 Marc Egnal (1998, 12) has said as much—“there was a strong correlation between . . . prices of the chief staples and the well-being of the colonists”—but he refers to evidence from the settled, coastal regions to prove the point. 12 North 1958; Harley 1988. 13 Allen, Murphy, and Schneider 2012.

52

•

CHAPTER 3

Table 3-3 Trends in the Colonies’ Terms of Trade, 1700–1776 Annual % rise in terms of trade (Px /Pm)

Commodities playing the biggest role

New England

+ 0.063

Fish

Middle colonies

+ 0.098

Flour and wheat

Upper South

+ 0.659

Corn, flour, and wheat

Lower South

+ 0.749

Rice and especially indigo at period end

export prices (PX) should have risen in the American colonies over the long run. Table 3-3 documents each region’s net barter terms of trade (PX/PM) and shows that fact confirms theory. Despite their export-price volatility, all four colonial regions underwent a rise in their terms of trade, but the improvement was only significant for the Upper and Lower South (0.66 and 0.75 percent per annum, respectively). While these terms of trade trends were not as steeply upward as those observed for nineteenthcentury commodity exporters, they certainly rose enough to have left their mark on growth.14 Th is suggests one reason why the southern colonies had so much higher per capita incomes by 1774. Since the literature indicates that per capita income grew at something like 0.61 percent per annum in the rich, coastal, staples districts, and even faster growth in the South’s terms of trade (averaging 0.71) is observed, this suggests that most of the per capita income growth in the staple districts of the Upper and Lower South was driven by secular terms of trade improvements, not by labor productivity growth. Since the terms of trade improved slowly, if at all, in the middle colonies and New England, whatever increases in income per capita those northern colonies achieved must have been due to labor productivity growth alone.

14 These net barter terms of trade improvements for eighteenth-century colonial America were much less than those for commodity exporters in the nineteenth century, when their growth rates averaged 1.4 percent per annum (Williamson 2011, 36, table 3.1)—twice that of the eighteenth-century colonial Lower and Upper South.

WHEN DID COLONIAL AMERICA GET RICH?

•

53

Rapid Population Growth: Fertility and the Dependency Rate The colonies had some of world history’s highest population growth rates, not only in the initial settlement phases, but also all the way up to the revolution. Figure 3-1 plots colonial population growth by region from 1610 to 1780.15 Between 1700 and 1780, population grew at 2.9 percent per annum in New England and in the middle colonies, and at 2.4 percent per annum in the South. These rates were well above those in the rest of the world, and they were six to seven times the average rate for western Europe.16 Should this rapid population growth have raised or lowered the colonial levels of income per person? Economists have long ago realized that the rate of population growth itself has no clear implication for either the level or rate of economic growth. Rather, its net impact depends on whether the high population growth raised or lowered the share of the population that was of working age—say, fifteen to sixty-four. High population growth fed by rapid net immigration would tend to raise income per capita, because immigrants tend to consist heavily of young adults ready to work. On the other hand, rapid population growth fed by high rates of natural increase (births minus deaths as a share of population) would cut the labor force share by raising the dependency rate. It would do so by raising either the share of children (if fertility were high) or retired elderly (if adult life expectancy were high). The former did most of the work in colonial British America. Those extraordinary rates of natural increase were driven by early marriage and high fertility within marriage as well as low child mortality (outside the coastal South). As early as 1751, Benjamin Franklin attributed all these features to the abundance of land, and half a century later Malthus agreed.17 Subsequent quantitative estimates have also found that Americans had lower crude death rates and longer life expectancies than did the Europeans.18 Yet the colonies also had, for the standards of that time, high rates of immigration, and these would have served to lower dependency rates. 15

Figure 3-1 is reproduced from McCusker and Menard 1985, 218. The average for western Europe, 1700–1820, was 0.43 percent per annum (Maddison 2010). 17 See Franklin (1751) 1959, 227–28; Malthus (1798) 1920, 105– 6. 18 Gemery 2000, 158– 69. This was not true of the coastal regions of the South, where the disease environment produced higher crude death rates. 16

54

•

CHAPTER 3

(log scale) 1,000,000

100,000

10,000

Total Mainland Colonies West Indies New England

1,000

Upper South Middle Colonies Atlantic Canada Lower South 100 1600

1625

1650

1675

1700

1725

1750

1775

Figure 3-1. Population of British America, 1610–1780

Which force dominated the age distribution and the dependency rate? Our clearest view is at the end of the colonial era. By 1774, the thirteen colonies had reached high dependency rates, implying that income per earner or per household must have been even higher relative to Europe than a comparison of incomes per capita would reveal. The 1774 age structure was extraordinary: in New England, 46 percent of the population consisted of children below age sixteen; in the Lower South, the figure was 52 percent; and the average across all thirteen colonies was 50 percent. These dependency burdens were high by any standard. WHEN DID COLONIAL AMERICA GET RICH?

•

55

For comparison, England in 1771 had only about 35 percent below age sixteen. More strikingly, in the 1980s, the child dependency share was 41 percent in the average Third World country and only 33 percent for mature, industrial countries.19 American age distributions and the dependency rates before 1774 are almost completely undocumented. We can, however, use Henry Gemery’s informed weighing of the meager colonial evidence to sketch the colonial patterns of natural increase versus net migration over time and space.20 Consider first the impact of natural increase on the dependency rate. As we noted above, it appears that death rates were higher in the disease environment of the coastal South, though fertility rates may have been similar to those in the North. It also appears that the chances of survival improved greatly in the South and had risen nearly to northern levels by the middle of the eighteenth century. Since much of this took the form of falling child mortality, the dependency rate must have risen in the South over time. Regarding immigration rates, there is rough consensus. For the thirteen colonies as a whole, the percentage rate of net (international) immigration had slowed down to low rates from 1690 onward, as one might expect from a settlement process.21 As for the geography of net immigration, our best guesses (i.e., those of Gemery) are especially tentative, since it is hard to measure both transatlantic migration and intracolonial migration. Fortunately, Georgia Villaflor and Kenneth Sokoloff have used muster roll evidence on the places of birth and current residence of those who fought in the French and Indian (Seven Years’) War and the Revolutionary War.22 Within the North, Bostonians left for Maine and New Hampshire, and New Englanders in general migrated to the middle colonies. From Maryland, Pennsylvania, and 19

Wrigley and Schofield 1981, 528–29; Bloom and Freeman 1986, 390, table 4. Gemery 2000. For a complementary survey of colonial population history, see Galenson 1996. One candle in the age-distribution darkness before the 1770s consists of the New York census on the white population. The share under age sixteen was 52.7 percent in 1703, 48.2 percent in 1723, 49.1 percent in 1746, 47.9 percent in 1749, 47.6 percent in 1756, and 46.1 in 1771 (Gemery 2000, 455). That is, the child share was consistently high back to 1723, and even a bit higher in 1703. 21 Gemery 2000, 178–79. As we will note later in this book, American immigration only became “mass” after the 1840s, when the cost of steerage had fallen to low levels and higher incomes in Europe made it possible for more to invest in the move (Hatton and Williamson 1998). 22 Villaflor and Sokoloff 1982. 20

56

•

CHAPTER 3

Virginia, the prevailing migration direction was south. The main region experiencing net emigration was New England, and the main recipients of net immigration were the Carolinas and New York. Our knowledge of colonial vital rates and dependency rates is sufficiently well-informed to make us confident about assessing their impact on income per capita. By 1774, the American colonies had reached exceptionally high dependency rates, implying, to repeat, that their incomes per earner or per household must have looked even better than their relative incomes per capita. That would have been especially true for New England, with its high natural increase and net emigration of young adults. Dependency rates in the Carolinas and New York were probably a little less elevated by global standards. The Geographic Battle between Ruralizing and Urbanizing Forces The final leading actor influencing income per capita trends across the colonial era was the urban share. As a general rule, cities have higher average incomes and more income inequality than the countryside. Development economists and historians have noted the implication that as a purely accounting matter, any forces that shift population toward the cities yield higher average incomes and higher income inequality. In this respect, colonial America was an exception. As figure 3-2 shows, the colonies were actually ruralizing after 1680. True, the cities were gaining in absolute population, but their share of the total colonial population was declining. Urbanization did not set in until after the revolution and during the early republic. Apparently, the rise of opportunities in the countryside and on the still-undeveloped frontier outran the rise of opportunities in Boston, Charleston, Newport, New York City, Philadelphia, and other smaller coastal and river towns. Other things being equal, the ruralization of the colonial economy between 1680 and 1790 would lead us to expect only modest colony-wide income per capita growth, as conjectured by the slow-growth camp in the colonial debate. Most of the forces surveyed in this section should have offset any positive per capita income growth forces. The volatility of the terms of trade should have weighed against the favorable trend in the terms of trade—a trend that was strong only for the southern colonies. The low share of colonial trade in domestic product should have dampened both WHEN DID COLONIAL AMERICA GET RICH?

•

57

Percent

20

10

All cities ≥ 2,500 Five main colonial cities 0 1650

1675

1700

1725

1750

1775

1800

1825

1850

1875

Figure 3-2. Urban Share of America’s Population, 1650–1870

these influences. To this absence of positive net growth forces, we have added two more: the high and rising dependency rate, a force that should have held down the rate of income per capita growth, and frontier settlement, which added low-income yeoman farms faster than coastal population growth. BACKCASTING INCOMES ACROSS THE COLONIAL ERA Aided by the informed quantitative guesses of others and by the likely roles of trade, commodity prices, demography, and frontier settlement as leading actors in the colonial economy, we now turn to our own conjectures about colonial income growth. The Method and Assumptions Starting from the 1774 benchmarks, how does one project, or backcast, to earlier and less documented times? As with all such extrapolations into the past, we have information on just a few forces driving income 58

•

CHAPTER 3

per capita growth. The method we use is similar to the controlled conjectures technique pioneered by David for estimating growth from 1860 back to 1800, and extended to the colonial era by Weiss and his collaborators.23 Prior to 1774, we have time series for wage rates and personal wealth—evidence that invites reapplication of our technique of adding own-labor and property incomes together to get the total income per person or per household. Our backcasts will represent the true income movements more faithfully, the smaller are the net errors from our making the assumptions listed in appendix H about missing information. Armed with these assumptions, we extend nominal incomes back over time. Table 3-4 lists the indicators that we employ to track nominal income movements. While the data permit annual series in some cases, a more realistic goal is to average the limited data over quarter centuries. Where possible, we trace back to a “ca. 1650” era that draws on incomplete data for 1638–1662. The next quarter century is an average of 1663–1687, and so on until we reach a “ca. 1770” benchmark averaging data for 1763–1774, followed by our 1774-only benchmark described in chapter 2. Regions differ in the availability of indicators that are closely tied to income as well as the underlying history being traced. Therefore, we proceed region by region, from north to south. New England New England offers the richest opportunity to follow household property income, thanks to a data set that recently became available. Gloria Main has supplied us with a large probate sample developed by herself and Jackson Turner Main in the 1970s and 1980s, and the data are now downloadable.24 The sample is both large (18,509 observations from 1631 to 1776) and broad in its coverage. Unlike most other probate samples, this one includes the value of real estate, the deceased wealth holder’s age at death, occupation, and other variables. Using regression techniques, we have held age constant by calibrating the (regressionpredicted) estate values to age forty-five, with historical interactions of 23

David 1967; for more recent, updated papers, see David 1996, 2005. To download the sample along with its variable defi nitions and some code values, see the folder “American Incomes ca. 1650–1870,” http://gpih. 24

WHEN DID COLONIAL AMERICA GET RICH?

•

59

Table 3-4 Indicators Guiding the Income Backcasts from 1774 to 1650 Income type

Source note

Indicator series

Boston

Free labor

NE-a

Boston seamen and male skilled craftsmen nominal wage rates back to 1725; male craftsmen 1725 back to 1650

Boston

Property

NE-b

Sample of 18,509 probates from colonial New England, 1631–1776; Boston subset of 1,492 probates

New England rural

Free labor

NE-c

New England farm labor wage rates for benchmark dates 1652–1769

New England rural

Property

NE-b

Sample of 18,509 probates from colonial New England, 1631–1776; non-Boston subset

New England slaves

Slave labor

NE-c

Assumed to be a fixed share of free farm wage

Middle colonies, urban

Free labor

MC-a

Philadelphia laborers and Philadelphia seamen, wages back to 1725, averaged

Middle colonies, urban

Property

MC-b

Philadelphia probated personal wealth by occupational group, 1685/1715–1756/1775

Middle colonies, all

Total income

MC-c

Production-side income per capita

Upper South, rural

Farm income

Ches-a

Gross farm income index, Chesapeake tobacco-and-grain farms, back to 1675

Lower South

All income

LS-a

Regional GDP per capita, 1720–1775, as estimated in Mancall, Rosenbloom, and Weiss 2003

Note: For further details, see appendix H.

Table 3-5 Predicted Wealth for Forty-Five-Year- Old Colonial New Englanders, by Time and Place, for Selected Occupations (among Those Having Positive Gross Assets) ca. 1650

ca. 1675

ca. 1700

ca. 1725

ca. 1750

ca. 1770

A. Gross wealth in bare-bones consumer bundles for one person Boston commerce, professions

305.6

297.7

354.2

—

—

319.2

Hinterland farmers

109.1

127.8

165.0

230.5

205.9

287.0

Hinterland artisans

75.2

62.5

70.4

62.4

50.2

50.4

Hinterland laborers

23.3

22.3

31.0

27.6

22.2

24.8

Hinterland widows

50.4

35.5

34.5

22.8

24.7

24.9

B. Cost of bare-bones bundle (£) One person

2.02

1.60

1.34

1.37

1.68

1.64

Four persons

6.38

5.03

4.22

4.33

5.28

5.15

C. Gross wealth in current £ sterling Boston commerce, professions

618.8

475.8

474.8

—

—

522.0

Hinterland farmers

220.9

204.2

221.1

316.5

345.3

469.4

Hinterland artisans

152.3

99.8

94.4

85.7

84.2

82.4

Hinterland laborers

47.2

35.7

41.5

37.9

37.2

40.5

Hinterland widows

102.0

56.8

46.2

31.3

41.5

40.7

place, time period, and occupation. Table 3-5 documents a notable pattern: from around 1650 to 1774, only farmers in the later-settled hinterland experienced great gains in average wealth and thus in property income.25 These hinterland farmers apparently kept improving the land, and adding livestock and other forms of capital, as if to prove Adam Smith right in his 1766 conjecture that “in the northern colonies . . . the lands are generally cultivated by the proprietors, which is the most favourable method to the progress of agriculture.”26 Their 25 Wealth here refers to gross assets rather than net worth. The Mains’ data give both, but we prefer gross wealth for the purposes of national product accounting and comparisons with other GDP estimates. 26 A. Smith (1766) 1978, 523.

WHEN DID COLONIAL AMERICA GET RICH?

•

61

Table 3- 6 Conjectural Estimates of Gross Income per Capita, 1650–1774 1650

1675

1700

1725

1750

1770

1774

A. In current pounds sterling (per capita) New England, all

7.0

7.2

7.7

8.3

10.3

11.6

11.3

Boston

10.9

10.1

10.1

11.2

13.2

11.4

9.8

Other New England

6.6

7.0

7.5

8.1

10.2

11.6

11.3

—

—

10.1

10.5

11.7

14.7

13.5

Philadelphia and New York, free

—

—

—

20.1

23.6

27.2

24.0

Other middle colonies

—

—

—

—

—

—

12.4

Chesapeake

—

18.4

18.3

16.0

16.0

18.4

16.6

Lower South

—

—

24.3

24.3

23.8

24.1

24.1

Charleston, free

—

—

—

—

—

—

119.0

Other Lower South

—

—

—

—

—

—

21.3

All thirteen colonies

—

—

13.1

12.7

14.2

16.5

15.6

Great Britain

7.6

9.3

11.5

11.9

12.9

15.21

15.7

Middle colonies (w/Delaware)

B. In units of the local bare-bones consumer bundle for one person New England, all

3.6

4.6

5.5

5.9

5.8

6.7

6.1

Boston

5.5

6.4

7.3

8.0

7.5

6.5

5.3

Other New England

3.4

4.4

5.4

5.8

5.7

6.7

6.1

—

—

8.2

7.9

8.2

8.2

8.6

Philadelphia and New York, free

—

—

—

15.2

16.6

15.2

15.3

Other middle colonies

—

—

—

—

—

—

7.9

—

18.8

16.1

13.3

12.4

12.3

12.0

Middle colonies (w/Delaware)

Chesapeake

(continued)

Table 3- 6 (cont.) 1650

1675

1700

1725

1750

1770

1774

—

—

21.3

20.2

18.5

16.1

17.4

Charleston, free

—

—

—

—

—

—

86.1

Other Lower South

—

—

—

—

—

—

15.4

All thirteen colonies

—

—

10.9

10.1

10.1

10.3

10.4

Great Britain

3.9

4.8

6.5

6.4

6.6

6.7

6.2

Lower South

Note: These estimates are for the total population, including slaves.

average wealth had almost tripled (rose 163 percent) in real terms by the 1770s, bringing them close to the average wealth of the upper classes in Boston. We can assemble the total income of New England back to 1650 by combining the property incomes inferred from the Mains’ sample with measures of labor earnings. The Mains’ data help us plot the trend in the average property incomes of New England households, assuming a 1774-fi xed share of households lacking wealth and fi xed net rates return and depreciation. Next, we need to add measures of free and slave labor earnings to these property incomes. These are constructed using the wage indicators listed for New England in table 3-4. The wage data are thinner than the probate data and can trace overall labor earnings only roughly. When the labor and property incomes are placed side by side, we find a New England trend that also shows up for Philadelphia: a rise in the share of income coming from property. This estimated rise in New England was gradual, up from 9.2 percent around 1650 to 14.6 percent in the 1770s. In Philadelphia, it rose from 8.7 to 15.7 percent in just half a century, between the 1720s and 1770s. Presumably, it marched upward even faster in the South, given the steep rise in slaves per white household. A rising property share is hardly a surprising outcome for a newly settled and prosperous region. Putting the total income picture for New England and the other colonial regions together yields the conjectural income history shown in table 3-6 and figure 3-3. The average income in New England advanced

WHEN DID COLONIAL AMERICA GET RICH?

•

63

Number of units of a bare-bones consumer bundle, for a household of 4 persons, at local prices

7

6

5

Lower South Chesapeake (see price caveat in text)

4

13 Colonies 3

Middle Colonies 2

Great Britain New England

1

0 1650

1675

1700

1725

1750

1770 1774

Figure 3-3. Purchasing Power per Capita in Th irteen American Colonies and Great Britain, 1650–1774

until around 1725, and then stagnated. This chronology agrees with previous scholarship.27 Even though it was the region with the most visible progress between 1675 and 1725, it remained the poorest, as we have already seen for 1774 in chapter 2. New England’s income estimates contain an urban–rural surprise. Table 3-6 implies that the New England countryside overtook Boston’s income per capita in the 1770s, contrary to the usual urban–rural pattern. The explanation for this apparent anomaly can be found in the population denominator. Income per capita was indeed lower, even though our 1774 estimates found that Boston had slightly higher incomes per household and slightly higher wage rates than the countryside. What dragged down Boston’s relative income per capita was its higher dependency rate in the 1770s. The available census data reveal 27 New England had high growth rates to 1680, and then slower to 1710, according to Terry Anderson (1975, 171; 1979, table 3). Jones (1980, 75) appears to agree. William Davisson’s (1967) local study of Essex County, Massachusetts, also emphasized seventeenth-century growth.

64

•

CHAPTER 3

that Boston came to have a lower share of adult males in the population than either its hinterland or the other main colonial cities. One reason is that the French and Indian War took an especially heavy toll on Boston’s male population, which disproportionately supplied troops to fight in the Canadian campaign.28 Furthermore, as we have seen earlier in this section, Boston suffered a net emigration of young adult males to other colonies. The 1770s, like the Revolutionary War that followed, stand out as a nadir in the relative economic position of Boston, from which it recovered in the early nineteenth century. Middle Colonies For the middle colonies—New Jersey, New York, and Pennsylvania—it is only for the city of Philadelphia that we can follow the same approach of combining labor income with property income trends before 1774. Yet we do have aggregate regional estimates from the production side.29 For Philadelphia, which serves as our proxy for the urban combination of Philadelphia and New York City (which was then just Manhattan), we have both useful wage rates and useful averages of probated personal wealth by occupation.30 There are limitations to even these data, however. The wage rates for Philadelphia laborers and seamen extend back only to 1725. The probated wealth averages go back further to 1685–1715, but even these cover only personal estate and not real estate, and the wealth at death has not been adjusted to make it a measure of wealth of the living. Subject to these limitations, table 3- 6 reports two key fi ndings for Philadelphia, our urban representative for the middle colonies. First, its wage rates and wealth were both consistently higher than those of Boston back to 1725, and probably earlier. But second, income per capita was stagnant in Philadelphia at that high level. Inequality probably rose 28

Nash 1979, 244– 45. These are due to the recent efforts of Rosenbloom and Weiss (2014). 30 Nash 1979; B. Smith 1981, 1984, 1990. As for the countryside in the middle colonies, we do have excellent studies of Chester County, Pennsylvania (Lemon and Nash 1968; Lemon 1972; Simler 1990, 2007). Yet these focused on inequality and household headships, without reporting useful time series on wealth or wages. 29

WHEN DID COLONIAL AMERICA GET RICH?

•

65

between the mid-eighteenth century and the revolution, to judge by the rise in property values and poor relief.31 The new estimates by Rosenbloom and Weiss suggest a slow rise of real income per capita for the Middle Atlantic region as a whole— perhaps 0.13 percent a year. Their slow-growth estimates have been incorporated into table 3-6 and figure 3-3. The Upper South Allan Kulikoff, Lois Green Carr, Russell Menard, and Lorena Walsh have offered some suggestive time series indicators for the Upper South, also called the Chesapeake.32 For this rural region, the indicator that we can follow back before 1774 is based on the gross incomes of a prototypical farm or plantation deriving 22 percent of its income from tobacco sales, 11 percent from grain sales, and the remaining 67 percent from the production of farm products that were consumed either on the farm itself or in the immediate vicinity.33 This gross farm income series includes the consumption, or income retained, by servants and slaves. The time series running back from 1774 to ca. 1675 (table 3- 6 and figure 3-3) summarizes what these authors have described in rich detail. In the Chesapeake’s tobacco-based heyday of the late seventeenth century, farmers did about as well as any group other than the even richer planters in the West Indies.34 Over the next century, income per capita in the Upper South fell by a third, despite the improvement in its terms of trade (between 1700 and 1776; see table 3-3). In spite of this 31

Nash 1976a, 1976b. See Kulikoff 1976, 1979, 1986; Carr, Menard, and Walsh 1991; Walsh 1999, 2010. 33 While we use typical farm incomes as our indicator here, there are many other series that might be woven together with additional data and then used to construct an alternative time line for aggregate incomes in the Chesapeake. We know that the slave share of total population rose, at least until 1750. Walsh (2010) offers several multiyear farm accounts. Kulikoff ’s (1976, 504–13) work suggests that the mean estate wealth rose in Prince George’s County, Maryland, yet returns from different counties fi nd an eighteenth-century drop in the shares of households owning land, though the share owning slaves rose (Kulikoff 1986, 135, 154). These clues indicate rising inequality, yet the best time series on aggregate incomes remain the ones we describe in the text as well as the “Chesapeake Income Clues” and “Chesapeake Backcasting” fi les, http://gpih. 34 See Burnhard 2001. 32

66

•

CHAPTER 3

secular decline, its average incomes in 1774 were still higher than those in the northern colonies or England. Furthermore, the decline in per capita income did not signal any institutional flaw in the Chesapeake but rather the kind of diminishing returns we should expect in a rich, newly settled region with relatively free entry of newcomers.35 The Lower South There are no income-side indicators whatsoever for the Lower South (the Carolinas and Georgia) that can span across the colonial era. To judge whether the Lower South enjoyed more or less prosperity before 1774, we turn to an indicator from the production side. Mancall, Rosenbloom, and Weiss have combined different output clues to estimate the growth of regional product from 1720 to 1740 and 1740 to 1770. We equate their 1770 benchmark with ours for 1774, and interpolate to get our 1725 and 1750 benchmarks.36 The implied result for the Lower South is high but declining prosperity from the 1720s to the eve of the revolution (table 3-6 and figure 3-3). It seems to us that the income per capita decline did not reflect any problems in the South Carolina low country, since the rich coastal area underwent fast wealth per capita growth (table 3-1, fast growth panel). Rather, it is explained by the rapid settlement in the less developed frontier of inland Georgia, North Carolina, and South Carolina. 35 A caveat must be attached here. Alternative price deflators could replace the Chesapeake’s real income drop with mere stagnation over the century ending with the revolution. The prices used to convert current-price Chesapeake incomes into constant prices are in some doubt. We have divided our estimates of the Chesapeake’s nominal income by the price of a bundle of staple consumer goods, using data supplied by Robert Allen. Th is price series disagrees with that of P.M.G. Harris (1996) and used by McCusker (in Carter et al. 2006, series Eg247). The disagreement is sharpest for 1675–1700, in which the Allen series shows a 15 percent consumer price rise while McCusker shows a 14 percent wholesale price drop. Using the Harris and McCusker series, one would fi nd no significant change in real income from 1675 to 1700. Until this issue is resolved, we resist extending our estimates back before 1700. 36 Mancall, Rosenbloom, and Weiss 2003. There is, however, some confl ict between their estimates and at least one series tracing private wealth for the Lower South across the late colonial era. Menard’s (1996, 277–78) averages for the wealth of probated estates in low-country South Carolina rise from £204 sterling in 1678–98 to £1,145 sterling in 1764, although Menard attaches caveats to these figures since they do not reflect the possible changes in social selectivity or the age of those probated.

WHEN DID COLONIAL AMERICA GET RICH?

•

67

THE AMERICAN COLONIES VERSUS ENGLAND, 1650–1774 Taken together, our estimates of income per capita trends tell a clear story about an eighteenth-century, colony-wide economic performance. Table 3-6 and figure 3-3 imply that the thirteen American colonies as a whole achieved no clear growth in income per capita, but sustained their prosperity and regional rankings over the century leading up to the revolution. The movements between episodes were not dramatic, aside from the reversal of northern colonial growth after 1750 associated with the turmoil and inflation between about 1770 and 1774. How prosperous were the thirteen colonies? Our conjectural estimates clearly imply that average colonial incomes (in real purchasing power terms) were further above that of Great Britain in 1700 than in 1774, even with slaves counted as low-income residents. The nominal, or current price, comparisons imply that the colonial advantage over the home country was only 7 to 13 percent between ca. 1700 and ca. 1770—a lead that had vanished by 1774. Yet when we switch from a simple exchange rate comparison to a real purchasing power comparison, the colonial advantage jumps to something like 50 percent for all the benchmark dates from 1700 to 1774. We think this striking result will withstand considerable error.37 As support for the magnitude of the income gap between the American colonies and the mother country, we also compare worker welfare ratios (purchasing power) that Robert Allen has designed for other 37 Britain’s 1700–1774 growth rate was higher than that of its American colonies as it got a running start on the Industrial Revolution (Broadberry et al. 2012). This paragraph’s estimates, such as the “something like 50 percent” advantage of the colonies in purchasing necessities, are sensitive to (a) the number of years used for averaging each benchmark’s prices, and (b) the mixture of different places’ costs of the same bundle on both sides of the Atlantic.

(a) Table 3- 6 said the purchasing power advantage was 54 to 68 percent, agreeing with our forthcoming article in the Economic History Review, table 6. That was based on averaging prices over long periods, usually a quarter century. Yet to keep the price average closer to the income benchmark year in turbulent times, we have recalculated the purchasing powers’ using only five-year averages. Th is yields an American advantage of 36 to 55 percent between 1700 and 1774, as will be shown in figure 5-2, table 10-1, and figure 10-2. (b) The American welfare ratios also depend on how one mixes the available series for Chesapeake, Massachusetts, and Philadelphia, and the England welfare ratios depend on how one mixes London and other English price series. Yet the colonial welfare ratios were higher on any of the available measures.

68

•

CHAPTER 3

Welfare ratio = number of units of a bare-bones consumer bundle for a household of 4 persons, at local prices (log scale)

Philadelphia

7

5

New England (labor composite)

4 3

York 2

Farm labor southern England 1 1650

1675

1700

1725

1750

1775

1800

1825

Figure 3-4. Workers’ Purchasing Power in England and the Colonies, 1650–1820 (Allen)

historical comparisons. Americans earned distinctly higher real wages in the eighteenth and early nineteenth centuries (figure 3-4). This comparison shows that the wage advantage over Britain was even greater than the income per capita advantage (table 3-6 and figure 3-3). There are two explanations for this. First, the greater equality in the colonies was driven in large part by labor scarcity and land abundance. Thus, the wage–rental and wage–income ratios should have been higher, implying the bigger wage advantage over Britain. Second, those high colonial child dependency rates implied bigger gaps per worker or per household than per capita. GETTING THE PRICES RIGHT The striking transatlantic living standard contrast owes much to the fact that the bundle of basic consumer goods was much cheaper in North America than in Britain. Land-intensive food products that delivered WHEN DID COLONIAL AMERICA GET RICH?

•

69

Annual cost of a bare-bones consumer bundle for one person (£ sterling)

4

London and southern England Boston

2

Philadelphia Chesapeake

0 1625

1650

1675

1700

1725

1750

1775

1800

1825

1850

1875

Figure 3-5. The Cost of Certain Necessities in England and America, 1632–1894 (see also appendix D)

calories and protein most cheaply, like grains, beans or peas, meat, and butter or oil, dominated the bundle. The nonfoods in the bundle included soap, linens/cottons, candles or lamp oil, and fuels like firewood or coal.38 Such common necessities were almost always cheaper in the colonies than in England (in terms of current sterling; see figure 3-5). To get the living standard comparisons with England right, dividing current price, nominal incomes by the cost of such consumption bundles is certainly superior to dividing by official exchange rates, since the latter fail to capture differences in the prices of things that do not enter international trade. 38 See Allen, Murphy, and Schneider 2012, including its online supplement. Note that resourceintensive fuels (wood and charcoal) and lighting (whale oils and candles) were also cheap in America, as was land and wood-intensive housing, although the latter was omitted from the bundle. For a family of four, this bundle is assumed to cost 3.15 times what it would for an adult male living alone.

70

•

CHAPTER 3

1900

Getting the living standard comparisons right also requires comparing relative prices from the same era. The issue of who was ahead of whom in any one era must be based on contemporaneous price comparisons, not by international price comparisons taken from the late twentieth century, as Maddison did. While Maddison delivered a rich harvest of estimated growth comparisons used extensively by economists, we stress once again that his long-span procedure of deriving levels of product per capita from late twentieth-century price structures may be very misleading.39 The only way to check for a cumulative, index number bias is to gather the price and quantity data across countries from earlier centuries. Marianne Ward and John Devereux have performed exactly those comparisons using data from the nineteenth century, and Douglas Campbell has recently confirmed those findings.40 These economists find that the United States had a lower cost of living and higher real purchasing power income per capita as early as 1830. Our work in this chapter extends this American income leadership back to 1700 or earlier. Table 3-6, figure 3-3, and figure 3-4 all show that colonial America first overtook Britain in purchasing power per capita at least two centuries earlier than the Maddison figures imply. The 1700 gap between Maddison’s and our Anglo-American comparison is huge; his estimates imply that “United States”/United Kingdom = 0.83, in stark contrast with our estimate of 1.66.41 Would better price data reverse the gap in purchasing power? Here “better” might mean “conceptually best” for price-adjusting GDP, or it might simply mean more complete in the sense of covering more goods and services. 39

These are called 1990 International Geary-Khamis dollars. Ward and Devereux 2003, 2004, 2006; Campbell 2015. 41 Maddison 2001, 246– 49. The Maddison and colleagues (2013; Bolt and van Zanden 2014) revision of Maddison’s GDP per capita estimates makes the gap even more glaring, if the “United States” is meant to refer to the thirteen colonies excluding Native Americans. The figures imply that the “United States” had only 0.64 of the per capita income of England in 1650, only 0.54 of Great Britain’s per capita income in 1720, and only 0.70 of Great Britain’s in 1775. The gap between Maddison’s and our estimates remains wide for all of the nineteenth century, as chapters 5 and 6 will note. Why had so much of the American “cheap cost-of-living” advantage disappeared between the late nineteenth century and the modern purchasing power parity benchmarks like the 1990 date used by Maddison? Th is question needs research, but we suspect that world globalization and the resulting price convergence will be the main explanation. 40

WHEN DID COLONIAL AMERICA GET RICH?

•

71

The concept of a “best” GDP deflator is clear enough: one would ideally divide current-price GDP by the prices of all goods and services bundled together in the GDP of one of the countries being compared. One might suspect that our comparisons of private consumption prices have missed two main categories of gross national expenditure: capital formation and government. Yet the “GDP deflators” offered by conventional measures of growth cannot have trends much different from the price trends of the bare-bones and respectability bundles. After all, consumption is two-thirds of gross national expenditure. Furthermore, the kinds of products purchased for capital formation and government services are more poorly documented than consumer prices, which are allowed to inform the indirect proxies often used for investment and government prices. So any comparisons can only be based on the broadest set of internationally available prices, either for our calculations or for the competing calculations that build GDP from the production side. When all the historically available prices are compared across the Atlantic, they still show most goods to have been cheaper in the American colonies than in England. Of the forty commodity price comparisons that are possible for either the period 1730–1753 or 1754–1774, only three had sterling prices that were at least 25 percent higher in the American colonies, while twenty-two had sterling prices that were at least 25 percent lower in the colonies. Similar results emerge for 1792– 1808 or 1840–1860. That is, for the wider range of consumer goods that can be compared across the Atlantic—just as for Allen’s food, fuel, and four other goods—American prices were lower than English prices, when both sets are expressed in sterling.42 Thus, using a wider range of homogeneous goods still shows Britain to have been a more expensive place for the average person to live than the American colonies, as did 42 For the extended price-comparison results, see appendix D. The 25 percent figure uses the English price as the comparison base. The three colonial cases with America–England ratios above 1.25 were Pennsylvania sugar in the period 1730–1753, and Massachusetts beans and cheese in the period 1754–1774. The data sources are Gregory Clark for England, Carroll Wright for Massachusetts, Anne Bezanson and colleagues for Pennsylvania, Lorena Walsh and colleagues for Maryland and Virginia (Chesapeake), and T. Adams for Vermont after 1790. See appendix D; “Price Comparisons between America and England, Specific Goods, ca. 1650— ca. 1870,” http://gpih.

72

•

CHAPTER 3

the Allen bare-bones costs used in table 3-6 and figures 3-3 through 3-5. The still-missing price data on heterogeneous goods and services, if somehow adjusted hedonically for quality, would presumably show many prices to be lower in England than in the colonies, since those heterogeneous goods and services tended to be consumption luxuries, capital goods, and government services.43 A simple pattern suggested by these welfare ratios and related work on early Canada is that sparsely populated, rich-land frontiers have an egalitarian price structure.44 The more homogeneous goods that loomed so large in the budgets of common folk were considerably cheaper in frontier North America than in the more developed parts of northwestern Europe. By contrast, at the top end of society, the richest Londoners or Parisians enjoyed cheaper (quality-adjusted) free domestic servants, fashion wear, carriages, fine wines, and tickets to Handel concerts. For such a lifestyle, the top 1 percent of colonists could not begin to compete in purchasing power. If those luxuries were the things your family demanded most in 1774, you would not have gained much consumer satisfaction by being the richest colonial family, living in Charleston, South Carolina. If cheap necessities made North America a “best poor man’s country” even in terms of the cost of living, while rich North Americans faced a much higher cost of luxuries than their counterparts in London, where was the switch-over point at which an individual faced about the same cost of living on both sides of the Atlantic? If it were only paupers who found things cheaper in North America, then our use of the barebones consumption basket in calculating the cost of living might seriously overstate the relative purchasing power of the North American 43 For comparisons of middle-class consumption baskets, such as those recently presented by H. M. Boot (1999, 649–55) for London in 1823–24, the transatlantic contrast might still show relative cheapness in America, partly because of the lower American prices for meat, a relative luxury. 44 Geloso (forthcoming) is fi nding that in New France (Quebec) between 1688 and 1760, the welfare ratios for ordinary unskilled workers were above those in France using an Allen-type, bare-bones consumption standard. By contrast, using the wage rates of more skilled workers and a more substantial “respectability” consumer bundle, it is not so clear that those on the Canadian frontier were better off than their French counterparts.

WHEN DID COLONIAL AMERICA GET RICH?

•

73

population as a whole. What do we know about those in the middle of  the distribution? For them, what is called a “respectability” consumption basket—one that would cost about three times as much as the bare-bones basket—might be more relevant. Stephen Broadberry and his co-authors have specified what such respectability would have cost in Great Britain at the start of the nineteenth century. Their respectability basket features more meat, and already-made bread in place of the raw oats in the bare-bones basket, and adds beer and cheese as well.45 We can compare the American and British prices of nine goods in this respectability bundle of consumer goods in the early nineteenth century, and tote up the costs in the two countries. It turns out that even for households that could afford the respectability bundle, Americans still paid about 16 percent less than their social counterparts in Britain, mainly because bread and meat were so much cheaper in land-abundant North America. So even for people who could afford such “respectable” household expenditures, and not just for the poorest, living was cheaper and real purchasing power was greater in America.46

COLONIAL RESULTS AND POSSIBLE ERRORS Our backcasts seem to be consistent with the implications of what is known about colonial demography, trade, staple export prices, and frontier settlement. They imply the following: • Between the late seventeenth century and the revolution, the mainland American colonists had higher average purchasing power than Britain or continental Europe in terms of basic consumer goods. That is, the colonial real income advantage over England in 1774 was already true in 1700 or even earlier. • The American income per household lead was even greater than its income per capita lead. This was because American colonists had very high fertility rates, and their children had relatively high survival rates. 45

Broadberry et al. 2015, 339. The nine goods in the comparison were bread, beans, beef, butter, cheese, eggs, soap, linen cloth, and candles. For the calculations, see appendix D, table D- 4. 46

74

•

CHAPTER 3

Thus, America had a child dependency rate much higher than Europe and even higher than the Third World today. • Our evidence supports the slow- or no-growth side of the colonial debate, as conjectured by Mancall, Rosenbloom, and Weiss. The support comes not only from our backcasting estimates but also from our survey of causal factors that should have kept colonial growth slow, once the initial discovery of tobacco and rice cultivation in the South and the fishing banks in New England and a few other staple opportunities had been exploited. The stimulus from exports, which was stronger for the South, must have tapered off as the export share of the colonial economy declined. Staple price volatility should also have dampened growth in the colonies, as it has done in developing countries more recently. Finally, high and rising child dependency rates should have held down income per capita growth. • Income inequality may have risen before the revolution, but probably not among free whites. A rapidly expanding frontier held down inequality among free colonists—that is, the movement of population away from coastal cities and towns, and to the inland frontier with lower income per capita.47 This ruralization was dramatically reversed in the nineteenth century, as we will see in chapter 5.

Our backcast estimates could be wrong, of course, since they are based on only limited evidence for the years before 1774, although the theoretical slow-growth predictions coming from colonial demography, trade, staple prices, and frontier settlement seem sound. At this point, the economist might ask for “sensitivity analysis” that “tests” whether the estimates could be wrong enough to give a false conclusion. Even though we find the income trends plausible, there is no objective way to put numerical bounds on the possible errors in the estimates given such a data-thin environment. Still, it’s not hard to see just how much error is needed to overturn our conclusions regarding the impact on the relative purchasing power of the colonies. Table 3-6 tells us how 47 Why would young men move to the still-undeveloped frontier if the income per capita was lower there? Potential migrants do not respond to income per capita differences but rather to income per worker differences within their social group and stage of the family formation cycle. Since most were farmers, the availability of cheap land and the possibility of lifetime capital gains certainly mattered.

WHEN DID COLONIAL AMERICA GET RICH?

•

75

large the error would have to be to overturn our conclusion that the colonies had higher incomes per capita than Britain. If our American and British estimates for 1774 are correct, then the net error for 1700 would have to be 67 percent (or 58 percent for 1725) to cancel the colonial advantage in purchasing power.48 Such large errors strike us as unlikely. Yet we welcome revisions based on better data. 48

76

As noted elsewhere, the British estimates are taken from Broadberry et al. 2012, 2015.

•

CHAPTER 3

CHAPTER 4

Losing the Lead The Cost of Revolution and Independence

T

he darkest times, when people suffer most, are also often the darkest for the historian trying to assess just how people lived through those times. So it was in the quarter century between 1774 and 1800, the latter our second benchmark year. Three tumultuous events took place between those dates: the Revolutionary War, the troubled Congress of Confederation of 1781–1789, and then the beginning of the federal republic. Since 1800 was the earliest benchmark following 1774 for which the evidence would support a social table, this chapter will not be able to assess the impact of those three events separately but rather only their combined impact. We can, however, shed light on a number of questions. How big was the economic cost of the Revolutionary War and the dysfunctional confederation? What happened to the American per capita income lead over Britain during this troubled quarter century? Which region was hardest hit during those years, the South or the North? What happened to the income per capita gap that favored the urban coastal regions over the rapidly filling subsistence-plus hinterland? And when our 1800 estimates are compared with those constructed by others for 1840, can we see the beginnings of modern economic growth? On the labor income side, our procedures for 1800 are roughly the same as those we applied to 1774 in chapter 2, although the supporting data are more abundant. The free-labor wage data are of comparable depth to those of 1774. The population data are on sounder footing given the availability of the second US census. Labor force participation rates (LFPRs) by location, age, sex, and slave/servant status rely once more on the work of Weiss. Occupations and assumptions about slave and servant

retained income (or more aptly, their “maintenance” or consumption) are also taken from the same kind of sources as for 1774, although they are of higher quality and more abundant.1 What makes the 1800 income estimates distinctive, however, is not the construction of labor incomes but rather that of property incomes.2 First of all, there is an important difference between the property income data sources for our first two social table benchmarks, 1774 and 1800. In 1798, Congress voted the nation’s first direct tax to finance a possible war against France. Each state was assigned a revenue quota proportional to its population. Given its responsibility to deliver this population-based total, each state was to tax its households according to real estate values and the numbers of able-bodied slaves held. Fortunately, tables of the assessed values have survived. So have some of the individual returns. In the District of Columbia, for example, former President George Washington was assessed for three houses and four lots worth a total of $1,100 in the place listed as “His City.”3 A handy feature of the 1798 wealth tax returns is that they were aggregated at the time.4 But a serious drawback of the 1798 return is that it does not report property income by occupation, making it impossible to add up labor and property income for each occupation/location cell in our 1800 social table. Thus, we cannot document the distribution of (total) income for 1800, although we can document the distribution of labor earnings as well as aggregate property income and total income by location. In short, while we cannot compare the distribution of total 1

On slave incomes in 1800 and other dates, see appendix C. For a detailed explanation of our 1800 property income estimates, see appendix A. 3 “Maryland Tax Lists, 1798–1805,” MS 807, General and Particular Assessment Lists of Lands, Lots, Bldgs., and Wharves for the District of Columbia, Maryland Historical Society, Baltimore. Washington apparently knew in advance the real estate potential of “His City,” and had advised his friend Benjamin Stoddert, later the secretary of the navy, of that potential: “After George Washington was elected president of the United States, he asked Stoddert to purchase key parcels of land in the area that would become the nation’s capital, before the formal decision to establish the federal city on the banks of the Potomac drove up prices there. Stoddert then transferred the parcels to the government. He also helped found the Bank of Columbia to handle purchases of land in the District of Columbia for the federal government.” Naval Historical Center, http://www.history.navy.mil/bios/stoddert.htm (accessed August 10, 2015). 4 As appendix A reports, we were able to project the 1798 wealth estimates to 1800 asset values using independent sources, including Blodget 1964. For the totals, see “US Realty Tax Returns for 1798,” http://gpih. 2

78

•

CHAPTER 4

income in 1800 with 1774, we can say quite a bit about changes in regional income gaps, the earnings distribution, the skill premium, urban–rural income gaps, and other component parts. Unlike those for 1774, our 1800 total income estimates in table 4-1 are not above those estimates constructed by others. In fact, our estimates for the new nation as a whole are in the lower half of several competing estimates (shown on the right-hand side of table 4-2). Our totals for the Lower South match those of Mancall, Rosenbloom, and Weiss, even though we use the income approach and they use the output approach.5 It might seem comforting that our 1800 estimates are so close to those constructed by others. Yet ours might have been a bit higher than most if we had been able to make all the adjustments that we feel are warranted. We are especially concerned about two such adjustments. The first potential adjustment—which can be quantified—is one already mentioned in chapter 2 and revisited here in table 4-3. Using the interest rate on public debt as a measure of the opportunity cost of assets, it appears that the net rate of return on property was higher in 1800 than in 1774, presumably in response to war and postwar inflation, financial disruption, and perhaps even productivity advances.6 As we will see, if the interest rate was around 8 percent in 1800 and around 6 percent in 1774, then the measured decline in real per capita income between the two dates would have been 14 percent rather than the bigger decline of 20 percent implied by the baseline estimates (tables 4-4 and 4-5). The second adjustment—which cannot be quantified—relates to an omission from the baseline 1800 estimates. We have no 1800 data or even guesstimates about farm operators’ pure residual profits, as distinct from returns to their land and other assets plus the implicit value of their own (and their family’s) farm labor. For 1774, we were able to use a few testimonies unearthed by Jackson T. Main to guesstimate that the farm profit residual was 16.3 percent of all farm operators’ income in New England, 19.7 percent in the Middle Atlantic, 30.5 percent in the South, and 25.7 percent for the thirteen colonies as a whole. We cannot apply these ratios to 1800, however, since we lack any delineation between farm operators 5 6

Mancall, Rosenbloom, and Weiss 2003. Homer and Sylla 1996, 274–96.

THE COST OF REVOLUTION

•

79

Table 4-1 Estimated American Incomes, Eastern Seaboard, 1800  New England

Middle Atlantic

South Atlantic

All thirteen colonies (fifteen states and District of Columbia)

Gross income, millions of current dollars ($4.44/£ sterling) FTEs, free own-labor income

73.65

84.20

87.77

245.62

Ditto, with part time

66.57

76.91

80.88

224.36

Slave retained earnings

0.07

2.10

37.34

39.51

Gross property income

21.39

47.83

89.77

158.99

Gross total income

95.11

134.13

214.88

444.12

Ditto, with part time

88.03

126.83

208.00

422.86

Relevant denominators Free labor force

334,685

380,162

402,504

1,117,351

Total labor force

335,500

404,900

835,590

1,575,990

Free population

1,231,671

1,423,924

1,428,695

4,084,290

Total population

1,233,011

1,464,548

2,222,221

4,919,780

Some average incomes (current 1800 dollars) Free own-labor income per member of free labor force full-time assumptions

220.06

221.48

218.06

219.82

part-time assumptions

198.90

202.31

200.94

200.80

per slave labor force member

85.89

84.89

86.22

86.15

per capita

52.24

51.69

47.06

47.29

per capita, free

71.42

87.60

119.45

93.86

per capita, total

71.39

86.60

93.60

85.95

Slave retained earnings

Gross total part-time incomes

Table 4-2 Alternative Estimates of National Income, 1774 and 1800, in Current and 1840 Dollars (Millions)

United States (original thirteen)

1774 (current $m)

1774 (1840 $m)

1800 (current $m)

1800 (1840 $m)

Source

142.2

152.5

—

—

GDP: McCusker 2000

131.7

141.6

—

—

Gross inc: Jones 1980

173.2

185.7

444.1

305.9

Gross income: Lindert and Williamson 2013b

164.1

176.3

422.8

291.3

Gross income, part time: Lindert and Williamson 2013b

508.7

350.3

GDP: McCusker 2000

United States (all) 150.3

161.6

515.5

355.5

GDP: Mancall and Weiss 1999

135–157

145–169

—

—

GDP: Gallman 1972

134.8

145.0

500.1

344.9

GDP: Goldin and Lewis 1980

132.6

142.6

—

—

Narrow GDP: Weiss 1992

430.9

297.2

Berry 1988

446.3

307.8

David 2005

510.4

351.5

GDP: Mancall, Rosenbloom, and Weiss 2003

470.7

324.2

Gross income: Lindert and Williamson 2013b

448.1

308.7

Gross income, part time: Lindert and Williamson 2013b (continued)

Table 4-2 (cont.)

New England

Middle Atlantic

South Atlantic

Lower South Atl.

1774 (current $m)

1774 (1840 $m)

1800 (current $m)

1800 (1840 $m)

Source

35.5

38.2

—

—

Income: Jones 1980

34.6

37.1

95.1

65.5

Gross income: Lindert and Williamson 2013b

36.5

39.3

—

—

Income: Jones 1980

39.7

42.6

134.1

92.4

Gross income: Lindert and Williamson 2013b

59.2

63.6

—

—

Income: Jones 1980

98.9

106.0

214.9

148.0

Gross income: Lindert and Williamson 2013b

22.0

23.7

93.5

64.4

GDP: Mancall, Rosenbloom, and Weiss 2003

—

—

94.1

64.8

Gross income: Lindert and Williamson 2013b

and free farm laborers in either the census or the Weiss labor force estimates on which we rely. As a result, the true decline from 1774 to 1800 might turn out to be a little less when and if an estimate of farmers’ residual profits around 1800 can be added in the future.

INCOME LEVELS AND ECONOMIC DECLINE, 1774–1800 Our estimates of labor and property incomes shed new light on the movements of American incomes over a quarter century of war, postwar economic adjustment, and national emergence. The levels and 82

•

CHAPTER 4

Table 4-3 Alternative Property Incomes and Total Incomes, 1774 and 1800 1774

$ millions in 1774 (at $4.44/£) New England

Middle colonies

South

All thirteen Colonies

Baseline estimate using 6% net rate of return Gross property incomes

4.840

8.372

23.830

37.042

Net property incomes

3.662

6.534

15.736

25.932

Total gross personal incomes

36.064

38.281

98.814

173.159

Total net personal incomes

34.886

36.444

90.719

162.049

Alternate estimate using 8% net rate of return Gross property incomes

6.061

10.550

29.075

45.685

Net property incomes

4.883

8.712

20.981

34.575

Total gross personal incomes

37.285

40.459

104.058

181.802

Total net personal incomes

36.106

38.622

95.964

170.692

1800

$ millions in 1800 Middle Atlantic

South

All fifteen states and District of Columbia

New England

Baseline estimate using 6% net rate of return Gross property incomes

21.391

47.829

89.772

158.993

Net property incomes

16.787

29.346

46.490

92.624

Total gross personal incomes

95.112

134.128

214.880

444.119

Total net personal incomes

90.508

115.645

171.598

377.750

Alternative estimate using 8% net rate of return Gross property incomes

26.987

57.611

105.269

189.867

Net property incomes

22.383

39.129

61.987

123.498

Total gross personal incomes

100.707

143.910

230.376

474.994

Total net personal incomes

96.103

125.427

187.094

408.625

Table 4-4 Real Income per Capita, 1774–1840 New England

Middle Atlantic

South Atlantic

All three regions

Gross personal income per capita (in McCusker’s 1840 prices) Baseline 1774

53.68

(73)

64.08

(87)

91.77 (124) 74.02

(100)

Baseline 1800

49.20

(85)

59.66

(101)

64.46 (107) 59.19

(100)

Alternative 1800

52.09

(85)

64.02

(102)

69.02 (107) 63.30

(100)

Weiss-Easterlin 1840

129.28 (118) 120.19 (109)

84.84 (77)

109.89 (100)

Per annum growth 1774–1800 (%) −0.33

−0.27

−1.35

−0.86

Using alternative 1800 −0.12

−0.00

−1.08

−0.60

2.44

1.77

0.69

1.56

Using alternative 1800 2.30

1.59

0.51

1.39

0.96

-0.12

0.60

Using baseline 1800

Per annum growth 1800–1840 (%) Using baseline 1800

Per annum growth 1774–1840 (%) 1.34

composition of total personal income are shown in tables 4-2 through 4-4. To facilitate comparisons of 1800 with 1774, we include 1774 in all three tables, retain the three region definitions used by Alice Hanson Jones, and make comparisons across the quarter century based on a geographically fi xed “nation.” In 1800, that “nation” is defined as the original thirteen colonies, which in 1800 included the easternmost fifteen states plus the District of Columbia.7 We start with a key result that stands out in the data: real income per capita dropped by about 20 percent over that quarter century (compare 7 That is, Connecticut, Delaware, the District of Columbia, Georgia, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, North Carolina, Pennsylvania, Rhode Island, South Carolina, Vermont, and Virginia.

84

•

CHAPTER 4

Table 4-5 Wage Gaps and Skill Premiums, 1774 and 1800 North

South

All thirteen colonies

Urban/rural male unskilled earnings

1.071

1.565

1.262

Average urban earnings/average rural earnings

2.154

4.078

2.793

Urban artisan/urban male unskilled earnings

1.388

1.094

1.269

Urban white collar/urban male unskilled earnings

6.271

8.376

6.933

Urban/rural male unskilled earnings

1.044

1.056

1.047

Average urban earnings/average rural earnings

1.585

1.095

1.348

Urban artisan/urban male unskilled earnings

2.229

2.074

2.017

Urban white collar/urban male unskilled earnings

2.047

1.935

2.002

1774

1800

the 1774 and 1800 baseline estimates in 1840 prices in table 4-4).8 This 20 percent drop was certainly much greater than it was in America’s recent Great Recession of 2007–2010, and nearly as big an economic disaster as the 1929–1933 drop in the Great Depression. If other authors who have reported evidence of brisk income gains across the 1790s are correct, then the Revolutionary War and confederation turmoil may have been America’s greatest income per person slump ever (in percentage terms), and was inflicted on a much poorer population than the Americans who suffered the Great Depression of the 1930s. The magnitude of the economic disaster between our benchmark years, 1774 and 1800, stands out clearly, but what about in between? Attempts to span that historical gap with numbers have had little success, explaining why the period is called a statistical dark age. As with late eighteenth-century 8 The percentage decline of 1774–1800 is sensitive, however, to the uncertainty about Main’s data on farm profits ca. 1774. For a discussion of the baseline versus alternative interpretation of the evidence on farm profits, see appendix A.

THE COST OF REVOLUTION

•

85

France, early nineteenth-century Latin America, early twentieth-century Russia, and late twentieth-century Africa, historians of the early United States have had great difficulty bridging the data gap across the revolutionary upheaval and early nation building. Using a variety of evidence, some economic historians have emphasized the strong growth experienced across the 1790s.9 An impressive performance of the American economy during that decade could be explained by “growth miracle” forces that always seem to follow destructive wars when only physical capital needs to be rebuilt. It could also be due to the policy wisdom of Alexander Hamilton and other founding fathers. Or it could be explained by the recovery of foreign markets. While we will offer our own informed guesses, we do not yet know which of these was most important. Of one thing we are certain, however: the more we come to accept a sanguine view of the 1790s, the more we must infer a truly huge economic disaster between 1774 and 1790. Charting the depths of the Revolutionary War and postwar depression requires an answer to this question: How deep would the per capita income loss have been from 1774 to 1790 if these other economic historians are right about the growth from 1790 to 1800, and if our estimate of the net decline from 1774 to 1800 is also correct? This question has eight possible conjectural answers, based on our two estimates for 1800 (“baseline” and “alternative” in table 4-3) times the four leading series documenting real income per capita growth from 1790 to 1800. The four series are those offered by Richard Sutch, Louis Johnston and Samuel Williamson, Thomas Berry, and John McCusker.10 All eight conjectures imply a big drop in income per capita between 1774 and 1790. Based on the Sutch estimates of growth across the 1790s and our alternative estimate for 1800, GDP per capita might have dropped by 18 percent over those sixteen years, 1774–1790, the lowest estimated fall. The largest estimated drop is 30 percent, based on Berry’s series and our baseline estimate for 1800. All these estimates certainly agree with the statement by McCusker and Menard that the “colonists paid a high cost for their freedom,” and by Kulikoff that the drop in incomes may have been “equal to the early years of the Great Depression.”11 9

The best examples include Berry 1968, 1988; Johnston and Williamson 2010; Sylla 2011. See series Ca11, Ca16, and Ca17 in Carter et al. 2006; McCusker 2000. 11 McCusker and Menard 1985, 374; Kulikoff 2005, 27. 10

86

•

CHAPTER 4

REVOLUTIONARY SHOCKS: WAR DAMAGE, DIMINISHED TRADE, AND A CRISIS AT THE TOP What could have caused such enormous income losses over so many years? There is good prima facie evidence that three related negative shocks could have been large enough to cause the deep depression between 1774 and 1790. The first was the economic destruction of the war itself as well as the impact of nearly two decades of hyperinflation and a dysfunctional financial system. Narratives listing the many forms of wartime economic destruction and disruption, with emphasis on the countryside, can be found in the qualitative literature.12 But some detail here might give the reader a better sense of the economic damage. We start with Kulikoff ’s summary: War tore the country apart; refugees and soldiers wandered the countryside; armies stole cattle and crops; people starved. Trade atrophied; money lost its value; jobs disappeared. . . . The war ended but the misery continued. .  .  . This depression seared the memories of all who lived through it.13

And the financially dysfunctional economy of the 1780s inhibited recovery: “State and national financial policies . . . kept small-scale farmers and artisans impoverished, reducing their ability to rebuild.”14 But the more permanent damage of the war was harder to restore. Immigration—so important to New York, Pennsylvania, and the rest of the Middle Atlantic—ceased, thereby increasing labor scarcity. In the South, slaves took advantage of the confl ict to run away, and many joined British forces. Indeed, the “loss of slaves . . . numbered 30,000, a tenth of all working-age slaves.”15 About a third of all white males over the age of sixteen served in the Continental Army, their state militia, or the British Army—reducing wartime private sector capacity—and 10 to 15 percent died. Such figures imply a reduction of the free labor 12

Mancall 1991, 130–59; Henretta 1991. See especially Kulikoff 2000, 256–80; 2005; 2014. Kulikoff 2014, 134. 14 Kulikoff 2014, 135. 15 For the estimate of thirty thousand runaway slaves, see Kulikoff 2014, 144. For a narrative of slave fl ight to Britain’s side in the Revolutionary War and especially the War of 1812, see A. Taylor 2013, 13–30, 132–390. 13

THE COST OF REVOLUTION

•

87

force by perhaps as much as 5 percent. Many also suffered wounds that made heavy farmwork impossible when peace returned. The second negative shock consisted of the disruption of overseas trade during the revolution and, after 1793, the Napoleonic Wars.16 As James Shepherd and Gary Walton have noted in their seminal work, the loss of trade in the 1780s was domestic as well as foreign, because the loose confederation that preceded the federal union allowed the new states to tax interstate trade, with disastrous effects on the magnitude of that trade as well as trade with the rest of the world.17 Indeed, “eleven of the thirteen colonies passed their own tariff laws during the 1780s.”18 Yet the larger and longer negative shocks to America’s trade involved Britain and its possessions. Available price and trade data show that the colonies, especially in the Lower South, suffered heavy volume and value losses in trade and shipping as the war deepened, and that they recovered only slowly and partially across the 1780s. Chesapeake tobacco exports fell from a hundred million pounds per year in the early 1770s to about nine million per year in the late 1770s and early 1780s. In 1781, exports of Carolina rice to Britain were only a fift h of the prewar levels. The middle colonies and New England suffered as well: By 1779 . . . Philadelphia tonnage shipped [had fallen] to one-seventeenth of the prewar average .  .  . forcing merchants out of business [and] fisheries—essential to the northern New England and Long Island economies—closed down as the British prevented American ships from going to the fishing banks.19

Between 1768–1772 and 1791–1792, commodity exports in real per capita terms rose by a trivial total of 1.2 percent in New England and a modest 9.9 percent in the Middle Atlantic, but fell by a spectacular 39.1 percent in the Upper South and an even bigger 49.7 percent in the Lower South, yielding a decline of 24.4 percent for the thirteen colonies as a whole.20 16

In particular, see O’Rourke 2006. Shepherd and Walton 1976. See especially Mittal, Rakove, and Weingast 2011; Shepherd 1993. 18 Irwin 2011, 94. 19 Kulikoff 2014, 136–37. 20 For the thirteen-colony totals, see Shepherd and Walton 1976, especially table 5 and the surrounding text. The estimate by Mancall, Rosenbloom, and Weiss (2008, table 1) for the Lower South refers to the twenty years 1770–90; their table 1 estimates an even larger 67 percent fall. 17

88

•

CHAPTER 4

The most painful of these shocks was the loss of well over half of all trade with England between 1771 and 1791. In addition, America lost imperial bounties like those on the South’s indigo and naval stores, and New England suffered from Britain’s reversal from colonial bounties to prohibitive duties on its whale oil exports. While these negative demand shocks to American commodity exports were large, especially for the Lower South, the initial share of exports in regional income was only about 6 to 7 percent in the early 1770s, according to the Shepherd and Walton export values per capita in 1768–1772 and our 1774 income estimates for the three main regions combined (chapter 3). A 24.4 percent per capita trade fall times a 6 to 7 percent initial share of trade in income equals no more than a 2 percent total fall in per capita income. The numbers are bigger for the South, where per capita exports fell by perhaps 45 percent and the initial trade share was 7.1 percent, implying a per capita income loss of more than 3 percent. These calculations only deal with foreign trade losses; the trade losses would be considerably higher if they included the decline in intercolonial and subsequent interstate trade between 1774 and 1790. Finally, these negative trade shocks created a move to more subsistence farming in the staples districts and presumably lower agricultural productivity.21 While we cannot be precise, it appears to us that most of the 20 percent fall in per capita income between 1774 and 1800 must have been due to war damage and postwar financial problems. The trade and world market shocks, while significant, were a smaller source of the disaster, and they also recovered more quickly. Perhaps this was predictable, given that those trade shares were so small nationwide. But trade was a far bigger share of the economies lying along the urban, coastal strips of each of the main regions—an issue we assess next. The third major negative shock involved what we call a “crisis at the top,” and it was felt primarily in the coastal cities and river towns. Obviously, this shock was related to the trade losses (after all, coastal and urban trade shares were much higher there than in the rural hinterland), but may have transcended them and therefore could have caused even greater income losses. America’s urban centers were severely 21

Kulikoff 2014, 11–12.

THE COST OF REVOLUTION

•

89

damaged by British naval attacks, blockades, occupation, and the eventual departure of skilled and well-connected loyalists, especially from Charleston, New York, and Savannah. In Richard Hildreth’s summary, “One large portion of the wealthy men of colonial times had been expatriated, and another part impoverished.”22 The damage to urban economic activity was considerable, and potentially enough to bring great declines in per capita incomes, even though the population kept growing. To identify the extent of the urban damage, one could start by noting that the combined share of Boston, Charleston, New York City, and Philadelphia in a growing national population shrank from 5.1 percent in 1774 to 2.7 percent in 1790, recovering only partially to 3.4 percent in 1800. To the degree that urbanization is a close correlate of levels of economic development, this big fall in the American urban population share certainly confirms what our income estimates document. There is even stronger evidence confirming an urban crisis (table 4-5): the ratio of urban to rural male, unskilled earnings fell from 1.3 to 1.0, with the biggest fall in the South, 1.6 to 1.1; the ratio of earnings per worker in urban jobs relative to that of rural jobs dropped from 2.8 to 1.3, with the South undergoing the biggest fall, from 4.1 to 1.1; and the ratio of average urban white-collar earnings to that of urban unskilled fell even more, from 6.9 to 2.0, with, once again, the biggest fall in the South, from 8.4 to 1.9. This evidence for an urban crisis during the quarter century 1774–1800 supports the view that America had not yet recovered from the revolutionary economic disaster even by 1800, and especially along the coastal strip of older and more urban settlements. Finally, while we cannot estimate the change in the inequality of total income, we can estimate earnings distributions and have found a greater equality of labor earnings in 1800 than in 1774. As was true of the Civil War (chapter 6), World War I, and World War II (chapters 7–9), the higher-paid occupations suffered more than the lower-paid during the revolution and early nation building.

22

90

Hildreth 1849, 3:465– 66, quoted in McCusker and Menard 1985, 365.

•

CHAPTER 4

LONGER-RUN GROWTH IMPLICATIONS How do our income estimates for 1800 fit into accepted narratives about the longer-run development of the young republic and its regions? Our new higher income estimates for 1774 have already reopened the debate about growth during the long colonial period (chapter 3). We are now in a position to use our 1800 income estimates to reopen another debate: America’s growth performance in the early nineteenth century both for the nation and its regions. We do so in two steps. Th is section compares our income estimates for 1774 and 1800 with those developed by Richard Easterlin and others for 1840, as an indirect test of the internal plausibility of two sets of estimates based on wholly different methods.23 Chapter 5 will take the next step, exploring the implications of our estimates for growth rates over the entire antebellum period up to 1860. Table 4-4 supplies our real per capita income growth estimates for each of the three regions and the three combined (the “nation” consisting of the thirteen original colonies), and it does so for 1774–1800, 1800– 1840, and 1774–1840. For the entire seven-decade period, real per capita incomes in the three-region “nation” grew at 0.6 percent per annum. While this rate was well above the eighteenth-century colonial performance, it still falls short of the modern economic growth hurdle.24 Over those seven decades, the South Atlantic fell behind the per capita income of the East Coast “nation” (from 24 percent above the national average in 1774 to 23 percent below the average in 1840), while New England (up from 27 percent below the average to 18 percent above) and the Middle Atlantic (up from 13 percent below the average to 9 percent above) pulled ahead (table 4-4). Between 1800 and 1840, per capita income in the North grew rapidly—2.3 to 2.4 percent per annum in New England, and 1.6 to 1.8 percent per annum in the Middle Atlantic—and industrial growth grew even faster in the Northeast, at 5 percent per annum.25 These rates 23

Easterlin 1960, 1961. As Douglas Irwin and Richard Sylla (2011, 4) remind us in the introduction to their Founding Choices, growth is considered modern if per capita income growth reaches 1 percent per annum or more for long stretches of time. 25 J. Davis 2004. 24

THE COST OF REVOLUTION

•

91

exceeded by far those in western Europe in the late eighteenth and early nineteenth century. By contrast, the South Atlantic grew only 0.5 to 0.7 percent per annum, pulling down the “national” average to a stillimpressive 1.4 to 1.6 percent per annum. By the modern economic growth standard suggested by Kuznets, these estimates argue that the United States had joined that growth club some time before 1840, and was perhaps one of the first nations to do so. How do the new growth rates in income per capita for 1800–1840 compare with the rates implied by the other estimates reported in table 4-2? Our growth rate for the South Atlantic of 0.51 to 0.69 per annum is about the same as that of the Mancall, Rosenbloom, and Weiss estimate for the Lower South, 0.53 to 0.79 percent. For the United States as a whole, Weiss offered three estimates ranging between 0.56 and 0.80 percent per annum, far below David’s estimate of 1.12 to 1.28 percent. Our estimates of 1.39 to 1.56 are even higher than David’s, although adding the currently missing farmers’ residual profits in 1800 would lower our 1800–1840 rates a bit.26 Table 4-4 suggests that the modest growth performance from 1774 to 1840 was driven by two special events: southern relative decline, and the economic disaster associated with the Revolutionary War and confederation turmoil. The decline of the South Atlantic region—its absolute income loss over the last quarter of the eighteenth century, followed by its relatively slow growth over the next four decades—stands out as a classic example of what has come to be called a “reversal of fortune.”27 The South Atlantic was already well behind the Northeast and the national average by 1840, having been well ahead of all other regions in 1774. Supporting this reversal of fortune, and noted in chapter 2, is the absence of any evidence that the colonial South had a large army of poor whites in 1774. We note again that the few local colonial censuses and tax records available reveal that nearly all southern white households around 1774 were assessed as having positive wealth. Why the reversal of fortune for the South? It could have been driven in turn by a falling 26 Mancall, Rosenbloom, and Weiss 2003; Weiss 1992, 27, table 1.2; David 1967, table 8; David 1996). For an excellent brief survey on growth estimates for early America, see Sylla 2011, 81–83. 27 Acemoglu, Johnson, and Robinson 2002.

92

•

CHAPTER 4

terms of trade, diminishing returns on those fertile coastal lowlands, institutional failure (stressed by Daron Acemoglu and James Robinson, and Stanley Engerman and Sokoloff, and others before them), exceptionally severe damage incurred in the Revolutionary War, or all four.

WHAT HAVE WE LEARNED? Constructing social tables for 1774 and 1800 has yielded a bountiful harvest. Here’s what we have learned about the 1774–1800 era. First, real per capita incomes in America fell severely between 1774 and 1790, perhaps by as much as 30 percent. The rapid growth after 1790 failed to make up for that steep decline during the war and early independence, since per capita income still fell by an estimated 20 percent between 1774 and 1800. While American per capita incomes were falling, they were rising in Britain (by 0.6 percent per annum between 1774 and 1800), engaged as it was in its Industrial Revolution.28 Thus, while the American colonies led the mother country in income per capita in 1774 (a lead of about 50 percent in real purchasing power), they probably had lost it by 1790 (American real per capita income was down by perhaps 30 percent and the British real per capita income was up by about 10 percent). By 1800, despite some recovery across the 1790s, the new nation’s incomes still averaged about 8 percent less than those of Great Britain. There is a modest positive side to this economic disaster associated with revolutionary war and early independence. Every country that shakes off colonial imperialism pays a heavy price, and in the growth and development literature that price is called “lost decades.”29 The positive side is that it took the Americans less than two lost decades of war and postwar adjustment to begin modern economic growth, while it took Latin America about five decades (the 1820s to 1870s) and Africa the same (the 1960s to 2000s). Second, the American North was undergoing fast, modern economic growth from 1800 to 1840, easily clearing the Kuznets per capita income growth bar of 1 percent per annum. Over those four decades, New 28 29

Broadberry et al. 2015, 241– 42, appendix 5.3. Bates, Coatsworth, and Williamson 2007.

THE COST OF REVOLUTION

•

93

England recorded a per capita income growth rate of more than 2.3 percent per annum, and the Middle Atlantic a rate of more than 1.6 percent. As we will see in the next chapter, these growth rates are much faster than those recorded by western European nations at that time. As economists seek explanations of comparative growth performance, perhaps the more relevant comparison is of others within the American Northeast, not with the United States as a whole. Third, there is the economic performance of the frontier versus the settled coastal areas to consider. We noted in chapter 3 that colonial America behaved like a dual economy. While the coastal economies in each colonial region achieved relatively fast income per capita growth by exploiting their export staples, the subsistence-plus hinterland did not. Thus, the income per capita gap between the two must have widened in the century before 1774. What went up, then went down: that is, between 1774 and 1800, the income per capita gap between the richer coast and the poorer inland frontier narrowed, as the former absorbed much bigger war- and foreign-market-related damage. In spite of the economic disaster, population growth and frontier settlement continued at a rapid pace. What was the impact of settlement on regional average incomes between 1774 and 1800? The data are not good enough to offer even imprecise answers to that question, but those for 1800– 1860 will suffice to address that question for the antebellum period of modern economic growth in the next chapter. Fourth, regional inequality was an important feature of the American colonies. Chapters 2 and 3 have shown that southern per capita income was far above that of New England and the middle colonies in 1774, partly driven by high and, in both the Upper and Lower South, rising terms of trade. Furthermore, poor whites were much less common in the South than in other colonies. Things had changed by 1800, and even more so by 1840. It appears that the South Atlantic—or what is called the Lower South and Upper South (or the Chesapeake) by colonial historians—underwent a reversal of fortune between 1774 and 1840, dropping from being the richest American region with the fewest poor to being the poorest region with the most poor. We know the South incurred relatively heavy war damage, and that its average slaveholding declined. We also know that foreign demand for its exports slumped hugely. Was this enough to explain the reversal of fortune, or do we 94

•

CHAPTER 4

need in addition to argue that southern institutions were inconsistent with modern economic growth?30 Finally, the early republic was probably an even more egalitarian place in 1800 than it was in 1774. Chapter 2 showed that incomes in the American colonies just prior to the revolution were more equally distributed than was true of Britain and the rest of Europe, or anywhere else on the planet that we can document. While it is true that we cannot measure income inequality in 1800, there is indirect evidence suggesting that America was an even more egalitarian place in 1800 than in 1774. We have seen that the gaps between the richer urban regions and the poorer rural hinterland declined between 1774 and 1800 (table 4-5). We have also documented a decline in the skill premium within urban areas. In addition, the rich South lost much of its per capita income lead over the North (table 4-4). All these forces should have produced a more equal distribution of earnings in 1800 than in 1774. The next chapter will show how much of this income-egalitarian America had disappeared by 1860. 30

For a key presentation of this institutional failure thesis, see Engerman and Sokoloff 2012.

THE COST OF REVOLUTION

•

95

CHAPTER 5

Unequal Economic Growth, 1800–1860

A

fter the Revolutionary War’s economic disaster and the struggles to form the new republic, the United States began its long nineteenth-century experience with modern economic growth, starting in the 1790s and reaching a peak by 1913. Thanks to the pioneering contributions of Easterlin, Gallman, Raymond Goldsmith, Kuznets, and other giants, we have learned much about America’s late nineteenth century. A subsequent wave initiated by David and Weiss has pushed our knowledge back into the antebellum period and earlier, but we still know far less about America’s early nineteenth-century experience with modern economic growth. Was American income per capita growth much faster than that of western Europe between 1800 and 1860, and if so, was it fast enough to have regained world income per capita leadership? How uneven was that growth between regions, classes, and skill groups? Did inequality rise with modern economic growth? If so, exactly which groups drifted farther apart in their incomes? Did the premiums on skills and schooling rise? Did the urban– rural income gap widen? Did the top income shares increase? Did property income shares rise, and did those incomes become more unequally distributed? Which regions were leaders, and which were followers? These questions have been actively debated for some time, but empirically based answers have been slow in coming, since the data have been so thin. Our social tables for 1800, 1850, and 1860 help fi ll that big gap so that we can now document America’s early experience with modern economic growth.

AMERICAN INCOMES IN 1850 AND 1860 Information about people’s incomes became more abundant from 1850 onward. The census began recording individual occupations, wealth, and other attributes, and the recent IPUMS now offers giant samples of individual records for each census region. Source materials on wages, farm economics, and slavery also expanded. Appendix E describes how we have fashioned this new information into income estimates. Our aggregate estimate of (gross) income is $3,025 million in current dollars for 1850 (table 5-1) and $5,246 million in current dollars for 1860 (table 5-2), using the baseline, part-time assumption about annual days worked.1 Note that the extensive settlement west of the Appalachians had reduced the original thirteen-colony income share to 59 percent of the US gross national income (GNI) in 1850 and 52 percent in 1860, whereas that share was almost 100 percent in 1774. Note also that human capital (excluding slaves) may have accumulated faster than conventional capital (which includes slave values), given that the gross property income share in GNI fell from 38 percent in 1800 (table 4-2) to 22 percent in 1860 (table 5-2). What was true of the United States as a whole was also true of the North, but not true of the South: between 1800 and 1860, the property income share fell from 24 to 17 percent in New England, and from 38 to 18 percent in the Middle Atlantic, but rose from 25 to 33 percent in the South Atlantic. The diminution of land settlement rates in the Northeast accounts for most of the difference, as we will see below. Our estimate of nominal GDP in 1860 is about 26 percent above that of the Weiss “broad” GDP estimate.2 What might account for the large gap between us? One possible explanation could be that our incomeside estimates have a more comprehensive coverage of the service sector. Another possible explanation may lie with price deflators. Recall that we estimate current price, or nominal, gross incomes. Others, like 1 The FTE, which assumes that everybody in the labor force worked 313 days a year, would be $5,712 million, or 7 percent higher than the part-time estimates. Yet we would not apply the FTE assumption to the economies of 1850 and 1860, even though we feel it could be valid for 1774 or 1800, because there is evidence that nonmarket home production had dropped significantly across the decades between 1774 and 1860. See Tryon (1917) 1966. 2 Weiss 1993a.

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

97

Table 5-1 Estimated American Personal Incomes in 1850 (Millions of Current Dollars) Free-labor income

Slaves’ income

Farm profits

Property income

Total income

New England

339.7

—

58.5

52.2

450.4

Middle Atlantic

655.8

—

89.0

127.5

872.2

East North Central

368.7

—

88.3

79.2

536.3

West North Central

71.2

2.5

13.0

11.1

97.9

1,435.4

2.5

248.9

270.0

1,956.9

South Atlantic

258.0

44.3

58.9

116.8

478.0

South Atlantic, no Florida

253.1

43.5

58.0

115.6

470.2

East South Central

166.4

33.9

41.2

101.1

342.6

West South Central

75.5

17.3

8.9

27.3

129.0

South

499.9

95.5

109.0

245.3

949.7

Mountain

11.9

—

1.5

0.7

14.1

Pacific

99.3

—

2.4

2.5

104.3

111.2

—

4.0

3.2

118.4

Original thirteen

1,248.6

43.5

205.5

295.2

1,792.8

United States

2,046.4

98.1

361.8

518.5

3,024.9

% of total

67.7

3.2

12.0

17.1

100.0

North

West

Weiss, build up their estimates from the output side, using production indexes to project forward and backward from the Gallman 1840 constant price base year—a key benchmark well worth a reexamination.3 To compare our estimates with the Weiss estimates, either our 1860 estimates need to be deflated to put them in 1840 prices or the Weiss 1860 estimates need to be inflated to put them in current prices. Choosing the right price deflator is likely to matter in any search for reconciliation between our direct estimates of nominal incomes and other 3

98

Weiss 1992, 1993a.

•

CHAPTER 5

Table 5-2 Estimated American Personal Incomes in 1860 (Millions of Current Dollars) Free-labor income

Slaves’ retained income

Farm profits

Property income

Total income

New England

422.5

0.0

69.4

98.0

589.9

Middle Atlantic

1,054.1

0.0

137.0

261.5

1,452.6

East North Central

636.3

0.0

176.0

188.0

1,000.2

West North Central

206.5

5.2

48.9

55.2

315.8

2,319.4

5.2

431.2

602.7

3,358.5

South Atlantic

333.9

61.7

84.7

238.3

718.5

East South Central

234.2

66.1

74.3

194.4

569.0

West South Central

144.5

38.7

29.7

102.1

315.0

South

712.5

166.5

188.7

534.8

1,602.5

Mountain

37.2

0.0

2.3

2.6

42.2

Pacific

212.9

0.0

15.2

14.5

242.6

250.1

0.0

17.6

17.1

284.7

United States

3,282.0

171.7

637.5

1,154.6

5,245.8

% of total

62.6

3.3

12.2

22.0

100.00

North

West

estimates of real GDP, which they have combined with price indexes to get nominal income estimates as a by-product.4 The eventual resolution of any differences between us should involve serious repair to the available price indexes to get a “true” price index for national GDP. That task has not yet been undertaken, however, either by us or others. Thus, “real” growth rates reported in this chapter are based on conventional 4 For the years up to the Civil War, readers can continue to ignore the subtle accounting distinction between the GNI we are measuring and GDP. They differ only by a statistical discrepancy that is small even today.

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

99

price series used by others. Specifically, we are using Sutch’s GDP deflator for 1800–1860, which is the same price deflator used by Weiss.5 It follows that any discrepancy between our nominal income estimates and those used by others will reflect a combination of unknown errors from the following three sources: (1) our nominal income estimates, (2) the price deflators used by us and previous scholars, and (3), current estimates of real output. To frame the differences that need to be resolved, the remainder of this chapter will address conflicts between “our” real growth rates implied by (1) plus (2) and other real growth estimates (3). We suspect that most of the differences between us are due to biases in the existing price indexes, rather than to errors in our nominal GDP estimates.6 We are about to report a revised history of real economic growth, but with the caveat that our new perspective on early American growth stresses the need to revisit existing views of price trends and not just the previous real growth estimates.

MODERN ECONOMIC GROWTH IN AMERICA, 1800–1860 One manifestation of America’s fast growth was the popularity of its debt with European investors. There was a surge of net capital inflows to America starting with US government securities like those used to finance the 1803 Louisiana Purchase, and after 1817, state bonds to 5 Carter et al. 2006, series Ca13. To make a long-run price deflator for GDP, Sutch attaches the Weiss price index for 1800–1860 to the best options for periods after 1860 and before 1800, as explained in his note to series Ca13. Across the Civil War decade, he used the David-Solar consumer price index. To link 1869 to series starting in the early twentieth century, he uses Balke-Gordon. For the seventeenth and eighteenth centuries, McCusker (2001) presents alternative series. He extends the David-Solar consumer price index back to 1774, its earliest year, but otherwise presents some volatile wholesale price indexes. As we noted in chapter 3, these disagree with the price movements of the Allen bare-bones consumer bundle. 6 For example, we think that the price deflators used by Weiss and others to get from 1840 current price to 1860 current price income estimates understate the rate of price inflation over those two decades, helping to account for some of the difference between us. We think the recent work of Ward and Devereux (2003, 2004, 2006) is consistent with the issues raised here. Like Allen’s work that we used in chapter 3, the research by Ward and Devereux has provided fresh direct price comparisons from the past, avoiding index number biases present in conventional price series, especially as they apply to services. The same approach of updating direct price comparisons as often as possible is also now embodied in what the new generation of the Penn World Tables now calls its “cgdpe” and “rgdpe” measures. See Feenstra, Inklaar, and Timmer (2015).

100

•

CHAPTER 5

finance the Erie Canal and other state infrastructure investments.7 The antebellum net capital inflow is well documented, but the new republic was established as a hot emerging market even before the 1820s: “Financial reforms of the 1790s gave the new Republic a modern fi nancial system—rare amongst the world’s nations at the time—and quickly converted what formerly had been the equivalent of junk bonds in default to high-grade securities.”8 Furthermore, the security markets were sufficiently active to have created a convergence of rates between London and American markets—a financial market integration that may have been complete by 1816.9 And international capital wasn’t the only factor seeking out high returns in the new republic. By the late 1840s, the cost of transatlantic steerage had fallen far enough to bring the passage to the New World within reach of ordinary workers.10 The result was the first “mass” migration in world history.11 By the 1820s, European resources were flowing with increasing magnitude toward a booming United States.

OVERALL GROWTH PERFORMANCE Table 5-3 supplies income per capita levels for 1800, 1850, and 1860 (in 1840 dollars), with the implied per annum growth rates. The estimates for the Atlantic census regions cover the space of the thirteen original colonies, which had become fifteen East Coast states and the District of Columbia, plus (by 1845) Florida.12 The rest of the country was west of the Appalachians, with its performance revealed by the difference 7

Wilkins 1989. Sylla, Wilson, and Wright 2006, 615. Capital inflows from Europe reached “huge proportions in the internal improvement era of the 1820s and 1830s, and in the railroad age that followed” (Sylla 1998, 89). See also North 1960; Williamson 1964, 99–124; Davis and Cull 1994; R. Wright 2002; Rousseau and Sylla 2005. 9 Sylla, Wilson, and Wright 2006. 10 The ratio of the passage fare to per capita income in Britain fell steeply from 0.495 in 1816– 21 to 0.205 in 1841 (Hatton and Williamson 2005, 41, table 3.3). 11 Hatton and Williamson 2005, chap. 3. 12 Here, as in chapter 4, the fi fteen states equal the original thirteen colonies, with Vermont detached from New York and Maine detached from Massachusetts. The fi fteen would become sixteen in 1863, when West Virginia was detached from Virginia. For convenience, we refer to the three Atlantic census regions as “East Coast” or “Eastern Seaboard” even though they include landlocked political units, such as the District of Columbia, Vermont, and West Virginia. 8

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

101

Table 5-3 Real Income per Capita Growth, 1800–1860 A. Regional levels of income per capita (in 1840 $) 1800

1850

1860

New England

56.44

167.28

178.51

Mid Atlantic

68.46

149.83

184.37

East North Central

—

120.13

136.83

West North Central

—

112.47

141.09

South Atlantic

73.99

104.81

126.90

East South Central

—

103.85

133.96

West South Central

—

140.70

171.20

Mountain

—

195.94

229.03

Pacific

—

1,009.05

525.81

Original thirteen Eastern Seaboard

67.95

137.90

163.85

United States

[67.95]

132.66

158.35

1800–50

1850–60

1800–60

New England

2.20

0.65

1.94

Mid Atlantic

1.58

2.10

1.66

East North Central

—

1.31

—

West North Central

—

2.29

—

South Atlantic

0.70

1.93

0.90

East South Central

—

2.58

—

West South Central

—

1.98

—

Mountain

—

1.57

—

Pacific

—

−6.31

—

Original thirteen Eastern Seaboard

1.43

1.74

1.48

United States

—

1.79

[1.42]

B. Regional growth rates (% per annum)

between the original thirteen and the total United States. For the entire six decades, the real per capita income growth of the East Coast was 1.48 percent per annum compared with 1.42 percent per annum for the United States as a whole. We are confident in our East Coast estimate but less so for the rest, or the “West” since it is so poorly documented for 1800. Yet the West was so small in 1800 that including it lowers the US growth rate below that of the East Coast by only a little. When we combine our estimated nominal incomes with the available price indexes, the resulting growth rate of real per capita income for the 1800 to 1860 years is considerably higher than those previously estimated. In percent per annum, these are: Abramovitz and David 0.93,  Weiss (broad GDP definition) 0.94, Weiss (narrow GDP definition) 1.06, Gallman 1.18, and David 1.29.13 Their estimates average a bit above 1 percent per annum, which barely passes the Kuznets test for achieving modern economic growth. As a reminder, Kuznets’s criterion was that per capita income growth must exceed 1 percent per annum over long periods. Our 1.48 percent per annum rate certainly passes the Kuznets test, and in that sense we suggest that America should be seen as one of the first, if not the first, to join the modern economic growth club. Johnston and Williamson recently estimated almost the same rate for 1790–1859, or 1.4 percent per annum.14 The rate of growth in our original-thirteen region on the Eastern Seaboard was also well above the western European average of 1.08 percent for 1820–1860 (table 5-4). It even outperformed the United Kingdom, implying catch-up growth after losing its big lead over the quarter century 1774–1800. Table 5-4 probably understates our case for catching up. The income per capita growth rate for Great Britain was considerably lower for 1800–1860 (0.69) than for 1820–1860 (1.06).15 If growth rates were also lower for 1800–1820 than for 1820–1860 for the other six western European countries in table 5-4, then there would be additional 13

Abramovitz and David 2000; Weiss 1992; Gallman 2000; David 1967. Johnston and Williamson 2008. Peter Rousseau and Sylla (2005, 21) cite Johnston and Williamson approvingly when writing recently that the US growth was “higher than that of other nations [and] earlier than most economic historians have thought.” 15 Broadberry et al. 2012. The British and continental growth rates would have been lower 1800–1860 than 1820–1860 since much of those fi rst two decades involved the so-called French Wars. See O’Rourke 2006. 14

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

103

Table 5-4 West European Growth Rates, 1800–1860 (% per Annum, per Capita) Austria

1820–1860

0.95

Belgium

1820–1860

1.39

France

1820–1860

1.29

Germany

1820–1860

1.06

Netherlands

1820–1860

0.65

Switzerland

1820–1860

1.19

United Kingdom

1820–1860

1.06

1800–1860

0.69

1820–1860

1.08

West Europe

reasons for thinking we have understated the American catch-up growth performance. According to Broadberry and his collaborators, the per capita income in Great Britain had reached £30.75 sterling in 1860, which at current exchange rates, was $149.13. The US figure reported in table 5-2 is $171.40, and $181.02 for the industrial Northeast. Thus, the United States had reclaimed its income per capita lead by 1860: the United States was about 15 percent higher, and labor productivity was higher still, since the labor participation rate was lower in the United States. Based on our purchasing power parity estimates, the United States—with its cheap food, fuel, and rents—had an even bigger 1860 per capita lead in real income, as we will see in a moment.16 We have already suggested that Maddison may have been more than two centuries off the mark when claiming that America only overtook Britain just after 1900.17 Still, he seems to have been right that there was a post1870 overtake, as we will confirm in chapters 6 and 7. But it looks as though America overtook Britain at least three times—by 1700, again by 1860, and again shortly before World War I. Of course, it also lost the lead twice before the twentieth century. Figure 5-1 plots the American 16 Gallman (1966, 5) thought the United States might have captured the lead in 1840. Ward and Devereux (2006, 252–53) estimate that the US GDP per capita was ahead of Britain by 30 percent as early as 1831 (in purchasing power parity terms). 17 Maddison 1995, 2001.

104

•

CHAPTER 5

50

GDP per capita in current £ sterling (log scale)

USA Great Britain

20

South Atlantic region All 13 colonies

10

New England

5 1650

1675

1700

1725

1750

1775

1800

1825

1850

1875

Figure 5-1. Nominal GDP per Capita, Britain and America, 1650–1870

experience relative to Britain from 1650 to 1870 using nominal exchange rates, and figure 5-2 does the same using Allen’s purchasing power parity adjustments. We need to add two qualifications. First, and as noted in chapter 4, we have been unable to measure pure profit residuals in 1800 for the dominant farm sector. Since we have farm pure profit residuals for 1860 (and 1774), the incompleteness of our measures for 1800 might understate that year’s income per capita and overstate the 1800–1860 growth rate, just as it might understate the recovery from the economic disaster of 1774–1790. Having said as much, we stress that the 1800 estimate plays no role in assessing per capita income relative to Great Britain in 1774 or in 1860. We have shown that American incomes exceeded those of the mother country at both dates. We have also shown that American income per capita fell dramatically between 1774 and 1800 while it rose in Britain. These facts make the American fast catch-up growth rates for 1800–1860 all the more plausible. Second, the acceleration and UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

105

GDP per capita in bare-bones bundles of consumer necessities (log scale)

16

South Atlantic region USA All 13 colonies 8

Great Britain New England 4

2 1650

1675

1700

1725

1750

1775

1800

1825

1850

1875

Figure 5-2. Real GDP per Capita, Britain and America, 1650–1870

high growth rates in the South between 1840 and 1860 have been debated in the literature ever since Easterlin presented his regional estimates.18 Based on an earlier literature led by Gavin Wright, we know that “the [1859–1860] price of cotton was above the level predicted on the basis of production and trend . . . by 7.6 to 15.9 percent. On a comparable basis, the cotton price for the year 1839–1840 was between 4.0 and 11.5 percent below its predicted value. . . . [Hence,] the choice of endpoints imposed by the census data [tilts upward] the southern growth path.”19 While the cotton states had their 1840–1860 growth exaggerated by the volatility of cotton prices, a properly adjusted growth rate would still have been quite impressive. To summarize thus far, real income per capita in our Eastern Seaboard, original thirteen region grew at 1.53 percent per annum over the 18 19

106

Easterlin 1960, 1961. G. Wright 1976, 326.

•

CHAPTER 5

1800–1860 period—well above the western European average of 1.08 percent for 1820–1860. The rate at which it outperformed the United Kingdom after 1800 implies that America would reclaim its world income per capita leadership by 1850.20

FILLING THE FRONTIER As we saw in chapter 3, the flow of Americans toward the lower-income frontier meant that income per capita grew more slowly for America as a whole than one would have guessed by tracking the rise of income and wealth in any location along the seaboard. To be sure, those who migrated toward the frontier were not making an economic mistake, since they were not lowering their incomes by moving.21 Yet by leaving, their migration produced an outcome that might seem paradoxical: income per capita grew more slowly for the colonies as a whole than it grew for any one place, either in seaboard cities or at any given location further west. As it turned out, more rapid growth in inland populations won the colonial tug of war: the rate of westward settlement was fast enough to outstrip any equilibrating tendency for the average incomes in the improving interior to achieve catch-up growth rates on incomes back east. Was this frontier colonial experience repeated during the first six or seven decades of the young republic? Tables 5-3 and 5-4 tell us that the answer is no: the frontier filling-up effect on overall US income growth rates was much weaker during the antebellum decades than during colonial times. That result may seem puzzling, since the filling-up forces certainly look as if they should have been more powerful in the antebellum decades; after all, the population growth rates in the West 20 To repeat the statement in the text above, we understate the case for catching up. The income per capita growth rate for Great Britain was considerably lower for 1800–1860 (0.69) than for 1820–1860 (1.06) (Broadberry et al. 2012). The lower (war-affected) figure is the relevant one since the American growth rate is also calculated for 1800–1860. 21 Nobody ever made a move based on income per capita differentials. There were more highpaying jobs and accumulated wealth in the settled East, pushing up its income per capita relative to the subsistence-plus interior. Migrants took their farm skills to the interior, where land was cheap and abundant.

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

107

Table 5-5 Regional Relative Incomes per Capita, 1800–1860 1800 Lindert and Williamson

1840 Easterlin 1960

1850 Lindert and Williamson

1860 Lindert and Williamson

New England

83.1

127.7

126.7

113.3

Middle Atlantic

100.7

118.5

113.5

116.5

East North Central

—

70.8

91.1

84.8

West North Central

—

78.5

85.4

85.0

South Atlantic

108.9

84.6

78.5

86.0

East South Central

—

84.6

78.2

82.9

West South Central

—

160.0

105.3

109.5

Mountain

—

—

78.6

130.5

Pacific

—

—

101.2

313.3

Original thirteen

100.0

109.8

103.5

105.6

United States

—

100.0

100.0

100.0

were more than three times that of the East— 6.34 versus 1.98 percent per annum. Why did this spectacular frontier filling up fail to reduce the overall growth rate much, as happened in colonial times? The answer can be found in the first panel of table 5-3 and in table 5-5. By the middle of the nineteenth century, the absolute levels of income per capita were no longer much lower on the frontier than on the Eastern Seaboard as they had been in colonial times. That is, the East and the West had become closer to equality in average incomes. Eastern per capita income was by 1850 only 9.2 percent higher than that of the West. How do we reconcile the fact that the East–West gap was so modest at the start of the nineteenth century when it was so big in 1774? We think most of the explanation lies with the asymmetrical economic impact of the Revolutionary War and early federation on the urban and staples-oriented East Coast versus the hinterland (chapter 4). We do not think it was due to some frontier catch up between 1774 and

108

•

CHAPTER 5

1800. But whatever the reasons, by 1800 the gap was small. Thus, the colonial fi lling-up effect had lost its ability to drag down economywide growth rates during the first six or seven decades of the young republic.

REGIONAL INEQUALITY AND MORE REVERSAL OF FORTUNE The absolute decline of South Atlantic per capita income over the last quarter of the eighteenth century and its relative decline over the first six decades of the nineteenth century offer a classic example of reversal of fortune over a remarkably short period of time. Tables 5-3 and 5-4 document the antebellum reversal. According to Easterlin, the South Atlantic was already behind the Northeast and the national average by 1840, and our 1850 estimates support his conclusion. Chapter 2 showed that the South Atlantic (then called Upper and Lower South) was well ahead of all other regions in 1774, but chapter 4 demonstrated that it had lost most of that lead by 1800 (then only 8.9 percent above the original thirteen average; see table 5-5). Income per capita in the South Atlantic had fallen from 24 percent above the East Coast average in 1774 to only 8.9 percent above it in 1800, to 15.4 percent below it in 1840 (according to Easterlin), 21.5 percent below in 1850, and 14 percent below in 1860. We also noted that this reversal of fortune draws support from the apparent absence of any large army of poor whites in the colonial South. The familiar image of the South as a repository for much of the nation’s poor whites—a topic of the next section and chapter 6—had become an attribute of the region by 1860. Why the Old South reversal of fortune? A benign part of the story seems to have been that the colonial South was still a labor-scarce frontier region with high returns to coastal land producing export crops, like indigo, rice, and tobacco. Its decline after 1774 was echoed in two other frontier cases many decades later. One was the dramatic relative decline of the West South Central income per capita between 1840 and 1860—from 60 percent above the US average to just 9.5 percent above it (table 5-5). The other was the loss of the Pacific region’s gold-discoverygenerated super-incomes after the 1850s and early 1860s (the Pacific

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

109

states were 213.3 percent above the US average in 1860, and the mountain states were 30.5 percent above). No doubt the southern reversal of fortune had multiple causes, and as we already confessed in chapter 4, we do not yet know what weights to attach to the decline of frontier super-returns, the exceptionally severe market and war damage incurred during the Revolutionary conflict, or  some fundamental institutional weakness associated with slavery. In spite of its plausibility and popularity, one piece of evidence arguing against the institutional weakness thesis is apparent in table 5-3: rapid South Atlantic per capita income growth of 2.91 percent per annum between 1850 and 1860 in our estimates, or 2.41 percent per  annum between 1840 and 1860, in Easterlin’s estimates. That is, across the final decades before the Civil War, the supposedly decaying South Atlantic had even faster income growth than the industrializing Middle Atlantic (2.22) and New England (0.82). How would the institutional weakness thesis explain the South Atlantic’s catching up during the two decades prior to the Civil War, having fallen behind up to 1840? That growth revival certainly wasn’t driven by a boom in the prices of the region’s new export staple, cotton, since its price did not rise over the full antebellum period (although the census date 1859–1860 recorded unusually high cotton prices and the census date 1839–1840 recorded unusually low cotton prices).22 Furthermore, Alan Olmstead and Paul Rhode have recently documented impressive productivity gains in cotton agriculture during this period, complicating a simple institutional weakness theory.23 What about market size? While income per capita was growing faster in the United States than among the European leaders in 1800–1860, its total income grew even faster (table 5-3). New England had income per capita growth rates almost double that of western Europe (1.96 versus 1.08), and the Middle Atlantic more than half again higher (1.68 versus 1.08). By 1860, that, plus rapid population growth, made both of these regional economies much bigger than many countries in western Europe, like Austria, Belgium, the Netherlands, and Switzerland. And the East Coast economy was bigger than most of western continental 22 23

110

Gray 1940, 1047. Olmstead and Rhode 2008, 2010.

•

CHAPTER 5

Europe. To the extent that these large domestic markets helped American firms exploit economies of scale before the world was globalized by falling intercontinental transport costs and trade liberalism in the late nineteenth century, the United States already had at least one manufacturing productivity advantage before the Civil War.24 That advantage must have contributed to its fast growth. Certainly observers looking on with envy from other countries thought so.

WIDENING INCOME GAPS WITHIN AMERICA, 1774–1860 There has been long-standing disagreement about nineteenth-century trends in American inequality, both in wealth and income. Some have argued that it was high and stable across the nineteenth century, best illustrated by the work of Lee Soltow.25 Others have argued that inequality increased markedly between the revolution and post–Civil War.26 A recent cross-section of ten countries shows mixed results for their nineteenth-century wealth distribution trends.27 Now we have the income inequality evidence to see which competing thesis wins for the American experience. While social tables are not an inherently superior way to estimate aggregate national income or product, they have the clear advantage of being able to reveal the inequality of income in historical settings where other evidence is lacking. Thus, our 1774 social table (chapter 2) was able to report a detailed description of American income inequality on the eve of the revolution. Now we will do the same for the eve of the Civil War. This is a significant advance over an earlier literature that could only draw regionally isolated sketches of wealth inequality and some wage ratios.28 We can therefore now explore the Kuznets hypothesis with American evidence: Did modern economic growth raise 24

O’Rourke and Williamson 1999. Soltow 1971, 1975, 1989. 26 See Williamson and Lindert 1980a, 1980b; Lindert 2000. 27 Roine and Waldenström 2014, table 5. 28 For a summary of the vast literature on early wealth inequality written in the 1960s and 1970s, see Soltow 1975; Shammas 1993; Williamson and Lindert 1980a, 1980b; Lindert 2000. For occupational wage ratios, see Williamson and Lindert 1980a; for an important update, see Margo 2000. 25

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

111

inequality in the young republic? And if it did raise inequality, what form did it take, and what were its causes? Piketty recently offered one view in his widely acclaimed Capitalism in the Twenty-First Century. By exploiting twentieth-century income tax evidence, Piketty, Atkinson, Saez, and their collaborators have traced a new inequality history of the twentieth and early twenty-first centuries using the share of income accruing to the top 1 or 10 percent.29 Piketty argues that this is due to property income growth, since property and wealth have always been highly concentrated in the hands of the rich few. He contends that income inequality can be predicted using various wealth-related indicators. One such indicator is the difference between the rate of return on private wealth (r) and the growth of total GDP (g). Consider its possible application to inequality trends in early US history. Piketty maintains that the return to wealth has typically been greater than the rate of growth of income. As we will see, and as Piketty acknowledged, this connection is weakened in newly settled regions like America where abundant and cheap land—a big share of preindustrial wealth—was available to all comers, including those close to the bottom of the income distribution. For industrial and postindustrial economies, Piketty argues plausibly that when national income growth rates (g) are high (as between 1945 and 1975), and the rate of return (r) is equal to g or lower (Piketty uses a 5 to 6 percent benchmark—a rate we have also documented for 1774–1860), then property incomes will grow (say, by 5 percent per annum) at the same rate as total incomes or even slower, yielding a stable or even falling property income share and inequality. Conversely, when growth is slow (as since 1975), the property income share, the top income shares, and inequality will all rise. Piketty does not deny that rising earnings inequality adds to rising income inequality, but he stresses the forces of property income growth. We have documented that income growth accelerated in America from about 3 percent per annum during 1700–1774 to 4.5 percent per annum during 1800–1860 (a per capita income growth of 1.5 percent and a population growth of almost 3 percent). Since modern economic growth induced a sharp rise in g in the decades following 1800, and given a stable 29

112

Piketty 2014; Atkinson, Piketty, and Saez 2011.

•

CHAPTER 5

r, Piketty’s (r − g) metric would have predicted falling property income shares and falling inequality between 1800 and 1860, in confl ict with the Kuznets prediction. As it turns out, the property income share did fall from about 38 to 22 percent between 1800 and 1860, in line with the Piketty prediction.30 True, the property income share rose in the South Atlantic from 25 to 33 percent, no doubt driven by rising slave values there; between 1841 and 1860, the value per slave rose at 6.1 and 5.7 percent per annum in the Lower and Upper South, respectively.31 But property income shares in total income fell in the industrializing Northeast (from 24 to 17 percent in New England, and from 38 to 18 percent in the Middle Atlantic). In short, based on falling property income shares, the Piketty prediction would have been no rise in the young republic’s inequality, and perhaps even a fall.32 In sharp contrast, Kuznets famously predicted rising inequality in the early stages of modern economic growth (the rise being the first part of what is now called the Kuznets curve).33 Furthermore, he thought that his Curve was generated by the combined impact of a rising urban– rural income gap as well as a rise in the richer (and more unequal) urban areas at the expense of the poorer (but less unequal) rural areas. In addition, labor and development economists have always stressed that industrialization and urbanization raise skill premiums and cause 30 A fuller statement of the estimated share of the nation’s property and real estate income alone as percentages of national income, excluding capital gains, are as follows:

1774

1800

1850

1860

1870

All property incomes

22.6

37.6

17.7

22.0

16.6

Real estate

11.2

12.3

8.5

12.7

6.7

Once again, note that the 1800 figures exclude pure farm profits in the national income denominator. 31 Fogel and Engerman 1974, 73, table B.9. 32 Piketty’s framework offers different late nineteenth- and twentieth-century predictions, as we will discuss more fully in chapters 8–10. Here we note only that the (r − g) predictor does not fit the inequality facts for 1800–1860. One that does better is his wealth–income ratio, which rose with inequality up to 1860, offering an even better fit with our new inequality figures than do the ones available to Piketty and Zucman (2015). So had Piketty relied on his wealth–income predictor of inequality rather than his featured (r − g) predictor, his prediction would have looked better. 33 Kuznets 1955.

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

113

wage stretching. This thesis has been explored for 1870 America, where regions with higher manufacturing employment shares revealed higher inequality.34 Claudia Goldin and Lawrence Katz have applied the argument to the American twentieth century with persuasive evidence in their recent The Race between Education and Technology.35 We can add another force that might have pushed up inequality. As financial institutions spread and deepened across America from 1790 to 1860, as European capital sought out booming investment opportunities there, so too did the incomes flourish of those at the top of the earnings distribution who had the financial skills to exploit those markets. Thus, there is no shortage of predictions that the American leap to modern economic growth between 1800 and 1860 should have been characterized by rising inequality, contrary to Piketty’s predictions based on (r − g).36 So which theory is right? Did inequality rise or not across the long antebellum era? And if it did rise, which of these forces caused it: a growing excess demand for skills and schooling, urbanization, booming financial markets, or all three? One could add another question to these two: Did increasing regional inequality—a rising gap between a poorer agrarian South and a richer industrializing North— reinforce any rising inequality forces? Household Income Inequality, 1774–1860 The next available benchmarks for charting the American income distribution are 1850 and 1860.37 Tables 5-6 and 5-7 allow us to compare those two years with 1774 from table 2-4. The main finding is unambiguous: inequality rose over this eighty-six-year span of American history—for the United States as a whole, for every region, among free households alone, and among slave and free combined. And income inequality rose a lot; among the original thirteen colonies, the 1774 34

Rosenbloom and Stutes 2008. Goldin and Katz 2008. 36 Soltow (1975) argued that there was no change in antebellum inequality at all. 37 Given this chapter’s focus on the 1800–1860 era, we would have preferred 1800 as the starting point rather than 1774. Th is has been possible for aggregate growth, as already discussed. But, and as we warned early in this chapter, we cannot do the same for inequality, since our 1800 social table lacks sufficient detail on property incomes. However, we can document changes in earnings inequality from 1800 onward. 35

114

•

CHAPTER 5

Table 5- 6 The Inequality of American Household Incomes, 1850 Region

Gini coefficient

Percent shares of total gross income Top 1

Top 5

Top 10

Top 20

Bottom 40

Mean household income (current $)

New England

0.417

6.7

18.8

29.7

46.3

14.6

828

Middle Atlantic

0.447

7.9

21.3

33.2

50.1

13.8

774

South Atlantic, w/Florida

0.574

13.8

32.2

44.3

61.2

8.7

505

South Atlantic, no Florida

0.575

13.9

32.3

44.5

61.4

8.8

503

East North Central

0.328

5.6

16.1

25.6

40.8

20.4

654

West North Central

0.399

6.2

18.4

29.1

45.2

16.2

588

East South Central

0.537

15.6

31.2

42.1

57.7

9.8

526

West South Central

0.516

14.0

29.1

40.8

57.5

12.2

628

Mountain

0.318

6.7

16.7

25.5

40.4

21.6

802

Pacific

0.432

7.2

21.2

32.3

48.7

14.7

3,635

All United States

0.487

10.4

24.9

36.5

52.7

11.7

672

Original thirteen

0.497

9.2

23.7

36.1

53.1

10.8

689

South Atlantic, w/Florida

0.479

11.3

27.2

38.5

53.0

13.7

770

South Atlantic, no Florida

0.480

11.3

27.3

38.6

53.1

13.6

772

West North Central

0.351

5.9

17.3

27.4

42.7

19.7

653

East South Central

0.439

12.8

28.1

37.6

51.0

16.7

781

West South Central

0.454

12.6

26.3

36.5

51.0

15.2

927

All United States

0.436

9.7

23.3

34.2

49.5

15.3

779

Original thirteen

0.449

8.6

22.1

33.7

50.0

13.9

786

A. All households

B. Free households only

Table 5-7 The Inequality of American Household Incomes, 1860 Region

Gini Percent shares of total income Mean coefficient household Top 1% Top 5% Top 10% Top 20% Bottom 40% income (current $)

A. All households New England

0.457

7.0

20.5

32.3

49.4

12.7

889

Middle Atlantic

0.488

9.1

23.3

35.7

52.9

11.8

977

South Atlantic, w/Florida

0.608

13.7

33.6

47.5

64.8

7.7

651

South Atlantic, no Florida

0.610

13.9

33.9

47.8

65.2

7.8

648

East North Central

0.399

7.0

19.1

29.6

45.5

16.6

761

West North Central 0.419

6.9

20.0

30.7

46.6

15.1

749

East South Central

0.552

12.5

31.4

44.4

60.8

10.8

705

West South Central 0.569

16.0

34.9

47.2

62.7

10.9

860

Mountain

0.515

10.2

26.2

39.2

56.1

11.6

1,003

Pacific

0.415

6.5

18.8

29.7

46.1

14.6

1,937

All United States

0.511

10.0

25.6

38.0

54.9

10.7

829

Original thirteen

0.529

9.9

25.5

38.4

56.0

9.3

852

B. Free households only South Atlantic, w/Florida

0.525

11.0

27.9

40.5

56.6

10.9

989

South Atlantic, no Florida

0.527

11.1

28.1

40.7

56.8

10.8

993

West North Central 0.388

6.6

19.2

29.5

44.9

17.1

806

East South Central

0.490

10.1

26.9

39.1

54.6

13.3

1,041

West South Central 0.523

13.6

31.1

42.5

57.4

11.9

1,295

All United States

0.474

9.4

24.1

35.9

52.1

13.0

944

Original thirteen

0.491

9.3

23.9

36.2

53.1

11.7

959

Gini coefficient of 0.441 rose to 0.529 in 1860 (tables 2-4 and 5-7). That is, the economic equality that Alexis de Tocqueville saw in his 1830s visit to America was fast disappearing: A manufacturing aristocracy . . . is growing up under our eyes. . . . The friends of democracy should keep their eyes anxiously fi xed in this direction: for if a permanent inequality of conditions . . . penetrates into [America], it may be predicted that this is the gate by which they will enter.38

Tocqueville’s famous warning proved even more broadly correct than he predicted. It wasn’t just a new manufacturing elite that arose in nineteenth-century America but also a property-owning and high-wage elite in all sectors of the economy. Income gaps widened on all fronts in the Eastern Seaboard (the New England Gini went from 0.354 to 0.457, and the Middle Atlantic Gini rose from 0.381 to 0.488). The widening was even more pronounced in the South, where there was less manufacturing. Among slave and free households combined, the South Atlantic Gini rose from 0.464 in 1774 to 0.608 in 1860. Inequality was greater still in the New South than in the Old (i.e., the South Atlantic in 1774): the 1860 Ginis were 0.552 in the East South Central, and 0.569 in the West South Central. What is especially notable about the South Atlantic was the enormous increase in income inequality among free households, where the Gini increased from 0.328 in 1774 to 0.527 in 1860. The top 1 percent share of income among free households in the South Atlantic almost doubled, rising from 6.3 to 11.1 percent, while the share going to the poorest 40 percent fell precipitously from 21.9 to 10.8 percent. Any historian looking for the rise of a poor white underclass in the Old South will find it between 1774 and 1860—not in the colonial period, and not just after the Civil War. Why did the income gaps widen over those nine decades more in the South than in the North? While we cannot offer a full explanation here, there are three likely candidates. First, the income gap between slave and free widened considerably. We will have more to say about this below, but we stress here that the widening free–slave gap fails to account for the steep rise in inequality among free southern households. 38

Tocqueville (1839) 1963, 161.

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

117

Second, as we have seen, the property income share increased in the South, but not in the North. Property incomes are much more unequally distributed than labor earnings, so we have one force accounting for the special steepness of the southern inequality rise.39 Finally, the antebellum South, like Latin America, had a more elitist approach to education.40 Relative to the North, its schooling was concentrated on children of the elite. Southern parents who could afford it paid much larger amounts for tuition than northern parents. Taxes for public schooling were lower in the South, and a smaller share of white children enrolled in school. The South also supplied fewer teachers per hundred children of school age. The fact that the South educated a smaller share of its youths for several decades, and spent so much private money on pupils from elite families, must have widened income gaps among southern whites in 1850 and beyond.41 Between the eve of the revolution and the eve of the Civil War, the early egalitarian condition in the North was moving westward, à la Frederick Jackson Turner, rather than disappearing. The relatively low Gini coefficients migrated from New England and the middle colonies to the East and West North Central; in 1860, their Ginis were 0.399 and 0.419, respectively—only a little higher than egalitarian Middle Atlantic in 1774. The rise in overall inequality was still in progress during the last decade of the 1774–1860 century. During the 1850s, every census region but the Pacific underwent a sharp rise in inequality, with or without slave households. The same was true for the nation as a whole (see the Gini coefficients in tables 5-6 and 5-7). But the 1850s did not reveal a uniform rise in the share held by the top 1 percent. The top 1 percent share dropped in all southern census regions, despite the continued rise in slave values. It rose in all northern census regions. We view this as an unresolved puzzle. The long rise in inequality before the Civil War was as great as the more familiar rise that America has experienced since the 1970s. Figure 5-3 maps the extent of American inequality since 1774 using 39 In 1860, the Gini for property incomes was 0.835 (table 5-8), while the figure for labor earnings was 0.473 (table 5-9). 40 See Go and Lindert 2010; Engerman and Sokoloff 2012, 121– 67. 41 See Go and Lindert 2010, especially table 2.

118

•

CHAPTER 5

Holland

Gini coefficient (pre-fisc household income)

0.6 England– Wales

Netherlands

Netherlands 0.5

United States

American 13 colonies United Kingdom

0.4

0.3 1725

1750

1775

1800

1825

1850

1875

1900

1925

Netherlands

1950

1975

2000

Figure 5-3. Income Inequality in America, Britain, and the Netherlands, 1732–2010

the Gini coefficient as an index. The two great surges in inequality were equally dramatic in overall magnitude, though the rise since the 1970s has occurred in four decades, not nine. The top 1 percent shares of total income tell the same story. To repeat, the 1774–1860 rise in inequality was similar in overall magnitude to the one America has experienced recently, but the first great inequality rise seems to have stretched across more decades. We say that it “seems to have” because inequality was probably even lower in 1800 than in 1774, implying an even steeper rise over the six decades after 1800. Why do we think inequality was lower in 1800? Since chapter 4 was able to document falls in the skill premium, the South–North income gap, and the urban– rural income gap between 1774 and 1800, we conjecture that the overall inequality of earnings or income fell. Still, we simply do not know whether inequality started rising in 1790, 1800, or 1820. When placed in international perspective, the two great surges in American inequality look quite different (figures 5-3 and 5-4). America UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

119

Percentage of pre-fisc income received by top 1 percent of households

30 England– Wales

UK households Netherlands

United States

20 Netherlands

Japan

Holland

UK adults

10 United States

Japan

American 13 colonies

0 1725

1750

1775

Japan

UK households

1800

1825

1850

1875

1900

1925

1950

1975

Netherlands

2000

Figure 5-4. Top 1 Percent’s Share of Income in Four Countries in Recent Centuries

started from much lower levels of income inequality than did industrializing European countries. Thus, the early and sharp rise in American inequality greatly exceeded any European rise, and by the early twentieth century America had joined the European inequality club, catching up with Britain, the European continent, and perhaps even Latin America.42 Had America reached its 1860 inequality levels a decade earlier? The answer appears to be almost. That is, the 1850 original thirteen Gini was already 0.529 in 1850, well above the 1774 figure of 0.437. Whatever those long-term inequality forces were, they did not suddenly appear a decade or so before the Civil War. 42

120

Williamson 2010, 2015.

•

CHAPTER 5

What could have caused the dramatic shift toward more unequal incomes between 1774 and 1860? The best way to attack the problem is by breaking down the observed income widening into its component parts—a great advantage of our social tables. As it turns out, similar widening tendencies show up in a variety of income gaps calculated over these decades—gaps in household property incomes and household labor incomes, and between cities and the countryside, North and South, the remuneration of free and slave labor, and skilled and unskilled labor. The rest of this chapter explores all these inequality dimensions up to the Civil War, and enlists them as guides to the most likely causal explanations of the great antebellum income inequality surge. Unequal Property Incomes, 1774–1860 Consider first the movements in property income inequality—the distribution of that income source alone—as well as the share of property income in the total income. Table 5-8 compares property incomes for 1774, 1860, and 1870, though we defer the details about 1870 until the next chapter. The distribution of property incomes became much more unequal across the century. For all households, the Gini coefficient on property incomes rose from 0.703 in 1774 to 0.835 in 1860, and the top 10 percent share rose from 48.8 to 72.7 percent. Rising inequality of wealth and property incomes was thus a major part of the rising income inequality in America between 1774 and 1860. Unequal Labor Earnings, 1774–1860 Did labor earnings inequality rise as well? Labor economists have long argued that earnings inequality explains much (or most) of the recent rise in American income inequality, since the distribution of human capital and schooling now matters so much.43 While changes in the 43 See, for example, Katz and Autor 1999; Goldin and Katz 2008; Acemoglu and Autor 2013. As noted above, other scholars disagree. See Piketty 2005, 2014; Leigh 2009; Atkinson, Piketty, and Saez 2011. They argue that changes in inequality across the twentieth century are driven in large part by changes in the top 1 or 10 percent share, and that these are driven mainly by changes in the property income share. The correlation between changes in inequality in the United States during 1985–2005 and changes in the top 10 percent share, however, is “only” 0.57 (Roine

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

121

Table 5-8 The Inequality of American Property Incomes, 1774–1870 Gini coefficient

Percent shares of all property incomes

Property income

Top 1%

Top 5%

Top 10%

Bottom 40%

Mean

Median

Property income % total income

1774 Total

0.703

13.7

34.1

48.8

1.5

74

23

22.6

1860 North

0.757

25.0

47.6

61.5

1.6

155

41

17.9

1860 South

0.899

34.1

68.8

84.2

0.0

235

4

33.4

1860 Total

0.835

32.0

59.2

72.7

0.3

182

25

22.0

1870 North

0.767

26.5

48.9

62.7

1.4

226

56

18.3

1870 South

0.855

32.4

61.2

75.5

0.0

88

8

11.1

1870 Total

0.808

29.6

53.6

67.8

0.4

185

35

16.6

Table 5-9 Earnings Inequality in America, 1774–1860 Free households

All households

Gini

Top 10%

Gini

Top 10%

1774

0.370

27.7

0.388

27.9

1850

0.422

34.5

0.459

36.2

1860

0.454

36.0

0.473

37.2

human capital distribution may have mattered less in the early nineteenth century, table 5-9 shows that earnings and income inequality rose together between 1774 and 1860, as they have since the 1970s. The Gini for earnings was, of course, lower than that of income and certainly less than that of property income, but earnings inequality rose (up 0.085) even more steeply than did income inequality (up 0.074). In addition, all earnings inequality measures were increasing—the top and Waldenström 2014, 33). Changes in the distribution of earnings, approximately the bottom 90 percent, matters almost equally.

122

•

CHAPTER 5

shares and the Ginis, and for free as well as all households. Thus, both property incomes and labor earnings became much more unequal from the eve of revolution to the eve of the Civil War.

URBANIZATION AND THE RISE OF AMERICAN INEQUALITY How did the rise in inequality relate to the growth of American cities? Recall that in the colonial era from about the 1670s to the 1770s, the American population actually “ruralized”—that is, the share living in cities and towns declined, before shifting toward urbanization from around 1790 onward.44 Chapter 3 conjectured that the colonial shift toward the rural frontier tended to suppress any upward trend in both average incomes and the inequality of incomes, below what one would see by looking only at trends along the older-settled East Coast. Would the rise of cities after independence have had the opposite effect, causing a steeper upward trend in average US incomes and inequality than one would infer by looking only at trends within cities or the countryside? Kuznets suspected as much. In his famous presidential address to the American Economic Association published in 1955, he conjectured that inequality might rise in the early phases of modern economic growth as employment shifted from low-paid rural to high-paid urban jobs. The initial urban shift would spawn new inequality, which would widen until the urban sector reached some intermediate share of the economy, after which it would have a leveling effect.45 But Kuznets had more in mind than the role of average income gaps between urban and rural activity. He also stressed that since cities were more unequal places than the countryside, urbanization should raise inequality on that score too.46 44

See figure 3-2. The idea is simple but elegant. When the fi rst person migrates from the low-wage but egalitarian rural area to uniformly high-wage urban employment, inequality rises. When the last person moves, the nation moves back to complete equality. During the early stages of modern economic growth, the rise in inequality is more dramatic the faster is urbanization. 46 Of course, it could be argued that the Kuznets urbanization effect is simply a statistical accounting of more fundamental economic forces, like the demand and supply of human capital and its distribution. 45

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

123

Here we estimate the impact of urbanization in three steps. First, we explore rural–urban wage gaps for unskilled labor alone. Did the rapid growth of urban labor demand outpace competing labor demand in the countryside by enough to offset the more rapid growth of urban labor supply, fed as it was by off-farm and foreign immigration? Second, we explore the average income gaps (not just wage gaps) between rural and urban locations. Did schooling and other skills accumulate faster in the cities, where they were more useful, thereby raising the average income gap and contributing to rising inequality? Third, we estimate the impact of urbanization on the rise in the total income inequality—that is, the impact of population shifts from more egalitarian rural areas to less egalitarian urban areas. Crucial to all these calculations is the pace and location of America’s urban transition, which accompanied early industrialization. We emphasize rates of urbanization and changes in the urban share, since land-abundant America was still much less urban than western Europe in 1860.47 Between 1800 and 1860, America’s urban share rose from 6 to 19.8 percent (table 5-11, below) and the annual growth rate of its urban population was 5.1 percent, almost double that of the rural population, at 2.7 percent.48 The urban transition was even faster in the Northeast, where the urban share rose from 9.3 to 35.7 percent, and its urban population grew annually at 4.7 percent, more than two and a half times that of its rural population, at 1.8 percent. Thus, the urban transition was certainly dramatic enough to give Kuznets’s hypothesis a chance to shine. And shine it does on all three counts. Urban–Rural Wage and Earnings Gaps The behavior of urban–rural wage gaps during modern economic growth has always been of interest to economic historians and development economists. Typically, and appropriately, they are calculated for 47 Europe (less Russia) was about 25 percent urban in 1860, while the leading countries were England (55 percent), the Netherlands (30 percent), and France (26 percent) (Hohenberg and Lees 1985, 219, figure 7.2). The share urban was considerably lower in 1860 United States, at about 20 percent. 48 The US census defi nes urban as locations with a population equal to and greater than twenty-five hundred.

124

•

CHAPTER 5

homogeneous unskilled labor. As such, the nominal wage gap was 73.2 percent for England in the 1830s, and higher in the south of England than in the north. Furthermore, the gap was on the rise in England between 1797 and 1851.49 Nor was the English experience uncommon, since in the 1960s and 1970s the gap averaged 41.4 percent in the Third World.50 The search for urban–rural wage gap explanations divides into two camps. One camp sees this as evidence of labor market failure: restrictions on mobility off the farm and away from the village, or an unwillingness to move, allows wages in the fast-growing city to rise above those of the slow-growing countryside. This might be called the disequilibrium or market failure view. The other camp takes the equilibrium view: namely, that city disamenities (like disease and crowding), food costs, and housing costs all rise with city size, thus requiring greater nominal wage compensation.51 Whichever view one favors, both camps predict rising urban–rural wage gaps during the onset of modern economic growth, consistent with the Kuznets hypothesis. As table 5-10 shows, the United States conformed to the historical standard. Between 1800 and 1860, the urban–rural wage gap for male common labor rose from 1 to 27 percent in the North, and from 8 to 28 percent in the South. Note, however, that the ratio of average male urban (excluding white-collar jobs) to rural earnings fell over these six decades, from 2.06 to 1.76 in the North, and from 2.31 to 2.08 in the South. Of course, average earnings were higher in the city than the countryside in part because the former had a much greater demand for higher-paid skilled artisans. But the decline in that city worker earnings premium (remember: without including white-collar employees in the skilled wage) during 1800–1860 reflects the impact of the new manufacturing technology that replaced artisans with unskilled manufacturing operatives and thus “hollowed out” the middle class. We will return to this below.

49

Williamson 1990, 182, table 7.1; 83, table 7.2. Squire 1981, 102, table 30. 51 There is certainly abundant evidence of rising crowding and disease in American cities in the early nineteenth century (Fogel 1986; Costa 2013). There is also plenty of evidence of rising city rents across the late nineteenth century, but the data only start with 1851 (Hoover 1960, 174, table A-3), and oddly enough, they show no evidence of a rise across the 1850s. 50

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

125

Table 5-10 Wage Gaps and Skill Premiums, 1800–1860 1800

1860

Urban–rural wage gap (ratio), unskilled labor

1.01

1.27

Urban–rural average earnings gap

2.06

1.76

Artisan/unskilled

2.48

1.70

Construction/unskilled

1.95

1.58

Urban–rural wage gap, unskilled labor

1.08

1.28

Urban–rural average earnings gap

2.31

2.08

Artisan/unskilled

2.48

2.28

Construction/unskilled

1.87

2.12

Urban–rural wage gap, unskilled labor

1.04

1.27

Urban–rural average earnings gap

2.15

1.82

Artisan/unskilled

2.48

1.83

Construction/unskilled

1.94

1.70

White collar/unskilled

2.42

3.94

Urban North–South wage gap

1.06

1.13

Urban Northeast–South Atlantic wage gap

1.06

1.38

Rural North–South wage gap

1.00

1.15

Rural Northeast–South Atlantic wage gap

1.00

1.38

North

Urban wage ratios

South

Urban wage ratios

United States

Urban wage ratios

Moving to Richer, More Unequal Cities As the urban population (and labor force) shares rose between 1800 and 1860, what were the inequality implications?52 Recall that Kuznets featured the inegalitarian effect of his plausible assumption that average incomes were higher in the nascent modern sector, here represented by America’s cities. For this reason alone, urbanization could have raised inequality up to 1860. It could also have made for more inequality if the rising cities were places where local incomes were inherently more unequal. There certainly are good reasons for expecting the cities to be more unequal. After all, that’s where all the high-wage white-collar and skilled labor was located, and that’s where the richest property owners lived, both of which lived with poor unskilled labor. The available data firmly support both premises—that gaps in average incomes favored the cities, and that the cities were much more unequal places. In 1860, average incomes were 35 percent higher in America’s cities than outside them. The 35 percent gap is wide, though back in 1774 the gaps were even wider, especially in the South Atlantic—by 35 percent in New England, 161 percent in the middle colonies, 348 percent in the South, and 177 percent in the thirteen colonies as a whole.53 In addition, incomes were a lot more unequal in the cities than in rural areas, both for the nation as a whole and within every single region (table 5-11). In 1860, when the data are best, the US urban and rural Ginis were 0.585 and 0.480, respectively (table 5-11), and the difference between them was about a fift h of the total inequality (= 0.105 versus 0.515, respectively), which is a significant magnitude indeed.54 Using the 1860 inequality weights, table 5-11 shows that the American Gini rose by 0.014 just due to urbanization. True, this is a small 52 All the results in this section refer to gaps among all households, both free and slave (or servant) households. Among free households alone, the patterns are the same, though with slightly narrower gaps. For the estimates, see “Urban and Rural Inequality 1860,” http://gpih. 53 These figures refer to household incomes. Recall that in terms of per capita incomes, Boston in 1774 had slightly lower per capita incomes than rural New England because it had much higher dependency ratios. And in the South in 1774, the urban–rural gap is somewhat overstated because Charleston’s slaves were not reported separately, forcing us to include them all in the rural aggregate. 54 Th is change-in-Gini accounting is merely suggestive of the Kuznets urbanization effect’s magnitude. Technically, such calculations cannot be applied using Gini coefficients, although they can be applied using the Theil inequality index.

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

127

Table 5-11 The Kuznets Urbanization Effect, 1800–1870, All Households A.

1860 Gini

Urban population share

Total

Urban

Rural

Urban– rural

1800

1830

1860

1870

New England

0.457

0.536

0.395

0.141

0.082

0.140

0.366

0.443

Middle Atlantic

0.488

0.586

0.402

0.184

0.102

0.142

0.354

0.442

East North Central

0.399

0.560

0.356

0.204

0

0

0.111

0.192

West North Central

0.419

0.578

0.332

0.246

—

0.036

0.154

0.217

South Atlantic

0.608

0.656

0.594

0.062

0.032

0.145

0.118

0.143

East South Central

0.552

0.595

0.543

0.052

—

0.015

0.059

0.088

West South Central

0.569

0.630

0.549

0.081

—

0.187

0.124

0.133

Mountain

0.515

0.520

0.508

0.012

—

—

0.103

0.121

Pacific

0.415

0.492

0.396

0.096

—

—

0.185

0.320

United States

0.511

0.585

0.480

0.105

0.060

0.116

0.198

0.255

B.

1800–1830

1830–60

1800–1860

1860–70

d(share)

d(gini)

d(share)

d(gini)

d(share)

d(gini)

d(share)

d(gini)

New England

0.058

0.008

0.226

0.032

0.284

0.040

0.077

0.011

Middle Atlantic

0.040

0.007

0.212

0.039

0.252

0.046

0.088

0.016

East North Central

0.000

0.000

0.111

0.023

0.111

0.023

0.081

0.017

West North Central

—

—

0.118

0.029

—

—

0.063

0.015

South Atlantic

0.113

0.007

−0.027

−0.002

0.086

0.005

0.025

0.002

East South Central

—

—

0.044

0.002

—

—

0.029

0.002

West South Central

—

—

−0.063

−0.005

—

—

0.009

0.001

Mountain

—

—

—

—

—

—

0.018

0.000

Pacific

—

—

—

—

—

—

0.135

0.013

United States

0.056

0.006

0.082

0.009

0.138

0.014

0.057

0.006

share of the total increase from 1774 (Gini 0.441; see table 2-4) to 1860 (Gini 0.511; see table 5-11), or about 20 percent (= 0.014 versus 0.070, respectively), so the rise in inequality within both American cities and its countryside was the more powerful force. Still, the Kuznets effect was much greater in the Northeast, where urbanization and industrialization was more rapid; the share was about 41 percent in the Middle Atlantic, and 44 percent in New England. We note in passing that the Kuznets effect was weak in the South, where urbanization was slow.

EXPLORING REGIONAL GAPS, SKILL PREMIUMS, AND HOLLOWING OUT Earnings inequality was on the rise in America between 1800 and 1860, and part of it was due to rising wage and income gaps between rural and urban locations. We now turn to those other gaps already previewed: namely, between North and South, skilled and unskilled, and slave and free, along with the possible “hollowing out” of the middle class as manufacturing technology replaced highly skilled (and well-paid) artisans with less skilled (and lower-paid) factory operatives. North–South Wage Gaps The literature on the rise of a North–South wage gap is extensive, especially that which assesses experience across the Civil War decade.55 Gavin Wright used Stanley Lebergott’s farm wage data, and based his analysis on four southern states (Arkansas, Georgia, Mississippi, and South Carolina).56 Robert Margo’s evidence is more comprehensive, including data from the US Department of Agriculture plus Margo’s own wage observations from army posts.57 As we will see in chapter 6, the North–South wage gap rose sharply across the Civil War decade: relative to the North, the South Atlantic nominal common labor wage fell from 0.81 to 0.55, and the South Central from 1.03 to 0.69.58 The increase 55

Margo 2000, 2002. G. Wright 1986, 76. 57 Blodget 1901, 1903; Lebergott 1964; Margo 2000. 58 Margo 2002, 34, table 1, panel B. 56

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

129

in the gap was just as dramatic for female domestics. In addition, the gap persists up to the 1890s, so it was not just a temporary, war-related event. But how far back does the rising North–South wage gap go? We saw in chapters 2 and 3 that it wasn’t a characteristic of colonial times, since southern wage rates were on par with, or even higher than, those in New England and the middle colonies. Chapter 2 confirmed that in 1774, farm wages were higher in the South, and urban unskilled wages about on par. Things had not changed all that much by 1800, when farm wages were on par between the two regions, and urban unskilled wages were only 6 percent higher in the North (table 5-10, above). But what happened thereafter, as modern economic growth took place? Table 5-10 reports that the gap rose between 1800 and 1860, and it did so both for the urban unskilled and farm labor. The ratio of wages in the Northeast relative to the South Atlantic grew from 1.06 to 1.38 for urban labor, and from 1.00 to 1.38 for farm labor. No such increase in the gap took place on the other side of the Appalachians, however, so the gap between the North (including the North Central) and the South (including the South Central) rose only from 1.13 to 1.15. Of course, some of the rising nominal wage gap between Northeast and South Atlantic could have been due to a rise in the relative cost of living in the more urban Northeast. Unfortunately, we lack information on regional cost-of-living trends covering the six decades before 1860. We do have some evidence of those trends for 1840–1880. Relative to New England (or the Middle Atlantic), the cost of living in the South Atlantic fell by 18.6 (or 16.2) percent, 11.1 (or 8.4) percent in the West South Central, and 3.5 (or 0.7) percent in the East South Central.59 Nevertheless, we simply do not know how much of this relative cost-of-living rise in the North—averaging about 5 percent per decade—took place during the Civil War decade and the one that followed, and how much before. 59 In an earlier work, we developed a crude device whereby the available urban cost-of-living data are converted to statewide estimates; see Williamson and Lindert 1980, 323–25, appendix I. In their pioneering study of regional costs of living during 1850–1880, Philip Coelho and James Shepherd (1974) were not able to report any cost-of-living estimates for the West South Central, and their South Atlantic series started only with 1866, and that was for Jacksonville alone.

130

•

CHAPTER 5

Earnings Gaps between Free and Slave Although we use the term “earnings” to describe the gap between free and slave, the slave did not, of course, receive a market wage but rather some subsistence portion of it. The gap between free and slave earnings could have risen for two reasons. First, the share of income produced that the slave was allowed to consume might have fallen over time, perhaps the result of rising slave productivity and more stable subsistence consumption. Second, due to occupational and schooling discrimination, the slave was not allowed to undergo the same human capital accumulation that free workers could and did. We don’t know whether the share of the slave’s marginal product retained for consumption rose or fell. As we note in appendix C, our estimates have postulated a slight rise in the share retained by slaves, from a bit below 50 percent for 1774–1800, before the British shut off the Atlantic slave trade, to 50 percent for 1850–1860. In what follows we assume a constant 50 percent share, with no change in slave occupations. Thus, the free–slave earnings gap would have been driven upward as free labor became more urban and more skilled. Using those assumptions, the free–slave earnings gap is estimated to have risen from 2.11 to 5.17 over the six decades, thereby producing another source of rising inequality. Did the Skill Premium Rise? As Goldin and Katz have shown for twentieth-century America, the skill premium—and the reward for schooling—is driven by supply and demand: when the supply of those skilled and schooled grows slowly and the demand grows fast, the rewards to scarcer skills and schooling goes up. Demand always seems to race ahead of supply during the early years of modern economic growth, and thus we expect the skill and schooling premium to have gone up, the wage structure to have stretched wider, and earnings inequality to have increased. While modern economists explore this issue across the whole skill and schooling range, economic historians have more limited evidence. Indeed, typically they have measured the skill premium using the wage rates of the building tradespeople—masons, carpenters, ship riggers, and such—relative to

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

131

their assistants or common labor.60 Such measures do not, however, speak to the value of schooling and literacy. Fortunately, our social tables arm us with better evidence since they document the premium that literate and numerate white-collar employees got relative to typically illiterate and less numerate unskilled labor.61 Table 5-10 documented quite clearly that modern economic growth in the young republic pushed up the skill premium dramatically. The ratio of urban white collar to urban common labor rose from 2.36 to 3.87 in the North (up by 64 percent) and from 2.48 to 4.25 in the South (up by 71 percent). This estimated rise in the skill premium from 1800 to 1860 is steeper than the increase that has recently been estimated for the 1820s to 1850s, up only 11 percent, but ours covers two or three more early nineteenth-century decades and has a much more comprehensive coverage of “high-skill” white collar.62 Was There a Hollowing Out of the Artisan Middle Class? The well-studied decline of the artisan during early industrialization has been given fresh legs in a paper by Jeremy Atack, Fred Bateman, and Margo as well as a more recent paper by Katz and Margo that exploits IPUMS census data for 1850–1910.63 Atack, Bateman, and Margo make their argument by looking at average wages by manufacturing establishment. Katz and Margo make theirs by estimating trends in both the 60 For perhaps the best illustration of this tradition, see Phelps Brown and Browne 1968. One of the present authors is guilty of what he now thinks is a misleading proxy. See Williamson 1980, 1982. 61 Katz and Margo (2013) use the wages of male clerks hired at army posts as their proxy for white-collar earnings. Since many forts were quite distant from urban America, one wonders about the effectiveness of their proxy. And was there really a “law of one wage” prevailing between military posts and the private economy? We do not have to adjust for differences in the days worked per year (higher for white collar), since by assumption we take them to have been constant within occupations over time. 62 Katz and Margo 2013, 58, table 5, panel B. The Katz and Margo estimates are based on clerks (relative to common labor) hired at army forts (Margo 2000). Our urban white-collar, average earnings estimates are calculated as an employment weighted average of male white merchants, shopkeepers, professionals, officials, teachers, clerks, and such. Yet our measured rise in the skill premium during 1800–1860 may be overstated since our white-collar earnings coverage is thinner for 1800, missing many high-wage occupations that are covered in 1860. 63 Atack, Bateman, and Margo (2004) are responding to earlier contributions. For more recent contributions on the issues discussed here, see Goldin and Katz 2008; Katz and Margo 2013. For a good example of the de-skilling literature, see Brown and Phillips 1986.

132

•

CHAPTER 5

ratio of artisan to common labor wages in manufacturing starting with the 1820s, and the occupational skill mix for manufacturing and economy-wide, starting in the 1850s.64 While our period is much longer, we also report both the artisan share in all nonfarm employment and earnings ratios between them. If there was a demand-led hollowing out of the artisanal class, it should have been reflected in a combination of falling relative pay and falling relative employment numbers for that class. That is, if factories were replacing the artisan in a big way, we should see artisanal employment falling relative to nonfarm employment and artisans’ earnings advantage eroding. We should also see plenty of evidence of it by 1860, by which time industrialization had advanced considerably, especially in the Northeast and the East North Central regions. We focus on skilled artisans, the workers considered in the deskilling literature and who were in the middle of the earnings distribution; in 1800, artisans’ average incomes were 29.3 percent above the nonfarm average and 64.8 percent above the economy-wide average. Consistent with the hollowing out hypothesis, table 5-10 documented an erosion in the relative earnings of artisans between 1800 and 1860: relative to the urban unskilled, artisans earnings fell from 2.48 to 1.70 in the North (down by 31 percent), 2.48 to 2.28 in the South (down by 8 percent), and 2.48 to 1.83 in the United States as a whole (down by 26 percent). Katz and Margo report a somewhat-smaller fall: 8 percent.65 Tables 5-10 and 5-11 together document how—in the words of Goldin and Katz—skilled artisan supply ran ahead of demand as manufacturing replaced those skills with unskilled operators. In the industrializing Northeast, the share of skilled artisans in the total nonfarm labor force rose from 20.8 percent in 1800 to 24.9 percent in 1860, and their share of blue collar and unskilled rose from 26.9 to 32.6 percent.66 In spite of declining relative incomes, skilled artisans were growing faster in the industrializing North during these decades of early modern economic growth (table 5-12). One could speak of a hollowing out of the 64

Katz and Margo 2013, 57, table 4, panels A and B. Katz and Margo 2013, 58, table 5, panel B. 66 Katz and Margo do not report their estimates for the Northeast alone, and their evidence does not cover 1800–1850, but their economy-wide artisan employment share falls modestly from 11.6 to 11.2 percent over the 1850s. Although table 5-11 does not report 1850, our artisan share for Northeastern urban employment falls from 28.5 to 24.9 percent. 65

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

133

Table 5-12 Was There Artisan Middle Class Hollowing Out between 1800 and 1860? 1800

1860

Northeast

0.208

0.249

South

0.270

0.183

Thirteen colonies

0.231

0.241

United States

—

0.206

Northeast

0.269

0.326

South

0.408

0.253

Thirteen colonies

0.315

0.318

United States

—

0.272

Share of skilled artisans in the total nonfarm labor force (males only)

Share of skilled artisans in the blue-collar and unskilled nonfarm labor force (males only)

middle class in the early republic, although it took the form of a significant decline only in relative incomes, not in relative employment numbers. A Final Force: Rewards at the Top during the Financial Revolution The fortunes of the US financial sector are likely to have moved in lockstep with overall inequality and the concentration of incomes at the top. Such was the case in the twentieth century, as we will demonstrate in chapters 8 and 9. For the years of modern economic growth before 1860, we cannot document trends in the rewards to financial actors, except for the following clue. Our social table for 1774 can document almost nobody in the financial sector; by contrast, the IPUMS 1860 sample for the Middle Atlantic region, where finance developed most extensively, tells us that 0.4 percent of the free population had occupations in that

134

•

CHAPTER 5

sector, and their gross wealth ($23,491) was 8.2 times the average wealth for all free households ($2,855). While we cannot offer any better evidence on rates of pay in the antebellum fi nancial sector, we can say a lot more about its impressive aggregate boom. Recent work, especially by Richard Sylla and his collaborators, has documented an enormous fi nancial expansion—in quantity, quality, and depth—in the young United States. Indeed, the splendid financial . . . accomplishments of the US from [the] 1790s to the 1830s when its fi nancial system came to equal, and even to surpass, the older, less liberal system of the UK . . . [were] the best in Europe.67

And it began quickly. Only a few years after ratifying its constitution, the United States had already “stabilized its public finances, restricted its [war] debts, introduced the . . . dollar as convertible currency, formed both a banking system and a national bank with branches, [and] established [an integrated] securities [market serving all] major cities.”68 What numbers we do have certainly confirm this view. Between 1800 and 1835, the capital in US state-charted banks rose by 8.6 percent per annum, and their numbers increased by 9.1 percent per annum, all considerably in excess of GDP growth.69 Of course, not all of those seven decades were smooth and without crisis. For starters, populist president Andrew Jackson vetoed the Bank of the United States recharter in 1832, and the financial boom was somewhat less spectacular over the next few decades. Nonetheless, between 1800 and 1860, the number of state banks rose by 6.8 percent per annum, well above the growth rate of real GDP. It seems reasonable to conclude that the financial boom was another force contributing to the observed inequality surge up to 1860.

67 Sylla 2009, 237. For example, the number of securities listed in the US securities market (mainly New York City and Philadelphia) in 1830 was 226, while it was 206 in London that same year (Sylla 2009, 225, table 7.2). 68 Rousseau and Sylla 2005, 3. 69 Sylla 1998, 86, table 1. Much of that expansion occurred in the four major port cities of Baltimore, Boston, New York City, and Philadelphia (Sylla 1998, 94).

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

135

INTERPRETING AMERICA’S EARLY EXPERIENCE WITH UNEQUAL MODERN GROWTH Between 1800 and 1860, America entered the modern economic growth club—perhaps the first country to do so. Its per capita income grew fast enough for it to overtake Britain and resume its position as the world’s per capita income and living standards leader. Combined with America’s fast population growth, even the Northeast had achieved a larger GDP and thus a domestic market that outdid all but a few European countries (which had poorer interregional market linkages). Over the same period, American inequality rose steeply. Almost all components of inequality change contributed to the overall widening of wage and income gaps. The distribution of property incomes became more unequal. Earnings inequality also increased, driven by urbanization (since urban inequality was larger than rural inequality), rising urban–rural earnings gaps, rising North–South income gaps, rising skill premiums, a hollowing out of the high-wage artisan in the middle, and a growing “earnings” gap between free labor and slaves.70 What do we know about the underlying forces that could have caused such a great widening of all kinds of income gaps during America’s first period of modern economic growth? The answers cannot be based on any econometric exercise, at least not yet. Rather, we can only stack these inequality movements side by side, and ask whether a few powerful forces could explain them all. What follows is a list of what needs to be explained: • income inequality rose in America as a whole and every region of the Eastern Seaboard—for both the entire and the free population • inequality of property incomes did the same • labor earnings inequality did the same • income inequality was greater in the cities than in the countryside, and the rate of urbanization was fast • the gap between average incomes in the city and the countryside widened 70 The only components that did not contribute to the rise in overall income inequality were shifts in factor shares and shifts between regions. The share of property income in total income did not rise, and the westward expansion was not to higher-income regions, except for the shift to the Pacific region.

136

•

CHAPTER 5

• the gap between urban and rural wage rates for common labor widened • the gap between northern and southern wage rates widened • the skill premium rose, as white-collar earnings increased faster than did that for unskilled labor

Why did all these inequality developments occur in America’s first experience with modern economic growth? We begin by noting one big fact that this book has exposed: the first great rise of inequality in America happened only between 1790 or 1800 and 1860, not earlier or later, at least not until the steep income widening that has taken place since the 1970s. What was so different about the early nineteenth century, when inequality rose dramatically, and the early twentieth (1913–1953), when it fell dramatically (chapter 8)? And what conditions did the 1800–1860 era share with the era of rising inequality since the 1970s (chapter 9)? We will suggest some leading candidates here, and will expand on them when the twentieth and twenty-first centuries come into view in chapters 8–10. We can first rule out a force that could not have played a role: namely, government redistribution policies were not introduced in the United States before the late twentieth century. Government budgets remained tiny until the Civil War. The inequality surge from 1800 to 1860 would have been just as steep, including or excluding the effects of taxes and transfers on the rich and poor since those effects were so small.71 So what’s left? We think there were three prime movers of American inequality between 1800 and 1860. Rapid labor force growth. The first likely candidate was rapid labor force growth, which raised the ratio of labor to improved land and other natural resources. America’s extraordinarily high rate of natural increase and high net immigration from the 1840s onward raised resource asset prices, especially the purchase price of land, relative to wages. This force could have made inequality rise by itself. The frontier was slowly

71 The only government budgetary item that affected income gaps between rich and poor was local public spending on education, which varied between tiny, pro-rich effects in the South and larger, pro-poor effects in the North. Such redistributions through education would have a small effect on the overall income distribution, however, given the small size of education budgets in that era.

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

137

fi lling up, while the labor force was growing much faster than at any time thereafter. Rapid technological progress favoring industry and cities. The early nineteenth century saw major productivity-enhancing advances in technology. While some of these were achieved in agriculture, such as the development of new seed varieties and the mechanization of the harvest, the rate of progress was greater in industry as well as such tertiary sectors as transportation, trade, and fi nance.72 One by-product of this bias against agriculture was to replace colonial ruralization (chapter 3) with rapid urbanization. The greatest gains went to city builders, city connectors, and city dwellers. Given that cities were already places of greater income inequality and higher average incomes before these developments, the effect was to accelerate the rise of inequality. Financial development. The financial stabilization of the 1790s secured investors against the massive uncertainties introduced by the Revolutionary War. Inflation was checked, the credit of government was established, and land became a more secure collateral for both individuals and governments, especially after 1815.73 The drop in risk-adjusted interest rates was aided greatly by the transatlantic financial integration we described at the beginning of this chapter. These financial developments contributed to rising inequality because the gains accrued disproportionately to the older, more established locations and those at the top holding financial wealth and, as we will see in a moment, financial skills. How did the rate of return figure in this causal transmission from improved financial institutions to inequality? Contrary to intuition, and contrary to Piketty’s emphasis on the positive correlation between the rate of return and inequality, the years from 1800 to 1860 were accompanied by a decline in the current returns on most traded assets. As of 1800, financial security was still not sufficient to have brought down risk premiums. Thus, chapter 4 considered 6 to 8 percent as the prevailing net rate of return. By 1850 and 1860, the prevailing net rate of return was only 5 percent. Financial improvements in America and the integration 72

Olmstead and Rhode 2008. On the initial 1790s’ wave of institutional improvements, see Irwin and Sylla 2011, chaps. 2, 4–8, and the sources cited there. On the long-run movements in realized rates of return on bonds, stocks, and commercial loans, see Sylla, Wilson, and Wright 2006. 73

138

•

CHAPTER 5

of Anglo-American capital markets drove the net rate down, leading to capitalization in asset values, thereby raising the wealth–income ratio. The wealthy pulled ahead of the rest of society, especially in the cities of the Northeast, on the strength of their savings, access to credit, and capital gains, not from any rise in the rate of return. And the wealthy pulled ahead for another reason: the development of the domestic financial sector—and its integration with Europe—required special skills, which were rewarded by increasingly large earnings. These three forces will also be featured later in this book, when we explain both the great egalitarian leveling from 1913 to 1970 (chapter 8) and the second great rise in inequality from 1970 to the present (chapter 9).

UNEQUAL ECONOMIC GROWTH, 1800– 1860

•

139

Some Notable (and Less So) Individuals in the US Censuses of 1850–1870 They were real people who answered the census takers’ questions about their wealth and occupations in 1850, 1860, and 1870, and the census tried to visit them all. Some became famous as national political leaders or popular writers. Some were superrich. Some were simply ancestors of the two authors. Here we display only their names, wealth, stated occupations, and places of current residence, even though the census also asked their ages, where they were born, where their fathers were born, and so forth. Some Individuals in the US Censuses of 1850–1870 Person

Real estate ($)

Personal estate ($)

Occupation given in census

Place

Anon, IPUMS richest

300,000

—

Bank manager

New York City

Stowe, Calvin E, and

—

—

Professor, Bowdoin

Brunswick, ME

Stowe, Harriet Beecher —

—

—

Brunswick, ME

Vanderbilt, Cornelius

300,000

—

Steam boat proprietor

New York City

Williamson, Triphemius

3,500

—

Farmer

Wilcox County, AL

Anon, IPUMS richest

300,000

300,000

Agent

St. Mary, LA

Douglass, Frederick

4,000

5,000

Editor, newspaper

Rochester, NY

Ingalls, Charles and Caroline

1,000

50

Farmer

Pepin, WI

Lee, Robert E.

80,000

—

US Army

Alexandria, VA

Lincoln, Abraham

5,000

12,000

Lawyer

Springfield, IL

Lindert, Frederick

200

80

Farmer

Carver County, MN

McCormick, Cyrus

278,000

1,750,000

Reaper factory

Chicago

Ogden, William Butler

1,500,000

1,000,000

Lawyer

Chicago

Stowe, Calvin E., and

—

1,000

Orthodox clergy

Andover MA

—

—

Andover, MA

In the 1850 census

In the 1860 census

Stowe, Harriet Beecher —

Person

Real estate ($)

Personal estate ($)

Occupation given in census

Place

Vanderbilt, Cornelius

—

—

—

Richmond, NY

Williamson, Triphemius

3,500

12,000

Farmer

Clarke County, AL

Anon, IPUMS richest

700,000

800,000

Retired, forty-fiveyear-old

New York City

Chesnut, James, Jr., and Mary

15,000

2,200

Farmer

Kershaw County, SC

Clemens, Samuel Langhorne (Mark Twain),

—

10,000

Imprinter, daily paper

Buffalo, NY

and Clemens, Olivia

14,000

8,000

Keeping house Buffalo, NY

Douglass, Frederick

9,000

13,000

Lecturer

Gale, William F.

5,000

5,500

Harness maker Stockbridge, MA

Ingalls, Charles and Caroline

—

200

Carpenter

Rutland, KS

Lee, Robert E.

—

25,000

President, Washington College

Lexington, VA

Lindert, Frederick

1,000

500

Farmer

Carver County, MN

Ogden, William Butler

3,000,000

850,000

Capitalist

Chicago

Stanford, Leland

90,000

8,653,180

President, Sacramento, CA Central Pacific

Stowe, Calvin E., and

20,000

10,000

Clergyman

Hartford CT

Stowe, Harriet Beecher —

—

Authoress

Hartford, CT

Vanderbilt, Cornelius

—

—

New York City

500

Farmer

Clarke County, AL

In the 1870 census

—

Williamson, Triphemius 500

Rochester, NY

Sources and notes: The sources are the 1850, 1860, and 1870 population censuses, accessed via Ancestry.com, February 2015. We are grateful to Richard Sutch for pointing out the entries for Stanford and Vanderbilt. The 1860 entry for William Ogden was reported in Soltow 1975, 5. An em dash means this entry was left blank in the original census.

CHAPTER 6

The Civil War Growth Lost, Freedom Gained, Inequality Maintained

D

id the Civil War cause a slowdown in American growth? If so, was the setback all Southern—caused by war damage, lost accumulation due to trading plowshares for guns there, and reductions in Southern labor supply due to slave emancipation—or was there a slowdown up North as well? These are old questions, raised famously by Charles and Mary Beard, and then elaborated on by Louis Hacker less than two decades later.1 While these historians were more interested in the extent to which the Civil War was an institutional and political watershed—shift ing political power up North and into the hands of pro-growth capitalist interests—Hacker in particular thought that the Civil War also induced a speedup in manufacturing growth during the war decade itself.2 Twenty-five years or so later, economic historians in the 1960s and 1970s challenged the Hacker speedup thesis and instead identified a slowdown.3 Gallman’s estimates of commodity production and national income have raised many questions about the propositions that the Civil War accelerated the course of economic growth and that the Civil War itself was one of rapid economic expansion.4 One of the present authors summarized what Gallman’s fi ndings implied: From 1860 to 1870, commodity output growth reached its lowest point anywhere in the nineteenth century, two percent per annum. The same is true of manufacturing value added. . . . The annual rate of growth of 1

Beard and Beard 1927; Hacker 1940. Hacker 1940, 252, 437–38. 3 Gallman 1960; Cochran 1961; Gilchrist and Lewis 1965; Engerman 1966; Williamson 1974. 4 Engerman 1966, 176. 2

per capita commodity output in the victorious North was only one percent during the war decade.5

Using quite different evidence from that cited by those writing in the 1960s and 1970s, this chapter will confirm the slowdown thesis. Indeed, not only will it document a slowdown but it will also document an absolute decline in per capita income in real terms. Furthermore, that decline was big enough so that the United States lost its pre–Civil War income per capita lead over Great Britain. While we noted in chapter 5 that the United States was well ahead of Britain in 1860, in 1870 it was no longer ahead in real purchasing power. Thus, for the rest of the nineteenth century, the United States had to regain lost leadership for the second time in its history. We all understand that emancipation was a powerful redistributive force in the South. But just how big was it? If southern inequality fell, how much of that fall was due to emancipation—property incomes at the top were slashed when whites lost their slave wealth while black incomes at the bottom rose—and how much was due to destructive wartime forces like those that were so powerful during the Revolutionary War and early federalist years? What were the components of black income gains? How much was due to the ability of emancipated slaves to retain all the income from their labor? How much of that was offset by a reduction in black household labor supply? And how much was due to blacks’ now-greater ability to move to higher-wage occupations and locations? How big were the emancipation gains for black Americans, as compared with their Great Migration to higher-wage northern cities from World War I to the 1960s? And how big compared with the impact of the civil rights movement since? And what happened to inequality up North? Did it continue its upward march that we documented in chapter 5 for the years 1800 to 1860? The traditional literature certainly thought so. Hacker and Wesley C. Mitchell thought that the war inflation across the 1860s eroded real wages.6 Since Mitchell had no evidence documenting any rising returns on financial assets or rents, it implied a rise in the share going to what Mitchell called the residual claimant—profits. Apart 5 6

Williamson 1974, 637. W. Mitchell 1903; Kessel and Alchian 1959.

THE CIVIL WAR

•

143

from real wages, Hacker and Mitchell did not have any income distribution information to test their rising inequality thesis. True, Hacker would have been helped by Rufus Tucker’s assessment of the 1863–1870 tax returns, but not by the subsequent work of Anna Schwartz, who could not document any significant increase in constant price corporate profits for 1864–1871.7 Until the estimates cited later in this chapter, no useful income distribution evidence had been added to the mix.8 This is the chapter’s agenda. We start with the income distribution for 1870.

AMERICAN INCOMES IN 1870 What we said previously in chapter 5 about our sources for 1850 and 1860 applies with even greater force for 1870. The 1870 census recorded individuals’ occupations, wealth, and other attributes in greater detail than earlier, and the recent IPUMS offers thousands of such individual records for each census region. Source materials on wages, white-collar earnings, farm economics, and earnings by race are also expanded.9 Appendix F describes how we have fashioned this information into our 1870 income estimates. Our aggregate (gross) income estimate for 1870 is $9,057 million in current dollars (table 6-1), which in war-inflated prices is 72.6 percent higher than 1860 ($5,246 million in current dollars; see table 5-2). Once again, that estimate uses the baseline part-time assumption about annual days worked. Before we explore the implications of the income estimates, we need to report that our 1870 nominal income estimate is well above the best GDP estimate in the Historical Statistics of the United States.10 As we pointed out in chapter 5, the large gap between us may be explained by flawed price deflators or their underreporting of the 7

Tucker 1938; Schwartz 1960. There is, however, added information on the distribution of total estate wealth held by adult males (Soltow 1975, 99–103). 9 The 1870 census reported black versus white respondents, while the 1850 and 1860 censuses reported slaves separately. Although the 1850 and 1860 IPUMS samples divided free population between the racial categories white, Negro, American Indian, and Chinese, our income results are based on aggregating the whole free population together. 10 Carter et al. 2006, series Ca10. 8

144

•

CHAPTER 6

Table 6-1 Estimated American Personal Incomes in 1870 (Millions of current dollars) Labor earnings

Farm profits

Property income

Total income

New England

804.2

87.7

189.9

1,081.8

Middle Atlantic

2,031.2

168.8

468.1

2,668.2

East North Central

1,395.0

235.0

382.9

2,012.9

West North Central

638.2

101.8

137.0

876.9

4,868.6

593.3

1,177.9

6,639.8

South Atlantic

570.0

170.6

96.2

836.8

East South Central

511.8

176.0

83.2

771.0

West South Central

271.6

105.3

29.9

406.8

1,353.4

451.9

209.3

2,014.6

Mountain

81.3

3.2

5.6

90.1

Pacific

235.8

21.9

54.8

312.5

317.1

25.1

60.4

402.6

6,539.1

1,070.3

1,447.6

9,057.1

North

South

West United States

service sector. Others build up their estimates from the real output side, using production indexes to project forward and backward from some constant price base year. For the inflationary Civil War decade, the price deflators will matter a great deal (for both the federal North and the confederate South) in any search for a reconciliation of our direct income estimates with that of others as well as the growth of real GDP they imply. What are the first-order messages coming from table 6-1? We start with Turner, who—by exploiting population density—argued that the frontier had stopped moving westward with the 1890 census.11 One could also contend that the share of US incomes generated west of the Appalachians had stabilized two decades sooner. In 1870, the income share of the Eastern Seaboard was 51 percent, down only trivially from 52 percent in 1860. Thus, the intensive development east of the 11

Turner 1921.

THE CIVIL WAR

•

145

Appalachians had finally matched the extensive development west of the Appalachians. One might also conclude from table 6-1 that human capital must have accumulated faster than conventional capital across the Civil War decade given that the gross property income share in GNI fell from 22 percent in 1860 (tables 5-1 and 5-2) to 16.6 percent in 1870, without any fall in the nominal rates of return. What was true of the United States as a whole, however, was driven mainly by events in the South, where the property share fell from about 33 percent in 1860 to 11.1 percent in 1870. This huge drop in the property share and the corresponding rise in the labor share was not due to some spectacular human capital accumulation down South but rather to a spectacular collapse in southern property, and hence in property incomes, due to slave emancipation and wartime destruction. Southern property incomes (in millions of current dollars) fell from 534.8 in 1860 to 209.3 in 1870. The Beards used this language to describe it: A complete destruction of . . . [assets] in possession of slave owners without compensation—[was] the most stupendous act of sequestration in the history of Anglo-Saxon jurisprudence.12

We will have more to say about the income redistribution down South in what follows below. In the North, the property share drifted upward modestly from about 18 to 18.3 percent of the total income.

THE COST OF WAR: ECONOMIC DECLINE AND LOST WORLD LEADERSHIP The Civil War caused by far the greatest wartime loss of life in the country’s history.13 Including the effects of the net immigration shortfall and a decline in births, the annual rate of population growth declined from 3.1 percent before the war to 2.4 percent across the 1860s. It also caused incomes to drop, even on a per capita basis. 12

Beard and Beard 1927: Volume II, 100. Engerman (1966, 193–94n17) reports an estimate that 4 percent of the 1865 population was lost during the Civil War, or 635,000 due to war deaths. In addition, there was a shortfall of immigration and births. The immigration shortfall has been estimated at 500,000 for the 1860s, and the white birth shortfall at 700,000 from 1860–1872 (Coale and Zelnick 1963, 25). 13

146

•

CHAPTER 6

Since the war was mainly fought on Southern soil, the losses were, of course, concentrated on the Southern white population. The South’s income loss on a per capita basis was severe even when we include the income of the emancipated black population, as illustrated by the 1860– 1870 plunge in the South Atlantic region’s income per capita in tables 6-2 and 6-3. The results for the defeated South are stunning: income per capita fell by 26 percent in the South Atlantic, 15 percent in East South Central, and 25 percent in the West South Central. The economic cost of the Civil War to the South is close to the total American economic losses associated with the Revolutionary War and early federalist struggles. Since the incomes of emancipated blacks rose, the decline in southern white incomes must have been even bigger, as we will confirm below. The victorious North did not prosper during the decade either. In contrast with the great gains in per capita incomes up North between 1800 and 1860, table 6-2 can barely report any progress at all across the decade. The North only managed a 6.3 percent total increase in real per capita income over the decade, and even the West, having reached diminishing returns on its mineral super-rents, underwent a 38.7 percent per capita income decline. As a result, the purchasing power of the average American’s income dropped enough (6 percent) to lose the lead over Great Britain by 1870. True, most of that is explained by the devastating losses in the South. But still, income per capita barely increased at all in the industrial North. Meanwhile, America’s western European competitors did not stand still but instead grew at a fast clip. The income per capita growth rate for seven western European countries (table 6-2, panel C) was 1.33 percent, close to America’s modern economic growth performance between 1800 and 1860. Great Britain was the leader of the pack, achieving an impressive growth rate of 2.59 percent per annum in its income per capita. Once again American income leadership proved vulnerable, as the Revolutionary War and early federalist years revealed. While American per capita income in 1860 was 17 percent higher than Britain in sterling and 50 percent higher in the ability to buy a bare-bones consumer bundle, it was no higher than Britain in 1870 (12 percent higher in sterling and only 3 percent higher in purchasing power). Leadership was only regained at the start of the twentieth century, and it was THE CIVIL WAR

•

147

Table 6-2 Real Incomes per Capita, 1860–1870 A. 1870 regional income (in 1840 dollars) and population Incomes (per capita)

Population (million)

New England

203

3.473

Middle Atlantic

199

8.777

East North Central

145

9.085

West North Central

149

3.844

172

25.179

South Atlantic

94

5.837

East South Central

115

4.397

West South Central

131

2.023

South

107

12.257

Mountain

191

0.308

Pacific

307

0.665

270

0.973

154

38.408

North

West United States

B. Regional income per capita growth rates, 1860–1870 % per annum

Total % change

New England

1.32

14

Middle Atlantic

0.74

8

East North Central

0.56

6

West North Central

0.55

6

South Atlantic

−2.99

−26

East South Central

−1.56

−15

West South Central

−2.62

−25

Mountain

−1.78

−16

Pacific

−5.24

−42

United States

−0.28

−3

Table 6-2 (cont.) C. West European per capita income growth rates, 1860–1870 (% per annum) Austria

0.47

Belgium

1.62

France

-0.08

Germany

1.16

Netherlands

1.43

Switzerland

2.09

United Kingdom

2.59

West Europe

1.33

subsequently reinforced by Europe’s damage and America’s prosperity during World War I. Only during the Great Depression of the 1930s did America again lose its lead over some other industrialized countries. In that depression, America’s income per capita fell a bit behind that of Switzerland, while Australia, Britain, the Netherlands, and New Zealand virtually, but only temporarily, caught up before World War II. During World War II, American incomes surged far above those of almost all countries, replicating the World War I years.14 The Civil War debates ignited by the Beards and Hacker stressed the performance of industry. While our social tables do not easily speak to that issue, other data have accumulated since the Beard and Hacker thesis was challenged in the 1960s and 1970s. To begin with, Hacker relied on manufacturing census information in current prices—not a good idea for assessing an inflationary decade, as Mitchell pointed out long before Hacker.15 The Frickey index of manufactured durables fell during the war years, but nondurables did not; the net result was “a much slower rise [in manufacturing output] from 1861 to 1865 than 14 Th is paragraph’s comparisons of “purchasing power” use the Allen bare-bones price deflators, whereas the preceding paragraph used conventional US price deflators. The comparisons of real income per capita in the 1930s are based on Maddison’s last estimates. See also table 10-1 and figure 10-2. 15 Hacker 1940, 437–38, appendix A; W. Mitchell 1903.

THE CIVIL WAR

•

149

from 1866 to 1870.”16 And according to Gallman, manufacturing output as a share of the total commodity output fell over the 1860s.17 To clinch the manufacturing-did-not-speed-up case, we now have Joseph Davis’s new US industrial production index, which documents a growth rate of 5.32 percent per annum between 1800 and 1860, falling to 4 percent across the 1860s.18 Why would we expect this slowdown in the North? There are at least three reasons, and they were all stressed in the literature of the 1960s and 1970s. They are also strikingly similar to the forces responsible for the related economic disaster between 1774 and 1790. First, all of those young adult males in the armies (and disabled after the peace) meant a reduction in the size of the effective labor force per capita and downward pressure on output per capita—a downward pressure not offset by more rapid productivity growth and capital accumulation. Second, labor productivity should have felt downward pressure as more women and older adult males had to take up some of the slack on farms and factories harmed by the departure of young adult males to battle. And labor productivity did fall: “the sharp decline in value-added per worker in manufacturing from 1860 to 1870, 13 percent, is a unique occurrence for the nineteenth century.”19 Finally, the “North could hardly have achieved significant rates of capital accumulation, ‘burdened’ as [it] was with war financing.”20 All these forces applied to the South as well, although there were two more that crippled the Southern economy. First, the Northern armies reduced the slave labor force by freeing blacks, which then allowed them to escape to free states. This source of slowdown would, of course, become permanent with the fall of the Confederacy. Second, the Northern blockade caused cotton exports to collapse. But while this force was big, it was only temporary since southern cotton exports recovered during the late 1860s.

16

Frickey 1947, 64; Williamson 1974, 638. Gallman 1960, 1966. 18 J. Davis 2004, 1189, table III. 19 Engerman 1966, 179. 20 Williamson 1974, 648. 17

150

•

CHAPTER 6

CIVIL WAR COSTS AND SOUTHERN WHITE INCOMES How much did the Civil War cost southern whites? Table 6-3 offers an assessment for each of the three southern regions, broken down into labor income, property income, and farm profits. We have already seen that real per capita income fell by 26.3 percent between 1860 and 1870 in the South as a whole (table 6-2). But since the incomes of emancipated blacks must have risen, southern whites must have undergone an even bigger fall in their average incomes. Indeed, they did. In constant 1840 dollars, their incomes fell on average by 30.5 percent (table 6-3), with all three southern regions experiencing similar drops. These are huge costs, but they must have fallen mainly on slaveholders and landowners, since property incomes absorbed most of those costs. In current prices, property incomes in the South fell by about 61 percent (tables 5-2 and 6-1). Property income from both realty (affecting landowners) and personalty (affecting slaveholders) fell in current prices—the former by 28.2 percent, and the latter by a huge 77.3 percent. Perhaps the best summary of the cost of the Civil War to the losers is this. Total southern per capita income in 1840 prices fell by 13 percent, but labor incomes (free and slave in 1860) rose by 6.6 percent while property incomes fell by an enormous 72.9 percent (tables 5-2 and 6-1). As another indicator, while 59 percent of the top 1 percent of the wealthiest American men in 1860 were southerners, the figure was only 18 percent in 1870.21 But they lost more than their wealth. Legislation increased the power of black tenants and farm laborers, suppressing returns for landowners. If they survived the war, the less propertied in the Confederate ranks didn’t pay much for the loss. Rather, their Confederate officers in front of them—“the flower of the South” from elite families—lost almost everything, including many of their lives. Finally, while inequality fell in the South, the economic gap between it and the North rose dramatically. In prewar 1860, the average per capita incomes in the South were 15.9 percent below those of the North, while the postwar 1870 gap was 37.8 percent. Thus, the North–South gap more than doubled. Interestingly, the rise in the North–South income per capita gap across the Civil War decade was almost the same 21

Soltow 1976, 101.

THE CIVIL WAR

•

151

Table 6-3 Southern Free and White Incomes, 1860–1870 (Millions of current dollars) 1860 free

(%)

1870 white

(%)

Labor income

333.9

50.2

440.8

63.6

Farm profits

84.7

12.9

158.0

22.8

Property income

238.3

36.3

94.4

13.6

From realty

72.0

11.0

58.4

8.4

From personalty

166.3

25.3

36.0

5.2

Total income

656.9

—

693.2

—

Income per capita (1840 dollars)

184.0

—

121.4

—

Labor income

234.2

46.6

391.3

61.9

Farm profits

74.3

14.8

159.8

25.3

Property income

194.4

38.7

81.2

12.8

From realty

58.6

11.7

46.8

7.4

From personalty

135.8

27.0

34.3

5.4

Total income

502.9

—

632.4

—

Income per capita (1840 dollars)

180.4

—

134.4

—

Labor income

144.5

52.3

198.7

62.2

Farm profits

29.7

10.7

91.8

28.7

Property income

102.1

37.0

29.0

9.1

From realty

39.0

14.1

16.5

5.2

From personalty

63.1

22.8

12.5

3.9

Total income

276.3

—

319.4

—

Income per capita (1840 dollars)

234.7

—

158.5

—

South Atlantic

East South Central

West South Central

Table 6-3 (cont.) (Millions of current dollars) 1860 free

(%)

1870 white

(%)

Labor income

712.6

49.6

1,030.8

62.7

Farm profits

188.7

13.1

409.6

24.9

Property income

534.8

37.2

204.6

12.4

From realty

169.6

11.8

121.8

7.4

From personalty

365.2

25.4

82.8

5.0

Total income

1,436.1

—

1,645.0

—

Income per capita (1840 dollars)

186.8

—

131.5

—

All South

as the change in the gap during the 1774–1800 era: namely, 20.8 points (table 4-4).22 It appears that relative to the North, the South suffered equal amounts during the two conflicts.

INCOME AND EARNINGS INEQUALITY IN THE NORTH AND ECONOMY-WIDE The Civil War decade brought a reduction in income inequality among southerners, even among white southerners, as we have just seen. By contrast, and consistent with the rising inequality thesis championed by the Beards and Hacker, income gaps continued their long march upward in the North. Table 6-4 offers the 1870 evidence to be compared with the 1860 evidence in table 5-7, revealing steep rises in inequality in each northern region. Gini coefficients rose in New England from 0.457 to 0.519, the Middle Atlantic from 0.488 to 0.519, the East North Central from 0.399 to 0.466, and the West North Central 0.419 to 0.478. These are big increases for just ten years—bigger than the average increases 22 The 1774–1800 change is calculated using 1860 and 1870 population weights applied to the original thirteen New England and the Middle Atlantic per capita incomes to get the North versus the South Atlantic per capita income gap. Between 1774 and 1800, we are measuring a loss in the southern per capita income lead over the North, while between 1860 and 1870 we are measuring an increase in the northern lead.

THE CIVIL WAR

•

153

0.519 0.519 0.466 0.478 0.530 0.487 0.479 0.449 0.498 0.511

New England

Middle Atlantic

East North Central

West North Central

South Atlantic

East South Central

West South Central

Mountain

Pacific

All United States

A. All households

Gini coefficient Top 20%

Next 40%

Bottom 40%

9.8

12.2

6.8

7.5

8.5

8.5

9.7

9.1

9.2

10.4

27.2

28.5

21.0

21.6

23.9

24.9

28.2

25.9

25.8

27.8

39.3

40.5

33.0

32.9

35.2

37.0

40.1

37.0

39.8

40.7

55.1

55.1

50.3

50.0

52.0

55.1

53.8

52.1

56.2

56.3

33.7

32.4

35.4

39.4

36.6

36.1

32.3

34.0

33.2

32.7

11.1

12.5

14.3

10.6

11.4

8.7

13.9

13.9

10.7

11.0

1,133

1,819

1,086

981

884

724

1,121

1,084

1,427

1,369

Mean income

Top 10%

Top 1%

Top 5%

Household

Percent shares of total income

Table 6-4 The Inequality of American Household Incomes, 1870

744

1,172

718

725

611

465

757

743

884

862

Median income

0.519 0.517 0.465 0.475 0.483 0.462 0.437 0.446 0.493 0.496

New England

Middle Atlantic

East North Central

West North Central

South Atlantic

East South Central

West South Central

Mountain

Pacific

All United States

B. White households only

9.4

12.6

7.0

7.0

7.8

7.5

9.6

9.1

9.2

10.4

26.3

28.7

21.2

19.9

22.7

22.4

28.0

25.9

25.6

27.8

38.5

40.4

33.0

30.7

33.6

34.0

40.0

37.0

39.6

40.7

54.1

54.7

50.1

46.6

49.5

50.3

53.7

52.1

56.0

56.2

33.9

32.4

35.3

40.7

38.1

39.2

32.3

34.0

33.2

32.7

12.0

12.9

14.6

12.6

12.4

10.5

14.1

13.9

10.8

11.1

1,230

1,818

1,067

1,245

1,102

956

1,142

1,093

1,444

1,374

819

1,185

718

1,079

833

720

772

749

894

867

per decade between 1800 and 1860. For example, while the top 10 percent share rose only by 1 to 4 points in the four northern census regions between 1850 (table 5-6) and 1860 (table 5-7), it rose by 4 to 10 points across the war decade. On the relative rise of wealthy northerners, the Beards and Hacker were right. Furthermore, this rise in inequality took place even though the federal government imposed taxes on high incomes, dividends, and interest income. It also imposed business and inheritance taxes. These taxes on profits, property incomes, and the rich were, of course, used to help finance the war, and they disappeared by 1870. But they could have created a disincentive to accumulate in the 1860s, and thus property shares and inequality up North could have been even higher in 1870 without them. The widening advantage of the top income groups in the North was eclipsed by the blow to wealthy southerners, and thus to inequality in the United States as a whole, which rose hardly at all; the US Gini stayed the same at 0.511, and the top 10 percent share increased only from 38 to 39.3 percent (tables 5-7 and 6-4). It appears that the northern income widening plus the additional increase in the North–South income per capita gap were just enough to offset the huge equalization of incomes in the South.23 Since so much of the leveling in the South was due to the disappearance of property income from those at the top, it would not be surprising if free labor earnings inequality rose more steeply than income inequality in the United States as a whole. And rise it did, steeply. The Gini coefficient for free labor earnings increased from 0.454 in 1860 (table 5-9) to 0.546 in 1870 (table 6-5). Moreover, all the components of rising earnings inequality that we listed in chapter 5 for 1800–1860 were still at work within the North during the Civil War decade. First, wage-rate gaps widened, between regions and occupations. Southern wage rates fell sharply relative to those in the North: between 1860 and 1870, the South Atlantic common labor wage fell from 0.81 to 0.55 relative to the North, and it fell in the South Central from 1.03 to 0.69.24 23 Our results for the changes in inequality between 1860 and 1870 are not radically different from those presented by Soltow (1975, 99–103) on the basis of his spin sample from the censuses, as we have mentioned above. It is difficult, however, to compare our apples with his oranges. We are measuring income among households, whereas he was measuring “total estate” wealth among adult males. 24 Margo 2002, 34, table 1, panel B.

156

•

CHAPTER 6

0.516 0.521 0.500 0.511 0.508 0.472 0.464 0.481 0.479 0.546

Middle Atlantic

East North Central

West North Central

South Atlantic

East South Central

West South Central

Mountain

Pacific

All United States

Gini coefficient

New England

A. All households

Region Top 20%

Next 40%

Bottom 40%

9.7

9.4

7.0

9.1

10.1

9.7

10.3

9.2

8.0

8.4

31.1

26.0

21.4

26.2

27.8

28.5

32.0

29.4

26.1

28.0

44.5

39.2

34.2

37.4

38.6

40.2

45.3

41.4

41.3

41.3

60.2

54.3

52.5

52.3

52.8

55.4

58.5

56.6

57.4

56.7

28.8

31.9

35.3

33.1

32.9

32.5

27.7

29.8

31.5

31.8

11.1

13.9

12.3

14.6

14.3

12.1

13.8

13.6

11.1

11.5

791

1,366

962

651

585

492

838

768

1,096

1,040

(continued)

420

832

553

431

390

306

483

437

662

662

Median

Mean

Top 10%

Top 1%

Top 5%

Labor earnings

Percent shares of labor earnings

Table 6-5 The Inequality of American Household Labor Earnings, 1870

0.516 0.520 0.501 0.512 0.512 0.494 0.482 0.479 0.468 0.544

Middle Atlantic

East North Central

West North Central

South Atlantic

East South Central

West South Central

Mountain

Pacific

All United States

Gini coefficient

New England

B. White households only

Region

Table 6-5 (cont.)

Top 20%

Next 40%

Bottom 40%

9.4

9.5

7.2

8.8

9.7

8.7

10.2

9.2

8.0

8.4

30.4

25.6

21.5

26.8

28.9

28.3

31.9

29.5

25.9

28.0

44.1

38.5

34.3

38.9

40.7

40.5

45.5

41.5

41.2

41.2

60.0

53.3

52.4

54.3

55.1

56.2

58.6

56.7

57.4

56.7

28.9

32.3

35.1

31.9

31.6

32.0

27.6

29.8

31.5

31.8

11.0

14.4

12.5

13.9

13.3

11.8

13.7

13.5

11.1

11.5

849

1,320

938

775

682

608

852

772

1,107

1,043

453

817

553

489

440

384

487

439

672

662

Median

Mean

Top 10%

Top 1%

Top 5%

Labor earnings

Percent shares of labor earnings

Within the Northeast, urban–rural wage gaps widened from 1.29 to 1.46 (male common labor), while the urbanization rate soared from 35.7 to 44.3 percent. Thus, wage data suggest that the Kuznets urban forces were adding to inequality in the North. The earlier literature’s comments on “a limited supply of skilled labor” during the Civil War decade are certainly consistent with the continued rise in the skilled wage and white-collar premiums shown by our data.25 The impressions just given from wage rate data are reinforced by the movements in the distribution of overall labor earnings.26 The earnings premium captured by the higher-skilled earners, such as the top 10 percent of earners, was slightly higher in 1870 than in 1860, and higher in the Northeast than in the North Central region. The main geographic change is that the earnings premiums of the higher paid converged across regions. That is, the North Central and East South Central, relative to other regions, were no longer as compressed in their earnings distributions as they were in 1860. Overall, it appears that the forces driving inequality among whites, both within the North and nationwide, were even stronger in the Civil War decade than over the six decades between 1800 and 1860. Without the powerful southern redistribution effects of the Civil War and emancipation, American inequality would have continued its upward march started early in the nineteenth century.

CIVIL WAR BENEFITS: EMANCIPATION The ambivalent economic status of the freed slaves has been aptly summarized by the phrase “one kind of freedom,” the famous title of a book by Roger Ransom and Sutch. In terms of their income prospects, some changes in the postbellum era should have given them strong income gains, and others would have reduced them. On the positive side, freed slaves could now receive something like 100 percent of the marginal product of their labor instead of the 50 percent that seems to have prevailed in the 1850s. True, as free farmers they had to share the value of what was produced with the landowner; 25 26

Williamson 1974, 647. Again, compare the 1870 results of table 6-5 with the 1860 results of table 5-8.

THE CIVIL WAR

•

159

a sharecropper typically got 50 percent. Yet it was now 50 percent of the value of the total output, not just 50 percent of the fraction of output that would be paid for labor in a free market. Had this been the only change in their fortunes, their incomes would have jumped from their meager rations under slavery even more.27 On the negative side, black incomes were held back by the depression that pervaded the entire postwar southern economy. For three-quarters of a century, the United States contained its own less developed country, and black Americans had the misfortune of being trapped in it until the World Wars. Black incomes were also reduced to the extent that emancipated slaves now exercised their right to drop out of the labor force, as did many children, elderly, and sick. Naturally, freed blacks worked less than slaves. As Ransom and Sutch have emphasized, “The reduction in the black labor supply did not go unnoticed. Throughout the South there were frequent complaints from planters of a ‘labor shortage.’ Most contemporaries estimated that the decline was even greater than our own conjecture.”28 Ransom and Sutch conjectured the following ranges of annual workdays by age-sex group in the rural South (the figures in parentheses are midpoint averages):29 Males 16+ Females, 16+ Children, 10–15

242–76 (259) 235–58 (247) 199–210 (205)

These rural, southern rates are below the “part-time” rates we have assumed for free family members in all regions: namely, 280 days for households with the head employed in construction trades, rural unskilled workers, and farm operators; and 222 days for households headed by free, urban unskilled laborers and zero-wealth HHs of 27 Th ree other changes would have raised the incomes of newly freed slaves a great deal more, but they were slow to arrive. First, blacks should have been free to get a rudimentary education, yet the schooling gains from the Reconstruction era were checked in later Jim Crow decades. So too was their freedom to migrate to higher-paying sectors and regions. And finally, the freedom to accumulate wealth could do little to raise income for several decades, starting, as blacks did, from zero wealth. Even the already free African American households in relatively prosperous 1860 had average wealth of less than 10 percent of the average white wealth. See Bodenhorn 2015, 144– 66. 28 Ransom and Sutch 1977, 46. 29 Ransom and Sutch 1977, 233, table C.1.

160

•

CHAPTER 6

unknown occupation. Recall that we have assumed that adult slaves worked 313 days per year. These figures imply that emancipated male slaves withdrew 17.3 percent of their labor from the market, and females 21.1 percent, for an average of 19.2 percent.30 Since Ransom and Sutch published One Kind of Freedom, the IPUMS census samples from 1850 onward have now developed a systematic way of coding whether or not each free individual was in the labor force. We have used the IPUMS labor force numbers to document the striking changes in labor force participation between 1860 and 1870. Table 6-6 unveils our updated perspectives, comparing the work rate of the emancipated nonwhite population with the work rates of whites and those nonwhites who were already free in 1860. The results speak more loudly about the effects of slavery and poverty than they do about any inherent racial propensities for work. Fully 61.5 percent of all slaves were in the labor force in 1860, consistent with our assumption that the slave labor force tended to equal the slave population over the age of ten.31 When emancipated, freed slaves stayed in the labor force much more like other free persons. The share of nonwhites at work fell from 61.5 percent under slavery, to 46.3 percent in the South and 52 percent in the non-South, averaging 46.9 percent for nonwhites nationwide in 1870. This implies a reduction in the labor participation rate of 15.2 percentage points. Nevertheless, blacks still worked more than whites did in 1870; 31.8 percent of the white population worked, though that percentage was a bit lower in the South and higher elsewhere. That free blacks continued to work more than free whites was probably due to their being poorer. How did these positive and negative influences on black incomes balance out? We can now offer some informed conjectures. Processing the 1870 data has not been easy, either for previous scholars or us. Racespecific incomes have proven exceedingly hard to document for any time before 1940, though others have worked the archives diligently to harvest good clues. To make comprehensive estimates for ex-slaves, we 30

The total withdrawal of black labor is bigger when children and the elderly are included. As we note in appendix C, this does not assume that the LFPR was 100 percent for those over age ten, and 0 percent for those younger. Rather, it assumes that the number of slaves over the age of ten who were not working, due to ill health or any other reason, was offset by the number younger than ten years who did work. 31

THE CIVIL WAR

•

161

Table 6- 6 Labor Force as a Percentage of Population, by Race and Region, 1860–1870 United States

South

Non-South

1860

1870

1860

1870

1860

1870

Whites

30.3

31.8

24.0

28.2

32.5

32.9

Nonwhites, total

59.0

46.9

59.2

46.3

55.7

52.0

Nonwhites, free only

35.4

46.9

21.9

46.3

52.7

52.0

Nonwhites, slaves

61.5

—

61.5

—

60.3

—

need information on their wealth, occupations, wage or salary earnings, and farm operator incomes, all by region. Fortunately, information on wealth and occupation are available in the 1870 census. Wealth in 1870 was essentially zero for most black households and low for the rest. We know their occupations in each census, and others have rightly used the occupational mix as the single-best indicator of black income disadvantage between emancipation and 1940. One factor that should have contributed to black income gains by 1870 was the fact that blacks got paid nearly what whites got paid within the same (low-paying) occupations. A dozen years ago, Margo surveyed this literature, all of which seems to support the earlier view of only modest racial discrimination in wage rates for farmhands around 1870.32 Margo himself reports a black–white wage ratio of 0.97 based on the 1880 Ransom and Sutch sample for both the South Atlantic and South Central.33 He also believes the same was true for common labor, but notes that the ratio may have been somewhat lower for skilled labor (as conjectured by Robert Higgs). These southern black–white wage ratios are pretty much what we have assumed for 1870. At the 184 farms in the Ransom and Sutch sample that hired both black and white farmhands, the former were paid 96 percent of what the latter got. In 1875, black teachers got paid 93 percent of the white teacher pay in the

32 Margo 2002, 16–18. See, for example, Higgs 1972, 1975; G. Wright 1986, 68; G. Wright 2013, 39–51. 33 Margo 2002, 18, 39, table 4.

162

•

CHAPTER 6

South.34 Other testimony also seems to fall into this 93 to 96 percent range for given occupations, both skilled and menial. We have applied these percentages to the all-races wage and salary rates reported by Lebergott, the Almanacs, and our other sources, thereby to estimate just how much less a black worker received, and how much more a white worker received, relative to the all-race average for each occupation.35 Our information requirements are more challenging for black farm operators, most of whom were southern tenants. The 1870 agricultural census is not adequate to allow a replication of our use of Lee Craig’s technique used in chapter 5 for estimating 1860 own-labor income for the South, inclusive of farm profits, by place and race. The best resource available for estimating southern farm operator incomes is the expanded Ransom and Sutch sample of southern farms in 1879–1880.36 Invoking some assumptions, we were able to use that sample to estimate the relationships of farm operator incomes by race to the local farm wage rates—relationships that we then applied to farm operators in the IPUMS sample for 1870.37 For southern blacks as a whole, our results in table 6-7 suggest big real income gains from emancipation, in addition to the value of leisure time and the immense human value of freedom itself. Their real income gains might have been something like 30 percent, despite the 21 percent decline in their rates of labor force participation. As we have noted, emancipation meant that free blacks captured something much closer to the marginal product of their labor rather than the 50 percent of it that slavery had let them consume in 1850 or 1860. And given the decline in southern white incomes, the ratio of black to white income per capita in the South roughly doubled from 1860 to 1870 (from 0.228 to 0.435; see table 6-7). In the North, nonwhites continued to have about half the average incomes of whites, as free nonwhites had done before the Civil War. What remains to be charted in the history of black 34 B. Smith 1984, 690. Smith also reports parity for teacher pay rates in 1880, and then much lower black rates in the Jim Crow depths of the 1890s. For 1890, however, Margo (1990, 26) finds an average ratio of 95 percent. 35 More generally, see our review of the evidence on black–white wage ratios in appendix F. Our calculations of these implicit black and white rates had to allow, of course, for the employment shares of blacks and whites in each occupation-sex-place group. 36 Ransom and Sutch 2001. 37 The complex estimation procedure and its vulnerability to error are described in appendix G.

THE CIVIL WAR

•

163

Table 6-7 The Progress of Black Incomes, 1860–1870 Real incomes per capita, in 1840 $ Blacks

Whites

Ratio, blacks–whites

1860

1870

1860

1870

1860

1870

New England

114.0

146.1

179.0

204.0

0.637

0.716

Middle Atlantic

95.5

93.4

185.8

200.6

0.514

0.466

East North Central

87.6

86.9

137.4

145.6

0.638

0.596

West North Central

42.6

75.9

144.4

151.9

0.295

0.499

75.0

89.4

163.2

173.9

0.459

0.514

South Atlantic

35.4

46.3

184.0

121.4

0.192

0.381

East South Central

46.3

64.3

180.4

134.4

0.257

0.478

West South Central

63.5

78.4

234.7

158.5

0.271

0.495

South

43.5

57.6

190.6

132.4

0.228

0.435

Mountain

233.5

137.1

237.8

186.3

0.982

0.736

Pacific

457.7

230.5

526.9

300.3

0.869

0.768

444.7

197.5

440.0

262.7

1.011

0.752

46.5

60.8

176.0

166.5

0.264

0.365

North

West All United States

incomes is the extent to which their incomes retreated again after 1870 with the suppression of their rights in the long Jim Crow era, but also their mid-twentieth century gains from the Great Migration and the civil rights movement. The next chapter places all this in context, offering a new quantitative history of African American relative incomes from the late eighteenth century to the early twenty-first.

THE CIVIL WAR DECADE TRADE-OFF The Civil War decade, like the Revolutionary War before it, had huge and conflicting impacts. Most obviously, people perished; more Americans died in the Civil War than in all of America’s other wars put 164

•

CHAPTER 6

together. Our new income estimates bring other major impacts into sharper focus. First, the Civil War and emancipation produced America’s greatest redistribution of wealth and income, taking much more from southern whites than did the revolutionary seizure of Loyalist assets. Second, the great rise in inequality from 1800 to 1860 continued in the North during the Civil War decade. Third, the Civil War cost America enough income to erase its purchasing power income advantage over the British. For the nation as a whole, the loss in average income may not have matched the Revolutionary War loss in percentage terms, but it was large enough to allow Britain to resume the lead. Fourth, this decade completed the South’s two-century reversal of fortune. That reversal accelerated during the Civil War decade, as it had during the Revolutionary War and early federalist years, and finished converting the nation’s richest region into its poorest. Finally, both upheavals delivered a new kind of freedom. The Revolutionary War gave white Americans their freedom from Britain, and the Civil War and emancipation gave African Americans the freedom they had been denied by the founding fathers. The two conflicts differed in the net benefits delivered to the victors. The Revolutionary War brought a significant economic loss to those gaining their freedom. The Civil War, by contrast, raised incomes of those gaining their freedom by about 30 percent. We turn next to the four decades that followed, so as to see whether the economic impact of the Civil War was reversed or reinforced.

THE CIVIL WAR

•

165

CHAPTER 7

Contending Forces American Incomes across the Late Nineteenth Century

S

cholars are much more confident about the pace and pattern of American economic growth from 1869 to 1913, thanks to the efforts of Kuznets and his followers. Moses Abramovitz, Maddison, and others have shown that American GDP per capita grew fast enough after the Civil War disaster to catch up with and overtake western Europe, and thus resume world income per capita leadership by the start of the twentieth century—for the third time in American history. We also know that the United States had become the world’s industrial leader, helped by its rising exploitation of natural resources.1 Less is known, however, about the inequality experience during this episode of mature modern economic growth. Some time ago, we described American inequality as moving along an uneven, high plateau between the Civil War and World War I.2 We can be more precise than that now, and are also better armed to assess the likely components of any aggregate inequality trends into those involving occupation, schooling, race, region, urban density, and income source—that is, labor versus property income. To understand the forces pushing income gaps up and down across the four decades between 1870 and 1913, and how those forces compared with antebellum America and late nineteenth-century western Europe, it is useful to identify what is likely to be exceptional about the late nineteenth-century American experience, and what is not.

1 2

Abramovitz 1986; Maddison 2001; G. Wright 1990. Williamson and Lindert 1980b, 75–92.

Three growth-inequality forces were not unique to America but rather were shared with the western European industrial leaders and perhaps every country that has experienced modern economic growth. First, urbanization continued, which implies more inequality pressure, in line with the Kuznets urban–rural effect we explored in chapter 5. Any slowdown in urbanization would have taken some wind out of American inequality sails and held it down near the 1870 levels. Second, America shared the Second Industrial Revolution with Europe, where the demand for white-collar skills ran ahead of supply and the blue-collar artisan was hollowed out of the middle of the earnings distribution—implying rising skill premiums and greater earnings inequality. Was the artisan de-skilling and rising white-collar scarcity any more powerful in America between 1870 and 1913 than it was between 1800 and 1860, or more than in Europe between 1870 and 1913? Third, as we will see, the property income (and wealth) share rose, and property income became more unequally distributed. What forces were exceptional about the late nineteenth-century American inequality experience, and did they offset or reinforce these modern economic growth forces shared by all the leading countries? It seems to us that four exceptional forces might have been at work. First, European immigrants poured into the lower-skill urban occupations, presumably pushing up skill premiums and raising earnings inequality in the cities. Europe experienced the opposite. These mass migrations should have made inequality trends steeper in the United States than in western Europe. However, there might have been another immigrant force at work, one that could have muted the within-city immigrant-inequality effect. In the antebellum decades, the rapid growth in urban labor demand had widened the urban–rural wage gap, in spite of migration off American farms. But the immigration of Europeans into American cities was still fairly modest. In the late nineteenth century, by contrast, the rising tide of European unskilled immigrant labor into American cities might have been large enough to help erode the urban–rural earnings gap for unskilled workers.3 3 These independent reasons included a glut of young adults created by a European demographic transition, the fi rst immigrants paying for the steerage of the next wave, and improved jobs at home, creating more family resources to invest in more emigrant moves. See Hatton and Williamson 2005.

CONTENDING FORCES, 1870–1914

•

167

Second, if Jim Crow racism eroded some of the emancipation gains in the South, it would have made for steeper inequality trends in the United States compared with western Europe. If the opposite were the case—that is, if Jim Crow effects were weaker than has been thought, and if the early Great Migration in the decade or so before World War I offset even that—then black economic progress relative to white would imply less steep inequality trends in the United States than in western Europe. In principle, there should have been a lot of scope for black American catch-up by their investment in schooling and skills. But did southern racism completely eliminate those catch-up possibilities? Third, did wealth accumulation speed up in the South, as rich property owners tried to reestablish their high, prewar wealth–income ratios? If so, did this serve to push southern inequality back up toward some high equilibrium? Such forces would also have made inequality trends steeper in the United States. Finally, we have the frontier and regional income gaps to consider. In previous chapters, we have learned a lot about the leveling effects of the frontier from 1650 to 1860. These effects were weaker in the antebellum decades than they had been in colonial times. When they completely disappeared in 1890, what difference did it make? Furthermore, was there any regional convergence in per capita incomes from 1870 to 1913? One might have predicted as much for North–South gaps, since wartime losers often experience growth miracles during the subsequent peacetime (e.g., Germany, Italy, and Japan after World War II). If so, it would have served to mute the rise in US inequality. Yet, as we will elaborate in chapter 8, the South did not gain ground on the North from 1870 to 1910, or even up to 1940. Rather, the convergence was between West and East, relating to the frontier. The mountain and Pacific states lost some of their natural resource super-rents after 1870, making their incomes converge on the US average.4 This West–East convergence also must be added to the scale in weighing the likely inequality trends between 1870 and 1910. These forces thus suggest opposing movements in the overall gaps between rich and poor. They seem to predict—on net—that America 4 See Easterlin 1960, 1961; Klein 2013. See also the behavior of the West South Central income series in our figure 8- 4 in the next chapter.

168

•

CHAPTER 7

might have experienced mixed inequality trends between the Civil War and World War I, unlike the steep antebellum rise of inequality. The forces shaping how income is distributed tend, however, to be those central to the process of economic growth. So economists think today, and so thought some historians who famously interpreted the effects of the Civil War. Let us next explore what is now known about the growth of the American economy between 1870 and 1910, to set the stage for the tug-of-war between the influences widening and narrowing income gaps in that era.

MODERN ECONOMIC GROWTH RESUMED, WORLD LEADERSHIP REGAINED The growth performance from 1870 to 1910 is reported in table 7-1 for the United States and eight western European industrial leaders. Only Switzerland grew faster, and by the tiniest of margins, 1.81 versus the US rate of 1.79 percent per annum. The United States grew more than half a percentage point faster than did the western European average, 1.22 percent, and almost twice as fast as the United Kingdom, 0.93 percent. Furthermore, the US post– Civil War performance implies an accelerating rate of growth, from the antebellum rate of 1.43 percent per annum to the post–Civil War rate of 1.79. This acceleration is certainly consistent with the political economy vision of the Beards, Hacker, and others who thought the shift of power to the northern victors generated a more pro-industrial and pro-growth federal policy agenda.5 In Hacker’s words, it “permit[ted] the unchecked advance of industrial capitalism.”6 The acceleration in economic growth was correlated with a shift in political power. In the late nineteenth century, northern, proindustrial interests dominated federal policy, unlike the antebellum tension between these interests and a powerful agrarian southern elite. Whether the correlation between growth and industrialists’ power was causal is another matter, and historians extensively debated that issue

5 6

Beard and Beard 1927, chap. 15; Hacker 1940. Hacker 1940, 250.

CONTENDING FORCES, 1870–1914

•

169

Table 7-1 GDP per Capita Growth, 1870–1910: Western Europe versus United States Real GDP per capita (1990 dollars)

GDP per capita growth

1870

1910

1870–1910 (% per annum)

Austria

1,863

3,290

1.43

Belgium

2,692

4,064

1.04

Denmark

2,003

3,705

1.56

France

1,876

2,965

1.15

Germany

1,839

3,348

1.51

Netherlands

2,757

3,789

0.80

Switzerland

2,102

4,311

1.81

United Kingdom

3,190

4,611

0.93

Western Europe

2,080

3,375

1.22

United States

2,445

4,964

1.79

many decades ago.7 We do know that accumulation rates rose between the 1850s and 1870s: “Not only did the American investment share in GNP rise dramatically (and permanently) between the 1850s and the 1870s, but the relative price of capital goods declined sharply over the period. This relative price change was pronounced and it was never again repeated [up to World War I].”8 It has been argued that much of this was attributable to the way the Civil War was financed by long-term debt and tariffs. The long-term debt impact may have petered out by the 1880s, but the impact of those high tariffs did not, and they pushed the relative price of capital goods downward, thus stimulating accumulation.9 7 Th is Beard and Hacker debate was intense during the late 1950s and 1960s (Sharkey 1959; Unger 1964; Salisbury 1962; Vartanian 1964; Scheiber 1965). It has been much quieter since, perhaps because it got deflected into an economic assessment of the Civil War decade itself rather than its long-run effects through institutional change. 8 Williamson 1974, 636. 9 From the 1860s to the turn of the century, the United States had the highest tariffs in the world (Coatsworth and Williamson 2004; Williamson 2006), and they were mainly on fi nished manufactured consumption goods, not on capital goods.

170

•

CHAPTER 7

In any case, US growth from 1870 to World War I was fast enough for America to overtake Britain as a leader in income per capita. On this point, our findings agree with Maddison’s oft-cited estimates. The US catch-up with and eventual overtake of the United Kingdom had two parts: the relatively poor British growth performance during 1870–1910 (0.93 percent per annum per capita)—which some historians call its Victorian and Edwardian “failure”), and the relatively fast US growth performance (which some call its Gilded Age “success”). American industrial production grew even faster than GDP.10 Yet Maddison’s view and ours differ on the earlier differences between the two countries’ average incomes. Maddison contended that the United States did not overtake Britain until the first decade of the twentieth century; according to his estimates, the United States was 8.3 percent behind the United Kingdom in 1900. Yet we have found that America not only was ahead by 1910 but also had held and lost its per capita income lead over Britain twice before. Maddison estimates a much bigger US lag behind Britain in 1870: 23.4 percent.11 By 1910, his fi xed 1990 purchasing power parity comparisons suggest that average incomes in the United States were 8 percent higher than in the United Kingdom and 47.1 percent higher than in western Europe. This is likely to understate the US income advantage since America was still a cheaper place for the average family to live (at least in 1894, according to our figure 3-5), and it had been so since 1632. In any case, whatever estimates we use, America may have lost the lead during the Civil War decade, but gained it back by World War I.

INCOME INEQUALITY FROM CIVIL WAR TO WORLD WAR I Now that we have a picture of how America’s incomes were distributed in 1870, what happened to inequality over the next four decades? Considerable fog surrounds the distribution of income around 1910, just before the income tax came to stay in 1913. The one brave guesstimate we have is that of Willford I. King, who estimated the whole size 10 Between 1870 and 1910, GDP grew at 3.93 percent per annum (Maddison 2010) while industrial production grew at 5.11 percent per annum (J. Davis 2004, 1189, table III). 11 Maddison 2010.

CONTENDING FORCES, 1870–1914

•

171

distribution of income among families in the United States for 1910.12 King’s income distribution was eclectic and mysterious. He wove 1901 worker survey data, 1902 Chicago wages, 1914 tax returns on top incomes, Wisconsin state income tax returns, and other odds and ends into a detailed set of estimates, using methods that were [in King’s own words] “mainly graphical and .  .  . too varied to describe here.”13 Six years later, King had second thoughts while generating new estimates with Mitchell and others.14 Almost a century later, Piketty and Saez augmented King’s original estimates.15 Here are the contrasting results for the top income shares for the year 1910 (showing the percent shares of the top groups of tax-reporting units or “families”):16 Top 10%

Top 5%

Top 1%

King 1915

35.6

27.7

16.7

Mitchell et al. 1921

41.8

31.9

19.0

Piketty-Saez 2003

40.6

30.7

17.8

As the second row shows, the National Bureau of Economic Research team of Mitchell, King, Frederick Macauley, and Oswald Knauth (1921) found considerably greater income concentration around 1910 than King’s 1915 book had revealed. Piketty and Saez also estimate higher top income shares than did King in 1915, not only by raising slightly the absolute incomes at the top, but even more by lowering the estimated national income in the denominator. Where King had estimated $30,529 million of “income among the families,” Piketty and Saez estimate a total income (excluding capital gains) of $27,630.9 million, or 9.5 percent lower.17 12

King 1915. Williamson and Lindert 1980, 90; King 1915, 122. 14 Mitchell et al. 1921. 15 Piketty and Saez 2003. 16 The only way to estimate a Gini for the entire distribution would be to accept the original King (1915) estimate, at least for the bottom 90 percent. We have not attempted this. Moreover, we have not used Lebergott’s (1976, 310–25) income distribution estimates for 1900, for three reasons. First, his incomes seem to be only labor earnings. Second, his family units are synthetic. And third, he only reports his estimates by income range. 17 Piketty and Saez 2003; Piketty 2014. It should be noted that the millennial edition of HSUS (series Ca10) gives $31,609 million for GDP, but it refers to a broader income concept. 13

172

•

CHAPTER 7

Table 7-2 Inequality Trends in America, 1870–1929 Gini

Income shares (%) received by Top 1%

Top 5%

Top 10%

Private wealth/ income

Wealth share held by top 1%

1870

0.511

9.8

27.2

39.3

4.21

—

1890

—

—

—

—

4.62

—

1900

—

—

—

—

4.62

—

1910

—

17.8

30.7

40.6

4.38

—

1913

—

18.0

—

—

4.70

—

1915

—

—

—

—

—

35.6

1920

—

14.5

—

—

3.60

35.2

1929

0.490

18.4

—

—

4.95

40.3

However tentative, what we now know about trends in American inequality between 1870 and 1929 is summarized in table 7-2. The rise in inequality from the Civil War to World War I was modest and mixed, confined mainly to concentration at the top. The overall Gini did not rise; rather, it drifted down from 0.511 in 1870 to 0.490 in 1929. In contrast, the ratio of private wealth to total private incomes rose from 4.21 in 1870, to 4.62 in 1890, and 4.70 in 1913. Piketty argues that such a rise would have been central to inequality trends.18 The top 1 percent share rose from 9.8 to 17.8 percent, while the rise in the top 10 percent share was much less, from 39.3 to 40.6, implying that there was greater inequality among the rich but not among the total population. In short, American inequality during 1870–1910 did not record the same steep, ubiquitous inequality trend that the country experienced up to the Civil War. After 1870, the trend was mixed and modest. What explains the absence of a steep rise in American inequality over those four decades or so up to World War I?

18

Piketty 2014.

CONTENDING FORCES, 1870–1914

•

173

CONTENDING INEQUALITY FORCES: THE CONVENTIONAL, “UNEXCEPTIONAL” FORCES Urban–rural inequality. We confirmed in chapter 5 (table 5-11) that for 1860, American cities had much more unequal incomes than did the countryside. This result certainly made sense, since the cities were where all the high-wage, white-collar and skilled labor was located, well-schooled individuals were best rewarded, and the richest property owners building industrial capital and the railroads lived, and were surrounded by poor, unskilled laborers, many of them recent foreign immigrants, packed into tenements. Since property income shares rose in the North, and the premium on skills and schooling rose there across the Civil War decade, one would have predicted that the urban–rural income gap was even bigger in 1870 than 1860. And so it was for the United States as a whole, although the urban–rural gap stayed about the same (and big) for northern regions. Table 7-3 (panel A) reports that the US urban–rural inequality gap (the difference between the incomebased Ginis) for 1870 was 0.149, and ranged between 0.176 and 0.211 (for an unweighted average of 0.191) in the North. Just a decade earlier (table 5-10 in chapter 5), those figures were 0.105 for the United States, and ranged between 0.141 and 0.246 (for an unweighted average of 0.194) in the North. The changes due to urbanization were smaller in the South and the mountain region. If the countryside had been like the cities, in terms of inequality and average income, the 1870 Gini would have been 0.591 not 0.511. Thus, rapid American urbanization from 1870 to World War I would have increased the total inequality, especially in the North, if other things had been equal. Panel B reports the Kuznets effect using the 1870 urban–rural inequality weights, and the impact is almost twice that of 1800–1860 (+0.030 versus 0.014), mainly because urbanization was almost twice as fast. And the impact is big: the US 1870 Gini of 0.514 would have been raised to 0.544 in 1910 simply due to the Kuznets urbanization effect. Urban–rural income gaps. To the extent that urban skill and schooling premiums and property income shares were on the rise, the income gaps between rural and urban locations are likely to have risen, too. But

174

•

CHAPTER 7

Table 7-3 The Kuznets Urbanization Effect, 1870–1910, All Households A.

1870 Ginis

Urban population share

Total

Urban

Rural

Urban–rural

1870

1910

New England

0.519

0.595

0.416

0.179

0.443

0.731

Middle Atlantic

0.519

0.590

0.414

0.176

0.492

0.712

East North Central

0.466

0.579

0.382

0.197

0.192

0.473

West North Central

0.478

0.604

0.393

0.211

0.217

0.369

South Atlantic

0.530

0.561

0.502

0.059

0.143

0.257

East South Central

0.487

0.616

0.449

0.167

0.088

0.204

West South Central

0.479

0.578

0.442

0.136

0.133

0.223

Mountain

0.449

0.496

0.444

0.052

0.121

0.360

Pacific

0.498

0.596

0.397

0.199

0.320

0.568

United States

0.511

0.591

0.442

0.149

0.255

0.455

B.

1870–1910 d(share)

d(gini)

New England

0.288

0.052

Middle Atlantic

0.220

0.039

East North Central

0.281

0.055

West North Central

0.152

0.032

South Atlantic

0.114

0.007

East South Central

0.116

0.019

West South Central

0.090

0.012

Mountain

0.239

0.012

Pacific United States

0.248 0.200

0.049 0.030

Table 7-4 Wage Gaps and Skill Premiums, 1870–1910 1870

1880

1890

1900

1910

1. Nonfarm versus farm (1910 = 100), a

106

123

123

119

100

2. Nonfarm versus farm (1910 = 100), b

110

123

121

107

100

3. Manufacturing versus farm (1910 = 100)

100

128

117

122

100

4. Clerks (1910 = 100)

—

—

84

—

100

5. Engineers (1880 = 100)

90

100

—

—

—

6. Public school teachers (1910 = 100)

80

85

79

92

100

7. Rural, public school teachers (1890 = 100)

77

101

100

—

—

8. Methodist ministers (1910 = 100)

97

116

116

115

100

9. Artisans (1890 = 100)

102

108

100

—

—

Urban–rural wage ratios

Skill premiums above unskilled

that was not true of wage gaps, at least as proxied by trends in the gaps between farm and unskilled nonfarm wages, or farm and manufacturing operatives’ wages. Table 7-4 cannot document any secular rise in either of these two urban–rural wage gap proxies. While they rose between 1870 and 1890, they fell thereafter, perhaps even showing a decline over the four decades as a whole. In short, there was no rising urban– rural wage gap to add to the Kuznets urbanization inequality forces from the Civil War to World War I. We suspect that the European immigrants flooding northern cities kept the urban–rural wage gap from increasing. Skill premiums. As we have seen, modern economic growth in the young republic pushed up the skill premium dramatically between 1800 and 1860.19 The ratio of urban white-collar to urban common labor rose by 64 percent in the North and 71 percent in the South. The demand for literate, numerate, and schooled labor raced ahead of supply, as city 19

176

See the discussion of table 5-9 in chapter 5.

•

CHAPTER 7

populations expanded, and firms got bigger and more sophisticated. More teachers were needed to cope with the rising demand for schooling, pushing up their incomes. One might expect the same for the decades between 1870 and 1910, and perhaps even more so as urbanization accelerated, as the Second Industrial Revolution raised the rate of manufacturing growth, and shifted toward activities calling for more capital and skills. Is that what we find? Considerable recent research on this issue has been stimulated by today’s concerns that technological change is advancing faster than workers can adapt to it, raising the premium on increasingly scarce skilled and schooled workers, and pushing the rest into low-wage jobs.20 The search for the sources of this phenomenon has been pushed back across the twentieth century and deep into the nineteenth. A recent paper has found evidence that capital–skill complementarity became a powerful force helping push up skill premiums only in the late nineteenth and early twentieth centuries, at least in manufacturing.21 This study was preceded by another that explored manufacturing over the years 1850– 1880, and a more recent one looks at a time span closer to this chapter’s post–Civil War focus, but again mostly in manufacturing.22 White-collar jobs in manufacturing increased as a share of total manufacturing employment from 1870 (4.8 percent) to 1910 (11.9 percent), from 10.6 to 19.7 percent as a share of total economy-wide employment, and even bigger shares for nonfarm and urban employment.23 Those increasing white-collar employment shares could, of course, have been pushed up by an augmented supply (lowering the skill premium) or an augmented demand (raising the skill premium). Which was it? Table 7-4 supplies the answers. The fragments of evidence there suggest economy-wide increases in the skill premium in the nonfarm sector. Artisan hollowing out. At the same time, urbanization and technological forces favored high-wage, white-collar workers, thus stretching the wage structure and fostering earnings inequality in the late 20

Goldin and Katz 2008; Brynjolfsson and McAffee 2011. Lafortune, Tessada, and Lewis 2014. 22 Atack, Bateman, and Margo 2004; Katz and Margo 2013. 23 Katz and Margo 2013, 57, table 4. 21

CONTENDING FORCES, 1870–1914

•

177

nineteenth century. We also reported in chapter 5 that there were hollowing out forces that were reducing artisanal premiums in the middle of the distribution, further augmenting earnings inequality in the cities. Yet we could not find any evidence supporting actual employment displacement of the artisan in the antebellum period. Can we find evidence of more powerful de-skilling forces at work in the post–Civil War decades? Martin Brown and Peter Phillips found plenty of de-skilling in one American industry—canning—across the nineteenth century. Atack, Bateman, and Margo found it more generally for all manufacturing for the decades 1850 to 1880.24 They showed that large manufacturing firms—ones using less artisanal production methods—paid lower average wages, reflecting the use of lower-skilled operatives. Larger firms also required more white-collar clerks, engineers, and professional managers.25 Katz and Margo report the employment figures for manufacturing in 1870–1910. As we saw above, the white-collar share rose by 7.1 percentage points, from 4.8 to 11.9 percent of the manufacturing labor force. Caitlin Rosenthal reports an economy-wide annual growth rate of 1.6 percent in clerical and accounting workers per capita. The skilled blue-collar share fell by 3.3 percentage points, and for the economy as a whole, the blue-collar artisan employment share fell 7 percentage points, from 35.5 percent in 1870 to 28.5 percent in 1910. Thus while it is easy to document artisanal and blue-collar hollowing out in the middle of the workers’ wage ranks, it is harder to fi nd the evidence to document a decline in their relative incomes, at least after 1880.26

24 Brown and Phillips 1986; Atack, Bateman, and Margo 2004. Indeed, Atack, Bateman, and Margo (2004, 186, table 3) found wage inequality rose in manufacturing between 1850 and 1880, with the gap between the top 10 percent of wage earners and the median widening markedly. 25 These authors are making use of what is now called the Goldin and Katz (1995) model where skill-intensity is determined by plant size, capital-intensity, and energy-intensity. In such a model, production operatives are used more intensively on machine-oriented tasks, and average wages are thus lower. 26 Katz and Margo 2013, 57–58, tables 4 and 5; Rosenthal 2012, 170, figure 4.4; E. Anderson 2001, figure I.

178

•

CHAPTER 7

CONTENDING INEQUALITY FORCES: THE MORE “EXCEPTIONAL” FORCES The impact of the unskilled mass immigration: Traditional views.27 The impact of immigration on native-born workers has been under debate ever since the Irish flooded British cities early in the fi rst Industrial Revolution. Over the two centuries since, interest grew when democracy gave labor the vote, which for the United States was already the case in the 1840s, when the first immigrant wave reached its shores. By the late nineteenth century, debate had come to focus on the effects of immigration on wage and employment outcomes for locals, both native born and previous immigrants. We know that they were unskilled relative to the American labor force receiving them, and that they became increasingly so as their source shifted from richer western European regions to poorer regions on the European periphery. We also know that their magnitudes were large, and that they settled mostly in the cities; between 1870 and 1910, the US labor force was augmented by about 37 percent by the mass immigration (see below), and by 1910, male immigrants were 24.3 percent of the adult male population.28 Everyone seems to agree that a mass immigration of this magnitude must have lowered unskilled wages and raised inequality (especially in the cities). However, we can’t seem to agree about the magnitude of its impact. The issue has become an important part of the research agenda on modern immigration, but it occupied economic historians and contemporaries interested in the age of mass migration much earlier.29 When making the historical assessment, it is crucial to control for the state of world capital markets. That is, we need to know whether capital inflows from Europe attenuated the wage, employment, and inequality impact of immigration in America. 27

Th is section draws heavily on Hatton and Williamson (2005, chap. 5; 2008). Goldin and Katz 1995, table 9. These are official US figures, but we have learned recently that they are significantly understated (Bandiera, Rasul, and Viarengo 2013). 29 Friedberg and Hunt 1995; Ottaviano and Peri 2012; US Immigration Commission 1911; Hourwich 1922; Jenks and Lauck 1926; Ferenczi and Willcox 1929. For a recent summary, see Ferrie and Hatton 2013. 28

CONTENDING FORCES, 1870–1914

•

179

So did capital chase after European emigrants in the late nineteenth century? The answer is yes.30 This fact implies that capital flows—helping finance capital deepening and augmenting job creation—softened the downward pressure on unskilled wages and the upward pressure on inequality in late nineteenth-century America. Given this fact, what, then, were the net effects of mass migration on inequality before World War I? Answers are often based on partial equilibrium analysis, but this seems inappropriate for such large migrations that are likely to have had substantial economy-wide effects. Over the last two decades, a number of studies have examined the effects of mass migration using computable general equilibrium (CGE) models.31 The calculations use multi-sector CGE, open economy models based on three factors of production: labor, capital, and land. A key characteristic of these CGE models is that land is assumed to be specific to agriculture and immobile. Furthermore, the model has three sectors: agriculture and manufacturing produce tradable goods (with manufactures imperfectly substitutable in international markets), while services are nontradable. These CGEs are calibrated with US data from the late nineteenth century and then counterfactuals (“what ifs”) are posed to estimate the impact of increases in the US labor force by the observed immigration. The results are these: had there been no US immigration from 1870 to 1910, the 1910 labor force would have been about 27 percent smaller.32 The counterfactual suggests that the US real wage for unskilled labor would have been 34 percent higher than it actually was. These estimated results are big, and their magnitudes depend largely on the assumption that land and capital are held constant at their 1910 levels, and that there are strong diminishing returns to labor. As a result, the US rate of return on capital would have fallen by almost 24 percent in the absence of immigration. 30 Hatton and Williamson 2008, table 1, figure 1. Thus, the late nineteenth-century evidence does not support the conventional Heckscher-Ohlin prediction that capital and labor flow in opposite directions. 31 See, for example, Taylor and Williamson 1997. 32 The counterfactual labor force takes account of the differences in labor participation rates between immigrants (higher) and the locals (lower) that result from the age and sex selectivity of migration. They also take into account the labor force contribution of migrant children. If the children are ignored, the US labor force would have been about 18 percent smaller in the absence of immigration.

180

•

CHAPTER 7

But what if capital had been perfectly mobile between countries? This assumption seems much more plausible than the assumption of world capital immobility made in the previous counterfactual. Suppose, therefore, we hold America’s rate of return constant, allowing capital inflows to mitigate the effects of diminishing returns on wages.33 Capital mobility reduces dramatically the effect of immigration on wages. In the absence of immigration, the US real wage would have been about 9 percent higher (in an economy with much less capital). While they are still substantial, the effects with international capital mobility are much smaller than those without capital mobility. The smaller impact is mainly because land, which was an important factor of production in the US economy before 1914, remains fi xed in the counterfactual.34 It should be added that if these simple models were made more realistic by assuming that immigrants were unskilled (as indeed they were), and that labor consisted of both skilled and unskilled, then the impact of immigration on inequality would be more pronounced. With more data, modern studies have often exploited the so-called spatial correlations approach, which seeks to isolate the effects of immigration by correlating wage changes with immigrant inflows into local areas within the receiving country. Debate about the validity of this method is well rehearsed, so we will stress only one point: if there is a national labor market in which nonimmigrant workers are mobile across localities, the effect of immigration will not be restricted to the cities or regions where immigrants first locate. If, in response to immigrant inflows, local workers move to other regions (or fewer move in than would have otherwise), then the wage and employment effects would be spread across the entire economy, and would not be limited to just the gateway cities. Is there any historical support for the spatial displacement hypothesis? One study estimates that for every hundred immigrants arriving in an East Coast (gateway) state in America during the late nineteenth century, forty locals were displaced to other states.35 An important feature 33 That is, we assume that the United States was a price taker in world capital markets: it could get all the financial capital it wanted at the world interest rate. 34 For the effects on land rents and the wage rental ratio, see O’Rourke, Williamson, and Hatton 1994. For a wider analysis of wage rental ratios in the Atlantic economy, see O’Rourke and Williamson 1999. 35 Hatton and Williamson 2008, tables 4 and 5.

CONTENDING FORCES, 1870–1914

•

181

of US development was its great westward settlement, which accelerated as the late nineteenth century progressed. Most historians see this westward migration as a land-induced pull, while we see it as both pull from the West and an immigration-induced push from the East. If the displacement effect had been one for one, we would observe no relative labor market effects of immigration in gateway cities and states at all— even though there may have been substantial effects at the national level. Thus, we should still observe some imprint of immigration on local wages and inequality. Thus, economic theory and economic history both tell us that unskilled immigration reduced unskilled wages and raised inequality in late nineteenth-century America when mass migration reached a peak. But this thinking may have ignored some historically relevant endogenous forces that might help explain any modest effects of immigration on inequality in the late nineteenth century. The impact of the unskilled mass immigration: Endogenous technology and output mix. There is no doubt that the mass immigration from 1870 to World War I suppressed unskilled wage growth and raised inequality in America’s cities. But recently, some additional reasons have been offered that suggest that the American economy found ways to absorb the immigrants with greater ease than previously thought. For example, recent evidence has shown that American agriculture shifted toward more labor-intensive crops in response to unskilled immigration. Furthermore, more recent evidence has been used to show that manufacturing firms adjusted their technologies to immigration-induced changes in relative labor costs by skill. This idea has now been applied to the late nineteenth-century mass immigration environment, and until late in the period, it appears that capital and unskilled labor were strong complements, and that “immigrant shocks” drove firms to search for technologies that absorbed the foreign unskilled.36 Still, it seems clear that immigrant mass migrations helped make America exceptional, at least compared to Europe. The immigrants were putting upward pressure on American inequality (especially in the cities), while the opposite was true of labor-emigrating Europe. 36 Lafortune, Tessada, and González-Velosa 2013; Lewis 2011, 2013; Lafortune, Tessada, and Lewis 2014.

182

•

CHAPTER 7

A Disappearing Frontier and Regional Convergence The frontier was a leveling force from 1650 to the Civil War, as we have said. Wealth was held more equally there, thanks to government policies giving the less wealthy good access to the new land. True, every region in America was undergoing urbanization in the late nineteenth century, so the Kuznets urban–rural inequality effect raised the aggregate inequality. In this there was nothing exceptional, since the same forces must have been at work in Europe. Yet the frontier served to mute the Kuznets effect, since western settlement implied a greater weight to the more rural, less unequal frontier. We know that these leveling frontier effects were already weaker in antebellum than in colonial times, at least in terms of per capita income growth. But what about the effect on aggregate inequality of filling the frontier between the Sierras and the Mississippi? If we define the frontier as the West North Central, the West South Central, and the mountain census regions, then the frontier’s share in the US population increased from 37.7 to 47 percent over the four decades. Since these regions were so rural (table 7-3), this filling up effect must have moderated the effects on rising inequality in America. This argument only exploits statistics, of course, and the frontier is likely also to have induced leveling effects east of the Mississippi. Turner and his followers have long stressed the role of the frontier option as a force that reduced the downward pressure on wages and the upward pressure on inequality in the settled regions, which sent their young emigrants west.37 In short, while we offer no explicit quantitative assessment of the effect, it seems clear that the American frontier was serving to reduce the power of the modern economic growth inequality raising forces even more between 1870 and 1910 than between 1800 and 1860. Consider another attenuating influence: regional convergence in per capita incomes. Were poor states catching up with rich states from 1870 to 1913? If so, it too would have reduced aggregate inequality forces. One would have predicted an erosion in North–South income gaps— since, to repeat, wartime losers always seem to have postwar growth miracles during peacetime periods—but is there reason to expect more 37

Turner 1921.

CONTENDING FORCES, 1870–1914

•

183

Table 7-5 Personal Income per Capita across Regions, 1880–1910 (US = 100) 1880

1890

1900

1910

Northeast

145

141

140

127

Midwest

85

87

90

94

South

56

58

57

65

West

195

175

154

146

ubiquitous convergence than that? There is an earlier and extensive literature on US regional inequality, and one of the present authors argued some time ago that American experience conformed to a regional Kuznets curve.38 More recently, Robert Barro and Xavier Sala-i-Martín analyzed the regional convergence part of the “curve” for the whole period from 1880 on. They found the following: Poor states tend to grow faster than . . . rich states. The rate of convergence is, however, not rapid: the gap between the typical poor and rich state diminishes at roughly 2 percent a year.39

Neoclassic growth theory predicts the convergence of poor regions upward to the mean and richer regions downward to the mean, and the Barro and Sala-i-Martín result is consistent with it. But 2 percent per year is a bit slow. The convergence of US regions in table 7-5 looks even slower than that estimated econometrically by Barro and Sala-i-Martín. The reason why table 7-5 documents slower convergence than what Barro and Sala-i-Martín estimate is that the period covered in this chapter excludes the acceleration in North–South convergence that started only with the 1930s. Barro and Sala-i-Martín used the Easterlin per capita income estimates before 1920, but since Easterlin’s seminal work, Alexander Klein has prepared new estimates, and we exploit that data in table 7-5. The biggest convergence of a rich region toward the American mean was for the West, but that convergence was also significant for the industrial Northeast, which was 45 percent above the national average in 1880, but 38 39

184

Easterlin 1960, 1961; Borts and Stein 1964; Williamson 1965. Barro and Sala-i-Martín 1991, 108.

•

CHAPTER 7

only 27 percent above it in 1910. Both the poorer Midwest and the evenpoorer South recorded a catch-up rise toward the mean—the former from 15 to 6 percent below the mean, and the latter from 44 to 35 percent below the mean. It should be added that Barro and Sala-i-Martín report that interregional migration from poor to rich states was contributing to the convergence. This, of course, is quite consistent with the Turner hypothesis. And shortly before World War I, the start of the Great Migration from the lower-wage South to the higher-wage border and northern cities was part of it.40 While modest, this regional convergence was another exceptional American force that served to attenuate the rise in US inequality during its Gilded Age compared with Europe and with the steep antebellum rise up to 1860 when the South fell behind. Racism from Emancipation to Jim Crow Another clearly exceptional feature of the American inequality experience from 1870 to 1913 was the Jim Crow regime of restrictions on African American safety, political voice, freedom to migrate, use of public facilities, incomes, and education.41 The spread of Jim Crow laws across the South was supported by pervasive intimidation. In place of the whipping and other punishments typical of slavery, the preferred instruments of intimidation after emancipation were the threats of imprisonment and lynching. A simple and blunt reminder of the enforcement of the new terror for blacks is the number of extralegal and widely publicized lynching events (figure 7-1). Even a small number of lynchings could intimidate a large population for a long time. As the figures meticulously gathered by reporters from that era remind us, the targeting of southern blacks was as obvious as the perpetrators meant it to be. With such fear a constant presence, it became easier to check the economic advance of blacks far into the twentieth century. The rate of 40 Easterlin 1960, 1961; Klein 2013; Barro and Sala-i-Martín 1991, 133–36, 129, figure 7; Collins 1997. 41 For two classic retellings of the story of the Jim Crow suppression of human rights from 1877 onward, see Woodward 1974; Kousser 1974. J. Morgan Kousser emphasizes the destruction of the voting rights that African Americans in the South possessed in the Reconstruction era, 1865–1877.

CONTENDING FORCES, 1870–1914

•

185

Lynchings per year (five-year averages)

200

All victims, USA 150

100

Blacks, USA

50

Blacks in eleven southern states 0 1880

1890

1900

1910

1920

1930

1940

1950

1960

Figure 7-1. Blacks and Other Victims of Lynching in the United States, 1882–1964

lynching then tapered off, due to a mixture of black submission and gradually rising legal prosecution. Did the rise of Jim Crow racism, especially in the South, reverse the income gains that blacks had enjoyed with emancipation, or did it simply check black economic progress? How did the income-suppressing effects of Jim Crow laws compare with the income-increasing forces, including the gains from the start of the Great Migration shortly before World War I? Did southern racism add to the rise of American inequality from 1870 to 1910? Economic historians have pieced together parts of the puzzle of black economic progress from the 1880s to 1950s. The literature on the Jim Crow era is large, and southern racial discrimination has been documented in land, labor, schooling and housing markets.42 Perhaps the largest component of discrimination against black earning power was the separate and unequal provision of public education. William Collins and Marianne Wanamaker have shown persuasively that 66 to 77 percent of the black–white income gap in 1910 is explained by 42 See, for example, Lebergott 1976, 121, table 3; Lebergott 1976, 124–25, tables 4–5; Ransom and Sutch 1977; Higgs 1977, 1982; Margo 1984, 1985, 1990; Collins and Margo 2003.

186

•

CHAPTER 7

discrimination in schooling provision.43 The mechanisms of schooling discrimination are now well understood. The strategy was to provide at least some publicly funded schooling, yet to dole out state and local education money unequally. The black schooling deficit varied over time and across states. It hit bottom sometime after 1890, when black voting rights were starting to be restricted. The most likely bottom date was around 1910, the second column in table 7-6. It was in that heyday of racist policies that black pupils, relative to white ones, endured the lowest public support, most crowded classrooms, and shortest school years. Table 7-6 also shows the gradual improvement in their relative schooling inputs over the next four decades. When southern states are ranked by black school quality in 1910, a pattern stands out clearly. Discrimination in school inputs was greater in the Black Belt of the Deep South. More blacks meant more schooling discrimination because it was in white self-interest, as summarized in 1917 by Thomas Jesse Jones: In practically all the Southern states the State funds are assigned to the counties on the basis of total population without regard to race. In this way a large Negro population is as much of an asset to county school systems as the white population. These funds are then divided between the races by the county board of education and supplemented by such local taxes as the county may decide to vote. The appropriations for Negro schools are therefore almost entirely dependent upon the local sentiment of the white school board.44

The greater the black share of the local population of children, the lower the effective “tax price” of extra education for whites. If there were two black children for each white child, each white child could receive two more dollars of school inputs per tax dollar taken away from each black child. But why did the local white school boards leave any money at all for black education? Why not give them zero dollars, or at least zero dollars from the taxes paid by whites? Margo has offered some good answers: 43 44

Collins and Wanamaker 2013, 3. Quoted in G. Wright 2013, 45.

CONTENDING FORCES, 1870–1914

•

187

Table 7- 6 Racial Differences in School Quality in the South, 1890–1950 ca. 1890

ca. 1910

ca. 1935

ca. 1950

Louisiana

0.50

0.17

0.27

0.62

South Carolina

—

0.19

0.28

0.64

Florida

0.49

0.28

0.41

0.80

Mississippi

0.50

0.28

0.23

0.31

Georgia

—

0.29

0.27

0.68

Alabama

0.99

0.31

0.33

0.76

Arkansas

—

0.42

0.45

0.62

Virginia

0.69

0.42

0.52

0.88

North Carolina

1.01

0.54

0.64

0.93

Maryland

0.65

0.59

0.78

0.95

Texas

—

0.63

0.50

0.83

Tennessee

—

0.67

0.57

0.69

Delaware

—

0.75

1.00

0.87

Mississippi

0.67

0.47

0.72

0.78

Florida

0.65

0.50

0.87

0.92

South Carolina

0.52

0.53

0.75

0.91

Louisiana

0.57

0.60

0.62

0.81

Alabama

0.79

0.69

0.71

0.88

Arkansas

0.93

0.76

0.78

0.83

Virginia

0.69

0.77

0.97

0.90

North Carolina

1.09

0.82

1.06

0.91

Tennessee

1.08

0.90

0.83

0.91

A. Black–white ratio of per pupil spending on instruction (ranked as in 1910)

B. Black–white ratio of teachers per hundred pupils (ranked as in 1910)

Table 7- 6 (cont.) ca. 1890

ca. 1910

ca. 1935

ca. 1950

Louisiana

3

−78

−47

−2

South Carolina

—

−42

−46

−6

Alabama

5

−34

−17

1

Florida

0

−21

−6

0

Virginia

9

−15

−5

0

North Carolina

2

−11

0

0

Mississippi

6

−10

−26

−22

C. Black–white differences in days of school per academic year (ranked as in 1910)

One reason might be that a modest amount of egalitarianism governed the actions of some boards; individual acts of white philanthropy were common. Some whites, according to Freeman, may have benefitted from a better educated black labor force. The courts and the Federal government were relatively ineffective avenues of redress for blacks after 1890 until the late 20s and early 30s, but probably insured at least a minimum level of expenditures on black schools.45

African Americans’ Relative Income Position in the Long Run This chapter’s examination of the late nineteenth and early twentieth centuries offers an appropriate moment to step back and examine how the suppression of African American relative economic progress in this period fits into the whole history of black income disadvantage. The years between emancipation and World War II were ones in which black progress should have been slow. It is also the most poorly documented: there is a long statistical darkness between slavery and, at long last, the reporting of labor earnings by race in the 1940 census. Fortunately, scholars have recently lit some new candles in this darkness.46 We can 45 46

Margo 1985, 89. See also Myrdal 1944; Freeman 1972. In particular, see Smith and Welch 1989; Collins and Wanamaker 2013.

CONTENDING FORCES, 1870–1914

•

189

Table 7-7 Black–White Income Gaps, 1774–2010 Ratios of labor earnings per worker, black–white

Ratios of incomes per capita, black–white

United States

Within non-South

Within South

United States

Within non-South

Within South

1774

0.323

0.525

0.258

0.380

0.530

0.256

1800

0.404

0.424

0.410

—

—

—

1850

0.327

0.319

0.334

0.246

0.474

0.226

1860

0.303

0.450

0.353

0.247

0.415

0.228

1870

0.412

0.558

0.525

0.365

0.513

0.435

1910

0.440

—

—

—

—

—

1930

0.470

—

—

—

—

—

1940

0.433

—

—

—

—

—

1950

0.552

—

—

—

—

—

1960

0.575

0.638

0.498

—

—

—

1970

0.644

0.661

0.567

0.557

—

—

1980

0.726

0.730

0.654

0.584

—

—

1990

0.707

0.741

0.685

0.591

—

—

2000

0.729

0.754

0.719

0.627

—

—

2009

—

—

—

0.647

—

—

2010

0.708

0.718

0.713

—

—

—

combine their black–white ratios of workers’ earnings in the census years 1910 and 1930–1960 with our own estimates of black and white incomes in 1774–1870, updating the series with official data for 1960–2010. The resulting picture of black–white income and earnings ratios for more than 350 years is shown in table 7-7 and figure 7-2. The relative income of African Americans has increased from about one-third for most of the slave era to 40-odd percent around 1870, and 70-odd percent in recent years. Of course, today’s 70-odd percent is not equality. 190

•

CHAPTER 7

1.0

0.8

Relative labor earnings per adult worker

0.6

0.4

0.2 1750

Relative income per capita 1775

1800

1825

1850

1875

1900

1925

1950

1976

2000

Figure 7-2. Black–White Ratios of Earnings per Worker and Income per Capita, 1774–2010

To interpret the movements of these black–white ratios, one must remember that a ratio can be moved by changes in either the numerator (black earnings or incomes) or the denominator (white earnings or incomes). Th is history begins with a downward movement in black relative incomes across the slave era of 1774–1860. The fall, however, is explained mainly by white progress rather than by greater slave exploitation. The treatment of slaves was probably no better in 1774 than later, since the existence of the Atlantic slave trade allowed slaveholders to work their slaves to death and use the profits to buy replacements. Instead, the fall in the black–white ratios between 1774 and 1860 was due to the rise in white incomes, adding to the rise in overall inequality. The first great jump in black relative incomes came with emancipation and Reconstruction. As chapter 6 and table 6-7 have shown, the income ratio nearly doubled in the South over the Civil War decade. Average southern black incomes gained by 32 percent, despite their 22 percent lower labor force participation, while average southern white income fell by 31 percent. The second great jump was much later, between 1940 and 1980. In this case, there was no decline in the denominators, as whites were making their best income progress in American history. Rather, black incomes rose even faster, thanks to two well-known CONTENDING FORCES, 1870–1914

•

191

events to which we will return in the next two chapters: namely, the Great Migration from the low-wage South to the high-wage North starting in World War I, and the civil rights movement in the 1960s and beyond.47 Between these two great leaps came the relative stagnation during the long Jim Crow era. Jim Crow in the South served to hold back blacks catching up on whites. Indeed, the 1870 black–white income ratio was not surpassed until 1950, shortly after World War II. What would have happened to black relative incomes had the country somehow maintained the Reconstruction era policies of black voting rights, less unequal schooling, less racial intimidation, and freedom to migrate? Such counterfactual questions cannot be answered with much confidence. Yet the progress of the black–white income ratios from 1950 to 1975 is something that could have happened almost a century earlier. It seems fair to suggest that renewed racism and segregation postponed progress toward racial income equality for eighty years.

A HIGH PLATEAU OF INEQUALITY WITH INCOME GAINS Most of the forces that this chapter has considered would have widened income gaps, if other things had been equal. Urbanization would have raised inequality. So too would the industrialization of that era, which tended to demand more capital and skills, raising both the return to property and the pay advantage of skilled workers. The rapid influx of less skilled immigrants, something exceptional about America, also served to enhance the income advantages of both capitalists and skilled workers. Other forces failed to contribute to rising income inequality. Between 1870 and 1910, there was no evident widening of the gap in average incomes between races, and thus no contribution to overall inequality on the racial front. The racial repression during the Jim Crow era certainly canceled progress toward racial equality, but there was no widening of the gaps that were already present in 1870. The expected postbellum 47 For some summaries of the 1940–1980 progress in blacks’ relative fortunes, see Smith and Welch 1989; Donohue and Heckman 1991; Collins 1997; Wright 2013; Carruthers and Wanamaker 2014.

192

•

CHAPTER 7

surge in southern wealth accumulation also failed to materialize, and as such, failed to contribute to rising wealth and income inequality. Finally, we have identified one force that contributed to income leveling between 1870 and 1910: the erosion of the western states’ early super-rents on natural resources caused frontier states’ incomes to regress toward the national average. Should all these forces have canceled out? While we still await better data to estimate income inequality trends over these forty years, what we know at the moment suggests that there was probably no clear trend in overall inequality—just a rise in the share of income in the hands of the top 1 percent, in an apparent response to their wealth accumulation. Thus, American income per capita overtook that of the United Kingdom between the Civil War and World War I without any obvious increase in overall inequality. America may have been an unequal country in 1910, but it was no more so than it was in 1870. Rapid economic growth certainly bred inequality before the Civil War, yet it appears that it did not do so between that war and the First World War.

CONTENDING FORCES, 1870–1914

•

193

CHAPTER 8

The Greatest Leveling of All Time

T

oday, scholars and the media see something distinctive about the period from the 1910s to 1970s that was not appreciated a few decades ago.1 We now know that virtually every industrialized country went through a pronounced decline in the share of income captured by those at the top.2 The decline has no parallel in the history of the nowindustrialized countries, and certainly not the United States. There was no inequality drop of a similar magnitude and duration before 1910, and none since 1970—either before or after adjusting “pre-fisc” incomes for taxes paid and transfers received. The Great Leveling wasn’t the only thing special about the era from the 1910s to the 1970s. While inequality plummeted, average incomes rose at a remarkable rate. Despite two world wars and the Great Depression, real income per capita tripled throughout Europe and North America, and it increased sevenfold in Japan.3 True, the rich got richer, but the percentage gains of the middle and lower classes eclipsed those of the rich. The World Top Incomes Database’s figures imply that the real income per family of the top 1 percent rose by 21.5 percent in the 1 If we were forced to pick specific years as turning points, the leveling era would best be dated from 1913 to 1973—that is, from the eve of World War I to the eve of the oil price shock induced by the Organization of Petroleum Exporting Countries. Yet the data often come at census dates, and the exact years of trend reversal differ by country. Thus, the text will refer to this era variously as 1910–1970, the 1910s to 1970s, or even 1913–1973. 2 For a summary of fi ndings, see Atkinson, Piketty, and Saez 2011. For an updated summary of historical trends, see Roine and Waldenström 2014. For the latest, downloadable data, see World Top Incomes Database, http://topincomes.g-mond.parisschoolofeconomics.eu/ (accessed August 16, 2015). 3 These income per capita ratios refer to 1973 over 1913, as posted on Maddison’s website. While the numbers may be subject to errors, even large errors would not erase these gains in per capita income.

United States, while average US family incomes grew by 180 percent, implying that the average real family income for the bottom 99 percent must have risen 214.3 percent—that is, it more than tripled. Will the bottom 99 percent ever have such good fortune again? The interpretative stakes are high. Understanding the causes of this combined leveling and strong growth would inform today’s policy debate. Consider three possible outcomes of an investigation of the Great Leveling era, and what each would imply: • The leveling was due to things we could control in the future. If we find that the leveling and growth were due largely to policies that could be replicated in the future, then the leveling era would offer a prescription. Did societies augment both growth and equality by following progressive policies, such as easing the lending constraints on the poor, raising the support for education, and improving the health of the labor force with preventive and outpatient services? Did more progressive fiscal policies help to foster all this?4 • The leveling was due to things we can forecast but not control. If instead, we find that the leveling along with the growth that accompanied it were caused by something understandable but beyond policy control, we can at least explore the likelihood that such conditions will reappear in the future. • The leveling was due to things we can neither forecast nor control. This is the least cheery prospect, although at least it allows us to doubt theories that view rising inequality as inevitable.

Our first step toward explaining the Great Leveling involves the use of an important global fact: while virtually all industrialized countries shared in this leveling, the magnitudes and timing differed. Thus, comparing the American experience with that of other countries is mandatory. We also need to start with the top of the income distribution, where new data have recently become available—data that have produced a new interpretation. Anthony Atkinson, Thomas Piketty, and Emmanuel Saez led a team that documented trends in the shares of income going to the top 10, 5, 1, or even 0.1 percent over the past century. We 4 In this book, “more progressive” policies are defined as those that transfer a greater share of rich people’s incomes to those with lower incomes. “More regressive” means taking a greater income share from the poor and giving it to the rich.

THE GREATEST LEVELING OF ALL TIME

•

195

will review their findings shortly. First, however, we must widen our view of the inequality movements themselves. What happened to inequality during the Great Leveling was much broader than just a decline in the top income shares. Inequality diminished even within the middle and lower ranks. And the Great Leveling was not just a manifestation of government’s redistribution from rich to poor with taxes and transfers. Incomes became more equal both before and after those taxes and transfers. All of these dimensions need to be explored before we can come up with causal explanations of the Great Leveling. We begin by reviewing what is now known about the movements in the income distribution from the 1910s to 1970s. To repeat, the equalization of incomes, like the forces that explain it, was much broader than past writings have acknowledged.

SHRINKING TOP INCOME SHARES By 1910, America’s top 1 percent income share had risen to about the same high levels as in western Europe and Japan. There followed a revolutionary fall in that share over the next half century or so, unlike anything experienced in any other documented period in history.5 The fall happened in every industrialized country for which we have data. Figure 8-1 underlines the pervasiveness of this drop. In each country shown there, the top 1 percent claimed 18 to 22 percent of the income around 1910, but only 8 to 9 percent around 1970. The World Top Incomes Database has also revealed that the twentieth-century swings in the top 1 percent share roughly match the swings in the top 10 percent share. This means that those in the neartop group, the ninetieth to ninety-eighth percentiles, did not experience much, if any, decline in their income shares. The top 1 percent dominated top share inequality trends. The top 1 percent shares plotted in figure 8-1 document a fall even before taking account of taxes and transfers. The rise of progressive redistribution, especially in Europe, brought an additional leveling to 5 For the Atkinson, Piketty, and Saez database on top income shares, see http://topincomes .g-mond.parisschoolofeconomics.eu/ (accessed August 17, 2015). For a summary and interpretation, see Piketty 2014, especially 274–76, 284–86, 355–56, 372–75.

196

•

CHAPTER 8

30

UK

20

Denmark

Percent

Japan

Canada, Australia, New Zealand

10

USA Five continentals (Denmark, France, Netherlands, Norway, Sweden) 0 1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

Figure 8-1. Shares of National Income Received by the Top 1 Percent, United States and Elsewhere, 1870–1970

disposable income over those fift y or sixty years. This is clear enough from the rise of both social spending as a share of GDP and progressive taxes to pay for a large part of them. The rise was not small. No country in the world spent more than 1.8 percent of its GDP on social transfers in 1910, while by 1970 the median share among the nineteen Organisation for Economic Co-operation and Development (OECD) countries was 14.8 percent, and the highest was 22.5 percent (the Netherlands). These funds were directed at those with less earning power, whereas the taxes paying for them came disproportionally from high incomes.6 Hence, the rise in progressive social spending strongly suggests that the equalization of post-fisc incomes, after taking account of those rising taxes and transfers, was even greater than the equalization of pre-fisc incomes shown here. 6 Social transfers here refer to public noncontributory pensions, public health care, unemployment compensation, family assistance (“welfare”), and public housing. They exclude government expenditures on education. For these defi nitions, see Lindert 2004, 1:12, table 1.2. To download Lindert’s data and estimates, see http://lindert.econ.ucdavis.edu/ (accessed August 18, 2015). For documentation of the fiscal progressivity of different kinds of social spending over the more recent period 1979–2004, see Immervoll and Richardson 2010, especially 34, 42.

THE GREATEST LEVELING OF ALL TIME

•

197

While income gaps narrowed in all industrialized countries, the timing and magnitudes varied, as did the proximate political and military causes. For some countries, it was mostly a matter of sharp inequality reductions during World War II. Such was certainly the case for Japan. As part of its war mobilization plan from 1937 on, Japan’s military government implemented measures that slashed the wealth of large landowners, and hiked the tax rates on individual and corporate incomes. Further confiscation followed the wartime defeat and devastation, and a foreign occupation bent on busting up Japan’s large concentrations of wealth (zaibatsu), with more progressive taxation and further land reform.7 Little wonder that the drop in Japan’s top 1 percent share was concentrated in the period 1937–1950. Perhaps the best analogue in the long American history we survey in this book would be the emancipation of slaves and the defeat of the Confederacy. In both the Japanese and Confederate crises, a polity that had been slow to liberalize had much of its top wealth suddenly confiscated and redistributed to those in the bottom 99 percent.

WHAT HAPPENED WITHIN THE LOWER 99 PERCENT? What was happening to inequality within the lower 99 percent? Anticipating this question, Jesper Roine and Daniel Waldenström have explored the post-1960 correlation between trends in the top income shares and that of other measures of income inequality, and it is significantly positive. Surveying international data for recent years from the Luxembourg Income Study, they find that the top 1 percent share has a 0.62 correlation with the overall Gini coefficient, and a correlation of 0.59 with the ratio of the ninetieth- to the tenth-percentile income. These correlations are hardly perfect, and there are some variations in the movements further down the ranks that require separate attention.8 Absent systematic data before 1960, tracing movements within the lower 90 percent is more difficult. One reason why the World Top Incomes Database chose to target the top 10 percent is that the tax-based data on the top incomes were more readily available before 1960 than were data distributing wages and incomes among the lower 90 percent. 7 8

198

See, along with the sources cited there, Moriguchi and Saez 2008. Roine and Waldenström 2014, table 4.

•

CHAPTER 8

4.0

Ratio

3.5

CPS series

Goldsmith– OBE series

The (middle / lower) income ratio

3.0

2.5

Wage ratios

Wage ratio, top decile /average (Piketty)

2.0

Wage gap 50/10

Wage gap 90/50 1.5 1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

2010

Figure 8-2. Inequalities among Non-Elite Incomes and Wages, United States, 1910–2010

Fortunately, many studies have documented the US experience with wage convergence during the Great Leveling. Their findings are summarized in figure 8-2. The United States was one of several countries where low-skilled wages gained ground on high-skilled wages during World War II. Th is sharp gain was not just due to wartime wage controls since American wages remained compressed for a whole generation after the wage controls were removed. Something more fundamental must have been at work. Goldin and Katz have documented that both the upper (fiftieth to ninetieth percentile) and lower (tenth to fiftieth percentile) gaps were already narrower in 1940 than in 1890, at least for male workers in several branches of manufacturing.9 There are some European data suggesting similar movements in the upper 9

Goldin and Katz 2008, 61– 62.

THE GREATEST LEVELING OF ALL TIME

•

199

and lower earnings groups.10 But apart from the United States, the paucity of data makes it impossible to get a clear picture of trends in labor earnings gaps before 1960. Occupational Wage Gaps More abundant than data on annual earnings by percentile ranks are occupational wage and salary rates, such as the ratio of the average pay for a skilled engineer to the average pay for a manual laborer.11 The United States shows an obvious erosion in the pay advantage of the more skilled relative to the less skilled.12 A striking example of movements in occupational pay ratios is revealed by figure 8-3’s pay history for a bundle of job categories in the highly paid finance sector between 1909 and 2006.13 What happened to rewards in these skilled occupations during the Great Leveling reflects the state of the world financial sector as international fi nancial flows dried up, security markets collapsed, and governments intervened with controls and regulations. The pay of these skilled financial agents shared the same tremendous drop in relative fortunes experienced by other skilled groups across the Great Leveling, from the stock market crash of 1929 up to the 1950s. They have also enjoyed an equally spectacular surge in their relative incomes since the 1970s—something we will explore in the next chapter. The fortunes of those in financial occupations followed those of the top 1 percent exactly. Their fortunes also followed 10 For a few observations on pre-1960 wage rates, see Atkinson 2008; Atkinson and Morelli 2014. Pre-1960 estimates of incomes indicate a decline in the middle-to-lower-income ratio for a couple of European countries. Thus, the ratio clearly declined in Denmark from 1939 to anytime in the period 1950–1963, as did the overall Gini coefficient. Sweden showed similar results during 1935–1948. The middle-to-lower-income ratio declined little in the Netherlands from the late 1920s to 1950s, with the main redistribution being the decline of the top decile share. See Morrisson 2000, 221, 228, 230. 11 A note of caution about the occupational pay ratios is warranted, however: we cannot assume that two occupations, one paid more than the other, represent fi xed ranking positions on the spectrum of earnings. If carpenters tended to be centered on the fortieth percentile and unskilled laborers tended to occupy the tenth percentile in 1910, we cannot assume that they occupied the same positions in 1970. More likely, both occupations would have drifted down the pay ranks in a modernizing economy. 12 For a summary of the evidence in this paragraph, see Williamson and Lindert 1980a, 78–99, 305–12. See, more recently, Goldin and Katz 2008, 57–71. 13 For the source of this study, see Philippon and Reshef 2012.

200

•

CHAPTER 8

(c)

(b)

0.20

+1.0

(a) 1.7

1.6 0.18

(a) Relative pay, financial occupations

0

1.5

1.4 0.16

(b) Index of deregulation

–1.0

(c) Relative education, financial occupations 0.14

1.3

1.2

–2.0 1.1

0.12

1.0

–3.0 1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

2010

Figure 8-3. Relative Salaries of Financial Occupations, with Some Correlates, 1909–2006

the degree to which the sector was unregulated—that is, unregulated until the crashes of 1929 and 2008 brought tighter financial regulations. And when it was unregulated, the financial sector attracted personnel with exceptionally high education. The experience of this sector offers an excellent illustration of the fall and later rise of a highly paid occupation across the twentieth century. Other skilled and white-collar earnings premiums also fell during the Great Leveling. College professors lost more than half their pay advantage over manual workers between 1908 and 1960, although elementary and secondary school teachers offer a partial exception: they lost a smaller part of their pay advantage over manual laborers up to 1950 (meeting the demands of the high school revolution), and they almost regained their 1910 pay advantage by 1970 (meeting the baby boomers’ demands for more education). Engineers lost between one-quarter and one-third of their pay advantage over the unskilled between 1904 and 1956. Clerical workers lost ground relative to production workers THE GREATEST LEVELING OF ALL TIME

•

201

between 1890 and 1957. Skilled manual workers, such as carpenters or lathe operators, lost some of their pay premiums over ordinary laborers between 1890–1910 and 1940. White-collar workers generally lost ground in both world wars, not regaining it after either war.14 Where we have adequate data, it appears that the Great Leveling also squeezed most of the skilled and white-collar occupations in other industrial countries. After late nineteenth-century stability, British and Australian wage premiums for the skilled manual occupations declined greatly from around 1910 to 1950s, although the results for the Continent are mixed. While the early work of Henry Phelps Brown found no skill premium declines in France or Germany, more recent work by Christian Morrisson found declines by about a third in several wage premiums for Belgium, France, Germany, and Sweden from 1910 to 1955.15 The European white-collar professions also suffered losses in relative pay status in the era of leveling. Such losses have been documented for physicians in Britain, Denmark, and Sweden; dentists in Sweden; lawyers in Britain and Denmark; university professors in Denmark, France, Germany, and Norway; and higher civil servants in Britain, France, Germany, and Norway. The ratio of average salaried pay to average wage pay was also down from 1890–1915 to 1946–1952. This evidence suggests that 1910–1970 stands out as an almostuniversal compression in occupational pay scales. In many, if not most, cases, the occupational rates were not dictated by government policy but rather by market forces. Thus, our search for causes of the Great Leveling within the lower 99 percent must focus on the market fundamentals that could have pushed the entire occupational wage structure toward equality even in the absence of changes in government wage-setting policies.

14 A short-run deviation from this long-run trend can be seen in the depths of the Great Depression, when skilled employees temporarily suffered less than everybody else because they tended to be on longer-term contracts, unlike less skilled workers. 15 Brown 1977; Morrisson 2000, 246. Some of Morrisson’s series are inter-quartile ratios much like our gaps between percentile ranks, while others are more traditional occupational pay ratios.

202

•

CHAPTER 8

Three Postponed Equalizations in America While this chapter emphasizes that the United States shared the same income leveling as other countries between 1910 and 1970, we will have a better understanding of how the American experience fits that of other industrial countries if we take account of the fact that three income equalization forces were delayed in America until after the 1940s or 1950s. One had not occurred at all by 1970, and the other two got under way only after 1940. First, the Great Leveling did not produce more gender equality in pay or work, despite women getting the vote and the big fall in fertility. No labor market forces favoring women were evident in the data. Goldin’s landmark study Understanding the Gender Gap showed that despite the rising tide of overall income equality, there was no significant change in female earning power relative to male—a ratio that remained essentially unchanged from 1890 to 1970. True, women surged into the labor force during World War II, with significant income gains, but this brief advance was largely reversed after the war. Young women returned to more domestic roles, and young men were favored in the GI Bill subsidies of their higher education. Thus, as late as 1970, women had no greater earning power relative to men than women had in 1890.16 The second postponed leveling force was specific to the United States. As we saw in chapter 7, the long, slow march toward racial income equality was halted during the first half of the Great Leveling by the persistence of Jim Crow laws and discrimination policies in the southern states. The march did not resume until World War II, the Great Migration of blacks out of the South, and the civil rights movement. The third postponed leveling force in America was spatial. The nation’s poorest region in 1870, the South lagged far behind the rest of the country for two or three generations following Reconstruction. Helped by two world wars, the region launched some income per capita convergence in the 1940s that continued for the rest of the twentieth century (figure 8- 4). The world wars raised southern incomes by bringing new government contracts to the South (e.g., shipyards in Mobile and 16

Goldin 1990, chaps. 3, 6, and 7.

THE GREATEST LEVELING OF ALL TIME

•

203

160

Relative income per capita (USA = 100)

140

West South Central

120

100

South Atlantic 80

East South Central 60

40 1760 1780 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000

Figure 8-4. The Relative Income per Person of Southern Regions, 1774–2010

Norfolk) and offering job opportunities to southerners emigrating to higher-wage regions (e.g., shipyards in California and Pennsylvania). Once the migration to cities had begun, its momentum was preserved by the robust growth of the postwar golden era, 1945–1973. The rising tide of southern business contributed to and was helped by the civil rights movement of the 1960s. Even the economic fortunes of southern whites were improved by removing the laws and practices that had perpetuated racial segregation.17 The rising prosperity of several southern states was in fact part of a larger geographic convergence of incomes across America. To appreciate that convergence, consider which states had average incomes that were at least 20 percent below the national average (call them “poor”) or 20 percent above it (call them “rich”). The geography of the Great 17

204

G. Wright 2013.

•

CHAPTER 8

Leveling shows up clearly in the state income data gathered annually since 1929 by the US Department of Commerce. In 1929, the rich states were a mixture of those containing the urban centers (cities where financial activities clustered) that had led American growth ever since the early nineteenth century, such as corporate-headquarter Delaware along with Illinois, New Jersey, and New York. This rich club also included California and Nevada. By 1970, this club of rich states had lost nearly all its members. The only places with incomes per capita at least 20 percent above the national average were Connecticut (123.5 percent) and the District of Columbia (135.3 percent). This convergence of rich states toward the national average fits the emphasis of Piketty and his research team. It is also consistent with the fall in financial sector incomes across the Great Leveling. Yet the rise of the bottom income states toward the national average was at least as dramatic as the relative fall of the rich states. Back in 1929, as many as twenty states had per capita incomes more than 20 percent below the national average. This included every southern state in the former Confederacy and several states in the western interior. By 1970, many of these had risen to average income status, offering more evidence that the bottom of the income ranks was gaining on the middle—a tendency not captured by the decline in the share of either the top 1 or 10 percent. The postwar convergence of regional incomes was not confined to America. The same convergence was evident in most industrialized countries that report the data.18 Per capita incomes also converged across regions in Canada, Finland, Japan, the Netherlands, Norway, Spain, Sweden, and West Germany. There was also regional convergence during and before World War II in all of the data-supplying countries except France and Italy, where regional inequality remained higher than in the United States. Thus, regional income gaps shrank during the Great Leveling in most industrialized countries—although the convergence was delayed in the United States until World War II.

18

Williamson 1965, 17, table 2.

THE GREATEST LEVELING OF ALL TIME

•

205

LIKELY CAUSES OF THE GREAT LEVELING What drove the Great Leveling in the industrialized world? Which forces were country-specific, and which ones affected all industrialized countries? Abundant historical evidence exists to explore these questions. True, history never offers many randomized experiments of the sort now preferred in health sciences and econometrics. Nor can we find many neat and convenient “natural experiments” in history, where only one causal variable shifted at a time. Nonetheless, we can move closer to the answers by examining the basic correlations offered up by the historical record. There are six fundamental forces that are likely to change pre-fisc income inequality in a market economy: • Political shocks including wars • Growth in the quantity of labor supplied19 • Growth in labor skills • Biased technological change, thereby shift ing the demand for labor, skills, and capital • Shifts in international trade competition • Shifts in the highly paid finance sector

Of these six, Thomas Piketty’s monumental Capital in the Twenty-First Century has emphasized the first, both as a determinant of incomes before taxes and a direct redistributor of incomes by changing taxes and transfers. His stress on the roles played by world wars, revolutions, and rising voting power for the masses looks qualitatively plausible. That is, it predicts the decline of top income shares in the wake of each left ward political shift or war, especially World War II, for the combatant countries. Yet he does not offer a story of how these same political and macroeconomic shocks should have affected inequality within the bottom 90 percent. We need to explain inequality movements that span the entire income spectrum, and the Great Leveling needs roles assigned to all six forces. 19 In what follows, “labor supply” is synonymous with the size of the working-age population aged fi fteen to sixty-four, the prime age range for paid work. Note that it is different from the size of the actual labor force. The labor force is more responsive to the state of the economy whereas the size of the working-age population is more exogenous.

206

•

CHAPTER 8

Uncontrollable Shocks No explanation of the Great Leveling can ignore the roles played by the shocks highlighted by Piketty: war, peacetime macroeconomic shocks (such as the Great Depression and some countries’ bouts of hyperinflation), and political shocks (especially the left ward shifts that expanded fiscal redistribution). The first two are understandable but are hard to control. The third is clearly within the control of the political process. Piketty is surely on the mark here. His explanation combines periods of diverse historical shocks into a single, long chaotic era from the 1910s to 1970s. World War I raised national income, only to have it dive when the war was over. Much of Europe had to deal with hyperinflation in the 1920s, which crippled the rich and middle classes. The Bolshevik and Ottoman revolutions confiscated a large chunk of private wealth, as did hyperinflations in several countries. The 1930s famously brought the Great Depression, hitting Canada, Germany, and the United States hardest. World War II devastated the European economies and Japan, but brought unprecedented prosperity and labor scarcity in the Americas and Australasia. The postwar era from 1945 to 1973 brought the highest growth rates in per capita income ever experienced by the industrialized world. All these diverse shocks seem to have had the effect of reducing the concentration of incomes, especially post-fisc incomes. How could such diverse shocks have led to similar results in so many countries? Most of the diverse shocks tended to share a common feature, the political shift toward the Left, whether it came from domestic origins or was imposed by foreign victors. The three postwar decades of rapid growth up to 1973—alias the “golden age of growth” in the Englishlanguage literature, or the trente glorieux in French—were also the years during which progressive fiscal redistribution took off. In the United States and Britain, maximum tax rates on top incomes and inherited wealth hit their all-time high.20 An even greater tax bite into top incomes came from the wartime inflation of nominal incomes, which pushed 20 For graphs of the top tax rates on incomes and inheritances since 1900, see Piketty 2014, 499, 503.

THE GREATEST LEVELING OF ALL TIME

•

207

rich Americans up into the income brackets that had been subject to high tax rates since the 1930s. On the European continent, the early postwar era brought a different kind of fiscal redistribution. There, the greater progressivity and falling inequality was achieved more by raising social expenditures rather than by making taxes more progressive. While Americans talk a great deal of the arrival of their own “welfare state” during Franklin Delano Roosevelt’s New Deal, social spending and redistribution never kept pace with northern Europe. The more important point, however, is that progressive fiscal redistribution took place on both sides of the Atlantic, and it reduced post-fisc inequality.21 So far the stage has been dominated by political shocks. What other forces might help explain the Great Leveling? The leading candidates are the labor market forces that affected earnings all up and down the income spectrum, even the top shares. Labor Supply Growth Every dimension of income inequality is clearly affected by the total supply of labor, and typically, modern analysts have ignored it. Indeed, labor supply strongly affects movements even in the top income shares. Recall that each of those top income shares is a ratio of top incomes to total incomes. So those top shares can be driven not only by forces affecting the average incomes of those at the top but also by average incomes in the denominator. The oversight of the labor supply, especially by economists, is even more surprising if one accepts Piketty’s premise that top incomes mean mainly capital incomes, and other incomes mean mainly labor incomes. Surely trends in any ratio of incomes per capitalist to earnings per 21 How did the fiscal redistribution affect pre-fisc inequality? Both theory and the empirical evidence have offered mixed verdicts. To the extent that fiscal leveling was accompanied by nonbudgetary policies of left ist governments, such as collective wage setting and tougher regulations on business, and to the degree that it reduced accumulation by top income groups, the rise of budgetary redistribution would also have equalized pre-fisc incomes. So Piketty plausibly conjectured about the early postwar era. On the other hand, it could have reduced the supply of risk takers and investments at the top, lowering the pretax incomes for managers and other top earners. The data suggest that the two opposing forces cancel each other out. Three studies covering twenty-four countries around the year 2004 have found that the correlation between the net redistribution of income and the inequality of pre-fisc income is insignificant (Lustig 2011; Wang, Caminada, and Goudswaard 2012; Ostry, Berg, and Tsangarides 2014; Solt 2009, 2014).

208

•

CHAPTER 8

laborer would be driven largely by the trends in the labor supply. This inequality argument goes back at least to David Ricardo and Karl Marx, with empirical analysis supplied by research since the 1970s.22 One can make good progress explaining trends in overall income inequality simply by contrasting movements in the total labor supply with trends in all the other leading macroeconomic factors so familiar to us—trends in the supply of land, capital, skills, and the impact of new technologies on the relative demand for all of these. In the United States and other English-speaking countries, the period between the 1910s and 1940s was famous for slow population growth. Quotas restricted immigration, and families were smaller. It’s little wonder that the late 1930s produced a gloomy English-language literature on “secular stagnation.” In an agricultural United States in the early nineteenth century, theory suggests that faster labor force growth meant a decline in the land– labor ratio, a rise in the ratio of rents to wages, and thus more inequality. For an industrial United States in the twentieth century, theory suggests that a slowdown in labor force growth should have caused a shift from slower wage gains and higher profit rates to faster wage gains and lower profit rates. Theory also suggests that slower labor force growth, whether caused by a drop in low-skilled immigration or fertility, should have compressed the wage structure. This is so because the slower labor force growth allows a faster buildup of skills per worker, bidding down the skill premium and lowering wage inequality. The labor force slowdown after 1910 was sharper in the younger countries of North America and Australasia than elsewhere. Table 8-1 documents that around 1910, the rate of growth of the US working-age population (ages fifteen to sixty-four) permanently switched from about 20 percent per decade to never being that fast again. By contrast, the labor force growth of most of continental Europe and Japan did not fall. By itself, the differential deceleration should have caused a greater leveling in those overseas English-speaking countries than in continental western Europe and Japan. To test this proposition, we need to know the country’s inequality trends between 1910 and 1970 as well as its 22 For a nonrandom selection of this more recent research, see Lindert 1978, chap. 7; Williamson and Lindert 1980a, especially chaps. 6–11; Lindert and Williamson 1985, especially 350–56; Hatton and Williamson 1998, 2005; Lindert and Williamson 2003.

THE GREATEST LEVELING OF ALL TIME

•

209

Table 8-1 Growth of the Fifteen- to Sixty-Four-Year- Old Population, 1870–1970 Growth in percent per decade 1870–1910

1910–1970

Australia

28.1

19.2

Canada

21.5

20.7

New Zealand

47.2

16.4

United States

27.3

14.2

United Kingdom

13.2

4.7

Ten continentals

5.4

9.4

Japan

8.7

15.8

Sources and notes: The ten continental countries are Denmark, Finland, France, Italy, the Netherlands, Norway, Portugal, Spain, Sweden, and Switzerland, weighted by 1980 population. For “1950,” the rates are average changes per decade for 1930–1950. See appendix H.

inequality trends before 1910. Unfortunately, as figure 8-1 has already shown, inequality trends before 1910 are documented for too few countries to assess the correlation. We thus cannot explore international differences in the working-age population growth rate and inequality trends from 1870–1910 to 1910–1970. Instead, inferences about the role of the slowdown in the labor supply growth can only be based on the US and UK evidence for 1910–1970 and before (the latter presented in earlier chapters), intercontinental comparisons within the 1910–1970 Great Leveling, and the experience since. The labor force growth slowdown shown in table 8-1 was driven by fertility decline and the shutting down of immigration. The interwar decades and the period since the 1970s saw many industrial countries produce fewer children than would be needed for even a stable population were it not for improvements in life expectancy. The historical decline was steepest in the United States. Of course, two world wars also contributed to the slowdown, what with marriage disruption and adult deaths. America’s great population slowdown was reinforced when the US government slammed the door on immigration in the wake of World War I. So great was the change in immigration policy that the foreignborn share of the US labor force dropped from over 21 percent in 1915 210

•

CHAPTER 8

Finland

–1

Australia

Top 1% income share, % rise per decade

Norway Denmark

Canada

USA New Zealand

Sweden France

Japan

–2

UK Netherlands

–3 0

5

10

15

20

Working age population, % rise per decade Figure 8-5. Changing Concentration of Income at the Top versus Labor Supply Growth, 1920–1970

to just 5.4 percent in 1970.23 Table 8-1 shows similar declines for Australia, Canada, and New Zealand, though all four countries continued to have faster population growth than most of Europe. Many of the European countries that recorded a rise in working-age population growth were countries of emigration before World War I but not after. Within the Great Leveling era, countries with slower labor force growth experienced deeper income inequality reductions (Figure 8-5). The five European countries that had less rapid growth in their workingage populations than the United States also had a greater reduction in their top 1 percent income shares. Australia and Canada, whose working-age populations grew faster than the United States, saw a bit less income leveling. While the international correlation is hardly perfect, it is significant.24 We will find a similar international pattern when in chapter 9 we explore inequality trends after 1970. 23

Goldin and Katz 2008, 308. In statistical terms, the correlation of 0.443 was significant at the 10 percent level for twelve countries. 24

THE GREATEST LEVELING OF ALL TIME

•

211

More Education for the Labor Force The unequal rewards to labor depend not only on the number of persons working or seeking work but also on their qualifications. These in turn depend on how well they are prepared by their home environment, their formal education, and the learning they get on the job. Of these three sources of skills, the one for which we have the most evidence is formal education, and that is the one we pursue here. Was American education rising faster between 1910 and 1970 than before 1910 or after 1970? Was the same true in other countries undergoing the Great Leveling? This issue certainly deserves more attention.25 Faster growth of education and skills should lower both the top income shares and the wage differentials within the bottom 99 percent. It should lower the top shares because it greatly raises the labor earnings power of everybody, and labor earnings constitute a much larger share of personal income in the lower 99 percent than they do in the top 1 percent. It should also lower the wage and income premiums received by schooled workers. After all, faster education growth expands the supply of more skilled workers, and their greater abundance should reduce their pay premium. Thanks to government agencies and careful scholarship, we can now construct a long-run chronology of adult educational attainment for the United States and a few other countries. Daniel Cohen and Marcelo Soto have already worked out the magnitudes for the whole world from 1960 to 2010, based on data from national governments. Collins and Margo have determined how many years of schooling US white and black adults achieved in 1940, and a few assumptions allow us to extend the 25 Piketty underemphasizes the role of education. True, Piketty (2014, 304– 6) does cite the Goldin and Katz analysis of the American “race between education and technology,” but he promptly dismisses it as “limited and naïve.” Elsewhere, he seems to concede more:

Historical experience suggests that the principal mechanism for convergence at the international level as well as the domestic level is the diff usion of knowledge. In other words, the poor catch up with the rich to the extent that they achieve the same level of technological know-how, skill, and education, not by becoming the property of the wealthy. (Ibid., 71) And later he notes, “The best way to increase wages and reduce wage inequalities in the very long run is to invest in education and skills” (ibid., 313). Yet education plays no further role in Piketty’s accounting of inequality.

212

•

CHAPTER 8

Table 8-2 Average Years of Schooling for the Fifteen- to Sixty-Four-Year- Old Population, United States, 1930–2010 Census year

Blacks plus whites (Collins and Margo)

All races (Cohen and Soto)

Growth in years of schooling, per decade

1930

8.23

—

—

—

1940

9.13

—

0.90

—

1950

9.80

—

0.67

—

1960

10.48

10.18

0.68

—

1970

11.29

11.27

0.81

0.89

1980

—

12.19

—

0.92

1990

—

12.62

—

0.43

2000

—

12.63

—

0.01

2010

—

13.24

—

0.61

national figures back to 1930. Table 8-2 and figure 8-6 report the US results.26 This evidence reveals something for the United States previously unnoticed. While educational attainment, measured as the average years of schooling completed by adults, advanced by nearly one school year per decade during the Great Leveling, a marked deceleration followed. The adult population’s average schooling grew at less than half that rate across the 1980s and came to a virtual standstill across the 1990s, before resuming its advance in the first decade of this century. We return to this schooling slowdown in the next chapter, where we explore the steep rise in inequality since the 1970s. To compare the advance in American education in the Great Leveling with earlier years or progress in other countries, indirect evidence must be used, since typically adult years of schooling were not recorded before 1940. Yet adult education can be inferred from the more abundant information on school enrollments back in their childhood years. Today’s adults would have experienced the enrollment, attendance, and grade completion rates recorded in census years ten, twenty, thirty, and 26

See Cohen and Soto 2007; Collins and Margo 2006.

THE GREATEST LEVELING OF ALL TIME

•

213

14

All races (Cohen–Soto)

Average years

12

Blacks plus whites (Collins–Margo) 10

8

6 1930

1950

1970

1990

2010

Figure 8- 6. Education Attainment in the United States, 1930–2010

forty years earlier. Table 8-3 reports this measure for the United States and other countries that recorded the necessary schooling data. America was always ahead of other countries in terms of school enrollments, giving it a clear advantage in labor earning power. Thus, during the Great Leveling the United States combined its slower labor force growth with rapid gains in education, particularly during the high school movement featured in the work of Goldin and Katz.27 As such, the combination should have contributed to the Great Leveling up to 1970. How did other countries compare with the United States? The enrollment indicators in table 8-3 offer an answer, at least for the period up to World War II. For 1910 to 1940, nine other data-supplying countries raised their adult years of schooling at least as fast as the United States. Six of the nine—Belgium, Canada, France, Italy, Japan, and the United Kingdom—converged on American adult education levels. Indeed, these six (plus some others) have overtaken the United States and increased their education lead since 1980—a point to be explored at greater length in the next chapter. A corollary emerging from table 8-3 is that America’s great high school movement in the early twentieth century did not push the United States to global leadership in education. That had already been achieved 27 Unfortunately, we cannot say whether US education advanced more rapidly after 1910 than before.

214

•

CHAPTER 8

Table 8-3 The School Enrollment Histories of the Population of Ages Fifteen to Sixty-Four in Selected Countries, 1900–1940 Weighted lagged enrollments per 1,000 school-age children 1900

United States

1910

1920

1930

1940

Ratio, 1940 ÷ 1910

Attainment ratio, 2010 ÷ 1980

Converging toward United States in both eras?

951

984

1,011

1,045

1.10

1.08

Australia

698

793

857

859

895

1.13

1.08

— — —

Belgium

543

560

576

649

715

1.28

1.24

Yes

Canada

—

856

886

943

989

1.16

1.15

Yes

France

514

567

679

672

684

1.21

1.23

Yes

Prussia/Germany

769

786

797

—

—

—

1.01

—

Italy

325

358

400

445

507

1.42

1.38

Yes

Japan

—

364

490

589

683

1.88

1.17

Yes

Netherlands

—

653

673

695

733

1.12

1.12

—

Norway

650

676

698

729

749

1.11

1.10

—

United Kingdom

—

632

681

723

760

1.20

1.15

Yes

before 1910, despite Jim Crow policies that held back black education, and despite the lower education of immigrants. A Less Labor-Saving and Skill-Using Bias of Technological Change There was an additional force driving the Great Leveling. Technological change can favor or disfavor unskilled labor, or can do the same for skilled labor, affecting jobs and incomes. This has been stressed during the recent great wave of automation and outsourcing experienced since the 1970s. But the recent literature has not noted that this happened less in the Great Leveling than it did before 1910 or after 1970. The evidence has taken the form both of econometric estimates of aggregate THE GREATEST LEVELING OF ALL TIME

•

215

production functions and of productivity growth by sectors. The latter singled out the first two-thirds of the twentieth century as one in which unskilled labor-intensive sectors, such as agriculture, kept their productivity growth up with other sectors. For agriculture, the explanation lay partly with hybrid seed varieties introduced before World War II. Since total factor productivity did not fall behind in the unskilled labor-intensive sectors, technical progress did not produce an unskilled labor-saving bias due to induced output mix changes. In short, the Great Leveling decades were filled less with unskilled laborsaving technical change than in other periods.28 Goldin and Katz reach a similar conclusion about the race between technological bias and education: “On average from 1915 to 2005 . . . [n]either education nor technology won the race in the long run.” But from 1915 to 1960, or for most of the Great Leveling, the relative skill “supply ran ahead of demand by about 1 percent average annually.” For the period 1980 to 2005—which we will cover in chapter 9—“demand outstripped supply” in the market for skills.29 In short, the educationversus-technology race is a part of the inequality story. Adding a less labor-saving technological bias to our causal explanations of the Great Leveling has an additional advantage. It helps explain the fact that practically all industrialized countries shared the income leveling. After all, new knowledge and the ability to use it quickly diffuses globally. Trade Winds and Financial Storms The last two of our six fundamental forces driving income inequality inequality deal with trade and finance. Higher trade barriers diminished exports and imports during the Great Depression and World War II, followed by only a slow and partial recovery up to the 1970s. How should diminished trade have affected inequality within the advanced countries? 28 For intersectoral evidence on the locus of technological change, see Williamson and Lindert 1980a, 144– 46, 156–77, and the literature cited there. 29 Goldin and Katz 2008, 303. Note that the terms “demand” and “supply” refer to the market for adult skills, not the market for education. Thus, the 1980s’ slowdown in the acquisition of education is viewed as a “supply” force here, even though it may have been induced by demand for schooling.

216

•

CHAPTER 8

All we can offer here is the likely direction of effects from shift ing trade winds. The United States was already exporting skill- and capitalintensive products and importing labor-intensive products by the 1910s, so income inequality should have been reduced by the shrinkage in world trade. Less trade meant less income advantage for those earning returns from their capital and skills in export industries, and more returns for those employed in import-competing labor-intensive industries. The shrinkage of trade probably contributed to the equalization of American incomes up through World War II. After the war, the sluggishness of the trade revival would have continued to hold down the income advantage of skilled workers and hold up that of less skilled workers, were it not for an effect that favored American exports. World War II so damaged European and Japanese productive capacity that it shifted world demand for capital- and skill-intensive products toward the United States. The overall postwar effect of trade shifts on US inequality was therefore probably mixed. In contrast, the storms in the financial sector made a strong contribution to the Great Leveling, at least in America. The remarkable decline of incomes in the US financial sector (figure 8-3) suggests that whatever gave the unregulated financial sector its income boom before the 1929 crash, led to pessimism and tight regulations afterward—forces that suppressed the financial component of top incomes for a half century. Economists and economic historians have now established two links in this causal chain. The first link runs from credit booms, like those of the 1920s and 2003–2007, to crises in asset markets and the banking sector. In the Great Leveling era, the second link was a strong political move toward tighter regulation of the financial sector. As Barry Eichengreen has emphasized, the greater the financial crash, the tougher the subsequent tightening. As a result, incomes in the high-skill financial sector were held in check and financial storms were avoided until the 1970s.30

30 Eichengreen 2015. Many have conjectured that there is an intermediate link running from the credit boom to inequality, and then to the fi nancial crises. Perhaps the credit boom throws too much of the national income into the hands of top income groups that overspeculate with their surging wealth, triggering the crisis. Th is is an intriguing possibility. Thus far, this thesis leans only on circumstantial evidence: namely, that US inequality was high right before the crashes of 1929 and 2008. Michael Bordo and Christopher Meissner (forthcoming) found no such link in the historical experience of fourteen countries between 1920 and 2000.

THE GREATEST LEVELING OF ALL TIME

•

217

A NEW CAUSAL UNDERSTANDING There now is enough historical evidence to explain the Great Leveling. First, we know it was not just a drop in the top income shares, since it took place everywhere in the distribution. Second, we know that six basic forces played the main causal roles in explaining the leveling. One such force was the appearance of progressive New Deal fiscal policies— policies that were reinforced greatly during World War II when inflation pushed the incomes of the rich up into higher income tax brackets. If our task were simply to explain the rise in fiscal redistribution—that is, a rising gap between pre- and post-fisc inequality—the New Deal political shift and World War II would do the job quite nicely.31 But the tougher task is to explain trends in pre-fisc inequality—that is, the inequality of market incomes before taxes and transfers. Progressive redistributive policies do not explain much of this shift after 1910. Instead, we must appeal to fundamental factor-market changes that occurred over the Great Leveling after 1910: not only those military and political shocks, but also the great slowdown in labor supply growth, a rapid advance of education, the slowdown in technological bias against the unskilled, a much more antitrade world economy that deflected labor-intensive imports from American shores and suppressed American skill- and capital-intensive exports, and a retreating financial sector. Most of these forces reversed after the 1970s and thus offer an explanation for the rising inequality thereafter—a reversal assessed in the next chapter. 31 The structure of fiscal redistribution was not significantly more progressive in 1970 than in 1950, despite the large increase in social spending over those two decades (Reynolds and Smolensky 1977).

218

•

CHAPTER 8

CHAPTER 9

Rising Inequality Once More, since the 1970s

T

he Great Leveling stopped in the 1970s, and the trend has reversed ever since. As it happens, the countries that underwent the greatest reversal speak English. So say the top income shares for eleven advanced economies (figure 9-1).1 Since the episodic reversal in the 1970s, incomes have become more concentrated in the top 1 percent in Australia, Canada, New Zealand, the United Kingdom, and the United States. They have also become more concentrated on the European continent and in Japan, although much less so. In the last chapter, we dated the end of leveling variously, depending on the data source, at 1970, 1973, or the 1970s, but the start of a renewed rise is probably best centered on 1977. In any case, this chapter will continue to refer to that episodic turning point as “the 1970s.” A principal task will be to explain why the top shares have risen in the industrialized world between 1977 and 2007, or even up to 2010, after the top incomes had been hammered a bit in the Great Recession.2 The contrast with the previous Great Leveling is sharp. As with the Great Leveling, the challenge is to explain why there was such a pronounced change of trend in the inequality of incomes before taxes and transfers, and why it was more dramatic in some countries than others. Once again, we will consider the six prime suspects we rounded up in the last chapter: namely, political shocks, growth in the quantity of labor 1 For more countries and more years, see Roine and Waldenström 2014; World Top Incomes Database, http://topincomes.g-mond.parisschoolofeconomics.eu/ (accessed August 20, 2015). 2 We can be a bit more precise. Of the thirteen countries yielding top income shares for 1977 to 2007, only Denmark had a stable share. Of the eleven countries covered for 1977 to 2010, all eleven saw a rise in the top income share. The greatest increases occurred in English-speaking countries, except that the increase in New Zealand was smaller than in some continental European countries.

20

USA

15

Percent

UK Canada, Australia, New Zealand 10

Japan

Five continentals 5 1950

1960

1970

1977

1990

2000

2007 2012

Figure 9-1. Share of National Income Received by the Top 1 Percent, United States and Elsewhere, 1950–2012

supplied, growth in skills and schooling, biased technological change, shifts in world trade patterns, and the growth of the financial sector. All six of these forces probably contributed to the rise of inequality from the 1970s. To elaborate: • Faster growth of the overall labor supply made inequality rise more steeply in English-speaking countries than in Japan or continental western Europe. • Slower growth in education attainment is another distinguishing feature of some English-speaking countries, again helping to explain the steeper rise of their income inequality. • While labor supply and education trends help explain international differences in the steepness of the inequality rise, they cannot explain the huge shift between the leveling era (1910–1970) and the widening era (1970–2010). • This shared shift can only be explained by other shared forces—that is, rightward political shifts, an acceleration in the adoption of laborsaving technologies, the massive rise in labor-intensive imports from 220

•

CHAPTER 9

the emerging nations, and the explosion of financial activity after six decades of tighter regulation.

Before turning to the evidence for these causal forces, let us first broaden the range of income movements that deserve explanation. The post-1970 rise in inequality in the United States and other English-speaking countries, like the earlier leveling, took the form of widening gaps all up and down the income ranks, and not just a rise in the share going to the top 1 percent.

WIDENING GAPS AMONG THOSE BELOW THE TOP Overall inequality trends since the 1970s matched those of the top shares, since the gaps widened even among those who earned their incomes from ordinary work. In the United States, what the media and the Occupy movement have been claiming about the past decade or so has been true: the lower 99 percent has seen only small gains in real earnings, and the lower half has seen essentially none. True, Englishspeaking countries saw more widening than did other industrialized countries, but the fact remains that since the 1970s, no country has experienced a narrowing of income gaps—not among the bottom 90 percent, not among the top 10 percent, and not between the two. And most have experienced a widening. Thanks to the work of Atkinson and his collaborators, we can now document inequality within the lower 90 percent, using data on the labor earnings distribution that were generally absent before the 1970s.3 Three measures of the inequality of pretax labor earnings are now available. First, there is the measure of what we will call an “upper-gap” in the wage ranks: the average annual labor earnings of those in the top 10 percent of earners divided by the median labor earnings. This measure is available for eleven countries between 1977 and 2010. Second, there is another upper-gap measure: the average annual labor earnings in the ninetieth percentile of the wage ranks divided by the median labor earnings, available for nineteen countries between 1980 and 2004. Third, there is a “lower-gap” measure: median earnings divided by the 3

Atkinson and Morelli 2014; Atkinson 2008.

RISING INEQUALITY ONCE MORE

•

221

average annual earnings of those in the tenth percentile of the wage ranks, available for the same nineteen countries between 1980 and 2004. Each of these three measures documents an unmistakable rise in wage inequality. The first upper-gap measure of labor earnings inequality rose between 1977 and 2010 in all but one of the eleven countries.4 Between 1980 and 2004, all nineteen countries experienced an increase in the second upper-gap measure. The lower-gap measure rose in twelve of the nineteen countries between 1980 and 2004. Increases in labor earnings inequality have thus been even more widespread than increases in the top income shares since the 1970s. To understand these trends, we need to look at the changing labor market conditions, and not just at the forces changing wealth concentration in the hands of a few and income shares at the top.

THREE AMERICAN COUNTERCURRENTS The reopening of wide gaps in earnings and income since the 1970s could have been worse, at least in America. Even though the United States recorded the biggest inequality rise, we must first emphasize three sources that did not get worse. In fact, there was a net improvement on all three fronts. The three American inequality countercurrents are the same three that failed to join the Great Leveling forces in America from the 1910s to 1970s: racial, regional, and gender inequality. Let us next note how different the recent history was in these dimensions, in order to narrow our search for the sources of the overall inequality rise. Racial Gaps The last four decades have not seen any widening in black–white income gaps despite the rise in overall inequality. This good news is a bit dampened by the slowdown in the black–white income convergence that had been achieved between 1940 and 1970. When did the slowdown in black catch-up set in, and why? The time series data and the studies assessing them do not agree on the answer, 4

222

The exception is France, where it declined slightly.

•

CHAPTER 9

but the literature tends to point to the 1980s.5 Collins and Margo have emphasized the damaging effects of urban race riots on black property values from the late 1960s onward. Although the riots themselves were concentrated in the late 1960s, the effects on local property values were even greater in the 1970s and later. As Margo observed in an interview, “From 1940 to 1970, the value of homes owned and occupied by blacks in central cities jumped to 69 percent of the value of urban homes owned and occupied by whites, from 51 percent. By 1990, however, the ratio was down to a mere 53 percent, nearly as low as in 1940.”6 The riot-related damage was enlarged by widespread residential segregation, which made the damage race specific: a loss in home values concentrated on black homeowners cut their consumption of goods and services that were mainly local, and mainly a source of incomes for other blacks. In addition, John Bound and Richard Freeman have pointed to a perfect storm of employment factors that impacted the economic fortunes of black males in the 1980s. The storm was partly regional and partly sectoral: black males were disproportionately employed in Rust Belt industries in the Northeast and North Central, regions that were hit hard by rising competition from East Asia. Private sector union power was also in full retreat in the 1980s, and the combination of inflation and the conservative swing in politics eroded the purchasing power of legislated minimum wage rates. Some black males also received declining rewards from their rapid entry into college-graduate professions, while crime rates and incarcerations rose with the high school dropout rate of other black males. In spite of this slowdown, and in spite of the Great Recession of 2008– 2011, the income gap between blacks and whites has narrowed slightly since the 1970s—an event that implies less inequality, not more. The best period of relative income gains for blacks since the 1970s was the boom period 1992–2000.

5 We focus on the recent work of Collins and Margo (2007) and Bound and Freeman (1992) here, but see also the surveys by James Smith and Finis Welch (1989) and John Donohue III and James Heckman (1991). 6 Quoted in New York Times, December 30, 2004.

RISING INEQUALITY ONCE MORE

•

223

Regional Convergence A second countercurrent against the recent rise in American inequality was the convergence of state incomes on the US average. True, the convergence did not involve the urban Northeast. In 1970, only Connecticut and the District of Columbia had average incomes 20 percent above the rest of the country. By 2010, those two were joined by Maryland, Massachusetts, New Jersey, and New York. The new prosperity of the Northeast echoed that of the top 1 percent nationwide, being led in part by occupations in finance and high tech. While much of the Northeast pulled ahead of the rest of the country, a more important regional event was that the poorer South continued its rise toward the national average, erasing almost all the region’s income deficit that this book has traced from the early nineteenth century. Today, only Mississippi has an average income less than 80 percent of the national average. The rest of the South is in the near-average category, having converged greatly toward the US mean. Alabama has become a major producer of quality cars from Honda, Hyundai, and Mercedes-Benz, and has attracted engine assemblies for Navistar and Toyota. South Carolina has become a center of aircraft assembly. Florida continues to attract retirees and their pension incomes. And Charlotte has become one of the nation’s top centers of banking and finance. All these developments are consistent with Gavin Wright’s emphasis on how the civil rights revolution of the 1950s and 1960s helped deliver solid income gains for southern whites, while at the same time attracting both black and white migrants into the region.7 Gender Gaps: Swimming Upstream An even stronger offset to America’s recent rise of inequality has been the accelerated narrowing in the gender pay gap. One might have expected women to fall farther behind, given that they were disproportionately employed in lower-paying jobs in 1970. Contrary to that expectation, during these years of rising overall inequality, American women have managed to “swim upstream.” Figure 9-2 makes it clear 7

224

G. Wright 2013.

•

CHAPTER 9

that women have broken out of the long stasis in their relative pay. Over the eighty-five years between 1888 and 1973, American women received an average of only three dollars pay for every five dollars received by the average male worker. Yet women have made big gains in average pay since the 1970s—or perhaps since 1982—swimming upstream against a strong current of increasing inequality and a conservative resistance to social programs. Goldin offers an economic explanation of this recent narrowing of the American gender gap.8 An important permissive condition, of course, has been the acknowledgment of women’s equal career rights by governments and the legal system, as in the Civil Rights Act of 1964. Another fundamental impetus toward gender pay equality has come from a new technological bias that has gathered steam since the 1820s: the declining relevance of brawn. What the economy has demanded increasingly are jobs that require more intelligence, analytic skills, organizational skills, communications skills, and patience, with less demand for physical strength. By itself, this technological drift should have been raising women’s relative pay gradually ever since the early nineteenth century, so this seems an inadequate explanation of the delay in relative pay gains.9 Thus, much of the story has to be the twentieth century’s political shift toward granting more equal rights to women, both in elections and labor markets. For this explanation to hold, the advance in women’s relative pay should have visited all advanced countries in recent times, not just the United States. This is indeed the case: figure 9-3 reports progress in female–male pay ratios in six other countries (Australia, Japan, the United Kingdom, and three Scandinavian countries). One crucial test of the economic explanation of the gender pay trends will be provided in the next decade or two by Japan. Is the rise in Japanese women’s relative pay around 2000 (figure 9-3) an artifact of some change in Japan’s official data series, or will the country famous for the long hours put in

8

Goldin 1990, 2014. Figure 9–2 reports an even earlier rise in the gender wage ratio, from 1820 to 1888. Th is increase has been documented for the manufacturing sector only, and we do not pursue it here. Yet it speaks to the fall in the relevance of brawn. For a fuller discussion, see Goldin 1990, chaps. 2 and 3. 9

RISING INEQUALITY ONCE MORE

•

225

1.0

0.9

All sectors, full-time weekly 0.8

0.7

0.6

Manufacturing only

All sectors, year-round

0.5

0.4

0.3 1820

1840

1860

1880

1900

1920

1940

1960

1980

2000

Figure 9-2. Female–Male Wage Ratios in the United States, 1820–2010

1.0

Australia

0.9

0.8

Three Scandinavian countries

USA UK

0.7

Japan 0.6

Japan 0.5

0.4 1965

1975

1985

1995

Figure 9-3. Female–Male Wage Ratios, Several Countries, 1967–2006

2005

by corporate “salarymen” actually join the multinational shift toward more part-time and flexible hours favoring women? Females are now a slight majority of those graduating from universities in virtually all advanced countries. This international phenomenon cannot be the result of events in just the United States, even though the timing of the US shift toward the female college majority was partly conditioned by the phasing out of the male-biased GI Bill subsidies to veterans’ education after World War II (which continued through the Korean and Vietnam wars).10 More international in scope is the rising incentive for women to gain higher education credentials as earning power insurance. This insurance motive has followed in the wake of new certainties and uncertainties in women’s lives. Improved birth control gave them greater certainty of uninterrupted study and career for those seeking that path. The greater uncertainty came from the rise of divorce. With increasing risk of losing access to the income of a male breadwinner, and facing a lingering prospect of slower advance up the occupational ladder, women have turned increasingly to higher education as a way to strengthen their grasp on the middle rungs of that ladder.11 In short, three forces have bucked the powerful inequality tide in America since the 1970s: the erosions of racial pay gaps, gender pay gaps, and North–South income gaps. The great rise in US inequality since the 1970s would have been even greater had these three not offered a partial offset.

CAUSES OF THE RECENT WIDENING If it was not racial, regional, or gender income gaps, what was behind the widening of income gaps in the United States after the 1970s? The movements in relative incomes seem to have been shaped largely by the same six forces that caused the Great Leveling: the political shift s already emphasized by Piketty and others, plus labor force growth, educational progress, technological bias, international trade shifts, and the development of the financial sector. The political shift to the right 10 Although it was “means” tested, the GI Bill included low-interest mortgages and small business loans. Canada had a similar program. 11 Goldin, Katz, and Kuziemko 2006.

RISING INEQUALITY ONCE MORE

•

227

has been well covered by others, and we turn to add some needed emphasis on the other five forces.12 Labor Force Growth We start with the rate of growth of the working-age population. This has played a role in the widening of both earnings and income gaps since the 1970s, just as it did in the Great Leveling and during America’s leap to modern economic growth across the nineteenth century. Any assessment of the impact of labor force growth on inequality depends on the context. If we are comparing the growth of the American working-age population after 1910 with that before, it matters. If we are comparing the growth of the American working-age population between 1910 and 1970 with that since 1970, it matters little.13 If we are comparing the rates of growth in the US working-age populations with the rates in other countries since the 1970s, it matters. Working-age populations grew faster in Australia, Canada, New Zealand, and the United States—all of them countries where the income gaps widened greatly compared with western Europe or Japan, and all of them countries of heavy immigration (since their founding)—as summarized in table 9-1. Comparing countries thus gives us reason to believe in a positive unit impact of labor supply growth on inequality. That is, faster growth of the US working-age population contributed to greater inequality rise from the 1970s onward, just as the slowdown between the 1910s and 1970s contributed to a greater inequality fall. Our belief gains support not only from conventional theory but also from the broader evidence that countries with a faster-growing labor force had steeper inequality trends since 1970. Figure 9-4 provides such evidence. Australia and North America, where income gaps have widened so dramatically since 1970, differed significantly from continental Europe and Japan, where the rates of working-age population growth were so much lower.14 12 For the rightward shift in international perspective, see again Piketty 2014. For American evidence emphasizing political suppression of union power, see Levy and Temin 2007. 13 See tables 8-1 and 9-1. 14 For 1970–2010, the correlation coefficient of 0.49 was significant at the 7 percent level on a one-tail test. Compare this with the 1920–1970 coefficient of 0.44, which chapter 8 noted as significant at the 10 percent level.

228

•

CHAPTER 9

Table 9-1 Growth of the Fifteen- to Sixty-Four-Year- Old Population per Decade, 1910–2010 Growth in percent per decade 1910–1970

1970–2010

Australia

19.2

17.1

Canada

20.7

15.1

New Zealand

16.4

14.6

United States

14.2

12.5

United Kingdom

4.7

4.0

Ten continentals

9.4

6.4

Japan

15.8

3.2

* The ten continental countries are Denmark, Finland, France, Italy, the Netherlands, Norway, Portugal, Spain, Sweden, and Switzerland, weighted by 1980 population.

Yet labor supply growth cannot sweep competing explanations from the field. It might seem so, given the two international correlations reported in figure 9-4. Note, however, the major shift in the relationship. For any given rate of growth in the labor supply, the observed change in inequality jumped up between the two eras. The jump in the top centile share of all income was an extra 2.25 percent per decade at a historically slow rate of labor force growth, and about 2.75 percent per decade at a historically fast rate of labor force growth.15 Something else must have happened to all developed countries to explain the fundamental shift from the Great Leveling to the recent turnaround. We will end up attributing this global shift to powerful exogenous changes in technological bias, trade, finance, and politics. But before we offer a final assessment, credit must be given to some other 15 To see where these shifts come from, compare the positions of the two regression lines in figure 9- 4. At the slow rate of labor supply growth of 5 percent per decade, the predicted gain in the top centile share was minus 2 percent of the national income in the fi rst period (roughly, the French case), but switched to plus 0.25 percent in the later period (roughly the Swedish case), a jump of 2.25 percent per decade. At the fast rate of labor supply growth of 15 percent per decade, the predicted top centile gain was minus 1.50 percent in the earlier period (roughly the New Zealand and US cases), but it jumped to plus 1.25 percent in the second period (roughly the Canadian case), for a net upward shift of the curve by 2.75 percent per decade.

RISING INEQUALITY ONCE MORE

•

229

Rise of the top 1% share of all income , % per decade

3

USA

2

1

From 1970 to 2010

UK

Canada

Australia

New Zealand

0

France

–1

USA

New Australia Zealand Canada

France From 1920 to 1970

–2

UK

–3 0

5

10

15

20

Growth of the age 15 to 64 population, % per decade Figure 9-4. Changing Concentration of Income at the Top versus Labor Force Growth, 1920–1970 versus 1970–2010

forces that help explain the international differences in the inequality trends within the period since the 1970s. Education Falls Behind in the Race against Technology After 1970, even more than in the previous era of leveling, countries differed in the extent to which education kept up with technology. The United States suffered a slowdown in the growth of schooling attainment, even in the face of rising rates of return to higher education. The slowdown is best dated as occurring in the 1980s and 1990s (table 9-2).16 Why the slowdown, and why then? One factor was the rising tide of 16 For broad coverage of the slowdown in educational attainment from the 1970s on, see Goldin and Katz 2008; Gordon 2016, 620–27. We date the period of greatest slowdown as 1980s– 1990s in order to use the decadal census data on education attainment, as we did when drawing on the Cohen and Soto estimates in table 8-2 and figure 8-8. Goldin and Katz offer similar dates for the recent slowdown in the supply of the more educated labor force. For the supply of college graduates, their slow growth era is 1980–2005, and for high school graduates the slow growth era was 1990–2005 (Goldin and Katz 2008, 297, 305).

230

•

CHAPTER 9

Table 9-2 Average Years of Schooling for the Population of Ages Fifteen to Sixty-Four, in Twenty-Three Countries, 1960–2010 1960

1970

1980

1990

2000

2010

Australia

9.8

11.0

12.2

12.8

13.1

13.2

Austria

8.3

9.3

10.3

10.9

11.4

11.7

Belgium

7.4

8.3

9.2

10.0

10.8

11.4

Canada

9.1

10.4

11.6

12.4

13.1

13.3

Denmark

9.1

10.1

11.0

11.5

12.2

12.3

Finland

6.9

8.0

9.5

10.7

11.7

12.3

France

6.7

8.0

9.3

10.4

10.7

11.4

Germany

9.5

11.1

12.7

13.2

13.0

12.7

Greece

5.9

6.7

7.7

8.7

9.9

10.7

Ireland

7.2

8.0

8.9

9.5

10.2

10.6

Italy

5.8

6.8

8.0

9.1

10.3

11.0

Japan

9.5

10.4

11.2

11.9

12.6

13.1

Korea

5.0

6.8

9.1

11.0

12.3

13.3

Netherlands

8.3

9.3

10.3

10.7

11.3

11.5

New Zealand

9.0

9.9

10.7

11.0

12.1

12.5

Norway

9.1

10.3

11.6

12.3

12.5

12.7

Portugal

3.2

4.1

5.6

5.9

7.3

7.9

Singapore

4.2

5.8

5.8

7.1

9.8

11.2

Spain

5.8

6.5

7.4

8.4

9.5

10.3

Sweden

8.7

10.0

11.3

12.0

11.7

12.1

Switzerland

11.0

11.8

12.5

13.0

12.7

12.6

United Kingdom

9.1

10.3

11.6

12.3

13.1

13.3

United States

10.2

11.3

12.2

12.6

12.5

13.2

immigration. The share of the labor force born in other (less educated) countries, which had fallen from 21 to 5.4 percent during the Great Leveling, was back up to 15 percent by 2005. The key factor here, of course, is that immigrants tend to have fewer years of labor-market-useful schooling than do native born. Another reason for the decline in American relative rates of school completion may have been the operation of what economists call the “cobweb” model. In this model, people decide how much they will supply of something that takes time to deliver—like school completion—on the basis of the price or return it fetched in recent years, when potential students were deciding whether or not to commit more time to schooling. The lower the recent return, the more students dropped out. This in turn contributed to a schooling shortfall and consequent rise in the returns to extra schooling. The power of the cobweb model depends on whether the decisionmaking lags fit the data. If it is to explain why the schooling credentials of the adult labor force grew more slowly in the 1980s and 1990s, what are the earlier dates for which the changing population of young adults should have perceived lower returns to education when growing up? The cobweb model does not tell us what the precise length of decisionmaking lag is. Recent work by Goldin and Katz, though, suggests a time line on which the cobweb effect might have worked. Their estimates of the rate of return suggest that the lower returns reaped by young men from extra schooling attained in the 1950s through 1970s (figure 9-5) could have contributed to the slowdown in educational attainment in the 1980s and 1990s. If so, did the rise in returns to extra education for the graduation years between 1990 and 2000 increase the subsequent growth of accumulated schooling? We will return to this early twenty-first-century acceleration later in this chapter. Comparisons with other countries indicate that the slowdown in American adults’ education in the 1980s and 1990s could indeed have helped raise earnings inequality. The rate of schooling advance for ten data-supplying countries between 1977 and 2007 is associated with a lower inequality of labor earnings (figure 9-6). Since 1977, France and a few other countries with faster growth in adult education attainment managed to avoid rapid widening of earnings inequality. Faster growth in education also tended to check the rise of the top centile’s share of 232

•

CHAPTER 9

Rate of return for a year of school

0.15

0.13

0.11

College 0.09

0.07

High school 0.05 1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

Figure 9-5. Rates of Return for an Extra Year of School, for a Young Man in the United States, 1914–2005

the total incomes, though the correlation is not as strong as the effect of education on labor earnings shown in figure 9-6.17 The other education trend that matters is the disappearance of America’s long-standing international advantage in schooling. Figure 9-6 suggests as much for ten advanced countries, and table 9-2 supplies even more such data for twenty-three countries. All but Germany and Switzerland have been accumulating formal education faster than the United States since 1970. The contrast could be dismissed by the belief that the United States simply hit some education ceiling. That view receives little support from table 9-2, however. Even in the United States, the level of education attainment has accelerated in the most recently documented decade, contrary to the fi xed ceiling thesis. Furthermore, other countries have begun to surpass US education levels: Canada, Korea, and the United Kingdom have already done so, and Australia and 17 The negative correlation between adult education growth and the share of the top percentile in total incomes was statistically insignificant (−0.19) for thirteen countries in the 1970– 2010 period. The thirteen countries supplying the necessary data on top income shares for these years were the ten countries in figure 9- 6 minus Italy, but with the addition of Denmark, New Zealand, Norway, and Switzerland.

RISING INEQUALITY ONCE MORE

•

233

Rise in wage ratio (top 10% / median) 1977–2007 (percentages per decade)

8

Australia USA

6

UK

4

Sweden

Canada Italy

2

Japan Netherlands Finland

0

France –2 0.4

0.6

0.8

1.0

1.2

Growth in education attainment, 1970–2010 (extra schooling years per decade) Figure 9- 6. Rise in Wage Inequality, 1977–2007, Compared with the Growth in Adults’ Schooling Attainment, in Ten Countries

Japan have pulled about even. Thus, the recent schooling trends point to one reason why some other countries have achieved faster GDP growth and a slower rise in inequality. Are other countries also surpassing the United States in the quality of education? Those decrying America’s loss of schooling leadership also worry about a decline in US schooling quality relative to that achieved elsewhere. If correct, the link between slower educational advance and rising income inequality would be strengthened. Has American inequality been rising faster because Americans are learning less than foreign students for each year of schooling? There is good evidence about international quality differences today, but much less evidence about the past. Every four years since 1995, the Trends in International Mathematics and Science Study reports results

234

•

CHAPTER 9

for math and science in fourth and eighth grade, and every three years since 2000 the OECD’s Programme for International Student Assessment has tested fifteen-year-olds on their math, science, and reading achievement levels. Both cover dozens of countries, and both claim to have sampled school districts in a random and representative fashion.18 The United States has typically occupied only a middling rank on these international tests. The latest Programme for International Student Assessment tests, from 2012, place the United States twentysecond of the twenty-three countries listed in table 9-2, just ahead of Sweden and about even with Russia. There has been no trend over the past twenty years in America’s relative position, either in average test performance or the inequality of test performance.19 Did the relative quality of American education drop some time before the 1990s? Actually, we question whether there was ever a period in which American primary and secondary education were “better” than the majority of advanced nations. What made the United States a leader in education was not its quality but rather its enrollment and attendance rates, or what we might call the “quantity” of schooling per student year. There are some fragments of testimony supporting this longer-run conjecture about educational quality. Tocqueville commented that “the human mind [is] more superficially instructed in the United States than in Europe.” His travels in the 1830s showed him a broad but not deep education of free whites: If [the observer] singles out only the learned, he will be astonished to fi nd how few they are; but if he counts the ignorant, the American people will appear to be the most enlightened in the world. The whole population . . . is situated between these two extremes.20 18 One can question whether randomness has been achieved in all cases. For example, China is represented only by officially selected school districts in Hong Kong, Macao, and Shanghai, all of which scored well. Also, bear in mind that the international tests evaluate only some standard elements of curriculum learning, and that they test only those in primary and secondary school. We have no international standard for evaluating university education, other than the market verdict delivered by international students. 19 International math achievements actually date back earlier, to the 1960s. Yet those earlier tests covered too few countries and were too selective in their choice of school districts to offer a verdict on overall movements in the quality of learning. 20 Tocqueville (1839) 1963, 326–27.

RISING INEQUALITY ONCE MORE

•

235

In other words, a high share of free white Americans were schooled (e.g., literate and numerate), but with few being “learned.” This suggests that the quantity, not the quality, of schooling was what set early Americans apart. One mechanism by which schooling quality may have lagged behind quantity is what Carole Shammas has called “educational sprawl” in the antebellum United States.21 While counted as enrolled in school, free American children had mediocre attendance rates, spreading their schooling across the childhood years, with many interruptions in attendance. Such indirect evidence suggests that perhaps the quality of American learning per student year was never above that of western Europe. Given that assumption, America’s relative standing is best measured by the quantities we have reviewed in this and the preceding chapter: the United States’ lead in the average years of educational attainment has declined since 1970, offering another reason why the inequality of earnings and income rose more in America than in other advanced countries. Shifting Trade Winds One of the most popular explanations for the lack of wage gains in the lower ranks since the 1970s has been globalization. Many have argued that the wages of ordinary labor have failed to make much progress over a third of a century because we live in a world of increasing international mobility of goods, people, and capital. Three kinds of globalization are most relevant to income distribution in the United States and other OECD countries—imports, labor migration, and financial markets, and the first two have their main source in the Third World.22 Trade globalization as rising import competition. Trade globalization stems mainly from long-overdue institutional improvements in the emerging countries whose manufacturing competitiveness is finally catching up. While Hong Kong, Japan, South Korea, and Taiwan led the way a bit earlier, the rest of Asia waited until after the 1970s to shed the 21

Shammas 2014. We omit any discussion of the effects of internationalized investment on the home country’s income inequality, owing to the intractability of measuring such effects. For a broader survey of the impact of globalization on inequality within and between countries since 1820, see Lindert and Williamson 2003. 22

236

•

CHAPTER 9

barriers that a long history had imposed on their economic growth and export competitiveness. Only at the end of the 1970s did China initiate the economic reforms that have erased the blunders of the Great Leap Forward and the Cultural Revolution. And only in the early 1990s did India implement the reforms that removed the “license raj” restrictions on imports and new competitive businesses. The same could be said of Mexico and Southeast Asia. Such catching up and improved competitiveness have been major forces destroying many low-skilled jobs in the United States and western Europe. Understanding that trade globalization has sprung from belated growth in developing countries, and not from trade policy reform in advanced countries, helps sharpen explanations for job loss in manufacturing. It also helps us judge whether trade globalization will continue to contribute to American inequality in the future. Liberalizing our international trade policies cannot have caused much of the destruction of low-skilled manufacturing jobs because the European Union and the United States already had low trade barriers in the 1970s. Nor will further American trade liberalization have much effect on inequality in the years ahead. Rather, the surge in the supply of imports that create competition with low-skill manufacturing jobs in advanced countries stems from something that the OECD countries cannot and should not reverse: productivity growth in developing countries. How much has the economic success of developing economies contributed to the rise in American wage inequality through trade competition? This issue was debated intensely in the 1990s. An estimate that seems to capture the middle ground attributes 15 to 33 percent of the widening US wage gaps to the rise in import competition from developing countries. Note that this is a share of wage inequality only, not total income inequality, and only for the period up to the early 1990s. Within these limits, it leaves 67 to 85 percent of the explanation for other forces, like political shocks, labor force growth, the fi nancial boom, and education slowdown. The share awaiting other explanations would be even greater for the total income inequality.23 23 For a survey of the wage inequality literature up through 1996, see Cline 1997, especially table 2.3 and the surrounding text. For a more recent survey, with deeper coverage of certain econometric issues, see Feenstra 2000, particularly the editor’s introduction as well as

RISING INEQUALITY ONCE MORE

•

237

Globalization as rising immigrant competition. If rising trade competition was due to supergrowth in a few rapidly industrializing countries, the increase in immigration after the 1970s owes much to the slowness of wage progress in developing countries. Poor job prospects in less developed countries gave young adults a strong reason to migrate to Australia, Canada, New Zealand, the United States, and western Europe. And they did so in increasing numbers from the 1970s onward. As noted earlier, the US foreign-born share of the labor force rose from 5.4 percent in 1970 to over 15 percent around 2005. The arrival of extra immigrants has contributed to the rise of US earnings and income inequality by lowering the average education and skill levels, just as it did from the Civil War to World War I (chapter 7). The approximate effect on US wage inequality has been estimated by Goldin and Katz. They find that the impact of immigration on relative skill wage premiums from 1980 to 2005 was larger than during earlier post–World War II periods, when immigration was much smaller. They also estimate that immigration was responsible for only about a 2.4 percent rise in the college pay premium over high school. Still, immigration can explain a much larger share (43 percent) of the gains from fi nishing secondary school.24 The rising tide of immigrants into western Europe has served to increase inequality there too, even though faster educational advance, progressive fiscal redistribution, and other forces have held back the rising inequality implications there. Immigration has thus been part of the story of rising US inequality since the 1970s, much as rising immigration was also part of the inequality story between the 1860s and World War I (chapter 7), and falling immigration was part of the Great Leveling story from the 1910s to the 1970s (chapter 8). A Deregulated Financial Boom Since the 1970s, the fortunes of the financial sector have continued to move in close step with the overall concentration of incomes into the top 1 percent, just as they did in the first great inequality surge up to contributions by Matthew Slaughter and James Harrigan. For the 15 to 33 percent figure, see Feenstra and Gordon 1999. 24 Goldin and Katz 2008, 308–9.

238

•

CHAPTER 9

1860.25 That correlation was again positive during the Great Leveling from World War I to the 1970s, as we noted in chapter 8. First international financial flows were interrupted by World War I, then bond and security markets collapsed after 1929. The government intervened in the 1930s with controls and regulations, such as the Securities and Exchange Commission regulation of asset markets and the Glass-Steagall firewall between commercial deposit banking and investment banking. The late 1970s and 1980s brought deregulation and soaring financial returns: world financial markets opened up again, and international capital flows boomed, but the international debt crisis of 1982 followed. The savings and loan sector also imploded in the 1980s following poorly supervised investment behavior. The financial sector boomed again in the 1990s with easy monetary policy, the technology investment boom, the repeal of Glass-Steagall restraints on investment banking, and deregulation of financial derivative markets. Easy credit and deregulation of derivatives again brought soaring fortunes to the financial sector until the mortgagerelated crash of 2008.26 While these waves of crises support the need for greater regulation, our immediate point is simply that the financial sector’s enormous gains continue to play a key role in the rise of the top 1 percent’s share of all incomes—not only from the 1970s to 2007, but even up to today, the Great Recession notwithstanding.

SUMMARY: WHY THE WIDENING? The rise of inequality since the 1970s has been much more broadly based than one would gather from observers who follow just the renewed concentration of income in the hands of those at the top. Inequality has also increased among those in the lower 90 or 99 percent of the income ranks, and real incomes at the bottom have barely risen. In addition, the 25 See the Phillipon and Reshef history of fi nancial sector pay, and its correlation with fi nancial deregulation since 1909, in figure 8-3 above. 26 On the international lending boom and the international debt crisis of 1982, see Obstfeld and Taylor 2003, 2004; James 2014; Sachs 1988; Eichengreen and Lindert 1990. For an autopsy on the savings and loan crisis, see Kane 1989. For perceptive chronicling and assessment of the subsequent financial booms from the 1990s through 2007, and the ensuing crisis, see Madrick 2011, 222– 404; Mian and Sufi 2014, 20– 41; Eichengreen 2015, 64–88, 184–212, 281–336.

RISING INEQUALITY ONCE MORE

•

239

recent rise of inequality has been much more visible in the Englishspeaking countries than in continental Europe or Japan. The underlying causes have come together like a perfect storm. Each of the six main causal forces played a role, either in explaining why US inequality turned upward in the 1970s or why inequality rose so much more in America than in other countries. Political shifts help to explain both the change in trend and the geographic differences: the rightward shift in economic policy, like the rise in inequality, was centered on the countries that voted for Margaret Thatcher and Ronald Reagan, stopping their long twentieth-century trend toward regulation, union power, the welfare state, and the New Deal. The rising rate of growth of the working-age population helps explain the geography of the switch to rising inequality, but not the turning point. The greatest widening of income gaps did indeed take place where the labor supply grew fastest, yet it is not true that all industrialized countries had faster labor supply growth in the new widening era than in the leveling era up to the 1970s. Similarly, the contribution of education to skills growth seems to have left a more visible imprint on the geography of the new inequality trend than on the contrast between the leveling era and the rising inequality since. In particular, education (school years attained, and perhaps even their quality) has risen more slowly in the United States than it has in other advanced countries, suggesting another reason why the average labor earnings have lagged behind and become so much more unequal. Two other factors—trends in technological bias, and shifts in import competition—are of greater help in explaining the turning point around the 1970s than they are in explaining any differences between countries in their inequality trends. There was a clear shift in technological change after the 1970s toward more automation and greater rewards for those who possess capital and skills. This technology is shared by all advanced countries, and thus plays little role in differentiating trends in income distribution. Similarly, the awakening of Asian trade competitiveness since the 1970s helps explain the end of the income-leveling era in all advanced countries. They all shared in the income distribution effects of this rise in trade competition from abroad.

240

•

CHAPTER 9

Finally, the rise of the financial sector, aided and abetted by the retreat from government regulation, clearly contributed to the shift in US distribution trends. Whether it did the same in other countries is less certain, since the literature has shed more light on the American experience. Will the American inequality rise continue? Will its inequality history continue to be exceptional? And how does American inequality relate to its world economic leadership? We turn to these questions in the next chapter.

RISING INEQUALITY ONCE MORE

•

241

CHAPTER 10

Inequality and Growth History Lessons for the Future

T

he new income history charted in this book reveals one way in which the United States is truly exceptional among advanced countries, and one way in which it is not. The widening of its income gaps between the rich and the poor is exceptional, but its per capita income growth is not. Both results will help in shaping our forecasts for the first half of the twenty-first century. What is clearly exceptional is that prolonged rise of income inequality from the colonial times to the present, with only two temporary interruptions. The much-discussed widening of our income gaps since the 1970s is only part of this longer history. What in the eighteenth century was called “the best poor man’s country”—when income was much more equally distributed than in Europe—might today be called “the best rich man’s country”—that is, the most unequal society in the postindustrial community.1 Will its inequality continue to pull away from that in Europe, Japan, and the rest? American inequality has been exceptional not only in the size of the gaps between what the top and the bottom receive in the marketplace but also in the lack of political will to transfer income from the rich to the poor. The American political process has not been nearly so progressive in its income transfers as has northern Europe. Furthermore, since the 1960s, the United States has also made relatively low investments in its young compared to its old.2

1 Among the countries having a greater post-fisc inequality than the United States, the only rich country that might be called “postindustrial” would be Singapore. 2 Here we refer to ratios of (subsidies per person in the targeted age group) to (GDP per capita), using international comparisons for 2010. Th ree other OECD countries that are similar to

Perhaps more surprisingly, Americans have not prospered any faster in terms of their average purchasing power per person than have residents of other rich countries. While American real income per person has maintained its lead, this chapter will show that its income advantage today is not much different than it was in the previous three centuries.

WILL AMERICAN INCOME GAPS CONTINUE TO WIDEN? As we saw in chapters 8 and 9, six forces shaped income inequality trends across the twentieth century, and the same six should have a similar impact on inequality in this century. What makes inequality predictions difficult, however, is the uncertainty that surrounds the future of some of these six basic forces: political shocks, including wars; growth in the working-age population; growth in labor skills; changes in import competition; biased technological change; and changes in the financial sector. All the usual forecasting dangers apply, so our inequality projections must be cautionary, conditional, and limited to the next few decades. Political Shocks and Wars Political change helps explain both the timing and the geography of inequality. No explanation of why the Great Leveling occurred between World War I and the 1970s can ignore the rise of mass voting power, the displacement of empires, and the leftward shift in political preferences. The massive income redistribution from white southerners to emancipated blacks in the 1860s was also implemented in the political arena. The same can be said of the high school movement in the early twentieth century and the immigration quotas imposed during the 1920s and kept in place until the 1960s. The geography of inequality over the last two centuries—especially the North–South gaps—also reflects political forces. That the United Kingdom and the United States have exhibited a greater rise of inequality than other advanced countries correlates with a shared rightward shift during the Reagan and Thatcher years. the United States in tilting toward favoring the elderly are Greece, Italy, and Japan. See Lindert 2014; Arroyo Abad and Lindert, forthcoming, figure 3.

INEQUALITY AND GROWTH

•

243

Since future political influences on inequality are highly uncertain, we turn to the remaining five basic forces for answers to the conditional question: If the politics of redistribution remains the same, what will the inequality trends be like up to 2050? Slower Growth of the Labor Force Demographers can generate good population forecasts for a generation or two into the future. While there are always uncertainties about future birthrates, death rates, and especially, immigration rates, demographers can predict the size of any age group in 2050 far more accurately than anybody can predict political or technological change. The US Census Bureau projections of the fifteen- to sixty-four-year-old age group are added to the American population history since 1774 plotted in figure 10-1. That history reminds us that American labor supply growth, which was once several times faster than that of western Europe, has slowed down dramatically over two and a half centuries. The Census Bureau predicts that the slowdown will continue. The workingage population will grow only modestly from now until 2050. These working-age population projections are driven primarily by the postwar baby boom. The first baby boom birth cohort, born in 1945, started to pass age sixty-five in 2010. The last cohort, born in 1963, will celebrate its sixty-fift h birthday in 2028. The rapid outflow of baby boomers from the under-sixty-five age group should generate a slowdown in the growth of the working-age population. These projections also imply that once all the baby boomers have passed the age of sixtyfive, the rate of growth of the working-age population might pick up again, however slightly. But it will continue to grow much more slowly than in the American past, and about as slowly as it is growing today in Europe and Japan. Of course, the population in the fifteen to sixty-four age range is not necessarily the same as the labor force. More and more of those in their late teens and early twenties will elect to stay in school. Moreover, the retirement age could rise above sixty-five. Even so, these demographic forces should lower labor force growth rates and thus dampen forces contributing to the widening of income gaps for the reasons we have already given in previous chapters. 244

•

CHAPTER 10

40

Percent increase per decade

30

United States 20

10

0 1800 (from 1774)

1860 1870 (from 1800)

1910

1970

2010 2030 2050 End of period

Figure 10-1. Growth in the Working-Age Population, United States, 1774–2050

The Census Bureau’s projections also may have missed something that could make labor force growth slow down even more. Using direct extrapolations from the recent past, the Census Bureau estimates that US gross immigration will rise by 51 percent from 2014 to 2030, implying that the foreign-born share will rise. However, there are many conditions in emigrant source countries that are likely to check that extrapolation. The determinants of Third World emigration include each sending country’s cohort size for those in prime migration age, its income per capita, its education, its poverty rate, the existing share of the sending country’s population that is already in the host country, and the immigration policies of the United States and other host countries. Armed with an international panel of migration histories, Timothy Hatton and one of the present authors generated explanations for recent migration flows, and then used them to predict immigration up to 2030–2034.3 The recent history suggests that Asia, Latin America, and the Middle East will send fewer migrants (as a percent of the US population) 3

Hatton and Williamson 2010.

INEQUALITY AND GROWTH

•

245

to the United States in the years ahead. Migration from sub-Saharan Africa will grow somewhat, but that region still supplies less than 10 percent of the US total immigration. Hatton and Williamson suggest that immigration to the United States may drop by 6 to 7 percent up to 2030–2034. These calculations reinforce our prediction that the US will see slower labor force growth and weaker pro-inequality forces in the near future. The possibility that the supply of immigrants will decline seems less likely for Europe than for North America. Migration from sub-Saharan Africa and the Middle Eastern war zone is still on the earlier upslope of what Hatton and Williamson call the migration life cycle. Since it is closer to the source, western Europe will continue to feel rising immigrant pressure the most. Western Europe already faces a hard choice in its immigration policy. Any tightening of restrictions would produce more domestic equality only by making average incomes more unequal internationally. The other outcome would be that the rising tide of immigration into Europe would be greater than tighter restrictions can hold back. In this case, greater immigration would be a mechanism generating more income inequality in Europe, thereby making it more like the United States. Labor Skills: A Rise in US Schooling Growth? Might working Americans get more schooling in the future? The trends since 2000 suggest as much, since they show an upturn in educational attainment growth after the slowdown in the 1980s and 1990s (figure 8-8, table 8-2, and table 9-2). The share of GDP spent on education also rose by 1.1 percentage points after 2000. Of this, the majority was funded publicly.4 A continuation of this recent trend would serve to slow down the rise of inequality. Its effects will be delayed, of course, since extra 4 For the shares spent on education, see Organisation for Economic Co-operation and Development 2013, tables B2.1 and B4.2). Recent media coverage has emphasized the privatization of universities through the elevation of fees, especially in Chile, South Korea, the United Kingdom, and the United States. Offsetting this policy shift in the United States, however, has been a rise in the generosity of mean-tested fee remissions, both by the government and top US universities. Meanwhile, primary and secondary education continues to have its slowly expanding budgets funded mainly by taxpayers (Organisation for Economic Co-operation and Development 2013, sec. B5).

246

•

CHAPTER 10

student learning today does not affect the distribution of earnings until some years later, when those students participate in the labor market. The chances of accelerating public commitment to education are a bit better in some other developed countries than they are in the United States. Such has already been the trend for some decades. If this continues, a number of countries will enjoy better equality and growth outcomes than the United States. Import Competition and International Trade Trends in the world trade environment seem to be the most predictable source of rising US inequality. Pessimists claim that low-skill jobs in the United States have been lost to foreign trade competition. The pessimists are correct, but we should be clear about the causes of this rising lowskill job competition. As we noted in chapter 9, what has been destroying American low-skill jobs since the 1970s has not been a shift toward free trade policies. The United States and other rich countries already liberalized international trade so much by the 1970s that there was little left to liberalize. While proposals for new trade agreements still draw heated opposition from threatened labor groups, the basic fact remains that the lower-skill jobs in high-income countries are in danger, even though trade barriers fell long ago. The threat to low-skill jobs in US industries producing tradable goods has not been trade policy. Instead, such jobs are being lost to newly exporting countries, particularly in Asia, because these countries are finally repairing their institutions and becoming more competitive. The leading conqueror of international markets in manufactured goods is China, the new “workshop of the world.” Consumers in Europe, North America, and elsewhere have become accustomed to seeing “made in China” on every manufactured good. China’s rising competitiveness and export growth have been unprecedented. Yet there is no reason to think that China has discovered economic institutions that the United States should emulate, or that China will surpass American GDP per capita any time soon. After all, their comparative advantage in labor-intensive manufactures is eroding and will continue to do so. Indeed, their competitive success has caused wage rates to soar in INEQUALITY AND GROWTH

•

247

Chinese industry, thereby reducing their competitive edge over American workers in the labor-intensive industries that produce tradable goods. But marching right behind China are many still-poorer Asian and African countries ready to take China’s place at the bottom of the low-skilled, labor-intensive end of the industrial ladder. As soon as these latecomers join the modern economic growth club, their rising exports will continue to widen pay gaps within rich countries like the United States. Biased Technological Change Will technological change continue to favor the skilled and disfavor the unskilled, as it has since the 1970s? Media coverage of automation, robots, and the digital age suggest that the verdict is already in, and that we are doomed to live in a world where a small group of high-skilled winners takes all. Yet we should reserve judgment on whether to expect the further displacement of low-skilled labor from future exogenous (independently caused) technological changes. We see nothing predictable about future breakthroughs in information technology, for example, that will disfavor the hiring of low-skilled labor more than of skilled labor. Yes, jobs have been lost more rapidly in the lower-skilled and older job categories. Yes, there has been a trend toward outsourcing low-skilled jobs to other countries. Nevertheless, in our view this is a development driven largely by the belated upsurge in foreign productivity and competitiveness, and not simply an exogenous global change in technology.5 For example, while increased outsourcing since the 1970s can be partially explained by the arrival of new information technologies that allow the long-distance monitoring of Asian production, we think there is a more powerful explanation. Had today’s more stable and efficient institutions prevailed in China a hundred years ago, the production outsourced to China could have been monitored a hundred years ago, even without computers and satellites. Telegraph and telephone could have done the job. Thus, we see part of the apparent labor-saving bias of technological change since the 1970s as an induced response to the international 5

248

Feenstra 2000; Feenstra and Gordon 1999, 2003.

•

CHAPTER 10

changes just discussed, and only part of it as an exogenous—and still hard to predict—determinant of future trends in inequality.6 The State of the Financial Sector Long-run trends in the inequality impact of the financial sector are hard to predict. It certainly rises in “quantity,” relative to the rest of the economy, as measured by the long upward rise in the wealth–income ratio. Yet its “price,” the rate of return on capital, keeps dropping. We see no long-run rise in the share of the financial sector in the economy, or its effect on overall inequality. Still, even without an ability to predict the secular trend effect of the financial sector on inequality, we are much more certain that it will contribute to volatile swings in GDP and inequality. Throughout the history of capitalism, the tension between those who trumpet the productivity gains from active, innovative, and free fi nancial markets and those who point to the damage caused by financial booms and busts was present even before the Industrial Revolution. Will the Great Recession of 2008–2011, triggered by an underregulated financial market, generate a political and institutional response much like that of the 1910s–1970s era? We accept Eichengreen’s prediction that the financial sector will remain underregulated and unstable. We also accept his reason for believing so: the very success of the Federal Reserve’s vigorous lending, the federal government’s stimulus package, and bailouts in keeping the Great Recession of 2008–2011 so much milder than the Great Depression of the 1930s has sapped political will for tighter regulation. Hence, we, like Eichengreen, foresee recurring rounds of financial booms stoking an inequality rise, followed by financial busts when inequality drops or stops rising.7 Our predictions about income inequality up to 2050 are thus a guarded mixture based on the movements in these six basic forces: 6 For an informed weighing of the predictable components of technological change (and factor bias) in the near future, see Gordon 2016, 579– 604. 7 On the rise of wealth–income ratios, see Piketty 2014, 164–98; World Top Incomes Database, http://topincomes.g-mond.parisschoolofeconomics.eu/ (accessed August 23, 2015). On the long history of fi nancial boom and bust, see Kindleberger 1978; Neal and Williamson 2014. On the shortfall of the regulatory response to the Great Recession, see Eichengreen 2015, 281–336.

INEQUALITY AND GROWTH

•

249

• Unpredictable political shocks, including wars • A slowdown in growth in the working-age population, which should lower inequality slightly up to 2028 • Renewed growth in labor skills, which should lower inequality slightly • Trade competition from lower-wage countries, which should worsen inequality for a long time • Biased technological change, which we consider largely unpredictable, but partly a by-product of the previous force • Unpredictable shifts in the finance sector, which should cause swings in inequality

The balance of these forces will fluctuate, creating episodes of rising and falling inequality. American history tells us to expect inequality episodes, not a monotonic march driven by some law of capitalist development.

HOW HAS AMERICA’S INTERNATIONAL LEADERSHIP CHANGED? What impacted America’s domination as a rich country has been nearly the opposite of what happened to the rich within America. It was during the Great Leveling era that international GDP per capita gaps reached their greatest heights, with the United States at the top, well above Britain. The renewed rise of income inequality in the United States since the 1970s was accompanied by a fall in America’s global advantage in income per capita. The earlier trends in the eighteenth and nineteenth centuries also produced a contrast between the position of this rich country and that of the rich within it: while inequality within the country was rising dramatically over those earlier centuries, our average income advantage over northwestern Europe stayed much the same. Our new history of American incomes allows us to take a fresh look at the history of this country’s leadership. The Long View: American versus British Incomes since 1700 How American real incomes compare with Britain can now be traced over more than three hundred years. The result is not what previous scholarship has taught. 250

•

CHAPTER 10

When an economic history course begins to discuss economic leadership, the curtain rises with Britain’s Industrial Revolution, in which the Americans play little part. By the end of the Industrial Revolution around 1850, Britain, we are told, was the clear GDP per capita leader.8 Then, the instructor explains, Britain was awakened by America’s unexpected innovations on display at the Crystal Palace Exhibition in 1851. We are also told that America was starting to catch up. The AngloAmerican comparisons for the late nineteenth century tell us that America was overtaking British in manufacturing technology and trade competitiveness. Britain fell farther behind in the two world wars and the first postwar generation, when its industry was seen as particularly anemic and uncompetitive. If one were to bundle these narratives together into a long history of relative income growth, the impression would be that Britain was well ahead up to 1850, but that America has outperformed Britain ever since. What is missing from this traditional comparison of American and British income per capita is the great depth of America’s three economic crises—the Revolutionary War and early independence, the Civil War, and the Great Depression of the 1930s. In all three, income per capita fell dramatically in the United States relative to Britain. To see how we have come to our revisionist position, let us again pick up the tool we used earlier in this book when comparing American income per person with the United Kingdom up to 1870. The tool is used to divide each country’s nominal income by the cost of a fi xed bundle of consumer goods and services in the same year. Th is comparison of purchasing power in the same year for all countries makes sense because it stays within the same historical context. It avoids using the cost of bundles from some distant time and place, like that of 1990 international (Geary and Khamis) dollars.9 When comparing the ability of Americans and others to purchase what they wanted in, say, 1860, this comparison is faithful to the fact that 8 And Britain may have been so even before the Industrial Revolution since recent research has repeatedly lowered the estimated growth rate during the Industrial Revolution, not by cutting the estimated output per capita in 1830 or 1850, but by raising the estimates of earlier levels, such as in 1700 or 1750. See, for example, Crafts 1987; Crafts and Harley 1992; Broadberry et al. 2015. 9 We are referring here to Maddison’s (2010) extensively used database.

INEQUALITY AND GROWTH

•

251

Table 10-1 American Real Income per Capita, Relative to Great Britain, 1700–2010 Purchasing power of American income per capita, relative to Great Britain = 100 (a) Using a bare-bones consumer price deflator, Philadelphia

(b) Using a respectability consumer price deflator, Massachusetts and Pennsylvania

1700

136

1725

155

1750

149

1774

152

1800

92

87

1850

141

124

1860

150

138

1870

103

(c) Using expenditure-based national purchasing power

1950

164

1990

152

2010

130

they used candles rather than fluorescent lighting, and ate mutton rather than pizza.10 American average purchasing power relative to Britain can be found in table 10-1 and figure 10-2. For the period 1700–1870, we can compare America and Great Britain in terms of the ability to buy a bundle of staple goods, which Allen has called a bare-bones or bare subsistence bundle. To avoid the danger of viewing all national income in terms of its ability to buy a poor household’s bare-bones market basket, we also display the American and British comparisons in terms of the ability to 10 Here, as in chapter 5, we are applying the same measurement strategy as Ward and Devereux (2003, 2004, 2006), Robert Allen and his coauthors (2010, 2012), and the direct-price comparisons in Feenstra, Inklaar, and Timmer (2015). Ward and Devereux’s comparisons (2003, 840; 2006, 252) also yield a loss of US income per person leadership around 1870, and a peak in that leadership in the 1950s, even though they are using quite-different data sets.

252

•

CHAPTER 10

200

American income per capita, relative to Britain = 100

175

(a) Using bare-bones bundle of goods 150

125

(c) Using national expenditure deflator

100

(b) Using respectability bundle of goods 75 1700

1730

1760

1790

1820

1850

1880

1910

1940

1970

2000

Figure 10-2. American Real Income per Capita, Relative to Britain, 1700–2011

buy a more respectable living standard. As noted in chapter 3, this second kind of comparison is available just for a few nineteenth-century dates.11 The price deflator for the years from 1950 on is based on the entire bundle representing all national expenditures on goods and services. Our history of comparative purchasing power yields some striking results (figure 10-2). In spite of all the familiar stories of overtaking and surpassing Britain, the United States advantage in purchasing power is today no greater than before independence. But things were different in between. The Americans spent most of the intervening three centuries growing faster than Britain, yet suffered those three massive setbacks visible in figure 10-2: the Revolutionary War era, the Civil War decade, and the Great Depression of the 1930s—a depression much deeper in the United States than in Britain.12 11

For the relative prices of the respectability bundle of nine goods, see appendix D, table D-4. If the drops in figure 10-2 seem abrupt, note two features of this chart. First, it compresses 312 years into a single graph, so that a drop occurring over a decade (e.g., 1860–1870) or a quarter century (1774–1800) might seem almost instantaneous. Second, the first two drops in this American–Britain ratio reflect British progress as well as American decline. 12

INEQUALITY AND GROWTH

•

253

Postwar Convergence among the Leaders For the years since the early twentieth century, figure 10-3 compares the purchasing power of average US incomes with those of many other countries, not just the United Kingdom. Even among this larger group, it is clear that World War II created the biggest gap between America and the rest. The exceptional American moment arrived in 1945, at the end of World War II. Equally obvious is the reason why: Asia and Europe were destroyed in that war. Yet across the postwar decades, Europe and Japan gained on the United States: France, Japan, Sweden, and the United Kingdom all were catching up on US income levels until the late 1970s. Since then, the gap has been fairly stable. Were there other countries, not shown in figure 10-3, with higher average incomes than the United Kingdom, or even than the United States? Switzerland, another country not ravaged by World War II, has had a slightly higher average purchasing power than the United States in most years since 1930.13 As of 1980, Switzerland was still the only rich country richer than the United States, although five other countries (Australia, Canada, Iceland, New Zealand, and Sweden) were ahead of Britain. As of 2011, the residents of Norway and Singapore also had greater average purchasing power than residents of the United States. But the central message from figure 10-3 is this: Japan and western Europe converged on the United States up until the 1970s, although there’s been little convergence since. One might ask whether different consumer bundles, or even different income concepts, might change this comparative income history. While we do not delve deeply into this question here, we can show that a study based on a different consumer bundle and unskilled wage rates yields a similar leadership history. Figure 10- 4 compares the purchasing power of wages over a century and a half, drawing on a study by one of the present authors. Here too, the lead of the United States over the United Kingdom is much the same today as it was 13 The Swiss–US comparison is from Maddison (2010) for 1920–1949, and Penn World Table 8.1 (http://www.rug.nl/research/ggdc/data/pwt/pwt-8.1 [accessed August 23, 2015]) for 1950– 2011. Th is paragraph’s list of richest countries ignores microstates based on oil or fi nancial havens, such as Bermuda, Brunei, the Emirates, or Hong Kong, though we do include Singapore.

254

•

CHAPTER 10

World War II

200

Real incomes per capita, relative to United Kingdom = 100

United States 150

Sweden 100

France Japan 50

China India 0 1920

1930

1940

1950

1960

1970

1980

1990

2000

2010

Figure 10-3. American Income Leadership and Convergence, 1920–2011

before the world wars. Again, those wars, especially World War II, catapulted the United States into exceptional world economic dominance. Overall, the leading countries have been converging since the Second World War, but that convergence has been more dramatic for real wages than for income per capita, especially after the 1970s.14 The steeper rise in US inequality since the 1970s is quite consistent with that result. Thus, the United States has now fallen back to the moderate income advantage that it had from colonial times to 1914. The two world wars made America economically dominant in the twentieth century, and since then it has fallen back to the long-run peacetime norm.

14 What other data-supplying countries had unskilled real wage rates above that of the United Kingdom? Two other migrant-destination countries did: Australia from 1870 (or earlier) through 1945, and Canada since 1871. Within western Europe, the list includes Norway since 1910, Denmark since 1919, the Netherlands since 1922–1945, Sweden since 1946, and Germany since 1969. See, again, Williamson 1995.

INEQUALITY AND GROWTH

•

255

Purchasing power of an urban pre-fisc unskilled wage rate, relative to UK = 100

World War I

World War II

3.0

2.0

USA Sweden UK = 100

1.0

France 0.0 1825

1850

1875

1900

1925

1950

1975

2000

Figure 10-4. Real Wages Relative to the United Kingdom, 1830–1988

Global Economic Leadership How does this twentieth-century convergence among the rich countries fit into the global history of international per capita income gaps? Is there a hollowing out where the leaders form one rich club, leaving poor countries farther behind? The answer has been yes during the globalization waves of both the nineteenth and twentieth centuries; in 1870– 1914 and the last quarter of the twentieth century, gaps opened up between the richer West and the poorer rest. Yet, and as we all know, China and India both began to experience rapid growth in per capita income starting in the 1980s and 1990s. Their populations are so large that any judgment as to whether income gaps between rich and poor countries are shrinking depends mainly on the performance of those two countries. According to that criterion, the early twenty-first century has brought some global convergence: China and India are gaining on the United States.15

15

256

Lindert and Williamson 2003; Milanovic 2005, 2009, 2015.

•

CHAPTER 10

RAISING THE AMERICAN INCOME FLOOR Is there anything the United States could do to diminish its upward inequality trend without eroding its income leadership? Tying the equality and growth goals together, as we have done implicitly throughout this book, is especially appropriate for the decades since the 1970s. The share of those in poverty is greater in the United States than in most rich countries, even using a common absolute consumption standard to define the same poverty line for all countries.16 Having greater absolute poverty than many other rich countries while also having a higher average income prompts a search for ways to promote equality without harming growth. Assembling a Tool Kit for Raising the Income Floor So what could be done to diminish income inequality without reducing the average income per person? The direct transfer approach would guarantee a minimum income for all and a tax on the rich to pay for it. That direct transfer route, however, probably would reduce work effort, investment, risk taking, and hence GDP, as suggested even by the literature that shows no direct effect on these by the overall welfare state package. One must always keep in mind that the likely “GDP growth free lunch” of the welfare state refers to the whole welfare state package, in which different negatives and positives cancel out. Thus, while direct transfers might be part of a policy package that makes people more equal without harming growth, it’s the rest of the package that is crucial to supporting growth. History suggests a set of policies consistent with egalitarian growth— a set drawn from the basic causal forces we have identified to explain twentieth-century inequality trends.17 Among the many policies that could be used to promote income equality without compromising 16 Th is remarkable result has emerged consistently from the Luxembourg Income Study ever since the appearance of estimates by Timothy Smeeding and his colleagues (1990). For updates and time series on relative poverty rates, see http://www.lisdatacenter.org/ (accessed August 23, 2015). 17 For a broader menu of egalitarian growth options than we give here, see Gordon 2016, 643–52.

INEQUALITY AND GROWTH

•

257

economic growth, we emphasize three that have emerged from our new history of American incomes: • Improve public education in order to raise the floor on labor skills • Tax inheritances so as to diminish inequality persistence across generations • Regulate the fi nancial sector to reduce GDP volatility that taxes the nonrich to bail out the rich, and makes the poor also pay by unemployment

Improvements in public education. The obvious way to equalize incomes while also raising the average is to raise schooling achievement at the bottom by improving public education at the preprimary, primary, and secondary levels. The United States once led in this policy strategy, which allowed this country, along with Canada and Prussia, to pull ahead of other countries in the nineteenth century. Since the 1970s, as we have seen in chapter 9, others have caught up to the United States in cumulative years of schooling, and are ahead of the United States in achievement test scores. All studies of the social returns on early schooling continue to yield high rates of return. One might infer that the United States should ramp up its public spending on education at the preprimary, primary, and secondary levels. The inference is valid, but history points to some caveats. Do extra public expenditures on mass schooling really keep children in school longer and make them learn more effectively? The answer has clearly been yes, both for the United States and other countries. Extra public money did raise student performance in the United States, at least for those attending secondary school in the 1960s or earlier. Since then, however, it is not clear that spending more each year has delivered either extra years of schooling or better learning. The “does money matter” debate remains unresolved for the United States.18 It is for this reason that we have titled this first tool “improvements in public education” rather than “increased budgets for public education.” Whether 18 For the positive effect of extra public school spending on students up to the secondaryschooling cohorts of the 1960s, see Loeb and Bound 1966. For a still-apt summary of the “does money matter” debate on school fi nance, see Burtless 1996. For a recent study fi nding that court-ordered expenditure increases do raise student performance, see Jackson, Johnson, and Persico 2014, 2015.

258

•

CHAPTER 10

or not the improvements involve expanding budgets, there can be little doubt that one way to raise the income floor is to raise the floor on their learning. Inheritance taxation. This type of taxation continues to be urged as an egalitarian mechanism, and rightly so. Its egalitarian effect is noticeable, especially for the strength of its effect as a commitment to equality of opportunity. Any society wanting to claim that “in our country, individuals make their own way, with their own hard work and abilities,” should favor a high level of inheritance taxation in order to honor that claim. As Andrew Carnegie famously declared, The parent who leaves his son enormous wealth generally deadens the talents and energies of the son, and tempts him to lead a less useful and less worthy life than he otherwise would. . . . [W]ealth left to young men, as a rule, is disadvantageous.

Recent empirics seem to confirm this “Carnegie effect”: namely, that excusing heirs from taxation has significantly negative effects on growth by cutting their work and income19 Three East Asian countries and Germany have some of the world’s highest rates of inheritance taxation plus a more equal distribution of earnings, with no obvious growth disadvantage. The marginal tax rate on inheritances is 50 percent in Germany, Japan, Korea, and Taiwan.20 By contrast, the top rate is only 45 percent in France; 40 percent in the United Kingdom and the United States; followed by lower top rates for Denmark, Finland, Italy, Norway, and Switzerland; and zero in the Netherlands, New Zealand, and Sweden.21

19 For this famous passage, see Carnegie 1901, 54–55. For a study of the different negative effects of inheritance on heirs’ labor income and working hours, and the positive effect on their take-up of early retirement options, see Bø, Halvorsen, and Thoresen 2015, and earlier studies cited there. 20 Note an upcoming test case of East Asian taxation of top inheritances: when Korea’s ailing patriarch Lee Kun-hee of Samsung dies, his heirs may have to pay $6,000 million at the statutory 50 percent inheritance tax rate. 21 Ernst and Young 2013. The Swiss system is administered at the canton level, with rates ranging from 0 to 55 percent. As for top income taxes, our three East Asians have top rates above the world median and are similar to other leading countries. Examples from 2010 were: Japan = United Kingdom = 50 percent; Taiwan = Switzerland = 40 percent (with France at 41 percent); and Korea = United States = 35 percent (KMPG Global 2015).

INEQUALITY AND GROWTH

•

259

The historical experience of the leading industrialized countries yields no evidence that higher inheritance taxation or, for that matter, higher marginal tax rates on top wealth have slowed down GDP growth. Indeed, as we noted in chapter 8, Piketty has shown that such top tax rates peaked in the era of fastest growth in GDP per capita. For the United States, those top marginal tax rates were as high as 77 percent on inheritances (1942–1977) and 90 percent on incomes (1952–1964). Experience from the fast-growing postwar years shows no lost GDP from taxing the wealthy heavily. Regulation of the financial sector. We see little reason to doubt that tighter financial regulations would reduce inequality and raise the longrun trend in national income per capita. Recall from chapter 5 that financial development outpaced GDP growth before the Civil War, correlating with rising inequality. Chapters 8 and 9 further noted the tight connection between financial deregulation, higher incomes in the financial sector, and a higher share of the top 1 percent in national incomes. For a verdict tied more closely to regulation itself, note the half century of financial stability and relatively equal incomes ushered in by the 1933 Glass-Steagall Act and related legislation setting up closer auditing and an enforced separation of commercial banking from financial investment. Until the repeal of the Glass-Steagall Act in 1999, the only financial crises were the international debt collapse of 1982, and the savings and loan meltdown of the 1980s, both of them caused in part by deregulation. While more income equality is not a direct outcome of tighter financial regulation, the income floor under those near the bottom would be raised by the prevention of unemployment caused by financial busts. What is required is the political will to go beyond the partial regulatory provisions of the Dodd-Frank bill of 2010. Reducing the costs of financial busts at the expense of curbing financial booms or bubbles would help reduce inequality over the long run, just as it did from 1933 into the 1970s. It might also raise income per capita growth by minimizing downward departures from full employment GDP.

260

•

CHAPTER 10

No Necessary Conflict between Equalizing Incomes and Raising Them To lay a factual foundation to the argument for raising the American income floor, we need to sweep away the remnants of an older view that policies cannot promote both equality and growth. The older view assumed an “efficiency-equity trade-off.” If such were true, then nothing could be done to foster economic growth without the collateral damage of greater inequality, or greater equality without the collateral damage of less growth. History does not confirm such a trade-off. To remember why, fi rst consider a simple point about the political process—one that we noted long ago.22 A dominant historical outcome has been that vested interests have blocked initiatives that would promote growth and/or equality.23 A conspicuous example is the suppression of mass public schooling—an investment that clearly promotes both equality and growth.24 Our second consideration comes from the numbers: history does not record any correlation—negative or positive—between income equalization and economic growth, either in our new American history over the past 360 years or world history over the past 150 years. The correlation does not emerge, regardless of whether “growth” means the GDP per capita growth rate or its absolute level, and regardless of whether “equalization” means the share of social spending in GDP, some measure of policy-induced redistribution, the level of pre-fisc income inequality before taxes and transfers, or even the rate of change in any of these. Economists have explored the effects on income per capita growth of three kinds of egalitarian variables: tax-based social spending and its composition; fiscal redistribution, measured by the gap between preand post-fisc inequality; and the greater equality of pre-fisc incomes before taxes and transfers. An empirical literature using contemporary 22

Lindert and Williamson 1985, 369–70. For a summary history of the prevalence of vested interest obstructionism and the power of “extractive institutions,” see Acemoglu and Robinson 2012; Engerman and Sokoloff 2012. 24 On the blockage of efficient and egalitarian education, see Lindert 2004, chap. 5; Go and Lindert 2010; Ramcharan 2010. 23

INEQUALITY AND GROWTH

•

261

world evidence finds that the growth effect of equalizing incomes is not significant. History agrees. American experience does not reveal any clear effect on GDP of greater tax-based social spending or more progressive redistribution from rich to poor. Indeed, recent analyses suggest that greater pre-fisc equality has a positive effect on growth.25 This result supports the argument that egalitarian investments in human capital simultaneously achieve more equality and more growth. While these statistical results can be and have been debated, they do not support any claim that equalizing incomes must lower growth. American income history offers no support either. *** If there were any fulcrum at which historical insight might be applied to move inequality, it would be political. As we have said, no nation has used up all its political opportunities for leveling income without harming economic growth. Improving education, taxing large inheritances, and taming financial instability with regulatory vigilance—the opportunities are there, like hundred dollar bills lying on the sidewalk. Of course, the fact that they are still lying there testifies to the political difficulty of bending over to pick them up. 25 For important results on types of expenditure and taxation, see Ostry et al. 2014; Kneller, Bleaney, and Gemmell 1999; Gemmell, Kneller, and Sanz 2011, 2014a, 2014b. These studies fi nd that public investments, including human investments, fi nanced by indirect taxes have positive growth effects, while pure transfers fi nanced by direct taxation have negative effects, without any overall significant effects of just larger government budgets. For a summary and extension of earlier tests, see Lindert 2004, chaps. 10 and 18. On the growth effects of greater fiscal progressivity, Ostry, Berg, and Tsangarides 2014 get a near-zero result. On a positive growth effect of more equality in pre-fisc incomes, see Easterly 2007; Ostry, Berg, and Tsangarides 2014.

262

•

CHAPTER 10

APPENDIX A

A Guide to the 1774 and 1800 Estimates

H

ow did we get the income numbers that chapters 2 through 4 have displayed for 1774? Most of this appendix guides the reader through the steps we used in measuring national income, its component parts, and its distribution among American colonial households in 1774. We illustrate the buildup of total incomes for six types of colonial households on the eve of the revolution. Each step of the way, we link the six illustrations to the fuller files available at http://gpih.ucdavis.edu (hereafter cited by the file name and just http://gpih). Then we turn more briefly to the 1800 estimates, which use different sources to build similar income aggregates.

THE INCOME CONCEPT National income, or national product, is the pie that the economy produces and divides up each year. It can be measured in any of several ways. Since the world has become most familiar with GDP, we have chosen to measure a pie that closely resembles it. The income concept that most accurately describes our measures is gross national income (GNI). Note the three operative words. “Gross,” in that depreciation is included and not netted out. “National,” meaning that we follow the incomes of persons whose primary residence is in this country or colony, and not the incomes generated by what is produced here, which would be “domestic.” And “incomes,” rather than product, because we want to focus on the division of the national pie among rich, middle, and poor households, instead of the division of production among such output sectors as agriculture, industry, and services.

Yet in practice, there is little harm in casually equating our GNI with GDP and also some other national aggregates. Even today, with our elaborate corporate forms of business and the globalization of ownership, the different concepts stack up about the same. Even for 2010, if we take the size of GDP as 100, then our GNI equals 100.3, or essentially the same magnitude. Other concepts stack up similarly: gross national product (GNP) was 101.3, and gross personal income was 98.3. Back before 1870, the match was even better, since the household sector earned practically the whole of GDP or GNP. Another convenient feature of the early data was that there was little need to distinguish between pre- and post-fisc measures of income. That is, the disposable (post-fisc) income one received after paying taxes and receiving transfers and the benefits of government services was almost exactly the same as what one received in the marketplace before dealing with the government (pre-fisc). The reason is simply that government was then so tiny, especially in Britain’s tax-free colonies on the American mainland.1 Thus, we can casually talk about “income” as though it referred to any of these familiar accounting concepts before the twentieth century. We have assembled the GNI as the sum of the parts listed on the righthand side of figure A-1. The dominant share, 69 percent for the thirteen colonies as a whole in 1774, consisted of free own-labor earnings—a concept that here includes farm operators’ residual profits.2 To these will be added property incomes and slaves’ retained earnings.

BUILDING THE INCOMES OF SIX KINDS OF HOUSEHOLDS IN 1774, STEP BY STEP To describe each of these components by discussing its raw-data inputs, listed on the left-hand side of figure A-1, let us follow six kinds of households from the larger set of ninety-four household types whose incomes 1

For the tiny magnitudes of American colonial taxation, see Perkins 1994; Rabushka 2008. Logically, the same concept should have included the residual profits, above all factor incomes, for nonfarm enterprises as well as farms. Yet it is only for farms that we are given anything resembling data on a pure-profit residual for 1774 and 1860. 2

264

•

APPENDIX A

New occupation counts + Wage-salary rates by occupation (esp. from J. T. Main) + Some assumptions about work days/year + Farmer profit residuals

Nominal own-labor incomes of the free

69%

+ New occupation counts + Alice Hanson Jones’s wealth estimates by occupation + Assumed gross rates of return

Nominal property incomes

23%

+ Slaves’ retained earnings

8%

= Gross national income (GNI) Figure A-1. Assembling the Parts of National Income in 1774

are estimated in detail online.3 Three of these six illustrative types are northern, and three are southern. Specifically, the six are: • A Boston common laborer • A New England small-town shopkeeper • A propertied widow in Philadelphia • A middling rural southern planter, where middling means that the planter’s household ranked between the fortieth percentile and the eightieth percentile in terms of farm wealth • A free southern male household head (HH) for whom no occupation was given, but who was listed as having positive wealth • A slave household in the rural South 3 Follow the blue-shaded cells for these six cases, in “American Incomes 1774, Baseline PartTime Assumptions,” Worksheet (3), http://gpih. These cells report the same numbers as are presented in table A-1 here.

THE 1774 AND 1800 INCOME ESTIMATES

•

265

We follow the data-input steps listed on the left-hand side of figure A-1, showing how these steps produce the numbers given for our six illustrative kinds of households in table A-1 below. Counting occupations of free household heads in 1774. It is important to count household heads (HHs), both because eighteenth-century data sources tended to do so and because of our interest in economic inequality. Such counts emerge from a mixture of local censuses, urban directories, probates, and tax lists. The occupational detail is best for cities. Thus, table A-1 is perhaps close to the mark in suggesting that Boston common laborers headed 164 households, or that there were 2,560 widows with positive property who headed households in Philadelphia. For most small towns and the countryside, we lack direct observations of occupations, and must make guesses based on “cloning” the occupational mixes offered for Lancaster and rural Chester County in Pennsylvania, with some help for all regions from Main’s readings of the local press and from Jones’s occupational estimates for her probate work.4 We therefore attach less certainty to the estimated numbers of Yankee small-town shopkeepers (2,560 in table A-1). Of course, in the colonial countryside, most free HHs were farmers or planters, helping us weigh this occupational category. Our 40,503 “middling” southern planters were the middle 40 percent (fortieth percentile up to 79 percent, ranked by property). The 6,922 southern free male HHs for whom we have no occupational data is a provisional number based on local tax lists and Jones’s guesses. Finally, for slaves, our sixth group in table A-1, we know the total population numbers quite well, thanks to local censuses. The harder part is judging how many households or houses the slaveholders grouped their slaves into. Slaves were generally not allowed to live as family units, nor did the censuses ever count slave families. As elaborated in appendix B, we use some early nineteenth-century information to estimate the number of slave households as 35 percent of the population age ten and up. Counting the 1774 labor force. To know which parts of the population were employed, one starts with the population counts themselves. The 4 For a summary list of the occupational sources used, see table 2-1; Lindert and Williamson 2013b, table 1. For cities, see “Charleston Population and Occupations, 1790, 1774,” “City Populations and Occupations, 1746–1795,” and “Philadelphia Occupations, 1772, per B. G. Smith,” http://gpih.

266

•

APPENDIX A

Table A-1 Six Examples of Calculating Household Incomes in 1774 (Incomes in Dollars at $4.44 per Pound Sterling) Type of household head (HH)

Notes

(a) Boston common laborer

(b) Yankee small-town shopkeeper

(c) Philadelphia widow (wealth > 0)

(1) Number of such households

164

2,560

231

(2) Persons in labor force

369

4,525

231

2.25

1.77

1.00

(3a) Per household, full time

310.09

391.54

157.16

(3b) Per household, part time

219.94

391.54

157.16

0.71

1.00

1.00

(3c) Total value, full time

50,855

1,002,385

36,236

(3d) Total value, part time

36,070

1,002,385

36,236

(4a) Per household

12.26

132.64

46.50

(4b) Total value

2,011

339,574

10,721

(5a) Per household, full time

322.36

524.18

203.66

(5b) Per household, part time

232.20

524.18

203.66

(5c) Total value, full time

52,866

1,341,959

46,957

(5d) Total value, part time

38,081

1,341,959

46,957

Ratio, labor force to HH Own-labor earnings

Part to full ratio

NIPA-type property income

Total income

(continued)

Table A-1 (cont.) Six Examples of Calculating Household Incomes in 1774 (Incomes in Dollars at $4.44 per Pound Sterling) Type of HH

Notes

(d) Southern, middling, rural planter

(e) Southern free male, no occupation, wealth > 0

(f) Rural southern slave

(1) Number of such households

40,503

6,922

88,938

(2) Persons in labor force

58,338

6,922

228,698

1.44

1.00

2.57

(3a) Per household, full time

403.90

106.84

117.10

(3b) Per household, part time

359.47

106.84

117.10

0.89

1.00

1.00

(3c) Total value, full time

16,359,251

739,538

10,414,917

(3d) Total value, part time

14,634,474

739,538

10,414,917

(4a) Per household

182.48

18.12

0.00

(4b) Total value

7,391,004

125,438

0

(5a) Per household, full time

586.38

124.96

117.10

(5b) Per household, part time

541.95

124.96

117.10

(5c) Total value, full time

23,750,255

864,976

10,414,917

(5d) Total value, part time

22,025,477

864,976

10,414,917

Ratio, labor force to HH Own-labor earnings

Part to full ratio

NIPA-type property income

Total income

Note: NIPA stands for national income and product accounts, as noted in chapters 1 and 2.

population censuses for seven colonies provided breakdowns by slave/ free status and crude age category (e.g., just over versus under sixteen years of age).5 We then adopted Weiss’s plausible estimates of labor force participation rates (LFPR) in 1800, with helpful breakdowns by state, gender, slave/free, and three age groups (those under ten, those ten to sixteen, and those sixteen and up).6 Placing the nonheads in households. Of those who were in the labor force, one faces a double task. First, one must decide—or assume—what occupations they held, even though occupational labels were typically given only to HHs. The nonheads were generally either unskilled or tenant farmers, or both, but with exceptions. Second, one must decide how to allocate them among households defined by the head’s occupation. Here we had to make some particularly arbitrary assumptions. Some are obvious enough, such as assuming that working slaves lived only in slave households. Among the free, we generally allocated those in the ten to fifteen age groups to the unskilled-occupation category of the same sex, while we allocated those nonheads who were sixteen years or older to the prevailing local mix of commercial, craft, farmer, and unskilled occupations.7 Wages and salaries by occupation in 1774. As already noted, we drew different occupations’ pay rates for free workers from the few available sources, most notably Main’s compendium The Social Structure of Revolutionary America. And as discussed further in appendix B, we adjusted these for the number of days worked per year, and whether or not the workers received board and/or lodging separately from their cash pay.

5 The six colonies not supplying age distributions were generally southern: Delaware, Georgia, North Carolina, Pennsylvania, South Carolina, and Virginia. For these we had to assume age distributions, guided by their initial federal censuses and the nearest colonies supplying age brackets for the 1770s. 6 See “American Labor Force by Colony” and “Estimated Mix of Occupations, 1774, by Region,” http://gpih. These labor force participation rates, kindly supplied by Weiss himself in Excel form, represent slightly updated versions of the rates he described in Weiss 1992. By “labor force,” we mean all persons whose labor participated in generating product sold in significant part (or for slaves, demanded in significant part) outside the household. Note that we equate labor employment with labor force participation, since the colonial data sources lacked any clear distinction between the two. 7 “Estimated Mix of Occupations, 1774, by Region,” Worksheet (3), http://gpih.

THE 1774 AND 1800 INCOME ESTIMATES

•

269

For a male common laborer working in Boston around 1774, our first illustration, we have averaged six wage estimates from four sources. Four of the six estimates were daily rates, and required adding board, while two were annual rates. The simple average pay of these was $125.37 full time (313 days), or $89.01 for a part-time year of 222 days.8 Estimating 369 members of the labor force, with the same unskilled pay rates as the 164 unskilled HHs, implies table A-1’s $310.09 per household full time, or $219.94 per household part time. The case of the Yankee small-town shopkeeper is one for which we cannot observe a wage or salary directly, since HHs in the commercial sector were self-employed. Main offers us only two income observations on anybody close to this category, and does so only for two Bostonians. How to estimate the labor component for New England shopkeepers outside Boston? For such cases, we had to use a social ladder technique drawing on their relative wealth positions between higher and lower groups. A shopkeeper tended to be in between wealthier top professions (local officials, lawyers, and the like) and less wealthy artisans. Apportioning the shopkeepers’ imagined labor-and-profit earnings between those observed for higher professionals and artisans in the same proportions as their wealth observed by Jones yields the $391.54 as their annual labor earnings, inclusive of pure entrepreneurial profits. Similarly, for our average Philadelphia widow in table A-1, we again lack direct wage rates. Indeed, we have no occupation for her other than “widow,” though she could well have been an innkeeper, and we know from Jones’s data that she had some positive wealth. We have to position her relative to others of neighboring economic strata, for which her relative wealth was a clue. Her labor earning power, like her wealth, was seemingly above that of a younger female HH earning $59.94 a year if employed full time in Philadelphia, yet below a similarly unskilled male in Philadelphia ($318.91). Giving her the same share of the distance between these two other groups, in earnings as in wealth, we estimate her earnings at $157.16. In fact, to judge from a few wealth returns, we estimate the same for a Philadelphia male who had some wealth but no 8 To these six observations for Boston common labor in the 1770s, we could have added William Weeden’s (1890) data recording about two shillings a day in the 1770s, which would work out to a similar wage rate.

270

•

APPENDIX A

stated occupation. We will use the same procedure for table A-1’s rural southern male who has some wealth but no stated occupation. Free farm labor earnings and farm profits in 1774.9 For any farm operator, such as the southern middling rural planter in table A-1, we again have the task of valuing the labor of somebody self-employed, along with the labor of their household members. Our first step is to partition all their own-labor earnings between the farm-labor component and all the rest of their efforts. For the unskilled labor earnings component, we multiply the total number of household members’ days of labor by the local farm wage. Of course, the farm wage must be adjusted for the payment or nonpayment of board. Next one must put a value of earnings on the residual own-labor category for farm operators. This residual profit was the reward for their risk taking and the farm management skills exercised either by the operators themselves or hired overseers within the household. Note that this return was over and above the competitive return on such marketable farm assets as real estate, equipment, livestock, product inventories, and slaves. Valuing these farm profits is exceedingly difficult. For 1860, we have the benefit of a regression technique for inferring such a farm profit residual from a large agricultural sample, thanks to a study by Lee Craig, as we will see in appendix E.10 For 1774, our only source is a section of Main’s book that wrestles with the returns that farmers got other than just their labor or normal return on farm assets.11 His gleanings from newspapers, personal journals, and estate accounts are not easy to interpret, presumably because the primary sources themselves were designed to serve different purposes. We interpret these passages of Main’s book as correctly trying to refine the “income” measure by excluding property income and the wages of hired labor. Our interpretation is supported by his referring to several of these estimates as ones that could apply equally to a tenant farm operator, who had to pay for the use of land. His passages seem to suggest a profit per farm of £16 ($71.04) for 9 See “Own-Labor Incomes, 1774,” Worksheet (2), or “American Incomes, 1774, Baseline Part-Time Assumptions,” Worksheet (2), http://gpih. 10 Craig 1993. 11 J. Main 1965, 104–14. For enormous plantations in the Chesapeake area, one can also consult the wide gaps between recorded revenues and recorded expenses for several estates; for details, see Walsh 2010, 449–56, 556–71.

THE 1774 AND 1800 INCOME ESTIMATES

•

271

New England, £19.27 ($85.58) for the Middle Colonies, and £45 ($200) for the South.12 These profits have considerably enhanced our baseline estimates of free own-labor earnings, both overall and in table A-1’s illustration of a middling southern planter. They are large, though not radically higher as a share of all farm income than we will find for the northern farms of 1860 studied by Craig (appendix E) and for southern farms in the Ransom and Sutch sample for 1880 (appendix G). While adding the profits to the imputed farm labor earnings is consistent with other practice, it occurs to us that the primary sources might have meant that such profits or income were a remuneration for the farm operator’s own labor as well as for risk taking and management skills. If so, one could argue for subtracting his imputed farm wage in order to avoid double counting the reward to his labor when adding in the profit. This interpretation would imply that after adding in profits, one should deduct an annual farm wage of $137 per farmer for New England, $120 for the Middle Colonies, and $110 for the South. Th is would be a sizable reduction, given that free farm operators were the largest single occupational group in each colonial region. As table A-2 reveals, this alternative interpretation would reduce GNI by as much as 12.4 percent. Still, it would not remove any of the contrasts being featured for 1774, such as the greater purchasing power of the mainland colonies relative to the mother country, or low level of income inequality. Nor does it reduce the income advantage of the southern colonies. We retain our baseline interpretation that the profits described in Main’s sources should be added to the imputed free-labor earnings to derive free own-labor incomes. Slaves’ retained incomes, 1774 and 1800.13 The slaves’ own-labor incomes for 1774 are built on those for 1800 and later years, which are described more fully in appendix C and several downloadable http://gpih files cited there. That appendix begins by appraising the literature on the 12 Our Middle Colonies figure of £19.27 ($85.56) is based on his separate estimates for Delaware, New Jersey, New York, and Pennsylvania. For the “Middle Colonies,” probably excluding Delaware, he alternatively gave an average profit of £24 (J. Main 1965, 108). 13 We use the term “retained incomes” for continuity with our later placement of these incomes into the income-distribution ranks. It should be remembered, however, that incomes were the same thing as consumption for slaves, who are hardly allowed to save anything. And the adjective “retained” is not meant to imply any decision making on the part of the slaves themselves, unlike corporations’ “retained earnings.”

272

•

APPENDIX A

Table A-2 Free Farm Incomes in 1774: Baseline Results and an Alternative (Dollar Values in Thousands) New England

Middle colonies

South

Thirteen colonies

Aggregate farm free-labor earnings

16,164

5,912

21,116

43,193

Gross property income

4,519

5,370

25,270

35,159

Farm profits

4,023

2,763

20,328

27,114

Total farm income, baseline

24,707

14,045

66,715

105,466

Free farm households

56,628

32,288

101,642

190,558

Free farm labor force

100,103

44,848

146,600

291,551

GNI (full time)

36,059

38,281

98,814

173,154

GNI (part time)

33,038

36,653

94,095

163,786

Minus farmer’s imputed labor earnings, full time

−7,762

−3,871

−11,181

−22,813

Total farm income, alternative

16,945

10,174

55,534

82,653

Full time

−21.5

−10.1

−11.3

−13.2

Part time

−20.9

−9.4

−10.6

−12.4

A. Farm incomes, baseline

B. Denominators

Percent change in GNI implied by the alternative

Notes: See “Own-Labor Incomes 1774,” Worksheet (2), http://gpih. The middle colonies here include Delaware, New Jersey, New York, and Pennsylvania, conforming to Jones’s practice. Part-time labor earnings = 0.89 times full-time labor earnings. Gross property income is derived from real estate, reproducible nonhuman capital, and slaveholding. See “Property Income Files for 1774,” http://gpih. The table excludes slave HHs and slave retained earnings.

“exploitation rate” around 1860—a literature magnified by some claims that Robert Fogel and Stanley Engerman made in Time on the Cross. From there we work back to 1800, using information about antebellum trends, and drawing on the useful Fogel and Engerman sample of slave hires and slave asset prices in Queen Anne’s County, Maryland, in THE 1774 AND 1800 INCOME ESTIMATES

•

273

1796–1804. For our 1800 benchmark, we compared different ways of calculating slaves’ retained earnings and the implied share of their marginal product extracted by slaveholders. When combined with the asset valuations of slaves, these implied an annual rate of return of 11 percent, including depreciation.14 For 1774, the estimates of slaves’ retained earnings start with applying the 1800 shares of retained earnings, or one minus the exploitation rate discussed in appendix C, to the 1774 free-labor wage rates representing slaves’ marginal product by occupation and region.15 These retained incomes per earner need to be checked against other information from the 1770s. The main competing data specific to that early setting are slave prices collected by other scholars. These are used for cross-checking because our derived slave incomes plus the exploitation rate will imply an annual flow of property income to slaveholders. We explore whether the ratio of our implied property incomes from slaveholding to the asset valuations of slaveholding yield a plausible rate of return. The comparison does not yield a perfect fit. The crucial rate here, the implied southern rural 15.8 percent rate of return, is above the 11 percent rate of return from Maryland slave hires in 1796–1804. We cannot yet judge whether this represented a true change from 1774 to 1800, or some net error. For now we simply note that our 1774 estimate might understate the slaves’ retained incomes.16 Property incomes in 1774. To generate property incomes for 1774, we start with the wealth data in Jones’s sample of 919 probated estates, reweighted to give an estimated population of living HHs. Part of that wealth would have generated the kinds of property incomes that enter into the national income and product accounts (NIPA). For each such NIPA-type asset, we apply a net rate of return plus a rate of depreciation, to derive net and gross property incomes. The net rate of return is taken as 6 percent for both 1774 and 1800, though we will use 5 percent

14 For the resulting estimates of slaves’ retained incomes in 1800, see appendix C; “OwnLabor Incomes, 1800,” Worksheet (3), http://gpih. 15 For the 1774 estimates, see “Own-Labor Incomes, 1774,” Worksheet (4), http://gpih. 16 Such a bias, if it exists, would mean that we are understating not only slaves’ retained earnings but also national income. The understatement of national income would arise because we are sticking with the 11 percent rate of property return for slaveholders, not the 15.8 percent implied by our low estimate of slaves’ retained earnings.

274

•

APPENDIX A

for 1850–1870, given the movements of interest rates over time.17 The rates of depreciation, or “capital consumption allowances,” are set as follows for NIPA-type assets: 0 percent for financial assets, real estate, producers’ perishables, business inventory, and crops in the field; 5 percent for slaves in the labor force; and 10 percent for producers’ durables, business equipment, and livestock. We ignore any returns on the following non-NIPA asset categories given in Jones’s data set: consumers’ durables, consumers’ perishables, cash, bad financial assets, household equipment, and apparel.18

FREE-LABOR EARNINGS AND PROPERTY INCOMES IN 1800 Our source materials for ca. 1800 are summarized in table A-3, using the same format as in table 2-1. Free-labor earnings. The primary sources and procedures for quantifying the rates of pay are the same for 1800 as for 1774. The secondary sources are different, however. No longer can we use Main’s Social Structure of Revolutionary America. For the most common kinds of wages and salaries, we can still draw on several scholars’ work, though we lack good coverage of the higher professions’ salaries.19 Property incomes ca. 1800. In 1798, Congress passed its first federal direct tax. Each state had to pay a quota based on its total population (free persons plus three-fift hs of all “others”). Within each state, this one-off tax was levied on real estate wealth and numbers of slaves. The revenue was to fund a possible conflict with France.20 The 1798 federal tax returns remain by far the most useful source available for estimating the property income component of 1800 national income. True, one might view these returns with some suspicion. Can we trust the quality of the data collected by tax authorities representing a new nation that 17 Farley Grubb (2013) has found many sources that seem to support using the 6 percent interest rate, or net rate of return, shortly before and after the revolution. See also Homer and Sylla 1991, 276–99, 286–88. 18 Jones (1977, 1980) chose these asset categories. 19 See “Wage Rates, 1800,” http://gpih; D. Adams 1968, 1970, 1982, 1986, 1992; T. Adams 1944; Blodget 1964; Lebergott 1964; Simler 1990; B. Smith 1990, 110–21; C. Wright 1885. 20 For the best introduction to the quantitative dimensions of the 1798 direct tax returns, see Soltow 1989. For the underlying political history, see Einhorn 2006.

THE 1774 AND 1800 INCOME ESTIMATES

•

275

Table A-3 Main Data Inputs for 1800 Income Estimates Data sources and adjustments for occupational shares A. Population, labor force US census, labor force participation rates for 1800 supplied by Weiss, expanding on his estimates in Weiss 1992. B. Occupations of HHs and the labor force • City directories and tax lists for Baltimore 1799, Boston 1800, Charleston 1800, Hartford 1799, New York City 1799, Norfolk 1801, Philadelphia 1800 • Town directories and tax lists for Lancaster, PA, 1800; Lexington, KY, 1806; Pittsburgh 1815 • Rural tax lists from Burke County, GA, 1799; Chester County, PA, 1799–1802 C. Free-labor earnings and farm profits Same sources and methods as in table 2-1, except no coverage of ca. 1800 by Jackson Turner Main (1965), and we can offer no estimate of farm profits D. Slave retained earnings Sources and methods explained more fully in appendix C E. Property income The 1798 direct tax on real estates and slaves, via Timothy Pitkin (1817) and the Connecticut State History Museum. See also Soltow 1989; Einhorn 2006. We assumed similar local ratios of real estate and slave valuations to total property as in 1774. Note: See the details of these data sources and their use in http://gpih.

had just fought a bloody and expensive war to shed its imperial government partly over tax issues? This suspicion turns out to be warranted, especially given evidence that properties had already been underassessed in local tax returns from the previous two decades. For example, warned by Gerard Warden’s investigation of the Massachusetts 1771 tax rolls, we found implausibly low assessments not only in those rolls but also in the Philadelphia 1772 returns supplied to us by Billy Gordon Smith and the 1786 New York City returns supplied to us by Herbert Klein.21 While these city tax rolls were useful for identifying urban occupational distributions—including occupations revealed by the presence or absence of each asset type—they are not useful for estimating market values of assets. Thus, while the 1798 federal returns 21

276

Warden 1976.

•

APPENDIX A

are the superior source from which to construct 1800 property incomes, we must identify and adjust for their likely biases before using them. The 1798 direct tax probably underassessed real estate market values by something like 15.5 percent in New England and the Middle Atlantic—a figure based on a contemporary study of marketed real estate in Connecticut in that same year. We know this because Lee Soltow found correspondence in the Oliver Wolcott papers showing that for 518 Connecticut properties sold in 1798, the average ratio of federalassessed to market value was 0.845.22 Thus, we have raised our 1800 property income estimates by this 15.5 percent underassessment in New England and the Middle Atlantic, and also the 7 percent rise in average asset values from 1798 to 1800 suggested by the contemporary Samuel Blodget.23 In the South, the federal tax authorities undercounted the numbers of taxable slaves. They also appear to have underassessed rich households’ realty by even more than that 15.5 percent underassessment we have identified for the North. Elsewhere, we lay out the peculiarities of southern assessment for the 1798 tax and present our preferred estimates for the extent of the underassessment.24 In addition, the federal tax returns taxed slaveholding only as a fixed amount per slave in the able-bodied age category, and not on the basis of slave values. We have thus added slave values based on the market prices in the Fogel and Engerman sample for 1796–1804, as described in appendix C. The adjustment for valuing the undercount of slaves, plus the extra southern underassessment of real estate, raises real estate plus slave property income values by 30.1 percent for the South Atlantic. This combined with the nationwide underassessment of real estate by 15.5

22

Soltow 1989, 37, 256–57. Blodget (1806) 1964. 24 Lindert and Williamson 2013a, supplementary materials, appendix 3. Current calculations by Frank Garmon Jr. of the University of Virginia suggest that the undervaluation of real estate was about twice the 15.5 percent for the southern states from Maryland and Delaware down through South Carolina. On the other hand, he does not fi nd high undervaluations for Georgia, Kentucky, or Tennessee. The authors thank him for sharing these calculations from his ongoing project. 23

THE 1774 AND 1800 INCOME ESTIMATES

•

277

percent raises real estate plus slave property income by 40.4 percent for the South Atlantic, or 27.7 percent for the whole Eastern Seaboard. Since the 1798 federal returns covered only real estate and slaves, we had to use the same ratios of total property to (realty plus slave values) obtained from the 1774 evidence to inflate them to total property. We apply region-specific ratios to each of the three regions—New England, the Middle Atlantic, and the South.25 The resulting estimates of total property and property income around 1800 allow us to reconstruct the total GNI. We are also able to offer new estimates of the inequality of the total labor earnings and the inequality of property incomes, but not income and property together, as noted earlier. 25

For the detailed calculations, see “American Property Income Totals, 1798–1800,” http://

gpih.

278

•

APPENDIX A

APPENDIX B

Salaries, Payment in Kind, and Workdays

T

his appendix clarifies our approach to several labor market dimensions that affect our measurements of incomes and their distribution for 1774–1870.

WHITE-COLLAR SALARIES 1774 and 1800 Salaried white-collar occupations were mostly in cities and towns during the colonial and early federalist years, and their annual or monthly salaries are relatively well documented for the following: clerks for merchants, clerks for government offices, and shopkeepers and innkeepers; schoolmasters, headmasters, academy ushers, professors (Yale and Harvard), and private tutors; judges, attorneys, and lawyers; soldiers and officers; ship captains; and ministers, doctors, and treasurers. Our prime sources are Main and Stanley Lebergott.1 1850–1870 For the mid-nineteenth century, our sources are richer for white-collar jobs, and rely mainly on American Almanacs around our benchmark dates 1850 and 1860, with some use of government pay rates around 1870. Apart from those already listed above, for 1850–1870 we can add

1 J. Main 1965; Lebergott 1964. For details and additional sources, see “Wage Data Survey, 1774” and “Wage Data Survey, 1800,” http://gpih.

many more occupations—like public school teachers, other public employees, and the military—and thicken the samples for each.2

PAYMENT IN KIND Laborers working for farmers, crews living on ships, and domestics and others living with employers received in-kind income. According to Lebergott, “The most common method of wage payment in agriculture was monthly.”3 While board and lodging were included in the contract (and sometimes washing, mending, fuel, candles, and even borrowing the employers’ horse), the monthly dollar wage quotes refer only to the cash payments. But in-kind payments were not limited to contract labor living with employers. For instance, ministers received free lodging, and private tutors lived with the families of their students, as did female public school teachers in the 1850s onward. To take another example, day laborers typically received a big midday meal and even whiskey for “hard work.”4 Many secondary sources supply in-kind valuations by occupation and location.5 The ratio of in-kind values to cash payments varied—for example, female domestics were at 0.618, farm laborers and seamen were at 0.516, and ship captains were at 0.300. Where appropriate, we have applied such ratios to the dollar pay data for our benchmark estimates of 1774, 1800, and 1850–1870.

ADULTS’ ANNUAL WORKDAYS The Full-Time Employment Assumption Different occupations report wages and salaries for different time spans. Most professionals received annual rates of pay, as in Main’s 2 See “Wage Data Summary, 1850,” “Wage Data Summary, 1860,” and “Wage Data Summary, 1870,” http://gpih. See also appendixes E and F, especially for public school teachers. 3 Lebergott 1964, 257. 4 For lengthy descriptions of how we valued these in-kind payments, see “Wage Data Survey, 1774” and “Wage Data Survey, 1800,” http://gpih. 5 See Adams 1968, 1982; Lebergott 1964; J. Main 1965; Rothenberg 1985, 1988, 1992; C. Wright 1885.

280

•

APPENDIX B

white-collar pay for the late colonial era, or the American Almanac entries from the 1830s to the 1870s. Teacher pay was recorded by the school year (and required constructing second occupation pay for the “vacation” period). Farm operators might have worked a full year, though their work intensity varied by season and rainfall. Most hired labor wages, however, were reported by the day, week, or month. We have little information on how many days or months people worked per year, and what secondary income they could fi nd when not employed in their primary job. Our reading of Lebergott, Main, and others leads us to believe that something close to a full-time employment of 313 days a year could be a valid assumption for all employed persons in 1774 and 1800, even wage labor.6 Our preference would be to assume that in the early economy, with so much work being done at home, people put in a six-day week (the 313 days) year-round. Some of our calculations, including those on income inequality, are based on this full-time assumption for 1774. Yet we would not make the same full-time assumption about the mid-nineteenth century or later. Even for 1774 and 1800, we recognize the need to offer different assumptions about the work year for the purpose of sensitivity analysis. The Part-Time Alternative Assumption We recognize that others may prefer to assume that wage labor did not fi nd work for some of those 313 days, either in the colonial era or since. Th is might fit the NIPA conventions better than the full-time assumption, since NIPA typically omits the value of most work performed in the informal sector or at home. Accordingly, we offer this interpretation of the likely locus of reduced hours and reduced implicit pay rates. First, consider adjustments for unemployment. In fact, the adjustment is really only relevant for wage earners out for day hires. It was certainly not relevant for slaves, and probably not relevant for fishermen, farmers, and farm laborers on contract, saddlers, tinsmiths, potters, weavers, shoemakers, and other artisans, or male and female 6

Lebergott 1964; J. Main 1965.

SALARIES, PAYMENT IN KIND, WORKDAYS

•

281

domestics. Taken as a whole, such occupations accounted for most of the American labor force in 1774 and 1800.7 But that (small) part of the labor force most vulnerable to unemployment—urban and nonfarm, rural unskilled laborers—got much bigger from 1800 to 1860 as wage labor became more common and cities grew in importance.8 Yet 1800, 1860, and 1870 were not crisis years.9 The average unemployment rates for the decades 1800–1809 and 1850–1859 were, respectively, 2 and 4.5 percent.10 This implies that nonfarm hired labor worked 307 days per year in 1800 (313 * 0.98) and 299 in 1860 (313 * 0.955). These business-cycle-adjusted figures may underestimate the impact of unemployment since they do not include lost work due to bad weather for outside work and dry months reducing waterpower for mills. What about underemployment? When it was off-season in any occupation, the fisherman could become a carpenter, the farm laborer a construction worker, the female teacher a domestic shop worker, or later a factory operative, and so on. No doubt these secondary occupations were of lower earnings capacity than these people’s primary ones. Otherwise, they would have been their primary occupation. So how do we adjust for the “underemployment” rate (something we have already done for male and female teachers)? Were off-season secondary jobs 10, 20, or 25 percent less productive? And what was the length of the offseason or weather-induced layoffs? Contemporaries repeatedly asserted “that 4 months’ work a year was lost because of bad weather,” implying a reduction in the 313 full time days to 209 part-time days per year, a 33 percent reduction in days worked at the primary occupation.11 “ ‘All outdoor operatives,’ . . . wrote George Nettles in 1850, ‘are suspended during the 4 or 5 winter months,’ but many go into the South for employment,” implying a maximum reduction in the full-time 313 days per year to 183 days, or a 42 percent reduction.12 This may have been true of gristmills, oil mills, and ironworks, but off-season secondary jobs took up much of the slack.13 If we take the 7

Here, we are paraphrasing Lebergott (1964, 171–72). See chapter 5; Lebergott 1964, 172. 9 J. Davis 2004. 10 Lebergott 1964, 189, table 4– 4. 11 Quoted in Lebergott 1964, 144. 12 Quoted in Lebergott 1964, 170. 13 Lebergott 1964, 171. 8

282

•

APPENDIX B

maximum lost months figure of 42 percent and apply the maximum underemployment figure of 25 percent to it (0.42 * 0.25 = 0.105), then the 313 full-time days at primary work might be reduced to a primary job equivalent of 280 days (313 * [1 − 0.105] = 280). A Final Assessment for Adult Work Days per Year As we say in chapter 2, the alternative days worked per year “parttime” assumptions that seem most plausible to us are: 313 days for those employed in the professions, commerce, and skilled manufacturing artisanal jobs, and slave households; 280 days for those employed in construction trades (including shipbuilding), rural unskilled, and farmers. What about free, urban unskilled labor? Might the loss of work have been worse for the urban unskilled? Did they have a secondary occupation when unemployed in the primary occupation? Suppose not, so that they became the vagrants and beggars when unemployed. Do they get reduced by the four months, from 313 to 209 days? This seems doubtful. Hired rural unskilled males, who had farm-related alternative employment in the off-season, earned 79.2 percent (i.e., $127.86/161.40) of that which urban unskilled workers got, as a thirteen-colony average. Suppose that the entire urban wage premium was a Todaro-type compensation for less regular employment. Then the urban unskilled work year was only 79.2 percent of the rural unskilled work year, or 0.792 * 280 = 222 days a year. This array of annual working days—222 for the urban unskilled, 280 days for rural unskilled, farmers, and construction workers, and a full 313 days for slaves, professionals, merchants, and skilled artisans—is certainly wide enough to encompass such estimates for England; England in 1760 and 1771 averaged 278 days, and England in 1800 averaged 280 days.14 The part-time assumption would seem to correspond to the Weiss “narrow” definition of GDP.15

14 15

For these English estimates, see Broadberry et al. 2015, 264. Weiss 1992, 1994.

SALARIES, PAYMENT IN KIND, WORKDAYS

•

283

CHILDREN’S WORK AND EARNINGS Most of the available pay rates for early America refer to adult males or employees of unspecified gender. Those for adult females are identified only for a few occupations—public primary school teachers, manufacturing workers, domestic servants, and slaves. Except for manufacturing factories starting in the early antebellum period, those for children are typically not identified at all. This appendix explains how we used limited data to quantify the LFPR and earnings of minors. Data limitations forced us to devise different solutions for different benchmark dates. For 1774 and 1800 We lean on the region-specific labor force participation rate (LFPR) estimated by Weiss for the year 1800. For the free population, there are separate rates for children under ten (LFPR = 0), boys ten to fifteen, girls ten to fifteen, males sixteen and up, and females sixteen and up. All are well below one-half, except for the LFPR of males sixteen and over, which hovers around 0.9. For slaves, we use the rate of 0.9 for both sexes, ages ten and up. For children in the free labor force, the wage rate data do not allow us to distinguish by age. This would not yield any bias if the wage rates data were for the true average age mix. Yet to the extent that the wage rate data refer to adult males, applying the same wage to children will overstate earnings to some degree. For 1850–1870 For our benchmark years based on the IPUMS samples from the census, labor force participation is defined by the IPUMS itself. Thus for any age group, the LFPR is simply the rate that emerges from the weighted sample for any subgroup. For those minors in the labor force, we must estimate wage rates for those working on farms, those who are apprentices, and all others. For children working within the household of a farm operator, we assumed that their marginal products related to those of an adult male in ratios 284

•

APPENDIX B

derived statistically for northern farms from work by Lee Craig.16 For children hired out for other rural labor, the same ratios were applied to the adult hired-hand wage rates, as noted for 1850–1860 in chapter 5 and described for 1870 in appendix G. To extend Craig’s patterns for the North to the South and West, we used interregional ratios of wage rates for hired farmhands, applying these regional matchings: Marginal product in South or West

equals

Craig coefficient for the corresponding northern region

times

Adult farm wage ratio, (South or West region)/ (corresponding northern region)

Here the “corresponding northern region” is assigned on an east-towest basis: • South Atlantic uses Craig’s Northeast coefficient and relative wage • East South Central uses the Midwest coefficient and relative wage • West South Central uses the frontier coefficient and relative wage • Mountain or Pacific uses the frontier coefficient and relative wage

Regarding apprentices in 1850–1870, we have no direct data on their rates of pay, which would in any case include some hard-to-document in-kind payment. Yet we do know some of their other attributes from the 1860 IPUMS. For a set of two hundred apprentices in the Middle Atlantic, almost all were male (97.5 percent), and few were HHs (15 percent). They averaged 18.1 years of age, and had almost no reported wealth (average = ten dollars in total property). Apprentices’ marginal product and their pay net of direct costs to the master should have been below the local pay of a free adult, urban unskilled worker. So if we considered them to be in training for something paying like the region’s adult male building trade pay, then the apprentice pay share for a near-twenty-one-year-old should be less than the region’s urban (common male/artisan) rate of pay. To break these regional averages down by age, we can take advantage of the fact that the IPUMS gives the age of each apprentice. Jan Luiten van 16

Craig 1991.

SALARIES, PAYMENT IN KIND, WORKDAYS

•

285

Zanden, based on the western European experience, assumes these ratios of apprentices’ imputed income to the wage rates for male unskilled nonfarm labor, across the typical seven years of apprenticeship from the fourteenth birthday to the twenty-first: age fourteen = 0.20, age fifteen = 0.40, age sixteen = 0.60, age seventeen = 0.70, age eighteen = 0.80, age nineteen = 0.90, and age twenty = 1.00.17 So assuming that the apprentice received his in-shop current marginal product, and assuming that it equaled the unskilled wage rate for his age group, with his parents’ paying the separate direct payment to the master simply covering the master’s training expense, we can apply van Zanden’s relatives to the regional, adult male, unskilled urban wage, yielding marginal products and implicit pay for the apprentice. In our files estimating the incomes of individuals by region, we use assumed values for apprentices grouped by region, urban/rural, and age.18 The age ranges are the ones suggested by van Zanden: namely, age ≤ fourteen years old, age = fifteen, sixteen, seventeen, eighteen, nineteen, and age ≥ twenty. As for children working in the manufacturing sector, we have followed Clarence Long, applying a ratio of 0.33 for (child manufacturing wage/adult male manufacturing wage) if the child worker is zero to fifteen years old, but 1.00 if the child is sixteen or older.19 For children in manufacturing, we do not distinguish by sex (male wage = female wage under sixteen). Meanwhile, we apply a ratio of 0.506 for (adult female manufacturing wage)/(adult male manufacturing wage) wherever we lack separate estimates for adult females versus adult males. 17

van Zanden 2009, 134. For our resulting estimates, see “IPUMS, 1870, All Regions Wage Data,” Worksheet “Notes on Tough Cases,” http://gpih. 19 Long 1960, 144. 18

286

•

APPENDIX B

APPENDIX C

Estimating Slaves’ Retained Earnings, Colonial Times to 1860

H

ow much of their marginal product were slaves allowed to retain as “income” for their own consumption? This appendix notes some ways to define those incomes, and then explains how we have tried to measure them over the long slave era. We start at 1860, which was the final benchmark date of the slave era, and the focal point for the heated debate over the 1974 publication of Time on the Cross by Fogel and Engerman. From there, we move back to 1850, where our estimates of slave incomes reapply one of the methods described for 1860. Thereafter we use information on trends in the antebellum slave economy to arrive at estimates for 1800, helped again by the Fogel and Engerman samples of slave assessments and slave hires. Finally, we revisit the subject of slave incomes in the colonial era, already introduced in appendix A.

WAYS TO DEFINE SLAVE “INCOMES” Eclectic evidence can be used to support three alternative ways of measuring the annual “income,” or consumption, that slaves had to live on:1

1 Note that we are treating slaves as a part of the income-receiving human population. In this respect we are following Gallman’s preferred approach to national product accounting. As Sutch and Rhode note in the Historical Statistics (Carter et al. 2006, 3–16), one could alternatively follow a suggestion by Kuznets that slaves could be treated as intermediate goods, such as livestock. As Sutch and Rhode explain, that would mean removing slaves’ consumption, as an intermediate input rather than a fi nal good, and treating any accumulation of slave value as a type of

(1) Direct observations of typical slave consumption (2) Conjectures in the historical literature on the “exploitation rate,” and thus on the share of their marginal product that slaves retained (3) The difference between their likely marginal product and the rental value of exploiting them for one year

The three measures should give similar answers. Let “s” be the slave’s retained income or consumption—that is, income measure (1) above. A second approach starts with the value of the marginal product of the slave’s labor, “w,” which should have matched the wage that a free laborer would have received per annum for the same occupation and same place—a qualification to which we return below. One can then defi ne the exploitation rate Ex = [w − s]/w. If Ex and w are documented, then (2) slave income s = w (1 − Ex). For a third approach, when a slave was hired out by a slaveholder in the active slave rental market documented by Fogel and Engerman and others, the rental rate R would approximate [w − s], over and above the s that the lessee had to spend for the slave’s upkeep.2 Hence, (3) one can estimate slaves’ retained income as s = w − R. In symbols, then, our three alternative measures are s directly measured as consumption or “maintenance” s = w (1 − Ex) s=w−R

Different alternatives are workable depending on the evidence available at different antebellum dates. For 1860, we will note two plausible alternatives out of these three. For 1850, only one of the three will work. All three are available for 1800. For the colonial era, we use some backward extrapolations from 1800 based on slave prices.

capital formation. We have not followed the Kuznets approach here, since we are interested in the division of incomes among different human groups. 2 As cited in the text, the source is the pair of downloadable Inter-university Consortium for Political and Social Research data sets referenced here as Fogel and Engerman 2006. The rental rate R could depart from w − s owing to the asymmetrical risks faced by the lessor (the lessee could mistreat the slave or otherwise violate the contract) and the lessee (the lessor may have misrepresented the slave’s attributes). We assume that such risks balance out, so as to keep R = w − s.

288

•

APPENDIX C

DEFINING SLAVE “HOUSEHOLDS” One other conceptual issue must be confronted before slave incomes can be estimated: How is a slave “household” defi ned? This question arises because household members shared incomes, and something like the household is the unit most frequently used in the inequality literature. What was a slave household, and what incomes did its members share? Slave households were clearly not family units, since slaveholders often sold off parts of families, even separating parents from children.3 Rather, slaves shared cabins, and their heating and cooking equipment, at the will of the slaveholder. How many slaves were there per cabin? The census materials did not count slave cabins (“houses”) until 1860. For the 1860 census, the owner usually reported the number of slave cabins, and we can compare the implied population per cabin with the slave household sizes implied by our already having assumed that 35 percent of all slaves age ten and older were HHs. Table C-1 shows the comparison. The two sources imply similar average slave household size, even though we had not yet used the partial 1860 enumeration of slave cabins at the time we set our assumption about HHs for all dates between 1774 and 1860. Thus, we continue to assume that HHs were 35 percent of the slave population age ten and older.4 Recall that we have also assumed, following Weiss, that slave labor force participation was 90 percent of the slaves ten or older. These two assumptions imply for all dates up to 1860 that the average slave household had more workers (2.57) but fewer persons than the average free household. Table C-1 illustrates the differences between free and slave household sizes for the South in 1860.

3

See, for example, Sutch 1975a, 1975b; Gutman and Sutch 1976. Interestingly, the numbers here are defi nitely lower than the 5.2 slaves per dwelling that Fogel and Engerman considered characteristic of the large plantations. They wrote, however, before the arrival of the IPUMS sample made it easier to make such census calculations. See Fogel and Engerman 1974, 115. For a critique of the Fogel and Engerman view of the quality of slave dwellings, which we do not explore here, see Sutch 1976, 292–98. 4

ESTIMATING SLAVES’ RETAINED INCOMES

•

289

Table C-1 Slave Household Size in 1860 South Atlantic

East South Central

West South Central

Missouri

Slaves per Lindert and Williamson’s assumed slave household

4.2

4.2

4.1

4.3

Slaves per census-recorded slave house

4.3

4.3

4.1

4.4

Average size of free households, South, 1860

5.2

5.4

5.2

5.3

Percent of slaveholders reporting slave houses

87

89

84

77

Percent of slave population over age ten

64

66

66

70

Memoranda

Source: IPUMS sample of the 1860 population census. Notes: In the row “Slaves per census-recorded slave house,” both the numerator and the denominator are restricted to those slaveholders who reported their slave cabin totals. The “Missouri” free household average is actually for the entire West North Central region.

TWO CALCULATIONS OF SLAVES’ RETAINED INCOMES CA. 1860 For the 1860 benchmark, we can estimate a slave’s retained earnings (s) following either the exploitation rate path, where s = w(1 − Ex), or the slave rental path, where s = w − R. Here we follow both paths to check for similarity between the resulting estimates.5 To do so, we need, by age and sex, the free wage rate (w), the exploitation rate (Ex), and the slave rental rate (R). We begin by exploring the annual free wages for the occupations in which slaves were engaged, and what little is known about the distribution of the slave labor force across these occupations. For all dates, it seems clear that the majority of slaves were employed on farms and 5 We could also have made a direct estimate of slave consumption, using the detail about diet, housing, and clothing in Sutch 1976. That would have required adding some assumptions and gathering detailed prices—a task we have not undertaken here.

290

•

APPENDIX C

plantations, and these were almost entirely farm laborers (as opposed to overseers, house servants, or artisans).6 In the end, however, the sources of pay and property information did not allow us to incorporate estimates of both the occupational distribution and the occupationspecific pay rates except for our 1800 benchmark. For 1774, and the census returns of 1850 and 1860, slave occupational distributions were not available in a form that we could usefully combine with the rest of the data. Accordingly, we had to base all estimates of the annual freelabor income w, and therefore slave retained incomes s, on the annual earnings of free farmhands. Is it reasonable to assume that the annual marginal product of a working slave approximated that of a free worker in the same place and same occupation? The issue was debated hotly in the wake of the 1974 publication of Time on the Cross. In that book, Fogel and Engerman went out of their way to stress the relatively high productivity of slaves, asserting that the slave “was harder-working and more efficient than his white counterpart.”7 The criticism of Time on the Cross, consisting mainly of solid empirical rebuttals, includes something less easy to quantify: a belief in the lower productivity of slaves at work. This lower productivity was thought to have shown go-slow resistance, if one wanted to emphasize slaves’ self-assertion, and/or lower morale and poorer health.8 Our middling assumption—namely, that a slave’s annual product was neither higher nor lower than that of a free counterpart—is based on a balancing of two likely differences. First, slaves worked more days per year on the slaveholders’ account than a free hired laborer of the same attributes and location. Paul David and Peter Temin estimate that emancipation may have reduced blacks’ annual workdays by sixteen to twenty-two percent, apparently for comparable southern locations. Their estimate includes both the cut in labor force participation, which we 6 The slaves’ occupational mix is explored in “Own-Labor Earnings, 1774,” Worksheet (4), http://gpih; “Own-Labor Earnings, 1800,” Worksheet (3), http://gpih; “Slave Occupational Distribution, 1800,” http://gpih, which also discusses the distribution for 1859. For fuller counts of slaves occupations on the eve of the Civil War, see Gutman 1975, 48–82; Olson 1992b. 7 Fogel and Engerman 1974, 199–200. 8 For narrative evidence on slaves’ work resistance, see Genovese 1974, 284–324. Note that some of the narrative evidence merely echoed the slaveholders’ and overseers’ irritation that they could not get more work out of their slaves, with no evidence on the productivity of a free worker in the same situation.

ESTIMATING SLAVES’ RETAINED INCOMES

•

291

would not include in the w measure, and a reduction in hours, which we would include.9 Lacking an easy way to disentangle these two components, we merely note that the extra annual work hours of a slave would be a smaller percentage than David and Temin’s 16 to 22 percent. Balanced against this would be the unknown amount of reduction in slaves’ productivity per hour of work. Our only workable way of balancing these two opposing effects is to assume that slaves and free laborers had the same marginal product per year.10 Our estimates of farm labor rewards in 1860 were nonetheless improved by Craig’s plausible estimates of the marginal product of free farm labor by age, sex, and region.11 While Craig’s regression estimates from the 1860 censuses of agriculture and population refer only to northern regions, they show an east–west pattern relating the era of land settlement to the relative products of boys and girls that apparently also held for free labor in the South. These age patterns can yield estimates of slave retained earnings, hooked to the different regions’ Lebergott wage rates for farm laborers.12 That is, for each slave state, we use the farm wage rate expression w = (Craig’s marginal product for that age-sex group, in the comparable northern region)*(Lebergott’s average farm wage for the slave state)/(Lebergott’s average farm wage for the comparable northern region).13 We will later use a similar procedure for 1850,

9

David and Temin 1976, 202–14, especially 211–12. Our approach makes a comparison quite different from those made in Fogel’s Without Consent or Contract volumes. Leaning on Olson’s (1992a) data, Fogel (1989, 89) concludes that slaves worked a shorter year than the “free work year” because “free northern farmers averaged 3,200 hours per year”—a figure that would resemble the 313 days per year that we apply to certain occupations in certain benchmark years. Fogel and Olson, however, are comparing a slave with a northern farm operator, whose recorded work possibly included help from family members. They also include a large share of northern dairy farming, a full-year operation. No evidence is presented on the pay or work hours of free southern farm labor, though Lebergott’s sources report that it received lower annual pay than in the North; see “Wage Data Summary, 1860,” http://gpih. Our marginal product assumption, by contrast, compares what the slave worked on the slaveholder’s account with what a nonfamily, southern, free farmhand was paid when hired, which would apply to a shorter year than that worked by the farm operator. 11 Craig 1991. 12 Lebergott 1964. 13 If one wonders why we did not simply use Lebergott’s farm wage rates, skipping the link to Craig’s study, the reason is that Lebergott’s averages are not specific to age-sex groups, unlike Craig’s estimates. 10

292

•

APPENDIX C

again borrowing ratios of farm marginal products estimates by Craig’s study of 1860. To follow the exploitation rate (Ex) path, one must distill a best guess from the Time on the Cross debate. Fogel and Engerman claimed that in 1859, slaves retained 90 percent of what they produced, and thus that “the material . . . conditions of the lives of slaves compared favorably with those of free industrial workers in the decades before the Civil War.”14 Work by other scholars has since slashed the Fogel and Engerman retention estimate of 90 percent for 1859 to about 50 percent, summarized by Jenny Bourne’s assessment that “current estimates suggest that the typical slave received only about fift y percent of the extra output that he or she produced.”15 The 50 percent figure is suggested both by Sutch and Richard Vedder.16 Vedder estimated an exploitation rate (Ex) between 43.2 and 72.2 percent, for an average of 57.7 percent, well above the more benign Fogel and Engerman 10 percent exploitation rate. Guided by this literature, our first set of estimates of slaves’ retained earnings in 1860 uses Bourne’s 50 percent share, with s = 0.50 * w. These are displayed in the first panel of table C-2. The alternative estimates of slaves’ retained earnings in 1860 are displayed in the second panel of the same table. This panel shows s = w − R, subtracting the average rental values of hired slaves in 1858–1860 from the free farm wage rate. Happily, the two alternatives yield fairly similar results when they are applied to the individual slaves of the 1860 IPUMS “flat” slave sample: 48.7 percent is implied by the rental approach versus the assumed 50 percent. We use the rounded 50 percent slave retention share estimate for both 1850 and 1860, in part because this is the only option available for 1850.

THE ONE ALTERNATIVE FOR 1850 We can pursue only one of the three alternatives for 1850. Direct estimates of slave consumption were not available, though we imagine that they could be extracted from a combination of sources. More 14

Fogel and Engerman 1974, 5– 6; Whaples 1995, 146– 47. Bourne 2008. 16 Sutch 1975b; Vedder 1975, 455. 15

ESTIMATING SLAVES’ RETAINED INCOMES

•

293

Table C-2 Slave Retained Earnings by Region, for 1850 and 1860 Slaves’ retained earnings ($)

Slave population Total slave population

Estimated households

Labor force

Assumed, or implied, retained earnings ratio

(1) 1860, using half the free farm wage rate South Atlantic

61,655,556

1,841,595

439,344

1,255,148

50.0

Florida

2,547,674

76,349

17,420

49,772

50.0

Upper South Atlantic

20,047,977

611,812

148,897

425,386

50.0

Lower South Atlantic, no Florida

39,059,904

1,153,434

273,027

779,990

50.0

66,141,541

1,367,341

324,014

925,782

50.0

Upper East South Central (KY, TN)

22,318,894

527,715

123,578

353,093

50.0

Lower East South Central (AL, MS)

43,822,647

839,626

200,436

572,689

50.0

West South Central

38,677,413

621,857

152,822

436,633

50.0

Missouri (West North Central)

5,239,195

123,424

28,924

74,375

50.0

All slaves, 1860

171,713,704

3,954,217

945,103

2,691,938

50.0

East South Central

(2) 1860, using free farm wage rate minus average slave rentals South Atlantic

67,684,069

1,841,595

439,344

1,255,148

54.8

Florida

2,679,096

76,349

17,420

49,772

52.8

Upper South Atlantic

23,714,450

611,812

148,897

425,386

58.6

Lower South Atlantic, no Florida

41,290,523

1,153,434

273,027

779,990

52.8

51,996,112

1,367,341

324,014

925,782

40.6

Upper East South Central (KY, TN)

26,500,804

527,715

123,578

353,093

59.3

Lower East South Central (AL, MS)

25,495,308

839,626

200,436

572,689

29.1

West South Central

31,655,032

621,857

152,822

436,633

40.9

East South Central

Table C-2 (cont.) Slaves’ retained earnings ($)

Slave population Total slave population

Estimated households

Labor force

Assumed, or implied, retained earnings ratio

Missouri (West North Central)

6,232,211

123,424

28,924

74,375

59.3

All slaves, 1860

157,567,423

3,954,217

945,103

2,691,938

48.7

(3) 1850, using half the free farm wage rate South Atlantic

44,292,825

1,605,623

383,091

1,094,547

50.0

Florida

791,546

26,818

6,292

17,976

50.0

Upper South Atlantic

16,357,149

587,133

142,653

407,554

50.0

Lower South Atlantic, no Florida

27,144,130

991,588

234,147

669,017

50.0

33,907,314

1,082,365

256,019

731,482

50.0

Upper East South Central (KY, TN)

13,002,923

453,086

105,914

302,631

50.0

Lower East South Central (AL, MS)

20,904,391

629,279

150,105

428,872

50.0

West South Central

17,339,027

338,925

84,945

242,699

50.0

Missouri (West North Central)

2,544,323

89,316

20,561

58,761

50.0

All slaves, 1850

98,083,490

3,116,229

744,616

2,127,489

50.0

East South Central

Notes: 1. The Upper South Atlantic region consists of Delaware, the District of Columbia, Maryland, and Virginia inclusive of today’s West Virginia. The Lower South Atlantic thus consists of Georgia along with North and South Carolina, with or without Florida, as indicated in the table. 2. The totals were calculated from the 1850 and 1860 “flat” IPUMS slave samples, using the assumptions described in the text. In particular, the earning power of a person in each age and sex group was assumed to be in the same relation to that of an adult male nineteen to fi ft y-four years old, as in Craig’s (1991) northern farm sample for 1860. 3. The whole-region ratios were then derived as household-weighted averages of the subregional ratios. 4. The labor force is assumed to consist of 90 percent of all slaves age ten and higher. Th is falsely counts some infirmed and elderly, but falsely omits a probably similar number of young labor force participants. 5. The peculiarly low earnings retention rate for the lower East South Central region in panel 2 was due to a high rental rate for 1858–1860 in the Fogel and Engerman sample. 6. These total retained earnings are higher than those displayed by Sutch in HSUS, Series Ca233, which gives $94,400,000 for 1850 and $114,000,000 for 1859. 7. In the second panel, the average twelve-month slave rental rate for 1858–1860 was calculated from the Fogel and Engerman slave hire sample, Inter-university Consortium for Political and Social Research fi le 07422. The states, and their numbers of observations, were: Georgia (467 slave hires), Louisiana (187), Maryland (2,599), Mississippi (155), North Carolina (7,082), South Carolina (6), Tennessee (6,350), and Virginia (3,407), making up a total of 20,253 slave hires.

important, we cannot make use of the Fogel and Engerman sample of slave rentals for these dates. While we were able to sample a few thousand such hires in their Inter-university Consortium for Political and Social Research sample for 1848–1850, the slave hire data had three problematic omissions for this time period: slave ages were seldom reported for these years in the Fogel and Engerman sample; slave occupation was almost never recorded in the Fogel and Engerman samples, and never recorded in the census; and the Fogel and Engerman hired slave sample was small (n = 2,201 for these three years), and 83 percent of the observations were from North Carolina and Tennessee. These omissions impeded our attempts to combine rental data with good wage or consumption data and the 1850 IPUMS “flat” slave sample. Thus, our 1850 estimates are based on the assumption that slaves retained 50 percent of their marginal product, so that s = 0.5 w.17 As with 1860, our choice for 1850 for w is a free-labor farm wage rate. Again, we used Lebergott’s farm wage averages by state. For slave children in the labor force, we again proportioned their farm wage to that of a prime-age adult male using the Craig coefficients. The resulting estimates are shown in the third panel of table C-2.

ESTIMATING SLAVE CONSUMPTION IN 1800, DIRECTLY AND INDIRECTLY A first pass at direct estimation. Aided by testimony on slave diets, housing, and other conditions, Mancall, Rosenbloom, and Weiss infer slaves’ consumption levels by a somewhat direct route.18 Their findings can be summarized by these quotes: In the nineteenth century, the value of a slave’s diet equaled about 75 percent that of a free person. The information we have found for the colonial period would put the relative value anywhere between 20 percent and 75 percent. 17 These estimates are summarized in table C-2 above, and presented in more regional detail in “Slave Retained Earnings 1850, 1860,” http://gpih. 18 Mancall, Rosenbloom, and Weiss 2003.

296

•

APPENDIX C

In estimating the diet of a slave, we assume that its value increased from around 50 percent of a [free] colonist’s diet in 1700 to 75 percent in 1800.19

Note that the latter statement interpolates to a share of 67 percent in 1774. Also note that Mandall, Rosenbloom, and Weiss are here referring only to the value of food (diet or subsistence), and not to nonfood consumption like shelter, fuel, clothes and extras, which would have loomed much larger in the free laborer’s budget. For example, in 1874– 1875 Massachusetts, the poorest families (roughly comparable to the average family in 1800) spent 64 percent on “subsistence,” leaving 36 percent for clothing, rent, fuel, and sundries.20 Mandall, Rosenbloom, and Weiss report the estimates summarized in table C-3.21 We use these components to construct the total consumption estimates for 1800, shown in table C-4. The exploitation rate in 1800. One study that attempted to take the exploitation rate (Ex) from 1859 back to 1800 assumed flat subsistence for slaves in 1800, and then combined this assumption with information on productivity. Unfortunately, what we have since learned about agricultural productivity in slave-using sectors implies something implausible about the slave retention rate back in 1800. Specifically, Vedder reported a 1859 exploitation rate figure of 66.7 percent and a 1849 exploitation rate figure of 48 percent, concluding that the “observed rising rate of slave exploitation over [that decade of ] time . . . reflects rising marginal productivity [of slaves] and a constant [subsistence].”22 If his two inferences were correct, this would imply that the slave value marginal product rose across the decade 1849–1859 at 4.5 percent per annum. Could Vedder’s two results have both lasted over the whole antebellum period? That is, could slaves’ labor productivity really have grown so fast for more than half a century, while slaves’ retained incomes remained at a constant subsistence level? The work of other scholars seems to reject the joint long-run persistence of that high 4.5 percent productivity growth rate and constant subsistence consumption for slaves. 19

Mandall, Rosenbloom, and Weiss 2010, 399; see also ibid., 417n37. US Census Bureau 1975, 322. 21 Mandall, Rosenbloom, and Weiss 2010, 396–97, table 14.1. 22 Vedder 1975, 456. 20

ESTIMATING SLAVES’ RETAINED INCOMES

•

297

Table C-3 Slave Consumption Components in 1800 (Mancall, Rosenbloom, and Weiss) United States

Lower South

North

Per capita, including slaves

29.37

27.66

29.80

Per slave, at 75% of free

23.84

23.80

22.94

Per capita, including slaves

6.64

6.10

6.63

Per slave, at 75% of free

5.39

5.25

5.10

(C) 1800 shelter per capita, free population only

4.80

—

—

(D) 1800 shelter per capita, slaves (based on rural areas)

0.60

—

—

(E) Slaves’ share of labor force

0.304

0.514

0.103

(A) 1800 food consumption

(B) 1800 firewood consumption

Sources: Mancall, Rosenbloom, and Weiss 2003; US Census Bureau 1976, 322. Notes: All consumption values in 1840 dollars using the David-Solar CPI. North = New England plus the three main Middle Atlantic states (NJ, NY, and PA).

Olmstead and Rhode have mapped slave productivity growth in cotton agriculture for 1800–1860.23 Using recent agricultural slave labor force estimates (slaves age ten and up) and four-hundred-pound bale output estimates, cotton bales per worker in the Old South grew at 1.57 percent per annum from 1800 to 1860.24 The Old South is defined as Georgia, North Carolina, South Carolina, and Virginia, the relevant region for our 1800 estimates; it grew much slower than the New South (South Central regions).25 This may overstate average labor productivity growth, since the plantations became more specialized in cotton over time.26 Still, even as late as 1880, cotton made up only 16 percent of improved acres in the South (with corn at 31 percent; the other crops were barley, buckwheat, hay, hops, Irish potatoes, oats, rice, rye, sweet potatoes, tobacco, and wheat). Thus, what about a broader crop-based index of labor productivity? Rhode offers such an index based on 23

Olmstead and Rhode 2010. Craig and Weiss 1998; Olmstead and Rhode 2010, 37, table 1. 25 Olmstead and Rhode, 2010, 4. 26 Olmstead and Rhode 2010, 5. 24

298

•

APPENDIX C

Table C-4 Total Consumption of Food, Fuel, Shelter, and Clothes per Slave Worker in 1800 In dollars of 1800

1840

Rural farm (23.80 + 5.25 + 0.60 = 29.65; + 5% for clothes)

45.18

31.13

Rural nonfarm (29.65 + 10% for better clothes and extras)

47.34

32.62

Small town (29.65 + 10% for higher prices and rents = 32.62; + 10% for better clothes and extras)

52.08

35.88

Big city (29.65 + 20% for higher prices and rents = 35.58; + 20% for better clothes and extras)

61.97

42.70

Rural farm (22.94 + 5.10 + 0.60 = 28.64; + 5% for clothes)

43.64

30.07

Rural nonfarm (28.64 + 10% for better clothes and extras)

45.72

31.50

Small town (28.64 + 10% for higher prices and rents = 31.50; + 10% for better clothes and extras)

50.29

34.65

Big city (28.64 + 20% for higher prices and rents = 34.37; + 20% for better clothes and extras)

59.86

41.24

Lower South

Middle colonies + New England

cotton, molasses, rice, sugar, and tobacco for 1800–1860, and it grows at 2.3 percent per annum for the whole South (Old plus New South).27 Applying the same discount to the slower-growing Old South that Olmstead and Rhode report for cotton productivity growth (2.44 percent per annum for the whole South and 1.57 percent per annum for the Old South, or 0.643 discount) implies 1.48 percent per annum over the six decades.28 Hence, 1.5 percent seems like a fair estimate. Earlier work by Alfred Conrad and John Meyer, Frank Whartenby, and Lebergott offers confirming evidence of fast slave productivity growth.29 Assuming an 1859 exploitation rate (Ex) of 57.7 percent, constant subsistence, and slave productivity growth of 1.5 percent per annum, then slaves in 1800 would have received an estimated 103.3 percent of 27

Rhode, communication with authors, August 22, 2010. Olmstead and Rhode 2010, 37, table 1. 29 Conrad and Meyer 1958; Whartenby 1977; Lebergott 1964. 28

ESTIMATING SLAVES’ RETAINED INCOMES

•

299

their marginal product, and the Ex would have been negative. Since we know it was not, subsistence must have grown considerably over the six decades from lower levels in 1800. For example, if the exploitation rate in 1800 was also 57.7 percent, then slaves’ absolute “subsistence” consumption would have grown at the same 1.5 percent a year, or more than doubled between 1800 and 1860. A recent survey offers evidence suggesting that slave consumption per capita grew at 0.65 percent per annum over the eighteenth century.30 Clearly, slave “subsistence” (i.e., their consumable incomes) cannot have been a Malthusian constant, and must have been rising before 1860. In short, it appears quite likely that slaves’ consumption rose along with their productivity between 1800 and 1860. We have used the following rates of slave income retention in 1800, which tended to hover around the s/w = 42.3 percent implied by the previous paragraph’s summary interpretation of the literature on the exploitation rate: South

North

Farm

41.4

40.1

Rural nonfarm

44.3

43.9

Small town

47.5

47.1

Big city

52.7

52.3

These shares are an average of two quite-different shares of slave income retention (1–Ex): low ones derived by comparing the direct estimates of slave consumption in table C-4 above, and the higher ones to which we now turn. The share of slave earnings retained for those who were hired out, 1796–1804. For the ca. 1800 benchmark, our third alternative is to estimate slaves’ retained incomes (consumption) as the difference between their likely marginal product (w) and the rental R = (w − s) that a renter would willingly pay for a slave, over and above their consumption (s). We assume that given skills and occupations, slaves and free labor were close substitutes, if not perfect substitutes, such that these African Americans would have received the earnings of free labor had they been 30

300

Mancall, Rosenbloom, and Weiss 2003.

•

APPENDIX C

Table C-5 Slave Earnings Retention Rates, Based on Hire Market Data for 1796–1804 Annual hire rental, rural Maryland

Earnings retained (1-Ex in parentheses) Lower South farm

Middle colonies farm

All slaves hired

$28.44

$45.18 (61.4%)

$43.64 (60.5%)

All male slaves hired

$29.62

$45.18 (60.4%)

$43.64 (59.6%)

All male “fellows”

$39.42

$45.18 (54.3%)

$43.64 (52.5%)

free—an assumption supported by qualitative evidence as well as testimony following emancipation.31 The Fogel and Engerman Inter-university Consortium for Political and Social Research file on slave hires reports average annual hire rates in Queen Anne’s County, Maryland, in 1796–1804 for 207 slaves, and we combine these with the consumption per slave estimates reported in table C-4 above. For farmwork, these hires were as shown in table C-5. These slave retention rates are not far from those that the literature following Time on the Cross has estimated for 1849–1859, as cited above, with averages of 58.7 percent for the Lower South and 57.5 percent for the Middle Colonies (or Middle Atlantic). Nonetheless, given the thinness of this 1796–1804 sample, our estimates for the 1800 benchmark average these shares together with the lower ones cited above based on the Ex literature, yielding the compromise slave income retention rates cited above (41.4 to 52.7 percent for southern slaves, and 40.1 to 52.3 percent for northern slaves). Relative to the slave retained income estimates for later benchmark dates for 1850 and 1860, this one for 1800 has biases in both directions: it allows a higher wage rate to be used for the 17 percent of slaves employed outside agriculture, yet it applies a lower retention share (s/w) than do the estimates for 1850 and 1860.32 31

Goldin 1976, 28–30. While we lack direct evidence confi rming that the slave retention share was lower in 1800 (like an estimated share around 41 percent) than the 50 percent we used for 1850 and 1860, the implied gradual rise in the share slaves retained seems plausible. After all, in 1800 the transatlantic slave trade had not yet been abolished by Britain, allowing slaveholders the luxury of 32

ESTIMATING SLAVES’ RETAINED INCOMES

•

301

Note that the earnings–retention ratios (1–Ex) for 1800 are not based on the free farm labor wage rate. This is because for 1800, the available free farm wage rate quotes do not refer to comparable labor on the twelve-month basis that prevailed among slave rentals in rural Maryland for 1796–1804. The free wages in this case were based on daily wage rates in western Virginia.33 The employers were not committed to hire and feed the workers year-round. Thus, multiplying the daily wage rate by 313 days and adding twelve months’ board would overstate the annual marginal product of a hired free laborer, who in reality had to find off-season work at lower pay. The resulting estimates for 1800 (table C-5) as well as 1850 and 1860 (table C-2) show some consistency with the differences one might expect across time and space. For 1860, it is also true that the share of their earnings that slaves retained around 1860 was about the same whether we used the approach of taking the difference between the free wage rate and the slave hire rental rate, or we simply assumed a 50 percent slave retention rate in accordance with the literature in the wake of Time on the Cross. Despite this consistency, one must remain alert to a range of possible shares for the retention of slave earnings. This need for sensitivity analysis is especially important when chapter 6 compares the earnings that slaves may have retained around 1860 with the labor earnings of free blacks in 1870. BACKCASTING SLAVE RETAINED INCOMES TO THE COLONIAL ERA By the time we reach back to 1774, we have run out of fresh data on direct consumption (despite a few hints by Main), the exploitation rate, or slave rental prices.34 We are thus left with second-best options. The one we use is to combine the 1774 wage rates for free southern farm labor with the compromise retained income shares we have used for 1800—namely 1–Ex = around 41 percent.35 working slaves harder and with higher mortality in the knowledge that they could still be replaced by fresh imports of slaves from Africa. 33 Adams 1992. 34 J. Main 1965. 35 This has been roughly checked against the implicit rental rates implied by applying a gross rate of return to the 1774 slave prices. See “Own-Labor Incomes 1774,” Worksheet (4), http://gpih.

302

•

APPENDIX C

In chapter 3, we could have extrapolated the 1774 slave incomes back to the seventeenth century (for the Chesapeake) or the early eighteenth century (for the lower South), but there was no need to do so. For the Chesapeake, our indicator of colonial income trends was a measure of the total farm income, for which slave incomes were already implicitly counted.36 For the Lower South and the Middle Colonies, slave incomes are hidden within the Mancall, Rosenbloom, and Weiss indexes of the total income as measured from the product side. 36

See “Backcasting Chesapeake Incomes from 1774 Back to 1675,” http://gpih.

ESTIMATING SLAVES’ RETAINED INCOMES

•

303

APPENDIX D

American versus British Prices, 1640–1875

Table D-1 Ratios of American to English Prices, Twenty-Two Commodities, 1640–1774 Massachusetts, 1640–1660

Pennsylvania, 1730–1753

Maryland and Virginia, 1730–1753

Massachusetts, 1754–1774

Pennsylvania, 1754–1774

Maryland and Virginia 1754–1774

Barley

1.66

—

—

—

—

—

Beans

—

—

—

1.31

—

—

Beef

—

0.39

0.62

0.57

0.47

0.58

Bread

—

—

—

—

—

—

Butter

—

—

—

1.05

—

0.81

Candles tallow

—

—

—

—

—

—

Cheese

—

—

—

1.45

—

—

Coffee

—

—

—

—

—

0.22

Cotton

—

—

—

—

1.21

—

Eggs

—

—

—

0.68

—

0.91

Flour

—

0.54

—

—

0.60

—

Hay

—

—

—

—

—

—

Milk

—

—

—

—

—

—

Mutton

—

—

—

0.81

—

—

Nails

—

—

—

—

—

—

Oats

—

—

—

—

—

0.72

Paper, 8.5" x 11"

—

—

—

—

—

—

Table D-1 (cont.) Massachusetts, 1640–1660

Pennsylvania, 1730–1753

Maryland and Virginia, 1730–1753

Massachusetts, 1754–1774

Pennsylvania, 1754–1774

Maryland and Virginia 1754–1774

Pepper

—

—

—

—

0.94

—

Pig iron

—

—

—

—

—

—

Pork

—

0.55

0.48

1.17

0.60

0.68

Potatoes

—

—

—

—

—

—

Rice

—

0.44

—

—

0.65

—

Rye

0.93

—

—

0.87

—

0.52

Shoes

—

—

—

1.19

—

—

Soap

—

—

—

—

—

—

Stockings

—

—

—

—

—

—

Sugar

—

1.48

—

0.83

1.13

1.05

Tea

—

—

—

—

0.34

0.76

Tobacco

—

0.09

—

—

0.10

—

Wheat

0.87

0.78

1.03

—

0.88

0.87

Wool

—

—

—

—

—

—

Numbers of comparisons

Total

Total

3

7

3

10

10

10

43

America/ Great Britain ≥ 1.20

1

1

0

2

1

0

5

America/ Great Britain ≤ 0.80

0

6

2

2

6

6

22

Sources for tables D-1 through D-3: For England, see Gregory Clark, http://gpih. For Massachusetts, see C. Wright 1885. For Pennsylvania, see Bezanson, Gray, and Hussey 1935. For Maryland and Virginia (Chesapeake) and West Virginia, see D. Adams 1986, 1992. For Vermont, see T. Adams 1944. Notes for tables D-1 through D-3: Th ree commodities ostensibly providing data in both countries were omitted because of difficulties with the comparability of products or units of measurement. For all three tables, these commodities were cotton cloth, salt, and wine. The strikingly low West Virginia prices of butter have been rechecked. They are indeed as the source implies. The term “West Virginia” was chosen by Donald Adams to refer to the locus of his primary data sources, which corresponded to the region that became West Virginia in 1863.

AMERICAN VERSUS BRITISH PRICES

•

305

Table D-2 Ratios of American to English Prices, Thirty Commodities, 1792–1808 Vermont, 1792–1808

Massachusetts, 1792–1808

Pennsylvania, 1792–1808

Maryland and Virginia, 1792–1808

West Virginia, 1792–1808

Barley

—

0.85

—

—

—

Beans

—

1.23

—

—

—

Beef

0.31

0.58

0.46

0.73

0.37

Bread

—

—

—

—

—

Butter

0.73

1.25

0.85

1.15

0.11

Candles tallow

—

—

0.90

—

—

Cheese

0.79

1.06

—

—

—

Coffee

0.30

—

0.22

0.31

1.26

Cotton

—

—

0.81

—

—

Eggs

0.69

1.31

—

1.99

—

Flour

—

1.16

0.75

—

2.77

Hay

0.31

—

—

—

—

Milk

—

0.86

—

—

—

Mutton

—

0.55

—

—

—

Nails

1.98

—

1.30

—

—

Oats

0.61

0.47

0.61

—

—

Paper, 8.5" x 11"

—

0.67

—

—

—

Pepper

—

—

0.58

—

0.94

Pig iron

—

—

1.05

—

—

Pork

0.45

1.00

0.66

0.80

1.15

306

•

APPENDIX D

Table D-2 (cont.) Vermont, 1792–1808

Massachusetts, 1792–1808

Pennsylvania, 1792–1808

Maryland and Virginia, 1792–1808

West Virginia, 1792–1808

Potatoes

0.76

0.95

—

—

—

Rice

—

1.17

1.01

—

—

Rye

0.59

1.04

0.70

0.77

—

Shoes

1.41

1.19

—

—

—

Soap

—

0.05

0.09

—

—

Stockings

—

3.12

—

—

—

Sugar

1.04

0.78

1.36

1.18

3.64

Tea

0.46

0.44

0.25

1.23

0.07

Tobacco

—

—

0.10

—

—

Wheat

0.76

—

1.03

0.98

—

Wool

2.27

—

—

—

—

Numbers of comparisons

Total

Total

16

20

18

9

8

71

America/ Great Britain ≥ 1.20

3

3

2

1

3

12

America/ Great Britain ≤ 0.80

12

7

10

3

3

35

AMERICAN VERSUS BRITISH PRICES

•

307

Table D-3 Ratios of American to English Prices, Thirty Commodities, 1840–1875 Vermont, 1840–1860

Massachusetts, 1840–1860

Pennsylvania, 1840–1860

West Virginia, 1840–1860

Vermont, 1866–1875

Pennsylvania, 1866–1875

Barley

—

0.84

—

—

—

—

Beans

—

2.01

—

—

—

—

Beef

0.36

0.92

0.38

0.44

0.55

0.42

Bread

—

—

0.77

—

—

1.18

Butter

0.78

1.14

0.57

0.13

1.32

0.94

Candles tallow

—

—

0.89

—

—

—

Cheese

0.56

0.90

—

—

1.13

—

Coffee

0.33

—

0.24

2.32

0.96

0.61

Cotton

—

—

0.96

—

—

1.00

Eggs

0.86

1.43

—

—

1.20

—

Flour

0.68

1.00

0.72

2.85

0.94

0.97

Hay

0.44

—

—

—

0.73

—

Milk

—

0.89

—

—

—

—

Mutton

—

0.65

—

—

—

—

Nails

0.87

—

0.65

—

0.83

0.67

Oats

0.74

0.50

0.63

0.97

0.86

Paper, 8.5 " x 11"

—

0.90

—

—

—

—

Pepper

—

—

0.27

0.71

—

1.13

Pig iron

—

—

2.33

—

—

3.31

Pork

0.51

0.88

0.52

0.70

0.50

0.51

Potatoes

0.49

1.13

—

—

0.11

—

Rice

—

1.34

1.01

—

—

3.40

Rye

0.86

1.29

0.79

—

—

—

Shoes

0.80

0.85

—

—

1.36

—

Soap

—

—

0.09

—

—

0.12

308

•

APPENDIX D

Table D-3 (cont.) Vermont, 1840–1860

Massachusetts, 1840–1860

Pennsylvania, 1840–1860

West Virginia, 1840–1860

Vermont, 1866–1875

Pennsylvania, 1866–1875

Stockings

—

—

—

—

—

—

Sugar

0.79

0.71

0.91

2.10

1.51

1.54

Tea

0.60

0.50

0.22

0.10

1.61

0.62

Tobacco

—

—

0.07

—

—

0.18

Wheat

1.25

—

1.07

—

2.17

1.73

Wool

1.79

—

—

—

1.46

—

Numbers of comparisons

Total

Total

17

18

19

8

16

17

95

America/ Great Britain ≥ 1.20

2

4

1

3

7

3

20

America/ Great Britain ≤ 0.80

12

4

13

5

4

7

45

For detailed calculations, see “Price Comparisons between North America and England, Specific Goods, c1650-c1870,” main data series, North America, http://gpih.

AMERICAN VERSUS BRITISH PRICES

•

309

Table D-4 American versus British Costs of a Respectability Bundle of Consumer Goods, 1800–1858 American costs as percentage of British 1800

1850

1858

Bread

81

87

76

Beans

129

220

224

Beef

43

44

38

Eggs

114

139

142

Butter

118

130

124

Cheese

108

92

100

Soap

126

112

115

Linen cloth

157

147

133

Candles

122

128

188

Overall costs of these nine items in the respectability bundle

84

91

85

Sources: For England, see Broadberry et al. 2015, 339; Clark, http://gpi. For America in general, see http://gpih. For Pennsylvania, see Bezanson, Gray, and Hussey 1935. For Massachusetts, see C. Wright 1885. Notes: The expenditure weights are those for Great Britain 1801–1803. Whenever the price data permitted for Great Britain, we used five-year averages for prices in 1798–1802, 1848– 1852, and 1856–1860. In some cases, the prices refer only to a subset of those years. The American prices are a mixture of Pennsylvania and Massachusetts.

310

•

APPENDIX D

APPENDIX E

A Guide to the 1860 Income Estimates, and Some Modifications for 1850

W

hile the benchmark estimations for 1850 and 1860 follow the same basic approach as for 1774 and 1800, the underlying evidence is vastly improved. The labor force and its occupations are now detailed in the censuses, and sampled in machine-readable form by the IPUMS project. As with 1774, we have been able to exploit the occupational detail to link own-labor income with wealth and its associated property income. Wages and white-collar earnings are better documented covering more occupations and locations from sources unavailable for the earlier years. Even slaves were recorded as individuals in the census, and the Fogel and Engerman sampling of slave hires and assessed values is more abundant than for earlier dates, as we noted in appendix C. From 1850 through 1870, the census and the IPUMS samples also recorded individuals’ asset ownership. To describe how we harvested the abundant data, we begin with 1860, proceeding from free-labor earnings to farm profits, and then to property incomes. The remainder of this appendix will describe some special assumptions we had to make for 1850, for which the census recorded neither the values of personal estate nor female occupations.

FREE-LABOR EARNINGS IN 1860 For job-specific wage and salary data from which to construct labor incomes for 1850 and 1860, we draw on several authors’ gleanings from the Weeks and Aldrich Reports, Lebergott’s wage series, Robert Margo’s rich data on civilian wage rates at army posts, and the

Table E-1 Marginal Labor Products and Profit Residuals for Northern Farm Households, 1859–1860 All northern states

Northeast

Midwest

Frontier

Profit intercept

$169.26 (29.48)

$234.55 (66.79)

$164.85 (26.83)

$117.57 (49.61)

Children age zero to six

−19.80 (7.39)

−20.82 (18.29)

8.59 (6.40)

−6.41 (12.43)

Children age seven to twelve

16.38 (10.52)

22.81 (25.91)

27.76 (8.87)

27.12 (18.96)

Teenage females

22.01 (16.06)

22.95 (37.91)

9.75 (13.79)

17.53 (29.74)

Teenage males

58.31 (15.03)

111.03 (35.95)

47.57 (12.74)

49.03 (28.84)

Adult females

152.63 (13.49)

154.08 (28.04)

70.25 (12.63)

147.28 (30.04)

Adult males age nineteen to fifty-four

229.09 (16.69)

294.77 (38.88)

186.44 (14.89)

193.66 (26.84)

Adult males age fifty-five and older

175.60 (24.65)

145.95 (48.52)

121.94 (24.17)

135.44 (55.07)

Farm size was ≥ eighty acres

269.26 (18.83)

422.52 (44.45)

223.61 (16.76)

221.98 (31.11)

Number of farms

8,496

3,130

4,347

1,019

Source: See Craig 1991, 74, table 3. Notes: The dependent variable is the value of the free farm family’s labor’s output. The estimation technique is generalized least squares. The figures in parentheses are the standard errors.

American Almanac’s reports of salaries for a wide variety of male and female white-collar occupations. We continue to apply our assumptions about the part-time work year for various daily and monthly hire occupations.1 Our adjustments for children’s lower pay rates, work hours per year, and in-kind payment are as described in appendix B. 1 See Lebergott 1964; Margo 1990; American Almanacs, 1856– 60; Craig 1991; “Wage Data Summary, 1860” and “Own-Labor Incomes, 1860,” http://gpih.

312

•

APPENDIX E

FARM OPERATOR INCOMES AND PROFITS, 1860 The task of estimating the own-labor income for farm operator households is rendered much easier for 1860 than for 1774, thanks to a sophisticated study by Craig.2 Craig’s objective was to determine the marginal products of farm family members, especially children, in different parts of the North. In the process of making his points about child labor on the northern frontier, the Northeastern Seaboard, and farms in between, Craig distilled plausible measures of what our study calls own-labor income for a free farm household. In his study, as in our farm estimates, own-labor income equals the marginal products of all family members plus a residual farm profit that depended on farm size. Table E-1 shows the marginal products and residual profit terms he produced from regressions on northern farm incomes in the 1860 census of agriculture and other sources. The first and last terms capture the profit residual, and those in between can be interacted with the numbers of family members in the age-sex categories to derive family labor product. That works for the North, the region sampled by Craig. Extending it to farm operator households in the South and West is less straightforward, and calls for an assumption. We calculated southern and western farm marginal product coefficients by assuming that Marginal product in southern or western region

equals

Craig coefficient for corresponding northern region

times

The adult farm wage ratio, (South or West region)/ (corresponding northern region)

Here the “corresponding northern region” is assigned on an east-towest basis: • South Atlantic uses Craig’s Northeast coefficient and relative wage • East South Central uses Craig’s Midwest coefficient and relative wage 2

Craig 1991.

A GUIDE TO THE 1860 INCOME ESTIMATES

•

313

• West South Central uses Craig’s frontier coefficient and relative wage • Mountain or Pacific uses Craig’s frontier coefficient and relative wage

The case for our using ratios of regional adult farm wage rates to index farm family returns from the North to the South and West rests on the fact that much of what is being estimated is a labor return, and a family member’s labor has an opportunity cost proportional to the wage rate for hired labor.3 The derived marginal products for each family member and levels of residual profits are then applied to each free farm family in the 1860 IPUMS sample, reweighted to approximate the whole population, and summed up to derive the own-labor income of each set of farm families represented by an IPUMS sample observation.

PROPERTY INCOMES IN 1860 As with 1774 and 1800, our 1860 estimates of property income are based on household wealth times a rate of gross annual return. The 1860 IPUMS sample delivers more information than the Jones sample in one respect, and less information in another. It is a richer sample because it yields tens of thousands of observations for each region versus Jones’s 919 observations for the whole of the thirteen colonies in 1774. Its asset detail is less generous, however. Jones gave separate asset values for slaves and servants, real estate, financial assets, producers’ durables, producers’ perishables, business inventories, livestock, business equipment, and crops. By contrast, the census of 1860 (and 1870) asked people only to report their real estate value and the value of their personal estate, or “personalty,” which was the gross value of all assets other than real estate. False zeros. The censuses that asked people about their wealth may have missed part of it because people failed to state any wealth even though their occupation should have implied that they had at least some personal estate, if not any ownership of real estate. The problem of not 3 We note that ours is not the only plausible way to index farm household incomes between regions. One could alternatively mix the interregional farm wage ratios we use here with the interregional ratios of farm real estate values per farm.

314

•

APPENDIX E

revealing one’s true positive personal estate is dramatized by the case of Cornelius Vanderbilt. As we note in the illustrative box between chapters 5 and 6, the censuses of 1860 and 1870 left the wealth columns blank for multimillionaire Vanderbilt, even though he had declared owning three hundred thousand dollars in real estate back in 1850. How widespread was this disturbing practice? In the Middle Atlantic, where Vanderbilt lived, his occupation group “managers, officials, and proprietors” usually reported positive personal estates. Still, those reporting no personal estate were 17.3 percent of all such Middle Atlantic managers in the 1860 census and 15.5 percent in the 1870 census. So the issue of false zeros is not to be swept under the rug. False zeros may have caused us to underestimate national income for 1850, 1860, and 1870.4 Offsetting overstatements of productive wealth. Yet in two offsetting ways, the same census questions that invited false zeros may also have overstated the property incomes that contribute to national income. First, they may have captured some personal estate, such as consumer durable wealth, that does not contribute to the conventional (NIPA) measures of national income. Second, the census takers were aware that their measures of real estate and personal estate double counted mortgage lending. For the borrowers, the value of real estate did not deduct the borrowed funds. For the lenders, however, including the owners of lending banks, the census question invited them to include their holdings of mortgages in their personal estate. The instructions to the enumerators made it clear in the 1850 census that under heading 8 insert the value of real estate owned by each individual enumerated. You are to obtain the value of real estate by inquiry of each individual who is supposed to own real estate, be the same located where it may, and insert the amount in dollars. No abatement of the value is to be made on account of any lien or incumbrance thereon in the nature of debt.

4 We did this, though, without any clear implications for the measurement of inequality. The share of zeros tended to be lower for the top decile of the households ranked by income, yet it is hard to determine how the prevalence of false zeros correlated with the overall income. The share of false zeros may have been lower toward the bottom of the ranks simply because true zeros were more prevalent for the less well off.

A GUIDE TO THE 1860 INCOME ESTIMATES

•

315

This remained the rule in the 1860 and 1870 censuses, even though these two censuses counted mortgages within lenders’ personal estates. Judging from the share of farm mortgages in farm real estate value in 1910, the share of mortgage values back in 1870 could have been anything between 0 and 5 percent.5 Rates of return. Again, we convert the wealth data into property incomes using the rates of return commonly quoted at that time. For real estate wealth, we use a net rate of return of 5 percent, based on reported interest rates. For the net rates of return on other assets, we had to reckon the approximate shares that slaves, other productive assets, and consumer durables were in total personal estate. Drawing on the wealth portfolio research of Goldsmith for the United States in 1850 and 1880, we estimate a net rate of return on personal estate of 7.8 percent for the South and 7.3 percent for the non-South, with the difference explained by slaveholding, for which a depreciation rate of 11 percent was assumed to convert net to gross returns.6

SPECIAL ADJUSTMENTS FOR ESTIMATING INCOMES IN 1850 Incomes in 1850 were estimated in the same ways as for 1860, using the same sources. The only differences are that two kinds of information available for 1860 are missing for the preceding census. Filling in the unsolicited “personal estate” values. The 1850 census innovated by asking about the values of real estate that each household member owned, but it failed to ask them about the value of their personal estate. Only the 1860 and 1870 censuses did economic historians the favor of asking for both the value of real estate and personal estate. We are left with the task of predicting the values of unreported personalty for free households in the IPUMS sample for 1850. Our basic prediction strategy was to find how personalty in 1860 depended on variables that the census and the IPUMS reported for both 1850 and 1860, and use these patterns to predict the personalty of each 5 On the measurement issue, see US Census 1870, 3–5. On the mortgage shares from 1910 onward, see US Department of Agriculture 1973. 6 Homer and Sylla 1996; Goldsmith 1985.

316

•

APPENDIX E

free IPUMS-sample household in 1850. Once the predicted values of personalty were then assigned their rate of return combined with the observed incomes from realty and own-labor income, we have an estimate of each household’s total income for 1850. Our technique for predicting personal estate using 1860 patterns has been shaped by two obvious features of the wealth variables. First, both real estate and personal estate are continuous variables. Our technique should make use of the continuous nature of real estate values (and years of age) by using regression techniques to determine slopes. Second, both of these wealth variables also took the value zero for most free individuals and even for a large share of free households. The frequency of zeros means that the values of the dependent variable, personal estate, are censored to be nonnegative. Accordingly, we divided the 1860 IPUMS samples of free households into two groups—those with positive real estate (the “1860r” regional samples), and those without any (the “1860n” regional samples)—so that we could extract the relationship of realty value to personalty value in the former. Within each subsample thus defined by region and realty ownership, we ran tobit censored-at-zero regressions to predict personalty.7 The variables used to predict the value of personal estate within each region for 1860 are: • Real estate (a fourth-order polynomial) for the 1860r samples only • Age of the HH (a cubic function) • Urban residence • Farm household • Nonwhite • The number of extra labor force participants beyond the first (if labor force > 0)

7 We suspect that our technique is subject to two opposing biases when it comes to calculating the inequality of income in 1850. On the one hand, we have probably made property and total incomes look too equal, because our tobit regressions never predict zeros and underpredict extremely high values. On the other hand, regressing personalty on the existence and the value of real estate overstates the correlation of the two. We can only hope that these two biases canceled each other for 1850. In any case, the net bias is limited to income from personal estate, and not to realty or own-labor income.

A GUIDE TO THE 1860 INCOME ESTIMATES

•

317

• The HH’s occupational category other than menial male laborer (professional, commerce, manufacturing craft, building trade, farm operator, female head with an occupation, female without an occupation, or male with no given occupation) • Subregional categories within the census region (upper New England within New England, Chesapeake within South Atlantic, District of Columbia, Upper East South Central (Kentucky and Tennessee) within the East South Central, and Pacific within the combined Mountain Pacific region)

The regression-predicted values were projected back to 1850, implicitly incorporating the decadal change in realty values and the decadal shift in the probability of having any realty. The predicted values of personal estate were then calculated for each household in the 1850 sample, en route to calculating the levels and inequality of total income. Filling in female occupations and earnings. The 1850 census was also inferior to the 1860 census when it came to reporting occupations for females. Here too, 1850 was a transitional and experimental census. Again, as with the missing personal estate values, we use the strategy of conditioning the missing variable on other variables that the census and the IPUMS reported for both 1850 and 1860. In this case, however, we did not employ the regression technique used to predict personal estate values for 1850. We instead used only a large matrix of categories that conditioned whether an individual female was in the paid labor force, and if she were, what pay she would have received. Using this approach, we generate conditional mean pay in 1860 from a large number of categories that were reported in both 1850 and 1860. First, for each of the nine regions in 1860, we classified free females age fifteen to sixty-four into twenty-four categories defined by: • The household’s type of residence: urban nonfarm, rural nonfarm, or farm • Whether she owned a positive value of real estate • Whether she was the HH • Whether she was literate

For each of the twenty-four categories, the 1860 census yielded two averages that we carried over to 1850. One is the share of such females who 318

•

APPENDIX E

were in the paid labor force. The other is a predicted annual rate of pay for such a paid individual. To help us rescale from 1860 pay rates back to 1850 pay rates, we divided each 1860 pay rate by the regional average annual pay for rural nonfarm female domestics. We then adjusted the pay rates for her age. Using the pay rate for an adult female between age twenty and sixty-four as 1.00, we applied these fractional pay rates for nonfarm teenage females: 0.20 for employed females up through age fourteen, 0.40 for fifteen-year-olds, 0.60 for sixteen-year-olds, 0.70 for seventeen-year-olds, 0.80 for eighteen-year-olds, and 0.90 for nineteen-year-olds. For females employed in farm households, we used the Craig coefficients for 1860, from table E-1 above. We then applied these categorical averages to free individuals in the 1850 IPUMS sample, generating predictions relative to each region’s wage rate for an adult female, free domestic servant. The predicted cell averages for 1850 often take on low values because they are based on 1860 averages combining many zeros with some positive values.8 8 For the resulting 1850 estimates, see “Regional Income Totals, 1850” and “American Income Inequality, 1850,” http://gpih.

A GUIDE TO THE 1860 INCOME ESTIMATES

•

319

APPENDIX F

A Guide to the 1870 Income Estimates

M

ost of the keys for unlocking household incomes are the same for 1870 as for 1850 and 1860, as described in chapter 5 and appendix E. Again, we can use the IPUMS for occupations, property values, and other attributes. Similar sources are available for wage and salary rates, though we were forced to use an 1875 American Almanac, since that issue was the closest we could find for 1870.1 In one respect, the task becomes easier for 1870: thanks to slave emancipation, our estimation can avoid guesswork about the levels of consumption that slaves were allowed. Here is the simplest way of stating how the 1870 income estimation procedures resemble the 1860 procedures, for the vast majority of cases: for everybody in the 1870 IPUMS sample who was not living in a farm operator’s household, incomes were estimated in the same way as for the free population in 1860. To illustrate the procedure, let us review how income was calculated for a concrete household from the 1870 IPUMS sample. To show the prevalent similarity to free households in 1860, we use the case of a black southern household engaged in agriculture, but one that is not the household of a farm operator. Consider IPUMS serial number 634401, a Georgia household headed by a nonwhite female, with five sons who are farmhands. Their individual attributes and labor earnings were as follows: 1 The Warren and Pearson wholesale price index fell by 9 percent between 1870 and 1875 (Historical Statistics 1975, E 52, 291). But most white-collar occupations involved annual contracts that were revised only irregularly, and with a lag in the short run and medium term. These price and wage contract facts encourage our optimism that these 1875 observations did not differ much from 1870.

• Child, twenty-eight, male, illiterate, in the IPUMS occupational group “farm laborers, wage workers, agriculture,” earning an estimated $146 over a 280-day year2 • Child, nineteen, male, illiterate, in the same IPUMS occupational group, again earning $146 • Head/householder, forty-five, female, in the IPUMS group “keeps house/housekeeping at home/housewife,” with zero labor income of the national product accounting type • Child, twenty-two, male, illiterate, in the same IPUMS group as his brothers, earning $146 • Child, twenty-six, male, literate, in the same IPUMS group as his brothers, earning $146 • Child, sixteen, male, illiterate, in the same IPUMS group as his brothers, earning $54, given his younger age group

Since her household had no farm of its own, it had no farm profits to estimate. And since she reported no assets at all, there was no property income. The total household income therefore equaled the labor earnings, or $638 for the year. If only the estimation were so easy for the farm operator households of 1870. Unfortunately, new challenges arise for the incomes of 1870 farm operators. Craig’s estimated patterns of their own-labor incomes, which were so convenient for 1860, might not apply to 1870, in view of the Civil War disruption to the entire economy, especially in the South. Farm operators’ incomes for the South are harder to unlock because the 1870 census of agriculture offered nothing like the coverage exploited by Craig for the North in 1860. Our struggle with the estimation of farm operator incomes requires its own separate explanation, in appendix G. The remainder of appendix F will focus on the other main challenge introduced by emancipation: determining the racial differences in rates of pay.

2 The farm wage rates are based on Lebergott’s data and the assumption that blacks got 96 percent of the wage rate received by white farmhands. The evidence for this assumption is reviewed below.

A GUIDE TO THE 1870 INCOME ESTIMATES

•

321

WAGE RATES BY RACE CA. 1870 One of the biggest data gaps in US income history is that relating to black incomes between 1861 and 1940. We have only scraps covering these eight decades. We know that black incomes were held down by two huge disadvantages right after emancipation: blacks had almost no income from property, and they were relegated to lowerpaying occupations. Did they also suffer wage discrimination within occupations? This section focuses narrowly on the part of the larger income gap that arose from pure wage differences for given occupations in and around 1870. It appears that racial wage discrimination within any given occupation was small.

Unskilled Workers in General Gavin Wright and others have cited statements supporting the null conclusion of equal pay within each of the main occupational categories. For example, Wright notes that the southern unskilled white wage was almost as low as the wage for blacks. Planters told the Industrial Commission in 1900: “I think we give them about the same thing. . . . If there is any difference I don’t know it.” USDA farm wage surveys taken between 1899 and 1902 confirm that racial differences were small, averaging 8 percent on a state basis. An earlier survey by the Bureau of Labor Statistics of North Carolina (1887) [found] that in ninety-four of the ninety-five counties [surveyed], the landlords responded no .  .  . to the question whether there were racial wage differences. And “in seventy-seven of the ninety-five counties, the tenants and laborers also indicated that no differences prevailed in wages paid to whites and blacks. . . . “[Racial] equalization [of ] the unskilled wage was the rule more often than not, an equilibrium reflecting the competitive pressures of a market. . . . “[T]hose blacks who entered the wage labor market did about as well as the white laborers in the same market.” “Specific responses to the 322

•

APPENDIX F

question of whether wages differed for blacks and whites in a given locality almost always were negative.”3

Similarly, when studying the wage rates paid, without board, for ordinary laborers in twelve southern states, 1898–1902, Robert Higgs found that whites got paid 8 percent more than blacks, both in and out of the harvest season.4 Thus for southern unskilled labor, blacks received 92.6 percent (or 1/1.08) of what whites received for the same labor. Teachers’ Salaries Howard Rabinowitz studied the racial discrimination among teachers. His conclusion was that there was no wage discrimination ca. 1870 in the southern public school sector, but that it emerged after black teachers and administrators took over black schools in the 1890s. After this, the labor market segmented and the wage gaps got quite large.5 Yet even in 1890, black teachers still received 95 percent of the salary of comparable white teachers, according to Margo’s study of six southern states.6

Racial Wage Differences for Hired Farm Labor, 1879–1880 The sector employing the greatest number of blacks in the late nineteenth century was, of course, agriculture. Fortunately, southern agriculture has received an exhaustive study for 1880. Ransom and Sutch 3

G. Wright 1986, 68, 70, 96. Higgs 1972, 1975, 1977. 5 For details on his evidence regarding roughly equal pay ca. 1870, see Rabinowitz 1974, 582–86. The Rabinowitz sample consisted of the following five large cities: Atlanta, GA; Montgomery, AL; Nashville, TN; Raleigh, NC; and Richmond, VA. “Educational developments in these cities mirrored those taking place elsewhere in the region” (ibid., 566). Among the fi ndings are the following. “As late as August 1874 the salaries of teachers .  .  . in Atlanta’s white and Negro schools were similar” (ibid., 583). “The scant information available for Raleigh [1883–84] suggests that teachers of white pupils enjoyed a slight edge” (ibid., 584). “During the next few years [after 1874] there was no indication of discriminatory compensation [in Richmond]” (ibid., 584). “Nashville . . . maintained a roughly equal pay scale throughout most of the 1880s” (ibid., 585). Up to 1874, “the wages were approximately the same [in Montgomery]” (ibid., 585). By the 1890s, “it was cheaper to hire black teachers” throughout the South, and Rabinowitz’s evidence here is quite persuasive (ibid., 582–86). 6 Margo 1990, 26, table 2.6. 4

A GUIDE TO THE 1870 INCOME ESTIMATES

•

323

have amassed a sample of 10,265 farms from the US Agricultural Census of 1880, covering nine southern states.7 As a clue about the differences in weekly pay rates for black versus white hired farm laborers, we have trimmed the sample in several ways, guided by warnings given in the Ransom and Sutch data fi les. Starting from the 10,265 farms, • We restricted the sample to the 5,891 cases of high-quality wage data (their “wagequ” = 1). • We restricted the sample further to include only observations used by Ransom and Sutch for a similar purpose. They say in their hired labor notes: “To convert the wage bill data into an estimate of hired labor, we used the farms that reported both weeks of labor and wages to get an estimate of the wage rate on that farm.” That brings us down to 1,172 observations. • Setting aside 50 farms where there seem to have been recording errors in the data on labor hiring and pay, we then divided the sample further, separating 336 farms where only whites were hired, 602 that hired only blacks, and 184 farms where both were hired. This separation allows us to reveal weekly pay rates for whites and blacks separately.

Once these rates of pay were estimated from the hiring-only-whites farms and the hiring-only-black farms, we then ran a cross-check to see whether our implied white and black wage rates correctly predicted the total wage bills for farms reporting separate weeks of hiring whites and blacks, but only a single aggregate for the wages paid. That cross-check was fairly successful: for the 184 farms hiring both races, we got an (unweighted) aggregate wage bill that was only 2.9 percent higher than the reported total, and the correlation between predicted and average wages was 0.905 for these 184 farms. Our estimated rates of pay, in dollars per week for 1879–1880, compare with Lebergott’s race-blind averages, as shown in table F-1. The black–white pay ratio of 0.96 seems usable for farmhands in general, given other scraps of evidence available. Similarly, 0.95 seems reasonable for nonfarm occupations. 7 Ransom and Sutch 2001. A subset of this sample is the cotton South sample that they used in their book One Kind of Freedom. Sutch has kindly supplied the fuller sample to us.

324

•

APPENDIX F

Table F-1 Southern Farm Wage Rates by Race, 1879–1880 Lebergott average, no race given, monthly $, with board

Weekly $ = month/4.346, with board

From the Ransom and Sutch sample (dollars per week) Whites

Blacks

Black–white ratio

South Atlantic

8.81

2.03

—

—

—

Delaware

—

—

—

—

—

Maryland

10.10

2.32

—

—

—

Virginia

8.43

1.94

—

—

—

West Virginia

11.71

2.69

—

—

—

North Carolina

8.78

2.02

2.35

2.56

1.09

South Carolina

7.95

1.83

2.23

1.70

0.76

Georgia

8.47

1.95

2.69

2.92

1.09

Florida

9.26

2.13

2.22

2.57

1.16

East South Central

10.16

2.34

—

—

—

Kentucky

10.16

2.34

—

—

—

Tennessee

9.58

2.20

3.19

2.85

0.89

Alabama

9.38

2.16

2.28

2.19

0.96

Mississippi

10.24

2.36

3.06

2.79

0.91

West South Central

12.90

2.97

—

—

—

Arkansas

13.03

3.00

—

—

—

Louisiana

12.26

2.82

3.33

2.85

0.86

Texas

13.31

3.06

3.49

3.32

0.95

2.28

2.65

2.55

0.96

(Unweighted average)

(Weighted by numbers of farms in the state)

Nine-state average

Finally, in applying such ratios to the race-blind averages, we calculate the white wage (wa) and the black wage (wb) thus: using the identity for the observed overall race-blind wage rate, W = wa(1 − B) + wbB, where B is the nonwhite share of all workers in this category, we get for farm laborers wa = W/(1 − 0.04 B) and wb = 0.96 W/(1 − 0.04 B). Similarly, we apply the same equation for nonfarm workers, but with 0.95 in place of 0.96, and using the nonfarm occupations’ separate blackemployment shares (B). Since there is no obvious reason to let the racial wage ratio (0.96 for farm, and 0.95 for nonfarm) vary by region, we apply them in all regions, assuming in addition that the same ratios applied to all nonwhites, and not just to blacks. The regions will, of course, differ in the overall income ratios of nonwhites to whites, because the shares of nonwhites in different occupations will vary, as will the relative nonwhite disadvantage in property incomes. Of course, for exclusively white occupations, wa = W, and for exclusively black occupations, wb = W.

326

•

APPENDIX F

APPENDIX G

Farm Operators’ Incomes in 1870

T

his appendix describes our strategy for estimating own-labor income for all farmer-headed households in the IPUMS sample in 1870.1 For the non-South, it adapts Craig’s estimation procedure that we used for 1860, as described in appendix E.2 The more severe data constraints for the South in 1870 require that we borrow some 1880 patterns within the region to shape estimates for 1870. Of all the estimation projects for the years since independence, this relating to 1870 farm operator incomes is the shakiest and the most in need of further research. Fortunately, this is only true of farm operator households for 1870. Our task is to estimate the earnings or marginal labor products of different farm households in 1870, by region and household attributes, and in particular to extract estimates of the residual profits received by farm operators for their risk taking and entrepreneurship.

NORTHERN AND WESTERN FARM OPERATORS Arguably, the relative labor and profit earnings of different farm households in the North and West in 1870 could have resembled patterns in Craig’s 1860 sample. We could apply the same procedures for 1870 as we did for 1850, again linking their earnings patterns to those of Craig’s 1860 patterns by rescaling according to movements in the adult male farm wage rate. More precisely, for a household labeled as a farm in the 1 Reminder: in this book and for free households, the phrase “own-labor income” refers to the value of one’s labor earnings plus the pure profit residual. 2 Craig 1991.

1870 IPUMS census sample, each extra member could have added the same marginal product relative to an adult male laborer of the same region, and farm profits could have had the same ratios to the adult male wage, as in 1860. Table G-1 presents our estimates for the northern and western regions. To illustrate our procedures for northern farm operators, consider the case of IPUMS serial number 874501, an Illinois white adult male farm owner, with a household of seven, including a hired farmhand. The HH was listed by the IPUMS as having the occupation “farmers (owners and tenants).” Since he was considered a farm operator, his total income includes “farm profit.” The household members and their attributes given by the IPUMS were • Spouse, thirty-one, female, literate, in the IPUMS occupational group “keeps house/housekeeping at home/housewife,” imputed earnings = $68.38 • Child, twelve, male, literate, imputed earnings = $27.02 • Head/householder, forty, male, literate, “farmers (owners and tenants),” imputed labor earnings = $202.87 • Child, four, female, illiterate, imputed earnings = $8.36 • Other, no relative, twenty-one, male, literate, “farm laborers, wage workers,” imputed earnings = $244.22 • Child, seven, female, illiterate, imputed earnings = $27.02 • Child, two, male, illiterate, imputed earnings = $8.36

The farmer’s individual labor income of $202.87, as given in table G-1, was based on Craig’s 1860 estimates and the general rise in wages in Illinois from 1860 to 1870. Imputing labor earnings to the other household members, again based on Craig’s technique, gives $586.25 in total household labor earnings. Since the Illinois farmer appears to have worked more than eighty acres, the Craig technique assigns him a “large farm” profit, which we estimate at $422.69 for the East North Central region. In terms of property income, this farmer earned an imputed rental of $240 on his $4,800 of declared real estate, at the assumed 5 percent rate of annual return, and $102.71 on his $1,407 of personal estate, at the assumed 7.3 percent rate. Adding this $342.71 of property returns

328

•

APPENDIX G

Table G-1 Marginal Products of Farm Labor and Pure Farm Profit (in Dollars) for the Non-South in 1870 New England

Middle Atlantic

East North Central

Missouri

Children age zero to six

−25.68

−23.68

9.35

−6.39

Children age seven to twelve

28.13

25.95

30.21

27.05

Teenage females

28.30

26.11

43.25

17.48

Teenage males

136.93

126.30

51.76

48.90

Adult females

190.02

175.26

76.44

146.90

Adult males age nineteen to fifty-four

363.53

335.30

202.87

193.16

Adult males, fifty-five and older

180.00

166.02

132.69

135.09

Small farm profits, HHs

289.27

266.80

179.38

117.27

Large farm profits, HHs

621.34

573.08

422.69

338.68

Nebraska

Other West North Central

Mountain

Pacific

Children age zero to six

−7.46

−7.17

−6.06

−11.57

Children age seven to twelve

31.54

30.33

25.65

48.95

Teenage females

20.39

19.60

16.58

31.64

Teenage males

57.03

54.83

46.37

88.50

Adult females

171.31

164.69

139.29

265.84

Adult males age nineteen to fifty-four

225.25

216.55

183.15

349.56

Adult males, fifty-five and older

157.54

151.45

128.09

244.47

Small farm profits, HHs

113.37

131.47

111.19

212.21

Large farm profits, HHs

371.57

379.69

321.12

612.89

Notes: Rates assumed for the 1870 IPUMS census sample, nonsouthern states only. Each estimated coefficient equals the value we used for 1860, based on Craig (1991, table 3) times the state-specific 1860 to 1870 ratios of the Lebergott wage rates for adult male farm labor. Within the West North Central region, we separate Missouri and Nebraska from the others (Minnesota, Iowa, and Kansas) in order to display our assumed values for these two states not covered in the Craig 1860 sample.

to the household labor earnings ($586.25) and his estimated residual profit ($422.69) yields a total income of $1,351.65 for this Illinois farm household.

SOUTHERN FARM OPERATORS’ INCOMES FOR 1870: THE LONG, INDIRECT ROUTE Southern farm incomes are the hardest kind of income to estimate reliably for 1870 and demand special attention. Comparisons with 1860 are hazardous, given the profound effects of both the Civil War and emancipation on southern agriculture. Nor can we simply assume a relationship of southern to northern farm income patterns of the sort in table G-1. True, comparing southern farms with those in the North is made a bit easier for 1870 because all laborers were free, unlike in 1860. Yet there were surely structural differences between the regions, due to differences in soil and climate, the race and literacy of farm operators, and the functioning of capital markets. One logical road is blocked by data quality problems for 1870. Logically, one might have replicated Craig’s work using the 1870 agricultural census. Unfortunately, the 1870 agricultural census was weak and underreported, especially in the South. Thus, we need a different approach to southern farm incomes. The Strategy for Indirect Estimation Our strategy for contrasting southern and other regions’ farm earnings and profits bears a slight resemblance to the assumption we used about Southern farms for 1860. There we started with a proportionality assumption: For 1860, in each Southern region, we assume that the wage rate of each demographic type of farm laborer was in the same proportion to that of an adult male free farm laborer as in a corresponding Northern region, and the profit per farm was also proportional to that same adult wage, South versus North. For 1870, we use a similar proportionality idea, but here we assume proportionality only between the East North Central region and a

330

•

APPENDIX G

baseline southern farmer in Tennessee alone. Here are the specific assumptions about labor earnings and profits on southern farms in 1870: • In 1870, the wage (or marginal product) of an adult male, white member of a farm’s labor force in Tennessee differed from that in a nearby northern region in proportion to the Lebergott wage rates.3 • Profits for white farm operators in Tennessee also differed from those in the nearby North in the same wage-based ratio. • Within the South in 1870, farm wages and profits differed by state, race, tenancy type, and household size according to patterns revealed by Ransom and Sutch for 1880, to which we turn shortly.

We use Tennessee as a doorway to the variegated South. Of all the southern states for which Ransom and Sutch have sampled farms (see below), Tennessee was one of the most grain oriented, like the East North Central region. In 1880, it derived only about 15 percent of its crop value from cotton, and not much from livestock. Tennessee will therefore be linked to the East North Central patterns of marginal productivity and own-labor income, as shown by the Craig-type coefficients in table G-2. An Overview of the Whole Procedure, for One Georgian Farmer The numbers given for labor earnings and profits of southern farm operators in 1870 emerge only after one hacks through a peculiarly dense thicket. We first map the estimation steps by flying over the thicket, taking an aerial snapshot of the farm operated by a white Georgian in 1870, with a general description of the steps for estimating income, before the rest of this appendix hacks through the thicket itself.

3 Lebergott 1964. In 1870, the average annualized 1870 farm wage rate for Tennessee was $12.86 a month “with board” versus $16.94 “with board” in the East North Central (ibid., 539). Multiplying by 1.5 for the value of board and assuming a work year of 280 days yields our farmhand wage rate of $207.07 versus $272.77 for the East North Central region—a ratio of 0.759. For a southern tenant farm operator, as opposed to a farmhand in the East North Central, this ratio should be 0.873, according to our regression results below.

FARM OPERATORS’ INCOMES IN 1870

•

331

Table G-2. Marginal Products of Farm Labor and Pure Farm Profit (in Dollars) for Tennessee in 1870 Children age zero to six

$8.16

Children age seven to twelve

$26.36

Teenage females

$37.75

Teenage males

$45.18

Adult females

$66.72

Adult males age nineteen to fifty-four

$177.07

Adult males, fifty-five and older

$115.81

Small farm profits

$156.56

Large farm profits

$368.93

Notes: Rates assumed for farm operator households in the 1870 IPUMS census sample, Tennessee only, extrapolating with Craig’s technique from the 1860 Census of Agriculture. These are race-blind averages, since they are based on Lebergott’s race-blind wage rates.

The head in the household with IPUMS serial number 634201 is a white adult male farm owner from Georgia. His household members and the final incomes we attribute to them are: • Head/householder, forty-nine, male, literate, in the IPUMS group “farmers (owners and tenants),” with imputed individual labor earnings of $194.94 • Child, female, eighteen, nonwhite [sic] illiterate, with imputed labor earnings of $37.18 • Child, female, sixteen, illiterate, with imputed labor earnings $37.18 • Child, male, nineteen, illiterate, in the IPUMS group “farm laborers, wage workers,” with imputed labor earnings of $145.81 • Spouse, female, forty-two, literate, in the IPUMS group “imputed keeping house (1850–1900),” imputed labor earnings $65.71 • Child, male, ten, illiterate, with imputed labor earnings of $25.96

The income summary is: Farm operator’s own-labor income His individual labor earnings His residual profits 332

•

APPENDIX G

$1,000.51, of which $194.94 $805.57

Labor earnings of other household members

$311.84

Total labor earnings Farm profits (again) Property income Total income

$506.78 $805.57 $76.50 $1,388.85

One easy part of the calculation was the estimation of the other household members’ labor earnings of $311.84. They were determined by the Lebergott wage series for prime-age farmhands in each state in 1870, and by inter-age-group wage ratios like those in tables G-1 and G-2. The only other easy part was the estimation of the farmer’s return on property—easy, that is, once one accepts the assumed rates of return already mentioned in chapter 6 and appendix F. Our Georgia farmer earned $40 in imputed income from his declared $800 of real estate (at the assumed 5 percent) and $36.50 on his declared personal estate of $500 (at the assumed rate of 7.3 percent), or $76.50 in property income. The hard part relates to the Georgia farm operator’s own-labor incomes—that is, the imputed value of his physical labor plus his residual farm profits. Our goal of marching through the thicket will be to estimate his overall “own-labor income.” We will use regressions to infer how southern farm operators’ own-labor incomes related to their age, race, literacy, landownership status, and state of residence in the Ransom and Sutch sample of 1879–80. Then, using that state’s change in farm wage ratios between 1870 and 1880, we will come up with an own-labor income of $1,000.51 for this farmer in 1870. From that $1,000.51, we now subtract his $194.94 of labor earnings, again estimated in the manner of tables G-1 and G-2. Th is gives us the $805.57 of his residual farm profits, his return for management skills and taking risks in an uncertain postbellum southern economy. That is the aerial overview of the procedure for estimating southern farm operators’ incomes. We now march through the thicket itself.

FARM OPERATORS’ INCOMES IN 1870

•

333

ESTIMATING OWN-LABOR INCOME OF SOUTHERN FARMERS, STEP BY STEP FROM 1880 BACK TO 1870 Fortunately, Ransom and Sutch have developed a useful sample of 10,265 southern farms from the 1880 census returns, and Sutch helpfully sent us a copy of their (“turbo-charged”) sample in September 2012.4 We will run regressions on their 1879–1880 sample to estimate the effects of certain variables on the farm household’s own-labor income, including pure profits, within the South. These regressions will then yield ways of predicting how southern households’ own-labor income (i.e., labor earnings plus profits) should have compared with those in grain-oriented Tennessee, our chosen gateway state more like those of the East North Central region than the cotton South. First, we must adjust the census data in the Ransom and Sutch sample of 10,265 southern farms to prepare for the next section’s estimation of the ratio of farm profits to the imputed value of farm household labor earnings. Some adjustments are conceptual, and some simply perform housecleaning on the census data. Perhaps the biggest challenge is to answer the question, “If the owner is not the same household as the tenant or user of a particular property, who was supposed to report the values and physical quantities related to that property?” In the case of farm real estate, is the census showing us data on the owner’s total farm holdings, or just on the holding on which he is resident? And should his tenants or users of his equipment or livestock also report data on the same properties? The enumerators’ instructions explicitly called for the enumerator to attribute the farm property to only one person, who could be the owner or the tenant, judging from these 1850 instructions, which were apparently still operative in 1880:5 Under heading 1, entitled “Name of individual managing his farm or plantation,” insert the name of the person residing upon or having charge of the farm, whether as owner, agent, or tenant. When owned or managed by more than 1 person, the name of 1 person only should be 4 For its description and use, see Ransom and Sutch 2001. In appendix F, we used a trimmed part of the same Ransom and Sutch sample to comment on racial differences in the wage rates paid to farmhands. 5 US 1900 Census, 744.

334

•

APPENDIX G

entered. . . . [Regarding] improved land . . . “connected with or belonging to the farm which the assistant marshal is reporting[,] it is not necessary that it should be contiguous, but it must be owned or managed by the person whose name is inserted in the column [and similarly for unimproved land].”

Thus, the census was designed to avoid the double counting of lands and their direct outputs. The choice between interviewing the owner or the tenant was apparently made on the basis of residence. This is somewhat reassuring, yet ambiguities remain. In particular, for the farms attributed to resident owners, we are never told whether that farm was worked by a tenant, and must make some guesses.

A Troubled Starting Point Gross farm production value (“proval”). Estimating farm household income starts with a measure of gross farm output. Its main embodiment in the census was the estimated dollar “value of all farm production (sold, consumed, or on hand) for 1879,” abbreviated as proval in the Ransom and Sutch data file. At face value, this seems to approximate the gross value of farm product. Yet inferring the farm household’s gross farm output value or gross income from the reported proval is not straightforward. The first reason for suspecting the understatement of household incomes is that the census of agriculture misses the household’s nonfarm incomes. Fortunately, we believe that this will cause only slight understatements when we come to transfer farm-based coefficients to the IPUMS sample from the 1870 population census. As long as our use of the 1880 farm sample correctly links farm own-labor earnings to different attributes of the farm operator, the population census plus wage rate information can pick up the nonfarm earnings of all nonheads in the household. We find a second hint of understatement in a large number of cases in which the farm product values look too low even as strictly farm incomes. These are cases where the proval is suspiciously low given the amounts of crops harvested, the numbers of farm animals reported, or the acres of land reported. We must choose between two basic lines of interpretation: FARM OPERATORS’ INCOMES IN 1870

•

335

• Incorrect omissions. The respondent just didn’t want to bother with estimating the value of the farm’s output, which required more thought and might have been used for a direct tax. • Correct omissions, based on the split between farm owners and farm operators. To avoid double counting the output value, the enumerators may well have made decisions to attribute part of the output to the owner and part to the operator. Such a split could occur quite naturally on farms that were sharecropped. In cases of cash rental, the renting tenant presumably should have reported the gross value of output.

We will side with the second, more charitable interpretation in most cases, but not all. A large, red warning light is flashed by a crop-value test. Logically, the gross product measure (proval) should be at least as great as the sum of its reported parts. Yet we find that in more than 40 percent of the cases, the proval is exceeded by the calculated value of two to four main crops alone (cropval), which we valued at US or Pennsylvania prices. These two to four crops are corn, cotton, and the household’s two greatest-acreage additional crops for which we have national price data for 1879, when any such additional crops are listed.6 In principle, the gross value of product (proval) should also have included the valuation of final animal products sold or consumed by the farm household. The census did not directly measure the outputs of animal products at the level of the individual household. Rather, it gives us an asset value of all livestock and counts of different types of animals (cattle, horses, milk cows, mules, oxen, sheep, and swine). Here again, we find cases where the announced total (livestock asset) value is exceeded by the values implied by the animal counts and the national wholesale prices of live animals. Still, our measure derived from animal counts has used the national urban prices of livestock for sale, which would overstate the farm gate prices of the existing stock. We acquit the reported values of livestock from any clear understatement of livestock asset values, though we note these discrepancies.

6 The additional crops for which we have national wholesale prices, aside from corn and cotton, are barley, molasses, oats, rice, rye, tobacco, and wheat. For regional export crops like cotton, rice, and tobacco, the national price presumably overestimates the local farm gate price. Yet the opposite would presumably be true of corn and other grains.

336

•

APPENDIX G

Table G-3 Twelve Kinds of Agricultural Census Returns for 1879–1880 Category and number of cases*

Outcome for reported gross farm value versus value of main crops

How we interpret such cases, and how we calculate gross farm income

Owners #1

0 = proval ≤ cropval

Omit from sample, for lack of evidence of positive household labor income

Owners #2 (n = 1,465)

0 < proval ≤ cropval

Accept proval as the owner’s share of crop value, and add returns on nontilled lands, implements, and livestock

Owners #3

0 < proval > cropval

Accept proval as gross farm income

Renters #1

0 = proval = cropval

Omit from sample, for lack of evidence of positive household labor income

Renters #2 (n = 559)

0 ≤ proval < cropval

A cash tenant whose gross farm income = cropval—rent paid on farm realty*

Renters #3 (n = 474)

0 < proval > cropval

A cash tenant whose gross farm income = proval—rent paid on farm realty**

Shares #1

0 = proval = cropval

Omit from sample, for lack of evidence of positive household labor income

Shares #2 (n = 870)

0 ≤ proval < cropval

Accept half of cropval as the sharecropper’s gross farm income

Shares #3 (n = 853)

0 < proval > cropval

Accept half of proval as the sharecropper’s gross farm income, assuming he shared in more than just the leading crops

Notes: “Proval” = reported gross value of the farm’s total production. “Cropval” = calculated value of the farm’s leading crops—namely, corn, cotton—and zero to two others. The use of a half share in the last two sharecropping categories is based on Ransom and Sutch 1977. * These add up to 6,753 cases rather than the 10,265 supplied by Ransom and Sutch. The other 3,512 cases had to be edited out due to various shortcomings in the original census data. ** The rent is calculated at the 5% rate of return on land, helped by the fact that the value of the farm was given in each of these cases. For fi fteen rented farms (e.g. for farm 5565), we assumed that the tiny crop values were reported net of rent paid, and that the tenants were mainly nonfarmers.

How should one deal with the many cases in which the reported gross farm product (proval) is less than the calculated value of a few crops alone (cropval)? Reflecting on how the census interview responses could have turned out that way, we classify the households into the interpretative categories shown in table G-3 below, which then lists how FARM OPERATORS’ INCOMES IN 1870

•

337

we propose to deal with each category. For this and other reasons, discussed below, we were forced to drop a total of 3,512 farms from the statistical sample used below to estimate the determinants of the farm household’s own-labor income, simply because those 3,512 cases gave no clues about that labor income. We make use of the remaining 6,753 farms. Our assumptions are based on guesswork about how the product value anomalies arose, so they can of course be challenged. In particular, note our use of a half share of gross crop value or gross total production value in the last two categories of table G-3. While the half share seems to have been the convention in postbellum sharecropping, we have made the additional assumption that the resident sharecropper told the census marshal’s assistant the total crop value or total farm production value, not just his half.7 Our assumption seems to fit the census’s objective of capturing total output, but it is still a guess. From Gross Product Value to Value Added by the Farm Operator’s Household From the gross farm output for the surviving household observations, we should be able to derive the value added by the farm household: Value added by the farm operator’s household = gross farm product minus purchases of material inputs

In the census as well as in the Random and Sutch sample, the only material inputs to purchase consisted of reported fertilizer purchases. This was zero in a majority of cases, yet deserves to be subtracted. We could not make use of the census data on the cost of fencing. One problem is that the census did not say how much of this was purchased rather than built by the operator’s household. Another problem is that the question on fencing, newly introduced with the 1880 census, seems to have caused confusion. Some appear to have given cost numbers so high that they could not have been an annual flow of fencing costs. We had to set aside the data on fencing costs.8 7

Ransom and Sutch 1977. Another possibility about material inputs that we have considered and set aside is that we might subtract the part of grain and grass output that is used as fodder for the farm’s animals. We are checked from such subtractions mainly by the absence of data. 8

338

•

APPENDIX G

From Value Added to Own-Labor Income, including Residual Profit The next step is to derive Own-labor income = the farm household’s value added minus out-ofpocket costs of hired labor and the cost of property use, whether the property was rented in or owned

The wages for hired labor posed the other biggest problem with the 1880 data, aside from the tension between proval and cropval already discussed. The respondents had difficulty interpreting the agricultural census questions about the wage bill and the number employed. Ransom and Sutch have carefully noted numerous farms with unreliable wage bills (their wage quality codes two, four, and five). We concur, and have been forced to omit over three thousand farms on these grounds alone. We have lingering suspicions about a few dozen cases that Ransom and Sutch rated as having high-quality data (wage quality coded as one). Perhaps the respondents were estimating the labor earnings of persons within the farm household, as opposed to just those hired from outside. Yet we cannot know, and will simply accept these few dozen cases as part of the sample. Once we have derived own-labor income, we will eventually sort out residual profits from other components in order to make visible the profit component, as we have done for 1774, 1850, 1860, and the nonsouthern states in 1870. This need not, and cannot, be done within the Ransom and Sutch sample for the 1880 South, however. Rather, once we have estimated the southern determinants of own-labor income and applied them to 1870 farm households in the IPUMS, we can derive the farm profit residual by subtracting the estimated 1870 labor earnings of farm operator households. Production-Function Determinants of Southern Farm Operators’ Earnings, 1879 To find patterns in the own-labor earnings, we expand on Craig’s production-function algebra, with a more explicit introduction of the residual factor of the farm operator’s risk-taking entrepreneurship. As we noted in appendix E, Craig began with the equation for the FARM OPERATORS’ INCOMES IN 1870

•

339

exhaustion of aggregate product in competitive markets with an agricultural production function that is homogeneous of degree one. From his initial accounting equation, we separate out an entrepreneurial residual term wN X N, where the XN amount of entrepreneurship depends on the farm operator’s individual attributes, and the wN rate of reward is derived as the accounting residual. The initial equation is Y=

N −1

∑w X + w i

i

N XN ,

i=1

where Y is the nominal value of the total product, the Xis are inputs, and the wis are their marginal products competitively priced.9 Since we share Craig’s emphasis on the farm household, we divide the inputs into those purchased from elsewhere (i = 1, . . . N − 2) and those supplied within the household (i = N − 1 for all household labor, and N for the farm operator’s entrepreneurship): Y* =Y −

N −2

∑w X = w i

i

N −1 X N −1 + w N X N

i=1

Here, as in Craig’s study, Y* is own-labor earnings—that is, the measured household, labor, and profit component of the total farm product derived by subtracting out the returns to all inputs other than farm labor and the operator’s entrepreneurship. The farm operator’s return above the market value of the unskilled part of his labor, or wN , is derived as a residual—one that should depend on attributes of the farm operator. We set the farm operator input of XN at 1, and standardize these measures in a way that allows us to project 9 In pricing the nonhuman property inputs, we assume that their implicit market rates of return include zero expected capital gains. Th is differs from Craig’s assumptions, in quantity though not in concept. Craig’s measure of capital gains implicitly assumed that investors in property expected a continuation of the previous decade’s gain: “Capital gains are defi ned as the product of the average percentage increase in the value of land between 1850 and 1860 by state and the value of real estate” (Craig 1991, 72). For 1870 or 1880, we introduce no expected capital gains on property. Southern farmers could not have formed any clear asset-price expectations on the basis of the Civil War decade 1860–1870. That decade brought huge capital losses. The decade 1870–1880 brought ex post nominal capital gains, but no clear change in real values per acre, and there were modest real gains per acre in the South between 1880 and 1900.

340

•

APPENDIX G

their economic content to other states and the 1870 census. The standardization divides each term by twelve months’ average farm wage for the state in which the farm is located.10 We interpret the labor input XN−1 as depending on the household size, with wN−1 as its coefficient of marginal productivity relative to the state’s average farm wage. Our dependent variable, the farm’s labor-and-profit term, further depends on observed attributes of the farm operator. Hence, the equation to be fitted involves these variables: Dependent variable Y* = the current dollar value of the farm household’s own-labor earnings, including residual profits, divided by twelve months’ average farm wage for that state. Independent variables influencing farm own-labor earnings = a polynomial expression for labor XN–1 involving: numhse = the number of persons in the household11 numhsesq = the number of persons in the household, squared numhsecu = the number of persons in the household, cubed with or without fi xed effects for nine state dummies representing the local market conditions in Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Texas, and Virginia. (The omitted sample state = Tennessee.) Independent variables influencing farm’s entrepreneurial product, or Y*−wN−1 XN−1: literate = 1, if farm operator is literate under19 = 1, if farm operator is younger than age 19 over54 = 1, if farm operator is older than age 54

10

Lebergott 1964, 539. Our handling of household members’ labor contributions is constrained by the differences between the 1880 farm sample and the population census. The agricultural census, as used for the southern farm sample, did not list individuals other than the HH. Furthermore, its counts of the number of workers in the home seem flawed, as Ransom and Sutch have noted. We therefore have only the number of persons in the household (numhse) as a clue to the productive contributions of household members to the farm. Th is number (numhse, in polynomial form) is used as the best way to control for the productive contributions of extra household members, helping us to estimate the residual farm profits. When the resulting coefficients are transferred to the 1870 IPUMS, we will not transfer the household size effects, instead replacing them with the better information from the population census on each household member’s occupation and labor force status and the predicted wage for their age-sex category. 11

FARM OPERATORS’ INCOMES IN 1870

•

341

owner = 1, if the farm operator is coded as owner in the Ransom and Sutch sample, or has real estate worth more than ten dollars in the 1870 IPUMS12 nonwhite = 1, if the farm operator is not white. ownnotwh = the product of owner and nonwhite ov54ownr = the product of owner and older than age 54 farmsize, a binary that = 1, if the value of realty (farmval in the Ransom and Sutch sample) is greater than the state average per farm (avelbpac), and owner = 1 (patterned after Craig’s largefarm variable)

The resulting regression equations, equipped with some interaction terms, are shown in table G-4 below. Some of these terms might be interpreted as reflecting one or another type of factor income. For example, one might interpret the constant term and the household size polynomial as reflecting a return to unskilled labor, whereas the attributes of the farm operator might be thought of as reflecting the returns to generalized skill, risk taking, and entrepreneurship. The state fi xed effects, however, cannot be divided into different types of factor returns. As a corollary, farm profits cannot be separated from the other incomes of the farm operator household just by using this otherwisehelpful Ransom and Sutch southern sample for 1880. Rather, we will later divide the own-labor incomes of southern farms for 1870 between profits and returns to labor by subtracting farm labor earnings implied by the farm household’s age-sex composition from the total own-labor incomes estimated with the procedures described in this appendix.

Hooking Tennessee to the Midwest and the Rest of the South in 1870 The second equation in table G-4 below can now link Tennessee to other southern states and the 1870 census, since its variables are chosen so as to fit the set of IPUMS variables from the population census to the Ransom and Sutch farm sample variables, as closely as possible. As 12 Among tenants (nonowners), the 1880 Ransom and Sutch sample distinguishes cash renters from sharecroppers. We are forced to lump the two groups together for our present purposes, because the 1870 population census does not offer this distinction.

342

•

APPENDIX G

Table G-4 Determinants of a Farm’s Own-Labor Income in the Ransom and Sutch Sample of Southern Farms, 1879–1880 Equation (1)

Equation (2)

Literate

0.954 (0.324)**

1.008 (0.32)**

Nonwhite

0.566 (0.435)

0.031 (0.44)

Owner

2.440 (0.365)**

2.354 (0.37)**

Owner, nonwhite

−1.805 (0.602)**

−0.567 (4.54)

Under nineteen

−0.360 (0.08)**

0.485 (0.52)

Over fifty-four

0.611 (1.18)

−1.700 (0.60)**

Owner, < nineteen

−0.291 (0.05)**

−0.658 (5.29)

Owner, > fifty-four

−0.860 (1.34)

−0.670 (0.64)

Numhse

0.14 (0.51)

Numhse ^ 2

Equation (2), continued

Attributes of the farm operator

State dummies Alabama

1.710 (0.48)**

Florida

0.752 (1.06)

Georgia

1.631 (0.56)**

Louisiana

3.277 (0.66)**

Mississippi

2.092 (0.51)**

North Carolina

−0.190 (0.69)

South Carolina

3.232 (0.57)**

0.205 (0.28)

Texas

0.154 (0.56)**

−0.032 (1.01)

−0.038 (0.03)**

Virginia

0.777 (0.59)

Numhse ^ 3

0.003 (0.001)**

0.003 (0.001)**

—

—

Constant

0.775 (1.00)

Household size polynomial

2

(N), R

(6,753) 0.03

Constant 2

(N), R

−0.727 (0.86) (6,573) 0.04

Notes: Dependent variable = own-labor income divided by the 1880 state average farm wage (annual, from Lebergott). Absolute standard errors in parentheses. Numhse = number of persons in the household. The omitted categories: an illiterate white tenant farm operator age nineteen to fi ft y-four in Tennessee. (The implicit household size is zero, but can be converted to the more sensible size of one by subtracting the sum of the three Numhse polynomial coefficients.)

previewed in table G-2 earlier, a literate Tennessee white tenant farmer living alone was estimated to have $177.07 in annual own-labor income for 1870, or 1.1474 times the twelve-month average farm wage for that state in 1870. We can use the patterns revealed in the Ransom and Sutch sample for 1880 to add the contrasts between such a Tennessee farmer and other farm operators in the South. For other farm operators, we add to this Tennessee farmer’s 1.1474 times the annual farm labor wage the following extra terms: Southern farm operators’ individual own-labor earnings (including profits) as a multiple of the average farm cash wage “with board” in their state, in either 1880 or 1870, which = the literate Tennessee white tenant’s 1.147413 – (1.008, if illiterate) + (0.031, if nonwhite) + (2.354, if an owner) – (1.700, if a nonwhite owner) – (0.567, if under age 19) + (0.485, if over age 54) – (0.658, if an owner under age 19) – (0.670, if an owner over age 54) + (1.710, if in Alabama) + (0.752, if in Florida) + (1.631, if in Georgia) + (3.277, if in Louisiana) + (2.092, if in Mississippi) – (0.190, if in North Carolina) + (3.232, if in South Carolina) + (0.154, if in Texas) + (1.508, if in Arkansas) + (0.777, if in the Upper South Atlantic—that is, Delaware, the District of Columbia, Maryland, Virginia, or West Virginia)14

13 In the baseline case of Tennessee, 83.4 percent of white farm tenants in the 1880 Ransom and Sutch sample were literate. Among all Tennessee farm operators, white or black and owner or tenant, the literacy rate was 79.5 percent. 14 We extrapolate the regression results in equation (2) of table G- 4 to cover southern states not included in the Ransom and Sutch sample. Thus, the 0.777 added for all of the Upper South Atlantic is actually the coefficient for Virginia. For Arkansas, the 1.508 is the simple average of the coefficients for Louisiana, Mississippi, and Texas. We equate Kentucky to the baseline case of Tennessee, thereby omitting any state fi xed effect. Note that since these coefficients are adjustments to own-labor income in ratio to the average annual state farm wage, then we need to multiply later by the state farm wage rates in each separate state. For example, if a farm operator in the 1870 IPUMS sample is from Maryland, multiply by Maryland’s twelve-month farm wage, and not Virginia’s, even though the coefficient used here is borrowed from Virginia.

344

•

APPENDIX G

We will omit the numhse polynomial terms for farm household size, since we will later use the Craig-based relatives for extra household members to scale the earnings of different nonheads of households (e.g., children in the seven to twelve age range) to those of an adult male (in the ratios implied for Tennessee 1870 in table G-2 above). These values for nonheads are then added onto the individual own-labor earnings (including profits) for the farm-operating HH, to assemble the farm household total own-labor earnings. So for the case of our Georgia farm owner in 1870, this 1879–1880 regression pattern was used only to estimate the farm operator’s individual own-labor earnings as a multiple of the Lebergott farm wage rate for Georgia. The regression-predicted multiple was 5.1364 in his case, and the Georgia average wage for twelve months, ignoring board value, was $194.94. The product of these was his own-labor income, leaving that $805.57 of residual profit. As a side note, the regressions in table G-4 yield predictions that in some cases imply negative own-labor income, both for the 1880 Ransom and Sutch sample itself and the 1870 IPUMS one. Fortunately, the negative income anomaly applies only to some nearly empty cells. Farm income is predicted to be negative only for illiterate nonwhite farm owners who are young or old, and only in three states. That these are rare is clear from the rarity of such cases: Number of such cases In Ransom and Sutch, 1880

In IPUMS, 1870

Illiterate, nonwhite farm owner, under nineteen

0

0

Illiterate, nonwhite farm owner, over fift y-four

3

1

Illiterate, nonwhite farm owner, under nineteen

0

0

Illiterate, nonwhite farm owner, over fift y-four

11

6

Tennessee Out of 44,252 East South Central population

North Carolina Out of 58,777 South Atlantic population

(continued) FARM OPERATORS’ INCOMES IN 1870

•

345

Number of such cases In Ransom and Sutch, 1880

In IPUMS, 1870

Illiterate, nonwhite farm owner, under nineteen

0

0

Illiterate, nonwhite farm owner, over fift y-four

8

3

Texas Out of 20,566 West South Carolina population

Sum = 22 Out of 6,753 total farms

The negative farm income prediction is rare enough that it could be a realistic prediction of the rare cases of truly negative profits, even though the actual cases of negative profits would often not match these categories, either in 1880 or 1870. Alternative Shaky Bridges to Tennessee The bridge from southern farms in 1880 to the East North Central region in 1870 is perhaps the shakiest part of our estimates for 1870. It is made even shakier by our extrapolating Craig’s marginal product estimates from 1860 to 1870 for the non-South. As Sutch has advised us, we need to offer some sensitivity analysis here to provide a sense of how some plausible ranges of error could affect our overall results for regional average incomes and inequality. While we cannot weigh all possible errors here, we can focus on the fragile Tennessee bridge described at the start of this appendix. Suppose that our entire set of South incomes for farm operators could be in error, in either direction, by the specific number 14.74 percent, the guesstimated shortfall of Tennessee’s marginal labor product compared with the marginal labor product on farms in the East North Central region. Here are two alternative errors. First, suppose that our Tennessee farmer prototype, and all other southern farm operators with him, really did not get that premium of 14.74 percent over the Lebergott “with board” cash farm wage. That is, suppose he only earned $12.86 a month 346

•

APPENDIX G

Table G-5. Alternative Bridges to Tennessee: Three Different Sets of Marginal Products for a Tennessee Tenant in 1870 Baseline Bridge A

Bridge B

Bridge C

Children age zero to six

7.10

8.16

9.36

Children age seven to twelve

22.93

26.36

30.25

Teenage females

32.83

37.75

43.31

Teenage males

39.29

45.18

51.84

Adult females

58.03

66.72

76.55

Adult males age nineteen to fifty-four

154.01

177.07

203.17

Adult males, fifty-five and older

100.73

115.81

132.88

Small farm profits

136.17

156.56

179.64

Large farm profits

320.88

368.93

423.31

Notes: All values in dollars per year. Baseline bridge B is our preferred option, already given in table G-2 above.

for twelve months, but with his somehow being cheated out of the food that should have come with it. Meanwhile, assume no such reduction for farm operator incomes in the East North Central region. Call this lower value of southern farm income our “bridge A” from the East North Central to Tennessee and the rest of the South. Table G-5 contrasts this low version of southern farm income with our baseline “bridge B” in table G-2 above. The opposite error of having understated the prosperity of southern tenant farmers in 1870 by 14.74 percent would lead us across the upward-sloping “bridge C” to Tennessee, using the coefficients shown on the right in table G-5. How would these alternative errors have affected our estimates of regional incomes and inequality for 1870? Table G-6 shows the results. For the average per capita income of the South, there is only a span of less than 3.5 percent between our overestimating the relative income of southern farm operators by 14.74 percent (bridge A) and our underestimating it by the same amount (bridge C). For the whole United States, the span between overstating and understating income per capita is FARM OPERATORS’ INCOMES IN 1870

•

347

Table G- 6 The Limited Sensitivity of 1870 Income Results to Different Income Intercepts for Tennessee Farm Operators Baseline Bridge A

Bridge B

Bridge C

1.0000

1.1474

1.2948

South

$ 201.67

$ 205.18

$ 208.68

All United States

$ 231.67

$ 232.11

$ 233.41

South

0.427

0.430

0.426

All United States

0.363

0.370

0.370

Assumed southern intercept Income per capita

Black–white ratio of income per capita

Notes: The baseline intercept 1.1474 refers to the regression-based ratio of (own-labor income of a literate white tenant farm operator in Tennessee)/(average twelve-month wage for farm labor in Tennessee), which builds our “bridge” to the non-South farm incomes. The table’s two “0.370” ratios are not identical. That for the baseline bridge B = 0.36961, and that for bridge C = 0.36997.

less than 1 percent. One may wonder how 14.74 percent differences in either direction would yield much smaller differences in income per capita. The answer is simply that farm operator incomes were less than half of the total income, even in the relatively rural South. Table G-6 adds the effects on black–white income ratios for the South and the United States. These are also similar regardless of the range of assumptions explored here—similar enough to leave all qualitative conclusions unaffected.

348

•

APPENDIX G

APPENDIX H

Sources and Notes to Tables and Figures in Main Text

T

o avoid cluttering the text with references and citations, this section details the data sources for the figures and tables.

CHAPTER 2 Table 2-1: For further details on the derivation of the occupational shares, see “American Incomes, 1774, Baseline Part-Time Assumptions,” Worksheets (2) and (3), http://gpih. The same estimates are available through the Journal of Economic History’s supplementary materials link for Lindert and Williamson 2013a. For other supporting fi les of occupational data, see “Occupation Codes and Counts c1774,” http:/gpih. Table 2-2: The text supplies definitions of the full-time (FTE) and parttime estimates, and how they treat days worked per year. The slave retained earnings estimates include indentured servants in Maryland, the one colony that counted them in the local census. As noted in chapter 1, all pre-1870 estimates exclude Native Americans. Delaware is here included with the Middle Colonies for both years, following Jones’s sample design. Table 2-3: Gross and net incomes = our full-time (FTE) and part-time estimates of personal income, gross and net of depreciation. This culled set of other estimates omits old ones, and if a modern source offers more than one estimate, this set selects the most recent. It also selects the highest in the Jones range, as recommended by Gallman and Weiss.

McCusker’s (2001) price deflators = 97 for 1774, 151 for 1800 (if 1860 = 100), or 93.3 for 1774 (if 1840 = 100). The western states included in the Lindert and Williamson “United States (all)” estimates are Kentucky and Tennessee, plus Mississippi for labor incomes only. Table 2-4 and figure 2-2: See “American Incomes, 1774, Baseline PartTime Assumptions,” http://gpih, where the Lorenz curves and inequality parameters are derived and compared on the last five worksheets. Note: Yes, median was greater for New England.

CHAPTER 3 Assumptions behind the Backcast from 1774 to the Seventeenth Century • For New England and the urban Middle Colonies, we assume that the 1774 occupational mix within each region and by urban/rural location, applied to all earlier years as well; free-labor incomes for all occupations moved in proportion with the wage series for craftsmen and seamen (urban), or farm laborers (rural); unemployment rates and the resulting deviations from wage-based estimates of free-labor income were comparable at all benchmark dates; the net rate of return on income-producing wealth remained at 6 percent, from ca. 1774 back to ca. 1650; and depreciation rates on different kinds of assets were fi xed at the rates assumed for 1774. • For the Middle Colonies as a whole, we assume that there was no change in the ratio of the Rosenbloom and Weiss (2014) estimates of real incomes per capita to the true values, 1720–1774. • For the Upper South, gross farm income per farm was in the same ratio to the total regional income over the whole period, ca. 1675–1774. • For the Lower South, there was no change in the ratio of the Mancall, Rosenbloom, and Weiss (2003) estimates of real incomes per capita to the true values, 1720–1774. • In all regions, slaves’ retained earnings (i.e., actual consumption) kept the same shares of the corresponding free-labor earnings in the earlier years as they did in 1774. 350

•

APPENDIX H

Table 3-1: The slow-growth estimates all use the “controlled conjecture” method about sectors and productivity growth. The fast-growth estimates are only those that use extensive data. Furthermore, the list excludes two extreme outliers—one high (Ball 1976; 1.27 percent per annum) and one low (Waters 1976; –0.30 percent per annum). Table 3-2: The source is Jacks, O’Rourke, and Williamson 2011, tables 3 and 5. Table 3-3: Calculated from the slope in a regression on time. Table 3-4: NE-a: The New England artisans figures are from Gloria Main 1977, with interpolations between her broad averages. The Boston seamen’s monthly wage is from Nash 1979, 392–94. Robert Allen and we have made some use of William Weeden’s (1890) Boston wage data in choosing how to interpolate Main’s series. The Weeden data are quite sparse, however. NE-b: Gloria Main’s and Jackson Main’s colonial New England probate sample, http://gpih. Regression-adjusted to age forty-five. For a related occupation-specific series on Boston probated wealth from 1685–1715 to 1756–1775, see Nash 1979, 397–98. NE-c: Gloria Main’s (1977) adult male farm wage rate, with interpolations between her broad averages. MC-a: Middle Colonies series indexed to the simple average of Philadelphia laborer wage rate and seaman wage rate, 1774 = 100, from Nash 1979, 392–94. See also the Smith (1981, 1990) series for Philadelphia, 1750–1775. MC-b: The Middle Colonies urban wealth indexes are derived by using the Philadelphia probated personal estate data by occupation in Nash 1979, 396–98. Realty not included. The adjustments for probate bias were necessarily less complete in this case than in the case of the Mains’ probate sample for New England. We could reweight to the occupational group weights of the living, but we could not adjust for differences in age at death. Thus, the average wealth for a given occupation in a given period is an average for probates at death, not probates for a given age. MC-c: Conjectural estimates for the Middle Colonies from Rosenbloom and Weiss 2014. SOURCES AND NOTES TO TABLES AND FIGURES

•

351

Ches-a: For the sources for these Chesapeake or Upper South figures, see Carr, Menard, and Walsh 1991; Walsh 1999, 2010. Guided by the literature just cited, we assume that 22 percent of the gross farm income came from tobacco sales, 11 percent from grain sales, and the remaining 67 percent from the production of products consumed on local farms. This farm-consumed production was assumed to grow at 0.1 percent per annum per capita. For our calculations, see variant 4 in “Chesapeake Income Clues, 1650–1774a (Revised),” http://gpih. LS-a: For the source of our Lower South figures, see Mancall, Rosenbloom, and Weiss 2003.

Table 3-5: For the underlying probate data, see “Mains’ 18,509 Probates for Colonial New England, 1631–1776,” http://gpih. The costs of bare-bones consumer bundles are annual series underlying Allen, Murphy, and Schneider 2012, kindly supplied by Allen. The time periods are quarter centuries centered on the year shown (e.g., “ca. 1650” = probates from 1638 through 1662), except that “ca. 1770” corresponds to 1763–1776. In the first period (ca. 1650), there were only twenty-five widows in the probate data set. The “hinterland” consists of all sampled towns founded later than 1638, thus excluding Boston, Hartford, New Haven, and eastern coastline and river towns. For more extensive probate regression results, see “Mains’ New England Probate Data, Regression Equation” and “Mains’ New England Probate Backcast Results,” http://gpih. Table 3- 6 and figure 3-3: For the methods of derivation, see the text, table 3-4, and the regional “backcasting” files, http://gpih. The underlying data were averaged over varying time periods, shaped by data availability. The periods over which incomes and the cost of living were averaged are 1638–1662 for “1650,” 1663–1687 for “1675,” 1688–1712 for “1700,” 1713–1737 for “1725,” 1738–1762 for “1750,” 1763–1774 for “1770,” and finally just 1774 alone. These longer-period averages will differ from corresponding numbers in table 10-1 and figure 10-2, which are derived using five-year price averages as consistently as possible. For the lower South, the estimates are those of Mancall, Rosenbloom, and Weiss 2003; the year “1725” is actually 1720, and “1750” is 1740.

352

•

APPENDIX H

Similarly for the Middle Colonies as a whole, taken from Rosenbloom and Weiss 2014. All colonial costs of purchasing a bare-bones consumer bundle are based on annual data kindly supplied by Allen, underlying the averages described in Allen, Murphy, and Schneider 2012. The Great Britain series is the Broadberry et al. 2011 nominal GDP divided by the annual purchasing power parity data for London and southern England underlying Allen and Weisdorf 2011. Figure 3-1: McCusker and Menard 1985. The population totals do not include Native American populations outside European settled areas. Figure 3-2: Carter et al. 2006, series Aa22, Aa700, Eg1, and Eg60– 64, drawing on Bridenbaugh 1938, 1955. Figure 3-3: Same as for table 3-6. Figures 3-4 and 3-5: Annual series supplied by Allen. Their nature is explained in Allen, Murphy, and Schneider 2012. Allen’s bare-bones and respectability English cost-of-living bundles for 1290 to 1801–1803 are detailed in Broadberry et al. 2015, 333–39. CHAPTER 4 Table 4-1: Native Americans are excluded. The 1774 entries repeat the figures in table 2-2. See appendix A for a detailed description of the 1800 estimates. The baseline estimates use the full-time assumptions of 313 days per year. The part-time assumptions are described at length in chapter 2. Delaware is here included with the Middle Colonies for both years, following Jones’s sample design. Table 4-2: The 1774 estimates repeat the figures in table 2-3. Baseline and part-time are defined in the notes to table 4-1. Our culled set of alternative or competing estimates omits old estimates, and if a modern source offers more than one estimate, we have selected the most recent. We have also selected the highest in the Jones range, as recommended by Gallman and Weiss.

SOURCES AND NOTES TO TABLES AND FIGURES

•

353

McCusker’s (2001) price deflators: if 1860 = 100, 1774 = 97, and 1800 = 151; if 1840 = 100, 1774 = 93.3, and 1800 = 145.2. The western states included in the Lindert and Williamson “United States (all)” estimates are Kentucky and Tennessee, plus Mississippi for labor incomes only. Table 4-3: The 1774 estimates repeat the figures in table 2-4. The 1800 property and total estimates can found in appendix A, and “Property Totals, 1798–1800,” “Own-Labor Incomes, 1800,” and “1800 Total Incomes, Using Baseline Part-Time Assumptions,” Worksheet (4), http:// gpih. The estimates displayed here in table 4-3 use the full-time definition of personal income; using the part-time definition would give similar absolute effects of differences in the rates of return on the overall incomes. Table 4-4: The 1774 estimates repeat the figures in table 2-5. The figures in parentheses are percentages of the all-three-regions average. All estimates for 1774 and 1800 use the part-time work year assumption to conform with the procedures that seem implicit in the Weiss and Easterlin 1840 estimates. The “baseline” estimates for 1774 and 1800 use a 6 percent net rate of return, whereas the alternative estimates for 1800 use 8 percent. The 1840 estimates start with Weiss’s (1992, 27, table 1.2) national estimates, and derive regional relatives from the state-level relatives in Easterlin (1960, 87–98). The three-region totals are derived from the regional averages. In 1800, the South Atlantic excludes Florida, and the Middle Atlantic includes Delaware and the District of Columbia. The price deflator is the McCusker composite price index up to 1800, spliced onto the Weiss index from 1800 onward (Carter et al. 2006, series Ca13). See notes to table 4-2. Table 4-5: All calculations are based on part-time annual earnings, and they include slaves. CHAPTER 5 Table 5-1: The totals refer to the entire population, free plus slave. All labor earnings use part-time assumptions. See text. The South now 354

•

APPENDIX H

includes Florida. Slave incomes in the West North Central were earned in Missouri. See “Regional Income Totals in 1850,” http://gpih. Table 5-2: See notes to table 5-1 and “Regional Income Totals in 1860,” http://gpih. Table 5-3: See the sources cited in American Incomes folder, http://gpih. The growth rate for western Europe is an unweighted average of the rates for the seven countries, and all but the United Kingdom are taken from Maddison 2010. The UK figures refer to Great Britain, and are from Broadberry et al. 2012. Table 5-4: For 1800, the original thirteen = 100; for 1840–1860, United States = 100. Lindert and Williamson is taken from American Incomes folder, http://gpih. The Easterlin relatives are built as follows: the constant-price income per capita estimate for the whole United States in 1840 is from the Weiss (1993b) “broad” measure. These totals then were multiplied by regional relative income per capita as implied by Easterlin (1960, 97–98, table A1, variant A). All other national income estimates are our own for part-time national personal income. Table 5-5: See “American Income Inequality, 1850,” http://gpih. Table 5-6: See “American Income Inequality, 1860,” http://gpih. Table 5-7: Taken from the Lindert and Williamson fi le “Property Inequality, 1860, and Comparisons,” and tables 2-2, 5-2, and 6-1. Table 5-8: Taken from the following http://gpih files: 1774: “American Incomes, 1774, Baseline Part-Time Assumptions,” Worksheet (3) 1850: “American Income Inequality, 1850” 1860: “American Income Inequality, 1860”

Table 5-9: The wage data sources are those underlying the social tables for 1800 and 1860. See “Wage Rate Data, 1800” and “Wage Data Summary, 1860,” http://gpih. Rural unskilled work is represented by farm labor. Artisan and construction wages are employment-weighted averages by region: North = New England, Middle Atlantic, East North SOURCES AND NOTES TO TABLES AND FIGURES

•

355

Central, and West North Central; South = South Atlantic, East South Central, and West South Central; United States = North, South, and other. All earnings are for males, and include in-kind payment and full-time assumptions. Average urban earnings exclude white collar, for want of sufficient data on high-level professional salaries for 1800. Table 5-10: The 1860 urban and rural inequality estimates are from unreported extra Stata runs decomposing the IPUMS incomes used in the estimates of table 5-6. The urbanization rates are taken from the census. The d(Gini) is calculated as the 1860 urban–rural inequality difference times the change in the urban share. Table 5-11: The data underlying the social tables for 1800 are from “Own-Labor Incomes, 1800,” http://gpih, and for 1860 are from “OwnLabor Incomes, 1860,” http://gpih. The blue-collar category includes skilled workers in construction and male skilled artisans. The white-collar category includes all such male workers. The unskilled category includes males only. Figures 5-1 and 5-2: The nominal GDP for the thirteen colonies, the United States, and the Atlantic regions are those developed in this and earlier chapters, with supporting files at http://gpih. The nominal GDP and population of Great Britain are from Broadberry et al. 2012, with the real series also available in Broadberry et al. 2015, 226–44. The cost-of-living comparisons used in figure 5-2 are based on the Allen annual bare-bones series for London, Philadelphia, and South England, as in chapter 3, and not the Sutch price series used in table 5-3. The dollar-sterling exchange rates, starting with 1791, are from Carter et al. 2006, series Ee618, contributed by Lawrence Officer. Figures 5-3 and 5- 4: The defi nitions of the population units changed somewhat over time. See sources for details. These are pre-fisc estimates—that is, they refer to income before taxes and transfers. The American colonies, 1774, and United States, 1850–1870, use the sources cited in the text. The figures for England and Wales and the United Kingdom in the seventeenth through nineteenth centuries are from Lindert and 356

•

APPENDIX H

Williamson 1982, 1983b, as updated in “Early Income Distributions,” http://gpih. For the figures for Holland in 1732, see van Zanden 1995. For the figures for the Netherlands in 1798 and 1914, see Soltow and van Zanden 1998, 35, 172. The Goldsmith and US Office of Business Economics series = shares of pretax money income received by “consumer units,” or roughly households. These estimates evolved through a long chain of revisions centering on the work of Selma Goldsmith and the US Office of Business Economics. The resulting estimates are now presented in Carter et al. 2006, Series Be21-26. Key links in the earlier chain are Budd 1967, xiii; Goldsmith 1967; Radner and Hinrichs 1974. The sources and numerical results are also cited in Lindert 2000. The United Kingdom, 1938–1954, estimates are from the Inland Revenue’s Survey of Personal Incomes. See Royal Commission 1977, 240–43. The income inequality series for Britain and the Netherlands, 1985–2010, are from the OECD. Figure 5-4’s series for Japan is from Moriguchi and Saez 2008, 720. The downloaded data underlying Atkinson, Piketty, and Saez 2011, starting in the 1960s and using the Gini, excludes capital gains, with post-2006 data spliced onto this series using the OECD series. The data are updated past 2005 by splicing the OECD pre-fisc Gini’s onto the Atkinson, Piketty, and Saez series.

CHAPTER 6 Table 6-1: The estimates are detailed at “1870 Regional Income PartTime,” http://gpih. For an analysis of the limited sensitivity of these totals to the “Tennessee bridge” linking the northern and southern estimates of farm operator incomes, see appendix G, and particularly table G-6. We think that the estimates of farmers’ residual profits might still be a bit high in some regions, particularly in the North. Table 6-2: The growth rate for western Europe is an unweighted average of the rates for the seven countries, and all but the United Kingdom are SOURCES AND NOTES TO TABLES AND FIGURES

•

357

taken from Maddison 2010. The UK figures refer to Great Britain, and are from Broadberry et al. 2012; 2015, 226–44. Here and in tables 6-3 and 6-6, the price deflators are 106 for 1860 and 153.1 for 1870, where 1840 = 100. These are based on the Weiss series for 1840–1860, and a ratio of 1.44 for 1870/1860, a compromise between the David and Solar 1977 consumer price deflator and the wholesale price index for 1860–1870. Table 6-3: The data are from “Regional Income Totals, 1870,” http:// gpih, which also gives per capita incomes, both nominal and real. Table 6-4: IPUMS 1870; “American Income Inequality, 1870,” http://gpih. Table 6-5: IPUMS 1870; “Labor Earnings Inequality, 1870,” http://gpih. Table 6-6: The source is the pair of IPUMS census samples for 1860 and 1870. In the South, nonwhites consisted almost exclusively of African Americans. In the non-South, they were a smaller group dominated by free blacks, American Indians, and Chinese on the Pacific Coast. The non-South slaves were those in Missouri, which is part of the West North Central region, despite being a slave state in 1860. Table 6-7: The data come from “1850–1870 Incomes by Race” and “1850–1870 Men’s Earnings by Race,” http://gpih, using the source materials cited there. The deflators used (1840 = 100) were 106 for 1860 and 153.1 for 1870, as in tables 6-2 and 6-3. A caution on the Mountain and Pacific regions: western prices were denominated in convertible gold dollars, not in greenbacks over this decade of greenback inconvertibility. This table therefore probably understates the already-high real incomes of the West relative to other regions. See Greenfield and Rockoff 1996; San Francisco price series at http://gpih. Our thanks to Hugh Rockoff for alerting us to this western price effect.

CHAPTER 7 Table 7-1: Maddison “horizontal-file_02 2010,” downloadable at Maddison 2010. Table 7-2: 358

•

APPENDIX H

1870: our own estimates of Gini and top shares, table 6-4. 1880–1900: wealth/income, Piketty and Zucman 2014; World Top Incomes Project Web site. 1910: top shares, Piketty and Saez 2003; wealth/income, Piketty and Zucman 2014. 1913–1929: top shares, Atkinson, Piketty, and Saez 2011; wealth/income, Piketty and Zucman 2014. 1915–1929: top wealth shares, Kopczuk and Saez 2004. 1929: the US income distribution among “consumer units” in 1929 is from the Goldsmith and US Office of Business Economics series. See Radner and Hinrichs 1974, 27, which cites the 1929–1954 estimates in Goldsmith and the US Office of Business Economics. The sources and numerical results are detailed in Lindert 2000.

Table 7-3: The 1870 urban and rural inequality estimates are from the IPUMS sample, as described in chapter 6, and its totals are from table 6-5. The urbanization rates are taken from the census. The d(Gini) is calculated as the 1870 urban–rural inequality difference times the change in the urban share. Table 7-4: 1. The farm series used in the urban/rural ratio “a” is the monthly farm labor with board, Carter et al. 2006, series Ba4234, based on Lebergott 1964. The nonfarm series is the unskilled wage, Carter et al. 2006, series Ba4218, from David and Solar 1977. 2. The farm series used in the urban/rural ratio “b” is the same, but the nonfarm series is the annual earnings, Carter et al. 2006, series Ba4282, from Lebergott 1964. 3. The farm series is the same, but the manufacturing series is hourly, Carter et al. 2006, series Ba4290, from Haines 1992, and 1910 is actually 1907. 4. Clerks in manufacturing include clerical staff and some managerial personnel versus production workers (Goldin and Katz 1995, table 5). 5. Relative to urban common labor, except for row 7, which uses farm labor (Williamson and Lindert 1980a, 305–9).

Table 7-5: Klein 2013, 234, table 14, panel A. SOURCES AND NOTES TO TABLES AND FIGURES

•

359

Table 7-6: Collins and Margo 2006, 135, citing Margo 1990, 21–22. The per pupil spending reflects teachers’ salaries per student in the average daily attendance in public schools. The black–white ratio of teachers per hundred students is the reciprocal of the average class size, which is also based on the average daily student attendance relative to the number of teachers. The data on school days are based on elementary and secondary public schools. Table 7-7 and figure 7-2: Note that the US average is not just a weighted average of the regional ratios, since it also reflects the width of the gap between regional average incomes. Sources for 1774: see the http://gpih file for this table, plus “American Incomes, 1774, Part-Time” and “Labor Force by Colony, ca. 1774,” http:// gpih. The colonial data permit this comparison of labor force earnings and income per capita by race. All members of the labor force are included, both free and slave. For 1800, we compare the earnings per member of the labor force, slave versus free. See “Own-Labor Earnings, 1800,” Worksheets (2)–(4), http://gpih. For 1850–1860, both measures combine the IPUMS free samples and slave samples, and aggregate free blacks and slaves together. The free-labor earnings refer to men with occupations, age twenty to fift y-nine. For 1910 and 1930, William Collins and Marianne Wanamaker (2013) estimate the average earnings for men age twenty to sixty who have positive earnings. As in our calculations for 1870 or earlier, they give people’s predicted earnings (their “scores”) based on their occupation, sex, and adult status. Collins and Wanamaker are able to distinguish those who are currently employed, thanks to the 1930 census. That is not possible for the 1850–1870 censuses or earlier materials, so that our earlier estimates refer to earning power, even of those not employed. James Smith and Finis Welch (1989, 522, table 2) provide the 1940– 1980 men’s earnings averages. Alternatively, Thomas Maloney (1994, 358) gives a ratio of 0.48 for 1940 and 0.61 for 1950. 360

•

APPENDIX H

For 1960–2010, Oscar A. Méndez Medina, working on the present project, estimated the black–white earnings ratios using the public use sample and the instructions described by Smith and Welch (1989, 522). Méndez Medina’s non-South and South estimates are presented here. For the United States as a whole, he gets different black–white earnings ratios for 1960–1980, despite following the procedures described by Smith and Welch. The table shows the Smith and Welch estimates for these three dates. The Méndez Medina estimates for the whole United States are 1960 = 0.558, 1970 = 0.624, and 1980 = 0.689. For 1990–2010, the table displays the Méndez Medina estimates. For the 1990–2009 men’s wages, we replicated the 1940–1980 procedures of Smith and Welch (1989, 522). Their thresholds for eliminating low and high earnings rates were extrapolated according to the 1980– 2010 progress of full-time earnings per male age twenty-five and up. For 1970–2009 relative incomes per capita, see the US Census Bureau, 2012 Statistical Abstract of the United States, http://census.gov/ compendia/statab, table 704 (accessed September 15, 2015). Figure 7-1: Carter et al. 2006, series Ec251–53. For the figures for 1882– 1951, see Wise 1952, 278. For 1952–1964, see unpublished estimates from Tuskegee Institute, Alabama, Department of Records and Research; compiled in US Census Bureau 1975, series H1168–70, 422.

CHAPTER 8 Table 8-1: Greek population growth accelerated in the 1920s and 1930s due to the heavy influx of immigrants under the 1923 population exchange agreement between Greece and the new government of Turkey. The US estimates up to 1870 are our own, and are described in earlier chapters and at http://gpih. For the other pre-1950 estimates of the age fifteen to sixty-four population, see the various volumes of International Historical Statistics edited by Brian Mitchell. For 1950–2010 estimates of age group populations, see http://esa.un.org/wpp/Excel -Data/population.htm (accessed May 17, 2013). For the projections of the age sixteen to sixty-four population from 2010 to 2050, see Statistical Abstract of the United States 2012. Territorial changes affect the SOURCES AND NOTES TO TABLES AND FIGURES

•

361

growth rates. Accordingly, we have omitted the shift from the AustroHungarian Empire to Austria in the World War I decade, and German losses of territory in both world wars. We ignore the population gains contemporaneous with the territorial gains of France, Greece, and Italy in the 1919 Versailles Treaty. The populations of Ireland and the United Kingdom are estimated for the post-1921 boundaries throughout. Table 8-2: Collins and Margo 2006, mainly 135, table 12; Carter et al. 2006, series Aa 287– 456; Cohen and Soto 2007. For the underlying spreadsheet file, see http://soto.iae-csic.org/Data.htm (accessed June 18, 2009). The Cohen and Soto estimates cover those age fifteen to sixtyfour who were not currently in school. To convert the census figures and the Collins and Margo attendance rates into approximate years of education for adults, we made the following assumptions: • The average attainment is achieved by the cohort as it passes age twenty. This overstates the attainment for those age twenty to twenty-four. On the other hand, we also ignore any effect of adult survival on the average attainment rate, which may understate the attainments of those at higher ages. • For ages fi fteen to nineteen, we assume that the attainment is the minimum of eleven years, or the average attainment achieved at ages twenty-five to twenty-nine. • For the birth cohorts over age fift y in 1930—that is, those born before 1880—we assume that their attainment has the same ratio to that of the birth cohort 1880–84 as implied by the cohorts’ respective rates of school attendance at ages five to nineteen in the lower panel of table 5 in Collins and Margo 2006. • We assume that both the population and the attendance rate are uniformly distributed within a five-year age bracket.

Table 8-3: The main source is Lindert 2004, tables A-1 and A-3, downloadable as Excel fi les from http://lindert.econ.ucdavis.edu (accessed September 29, 2015). The attainment ratios for 1980 and 2010 are from Cohen and Soto 2007. The “weighted lagged enrollments” (WLE) measure is a crude proxy for the education attainment of native adults. It is a synthetic measure 362

•

APPENDIX H

of primary and secondary enrollments per thousand children age five to fourteen when the age fifteen to sixty-four population was of school age. The WLE measure starts from gross enrollments in private plus public schools (both primary and secondary) per thousand children age five to fourteen. For each census year, it is a distributed lag function of such enrollment rates around earlier decadal census years. Taking the fifteen- to sixty-four-year-old population of 1940 as an example, the lag function is of the form WLE for 1940 = (0.20 times enrollment rate for 1900) + (0.23 times enrollment rate for 1910) + (0.27 times enrollment rate for 1920) + (0.30 times enrollment rate for 1930). The private school coverage is incomplete in some cases: the rates for Canada and France cover only enrollments in public schools; the rates for Italy and the Netherlands exclude private secondary schools; the rates for Norway exclude the shrinking share of private elementary schools (in 1890, these added only 11.1 per thousand to the 287.4 per thousand in public primary schools); the figures for the United Kingdom include only some private schools; the US private enrollment rates for 1870 and 1880 were extrapolated back from private–public ratios in 1890 and later. Figure 8-1: The US estimates for 1870 are reported in chapters 5–7. The UK estimate for 1870 is our revision of Baxter’s distribution for 1867, which can be downloaded from http://gpih. The UK estimate for 1910 comes from our hybrid distillation of the Bowley, Stamp, and Routh distribution among taxpaying units in 1911, also downloadable from http://gpih. All other observations have been downloaded from the top 1 percent file at http://topincomes.g-mond.parisschoolofeconomics.eu/ (accessed July 30, 2014). Figure 8-2: Middle–lower ratio = the ratio of the average income of the 41 to 95 percentiles divided by the average income of the 0 to 40 percentiles in the household income ranks. As cited in the notes to figures 5-3 and 5-4 above, the Goldsmith and US Office of Business Economics series = shares of pretax money income received by households. These estimates evolved through a long chain of revisions centering on the work of Selma Goldsmith and the US Office of Business Economics. The resulting estimates are now presented in Carter et al. 2006, Series Be21-26. Key links in the earlier chain SOURCES AND NOTES TO TABLES AND FIGURES

•

363

are Budd 1967, xiii; Goldsmith 1967; Radner and Hinrichs 1974. The sources and numerical results are also cited in Lindert 2000. CPS money income = US Census Bureau, Current Population Survey, Money Incomes of Households. The money income series includes receipts of cash transfer payments, but not of in-kind payments. See http://www.census.gov/compendia/statab/cats/income_expenditures_ poverty_wealth/household_income.html (accessed October 31, 2013). For the wage ratio, top decile/average, see table S8.1, http://piketty .pse.ens.fr/files/capital21c/en/xls/ (accessed September 1, 2015). Wage ratio 90/50 = Atkinson 2008, 411–24, series for the ninetieth percentile wage relative to the median. The series starts with the Labor Market Survey wage of the ninetieth percentile for 1973–2000. To this it splices an Economic Policy Institute wage series for 2000–2006, a CPS male wage for 1967–1973, and a CPS series for workers of both sexes for 1939–1967. Wage ratio 50/10 = same sources and procedures as for the wage ratio 90/50, except that it takes the ratio of the median wage to the tenth percentile wage. See also Margo’s series on the log differential of P90 versus P10 among males age eighteen to sixty-five, with constraints, 1939–1989, in Carter et al. 2006, series Ba4426 and Ba4427. Figure 8-3: Phillippon and Reshef 2012. This relative pay series traces the ratio of the average pay for all financial sector employees to the average pay for those hired in agriculture and industry. The educational attainment of those employed in the financial sector is measured relative to that of the entire private nonfarm sector. Figure 8-4: For 1774, 1800, and 1850–1870, our own estimates underlying the regional totals given in chapters 2–5. These can be downloaded from “American Incomes, ca. 1650–1870,” http://gpih. For 1840 and 1880–1929, see Easterlin 1960, 1961. For 1929–2010, see the US Department of Commerce estimates of personal income per capita published first by the Office of Business Economics and then the Bureau of Economic Analysis, as summarized in the US Census Bureau’s Statistical Abstract of the United States up through 2012. Figure 8-5: See the sources and notes to table 8-1 in this appendix for the estimates of the labor supply growth. As in figure 8-1, the top income 364

•

APPENDIX H

shares have been downloaded from the top 1 percent file at http:// topincomes.g-mond.parisschoolofeconomics.eu/ (accessed July 30, 2014). Figure 8- 6: Collins and Margo 2006; Cohen and Soto 2007, supplemented by the data file http://soto.iae-csic.org/Data.htm (accessed June 18, 2009). See also the notes to table 8-2 above. CHAPTER 9 Table 9-1: See the sources and notes to table 8-1 in the previous chapter. Table 9-2: See http://soto.iae-csic.org/Data.htm (accessed June 18, 2009). See also Cohen and Soto 2007.The estimates here refer to those persons age fifteen to sixty-four who were not enrolled in school. Figure 9-1: All observations have been downloaded from the top 1 percent fi le at http://topincomes.g-mond.parisschoolofeconomics.eu/ (accessed July 30, 2014). Figure 9-2: Goldin 1990; Carter et al. 2006, 2–294; 2012 Statistical Abstract of the United States 2012. Figure 9-3: Francine Blau 2012, 372, citing the ILO LABORSTA database, http://laborsta.ilo.org (accessed September 2, 2015). For Denmark, Norway, and Sweden, the figures for 1990 and earlier refer to the manufacturing sector. A new series starts for Japan in 2000. Figure 9-4: See the sources and notes to figure 8-5. Figure 9-5: Goldin and Katz 2008. Figure 9- 6: For the labor force growth, see the sources and notes to tables 8-1 and 9-1 and figures 8-5 and 9-4. For the growth in education attainment, see http://soto.iae-csic.org/ Data.htm (accessed June 18, 2009).

SOURCES AND NOTES TO TABLES AND FIGURES

•

365

CHAPTER 10 Table 10-1: The figures for 1700–1870 use five-year averages for prices and British national product, as consistently as the data allow. Accordingly, the numbers here will differ from those based on the longerperiod averages in table 3-6 and figure 3-3. As in chapters 3 and 5, the sources for the years up through 1870 are as follows: Nominal national product up through 1870 = our own estimates for America, and for Great Britain, Broadberry et al. 2011, 2015. For 1700– 1774, see table 3-6. For 1850–1870, see tables 5-1, 5-2, and 6-1. The population estimates are from Carter et al. 2006. The individual series are derived as follows: (a) American bare-bones price deflator = the cost-of-living measured for the bare-bones bundle for a family of four in Philadelphia, supplied to the authors by Allen. See “Thirteen Colonies, up to ca. 1774” and “Allen Subsistence Bundle Costs, Colonial America versus England, as of 2013,” http://gpih. The Philadelphia deflator uses his Philadelphia series 1720–1870, with extrapolation back to 1650, splicing the Massachusetts deflator onto the Philadelphia series. The Philadelphia series is preferred for its longer time span, given that the Chesapeake series stops with 1800 and the Massachusetts series stops with 1860. The British price deflator = Allen’s series for London and southern England. (b) The respectability price series is based on the consumer expenditure weights given for Great Britain 1801–3, in Broadberry et al. 2015, 339. It refers to a budget about triple that of the bare-bones budget suggested by the same authors, yet still refers to somebody in the low-skilled ranks, near the poverty line. From that bundle we were able to use the expenditure weighted prices for nine commodities: beans, beef, bread, butter, candles, cheese, eggs, linen cloth, and soap for the benchmark years 1800, 1850, and 1860 only. The British prices used in comparing the costs of the respectability bundle are those supplied by Clark’s file of English prices, 1209–1914, in http://gpih. The American prices are a mixture of Philadelphia prices (beef, bread, butter, and soap) and Massachusetts prices (the other five 366

•

APPENDIX H

commodities, with adjustments for linen cloth). See the file “Respectability Bundle Costs Am-Br,” http://gpih. (c) For 1950–2010, we use the expenditure-based income measure (cgdpe) per capita from Penn World tables 8.1, http://www.rug.nl/research/ggdc/data/pwt/pwt-8.1 (accessed April 17, 2015), and our ratios refer to the United Kingdom, not to Great Britain.

Figure 10-1: The 1774 population of fifteen- to sixty-four-year-olds is our own estimate as described and documented in chapters 2 and 3. The numbers for the 1800–2000 censuses are from Carter et al. 2006. The projections of the age sixteen to sixty-four population from 2010 to 2050 are from Statistical Abstract of the United States 2012. Figure 10-2: For 1700–1870, series (a) and (b), see the notes to table 10-1. (c) For 1950–2011, we have used the expenditure side, purchasing power parity measures of GDP per capita from Penn World Tables 8.1, http://www.rug.nl/research/ggdc/data/pwt/pwt-8.1 (accessed April 17, 2015). For 1920–1950, lacking any Penn World tables for that era, we spliced Maddison’s estimates onto the Penn World tables estimates at 1950. Our 1920–1949 figures thus differ from Maddison’s because ours use updated price weights based on 2005 prices. Figure 10-3: Same as for series (c) in figure 10-2. Figure 10-4: Williamson 1995, tables A1.1–A2.3. The cost of living deflators are presented and documented in his tables A3.1–A3.4.

SOURCES AND NOTES TO TABLES AND FIGURES

•

367

REFERENCES

Abramovitz, Moses. 1986. “Catching Up, Forging Ahead, and Falling Behind.” Journal of Economic History 46: 385– 406. Abramovitz, Moses, and Paul A. David. 2000. “American Macroeconomic Growth in the Era of Knowledge-Based Progress: The Long-Run Perspective.” In The Cambridge Economic History of the United States: Volume III: The Twentieth Century, edited by Stanley Engerman and Robert Gallman, 1–92. Cambridge: Cambridge University Press. Acemoglu, Daron, and David Autor. 2013. “Skills, Tasks, and Technologies: Implications for Employment and Earnings.” In Handbook of Labor Economics, Volume 4, edited by Orley Ashenfelter and David Card, 1043–171. Amsterdam: North-Holland. Acemoglu, Daron, Simon Johnson, James A. Robinson, and Y. Thaicharoen. 2003. “Institutional Causes, Macroeconomic Symptoms: Volatility, Crises, and Growth.” Journal of Monetary Economics 50, no. 1: 49–122. Acemoglu, Daron, Simon Johnson, and James Robinson. 2002. “Reversal of Fortune: Geography and Institutions in the Making of the Modern World Income Distribution.” Quarterly Journal of Economics 117, no. 4 (November): 1231–94. Acemoglu, Daron, and James Robinson. 2006. “De Facto Political Power and Institutional Persistence.” American Economic Review 96, no. 2: 325–30. ———. 2012. Why Nations Fail: The Origins of Power, Prosperity, and Poverty. New York: Crown Business. Adams, Donald R. 1968. “Wage Rates in the Early National Period: Philadelphia, 1785– 1830.” Journal of Economic History 28, no. 3 (September): 404–26. ———. 1970. “Some Evidence on English and American Wage Rates, 1790–1830.” Journal of Economic History 30, no. 3 (September): 499–520. ———. 1982. “The Standard of Living during American Industrialization: Evidence from the Brandywine Region, 1800–1860.” Journal of Economic History 42, no. 4 (December): 903–17. ———. 1986. “Prices and Wages in Maryland, 1750–1850.” Journal of Economic History 46, no. 3 (September): 625– 45. ———. 1992. “Prices and Wages in Antebellum America: The West Virginia Experience.” Journal of Economic History 52, no. 1 (March): 206–16. Adams, T. M. 1944. Prices Paid by Vermont Farmers [1790–1940]. Bulletin 507, with Supplement, Vermont Agricultural Experiment Station, Burlington. Allen, Robert C., Jean-Pascal Bassino, Debin Ma, Christine Moll-Murata, and Jan Luiten van Zanden. 2010. “Wages, Prices, and Living Standards in China, 1739–1925:

In Comparison with Europe, Japan, and India.” Economic History Review 64, no. S1 (January): 8–38. Allen, Robert C., Tommy E. Murphy, and Eric B. Schneider. 2012. “The Colonial Origins of Divergence in the Americas: A Labor Market Approach.” Journal of Economic History 72, no. 4 (December): 863–94, supplementary materials. Allen, Robert C., and Jacob Weisdorf. 2011. “Was There an ‘Industrious Revolution’ before the Industrial Revolution? An Empirical Exercise for England, c. 1300–1830,” Economic History Review 64, no. 3 (August): 715–29. Anderson, Edward. 2001. “Globalisation and Wage Inequalities, 1870–1970.” European Review of Economic History 5, no. 1 (April): 91–118. Anderson, Terry L. 1975. “Wealth Estimates for the New England Colonies, 1650– 1709.” Explorations in Economic History 12: 151–76. ———. 1979. “Economic Growth in Colonial New England: ‘Statistical Renaissance.’” Journal of Economic History 39, no. 1 (March): 243–357. Arroyo Abad, Leticia, and Peter H. Lindert. Forthcoming. “Fiscal Redistribution in Latin America since the Nineteenth Century.” In Inequality in Latin America in the Long Run, edited by Luis Bértola and Jeff rey G. Williamson. Atack, Jeremy, Fred Bateman, and Robert Margo. 2004. “Skill Intensity and Rising Wage Dispersion in Nineteenth-Century American Manufacturing.” Journal of Economic History 64, no. 1: 172–92. Atkinson, Anthony B. 2008. The Changing Distribution of Earnings in OECD Countries. Oxford: Oxford University Press. Atkinson, Anthony B., and Salvadore Morelli. 2014. Chartbook of Economic Inequality. http://www.chartbookofeconomicinequality.com (accessed August 18, 2015). Atkinson, Anthony B., Thomas Piketty, and Emmanuel Saez. 2011. “Top Incomes in the Long Run of History.” Journal of Economic Literature 49, no. 1 (March): 3–71. Ball, Duane E. 1976. “Dynamics of Population and Wealth in Eighteenth-Century Chester County, Pennsylvania.” Journal of Interdisciplinary History 6, no. 4 (Spring): 621– 44. Bandiera, Oriana, Imran Rasul, and Martina Viarengo. 2013. “The Making of Modern America: Migratory Flows in the Age of Mass Migration.” Journal of Development Economics 102: 23– 47. Barro, Robert J., and Xavier Sala-i-Martín. 1991. “Convergence across States and Regions.” Brookings Papers on Economic Activity 1: 107–82. Bates, Robert H., John H. Coatsworth, and Jeff rey G. Williamson. 2007. “Lost Decades: Postindependence Performance in Latin America and Africa.” Journal of Economic History 67 (December): 917– 43. Beard, Charles, and Mary Beard. 1927. The Rise of American Civilization. New York: Macmillan. Berry, Thomas S. 1968. Estimated Annual Variations in Gross National Product, 1789 to 1909. Richmond, VA: Bostwick Press. ———. 1988. Production and Population since 1789: Revised GNP Series in Constant Dollars. Richmond, VA: Bostwick Press. Bezanson, Anne, Robert D. Gray, and Miriam Hussey. 1935. Prices in Colonial Pennsylvania. Philadelphia: University of Pennsylvania Press. Blattman, Christopher, Jason Hwang, and Jeff rey G. Williamson. 2007. “The Impact of the Terms of Trade on Economic Development in the Periphery, 1870–1939: 370

•

REFERENCES

Volatility and Secular Change.” Journal of Development Economics 82 (January): 156–79. Blau, Francine. 2012. The Economics of Women, Men, and Work. 6th ed. Englewood Cliffs, NJ: Prentice-Hall. Blodget, James H. 1901. Wages of Farm Labor in the United States: Results of Eleven Statistical Investigations, 1866–1899. Washington, DC: US Government Printing Office. Blodget, James H. 1903. Wages of Farm Labor in the United States. Department of Agriculture, Bureau of Statistics, Miscellaneous Series, Bulletin No. 26. Washington, DC: US Government Printing Office. Blodget, Samuel. 1964. Economica: A Statistical Manual for the United States of America. New York: August Kelley. First published 1806. Bloom, David E., and Richard B. Freeman. 1986. “The Effects of Rapid Population Growth on Labor Supply and Employment in Developing Countries.” Population and Development Review 12, no. 3 (September): 381– 414. Bø, Erlend E., Elin Halvorsen, and Thor O. Thoresen. 2015. “Heterogeneity of the Carnegie Effect.” CESifo Working Paper No. 5339, Munich, May. Bodenhorn, Howard. 2015. The Color Factor: The Economics of African-American Well-Being in the Nineteenth-Century South. New York: Oxford University Press. Boot, H. M. 1999. “Real Incomes of the British Middle Class, 1760–1850: The Experience of Clerks at the East India Company.” Economic History Review 52, no. 4 (November): 638–68. Bordo, Michael D., and Christopher M. Meissner. Forthcoming. “Does Inequality Lead to Financial Crisis?” Journal of International Money and Finance. Borjas, George J. 2003. “The Labor Demand Curve Is Downward Sloping: Reexamining the Impact of Immigration on the Labor Market.” Quarterly Journal of Economics 118 (November): 1335–74. Borts, George H., and Jerome L. Stein. 1964. Economic Growth in a Free Market. New York: Columbia University Press. Bound, John, and Richard B. Freeman. 1992. “What Went Wrong? The Erosion of Relative Earnings and Employment among Young Black Men in the 1980s.” Quarterly Journal of Economics 107 (February): 201–32. Bourne, Jenny. 2008. “Slavery in the United States.” EH.Net Encyclopedia, edited by Robert Whaples, March 26. https://eh.net/encyclopedia/slavery-in-the-united-states/ (accessed August 26, 2015). Bridenbaugh, Carl. 1938. Cities in the Wilderness: The First Century of Urban Life in America, 1625–1742. New York: Ronald Press. ———. 1955. Cities in Revolt: Urban Life in America, 1743–1776. New York: Alfred A. Knopf. Broadberry, Stephen, Bruce Campbell, Alexander Klein, Mark Overton, and Bas van Leeuwen. 2012. “British Economic Growth, 1270–1870: An Output-Based Approach.” Paper presented at the World Economic History Congress, Stellenbosch, South Africa, July. ———. 2015. British Economic Growth, 1270–1870. Cambridge: Cambridge University Press. Brown, E. Henry Phelps. 1968. A Century of Pay: The Course of Pay and Production in France, Germany, Sweden, the United Kingdom, and the United States of America, 1860–1960. New York: St. Martin’s Press. REFERENCES

•

371

———. 1977. The Inequality of Pay. Berkeley: University of California Press. ———. 1988. Egalitarianism and the Generation of Inequality. Oxford: Clarendon Press. Brown, Martin, and Peter Phillips. 1986. “Craft Labor and Mechanization in Nineteenth-Century Canning.” Journal of Economic History 46: 743–56. Brynjolfsson, Erik, and Andrew McAffee. 2011. Race against the Machine: How the Digital Revolution Is Accelerating Innovation and Irreversibly Transforming Employment and the Economy. Lexington, MA: Digital Frontier Press. Budd, Edward C. 1967. “Introduction.” In Inequality and Poverty, edited by Edward C. Budd, viix–xxiv. New York: Harper and Row. Burnhard, Trevor G. 2001. “Prodigious Riches: The Wealth of Jamaica before the American Revolution.” Second series, Economic History Review 54, no. 3 (August): 506–24. Burtless, Gary, ed. 1996. Does Money Matter? The Effect of School Resources on Student Achievement and Adult Success. Washington, DC: Brookings Institution. Campbell, Douglas L. 2015. “Th rough the Looking Glass: A WARPed View of Real Exchange Rate History.” New Economic School (NES), Moscow, August. Carnegie, Andrew. 1901. The Gospel of Wealth and Other Timely Essays. New York: Century Company. Carr, Lois G., Russell R. Menard, and Lorena S. Walsh. 1991. Robert Cole’s World: Agriculture and Society in Early Maryland. Chapel Hill: University of North Carolina Press. Carruthers, Celeste K., and Marianne H. Wanamaker. Forthcoming. “Separate and Unequal in the Labor Market: Human Capital and the Jim Crow Wage Gap.” Carter, Susan B., Scott Sigmund Gartner, Michael R. Haines, Alan L. Olmstead, Richard Sutch, and Gavin Wright, eds. 2006. The Historical Statistics of the United States: Millennial Edition. 5 vols. New York: Cambridge University Press. Cederberg, Herbert Renando. 1977. An Economic Analysis of English Settlement in North America, 1583–1635. New York: Arno Press. Cline, William R. 1997. Trade and Income Distribution. Washington, DC: Institute for International Economics. Coatsworth, John H., and Jeff rey G. Williamson. 2004. “Always Protectionist? Latin American Tariffs from Independence to Great Depression.” Journal of Latin American Studies 36, pt. 2 (May): 205–32. Cochran, Thomas C. 1961. “Did the Civil War Retard Industrialization?” Mississippi Valley Historical Review 48 (September): 197–210. Coelho, Philip R. P., and James F. Shepherd. 1974. “Differences in Regional Prices: The United States, 1851–1880.” Journal of Economic History 34, no. 3 (September): 551–91. Cohen, Daniel, and Marcelo Soto. 2007. “Growth and Human Capital: Good Data, Good Results.” Journal of Economic Growth 12 (March): 51–76. Collins, William J. 1997. “When the Tide Turned: Immigration and the Delay of the Great Black Migration.” Journal of Economic History 60, no. 3: 756–81. Collins, William J., and Robert A. Margo. 2003. “Historical Perspectives on Racial Differences in Schooling in the United States.” NBER Working Paper 9770, National Bureau of Economic Research, Cambridge, MA, June. ———. 2006. “Historical Perspectives on Racial Differences in Schooling in the United States.” In Handbook of the Economics of Education, edited by Eric Hanushek and Finis Welch, 107–54. Vol. 1. Amsterdam: Elsevier. 372

•

REFERENCES

———. 2007. “The Economic Aftermath of the 1960s Riots in American Cities: Evidence from Property Values.” Journal of Economic History 67, no. 4 (December): 849–83. Collins, William J., and Marianne H. Wanamaker. 2013. “Selection and Economic Gains in the Great Migration of African-Americans: New Evidence from Linked Census Data.” NBER Working Paper 19124, National Bureau of Economic Research, Cambridge, MA, June. Conrad, Alfred H., and John R. Meyer. 1958. “The Economics of Slavery in the Ante Bellum South.” Journal of Political Economy 66 (April): 95–130. Costa, Dora. 2013. “Health and the Economy in the United States, from 1750 to the Present.” Working Paper 19685, National Bureau of Economic Research, Cambridge, MA, November. Crafts, N.F.R. 1987. “British Economic Growth, 1700–1850: Some Difficulties of Interpretation.” Explorations in Economic History 24, no. 3: 245– 68. Crafts, N.F.R., and Knick Harley. 1992. “Output Growth and the British Industrial Revolution: A Restatement of the Crafts-Harley View.” Economic History Review 45: 703–70. Craig, Lee A. 1991. “The Value of Household Labor in Antebellum Northern Agriculture.” Journal of Economic History 51, no. 1 (March): 67–81. Craig, Lee A., and Thomas Weiss. 1998. “Rural Agriculture Workforce by County, 1800 to 1900.” http://eh.net/databases/agriculture/ (accessed August 27, 2015). David, Paul A. 1967. “The Growth of Real Product in the United States before 1840: New Evidence and Controlled Conjectures.” Journal of Economic History 27, no. 2 (June): 151–97. ———. 2005. “Real Income and Economic Welfare Growth in the Early Republic, or, Another Try at Getting the American Story Straight.” Discussion Papers in Economic and Social History, no. 5, Oxford University. First version 1996. David, Paul A., and Peter Solar. 1977. “A Bicentenary Contribution to the History of the Cost of Living in America.” Research in Economic History 2: 1–80. David, Paul A., and Peter Temin. 1976. “Slavery: The Progressive Institution.” In Reckoning with Slavery: A Critical Study in the Quantitative History of American Negro Slavery, edited by Paul A. David, Herbert G. Gutman, Richard Sutch, Peter Temin, and Gavin Wright, 165–230. New York: Oxford University Press. Davis, Joseph H. 2004. “An Annual Index of U.S. Industrial Production, 1790–1915.” Quarterly Journal of Economics 119, no. 4 (November 2004): 1177–215. Davis, Lance E., and Robert J. Cull. 1994. International Capital Markets and American Economic Growth, 1820–1914. Cambridge: Cambridge University Press. Davisson, William I. 1967. “Essex County Price Trends: Money and Markets in 17th Century Massachusetts.” Essex Institute Historical Collections 103, no. 2 (April): 144–88. Donohue, John J., III, and James A. Heckman. 1991. “Continuous versus Episodic Change: The Impact of Civil Rights Policy on the Economic Status of Blacks.” Journal of Economic Literature 29, no. 4 (December): 1603– 43. Easterlin, Richard A. 1960. “Interregional Differences in per Capita Income, Population, and Total Income, 1840–1950.” In Trends in the American Economy in the Nineteenth Century, edited by Conference on Research in Income and Wealth, 73–140. Princeton, NJ: Princeton University Press. REFERENCES

•

373

———. 1961. “Regional Income Trends, 1840–1950.” In American Economic History, edited by Seymour E. Harris, 525– 47. New York: McGraw-Hill. Easterly, William. 2007. “Inequality Does Cause Underdevelopment: Insights from a New Instrument.” Journal of Development Economics 84, no. 2: 755–76. Egnal, Marc. 1998. New World Economies. New York: Oxford University Press. Eichengreen, Barry J. 2015. Hall of Mirrors. New York: Oxford University Press. Eichengreen, Barry J., and Peter H. Lindert, eds. 1990. International Debt Crisis in Historical Perspective. Cambridge, MA: MIT Press. Einhorn, Robin. 2006. American Taxation, American Slavery. Chicago: University of Chicago Press. Engerman, Stanley L. 1966. “The Economic Impact of the Civil War.” Explorations in Economic History 3, no. 2: 176–99. Engerman, Stanley L., and Kenneth L. Sokoloff. 2012. Economic Development in the Americas since 1500: Endowments and Institutions. New York: Cambridge University Press. Ernst & Young. 2013. International Estate and Inheritance Tax Guide 2013. https:// webforms.ey.com/ (accessed January 24, 2015). Feenstra, Robert C., ed. 2000. The Impact of International Trade on Wages. Chicago: University of Chicago Press. Feenstra, Robert C., and Gordon H. Hanson. 1999. “The Impact of Outsourcing and High-Technology Capital on Wages: Estimates for the United States, 1979–1990.” Quarterly Journal of Economics 114 (August): 907– 40. ———. 2003. “Global Production Sharing and Rising Inequality: A Survey of Trade and Wages.” In Handbook of International Trade, edited by Kwan Choi and James Harrigan, 146–87. Oxford: Basil Blackwell. Feenstra, Robert C., Robert Inklaar, and Marcel P. Timmer. 2015. “The Next Generation of the Penn World Table.” American Economic Review. http://www.ggdc.net /pwt (accessed October 1, 2015). Ferenczi, Imre, and Walter F. Willcox. 1929. International Migrations, Vol. I. New York: National Bureau of Economics Research. Ferrie, Joseph P., and Timothy J. Hatton. 2013. “Two Centuries of International Migration.” In Handbook of the Economics of International Migration, Volume 1A: The Immigrants, edited by Barry R. Chiswick and Paul W. Miller, 53–88. Amsterdam: North-Holland. Fogel, Robert W. 1986. “Nutrition and the Decline in Mortality since 1700: Some Preliminary Findings.” In Long-Term Factors in American Economic Growth, edited by Stanley L. Engerman and Robert E. Gallman, 439–527. Chicago: University of Chicago Press. ———. 1989. Without Consent or Contract. New York: W. W. Norton. Fogel, Robert W., and Stanley L. Engerman. 1974. Time on the Cross: The Economics of American Negro Slavery. Boston: Little, Brown and Company. ———. 2006. “Slave Sales and Appraisals, 1775–1865” and “Slave Hires and Appraisals, 1775–1865” [computer fi le]. ICPSR07421-v3. Ann Arbor, MI: Inter-university Consortium for Political and Social Research. First published 1976. Franklin, Benjamin. (1751) 1959. “Observations concerning the Increase of Mankind, Peopling of Countries, Etc.” In The Papers of Benjamin Franklin, Volume 4, edited by Leonard W. Larabee. Reprint, New Haven, CT: Yale University Press. 374

•

REFERENCES

Freeman, Richard B. 1972. “Black/White Economic Differences: Why Did They Last So Long?” Paper presented at the Cliometrics meetings, Madison, WI, April. Frickey, Edwin. 1947. Production in the United States, 1860–1914. Cambridge, MA: Harvard University Press. Friedberg, Rachel M., and Jennifer Hunt. 1995. “The Impact of Immigrants on HostCountry Wages, Employment, and Growth.” Journal of Economic Perspectives 9: 23– 44. Galenson, David W. 1981. White Servitude in Colonial America: An Economic Analysis. Cambridge: Cambridge University Press. ———. 1996. “The Settlement and Growth of the Colonies: Population, Labor, and Economic Development.” In The Cambridge Economic History of the United States: Volume I: The Colonial Era, edited by Stanley L. Engerman and Robert E. Gallman, 135–207. Cambridge: Cambridge University Press. Gallman, Robert E. 1960. “Commodity Output, 1839–1899.” In Trends in the American Economy in the Nineteenth Century, edited by Conference on Research in Income and Wealth, 13–72. Princeton, NJ: Princeton University Press. ———. 1966. “Gross National Product in the United States, 1834–1909.” In Output, Employment, and Productivity in the United States after 1800. New York: National Bureau of Economic Research. ———. 1972. “The Pace and Pattern of American Economic Growth.” In American Economic Growth: An Economist’s History of the United States, edited by Lance E. Davis, Richard A. Easterlin, and William N. Parker, 15–60. New York: Harper and Row. ———. 1982. “Influences on the Distribution of Landholdings in Early Colonial North Carolina.” Journal of Economic History 42, no. 3 (September): 549–75. ———. 2000. “Economic Growth and Structural Change in the Long Nineteenth Century.” In The Cambridge Economic History of the United States: Volume II: The Long Nineteenth Century, edited by Stanley L. Engerman and Robert E. Gallman, 1–55. Cambridge: Cambridge University Press. Geloso, Vincent. Forthcoming. “The Seeds of a Little Divergence: Living Standards in Canada under French Rule, 1688 to 1760.” PhD diss., London School of Economics. Gemmell, Norman, Richard Kneller, and Ismael Sanz. 2011. “The Timing and Persistence of Fiscal Policy Impacts on Growth: Evidence from OECD Countries.” Economic Journal 121: F33–F58. ———. 2014a. “Does the Composition of Government Expenditure Matter for LongRun GDP Levels?” Victoria University of Wellington, Working Papers in Public Finance, Paper 10/2014, August. ———. 2014b. “The Growth Effects of Tax Rates in the OECD.” Canadian Journal of Economics 47, no. 4 (November): 1217–55. Gemery, Henry. 2000. “The White Population of the Colonial United States, 1607– 1790.” In A Population History of North America, edited by Michael Haines and Richard Steckel, 143–90. Cambridge: Cambridge University Press. Genovese, Eugene D. 1974. Roll, Jordan, Roll. New York: Pantheon. Gilchrist, David T., and W. David Lewis, eds. 1965. Economic Change in the Civil War Era. Greenville, DE: Hagley Foundation. Global Price and Income History Group [cited as http://gpih]. 2015. “Downloadable Data on Incomes and Prices, 1200–1950.” http://gpih.ucdavis.edu (accessed September 2, 2015). REFERENCES

•

375

Go, Sun, and Peter H. Lindert. 2010. “The Uneven Rise of American Public Schools to 1850.” Journal of Economic History 70, no. 1 (March): 1–26. Goldin, Claudia. 1976. Urban Slavery in the American South, 1820–1860: A Quantitative History. Chicago: University of Chicago Press. ———. 1990. Understanding the Gender Gap: An Economic History of American Women. New York: Oxford University Press. Goldin, Claudia, and Lawrence Katz. 1995. “The Decline of Non-Competing Groups: Changes in the Premium to Education, 1890 to 1940.” NBER Working Paper 5202, National Bureau of Economic Research, Cambridge, MA, August. ———. 2008. The Race between Education and Technology. Cambridge, MA: Belknap Press. Goldin, Claudia, Lawrence Katz, and Ilyana Kuziemko. 2006. “The Homecoming of American College Women: The Reversal of the College Gender Gap.” Journal of Economic Perspectives 20, no. 4 (Fall): 133–56. Goldin, Claudia, and Frank D. Lewis. 1980. “The Role of Exports in American Economic Growth during the Napoleonic Wars, 1793–1807.” Explorations in Economic History 17, no. 1 (January): 6–25. Goldsmith, Raymond W. 1985. Comparative National Balance Sheets: A Study of Twenty Countries, 1688–1978. Chicago: University of Chicago Press. Goldsmith, Selma F. 1967. “Changes in the Size Distribution of Income.” In Inequality and Poverty, edited by Edward C. Budd, 65– 80. New York: Harper and Row. Gordon, Robert J. 2016. The Rise and Fall of American Growth. Princeton, NJ: Princeton University Press. Gray, Lewis C. 1940. History of Agriculture in the Southern United States to 1860: Volume II. Washington, DC: Carnegie. Greenfield, Robert L., and Hugh Rockoff. 1996. “Yellowbacks Out West and Greenbacks Back East: Social- Choice Dimensions of Monetary Reform.” Southern Economic Journal 62, no. 4 (April): 902–15. Grubb, Farley. 1985. “Immigrant Servant Labor: Their Occupational and Geographic Distribution in the Late Eighteenth-Century Mid-Atlantic Economy.” Social Science History 9 (Summer): 249–75. Gutman, Herbert G. 1975. Slavery and the Numbers Game: A Critique of Time on the Cross. Urbana: University of Illinois Press. Gutman, Herbert G., and Richard Sutch. 1976. “The Slave Family: Protected Agent of Capitalist Masters or Victim of the Slave Trade?” In Reckoning with Slavery: A Critical Study in the Quantitative History of American Negro Slavery, edited by Paul A. David, Herbert G. Gutman, Richard Sutch, Peter Temin, and Gavin Wright, 94–133. New York: Oxford University Press. Hacker, Louis. 1940. The Triumph of American Capitalism. New York: Columbia University Press. Haines, Christopher. 1992. “Comparable Indices of Wholesale Prices and Manufacturing Wage Rates in the United States, 1865–1914.” Research in Economic History 14: 269–92. Harley, C. Knick. 1988. “Ocean Freight Rates and Productivity, 1740–1913: The Primacy of Mechanical Invention Reaffirmed.” Journal of Economic History 48, no. 4 (December): 851–76. 376

•

REFERENCES

Harris, P.M.G. 1996. “Inflation and Deflation in Early America, 1634–1860: Patterns of Change in the British American Economy.” Social Science History 20, no. 4 (Winter): 469–505. Hatton, Timothy J., and Jeff rey G. Williamson. 1998. The Age of Mass Migration: Causes and Economic Impact. Oxford: Oxford University Press. ———. 2005. Global Migration and the World Economy: Two Centuries of Policy and Performance. Cambridge, MA: MIT Press. ———. 2008. “The Impact of Immigration: Comparing Two Global Eras.” World Development 36, no. 3 (March): 345– 61. ———. 2010. “Are Third World Emigration Forces Abating?” World Development 39, no. 1 (January): 20–32. Henretta, James A. 1991. The Origins of American Capitalism. Boston: Northeastern University Press. Higgs, Robert. 1972. “Did Southern Farmers Discriminate?” Agricultural History 46: 325–28. ———. 1975. “Did Southern Farmers Discriminate? Interpretive Problems and Further Evidence.” Agricultural History 49: 445– 47. ———. 1977. “Firm-Specific Evidence on Racial Wage Differentials and Workforce Segregation.” American Economic Review 67: 236– 45. ———. 1982. “Accumulation of Property by Southern Blacks before World War One.” American Economic Review 72: 725–37. Higman, B. W. 1996. “Economic and Social Development of the British West Indies, from Settlement to ca. 1850.” In The Cambridge Economic History of the United States: Volume I: The Colonial Era, edited by Stanley L. Engerman and Robert E. Gallman, 297–335. Cambridge: Cambridge University Press. Hildreth, Richard, 1849. The History of the United States of America, from the Discovery of the Continent to the Organization of Government under the Federal Constitution. 3 vols. New York: Harper and Brothers. Hnatkovska, Viktoria, and Norman Loayza. 2005. “Volatility and Growth.” In Managing Economic Volatility and Crises, edited by Joshua Aizenmann and Brian Pinto. Cambridge: Cambridge University Press. Hoff man, Philip T., David Jacks, Patricia Levin, and Peter H. Lindert. 2002. “Real Inequality in Western Europe since 1500.” Journal of Economic History 62, no. 2 (June): 322–55. Hohenberg, Paul M., and Lynn H. Lees. 1985. The Making of Urban Europe, 1000–1950. Cambridge, MA: Harvard University Press. Homer, Sidney, and Richard Sylla. 1996. A History of Interest Rates. 3rd ed. rev. New Brunswick, NJ: Rutgers University Press. Hoppit, Julian. 1996. “Political Arithmetic in Eighteenth-Century England.” Economic History Review 49, no. 3 (August): 516– 40. Hoover, Ethel D. 1960. “Retail Prices after 1850.” Trends in the American Economy in the Nineteenth Century, edited by Conference on Research in Income and Wealth, 141–90. Princeton, NJ: Princeton University Press. Hourwich, Isaac. 1922. Immigration and Labor: The Economic Aspects of European Immigration to the United States. 2nd ed. New York: Huebsch. Huff, Gregg. 2002. “Boom-or-Bust Commodities and Industrialization in Pre–World War II Malaya.” Journal of Economic History 62, no. 4: 1074–115. REFERENCES

•

377

Immervoll, Herwig, and Linda Richardson. 2010. “Redistribution Policy and Inequality Reduction in OECD Countries: What Has Changed in Two Decades?” OECD Social, Employment, and Migration Working Papers No. 122. Paris: OECD. Irwin, Douglas A. 2011. “Revenue or Reciprocity? Founding Feuds over Early U.S. Trade Policy.” In Founding Choices: American Economic Policy in the 1790s, edited by Douglas A. Irwin and Richard Sylla, 89–120. Chicago: University of Chicago Press. Irwin, Douglas A., and Richard Sylla, eds. 2011. Founding Choices: American Economic Policy in the 1790s. Chicago: University of Chicago Press. Jacks, David, Kevin H. O’Rourke, and Jeff rey G. Williamson. 2011. “Commodity Price Volatility and World Market Integration since 1700.” Review of Economics and Statistics 93, no. 3 (August): 800–813. Jackson, C. Kirabo, Rucker C. Johnson, and Claudia Persico. 2014. “The Effects of School Finance Reforms on the Distribution of Spending, Academic Achievement, and Adult Outcomes.” NBER Working Paper 20118, National Bureau of Economic Research, Cambridge, MA, May. ———. 2015. “The Effects of School Spending on Educational and Economic Outcomes: Evidence from School Finance Reforms.” NBER Working Paper 20847, National Bureau of Economic Research, Cambridge, MA, January. James, Harold. 2014. “International Capital Movements and the Global Markets.” In The Cambridge History of Capitalism: Volume II: The Spread of Capitalism: From 1848 to the Present, edited by Larry Neal and Jeff rey G. Williamson, 264–300. Cambridge: Cambridge University Press. Jasanoff, Maya. 2011. Liberty’s Exiles: American Loyalists in the Revolutionary World. New York: Alfred A. Knopf. Jenks, Jeremiah Whipple, and William Jett Lauck. 1926. The Immigration Problem. 6th ed. New York: Huebsch. Johnston, Louis D., and Samuel H. Williamson. 2008. “The Annual Real and Nominal GDP for the United States, 1790–1928.” MeasuringWorth, http://www.measuringworth.org/usgdp/ (accessed August 10, 2015). ———. 2010. “What Was the U.S. GDP Then?” MeasuringWorth, http://www.measur ingworth.org/usgdp/ (accessed August 10, 2015). Jones, Alice Hanson. 1977. American Colonial Wealth. 3 vols. New York: Arno Press. ———. 1980. Wealth of a Nation to Be. New York: Columbia University Press. Kane, Edward J. 1989. The S&L Insurance Mess: How Did It Happen? Washington, DC: Urban Institute Press. Katz, Lawrence F., and David Autor. 1999. “Changes in the Wage Structure and Earnings Inequality.” In Handbook of Labor Economics, Volume 3, edited by Orley Ashenfelter and David Card, 1463–555. Amsterdam: North-Holland. Katz, Lawrence F., and Robert A. Margo. 2013. “Technical Change and the Relative Demand for Skilled Labor: The United States in Historical Perspective.” NBER Working Paper 18752, National Bureau of Economic Research, Cambridge, MA, February. Kessel, Reuben A., and Armen A. Alchain. 1959. “Real Wages in the North during the Civil War: Mitchell’s Data Reinterpreted.” Journal of Law and Economics 2 (October): 95–113. Kindleberger, Charles P. 1978. Manias, Panics, and Crashes: A History of Financial Crises. New York: Basic Books.

378

•

REFERENCES

King, Willford I. 1915. The Wealth and Income of the People of the United States. New York: Macmillan. Klein, Alexander. 2013. “New State-Level Estimates of Personal Income in the United States, 1880–1910.” Research in Economic History 29 (annual): 191–255. KMPG Global. 2015. https://home.kpmg.com/xx/en/home/services/tax/tax-tools-and -resources.html (accessed January 31, 2015). Kneller, Richard, M. Bleaney, and Norman Gemmell. 1999. “Fiscal Policy and Growth: Evidence from OECD Countries.” Journal of Public Economics 74: 171–90. Kopczuk, Wojciech, and Emmanuel Saez. 2004. “Top Wealth Shares in the United States, 1916–2000: Evidence from Estate Tax Returns.” National Tax Journal 57, no. 2: 445–87. Kousser, J. Morgan. 1974. The Shaping of Southern Politics: Suff rage Restriction and the Establishment of the One-Party South, 1880–1910. New Haven, CT: Yale University Press. Kulikoff, Allan. 1976. “Tobacco and Slaves: Population, Economy, and Society in Eighteenth-Century Prince George’s County, Maryland.” PhD diss., Brandeis University. ———. 1979. “The Economic Growth of the Eighteenth-Century Chesapeake Colonies.” Journal of Economic History 39, no. 1 (March): 275–88. ———. 1986. Tobacco and Slaves: The Development of Southern Cultures in the Chesapeake, 1680–1800. Chapel Hill: University of North Carolina Press. ———. 2000. From British Peasants to Colonial American Farmers. Chapel Hill: University of North Carolina Press. ———. 2005. “The Economic Crisis of the Revolutionary Era.” Paper presented at the annual conference of the Program in Early America Economy and Society, Philadelphia, October 28. ———. 2014. “ ‘Such Things Ought Not to Be’: The American Revolution and the First National Great Depression.” In The World of the American Revolution and Republic: Land, Labor, and the Conflict for a Continent, edited by Andrew Shankman, 134– 64. New York: Routledge. Kuznets, Simon. 1955. “Economic Growth and Income Inequality.” American Economic Review 45, no. 1: 1–28. Lafortune, Jeanne, José Tessada, and Carolina González-Velosa. 2013. “More Hands, More Power? The Impact of Immigration on Farming and Technological Choices in US Agriculture in the Early Twentieth Century.” Working Paper 431, Instituto de Economia, Pontificia Universidad Catolica de Chile, March. Lafortune, Jeanne, José Tessada, and Ethan Lewis. 2014. “People and Machines: The Evolving Relationship between Capital and Skill in Manufacturing, 1850–1940, Using Immigration Shocks.” Paper presented to the Development of the American Economy and National Bureau of Economic Research summer workshop, Cambridge, MA, June 20. Lebergott, Stanley. 1964. Manpower in Economic Growth: The American Record since 1800. New York: McGraw-Hill. ———. 1976. The American Economy: Income, Wealth, and Want. Princeton, NJ: Princeton University Press.

REFERENCES

•

379

Leigh, Andrew. 2009. “Top Incomes.” In The Oxford Handbook of Economic Inequality, edited by Wiemer Salverda, Brian Nolan, and Timothy M. Smeeding, 150–76. Oxford: Oxford University Press. Lemon, James T. 1972. The Best Poor Man’s Country: A Geographic Study of Early Southeastern Pennsylvania. Baltimore: Johns Hopkins University Press. Reissued 2002. Lemon, James T., and Gary B. Nash. 1968. “The Distribution of Wealth in EighteenthCentury America: A Century of Changes in Chester County, Pennsylvania, 1693– 1802.” Journal of Social History 2: 1–24. Levy, Frank, and Peter Temin. 2007. “Inequality and Institutions in 20th Century America.” NBER Working Paper 13106, National Bureau of Economic Research, Cambridge, MA, April. Lewis, Ethan. 2011. “Immigration, Skill Mix, and Capital-Skill Complementarity.” Quarterly Journal of Economics 126, no. 2 (May): 1029– 69. ———. 2013. “Immigration and Production Technology.” Annual Review of Economics, 5, no. 1: 165–91. Lillard, Lee A. 1977. “Inequality: Earnings vs. Human Wealth.” American Economic Review 67, no. 1 (March): 42–53. Lindert, Peter H. 1978. Fertility and Scarcity in America. Princeton, NJ: Princeton University Press. ———. 2000. “Three Centuries of Inequality in Britain and America.” In Handbook of Income Distribution, edited by Anthony B. Atkinson and François Bourguignon, 1: 167–216. Amsterdam: Elsevier Science. ———. 2004. Growing Public: Social Spending and Economic Growth since the Eighteenth Century. 2 vols. New York: Cambridge University Press. ———. 2014. “Private Welfare and the Welfare State.” In The Cambridge History of Capitalism: Volume II: The Spread of Capitalism, edited by Larry Neal and Jeff rey G. Williamson, 464–500. Cambridge: Cambridge University Press. Lindert, Peter H., and Jeff rey G. Williamson. 1982. “Revising England’s Social Tables, 1688–1812.” Explorations in Economic History 19, no. 4 (October): 385– 408. ———. 1983a. “English Workers’ Living Standards during the Industrial Revolution: A New Look.” Second series, Economic History Review 36, no. 1 (February 1983): 1–25. ———. 1983b. “Reinterpreting Britain’s Social Tables, 1688–1913.” Explorations in Economic History 20, no. 1 (January): 94–109. ———. 1985. “Growth, Equality, and History.” Explorations in Economic History 22, no. 4 (October): 341–77. ———. 2003. “Does Globalization Make the World More Unequal?” In Globalization in Historical Perspective, edited by Michael D. Bordo, Alan M. Taylor, and Jeff rey G. Williamson, 227–70. Chicago: University of Chicago Press. ———. 2013a. “American Incomes before and after the Revolution.” Journal of Economic History 73, no. 3 (September): 1– 40. ———. 2013b. “American Incomes before and after the Revolution.” NBER Working Paper 17211, National Bureau of Economic Research, Cambridge, MA. First published July 2011. Loeb, Susanna, and John Bound. 1996. “The Effect of Measured School Inputs on Academic Achievement: Evidence from the 1920s, 1930s, and 1940s Birth Cohorts.” Review of Economics and Statistics 78, no. 4 (November): 653– 64. 380

•

REFERENCES

Long, Clarence D. 1960. Wages and Earnings in the United States, 1860–1890. Princeton, NJ: Princeton University Press for the NBER. Lustig, Nora. 2011. “Fiscal Policy and Income Redistribution in Latin America: Challenging the Conventional Wisdom.” Tulane University Economics Working Paper 1124, October. Also EcIneq Working Paper 227, November. Maddison, Angus. 1995. Monitoring the World Economy, 1820–1992. Paris: Organisation for Economic Co-operation and Development. ———. 2001. The World Economy: A Millennial Perspective. Paris: Organisation for Economic Co-operation and Development. ———. 2007. http://www.ggdc.net/maddison/Historical . . . /horizontal-fi le_03–2007 (accessed August 5, 2015). ———. 2010. http://www.ggdc.net/maddison/Historical_Statistics/horizontal- fi le _02–2010.xls (accessed August 2, 2015). Maddison Project. 2013. http://www.ggdc.net/maddison/maddison-project/home.htm (accessed June 1, 2013). Madrick, Jeff. 2011. Age of Greed: The Triumph of Finance and the Decline of America, 1970 to the Present. New York: Alfred A. Knopf. Main, Gloria L. 1977. “Inequality in Early America: The Evidence from Probate Records of Massachusetts and Maryland.” Journal of Interdisciplinary History 7, no. 4 (Spring): 559–81. Main, Gloria L., and Jackson Turner Main. 1988. “Economic Growth and the Standard of Living in Southern New England.” Journal of Economic History 48, no. 1 (March): 27– 46. Main, Jackson Turner. 1965. The Social Structure of Revolutionary America. Princeton, NJ: Princeton University Press. Maloney, Thomas H. 1994. “Wage Compression and Wage Inequality between Black and White Males in the United States, 1940–1960.” Journal of Economic History 54, no. 2 (June): 358–81. Malthus, Thomas Robert. (1798) 1920. An Essay on the Principle of Population. Reprint, London: Royal Economic Society. Mancall, Peter C. 1991. Valley of Opportunity: Economic Culture along the Upper Susquehanna, 1700–1800. Ithaca, NY: Cornell University Press. Mancall, Peter C., Joshua L. Rosenbloom, and Thomas Weiss 2003. “Conjectural Estimates of Economic Growth in the Lower South, 1720 to 1800.” In History Matters: Economic Growth, Technology, and Population, edited by William Sundstrom and Timothy Guinnane, 389– 424. Stanford, CA: Stanford University Press. ———. 2008. “The Role of Exports in the Economy of Colonial North America: New Estimates for the Middle Colonies.” NBER Working Paper 14334, National Bureau of Economic Research, Cambridge, MA, September. Mancall, Peter C., and Thomas Weiss. 1999. “Was Economic Growth Likely in Colonial British North America?” Journal of Economic History 59, no. 1 (March): 17– 40. Margo, Robert A. 1984. “Accumulation of Property by Southern Blacks before World War One: Comment and Further Evidence.” American Economic Review 74: 768–76. ———. 1985. Disenfranchisement, School Finance, and the Economics of Segregated Schools in the United States South, 1890–1910. New York: Garland Publishing. ———. 1990. Race and Schooling in the South, 1880–1950: An Economic History. Chicago: University of Chicago Press. REFERENCES

•

381

———. 2000. Wages and Labor Markets in the United States, 1820–1860. Chicago: University of Chicago Press. ———. 2002. “The North–South Wage Gap, before and after the Civil War.” NBER Working Paper 8778, National Bureau of Economic Research, Cambridge, MA, February. McCusker, John J. 2000. “Estimating Early American Gross Domestic Product.” Historical Methods 33, no. 3 (Summer): 155–59. ———. 2001. How Much Is That in Real Money? A Historical Commodity Price Index for Use as a Deflator of Money Values in the Economy of the United States. 2nd ed. Worcester, MA: American Antiquarian Society. ———. 2006. “Colonial Statistics.” In Historical Statistics of the United States: Earliest Times to the Present, Millennial Edition, edited by Susan G. Carter, Michael Haines, Alan Olmstead, Richard Sutch, and Gavin Wright, 5- 627–50. Cambridge: Cambridge University Press. McCusker, John J., and Russell R. Menard. 1985. The Economy of British America, 1607–1789: Needs and Opportunities for Study. Chapel Hill: University of North Carolina Press. Menard, Russell R. 1996. “Economic and Social Development of the South.” In The Cambridge Economic History of the United States: Volume I: The Colonial Era, edited by Stanley L. Engerman and Robert E. Gallman, 249–95. Cambridge: Cambridge University Press. Mian, Atif, and Amir Sufi. 2014. House of Debt: How They (and You) Caused the Great Recession, and How We Can Prevent It from Happening Again. Chicago University of Chicago Press. Milanovic, Branko. 2005. World Apart: Measuring International and Global Inequality. Princeton, NJ: Princeton University Press. ———. 2009. “Global Inequality Recalculated: The Effect of New 2005 PPP Estimates on Global Inequality.” Washington, DC: World Bank. ———. 2015. “Income Inequalities between the Populations of the World Have Decreased since the Early 2000s.” http://www.inequalitywatch.eu/spip.php?article101 &id_mot=70&lang=en (accessed April 27, 2015. Milanovic, Branko, Peter H. Lindert, and Jeff rey G. Williamson. 2011. “Pre-Industrial Inequality.” Economic Journal 121 (March): 255–72. Mill, John Stuart. 1848. Principles of Political Economy. 1st ed. London: Longmans, Green, and Company. Mitchell, Wesley C. 1903. A History of the Greenbacks. Chicago: University of Chicago Press. Mitchell, Wesley C., Willford I. King, Frederick R. Macauley, and Oswald W. Knauth. 1921. Income in the United States: Its Amount and Distribution, 1909–1919. Vol. 1. New York: Harcourt, Brace. Mittal, Sonia, Jack N. Rakove, and Barry R. Weingast. 2011. “The Constitutional Choices of 1787 and Their Consequences.” In Founding Choices: American Economic Policy in the 1790s, edited by Douglas A. Irwin and Richard Sylla, 25–56. Chicago: University of Chicago Press. Moriguchi, Chiaki, and Emmanuel Saez. 2008. “The Evolution of Income Concentration in Japan, 1886–2005: Evidence from Income Tax Statistics.” Review of Economics and Statistics 90, no. 4: 713–34. 382

•

REFERENCES

Morrisson, Christian. 2000. “Historical Perspectives on Income Distribution: The Case of Europe.” In Handbook of Income Distribution, Volume 1, edited by Anthony B. Atkinson and François Bourguignon, 217–59. Amsterdam: Elsevier Science. Myrdal, Gunnar. 1944. An American Dilemma. New York: Harper. Nash, Gary B. 1976a. “Poverty and Poor Relief in Pre-Revolutionary Philadelphia.” William and Mary Quarterly 33: 3–30. ———. 1976b. “Urban Wealth and Poverty in Pre-Revolutionary America.” Journal of Interdisciplinary History 6: 545–84. ———. 1979. The Urban Crucible: Social Change, Political Consciousness, and the Origins of the American Revolution. Cambridge, MA: Harvard University Press. Neal, Larry, and Jeff rey G. Williamson, eds. 2014. The Cambridge History of Capitalism: Volume II: The Spread of Capitalism: From 1848 to the Present. Cambridge: Cambridge University Press. North, Douglass C. 1958. “Ocean Freight Rates and Economic Development, 1750– 1913.” Journal of Economic History 18 (December): 538–55. ———. 1960. “The United States Balance of Payments, 1790–1860.” In Trends in the American Economy in the Nineteenth Century, edited by Conference on Research in Income and Wealth, 573– 627. Princeton, NJ: Princeton University Press. Obstfeld, Maurice, and Alan M. Taylor. 2003. “Globalization and Capital Markets.” In Globalization in Historical Perspective, edited by Michael D. Bordo, Alan M. Taylor, and Jeff rey G. Williamson, 121–83. Chicago: University of Chicago Press. ———. 2004. Global Capital Markets: Integration, Crisis, and Growth. Cambridge: Cambridge University Press. Olmstead, Alan, and Paul Rhode. 2008. Creating Abundance: Biological Innovation and American Agricultural Development. Cambridge: Cambridge University Press. ———. 2010. “Productivity Growth and Regional Dynamics of Antebellum Southern Development.” In Economic Evolution and Revolution in Historical Time, edited by Paul W. Rhode, Joshua L. Rosenbloom, and David F. Weiman, 180–213. Stanford, CA: Stanford University Press. Olson, John F. 1992a. “Clock Time versus Real Time: A Comparison of the Lengths of the Northern and Southern Agricultural Work Year.” In Without Consent or Contract: The Rise and Fall of American Slavery, Markets and Production: Technical Papers, Volume 1, edited by Robert W. Fogel and Stanley L. Engerman, 216– 40. New York: W. W. Norton. ———. 1992b. “The Occupational Structure of Southern Plantations during the Late Antebellum Era.” In Without Consent or Contract: The Rise and Fall of American Slavery, Markets, and Production: Technical Papers, Volume 1, edited by Robert W. Fogel and Stanley L. Engerman, 137– 69. New York: W. W. Norton. Organisation for Economic Co-operation and Development. 2013. Education at a Glance 2013. Paris: Organisation for Economic Co-operation and Development. O’Rourke, Kevin H. 2006. “The Worldwide Economic Impact of the French Revolutionary and Napoleonic Wars, 1793–1815.” Journal of Global History 1, no. 1: 123– 49. O’Rourke, Kevin H., and Jeff rey G. Williamson. 1999. Globalization and History. Cambridge: Cambridge University Press. O’Rourke, Kevin H., Jeff rey G. Williamson, and Timothy J. Hatton. 1994. “Mass Migration, Commodity Market Integration, and Real Wage Convergence.” In REFERENCES

•

383

Migration and the International Labor Market, 1850–1939, edited by Timothy J. Hatton and Jeff rey G. Williamson, 203–20. London: Routledge. Ostry, Jonathan D., Andrew Berg, and Charalambos G. Tsangarides. 2014. “Redistribution, Inequality, and Growth.” IMF Staff Discussion Note SDN/14/02X, February. Ottaviano, Gianmarco, and Giovanni Peri. 2012. “Rethinking the Effect of Immigration on Wages.” Journal of the European Economics Association 10, no. 1 (February): 152–97. Perkins, Edwin J. 1994. American Public Finance and Financial Services, 1700–1815. Columbus: Ohio State University Press. Phelps Brown, E. H. 1968. A Century of Pay. London: Macmillan. Phillippon, Thomas, and Ariell Reshef. 2012. “Wages and Human Capital in the U.S. Financial Industry: 1909–2006.” Quarterly Journal of Economics 127, no. 4: 1551–609. Piketty, Thomas. 2014. Capital in the Twenty-First Century. Cambridge, MA: Belknap Press. Piketty, Thomas, and Emmanuel Saez. 2003. “Income Inequality in the United States, 1913–1998.” Quarterly Journal of Economics 118, no. 1: 1–39. Piketty, Thomas, and Gabriel Zucman. 2014. “Capital Is Back: Wealth-Income Ratios in Rich Countries.” Quarterly Journal of Economics 129, no. 3 (August): 1255–1310. ———. 2015. “Wealth and Inheritance in the Long Run.” In Handbook of Income Distribution, Volume 2, edited by Anthony B. Atkinson and François Bourguignon, 1303– 68. Amsterdam: North-Holland. Pitkin, Timothy. 1817. A Statistical View of the Commerce of the United States of America: Its Connection with Agriculture and Manufactures: and an Account of the Public Debt, Revenues, and Expenditures of the United States. With a Brief Review of the Trade, Agriculture, and Manufactures of the Colonies, Previous to Their Independence. Accompanied with Tables, Illustrative of the Principles and Objects of the Work. 2nd ed. New York: James Eastburn and Company. Price, Jacob M. 1974. “Economic Function and the Growth of American Port Towns in the Eighteenth Century.” Perspectives in American History 8: 121–86. Rabinowitz, Howard N. 1974. “Half a Loaf: The Shift from White to Black Teachers in Negro Schools of the Urban South, 1865–1890.” Journal of Southern History 40, no. 4 (November): 565–94. Rabushka, Alvin. 2008. Taxation in Colonial America. Princeton, NJ: Princeton University Press. Radner, Daniel B., and John C. Hinrichs. 1974. “Size Distribution of Income in 1964, 1970, and 1971.” Survey of Current Business 54, no. 10 (October): 19–31. Ramcharan, Rodney. 2010. “Inequality and Redistribution: Evidence from U.S. Counties and States, 1890–1930.” Review of Economics and Statistics 92, no. 4 (November): 729– 44. Ransom, Roger L., and Richard Sutch. 1977. One Kind of Freedom: The Economic Consequences of Emancipation. Cambridge: Cambridge University Press. ———. 2001. “One Kind of Freedom: Reconsidered (and Turbo Charged).” Explorations in Economic History 38, no. 1 (January): 6–39. Reynolds, Morgan O., and Eugene Smolensky. 1977. Public Expenditures, Taxes, and the Distribution of Income: The U.S., 1950, 1961, 1970. New York: Academic Press.

384

•

REFERENCES

Roine, Jesper, and Daniel Waldenström. 2014. “Long-Run Trends in the Distribution of Income and Wealth.” Working Paper 2014:5, Uppsala Center for Fiscal Studies, Uppsala, Sweden, April. Rosenbloom, Joshua L., and Gregory W. Stutes. 2008. “Reexaming the Distribution of Wealth in 1870.” In Quantitative Economic History: The Good of Counting, edited by Joshua L. Rosenbloom, 146– 69. London: Routledge. Rosenbloom, Joshua L., and Thomas Weiss. 2014. “Economic Growth in the MidAtlantic Region: Conjectural Estimates for 1720 to 1800.” Explorations in Economic History 51 (January): 41–59. Rosenthal, Caitlin. 2012. From Memory to Mastery: Accounting for Control in America, 1750–1880. PhD diss., Harvard University. Rothenberg, Winifred B. 1985. “The Emergence of a Capital Market in Rural Massachusetts, 1730–1838.” Journal of Economic History 45, no. 4 (December): 781– 808. ———. 1988. “The Emergence of Farm Labor Markets and the Transformation of the Rural Economy: Massachusetts, 1750–1855.” Journal of Economic History 48, no. 3 (September): 537– 66. ———. 1992. From Market-Places to a Market Economy: The Transformation of Rural Massachusetts, 1750–1850. Chicago: University of Chicago Press. Rousseau, Peter L., and Richard E. Sylla. 2005. “Emerging Financial Markets and Early U.S. Growth.” Explorations in Economic History 42: 1– 46. Sachs, Jeffrey D., ed. 1988. Developing Country Debt and the World Economy. Chicago: University of Chicago Press. Salisbury, Stephen. 1962. “The Effects of the Civil War on American Industrial Development.” In The Economic Impact of the American Civil War, edited by Ralph Andreano, 2: 161– 68. Cambridge, MA: Schenkman. Scheiber, Harry N. 1965. “Economic Change in the Civil War Era: An Analysis of Recent Studies.” Civil War History 11 (December): 396– 411. Schwartz, Anna Jacobson. 1960. “Gross Dividend and Interest Payments by Corporations at Selected Dates in the 19th Century.” In Trends in the American Economy in the Nineteenth Century, edited by Conference on Research in Income and Wealth, 407– 48. Princeton, NJ: Princeton University Press. Shammas, Carole. 1993. “A New Look at Long-Term Trends in Wealth Inequality in the United States.” American Historical Review 98, no. 4: 412–32. ———. 2007. “The Housing Stock of the Early United States: Refi nement Meets Migration.” William and Mary Quarterly 64, no. 3 (July): 549–90. ———. 2014. “Did Democracy Give the U.S. an Edge in Primary Schooling?” Paper presented at the All–University of California Conference in Economic History, Berkeley, May 3. Sharkey, Robert. 1959. Money, Class, and Party: An Economic Study of Civil War and Reconstruction. Baltimore: Johns Hopkins University Press. Shaw, William Howard. 1947. Value of Commodity Output since 1869. Study no. 48. New York: National Bureau of Economic Research. Shepherd, James F. 1993. “State Tariff Policies in the Era of Confederation.” Paper presented at the sixty-eighth annual meeting of the Western Economic Association International, Lake Tahoe, Nevada, June 21.

REFERENCES

•

385

Shepherd, James F., and Gary M. Walton. 1972. Shipping, Maritime Trade, and the Economic Development of Colonial North America. Cambridge: Cambridge University Press. ———. 1976. “Economic Change after the American Revolution: Prewar and Postwar Comparisons of Maritime Shipping and Trade.” Explorations in Economic History 13, no. 4 (November): 397– 422. Silver, Nate. 2014. “Be Skeptical of Both Piketty and His Skeptics.” http://fivethirty eight.com/features/be-skeptical-of-both-piketty-and-his-skeptics/ (accessed May 27, 2014). Simler, Lucy. 1990. “The Landless Worker: An Index of Economic and Social Change in Chester County Pennsylvania, 1750–1820.” Pennsylvania Magazine of History and Biography 114, no. 2 (April): 163–99. ———. 2007. “ ‘She Came to Work’: The Female Labor Force in Chester County, 1750– 1820.” Early American Studies 5, no. 2: 427–53. Smeeding, Timothy M., Lee Rainwater, Martin Rein, Richard Hauser, and Gaston Schaber. 1990. “Income Poverty in Seven Countries: Initial Estimates from the LIS Database.” In Poverty, Inequality, and Income Distribution in Comparative Perspective: The Luxembourg Income Study (LIS), edited by Timothy M. Smeeding, Michael O’Higgins, and Lee Rainwater, 57–76. Lanham MD: University Press of America. Smith, Adam. (1766) 1978. Lectures on Jurisprudence. Edited by R. L. Meek, D. D. Raphael, and P. G. Stein. Oxford: Clarendon Press. Smith, Billy Gordon. 1981. “The Material Lives of Laboring Philadelphians, 1750 to 1800.” Third series, William and Mary Quarterly, 38, no. 2 (April): 164–202. ———. 1984. “Inequality in Colonial Philadelphia: A Note on Its Nature and Growth.” Third series, William and Mary Quarterly 41, no. 4 (October): 629– 45. ———. 1990. The “Lower Sort”: Philadelphia’s Laboring People, 1750–1800. Ithaca, NY: Cornell University Press. Smith, Daniel Scott. 1973. “Population, Family, and Society in Hingham, Massachusetts, 1635–1880.” PhD diss., University of California at Berkeley. Smith, James P., and Finis R. Welch. 1989. “Black Economic Progress after Myrdal.” Journal of Economic Literature 27, no. 2 (June): 519– 64. Solt, Frederick. 2009. “Standardizing the World Income Inequality Database.” Social Science Quarterly 90, no. 2: 231– 42. ———. 2014. “The Standardized World Income Inequality Database.” http://myweb .uiowa.edu/fsolt/swiid/swiid.html (accessed October 1, 2015). Soltow, Lee. 1971. “Economic Inequality in the United States in the Period from 1790 to 1860.” Journal of Economic History 31: 822–39. ———. 1975. Men and Wealth in the United States, 1850–1870. New Haven, CT: Yale University Press. ———. 1989. Distribution of Wealth and Income in the United States in 1798. Pittsburgh: Pittsburgh University Press. Soltow, Lee, and Jan Luiten van Zanden. 1998. Income and Wealth Inequality in the Netherlands, 16th–20th Century. Amsterdam: Het Spinhuis. Squire, Lyn. 1981. Employment Policy in Developing Countries. Oxford: Oxford University Press. Stiverson, Gregory A. 1977. Poverty in a Land of Plenty: Tenancy in Eighteenth-Century Maryland. Baltimore: Johns Hopkins University Press. 386

•

REFERENCES

Sutch, Richard. 1975a. “The Breeding of Slaves for Sale and the Westward Expansion of Slavery, 1850–1860.” Race and Slavery in the Western Hemisphere: Quantitative Studies, edited by Stanley L. Engerman and Eugene D. Genovese, 173–210. Princeton, NJ: Princeton University Press. ———. 1975b. “The Treatment Received by American Slaves: A Critical Review of the Evidence Presented in Time on the Cross.” Explorations in Economic History 12 (October): 335– 438. ———. 1976. “The Care and Feeding of Slaves.” In Reckoning with Slavery: A Critical Study in the Quantitative History of American Negro Slavery, edited by Paul A. David, Herbert G. Gutman, Richard Sutch, Peter Temin, and Gavin Wright, 231– 301. New York: Oxford University Press. Sylla, Richard E. 1998. “U.S. Securities Markets and the Banking System, 1790–1840.” Federal Reserve Bank of St. Louis Review, May–June, 83–98. ———. 2009. “Comparing the UK and US Financial Systems, 1790–1830.” In The Origins and Development of Financial Markets and Institutions, edited by Jeremy Atack and Larry Neal, 209– 40. Cambridge: Cambridge University Press. ———. 2011. “Financial Foundations: Public Credit, the National Bank, and Securities Markets.” In Founding Choices: American Economic Policy in the 1790s, edited by Douglas A. Irwin and Richard E. Sylla, 59–88. Chicago: University of Chicago Press. Sylla, Richard E., Jack W. Wilson, and Robert E. Wright. 2006. “Integration of TransAtlantic Capital Markets, 1790–1845.” Review of Finance 10: 613– 44. Taylor, Alan M., and Jeff rey G. Williamson. 1997. “Convergence in the Age of Mass Migration.” European Review of Economic History 1: 27– 63. Taylor, Alan S. 2001. American Colonies: The Settling of America. New York: Viking Penguin. ———. 2013. The Internal Enemy: Slavery and War in Virginia, 1772–1832. New York: W. W. Norton. Tocqueville, Alexis de. (1839) 1963. Democracy in America. Reprint, New York: Alfred A. Knopf. Tryon, Rolla M. (1917) 1966. Household Manufactures in the United States. Reprint, New York: A. M. Kelley. Tucker, Rufus S. 1938. “Distribution of Income among Taxpayers in the United States, 1863–1935.” Quarterly Journal of Economics 3 (August): 547–87. Turner, Frederick Jackson. 1921. The Frontier in American History. New York: Holt. Ubelaker, Douglas H. 1992. “North American Indian Population Size: Changing Perspectives.” In Disease and Demography in the Americas, edited by John W. Verano and Douglas H. Ubelaker, 169–76. Washington, DC: Smithsonian Institution Press. Unger, Irwin. 1964. The Greenback Era: A Social and Political History of American Finance, 1865–1879. Princeton, NJ: Princeton University Press. United Kingdom, Royal Commission on the Distribution of Income and Wealth. 1977. Report No 5, Th ird Report on the Standing Reference (Cmnd. 6999). London: HMSO. US Bureau of Labor Statistics. 1929. The History of Wages in the United States from Colonial Times to 1928. Washington, DC: US Government Printing Office. US Census Bureau. 1975. Historical Statistics of the United States: Colonial Times to 1970. Bicentennial ed. Washington, DC: US Government Printing Office. REFERENCES

•

387

———. 2012. Statistical Abstract of the United States. http://census.gov/compendia /statab, table 704 (accessed September 15, 2015). US Internal Revenue Service. 1863–71. Annual Reports. http://www.irs.gov/uac/SOITax-Stats-Archive-1863-to-1999-Annual-Reports-and-IRS-Data-Books (accessed September 3, 2015). van Zanden, and Jan Luiten. 1995. “Tracing the Beginning of the Kuznets Curve: Western Europe during the Early Modern Period.” Second series, Economic History Review 48: 643– 64. ———. 2009. “The Skill Premium and the ‘Great Divergence.’ ” European Review of Economic History 13, no. 1 (April): 121–53. Vartanian, Pershing. 1964. “The Cochran Thesis: A Critique in Statistical Analysis.” Journal of American History 51 (June): 77–79. Vedder, Richard. 1975. “The Slave Exploitation (Expropriation) Rate.” Explorations in Economic History 12 (October): 453–57. Villaflor, Georgia C., and Kenneth L. Sokoloff. 1982. “Migration in Colonial America: Evidence from the Militia Muster Rolls.” Social Science History 6, no. 4 (Fall): 539–70. Wade, Richard D. 1964. Slavery in the Cities. New York: Oxford University Press. Walsh, Lorena S. 1999. “Summing the Parts: Implications for Estimating the Chesapeake Output and Income Subregionally.” Third series, William and Mary Quarterly 56, no. 1 (January): 53–94. ———. 2010. Motives of Honor, Pleasure, and Profit: Plantation Management in the Colonial Chesapeake, 1607–1763. Chapel Hill: University of North Carolina Press. Walsh, Lorena S., Ann Smart Martin, and Joanne Bowen. 1997. Provisioning Early American towns: The Chesapeake: A Multidisciplinary Case Study. Final Performance Report, National Endowment for the Humanities, Colonial Williamsburg Foundation. http://research.history.org/Files/HistRes/Provisioning.pdf (accessed April 10, 2014). Wang, Chen, Koen Caminada, and Kees Goudswaard. 2012. “The Redistributive Effect of Social Transfer Programmes and Taxes: A Decomposition across Countries.” International Social Security Review 65, no. 3: 27– 48. Ward, Marianne, and John Devereux. 2003. “Measuring British Decline: Direct versus Long-Span Income Measures.” Journal of Economic History 63, no. 3 (September): 826–51. ———. 2004. “Relative UK/US Output Reconsidered: A Reply to Broadberry.” Journal of Economic History 64, no. 3 (September): 879–91. ———. 2006. “Relative British and American Income Levels during the First Industrial Revolution.” Research in Economic History 23: 249–86. Warden, Gerard B. 1976. “Inequality and Instability in Eighteenth-Century Boston: A Reappraisal.” Journal of Interdisciplinary History 6: 585– 620. Waters, John J. 1976. “Patrimony, Succession, and Social Stability: Guilford, Connecticut, in the Eighteenth-Century.” Perspectives in American History 10: 131– 60. Weeden, William Babcock. 1890. Economic and Social History of New England, 1620– 1789. 2 vols. Boston: Houghton Mifflin. Weiss, Thomas. 1992. “U.S. Labor Force Estimates and Economic Growth, 1800–1860.” In American Economic Growth and Standards of Living before the Civil War, edited

388

•

REFERENCES

by Robert E. Gallman and John J. Wallis, 19–78. Chicago: University of Chicago Press. ———. 1993a. “Estimates of Gross Domestic Output for the United States, 1800 to 1860.” Unpublished paper, University of Kansas, July. ———. 1993b. “Long-Term Changes in U.S. Agricultural Output per Worker, 1800– 1900.” Economic History Review 46, no. 2 (May): 324– 41. ———. 1994. “Economic Growth before 1860: Revised Conjectures.” In American Economic Development in Historical Perspective, edited by Thomas Weiss and Donald Schaefer, 11–27. Stanford, CA: Stanford University Press. Whaples, Robert. 1995. “Where Is There Consensus among American Economic Historians? The Results of a Survey on Forty Propositions.” Journal of Economic History 55, no. 1 (March): 139–54. Whartenby, Frank G. 1977. Land and Labor Productivity in United States Cotton Production, 1800–1840. New York: Arno Press. Wilkins, Mira. 1989. The History of Foreign Investment in the United States, 1914–1945. Cambridge, MA: Harvard University Press. Williamson, Jeff rey G. 1964. American Growth and the Balance of Payments, 1820– 1913. Chapel Hill: University of North Carolina Press. ———. 1965. “Regional Inequality and the Process of National Development: A Description of the Patterns.” Economic Development and Cultural Change 13, no. 4, pt. 2 (July): 1–84. ———. 1974. “Watersheds and Turning Points: Conjectures on the Long-Term Impact of Civil War Financing.” Journal of Economic History 34 (September): 636– 61. ———. 1977. “Strategic Wage Goods, Prices, and Inequality.” American Economic Review 66, no. 1 (March): 29– 41. ———. 1980. “Earnings Inequality in Nineteenth- Century Britain.” Journal of Economic History 45, no. 3 (September): 457–76. ———. 1982. “The Structure of Pay in Britain, 1710–1911.” In Research in Economic History, edited by Paul Uselding, 7: 1–54. Greenwich, CT: Johnson Associates. ———. 1990. Coping with City Growth during the British Industrial Revolution. Cambridge: Cambridge University Press. ———. 1995. “The Evolution of Global Labor Markets since 1830: Background Evidence and Hypotheses.” Explorations in Economic History 32: 141–96. ———. 2006. “Explaining World Tariffs, 1870–1938: Stolper-Samuelson, Strategic Tariffs, and State Revenues.” in Eli Heckscher, International Trade, and Economic History, edited by Ronald Findlay, Rolf G. H. Henriksson, Håkan Lindgren, and Mats Lundahl, 199–228. Cambridge, MA: MIT Press. ———. 2010. “Five Centuries of Latin American Inequality.” Journal of Iberian and Latin American Economic History 28, no. 2 (September): 227–52. ———. 2011. Trade and Poverty: When the Third World Fell Behind. Cambridge, MA: MIT Press. ———. 2012. “Commodity Prices over Two Centuries: Trends, Volatility, and Impact.” Annual Review of Resource Economics 4: 185–207. ———. 2015. “Latin American Inequality: Colonial Origins, Commodity Booms, or a Missed 20th-Century Great Leveling?” Special issue, Journal of Human Development and Capabilities 16 (August): 24–41.

REFERENCES

•

389

Williamson, Jeff rey G., and Peter H. Lindert. 1980a. American Inequality: A Macroeconomic History. New York: Academic Press. ———. 1980b. “Long-Term Trends in American Wealth Inequality.” In Modeling the Distribution and Intergenerational Transmission of Wealth, edited by James D. Smith, 9–94. Chicago: University of Chicago Press. Woodward, C. Vann. 1974. The Strange Career of Jim Crow. 3rd rev. ed. Oxford: Oxford University Press. First published 1955. Wright, Carroll D. 1885. “Historical Review of Wages and Prices 1752–1860,” from the Sixteenth Annual Report of the Massachusetts Bureau of Statistics of Labor. Boston: Wright and Potter. Wright, Gavin. 1976. “Prosperity, Progress, and American Slavery.” In Reckoning with Slavery: A Critical Study in the Quantitative History of American Negro Slavery, edited by Paul A. David, Herbert G. Gutman, Richard Sutch, Peter Temin, and Gavin Wright, 302–36. New York: Oxford University Press. ———. 1986. Old South, New South: Revolutions in the Southern Economy since the Civil War. New York: Basic Books. ———. 1990. “The Origins of American Industrial Success, 1879–1940.” American Economic Review 80 (September): 651– 68. ———. 2013. Sharing the Prize: The Economics of the Civil Rights Revolution in the American South. Cambridge, MA: Belknap Press. Wright, Robert E. 2002. The Wealth of Nations Rediscovered: Integration and Expansion of the U.S. Financial Sector, 1780–1850. Cambridge: Cambridge University Press. Wrigley, E. A., and R. S. Schofield. 1981. The Population History of England, 1541–1871: A Reconstruction. London: Edward Arnold.

390

•

REFERENCES

INDEX

Abramovitz, Moses, 166 Acemoglu, Daron, 93 Africa, 86, 93, 246, 248 age distribution, 21n8, 55–56, 269n5 agriculture. See farming aircraft manufacturing, 224 Alabama, 224, 344 Allen, Robert, 68– 69, 72–73 apprentices, 284, 285 Arkansas, 129, 344 Articles of Confederation, 8, 77, 85 artisans, 17, 23, 24, 87, 270; hollowing out of, 125, 132–34, 136, 167, 177–78; working hours of, 282, 283 Asia, 223, 236–37, 240, 245– 48 Atack, Jeremy, 132, 178 Atkinson, Anthony, 2, 112, 195, 221 Australia, 202, 254; educational attainment in, 233–34; gender gap in, 225; during Great Depression, 149; immigration to, 211, 238, 255n; rising inequality in, 12, 219, 228 Austria, 110 automobile manufacturing, 224 baby boom, 244 Baltimore, 48, 135n69 Bank of the United States, 135 barley, 298 Barro, Robert, 184, 185 Bateman, Fred, 132, 178 beans, 40, 70 Beard, Charles, 142, 146, 149, 153, 156, 169 Beard, Mary, 142, 146, 149, 153, 156, 169 Belgium, 110, 202, 214 Berry, Thomas, 86 birth control, 227 Blodget, Samuel, 5n10, 277 Bolshevik revolution, 207

Boston, 47, 57, 70, 135n69; colonial laborers in, 265, 266, 270; dependency rate in, 64– 65, 127n53; migration from, 56, 65; poor whites in, 42; population of, 15, 90 Bound, John, 223 Bourne, Jenny, 293 bread, 74 Broadberry, Stephen, 74, 104 Brown, Henry Phelps, 202 Brown, Martin, 178 buckwheat, 298 building trades, 17, 23, 131–32, 160, 283, 285 butter, 70 California, 204, 205 Campbell, Douglas, 71 Canada, 7, 73; educational attainment in, 214, 233, 258; Great Depression in, 207; immigration to, 211, 228, 238, 255n; income convergence in, 205, 211; rising inequality in, 219, 228 candles, 70 canning, 178 Capital in the Twenty-First Century (Piketty), xv, 6, 112–13, 206 capital markets, 139, 179, 330 Carnegie, Andrew, 258 Carr, Lois Green, 66 Charleston, S.C., 17, 23, 48, 57, 73, 90, 127n53 Charlotte, N.C., 224 Chesnut, James, Jr., 141 Chesnut, Mary, 141 Chester County, Pa., 35 Child, Josiah, 48 child dependency, 9, 37n36, 45, 54–57, 69, 75 child labor, 284–86, 313 child mortality, 7, 44, 54, 56

Chile, 246n China, 52, 235n18, 237, 247– 48, 256 Civil Rights Act (1964), 225 civil rights movement, 164, 192, 203, 204. See also Jim Crow; segregation Civil War, 10, 129, 253, 321; benefits of, 159– 64; Confederate army during, 151; confl icting views of, 142– 44; cost of, 146–53, 340n; deaths from, 164– 65; farming disrupted by, 330; fi nancing of, 170; high-wage occupations during, 90; incomes after, 90, 144– 46, 152–53, 154–55; inequality and, 11, 111, 153–59 clerical workers, 132n61, 201–2, 279 coal, 70 cobweb model, 232 Cohen, Daniel, 212 Collins, William, 186–87, 212, 223 computable general equilibrium (CGE) models, 180 Connecticut, 16, 47, 205, 224, 277 Conrad, Alfred, 299 construction trades, 17, 23, 131–32, 160, 283, 285 corn, 40, 48, 336 cotton, 70, 106, 109, 150, 298–99, 331, 336 Craig, Lee, 163, 271, 272, 285, 292–93, 313–14, 321, 339– 40 credit boom, 217n, 239 crowding, 125 Crystal Palace Exhibition (1851), 251 David, Paul, 4–5, 8, 59, 92, 96, 291–92 Davis, Joseph, 150 death rates, 7, 10, 43, 44, 54, 56, 244 debt crisis, 239 deerskins, 49 Delaware, 48, 205, 344 Denmark, 200n10, 202, 219n2, 259 depreciation, 27, 32, 274, 275, 316 derivatives, 239 de-skilling, 167, 178 de-urbanization (ruralization), 15, 57–58, 75, 123, 138 Devereux, John, 71 disease, 10, 125 District of Columbia, 78, 140, 205, 224, 344 divorce, 227 Dodd-Frank Act (2010), 260 domestic servants, 130, 280, 282, 284, 319 Douglass, Frederick, 140

392

•

INDEX

Easterlin, Richard, 91, 96, 106, 109, 184 education: American slippage in, 12, 214, 233–35, 240; cost of, 246n; during Great Leveling, 212–15, 218; outlook for, 246– 47; pay scales in, 162– 63, 177, 201; political opposition to, 261; quantity vs. quality of, 234–36; recommendations for, 258–59, 262; rising demand for, 177; skill premium and, 131–32; southern inequities in, 10, 118, 137n, 160, 186–89; technology vs., 212n, 215–16, 230–32; of veterans, 203, 227; of women, 227. See also skill premium; teachers efficiency-equity trade-off, 261 Eichengreen, Barry, 217, 249 Engerman, Stanley, 93, 273, 277, 287–88, 291, 293, 296, 301, 311 engineers, 178, 200, 201 England: age distribution in, 56; American economic advantage over, 39, 68–74; inequality in, 14, 40, 42; living standards in, 43; New England vs., 48; urbanization in, 124n47; urban–rural wage gap in, 125. See also Great Britain entrepreneurship, 270, 327, 339– 42 Erie Canal, 101 exploitation rate, 273, 274, 288, 290, 293, 297, 299–300 farming, 14–15, 17; children working at, 284–85, 313; earnings from, 24, 27, 63, 79, 82, 105, 163, 264, 271–75, 313, 327– 48; subsistence, 33, 89; technological advances in, 138, 182, 216; volatility affecting, 49–50 Federal Reserve, 249 fencing costs, 338 fertility, 9, 44, 54, 56, 74, 209, 210, 244 fertilizer, 338 fi nancial markets, 101, 114, 179–80 fi nancial sector, 243, 258, 260; disruptions in, 217, 218; employment in, 224; inequality heightened by, 134–35, 138–39, 200–201, 227, 238–39, 241; outlook for, 249–50 Finland, 205, 259 firewood, 70 fish, 40, 44, 47, 50 Florida, 101, 224, 344 flour, 48, 50 Fogel, Robert, 273, 277, 287–88, 291, 293, 296, 301, 311 food costs, 125

France, 78, 86, 124n47; American tensions with, 275; as economic power, 10n17, 254; educational attainment in, 214, 232; regional inequality in, 205; skill premium in, 202; taxation of inheritances in, 259 Franklin, Benjamin, xv, 13, 54 Freeman, Richard, 223 French and Indian (Seven Years’) War, 56, 65 French Wars, 103n15 frontier economy: coastal economy vs., 94; colonial South as, 33–34, 67; diminishing returns in, 10, 193; fi lling-up effect of, 107–9; influences on growth in, 47–58; leveling effect of, 73, 75, 168, 183; price structure in, 73; subsistence living in, 33, 49, 94 Gallman, Robert, 4, 96, 98, 14, 150 game, 44 Garmon, Frank, Jr., 277n24 Gates, Bill, 2 Gemery, Henry, 56 Georgia, 67, 129, 298, 320–21, 331–33, 344, 345 Germany, 168, 202, 207, 233, 255n, 259 GI Bill, 203, 227 Gilded Age, 171, 185 Glass-Steagall Act (1933), 239, 260 Glen, James, 5n10 globalization, 71n41, 236–39, 240, 247, 250, 264 Goldin, Claudia, 114, 131, 133, 199, 203, 214, 216, 225, 232, 238 Gold Rush, 109–10 Goldsmith, Raymond, 96, 316 government employees, 279–80 government spending, 79, 100–101, 137 grains, 40, 48, 49–50, 66, 70, 331 Great Britain: American economic advantage over, 8–9, 10, 13, 42, 68, 69, 71, 76, 77, 93, 95, 103–7, 120, 136, 143, 147, 149, 165, 169, 171, 193, 250–55; American exports to, 47, 88, 89; consumer prices in, 40, 69–70, 72, 74, 304–10; as economic power, 10n17, 254; education in 214, 233, 246n; Great Leveling in, 202, 250; inequality in, 12, 119, 120, 219, 243; Irish migration to, 179; naval attacks by, 90; during Revolutionary War, 87; slave trade ended by, 131, 301–2n32; taxation in, 207, 259; trade shocks to, 88; after World War II, 254. See also England

Great Depression, 8, 149, 194, 207, 249; in America vs. Britain, 253; Revolutionary era vs., 85, 86; skilled vs. unskilled workers during, 202n14; trade barriers during, 216 Great Leveling: in America, 11–12, 194–218, 222, 227, 228, 239, 243; in Britain, 202, 250; worldwide, 250 Great Migration, 143, 164, 168, 185, 186, 192, 203 Great Recession, 85, 219, 223, 239, 249 Greece, 242– 43n2 Hacker, Louis, 142, 143– 44, 149, 153, 156, 169 Hamilton, Alexander, 86 Hatton, Timothy, 245– 46 hay, 298 Higgs, Robert, 162, 323 Hildreth, Richard, 90 Historical Statistics of the United States, 16 Holland, 120 “hollowing out,” 125, 129, 132–34, 136, 167, 177–78 Hong Kong, 236–37 hops, 298 households: counting of, 20–21; slaves grouped into, 266, 289–90 housing costs, 125 hyperinflation, 10, 87, 207 Iceland, 254 Illinois, 205, 328–29 immigration. See migration income tax, 2 indentured servants, 22, 23, 25, 33 India, 52, 237, 256 Indian Wars, 49 indigo, 34, 48–50, 89, 109 Industrial Revolution, 68n, 93, 179, 251 infant mortality, 7 inflation: antebellum, 100n6; during Civil War period, 143, 145, 149; minimum wage eroded by, 223; during Revolutionary period, 10, 27, 68, 79, 87, 138; tax bite worsened by, 207–8, 218; during World War II, 218 information technology, 248 Ingalls, Caroline, 140, 141 Ingalls, Charles, 140, 141 inheritance, 7, 258, 259– 60, 262 innkeepers, 270, 279 insurance, 49n8

INDEX

•

393

Integrated Public Use Microdata Series (IPUMS), 4, 97, 132, 134, 144, 161, 285, 311, 320 interest rates: during Civil War period, 274–75, 316; during colonial period, 26–27; on public debt, 79; after Revolutionary War, 138 Inter-university Consortium for Political and Social Research, 296, 301 investment banking, 239, 260 Irish immigration, 179 Italy, 168, 205, 214, 242– 43n2, 259 Jackson, Andrew, 135 Jamestown, Va., 43– 44 Japan, 196, 209, 219, 220, 240, 242– 43n2; educational attainment in, 214, 233–34; gender pay trends in, 225–27; income convergence in, 198, 205; labor force growth in, 228, 244; postwar growth in, 9, 168, 194, 236, 254; taxation of inheritances in, 259; wartime devastation in, 207, 217 Jim Crow, 160n27, 163n34, 164, 168, 185–89, 192, 203, 215. See also civil rights movement; segregation Johnston, Louis, 86, 103 Jones, Alice Hanson, 3, 17, 26, 28, 31–36, 84, 266, 270, 274, 314 Jones, Thomas Jesse, 187 Katz, Lawrence, 114, 132–33, 178, 199, 214, 216, 232, 238 King, Gregory, xv, 5, 14 King, Willford I., 171–72 Klein, Alexander, 184 Klein, Herbert, 276 Knauth, Oswald, 172 Korea, 233, 236, 246n, 259 Korean War, 227 Kousser, Morgan, 185n41 Kulikoff, Allan, 66, 86, 87 Kuznets, Simon, 1, 10, 20, 92, 96, 166; urbanization thesis of, 123–29, 159, 167, 175, 176, 183 Kuznets curve, 36, 113, 184 laborers, 17, 22; black, 151, 322–26; earnings of, 200–202, 208–9, 280, 322–26; sources on, 65, 82, 266; urban, 174, 265, 266, 270, 282, 283; working hours of, 160

394

•

INDEX

labor force participation rates (LFPR), 16, 22, 23, 77; in American colonies, 45; European vs. American, 104; of minors, 284; of nonwhites, 11, 161, 162; of slaves, 16, 34, 77, 161, 162, 163, 284, 289 labor unions, 223, 247 Latin America, 86, 93, 118, 120, 245– 46 Lebergott, Stanley, 129, 163, 279, 280, 292, 296, 299, 311, 324 Lee, Robert E., 140, 141 Leiden, 15 life expectancy, 7, 54, 210 Lincoln, Abraham, 140 Lindert, Frederick, 140, 141 linens, 70 literacy, 132, 330, 345– 46 livestock, 27, 47, 48, 49, 61, 271, 275, 331, 336 London, 15 Long, Clarence, 286 lost decades, 93 Loudoun County, Va., 35 Louisiana, 344 Louisana Purchase, 100 lower-gap measure, 221–22 Luxembourg Income Study, 198, 257n16 lynching, 185–86 Macauley, Fredrick, 172 Maddison, Angus, 8–9, 13, 71, 104, 166, 171 Main, Gloria Lund, 3– 4, 59, 63 Main, Jackson Turner, 3, 24, 36, 59, 63, 79, 266, 269, 270, 271–72, 279 Maine, 56 maize, 44 Malthus, Thomas Robert, 13, 54 Mancall, Peter, 45, 67, 75, 79, 92, 296–97, 303 Margo, Robert: on civilian wages at army posts, 311; on de-skilling in manufacturing, 178; on hollowing of artisan class, 132–33; on North–South wage gap, 129; on race riots, 223; on racial disparities in education, 187, 189; on racial disparities in pay, 162, 212, 323 maritime insurance, 49n8 market failure, 125 Marx, Karl, 209 Maryland, 22, 23, 48, 56–57, 224, 273–74, 301–2, 344 Massachusetts, 35, 44, 47, 224, 276, 297 McCormick, Cyrus, 140 McCusker, John, 28, 32–33, 86

meat, 70, 73n43, 74 Menard, Russell, 66, 86 merchants, 17 Mexico, 237 Meyer, John, 299 Middle East, 245– 46 migration, 124, 143, 176, 236; affordability of, 56n21, 101; asset prices raised by, 137; educational attainment and, 215, 232; toward frontier, 107, 118, 183; incentives for, 6, 13, 34, 39, 75n, 238, 245; intracolonial, 56–57, 65, 118; labor force swelled by, 228; life cycle of, 246; living conditions and, 174; outlook for, 244– 46; population growth fed by, 54; restrictions on, 209, 210–11, 243, 246; skill premium heightened by, 167, 179, 192; to southern states, 224; spatial displacement hypothesis and, 181–82; traditional views of, 179–82; wartime dampening of, 87, 146; after World War II, 204. See also Great Migration; ruralization; urbanization mills, 282 minimum wage, 223 Mississippi, 129, 224, 344 Mitchell, Wesley, C., 143– 44, 172 Mobile, Ala., 203 molasses, 40, 299 mortality, 7, 10, 43, 44, 54, 56, 244 mortgage lending, 315–16 Morrisson, Christian, 202 Napoleonic Wars, 88 national income and product accounts (NIPA), 14, 27, 274–75, 315 Native Americans, 7, 44 naval stores, 44, 48– 49, 89 Netherlands, 110, 149, 200n10, 205, 255n; taxation of inheritances in, 259; transfer payments in, 197; urbanization in, 124n47 Nettles, George, 282 Nevada, 205 New Deal, 208, 218, 240 New Hampshire, 47, 56 New Jersey, 48, 65, 205, 224 Newport, R.I., 57 New York City, 15, 42, 48, 57, 65, 90, 135n69 New York State, 21, 48, 56n20, 57, 65, 87, 205, 224 New Zealand, 149, 211, 219, 228, 238, 254, 259 Norfolk, Va., 48, 203– 4

North Carolina, 17, 19, 35, 57, 67, 296, 298, 344 Norway, 202, 205, 254, 255n14, 259 oats, 298 occupational statistics, 16–20 Occupy movement, 221 Ogden, William Butler, 140, 141 oil, 70 Olmstead, Alan, 110, 298–99 One Kind of Freedom (Ransom and Sutch), 161 Organisation for Economic Co-operation and Development (OECD), 197 Ottoman revolution, 207 outsourcing, 248 peas, 70 Pennsylvania, 16, 48, 56–57, 65, 87, 204 Petty, William, xv, 5, 14 Philadelphia, 57, 135n69, 265, 266, 270; as export hub, 48; inequality in, 65– 66; labor vs. property income in, 63, 65; poor whites in, 42; population of, 15, 90; taxation in, 276; trade shocks in, 88 Phillips, Peter, 178 Piketty, Thomas, xv, 2, 6, 112–13, 114, 138, 172, 173, 195, 205–9; education underemphasized by, 212n; political shift s stressed by, 227; on progressive taxation, 260 pitch, 49 planters, 14, 17 political arithmetic, 5, 14 population growth: antebellum, 10, 108, 110; Civil War dampening of, 146; coastal vs. interior, 44– 45; colonial, 13, 54–57; in early republic, 94; immigration and, 211; between world wars, 209 pork, 48 potatoes, 298 probate records, 4, 17, 26, 59, 63, 65, 266, 274 productivity, 53, 79, 104, 111, 150; agricultural, 89, 110, 138; preindustrial, 47; of slaves, 131, 291–92, 297–99; from technology, 138, 215–16 professionals, 17, 24, 178, 202, 270, 275, 279, 283 Prussia, 258 public debt, 79, 100–101, 170 public employees, 279–80

INDEX

•

395

Rabinowitz, Howard, 323 Race between Education and Technology, The (Goldin and Katz), 114 race riots, 223 racial segregation, 192, 203, 223. See also civil rights movement; Jim Crow Ransom, Richard, 159, 160, 161– 63, 272, 323–24, 334, 339 Reagan, Ronald, 240, 243 real property, 20, 151, 317–18, 333; appreciation of, 27, 271, 275; in census records, 17, 314–16; in probate records, 59, 65; taxation of, 78, 275, 277–78 regulation, 11–12, 201, 208n, 221, 249; during Great Leveling, 200, 217; loosening of, 12, 238–39, 240– 41; recommendations for, 258, 260, 262 rent seeking, 41 respectability bundles, 72–74, 309–10 retirement, 244 Revolutionary War, 8, 56, 165; cost of, 10, 77, 85, 87, 92–93, 96, 109, 143, 147, 251, 253; uncertainties caused by, 138; urban vs. rural impact of, 108 Rhode, Paul, 110, 298–99 Rhode Island, 47 Ricardo, David, 209 rice, 34, 44, 48–50, 75, 88, 109, 298, 299 Richmond, Va., 35 Robinson, James, 93 Roine, Jesper, 198 Roosevelt, Franklin D., 208 Rosenbloom, Joshua, 45, 66, 67, 75, 79, 92, 296–97, 303 Rosenthal, Caitlin, 178 Rothenburg, Winifred, 26–27 Ruggles, Steven, 4 rum, 40, 47 ruralization (de-urbanization), 15, 57–58, 75, 123, 138 Russia, 86, 235 rye, 298 Saez, Emmanuel, 2, 112, 172, 195 Sala-i-Martín, Xavier, 184, 185 Savannah, Ga., 48, 90 savings and loan crisis, 239, 260 Schwartz, Anna, 144 Second Industrial Revolution, 167, 177 Securities and Exchange Commission, 239

396

•

INDEX

segregation, 192, 203, 223. See also civil rights movement; Jim Crow services, 47, 144– 45, 180 Seven Years’ (French and Indian) War, 56, 65 Shammas, Carole, 236 sharecropping, 17, 160, 338 Shepherd, James, 88, 89 shipbuilding, 203– 4, 283 shopkeepers, 17, 132n62, 265, 270, 279 Silver, Nate, xv Singapore, 242n1, 254 skill premium, 11, 79, 95, 119, 126, 209; in early republic, 131–32, 136–37, 176–77; in Europe, 202; immigration and, 167, 179, 192; industrialization and, 113–14; labor force growth vs., 209; urbanization and, 113–14. See also education slaveholders: income of, 151, 274; residences of, 17; slaves grouped into households by, 266, 289–90; slaves mistreated by, 191, 301–2n32; taxation of, 275, 276, 277 slaves, 24, 36, 37, 63, 68, 110, 271; in Charleston, 127n53; children of, 9n, 21n8; consumption by, 25, 288, 293, 296–300, 320; depreciation of, 27, 274, 275, 316; emancipated, 11, 21n8, 143, 147, 151, 159, 160– 61, 163, 243, 302; emancipation of, 11, 142, 146, 150, 159– 63, 165, 185, 189, 191, 198, 320; exploitation rate of, 273, 274, 288, 290, 293, 297, 299–300; “households” of, 266, 289–90; importation of, 43, 44, 191; incommensurable losses to, 8, 38; inequality and, 13, 114, 117, 118, 121, 129, 131, 136; labor force participation of, 16, 34, 77, 161, 162, 163, 284, 289; living standards of, 43; mistreatment of, 191, 301–2n32; number of, 66n33, 94, 266, 294–95; productivity of, 131, 291–92, 297–99; rate of return on, 274; rental of, 288, 290, 293, 294–96, 300–302; retained earnings of, 5, 10, 11, 19, 22, 23, 33, 60, 66, 77–78, 131, 160, 190, 191, 264, 265, 268, 272–74, 276, 284, 287–303; runaway, 87; skilled vs. unskilled, 23, 290–91; sources on, 97, 144n9, 189, 266, 269, 311, 314; tax on holdings in, 275, 276, 277; value of, 113, 118, 143, 277; working hours of, 161, 283, 291–92 Smith, Adam, 61 Smith, Billy Gordon, 276

soap, 70 social spending, 208 Social Structure of Revolutionary America, The (Main), 3, 36, 269 social tables, 5– 6, 14, 25, 39, 41– 42, 93, 121, 149 Sokoloff, Kenneth, 56, 93 Soltow, Lee, 111, 277 Soto, Marcelo, 212 South Carolina, 16, 57, 67, 129, 224, 298, 344 Southeast Asia, 237 South Korea, 233, 236, 246n, 259 Spain, 205 squash, 44 Stanford, Leland, 141 Stoddert, Benjamin, 78n3 Stowe, Calvin E., 140, 141 Stowe, Harriet Beecher, 140, 141 sugar, 299 Sutch, Richard, 86, 100, 159– 63, 272, 293, 323–24, 334, 339, 346 Sweden, 200n10, 202, 205, 235, 254, 255n, 259 sweet potatoes, 298 Switzerland, 110, 169, 233, 254, 259 Sylla, Richard, 135 Taiwan, 236, 259 tar, 49 tariffs, 88, 170 taxation: during Great Leveling, 196; of inheritances, 259– 60, 262; of incomes, 156; of interstate trade, 88; of property, 78, 275–77; progressive, 137, 156, 196–97, 207, 218, 257, 260; records of, 15, 17, 19, 21, 266; for schools, 118, 187; of slaveholders, 275, 276, 277; wars financed by, 78 teachers: in-kind payments to, 280; rising demand for, 177; secondary occupations of, 282; southern, 118, 162– 63; wages of, 162– 63, 177, 201, 281, 284, 323. See also education technology, 208, 209, 218, 239, 240, 243; artisan class hollowed out by, 125, 129; education vs., 212n, 215–16, 230–32; employment in, 224; immigration-induced changes in, 182; inequality heightened by, 177–78, 229, 250; outlook for, 248– 49; productivity enhanced by, 138, 215–16; women’s pay boosted by, 225 Temin, Peter, 291–92

Tennessee, 296, 331, 334, 342– 48 terms of trade, 51, 52, 53, 57, 66, 93, 94 Texas, 344 Thatcher, Margaret, 240, 243 Time on the Cross (Fogel and Engerman), 273–74, 277, 287, 291, 293 tobacco, 34, 44, 48, 50, 66, 75, 88, 109, 298, 299 Tocqueville, Alexis de, 117, 235–36 “top coding,” 2n2 trade globalization, 71n41, 236–39, 240, 247, 250, 264 transfer payments, 137, 194, 196, 257 transportation, 52, 138 Tucker, Rufus, 144 Turner, Frederick Jackson, 118, 145, 183, 185 turpentine, 49 underemployment, 282–83 Understanding the Gender Gap (Goldin), 203 unemployment, 258, 282 unions, 223, 247 United Kingdom. See Great Britain upper-gap measures, 221–22 urbanization, 56–57, 58, 136, 138, 282; after Civil War, 167, 174; inequality heightened by, 192; Kuznets’s thesis of, 123–29, 159, 167, 175, 176, 183; skill premium and, 177. See also ruralization usury laws, 32 Vanderbilt, Cornelius, 140, 141, 315 van Zanden, Jan Luiten, 285–86 Vedder, Richard, 293 Vietnam War, 227 Villaflor, Georgia, 56 Virginia, 35, 36, 48, 57, 298, 344 volatility, 44, 47, 49–52, 53, 57, 75 voting rights, 185n41, 187, 192 wage gaps, 85, 126, 176, 200–202 Waldenström, Daniel, 198 Wales, 14, 39, 40, 42 Walsh, Lorena, 66 Walton, Gary, 88, 89 Wanamaker, Marianne, 186–87 Ward, Marianne, 71 Warden, Gerard, 276 wars, 206, 242, 250 Washington, D.C., 78, 140, 205, 224, 344 Washington, George, 78

INDEX

•

397

Weiss, Thomas: growth rate estimates of, 4–5, 45, 59, 66, 67, 75, 79, 92, 96, 97–98, 100, 283; labor force estimates of, 16, 22, 77, 82, 269, 289; slave consumption estimates of, 296–97, 303 welfare state, 208, 240, 257 West Germany, 205 West Indies, 7, 10n16, 43, 47, 66 West Virginia, 344 whale products, 47, 70n, 89 Whartenby, Frank, 299 wheat, 44, 48, 50, 298 Williamson, Jeff rey, 246 Williamson, Samuel, 86, 103 Williamson, Triphemius, 140, 141 Wolcott, Oliver, 277

398

•

INDEX

wood products, 47, 49, 51 working hours, 24–25, 33, 280–83; of slaves, 161, 283, 291–92; in South, 34, 160 World Top Incomes Database, 194, 196, 198 World War I, 90, 149, 207, 210 World War II, 9, 90; American income surge from, 149, 255; devastation wrought by, 207, 217; growth miracles after, 168; inequality reductions during, 198, 199, 203, 205, 206, 217; inflation during, 218; trade barriers during, 216; veterans of, 227 Wright, Carroll, 23 Wright, Gavin, 106, 129, 224, 322 Zucman, Gabriel, 6

THE PRINCETON ECONOMIC HISTORY OF THE WESTERN WORLD Joel Mokyr, Series Editor Growth in a Traditional Society: The French Countryside, 1450–1815 by Philip T. Hoff man The Vanishing Irish: Households, Migration, and the Rural Economy in Ireland, 1850–1914 by Timothy W. Guinnane Black ’47 and Beyond: The Great Irish Famine in History, Economy, and Memory by Cormac Ó. Gráda The Great Divergence: China, Europe, and the Making of the Modern World Economy by Kenneth Pomeranz The Big Problem of Small Change by Thomas J. Sargent and François R. Velde Farm to Factory: A Reinterpretation of the Soviet Industrial Revolution by Robert C. Allen Quarter Notes and Bank Notes: The Economics of Music Composition in the Eighteenth and Nineteenth Centuries by F. M. Scherer The Strictures of Inheritance: The Dutch Economy in the Nineteenth Century by Jan Luiten van Zanden and Arthur van Riel Understanding the Process of Economic Change by Douglass C. North Feeding the World: An Economic History of Agriculture, 1800–2000 by Giovanni Federico Cultures Merging: A Historical and Economic Critique of Culture by Eric L. Jones The European Economy since 1945: Coordinated Capitalism and Beyond by Barry Eichengreen War, Wine, and Taxes: The Political Economy of Anglo-French Trade, 1689–1900 by John V. C. Nye A Farewell to Alms: A Brief Economic History of the World by Gregory Clark Power and Plenty: Trade, War, and the World Economy in the Second Millennium by Ronald Findlay and Kevin O’Rourke Power over Peoples: Technology, Environments, and Western Imperialism, 1400 to the Present by Daniel R. Headrick Unsettled Account: The Evolution of Banking in the Industrialized World since 1800 by Richard S. Grossman

States of Credit: Size, Power, and the Development of European Polities by David Stasavage Creating Wine: The Emergence of a World Industry, 1840–1914 by James Simpson The Evolution of a Nation: How Geography and Law Shaped the American States by Daniel Berkowitz and Karen B. Clay Distant Tyranny: Markets, Power, and Backwardness in Spain, 1650–1800 by Regina Grafe The Chosen Few: How Education Shaped Jewish History, 70–1492 by Maristella Botticini and Zvi Eckstein Why Australia Prospered: The Shifting Sources of Economic Growth by Ian W. McLean Cities of Commerce: The Institutional Foundations of International Trade in the Low Countries, 1250–1650 by Oscar Gelderblom Lending to the Borrower from Hell: Debt, Taxes, and Default in the Age of Philip II by Mauricio Drelichman and Hans-Joachim Voth Power to the People: Energy in Europe over the Last Five Centuries by Astrid Kander, Paolo Malanima, and Paul Warde Fragile by Design: The Political Origins of Banking Crises and Scarce Credit by Charles W. Calomiris and Stephen H. Haberi The Son Also Rises: Surnames and the History of Social Mobility by Gregory Clark Why Did Europe Conquer the World? by Philip T. Hoff man The Rise and Fall of American Growth: The U.S. Standard of Living since the Civil War by Robert J. Gordon Unequal Gains: American Growth and Inequality since 1700 by Peter H. Lindert and Jeff rey G. Williamson