Research in Economic History, Volume 25 (Research in Economic History)

LIST OF CONTRIBUTORS Giovanni Caggiano Department of Economics, University of Glasgow, Scotland, UK Klas Fregert Dep...

Author: Alexander J. Field | Gregory Clark | William A. Sundstrom

118 downloads 1886 Views 3MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

LIST OF CONTRIBUTORS Giovanni Caggiano

Department of Economics, University of Glasgow, Scotland, UK

Klas Fregert

Department of Economics, Lund University, Lund, Sweden

Roger Gustafsson

The National Board of Housing, Building, and Planning, Sweden

Gregg Huff

Department of Economics, University of Glasgow, Scotland, UK

Drew Keeling

Department of History, University of Zurich, Zurich, Switzerland

John Komlos

Department of Economics, University of Munich, Munich, Germany

Peter C. Mancall

Department of History, University of Southern California, Los Angeles, CA, USA

Gary Richardson

Department of Economics, University of California, Irvine, CA, USA

Joshua L. Rosenbloom

Department of Economics, University of Kansas, Kansas, USA

Thomas Weiss

Department of Economics, University of Kansas, Kansas, USA

vii

EDITOR’S INTRODUCTION Research in Economic History, Volume 25, includes six chapters covering a range of geographic areas and tackling a range of issues in economic history. The ﬁrst two address United States topics, one analyzing data from the eighteenth and the other from the twentieth century. Both have a macroeconomic focus. Peter Mancall, Josh Rosenbloom, and Tom Weiss consider growth in colonial America, while Gary Richardson examines the role of bank failures in propagating the Great Depression. Two chapters cover European topics. John Komlos examines the heights of rich and poor youth in England in the late eighteenth and early nineteenth century with a view to what these data can tell us about inequality and trends in the standard of living. Klas Fregert and Roger Gustafson provide a synoptic view of public ﬁnances in Sweden from the eighteenth to the twentieth century. In the ﬁfth chapter, Drew Keeling studies the economics of the steamship industry that facilitated migration between Europe and the United States between 1900 and 1914. Finally, Gregg Huff and Giovanni Caggiano use newly assembled data to look at the integration of labor markets in Southeast Asia in the late nineteenth and early twentieth century in the context of a globalizing economy. We begin with Mancall, Rosenbloom, and Weiss on the economy of the lower South between 1720 and 1774. Scholars have generally recognized the lower South (Georgia and the Carolinas) as one of the wealthiest regions in British mainland North America, and the conventional wisdom has attributed growth in per capita output principally to rice and indigo exports. Mancall, Rosenbloom, and Weiss question the standard interpretation. They argue ﬁrst that overall export growth, although robust, was slower than the growth of exports of rice and indigo considered alone. Second, population grew faster than exports so exports per capita actually declined between the 1720s and the 1770s. Finally, the impact of rice and indigo exports was concentrated in a small geographical area without strong linkages to the rest of the economy, and affected a relatively small number of people. The authors call for more attention to productivity trends in ix

x

EDITOR’S INTRODUCTION

domestic production in understanding the economic history of these colonies prior to the Revolution. Concerned with capacity utilization and unemployment rather than growth, Gary Richardson looks at the United States two centuries later, in the throes of the Great Depression. His contribution makes available detailed data from the National Archives on the role of bank panics and failures in propagating the economic downturn. Up until now, most studies of the role of panics have depended upon a summary of data collected by the Federal Reserve System and published in the Federal Reserve Bulletin of September 1937. The data that Richardson has retrieved from the National Archives and that he publishes here are the same as those underlying that publication. But these tables provide much more information than has been easily accessible before on the causes of bank suspension and their regional incidence. They should prove to be a valuable resource for researchers. John Komlos ﬁres another salvo in the war between the optimists and the pessimists regarding the impact of the industrial revolution on living standards in England. He begins by re-examining heights of upper class boys from the Royal Military Academy at Sandhurst. The data were originally collected by Roderick Floud, but Komlos makes additional adjustments based on truncation due to a minimum height requirement. Since some students were fee-paying and others were not, he also looks at the inﬂuence of family income on height. His overall conclusion: upper class English youth were exceptionally tall, only an inch shorter than the average for the United States today. He then turns his attention to the English poor and data, also originally collected by Floud, on boys inducted into the Marine Society. The Marine Society, founded in 1756, was a charity intended to assist poor youth and provide mariners for the King’s Navy and the British merchant marine. Komlos has argued previously that although Floud adjusted for truncation at the lower end of heights, the sample was truncated at the upper end as well, because excessive height was also a barrier to successful service on ships. Komlos ﬁnds huge differences (6–9 inches) between the heights of rich and poor youth, emphasizing the class nature of British society and its implications for food availability. He ﬁnds that the gap widened in the late eighteenth century and in the 1830s and 1840s, reinforcing the pessimistic view that the industrial revolution adversely affected living standards, particularly those of the poor, during these periods. Our fourth chapter, by Klas Fregert and Roger Gustafson, provides a comprehensive look at Swedish public ﬁnances between 1719 and 2003. Their database includes estimates of government revenues, expenditures,

Editor’s Introduction

xi

deﬁcits, and debt. They test for the reliability and consistency of their series by examining the relationship between estimates of deﬁcits and estimates of changes in the debt. Differences between these, they ﬁnd, are serially uncorrelated around zero, giving them some conﬁdence in their procedures. Drew Keeling studies the economics of the passenger steamship business in the ﬁnal decades of the great era of transatlantic migration. That era came to an end with unrestricted U-boat warfare during the First World War. The US Immigration and Naturalization Acts of 1921 and 1924 then ﬁnished what the torpedoing of the Lusitania had begun. But prior to that, millions of Europeans migrated to the United States, and Keeling studies the role played by the passenger steamship companies in the decade and a half prior to the First World War. Perhaps surprisingly, rising immigration over this period was not fueled by declines in the real cost of transport. Transit costs actually rose, and steamship companies passed on productivity improvements in the form of qualitative changes in the conditions of transport. Keeling’s chapter is a fascinating study of how companies managed risk in an unforgiving economic environment. Steamship companies faced a combination of huge variability in demand over the year and over the business cycle, and very high ﬁxed costs. The result was that their load factors averaged only about 40 percent, compared with 78 percent for the modern air travel business. Our ﬁnal chapter, by Gregg Huff and Giovanni Caggiano, examines labor market integration in Southeast Asia in the late nineteenth and early twentieth century. Their focus is on new data on migration from labor abundant regions of South India and Southeastern China to the labor scarce regions of Burma, Malaya, and Thailand. They test to determine whether source and destination markets were integrated, using three criteria. Markets are viewed as integrated if (1) wages exhibited a common trend, (2) there was no increase in wage dispersion over time, and (3) equilibrating mechanisms restored long-term relationships following shocks. The authors ﬁnd that Southeast Asian labor markets were indeed integrated, as were labor markets in core industrial countries. On the other hand, integration did not extend between the labor markets of Southeast Asia and those of the core industrial countries, which exhibited divergent trends in unskilled real wages. Globalization appears to have been more effective in integrating commodity than labor markets. Research in Economic History welcomes innovative contributions to economic history in any area. We continue to have more ﬂexibility than other outlets in publishing data-rich papers running to somewhat greater length. Potential authors may submit their work in hard copy or (this is

xii

EDITOR’S INTRODUCTION

preferred) as attachments in an email addressed to the editor (aﬁeld@scu. edu). We are ﬂexible on matters of style and formatting for the ﬁrst round of submission. If it looks as if the paper has a reasonable likelihood of acceptance, we will ask that you prepare it according to JAI/Elsevier guidelines, and are happy at any stage to send you a copy of these guidelines as well as a recently prepared manuscript for use as a template. Authors may wish to consult a recent volume of REH for examples of house style. Alexander J. Field Editor

EXPORTS AND THE ECONOMY OF THE LOWER SOUTH REGION, 1720–1770 Peter C. Mancall, Joshua L. Rosenbloom and Thomas Weiss ABSTRACT Scholars have long emphasized that the Lower South was one of the most economically successful regions of British North America. The region had the highest levels of private wealth per capita in the colonies by 1774, and it has been argued that income per capita rose rapidly due to the rapid growth of rice exports. Here we present new and more comprehensive estimates of the region’s exports, which reveal a different result. While exports grew rapidly, they grew slower than rice and indigo alone, and slower than population. Here we explain why the extensive growth of exports and population did not lead to rapid growth of income per capita.

For the past generation scholars have emphasized that the Lower South – the Carolinas and Georgia – was perhaps the most economically successful region of British mainland North America. Planters, the primary economic actors, made extensive use of slave labor to generate a successful stapleexport sector, which by 1774 produced the highest levels of private wealth per capita in the mainland colonies.1 Until recently, however, our Research in Economic History, Volume 25, 1–68 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(07)25001-3

1

2

PETER C. MANCALL ET AL.

knowledge about the economy of the Lower South did not go much beyond such generalities. Writing in 1985, John McCusker and Russell Menard (1985, Ch. 8) decried the lack of knowledge about this region in their wideranging analysis of the economy of British America. Though subsequent work by others has broadened our knowledge, these studies continue to pivot on the centrality of exports as the primary engine of economic growth.2 Even while scholars have acknowledged the limitations of the ‘‘staples thesis,’’ it has nonetheless remained the primary organizing tool in most work on the economy of eighteenth-century British America.3 By focusing on the rapid growth of the rice and indigo trades of the Lower South, most scholars have concluded that standards of living for colonists in the region must have been rising rapidly. Writing in 1998, for example, Marc Egnal suggested that per capita incomes in the Lower South increased at an annual average rate of 0.9 percent per year from 1713 to 1775, implying that per capita incomes grew by more than 70 percent over the entire period (Egnal, 1998, p. 43). This ﬁnding ﬁts a consensus that the mainland colonies as a whole grew as rapidly as 0.6 percent per year between 1720 and 1774, the ﬁgure found in the recently published Historical Statistics of the United States (Carter, Gartner, Hainer, Olmstead, Sutch, & Wright, 2006, p. 631). But it is possible that this conventional view of the colonies of the Lower South exaggerates the rate of economic growth in the region because it is based on an incomplete understanding of the dynamics of regional exports.4 We believe that previous analysts have been misled in their assessment of the potential for growth because they focused on only a subset of the most successful exports and did not take into account the slower growth of other export commodities. Moreover, these scholars have given too little attention to the rapid growth of the region’s population, which was linked to the increase in aggregate exports. In what follows we present a new and much more comprehensive measure of regional exports, compare that with the growth of population, and set export performance in the context of the entire economy.5 Our more comprehensive measure of exports shows somewhat slower growth than that for rice and indigo alone, but still reveals that the export sector for the region was buoyant and successful, with the real volume of exports growing by more than 4 percent per year between 1720 and 1770. But these rapid rates of expansion are nonetheless inconsistent with any signiﬁcant increase of GDP per capita in the region for at least three reasons. First, regional population grew even more rapidly than did the real volume of exports, causing exports per capita to decline between 1720 and 1770. Population increased in part because the methods of production required large increases

Exports and the Lower South, 1720–1770

3

in the labor force (and hence population), and because many colonists chose to increase their family size. Second, the success of the two staple exports was conﬁned almost entirely to a limited geographic area and a small part of the region’s population. Third, production and productivity in the domestic sector, which were previously under-emphasized, were important determinants of the region’s economic performance.6 Of course, even if the per capita quantity of exports did not change, the nominal value of exports could have risen if the prices of key exports rose relative to imports over time, allowing the colonists to purchase more imports for any given quantity of exports. But, as we show below, the favorable effects of changes in the terms of trade were small. Drawing inferences about overall regional economic performance from exports requires embedding measures of exports within a model of the entire economy. After surveying the available export evidence we brieﬂy discuss a framework within which export data can be combined with evidence on the value of domestically consumed agricultural production to yield consistent estimates of per capita GDP for the Lower South. This exercise reveals that exports accounted for only about one-quarter of regional economic activity in the eighteenth century, and suggests that there was little if any growth in GDP per capita prior to 1800. This of course does not in any way contradict views of regional economic success. As we show, free colonists were doing very well economically, enjoying per capita incomes that were quite high even in comparison to income levels in the nineteenth century.

EXPORTS FROM THE LOWER SOUTH Exports were crucial to economic success in colonial British North America. That, at least, is the argument advanced by both historians and economists. Whether one looks at the literature about the colonies taken as a whole, or for any of the major regions, exports loom large as the primary engine of economic growth. According to the dominant theme found in textbooks as well as scholarly works, enterprising Europeans arrived in North America and through hard work and the good fortune of abundant land created a prosperous and burgeoning economy based on the export of agricultural staples. There is an appealing intuition to this argument. After all, extracting wealth from North America was one of the factors motivating the English since the age of Queen Elizabeth I.7 As a result, English and AngloAmerican authorities often kept careful track of exports from the colonies. And, on the face of it, there are grounds to support the argument: some

4

PETER C. MANCALL ET AL.

exports increased rapidly and provided the appearance of economic growth, while population and aggregate GDP expanded at unprecedented rates. The central role of exports in prior estimates of the southern economy follows from scholars’ extensive use of the records of low-country planters and those with whom they organized the regional rice trade. To be sure, exports – especially of rice – grew rapidly and represented a substantial output. But a comprehensive picture of regional export performance requires that we look beyond rice and indigo to consider exports of naval stores and deerskins. Although cotton was the region’s primary export in the nineteenth century, it did not emerge as a signiﬁcant contributor to exports until the 1790s. Consequently, it is not a factor in economic growth in the colonial period, but must be considered when measuring economic growth over the longer period to 1800. Table 1 summarizes data on the quantity and value of these ﬁve items along with evidence on the value of all other exports for 1768–1772, drawn from the American Inspector-General’s ledgers. According to James Shepherd this is ‘‘the only complete source for commodity trade for any years in the colonial period’’ (Shepherd, 1969, p. 9). At the end of the colonial period these ﬁve commodities accounted for 80 percent of the value of regional exports. Table 2 presents data on the output and growth of each of these commodities over the course of the eighteenth century. The top panel shows the quantity (three-year averages) of each of the ﬁve exports at decadal benchmark dates; the bottom panel summarizes the rates of growth of these exports over selected periods. In 1712, the quantity of rice exports topped 3 million pounds, and by 1720 had more than doubled to 8.1 million pounds and continued to climb, reaching a temporary peak of 34 million pounds in 1740. That initial period of expansion ended with the start of the international conﬂict known as the War of Jenkins’ Ear, which pitted Britain against Spain in a contest for who would control shipping from the Caribbean and Central America. This conﬂict, as well as King George’s War (1744–1748), substantially raised shipping and insurance costs and caused a sharp drop in the net prices received by rice farmers (Mancall, Rosenbloom, & Weiss, 2001, pp. 616–639). The depressed conditions persisted for most of the decade, during which planters experimented with other crops, including indigo, which emerged as an important complement to rice.8 With the return of peace in the late 1740s, rice prices recovered along with exports. In the early 1760s, rice prices began to increase again, and exports shot upward in the decade and a half before the Revolution.9 By 1770 exports had increased to over 70 million pounds. This expansion implies a compound average annual

Exports and the Lower South, 1720–1770

Table 1.

5

Quantity and Value of the Major Exports from the Lower South, 1768–1772. 1768

1769

1770

1771

1772

137,740 517,301 85,388 3,300

102,833 416,436 91,604 392,739 544

148,831 573,017 76,375 328,832 2,444

204,894 454,207 86,852 438,344 2,615

127,662 758,677 71,882 359,482 1,128

Value of major exports in pounds sterling Rice 322,878 Indigo 78,113 Naval stores 31,381 Deerskins – Cotton 170

284,216 75,375 32,856 39,838 25

260,584 103,430 28,860 31,731 105

282,985 106,285 33,035 42,241 98

377,001 196,118 32,412 34,563 45

432,542 538,000

432,310 551,000

424,710 534,000

464,644 593,000

640,139 800,000

80.4

78.5

79.5

78.4

80.0

Quantities of major exports Rice (bbl) Indigo (lb) Naval stores (bbl) Deerskins (lb) Cotton (lb)

Sum of the above Value of all exports from the region (pounds sterling) Major export share (%)

Notes: The value of all exports is an independent estimate made by contemporaries, and accepted by Shepherd and Walton (1972, pp. 93–95). Shepherd and Walton reported the 1769 ﬁgure for rice exports to southern Europe in hundredweight. We converted into barrels by dividing the reported value by the price per barrel (2.1814 pounds sterling) implicit in the Shepherd and Walton ﬁgures for rice exports to Great Britain and the West Indies in 1769. That conversion implies there were 5.25 hundredweight per bbl. Source: Shepherd and Walton (1972, Appendix IV, Tables 2–6 pp. 211–227).

rate of growth of nearly 5.7 percent from 1712 to 1770. The upward trajectory of rice exports came to an end with the American Revolution as substantial conﬂict in the region led to substantial material losses.10 In 1790, the next year for which we have data, exports were almost 42 million pounds, slightly less than they had been in 1760. Although the quantity of rice exported increased during the 1790s, in 1800 it remained well below the peak reached during the colonial period. Indigo production, by contrast, did not begin in earnest until the 1740s, but, as Table 2 reveals, during the next two decades the quantity of indigo exported from the Lower South expanded quite rapidly. In the 1750s, exports of indigo grew at the astonishing rate of 30 percent per year on average. The growth of rice and indigo exports obscures changes that were taking place in the production of other export commodities. While it is true that rice and indigo became the most important exports in the Lower South in

6

PETER C. MANCALL ET AL.

Table 2. Quantities of Major Exports from the Lower South, 1712–1800. Rice (lb)

Naval Stores (bbl)

Deerskins (lb)

Indigo (lb)

Cotton (1,000 lb)

Panel A: Three-year average benchmark figures by year of production 1712 3,168,625 9,506 179,350 1720 8,060,551 33,505 120,721 1730 19,131,450 30,564 229,244 1740 34,917,672 33,148 219,575 1750 34,123,207 66,594 285,387 28,933 1760 47,080,950 46,407 242,874 389,767 1770 78,227,450 81,500 375,553 595,300 y 1790 41,911,388 50,188 71,269 488,017 1800 45,275,731 37,772 146,227 4,790

112 18,681

Panel B: Average annual rates of change By decade 1712–1720 12.38 17.05 1720–1730 9.03 0.91 1730–1740 6.20 0.82 1740–1750 0.23 7.23 1750–1760 3.27 3.55 1760–1770 5.21 5.79 y 1790–1800 0.78 2.80

2

4.83 6.62 0.43 2.66 1.60 4.45

29.70 4.33

7.45

37.02

66.76

7.49 14.85

35.49

By subperiod 1720–1740 1740–1770 1770–1800

7.61 2.73 1.81

0.05 3.04 2.53

3.04 1.81 3.10

Over the long term 1712–1770 1720–1770 1720–1800

5.68 4.65 2.18

3.77 1.79 0.15

1.28 2.30 0.24

Notes: The rate of growth for indigo shown for the period 1740–1770 covers only the period from 1746, the ﬁrst year of production. Source: See the appendix to this chapter.

the eighteenth century, the rise of those trades came at the expense of others. Planters employed the labor under their control to maintain rice plantations and harvest indigo, which meant they put less labor toward the production of naval stores, beef and pork, or maize (Indian corn).11 And there was less

Exports and the Lower South, 1720–1770

7

emphasis on the trade in deerskins, most of which arrived in colonists’ hands from native suppliers. The much slower growth of deerskins and naval stores exports can be seen in Table 2.12 Between 1712 and 1770 the export of deerskins rose at 1.3 percent per year, while naval stores increased at 3.8 percent per year. The performance of these four items (rice, indigo, deerskins, and naval stores) taken together gives a close approximation to the behavior of total exports during the colonial period. In 1746–1747, these four items comprised 89.3 percent of exports from Charleston, whereas in 1768–1772 they made up 89.7 percent of the region’s exports.13 Table 3 presents our estimates of the real volume of all exports from the region expressed in 1840 prices at benchmark dates over the course of the eighteenth century.14 As the evidence in Table 3 makes clear, although total foreign exports rose at a rather rapid pace – 4.27 percent per year between 1720 and 1770 – this was substantially slower than the growth in rice exports alone. On the other hand, over the longer period, from 1720 to 1800, total exports grew more than rice exports alone because of the effects of the rapid growth of cotton production after 1790. Looked at in the broader context, it is clear that the success of rice exports in the colonial period has given a distorted picture about the role of exports more generally.

EXPORT PERFORMANCE IN CONTEXT While exports were crucial for low-country rice planters, an understanding of the impact of the export sector on the Lower South as a whole must look more broadly across the region, taking into account not only the full range of regional exports, but also the enormous economic activity devoted to non-export functions. Georgia, North Carolina, and the backcountry of South Carolina – which are bit players in others’ economic histories – need to be integrated into any assessments of regional economic patterns.15 The place to begin is with population. As can be seen in Fig. 1, the bulk of the region’s population in 1700 was concentrated in North Carolina, and though the colony exported some naval stores it remained largely insulated from foreign trade (Egnal, 1998, pp. 114–117). Georgia was established only in 1732, and prohibited slavery until 1749. Until it eliminated restrictions on slavery its population grew relatively slowly, but during the second half of the century it increased more quickly, expanding to account for about 15 percent of the region’s population.16 Although the growth of rice exports at the beginning of the eighteenth century contributed to the expansion of

8

PETER C. MANCALL ET AL.

Table 3.

Lower South Exports to Foreign Destinations, 1720–1800. Population (1,000s)

Agricultural Exports ($1,000s)

$ per capita

540.90 1,104.40 1,823.60 1,758.30 2,652.70 4,367.90 NA 3,867.20 9,510.20

13.64 18.41 16.15 12.28 12.65 12.63 NA 5.07 8.71

4.22 6.53 2.41 3.89 5.12

7.40 5.14 0.36 4.20 5.11

3.04 1.30 2.70 0.30 0.02

3.65

9.42

5.56

By subperiod 1720–1740 1740–1770 1770–1800

5.36 3.80 3.91

6.27 2.95 2.63

0.85 0.82 1.23

Over the long term 1720–1770 1720–1800

4.42 4.23

4.27 3.65

0.15 0.56

1720 1730 1740 1750 1760 1770 1780 1790 1800 Average annual rates of change By decade 1720–1730 1730–1740 1740–1750 1750–1760 1760–1770 y 1790–1800

39.7 60.0 112.9 143.2 209.8 345.8 516.2 762.4 1,091.4

Notes: Exports to foreign destinations are three-year averages valued in 1840 prices. See appendix to this chapter for details regarding the estimation.

South Carolina’s share of regional population, this ﬁgure peaked at 50 percent in 1730 and then declined. But even within South Carolina, a declining share of population was engaged in export production. The growth of rice exports produced in the low country coincided with the rapid expansion of settlement into the interior regions of the colony. For the most part, the settlement of the backcountry consisted of small independent farmers possessing few if any slaves, and producing only small quantities of marketable crops (Hughes, 1985, p. 119; Johnson, 1997, pp. 40–60; Klein, 1990, pp. 10–27).

Exports and the Lower South, 1720–1770

9

100%

80%

Share

60%

40%

20%

0% 1700 1710 1715 1720 1730 1740 1745 1750 1760 1770 1774 1780 1783 1790 1793 1800

Year NC

Fig. 1.

SC

GA

Distribution of the Population in the Lower South, 1700–1800. Source: U.S. Bureau of the Census (1975, Series Z:1–19).

As late as 1770, only 6,000 (8.7 percent) of South Carolina’s 76,000 slaves lived in the backcountry. In contrast, 30,000 (61 percent) of the colony’s 49,066 free inhabitants resided in this region (Coclanis, 1989, p. 68).17 That larger share of the population residing in the backcountry was virtually uninvolved in the export trade dominated by the low country. From 1768 to 1772, average annual exports per capita in Carolina’s low country averaged d3.7 for the entire population, and d17.1 for the white population. The backcountry ﬁgure was a mere d0.5 (Coclanis, 1989, p. 75). Nor does it seem that backcountry residents were producing food for the plantations since low-country slaves were largely self-sufﬁcient (Morgan, 1998, pp. 134–143).18 Residents of rice plantations also relied on hunting to make up for the fact that planters did not give them much meat, and they tended their own gardens, growing a variety of foods, including African foods (Morgan, 1998, pp. 138–141). When compared to the growth of population, the success of the export sector looks considerably less impressive, as Table 3 makes clear. While

10

PETER C. MANCALL ET AL.

exports were growing at 4.27 percent per year between 1720 and 1770, the region’s population was increasing at 4.42 percent per year. Consequently, the real volume of exports per capita was declining on average at 0.15 percent per year.19 Moreover, even the success of the rice industry pales when population growth is taken into account, with the volume of rice exports per capita having risen at only 0.3 percent per year between 1720 and 1770.20 Rather than serving as an engine of growth, the export sector is better understood as a source of short-run instability. Exports, especially rice and indigo, rose more rapidly than population in some decades and no doubt served to propel the region’s economy ahead in those years, but these times were intermingled with periods when exports grew slowly or even declined. The net result is that over the entire colonial period population grew more rapidly than exports, and thus exports per capita declined. There are two reasons why export growth could not keep ahead of population growth. First, the production of exports rose rapidly in the Lower South primarily because the labor input had increased rapidly, and that required an increase in the population. Second, economic success fueled other demographic changes that made growth in GDP per capita difﬁcult to achieve. In the Lower South of the eighteenth century, the methods of production yielded little in the way of improvements in agricultural output per worker.21 Instead, the increase in the volume of rice and indigo exports was predominantly the direct result of an increase in the amount of slave labor applied to the cultivation of those products. To some extent, the increased labor input to rice and indigo came from a reduced production of other goods, but to a far greater extent it required an increase in the slave population. And, because the natural rate of growth in the slave population was extremely low (although higher than in the Caribbean), the increase in the number of slave workers was achieved by importing them.22 Indeed, as Fig. 2 illustrates, during the colonial period decadal variations in slave imports into the region closely paralleled changes in the volume of exports. Although these imported slaves may have been of working age, and thus did not negatively affect the worker/population ratio, they nevertheless increased the population. Indeed, even the slave population grew faster than exports.23 Between 1720 and 1770, when the volume of exports from the Lower South to foreign destinations increased at 4.2 percent per year, the slave population was rising by 4.8 percent. Meanwhile, the composition of the free population was changing in ways that reduced the worker/population ratio. The overall labor force participation rate for the entire colony is a function of the participation rates for speciﬁc population groups and the relative importance of the

Exports and the Lower South, 1720–1770

11

25000 1770s 20000

Slave Imports

1760s 15000 1730s

1740s

10000

5000

1750s 0 -200 0

200

400

600 800 1000 1200 Change in Export Value

1400

1600

1800

2000

Fig. 2. Slave Imports as a Function of Change in Export Value by Decade, 1720–1770. Source: Morgan (1983, p. 87, 1998, p. 59). Export values are from Table 3.

different groups. In this economy, slaves had a higher participation rate than free persons, free males had a higher participation rate than free females, and those aged 10 and over had higher participation rates than those younger. Initially the colonial population of the Lower South consisted largely of males of working age. Over time, and in part as a result of economic success in the export trade, the number of women and children increased. More women migrated to the colonies, and successful farmers and planters married and chose to have more children. The trends in the composition of the population that inﬂuenced the labor force participation rate are shown in Fig. 3. As can be seen, the slave share of the regional population increased from 39 percent in 1720 to 43 percent in 1730, but thereafter the share dipped slightly and recovered, so that by 1774 the share was back at 43 percent, the same as in 1730. The growth of the slave share of the population in the 1720s would have helped push up the labor force participation rate in that decade, but the stability after that point implies no lasting impact on labor force to population ratios. The trends in the shares of the other population groups – the male share of the free population, the adult share of the male population, and the adult share of the slave

12

PETER C. MANCALL ET AL. 0.900 0.800 0.700

10 + share of slave population 10 + share of free males

Shares

0.600 0.500

Male Share

0.400 0.300

Slave Share

0.200 0.100 1720 1730 1740 1750 1760 1770 1774 1780 1783 1790 1793 1800 Year

Fig. 3. Composition of the Population of the Lower South, 1720–1800. Notes and Sources: The free and slave population for 1700–1780 are from U.S. Bureau of the Census (1975, Series Z:1–19). McCusker and Menard (1985, Table 8.1, p. 173) made slight revisions to the ﬁgures for 1710, 1720, and 1740, but did not report the white and black populations separately for each colony. The 1790 ﬁgures are from Rossiter (1909), and those for 1800 from Weiss (1992). For both years, the ﬁgures came originally from the US Census of 1800. The male shares of the population for 1790 were calculated from data in Rossiter (1909), those for 1800 are from the worksheets underlying Weiss (1992). The shares of the free and slave population aged 10 and over for 1800 are from the worksheets underlying Weiss (1992). The shares of males aged 10 and over in 1790 were estimated by assuming that the ratio of those aged 10 and over to those aged 16 and over in 1790 was the same as the ratio that existed in 1800. The data on those aged 16 and over in 1790 are from Rossiter (1909). The male shares of the population and the share of males aged 10 and over for 1700–1780 were estimated based on the evidence for 1790 and 1800, along with evidence on the distribution of the populations in selected colonies or parishes at various dates taken from the following sources: probate data underlying Chaplin (1993), Greene and Harrington (1932), Hughes (1985, p. 52), Jones (1980, pp. 117–120), Klingberg (1939, p. 496, 1941, pp. 58–60), Menard (1995, pp. 287–293), Morgan (1983, Table 5, p. 98, 1998, Table 31).

population – were all downward. That is to say, the components of the population that were increasing were those with lower participation rates, which resulted in a decline in the average participation rate for the entire colony. As a result of these demographic shifts the participation rate remained

Exports and the Lower South, 1720–1770

13

steady at 0.49 between 1720 and 1740, but declined thereafter to 0.47 in 1774. Unless it was offset by improvements in labor productivity, the decline in labor force participation rates implied by these demographic changes would have caused GDP per capita to decline by 4 percent between 1720 and 1774.24 These demographic shifts can be interpreted as an indication of economic success and prosperity, but the standard indicators of economic output, which measure the value of goods and services produced, do not reﬂect any increased value for this behavior. Indeed, the standard measure such as GDP per capita could decline as the output was shared with additional family members.

RE-INTERPRETING THE ECONOMIC PERFORMANCE OF THE LOWER SOUTH The focus on the export trade of the low country in the decades prior to the Revolution masks what was happening in the rest of the regional economy, where most production was taking place. Even if the export trade had grown faster than it did, and faster than the population, its impact would still have been restrained simply because the value of exports abroad likely accounted for only about one-quarter of regional economic activity.25 The dominant part of GDP was production for the domestic market. To understand the performance of the entire region requires more than an analysis of lowcountry rice exports. A more comprehensive assessment demands analysis of all the basic components of output. Elsewhere we have presented estimates of GDP for the region (Mancall et al., 2004). Table 4 presents estimates of output for the major components of the domestic sector underlying those GDP ﬁgures. Food produced for consumption within the region was the major component, comprising 50 percent or more in every year. And the value of food production exceeded the value of exports in every year. It is also worth noting that the value of non-agricultural output was nearly as large as the value of exports in every year, and grew at about the same rate over the colonial period, and somewhat faster after 1770. The details of the domestic sector are not, however, our chief interest. We are interested in placing exports in perspective by comparing them to the domestic sector and to GDP, as is done in Table 5. The economy of the Lower South expanded greatly between 1720 and 1770 with population and exports rising in excess of 4.0 percent per year, but there was no long-term upward movement in real GDP per capita. Within the colonial period the region experienced particular success in the 1720s and

14

Table 4.

Output of the Domestic Sector of the Lower South, 1720–1800.

Year

Food Consumed

Food Imported

Food Produced

Firewood

Shelter

Non-Agricultural Output

Domestic Sector Output

1720 1730 1740 1750 1760 1770 y 1800

900,112 1,358,163 2,610,737 3,412,969 5,115,313 8,588,774

29,977 101,417 316,694 330,577 448,192 359,567

870,135 1,256,746 2,294,043 3,082,391 4,667,121 8,229,207

163,263 247,544 464,242 803,642 1,203,615 1,986,491

91,275 134,749 268,655 349,983 521,275 888,289

480,527 701,709 1,244,090 1,563,137 2,376,259 3,941,888

1,605,200 2,340,748 4,271,030 5,799,154 8,768,270 15,045,875

30,184,719

1,145,964

29,038,755

6,658,683

3,758,473

13,767,211

53,223,122

12.96 12.06 0.43 3.09 2.18

3.74 6.20 3.00 4.24 5.84

4.25 6.49 5.64 4.12 5.14

3.97 7.14 2.68 4.06 5.47

3.86 5.89 2.31 4.28 5.19

3.84 6.20 3.11 4.22 5.55

By subperiod 1720–1740 1740–1770 1770–1800

5.47 4.05 4.28

12.51 0.42 3.94

4.97 4.35 4.29

5.36 4.97 4.11

5.55 4.07 4.93

4.87 3.92 4.26

5.01 4.29 4.30

Over the long term 1720–1770 1720–1800

4.61 4.49

5.09 4.66

4.60 4.48

5.12 4.74

4.66 4.76

4.29 4.28

4.58 4.47

Notes and Sources: Estimates are described in Mancall et al. (2004, pp. 389–424).

PETER C. MANCALL ET AL.

Average annual rates of change By decade 1720–1730 4.20 1730–1740 6.75 1740–1750 2.72 1750–1760 4.13 1760–1770 5.32

GDP, Major Components, and GDP per capita the Lower South, 1720–1800 (US Dollars, 1840 Prices).

Year

GDP (Narrowly Deﬁned)

Domestic Sector Output

Exports Abroad (Three-Year Averages)

Exports to Other North American Colonies

Exports Abroad as a Percentage of GDP

GDP per capita (Narrowly Deﬁned)

1720 1730 1740 1750 1760 1770 1780 1790 1800

2,247,403 3,571,884 6,305,971 7,723,909 11,804,613 19,901,062

1,605,200 2,340,748 4,271,030 5,799,154 8,768,270 15,045,875

64,370,772

53,223,122

540,879 1,104,445 1,823,638 1,758,314 2,652,693 4,367,879 – 3,867,239 9,510,205

101,223 126,690 211,302 166,441 383,650 487,308 – 1,143,626 1,637,091

24.0 30.9 28.8 22.6 22.4 21.9 NA NA 14.8

56.66 59.53 55.84 53.94 56.28 57.55 NA NA 58.98

Average annual rates of change By decade 1720–1730 4.74 1730–1740 5.85 1740–1750 2.05 1750–1760 4.33 1760–1770 5.36

3.84 6.20 3.11 4.22 5.55

7.40 5.14 0.36 4.20 5.11

2.27 5.25 2.36 8.71 2.42

0.50 0.64 0.34 0.42 0.22

By subperiod 1720–1740 1740–1770 1770–1800

5.01 4.29 4.30

6.27 2.95 2.63

3.75 2.82 4.12

0.07 0.10 0.08

5.29 3.91 3.99

Exports and the Lower South, 1720–1770

Table 5.

15

Year

Over the long term 1720–1770 1720–1800

GDP (Narrowly Deﬁned)

4.46 4.28

Domestic Sector Output

4.58 4.47

16

Table 5.

(Continued )

Exports Abroad (Three-Year Averages)

4.27 3.65

Exports to Other North American Colonies

3.19 3.54

Exports Abroad as a Percentage of GDP

GDP per capita (Narrowly Deﬁned)

0.03 0.05

PETER C. MANCALL ET AL.

Notes and Sources: GDP is the sum of output in the domestic and export sector, where the latter includes shipments abroad and shipments to other North American colonies. The estimates of exports abroad are three-year averages from Table 3. The estimates of output in the domestic sector are from Table 4. GDP as measured here is narrowly deﬁned to exclude land clearing and the value of home manufactures. If it were more broadly deﬁned to include those items, growth would be slower because both of those items were of greater importance early on in the period. All ﬁgures are real dollars expressed in terms of 1840 prices. These estimates are similar in concept to the earlier conjectures made for the early nineteenth century in that the scope of coverage is consistent as regards the extent to which marketed and non-marketed output is measured. See David (1967, pp. 151–197) and Weiss (1992, pp. 19–75, 1994, pp. 11–27). These GDP ﬁgures were estimated by extrapolating backward in time a base year value for 1800 that reﬂected a greater degree of market orientation than existed in 1720. They measure the value of the colonies’ output other than exports as though the same fraction were marketed in each year as prevailed in 1800. This makes sense when trying to gauge the output and standard of living of people whose economy may have been largely non-market oriented, but as a result the ﬁgures are conceptually different from the standard national income accounting measures wherein GDP is conﬁned for the most part to market transactions. In that scheme, GDP can increase with an increase in the extent to which output passes through markets, even when there is no increase in production. The present estimates minimize growth from that source. It may be that some of the divergence between the present estimates of growth in output per capita and those of previous researchers reﬂects the differences in concept. That is, the higher rates of growth estimated by others may reﬂect the increased output that is going through market channels rather than an increase in production. This, however, is only speculation, as other researchers have not made clear how much, if any, of their estimates reﬂect increased output versus increased market orientation. Of course, as Grubb (2004, p. 353) has argued, the increase in the marketed versus non-marketed activity can mean an increase in welfare per capita.

Exports and the Lower South, 1720–1770

17

1750s, with GDP per capita rising on average around one half of a percent per year, but these upward surges were offset by declines in GDP per capita in other decades. As a result of these offsetting performances, GDP per capita in 1770 was almost identical to that in 1720, and there was very little change up or down over the major subperiods of 1720–1740 and 1740–1770.26 The performance, moreover, reveals little consistency between the success of the export sector and that of the economy as a whole. Agricultural exports to foreign markets rose rapidly in some decades, but not all, and GDP per capita and even exports per capita did not always move in the same direction or with the same force. In the 1720s, there appears to have been a rather strong relationship between export growth and the growth of GDP per capita. In the 1720s, exports grew at 7.4 percent per year, the highest average rate achieved for any decade in the colonial period, and GDP per capita rose at 0.5 percent per year. In the 1740s, the relationship seems clear as well, but with unfavorable consequences: exports declined at 0.4 percent per year and GDP per capita declined at 0.3 percent per year. In the other decades of the colonial period, that is the 1730s, 1750s, and 1760s, the relationship is less clear. In the 1730s, exports increased at the strong rate of 5.1 percent per year, only slightly slower than in the previous decade, but GDP per capita declined at 0.6 percent per year. In the 1750s, exports grew more slowly than in any decade except the 1740s yet GDP per capita rose at 0.4 percent per year. In the following decade, exports grew faster than in the 1750s, but GDP per capita grew much more slowly at only 0.2 percent per year. These ﬁgures reveal that the domestic sector was an important inﬂuence on overall economic performance. Between 1720 and 1740, when exports abroad surged upward by 6.3 percent per year in the aggregate and 0.8 percent per year on a per capita basis, GDP was held in check by a slower growth of output in the more dominant domestic sector. Output there rose at an annual rate of only 5.0 percent because food production grew slowly between 1720 and 1740. Two related phenomena explain this situation. First, there was less need to produce food because the slave share of the population had increased, and on average slaves consumed less food. Second, the region imported more food, in all likelihood because landholders devoted more of their holdings to crops for export. The aggregate import of food rose at 12.5 percent per year, implying a $2 increase in the per capita value of food imports from other colonies between 1720 and 1740. Some of the growth of exports abroad was accomplished by shifting resources out of the production of food for the local market into the production of exports. After 1740 the food-producing sector was less of a drag on the region’s production, with the per capita value of food production rising in each decade. Had the export sector been able to

18

PETER C. MANCALL ET AL.

maintain the rapid pace of growth achieved in the 1720s and 1730s, or if it had slowed down less than it did, the region’s economic history might have been much different. But this was not the case. Export growth slowed so much after 1740 that the per capita value of exports declined at a rate of 0.8 percent per year through 1770.27 In these circumstances, the region was fortunate to have had the larger food and ﬁrewood sector, which plowed ahead steadily and kept GDP per capita from declining.

THE TERMS OF TRADE AND THE VALUE OF EXPORTS Economists are interested in constant price estimates of economic activity because it allows them to separate changes in productivity from the effects of shifts in relative prices. But from the point of view of the colonists, who engaged in trade to obtain goods produced more cheaply abroad, changes in the relative prices of imports and exports exerted an important inﬂuence on their well-being. If the terms of trade improved, the real quantity of imports that could have been obtained would have risen even if there were no increase in productivity or the quantity of exports produced.28 But, while relative prices of colonial imports and exports experienced large short-term swings, there was little long-term trend in these movements. The price of rice rose substantially in the 1720s and 1730s, but those gains disappeared when prices dropped during the War of the Austrian Succession in the 1740s. Prices only returned to their pre-war high after 1760 (Egnal, 1975, pp. 209–210). Indigo prices moved opposite to those for rice, soaring upward in the late 1740s when the crop was ﬁrst cultivated on a commercial scale, but declining after 1760. But neither of these export prices can adequately reﬂect the general movement of all export prices; changes in the terms of trade must take into account the different behavior of the various exports.29 The wholesale price indexes of South Carolina products prepared by Arthur Cole and George Rogers Taylor, which take into account the prices of all the major exports, provides an index to represent all exports.30 For the import price index, we have used the Gilboy–Schumpeter Price Index, as modiﬁed by John McCusker.31 English-manufactured goods made up the bulk of colonial imports, and this index as modiﬁed by McCusker is heavily weighted toward manufactured products.32 Although price data for imports into the Lower South would be preferable, they are not available for the entire period.33 Fig. 4a depicts the ratio of export to import prices over the colonial period. The terms of trade ﬂuctuated considerably, so it is difﬁcult to generalize about the impact on the colonists. Colonists beneﬁted from an

Exports and the Lower South, 1720–1770

19

1.8

Index (1797-1801=100)

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 17

17

17

17

17

17

17

17

17

17

17

72

67

62

57

52

47

42

37

32

27

22

(a)

Year

Index (1797-1801=100)

1.900 1.700 1.500 1.300 1.100 0.900 0.700 0.500 73 17 70 17 67 17 64 17 61 17 58 17 55 17 52 17 49 17 46 17 43 17 40 17 37 17 34 17 31 17 28 17 25 17 22 17

(b)

Year

Fig. 4. Ratio of Export Prices to Import Prices for Charleston (a) 1722–1775, (b) 1722–1773 (with Slave Prices Included in the Import Price Index). Sources and Notes: See discussion in text.

improvement in the terms of trade from 1722 through 1738 as the price of rice rose substantially, but those gains were lost over the next 10 years. With the rise in the price of indigo in the late 1740s the terms of trade improved brieﬂy, but fell for nearly a decade beginning in 1753. The terms of trade improved after the Seven Years’ War, but even with that recovery there was only modest improvement over the colonial period.34

20

PETER C. MANCALL ET AL.

The conventional terms of trade calculations depicted in Fig. 4a ignore slave imports, a vital component of the Lower South’s trade. A substantial portion of the region’s export proceeds was spent on slave imports, not on goods and services, and such importation has implications for the terms of trade and the real incomes and standards of living of colonists.35 The value of slave imports as a percentage of the value of exports ran between 10 and 31 percent.36 A second time series on the terms of trade that takes into account the importation of slaves by combining the import price index discussed above with an index of slave prices is shown in Fig. 4b.37 As seen there, this alternative version of the terms of trade showed little or no improvement from 1722 to 1773, with the ending year index being only 10 percent above that for 1722.38 The much slower rise in this terms-of-trade series reﬂects the fact that the price of slaves rose over the century, especially after 1760. In summary, though it is impossible to provide a deﬁnitive measure for the trend in the terms of trade during the colonial period, there was very little upward movement – at most a 30 percent increase over 50 years.39 Such modest improvements in the terms of trade allowed colonists to buy more producer and consumer goods, but this favorable effect would not alter much the picture of sluggish growth in output per capita because exports amounted to only 25 percent of GDP.40

THE STANDARD OF LIVING OF THE FREE POPULATION The preceding results indicate that the export trade had only a small impact on the standard of living in the Lower South for the population as a whole. While this calculation, which counts free and slave populations equally in the denominator of the per capita income calculation, is reasonable for assessing trends in regional productivity, it does not accurately reﬂect the distribution of regional production and hence well-being. As a ﬁrst approximation, then, it is worth considering changes in GDP per free person in the region. While this calculation raises the level of exports and GDP, it does not dramatically alter our conclusions about the relative lack of growth in per capita terms.41 The average value of exports per free person was substantially higher than that for the entire population – 80 percent higher in 1740 for example. Since the free population grew slightly slower than the total population during the period 1720–1770, exports per free person rose slightly from just under

Exports and the Lower South, 1720–1770

21

$22–$23.42 Gross domestic product per free person increased only slightly faster during the colonial period than it did for the entire population.43 For the period 1720–1770, GDP per free person rose at 0.14 percent per year versus the negligible 0.03 percent for the entire population (see Tables 5 and 6).44 Although the growth of income per free colonist in the Lower South was still sluggish, the colonists of the Lower South were well-off. The average GDP per capita (valued in prices of 1840) amounted to $75 in 1720, rose noticeably to $85 in 1730, and then fell back to $80 by 1770.45 The free white colonists may also have beneﬁted from bounties provided by England to encourage the importation of items that the British thought they sorely needed. These bounties were paid to encourage the import of selected items from the colonies. To some extent they were intended to offset higher costs of transporting items from the colonies than from the continent. The colonists beneﬁted from the increased demand for their product by English importers who received these subsidies, but they may also have beneﬁted from some portion of the bounty being passed on to the colonial producers. Because such a government subsidy is not included in the 1840 price used to value colonial output, the income ﬂowing to white colonists may not be fully captured in our measure.46 Two subsidized items were of some importance to colonists of the Lower South: naval stores and indigo. In the case of naval stores, the British needed an alternate source of supply when war in Scandinavia and actions by the Stockholm monopoly severely reduced imports of tar and pitch from the Baltic region. Beginning in 1705, the British authorized a bounty of d4 per ton on tar and pitch for imports from the colonies.47 The Act providing for these bounties lapsed in 1725, and ‘‘the bounties were discontinued during the four years following 1725,’’ but were restored with passage of a new Act in 1729 (Gray, 1958, p. 156; see also Williams, 1935, pp. 175, 184). The latter provided smaller bounties of d2 4s on tar and d1 on pitch.48 The bounty for indigo was also established in response to the impact of war. King George’s war (1739–1748) disrupted both the rice trade of the Lower South and the British importation of indigo from French colonies. During the War, the colonists experimented with indigo as an alternative export staple, and at the end of the War the British established a bounty to encourage the development of an alternative supply.49 Although the effectiveness of these bounties in calling forth additional production has not been fully determined, the colonists nevertheless may have received additional income that is not included in our estimate of the value of exports produced.50 And, in the case of naval stores, this was not an inconsiderable amount. In 1706, the initial year of payments, the bounty

22

PETER C. MANCALL ET AL.

Table 6.

Exports, Bounties, and Income per Colonist by Source.

Year

White Population

Exports to Foreign Destinations

Residual GDP (Excl. Exports)

Residual GDP Plus Exports

Residual Bounty Bounty on Indigo on Naval GDP Plus Exports and Stores Bounties

1720 1730 1740 1750 1760 1770 1780 1790 1800

24,795 34,000 62,781 82,384 119,628 190,216 305,873 521,859 738,961

21.81 32.48 29.05 21.34 22.17 22.96

53.58 52.77 50.44 52.03 55.05 57.85

75.39 85.26 79.49 73.37 77.23 80.81

4.72 0.88 0.83 0.82 0.45 0.52

12.87

59.29

72.16

Average annual rates of change By decade 1720–1730 3.21 1730–1740 6.32 1740–1750 2.75 1750–1760 3.80 1760–1770 4.75

4.06 1.11 3.04 0.38 0.35

0.15 0.45 0.31 0.57 0.50

1.24 0.70 0.80 0.51 0.45

By subperiod 1720–1740 1740–1770 1770–1800

4.75 3.76 4.63

1.44 0.78 1.91

0.30 0.46 0.08

Over the long term 1720–1770 4.16 1720–1800 4.33

0.10 0.66

0.15 0.13

0.03 0.39 0.18

80.11 86.14 80.32 74.22 78.07 81.51

72.16

15.41 0.66 0.12 5.76 1.31

0.73 0.70 0.79 0.51 0.43

0.26 0.06 0.38

8.33 1.57

0.01 0.05 0.41

0.14 0.05

4.33

0.03 0.13

27.86 7.19

Sources and Notes: See Tables 4 and 5, the text, and the appendix. All ﬁgures, except population, are expressed in US dollars at prices of 1840. Residual GDP was calculated by subtracting a maintenance allowance for slaves from the estimated value of GDP and also deducting the value of exports to foreign destinations. The maintenance allowance was equal to the value of the slave diet underlying the estimates of GDP, which increased over time from a value of $25.42 in 1720 to $29.11 in 1770, plus an allowance for housing, ﬁrewood, and all other items. The value of these three items varied slightly over time, and averaged $7.80 for the period 1720–1770. The bounties were estimated by multiplying the quantities exported of indigo, and of naval stores (tar, green tar, pitch, and turpentine) by the speciﬁed bounty per unit. The indigo bounty rate is from McCusker and Menard (1985, p. 187). The initial bounty of 6 pence per lb was reduced to 4 pence sometime in the 1750s or 1760s, but then restored to 6 pence in the early 1770s. Gray (1958, p. 293) noted that the bounty was reduced to 4 pence per lb in the early 1770s. The naval stores bounty rate is from Gray (1958, pp. 153–156). A ton was speciﬁed as being 8 bbl. Bounties were also provided for turpentine and rosin at d3 per ton, and for mast, yards, and bowsprits at d1, none of which were of much importance to the Lower South.

Exports and the Lower South, 1720–1770

23

amounted to only d554, but rose quickly to reach d10,135 in 1715, nearly tripled the following year, and peaked at d52,011 in 1718.51 On a per capita basis for the colonies as a whole, these do not amount to much, running between 3 and 5 pence per white person in all colonies, with a peak of around 30 pence per person in 1718. The export of naval stores, however, came disproportionately from the Lower South. From 1768 to 1772, years for which there exist export data for all colonies, the Lower South accounted for 68 percent of all naval stores exports. Given that this was a region with a relatively small white population, the bounty per person would have been a more noticeable amount, reaching as high as d1 Sterling (or $4.44) in 1720 (see Table 5). If the bounties on naval stores and indigo were fully passed on to the colonists in all years, they would have given a bigger boost to the colonists’ income early in the period, especially between 1716 and 1725, than near its end. But the consequence of such a shift would have been to slow the growth of per capita income. With or without the bounties, the free colonists in the Lower South were quite well-off in the mid-eighteenth century. Indeed, they were as well-off as the average American resident was in the opening decades of the nineteenth century, when GDP per capita for the nation was between $66 and $79 up through 1830. Not until 1840, when the value of GDP per capita reached $91 did the average American surpass the level that had been achieved by these free colonists as early as 1730.52 Of course those nineteenth-century ﬁgures pertain to the entire population, including slaves. When the comparison is made to the free population alone in 1840, the free colonists do not appear quite as well-off. The most pertinent comparison is with the free population in the South Atlantic region. That region, which includes the District of Columbia and the states of Delaware, Maryland, Virginia, North Carolina, South Carolina Georgia, and Florida, is not exactly comparable to the Lower South, but is more similar than the entire nation. Average income per capita for free persons in that region in 1840 was $96, higher than that for the colonists, but not by much.53 That ﬁgure suggests that per capita income in the region may have increased by only around 12–20 percent over the preceding century.

CONCLUSIONS These new estimates of exports, output in the domestic sector, and GDP provide little support for the view that export success led to intensive economic growth. Any picture of colonial success that rests on the

24

PETER C. MANCALL ET AL.

performance of the export sector alone is likely to misrepresent the true course of change, at least for the Lower South. This should not be too surprising. Exports were a relatively small part of the economy. They have received a great deal of attention because they were a primary reason for the initial English colonization of North America, at the heart of political debates, and generated quantiﬁable statistics. Moreover, exports from the South grew quite rapidly in the aggregate, so that sector would appear to have been dynamic and capable of generating sustained growth. Population, however, was simply growing quicker. The emphasis that previous research put on South Carolina, and in particular the low country’s export and economic success, has fostered a misleading view of the performance of the entire region. The low country was only a portion of the region, and its population became less numerically signiﬁcant over the course of the eighteenth century. The economic activity taking place in Georgia, North Carolina, and in the rapidly growing backcountry of South Carolina played a large role in shaping the region’s economic performance. The other primary demographic shift in the region – the growing proportion of the colonial and enslaved population under age 10 – also has received inadequate attention. Each of these factors suggests that reliance on the adult population of the low country – the population most responsible for exports – provides an imperfect picture of the regional economy. The argument that the economy of the Lower South experienced growth only in aggregate output and population but not growth in per capita income echoes the argument put forth by Daniel Scott Smith, who recognized that the distinctive feature of the early American economy was its rapid extensive growth (Smith, 1980, p. 17). In their comprehensive overview of the early American economy, John McCusker and Russell Menard argued that Smith’s view did not ﬁt the available evidence and that his ‘‘Malthusian’’ model remained untested. Smith’s thesis could not work if, as they suggested, per capita income growth in the colonies was at least as rapid as that in England, or that GDP grew as rapidly as the most successful exports.54 But there is good reason to question whether these premises are in fact valid. The absence of growth in per capita income does not mean that the economy of the Lower South was stagnant or unsuccessful. Quite the contrary: over the course of the eighteenth century, the colonies of the Lower South experienced rapid population growth. The number of free persons and slaves grew at 4.2 percent per year between 1720 and 1800, and the labor force grew slightly slower (3.9 percent per year) as the share of the

Exports and the Lower South, 1720–1770

25

population comprised of women and children rose. Such high rates of extensive growth were, as Smith argued, ‘‘extraordinary by any standard’’ (Smith, 1980, p. 17). Moreover, the region did experience some productivity growth in the export-oriented part of agriculture, which helped make possible that growth of population and GDP.55 The effects of this productivity advance were offset, however, by demographic shifts closely linked to the region’s rapid pace of extensive growth. In particular, rapid population growth was associated with an increase in the percentage of the population under the age of 10, which reduced the share of the population engaged in production, and held in check growth of GDP per capita. Under these circumstances, simply maintaining a constant level of per capita income should be seen as a remarkable achievement. The great need for labor in the colonies encouraged both migration and the importation of slaves over the course of the eighteenth century. It is all too easy, perhaps, to overlook the fact that the colonial economy was able to absorb the additional labor without experiencing declines in productivity. This stands in contrast to the inability of so many economies to absorb labor in the period after World War II, and differs from the experience of much of Europe between 1500 and 1750 when economic advance could occur only when population growth was held in check (Allen, 2003, pp. 406– 407). The success of the Lower South’s economy is in part a tribute to the abundance of land that enabled colonists to fend off diminishing returns, at least in agriculture, and perhaps in part to the transfer of land from Natives to newcomers at relatively low costs – low at least for colonists, who proﬁted from the steep decline in Native populations during the colonial era. Nevertheless, the existence of land does not mean that colonists used it in a particularly efﬁcient way; as the environmental historian William Cronon put it in his study of early New England, ‘‘the people of plenty were a people of waste’’ (Cronon, 1983, p. 170). That the per capita ﬁgure did not decline in the face of the rapid increase in population and labor is a notable accomplishment. The ﬁndings here conﬁrm the view that there was prosperity in the southern mainland colonies, evident by extensive growth of GDP and population rather than by growth in GDP per capita. Colonists had more children because they felt they could afford to maintain larger families and because they believed that their children would eventually become productive workers.56 Further, some colonists imported slaves because they saw them as productive investments in a land-abundant environment. Others migrated to the region because the economy looked attractive to them. These responses are signs of a productive economy. Indeed, the fact

26

PETER C. MANCALL ET AL.

that the region’s economy could forge ahead in the wake of such population increases attests to its capabilities. Careful attention to the changing demographic composition and residential patterns of the region reveals that the traditional view needs to be modiﬁed. Rather than concentrate on exports, which involved an increasingly small percentage of the colonial population, economic success needs to be measured by considering the entire population, including the vast majority who were not directly involved in export-oriented occupations. The staples thesis can no longer stand alone as the primary way to explain this economy. If we are to take seriously the effort to understand and measure economic performance in the mainland, we need to replace explanations pivoting on exports with newer, more encompassing models that include a wide range of economic behavior. One way to launch this new conceptualization of the Anglo-American economy is to pay greater attention to the entire population, even those whose efforts were never recorded by merchants or port authorities. Seen from this new angle, slow growth of income per capita is no longer an anomaly. It is, instead, a reasonable measure of an economy that succeeded despite massive population growth, a demographic trend that has undermined all too many other societies. This assessment inverts the idea that the low-country rice and indigo exports drove the regional economy in the way that, for example, sugar drove the economies of Jamaica and Barbados in the eighteenth century. There was little growth of the non-productive population in the Caribbean because the islands remained unpopular destinations for Britons or other Europeans who wanted to raise families, and the brutal conditions on sugar plantations retarded substantial natural growth among the transplanted African population (Dunn, 1972, pp. 300–334). The low country on the mainland, by contrast, became home to a growing population of European colonists, free African-Americans, and slaves. Longer life expectancy than on the islands, combined with a more favorable setting for colonists seeking land, led to the creation of stable families and, as a result, substantial growth of the non-productive share of the population.

NOTES 1. It was rivaled in this regard only by the Chesapeake. In both regions, much of the wealth was in the form of slaves. If measured by non-human wealth the differences among regions are reduced substantially, although not eliminated (see Jones, 1980, p. 54; Perkins, 1980, p. 154).

Exports and the Lower South, 1720–1770

27

2. To Egnal, the key to economic growth lay in trade between the colonies and Britain (Coclanis, 1989, pp. 73–93; Chaplin, 1993, Chs 7 and 8; Egnal, 1998, Ch. 1). 3. See Egnal (1998, p. 4) on the continued usefulness of the staples thesis, along with a discussion of its limitations. According to Egnal, the key features of the staples thesis are that ‘‘the export of primary products was the engine of growth for the colonial economy y [and that] the nature of these exports shaped the pattern of regional development’’ (Egnal, 1998, p. 5). 4. Recent estimates suggest that per capita incomes may have been stagnant from 1720 to 1770, and did not change appreciably between 1770 and 1800 (Mancall, Rosenbloom, & Weiss, 2004). 5. While we have produced a new, more comprehensive export series for the region, the construction of which is described in the appendix to this chapter, evidence on imports into the region remains limited to those from England and Scotland. See Carter et al. (2006, chapter Eg). 6. Even without this shift in perspective, it would be surprising to ﬁnd that exports from this slave plantation-based economy led to rapid growth of output per capita across the region. As McCusker and Menard have explained, when the export staple is a plantation crop it ‘‘will have only a minor impact on local industry or the size of the market’’ (McCusker & Menard, 1985, p. 26). Douglass North (1962) made the same case for the South in the period 1790–1860, when the region came to depend on the export of cotton. See also Galenson and Menard (1980, pp. 13–15). 7. See, for example, Hakluyt (the Elder) (1585) as discussed in Mancall (1995, pp. 33–44). 8. Because indigo had a much higher value relative to weight than did rice, it could much more easily bear the higher costs resulting from wartime conditions than could a bulky commodity like rice. But the volume of exports did not take off until Britain began to offer a bounty for indigo in 1749 making the crop commercially attractive. Moreover, since it could be grown on lands not suited to rice cultivation, and its peak labor demands did not coincide with those of rice cultivation, planters could add indigo without substantially reducing their commitment to rice (Gray, 1958, p. 289). 9. During this export upsurge, planters expanded rice cultivation into Georgia and the Cape Fear region of North Carolina. Despite that expansion, South Carolina remained by far the largest producer (Nash, 1992, p. 692; Dethlof, 1982, p. 235). 10. Perhaps the most important effect was the reduction in the slave population occasioned by the war. The conﬂict interrupted the importation of slaves, and resulted in signiﬁcant losses to the existing slave population. Although data are imprecise, it is estimated that approximately 25,000 slaves died, ran away, or were carried off by the British during the war (Morgan, 1983, p. 111; Gray, 1958, p. 596). The devastation caused by the war is apparent in the low levels of exports in the immediate post-war period. Although exports rose rapidly in the second half of the 1780s, even at their post-war peak in 1793, Charleston’s exports were well below the level of the early 1770s (Gray, 1958, pp. 1020–1023). 11. Between 1753/1754 and 1768–1772, the export of Indian corn increased at only 1.25 percent per year, while the export of beef and pork declined at 0.8 percent per year. The 1753 and 1754 data are from Burke (1777, pp. 259–261); those for 1768–1772 are from Shepherd and Walton (1972, Appendix IV, Tables 2–6, pp. 211–227).

28

PETER C. MANCALL ET AL.

12. According to Clowse (1971, pp. 178–179, 222–223), beef and pork exports began to decline when rice initially rose in importance. 13. In the 1740s, almost all the region’s exports would have passed through Charleston. The 1746–1747 shares are from Coclanis (1989, p. 81); the 1768–1772 ﬁgures are from Shepherd and Walton (1972, Appendix IV, Tables 2–6, pp. 211–227). If deerskins were excluded the remaining three items would not proxy the movement of total exports as well. The share comprised by rice, indigo, and naval stores rose from 67.3 to 82.5 percent. 14. To construct an index of real aggregate export performance, it is necessary to combine data on physical quantities of individual exports based on a constant set of prices that abstracts from general changes in the price level. The details of the derivation of these estimates are provided in the appendix to this chapter. Nash constructed a similar index, but it was limited to exports from South Carolina, did not include deerskins, and used average prices calculated over varying periods in the eighteenth century. In comparison to our calculations Nash (1992) shows a somewhat higher rate of growth. 15. Although Peter Coclanis (1989, pp. 71–77) focused on the rise and decline of the low country, he made careful and important distinctions between the impact of trade on that subregion and on South Carolina as a whole. He thus brought to the fore the need to better understand the economy of the backcountry, but he did not pursue the implications of the wide variations in economic performance across subregions for economic growth in the Lower South as a whole. 16. Although North Carolina was more populous than South Carolina, it had far fewer slaves for most of the eighteenth century. Slaves constituted a majority of South Carolina’s population for most of the eighteenth century. That colony contained 85 percent of the region’s slave population in 1700 and although the numbers of slaves in Georgia and North Carolina grew more rapidly than those in South Carolina, at the end of the century it still accounted for 43 percent of the region’s slaves (U.S. Bureau of the Census, 1975, Series A-7, Z-1–19). 17. Over the next 20 years, the backcountry’s share of both free and slave population increased, the latter having risen to slightly more than 29,000 by 1790 (an increase of over 400 percent) and accounting for 27 percent of the state’s slave population (Klein, 1990, p. 253). Although this shift of the population set the stage for the ascendance of cotton at the end of the century, rice planters remained the dominant employer of slaves, and the production of exports was concentrated on the slave-based plantations. The rapid expansion of cotton cultivation beginning in the early 1790s accelerated this shift, so that by 1810 close to 44 percent of the state’s slaves were living in the backcountry. 18. In the low country, ‘‘the discarded parts of Carolina’s chief staple, rice, served as cheap food’’ (Morgan, 1998, p. 35; Hughes, 1985, Ch. 6). 19. As discussed below, the picture looks slightly different when exports are compared to the free population alone. 20. The growth of exports was slower after 1770, which resulted in a poorer performance over the longer period from 1720 to 1800 when total exports per capita declined at 0.6 percent per year. Even rice exports per capita declined over the longer period, due to a fall in production after 1770.

Exports and the Lower South, 1720–1770

29

21. Although the shift in the locus of production from upland areas to the low country and the adoption of tidal irrigation suggest productivity gains might have occurred in rice farming, we estimate that long-term productivity improvements in the region’s entire farm sector – free and slave – were modest at best. Although the productivity of slaves may have increased, perhaps as fast as 0.9 percent per year between 1720 and 1770, the rate of advance for the entire farm sector fell in a range between 0.2 and +0.3 percent per year (Mancall, Rosenbloom, & Weiss, 2002a, pp. 390–424, Table 2, 2006, Table 1). 22. For comparisons of slave fertility and mortality in British North America with that of other western slave societies see Walsh (2000, p. 198–209), and Engerman (2000, p. 505–510). In almost every decade of the eighteenth century slave imports into the Lower South exceeded the increase in the stock of the slave population (see Mancall et al., 2001, Table 2). 23. The slave population did grow more slowly than rice exports alone which could imply that there was some productivity advance in rice production. (Nash, 1992, Table 6, p. 689; Mancall et al., 2002a, Table 2). On the other hand, it may simply reﬂect the re-allocation of slave labor into rice cultivation and out of other, less proﬁtable, activities. 24. After 1770, all the changes in the composition of the population, especially the decline in the slave share of the population, worked to reduce the labor force participation rate from 0.47 in 1774 to 0.39 in 1800. With no increase in labor productivity, GDP per capita would have declined by 17 percent in the last quarter of the century. 25. McCusker (2000, pp. 155–162) put the share at 25–30 percent for the colonies as a whole. See Table 5 for our estimates of GDP for the Lower South and a comparison to foreign exports. 26. Production for the domestic market helped to buoy up the growth of GDP after 1770, but even still, the GDP per capita ﬁgure 1800 was only slightly higher than that for 1720 and below that for 1730. 27. And exports per capita declined even faster thereafter. By 1800, the per capita value of exports was only 60 percent the value achieved in 1720. 28. This was a point stressed by North (1961, 1962) in his staple-export model, and was a key part of Egnal’s (1975, pp. 199–214) estimate that exports stimulated colonial economic growth to a rate of 0.5 percent per year. 29. Coclanis (1989, pp. 107–108) shows a more continuous times series of both rice and indigo prices, and presents a comparison of the index of rice prices and an index of English wholesale commodity prices, which shows that over the period 1722–1775, the price of rice rose relative to the price of English goods, and likewise for indigo prices over the period 1747–1775. He did not present the average increase over the periods, nor did he combine the two export prices to get a combined export price. 30. The indexes are available in U.S. Bureau of the Census (1975, Series E92–E95), with further details available in Taylor (1932, pp. 356–377, 848–876). 31. McCusker (1992, pp. 334–343) shifted the original observations to a calendar year basis, and shifted the base year from 1700/1701 to 1700–1702. According to the evidence for 1746–1775 shown in Egnal (1998, Appendixes A and B), prices that

30

PETER C. MANCALL ET AL.

English exporters charged Philadelphia merchants for speciﬁc textile goods, such as linen checks and buckrams, as well as prices that Philadelphia merchants charged shopkeepers for those same goods, changed in line with this CPI index. And, as Smith (1995, Table 7) has shown, the decadal changes in this modiﬁed CPI index were very similar to the changes evident in the ‘‘specially calculated export price index’’ he produced, which was made up entirely of manufactured exports. 32. McCusker took a simple average of the wholesale price indexes for producer and consumer goods. In McCusker’s version of the index, producer goods made up 50 percent of the index. The consumer goods category, which made up the other 50 percent, included ﬁve groups of products: cereals, animal products, beverages and condiments, candles and coal, and clothing. The cereal category included ﬂour, a manufactured item (Schumpeter, 1938, p. 32). 33. The exact composition of imports is unkown. Most imports from Britain were manufactured goods, but less is known about imports from elsewhere. According to Shepherd and Walton (1972), some of the more important imports from Southern Europe and the West Indies were Muscovado sugar, West Indian rum, molasses, and salt, but they could identify only 19–33 percent of the Lower South’s imports from these two regions in the years from 1768 to 1772. 34. There is so much variation in the series that no regression estimates of the trend are signiﬁcant. The 1775 index of 1.09 was roughly 30 percent above the 1720 index, but well below the peak of 1.65 found for 1738. 35. Even though it is not conventional to include the cost of labor in the terms of trade, because colonists viewed slaves as a form of agricultural capital and thus an imported item it seems appropriate to include them in these calculations. 36. In the 1740s, when the rice market was severely depressed and there was a prohibitive duty of slave imports up through 1744, the ﬁgure amounted to only 2 percent (Mancall et al., 2001, Tables 1 and 2). These shares are based on the values of slaves and exports expressed in 1840 prices. 37. The weights given to each are based on the import data for 1768–1772, which indicate that slave imports comprised 27.7 percent of all imports into the Lower South (U.S. Bureau of the Census, 1975, Series Z: 287, 290). An alternative way to treat the impact of slave imports would be to calculate the effect of the terms of trade on only the fraction of export revenues that were spent on merchandise imports. 38. There is so much variation in the series that no regression estimate of the trend is signiﬁcant. 39. The terms of trade would have improved even less if we had included prices of agricultural imports. Wholesale prices for New York and Philadelphia, which reﬂected primarily agricultural goods, rose faster than the price of Englishmanufactured commodities between 1720 and 1775 (Cole, 1938, pp. 124–125, 148–149, 153). 40. If we adjust GDP per capita for changes in the terms of trade over the period 1720–1770, the growth rate is raised from 0.03 to 0.17 percent per year when slave imports are ignored, but only to 0.07 when the terms of trade include slave prices. 41. The lack of growth in aggregate income per capita is nonetheless entirely consistent with the increasing wealth of low-country planters, and others engaged in the export trade. As the aggregate economy expanded this diminishing fraction of

Exports and the Lower South, 1720–1770

31

the population may well have enjoyed a rising standard of living, while the bulk of regional residents experienced little change in their living standard. 42. For the longer period 1720–1800, however, the free population grew faster than the total population with the consequence that exports per free person declined at an average annual rate of 0.7 percent per year. 43. The amount of GDP ﬂowing to the colonists was estimated by subtracting from total GDP an estimate of maintenance for the slave population. 44. For the longer period 1720–1800, however, the ﬁgure fell at 0.05 percent per year instead of growing at 0.04 percent when the entire population is taken into consideration. 45. The value declined further after the Revolutionary War, with the 1800 ﬁgure being below that for 1720. These values are somewhat below those presented by Perkins (1980, p. 154), which he derived by multiplying average wealth estimates by an assumed wealth/output ratio. 46. We do not know which portion of the bounty, if any, was passed on to the colonists, so our calculations are made to illustrate the consequences of these bounties in the event that all of it was passed on to the colonists. 47. A ton was speciﬁed as being 8 bbl. Bounties were also provided for turpentine and rosin at d3 per ton, and for mast, yards, and bowsprits at d1, none of which were of much importance (Gray, 1958, pp. 153–156; Williams, 1935, pp. 173–174). 48. For tar that met the speciﬁcations of the 1722 Act, the bounty would be d4. But there were complaints about the quality of tar from the colonies, and ‘‘London merchants petitioned the Board of Trade in 1769 to drop green tar from the bounty list, ‘there being little or no tar of such quality imported or used in the Manufactures of this Kingdom’’’ (Williams, 1935, pp. 184–185). 49. According to McCusker and Menard (1985, p. 187), the initial bounty of 6 pence per lb was reduced to 4 pence sometime in the 1750s or 1760s, but then restored to 6 pence in the early 1770s. Gray (1958, p. 293) reports that the bounty was reduced to 4 pence per lb in the early 1770s. 50. McCusker and Menard (1985, pp. 179–180, 187) are of the opinion that the bounties stimulated the production of naval stores up through the 1720s, whereas for indigo ‘‘the bounty’s impact has been exaggerated.’’ In the latter’s case, duties on foreign indigo likely had a more favorable effect. 51. Over the course of the 70 years of payments, the cost totaled d1,471,719 (Albion, 1926, Appendix B, p. 418). 52. These ﬁgures are all expressed in prices of 1840 (Weiss, 1992, Table 1.6). 53. The average per capita income for the entire free population in the United States in 1840 was $105, above that for the free colonists of the Lower South (Fogel & Engerman, 1971, p. 335). These ﬁgures were reported in prices of 1860, but they would be the same if expressed in 1840 prices because there was no change in the GDP price deﬂator between the two dates (Gallman, 1966, Table A-3). 54. McCusker and Menard (1985, pp. 32–34, 55) acknowledged that Smith’s approach would be consistent with a constant per capita income, but argued it fails because in their estimation per capita income had increased. 55. Slave productivity, as measured by the real price of slaves, rose at 0.63 percent per year between 1720 and 1770 (Mancall et al., 2002a, 2002b, pp. 390–424, Table 6).

32

PETER C. MANCALL ET AL.

56. With a rise in income, households can choose to have more goods and more children. If the prices of goods were relatively high, as they likely were in colonial America, households might have chosen to substitute children for durable goods. Further some might have chosen to increase the number of children rather than try to provide greater quality for a smaller number. See Easterlin (2004, Ch. 8), for a discussion of these issues. 57. We have used a terminal date of 1803 so that we could link to an established ﬁgure for domestically produced exports reported in Pitkin (1816 [reprinted 1967]). 58. These records are catalogued in the Public Record Ofﬁce as Customs 16/1. 59. The sterling values are from Shepherd and Walton (1972). The $4.44 ﬁgure used to convert to dollars was the ofﬁcial exchange rate (Davis & Hughes, 1960). 60. We did have to ﬁll in a few gaps in the data in order to avoid spurious ﬂuctuations. These are explained in the text. 61. The bounties were ﬁrst paid in 1706. 62. This is the same thing as assuming that the ratio of NC’s exports per bounty remained constant at the 1720 value. 63. Clowse (1981, pp. 57–58, Table B.21) did likewise in reporting exports for Charleston. 64. The series on naval stores exports terminates in 1712, so we have used that date as the terminal date for the extrapolated volume index series. We have extended the index to 1710 based on the export of only rice and deerskins. By excluding naval stores, or any substitute export, we think the growth in the volume index between 1710 and 1712 is overstated. 65. The series based on Nash’s index grew faster at times and over the full period 1720–1770, but his index pertains to only South Carolina, not the entire region. 66. This price was the average value of a lb of deerskin exports for 1769–1772 calculated from the data reported by Shepherd and Walton (1972, Appendix IV). 67. The 1701 ﬁgure covers only exports to England, not to all destinations, so probably understates the total volume of exports. As a result, rates of growth between 1701 and other years are probably upper bounds. 68. The increase was 37.1 percent using the sterling prices of 1768–1772 and 37 percent using the prices of 1791–1792 (Shepherd & Walton, 1976, p. 412, footnote 24). 69. If we use the average export value for 1790–1792 the increase from the average value for 1768–1772 was 20 percent. Using the average values the increase was 11 percent. 70. He did point out that the increase in exports was far below the rapid increase in population, and he was commenting on the colonies as a whole, not the Lower South speciﬁcally. 71. There is some question as to whether the value of re-exports was $39 or $49 million in 1800, and thus whether the value of domestically produced exports, which was obtained by subtraction of the re-exports from the total value of exports, was $31.7 or $21.7. North (1961, p. 221) reported re-exports as the larger ﬁgure, whereas the ﬁgure is reported as $39 million in U.S. Bureau of the Census (1975, Series U-192). Irwin (2003, Table 1, p. 509) argues that the lower value of re-exports is the correct one as the Treasury Report of 1837 shows the $39 million ﬁgure.

Exports and the Lower South, 1720–1770

33

72. In all these estimates we have assumed that all the domestically produced exports from the Lower South were agricultural products. 73. A minor problem is that the index for the period before 1791 includes corn. 74. For the period 1796–1803, changes in the price of the two commodities were substantially different; the price of rice rose by about 26 percent while the price of cotton fell by 52 percent.

ACKNOWLEDGMENTS The research reported here is part of the NBER’s program on the Development of the American Economy. Any opinions expressed are those of the authors and not those of the National Bureau of Economic Research. This research was funded in part by the National Science Foundation (SBR9808516). An earlier version of this paper was presented at the Social Science History Association annual meeting, Alexander Humboldt University, Tubingen University, UC Davis, UCLA, Williams College, and the American Origins Seminar of the USC-Huntington Early Modern Studies Institute. We thank participants at those venues, as well as Lou Cain, Marc Egnal, Stan Engerman, and Carole Shammas for their comments. We also thank the Center for Economic Studies at the University of Munich for providing Tom Weiss uninterrupted time to work on the paper. We are also grateful for the research assistance provided by Jasonne Grabher O’Brien and Maril Hazlet.

REFERENCES Albion, R. G. (1926). Forest and sea power. Cambridge, MA: Harvard University Press. Allen, R. (2003). Progress and poverty in early modern Europe. Economic History Review, LVI, 403–443. Bezanson, A., Gray, R., & Hussey, M. (1937). Wholesale prices in Philadelphia, 1784–1861. Philadelphia: University of Pennsylvania Press. Bjork, G. C. (1964). The weaning of the American economy: Independence, market changes, and economic development. The Journal of Economic History, 24, 541–560. Bruchey, S. (1967). Cotton and the growth of the American economy, 1790–1860. New York: Harcourt Brace & World. Burke, E. (1777). An account of the European settlements in America (6th edn). London: J. Dodsley. Carter, S. B., Gartner, S. S., Haines, M. R., Olmstead, A. L., Sutch, R., & Wright, G. (Eds). (2006). Historical statistics of the United States: Earliest times to the present. New York: Cambridge University Press. Chaplin, J. (1993). An anxious pursuit, agricultural innovation and modernity in the Lower South, 1730–1815. Chapel Hill: University of North Carolina Press.

34

PETER C. MANCALL ET AL.

Clowse, C. (1971). Economic beginnings in colonial South Carolina. Columbia: University of South Carolina Press. Clowse, C. (1981). Measuring Charleston’s overseas commerce: Statistics from the Port’s Naval list. Washington, DC: University Press of America. Coclanis, P. (1989). Shadow of a dream: Economic life and death in the South Carolina low country, 1670–1920. New York: Oxford University Press. Cole, A. H. (1938). Wholesale commodity prices in the United States, 1700–1861 (Vol. 1). Cambridge, MA: Cambridge University Press. Cronon, W. (1983). Changes in the land: Indians, colonists, and the ecology of New England. New York: Hill and Wang. David, P. (1967). The growth of real product in the United States before 1840: New evidence, controlled conjectures. Journal of Economic History, 27, 151–197. Davis, L., & Hughes, J. (1960). A dollar sterling exchange, 1803–1895. Economic History Review, 13, 52–78. Dethlof, H. C. (1982). The colonial rice trade. Agricultural History, 56, 231–243. Dunn, R. S. (1972). Sugar and slaves; the rise of the planter class in the English West Indies, 1624–1713. Chapel Hill: University of North Carolina Press. Easterlin, R. (2004). The reluctant economist. New York: Cambridge University Press. Egnal, M. (1975). The economic development of the thirteen continental colonies, 1720 to 1775. William and Mary Quarterly, XXXII(3), 191–222. Egnal, M. (1998). New world economies: The growth of the thirteen colonies and early Canada. New York: Oxford University Press. Engerman, S. (2000). Population history of the Caribbean. In: M. Haines & R. Steckel (Eds), A population history of North America (pp. 505–510). Cambridge: Cambridge University Press. Fogel, R., & Engerman, S. (Eds). (1971). The reinterpretation of American economic history. New York: Harper and Row. Galenson, D., & Menard, R. R. (1980). Approaches to the analysis of economic growth in colonial British America. Historical Methods, 13, 3–18. Gallman, R. (1966). Gross national product in the United States, 1834–1909, In: S. B. Dorothy (Ed.), Studies in income and wealth (Vol. 30, pp. 3–76). New York: National Bureau of Economic Research. Gray, L. C. (1958). History of agriculture in the Southern United States to 1860. Gloucester, MA: Peter Smith. Greene, E. B., & Harrington, V. D. (1932). American population before the federal census of 1790. New York: Columbia University Press. Grubb, F. (2004). The circulating medium of exchange in colonial Pennsylvania, 1729–1775. Explorations in Economic History, 41, 353. Hakluyt (the Elder), R. (1585). Inducements to the liking of the voyage intended towards Virginia in 40. and 42. degrees. In: P. C. Mancall (Ed.), Envisioning America: English plans for the colonization of North America (pp. 33–44). Boston: Bedford Books of St. Martin Press. Holmes, G. K. (1912). Rice crop of the United States, 1712–1911. U.S. Department of Agriculture, Bureau of Statistics, Circular no. 34. Washington, DC: U.S. Government Printing Ofﬁce. Hughes, K. (1985). Populating the back country. Ph.D. dissertation, Florida State University. Irwin, D. (2003). New estimates of the average tariff of the United States, 1790–1820. Journal of Economic History, 63, 506–513.

Exports and the Lower South, 1720–1770

35

Johnson, G. (1997). The frontier in the colonial south. Westport, CT: Greenwood Press. Jones, A. H. (1980). Wealth of a nation to be: The American colonies on the eve of the Revolution. New York: Columbia University Press. Klein, R. N. (1990). Unification of a slave state. Chapel Hill: University of North Carolina Press. Klingberg, F. J. (1939). The Indian frontier in South Carolina as seen by the S.P.G. missionary. Journal of Southern History, 5, 479–500. Klingberg, F. J. (1941). An appraisal of the Negro in colonial South Carolina; a study in Americanization. Washington. Mancall, P. C. (Ed.) (1995). Envisioning America: English plans for the colonization of North America, 1580–1640. Boston: Bedford Books. Mancall, P. C., Rosenbloom, J., & Weiss, T. (2001). Slave prices and the economy of South Carolina, 1722–1809. Journal of Economic History, 61, 616–639. Mancall, P. C., Rosenbloom, J., & Weiss, T. (2002a). Agricultural labor productivity in the Lower South, 1720–1800. Explorations in Economic History, 39, 390–424. Mancall, P. C., Rosenbloom, J., & Weiss, T. (2002b). Estimates of deerskin exports from the Lower South. Photocopy, Lawrence, KS: University of Kansas. Mancall, P. C., Rosenbloom, J., & Weiss, T. (2004). Conjectural estimates of economic growth in the Lower South, 1720 to 1800. In: T. W. Guinnane, W. A. Sundstrom & W. C. Whatley (Eds), History matters: Essays on economic growth, technology, and demographic change (pp. 387–424). Stanford: Stanford University Press. Mancall, P. C., Rosenbloom, J., & Weiss, T. (2006). Slave prices, the African slave trade, and productivity in eighteenth-century South Carolina: A reply. Journal of Economic History, 66, 1066–1071. McCusker, J. (1992). How much is that in real money. In: Proceedings of the American Antiquarian Society (Vol. 101, pt 2, pp. 334–343). McCusker, J. (2000). Estimating early American gross domestic product. Historical Methods, 33, 155–162. McCusker, J., & Menard, R. (1985). The economy of British America, 1607–1789. Chapel Hill: University of North Carolina. Menard, R. R. (1995). Slave demography in the lowcountry, 1670–1740: From frontier society to plantation regime. South Carolina Historical Magazine, 96, 280–303. Merrens, H. R. (1964). Colonial North Carolina in the eighteenth century. Chapel Hill: University of North Carolina Press. Morgan, P. D. (1983). Black society in the lowcountry, 1760–1810. In: I. Berlin & R. Hoffman (Eds), Slavery and freedom in the age of the American Revolution. Charlottesville, VA: University of Virginia Press. Morgan, P. D. (1998). Slave counterpoint: Black culture in the eighteenth-century Chesapeak and lowcountry. Chapel Hill: University of North Carolina Press. Nash, R. C. (1992). South Carolina and the Atlantic economy in the late seventeenth and eighteenth centuries. Economic History Review, 45, 677–701. North, D. (1961). Early national income estimates for the U.S. Economic Development and Cultural Change, 9, 387–396. North, D. (1962). Economic growth of the U.S. 1790–1860. New York: Prentice Hall. Perkins, E. (1980). The economy of colonial America. New York: Columbia University Press. Pitkin, T. (1816). A statistical view of the commerce of the United States. New York: Augustus M. Kelley Publishers.

36

PETER C. MANCALL ET AL.

Rossiter, W. A. (Ed.) (1909). A century of population growth. Baltimore: Genealogical Publishing Company. Saunders, W. L. (1886). Colonial records of North Carolina (Vol. II). Raleigh: P.M. Hale. Schumpeter, E. B. (1938). English prices and public ﬁnance, 1660–1822. Review of Economic Statistics, 20, 21–37. Shepherd, J. (1969). Commodity exports from the British North American colonies to overseas areas, 1768–1772: Magnitude and patterns of trade. Paper No. 258. Purdue University, Krannert Graduate School of Industrial Organization. Shepherd, J., & Walton, G. (1972). Shipping, maritime trade, and the economic development of colonial North America. Cambridge: Cambridge University Press. Shepherd, J., & Walton, G. (1976). Economic change after the American Revolution: Pre- and post-war comparisons of maritime shipping and trade. Explorations in Economic History, 13, 397–422. Smith, D. S. (1980). A Malthusian-frontier interpretation of United States demographic history before c. 1815. In: W. Borah, J. Hardoy & G. A. Stelter (Eds), Urbanization in the Americas: The background in comparative perspective (pp. 15–24). Ottawa: History Division, University of Manitoba. Smith, S. D. (1995). Prices and the value of English exports in the eighteenth century: Evidence from the North American colonial trade. Economic History Review, 48, 575–590. Taylor, G. R. (1932). Wholesale commodity prices at Charleston, South Carolina, 1732–1791, and wholesale commodity prices at Charleston, South Carolina, 1796–1861. Journal of Economic and Business History, IV, 356–377 and 848–876. U.S. Bureau of the Census. (1975). Historical statistics of the United States. Washington, DC: U.S. Government Printing Ofﬁce. Walsh, L. S. (2000). African American colonial population. In: M. Haines & R. Steckel (Eds), A population history of North America (pp. 198–209). Cambridge: Cambridge University Press. Weiss, T. (1992). U.S. labor force estimates and economic growth, 1800–1860. In: R. Gallman & J. Wallis (Eds), American economic growth and standards of living before the Civil War (pp. 19–75). Chicago: University of Chicago Press. Weiss, T. (1994). Economic growth before 1860: Revised conjectures. In: T. Weiss & D. Schaefer (Eds), American economic development in historical perspective. Stanford: Stanford University Press. Williams, J. (1935). English mercantilism and Carolina naval stores, 1705–1776. Journal of Southern History, 1, 173–174.

Exports and the Lower South, 1720–1770

37

APPENDIX. ESTIMATES OF DOMESTICALLY PRODUCED EXPORTS FOR THE LOWER SOUTH, 1710–1803 Although the historical record for the colonial period is more complete as regards exports than other things, it nevertheless is not fully complete, nor is it complete enough for our purposes. A complete record would cover all exports from each colony or colonial region, not only one or two major staple exports, and cover exports to all destinations, not just Great Britain or England, which is readily available. And, it would distinguish exports produced in the region from re-exports of goods produced elsewhere. In order to estimate GDP for the region, we had to construct the series on domestically produced exports shipped to all destinations. The series we produced is shown in Tables A1 and A2. In Table A1 we present benchmark estimates of the real value of all domestically produced exports to all destinations, as well as the per capita values of those exports and the average rates of change. We have shown the real value of exports under two different dating schemes. We have shown the export ﬁgures by year of export, which is the more common or traditional dating scheme. We have also shown the value of exports by year of production, because we are ultimately interested in estimating the region’s production. Thus, we have shifted the estimated values of exports backwards one year in order to have a series that is more closely aligned with the time of production than with the time of export. Table A2 shows the annual values of these variables. The derivation of this series on the real value of all exports to all destinations is explained below and laid out in Tables A3–A12.

DERIVATION OF THE SERIES ON DOMESTICALLY PRODUCED EXPORTS We constructed the estimates in parts. First, we established a benchmark ﬁgure for 1770 based on the estimates of Shepherd and Walton (1972). We use the 1770 date as a reference point, but in fact the evidence is for the ﬁveyear period 1768–1772. We then extended that ﬁgure backward to 1710 and forward to 1803. The former extension was done using an index of the volume of the four most important exports. The latter extension was done in two parts: we ﬁrst linked the 1770 benchmark ﬁgure with data for 1790–1792 using the method set out by Shepherd and Walton (1976) and

38

Table A1.

Benchmark Estimates of the Real Value of Domestically Produced Exports from the Lower South, in 1840 Prices.

Attributed to the Year of Production

Benchmark values 1710 121,878 1720 501,486 1730 1,201,512 1740 1,976,590 1750 1,691,743 1760 3,086,058 1770 4,369,062 1780 na 1790 4,557,450 1800 9,576,713

Per capita value

Three-year average values

Per capita value

4.69 12.64 20.03 17.50 11.82 14.71 12.63 na 5.98 8.77

199,509 540,879 1,104,445 1,823,638 1,758,314 2,652,693 4,367,879 na 3,867,239 9,510,205

7.47 13.60 18.31 16.48 12.21 12.61 12.71 na 5.03 8.70

26,003 39,663 60,000 112,936 143,184 209,760 345,819 762,417 1,091,394

Attributed to the Year in which Exported

Annual value of exports

Per capita value

– 469,930 1,057,817 2,199,092 1,459,722 2,368,345 4,375,343 na 1,818,252 8,359,874

– 11.85 17.63 19.47 10.19 11.29 12.65 na 2.38 7.66

Three-year average values

Per capita value

446,687 1,040,001 1,964,338 1,491,341 2,619,366 4,260,925 na 3,187,851 8,148,786

11.24 17.14 17.59 10.37 12.41 12.38 na 4.08 7.44

PETER C. MANCALL ET AL.

Annual value of exports

Population of the Lower South

4.71 1.34 3.85 2.22 1.51 3.91

7.40 5.14 0.36 4.20 5.11 9.42

3.02 1.05 2.95 0.32 0.08 5.6

4.23 6.53 2.40 3.89 5.13 3.65

8.45 7.59 4.02 4.96 6.33 16.481

4.05 1.00 6.27 1.03 1.14 12.38

8.82 6.57 2.72 5.79 4.99 9.84

4.31 0.26 5.15 1.81 0.02 6.2

1.64 1.08

6.27 2.95 2.63

0.96 0.86 1.26

5.37 3.80 3.90

8.02 2.32 2.18

2.52 1.43

7.69 2.61 2.18

2.26 1.16 1.68

0.00 0.46

4.27 3.65

0.14 0.56

4.43 4.23

4.56 3.66

0.13 0.54

4.61 3.70

0.19 0.51

Notes: The derivation of the export series is explained in the accompanying text and tables. The population of the Lower South includes that in Georgia, North Carolina, South Carolina and Tennessee. The benchmark ﬁgures for the free and slave population for 1700–1780 are from U.S. Bureau of the Census (1975, Series Z: 1–19). The Black population includes both slaves and free Negroes. John McCusker and Russell Menard made slight revisions to the ﬁgures for 1710, 1720 and 1740 (Table 8.1, p. 173), but did not report the white and black populations separately for each colony. Their ﬁgures, as well as those shown in U.S. Bureau of the Census, differ slightly from the ﬁgures reported by Wood for some of these years. The 1790 ﬁgures are from the U.S. Census of 1800, as shown in Rossiter (1909). The ﬁgures were taken from an electronic ﬁle provided by Michael Haines. The 1800 white and slave population ﬁgures are those underlying Weiss’s (1992) labor force estimates, and came originally from the U.S. Census of 1800.

Exports and the Lower South, 1720–1770

Average annual rates of change 1720–1730 9.13 1730–1740 5.10 1740–1750 1.54 1750–1760 6.20 1760–1770 3.54 1790–1800 7.71 By subperiod 1720–1740 7.10 1740–1770 2.68 1770–1800 2.65 Over the long term 1720–1770 4.42 1720–1800 3.76

39

40

PETER C. MANCALL ET AL.

Table A2. Annual Estimates of the Real Value of Domestically Produced Exports from the Lower South (in 1840 Prices). Attributed to the Year of Production

1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750

Annual value of exports

Per capita value

Three-year average values

121,878 277,140 304,382 266,991 260,987 307,976 314,515 338,527 368,645 469,930 501,486 651,222 656,350 658,381 675,854 761,553 780,681 748,107 860,672 1,057,817 1,201,512 1,054,007 1,356,092 978,667 1,300,780 1,379,433 1,224,502 1,025,509 1,717,334 2,199,092 1,976,590 1,295,233 1,982,678 2,141,345 1,713,774 1,610,289 1,607,178 1,561,891 1,322,560 1,459,722 1,691,743

4.69 10.25 10.82 9.12 8.55 9.68 9.48 9.77 10.18 12.40 12.64 15.77 15.26 14.69 14.47 15.65 15.39 14.15 15.61 18.39 20.03 16.52 20.00 13.49 16.76 16.69 13.84 10.92 17.16 20.70 17.50 11.22 16.81 17.76 13.90 12.76 12.45 11.81 9.77 10.52 11.82

199,509 234,467 282,838 277,453 278,651 294,492 320,339 340,562 392,367 446,687 540,879 603,020 655,318 663,528 698,596 739,363 763,447 796,487 888,866 1,040,001 1,104,445 1,203,870 1,129,588 1,211,846 1,219,627 1,301,571 1,209,814 1,322,448 1,647,311 1,964,338 1,823,638 1,751,501 1,806,419 1,945,933 1,821,803 1,643,747 1,593,119 1,497,210 1,448,058 1,491,341 1,758,314

Per Capita Value

7.47 8.59 10.06 9.50 9.12 9.24 9.64 9.81 10.78 11.74 13.60 14.56 15.24 14.81 14.94 15.17 15.06 15.05 16.05 18.01 18.31 18.85 16.67 16.75 15.65 15.76 13.82 13.97 16.26 18.45 16.48 15.18 15.27 16.16 14.81 13.04 12.34 11.34 10.70 10.70 12.21

Population of the Lower South

26,003 27,037 28,130 29,286 30,512 31,812 33,193 34,661 36,223 37,887 39,663 41,304 43,021 44,818 46,698 48,667 50,728 52,887 55,148 57,517 60,000 63,788 67,818 72,562 77,632 82,632 88,453 93,949 100,099 106,233 112,936 115,389 117,935 120,576 123,319 126,167 129,126 132,201 135,399 138,724 143,184

Annual Value by Year of Export

121,878 277,140 304,382 266,991 260,987 307,976 314,515 338,527 368,645 469,930 501,486 651,222 656,350 658,381 675,854 761,553 780,681 748,107 860,672 1,057,817 1,201,512 1,054,007 1,356,092 978,667 1,300,780 1,379,433 1,224,502 1,025,509 1,717,334 2,199,092 1,976,590 1,295,233 1,982,678 2,141,345 1,713,774 1,610,289 1,607,178 1,561,891 1,322,560 1,459,722

Exports and the Lower South, 1720–1770

41

Table A2. (Continued ) Attributed to the Year of Production

Per Capita Value

Population of the Lower South

Annual Value by Year of Export

Annual value of exports

Per capita value

Three-year average values

1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776

2,123,479 1,271,213 2,517,878 3,149,450 2,495,034 2,733,848 2,612,939 2,403,697 2,368,345 3,086,058 2,503,678 3,487,769 3,374,881 3,563,429 3,122,350 3,511,703 3,908,526 4,038,370 4,375,343 4,369,062 4,359,232 4,348,323 4,102,437

14.28 8.23 15.71 18.91 14.42 15.21 13.99 12.38 11.74 14.71 11.37 15.09 13.91 13.98 11.66 12.47 13.20 12.97 13.36 12.63 12.14 11.66 10.58 – – –

1,695,478 1,970,857 2,312,847 2,720,787 2,792,777 2,613,940 2,583,495 2,461,660 2,619,366 2,652,693 3,025,835 3,122,109 3,475,360 3,353,554 3,399,161 3,514,193 3,819,533 4,107,413 4,260,925 4,367,879 4,358,872 4,269,997 4,225,380

11.44 12.74 14.28 16.35 16.18 14.54 13.86 12.70 12.95 12.61 13.73 13.46 14.33 13.18 12.70 12.44 12.88 13.18 12.99 12.71 12.14 11.46 11.12

148,665 154,374 160,322 166,521 172,984 179,724 186,756 194,095 201,757 209,760 220,140 231,102 242,684 254,925 267,868 281,558 296,044 311,378 327,616 345,819 359,064 372,970 387,589 402,979 419,211 436,360

1,691,743 2,123,479 1,271,213 2,517,878 3,149,450 2,495,034 2,733,848 2,612,939 2,403,697 2,368,345 3,086,058 2,503,678 3,487,769 3,374,881 3,563,429 3,122,350 3,511,703 3,908,526 4,038,370 4,375,343 4,369,062 4,359,232 4,348,323 4,102,437

1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802

1,818,252 4,557,450 5,226,014 5,132,654 5,006,295 5,697,105 6,003,413 4,914,175 6,400,662 6,509,771 8,359,874 9,576,713 10,594,028 14,266,148

5.98 6.63 6.30 5.94 6.52 6.63 5.24 6.58 6.44 7.96 8.77 9.35 12.13

3,867,239 4,972,039 5,121,654 5,278,685 5,568,938 5,538,231 5,772,750 5,941,536 7,090,102 8,148,786 9,510,205 11,478,963 12,430,088

5.03 6.30 6.29 6.25 6.36 6.13 6.15 6.09 6.99 7.73 8.70 10.09 10.74

762,417 788,101 815,087 843,471 873,357 904,859 938,100 973,218 1,010,360 1,049,693 1,091,394 1,132,867 1,175,916

1,818,252 4,557,450 5,226,014 5,132,654 5,006,295 5,697,105 6,003,413 4,914,175 6,400,662 6,509,771 8,359,874 9,576,713 10,594,028

Note: See Table A1.

42

PETER C. MANCALL ET AL.

then extrapolated the 1790–1792 data to 1803 using a volume index of the eight most important exports.57

BENCHMARK FIGURES FOR EXPORTS OF 1768–1772 (PRODUCTION YEARS 1767–1771) The estimates for 1768–1772 were calculated from the evidence assembled by James Shepherd and Gary Walton from the ‘‘American InspectorGeneral’s Ledgers.’’58 Shepherd (1969, p. 9) describes this as ‘‘The only complete source for commodity trade for any years in the colonial period.’’ Shepherd and Walton presented the total value of exports in pounds sterling from each colony of the Lower South, as well as the quantities and the sterling values of many selected commodities that were exported from each of those colonies in each year 1768–1772 (1972, Appendix IV, Tables 2–6). We wanted a series valued in constant prices in order to gauge the growth of real output produced for export. And, we wanted the series valued in prices of 1840 so that it could be combined with other estimates of real output for the region and then linked to estimates of real GDP for the nineteenth century. We used 1840 prices to value the quantities of the eight most important exports reported by Shepherd and Walton in each of the ﬁve years, and inﬂated that ﬁgure to obtain the value of all exports in 1840 prices. The eight items and their 1840 prices are shown in Table A3. These eight items comprised an average of 85 percent of the current price value of all exports from the region for the period 1768–1772, with the percentage having remained fairly constant over the period (see Table A4, column 3). The value of the eight items in 1840 prices (Table A4, column 4) was divided by those current price shares to obtain the value of all exports from the region in 1840 prices (Table A4, column 5). We have also shown in Table A4 alternative estimates of the exports valued in 1840 prices in each of these ﬁve years. We derived these by converting the current price values from sterling to dollars at the exchange rate of $4.44, and then deﬂating to obtain the values in 1840 prices.59 In the ﬁrst instance we deﬂated by the David–Solar–McCusker Price Index, which on a base of 1840 had a three-year average value of 92.6 in 1770. In the second case, we used Cole’s data to construct an export price index for the Lower South, which on a base of 1840 had a three-year average value of 161.7 in 1770. As can be seen in Table A4, these alternative values differ noticeably from the ﬁgures we calculated using 1840 prices, and differ noticeably from each other,

Exports and the Lower South, 1720–1770

Table A3. Export

Rice (per lb) Indigo (per lb) Naval stores (per bbl) Deerskins (per lb) Cotton (per lb) Boards (per M feet) Staves (per M) Tobacco (per cwt)

43

Prices of the Major Export Items of the Lower South at Selected Dates.

1767–1771 Average Price

1791

1791

1800

1840

Average Price for 1839–1841

Pounds sterling

US dollars

Implicit Export Prices

Prices from A.H. Cole

0.004 0.211

$0.02 $0.94

$0.027 $1.15

$0.03 $1.05

$0.036 $0.580

$0.033 $1.530

$0.037 $1.443

0.453

$2.01

$2.00

$1.92

$2.79

$2.12

$2.12

0.098

$0.44

$0.50

$0.50

$0.41

$0.47

$0.46

0.043

$0.19

$0.25

$0.44

$0.27

$0.08

$0.10

2.928

$13.00

$6.00

$13.32

$11.14

$29.17

$29.42

3.240

$14.39

$12.72

$14.24

$22.76

$21.09

$21.03

0.913

$4.06

$3.67

$3.47

$9.50

$9.58

$4.729

Notes: 1767–1771: Shepherd and Walton (1972, Appendix IV, Tables 2–7). The Shepherd–Walton ﬁgures pertain to 1768–1772. We have shifted the dating back a year to better represent the year in which the exports were produced. 1791 implicit export price: (New American State Papers, 1884, pp. 200–202; U.S. Congress, 1884, Table 2). 1791 and 1800: Cole (1938, Statistical Supplement). The prices for rice, indigo, and cotton are three-year averages. 1840: All prices except those for indigo and deerskins are from Cole (1938, Statistical Supplement). The prices of rice, naval stores, boards, staves and headings, and tobacco are the average of the monthly prices reported for the year 1840. The cotton price is the average for 1839–1841, which was higher than the price for 1840. For indigo, we used the average value per pound for indigo exports in 1840 (U.S. Congress, 1884, p. 37) rather than the price reported by Cole which referred to the price of Venezuelan indigo in New Orleans. The price of deerskins for 1840 was derived from data in Bezanson, Gray, and Hussey (1937). We extrapolated the 1791 implicit export price to other years based on the change in their index of prices in Philadelphia.

Current Price Value of Exports in Pounds Sterling Value of eight major exports

Total value of exports

Eight exports’ share of total exports (%)

Estimates Based on Prices of 1840 Value of eight major exports ($S)

Total value of exports ($S)

Current Price Value of Exports in Dollars ($S)

44

Alternative Estimates of Exports in Constant Prices by Year of Production, 1767–1771.

Table A4.

Estimates Obtained by Deﬂation

Deﬂated by David–Solar– McCusker Price Index (three-year average values)

Deﬂated by Cole’s price index (three-year average values)

Index

$ value

Index

$ value

1767 1768 1769 1770 1771

457,563 465,250 450,920 511,770 688,812

538,000 551,000 534,000 593,000 800,000

85.0 84.4 84.4 86.3 86.1

3,324,158 3,409,894 3,694,625 3,770,582 3,753,364

3,908,526 4,038,370 4,375,343 4,369,062 4,359,232

2,388,720 2,446,440 2,370,960 2,632,920 3,552,000

89.10 90.71 92.63 97.76 98.08

2,680,866 2,697,135 2,559,652 2,693,348 3,621,647

159.1 158.6 161.7 179.2 191.4

1,501,494 1,542,918 1,465,898 1,468,928 1,855,433

Average: 1767–1771 Percentage increase 1767–1771

514,863

603,200

85.4

$ 3,590,525

4,210,107

2,678,208

93.7

2,850,530

170.0

1,566,934

51

49

1

13

12

49

10

35

20

24

PETER C. MANCALL ET AL.

Notes and source: Columns 1 and 2: Shepherd and Walton (1972, Appendix IV, Tables 2–6); U.S. Bureau of the Census (1975, Series E: 92, 95). Column 3 equals column 1 divided by column 2. Column 4 equals the quantities of the eight most important regional exports: rice, indigo, naval stores, deerskins, cotton, boards, staves and headings, and tobacco valued in prices of 1840. The quantities are from Shepherd and Walton (1972, Appendix IV, Tables 2–6); the prices are shown in Table A3 above. Column 5: The total value of exports in 1840 prices equals those in column 4 divided by the shares shown in column 3. Column 6: The current price value of all exports in dollars equals the pound sterling values in column 2 converted at the ofﬁcial exchange rate of $4.44 established by the revenue act of July 31, 1789 (Davis & Hughes, 1960, p. 54). Column 7: This is a composite index constructed by McCusker (2000, Table 1). McCusker extended the Brady–David–Solar Consumer Price Index backward to 1720 based on Bezanson’s index of Philadelphia commodity prices. Whereas the original index used 1860 as the base year, we have shifted the index to a base year of 1840. We have used three-year average values of the index for deﬂation. Column 9: This is an index of Charleston’s export prices, prepared by George Rogers Taylor (U.S. Bureau of the Census, 1975, Series E: 92–95). We have used three-year average values of the index for deﬂation. The ﬁgures in columns 8 and 10 were obtained by deﬂating the ﬁgures in column 6 by the price indices in columns 7 and 9, respectively.

Exports and the Lower South, 1720–1770

45

and demonstrate the difﬁculties of attempting to put a precise value on real exports. It is worth pointing out that the estimated growth in our export series before or after this period will not be affected by our choice of benchmark ﬁgures, even though the benchmark ﬁgures differ widely across these series. In our time series estimates, growth before and after 1768–1772 is based on the behavior of volume indexes constructed independently of the benchmark ﬁgures.

EXPORTS BEFORE AND AFTER THE BENCHMARK PERIOD We extended our benchmark estimates backward in time to produce an annual series running back to 1710. We extended the series forward in time as well, but the annual series is conﬁned to the period 1790–1803. In both cases, we estimated the value of domestically produced exports by using a volume index of change in the region’s exports. A volume index is a real index of exports in which quantities of the individual exports are weighted by a constant set of prices and summed. The index measures the change in that weighted sum. Our index for 1790–1803 is based on the same eight major exports that were used to construct the benchmark ﬁgures for 1767–1771; our index for the period before 1767 is based on four of those major exports. The issue is how to weight these available data in order to best approximate the behavior of total exports.

EXPORTS IN THE PRE-1767 PERIOD In the colonial period, there were four major items exported internationally from the Lower South: rice, naval stores, deerskins, and indigo. The quantity exported of each of the ﬁrst three of these was available for most years back through 1712.60 Indigo was not produced until 1746, but an annual series could be compiled from then on. Export quantities for rice, naval stores, and indigo were taken from the U.S.Census Bureau (1975). The rice ﬁgures (Series Z: 481) purport to show total exports from ports in South Carolina and Georgia so provide the coverage we were seeking. The data for indigo (series Z: 433–434) cover both South Carolina and Georgia for 1753–1770 but only South Carolina for the

46

PETER C. MANCALL ET AL.

remaining years. We compiled a new series on deerskin exports, combining the evidence available from several sources for South Carolina and Georgia. Those estimates are described in a separate working paper (Mancall et al., 2002b). For naval stores, the U.S. Bureau of the Census (1975, Series Z: 500–503) reports exports from Charleston and so covers exports from South Carolina, one of the two major sources in the region. The data reported for some years cover only a few months, so in those instances we inﬂated each of those to full years assuming that the exports in each month were proportionate to the months for which data are available. Annual exports from the other major producing colony, North Carolina, are not available, but exports were reported at selected dates. Shepherd and Walton (1972) provide ﬁgures of exports to overseas destinations for 1768–1772, all of which went to Great Britain. Merrens (1964, Table 3) reports exports to all destinations for those same years, including other North American colonies. A comparison of the two ﬁgures indicates that for the ﬁve years, exports overseas accounted for 66 percent of the total. Exports of naval stores were also reported for 1720 and 1753 (Saunders, 1886, p. 396; Burke, 1777, p. 261). Given that the chief interest of the Burkes was on the trade between the colonies and England (see vol. 1, Preface), it is likely that all 84,012 bbl reported were shipped overseas. But in order not to minimize the growth of naval stores exports we assumed the total included exports to other colonies, and applied the same percentage (66 percent) as calculated for 1768–1772. We made the same assumption for the 1720 ﬁgures. We interpolated between 1720 and 1753, and also between 1753 and 1770, to obtain annual ﬁgures for other years. North Carolina produced a large portion of total naval stores exports (averaging 58 percent for 1768–1772) and thus received the bulk of the bounties, so we used changes in the value of bounties paid on American naval stores to interpolate changes in exports (Albion, 1926, see the appendix). Because North Carolina’s export of naval stores increased faster over time than did that of all colonies, we also used changes in the ratio of North Carolina’s exports to the bounties paid. The ratio could be calculated for the benchmark years of 1720, 1752, and 1768–1772 (ﬁve-year average), and then was assumed to increase at a constant rate between benchmark dates. The estimated exports of naval stores from North Carolina in the intervening years are thus the product of the interpolated ratio and the reported value of bounties paid to all colonies. This procedure yielded the result that North Carolina’s share of all bounties rose over time from 7 percent in the years 1720–1722 to around 60 percent in 1768–1772. We assumed that the share in years 1710–1720 remained constant at the 7 percent ﬁgure found for 1720–1722

Exports and the Lower South, 1720–1770

47

and calculated the amount of bounty paid to North Carolina.61 Dividing that amount by the bounty per barrel yielded the estimate of naval stores exported from the colony in each year.62 Most of the export data reﬂects exports made between the fall of one calendar year and the fall of the next year (e.g., November 1, 1734–October 31, 1735). This ‘‘export year’’ reﬂects sales of the crop harvested in the fall of the calendar year on which it begins. The U.S. Bureau of the Census (1975) generally attributes exports to the year from which most of the data come (in the example, exports would be listed as 1735; but correspond to the crop grown and harvested in 1734). Gray (1958) generally adopts a ‘‘crop year’’ dating scheme, so he would list the exports as being for 1734.63 The choice is in some sense arbitrary, but it needs to be consistent. We have adopted a ‘‘crop year’’ dating scheme, which means that we have changed the dates of all the data from U.S. Bureau of the Census to be one year earlier than in the source. In Table A5, we have presented benchmark ﬁgures on the quantities of these major exports that underlie our volume indexes, along with the rates of growth of each of those. As can be seen, rice exports grew quite rapidly from 1712 onward, and quite rapidly in almost every decade. Only in the 1740s did rice exports decline and only in one other decade, the 1750s, was the average rate of growth below 5 percent. Exports received a substantial boost in the 1750s by the growth of indigo production, but the other major items that were exported over the entire period did not fare so well. Over the longer term, naval stores rose on average at only 0.55 percent per year between 1712 and 1770, while deerskins increased at 1.3 percent per year over that same period. Although rice was the dominant export in value terms (see Table A6), naval stores and deerskin exports were not unimportant, especially so in the earlier decades of the century. Thus, the increase in the real value of exports over the period 1720–1770 was held in check by these slower growing items, and the average rate of growth for total exports in constant prices must have been less than the 4.65 percent shown by rice alone. We considered a number of alternative weights to use in order to construct a volume index for the years before 1767 based on these four exports. These weights, which are based on the composition of exports in selected years, are summarized in Table A6. We also show there the implicit shares of each export yielded by valuing the export quantities by the prices of 1840. The volume indexes that we have calculated based on various weighting schemes are summarized in Table A7. The ﬁrst series shown in Table A7 is that derived by weighting the quantities of each export by its price in 1840 (prices are shown in Table A3)

48

Table A5.

Quantities of Major Exports from the Lower South, 1712–1770.

Rice (lbs)

Deerskins (lbs)

Indigo (lbs)

9,506 33,505 30,564 33,148

179,350 120,721 229,244 219,575 352,526 285,387 242,874 375,553

– – – – 112,933 28,933 389,767 595,300

66,594 46,407 81,500

Average rates of change 1712–1747 1747–1770 1712–1770

6.17 4.94 5.68

4.34 2.92 3.77

1.95 0.28 1.28

1720–1740 1740–1770 1720–1770

7.61 2.73 4.65

0.05 3.04 1.79

3.04 1.81 2.30

1720–1730 1730–1740

9.03 6.20

0.91 0.82

6.62 0.43

7.49

PETER C. MANCALL ET AL.

Three-year average benchmark values 1712 3,168,625 1720 8,060,551 1730 19,131,450 1740 34,917,672 1747 25,797,547 1750 34,123,207 1760 47,080,950 1770 78,227,450

Naval Stores (bbls)

0.23 3.27 5.21

7.23 3.55 5.79

2.66 1.60 4.45

29.70 4.33

Notes and Sources: Naval Stores: The total is the sum of estimates of exports from South Carolina and from North Carolina. The South Carolina exports for 1712–1724 are from Clowse (1981, p. 65). We estimated the values for 1713–1716 by interpolating between the reported values for 1712 and 1717. For 1724–1777, data are from U.S Bureau of the Census (1975, series Z: 500, 501, 503). The reported data for some years covered only a few months. We have inﬂated each of these to full years assuming that the exports in each month were proportionate to the months for which data are available. The exports of naval stores from North Carolina were reported for only a few selected dates. Shepherd and Walton provide ﬁgures of exports to overseas destinations for 1768–1772, all of which went to Great Britain. Exports of naval stores were also available for 1720 (Saunders, 1886, Vol 2, p. 396) and for 1753 (Burke, 1777, Vol 2, p. 261). As described in the accompanying text, we interpolated between 1720 and 1752, and also between 1752 and 1770, to obtain annual ﬁgures for other years. We then shifted these export estimates backward a year to attribute them to the year of production, and combined them with the exports from Charleston (displaced to the year of production) to arrive at the total for the Lower South. Rice: 1710–1773, U.S. Census Bureau (1975, Series Z 481), which shows total exports from ports in South Carolina and Georgia; after 1782, export data are for Charleston only, and we inﬂated these using the ratio of South Carolina and Georgia to Charleston exports that prevailed from 1768 to 1773 (1.23) to inﬂate Charleston exports to their regional level. Data for 1782–1788 are from U.S. Census Bureau (1975, Series Z 483). The reported data for 1713–1774 and 1783–1789 were shifted back a year to represent the crop year (i.e., the year of production), a method espoused by Clowse (1981, pp. 57–58, Table B.21). Indigo: 1746–1787, U.S. Census Bureau (1975, Series Z: 433–434). For 1753–1770 data for Georgia are available. For the remaining years the data are for South Carolina alone. For 1789–1800 data are from Gray (1958, p. 1024). The reported data for all years were shifted back a year to reﬂect the crop year. Gray reports quantities in chests or casks, both of which were assumed to weigh 350 pounds. Deerskin: Exports were compiled by us (Mancall et al., 2002b).

Exports and the Lower South, 1720–1770

1740–1750 1750–1760 1760–1770

49

50

PETER C. MANCALL ET AL.

Table A6.

Export Shares at Various Dates and Valued in Different Prices. Rice (%)

Naval Stores (%)

Deerskins (%)

Indigo (%)

Cotton (%)

61.1

2.4

24.9

11.6

0.0

63.3

22.5

6.6

7.6

0.02

65.3 74.7

21.1 5.5

6.2 2.9

7.4 16.9

0.02 0.01

59.3

33.7

4.4

1.6

1.1

61.3 69.9

31.9 3.6

4.1 0.6

1.5 25.5

1.2 0.4

3.0

1.6

0.2

47.7

2.7

1.5

0.3

50.0

3.5

1.1

14.0

16.8

Shares implicit in the series based on 1840 prices 1712 50.2 9.6 1720 67.6 18.5 1730 78.6 8.2 1740 86.1 5.9 1747 69.8 7.4 1750 78.6 9.6 1760 71.6 4.6 1770 73.2 4.9

40.3 13.9 13.2 8.0 13.4 9.4 5.4 4.9

9.5 2.3 18.5 17.0

Shares in 1747–1748 Valued in SC currency Average shares 1767–1771 Valued in 1767–1771 prices In 1791 prices In 1840 prices Average share 1791–1792 Valued in 1767–1771 prices In 1791 prices In 1840 prices

Average share in 1800 valued in 1840 prices Three-year average: 47.5 1799–1801 Five-year average: 45.4 1798–1802 Average share, 1790–1800 In 1840 prices

Shares with cotton included 1770 1790 1800 Average share, 1712–1773

64.6

73.3 68.4 41.2

4.9 5.3 2.2

4.9 1.6 1.9

16.9 24.1 0.1

69.7

11.8

12.1

14.9

0.01 0.59 54.60

Notes and Sources: The 1747–1748 ﬁgures are from Coclanis (1989, p. 81). Those for 1767–1771 are from Shepherd and Walton (1972, Appendix IV, Tables 2–6). The 1791–1792 current price ﬁgures for the region equal the quantities (New American State Papers, pp. 163ff.) times prices (Table A3 above). The values in 1840 prices were calculated by weighting the quantities of each export by prices shown in Table A3 (see Tables A4 and A8 for the sources of the quantity data).

Alternative Estimates of Volume Indexes of Exports from the Lower South, 1712–1770.

Based on 1840 Prices

Nash’s Index for SC

Benchmark values: three-year averages 1712 0.065 1720 0.124 100.1 1730 0.253 177.6 1740 0.417 322.8 1747 0.343 1750 0.402 351.6 1760 0.607 659.0 1770 0.993 960.0

Extended Nash Index

0.089 0.145 0.282 0.441 0.391 0.395 0.660 1.052

Average rates of change between benchmark values 1712–1747 4.88 6.48 4.32 1747–1770 4.73 4.40 1712–1770 4.82 4.35 1720–1740 6.27 6.03 5.73 1740–1770 2.93 3.70 2.94 1720–1770 4.25 4.62 4.05 1720–1730 7.40 5.90 6.88 1730–1740 5.14 6.16 4.59 1740–1750 0.36 0.86 1.11 1750–1760 4.20 6.48 5.28 1760–1770 5.04 3.83 4.77

Based on 1767–1771 Shares in Current Price

Based on 1767–1771 Shares in 1840 Prices

With reweighting

With reweighting

Based on 1747– 1748 Shares

0.067 0.163 0.258 0.385 0.375 0.476 0.581 0.963

0.071 0.173 0.273 0.408 0.375 0.476 0.581 0.963

0.045 0.102 0.208 0.354 0.326 0.381 0.619 1.010

0.052 0.119 0.244 0.413 0.326 0.381 0.619 1.010

0.138 0.150 0.306 0.429 0.433 0.442 0.602 0.971

5.04 4.18 4.70 4.39 3.10 3.62 4.69 4.09 2.14 2.01 5.19

4.87 4.18 4.60 4.39 2.90 3.50 4.69 4.09 1.55 2.01 5.19

5.85 5.04 5.53 6.43 3.56 4.70 7.43 5.43 0.76 4.97 5.02

5.38 5.04 5.25 6.43 3.02 4.37 7.43 5.43 0.80 4.97 5.02

3.31 3.58 3.42 5.40 2.76 3.81 7.40 3.44 0.29 3.13 4.91

51

Notes and Sources: Nash’s index for SC from Nash (1992, p. 699). The rate of change shown for the Nash Index for 1712–47 is for the period 1710–50. See the text for a discussion of how the various indices were calculated. We have used 1712 as the terminal date for the extrapolated volume index series because the series on naval stores exports terminates in 1712. We could extend the index to 1710 based on the export of only rice and deerskins, but with the exclusion of naval stores or any substitute export, the growth in the volume index between 1710 and 1712 would be greatly overstated.

Exports and the Lower South, 1720–1770

Table A7.

52

PETER C. MANCALL ET AL.

and is the series we have used to extrapolate the benchmark ﬁgures back to 1712.64 As can be seen, this index yields the highest rates of growth for the region over the colonial period, as well as the highest in most decades and subperiods.65 This index shows a decline in real exports during the 1740s, a result which captures the impact of the decline in rice exports before the favorable effect of increased indigo exports takes hold. That decline or slowing down is captured as well in most of the other indexes. The rate of growth shown by our volume index between 1720 and 1770 is consistent with the fact that real exports in total must have grown more slowly than the quantity of rice exported. The latter rose at 4.6 percent per year between 1720 and 1770, our volume index implies that total exports grew at 4.25 percent per year over that same period. Indeed, the rates of growth implied by this index seem consistent with the rate of growth of rice exported in each subperiod. In all subperiods the total grew somewhat slower than rice exports with three exceptions: in the 1740s when total exports declined somewhat more rapidly than rice, in the 1750s, and between 1740 and 1770. In these last two cases the growth of total exports was pushed up by the increased production of indigo. The other series shown in Table A7 are also based on the exports of the same three or four major items used in our volume index, but with different weights given to each export in the various series. The ﬁrst two alternative indexes are based on valuing the major exports in terms of hundredweights (cwts) of rice. To construct such an index, the exports of the other products – deerskins, indigo, and naval stores – were converted to their equivalent in rice using a ﬁxed set of relative prices. Nash (1992, p. 699) calculated such an index for South Carolina based on three exports: rice, naval stores, and indigo. The series labeled Extended Nash Index was constructed by us to broaden Nash’s index to cover exports from the other colonies in the region and to include deerskin exports. Nash (1992, p. 699) calculated the relative prices of rice, naval stores, and indigo for the pre-1773 period. For rice he used the average price for 1733–1774, for indigo prices the average price for 1747–1774, and for naval stores he used a weighted index of prices of pitch, tar, and turpentine averaged for 1733–1774. Unfortunately price data for deerskins are very limited, and we used the average price of deerskin exports for the 1767–1771 period.66 Although we did not ﬁnd many prices of deerskins, those we did ﬁnd, with the exception of a price quoted in the correspondence of Henry Laurens, were higher than the price for 1769–1772 and thus would have given a greater weight to deerskins in the index. A greater weight may have been appropriate, but because deerskins were among the slower growing exports we did not want to assign too high a

Exports and the Lower South, 1720–1770

53

weight to them and thereby bias downward the growth of the volume index. Thus we chose the relatively lower average price for 1769–1772 to give a lower weight to deerskins in the index. The prices are Rice Indigo Naval stores Deerskins

=6s. 10d. per cwt=1 volume unit =3s 7d. per lb=0.52 volume unit =6s. 10.75d. per bbl=1.01 volume unit =1s. 11.5d. per lb=0.29 volume units

The Extended Nash Index is based on the quantities of four exports back to 1746, but only three exports from 1746 back to 1712. The index value for 1746 based on the four exports was extrapolated back to 1712 based on the change in the index for three items. The other alternative indexes are based on the relative importance of each of the four major exports in 1767–1771 and 1747–1748. The relative importance varied depending on whether the items were valued in current prices or in constant prices of 1840. Moreover, in each of the series based on the 1767–1771 share weights we produced two variations, one without reweighting before 1746, the other with reweighting in order to adjust for the fact that indigo was not produced before 1746. Another check on our series is to gauge the growth of real exports in the colonial period by deﬂating the value of exports in current prices. This, however, can only be done for a few subperiods because the current price data are available for only a few years. Moreover, the price data available are not ideally suited to the task of deﬂating exports. One of those, the price of rice covers only one export; the other, the David–Solar–McCusker Price Index is a more general consumer price index, not an index of producers prices. Nevertheless, these deﬂated values can give an approximation of the likely rate of increase in the real value of exports. As can be seen in Table A8, exports deﬂated by the David–Solar–McCusker index increased between 1712 and 1747 at an annual average rate of 5.7 percent, between 1747 and 1770 at an annual rate of 4.3, and for the longer period 1712–1770 the average rate of growth was 5.1 percent per year.67 Use of the other indices gives different results. For the period 1747–1770, when deﬂated by the price of rice, the real value of exports rose at only 3.2 percent per year, whereas deﬂated by the Cole index, the rate of growth was 3.7 percent per year. Although these estimates are imperfect and limited in scope, they do provide some perspective for assessing the growth measured by using export quantities. The rate of growth in our volume index corresponding to the years for which we have

Value of Exports in Current Prices (d Sterling)

Real Value of Exports Estimated by Deﬂating Current Price Values, Selected Dates. Value of Exports in Current Prices (US$s)

Price Deﬂators Price of rice (cents per lb)

David–Solar– McCusker index (1840=100)

Real Value of Exports Cole’s export price index

Deﬂated by price of rice

Three-year average Three-year average benchmark 1712 24,905 1747 161,365 1770 559,333 Average rates of change 1712–1747 5.48 1747–1770 5.55 1712–1770 5.51

54

Table A8.

values 110,578 716,461 2,483,440

5.48 5.55 5.51

1.04 1.73

2.26

74.0 69.9 92.6

0.16 1.23 0.39

Deﬂated by David–Solar– McCusker index

Based on Our Volume Index Deﬂated by (Three-Year Cole’s Average) index

Prices of 1790

107.1 161.7

1,693,918 3,512,900

1.81

3.22

157,499 1,080,328 2,826,723

5.66 4.27 5.10

Prices of 1840

762,619 1,750,848

3.68

282,838 1,497,210 4,367,879

4.88 4.77 4.83

PETER C. MANCALL ET AL.

Notes and Sources: The value of exports in dollars equals the sterling values converted at $4.44 per d. The value of exports in sterling for 1701 is the three-year average for 1700–1702 taken from the ‘‘ofﬁcial series’’ (U.S. Bureau of the Census (197), Series Z: 223). That series was valued in prices of 1700–1702, so the values for these particular years should thus be close to current price values. The value of exports in sterling for 1747 is from Coclanis (1989, p. 107). The prices of rice are from Cole (1938, p. 154) converted at the exchange rate of $4.44 per d. The composite index was constructed by McCusker (2000, Table A1). McCusker extended the Brady–David–Solar Consumer Price Index backward to 1720 based on Bezanson’s index of Philadelphia commodity prices. Whereas the original index used 1860 as the base year, we have shifted the index to a base year of 1840. We have used three-year average values of the index for deﬂation. Cole’s export price index is that for Charleston’s export prices, prepared by George Rogers Taylor (U.S. Bureau of the Census (1975), Series E: 92–95). We have used three-year average values of the index for deﬂation.

Exports and the Lower South, 1720–1770

55

been able to calculate deﬂated values are shown in the last column of the table. As can be seen our estimates show faster growth than all three of the deﬂated series in the period 1747–1770. For the longer time period, 1701–1770, our index shows somewhat slower growth than the series deﬂated by the David–Solar–McCusker Price Index, but the difference is not great.

EXPORTS IN THE POST-1770 PERIOD We extrapolated the 1767–1771 benchmark data forward in two stages. We ﬁrst linked the benchmark data to 1790–1792, and then extrapolated that ﬁgure forward to 1803 based on a volume index of change in the most important exports. There is not enough data available to construct an annual export series running from 1772 to 1791. Instead we have made an estimate for 1790, 1791, and 1792 following the method laid out by Shepherd and Walton (1976, pp. 397–422). They estimated the increase in real exports between 1768–1772 and 1791–1792 by valuing the quantities of the most important exports in constant prices. For the 13 colonies and states they calculated such an increase in two ways; using on the one hand, the prices of 1768–1772 in pounds sterling, and on the other, the average price for 1791–1792 in dollars. The choice of price weights did not matter. The increase in real exports for the selected commodities was 37 percent.68 Those selected commodities comprised 88 percent of the current price value of all exports in 1768–1772, and 86 percent in 1791–1792. Their estimates by region were not presented in comparable detail. They were calculated by deﬂating the reported current price values by the Paasche Price Index implicit in the calculations for the 13 colonies and states. For the Lower South they showed a growth of real exports between 1768–1772 and 1791–1792 of only 5.6 percent. Rather than accept this estimate, we made more detailed calculations of the sort they had made for the colonies as a whole. These calculations are shown in Table A9, with the dating shifted backward one year. We obtained somewhat different results from those of Shepherd and Walton, but not greatly so and the result depends on the choice of price weights. If we use the sterling prices of 1767–1771 we ﬁnd that real exports rose by 8 percent between 1767–1771 and 1790. If we use the prices of 1791 as the weights, the real value of exports declined by 1 percent between those dates.69 We have chosen to use the estimates based on the prices of 1767–1771 to link the colonial export series with the post-1790 period, and thus impute

56

Table A9.

Estimate of the Value of the Major Exports from the Lower South in Constant Prices in 1767–1771 and 1790–1792. Major Export Items

Rice

Indigo

Naval stores

Deerskins

Real exports valued in prices of 1768–1772 in pounds sterling 1767–1771 308,228 109,330 31,878 37,077 Average

Sum of Eight Items

Index of Real Exports (1768– 1772=100)

Value of Real Exports in Prices of 1840

Cotton

Boards

Staves and headings

Tobacco

87

17,140

8,733

10,621

523,114

100

4,210,107

239,593 269,454 280,972

116,008 157,936 175,524

16,787 21,344 16,854

4,563 13,696 11,355

8,202 5,899 20,541

39,104 37,592 46,981

10,722 11,892 11,542

130,625 130,680 73,301

566,273 649,343 637,743

108 124 122

4,557,450 5,226,014 5,132,654

1790–1792 Average

263,339

149,823

18,328

9,871

11,547

41,226

11,385

111,535

617,786

118

4,972,039

Real exports valued in prices of 1791 1767–1771 1,676,608 543,384 Average

164,607

190,312

502

77,975

38,373

35,519

2,721,069

100

4,210,107

1790 1791 1792

86,682 110,211 87,030

23,419 70,297 58,285

47,250 33,983 118,340

177,895 171,017 213,730

47,112 52,255 50,716

436,856 437,039 245,144

2,699,057 3,125,459 3,173,969

99 115 117

4,176,049 4,835,790 4,910,845

1,303,270 1,465,695 1,528,348

576,573 784,964 872,377

PETER C. MANCALL ET AL.

1790 1791 1792

1,432,438

744,638

94,641

Percentage increase between 1767–1771 and 1790 1767–1771 22 6 47 Prices 1791 Prices 22 6 47

50,667

66,524

187,547

50,028

373,013

88

9,311

128

23

1,130

8

8

8

88

9,311

128

23

1,130

1%

1

1

13,151

141

30

950

18

18

18

13,151

141

30

950

10

10

10

Percentage increase between 1767–1771 and 1790–1792 average 1768–1772 15 37 43 73 Prices 1791 Prices 15 37 43 73

2,999,495

110

4,640,895

Notes and Sources: The values of the individual exports are the products of the quantities (Tables A8 below and Shepherd and Walton, 1972, Appendix IV, Tables 2–7) times the prices of 1768–1772 or 1791 (see Table A3 above). The index of real exports is the index of the sum of the eight items with 1767–1771 set as the bas year. The value of exports in 1840 prices for 1767–1771 comes from Table A4. The value in other years equals that base year value times the index for each year. The sum of the eight items in pounds sterling for 1767–1771 shown in this table differs slightly from the ﬁgure shown in Table A4 because the latter is the average of the ﬁve annual totals.

Exports and the Lower South, 1720–1770

1790–1792 Average

57

58

PETER C. MANCALL ET AL.

some growth in the real value of exports. Although foreign trade was subject to much disruption after the Revolutionary War, Gordon Bjork (1964, p. 560) concluded from his examination of the period that ‘‘there was a modest increase in exports between the pre- and post- revolutionary period.’’70 He was commenting on the colonies as a whole, not the Lower South, but the calculations made by Shepherd and Walton indicated that some small increase in exports had taken place for the region. The index shown in Table A9 for the eight items was used to extrapolate to 1790, 1791, and 1792, the 1767–1771 value of all exports (expressed in 1840 prices). Those values are shown in the last column in Table A9.

ESTIMATES FOR 1790–1803 Although export statistics are available for the years after 1790, the reported values are in current prices and include re-exports. We do not know exactly how large re-exports were for any region, but for the nation they were quite large from 1793 to 1806, and varied in importance from year to year. And, over time, inﬂation contributed noticeably to the increase in the value of exports. Because our goal is to estimate real output in the Lower South, we wanted a series on only domestically produced agricultural exports in constant prices, so we needed to eliminate the inﬂuences of inﬂation and reexports. An indication of the extent to which these two phenomena inﬂuenced the value of exports from the Lower South, and the increase in the value over time, can be seen from a glance at the trade statistics shown in Table A10a. There we show the extent to which the increase in the nominal value of all exports from the region between 1790 and 1801 can be accounted for by the growth of the eight major exports known to have been produced in the region. The portion unaccounted for – labeled ‘‘share not identiﬁed’ – is an approximation of the value of re-exports of items not produced within the region. As can be seen, that share varied widely over the decade, and amounted to as much as 53 percent of the total value of the region’s exports in 1796. Between 1790 and 1800, the nominal value of exports from the Lower South soared from $3.6 to $13.6 million. There was of course a cotton boom, but this can account for only a fraction of the increase. Cotton exports valued in current prices rose from virtually nil ($47,000 in 1790) to $4.4 million in 1800. Rice, which had been the region’s chief export, did not fare so well; the nominal value of rice exported increased by only $408,000 between 1790 and 1800. In other words, the two major agricultural exports

Estimate of the Value of the Major Exports from the Lower South in Current Prices, 1767–1771 and 1789–1803. Major Export Items

Total Exports from the Lower South, Inclusive Re-Exports

Value of Exports Not Identiﬁed

Share Not Identiﬁed (%)

Rice

Indigo

Naval stores

Deerskins

Cotton

Boards

Staves and headings

Tobacco

Sum of Eight Items

1767–1771 Average

1,368,532

485,424

141,540

164,624

387

76,102

38,775

47,156

2,322,628

2,678,208

355,580

13

1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803

143,341 1,478,710 1,465,695 1,650,322 1,630,144 1,222,108 1,989,413 2,274,874 1,192,194 840,803 888,320 1,887,395 2,033,579 1,645,115 2,150,894

97,104 581,911 784,964 1,008,080 537,078 510,466 378,244 97,086 59,393 15,414 3,953 2,013 3,230 0 240

80,542 153,360 110,211 78,619 62,808 113,200 120,844 118,143 115,972 133,557 75,575 118,597 94,803 284,348 222,094

14,068 24,574 70,297 58,285 60,840 60,953 48,169 50,390 99,936 129,306 49,361 55,462 62,614 106,996 151,382

3,000 47,250 33,983 156,178 496,946 1,840,815 3,196,775 1,096,171 2,102,677 2,499,672 4,755,000 4,425,813 5,811,731 5,822,089 5,006,418

150,486 163,071 171,017 148,912 104,426 143,782 220,266 119,406 148,797 252,592 240,179 194,419 219,520 213,718 195,203

77,928 44,321 52,255 54,484 45,414 63,155 87,389 98,597 86,877 91,951 51,530 101,379 79,937 83,834 77,335

643,844 497,653 437,039 222,697 367,595 333,841 455,356 477,632 647,835 1,134,909 791,125 657,320 521,437 560,261 656,039

1,210,314 2,990,851 3,125,459 3,377,576 3,305,251 4,288,320 6,496,456 4,332,299 4,453,680 5,098,204 6,855,042 7,442,398 8,826,851 8,716,361 8,459,604

na 3,582,928 3,709,066 3,415,256 4,078,236 4,453,327 7,186,639 9,241,694 7,690,326 8,493,837 10,611,695 13,607,577 16,964,298 13,153,706 11,134,597

592,077 583,607 37,680 772,985 165,007 690,183 4,909,395 3,236,646 3,395,633 3,756,653 6,165,179 8,137,447 4,437,345 2,674,993

17 16 1 19 4 10 53 42 40 35 45 48 34 24

Changes between 1790 and 1800 Values 408,685 (579,899) Percentage 28 100

(34,763) 23

30,888 126

4,378,563 9,267

31,348 19

57,058 129

159,667 32

4,451,547 149

10,024,649 280

5,573,102 941

Exports and the Lower South, 1720–1770

Table A10a.

Notes and Sources: The values of the individual exports equal the product of the quantities and prices shown in Table A10B. Total exports from the Lower South including re-exports are from Pitkin (1816, reprinted 1967, Table I, pp. 51–54). The value of exports not identiﬁed equals the total minus the sum of the eight items.

59

Quantities and Prices of the Major Exports from the Lower South, 1767–1771 and 1789–1803. Rice

Indigo

Naval Stores

Deerskins

Cotton

Boards

Staves and Headings

Tobacco

72,545,556

543,928

85,554

379,849

2,008

5,854

2,695

11,628

1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803

5,923,180 56,391,495 63,419,490 66,130,427 60,826,279 45,096,248 55,415,402 54,684,465 52,289,190 48,045,872 45,554,863 49,023,258 41,249,072 42,290,867 40,506,484

101,150 577,150 785,750 873,250 636,650 753,900 425,950 171,500 96,121 19,838 6,892 3,400 4,079 0 221

48,229 45,053 57,282 45,234 30,700 52,274 54,434 45,485 34,481 46,277 38,358 42,483 32,475 59,344 57,799

26,758 46,742 140,307 116,332 121,433 124,840 106,763 97,978 168,869 305,626 152,033 136,099 150,549 253,606 358,813

12,000 189,000 135,930 473,360 1,529,910 5,899,440 5,648,975 3,447,080 8,377,200 8,388,160 15,388,350 17,774,350 22,880,832 33,460,284 30,341,928

11,392 13,356 12,839 16,046 8,567 9,927 13,128 10,532 12,770 13,804 16,772 17,455 19,709 19,306 18,520

4,361 3,309 3,670 3,562 3,234 3,595 4,144 4,441 3,363 4,074 2,321 4,455 3,571 4,228 4,147

158,582 143,010 143,070 80,251 102,847 81,728 92,394 77,868 91,791 128,609 105,329 138,901 104,045 115,518 111,571

0.0042 $0.019

0.2010 $0.892

0.3726 $1.654

0.0976 $0.433

0.0434 $0.193

2.9279 $13.00

3.2402 $14.39

0.9134 $4.06

$0.024 $0.026 $0.023 $0.025 $0.027 $0.027 $0.036 $0.042 $0.023 $0.018

$0.960 $1.01 $1.00 $1.15 $0.84 $0.68 $0.89 $0.57 $0.62 $0.78

$1.670 $3.40 $1.92 $1.74 $2.05 $2.17 $2.22 $2.60 $3.36 $2.89

$0.53 $0.53 $0.50 $0.50 $0.50 $0.49 $0.45 $0.51 $0.59 $0.42

$0.250 $0.250 $0.250 $0.330 $0.325 $0.312 $0.566 $0.318 $0.251 $0.298

$13.210 $12.21 $13.32 $9.28 $12.19 $14.48 $16.78 $11.34 $11.65 $18.30

$17.870 $13.39 $14.24 $15.30 $14.04 $17.57 $21.09 $22.20 $25.83 $22.57

$4.060 $3.48 $3.05 $2.78 $3.57 $4.08 $4.93 $6.13 $7.06 $8.82

PETER C. MANCALL ET AL.

Quantities of exports 1767–1771 Average

Current prices of exports 1767–1771: d Sterling 1767–1771: US dollars 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798

60

Table A10b.

$0.020 $0.039 $0.049 $0.039 $0.053

$0.57 $0.59 $0.79 $0.82 $1.09

$1.97 $2.79 $2.92 $4.79 $3.84

$0.32 $0.41 $0.42 $0.42 $0.42

$0.309 $0.249 $0.254 $0.174 $0.165

$14.32 $11.14 $11.14 $11.07 $10.54

$22.20 $22.76 $22.39 $19.83 $18.65

$7.51 $4.73 $5.01 $4.85 $5.88

Notes and Sources: The data for 1767–1771, both price and quantity, are the values reported by Shepherd and Walton for 1768–1772 (1972, Appendix IV, Tables 2–A). The reported data were shifted back a year to reﬂect the crop year. The sterling values were converted at $4.44 per pound sterling. Rice: For 1789–1800 we used quantity data from Gray (1958, pp. 1020–1023, Table 37). For 1801–1803 we extrapolated the 1800 ﬁgure forward on the basis of the change in rice exports for the U.S. taken from Holmes (1912, pp. 5–7). The Lower South’s share of U.S. exports implicit in this calculation is 86 percent, about equal to the 87.5 percent share for 1791–1792 based on tierces exported (New American State Papers, vol. 1, pp. 163ff.). Indigo: 1789–1800, from Gray (1958, p. 1024). Gray reports quantities in chests or casks, both of which were assumed to amount to 350 pounds. Deerskin: Exports were compiled by us (see Mancall et al., 2002b). Cotton: Bruchey (1967, Table 3.A). Lower South exports were estimated based on total U.S. exports and the share of the Lower South in total U.S. production. According to Bruchey, in 1791 the Lower South accounted for 100 percent of cotton production. By 1801 this ﬁgure had fallen to 85 percent. We interpolated linearly between these ﬁgures to get the shares for other years. Naval Stores, boards, staves and headings, and tobacco: The quantity ﬁgures shown as being produced in 1790 and 1791 are the export ﬁgures for 1791 and 1792 (New American State Papers, Commerce and Navigation, Vol. 1, pp. 163ff.). The ﬁgures for other years were estimated based on changes in U.S. exports of each item. The U.S. ﬁgures are from U.S Congress (1884, Table 2, pp. 16–23). The quantities reported in the sources were shifted back one year to better represent the crop or production year. This is a method espoused by Clowse (1981, pp. 57–58, Table B.21). The price data are from Cole (1938), except for deerskin prices. In any year for which Cole did not report a price for Charleston, we estimated a price by extrapolating the nearest Charleston price by the change in the Philadelphia price for that same export item. The deerskin price for 1791 is the price implicit in the exports of deerskins for that year (New American State Papers, Commerce and Navigation, Vol. 1, p. 158). The deerskin price in other years was estimated by extrapolating the 1791 ﬁgure on the change in the Bezanson Price Index for deerskins in Philadelphia (Bezanson et al., 1937).

Exports and the Lower South, 1720–1770

1799 1800 1801 1802 1803

61

62

PETER C. MANCALL ET AL.

of the region can account for only around 40 percent of the increase in the region’s total exports expressed in nominal prices. The remaining major exports cannot explain the increase. Indeed, as a group the value of these other exports declined slightly; pulled down for the most part by the fall in indigo exports. In other words, a very large increase in total exports, more than $6 million or about 60 percent of the increase, remains unexplained by the growth of the eight most important regional products. It is possible that other regional products not among these eight, such as wheat and ﬂour, might account for some of the increase, but more likely most of the increase was due to an upsurge in the re-export of items, such as coffee and sugar, that were not produced in the region. The national ﬁgures suggest this was the case. For the nation, the nominal value of exports rose from $20.2 million in 1790 to $70.8 million in 1800, about the same percentage increase as took place in the Lower South. Re-exports, which were recorded for the nation, rose by $38.8 million accounting for about three-fourths of the total increase. Domestically produced exports rose by only $11.8 million.71 Because our interest is in the value of real exports produced in the region, it was necessary that we take account of the two phenomena of inﬂation and re-exports. We ﬁrst attempted to do this by using the available published statistics. We deducted an estimate of the value of re-exports from the reported value of all exports for the region in order to obtain a ﬁgure representing the value of agricultural exports produced in the region. We then adjusted those current values of domestically produced exports for price changes in order to obtain the real values. The results are shown in Table A11. The current value of exports shown there is net of our estimate of reexports from the region. We derived re-exports from the region as the sum of estimates for each of three states (GA, NC, and SC), adjusting each state’s series for the value of re-exports likely to have been included in the reported totals. For 1791–1802, the reported export ﬁgures combined the value of domestically produced exports and re-exports. Beginning in 1803, the values of the two types of exports were reported separately by the state. We used the relative shares for 1803–1810 for each state to distribute the totals reported for each state in the earlier years. For Georgia and North Carolina, domestically produced exports comprised 99 or 100 percent of total exports in every year after 1803, so we assumed all exports in each of these states for the years 1791–1802 were produced domestically. The domestically produced share of exports in South Carolina was less than 100

Real Value of Exports from the Lower South, 1790–1800. Estimated by Deﬂating the Current Price Values

Dollar Value of Exports in Current Prices

Price deﬂator David–Solar– McCusker price index

Cole’s export price index

(1840=100) Three-year average benchmark values 1790 3,582,928 1800 10,255,397 Average rates of change 1790–1800 11.09

105.8 145.2 3.22

Real value of exports deﬂated by North’s export price index

David–Solar– McCusker price index

(1790=100)

189.3 171.6 0.98

100.0 145.9 3.85

Cole’s export price index

Prices of 1840

North’s export price index

Prices of 1790

3,386,510 7,062,946

1,892,725 5,976,339

3,582,928 7,029,059

7.63

12.18

6.97

Exports and the Lower South, 1720–1770

Table A11.

Notes and Sources: See text for a discussion of how the current dollar value of exports produced in the region were obtained. The David–Solar–McCusker Price Index was constructed by McCusker (2000, Table A.1). McCusker extended the Brady–David–Solar Consumer Price Index backward to 1720 based on Bezanson’s index of Philadelphia commodity prices. Whereas the original index used 1860 as the base year, we have shifted the index to a base year of 1840. Cole’s export price index is that for Charleston’s export prices, prepared by George Rogers Taylor (U.S. Bureau of the Census (1975), Series E: 92–95). North’s export price index taken from North (1961, Table C.III, p. 221).

63

64

PETER C. MANCALL ET AL.

percent in each year and varied over the period 1803–1810. The share dropped between 1803 and 1805–1807 and then rose. We extrapolated the domestic share for South Carolina back to 1791 based on the change in the domestically produced share for the United States and an assumed ratio of South Carolina’s share to that for the United States. In order to bias upward the 1800 ﬁgure for the Lower South – and thus bias upward the rate of growth that had taken place before 1800 – we used the higher average ratio found for the subperiod 1805–1807 rather than the average for the longer term 1803–1810. For 1790, we had to resort to an alternative method of estimation because even the total export ﬁgures by state were unavailable; only national ﬁgures were reported but they did provide the breakdown between domestically produced exports and re-exports. We estimated the value of domestically produced exports for the region as 18 percent of the ﬁgure reported for the United States, this percentage being the average for the years 1791–1796.72 These net values were then deﬂated by several available, but imperfect, price indexes to obtain estimates of the real value of exports produced in the region. The results using three different price series as deﬂators are shown in Table A11. All the deﬂated series show substantial increases over the decade of the 1790s, with the average annual rates of growth ranging from 6.97 to 12.18 percent per year. The Taylor–Cole Export Price Index is conceptually the more appropriate deﬂator as it measures changes in prices of the chief exports from the region. For the period after 1796 the price index includes all South Carolina’s export staples except deerskins, while before 1791 it covers the four chief export staples: rice, deerskins, naval stores, and indigo. The real value series based on that deﬂator may be the best of the three, but there are several shortcomings. The most important ﬂaw is that the index constructed for the period 1796–1812 gives cotton a weight of 65 percent. This large a weight may be suitable for much of the period, but not for the 1790s when cotton was only beginning its rise to prominence. As was seen in Table A7, and again in Tables A10a and 10b, at the beginning of the 1790s cotton accounted for at most 2 percent of the region’s exports. Its share rose substantially thereafter, but even at the end of the decade its share was not above 50 percent, well below the 65 percent weight in the Taylor–Cole Index. Another shortcoming is that there is no coverage between 1791 and 1796, and the index for the post-1796 rests on a different base year than that for the period before 1791.73 We have linked the two series based on the change in the price of rice between 1791 and 1796 in order to put them on the same base year in both periods. Given that the Taylor–Cole Index gives

Exports and the Lower South, 1720–1770

65

such a large weight to cotton, interpolating by changes in the price of rice may not be ideal, but it seemed the only practical alternative.74 Although imperfect, the series deﬂated by the Cole index gives some measure of the likely increases in the real value of domestically produced exports. Given the extremely high weight given to cotton, this series most likely gives an upwardly biased measure of growth in the region’s exports during the 1790s. We also estimated the real value of domestically produced exports in the 1790s by applying the same method that was used to link 1767–1771 with 1790–1792. That is, we ﬁrst derived an index of change in the volume of the eight most important exports and then extrapolated the benchmark ﬁgure for 1791 (expressed in prices of 1840) on the change in that index. This too does not lend itself to unequivocal results because of the wide variation in the behavior of the chief exports and the large changes in their relative importance over time. As was seen in Table A10b, the pattern of increase or decrease was not uniform among the chief exports. The quantity of cotton surged while the quantity of rice declined substantially between 1790 and 1800. The quantity of naval stores, which had fallen off substantially between 1770 and 1790, remained fairly constant during the 1790s. The quantity of deerskins, which had declined precipitously between 1770 and 1790, rose substantially during the 1790s although at the end of the century the quantity exported was still well below the 1768–1772 average. The quantity of wood products, both boards as well as staves and headings, remained roughly constant during the 1790s, while tobacco which had increased noticeably between 1770 and 1790 declined during the 1790s. Indigo exports fell throughout. Based on these conﬂicting patterns, one cannot say with much conﬁdence that real exports in total increased or decreased over the closing decade of the century. It should also be pointed out that the quantity ﬁgures for the region’s export of naval stores, boards, staves and headings, and tobacco in the years after 1792 were estimated by assuming that the region’s share of the U.S. exports of those products equaled the share for 1791 and 1792. The consequence of this is that the behavior of these exports is heavily inﬂuenced by what was happening in the other regions that exported these products. The behavior of exports from those other regions may not represent well the growth in the exports from the Lower South. This is perhaps most obvious in the case of tobacco for which the Lower South was a minor exporter, but it may be true for naval stores and wood products as well. The Lower South accounted for only 13 percent of the nation’s export of tobacco in 1791 and 1792, whereas it accounted for 31 percent of the wood products and 39 percent of the naval stores.

66

PETER C. MANCALL ET AL.

Despite these shortcomings, we weighted the quantities of these eight exports by prices to obtain an index of the change in the volume of the eight exports combined that took place between 1791 and 1800. In fact, we constructed two indexes: one in which the quantities were weighted by prices at the beginning of the period (1791 prices) and a second using prices from the end of the period. The two estimates are shown in Table A12. The two index series behave very similarly, especially up to 1798. They diverge somewhat after that point, with the index weighted by 1791 prices rising more rapidly. Nevertheless, for the entire decade the two indexes give fairly similar results. Real exports increased substantially between 1791 and 1800, by 106 percent when weighted by 1791 prices and by 83 percent when weighted by prices of 1800. The average annual rates of change between 1790 and 1800 were 9.1 and 7.7 percent. These rates are also very similar to those obtained by deﬂating the estimated current price value of domestically produced exports by either the David–Solar–McCusker Price Index or North’s Export Price Index (see Table A11), but are below the rate of change implied when the current price ﬁgures were deﬂated by Cole’s Export Price Index. Based on these comparisons, the Cole-deﬂated series appears to be an outlier, and we think this is for the reason we stated earlier that it simply gives too great a weight to cotton during this decade. We have used the index based on the 1791 prices to extrapolate the real value of all exports expressed in prices of 1840. Those estimates are also shown in Table A12. We also show there the per capita value of exports expressed in prices of 1840. The per capita ﬁgures held quite steady until 1798, then increased sharply.

Rice

Estimate of the Value of the Major Exports from the Lower South in Constant Prices, 1789–1801. Indigo

Per Capita Total Index of Value of Exports, Real Exports in Exclusive Exports (1791=100) Re-Exports, 1840 Prices in 1840 Prices

Deerskins

Cotton

Boards

Staves and Headings

Tobacco

Sum of Eight Items

Estimates using prices of 1791 1789 136,891 101,049 1790 1,303,270 576,573 1791 1,465,695 784,964 1792 1,528,348 872,377 1793 1,405,763 636,013 1794 1,042,224 753,146 1795 1,280,712 425,524 1796 1,263,819 171,329 1797 1,208,461 96,025 1798 1,110,393 19,818 1799 1,052,824 6,885 1800 1,132,982 3,397 1801 953,312 4,075 1802 977,389 0 1803 936,150 221

92,792 86,682 110,211 87,030 59,067 100,576 104,731 87,513 66,342 89,038 73,801 81,737 62,481 114,178 111,205

13,406 23,419 70,297 58,285 60,840 62,547 53,491 49,089 84,607 153,125 76,172 68,189 75,428 127,062 179,772

3,000 47,250 33,983 118,340 382,478 1,474,860 1,412,244 861,770 2,094,300 2,097,040 3,847,088 4,443,588 5,720,208 8,365,071 7,585,482

151,740 177,895 171,017 213,730 114,110 132,225 174,860 140,291 170,098 183,871 223,399 232,504 262,522 257,157 246,689

62,088 47,112 52,255 50,716 46,052 51,191 58,995 63,234 47,882 58,004 33,048 63,432 50,843 60,191 59,038

484,425 436,856 437,039 245,144 314,168 249,655 282,239 237,865 280,395 392,864 321,750 424,303 317,828 352,874 340,819

1,045,391 2,699,057 3,125,459 3,173,969 3,018,491 3,866,424 3,792,796 2,874,910 4,048,110 4,104,154 5,634,966 6,450,130 7,446,697 10,253,923 9,459,376

33 86 100 102 97 124 121 92 130 131 180 206 238 328 303

1,747,975 4,557,450 5,226,014 5,132,654 5,047,154 6,464,965 6,341,853 4,807,075 6,768,758 6,862,467 9,422,106 10,785,125 12,451,463 17,145,365 15,816,820

$2.39 $5.98 $6.63 $6.30 $5.98 $7.40 $7.01 $5.12 $6.95 $6.79 $8.98 $9.88 $10.97 $14.54 $12.90

Estimates using prices of 1800 1789 228,042 59,881 1790 2,171,073 341,673 1791 2,441,650 465,164 1792 2,546,021 516,964 1793 2,341,812 376,897 1794 1,736,206 446,309 1795 2,133,493 252,162

134,638 125,772 159,911 126,277 85,704 145,931 151,961

10,904 19,048 57,176 47,406 49,485 50,873 43,507

2,988 47,061 33,847 117,867 380,948 1,468,961 1,406,595

126,884 148,755 143,003 178,720 95,418 110,566 146,217

99,231 75,296 83,515 81,057 73,601 81,814 94,288

750,459 676,766 677,050 379,772 486,702 386,759 437,238

1,413,027 3,605,444 4,061,318 3,994,083 3,890,567 4,427,420 4,665,462

35 89 100 98 96 109 115

1,818,252 4,557,450 5,226,014 5,132,654 5,006,295 5,697,105 6,003,413

$2.49 $5.98 $6.63 $6.30 $5.93 $6.52 $6.63

67

Naval Stores

Exports and the Lower South, 1720–1770

Table A12.

68

Table A12. (Continued )

1796 1797 1798 1799 1800 1801 1802 1803

Per Capita Total Index of Value of Exports, Real Exports in Exclusive Exports (1791=100) Re-Exports, 1840 Prices in 1840 Prices

Rice

Indigo

Naval Stores

Deerskins

Cotton

Boards

Staves and Headings

Tobacco

Sum of Eight Items

2,105,352 2,013,134 1,849,766 1,753,862 1,887,395 1,588,089 1,628,198 1,559,500

101,528 56,904 11,744 4,080 2,013 2,415 0 131

126,979 96,260 129,191 107,082 118,597 90,658 165,668 161,354

39,927 68,816 124,546 61,955 55,462 61,350 103,347 146,220

858,323 2,085,923 2,088,652 3,831,699 4,425,813 5,697,327 8,331,611 7,555,140

117,311 142,235 153,752 186,806 194,419 219,520 215,033 206,280

101,062 76,526 92,705 52,818 101,379 81,258 96,199 94,357

368,495 434,381 608,616 498,448 657,320 492,372 546,664 527,988

3,818,977 4,974,178 5,058,970 6,496,750 7,442,398 8,232,989 11,086,722 10,250,970

94 122 125 160 183 203 273 252

4,914,175 6,400,662 6,509,771 8,359,874 9,576,713 10,594,028 14,266,148 13,190,720

11.28

57.52

2.71

3.02

0.29

9.10

9.10

9.00

5.16

11.28

57.52

2.71

3.02

0.29

7.52

7.52

7.71

3.91

Notes and Sources: The values of the individual exports equal the quantities times the prices (see Table A10B for data). Total exports valued in 1840 prices for 1790–1792 are from Table A9 above. Total exports valued in 1840 prices for subsequent years equal the 1791 value times the index of real exports.

PETER C. MANCALL ET AL.

Average annual rate of change, 1790–1800 Using 1.39 40.16 0.59 prices of 1791 Using 1.39 40.16 0.59 prices of 1800

$5.24 $6.58 $6.44 $7.96 $8.77 $9.34 $12.09 $10.76

QUARTERLY DATA ON THE CATEGORIES AND CAUSES OF BANK DISTRESS DURING THE GREAT DEPRESSION, 1929–1933 Gary Richardson ABSTRACT During the contraction from 1929 to 1933, the Federal Reserve System tracked changes in the status of all banks operating in the United States and determined the cause of each bank suspension. This chapter introduces that hitherto dormant data and presents aggregate series constructed from it. The new data series will supplement, and in some cases, supplant the data currently used to study banking panics during the period, which were published by the Federal Reserve Board of Governors in 1937.

Miss Jones, Division of Bank Operations, requests that she be notiﬁed should any of this data be considered for destruction. She considers this valuable from a historical standpoint as it is the only record the Board has prior to 1933. Plans are now under way for making a card record, and Miss Jones wants to be sure this record is complete prior to destruction. –Esther Crews, June 11, 19411

Research in Economic History, Volume 25, 69–147 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(07)25002-5

69

70

GARY RICHARDSON

INTRODUCTION The causes, consequences, and possibilities of preventing the banking panics of the Great Contraction have been debated for seven decades. The debate’s factual foundations rest upon data published in the Federal Reserve Bulletin. The September 1937 issue contains the only comprehensive collection of statistics on suspended banks, and is the sole source of aggregate bank failure rates (Board of Governors of the Federal Reserve System, 1937, hereafter FRB’37).2 Scholars studying the contraction continually redeﬁne, reinterpret, and reveal new relationships between data from FRB’37 and measures of industrial, commercial, and ﬁnancial activity. The principal reason the debate continues may be the primary source of evidence. FRB’37 provides imperfect information about bank distress. It distinguishes neither temporary from terminal suspensions, nor voluntary from involuntary liquidations, nor institutions afﬂicted by illiquidity from banks suffering insolvency. It contains information neither on the causes of bank suspensions nor the number of bank mergers. The smallest period of aggregation at the national level is the month and at the Federal Reserve district level is the year. Key terms remain undeﬁned, leaving much open to interpretation. This chapter introduces new statistical series that provide precise, detailed, aggregate information about categories of bank distress and causes of bank suspensions. The source for the new series is the same as for the old series.3 From 1929 to 1933, the Board of Governors collected data on changes of status for all banks operating in the United States, both members of the Federal Reserve System and nonmembers, state and national, incorporated and private. The Board also analyzed the cause of each bank suspension. The Division of Bank Operations recorded this information on the St. 6386 series of forms. Form St. 6386a reported bank consolidations. Form St. 6386b reported bank suspensions. Form St. 6386c reported all other bank changes.4 Facsimiles of the forms appear in the appendix. The St. 6386 series comprehensively covered the commercial banking industry from January 1929 through the national banking holiday in March 1933. Observations existed for a wide array of events affecting banks. These events included the major, such as openings, closings, reopenings, receiverships, and consolidations, and the minor, such as changes in Federal Reserve membership, capital stock, charter type, and even street address. The forms also included ﬁnancial information for each bank on the date of each transaction. The complete series of St. 6386 forms survives in the National Archives of the United States.

Categories and Causes of Bank Distress, 1929–1933

71

The remainder of this chapter introduces this recently rediscovered source. Section 1 reviews the relevant literature, highlights FRB’37’s role in academic debates, discusses how the extant evidence shapes academic opinions about the contraction, and elucidates the utility of the archival evidence. Section 2 describes the data, deﬁnes key concepts, and discusses issues important for interpreting empirical work based upon this source. Section 3 describes the original sources of information and the procedures that the Federal Reserve used to validate and cross-check the evidence. The purpose of the extended discussion is to reassure the reader of the veracity of the Federal Reserve’s observations. Section 4 compares the archival data to the extant aggregate series and to studies based on microeconomic evidence. The comparisons show that aggregate series constructed from the archival data match equivalent series published in FRB’37 and that the Division of Bank Operation’s conclusions about the causes of suspensions match those of modern scholars using historical and econometric methods. Section 5 presents new quarterly series on the categories and causes of bank distress. The new series supplement the FRB’37 series most often used by scholars to study the contraction. Section 6 presents data on the categories of bank distress aggregated at the Federal Reserve District level. This district-level data enable scholars to study the geographic distribution of bank distress during the Great Depression. Section 7 discusses potential uses of the new data.

1. REVIEW OF THE LITERATURE Scholars have long studied the banking crises of the Great Contraction. The initial studies analyzed data from FRB’37 aggregated at the national level. Later studies scrutinized FRB’37 data aggregated at lower levels. Recent studies combine data from FRB’37 with data drawn from the balance sheets of national and Federal Reserve member banks. This section reveals correlations between the sources that scholars have analyzed and the conclusions that scholars have reached. Scholars ask three general questions about the banking system during the contraction of the early 1930s. Why did bank suspensions surge at certain points in time? How did bank suspensions affect commercial and industrial activity? Could Federal Reserve intervention have prevented (or did its actions trigger) this crisis? Despite 70 years of analysis, debate persists about the answer to each inquiry. Concerning the reasons that banks failed, some scholars conclude that a contagion of fear, a ﬂight to cash holdings, and withdrawals en masse drained

72

GARY RICHARDSON

deposits from banks and pushed ﬁnancial markets toward collapse. Illiquidity of assets and Federal Reserve inaction exacerbated the credit crunch. Milton Friedman and Anna Schwartz (1963) argue that the Federal Reserves’ failure to act as a lender of last resort, prevent banking panics, and stem the decline of the money supply transmogriﬁed what would have been an ordinary recession into the Great Depression. They reached this conclusion by examining seasonally adjusted series of deposits in suspended banks from FRB’37 to determine the dates of banking panics and correlating those dates with changes in monetary aggregates. Elmus Wicker (1996) uses district-level data on suspensions from FRB’37 to illuminate the regional patterns of panics. Other scholars conclude that banks failed because the economy contracted, loan default rates rose, and asset values declined. Deteriorating fundamentals forced banks into insolvency, continuing a process of liquidation that began during the 1920s. Peter Temin (1976) argued that real, rather than monetary, forces caused banks to fail. Temin supported his thesis by regressing statelevel FRB’37 suspension data on explanatory variables. Eugene N. White (1984) advanced this line of argument by examining a sample of national banks. White regressed national banks’ fates (survival or liquidation) on their ﬁnancial characteristics, aggregate variables, and FRB’37 state-level suspension rates. The most recent and comprehensive work in this line of research (Calomiris & Mason, 2003) regresses time-to-liquidation for individual Fed member banks on an array of bank characteristics, aggregate variables, and state-level FRB’37 suspension rates. Concerning the consequences of the banking crises, some scholars believe that banking panics had monetary effects. Panics eroded depositors’ conﬁdence, induced further withdrawals, forced banks to liquidate assets at deep discounts, lowered asset prices, encouraged banks to hold excess reserves, and reduced the money multiplier. This vicious cycle reduced the money supply and turned what would have been a typical recession into a cataclysmic contraction. Friedman and Schwartz (1963) contains the seminal statement of the monetarist position, whose roots date back at least to the Chicago monetary tradition which arose during the depression. Friedman and Schwartz employed data from FRB’37. The early monetarists did not have access to these data. Instead, they used data drawn from earlier editions of the Federal Reserve Bulletin. Other scholars see bank failures as symptoms of ongoing events with no special role in the propagation of the downturn. This view arises from Keynesian, classical, real business cycle, and other macroeconomic models. Temin’s (1976) spending hypothesis is the seminal statement of the stark

Categories and Causes of Bank Distress, 1929–1933

73

Keynesian conception of the contraction, which sees bank failures as a symptom of the autonomous decline in consumption and investment expenditure, and which dates back at least to the publication of Keynes’ General Theory. Another set of scholars maintains that bank panics inﬂuenced economic activity by disrupting ﬁnancial intermediation. Bank failures increased the cost of credit intermediation, dislocated the ﬁnancing of small and medium ﬁrms, disrupted current production, and curtailed investment spending. This ﬁnancial acceleration deepened the depression. In the seminal articles in this line of research, Ben Bernanke (1983) regresses growth rates of industrial production on ﬁrst differences of deposits in suspended banks (drawn from FRB’37) to show that suspensions increased the cost of credit intermediation. Concerning the possibilities of preventing the banking panics, some scholars argue that the Federal Reserve could have done little to aid ailing banks. Fundamental forces pushed banks into insolvency; monetary intervention could not pull them out. Liquidity assistance could not eliminate loan losses. Open-market expansion – even on a massive scale – could not lift the economy out of the liquidity trap (Temin, 1976). Other scholars argue that the Federal Reserve could not aid ailing banks directly, since illiquidity and contagion caused few banks failures, but that massive open-market expansions, such as those that the Roosevelt administration implemented after abandoning the gold standard, could reignite economic progress, and thus indirectly alleviate the banking situation (Calomiris & Mason, 2003; Eichengreen, 1992; Romer, 1993; Temin, 1989). A ﬁnal and inﬂuential set of scholars concludes that the Federal Reserve’s sins were of commission as well as omission. The Federal Reserve not only neglected to aid ailing banks, but by raising interest rates, reducing the monetary base, and restricting discount lending, it weakened all banks, and created conditions conducive to panics. As evidence, these scholars highlight the Federal Reserve’s monetary contraction in 1928 and the Federal Reserve’s defense of the gold standard in 1931 (Friedman & Schwartz, 1963; Meltzer, 2003). The policy views of all of these schools of thought stem from their views on the causes of the banking crisis. Scholars who think that banks failed because they lacked liquidity believe that the Federal Reserve could have alleviated the situation by acting as a lender of last resort. Scholars who think that banks failed because the economy contracted and borrowers defaulted believe that the Federal Reserve could not have saved the situation by acting as a lender of last resort, but might have aided the banks indirectly by

74

GARY RICHARDSON

reinﬂating the economy. As I noted previously, these beliefs arise largely from the data which the scholars examine. The debate about policy, therefore, indirectly reﬂects the use of data sources. A correlation exists between the sources that scholars utilize and the conclusions that scholars reach. Scholars who study aggregate data tend to view illiquidity and the withdrawal of deposits as the root cause of the banking crisis. Scholars who study microdata tend to view insolvency and the declining value of investment portfolios as the root of the crisis. The correlation exists for two reasons. First, different sources illuminate different dimensions of the banking industry and highlight different points in time and space, leading to different interpretations of events. Each source portrays a portion of a complex, dynamic mosaic. None portrays the entire picture. Second, the extant sources share several weaknesses. All of the sources provide imperfect information about changes in banks’ status. They distinguish neither temporary from terminal suspensions, nor voluntary from involuntary liquidations, nor institutions afﬂicted by illiquidity from banks suffering insolvency. They describe neither the causes of bank suspensions nor the number of bank mergers. They neglect consolidations of banks in ﬁnancial difﬁculties and seeking to avoid suspension. Unclear and inconsistent terminology leaves much of the information in the sources subject to interpretation. Reconciling scholars’ views requires comprehensive, deﬁnitive data. The remainder of this chapter turns to that task.

2. DATA ORIGINS AND DEFINITIONS This section discusses the data on bank distress collected by the Federal Reserve Board of Governors during the early 1930s. The discussion emphasizes the methods that Federal Reserve agents employed to ascertain the causes of bank suspensions and the lexicon that the Board of Governors devised during the 1920s to facilitate their analysis. The discussion cites the memos and training materials distributed to Federal Reserve employees undertaking this task and outlines the algorithms that they employed when analyzing the evidence. During the 1920s, the Federal Reserve Board of Governors embarked on an ambitious project: the creation of standard statistical reports for all banks operating in the United States, both members of the Federal Reserve System and nonmembers, national and state, incorporated and private. In August 1925, the Board began collecting information on the causes of bank suspensions. The Division of Bank Operations introduced Form X-4401,

Categories and Causes of Bank Distress, 1929–1933

75

‘‘Notiﬁcation of Bank Suspension or Insolvency,’’ and Form X-4402, ‘‘Notiﬁcation of Termination of Insolvency or Suspension.’’ The forms remained in use for four years. During that period, the Board of Governors strove to improve the accuracy of the information that they collected. The process was interactive and ongoing. The Board circulated forms to agents around the country and asked them to suggest improvements. They solicited comments from bureaucrats, bankers, examiners, and academics. They amended the forms to make them more concise and complete. Correspondence between the Board of Governors and the Federal Reserve district banks announced upcoming modiﬁcations and important decisions concerning the coding of ambiguous cases. A series of memos deﬁned key terms and explained how to ﬁll out the forms.5 Analysts strove to standardize the deﬁnitions of the words with which they worked. Standardization was important, because deﬁnitions of banking terms varied across time and jurisdictions. Each state banking bureau published its own report and used terms as it saw ﬁt. Explicit deﬁnitions seldom appeared. This lexical ambiguity caused confusion. Multiple deﬁnitions existed for frequently used terms. Many varied according to context. For example, the nine letters l-i-q-u-i-d-a-t-e referred in some circumstances to the sale of assets, in other circumstances to a change in a corporate charter, and at other times to the dissolution of a ﬁnancial institution and its departure from the banking business. At the end of the 1920s, as the Federal Reserves’ lexicon approached its ﬁnal form and its research staff assimilated the lessons that they had learned, the Board of Governors authorized a comprehensive revision of the panoply of forms used to gather information on the status of banks. In 1929, the Board of Governors introduced the documents devised via that endeavor: the St. 6386 series of forms. Form St. 6386a reported bank consolidations. The Federal Reserve deﬁned consolidation in a de facto rather than de jure sense. A consolidation is the corporate union of two or more ongoing banks into one bank which continues under a single charter, either new or old. The term is used not in a legal or technical sense, but in an economic sense, the essential feature being that the business of two or more going banks becomes concentrated under one charter and one management. The method of effecting the consolidation, whether by purchase of assets, assumption of deposit liabilities, exchange of stock, or other procedure, varies with circumstances and is unessential for the present purpose. What is here described as a consolidation is frequently designated by the terms ‘‘merger,’’ ‘‘absorption,’’ ‘‘amalgamation,’’ ‘‘combination,’’ or ‘‘purchase,’’ according to different points of view. It includes those cases where one bank absorbs another and turns it into a branch. It also includes

76

GARY RICHARDSON those cases where a given bank is absorbed by two or more banks, which distribute its business among themselves. A consolidation is not a simple transaction like a conversion, a liquidation, a primary organization, etc., but is inevitable complicated by the fact that these other transactions are incidental to it. That is, a consolidation as often as not entails a voluntary liquidation and the issuance of a new charter; but these attendant circumstances should not obscure the important fact that a concentration of banking has been effected by the transaction as a whole. (Goldenweiser et al., 1931, pp. 87–88)

Form St. 6386a recorded these transactions and the attendant circumstances. Details included the name of the absorbing bank, the fate of ofﬁces and charter of the bank being absorbed, and whether the banks consolidated because they faced ﬁnancial difﬁculties or to avoid suspension. The latter information helped distinguish between the categories of consolidation and suspension. The distinction involved several simple rules. First, a bank that merged with another bank was classiﬁed as a consolidation if it was ‘‘not at any time closed to depositors, even though the reason for the consolidation or succession may have been ﬁnancial difﬁculties encountered by the bank’’ (Board of Governors of the Federal Reserve System, 1929). Second, a transaction was classiﬁed as a consolidation if the assets and liabilities of a weak bank were transferred to another bank, and then the denuded bank suspended and surrendered its charter (Goldenweiser et al., 1931, p. 88). Third, a transaction was classiﬁed as a terminal bank suspension if at the time of absorption or merger the bank had been closed for at least one business day. Form St. 6386b reported bank suspensions. According to the Federal Reserve’s ﬁnancial lexicon, a bank suspension occurred when payments were halted to the public either temporarily or permanently by supervisory authorities or by the bank’s board of directors on account of ﬁnancial difﬁculties, regardless of whether or not the bank is ultimately classed as a suspension by the supervisory authorities. (Board of Governors of the Federal Reserve System, 1929)

and regardless of whether or not the depositors ultimately suffered losses. A suspension had to last overnight. A bank that closed its doors in the morning and reopened them in the afternoon did not ﬁt into the classiﬁcation. A suspension had to include the cessation of normal banking business. A bank that slowed withdrawals by holding depositors to the agreements that they signed when opening accounts and requiring them to provide 30, 60, or 90 days notice of savings withdrawals did not ﬁt into the classiﬁcation. A bank which, without actually closing, obtained agreements from depositors to waive a portion of their deposits or to defer the withdrawal of a portion of their deposits was not classiﬁed as a suspension.

Categories and Causes of Bank Distress, 1929–1933

77

A bank that closed during a special holiday declared by civil authorities and remained closed only during such holiday or part thereof was not classiﬁed as a suspension. Form St. 6386c reported changes in the status of ﬁnancial institutions, including voluntary liquidations and the reopening of suspended banks. A voluntary liquidation occurred when a bank’s directors paid off all depositors and creditors in full, sold the remaining assets, distributed proceeds to stockholders, annulled the bank’s charter, and abandoned the banking business. Some voluntary liquidations were orderly affairs arranged months in advance. Others resembled suspensions. A sudden shock, such as the death of the founder, forced a bank to close its doors, and the discouraged board of directors decided to repay depositors rather than attempt to reopen the institution. The latter case was classiﬁed as a voluntary liquidation, rather than a suspension, as long as the winding up of the bank’s business did not require the intervention of regulators or courts, and as long as the owners of the bank could repay all creditors the full value of their claims. A reopening occurred when a suspended bank resumed operations. A bank could resume operations voluntarily at any time until regulators used legal powers to take authority over the institution or the courts appointed a receiver to liquidate the bank. After either event, a bank could reopen only with the consent of judicial authorities. In certain circumstances, reopening also required the agreement of creditors such as depositors and stockholders. In many cases, reopenings entailed the reorganization of affairs. Depositors waived rights to portions of their deposits. Stockholders waived their original capital and paid assessments of up to 100% of the value of their stock. In most cases, these reorganizations appear to have been voluntary agreements among the interested parties, facilitated by banking authorities on some occasions and laws in some states. A reopening could involve an array of incidental transactions, such as changes in the name, charter, capital stock, and legal status of a bank. As with consolidations, the entire package of transaction was recorded as a single event. Rules indicated how to handle confusing cases. For example, A reopening consequent upon consolidation of two or more suspended banks should be classed as a single reopening. If a suspended national bank reopens as a State bank, however, or vice versa, the transaction must be accounted as a primary organization and not as a reopening. (Board of Governors of the Federal Reserve System, 1929)

Thus, if two banks suspended operations, merged a few days later in the morning, and reopened that afternoon, the Federal Reserve recorded two suspensions, one reopening, one liquidation, and no consolidations. The

78

GARY RICHARDSON

bank whose charter continued in use was listed as reopened. The bank whose charter lapsed was listed as liquidated. If two banks suspended operations, reopened a few days later in the morning, and merged that afternoon, however, the Federal Reserve recorded two suspensions, two reopenings, and one consolidation. If a national bank suspended operations, changed to a state charter, and then reopened for business with exactly the same staff, depositors, and borrowers, the Federal Reserve recorded one national bank suspension, one national bank liquidation, and the opening (i.e. primary organization) of a newly state-chartered bank. If a similar national bank suspended operations, reopened for business, and then adopted a state charter, however, the Federal Reserve recorded one suspension, one reopening, and one conversion from national to state charter. Completing forms St. 6386c and St. 6386a involved implementing algorithms to determine how to classify complex transactions. Completing the 6386b form required additional knowledge, judgment, and effort. The 6386b form attributed bank suspensions to one or more of ﬁve common causes (deﬁnitions from Board of Governors of the Federal Reserve System, 1930).6 The ﬁrst was slow, doubtful, or worthless paper. The term worthless paper indicated an asset with little or no value. The term doubtful paper meant an asset unlikely to yield book value. Examples included loans to businesses in ﬁnancial difﬁculties and securities such as stocks and bonds which had depreciated since purchase. The term slow paper meant an asset likely to yield full value in time, but whose repayment lagged or that could not be converted to full cash value at short notice. An example was a loan to a proﬁtable corporation with cash-ﬂow problems struggled to make timely payments. The term structure prevented the bank from calling in the unpaid balance. Uncertainty over the value of the stream of payments prevented the bank from selling the loan at face value. The second common cause of suspension was failure of a banking correspondent. Correspondents were banks with ongoing relationships facilitated by deposits of funds. Services rendered by correspondents included extending lines of credit, clearing checks, absorbing exchange charges, redeeming coupons from bonds, conducting wire transfers, supplying coins and cash, and facilitating investments in stocks and bonds. A typical situation involved a small state bank outside a reserve city (called a country bank) that deposited funds in a large national bank in a reserve city (called a city bank) and received services in return. Such deposits often formed a large portion of country banks’ legal reserves. In many cases, the suspension of the city bank precipitated the suspension of the country bank, because the latter lost its reserves and linkages to the wider ﬁnancial system.

Categories and Causes of Bank Distress, 1929–1933

79

The third and fourth common causes of suspension were the failure of a large debtor and defalcation. Of particular interest were debtors with connections to the bank via overlapping ownership, interlocking directorates, or intermingled management. Defalcation was a monetary deﬁciency in the accounts of a bank due to fraud or breach of trust for which the management was legally liable. Embezzlement was a common example. The ﬁfth common cause of suspension was heavy withdrawals. Some banks experienced such large withdrawals that they could no longer continue operations. These deposit declines could be dramatic events in which multitudes of small-account holders lined up in the streets outside of banks hoping to empty their accounts. But, long queues of despondent depositors could be a symptom, rather than a cause, of a bank’s demise. Signiﬁcant deposit losses usually occurred before ordinary men and women lined up on the streets outside of their banks. Businesses, banks, and wealthy individuals possessed the ability to transfer funds quickly via wire or check to other ﬁnancial institutions. They often also possessed better-than-average information about ﬁnancial events and transferred large sums out of banks in weeks before ordinary individuals panicked over the safety of their savings or banks suspended payments on deposits. Researchers at the Federal Reserve called these events invisible runs. The Board of Governors spent several years reﬁning this checklist and teaching representatives how to complete the form. Instructions were straightforward. If acts for which management faced criminal liability precipitated the suspension of payments, check the box labeled defalcation. If the failure of a correspondent or debtor precipitated the suspension of payments, check the appropriate box and note the name of the institution or individual. If withdrawals reduced the liquidity of a bank to the point that it could not continue operations or forced the bank to sell assets at ﬁre sale prices, and thus to choose between suspending payments or suffering insolvency, check the box for heavy withdrawals. If bad assets accumulated to such an extent that a bank could not continue operations, either because it could not maintain the necessary cash ﬂow, because it could no longer absorb the losses, or because auditors determined it to be insolvent, check the box for slow, doubtful, and worthless paper. Place the check in the column labeled primary cause if that factor alone forced the bank to suspend operations. Place the check in the column labeled contributing cause if that factor aggravated or accelerated the bank’s demise. Agents often checked multiple boxes, indicating that multiple causes contributed to a bank’s suspension. The most common combination was heavy withdrawals and slow, doubtful, and worthless paper. That should be no

80

GARY RICHARDSON

surprise. Withdrawals forced a bank to suspend payments only if the bank could not convert assets to cash quickly enough to cover claims against it. Bad assets would not force a bank out of business if it retained sufﬁcient deposits and eventually covered its losses. Runs and frozen assets, in other words, could be ﬂip sides of the same coin. Differences did exist, however, between the asset and liability side of the balance sheet. Agents attempted to communicate these distinctions by marking primary and contributing causes in particular patterns. If the problems precipitating suspension arose on the asset side of a bank’s balance sheet, agents checked slow, doubtful, or worthless paper as the primary cause and withdrawals as a contributing cause or not at all. Checking paper as the primary cause indicated that problems with the bank’s assets necessitated suspension regardless of other factors. The banks’ assets possessed little value. Collections had fallen far behind scheduled. The capital was impaired. The bank was or would soon be insolvent. Checking withdrawals as a contributing cause indicated that depositors’ demands exacerbated the situation. Withdrawals could complicate asset problems in several ways. A run on a bank on the edge of insolvency might hasten its demise. Fears of runs might force a bank to hold large reserves of liquid assets, such as cash, rather than large portfolios of remunerative assets, such as loans, preventing the bank from earning proﬁts and writing off bad debts. In an effort to satisfy depositors’ demands, a bank might sell so many of its good assets that only bad assets remained, or a bank might borrow against so many of its better investments that it could not earn enough proﬁts to retire its substandard loans. If the problems precipitating suspension arose on the liability side of the bank’s balance sheet, agents checked heavy withdrawals as the primary cause and paper as a contributing cause or not at all. Checking withdrawals as the primary cause indicated that deposits declined to such an extent that the bank could not continue operations. Its reserves of cash were depleted. It could no longer convert assets to currency. It could not satisfy depositors’ demands. Checking paper as the contributing cause indicated that the condition of the bank’s assets exacerbated the situation. Doubtful and worthless assets could neither be sold nor rediscounted. Slow assets could not be converted to cash quickly enough to alleviate a run. Substandard assets of any kind reduced banks’ liquidity and prevented them from converting resources to cash. If problems arose equally on both sides of the balance sheet, agents checked both withdrawals and paper as primary causes. In this case, the bank’s position was doubly difﬁcult. Its assets appeared so problematic that

Categories and Causes of Bank Distress, 1929–1933

81

insolvency approached, and its depositors withdrew so much of their savings that illiquidity loomed. Checking two primary causes indicated that either factor alone would have caused the closure of the bank. For the depression as a whole, the suspensions of 2,106 banks (38%) were attributed to a single cause. The suspensions of 2,552 banks (46%) were attributed to two causes. In three out of four cases, the pair included one primary and one contributing cause. In one out of four cases, the pair included twin primary causes. The suspensions of 544 banks (10%) were attributed to three causes. These cases were evenly split between banks with two primary and one contributing cause and one primary and two contributing causes. The suspensions of 57 banks (1%) were attributed to four causes. The suspensions of 278 banks (5%) were not attributed to causes on the checklist. In most of these cases, the Federal Reserve agent who completed the form included a written paragraph describing the factors leading up to the bank’s failure. Table 1 summarizes the issues discussed in this section. Column (1) indicates the changes in status tracked by the Division of Bank Operations. Column (2) indicates whether a change in status was a form of bank distress. Column (3) indicates which of the changes in status involved the creation or dissolution of a bank (or several banks) charters. Column (4) indicates type of St. 6386 form (a, b, or c) on which the information should have been reported. Column (5) contains a concise deﬁnition of the change in status. For exact deﬁnitions and a clariﬁcation of borderline cases, please consult the discussion above.

3. DATA SOURCES AND CROSS-VALIDATION This section delves into details of the data-collection process and the procedures that the Federal Reserve used to validate and cross-check the evidence. The purpose of the extended discussion is to reassure the reader of the veracity of the Federal Reserve’s observations, illuminate their value for historical research, and elucidate the information available in the database, so that readers may effectively employ it in their own research. The Board of Governors strove to gather information about causes of suspension from the man on the spot who knew the facts of the issue at hand. Annotations on the forms reveal these efforts. Comments on hundreds of forms indicate that the information agrees with the receiver’s report or was taken ‘‘From receiver’s report as of date of closing,’’ a phrase often imprinted with an ink stamp. Federal Reserve agents often communicated

82

GARY RICHARDSON

Table 1.

Changes in Bank Status, Deﬁnitions, and Sources, January 1929 to March 1933.

Change in Status

Distress?

Change in Charter?

Source St. 6386

Brief Deﬁnition. A Bank that y

(1)

(2)

(3)

(4)

(5)

Suspension, terminal

Yes

Yes

b

Suspension, temporary

Yes

Ceased operations and entered liquidation under supervision of a receiver Ceased operations temporarily and then reopened for business

Voluntary liquidation

Yes

Yes

c

Ceased operations and rapidly repaid depositors the full value of deposits

Consolidation in ﬁnancial difﬁculties

Yes

Yes

a

Consolidation

Yes

a

Merged with another bank while at risk of suspension or liquidation Merged with another bank while in a sound ﬁnancial position

Organization of new bank

Yes

c

Acquired a charter and opened its doors to depositors

c

Joined the Federal Reserve System

c

Departed from the Federal Reserve System Surrendered state charter but continued under a new, national charter Surrendered national charter but continued under a new, state charter

Conversion – nonmember to member Conversion – member to nonmember Conversion – state to national charter Conversion – national to state charter Change in capital stock Change in street address

b, c

Yes

c

Yes

c

c c

Increased or decreased the amount of paid-up capital Headquarters moved to a new location

Sources: National Archives and Record Administration (see text for details). Board of Governors of the Federal Reserve System (1929, 1930, 1931).

directly with receivers, requesting additional information or inquiring about complex or unclear cases. Receivers’ replies appear on many forms. For example, the receiver for Mayo’s Money Exchange Bank of San Antonio, Texas, stated ‘‘the primary cause of failure was due to investments of funds

Categories and Causes of Bank Distress, 1929–1933

83

in stocks and bonds representing real estate investments.’’7 The Central Bank and Trust Company of Chicago, Illinois, which was the liquidation trustee of the Mecca Bank of Mecca, Indiana, stated the bank should be classiﬁed as a voluntary liquidation because ‘‘all of the creditors will be paid off in full within a reasonable period.’’8 A letter dated 6 October 1932 and written by F. E. J. Bower, the receiver for the First National Bank of Bradley Beach, New Jersey, described the ‘‘reorganization plan as follows: Creditors were to waive 40% of claims against the closed bank, total amounts of such waiver equal 90% of present liabilities. Old stock of bank ($100 par) to be surrendered and then resold at $25 per share in this way raising $125,000.’’9 The Division of Bank Operations also consulted call reports and examination reports and communicated directly with bank examiners and banking departments. Examples abound. The examiner of the Central National Bank in Bartleaville, Oklahoma, which suspended on 22 March, 1930, classiﬁed ‘‘$235,000 of assets as worthless and $164,000 as doubtful.’’10 The examiner of the First National Bank of Bixby, Oklahoma, which suspended on 18 February 1929, stated ‘‘the suspension of this bank is chargeable directly to top heavy and other ill-advised loans made during the period of inﬂation following the recent war, together with the incompetency of succeeding ofﬁcers in the matter of enforcing any effective policy of collection over the following period of deﬂation.’’11 On 28 February 1931, a letter from the Bank Commissioner of Connecticut stated the Central Fairﬁeld Trust Company of Norwalk, now the Merchants Bank and Trust Company, resumed business on Friday, February 20, 1931. The name of the institution was changed by special act of the legislature. While the name has been changed, it is still the same corporation which has resumed business upon the discharge of the temporary receiver appointed by the courts.12

In January 1930, a telegram from the Indiana State Banking Department explained the complicated series of transactions surrounding the suspension of the Argos State Bank of Argos, Indiana, and its reopening under a new charter.13 In March 1933, the Utah State Bank Commissioner reported that Gunnison Valley Bank of Gunnison, Utah, would reopen shortly.14 On 27 March 1930, an employee of the Division of Bank Operations with the initials M.F. called Mr. Gough of the Comptroller of Currency after ‘‘examination of agent’s reports proved inadequate.’’ Mr. Gough explained that the myriad of transactions reported for the town of Bartow, Florida, pertained to the closure of a single institution, the Polk County National Bank, which unsuccessfully attempted to shed dubious assets, change its charter, and establish itself as a new bank under a different name in hopes of saving itself.15

84

GARY RICHARDSON

The Division of Bank Operations also contacted institutions and individuals with local knowledge of particular banks. Local courts provided access to liquidation records, names and addresses of court-appointed receivers, and opinions about suspended banks under their jurisdiction. Local postmasters provided another means to attain such information. For example, on 10 December 1930, a letter from the postmaster at Sandia, Texas, informed the Federal Reserve that W.T. Mumme, the private bank in the town, ceased operations and liquidated voluntarily.16 For country banks, a good source of information was their banking correspondent. Correspondents in reserve cities held deposits of county-bank clients, provided clients with services such as check clearing and wire transfers, and monitored clients’ ﬁnancial health and creditworthiness. Correspondents frequently replied to inquiries concerning the closures of the clients. Many examples exist. The correspondent of Juan McKeyes and Company, a private bank in Lawton, Michigan, wrote that McKeyes suspended operations on 9 May 1932 due to the ﬁnancial consequences of grape crop failures in the area, and reopened on 2nd June of that year after its depositors of their own volition had signed waiver agreements providing that no funds except such as are deposited after the bank reopened would be withdraw for a period of ﬁve years. Under the circumstances the court having jurisdiction over its affairs dismissed the receivership petition.17

The Continental Illinois Bank and Trust Company, which was the correspondent of the Commercial Bank of Wapellow, Iowa, stated that the latter closed its doors on 13 October 1931, because the bank unwisely expanded loans to care for needs of borrowers in communities where banks had suspended and found itself unable to meet shrinkage in its deposits due to unrest among its own customers.18

The Valley National Bank of Des Moines, Iowa, stated that its client, the Farmers and Merchants Bank of Scranton, Iowa, suspended operations primarily because it was ‘‘loaded with slow, doubtful, and worthless paper.’’19 Banks provided information about their own conditions. Reopened banks answered inquiries about the cause of their temporary suspensions and the conditions under which they reopened. Banks entering into voluntary liquidation released public statements explaining how and why. Even the management of failed banks provided information in instances when Federal Reserve agents contacted them directly. For example, in the fall of 1930, Harry M. Wilcox, the President of the Citizens Bank of North Adams, Michigan, replied to an inquiry about the cause of his bank’s suspension by

Categories and Causes of Bank Distress, 1929–1933

85

stating ‘‘general business depression and decline in land value caused the bank to close its doors.’’20 Federal Reserve agents could gather this information because they possessed the legal authority to compel truthful testimony. The law required bankers to answer their inquiries. Untruthful respondents could be prosecuted for perjury. Federal Reserve agents seeking answers from reticent witnesses could ask the Ofﬁce of the Comptroller of Currency (OCC) for assistance. OCC examiners could compel a response, since they possessed investigative powers identical to secret service agents. Federal Reserve agents consulted the publications of the state banking departments.21 Most banking departments published annual reports. A few published biennially. Many banking departments also published summaries of bank changes in the interim. Periodicity varied across states. Indiana published a monthly bulletin. The New York State Banking Department published a weekly bulletin with detailed descriptions of the causes of each state bank suspension. The New York Superintendent of Banks also released an announcement at the time of each bank suspension, stating the reasons why it was ‘‘deemed unsafe and inexpedient to permit the institution to continue in business.’’ The Federal Reserve banks also compiled monthly summaries of bank changes, which they forwarded to the Board’s Division of Bank Operations. Federal Reserve agents also consulted an array of business periodicals and popular dailies. Citations to the following appear on the 6386 forms:

Commercial and Financial Chronicle Bankers’ Monthly New York Sun New York Times Polk’s Bankers Encyclopedia Rand McNally Bankers Directory Reports of R.G. Dunn & Company

Numerous newspapers from locales around the United States were also cited, but the typists who transcribed the newspaper clippings merely stated that their source was a ‘local newspaper’ without additional bibliographic details. Each of these sources possessed its own strength. Federal Reserve ofﬁcers had direct and continuous contact with the management of member banks and the ability to comment credibly on their competency. Correspondents could also provide vivid and valuable testimony on this issue. Correspondents had ongoing relationships with the management of client banks, and tracked the quality of their clients’ portfolios in case decisions had to be made about the extension of credit. Call, examiners’, and receivers’ reports

86

GARY RICHARDSON

yielded ﬁnancial data including the kind, quality, and quantity of assets possessed by a bank on the date of suspension and the patterns of deposit inﬂows and outﬂows before that date. Examiners of all national and most state banks classiﬁed assets according to the categories good, slow, doubtful, and worthless, which the Federal Reserve used on its suspension forms. Receivers of national and state banks also listed assets and estimated values. Discussions with examiners, receivers, correspondents, and bankers provided expert opinions about phenomena that they observed, often over a period of years. Particularly useful are their opinions about the quality of the management and the importance of correspondent linkages. Newspapers provided detailed descriptions of events that interested their readers, such as defalcations and bank runs, and on important ﬁnancial events in their hometowns, such as the closing and reopening of local banks. Much of the information provided by these sources could come from nowhere else. The Federal Reserve Board incorporated all of this information into the 6386 forms by creating a nationwide reporting network. The law required member banks to report changes in status to their local Federal Reserve Bank. Similar regulations required national banks to report such information to the OCC. Federal Reserve agents at district and branch banks completed initial drafts of the forms from information at their disposal and forwarded the forms to the Board of Governors in Washington, where the Division of Bank Operations checked the information against available sources (including copies of the materials submitted to the OCC and routinely forwarded to the Federal Reserve), sorted and tabulated the forms, compiled aggregate statistics, and disseminated the results. Nonmember banks reported to state authorities, which cooperated in the Federal Reserve’s data-collection endeavors by forwarding the relevant information. The Federal Reserve cultivated contacts with state agencies, and during the Depression, subsidized the salaries of state employees who worked part time on Board businesses. In complicated, unclear, and unresolved cases, the Division requested additional information and researched the events in greater depth. The deﬁnitive determination of the cause of a bank suspension often took several months. The archives contain preliminary, in-process, and ﬁnal reports for many banks. These duplicates accumulated as agents incorporated additional information into their analysis and updated their conclusions. Information ﬂowed to the Division of Bank Operations as institutions and individuals replied to Federal Reserve requests for information. In several cases, the materials collected by the Division, such as examiners’ reports, telegrams from receivers, press clippings, etc., remain clipped to the 6386

Categories and Causes of Bank Distress, 1929–1933

87

forms, although routine procedure appears to have been to discard the source material after reaching a deﬁnitive determination of the cause of a bank’s failure. The Board’s efforts continued throughout the Depression. After the national banking holiday, the Board joined forces with the OCC and the Federal Deposit Insurance Corporation. With funding from the Works Project Administration, this regulatory triumvirate strove to construct a database containing information from all banks operating in the United States since 1929. As part of the process, the Board of Governors rechecked the 6386 forms against the records of the state banking bureaus, the OCC, and the Rand McNally and Polk corporations. The Board published the initial results of this reconciliation in the Federal Reserve Bulletin for September 1937. Further publications were planned, but the project lapsed as the Board allocated resources toward other projects.

4. THE ACCURACY OF THE ARCHIVAL INFORMATION The previous sections describe how the Board of Governors gathered information about events effecting banks and how the Division of Bank Operations determined the cause of each bank’s demise. The method involved gathering ﬁnancial data for each institution; talking with the regulators, examiners, receivers, correspondents, and bankers who knew the facts of the issue at hand; applying a set of algorithms devised to determine the cause of the failure; and reporting the results in a lexicon devised to convey the conclusions clearly, concisely, and consistently. This section compares the archival evidence to tabulations previously published in FRB’37, a wide array of historical sources, and numerous scholarly studies. The comparisons demonstrate four important points. First, the archives retain the original dataset in its entirety. Second, the archival information on events effecting banks is accurate. The information matches facts revealed by other sources. Third, the archival information on the causes of bank suspensions is accurate. The information matches conclusions drawn from historical studies of particular institutions and econometric studies of the broader ﬁnancial system. Fourth, the archival evidence contains information previously unavailable to scholars. The initial question one should ask about the archival evidence is: do ﬁgures from it correspond closely to the ﬁgures published in the FRB’37?

88

GARY RICHARDSON

600

500

Archival Data Published Data

400

300

200

100

0 Jan-29

Jul-29

Jan-30

Jul-30

Jan-31

Jul-31

Jan-32

Jul-32

Jan-33

Fig. 1. Comparing Archival and Published Data. Number of Bank Suspensions, by Month, January 1929 to March 1933. Sources: Archival data from National Archives and Record Administration (see text for details). Published data from Federal Reserve Bulletin (September 1937, Table 12, p. 907).

Figs. 1 and 2 answer that question. In Fig. 1, the solid line is the number of bank suspensions each month according to FRB’37. This tally includes all banks: national, state, and private. The dotted line represents the archival data. It is calculated from the St. 6386 forms in the National Archives using the deﬁnitions, rules, and algorithms established by the Division of Bank Operations in the 1930s. The lines correspond closely. The coefﬁcient of correlation for the two series is 0.996. In Fig. 2, the solid line is total deposits in the banks that suspended each month according to FRB’37. This tally includes all banks and both temporary and terminal suspensions. The dotted line represents the archival data. It is calculated from the St. 6386 forms in the National Archives using the deﬁnitions, rules, and algorithms established by the Division of Bank Operations in the 1930s. The lines correspond closely. The correlation coefﬁcient for the two series is 0.998. Similar exercises on other cross-tabulations duplicate the analogous series published in 1937. The replication of the FRB’37 series demonstrates that

Categories and Causes of Bank Distress, 1929–1933

89

600

Millions of Dollars

500

Archival Data Published Data

400

300

200

100

0 Jan-29

Jul-29

Jan-30

Jul-30

Jan-31

Jul-31

Jan-32

Jul-32

Jan-33

Fig. 2. Comparing Archival and Published Data. Deposits in Suspended Banks, Millions of Dollars, by Month, January 1929 to March 1933. Sources: Archival data from National Archives and Record Administration (see text for details). Published data from Federal Reserve Bulletin (September 1937, Table 13, p. 909).

the archival data remains comprehensive. The dataset retains observations for all (or almost all) of the banks represented by the FRB’37 series. A discrepancy between the two series exists during the ﬁrst ﬁve months of 1929. In each month, the archival data contains 15–30 fewer suspensions than the published tabulations. I have not been able to determine the source of this discrepancy. I suspect the cause lies in the Federal Reserves transition to a new reporting system beginning in January 1929. After July 1929 (i.e. after the start of the contraction), the archival and published series are nearly congruent. From July 1929 to February 1933, the correlation coefﬁcient between the archival and published series is 0.999. The slight discrepancies after July 1929 appear to be due to duplicate entries in the dataset and complicated or unclear cases where properly applying the coding algorithms is difﬁcult. The most common complication was a bank that suspended operations and then consolidated with another institution. If the suspended bank reopened for business (for even a brief moment) prior to consolidation, then the Division of Bank Operation’s algorithms reported these events as a bank suspension (temporary), a

90

GARY RICHARDSON

reopening, and a consolidation (i.e. three transactions on three separate dates). If the banks did not ofﬁcially reopen, then the algorithm classiﬁed this transaction as a terminal suspension (on the original date of suspension). In approximately 25 cases, information in the St. 6386 forms did not clearly indicate the order of events and the appropriate classiﬁcation of the transaction. In several cases, preliminary St. 6386 forms contradicted revised St. 6386 about the relevant details. Reference to contemporary publications – such as newspapers and state banking reports – did not clarify the situation. So, in all these cases, I have assumed that the bank did reopen prior to consolidation, and that the events should be recorded as a temporary suspension, a reopening, and a consolidation. Additional reassurance of the accuracy of the archival information comes from state-level databases which I have constructed by combining information from government (state examiners, Fed, and OCC reports) and commercial sources (Rand McNally’s Bankers Directory and Polk’s Bankers Encyclopedia). For the four states that I have completed, Florida, Mississippi, New York, and Tennessee, information about changes in bank status in the state-level databases and St. 6386 forms match at rates over 99.5%. The few discrepancies appear to be temporary suspensions (which were not recorded in many sources) and incorrect St. 6386 forms (e.g., misspellings of bank names and locations and incorrect transaction dates). The striking similarities between datasets derived from these independent sources suggest that the archival dataset contains observations for all banks that suspended operations during the Great Contraction. Does the quality of archival information about the causes of bank suspensions equal the quality of information on the timing and nature of changes in bank status? A thorough investigation of the evidence indicates that the answer is yes. This conclusion rests on three facts. First, for individual institutions whose demise has been the subject of academic analysis, the conclusion of scholars coincides closely with the cause of suspension indicated on the relevant St. 6386b form. For example, Wicker (1980) and McFerrin (1939) attribute the demise of the Bank of Tennessee (Nashville, TN), the principal bank of the Caldwell conglomerate, to an ill-informed strategy that led to large investment losses. Wicker (1980) and McFerrin (1939) attribute the demise of the American Exchange Trust Company (Little Rock, AR), a Caldwell afﬁliate, to runs on the institution. The corresponding St. 6386 forms attribute the Bank of Tennessee’s suspension to ‘‘depreciation in value of securities’’ and excessive debts on real estate and bills payable. The corresponding St. 6386 forms attribute the American Exchange Trust Company’s suspension to ‘‘heavy

Categories and Causes of Bank Distress, 1929–1933

91

withdrawals due to rumors caused by failure of Caldwell and Company, Nashville, Tenn.’’22 Friedman and Schwartz (1963) and Joseph Lucia (1985) attribute the failure of the Bank of United States to its large, underperforming real estate portfolio, which threatened its solvency and prevented it from merging with other institutions, and to ‘‘runs on several of the bank’s branches’’ (p. 310, n. 9). The corresponding St. 6386 form attributes the suspension to the primary cause of ‘‘slow, doubtful, and worthless paper’’ with the contributing cause of ‘‘heavy withdrawals.’’23 Second, for broader events that have been the subject of academic analysis, the academic conclusions coincide closely with the causes of suspension indicated on the relevant St. 6386b forms. For example, Friedman and Schwartz (1963) attribute the surge in suspensions during March 1931 to the public’s ‘‘resumed conversion of deposits into currency.’’ The St. 6386 forms paint a similar picture. Federal Reserve agents reported that heavy withdrawals were the primary cause of almost of the suspensions and stated laconically on most of the forms that the bank closed its doors after runs depleted its cash reserves.24 Calomiris and Mason (1997) attribute the Chicago banking panic of June 1932 to depositors’ confusion about bank asset quality. Depositors withdrew funds widely from banks which they feared might be insolvent, because they did not know which institutions were, in fact, insolvent. The St. 6386 forms tell a similar tale. For the preponderance of the banks which failed during the panic, examiners concluded that the primary cause of suspension was doubtful and worthless assets. Runs on these institutions were a contributing cause of their closure. Written comments on the suspension forms indicate that banks suffered from declines in the value of securities, bonds, and real estate and from large loan losses.25 Third, conclusions drawn from econometric studies of the causes of bank suspensions correspond closely with the causes of suspension indicated on the relevant St. 6386b forms. One example is the study by Calomiris and Mason (1997) discussed above. Another example comes from my study of bank suspensions in the state of Mississippi. In 1929, the industrial depression and the failure of local businesses caused the closure of banks in the surrounding community. These closures were concentrated in the southern half of the state under the jurisdiction of the 6th Federal Reserve District. In 1930, a banking panic swept through the state, forcing large numbers of banks to suspend operations in the 6th District and an even higher percentage to suspend operations in the northern half of the state, which was under the jurisdiction of the 8th Federal Reserve District.26

92

GARY RICHARDSON

Table 2. Comparing Econometric and Archival Evidence on Causes of Suspension Probit Predictions, Mississippi, 1929 and 1930. Predicted

6th District (%) 8th District (%)

Actual

1929 (i)

1930 (ii)

1929 (iii)

1930 (iv)

7.1 3.0

4.5 3.0

7.1 3.0

14.2 39.5

Source: Richardson and Troost (2006).

Table 2 presents predictions from a parsimonious probit model using bank and county characteristics in July 1929 to predict suspensions between July 1929 and June 1930. The table also uses the coefﬁcients from the 1929 regression and characteristics of banks and counties in July 1930 to predict suspension rates between July 1930 and June 1931. Column (i) indicates the average predicted probability of suspension for 1929. Column (iii) indicates the actual suspension rate in 1929. The null hypothesis that the former equals the latter cannot be rejected, demonstrating that the model ﬁts the data reasonably well. Column (ii) indicates the average predicted probability of suspension for 1930. The prediction for the 8th District changes little, because the balance sheets of banks in the 8th District changed little between July 1929 and July 1930. The prediction for the 6th District falls substantially, because the 6th Districts high failure rate for 1929 was driven by adverse shocks in particular counties. In 1930, fewer banks operate in those counties (in fact, almost all of the banks in those counties failed). In the remainder of the district, the balance sheets of banks, and thus the predicted probability of failure, changed little. Comparing Columns (ii) and (iv) shows that our model of fundamentals which ﬁt the data well for 1929 does not predict the events that occurred in the following year. The standard interpretation of these regression results is that fundamentals caused banks to fail in 1929 while unpredictable panics caused the failures one year later. The archival evidence concurs with this conclusion. The St. 6386 forms indicate that the most common cause of suspension in 1929 was problematic assets. The St. 6386 forms indicate that the most common cause of suspension in 1930 was heavy withdrawals. Half of the suspensions were temporary. In sum, in the Mississippi case, the causes of suspension indicated on the St. 6386 forms are consistent with the econometric evidence. The veracity of the archival evidence raises the question: what information is available in the new dataset? What does the data reveal?

Categories and Causes of Bank Distress, 1929–1933

93

160 All Observations

140

New Observations + Temporary Suspensions

Number of Banks per Week

New Observations

120 100 80 60 40 20 0 Jan-29

Jan-30

Jan-31

Jan-32

Jan-33

Fig. 3. Comparing Archival and Extant Data. Notes: Solid black area indicates the number of voluntary liquidations plus consolidations due to ﬁnancial difﬁculties. The gray area indicates the number of temporary suspensions. The vertically striped area indicates the number of terminal suspension. The sum of these three categories is the total number of new and extant observations (i.e. all observations). Source: National Archives and Records Administration (see text for details).

Fig. 3 summarizes the available information. The solid black area at the bottom indicates the new observations on bank distress. The new observations consist of voluntary liquidations and consolidations due to ﬁnancial difﬁculties. The gray area indicates the number of temporary bank suspensions. The FRB’37 reported these suspensions, but did not differentiate them from permanent suspensions. The vertically striped area indicates the number of terminal suspensions. The entire shaded area indicates the total number of bank changes due to ﬁnancial difﬁculties that occurred each week from January 1929 to March 1933. For all these observations, the archival database adds an array of information including the cause of the suspension, the date of the suspension, the ﬁnancial condition of the bank on that date, the identity of the institution or individuals who initiated the change, and in many cases, written comments concerning the circumstances leading to and/ or resulting from the event.

94

GARY RICHARDSON

Three patterns in the new data should be noted. First, the new observations are clustered between the fall of 1930 and winter of 1932. During that period, bank suspensions surged above pre-depression levels in repeated waves. Second, the archival evidence allows us to view the data without chronological aggregation. Weekly and daily series of suspensions display volatility much greater than the monthly series published in FRB’37. On several occasions, the number of bank suspensions rose (or fell) overnight by a factor of 20 or more. Such wild swings may be one reason that bank failures made such an impression on public perceptions. Third, the banking holiday of March 1933 does not appear in the archival evidence. Why not? The Federal Reserve’s reporting system was not designed to (and for that reason did not) track the status of banks that closed during state or national banking holidays. The Division of Bank Operations provided an approximation (4,000) of the number of banks that did not reopen within three months of the national banking holiday, which was published (with caveats) in FRB’37. The exact number is unknown. This lacuna in the evidence is a problem universal to all datasets. No extant source describes this event in detail.

5. CHRONOLOGICAL DATA ON CATEGORIES OF DISTRESS AND CAUSES OF SUSPENSIONS The previous sections described the nature and properties of the archival evidence. This section presents aggregate series constructed from this rediscovered source. The series are designed to illuminate the nature of the banking crises during the early 1930s, to supplement (and in some cases replace) the series in FRB’37, and to enable economic historians and macroeconomists to accurately assess the causes and consequences of the contraction. Tables 3 and 4 present annual hazard rates for categories and causes of distress. Table 3 focuses on permanent bank changes. Table 4 focuses on the causes of temporary suspensions. In the tables, row (a) indicates the total number of banks suspending operations. Row (a.1) indicates the total number of banks suspending operations due to problems with assets and liabilities. Imbalances often arose on one side of the balance sheet, but could arise simultaneously on both sides. Rows (a.1.1)–(a.1.5) display the possible combinations: (a.1.1) slow, doubtful, or worthless assets listed as the primary cause of suspension; (a.1.2) assets listed as the primary cause and heavy

Categories and Causes of Permanent Bank Changes, Annual Hazard Rates, January 1929 to March 1933. 1929 %

Banks at beginning of year (a) Terminal suspensions due to (a.1) Assets and liabilities (a.1.1) Assets (a.1.2) Assets (p) and withdrawals (c) (a.1.3) Withdrawals (p) and assets (p) (a.1.4) Withdrawals (p) and assets (c) (a.1.5) Withdrawals (a.2) Closure of correspondent (a.3) Defalcation and mismanagement (a.4) Other causes (b) Consolidations due to difﬁculties (c) Voluntary liquidations (d) Net changes for other reasons Banks at end of year

1930 #

%

1931 #

%

1932 #

%

1933 #

%

Total #

%

#

24,198 1.79 433 0.88 212 0.39 95 0.17 42 0.05 11 0.13 32 0.13 32 0.09 22 0.51 123 0.31 76 0.22 54 0.16 39 6.20 1,500

22,172 4.86 1,078 3.18 704 1.32 293 0.51 114 0.27 59 0.51 113 0.56 125 0.35 78 0.59 130 0.75 166 0.61 136 0.39 86 6.75 1,497

19,375 10.11 1,958 7.70 1,491 2.22 430 1.29 250 0.70 135 2.24 434 1.25 242 0.40 77 0.73 142 1.28 248 1.59 309 0.57 110 0.43 84

17,082 7.61 1,300 5.91 1,010 1.98 338 1.25 214 0.42 71 1.46 250 0.80 137 0.29 50 0.43 74 0.97 166 0.88 151 0.58 99 6.39 1,092

14,440 2.91 420 2.15 310 0.60 86 0.40 58 0.13 19 0.73 106 0.28 41 0.14 20 0.11 16 0.51 74 0.26 38 0.26 37

24,198 21.44 5,189 15.40 3,727 5.13 1,242 2.80 678 1.22 295 3.86 935 2.38 577 1.02 247 2.00 485 3.02 730 2.84 688 1.53 371 16.55 4,005

22,172

19,375

17,082

14,440

13,945

13,945

Categories and Causes of Bank Distress, 1929–1933

Table 3.

Notes: (p) indicates a primary cause and (c) indicates a contributing cause. Statistics for 1933 include only the months January–March. The initial entry is the number of banks operating at the beginning of January 1929. For each year, the percentage column indicates the fraction of the banks operating in that year which experienced each event. For the total column, the percentage indicates the fraction of banks which experienced each event at some point in time from January 1929 to March 1933. Source: Data on banks in operation from Wicker (1996, Table 1.1, p. 2) and Federal Reserve Bulletin (September 1937, Table 12, p. 907). Suspension data from National Archives and Records Administration (see text for details).

95

96

Table 4.

Causes of Temporary Bank Changes, Annual Hazard Rates, January 1929 to March 1933. 1929 %

Banks at beginning of year (a) Temporary suspensions due to (a.1) Assets and liabilities (a.1.1) Assets (a.1.2) Assets (p) and withdrawals (c) (a.1.3) Withdrawals (p) and assets (p) (a.1.4) Withdrawals (p) and assets (c) (a.1.5) Withdrawals (a.2) Closure of correspondent (a.3) Defalcation and mismanagement (a.4) Other causes

1930 #

24,198 0.29 70 0.12 30 0.03 7 0.01 2 0.02 5 0.02 6 0.04 10 0.02 6 0.09 21 0.05 13

%

1931 #

22,172 1.25 278 0.65 144 0.10 23 0.13 28 0.05 12 0.14 32 0.22 49 0.30 67 0.11 24 0.19 43

%

1932 #

19,375 1.63 316 1.18 229 0.14 27 0.23 44 0.14 28 0.27 53 0.40 77 0.10 20 0.09 17 0.26 50

%

1933 #

17,082 0.94 161 0.67 114 0.09 16 0.20 35 0.02 3 0.26 44 0.09 16 0.11 18 0.01 2 0.16 27

%

Total #

14,440 0.33 48 0.19 27 0.03 5 0.02 3 0.02 3 0.07 10 0.04 6 0.03 4 0.12

17

%

#

24,198 3.70 873 2.31 544 0.33 78 0.47 112 0.22 51 0.61 145 0.67 158 0.49 115 0.27 64 0.64 150

GARY RICHARDSON

Notes: (p) indicates a primary cause and (c) indicates a contributing cause. Statistics for 1933 include only the months January–March. The initial entry is the number of banks operating at the beginning of January 1929. For each year, the percentage column indicates the fraction of the banks operating in that year which experienced each event. For the total column, the percentage indicates the fraction of banks which experienced each event at some point in time from January 1929 to March 1933. Source: Data on banks in operation from Wicker (1996, Table 1.1, p. 2) and Federal Reserve Bulletin (September 1937, Table 12, p. 907). Suspension data from National Archives and Records Administration (see text for details).

Categories and Causes of Bank Distress, 1929–1933

97

withdrawals listed as a contributing cause; (a.1.3) both withdrawals and assets listed as primary causes; (a.1.4) heavy withdrawals listed as the primary cause and assets as the contributing cause; and (a.1.5) withdrawals alone as the primary cause of suspension. Row (a.2) indicates the number of banks suspending due to the closure of a correspondent. Row (a.3) indicates the number of banks suspending due to defalcations, mismanagement, excessive loans to ofﬁcers and directors, excessive investment in building and facilities, and similar circumstances. Row (a.4) indicates the number of banks suspending due to other or multiple causes. The latter consists in part of complex cases which do not ﬁt into the mutually exclusive categories above, often because they spanned two or more classiﬁcations. An example is a poorly managed bank which failed to enforce collections on its slow farm loans and which experienced runs after local newspapers revealed that its president embezzled funds from savings accounts. Row (b) indicates the number of consolidations due to ﬁnancial difﬁculties. Row (c) indicates the number of voluntary liquidations. Row (d) indicates the number of net changes for reasons unrelated to ﬁnancial difﬁculties. This series consists almost entirely of the number of mergers minus the number of new openings. Table 5 (located at the end of this chapter) indicates the number of banks in distress by category and quarter. The quarters are deﬁned as winter (January, February, and March), spring (April, May, and June), summer (July, August, September), and fall (October, November, and December). Columns (1)–(3) refer to the three categories of distress – terminal suspensions, voluntary liquidations, and consolidations due to ﬁnancial difﬁculties – in which banks departed from the banking business. Column (4) indicates the total of the ﬁrst three columns. Column (5) indicates the number of temporary bank suspensions – a form of distress in which banks remained in business. Column (6) indicates the total number of banks in distress (i.e. Columns (1)+(2)+(3)+(5)). The ﬁgures in Table 5 can be compared to Tables 1 and 12 of FRB’37. Those tables report total number of bank suspensions, which is equivalent to the sum of Columns (1) and (5) in Table 5 of this chapter. Scholars studying the macroeconomic effects of bank suspensions should rely on the FRB’37 series for the years 1927–1928 and 1933–1936, for which there is little evidence of temporary bank suspensions. Scholars should interpret results for the period 1921–1926 with caution, because narrative reports of temporary suspensions exist, but data distinguishing temporary from permanent suspensions does not. Scholars should employ bank suspension data from this chapter when studying the contraction of the early 1930s (i.e. summer 1929 to winter 1933), because the series of temporary suspensions peaks in the

98

Table 5.

Number of Banks in Distress by Category and Quarter, Winter 1929 to Winter 1933. Suspensions, Terminal

Voluntary Liquidations

Consolidations Due to Financial Difﬁculties

Total, Distressed Departures

Suspensions, Temporary

Total, Banks in Distress

(1)

(2)

(3)

(4)

(5)

(6)

Winter Spring Summer Fall

102 92 96 141

1 1 14 23

2 3 19 29

105 96 129 193

9 17 25 20

114 113 154 213

1930

Winter Spring Summer Fall

229 174 185 489

30 17 19 21

38 19 15 52

297 210 219 562

32 34 21 190

329 244 240 752

1931

Winter Spring Summer Fall

316 306 496 844

34 12 23 42

65 57 58 119

415 375 577 1,005

63 13 53 189

478 388 630 1,194

1932

Winter Spring Summer Fall

435 282 244 340

45 20 12 22

67 36 21 25

547 338 277 387

81 23 38 21

628 361 315 408

1933

Winter

420

38

34

492

48

540

5,191

374

659

6,224

877

7,101

Total

Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

1929

Categories and Causes of Bank Distress, 1929–1933

99

fall of 1930 and the fall of 1931. Those quarters contain breaks in trend for many macroeconomic time series (including money supply, interest rates, and output), and those quarters tend to be the outliers that identify relationships in most macroeconomic models. Table 6 reports deposits of banks in distress by category and quarter in thousands of dollars. Figures from this table can be compared to Tables 2 and 13 of FRB’37, which report total deposits in suspending banks, which is equivalent to the sum of deposits in banks suspending temporarily and terminally (i.e. the sum of Columns (1) and (5)). Scholars studying the macroeconomic effects of deposits in suspending banks should treat these data in the same manner as they treat data on the number of suspensions. For the years 1927–1928 and 1933–1936, rely on FRB’37. For the period 1921–1926, interpret results with caution. For the period winter 1929 to winter 1933, employ the new data separating total deposits in suspending banks into terminal and temporary components. Scholars studying data on deposits should keep in mind an additional caveat. Form St. 6386a, which reported bank consolidations, and Form 6386c, which reported voluntary liquidations, did not contain a space for information on deposits. On occasion, Federal Reserve agents included this information. For that reason, dollar values of deposits appear in the spreadsheet in Columns (2) and (3), but those ﬁgures are far from complete, and do not systematically represent dollar values of deposits in banks consolidating due to ﬁnancial difﬁculties or undergoing voluntary liquidation. Table 7 reports loans and investments in banks in distress by category and quarter in thousands of dollars. Figures from this table can be compared to Table 3 of FRB’37, which reports loans and investments in suspending banks, which is equivalent to the sum of loans and investments in banks suspending temporarily and terminally (i.e. the sum of Columns (1) and (5)). Scholars studying the macroeconomic effects of loans and investments should treat these data in the same manner as they treat data from the previous two tables. For the years 1927–1928 and 1933–1936, rely on FRB’37. For the period 1921–1926, interpret results with caution. For the period 1929 to winter 1933, use this table to separate loans and investments in suspending banks into their total and temporary components. The data on loans and investments in Table 7 have an advantage over the data on deposits in Table 6. All three St. 6386 forms report this information for banks on the date of a change in status (i.e. suspension, liquidation, or consolidation). This information reveals, therefore, the relative signiﬁcance of different categories of distress, as measured by the productive assets of the banks involved. The bottom line of Table 7 illuminates this issue. During

Deposits of Banks in Distress, by Category and Quarter, in Thousands of Dollars, Winter 1929 to Winter 1933. Suspensions, Terminal

Voluntary Liquidations

Consolidations Due to Financial Difﬁculties

Total, Distressed Departures

Suspensions, Temporary

Total, Banks in Distress

(1)

(2)

(3)

(4)

(5)

(6)

Winter Spring Summer Fall

19,115 25,662 55,200 39,811

0 0 744 73

0 0 275 1,200

19,115 25,662 56,219 41,083

2,260 5,148 10,806 8,045

21,375 30,810 67,026 49,128

1930

Winter Spring Summer Fall

75,196 90,169 69,145 524,594

426 91 44 100

232 17,097 23 273

75,854 107,356 69,212 524,968

12,694 14,225 10,315 89,142

88,547 121,581 79,527 614,109

1931

Winter Spring Summer Fall

133,745 271,015 429,473 561,001

107 241 770 242

2,540 0 1,384 4,879

136,393 271,256 431,627 566,122

21,173 4,487 37,208 251,588

157,566 275,743 468,835 817,709

1932

Winter Spring Summer Fall

263,645 199,350 83,443 130,903

0 10,737 57 0

842 0 525 331

264,487 210,087 84,025 131,234

38,536 9,259 16,316 7,699

303,023 219,346 100,340 138,933

1933

Winter

191,520

5,050

510

197,080

16,156

213,236

3,162,988

18,681

30,111

3,211,780

555,054

3,766,834

Total

Notes: Figures for deposits Columns (2) and (3) are incomplete, because the relevant St. 6386 forms did not request information on deposits. Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

1929

100

Table 6.

Loans and Investments of Banks in Distress, by Category and Quarter, in Thousands of Dollars, Winter 1929 to Winter 1933. Suspensions, Terminal

Voluntary Liquidations

Consolidations Due to Financial Difﬁculties

Total, Distressed Departures

Suspensions, Temporary

Total, Banks in Distress

(1)

(2)

(3)

(4)

(5)

(6)

1929

Winter Spring Summer Fall

11,320 10,722 39,710 37,132

113 31 750 5,565

316 1,189 9,123 16,949

11,749 11,942 49,582 59,646

1,688 2,222 2,616 5,463

13,437 14,163 52,198 65,109

1930

Winter Spring Summer Fall

77,680 92,964 71,922 557,996

3,999 5,415 2,735 1,793

34,862 24,391 26,790 55,581

116,540 122,770 101,447 615,370

13,217 15,593 9,723 99,463

129,758 138,363 111,170 714,833

1931

Winter Spring Summer Fall

147,393 310,684 495,800 698,744

4,678 3,760 5,619 6,663

53,402 298,349 95,067 725,454

205,473 612,793 596,485 1,430,860

22,934 4,788 40,649 290,563

228,407 617,581 637,135 1,721,423

1932

Winter Spring Summer Fall

296,799 230,213 99,403 177,880

8,348 37,628 1,620 1,656

151,018 142,260 28,216 17,678

456,165 410,100 129,239 197,215

39,019 10,516 18,260 9,499

495,183 420,616 147,499 206,714

1933

Winter

233,555

18,189

44,623

296,366

20,458

316,824

3,589,915

108,560

1,725,267

5,423,742

606,670

6,030,412

Total

101

Source: National Archives and Records Administration (see text for details).

Categories and Causes of Bank Distress, 1929–1933

Table 7.

102

GARY RICHARDSON

the contraction, loans and investments of the 5,191 banks suspending permanently amounted to $3.6 billion. Loans and investments of the 877 banks suspending temporarily amounted to $607 million. Loans and investments of the 659 banks consolidating due to ﬁnancial difﬁculties amounted to $1.7 billion. Most of these consolidations occurred during panics in the period beginning in the fall of 1930 and continuing until the winter of 1932. The peak occurred during the fall of 1931, after Britain abandoned the gold standard and the Federal Reserve raised the discount rate to stem ﬁnancial outﬂows. During that quarter, banks with assets totaling $725 million consolidated in the face of ﬁnancial difﬁculties. Banks with loans and investments totaling $698 million suspended operations terminally. Banks with loans and investments totaling $291 million suspended operations temporarily. Tables 8–10 report the causes of terminal bank suspensions. The columns of the table have the same deﬁnitions as the rows of Table 3, and are labeled accordingly. Table 8 indicates the number of terminal suspensions by cause and quarter. Table 9 indicates deposits in banks suspending permanently in millions of dollars. Table 10 indicates loans and investments in banks suspending permanently in millions of dollars. Tables 11–13 report the causes of temporary bank suspensions. The columns of the table have the same deﬁnitions as the rows of Table 3. Table 11 indicates the number of temporary suspensions by cause and quarter. Table 12 indicates deposits in banks suspending temporarily in millions of dollars. Table 13 indicates loans and investments in banks suspending temporarily in millions of dollars. Tables 8–13 have no counterpart in FRB’37. The Federal Reserve never published data on the causes of suspensions during the Great Contraction. Scholars interested in the macroeconomic implications of these series must restrict their attention to the early 1930s. Reasonable assumptions, however, should allow them to extend the series forward. The number of banks suspending operations dropped dramatically after March 1933. In 1934, only 57 banks suspended operations. In 1935 and 1936, the number of suspensions was 34 and 44, respectively. In 1934, Congress established the Federal Deposit Insurance Corporation. In the years that followed, depositors returned funds to ﬁnancial institutions, and banks held tremendous quantities of cash reserves. In this environment, it seems probable that unlucky investments, idiosyncratic asset shocks, and mistakes of management caused the bulk of bank failures, and that bank runs and contagion through correspondent networks played a lesser part. So, macroeconomists may be able to safely assume that the bulk of bank suspensions during the

Number of Terminal Bank Suspensions by Cause and Quarter, Winter 1929 to Winter 1933. Cause of Terminal Suspension Assets and/or withdrawals Assets and liabilities

Assets

(a.1)

(a.1.1)

Assets (p) and Assets (p) and withdrawals withdrawals (c) (p)

Withdrawals (p) and assets (c)

Withdrawals

Closure of correspondent

Defalcation

Other causes

Total, terminal suspensions (a)

(a.1.2)

(a.1.3)

(a.1.4)

(a.1.5)

(a.2)

(a.3)

(a.4)

1929

Winter Spring Summer Fall

55 56 50 96

30 30 21 44

11 11 3 19

1 1 5 10

10 9 10 8

3 5 11 15

6 1 15 3

13 7 8 12

28 28 23 30

102 92 96 141

1930

Winter Spring Summer Fall

145 106 133 327

70 46 68 109

25 7 12 77

13 4 11 31

15 22 17 59

22 27 25 51

8 11 3 58

21 21 17 21

55 36 32 83

229 174 185 489

1931

Winter Spring Summer Fall

258 240 346 651

94 66 134 150

53 21 61 111

23 25 30 51

36 102 65 238

52 26 56 101

3 5 34 42

14 15 18 24

41 46 98 127

316 306 496 844

1932

Winter Spring Summer Fall

341 227 197 248

91 82 70 86

46 41 49 85

27 15 15 15

108 57 40 49

69 32 23 13

17 7 5 21

17 13 5 11

60 35 37 60

435 282 244 340

1933

Winter

Total

304

83

53

20

107

41

21

5

90

420

3,780

1,274

685

297

952

572

260

242

909

5,191

103

Notes: (p) indicates a primary cause and (c) indicates a contributing cause. Source: National Archives and Records Administration (see text for details).

Categories and Causes of Bank Distress, 1929–1933

Table 8.

Deposits in Terminal Bank Suspensions, by Cause and Quarter, in Millions of Dollars, Winter 1929 to Winter 1933.

104

Table 9.

Cause of Terminal Suspension Assets and/or withdrawals Assets and liabilities

Assets

Assets (p) and withdrawals (c)

Assets (p) and withdrawals (p)

Withdrawals (p) and assets (c)

Withdrawals

Closure of correspondent

Defalcation

Other causes

Total, terminal suspensions

(a.1)

(a.1.1)

(a.1.2)

(a.1.3)

(a.1.4)

(a.1.5)

(a.2)

(a.3)

(a.4)

(a)

Winter Spring Summer Fall

8.2 16.4 18.9 29.1

3.6 4.8 5.9 9.5

1.3 1.8 1.1 10.5

0.1 0.0 2.6 2.5

2.9 9.2 7.2 3.0

0.2 0.5 2.1 3.6

0.8 2.1 24.5 0.4

2.8 1.7 1.5 2.4

7.3 5.6 10.4 8.0

19.1 25.7 55.2 39.8

1930

Winter Spring Summer Fall

49.6 67.8 52.4 461.8

24.6 24.4 19.4 51.7

8.2 22.1 2.8 246.9

7.1 1.8 16.9 13.1

3.7 7.8 6.6 76.5

6.1 11.6 6.6 73.7

1.0 6.5 0.7 16.7

12.2 5.7 4.5 5.7

12.3 10.2 11.6 40.5

75.2 90.2 69.1 524.6

1931

Winter Spring Summer Fall

109.5 236.3 306.9 449.2

32.3 24.7 35.2 40.9

22.8 10.4 60.0 100.7

16.1 27.9 31.7 36.7

19.8 133.8 106.6 176.2

18.5 39.5 73.4 94.7

5.1 1.3 17.3 28.6

6.0 3.3 10.7 9.9

13.1 30.1 94.7 73.3

133.7 271.0 429.5 561.0

1932

Winter Spring Summer Fall

187.7 184.1 64.7 86.7

44.9 40.0 14.9 32.5

21.7 24.8 13.9 21.5

14.2 2.1 12.0 7.7

61.1 93.1 9.6 22.9

45.9 24.1 14.3 2.0

4.3 1.4 0.7 16.4

4.4 3.3 1.5 3.2

67.2 10.5 16.6 24.6

263.6 199.4 83.4 130.9

1933

Winter

Total

129.6

20.5

21.1

6.1

62.5

19.4

36.6

2.0

23.3

191.5

2,458.8

429.8

591.6

198.8

802.5

436.2

164.2

80.7

459.2

3,163.0

Notes: (p) indicates a primary cause and (c) indicates a contributing cause. Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

1929

Loans and Investments in Terminal Bank Suspensions, by Cause and Quarter, in Millions of Dollars, Winter 1929 to Winter 1933. Cause of Terminal Suspension Assets and/or withdrawals Assets and liabilities

Assets

Assets (p) and withdrawals (c)

Assets (p) and withdrawals (p)

Withdrawals (p) and assets (c)

Withdrawals

Closure of correspondent

Defalcation

Other causes

Total, terminal suspensions

(a.1)

(a.1.1)

(a.1.2)

(a.1.3)

(a.1.4)

(a.1.5)

(a.2)

(a.3)

(a.4)

(a)

1929

Winter Spring Summer Fall

6.3 5.7 13.3 29.3

2.9 3.3 9.3 8.8

1.0 0.9 1.1 12.0

0.0 0.4 1.7 1.2

2.1 0.6 0.3 3.7

0.2 0.4 1.0 3.6

0.8 2.0 18.1 0.4

1.6 0.6 1.3 0.9

2.6 2.5 7.1 6.5

11.3 10.7 39.7 37.1

1930

Winter Spring Summer Fall

53.6 71.2 55.5 498.5

27.2 26.6 19.2 55.4

9.3 22.4 3.4 265.7

6.5 2.1 17.8 11.0

4.0 7.5 7.8 86.7

6.8 12.6 7.3 79.6

1.1 5.6 0.3 18.6

10.3 6.0 4.5 6.2

12.6 10.2 11.6 34.7

77.7 93.0 71.9 558.0

1931

Winter Spring Summer Fall

124.3 265.9 361.8 573.0

36.2 28.7 54.2 50.3

26.6 10.6 75.5 113.1

17.6 30.4 34.5 42.1

23.1 153.8 118.3 258.7

20.9 42.3 79.2 108.9

0.3 1.6 19.1 29.4

7.0 3.5 11.8 10.9

15.8 39.6 103.1 85.5

147.4 310.7 495.8 698.7

1932

Winter Spring Summer Fall

221.6 211.6 77.6 122.8

55.4 45.8 16.2 47.8

27.1 29.6 16.9 28.6

16.2 3.7 15.3 12.5

75.8 105.5 13.7 31.6

47.1 27.0 15.5 2.3

5.8 1.5 0.8 18.1

5.4 4.2 2.0 3.4

64.0 12.9 19.0 33.6

296.8 230.2 99.4 177.9

1933

Winter

Total

164.6

26.8

27.2

7.9

79.4

23.4

37.7

2.7

28.6

233.6

2,856.6

514.2

671.0

220.8

972.5

478.2

161.3

82.3

489.8

3,589.9

105

Notes: (p) indicates a primary cause and (c) indicates a contributing cause. Source: National Archives and Records Administration (see text for details).

Categories and Causes of Bank Distress, 1929–1933

Table 10.

Number of Temporary Bank Suspensions, by Cause and Quarter, in Thousands of Dollars, Winter 1929 to Winter 1933.

106

Table 11.

Cause of Temporary Suspension Assets and/or withdrawals Assets and liabilities

Assets

Assets (p) and withdrawals (c)

Assets (p) and withdrawals (p)

Withdrawals (p) and assets (c)

Withdrawals

Closure of correspondent

Defalcation

Other causes

Total, terminal suspensions

(a.1)

(a.1.1)

(a.1.2)

(a.1.3)

(a.1.4)

(a.1.5)

(a.2)

(a.3)

(a.4)

(a)

Winter Spring Summer Fall

3 6 11 13

2 3 4 1

0 0 0 2

0 1 0 4

0 0 4 3

1 2 3 3

0 0 6 0

1 2 2 1

5 9 6 6

9 17 25 20

1930

Winter Spring Summer Fall

20 20 14 84

6 4 4 7

0 1 1 25

5 1 0 3

3 9 5 17

6 5 4 32

4 4 0 59

0 6 2 4

8 4 5 43

32 34 21 190

1931

Winter Spring Summer Fall

45 11 38 138

6 3 4 18

10 1 9 23

3 2 4 18

6 3 6 40

20 2 15 39

3 0 3 14

4 0 0 1

11 2 12 36

63 13 53 189

1932

Winter Spring Summer Fall

61 18 21 14

7 3 4 2

12 8 10 5

3 0 0 1

26 5 6 6

13 2 1 0

5 1 9 3

0 0 0 0

15 4 8 4

81 23 38 21

1933

Winter

Total

27

5

3

3

10

6

4

0

17

48

544

83

110

48

149

154

115

23

193

877

Notes: (p) indicates a primary cause and (c) indicates a contributing cause. Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

1929

Deposits in Temporary Bank Suspensions by Cause and Quarter, Winter 1929 to Winter 1933. Cause of Temporary Suspension Assets and/or withdrawals Assets and liabilities

Assets

Assets (p) and withdrawals (c)

Assets (p) and withdrawals (p)

Withdrawals (p) and assets (c)

Withdrawals

Closure of correspondent

Defalcation

Other causes

Total, terminal suspensions

(a.1)

(a.1.1)

(a.1.2)

(a.1.3)

(a.1.4)

(a.1.5)

(a.2)

(a.3)

(a.4)

(a)

1929

Winter Spring Summer Fall

0.9 2.5 5.8 5.4

0.4 2.1 2.4 0.2

0.0 0.0 0.0 2.0

0.0 0.3 0.0 1.3

0.0 0.0 1.9 1.3

0.5 0.1 1.5 0.5

0.0 0.0 2.4 0.0

0.2 0.3 0.5 0.2

1.2 2.4 2.1 2.5

2.3 5.1 10.8 8.0

1930

Winter Spring Summer Fall

8.8 8.1 6.7 37.1

1.1 0.8 1.2 1.4

0.0 0.4 0.1 7.9

1.6 0.9 0.0 0.7

0.5 3.8 1.4 8.5

5.6 2.2 4.0 18.7

0.8 1.0 0.0 15.2

0.0 3.0 0.8 5.6

3.1 2.2 2.8 31.2

12.7 14.2 10.3 89.1

1931

Winter Spring Summer Fall

14.9 3.8 30.3 147.2

2.9 0.4 1.6 13.9

1.7 0.5 3.5 26.1

0.8 1.2 3.4 37.1

3.6 0.7 10.5 41.6

5.9 1.0 11.3 28.4

0.6 0.0 1.8 38.8

1.3 0.0 0.0 0.3

4.3 0.7 5.0 65.3

21.2 4.5 37.2 251.6

1932

Winter Spring Summer Fall

30.3 7.9 10.6 4.9

2.6 0.4 1.6 0.2

8.4 4.4 2.6 2.0

0.6 0.0 0.0 0.1

13.5 2.7 6.2 2.5

5.2 0.3 0.3 0.0

3.7 0.0 3.8 2.0

0.0 0.0 0.0 0.0

4.5 1.3 2.0 0.7

38.5 9.3 16.3 7.7

1933

Winter

Total

11.5

1.2

1.9

0.8

2.5

5.2

0.6

0.0

4.0

16.2

336.7

34.2

61.5

48.9

101.3

90.7

70.8

12.1

135.4

555.1

107

Notes: (p) indicates a primary cause and (c) indicates a contributing cause. Source: National Archives and Records Administration (see text for details).

Categories and Causes of Bank Distress, 1929–1933

Table 12.

Loans and Investments in Temporary Bank Suspensions by Cause and Quarter, Winter 1929 to Winter 1933.

108

Table 13.

Cause of Temporary Suspension Assets and/or withdrawals Assets and liabilities

Assets

Assets (p) and withdrawals (c)

Assets (p) and withdrawals (p)

Withdrawals (p) and assets (c)

Withdrawals

Closure of correspondent

Defalcation

Other causes

Total, terminal suspensions

(a.1)

(a.1.1)

(a.1.2)

(a.1.3)

(a.1.4)

(a.1.5)

(a.2)

(a.3)

(a.4)

(a)

Winter Spring Summer Fall

0.6 1.6 1.5 4.6

0.0 1.0 0.7 0.0

0.0 0.0 0.0 1.9

0.0 0.0 0.0 1.2

0.0 0.0 0.7 0.9

0.5 0.5 0.0 0.5

0.0 0.0 0.0 0.0

0.1 0.3 0.3 0.1

1.0 0.3 0.9 0.7

1.7 2.2 2.6 5.5

1930

Winter Spring Summer Fall

9.2 9.2 6.1 39.8

1.1 1.0 0.9 2.0

0.0 0.5 0.1 9.5

1.8 0.9 0.0 0.8

0.7 4.4 1.6 9.8

5.7 2.4 3.4 17.7

0.7 0.9 0.0 15.4

0.0 3.2 0.9 5.0

3.3 2.3 2.7 39.2

13.2 15.6 9.7 99.5

1931

Winter Spring Summer Fall

16.9 4.1 33.6 162.5

3.0 0.5 1.6 10.5

2.0 0.6 3.8 30.6

0.8 1.2 3.9 42.7

4.0 0.7 11.9 49.0

7.1 1.1 12.4 29.6

0.5 0.0 1.8 42.1

1.3 0.0 0.0 0.3

4.2 0.7 5.2 85.6

22.9 4.8 40.6 290.6

1932

Winter Spring Summer Fall

30.2 9.1 11.4 6.0

2.9 0.4 1.8 0.3

4.4 4.8 2.9 2.7

0.7 0.0 0.0 0.2

16.7 3.5 6.4 2.8

5.6 0.3 0.3 0.0

4.2 0.1 4.1 2.2

0.0 0.0 0.0 0.0

4.6 1.4 2.8 1.4

39.0 10.5 18.3 9.5

1933

Winter

Total

16.3

1.5

2.0

1.0

3.2

8.6

0.6

0.0

3.5

20.5

362.6

29.1

66.0

55.3

116.4

95.9

72.5

11.7

159.9

606.7

Notes: (p) indicates a primary cause and (c) indicates a contributing cause. Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

1929

Categories and Causes of Bank Distress, 1929–1933

109

late 1930s stemmed from fundamental forces rather than the panics that appear to have plagued the banking system at some points in the past.

6. BANK DISTRESS BY CATEGORY, QUARTER, AND FEDERAL RESERVE DISTRICT Recent research emphasizes the importance of analyzing the geographic distribution of the categories of bank suspensions. To facilitate that analysis, this section reports data aggregated at the level of the Federal Reserve District. Tables 14–18 report the numbers of banks in distress by category, quarter, and Federal Reserve District. Table 14 reports the number of banks Table 14.

Number of Banks in Distress by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

2

1929 Winter Spring Summer Fall 1930 Winter Spring Summer Fall

4

5

6

7

2 4

3 5 3 6

3 11 18

32 28 44 15

26 17 30 45

8 16 22 18

37 21 7 97

47 39 11 65

78 61 73 97

8

9

10

11

12

Total

11 8 12 25

21 28 16 25

10 19 30 58

6 1 2 14

5 4 4 3

114 113 154 213

44 37 40 267

39 24 20 79

43 33 41 48

18 6 8 44

6 2 8 10

329 244 240 752

84 26 45 16 58 28 43 104 61 147 95 126

22 14 25 53

12 478 11 388 22 630 48 1,194

2 3 1 7

7 9

6 2 2 11

1931 Winter 2 Spring 7 Summer 1 Fall 36

10 9 36 57

13 23 23 90

38 30 32 30 61 44 88 121

62 18 20 69

134 142 190 264

1932 Winter 12 Spring 8 Summer Fall 1

16 7 2 3

35 13 17 6

33 11 15 12

42 17 20 7

40 18 17 38

167 135 117 131

93 31 22 37

21 25 30 52

82 46 38 68

24 12 5 18

63 38 32 35

628 361 315 408

9

13

22

16

27

148

120

33

78

17

56

540

1933 Winter Total

1

1

3

81 166 260 393 521 590 1,855 1,037 696 854 289 359 7,101

Source: National Archives and Records Administration (see text for details).

110

GARY RICHARDSON

Table 15. Number of Banks in Terminal Suspension by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1 1929

1930

1931

2

Winter Spring Summer Fall

3

1 2

Winter Spring Summer Fall

1 3 1 4

Winter Spring Summer Fall

1 1 25

2 1

4

5

6

3 5 1 4

1 10 15

28 25 30 11

26 15 6 60

7

8

9

10

11

12

Total

22 14 14 30

8 7 6 12

20 25 13 20

10 11 18 43

6

3

5 4 3 1

102 92 96 141

34 29 7 47

54 42 62 78

32 25 31 150

36 22 16 70

27 22 28 38

10 4 8 13

2 2 8 9

229 174 185 489

3 5

7

5 9 15 8

5 7 20 40

10 16 14 54

28 23 48 54

23 26 32 73

46 13 16 40

91 120 154 202

47 11 36 116

19 48 95 78

28 24 41 98

9 7 20 28

9 11 19 36

316 306 496 844

1932

Winter Spring Summer Fall

4 5

5 4 1

12 6 6 2

25 9 13 9

26 13 13 6

18 10 11 29

143 118 102 116

59 26 15 31

19 22 29 52

61 33 32 56

17 6 4 10

46 30 18 29

435 282 244 340

1933

Winter

1

6

4

18

11

20

123

103

31

53

6

44

420

46

96

137

277

356

414

1,485

715

615

623

151

276

5,191

Total

Source: National Archives and Records Administration (see text for details).

experiencing any form of distress. It reveals the district-level distribution of the national-level information reported in Table 6, Column (6). Table 15 reports terminal bank suspensions. It illuminates the district-level distribution of the national-level information reported in Table 6, Column (1). Table 16 reports temporary bank suspensions. It reveals the district-level distribution of the national-level information reported in Table 6, Column (5). Table 17 reports banks consolidating due to ﬁnancial difﬁculties. It reveals the districtlevel distribution of the national-level information reported in Table 6, Column (3). Table 18 reports banks liquidating voluntarily. It reveals the district-level distribution of the national-level information reported in Table 6, Column (2). Tables 19–23 report total loans and investments in millions of dollars for distressed banks by category, quarter, and Federal Reserve District. Table 19 reports the sum of loans and investments in all banks experiencing distress.

Categories and Causes of Bank Distress, 1929–1933

111

Table 16. Number of Banks in Temporary Suspension by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1 1929

1930

1931

1932

1933 Total

2

3

4

Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall

5

6 1

3

1 1

11

1 3 2 1

8 8 9

1

3 7 4 6

3 1

11 12 5 8

1 5 3 112

2

3

6

28

12

17 5 19 35

5 17

6 5 8 9

9 3 4 7

18 12 12 2

18 1 5 3

1 1 1

4 4 1 36

1 6 17

4 45

1

3

2

2 1

11 3

1

8

10

7 4 4 10

13

1 5

12

9

3 1

1 6

10

8

4 3 1 7

1 1 2

6

Winter

7

1

2 8

2

7 16

12

Total 9 17 25 20

1

5

32 34 21 190

2 3 1 18

1 3

63 13 53 189 81 23 38 21

8 3 4 5

1

1

8 1 10 2

1

2

3

9

8

1

17

2

5

48

32

117

83

180

211

50

105

38

31

877

Source: National Archives and Records Administration (see text for details).

It indicates the district-level distribution of the national-level information reported in Table 7, Column (6). Table 20 reports loans and investments of banks suspending terminally. It indicates the district-level distribution of the national-level information reported in Table 7, Column (1). Table 21 reports loans and investments of banks suspending temporarily. It reveals the district-level distribution of the national-level information reported in Table 7, Column (5). Table 22 reports loans and investments of banks consolidating due to ﬁnancial difﬁculties. It reveals the district-level distribution of the national-level information reported in Table 7, Column (3). Table 23 reports loans and investments for banks liquidating voluntarily. Its reveals the district-level distribution of the national-level information reported in Table 7, Column (2). Tables 24–28 report median loans and investments of banks in distress by category, quarter, and Federal Reserve District. The ﬁgures are reported in

112

GARY RICHARDSON

Table 17. Number of Banks Consolidating Due to Financial Difﬁculties by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1 1929

1930

1931

1932

1933 Total

2

3

4

5

6

1

2 2

1

2

Winter Spring Summer Fall

1 2

2 2

Winter Spring Summer Fall

3 6 6 9

5 2

2 2

2 2

4 1 2 4

1

5

5 2 14 11

3 7 6 28

8 8 5 17

5 2

7 2

1

3

21 7 11 4

2 50

Winter Spring Summer Fall Winter Spring Summer Fall

7

Winter 20

7

8

9

10

11

12

Total 2 3 19 29

4 5

6 7

4 5

1 5

1

4 3

10 2 2

1

2

6

7 3 2 8

19

1

38 19 15 52

1 4 7 3

5 3 3 8

23 17 12 21

2

1 7

65 57 58 119

5 1 2 2

2 3 2

1 1 1

1 3

1

7 4 4 3

67 36 21 25

9

3

2

2

5

34

110

79

34

35

659

1 2

1

9 3 6 5

9 3 3 2

1

7 6

4

3 1 11 11 2 1

11

38

6

5

123

55

6

62

47

Source: National Archives and Records Administration (see text for details).

thousands of dollars. Table 24 reports the median for banks experiencing all forms of distress. Table 25 reports the median for banks suspending terminally. Table 26 reports the median for banks suspending temporarily. Table 27 reports the median for banks consolidating due to ﬁnancial difﬁculties. Table 28 reports the median for banks liquidating voluntarily. Tables 29–33 report average loans and investments of banks in distress by category, quarter, and Federal Reserve District. The ﬁgures are reported in thousands of dollars. Table 24 reports the average for all banks experiencing distress. Table 25 reports the average for banks suspending terminally. Table 26 reports the average for banks suspending temporarily. Table 27 reports the average for banks consolidating due to ﬁnancial difﬁculties. Table 28 reports the average for banks liquidating voluntarily. In many cases, the averages reported in Tables 29–33 exceed the medians reported in Tables 24–28, in some case by a substantial margin. This occurs because

Categories and Causes of Bank Distress, 1929–1933

113

Table 18. Number of Banks in Voluntary Liquidation by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1 1929

1930

1931

1932

1933 Total

2

3

4

5

Winter Spring Summer Fall

6 1 1 1 2

Winter Spring Summer Fall

1

Winter Spring Summer Fall

1

Winter Spring Summer Fall

2 1

1

2

1 1

1

Winter

1 1 1

8

5

9

10

11

5

1 4

1

6 4 4 3

7 4 6 3

1 2 1 1

3 2 7 4

4 1

5 2 1 9

3 5 6

7 5 1 3

1 2 1 7

8 1 7 7

4 4

1 6

12

Total

2

1 1 14 23

1

30 17 19 21

2 1 1 5

3

34 12 23 42

6 4 2 6

6 2 1 7

2 3 1

45 20 12 22

7

1 1 2

3

1

3 1 2 1

12 4 1 2

5 2 3 2

5 2 1 3

2

1

4

10

4

1

6

9

2

38

14

55

67

56

25

64

53

17

374

1 5

8

11 3

2

2 3

7

5

1

1

1 2

Source: National Archives and Records Administration (see text for details).

smaller banks failed more often than larger banks and because in a few cases banks that failed were many times the median size. Comparing the averages reported in Tables 29–33 with averages calculated by dividing the total of loans and deposits reported in Tables 19–23 by the number of banks with the number of banks reported in Tables 14–18 reveals an important point. The average calculated directly from the data differs from the ‘average’ calculated by dividing corresponding cells in the tables for investments in banks and numbers of banks. The discrepancy arises because information on loans and investments is missing for some banks. Roughly 4% of all banks did not report this information including: 211 banks suspending terminally, 40 banks suspending temporarily, 15 banks consolidating due to ﬁnancial difﬁculties, and 46 banks liquidating voluntarily. Despite these errant observations, information on loans and investments has advantages. All three St. 6386 forms report these loans and investments

114

Table 19.

Loans and Investments of Banks in Distress in Millions of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

2

6

1,284 6,106

1,018 638 4,988 4,422

246 4,479 10,639

2,036 5,491 22,868 3,743

4,790 988 1,852 16,001

1,176 1,432 3,986 5,103

3,112 4,058 4,126 5,239

4,232 1,977 21,422 60,357

22,173 26,266 14,974 25,574

10,587 9,524 6,648 72,330

17,992 39,330 4,296 46,205

31,182 10,025 23,324 11,060 29,578 9,726 51,086 171,297

1931 Winter 3,323 27,215 12,210 29,354 Spring 10,720 30,049 37,435 60,474 Summer 181 122,180 48,414 194,214 Fall 155,558 128,077 199,346 290,934

11,016 13,544 48,411 78,500

23,830 19,442 4,587 72,607

71,616 410,053 117,152 607,592

1932 Winter 44,222 Spring 139,644 Summer Fall 739

34,321 29,379 1,880 4,996

82,408 14,124 25,528 9,811

68,758 5,751 9,253 27,161

44,854 5,938 17,073 2,207

14,745 3,997 5,702 11,961

1933 Winter

12,657

52,044

10,178

9,795

26,702

1930 Winter Spring Summer Fall

Total

476 33 13,359 1,049 1,631 8,703 241,352

1,494

4

7

8

9

10

11

12

Total

1,018 911 750 10,396

0 0 4,040 463

287 399 3,824 2,997

13,437 14,163 52,198 65,109

4,899 3,854 2,903 13,020

14,176 7,086 10,687 7,371

12,572 742 1,468 10,717

1,413 1,841 6,788 6,822

129,758 138,363 111,170 714,833

23,262 4,268 10,519 83,271

7,422 13,717 24,930 18,377

9,426 4,754 18,208 26,453

5,878 7,103 28,535 29,736

104,200 156,066 49,929 76,953

31,539 7,207 5,228 17,867

3,525 7,011 7,924 11,683

15,517 11,574 6,205 10,322

8,478 4,028 541 2,964

42,617 35,898 18,236 30,049

495,183 420,616 147,499 206,713

87,268

34,663

8,165

16,072

4,662

53,124

316,824

3,855 228,407 6,021 617,581 19,804 637,134 30,973 1,721,423

379,467 633,769 576,698 796,130 345,790 325,533 1,839,631 431,629 143,962 170,925 121,929 264,948 6,030,412

Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

5

1929 Winter Spring Summer Fall

3

Loans and Investments of Banks in Liquidation in Millions of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

2

1929 Winter Spring Summer Fall

3

4

5

6

7 3,786 456 1,310 10,495

531 1,119 1,133 2,934

3,112 3,764 2,388 4,007

17,739 9,285 12,508 8,552 25,755 7,655 39,400 109,627

1,018 638 3,700 1,467

4,318 7,198

1,568 4,344 21,741 1,270

2,377 4,435 11,300 2,565

9,215 7,870 4,655 57,674

13,525 33,362 3,531 30,736

1931 Winter 2,446 Spring Summer 181 Fall 101,132

11,905 10,678 20,953 19,244 22,289 48,174 69,493 42,953 172,726 57,492 127,579 121,867

8,662 9,241 31,180 39,469

19,868 40,794 7,222 178,215 4,178 87,211 43,960 102,910

1932 Winter Spring Summer Fall

35,480 24,489

10,931 7,442 1,690

23,627 5,623 5,431 3,509

11,223 3,343 7,913 24,361

34,200 5,185 9,994 2,207

9,808 96,056 2,648 130,860 4,775 44,469 8,493 71,493

1,494

9,226

23,449

9,376

8,202

1930 Winter Spring Summer Fall

1933 Winter Total

1,201 3,207 285 13,359 1,049 963 1,512 236,224

747 1,900 55,967

26,102

51,773

8

9

10

11

12

Total

1,018 1 109 6,384

287 399 3,810 170

11,320 10,722 39,710 37,132

4,438 3,799 2,265 10,245

8,382 11,286 4,707 631 6,493 1,468 5,924 1,539

402 1,841 6,788 6,583

77,680 92,964 71,922 557,996

15,170 3,321 9,151 40,801

5,533 10,394 22,864 14,493

5,277 3,850 4,192 2,370 11,867 25,743 21,677 9,942

2,258 6,021 18,253 17,423

147,393 310,683 495,800 698,744

24,497 6,841 3,719 17,224

3,404 4,277 7,863 11,683

13,109 4,536 5,329 9,248

7,547 1,061 541 2,213

26,916 33,908 7,680 27,449

296,799 230,213 99,403 177,880

33,092

8,059

12,342

2,636

47,803

233,554

Categories and Causes of Bank Distress, 1929–1933

Table 20.

181,427 424,609 328,161 447,435 239,270 237,131 915,229 294,652 122,588 120,595 70,828 207,991 3,589,915

115

Source: National Archives and Records Administration (see text for details).

116

Table 21.

Loans and Investments of Banks in Temporary Suspension in Millions of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

1931

1933 Total

3

4

Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall

5

6

7 39

7,191

489 2,772

1,004 532

645 313

2,509

230

1,600 2,038 764 3,805

9,292 8,109 2,963 3,704

56 1,379 361 61,099

1,228

2,703

7,483

26,024

8,148 2,647 18,486 68,729

802 5,798

4,552 481 606 3,423

6,595 4,871 5,238 2,052

4,049 106 848 285

173 161 890

1,079

1,168 273 444

389 1,305 1,993 14,555

1,051

24,811

441 19,094

867 6,391

7,842 82,432

5,715 34,383

3,211

3,500

1,359

877 2,333

9,416 1,528

1,160

Winter 35,213

26,296

8,617

8

9

10

11

12

Total 1,688 2,222 2,616 5,463

294 1,147 1,081

910 229 753

413

344 1,532 2,561 985

2,156

13,217 15,593 9,723 99,463

3,098 3,205

490 450 859 10,864

525 5,962

22,934 4,788 40,649 290,563 39,019 10,516 18,260 9,499

319 2,751 1,831 1,691 2,013 2,870 1,758

765 62

359

1,059 533 727 540

628

138

3,773 434 9,314 1,900

147

1,160

538

11,251

971

106

3,307

1,685

1,293

20,458

97,727

72,896

47,653

156,128

84,424

16,274

19,783

18,099

23,560

606,670

Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

1932

2

Loans and Investments of Banks Consolidating Due to Financial Difﬁculties in Millions of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

1931

1932

1933 Total

2

3

4

Winter Spring Summer Fall

83 2,899

1,288 2,955

Winter Spring Summer Fall

179 2,300

3,485 77 21,422 4,390

19,796 20,663 3,401 22,565

29,615

15,310 10,805 52,096 51,491

1,532 15,146 4,154 62,426

5,262 111,215

18,893 9,198

739

Winter Spring Summer Fall Winter Spring Summer Fall

10,720

Winter 157,551

5

6

73

316 1,116

2,551

468

722 349

2,159 90

101

7,350 11,426 13,554 86,636

4,996

57,422 8,501 20,097 6,302

2,267 167,535

7

8

9

10

11

12

Total 316 1,189 9,123 16,949

514 2,889

2,853 1,697

413 3,254

3,958 236

14

455 110

5,305 801 1,266

93

544

11,589

2,303 2,301 285 7,612

6,636

239

34,862 24,391 26,790 55,581

1,026 4,303 11,516 4,648

785 12,071 399 2,020

22,223 229,192 10,513 435,497

298

184 7,211

53,402 298,349 95,067 725,454

55,903 76 1,340 1,640

1,064 732 4,895

69 590

579 640

300

7,807 1,173 1,243 694

151,018 142,260 28,216 17,678

28,595

655

395

211

3,999

44,622

236,531

249,248

32,375

23,108

1,725,267

237 149

390

3,414 562 704 884

1,464 2,813 1,901 8,567

976

1,118 6,320

2,839

1,311 46 36,069 2,901 641

3,007

31,672

8,169

332

725,724

45,551

24,552

28,507

117

Source: National Archives and Records Administration (see text for details).

2,914

Categories and Causes of Bank Distress, 1929–1933

Table 22.

118

Table 23.

Loans and Investments of Banks in Voluntary Liquidation in Millions of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

1931

1933 Total

3

4

5

Winter Spring Summer Fall

6

7

113 31 48 2,005

Winter Spring Summer Fall

191

Winter Spring Summer Fall

877

Winter Spring Summer Fall

269 3,940

33

6

1,848 406 575 370

229 1,019 1,711 421

48 55 82 23

144 45 367 462

428 50

474 149 9 603

452

57 320 53 624

736

942 457

312 948 520 603

316 278 321 45

970 19,695 222 401

92 260 20 358

121

38

62

16,076

268

1,249

9,078

42,550

7,002

100

Winter

874 91 755

1,164 5,277

15,329

3,390

1,720

11

592 151

708 3,840 75

997 12,739 190

10

242

261

440 2,950

9

28 108

56

150

8

173 21 656

Source: National Archives and Records Administration (see text for details).

61

2,187

82 226

12

Total

2,827

113 31 750 5,565

108

3,999 5,415 2,735 1,793

1,597

2,538 686

74 1,469 32 363

4,678 3,760 5,619 6,663

231 185 149 234

302 127 0 612

4,121 383 6

8,348 37,628 1,620 1,656

211

341

29

18,189

5,995

4,494

10,290

108,560

386

842 376

GARY RICHARDSON

1932

2

Median Loans and Investments of Banks in Distress in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

2

Winter Spring Summer Fall

4

5

6

7

8

9

10

11

12

Total

642 1,450

509 638 877 894

123 393 385

92 175 543 176

319 192 160 282

106 199 249 102

105 115 131 97

243 8 20 121

2,020 33

287 399 1,912 1,191

120 127 186 166

910 404 350 661

117 281 381 301

151 400 141 178

214 275 218 273

103 220 121 190

107 124 125 127

136 119 121 127

189 101 150 141

236 921 237 396

148 243 185 199

Winter Spring Summer Fall

238 1,140 1,049 2,796

202 1,716

696 989 10,711 676

1931

Winter Spring Summer Fall

1,661 903 181 2,775

1,446 1,882 1,822 1,180

876 475 1,260 961

654 580 461 630

246 263 320 336

135 510 188 209

245 744 294 239

150 134 150 178

150 155 191 132

124 138 156 134

97 267 376 224

229 263 284 230

211 379 251 282

1932

Winter Spring Summer Fall

1,120 6,942 739

1,686 2,647 940 1,076

792 775 1,339 1,075

500 329 214 441

182 128 305 156

137 83 312 145

291 464 232 220

157 131 139 96

108 182 133 155

123 126 115 91

106 142 140 97

344 232 318 332

244 254 218 168

Winter

1,494

1,164

1,475

184

173

131

181

154

172

128

85

221

173

2,245

1,405

908

523

270

174

277

160

135

125

151

278

224

1933 Total

33

3

119

Source: National Archives and Records Administration (see text for details).

Categories and Causes of Bank Distress, 1929–1933

Table 24.

120

Table 25.

Median Loans and Investments of Banks in Terminal Suspension in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

Winter Spring Summer Fall 285 1,140 1,049 1,512

Winter Spring Summer Fall

2,446

1932

Winter Spring Summer Fall

1933

Winter

Total

3

4

1,201 1,604

6

7

8

9

10

509 638 3,700 734

412 382

77 163 870 122

323 100 506 292

49 190 201 147

105 118 113 105

243 1 8 113

175 281 375 333

163 421 169 176

240 172 210 261

150 271 121 211

97 130 125 116

131 86 121 130

11

12

Total

287 399 3,810 170

117 134 181 166

370 160 150 73

201 921 237 463

161 243 181 192

481 1,716

676

601 294 457 236

1,533 1,296 1,796 896

911 545 2,320 961

609 575 463 518

266 269 242 336

135 472 193 209

218 714 273 230

149 164 162 179

144 128 186 132

107 132 168 131

245 292 373 203

229 263 267 221

214 349 238 250

4,559 4,550

1,910 1,605 1,690

903 714 974 1,755

443 329 167 433

258 128 268 156

155 64 413 102

301 523 230 221

129 140 149 96

144 151 136 155

142 122 115 105

119 142 140 100

305 232 245 356

224 246 201 165

1,494

1,284

6,290

221

108

140

179

162

175

126

270

211

177

2,446

1,312

980

461

273

185

267

164

133

121

192

260

215

181 3,324

747 1,900

5

Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

Winter Spring Summer Fall

1931

2

Median Loans and Investments of Banks in Temporary Suspension in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

1931

1932

1933 Total

2

3

4

Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall

3,595

489 2,772

5

319 532

323 313

530

230

61 299 1,993 262

125 260 113 175

284 609 575 402

56 231 132 178

1,051

428

348

183

380

273 411 617 413

123 133

521 108 97 506

264 359 300 1,026

203 106 123 111

672 1,575

1,421 351

3,211

1,216

679

438 2,333

165 119

1,160

814

39

1,168 273 444

867 760

973

8

287

441 973

2,809

7

173 161 445

2,766

Winter

6

9

10

88 573 1,081

10 5 230

206

172 209 408 134

404

191 370 326 207

253 204

245 181 859 292

525 750

217 212 423 362 271 301 261 198

159 356 294 253 242 192 879

11

12

Total 230 146 234 235

271 62

359

73 217 168 130

628

138

357 434 521 950

147

580

177

171

112

106

152

843

231

164

991

287

279

414

175

253

185

285

434

274

121

Source: National Archives and Records Administration (see text for details).

Categories and Causes of Bank Distress, 1929–1933

Table 26.

122

Table 27. Median Loans and Investments of Banks Consolidating Due to Financial Difﬁculties in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1 1929

1930

1931

1933 Total

3

4

Winter Spring Summer Fall

83 1,450

644 1,478

Winter Spring Summer Fall

90 1,150

524 77 10,711 994

1,975 2,743 348 1,144

1,975

1,358 5,403 2,483 1,985

321 427 581 1,073

1,004 55,608

1,862 4,599

739

Winter Spring Summer Fall Winter Spring Summer Fall

903

Winter 1,120

5

6

73

158 558

2,551

234

52 175

1,080 45

101

898 590 1,024 1,318

1,076

792 968 1,987 1,075

1,134 1,875

7

8

9

10

11

12

Total 158 490 258 265

216 425

490 314

150 216

3,958 52

14

75 55

187 401 633

93

272

753

103 300 143 449

222

239

187 226 238 400

1,026 146 1,340 100

134 1,765 174 134

426 1,308 312 833

149

184 450

321 660 656 807

5,587 76 670 820

532 211 2,448

69 590

579 200

98

300

828 278 284 239

585 412 670 300

874

328

197

106

347

422

841

868

203

340

435

237 75

390

184 167 101 148

58 543 798 4,284

976

131 546

305

377 46 464

641 180

190

Source: National Archives and Records Administration (see text for details).

343

166

348

261

384

184

217

GARY RICHARDSON

1932

2

Median Loans and Investments of Banks in Voluntary Liquidation in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

1931

1932

1933 Total

2

3

4

5

Winter Spring Summer Fall

6 113 31 48 1,003

Winter Spring Summer Fall

191

Winter Spring Summer Fall

877

Winter Spring Summer Fall

135 3,940

Winter 216

190

309

414

Total

1,414

113 31 48 40

33 203 62 92

48 27 82 23

58 45 64 73

46 50

62 75 9 49

71

57 160 53 73

80

66 50

31 26 520 41

37 1,469 32 31

842 188

26 35 321 22

123 9,847 60 200

24 130 20 12

61

55 49 75 25

41 64

2,061 44

52

6

50 62 97 31

38

31

141

55

27

43

29

50

44

47

100

41

56

41

108

58

27

87 21 656

57

108

58 70 80 56

799

48 78 87 60

38

123

Source: National Archives and Records Administration (see text for details).

61

101 81

82 33

12

108 87 144 58

874 46

1,164

11

6

75

755

10

592 45

56

997 12,739 190

9

31

42 134

220 1,318

8

14 6

33

150

7

Categories and Causes of Bank Distress, 1929–1933

Table 28.

124

Table 29.

Average Loans and Investments of Banks in Distress in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

2

Winter Spring Summer Fall

4

5

6

642 1,527

509 638 1,663 1,106

123 407 591

120 323 1,633 288

3,168 1,642 713 1,421

286 454 950 761

7

8

9

10

11

12

435 247 231 516

168 239 332 213

156 145 258 218

409 1,008 430 733

405 395 417 532

239 307 249 646

Total

255 228 58 226

2,020 46

287 399 1,912 999

210 215 614 368

126 168 145 167

338 221 267 154

698 124 183 249

235 921 970 758

405 579 481 969

238 4,453 1,049 2,901

326 26,817

846 989 10,711 6,706

1931

Winter Spring Summer Fall

1,661 1,531 181 4,575

2,721 3,339 3,394 2,247

1,018 1,628 2,105 2,240

815 1,951 3,237 3,344

380 467 1,126 649

391 1,080 229 1,084

543 2,929 630 2,310

277 267 270 566

285 236 242 193

214 176 298 210

267 546 1,141 561

350 547 900 659

487 1,617 1,024 1,451

1932

Winter Spring Summer Fall

3,685 23,274 739

2,145 4,197 940 1,665

2,355 1,086 1,502 1,635

2,084 523 617 2,263

1,094 349 899 368

410 250 356 315

632 1,165 427 592

350 249 238 483

168 280 264 225

194 257 163 154

353 336 135 174

676 945 570 859

804 1,192 473 512

1,494

1,406

4,003

485

612

1,068

598

301

255

214

311

966

606

5,198

3,864

2,262

2,106

675

633

1,039

427

209

216

448

768

889

Total

Winter

Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

Winter Spring Summer Fall

1933

33

3

Average Loans and Investments of Banks in Terminal Suspension in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

Winter Spring Summer Fall Winter Spring Summer Fall

285 4,453 1,049 1,512

Winter Spring Summer Fall

2,446

1932

Winter Spring Summer Fall

1933

Winter

1931

Total

2

3

4

1,201 1,604

509 638 3,700 734

6

7

8

9

10

432 480

121 310 2,174 141

473 152 437 552

106 224 189 267

156 151 184 200

255 1 16 177

354 525 776 994

436 1,150 504 668

335 313 422 512

290 342 247 731

123 181 142 148

310 214 232 156

11

12

Total

287 399 3,810 170

214 214 764 323

1,129 158 183 118

201 921 970 823

348 544 397 1,167

481 47,245

11,193

594 493 807 321

2,381 2,749 3,475 1,437

1,186 1,393 3,068 2,407

776 2,190 3,675 2,299

394 370 974 541

442 556 261 1,099

458 1,510 581 509

323 302 286 352

291 217 243 186

188 175 289 221

428 395 1,287 355

251 547 961 484

475 1,032 1,020 832

8,870 8,163

2,186 1,861 1,690

1,969 937 905 1,755

449 371 609 2,707

1,315 399 769 368

545 331 434 293

681 1,118 436 622

415 263 248 556

179 194 271 225

215 137 167 165

444 177 135 221

585 1,130 427 947

685 831 407 525

1,494

1,538

5,862

521

746

1,305

424

321

260

233

439

1,086

557

4,652

4,470

2,486

1,695

682

661

645

418

200

205

502

785

722

181 4,397

747 1,900

5

125

Source: National Archives and Records Administration (see text for details).

Categories and Causes of Bank Distress, 1929–1933

Table 30.

126

Table 31.

Average Loans and Investments of Banks in Temporary Suspension in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

1931

1933 Total

3

4

Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall

5

6 39

3,595

489 2,772

323 313

502

230

229 509 255 381

845 676 593 463

56 276 120 555

409

450

267

2,169

479 529 973 1,964

160 341

506 160 151 489

366 406 437 1,026

238 106 170 95

360

1,168 273 444

97 326 1,993 404

1,051 441 3,182

867 1,278

1,307 4,849

1,429 764

3,211

1,167

679

438 2,333

856 509

1,160

3,913

2,191

1,077

8

335 532

173 161 445

4,135

Winter

7

9

10

98 573 1,081

303 76 188

206

172 219 640 164

431

426 459 486 532

443 200

245 150 859 604

525 1,987

364 368 767 1,537 488 457 481 475

159 344 305 338 252 319 879

11

12

Total 281 247 291 420

255 62

359

132 178 182 135

628

138

472 434 931 950

147

580

179

1,250

162

106

220

843

259

465

3,054

623

662

897

412

332

230

489

760

729

Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

1932

2

Federal Reserve District 1 1929

1930

1931

1932

1933 Total

2

3

4

Winter Spring Summer Fall

83 1,450

644 1,478

Winter Spring Summer Fall

90 1,150

871 77 10,711 1,098

6,599 3,444 567 2,507

5,923

3,062 5,403 3,721 4,681

511 2,164 692 2,230

1,052 55,608

2,699 4,599

739

Winter Spring Summer Fall Winter Spring Summer Fall

1,531

Winter 7,878

5

6

73

158 558

2,551

234

144 175

1,080 45

101

919 1,428 2,711 5,096

1,665

2,734 1,214 1,827 1,576

1,134 3,351

7

8

9

10

11

12

Total 158 396 537 628

171 722

476 242

138 651

3,958 59

14

152 55

531 401 633

93

272

1,931

329 767 143 952

349

239

942 1,355 1,786 1,069

1,026 1,076 1,645 1,549

157 4,024 133 289

966 13,482 876 20,738

149

184 1,202

822 5,234 1,639 6,200

11,181 76 670 820

532 244 2,448

69 590

579 213

300

1,115 293 311 231

2,254 4,184 1,411 707

3,177

328

197

106

800

1,487

2,150

3,195

952

680

2,679

237 75

390

379 187 117 177

163 938 634 4,284

976

160 1,053

710

437 46 3,279 264 641

273

880

166

5,998

949

486

402

620

127

Source: National Archives and Records Administration (see text for details).

1,361

Categories and Causes of Bank Distress, 1929–1933

Table 32. Average Loans and Investments of Banks Consolidating Due to Financial Difﬁculties in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933.

128

Table 33.

Average Loans and Investments of Banks in Voluntary Liquidation in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

1931

1933 Total

3

4

5

Winter Spring Summer Fall

6 113 31 48 1,003

Winter Spring Summer Fall

191

Winter Spring Summer Fall

877

Winter Spring Summer Fall

135 3,940

33

131

56

Winter

75 33

150

220 983

997 12,739 190

755

2,190

678

430

8

11

Total

6

308 102 192 123

38 255 342 140

48 27 82 23

48 45 61 116

107 50

151

57 160 53 89

105

37 1,469 32 73

842 188

61

40 70 321 22

194 9,847 74 200

31 130 20 119

38

31

1,786

67

125

193

774

135

61

95

363 98

82 38

12

592 50

188 91

Source: National Archives and Records Administration (see text for details).

10

48

45 190 520 201

87 21 656

9

14 36

95 75 9 75

874 46

1,164 1,055

177 960

7

1,414

113 31 150 253

108

138 338 182 94

799

151 342 255 167

64

58 62 75 39

50 64

2,061 128

102

6

232 1,980 162 83

42

49

29

606

107

94

686

332

GARY RICHARDSON

1932

2

Deposits of Banks in Suspension in Millions of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

2

1929 Winter Spring Summer Fall 1930 Winter Spring Summer Fall

3

4

6

7

1,851

228 3,940 7,061

4,256 11,905 41,565 1,163

8,512 3,985 5,233 17,430

1,662 1,537 963 3,121

3,291 4,051 3,538 4,765

2,409 5,438 14,057 2,925

8,361 8,771 6,131 62,203

12,535 35,460 3,692 39,886

24,879 9,773 18,860 8,998 27,972 6,896 41,579 151,804

1,185 1,458 1,170 2,136 2,707 13,412 916 2,036 19,114 227,337

560 1,550 41,311

5

8

9

10

11

1,257 4,072 5,991 10,094

637 1,096

12

Total

114

574 2,479 3,606 123

21,375 30,810 66,006 47,855

4,755 4,175 2,312 11,843

9,304 12,006 5,602 519 7,991 1,164 5,691 3,153

602 1,609 6,294 6,890

87,890 104,394 79,460 613,736

4,543 3,860 3,772 1,755 13,820 26,798 21,589 18,103

2,316 5,332 18,828 22,095

154,919 282,093 533,356 740,125

1931 Winter 2,312 Spring Summer 241 Fall 116,586

10,231 17,250 67,618 74,291

8,841 20,305 14,545 52,824 34,349 219,511 96,057 103,157

7,972 7,527 33,758 69,781

19,620 49,005 8,309 156,194 3,844 81,367 18,496 143,956

18,953 2,729 9,568 40,491

6,961 11,856 23,653 15,523

1932 Winter Spring Summer Fall

36,009 36,268

12,456 5,098 1,508

21,646 4,159 3,335 1,967

9,683 4,388 5,638 19,219

42,112 3,578 9,279 1,446

10,030 94,966 2,763 108,960 3,607 45,227 7,751 51,119

24,889 4,829 4,060 14,335

3,179 5,206 6,711 9,237

12,450 3,965 4,724 8,799

7,150 926 373 1,930

26,809 28,468 15,296 22,799

301,377 208,609 99,758 138,602

1,091

5,885

14,735

7,683

6,833

15,259

37,107

6,315

14,787

2,974

46,138

207,676

1933 Winter Total

48,868

Categories and Causes of Bank Distress, 1929–1933

Table 34.

228,655 423,710 246,361 471,731 278,980 240,140 928,111 341,715 127,373 138,450 82,558 210,257 3,718,041

129

Source: National Archives and Records Administration (see text for details).

130

Table 35.

Deposits of Banks in Terminal Suspension in Millions of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

2

1929 Winter Spring Summer Fall 1930 Winter Spring Summer Fall 1931 Winter Spring Summer Fall

1933 Winter Total

4

6

7

1,851

49 3,814 6,061

4,256 11,905 35,470 1,163

7,207 2,984 4,407 14,091

708 1,234 963 2,948

3,291 3,780 2,350 3,755

1,257 1,775 3,420 7,570

2,409 4,356 13,795 2,513

8,000 7,660 3,993 50,268

11,095 33,738 2,857 37,456

15,949 11,268 24,620 38,163

9,678 7,831 6,604 98,876

8,841 19,359 14,545 46,233 33,551 146,244 90,939 102,353

6,816 7,527 28,292 33,824

17,123 41,349 8,309 153,801 3,844 64,448 12,031 83,350 6,168 84,325 2,183 104,368 3,180 40,666 5,452 49,418

1,185 1,458 1,170 2,136 2,707 13,412 916 1,615 9,030 224,424 2,312 241 93,628

10,231 17,250 67,174 52,889

560 1,550 41,311

5

33,244 36,268

9,482 5,098 1,508

20,612 4,159 3,335 1,967

9,683 2,771 5,638 18,146

34,516 3,578 8,407 1,446

1,091

5,885

14,735

7,614

5,546

14,707

41,237

8

9

10

11 637

12

Total

114

574 2,479 3,606 123

19,115 25,662 55,200 39,810

4,385 4,175 1,983 9,487

8,932 11,137 4,095 474 5,303 1,164 4,773 1,402

344 1,609 6,294 6,890

75,196 90,169 69,145 524,594

12,372 2,729 8,798 35,859

5,215 10,221 21,766 12,822

4,543 3,268 3,772 1,297 10,856 26,046 18,693 8,665

2,316 5,332 18,211 15,947

133,745 271,015 429,473 561,001

21,288 4,756 3,305 14,066

3,179 3,658 6,711 9,237

11,625 3,506 4,036 8,393

6,480 926 373 1,721

23,043 28,080 6,283 21,057

263,645 199,350 83,443 130,903

35,969

6,315

12,129

1,713

44,579

191,520

192,850 395,556 239,410 385,608 209,797 210,937 781,652 267,985 112,331 114,679 65,417 186,767 3,162,988

Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

1932 Winter Spring Summer Fall

3

Deposits of Banks in Temporary Suspension in Millions of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

1931

1932

1933 Total

2

3

4

5

Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall

421 2,913

1,082 262 412

22,957

444 21,402

798 5,119

947 6,591 73,267 803

2,765

2,974

1,034

10,084

1,440 1,722 835 2,430

8,929 7,591 3,351 3,416

95 1,166 292 52,927

1,156

2,497

6,581

5,466 35,958

6,465

7,656 2,393 16,919 60,606

770 4,632

3,862 580 427 2,299

10,641 4,591 4,561 1,701

3,601 73 755 269

6,096

361 1,111 2,137 11,935

7,595 872

28,154

6,950

8 954 303

1,072

35,806

7 1,306 1,001 826 3,339

179 126 1,000

1,618

Winter

6

173

69

1,287

552

7,632

1,138

86,124

69,183

29,204

146,459

73,729

10

11

271 1,188 1,010

2,297 2,571 2,523

1,096

371

372 1,506 2,688 918

868 46 1,751

12,694 14,225 10,315 89,142

2,963 2,896

592 459 752 9,437

617 6,148

21,173 4,487 37,208 251,588 38,536 9,259 16,316 7,699

329 2,356 1,746 1,635 1,887 2,701 1,549

15,043

12

Total 2,260 5,148 10,806 8,045

258

825 458 688 406

670

209

3,766 388 9,013 1,742

2,659

1,261

1,559

16,156

23,771

17,141

23,490

555,054

131

Source: National Archives and Records Administration (see text for details).

9

Categories and Causes of Bank Distress, 1929–1933

Table 36.

132

Table 37.

Median Deposits of Banks in Suspension in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

Winter Spring Summer Fall

1,170 1,068

4

5

6

7

8

9

10

11

12

Total

114 180 429 77

267 200 180 285

120 143 165 112

117 118 125 124

82 139 190 146

81 1,096

420

114 285 360

91

86 307 1,803 123

135 149 215 171

Winter Spring Summer Fall

2,312

110 279 400 219

136 351 148 145

219 270 192 243

148 191 112 179

97 132 108 105

124 83 85 104

357 84 109 114

258 805 189 400

150 244 170 178

558 2,087

675

593 350 517 222

1,585 1,316 1,619 991

772 371 1,394 656

545 558 386 517

194 238 260 293

110 404 124 166

214 557 245 193

135 141 132 143

158 133 150 117

97 124 142 126

264 125 301 202

231 208 208 194

184 270 220 229

1932

Winter Spring Summer Fall

3,886 8,808

1,048 1,484 1,508

807 591 512 983

380 302 135 371

156 112 194 136

281 111 250 95

240 371 195 185

131 99 117 106

160 140 122 113

112 89 112 93

107 140 88 83

291 205 236 286

199 226 170 134

1933

Winter

1,091

711

3,435

141

130

127

142

152

169

108

201

201

147

2,736

1,079

683

409

228

166

229

142

127

115

145

223

189

241 2,803

121

560 1,550

Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

2,707 1,160 916 2,436

Total

3

327 150

Winter Spring Summer Fall

1931

2

Median Deposits of Banks in Terminal Suspension in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

Winter Spring Summer Fall

1,170 1,068 2,707 1,160 916 1,790

Winter Spring Summer Fall

2,312

1932

Winter Spring Summer Fall

1933

Winter

Total

3

4

5

6

7

8

9

10

11

12

Total

420

114 180 462 77

267 200 189 277

92 100 165 76

117 124 125 112

82 116 141 119

81

49 335 360

57

86 307 1,803 123

120 144 193 146

133 291 309 228

148 399 162 142

213 143 191 226

163 179 121 176

92 132 101 99

130 69 74 99

393 90 109 88

172 805 189 425

150 225 166 165

327 150

Winter Spring Summer Fall

1931

2

694 1,261

675

593 280 610 194

1,259 1,316 2,022 882

772 371 1,893 683

501 558 370 478

194 238 238 293

108 404 124 141

198 562 226 185

132 141 116 160

145 133 147 108

97 124 133 115

264 156 276 172

231 208 205 191

180 280 205 205

4,639 8,808

1,711 963 1,508

974 591 532 983

379 212 135 317

157 112 210 136

100 89 288 66

234 376 178 182

113 103 118 105

160 131 122 113

113 89 112 93

105 140 88 80

267 193 173 215

185 208 152 131

1,091

829

3,435

148

110

116

142

168

169

99

251

184

153

2,554

1,141

692

373

224

152

216

138

122

110

127

208

180

241 2,803

560 1,550

133

Source: National Archives and Records Administration (see text for details).

Categories and Causes of Bank Distress, 1929–1933

Table 38.

134

Table 39.

Median Deposits of Banks in Temporary Suspension in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

1931

1933 Total

3

4

Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall

2,780

421 2,913

5

387 303

65 271 2,137 192

134 253 135 161

264 533 401 372

95 196 106 184

947

421

324

154

354

282 331 558 362

160 113

439 122 58 269

256 327 291 851

163 73 95 106

644 1,992

1,441 285

2,765

1,096

517

402 1,618

156 80

1,072

613

292 501 826 392

1,082 262 412

798 522

1,058

8

394

444 1,058

2,773

7

179 126 500

2,792

Winter

6

173

69

644

180

142

113

908

237

274

393

166

Source: National Archives and Records Administration (see text for details).

9

10

11

75 594 1,010

298 224 181

1,096

185

186 209 501 119

357 46 340

180 195

296 108 752 268

165 296 277 252 231 225 774

241

12

Total 374 233 390 252

258

206 317 311 211

617 340 328 388

202 252 419 325

51 172 158 101

670

209

496 871

257 301 264 187

147

630

226

147

177

275

340

263

GARY RICHARDSON

1932

2

Average Deposits of Banks in Suspension in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

Winter Spring Summer Fall

1,170 1,068 2,707 4,471 916 2,731

Winter Spring Summer Fall

2,312

1932

1933 Total

3

4

5

6

7

463

114 358 415

213 518 990 106

277 462 876 662

395 292

Winter Spring Summer Fall

1931

2

121

560 1,550

8

9

10

11

12

Total

327 266 271 461

166 192 161 266

165 145 222 227

140 254 230 205

106 1,096 74

144 826 1,803 123

218 305 528 301

300 1,075 369 690

385 364 424 489

289 300 204 575

125 199 128 152

312 193 236 129

924 104 145 172

201 805 899 692

337 585 390 904

509 37,890

8,262

602 1,733 937 325

241 3,761

2,117 2,464 3,111 1,615

982 970 2,290 1,601

725 2,010 2,830 2,509

319 301 925 623

385 639 256 364

455 1,270 465 603

251 228 239 302

278 224 230 182

162 157 292 189

351 195 1,276 394

257 485 941 567

418 873 848 789

Winter Spring Summer Fall

7,202 7,254 .

1,330 3,167 1,508

1,546 693 482 983

388 439 434 1,922

1,138 275 580 241

371 230 256 215

597 845 394 440

332 186 203 422

167 217 231 178

183 110 131 147

397 154 93 175

496 918 546 735

591 724 352 387

Winter

1,091

942

3,684

404

496

616

390

347

204

217

338

942

453

4,083

3,821

1,735

1,592

601

497

559

371

193

193

436

692

622

135

Source: National Archives and Records Administration (see text for details).

Categories and Causes of Bank Distress, 1929–1933

Table 40.

136

Table 41.

Average Deposits of Banks in Terminal Suspension in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

Winter Spring Summer Fall

1,170 1,068

4

Winter Spring Summer Fall

2,312

1932

Winter Spring Summer Fall

1933

Winter

6

7

8

9

10

11

463

213 518 1,223 106

328 230 315 470

101 176 161 268

165 151 181 188

140 177 190 176

106

49 381 404 308 511 666 867

326 1,205 408 797

307 289 404 496

302 313 220 659

122 199 124 136

12

Total

57

144 826 1,803 123

210 295 594 286

331 186 189 126

1,114 118 145 108

172 805 899 766

334 534 382 1,084

538 44,885

8,262

602 484 985 314

2,046 2,464 3,359 1,322

982 970 2,397 1,684

717 2,010 3,047 1,931

310 301 884 463

381 639 256 301

465 1,303 421 415

269 248 251 309

274 213 229 164

162 157 265 191

363 216 1,302 309

257 485 958 443

433 900 871 666

8,311 7,254

1,896 1,275 1,508

1,718 693 556 983

387 308 434 2,016

1,328 275 647 241

343 243 289 188

594 892 399 433

367 190 220 454

167 166 231 178

191 106 126 150

381 154 93 172

501 936 349 726

609 715 342 387

1,091

981

3,684

423

504

735

338

360

204

229

286

1,013

460

4,192

4,120

1,814

1,418

596

529

533

380

183

185

439

687

617

241 3,745

560 1,550

5

Source: National Archives and Records Administration (see text for details).

GARY RICHARDSON

2,707 4,471 916 2,258

Total

3

395 292

Winter Spring Summer Fall

1931

2

Average Deposits of Banks in Temporary Suspension in Thousands of Dollars by Quarter and Federal Reserve District, Winter 1929 to Winter 1933. Federal Reserve District 1

1929

1930

1931

1932

1933 Total

2

3

4

Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall Winter Spring Summer Fall

3,361

421 2,913

5

206 431 278 270

812 633 670 427

95 233 97 473

385

416

235

646

450 479 890 1,732

154 272

429 193 107 328

626 383 380 851

212 73 151 90

469

1,082 262 412

90 278 2,137 332

947 798 1,024

1,099 4,310

1,367 799

2,765

991

517

402 1,618

690 291

1,072

2,346

869

8 318 303

444 3,567

3,581

7 326 501 826 477

179 126 500

3,826

Winter

6

173

69

644

184

848

190

2,691

591

374

823

354

10

11

90 594 1,010

383 321 280

1,096

185

186 215 672 153

289 46 350

409 418 516 472

423 181

296 153 752 524

617 2,049

336 345 702 1,345 494 403 429 385

165 295 291 327 236 300 774

313

12

Total 323 368 432 402

258

118 153 172 102

670

209

471 388 901 871

177

630

312

376

243

451

758

647

137

Source: National Archives and Records Administration (see text for details).

9

Categories and Causes of Bank Distress, 1929–1933

Table 42.

138

GARY RICHARDSON

for banks on the date of a change in status (i.e. suspension, liquidation, or consolidation). This information reveals, therefore, the relative signiﬁcance of different categories of distress, as measured by the productive assets of the banks involved. Scholars studying the macroeconomic effects of loans and investments should treat these data in the same manner as they treat data from the previous Tables 5–7. For the years 1927–1928 and 1933–1936, rely on FRB’37. For the period 1921–1926, interpret results with caution. For the period 1929 to winter 1933, use this table to separate loans and investments in suspending banks into their total and temporary components. Scholars studying data on deposits, reported in Tables 34– 42, should keep in mind an additional caveat. Form St. 6386a, which reported bank consolidations, and Form 6386c, which reported voluntary liquidations, did not contain a space for information on deposits. On occasion, Federal Reserve agents included this information, but the coverage was fragmentary. For that reason, the tables that follow include disaggregated data neither on deposits in banks consolidating due to ﬁnancial difﬁculties nor liquidating voluntarily. The tables only include information on banks in suspension. Tables 34–36 report total deposits of banks in suspension. Tables 37–39 report median deposits of banks in suspension. Tables 40–42 report average deposits at banks in suspension.

7. DISCUSSION The evidence introduced in this chapter has the potential to resolve many debates concerning the causes and consequences of the banking crises during the Great Depression (see for example, Richardson, 2006b, 2006c, 2006d, 2007; Richardson & Chung, 2006a, 2006b; Richardson & Van Horn, 2007). Academic debate continues because scholars lack detailed, comprehensive data on the nature and timing of events. The St. 6386 database provides such evidence. Quarterly series of data constructed from the St. 6386 database make this information available to scholars studying the macroeconomic aspects of the contraction. These series supplement those published in the Federal Reserve Bulletin of 1937, enabling scholars to make important distinctions between different forms of bank distress. For example, scholars can now distinguish temporary from terminal suspensions. These two types of suspensions have different macroeconomic implications. Temporary suspensions indicate

Categories and Causes of Bank Distress, 1929–1933

139

periods of illiquidity when solvent banks temporarily suspend the conversion of deposits to currency and when depositors demand for currency to increase quickly for the ﬁnancial system to respond. Temporary suspensions reﬂect periods where the monetary problems described by Friedman and Schwartz (1963) afﬂict the economy. Terminal suspensions indicate periods when banks cease operations, depositors lose access to their savings, and borrowers lose access to sources of credit. Terminal suspensions reﬂect periods of disintermediation where the nonmonetary dynamics outlined by Bernanke (1983) afﬂict the economy. The new series also enables scholars to observe events that could not be observed in the past. For example, now scholars can observe consolidations due to ﬁnancial difﬁculties (usually consummated quickly in a rush to avoid suspension). The concentration of these consolidations during the fall of 1931 may alter interpretations of that event. Could it be possible, for example, that by forcing banks to consolidate, the banking crisis following Britain’s departure from the gold standard reduced the competition among ﬁnancial institutions, raising interest rates on commercial loans, even though the banks remained open for business? The author’s goal is to disseminate the new data as widely as possible and quickly as practicable. The quarterly series introduced in this chapter will be posted at the Inter-University Consortium for Social and Political Research (ICPSR) and the author’s web page (currently at the University of California in Irvine). The computerization, cleaning, and processing of the newly recovered data series continues. The endeavor consumes considerable resources and takes considerable time. The author has recently received a three-year grant to assist with its entry, tabulation, and dissemination of the data. The grant runs through the academic year 2008–2009. By the end of the grant period, the author hopes to create, clean, and double-check series reporting monthly and weekly data on the categories and causes of bank distress. Further grants may enable the author to process the entire body of microdata and to disseminate the original archival forms.

NOTES 1. Memo in the ﬁrst ﬁle in the ﬁrst box of St. 6386 forms in the National Archives. The card record, a compact method of recording information compatible with tabulating machines and suitable for long-term storage, was never completed. Plans called for cards to describe each bank suspension, liquidation, merger, reopening, and other change of status (e.g. withdrawal from Federal Reserve membership) which occurred in the United States between 1929 and 1933. Cards would also report

140

GARY RICHARDSON

the cause of each suspension and describe each consolidation due to ﬁnancial difﬁculties. Work had barely begun when the Japanese bombed Pearl Harbor. The Federal Reserve prioritized wartime tasks. Historical studies were forsaken. The Division of Bank Operations cleared its decks by sending records from the last decade to the Board’s Central Subject File, including those related to the Division’s inchoate endeavor to encode data from the 1930s. The Division never returned to the task (see Richardson, 2006a for details). 2. Reprints of these series (or cross-tabulations based upon them) appeared in Banking and Monetary Statistics (Board of Governors of the Federal Reserve System, 1943) and Historical Statistics of the United States (United States, Bureau of Census, 1975). Precursors to these series appeared in various publications of the Board of Governors and in Bank Suspension in the United States (Goldenweiser et al., 1931). Studies of the banking panics published prior to 1938 employ these predecessor series. 3. The St. 6386 forms were the basis of the Federal Reserve’s bank data-collection system from 1929 to 1933. All facts and ﬁgures about bank distress published by the Board of Governors or the Federal Reserve district banks and much of the material published by state banking authorities originated with this source. 4. The surviving forms may be found in the National Archives (1930–1933). The forms are ﬁled alphabetically by state, name of town or city, and name of bank. Multiple entries for individual banks appear in chronological order. 5. Board of Governors of the Federal Reserve System (1929, 1930, 1931) and Young (1929). 6. Note that a sixth option existed. Agents completing St. 6386b forms could check a box entitled ‘‘other cause’’ and add a description of the phenomenon. The form also contained space for written remarks. About one-third of all forms contained such additional and/or extended comments. 7. NARA, Bank Changes, Mayo’s Money Exchange Bank, San Antonio, TX, 24 February 1931. 8. NARA, Bank Changes, Mecca Bank, Mecca, IN, 12 October 1931. 9. NARA, Bank Changes, First National Bank, Bradley Beach, NJ, 24 December 1931. 10. NARA, Bank Changes, Central National Bank, Bartleaville, OK, 22 March 1930. 11. NARA, Bank Changes, First National Bank, Bixby, OK, 18 February 1929. 12. NARA, Bank Changes, Central Fairﬁeld Trust Company, Norwalk, CT, 1 December 1930. 13. NARA, Bank Changes, Argos State Bank, Argos, IN, 16 January 1930. 14. NARA, Bank Changes, Gunnison Valle Bank, Gunnison, UT, 18 March 1933. 15. NARA, Bank Changes, Polk County National Bank, Bartow, FL, 28 June 1929 and Polk County National Bank, Bartow, FL, 8 November 1929. 16. NARA, Bank Changes, W.T. Mumme, Sandia, TX, 10 December 1930. 17. NARA, Bank Changes, McKeyes, Lawton, MI, 9 May 1932. 18. NARA, Bank Changes, Commercial Bank, Wapellow, IA, 13 October 1931. 19. NARA, Bank Changes, Farmers and Merchants Bank, Scranton, IA, 18 November 1930. 20. NARA, Bank Changes, Citizens Bank, North Adams, MI, 24 July 1930.

Categories and Causes of Bank Distress, 1929–1933

141

21. Katherine Tunis, head of the research library of the Board of Governors, reports that the Board’s library retained copies of all publications from the state banking departments until the late 1970s, when to conserve shelf space, the library sent all these publications to the libraries of the Federal Reserve district banks. 22. NARA, Bank Changes, Bank of Tennessee, Nashville, TN, 7 November 1930; American Exchange Trust Company, Little Rock, AR, 17 November 1930. 23. NARA, Bank Changes, Bank of United States, New York, NY, 11 December 1930. 24. NARA, Bank Changes, Commerce Trust and Savings Bank, Chicago, IL, 28 May 1931; South Side Savings Bank and Trust Company, Chicago, IL, 6 June 1931; Inland-Irving National Bank, Chicago, IL, 8 June 1931; Washington Park National Bank, Chicago, IL, 8 June 1931; Foreman-State National Bank, Chicago, IL, 8 June 1931; Foreman-State Trust and Savings Bank, Chicago, IL, 8 June 1931; Second North Western State Bank, Chicago, IL, 9 June 1931; North-Western Trust and Savings Bank, Chicago, IL, 9 June 1931; Armitage State Bank, Chicago, IL, 9 June 1931; Auburn Park Trust and Savings Bank, Chicago, IL, 9 June 1931; Brainerd State Bank, Chicago, IL, 9 June 1931; Chatham State Bank, Chicago, IL, 9 June 1931; Chicago Lawn State Bank, Chicago, IL, 9 June 1931; Elston State Bank, Chicago, IL, 9 June 1931; Ridge State Bank, Chicago, IL, 9 June 1931; Stony Island State Savings Bank, Chicago, IL, 9 June 1931; West Englewood Trust and Savings Bank, Chicago, IL, 9 June 1931; West Highland State Bank, Chicago, IL, 9 June 1931; West Lawn Trust and Savings Bank, Chicago, IL, 9 June 1931; Cragin State Bank, Chicago, IL, 10 June 1931; Fullerton State Bank, Chicago, IL, 14 June 1931; First National Bank, Downers Grove, IL, 17 June 1931. 25. NARA, Bank Changes, Midway State Bank, Chicago, IL, 14 May 1932; Douglass National Bank, Chicago, IL, 21 May 1932; Citizens State Bank of Chicago, Chicago, IL, 25 May 1932; Papanek-Kovac State Bank, Chicago, IL, 1 June 1932; Sheridan Trust and Savings Bank, Chicago, IL, 6 June 1932; Peoples Trust and Savings Bank, Chicago, IL, 10 June 1932; Madison Square State Bank, Chicago, IL, 14 June 1932; United American Trust and Savings Bank, Chicago, IL, 14 June 1932; Alliance National Bank, Chicago, IL, 15 June 1932; Division State Bank, Chicago, IL, 15 June 1932; Empire Trust and Savings, Chicago, IL, 17 June 1932; Logan Square State and Savings Bank, Chicago, IL, 17 June 1932; Devon Trust and Savings Bank, Chicago, IL, 18 June 1932; North Avenue State Bank, Chicago, IL, 18 June 1932; Prudential State Savings Bank, Chicago, IL, 18 June 1932; Reliance Bank and Trust Company, Chicago, IL, 18 June 1932; The Bowmanville National Bank, Chicago, IL, 20 June 1932; Peoples National Bank and Trust Company of Chicago, Chicago, IL, 21 June 1932; Phillip State Bank and Trust Company, Chicago, IL, 21 June 1932; Standard National Bank of Chicago, Chicago, IL, 21 June 1932; Chatﬁeld Trust and Savings Bank, Chicago, IL, 22 June 1932; Commonwealth Trust and Savings Bank, Chicago, IL, 22 June 1932; Cottage Grove State Bank, Chicago, IL, 22 June 1932; First Englewood State Bank, Chicago, IL, 22 June 1932; Jefferson Park National Bank, Chicago, IL, 22 June 1932; South Shore State Bank, Chicago, IL, 22 June 1932; Woodlawn Trust and Savings Bank, Chicago, IL, 22 June 1932; Hyde Park-Kenwood National Bank, Chicago, IL, 23 June 1932; Jackson Park National Bank, Chicago, IL, 23 June 1932; Universal State Bank, Chicago, IL, 23 June 1932; West Irving State Bank, Chicago, IL, 23 June 1932; Central

142

GARY RICHARDSON

Manufacturing District Bank, Chicago, IL, 24 June 1932; Chicago Bank of Commerce, Chicago, IL, 24 June 1932; Kaspar-American State Bank, Chicago, IL, 24 June 1932; Ravenswood National Bank, Chicago, IL, 24 June 1932; The Midland National Bank of Chicago, Chicago, IL, 24 June 1932; The National Bank of Woodlawn, Chicago, IL, 24 June 1932; The South Ashland National Bank of Chicago, Chicago, IL, 25 June 1932; Congress Trust and Savings Bank, Chicago, IL, 28 June 1932. 26. Richardson and Troost (2006).

ACKNOWLEDGMENTS I thank Shagufta Ahmed, Shaista Ahmed, Jacqueline Chattopadhay, ChingYi Chung, Li Fan, Nathan Montgomery, Yuiichi Inomata, Mark Ng, Mitra Pai, Brandon Tsang, Ian Wagner, and Eve Wang for research assistance. I thank Reid Click, Deborah Kauffman, Ed and Edwin Richardson, and Gloria Richardson for accommodations near the National Archives. I thank Erik Heitﬁeld for the loan of photographic equipment and additional assistance. I thank Dan Bogart, William Branch, Mark Carlson, Milton Friedman, Michelle Garﬁnkel, Joseph Mason, and Kris Mitchener for extensive comments on earlier drafts of this chapter. I thank numerous friends and colleagues for comments, advice, and encouragement.

REFERENCES Bernanke, B. S. (1983). Nonmonetary effects of the ﬁnancial crisis in the propagation of the Great Depression. American Economic Review, 73(June), 257–276. Board of Governors of the Federal Reserve System. (1929). Memorandum regarding preparation of reports of changes in status of member and nonmember banks. Memo, November 18. National Archives, Record Group 82, Federal Reserve Central Subject File, 421.113. Board of Governors of the Federal Reserve System. (1930). Committee on branch, group and chain banking, bank changes – definitions of terms. Memo, November 5. National Archives, Record Group 82, Federal Reserve Central Subject File, 421.113-1. Board of Governors of the Federal Reserve System. (1931). Committee on branch, group and chain banking. Comments regarding preparation of form F. Memo, November 23. National Archives, Record Group 82, Federal Reserve Central Subject File, 421.113-1, Forms and Instructions. Board of Governors of the Federal Reserve System. (1937). Bank Suspensions in the United States. Federal Reserve Bulletin. Washington, DC: Board of Governors of the Federal Reserve System. Board of Governors of the Federal Reserve System. (1943). Banking and monetary statistics, 1914–1941. Washington, DC: Federal Reserve System.

Categories and Causes of Bank Distress, 1929–1933

143

Calomiris, C. W., & Mason, J. R. (1997). Contagion and bank failures during the Great Depression: The June 1932 Chicago banking panic. American Economic Review, 87(5), 863–883. Calomiris, C. W., & Mason, J. R. (2003). Fundamentals, panics, and bank distress during the depression. American Economic Review, 93(5), 1615–1646. Eichengreen, B. (1992). Golden fetters. New York: Oxford University Press. Friedman, M., & Schwartz, A. J. (1963). A monetary history of the United States, 1867–1960. Princeton: Princeton University Press. Goldenweiser, E. A., et al. (1931). Bank suspensions in the United States, 1892–1931. Volume 4. Material prepared for the information of the Federal Reserve System by the Federal Reserve Committee on Branch, Group, and Chain Banking. Lucia, J. (1985). The failure of the bank of the United States: A reappraisal. Explorations in Economic History, 22, 402–416. McFerrin, J. B. (1939). Caldwell and Company. Chapel Hill: University of North Carolina Press. Meltzer, A. H. (2003). A history of the Federal Reserve, volume 1, 1913–1951. Chicago: University of Chicago Press. National Archives, Record Group 82, Federal Reserve Central Subject File, ﬁle number 434.-1, ‘‘Bank Changes 1921–1954 Districts 1929–1954 – Consolidations, Suspensions and Organizations-St. 6386 a,b,c, (By States) 1930–1933.’’ Richardson, G. (2006a). The records of the Federal Reserve Board of governors in the National Archives of the United States. Financial History Review, 13(1), 123–134. Richardson, G. (2006b). Bank distress during the Great Depression: The illiquidity-insolvency debate revisited. National Bureau of Economic Research Working Paper No. w12717. Cambridge, MA. Richardson, G. (2006c). Correspondent clearing and the banking panics of the Great Depression. National Bureau of Economic Research Working Paper No. w12716. Cambridge, MA. Richardson, G. (2006d). Bank distress during the Great Contraction, 1929 to 1933, new evidence from the archives of the Board of Governors. NBER Working Paper w12590. Richardson, G. (2007). The collapse of the United States banking system during the Great Depression, 1929 to 1933, new archival evidence. Australasian Accounting, Business, and Finance Journal, 1(1), 39–50. Richardson, G., & Chung, C.-Y. (2006a). Deposit insurance and the composition of bank suspensions in developing economies: Lessons from the state deposit insurance experiments of the 1920s. National Bureau of Economic Research Working Paper No. w12594. Cambridge, MA. Richardson, G., & Chung, C.-Y. (2006b). Deposit insurance altered the composition of bank suspensions during the 1920s: Evidence from the archives of the Board of Governors. Contributions to Economic Analysis & Policy, 5(1), 1–34. Richardson, G., & Troost, W. (2006). Monetary intervention mitigated banking panics during the Great Depression: Quasi-experimental evidence from the Federal Reserve Border in Mississippi, 1929 to 1933. National Bureau of Economic Research Working Paper No. w12591. Cambridge, MA. Romer, C. (1993). The nation in depression. The Journal of Economic Perspectives, 7(2), 19–39. Temin, P. (1976). Did monetary forces cause the Great Depression? New York: W.W. Norton. Temin, P. (1989). Lessons from the Great Depression. Cambridge, MA: MIT Press.

144

GARY RICHARDSON

United States, Bureau of the Census. (1975). Historical statistics of the United States, colonial times to 1970. Bicentennial edition. Washington: U.S. Department of Commerce, U.S. Government Printing Ofﬁce. White, E. (1984). A reinterpretation of the banking crisis of 1930. Journal of Economic History, 44, 119–138. Wicker, E. (1980). A reconsideration of the causes of the banking panic of 1930. Journal of Economic History, 40, 571–583. Wicker, E. (1996). The banking panics of the Great Depression. Cambridge: Cambridge University Press. Young, R. A. (1929).Memorandum regarding preparation of reports of changes in status of member and nonmember banks. Memo, November 18. National Archives, Record Group 82, Federal Reserve Central Subject File, 421.3, Group and Chain Banking (1924–1929).

Categories and Causes of Bank Distress, 1929–1933

145

APPENDIX. ST. 6386 FORMS F. R. Board Form St. 6386a November 1929

BANK CONSOLIDATIONS (Include absorptions, mergers, etc.)

Effective data of consolidation ________________ Name and Location of Banks

Member or nonmember

Federal Reserve District ____________ Capital

Surplus and profits

Loans and investments

(a) Banks entering into consolidation:

(b) New or consolidated bank:

Is the consolidated institution a newly chartered bank? ____. If not, give the name of the bank (before consolidation) under whose charter the consolidated institution is to operate:

Give the names of banks, if any, that were in financial difficulties at time of consolidation:

Disposition made of the head office of each bank entering into consolidation:

Did any of the banks entering into consolidation have branches? ______ Were any of the banks affiliated with banking groups or chains? _______ (Report details on Forms St. 6386d and St. 6386e) Terms of consolidation, if known, etc.

146

GARY RICHARDSON

F. R. Board Form St. 6386b November 1929

BANK SUSPENSIONS

Date of suspension _____________________

Federal Reserve District _____________

Name and location of bank _____________________ Member or nonmember _____________ Capital $ ________________

Loans and Investments $________________ Gross Deposits $

______________ Borrowings from Federal Reserve Bank $ _______________ From other banks $ _________________ Condition figures are as of ______________________ Closing directed by __________________________________________________________________ Causes of suspension: Check in the appropriate column those of the following which apply, either as primary or contributing causes, amplifying the indicated causes with such supplementary data as may be available. Primary cause 1. Slow, Doubtful or worthless paper 2. Failure of banking correspondent (Name of failed correspondent) 3. Failure of other large debtor (Name of failed debtor and connection with bank, if any) 4. Defalcation 5. Heavy withdrawals 6. Other causes (specify) Remarks:

Secondary cause

Categories and Causes of Bank Distress, 1929–1933

F. R. Board Form St. 6386c Nov.1929

147

ORGANIZATIONS, LIQUIDATIONS, CONVERSIONS, AND OTHER CHANGES IN STATUS OF MEMBER AND NONMEMBER BANKS

Effective date of change_______________________

Federal Reserve District _____________

Character of change__________________________________________________ Name and location of Member or bank before change ________________________________________ nonmember ___________ Name and location of Member or bank after change _________________________________________ nonmember ___________ Capital $________________

Surplus Loans and and profits $________________ investments $ _______________

Condition figures are as of ____________________________________________________________ Is the above change in status final or merely preliminary to a further change, and if preliminary, what further change is contemplated? Is the case of conversions and successions, were financial difficulties responsible to any extent for the change in status? Remarks: (In the case of suspended banks reopened for business, give the change in capital account, the assessment paid by stockholders, etc.)

ON ENGLISH PYGMIES AND GIANTS: THE PHYSICAL STATURE OF ENGLISH YOUTH IN THE LATE 18TH AND EARLY 19TH CENTURIES John Komlos ABSTRACT The heights of lower- and upper-class English youth are compared to one another and to their European and North American counterparts in the late 18th and early 19th centuries. The height gap between the rich and poor was the greatest in England, reaching 22 cm at age 16. The poverty-stricken English teenagers were among the shortest for their age so far discovered in Europe or North America; in contrast, the English rich were the tallest in the world in their time: only 2.5 cm shorter than today’s US standard. Height of the poor declined in the late 18th century, and again in the 1830s and 1840s conforming to the general European pattern, while the height of the wealthy tended rather to increase until the 1840s and then levelled off. The results support the pessimistic view of the course of living standards among the ultra-poor in the Industrial Revolution period.

Research in Economic History, Volume 25, 149–168 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(07)25003-7

149

150

JOHN KOMLOS

The fact that England was a rather unequal (and hierarchical) society throughout the Industrial Revolution period and even before is quite well known (Lindert, 2000a). Yet, for generations until of late there has been a lively and sometimes vociferous controversy concerning the secular change in the standard of living of British workers during the Industrial Revolution. Nonetheless, the consensus seems to have emerged slowly and with some difﬁculty that the living standard of the mass of the population declined in the course of the 18th century. Peter Lindert (2000a) puts the beginning of the decline in the 1740s and illustrates it with a soaring cost-of-living index which increased by a factor of 2.8 between 1750 and 1810. As a consequence, real wages of London masons declined by at least a third between c. 1740 and 1800 (Allen, 2001, p. 434), and those of farm laborers by 18 percent during the same period (Clark, 2001, p. 496). Many diverse indicators corroborate these ﬁndings. The decline in physical stature, for example, points in a similar direction (Komlos, 1993, 1998, 2004). Moreover, inequality increased between 1750 and 1800 and while it began to decrease slowly thereafter, it did not reach its 1750 level again until c. 1890 (Hoffman, Jacks, Levin, & Lindert, 2002, p. 339). Feinstein’s (1998) seminal analysis starts a bit later, in the 1770s, so he does not quite document the substantial decline in purchasing power that took place until then, but concludes cautiously that thereafter living standards more-or-less stagnated for several generations: ‘‘For the majority of the working class the historical reality was that they had to endure almost a century of hard toil with little or no advance from a low base before they really began to share in any of the beneﬁts of the economic transformation they had helped to create’’ (p. 652). Mokyr (2003) essentially echoes these sentiments in his more recent overview: ‘‘Those who lived through the Industrial Revolution did not enjoy its fruits’’ (p. 53). Importantly in this regard, estimates of consumption conﬁrm this perspective (Clark, Huberman, & Lindert, 1995; Mokyr, 1988), and so do mortality statistics: ‘‘mortality levels in Britain’s industrializing cities deteriorated substantially during the second quarter of the nineteenth century’’ (Szreter & Mooney, 1998). In sum, the optimists seem to have been rather quiet of late. We reinforce the pessimistic assessment of living standards during the Industrial Revolution with data on physical stature of segments of the population at the two extremes of the socio-economic hierarchy: the ultrapoor and the ultra-rich. While these are by no means representative elements of the population at large, and while they have not been at the center of the living-standard debate, a comparative perspective on these two groups should enable us to gain further insights into the social dynamics of this

On English Pygmies and Giants

151

crucial period in British economic history. One of the reasons that ‘‘economic historians have pushed their study of growth and inequality beyond the usual income measures is that they ﬁnd annual income, or even annual consumption, far too narrow a concept to represent well-being or living standards’’ (Lindert, 2000b). Another important advantage of anthropometric history is the insight it affords into the living conditions of the segments of a population for whom conventional economic indicators are frequently – or even generally – unavailable, for example, children and youth, and this chapter is precisely about these groups upon whom very little work has actually been done. Understandably, most wage statistics pertain to male adults in that period. The welfare of children and youth depended, in the main, upon overall family socio-economic circumstances as well as upon resource allocation within the family. To what extent family income beneﬁts the children of a household is not at all clear even in contemporary societies; in a historical context such evidence is even more difﬁcult to interpret. The standard economic assumptions pertaining to the relationship between income and welfare do not hold easily for dependent groups, i.e., for those who do not have a personal source of income. As a consequence, anthropometric records on minors are of considerable value, particularly since these can often be decomposed by social status, gender, and age. To shed more light on well-being of children and youth in early-industrial Britain, we compare and contrast the physical stature of those who were at the absolute top of the social ladder with those who were at the absolute bottom. We conclude by comparing them to their cohorts in other countries in the late 18th and early 19th centuries.

THE HEIGHT OF ENGLISH GENTRY BOYS We ﬁrst analyze the height of upper-class students at the prestigious Royal Military Academy at Sandhurst in the 19th century (Floud, Wachter, & Gregory, 1990, p. 107).1 The data were collected by Roderick Floud (1986a) and analyzed in Floud et al. (1990, pp. 174–178), without, however, considering the effect of a minimum height requirement (MHR) for being accepted into the institution. Floud et al. (1990) suggest that ‘‘it seems that upper-class boys could pass the standard with ease and the observed distributions are very close to normal’’ (p. 174). Yet their calculations did reveal some unexpected ﬂuctuations in the estimated height of the youth, which could be due to changes in the height requirements.2 They did not

152

JOHN KOMLOS

publish the height distributions themselves, but a re-examination of the Sandhurst data indicates that de facto there were both minimum and maximum height requirements for gaining entrance into the Academy, which were not enforced consistently over time. The height distributions, particularly those of the younger students, often depart quite obviously from the expected bell-shaped (or ‘‘normal’’) curve3 (Fig. 1). This suggests that height requirements were enforced from time to time by the examiners perhaps on an ad hoc basis (or perhaps by the parents themselves as a way of selecting into the Academy), even if their actions might have followed informal procedure, rather than ofﬁcial ones. As a consequence, in such a truncated distribution the calculation of simple means is inaccurate, and the appropriate statistical procedure to estimate mean heights, and their correlates, is truncated regression4 (A’Hearn, 2004; Komlos, 2004). In addition, Floud et al. did not consider the effect of family income on the height of the students. Insofar as the fees paid by the student’s families is available, it can be used as a proxy for family economic circumstances, and consequently, it is used as an independent variable in the determinant of physical stature (Table 1). For these two limitations of the original

Fig. 1.

Height Distribution of 13-Year-Old Boys, Sandhurst Military Academy (Inches). Source: Own Calculations Based on Data in Floud (1986b).

On English Pygmies and Giants

Table 1.

153

Characteristics of the Sandhurst Sample.

Age

N

13 14 15 16 17 18 19 20 W20 Total

1,235 1,687 544 513 1,086 2,213 2,240 820 494 10,832 Fees Paid Before 1858

Fee category 1 2 3 Total

After 1858 N

Fees (pounds)

N

645 1,027 1,808 3,480

0 W50 50–99 100– Unknown Total

756 700 1,706 3,677 513 7,352

Source: Own Calculations Based on Data in Floud (1986a).

analysis it is worth revisiting these data to estimate the trends taking into account the fees paid and the height requirements which could inﬂuence the estimates. The students are divided into two groups based on the fee paid: lesser – category 1 and 2; higher – category 3.5 We examine the distributions for three time periods by recruitment years: during the Napoleonic Wars (1807–1816) (referred to as Period 1), between 1817–1836 (Period 2), and after 1836 (Period 3) in order to allow for changing truncation points which differed for different fee-category students (Fig. 1). For example, in Period 2 there was an MHR imposed on 13-year-olds at 56 inches (142.2 cm), whereas in Period 3 the requirement was raised to 57 inches (144.8 cm) but only for fee category 1 and 2 students, not for fee category 3 students (Table 2 and Fig. 1). Similar patterns are found in Periods 1 and 3. After having determined the truncation points of Table 2, we estimate the height of the students by age and fee category using truncated regression program in STATA7.

154

JOHN KOMLOS

Table 2. Age

13

14

15

16 17

Period

1807–16 1817–36 1837–57 1837–57 1807–16 1807–16 1817–36 1837–57 1807–16 1817–36 1817–36 1837–57 1855–70 1855–70 1839–43 1844–53 1844–53 1854–76

Height Requirements of Sandhurst Students. Fee

1, 2, 3 1, 2, 3 1, 2 3 1, 2 3 1, 2, 3 1, 2, 3 1, 2, 3 1, 2 3 1, 2, 3 1, 2 3 1, 2, 3 1, 2 3 1, 2, 3

Lower

Upper

inches

cm

inches

cm

– 56 57 – 58 57 58 58 – – – – – 65 – 63 – 64

– 142.2 144.8 – 147.3 144.8 147.3 147.3 – – – – – 165.1 – 160.0 – 162.6

65 – – – – – – 65 – 70 70 69 – – 71 72 72 72

165.1 – – – – – – 165.1 – 177.8 177.8 175.3 – – 180.3 182.9 182.9 182.9

Source: Own Calculations Based on Data in Floud (1986a).

There were differences in height among the Sandhurst students by fees paid (Fig. 2). Those who paid higher entrance fees were invariably taller: for example among 13-year-olds by 3.3 cm (1.3 inches) on average (Table 3). This indicates that even among the upper classes income mattered to height, and rural residence by itself did not sufﬁce to provide optimal conditions for human growth. Our estimated trends are quite similar to those of Floud et al. on average, although the new estimates are much less volatile (Fig. 3). The estimated heights of these upper-class youth did not decline in the late 18th century as did those of the average adult population (Fig. 4) (Komlos, 1989, 1993, 1998). This is similar to the pattern found among aristocratic and middle-class German youth of the late 18th century (Komlos, 1990; Komlos, Tanner, Davies, & Cole, 1992). Sandhurst students tended to be taller in 1840 than in 1795, even if the trends were not uniform and were interspersed with cycles.6 A regression of average heights on time and ages between 1795 and 1840 yields an average annual increase of 0.7 mm per annum for ages 13–15.7 Thus, upper-class height trends departed substantially from those of the rest of the population in the late 18th and

On English Pygmies and Giants

155

62 156.2 154.2

60

152.2

59

150.2

cm

inches

61

148.2

58

146.2 57

144.2

56 1790

142.2 1800

1810

1820

1830

1840

Quinquennium of Birth Fee=1&2

Fig. 2.

Fee=3

Average

Height of 13-Year-Old Sandhurst Students. Source: Own Calculations Based on Data in Floud (1986a).

Table 3. Estimated Height of Sandhurst Students by Fees Paid. Age

Fees Paid Groups 1&2

13 14 15 16 17 18 19 W19

Difference

Average

Group 3

Growth Velocity

All Fees

inches

cm

inches

cm

inches

cm

inches

cm

inches

cm

59.1 61.4 64.2 67.0 67.9 68.2 68.4 68.0

150.1 156.0 163.1 170.2 172.5 173.2 173.7 172.7

60.4 62.4 64.5 67.8 68.3 68.5 68.9 68.9

153.4 158.5 163.8 172.2 173.5 174.0 175.0 175.0

1.3 1.0 0.3 0.8 0.4 0.3 0.5 0.9

3.3 2.5 0.8 2.1 1.1 0.8 1.3 2.4

59.6 62.0 64.4 67.4 68.1 68.4 68.7 68.5

151.5 157.5 163.5 171.1 173.1 173.8 174.4 174.1

– 2.4 2.4 3.0 0.7 0.3 0.3 –

– 6.0 6.0 7.6 2.0 0.7 0.6 –

Source: Own Calculations Based on Data in Floud (1986a).

early 19th centuries. This is plausible insofar as their income would have sufﬁced to compensate for the increased price of nutrients. Admittedly upper-class lads probably also beneﬁted from a cleaner environment as they grew up on their parent’s estates rather than in the polluted cities. Data on older students are available for ages 16–20 beginning with the birth cohorts of 1840, although the trend for those 20 and above cannot be

156

JOHN KOMLOS

66 65 64 63 62 61 60 59 58 57 56 1790

166.2 Age 15

162.2

Age 14

158.2 154.2

Age 13

cm

inches

estimated accurately due to the small number of observations. The height of 16–19-year-olds tended to be constant after the middle of the 19th century (Fig. 4). Sandhurst students were exceptionally tall at the time in international comparison. Their height at age 20, which can be considered their adult height, was c. 174 cm (68.5 inches), just 3 cm (1.2 inches) less than the

150.2 146.2 142.2 1830 1810 1820 Quinquennium of Birth

1800

Floud 13

Age 14

Floud 14

Age 15

1840 Age13

Floud 15

69 68 67 66 65 64 63 62 61 60 59 58 1790

174.3 171.3 168.3 165.3 162.3 159.3 156.3 153.3 150.3 147.3 1810 13

14

1830 1850 Quinquennium of Birth 15

16

17

cm

inches

Fig. 3. The Height of Sandhurst Students, Comparison with Floud et al.’s Estimates. Source: Own Calculations Based on Data in Floud (1986a).

1870 19

18

Fig. 4. Height of Sandhurst Students at Different Ages. Source: Own Calculations Based on Data in Floud (1986a).

On English Pygmies and Giants

157

average height of current British male youth. High-fee students were 1.6 cm (0.6 inches) shorter than today’s US standard (Fig. 5).8 Even low-fee-paying Sandhurst students were taller than most other students attending elite schools in Germany, France, and the United States (Fig. 6). Although the reference to the German youth (of the lower aristocracy) is to the birth cohorts of the 18th century, the 10 cm (3.9 inches) advantage of the low-fee-paying Sandhurst students at age 16, is nonetheless, very substantial. It is also quite

190

97% 95% 90%

185

75%

180 175

50% High-Fee

Low-Fee

25%

cm

170

10% 5% 3%

165 160 155 150 145 140

13

14

15

16

17

18

19

20

Age

cm

Fig. 5. Growth Proﬁle of Sandhurst Students Compared to Contemporary US Standards. Source: Own Calculations Based on Data in Floud (1986a).

175 170 165 160 155 150 145 140

13

14

15

16

West Point Cadets German Lower Aristocrats

Fig. 6.

17

18

19

20

Sandhurst Low Fee Students France, Polytechnique

Growth Proﬁles of Elite Students. Sources: Own Calculations Based on Data in Floud (1986a) and Komlos (1987, 1990, 1994).

158

JOHN KOMLOS

175

U.S.

170

69

cm

165

65 German

160

63 English

155

inches

67 U.S.

61

150

59 13

14

15

16

17

18

19

20

Age Sandhurst High Fee Students South Carolina Students

German Upper Aristocrats West Point Cadets

Fig. 7. Growth Proﬁles of Elite Youth, International Comparison. Sources: Own Calculations Based on Data in Floud (1986a), Coclanis and Komlos (1995), and Komlos (1987, 1990).

extraordinary that the Sandhurst students were taller than the cadets attending the West Point Military Academy as well as the Citadel, the Military Academy of Charleston, South Carolina (Fig. 7). This is unexpected, because the more propitious disease environment and the greater availability of nutrients meant that the average American adult male was at least 5–6 cm (2.0–2.4 inches) taller than its European, including British, counterparts in the 19th century (Komlos & Baur, 2004). In fact, average Americans were the tallest in the world, but no segment of the American population was as privileged as the European elite.9 The European elite was clearly capable of overcoming the disadvantages brought about by a higher level of urbanization, higher population density, a more virulent disease environment, and higher nutrient prices. In fact, the high-nobility in Germany was the only group who was as tall as or even taller, than the high-fee paying Sandhurst students10 (Fig. 7). Hence, only the sons of the hereditary princes and barons on the continent were as tall as the descendants of the British gentry attending the Sandhurst Academy.

THE HEIGHT OF THE ENGLISH ULTRA-POOR The records of the Marine Society provides important evidence on the height of poverty-stricken English boys in the late 18th and ﬁrst half of the

On English Pygmies and Giants

159

19th centuries, ﬁrst reported in Floud and Wachter (1982). Floud collected more than 50,000 observations of ‘‘poor children’’ ‘‘of the London slums’’ who entered this institution between the 1770s and 1870s for a life at sea (Floud et al., 1990, pp. 55, 105, 196) (Table 4). The initial analysis of these data (Floud et al., 1990) also showed implausibly large and confusing variations in the height estimates11 (Fig. 8). The ﬂuctuations were caused by Floud et al.’s disregard of the fact that both tails of the height distributions were truncated in some cases, not just the lower tail (Floud et al., 1990, p. 164; Komlos, 1993, 2004). This was often the case because very tall men

Table 4.

Characteristics of the Marine Society Sample.

Age 10 11 12 13 14 15 16 17 18 19 20 21 Other Total Date of Recruitment 1770s 1780s 1790s 1800s 1810s 1820s 1830s 1840s 1850s 1860s 1870s Source: Own Calculations Based on Data in Floud (1986b).

Number of Observations 17 93 600 9,527 12,392 15,150 10,435 2,103 655 213 38 11 42 51,276 Number of Observations 4228 4467 7289 6091 4327 5103 5851 4847 4344 3331 1398

62

157

60

152

58

147

56

142

54

137

52

132

50 1755

127 1775

1795 1815 Date of Birth

1835

Age 15

Age 14

Age 13

Fig. 8.

cm

JOHN KOMLOS

inches

160

1855 Age 16

Height of Poor English Boys: Floud and Wachter Esitimates. Source: Floud Wachter and Gregory (1990, p. 166).

20

Percent

15 N=643 10 5 0 49

51

52

53

54

55

56

57

58

59

60

61

62

63

64

Inches

Fig. 9.

Height Distribution of 16-Year-Old Boys Born in the 1760s. Source: Own Calculations Based on Data in Floud (1986b).

were not suitable to become sailors, as life at sea required a low center of gravity (Fig. 9). The use of truncated regression alleviates this problem mostly, and identiﬁes the following secular trends in the height of the children of the ultra-poor: heights declined between the birth cohorts of c. 1770 and 1795, increased until 1805, and then decreased slowly in the main but more substantially again in the 1830s and 1840s, as in most other parts of the Atlantic community (Fig. 10) (Komlos, 1993, 1998, 2004). The

Fig. 10.

Age12 Age18

Age13 Age19

Age14 Age20

Age15 Age21

Height of Marine Society Boys, Truncated Regression Estimates. Source: Own Calculations Based on Data in Floud (1986b).

160 155 150 145 140 135 130 125 120

63.2 61.2 59.2 57.2 55.2 53.2 51.2 49.2 47.2

1805 1850 1770 1795

10

Fig. 11.

Age11 Age17

cm

162 64 157 62 152 60 147 58 142 56 137 54 132 52 127 50 122 48 1750 1760 1770 1780 1790 1800 1810 1820 1830 1840 1850 1860 Date of Birth Age10 Age16

cm

161

11

12

13

14

15

16 Age

17

18

19

20

inches

inches

On English Pygmies and Giants

21

Height Proﬁles of Marine Society Boys. Source: Own Calculations Based on Data in Floud (1986b).

height proﬁles were shifting practically parallel to one another over time, with heights ending up in the mid-19th century at the level of c. 1795, the probable 18th-century nadir (Fig. 11).

DISCUSSION This evidence reveals an extremely deep divide that separated the social classes during the Industrial Revolution to an extent that is no longer

162

JOHN KOMLOS

imaginable today. The average difference between Sandhurst cadets and Marine Society boys was 16.3 cm (6.4 inches) at age 13, rising to 22 cm (8.7 inches) at age 16, indicating that the wealthy experienced an adolescent growth spurt earlier and their peak growth velocity was greater than those of the poor (Fig. 12). The difference between the two groups traces a ‘‘U’’shaped curve: the elite students enjoyed a 20.8 cm (8.2 inches) height advantage at the beginning of the period, declining to about 15 cm (5.9 inches) for most of the period under consideration, and rising again in the late 1830s to reach 22.6 cm (8.9 inches) in 184012 (Fig. 13). It appears

175 170

67.2 Sandhurst

160 cm

65.2 22.0 cm

63.2

18.8 cm

155

61.2

18.7 cm

150

59.2

16.3 cm

145

57.2

Marine Society

140

inches

165

55.2

135

53.2 13

14

16

15

17

18

Age

Fig. 12.

Height Proﬁle of Sandhurst and Marine Society Boys. Source: Own Calculations Based on Data in Floud (1986a, 1986b).

9 8 7 6 5 4 3 2

cm

20 15 10 5 1795

1805 13

Fig. 13.

1815 Date of Birth 14

15

1825

inches

25

1835 Average

The Height Advantage of Sandhurst Students over Marine Society Boys. Source: Own Calculations Based on Data in Floud (1986a, 1986b).

On English Pygmies and Giants

163

175 170 165 160 155 150 145 140 135 130 125

67.2 65.2 63.2 61.2 59.2 57.2 55.2 53.2 51.2 49.2

11

12

13

14

15

16

17

18

19

inches

cm

that harder times of the late 18th century and again in the 1830s and 1840s affected the nutritional well-being of the lower classes more adversely than those of the upper classes. All in all, the degree of inequality was about the same at the end of the period under consideration as at its beginning. Hence, these data do not conﬁrm Kuznets’ (1955) inverse ‘‘U’’ pattern of inequality for the extreme ends of the income/wealth distribution for this period. Kuznets had suggested that inequality would increase during the initial stages of modern economic growth and most authors do, in fact, report that inequality was increasing during the classical phase of the Industrial Revolution (Lindert, 1994; Williamson, 1985). This discrepancy is no doubt due to the fact that the comparison is between the ultra-poor and ultra-rich rather than the average level of inequality examined by Kuznets and his many followers. In addition, the period considered is but half a century, shorter than the length of time Kuznets envisioned. The poor Marine Society boys, many of them from London, were among the shortest for their age in Europe and North America ever recorded. Only Dutch orphans recorded in 1865 were somewhat shorter (Fig. 14) (Fredriks, 2004, p. 174), implying that the physical stature of Marine Society boys might have been typical of teenagers at the very bottom of the European social pyramid of the 19th century. At age 16 the Marine Society boys were 1.4–2.6 cm (0.6–1.0 inches) shorter than German servants (Komlos, 1990) and 5–8 cm (2.0–3.1 inches) shorter than US slaves (Engerman, 1976; Steckel, 1979). The tallest 16-year-old Marine Society

20

Age German US Slaves

Habsburg Sandhurst

Marine Society Dutch Orphans

Fig. 14. Height of Lower-Class Youth Compared to Sandhurst Students. Sources: Own Calculations Based on Data in Floud (1986a, 1986b), Fredriks (2004), Komlos (1986, 1990) and Steckel (1986).

164

JOHN KOMLOS

16 70 68 14 66 12 64 10 62 8 60 58 6 56 4 54 2 52 50 0 1750 1760 1770 1780 1790 1800 1810 1820 1830 1840 1850 1860

Years

inches

boys, born in the 1820s, were on average 155 cm (61 inches) tall, shorter than the 3rd centile of the modern US height reference value of 160 cm (63 inches). The more egalitarian nature of American society prevented such an European-type gap from emerging among the classes in North America. American apprentices (Komlos, 2001), for example, were 8.2 cm (3.2 inches) and American Slaves were 6.6 cm (2.6 inches) taller than German servants or the boys attending Habsburg military schools (Fig. 14). Among lower-class Americans, the Georgia convicts were the tallest and the slaves the shortest, with the difference between them at age 17 of about 5.2 cm (2.0 inches), but northern white apprentices were only 1.6 cm (0.6 inches) taller than slaves, while free blacks were merely 1.1 cm (0.4 inches) taller than slaves (Komlos & Coclanis, 1997). Freed slaves were but 3.5 cm (1.4 inches) shorter than the average northern soldier. There is an uncanny similarity between the length of solar sun-spot cycles and the height of Marine Society boys (Fig. 15) (Komlos, Corne´lissen, Woitek, Otsuka, & Halberg, 2004). While the correlation could well be spurious, the sun-spot cycles could have had an effect on agricultural conditions, and thus on real wages, as these are themselves strongly inﬂuenced by prices and harvest outcomes. The length of sun-spot cycles hit a nadir around 1790 while real wages hit a low point about

Date of Birth Age13

Age14

Age15

Age16

Sun Spot Period

Fig. 15. Height of Marine Society Boys and Solar Activity Cycles. Sources: Own Calculations Based on Data in Floud (1986b) and Komlos et al. (2004).

On English Pygmies and Giants

165

1800 (Allen, 2001). The rural cost of living showed similar tendencies (Clark, 2001).

CONCLUSION The re-examination of the evidence enables us to estimate the height trends and height proﬁles both of ultra-rich and ultra-poor English youth from the late 18th to mid-19th century more accurately than has been done previously. The results point to an extremely deep divide in English society between the gentry students of the Royal Military Academy at Sandhurst and the slum boys of the Marine Society.13 The extremes of income inequality during the early industrial era are illustrated more vividly by the differences in physical stature of these two groups of youth than possibly by any other measure. This has been less obvious in previous work, though admittedly Floud et al. (1990) did note in passing that ‘‘the [Sandhurst] cadets were very tall by the standards of the Marine Society’’, and that the ‘‘contrast between them is dramatic’’, i.e., ‘‘20 cm’’ at age 14 (pp. 174, 196, 225). Indeed, the differences were very substantial. In fact, the ‘‘Oliver Twists’’ of England were shorter than all other groups hitherto examined in Europe or North America save the Dutch orphans (Fig. 14). They were even shorter than American slaves for the ages for which data are available. Moreover, those who were further disadvantaged in this group, such as orphans, or who were raised in female-headed households were even shorter (Horrell, Humphries, & Voth, 1998). At the same time, the wealthy English youth were among the tallest in the world in the late 18th and early 19th centuries, equalled only by the German upper aristocracy. This was the case not only because of their income but also because of the propitious rural environment in which they lived. The weight of evidence examined here further supports the pessimistic assessment of the course of living standards during the Industrial Revolution. The difference in height between the English lower- and upper-class youth reached some 22 cm (8.7 inches) at age 16 and is thereby the largest difference in height between social groups ever recorded. This divide developed although the amount spent on poor relief doubled as a percentage of GDP in this period (Boyer, 2002; Lindert, 1998). The hidden costs of industrialization are thus vividly revealed by the great inequality in physical stature of these two socio-economic groups at the extreme ends of the income distribution. Hence, anthropometric history once again provides valuable insights into socio-economic inequality accompanying the Industrial Revolution.

166

JOHN KOMLOS

NOTES 1. The data were extracted from the records of the Academy by Roderick Floud (1986a) and deposited at the University of Essex’s data archive. There are about 10,000 data deposited in the Essex archive even though Floud et al. (1990, p. 133) mention that there were twice as many data recorded originally. Unfortunately, the occupation of the parents was not recorded, even though the information is apparently available. 2. They do report ‘‘substantial movements in the heights of the recruits over time’’, without, however, considering that these ﬂuctuations could be due to height requirements (Floud et al., 1990, p. 174). 3. The height distributions of 14–16-year-olds are not reported here, but are available from the author upon request. 4. Truncated regression estimates the true height of the population from which the sample is drawn, even though people shorter than the MHR (or taller than the maximum height limit) are underrepresented in the sample. This is done by assuming that the height is normally distributed. 5. After 1858, three fee categories were built: W50, 50–99, 100o (Table 1). 6. Some of the ﬂuctuations are possibly due to omitted variables, such as the regional provenance of the boys and the occupation of the parents. Coupled with the uncertainties associated with the determination of the height requirements, the missing variables prohibit a precise estimate of the trends. The provenance of the boys was available in the archival data but was not collected. 7. With t-statistics=4.0. 8. High-fee paying students were 2.4 cm (0.9 inches) taller as adults than low-fee paying students. 9. The students at the E´cole Polytechnique, whose admission policies appear to have been more merit based than those for Sandhurst, were as tall as the cadets of the West Point Military Academy (Fig. 9). 10. The only groups in America who were taller than the English Gentry were some tribes among the Plains Indians who were living close to buffalo herds (Steckel & Prince, 2001). 11. Height of a population tends to change minimally from year to year (on the order of a millimeter). Whenever estimates change by more than 2 cm per decade, they should be considered suspect. Floud et al.’s estimates vary as much as 2.5 cm per year – which is above a reasonable order of magnitude. 12. In contrast, Floud et al. (1990, p. 198) suggest that the ‘‘gap between them was narrowing towards the middle of the nineteenth century’’. 13. If one considers that the shorter youth among the lower classes probably had a higher than average mortality rate than the estimated gap between the classes could be considered as a lower bound of the true value.

ACKNOWLEDGMENT I appreciate comments from Francesco Cinnerella, Thomas Weiss, and Hans Th. Waaler on an earlier version of this chapter.

On English Pygmies and Giants

167

REFERENCES A’Hearn, B. (2004). A restricted maximum likelihood estimator for truncated height samples. Economics and Human Biology, 2(1), 5–19. Allen, R. (2001). The great divergence in European wages and prices from the middle ages to the ﬁrst world war. Explorations in Economic History, 38, 411–447. Boyer, G. (2002). English poor laws. In: R. Whaples (Ed.), Eh.Net Encyclopedia. http://eh.net/ encyclopedia/article/boyer.poor.laws.england Clark, G. (2001). Farm wages and living standards in the industrial revolution: England, 1670–1869. Economic History Review, 54(3), 477–505. Clark, G., Huberman, M., & Lindert, P. (1995). A British food puzzle, 1770–1850. Economic History Review, 48(2), 215–237. Coclanis, P., & Komlos, J. (1995). Nutrition and economic development in post-reconstruction South Carolina: An anthropometric approach. Social Science History, 19(1), 92–115. Engerman, S. (1976). The height of U.S. slaves. Local Population Studies, 16, 45–49. Feinstein, C. H. (1998). Pessimism perpetuated: Real wages and the standard of living in Britain during and after the industrial revolution. Journal of Economic History, 58, 625–658. Floud, R. (1986a). Long-term changes in nutrition, welfare and productivity in Britain. Heights and Ages of Sandhurst Recruits, 1808–1893 (computer ﬁle). UKDA study number 2133. Colchester, Essex: UK Data Archive (distributor), 7 July. Floud, R. (1986b). Long-term changes in nutrition, welfare and productivity in Britain physical and socio-economic characteristics of boys recruited into the marine society, 1770–1873 (computer ﬁle). UKDA study number 2134. Colchester, Essex: UK Data Archive (distributor), 7 July. Floud, R., & Wachter, K. W. (1982). Poverty and physical stature: Evidence on the standard of living of London boys 1770–1870. Social Science History, 6, 422–452. Floud, R., Wachter, K., & Gregory, A. (1990). Height, health and history: Nutritional status in the United Kingdom, 1750–1980. Cambridge: Cambridge University Press. Fredriks, M. (2004). Growth Diagrams, 1997. Amsterdam: Fourth Dutch Nation-wide Survey. Hoffman, P. T., Jacks, D. S., Levin, P. A., & Lindert, P. H. (2002). Real inequality in Europe since 1500. Journal of Economic History, 62(2), 322–355. Horrell, S., Humphries, J., & Voth, H.-J. (1998). Stature and relative deprivation: Fatherless children in early industrial Britain. Continuity and Change, 13(1), 73–115. Komlos, J. (1986). Patterns of children’s growth in east-central Europe in the eighteenth century. Annals of Human Biology, 13, 33–48. Komlos, J. (1987). The height and weight of west point cadets: Dietary change in antebellum America. Journal of Economic History, 47, 897–927. Komlos, J. (1989). Nutrition and economic development in the eighteenth-century Habsburg monarchy: An anthropometric history. Princeton: Princeton University Press. Komlos, J. (1990). Height and social status in eighteenth-century Germany. Journal of Interdisciplinary History, 20, 607–621. Komlos, J. (1993). The secular trend in the biological standard of living in the United Kingdom, 1730–1860. Economic History Review, 46, 115–144. Komlos, J. (1994). The nutritional status of French students. Journal of Interdisciplinary History, 24(3), 493–508. Komlos, J. (1998). Shrinking in a growing economy: The mystery of physical stature during the industrial revolution. Journal of Economic History, 58, 779–802.

168

JOHN KOMLOS

Komlos, J. (2001). On the biological standard of living of eighteenth-century Americans: Taller, richer, healthier. Research in Economic History, 20, 223–248. Komlos, J. (2004). How to (and how not to) analyze deﬁcient height samples: An introduction. Historical Methods, 37(4), 160–173. Komlos, J., & Baur, M. (2004). From the tallest to (one of) the fattest: The enigmatic fate of the size of the American population in the twentieth century. Economics and Human Biology, 2, 57–74. Komlos, J., & Coclanis, P. (1997). On the ‘puzzling’ antebellum cycle of the biological standard of living: The case of Georgia. Explorations in Economic History, 34(4), 433–459. Komlos, J., Corne´lissen, G., Woitek, U., Otsuka, K., & Halberg, F. (2004). Time structures, chronomes, of soldiers’ stature mimicking hale cycle of neonatal body length. Biomedicine & Pharmacotherapy, 58, 135–139. Komlos, J., Tanner, J. M., Davies, P. S. W., & Cole, T. (1992). The growth of boys in the stuttgart Carlschule, 1771–93. Annals of Human Biology, 19, 139–152. Kuznets, S. (1955). Economic growth and income inequality. American Economic Review, 45, 1–28. Lindert, P. (1994). Unequal living standards. In: R. Floud & D. McCloskey (Eds), The economic history of Britain since 1700 (Vol. 1, pp. 357–386). Cambridge: Cambridge University Press. Lindert, P. (2000a). When did inequality rise in Britain and America? Journal of Income Distribution, 9, 11–25. Lindert, P. (2000b). Early inequality and industrialization: An introduction. Journal of Income Distribution, 9, 5–9. Lindert, P. H. (1998). Poor relief before the welfare state: Britain versus the continent, 1780–1880. European Review of Economic History, 2, 101–140. Mokyr, J. (1988). Is there still life in the pessimist case? Consumption during the industrial revolution, 1790–1850. Journal of Economic History, 48(1), 69–92. Mokyr, J. (2003). The Industrial Revolution. In: J. Mokyr (Ed.), The Oxford encyclopedia of economic history (Vol. 3, pp. 49–56). Oxford: Oxford University Press. Steckel, R. (1979). Slave height proﬁles from coastwise manifests. Explorations in Economic History, 16(4), 363–380. Steckel, R. H. (1986). A peculiar population: The nutrition, health, and mortality of American slaves from childhood to maturity. Journal of Economic History, 46(3), 721–741. Steckel, R. H., & Prince, J. M. (2001). Tallest in the world: Native Americans of the great plains in the nineteenth century. American Economic Review, 91(1), 287–294. Szreter, S., & Mooney, G. (1998). Urbanization, mortality, and the standard of living debate: New estimates of the expectation of life at birth in nineteenth-century British cities. Economic History Review, 51(1), 84–112. Williamson, J. (1985). Did British capitalism breed inequality? London: Allen & Unwin.

FISCAL STATISTICS FOR SWEDEN 1719–2003 Klas Fregert and Roger Gustafsson ABSTRACT We construct yearly fiscal series for Sweden between 1719 and 2003 including expenditures, revenues, deficits and debt. We present measures for the fiscal branch of the central government as well as for the consolidated fiscal and monetary branch, which includes fiscal seigniorage. We evaluate the reliability and consistency of the series by calculating the difference between budget deficits and the change in debt to test if the differences are serially uncorrelated around zero, which we confirm.

1. INTRODUCTION Long-run data on government ﬁnances are central to understanding macroeconomic and political history. They are also needed to test macroeconomic hypotheses on ﬁscal policy, such as how sustainability is achieved. Yet, international data are scarce before the 20th century. Data covering the 18th century are, to our knowledge, only available for the United Kingdom and for the 19th century there exist data only for a handful of countries.1 The purpose of this chapter is to present and evaluate data for the central government of Sweden starting in 1719 according to modern conventions of ﬁscal measures. We present yearly data for the key ﬂow and stock measures, Research in Economic History, Volume 25, 169–223 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(07)25004-9

169

170

KLAS FREGERT AND ROGER GUSTAFSSON

with the total deﬁcit and the total debt as the bottom line. The aim is to construct measures that are broad, meaningful and widely used. In practice, this means constructing measures according to the principles of the current Swedish budget system. The chapter deals with the data: sources, reliability and calculations. Only brieﬂy will we allude to causes.2 Sweden is well suited for this attempt. First, it has been territorially and politically stable to an unusual degree since 1719 when a new constitution with a representative government was instituted. Second, yearly data are available in printed form for the whole period. For the 1719–1809 period, we use the monumental study by A˚mark (1961) based on archival material. For the period 1810–1820, we use Rathsman (1855). From 1821 onwards we use government publications. We focus on the main problems which, we believe, are the same for all countries. These are foremost the consolidation of budget and off-budget items, construction of consistent nominal amounts when there are currency reforms and interpreting the changing formats of the government accounts. We hope this work may inspire others to take up the challenges of constructing long ﬁscal series that connect to current data to facilitate the comparison between countries and enable researchers to use long panels. We suggest two methodological improvements on previous work. First, we argue that the difference between deﬁcits and changes in debt should be used to check their mutual consistency. Second, we show how to compute ﬁscal seigniorage consistent with other revenue measures. The chapter is organized as follows. Section 2 presents the deﬁnitions used in the calculations. Section 3 describes the ﬂow variables (expenditures, revenues and deﬁcits). Section 4 describes the construction and evolution of debt for the ﬁscal branch of the central government. Section 5 evaluates the consistency of the constructed series by comparing accumulated deﬁcits and the evolution of debt. Section 6 presents consolidated measures for the ﬁscal and monetary branches of the central government, including seigniorage. Section 7 concludes. A companion working paper (Fregert & Gustafsson, 2005) describes the data in more detail as well as the institutional setting. In the appendix we list all our data sources and give a table with the ﬁnal yearly data.

2. DEFINITIONS3 2.1. The Territory The Swedish territory has been constant since 1809. During the 18th century, the Swedish nation included Pomerania, Wismar and Finland.4

Fiscal Statistics for Sweden 1719–2003

171

Separate budgets were constructed for Pomerania and Wismar by their administrations, which had to be approved by the Swedish Parliament. Occasionally, funds were transferred from Sweden to Pomerania, but not the other way. These expenditures appear in the Swedish budgets as ‘‘payment to ﬁll the state deﬁcit in Pomerania’’. A separate budget was made up in Sweden for Finland during the period 1722–1792. Since we cannot separate Sweden and Finland between 1793 and 1809, Finland is included in the ﬁgures between 1722 and 1809.5 Furthermore, the year 1809 also represents a break in the data sources. Between 1815 and 1905, Sweden was in a union with Norway with the Swedish king as the head of state. The two countries had separate administrations and budgets. As Sweden ran foreign policy, the two countries shared the cost of the Swedish Ministry of Foreign affairs. 2.2. The Central Government The central government is part of the larger general government, which in turn is a part of the larger public sector as outlined in Fig. 1, according to standard conventions, such as the UN System of National Accounts (SNA) 1993. The deﬁning characteristic of the central government relative to public corporations and social security is that it is ﬁnanced mainly by taxes. A practical deﬁnition of the central government is the activities that appear in current ofﬁcial budgets. Thus, we do not include public corporations, such as the national railway system and the social security system, or local and state government. The public corporations’ net revenues, however, are included, as the central government is their owner. In ofﬁcial government accounts, the central bank is treated as any other public corporation, such that only its transfers to the treasury enters the budget and borrowing at the central bank is lumped together with other

Public sector Public corporations Other public corporations

General government

Central bank (Monetary branch)

Central government (Fiscal branch)

National Debt Office, NDO

Special funds, SF

Fig. 1.

The Public Sector.

Social security

Local and state government

Government proper, GP

172

KLAS FREGERT AND ROGER GUSTAFSSON

borrowing. This concept of the central government may be labeled the ﬁscal branch of the central government. Since the central bank has a special position as a source of ﬁnance for the ﬁscal branch of the central government, it may be preferable for some macroeconomic purposes to consolidate the monetary and the ﬁscal branch of the central government, which we do in Section 6. A large part of the practical work comes from the consolidation of different entities that today constitute the ﬁscal branch of the central government, but earlier were separated institutionally and therefore had separate ‘‘off-budget’’ accounts. Fig. 1 and Table 1 give an overview of how revenues and expenditures have been consolidated. The central government proper (Statsverket) with its own budget and accounts has always been a subset of our concept of the central government. The ofﬁcial budget has been consolidated gradually. The split was largest before 1809, when a number of special funds existed shown in Table 2. The special funds and the National Debt Ofﬁce had their own separate budgets and earmarked revenues. In 1809, the special funds disappeared as separate agencies, and the National Debt Ofﬁce’s accounts were included in the ofﬁcial budget. An old agency, the National Board of Trade (Kommerskollegium), which previously was ﬁnanced within the budget, appeared offbudget from 1809 to 1874.6 It has been included though the numbers are insigniﬁcant. The only major item outside the budget after 1809 has been government lending to public corporations for investment purposes and later also to private companies (private railways) and individuals (mortgage lending and student loans). Most of the lending was undertaken by the National Debt Ofﬁce, though the government proper occasionally also has conducted some

Table 1. Consolidation of the Central Government (1722–2003). Years 1722–1766 1766–1788 1789–1809 1810–1853 1854–1911 1912–2003

Revenues

Expenditures

GP+NDO+SF GP+SF GP+NDO+SF GP+SF GP+SF (1854–1873) GP

GP+NDO+SF GP+SF GP+NDO+SF GP+SF GP+SF (1854–1873)+NL (1854–1911) GP+NL (1938/39–1995/96)

Note: GP, Government proper; NDO, National Debt Ofﬁce; SF, Special funds; NL, Net lending.

Name in Swedish

Generalfo¨rra˚dskassan Krigsmanshuskassorna Konvojfonden

Landhja¨lpsfonden Manufakturfonden Manufakturdiskontfonden Slottsbyggnadsfonden Kommerskollegium

Special Funds.

Explanation

Years

Maximum Revenue, dsm (Year)

Maximum Debt, dsm (Year)

Funding of temporary military needs Pension fund for retired military Military convoys to escort Swedish ships ﬁnanced by export and import duties Subsidized lending to manufacturing ﬁnanced by import duties Successor to Landhja¨lpsfonden Lending agency for manufacturers Tax ﬁnancing of the new royal castle in Stockholm National Board of Trade

1721–1775 1722–1809 1726–1809

114,795 (1766) 109,284 (1809) 440,550 (1778)

46,887 (1740) 74,982 (1729) 2,107,253 (1776)

373,346 (1750)

3,380,381 (1757)

Fiscal Statistics for Sweden 1719–2003

Table 2.

1727–1739 1739–1766 1756–1776 1728–1777

114,828 (1756)

1809–1874

Source: A˚mark (1961).

173

174

KLAS FREGERT AND ROGER GUSTAFSSON

lending. Traditionally, this lending has been called net lending.7 We add net lending to the expenditure side for the periods 1854–1911 and 1938/39–1995/96. Net lending was included in the budget in the years 1912–1937 and again from 1997 onwards.

2.3. Government Accounts and Deficit Measures Until 1980, the government accounts have been presented in the form of ﬁnancial statements divided into two equal value columns of sources and uses of funds. This corresponds to the one-period central government budget constraint which shows sources of funds being equal to uses, T t þ CBT t þ Lt ¼ G t þ I t þ it Bt1 þ DAt þ AM t |ﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ ﬄ{zﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ} |ﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ{zﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ} Sources

(1)

Uses

The sources of funds consist of: T t is the government’s revenues from taxes, income from public corporations and sales of real assets (privatizations), CBT t the revenue transfers from the central bank and Lt the new loans. The uses of funds consist of: Gt the current expenditures (consumption and transfer payments), I t the real investments and long-term ﬁnancial investments (net lending), it Bt1 the nominal interest payments on government debt, DAt the increases in short-term ﬁnancial assets and AM t the amortization of debt (buy-backs and principal payments). The change in the government debt in turn is equal to new loans, Lt , minus amortization of government debt, AM t , so Eq. (1) may be written as, T t þ CBT t þ DBt ¼ Gt þ I t þ it Bt1 þ DAt |ﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ{zﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ} |ﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ{zﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ} Sources

(2)

Uses

We calculate the total budget deﬁcit as total expenditures minus total revenues, also commonly labeled as the government’s borrowing requirement.8 Rearranging Eqs. (1) and (2) gives the link between the deﬁcit and its ﬁnancing,9 DEF t ¼ ðG t þ I t þ it Bt1 Þ ðT t þ CBT t Þ ¼ Lt AM t DAt ¼ DBt DAt

(3)

A positive deﬁcit can thus be ﬁnanced by increasing the government debt or by selling ﬁnancial assets.10 If the increase in the debt, DBt, is larger than required borrowing, DEF t ; then short-term ﬁnancial assets, DAt , will

Fiscal Statistics for Sweden 1719–2003

175

increase. If the government is running a surplus, the government amortizes the debt or accumulates ﬁnancial assets. Financial asset changes naturally arise from the uneven timing of revenues, expenditures and borrowing operations. They can also arise from the planned delay of expenditures from one year to the next as some revenues are allocated to be used over more than one ﬁscal year. These means are then carried across years as ‘‘unspent balances’’ or ‘‘reservations’’. The change in short-term ﬁnancial assets is usually small and consists mostly of changes in the National Debt Ofﬁce’s and the central government proper’s checking accounts at the central bank. The source–use account statement is presented in Table 3. Identifying the deﬁcit amounts to separating revenues and expenditures from the ﬁnancing operations. By convention expenditures and revenues, the left-hand side of Eq. (3), are said to appear ‘‘above the line’’, and ﬁnancing operations, the right-hand side, appear ‘‘below the line’’, where the line is the total deﬁcit line drawn in Table 3. The table may also be simpliﬁed by netting sources and uses below the line and put the resulting net amount, that is, the total deﬁcit, on the source side as in Table 4. From a practical point of view, calculating the deﬁcit as expenditures minus revenues from source-use statement, amounts to identifying items above the line and ignoring the rest. Table 3.

The Government Budget and the Total Deﬁcit Line.

Sources Revenues: taxes, sales of real assets, income from public corporations, T t Transfers from the central bank, CBT t

Uses Current expenditure, Gt

Investment, I t Interest payments, it Bt1

Subtotal

New loans, Lt

Subtotal

Total Deﬁcit: Borrowing Requirement

Short-term ﬁnancial asset accumulation, DAt Amortization, AM t

Subtotal Total

Net: Uses–Sources

Subtotal Total

Total Surplus 0

176

KLAS FREGERT AND ROGER GUSTAFSSON

Table 4.

The Government Budget and the Total Deﬁcit Line with the Budget Deﬁcit on the Source Side.

Sources

Uses

Revenues: taxes, sales of real assets, income from public corporations, T t Transfers from the central bank, CBT t

Current expenditure, Gt Investment, I t Interest payments, it Bt1

Subtotal

Subtotal

Total deﬁcit: borrowing requirement Total

Net: Uses–Sources

Total Deﬁcit: Borrowing Requirement Total surplus

Total

0

Table 5. Conversion of Units in Government Accounts to SEK. Period

Currency Unit in Accounts

Conversion Rate Account Units to Kronor

1718–1776 1777–1789 1789–1803 1803–1809 1809–1857 1858–1873 1873–2003

Daler silvermynt (dsm) Riksdaler specie (rdr sp) Riksdaler specie (rdr sp) Riksdaler specie (rdr sp) Riksdaler banco (rdr bco) Riksdaler riksmynt (rdr rmt) SEK (Krona)

1 dsm=1/6 rdr sp=1/6 SEK 1 rdr sp=1 rdr rgs=1 SEK Floating exchange rate rdr sp and rdr rg 1 rdr sp=1.5 rdr rgs=1.5 SEK 1 rdr bco=1.5 rdr rgs=1.5 SEK 1 rdr rmt=1 rdr rgs=1 SEK

Note: The exchange rate 1789–1803 is taken from Ahlstro¨m (1972), Tables XII:5a and XII:5b.

2.4. Conversion to a Common Currency Unit We present nominal amounts in thousands of SEK (krona), the currency unit used in Sweden since 1873. The currency units used in the original sources before 1873 and their conversion rates into the previous units and SEK are presented in Table 5. It is possible to construct a series that links previous monetary units between 1777 and 1873 to SEK at a one-to-one basis as described in Heckscher (1942, p. 19). Thus amounts given in these earlier units may be labeled SEK. Amounts before 1777 are given in daler silvermynt (dsm), which are converted to SEK by multiplying nominal amounts by 6 as implied by the currency reform of 1777. While other

Fiscal Statistics for Sweden 1719–2003

177

Fixed 1803-1857: 1,5 rdr rg =1 rdr sp = 1 rdr bco

Absolute price level

Floating 1789-1803

rdr rmt

SEK

rdr rg rdr bco

rdr sp

rdr sp

rdr sp

1776

1789

1803

1809

1858

1873

1:1 sequence = SEK (krona) Government bookkeeping unit

Fig. 2. The Evolution of Monetary Units after 1776 (Deﬁnitions in Table 5). Abbreviations: rdr sp, riksdaler specie; rdr rg, riksdaler riksga¨lds; rdr bco, riksdaler banco; rdr rmt, riksdaler riksmynt.

conversions are possible this is a natural one as it corresponds to the common unit of account, as well as the available price index.11 The monetary units in the government accounts follow another sequence than the one-to-one sequence due to the use of parallel monetary units between 1789 and 1857. Hence we have converted the amounts in the government accounts to SEK by converting them to the units in the oneto-one sequence shown in Fig. 2. The government accounts were measured in riksdaler specie between 1855 and 1857. Riksdaler specie ceased to be the main medium of exchange and account soon after a new currency, riksdaler riksga¨lds, was introduced in 1789 at a one-to-one exchange rate with the riksdaler specie. Riksdaler riksga¨lds was subsequently relabeled riksdaler riksmynt in 1855 at one-to-one. Finally, riksdaler riksmynt was relabeled krona (SEK) in 1873 at one-to-one. Hence there is a one-to-one sequence from riksdaler specie to SEK, via riksdaler riksga¨lds and riksdaler riksmynt, from 1777 to 1873. Shortly after the introduction of riksdaler riksga¨lds in 1789, they lost value relative to the silver currency riksdaler specie and hence price levels in riksdaler specie and riksdaler riksga¨lds began to diverge. From 1789 we have converted the government ﬁgures in riksdaler specie to riksdaler riksga¨lds and hence SEK, by the ﬂoating exchange rate between them until 1803. In 1803, the riksdaler riksga¨lds became convertible to riksdaler specie at 1.5 riksdaler riksga¨lds per riksdaler specie, which we have used to convert the ﬁgures between 1803 and 1857.12

178

KLAS FREGERT AND ROGER GUSTAFSSON

2.5. Deflators Since we are dealing with very long time series, nominal values may for some purposes be of less interest. We therefore also show graphs with the nominal data deﬂated with prices, GDP or population.13 The references in this section to sources are given in the appendix on page 182. Consumer price level data has been calculated by A˚mark (1921) for the period 1732–1829. This series has been linked to the consumer price index (1830–2003) from Statistics Sweden (Statistiska centralbyra˚n, 2004a). This price index corresponds to the nominal series expressed in SEK.14 For the 1719–1731 period, we have used data from Jansson, Andersson Palm, and So¨derberg (1991, pp. 82–87) to construct an index that has been linked to the A˚mark index.15 The price level for this period should be considered to be very approximate, since it is constructed using very few goods and only include prices from the city of Stockholm. Price Level 1719-2003, index 1830=100, logarithmic scale.

GDP 1719-2003 (thousands of SEK), logarithmic scale. 10 000 000 000

10 000

1 000 100 000 000 100 1 000 000 10

1 1719 1754 1789 1824 1859 1894 1929 1964 1999

10 000 1719 1754 1789 1824 1859 1894 1929 1964 1999

Population 1719-2003 10 000 000

7 500 000

5 000 000

2 500 000

0 1719 1754 1789 1824 1859 1894 1929 1964 1999

Fig. 3. The Swedish Price Level, GDP and Population. Sources: The price level: 1719–1731, Jansson et al. (1991); 1732–1829, A˚mark (1921); 1830–2003, Statistiska centralbyra˚n (2004a). Population: 1719–1748, Andersson Palm (2001); 1749–2003, Statistiska centralbyra˚n (www.scb.se.). GDP: 1719–1799, Edvinsson (2005); 1800–1998, Krantz (2001); 1999–2003, Statistiska centralbyra˚n (2004b).

Fiscal Statistics for Sweden 1719–2003

179

Nominal GDP data at factor cost from 1800 to 1998 from Krantz (2001) have been linked forward with data from Statistics Sweden (Statistiska centralbyra˚n, 2004b) for the period 1999–2003. For the 18th century we have used data from Edvinsson (2005). Edvinsson presents real GDP that we transform to nominal ﬁgures using the linked CPI-index and then link to the series from Krantz. Data on population are taken from Andersson Palm (2001) for the period 1719–1748. Population data from 1749 until 2003 are available at the homepage of Statistics Sweden. The consumer price level, nominal GDP and population are shown in Fig. 3.

3. EXPENDITURES, REVENUES AND DEFICITS We present aggregate revenues and expenditures as potential measures of the claims of the central government of the total economy. We do not present any disaggregated data, since the classiﬁcation schemes in the government accounts have varied over time, which creates a number of difﬁculties for creating consistent long series. While we cannot rule out that we missed some transactions altogether, we believe this is a minor problem. A potentially larger problem is changes in the degree of netting between incomes and revenues. Ideally all revenues and expenditures should be recorded gross. The nettings have in particular occurred in transfer systems, which are ﬁnanced by special taxes where only net contributions have occurred in the budget. In recent years, the budget has gradually moved towards gross numbers for all transfer systems. A signiﬁcant modern example of netting is the transfers to local governments ﬁnanced by central government personal income taxes, where only the net revenues are recorded in the budget. We have calculated the net revenues for government corporations (affa¨rsdrivande verk, uppdragsverksamhet hos myndigheter) between 1810 and 1911 to conform to the data thereafter. The rationale is that these expenditures are truly beneﬁt ﬁnanced and therefore should be treated as commercial activities. Finally, it should be noted that netting does not affect the deﬁcit calculations as they affect revenues and expenditures by the same amount. Another problem is the degree to which the budget has been recorded on a cash or an accrual basis. This is a minor problem, since only the timing, not the amount, is affected. Our overall impression, documented below, is that the budget makers strove for cash-based accounting until 1993. Major differences occur on the expenditure side. Transfer payments were generally

180

KLAS FREGERT AND ROGER GUSTAFSSON

on a cash basis, while other expenditures were often on an accrual basis. An explicit attempt to use cash-based accounting also for the expenditure side was introduced in 1917. We divide the whole period into four sub-periods corresponding to different source materials. The ﬁrst is the period 1722–1809 where only ex ante budget data for expenditures and revenues exist.16 The second is the period 1810–1820. The third period begins in 1821 when ofﬁcial closed budget accounts began to be published and ends in 1911. A major revision of the budget system in 1912 marks the beginning of the ﬁnal period 1912–2003. Yearly data on total revenues, expenditures and deﬁcits are given in the Appendix. More detailed divisions are given by Fregert and Gustafsson (2005).

3.1. The Period 1722–1809 The general format of the government budget (Riksstaten), that is, ex ante measures, is given in A˚mark (1961, Tables 1, 16 and 26). It is a statement in the form of sources and uses of funds constructed so that sources equal uses. The terms used were requisitions (rekvisitioner) for sources of funds and ‘‘orders’’ (anordningar) for uses. The residual making the two sides equal is the accounting item ‘‘State deﬁcit’’ (statsbrist). It is simply the means lacking at the time of the construction of the budget, but anticipated to be ﬁlled with some combinations of loans, increased revenues or reduced expenditures. To deﬁne the measures of the total and primary deﬁcits, it is useful to formalize the bookkeeping identity. Using the abbreviations in Table 6, we have, TY M t þ LY M t þ KLY t þ Lt þ SDt ¼ PE t þ AM t þ it Bt1 þ LYSDt þ KNY t |ﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ{zﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ} |ﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ{zﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ} Sources

Uses

(4) The sources of funds consists of this year’s total funds ðTY M t þ LY M t þ Lt þ SDt Þ plus funds kept from last year ðKLY t Þ. The uses of funds consists of primary expenditures ðPE t Þ plus debt service ðAM t þ it Bt1 Þ, that is amortization and interest payments, plus payment for last year’s state deﬁcit ðLYSDt Þ plus the means kept to next year ðKNY t Þ. If the projected state deﬁcit is covered by borrowing, the total deﬁcit, computed as expenditures minus revenues, is, DEF t ¼ ðPE t þ it Bt1 Þ ðTY M t þ LY M t Þ ¼ Lt þ SDt AM t LYSDt ðKNY t KLY t Þ

(5)

Fiscal Statistics for Sweden 1719–2003

Table 6.

181

The Government Budget (Riksstaten) (1722–1809).

Sources (Rekvisitioner) This year’s means, TY M t (Lo¨pande a˚rets medel )

Uses (Anordningar)

Net: Uses–Sources

Primary expenditures, PE t Interest payment part of debt service, it Bt1 (Skuldfordringsstaten)

Subtotal Last year’s means, LY M t (Fo¨rra a˚rens medel ) Kept from last year, KLY t (Beha˚llet till fo¨rsta kvartalets behov (1722–1792)) (Beha˚llet i statens kassor (1793–1809)) New loans, Lt (La˚nemedel )

Subtotal

Total Deﬁcit

Amortization part of debt service, AM t (Skuldfordringsstaten) Last year’s state deﬁcit, LYSDt (Fyllnad i fo¨rega˚ende a˚rs statsbrist)

Kept to next year, KNY t (Beha˚llet till na¨sta a˚rs stat)

State deﬁcit, SDt (Statsbrist) Subtotal Total

Subtotal Total

Total Surplus 0

Notes: Debt service appears as a sum in the budgets and its division has to be estimated as described in the text. Primary expenditures are the sum of the categories: Civilian (Civila behov), Military (Fo¨rsvarsva¨sendet), Court (Hovet) and Payments to Pomerania (Fyllnad i Pommerska staten). Debt service is presented as a total in A˚mark (1961, Tables 16 and 60).

The right-hand side gives the deﬁcit from the ﬁnancing side, corresponding to DBt DAt in Eq. (3). The budget only shows total debt service, which must be divided into amortization (not included in the deﬁcit) and interest payments (included). For the period 1722–1777, we have approximated interest payments on central government debt from known interest rates and debt amounts for the different debt classes. From 1778, the interest payments have been presented yearly by the National Debt Ofﬁce.17 The item ‘‘payment of last year’s deﬁcit’’ appears as a balancing item on the use side during some years between 1765 and 1792. We are not sure how this item has appeared. If it represents expenditures which were included in the previous year’s expenditures but not paid that year, it constitutes arrears which should be put above the line as a true expenditure this year and subtracted from the previous year’s expenditures. Compared to putting the item below the line, some of the deﬁcit would be shifted forward in time to represent a cash, as opposed to an accrual, measure of the deﬁcit. If it represents, on the other hand, new expenditures not accounted for in the previous year’s budget, then

182

KLAS FREGERT AND ROGER GUSTAFSSON

it should be added to this year, without a deduction last year. The description in A˚mark (1961, p. 166) of the term as a balancing item suggest the ﬁrst interpretation and we have therefore not included it, that is, we have put it below the line as a purely ﬁnancial transaction. The cumulative deﬁcit in this case will be the same as if we have shifted it between the years. Since the ﬁgures are ex ante, the deﬁcit calculation will not be equal to the actual deﬁcit. A priori, there is no reason to believe that the budget ﬁgures systematically under- or overestimate the actual deﬁcit. First, the deﬁcit may be underestimated, as the budget was a political document and thus likely to err on the optimistic side. Second, the deﬁcit may be underestimated, as the revenues may be overestimated due to double counting between ‘‘this year’s means’’ (TY M t ) and ‘‘last year’s means’’ (LY M t ).18 Finally, the deﬁcit may be overestimated to the degree that the state deﬁcit is covered by expenditure cuts or revenue increases, instead of borrowing. A posteriori, the calculated budget deﬁcits are not systematically different from the change in debt, as shown in Section 5, which indicates that there were no systematic errors. Total revenues, expenditures and deﬁcits are shown in Figs. 4 and 5.19 The parliamentary government under the new constitution in 1719 managed to Real, 1722 prices (thousands of SEK)

Current prices (thousands of SEK) 30 000

30 000

20 000

20 000

10 000

10 000

0 1722

1734

1746

1758

1770

Expenditures

1782

1794

1806

0 1722

As fraction of GDP

0 .2

2 000

0 .1

1 000

1746

1758

Expenditures

Fig. 4.

1770

1782

1758

1770

1782

1794

1806

Revenues

Real per capita, 2003 prices (SEK) 3 000

1734

1746

Expenditures

0 .3

0 .0 1722

1734

Revenues

1794

Revenues

1806

0 1722

1736

1750

1764

Expenditures

Expenditures and Revenues (1722–1809).

1778

1792

Revenues

1806

Fiscal Statistics for Sweden 1719–2003

183

Current prices (thousands of SEK)

Real, 1722 prices (thousands of SEK)

15 000

15 000

7 500

7 500

0

0

-7 500

-7 500

-15 000

-15 000 1722

1734

1746

1758

1770

1782

1794

1806

1722

As fraction of GDP

1746

1758

1770

1782

1794

1806

Real per capita, 2003 prices (SEK)

0.2

2 000

0.1

1 000

0.0

0

-0.1 1722

1734

-1 000 1734

1746

1758

1770

1782

Fig. 5.

1794

1806

1722

1734

1746

1758

1770

1782

1794

1806

Total Deﬁcits (1722–1809) .

run surpluses to pay off the debt inherited from the despotic king Karl XII almost every year until Sweden’s participation in the Seven Years’ War 1757–1763 when deﬁcits occurred around 3 percent. Very large deﬁcits occurred in 1790 (20 percent of GDP) and 1792 (10 percent of GDP) in connection with a new war with Russia and then again in 1809.

3.2. The Period 1810–1820 For 1810–1820, we use the ﬁgures of Rathsman (1855) and adjust them to be consistent with later periods. The closed account ﬁgures given in Rathsman cover only a subset of the activities used in the accounts from 1821. For the overlapping year 1821, the sum of revenues in Rathsman is about 40 percent less than in the capital account. Since we lack information on the missing revenues, we chain Rathsman’s series to the new ﬁgures by multiplying his ﬁgures by 1.4. Regarding expenditures, Rathsman calculated expenditures for 1810, but for the 1811–1820 period he only submitted budget data, which only contains about half of the true expenditures during this period. Another complication during this period is the so-called war-fund

184

KLAS FREGERT AND ROGER GUSTAFSSON Current prices (thousands of SEK)

Real, 1810 prices (thousands of SEK)

300 000

300 000

200 000

200 000

100 000

100 000

0 1810

1824

1838

1852

1866

Expenditures

1880

1894

1908

0 1810

Revenues

As fraction of GDP

0.10

1 600

0.05

800

1838

1852

1866

Expenditures

Fig. 6.

1880

1852

1866

1880

1894

1908

Revenues

Real per capita, 2003 prices (SEK) 2 400

1824

1838

Expenditures

0.15

0.00 1810

1824

1894

1908

Revenues

0 1810

1824

1838

1852

1866

Expenditures

1880

1894

1908

Revenues

Expenditures and Revenues (1810–1911).

(1810–1817). The exact time of the revenues and expenditures of the war fund is uncertain. We use a summary printed in the ofﬁcial records of the parliament session in 1817–1818, Rikets sta¨nders revisorer (1817–1818). To calculate the total expenditures of the central government, we sum the recalculated revenue ﬁgures from Rathsman, the revenues of the ‘‘war-fund’’ and the revenues of the National Board of Trade (Kommerskollegium) and then add the change in government debt. That is, we calculate total expenditures residually as the sum of revenues plus the change in government debt ðGt þ I t þ it Bt1 ¼ T t þ CBT t þ DBt Þ. The data are presented in Figs. 6 and 7.

3.3. The Period 1821–1911 At the Riksdag in 1809–1810, it was decided to set up a general ledger (Rikshuvudbok) of the total Swedish government. This work was not completed until 1821 when the ﬁrst account was constructed (published in 1822). We use the published closed accounts (Capital-Ra¨kning till

Fiscal Statistics for Sweden 1719–2003

185

Current prices (thousands of SEK)

Real, 1810 prices (thousands of SEK)

75 000

75 000

50 000

50 000

25 000

25 000

0

0

-25 000 1810

1824

1838

1852

1866

1880

1894

1908

-25 000 1810

As fraction of GDP

1838

1852

1866

1880

1894

1908

Real per capita, 2003 prices (SEK)

0.04

600

0.02

300

0.00

0

-0. 02

-300

-0. 04 1810

1824

-600 1824

1838

1852

1866

1880

Fig. 7.

Table 7.

1894

1908

1810

1824

1838

1852

1866

1880

1894

1908

Total Deﬁcits (1810–1911).

The Capital Account from the General Ledger (Capital-Ra¨kning till Riks-Hufvud-Boken) (1821–1853).

Debit

Credit

Closing balance, credit t1=Opening balance debt, Bt1 Expenditures, G t þ it Bt1 Closing balance assets, ATotal t

Closing balance, debit t1=Opening NL balance assets, AR t1 þ At1 þ At1 Revenues, T t þ CBT t Closing balance debt, Bt

Total debit=Bt1 þ G t þ it Bt1 NL þAR þ At t þ At

NL Total credit=AR t1 þ At1 þ At1 þT t þ CBT t þ Bt

Riks-Hufvud-Boken 1821–1853, Kapital-Konto till Riks-Hufvud-Boken 1854–1911). The accounts were constructed as a combined balance sheet and expenditure–revenue statement called the Capital Account, which includes the government proper and the National Debt Ofﬁce. They were set up as in Table 7.

186

KLAS FREGERT AND ROGER GUSTAFSSON

The two sides in Table 7 are equal according to the bookkeeping identity, Bt1 þ G t þ it Bt1 þ ATotal ¼ ATotal t t1 þ T t þ CBT t þ Bt

(6)

where total assets at the end of period t, ATotal , are divided into real assets, t NL AR , net lending, A , and short-term ﬁnancial assets, At . t t The format of the accounts changed in 1854. First, instead of being presented as one account, the general ledger was divided in ten funds, which grew to over 80 in 1911. From each fund balances, expenditures and revenues were transferred to and summarized in the capital account. Second, the ledger was set up with net assets (or net debts), instead of assets and debts. To calculate the deﬁcit as the borrowing requirement as described by Eq. (3), we rearrange the statement into a ‘‘Sources and uses’’ statement as described in Table 8. Move the opening and closing assets to the Uses (expenditure/debit) side and likewise the opening and closing debt to the Sources (revenue/credit) side. The change in real assets and long-term ﬁnancial assets corresponds to NL investments, I t , that is, real investments and net lending, DAR t þ DAt . We then get the total deﬁcit as, NL ðT t þ CBT t Þ ¼ DBt DAt (7) DEF t ¼ G t þ it Bt1 þ DAR t þ DAt where the right-hand side shows the ﬁnancing of the deﬁcit as DBt DAt . In practice, however, this procedure is not possible, since the change in real assets and net lending reported in the capital accounts does not measure true investment expenditures, I t . First, assets appeared in the capital account long after the investments had been made. A prominent example is the national railways, which entered the general ledgers 1876 at a value of 165.5 million SEK. When new assets were introduced in the general ledger Table 8.

Capital Account Rearranged as Sources and Uses.

Sources Revenues, T t þ CBT t

Uses

Net: Uses–Sources

Expenditures, Gt þ it Bt1 NL Investment, DAR t þ DAt

Subtotal

Subtotal

Borrowing, DBt

Financial asset change, DAt

Total

Total

Total Deﬁcit

0

Fiscal Statistics for Sweden 1719–2003

187

these were balanced by an entry called ‘‘additional opening balance’’. Second, some assets were never included in the general ledger. Two examples are investments in telegraphs and telephones and in hydroelectric power and canals. The sum of these investments was during 1891–1911 approximately 68.5 million SEK. In addition, the government lent funds to investments in private railways. During the period 1854–1911, these investments totaled approximately 89.2 million SEK. We have therefore calculated the investment expenditures separately and added them to the other expenditures. Some investments were paid by the government proper and some by the National Debt Ofﬁce. The investment expenditures paid out from the National Debt Ofﬁce were the largest part.20 We calculate the total ﬁgure (for the period 1854–1911) to 598.2 million SEK. The part paid out from the government proper can be retrieved from the general ledgers.21 The problem is what to include as investment expenditures. The most obvious part is transfers to the fund of the national railways, which totals 133.4 million SEK during 1876–1911. We also add government appropriations, which was transferred to a number of different funds, to a total of 56.6 millions during 1865–1911.22 Expenditures and revenues have been recalculated to correspond with the level of netting in later periods, since all revenues and expenditures are given as gross ﬁgures, including government corporations. Thus the expenditures and revenues are overestimated relative to later periods. For example, all operational expenditures for the national railway system entered the expenditure side of the capital account, while operational revenues entered the revenue side. From 1912 only the operational surplus, that is, net revenues, entered the revenue side. We therefore must compute net revenues for all government corporations and similar activities that later are included as net revenues.23 Due to incomplete data for the period 1810–1854 we have approximated IB IB the interest payments according to it BIB t1 ¼ 0:04 Bt1 , where Bt is the interest-bearing debt, most of it to the Riksbank at 4 percent interest. From 1855 interest payments are given in the capital accounts to which we have added the interest payments of the National Board of Trade. The total expenditure, revenue and deﬁcit ﬁgures of the central government, that is, the government proper plus the National Board of Trade and the War Fund, are presented in Figs. 6 and 7 for the period 1810–1911.24 After the war in 1809, the budget was close to balanced almost every year until the 1850s. The period from the mid-1850s until 1910 is characterized by budget deﬁcits connected with the construction of the national railway system.

188

KLAS FREGERT AND ROGER GUSTAFSSON

3.4. The Period 1912–2003 Since 1912 the budget and the closed accounts have been presented as a source–use statement of ﬂows only.25 Between 1923/24 and 1995/96 the accounts were given for broken ﬁscal years with a new year beginning 1st of July. The ﬁscal year 1923 lasted from January 1st to June 30th and the ﬁscal year 1995/96 was running from 1st July 1995 to 31st December 1996. This explains the small ﬁgures for 1923 and the large ﬁgures for 1995/96.26 The structure of the budget between 1912 and 1937/38 is presented in Table 9. The consequence of the budget reform of 1911 was that all government revenues and government expenditures, including loans, amortizations and expenditures for increases in state capital assets (investment expenditures), Table 9. The Budget of the Swedish State (1912–1937/38). Sources (Swedish Term)

Revenue proper, RPt (Egentliga statsinkomster) Receipts from productive funds, RPF t (Inkomster av statens produktiva fonder)

Uses (Swedish Term)

Net: Uses– Sources

Expenditure current, EC t (Verkliga utgifter) Expenditure for increase of State Capital Assets, excl. amortization, IE t (Utgifter fo¨r kapitalo¨kning, exkl amortering)

Share in the proﬁt of the Riksbank, CBT t (Andel i Riksbankens vinst) Subtotal Capital assets employed (I anspra˚k tagna kapitaltillga˚ngar) Loans (La˚nemedel) Unspent balances from last year (Minskning av beha˚llningen a˚ reservationsanslagen)

Subtotal Amortization (Avbetalning a˚ statsskulden) Change in the cash fund (tillfo¨rt kassafonden) Unspent balances kept to next year (O¨kning av beha˚llningen a˚ reservationsanslagen)

Subtotal Total

Total Deﬁcit

Subtotal Total

Total Surplus 0

Notes: The items ‘‘Unspent balances from last year’’ and ‘‘Unspent balances kept to next year’’ refer to transfer of funds between years emanating from income items which the state could choose whether to use in the current or future ﬁscal years. Only the net amount is shown and appears as a positive number on either the source or the use side. The sum of this net amount and the change in the cash fund represent the net change in ﬁnancial assets (DA).

Fiscal Statistics for Sweden 1719–2003

189

were brought together in a single budget. The purpose was to achieve a more uniform arrangement of the budget where incomes and revenues were balanced against each other. Within the budget, current revenues and expenditures should balance and loans should only be used for capital expenditures. We calculate the deﬁcit as, DEF t ¼ ðEC t þ IE t Þ ðRPt þ RPF t þ CBT t Þ

(8)

where EC t is the expenditure current, IE t the expenditure for increase of state capital assets minus amortization of government debt RPt the revenue proper and RPF t the receipts from productive funds (net revenues from public corporations). In 1937 it was time for a new budget reform. The budget was now divided in two parts, a current or the so-called working budget (driftbudget) with revenues and current expenditures and the ‘‘capital budget (kapitalbudget)’’. The requirement that the current budget should be balanced yearly was abandoned, which instead should be balanced over the medium term. By removing this constraint, the working budget became a tool to stabilize business cycles. Table 10 describes the two budgets. To compute a total deﬁcit, it is necessary to consolidate the two separate budgets. This is done according to, DEF t ¼ ½EW Bt þ DSGC t þ ðCI t RI t Þ RW Bt

(9)

where EW Bt is the total expenditures in the working budget, DSGC t the ‘‘changes in standing government credits to public enterprises’’27 representing net lending outside the budget, CI t the capital investments, RI t the repaid investments and RW Bt the total revenue in the working budget. A summary table of the total budget appeared from 1965/66, shown as Table 11, with the deﬁcit on the source side and a new division of the categories above the line. Our method, applied from 1938/39 to 1979/80, gives the same deﬁcit result. In 1977 a proposal for a modernization of the government budget was put forward, which was adopted from the ﬁscal year 1980/81. The earlier system with separate working and capital budgets was replaced by a uniform state budget, as shown in Table 12. Items below the line are no longer explicit, but replaced by the borrowing requirement, that is, the total deﬁcit on the source side as in Table 4. The budget did not, however, include all net lending. Between 1980/81 and 1989/90 net lending, DSGC t , was included in the budget under the heading ‘‘other expenditure’’. From 1990/91, DSGC t was incorporated in the National Debt Ofﬁce’s net lending and disappeared from the budget.

190

KLAS FREGERT AND ROGER GUSTAFSSON

Table 10.

The Budget of the Swedish State (1937/38–1979/80).

Sources (Swedish Term) Current revenue (Egentliga statsinkomster) Receipts from State Capital Funds (Inkomster av statens kapitalfonder) Subtotal working budget, RW Bt

Uses (Swedish Term)

Net: UsesSources

Current expenditure (Egentliga statsutgifter) Expenditures on State Capital Funds (Utgifter fo¨r statens kapitalfonder) Subtotal Working Budget, EW Bt

Total Working Budget Deﬁcit

Capital Investments, CI t (Investeringsbemyndiganden) Repaid Investments, etc., RI t (Avga˚r kapitala˚terbetalning) Subtotal capital budget

Subtotal working and capital budget Capital meansa (Kapitalmedel ) Kept from last year (Reservationer till fo¨rega˚ende budgeta˚r) Change in cash fundb (Underskott att avfo¨ras a˚ statens budgetutja¨mningsfond )

Subtotal Capital Budget

Subtotal working and capital budget

Total

Total budget deﬁcit

Kept to next year (Reservationer till fo¨ljande budgeta˚r) Savings on capital budgeta (Besparingar som regleras inom riksga¨ldsfonden)

Subtotal

a

Total Capital Budget Deﬁcit

Subtotal Total

Total Surplus 0

Appears only in capital budget. Appears only in working budget.

b

In addition, the National Debt Ofﬁce began its own net lending in 1985/86 outside the budget. In 1997, all net lending was again included in the budget. Thus we must add net lending from the National Debt Ofﬁce during 1985/861995/96 to the expenditures. Finally, in 1997 a cash correction factor was added to the expenditure side (see the discussion in Section 5). By adding the National Debt Ofﬁce net lending and the cash correction factor, the budget deﬁcit is equal to the borrowing requirement for the central government. The deﬁcit is then straightforwardly calculated as total expenditures minus total revenues.

Fiscal Statistics for Sweden 1719–2003

Table 11.

191

Summary of the Total Budget of the Swedish State (1965/66–1979/80).

Sources (Swedish Term)

Uses (Swedish Term)

Current revenue (Skatter, avgifter m.m.) Receipts from State Capital Funds (Inkomster av statens kapitalfonder) Other ﬁnancing (Bera¨knad o¨vrig medelsfo¨rbrukning)

Net: Uses–Sources

Expenditure (Utgiftsanslag) Other uses of funds (Bera¨knad o¨vrig medelsfo¨rbrukning)

Subtotal

Subtotal

Budget deﬁcit (Underskott)

Budget Deﬁcit Budget Surplus

Total

Total

Table 12.

0

The Budget of the Swedish State (1980/81– ).

Sources (Swedish Term) Tax revenue (Skatter) Nontax revenue (Inkomster av statens verksamhet) Capital revenue (Inkomster av fo¨rsa˚ld egendom) Loan repayment (A˚terbetalning av la˚n)

Uses (Swedish Term)

Net: Uses–Sources

Expenditure (Utgiftsanslag) Other expenditure (1980/81– 1989/90) (O¨vrig medelsfo¨rbrukning) Adjustment to cash basis (1997–) (Kassama¨ssig korrigering) National Debt Ofﬁce net lending (1997–) (Riksga¨ldskontorets nettoutla˚ning)

Computed revenue (Kalkylma¨ssiga inkomster) Grants from the EU (1994/ 95–) (Bidrag fra˚n EU ) Subtotal

Subtotal

Total Deﬁcit (Borrowing Requirement)

Borrowing requirement (=total deﬁcit) Total

Total

0

192

KLAS FREGERT AND ROGER GUSTAFSSON Current prices (thousands of SEK), logarithmic scale

Real, 1912 prices (thousands of SEK), logarithmic scale

1000000000

1000000000

100000000

100000000

10000000

10000000

1000000

1000000

100000 1912 1924/25 1937/38 1950/51 1963/641976/77 1989/90 2003

Expenditures

100000 1912 1924/25 1937/38 1950/51 1963/641976/77 1989/90 2003

Revenues

Expenditures

Asfraction of GDP

Real per capita, 2003 prices (SEK)

0.75

3000

0.50

2000

0.25

1000

0.00 1912

1924/25 1937/38 1950/51 1963/64 1976/77 1989/90

Expenditures

Fig. 8.

2003

0 1912

1924/25 1937/38 1950/51 1963/64 1976/77 1989/90

Revenues

Expenditures

2003

Revenues

Expenditures and Revenues (1912–2003).

Current prices (thousands of SEK)

Real, 1912 prices (thousands of SEK)

250 000 000

250 000 000

150 000 000

150 000 000

50 000 000

50 000 000

-50 000 000

-50 000 000

-150 000 000

-150 000 000

1912 1924/25 1937/38 1950/51 1963/64 1976/77 1989/90 2003

1912 1924/25 1937/38 1950/51 1963/64 1976/77 1989/90 2003

As fraction of GDP

Real per capita, 2003 prices (SEK)

0.30

800

0.15

400

0.00

0

-0.15 1912

Revenues

1924/25 1937/38 1950/51 1963/64 1976/77 1989/90 2003

Fig. 9.

-400 1912

1924/25 1937/38 1950/51 1963/64 1976/77 1989/90 2003

Total Deﬁcits (1912–2003).

Fiscal Statistics for Sweden 1719–2003

193

The expenditure, revenue and deﬁcit ﬁgures for the 1912–2003 period are presented in Figs. 8 and 9. The large deﬁcits in the late 1970s/early 1980s and the early 1990s triggered by economic crises stand out and are larger as shares of GDP than during the two world wars. In contrast, the deﬁcits between 2 and 4 percent of GDP in the ﬁrst half of the 1930s, when Sweden as one of the ﬁrst countries experimented with expansionary ﬁscal policy, appear very moderate.

4. GOVERNMENT DEBT Currently there are three ofﬁcial measures of central government debt in Sweden. We report the total gross debt at par value, which is collected by the Swedish National Debt Ofﬁce (Riksga¨ldskontoret). The Swedish National Financial Management Authority (Ekonomistyrningsverket) reports the total gross debt at par value minus the holdings of government debt by government authorities, the so-called consolidated government debt.28 Finally, Statistics Sweden (Statistiska centralbyra˚n) reports the consolidated government debt at market value, according to EU and UN national account standards. There is no single best measure of government debt. Here we focus on the consistency between the change in debt and the deﬁcit. Consistent measures can be constructed with either at par values or actual values. If government debt initially is sold below par, that is, at a discount, the ﬂow of money from new bond sales is less than the change in debt at par value. The change in the debt at par value will correspond to the deﬁcit in this case, if the discount is included as an expenditure. This has been the case in the Swedish budget at least since 1912. By the same token, if the discounts are not included in the budget as an expenditure, the deﬁcit will be smaller than the change in debt at par value (and equal to the change in debt at market value). Robert Barro (1987) argued that the nominal value of the government debt in the United Kingdom gives an exaggerated picture of the true debt burden during the 18th and 19th century due to large discounts at the issue of new debt. In principle, the government can choose any par value on new debt issues and then sell at a discount.29 To calculate a more realistic debt measure, he instead used the cumulated deﬁcits as the measure of government debt from a benchmark in 1700. (Thus he assumed that the discounts are not included in expenditures.) We stick with the ofﬁcial debt ﬁgures at par, as they are consistent with the method in the government budget accounts, and we wish to study if there are other possible sources of discrepancies between the deﬁcit and the change in the debt. In addition, the

194

KLAS FREGERT AND ROGER GUSTAFSSON

discounts below par have been small so the difference between the evolution of the debt valued at par and at actual value is negligible. Between 1857 and 1960/61, the cumulative discount constituted less than half of a percent of the total debt (Riksga¨ldskontorets a˚rsbok 1960/61). We follow the ofﬁcial statistics of foreign debt which is converted to SEK at the rate when the bond was issued before 1988/89. After 1988/89 the debt is recorded at the actual exchange rate.30 During the years the composition of government debt has changed quite substantially. We prefer to use the ofﬁcial ﬁgures as much as possible. One problem is the ‘‘interest-bearing’’ versus ‘‘non-interest-bearing’’ debt. Up until 1809, we do not have any information how the debt was divided into interest-bearing and non-interest-bearing debt, why we include all debt. During the ﬁrst half of the 19th century, the non-interest-bearing component changed through political decisions several times. We therefore choose to include all debt, that is, interest-bearing plus non-interest-bearing, up until 1857. From the mid-1830s the non-interest-bearing part was quite small and stable. From 1858 we include only interest-bearing debt as presented by the National Debt Ofﬁce. We divide the description in three periods according to data availability. For the ﬁrst period 1719–1776, yearly data are not available. The second period covers 1777–1857 when the National Debt Ofﬁce presented yearly ﬁgures. The third period stretches from 1858, when the government began large-scale foreign borrowing, until today.31 Yearly data are given in the Appendix. 4.1. The Period 1719–1776 A National Debt Ofﬁce under the supervision of the parliament, Riksens sta¨nders kontor, was set up in 1719 to pay off the debt accumulated during the reign of Karl XII between 1697 and 1718. Almost no new loans were taken up before 1740 and budget surpluses were used to pay off this socalled old debt. From 1740 the government started taking up new loans to ﬁnance the war against Russia (1741–1743). The ‘‘old debt’’ was not paid off until well into the 19th century. The different parts of the debt have been estimated separately, using available data on the debt in certain years in combination with available data on new loans and amortization. We calculate the debt recursively back in time from known amounts, using the relation, Bt1 ¼ Bt Lt þ AM t

(10)

Fiscal Statistics for Sweden 1719–2003

Table 13.

195

Composition of Government Debt Incurred before 1719, the ‘‘Old Debt’’.

Swedish Name

Bankoga¨lden Fo¨rsa¨krings- och lo¨nesedlar

Nummerga¨lden Kronofo¨rpantningar

Diverse kreditorer

Explanation

Total 1719 (dsm)

Loans from the Riksbank Interest bearing promissory notes paid to government employees in exchange for token money (mynttecken and myntsedlar) issued 1715–1718. Number debt. Private loans classiﬁed in 11 groups in 1719 in order of priority. Swaps of income from government properties against ﬁxed down payment to the government for a limited period. Short-term loans from various creditors

6,910,796 11,049,911

24,139,180 2,301,358

2,413,481

The ‘‘old debt’’ can be divided into ﬁve parts as described in Table 13 with the initial values in 1719. The largest part consists of: debt to the Riksbank, the insurance- and salary-notes and the number debt. The Riksbank debt is known from the bank’s yearly balance sheet (given in Sveriges Riksbank, part V). The other debts were left to the newly formed National Debt Ofﬁce to pay off. The main books of the National Debt Ofﬁce are arranged in four-year periods, why we have interpolated to estimate yearly debt ﬁgures.32 We have calculated the initial value of debt in 1719 from known amounts in 1718. The amount of the so-called insurance notes, issued initially in 1719, was estimated. They were issued in exchange for token money (mynttecken and myntsedlar), issued for war ﬁnance 1715–1718. The amount in 1719 is valued at 50 percent of their initial value following their devaluation in 1719. The number debt consisted of different types of loans. It contained loans from churches, unpaid wages, bills, etc. The name ‘‘number debt’’ refers to its division into 12 groups (labeled 1–11 and one group ‘‘without number’’), which were to be paid off beginning with group 1. To this debt, interest arrears must be added, which had the lowest priority. The ‘‘new debt’’ can be divided into ﬁve parts as described in Table 14. To calculate the yearly debt ﬁgures, we use information on speciﬁc new loans and the amortization of old loans given in A˚mark (1961, chapter 10). Further information can be gained by studying the debt service ﬁgures in the proposed budgets. We also make use of the inventories of the total debt that were made in 1764 and in 1777.33

196

KLAS FREGERT AND ROGER GUSTAFSSON

Table 14. Composition of Government Debt Incurred after 1740. Swedish Name

Bankoga¨ld Utrikes la˚n Lotterila˚n Andra inrikes la˚n La˚n fra˚n publika kassor

Explanation

Start Year

Loans from the Riksbank Foreign debt Lottery loans Other domestic debt Loans from public depositories

1743 1759 1758 1751 1731

Maximum Amount, dsm (Year) 49,585,997 29,444,790 6,650,175 7,793,281 4,437,240

(1772) (1776) (1759) (1764) (1764)

The total debt is displayed in Fig. 10 below. The steady debt reduction after 1719 from budget surpluses resulted in a two-thirds reduction of the real capita debt by 1756, when a new modest real debt build-up began with the Seven Years’ War in 1757.34 In nominal terms, the total debt tripled between 1757 and 1776, most of it due to inﬂation reﬂecting that the largest part of the new debt was lent and monetized by the central bank. The increase in debt from the late 1760s was caused by new loans taken up to ﬁnance silver purchases in preparation of the introduction of a convertible currency in 1777 to replace the paper currency introduced in 1743.

4.2. The Period 1777–1857 Between 1777 and 1809, yearly debt ﬁgures were presented by the National Debt Ofﬁce, which are given in A˚mark (1961). We include the debt issued by Riksga¨ldskontoret after 1789 in the form of short-term notes, the so-called riksga¨ldssedlar (national debt notes), which quickly turned into a medium of exchange. This part of the debt has the same character as the loans from the Riksbank as discussed in Section 6. Between 1811 and 1815, the foreign debt was eliminated through an effective default by the Swedish parliament to compensate for the losses suffered by Sweden in the Napoleonic wars (see A˚mark, 1961, pp. 654–660). In addition to the debt handled by the National Debt Ofﬁce, the government proper had during the period 1808–1830 a debt to the Riksbank, which has to be added to the debt ﬁgures from the National Debt Ofﬁce. This debt, which arose to cover expenditures for the 1808–1809 war, was taken over by the National Debt Ofﬁce in 1830.

Fiscal Statistics for Sweden 1719–2003

197

Current prices (thousands of SEK)

Real, 1719 prices (thousands of SEK)

20 000

20 000

15 000

15 000

10 000

10 000

5 000

5 000

0 1719

1727

1735

1743

1751

1759

1767

1775

0 1719

As fraction of GDP 4 500

0.4

3 000

0.2

1 500

0.0

1727

1735

1743

1751

Fig. 10.

1759

1735

1743

1751

1759

1767

1775

Real per capita, 2003 prices (SEK)

0.6

1719

1727

1767

1775

0 1719

1727

1735

1743

1751

1759

1767

1775

Central Government Debt (1719–1776).

For most of the 1815–1850 period, we only have semiannual data. This is a minor problem, since there were no dramatic changes in the debt. (The only major changes were the repayment of the debt handled directly by the government proper, and for this part of the debt we have annual data from the balance sheet of the Riksbank.) We use the income–expenditure statement of the National Debt Ofﬁce on new loans and on amortizations to approximate the debt for the missing years. As can be seen from Fig. 11, the central government debt ﬂuctuated considerably in the 18th century and the ﬁrst decades of the 19th century.35 We see a decline during the late 1770s when the Riksbank debt was written off (see A˚mark, 1961, p. 616). Then there is a sharp rise in connection with the Russian war (1788–1790), when the debt rose over 200 percent. In 1808–1809 a new war with Russia caused the debt to rise by 50 percent. The period from 1811 to around 1830 was characterized by the repayment of the large war debts. For the remainder of the period the debt was small and stable. The debt in real terms, as a fraction of GDP, and real per capita show similar patterns as the nominal debt but starts to decline already in 1792 due to inﬂation and from 1815 they move in line with the nominal debt.36

198

KLAS FREGERT AND ROGER GUSTAFSSON Current prices (thousands of SEK)

Real, 1777 prices (thousands of SEK)

45 000

45 000

30 000

30 000

15 000

15 000

0 1777

1788

1799

1810

1821

1832

1843

1854

0 1777

As fraction of GDP 5 000

0.4

4 000

0.3

3 000

0.2

2 000

0.1

1 000

1788

1799

1810

1821

Fig. 11.

1832

1799

1810

1821

1832

1843

1854

Real per capita, 2003 prices (SEK)

0.5

0.0 1777

1788

1843

1854

0 1777

1788

1799

1810

1821

1832

1843

1854

Central Government Debt (1777–1857).

4.3. The Period 1858–2003 For this period we use the ofﬁcial data from the National Debt Ofﬁce and add, as in the previous period, the debt of the National Board of Trade. The total debt is shown in Fig. 12.37 Between 1923/24 and 1995/96 the state budget accounts were given for broken ﬁscal years with a new year beginning 1st of July, so we have chosen the debt ﬁgures at midyear (30 June) to conform with the budget data. As can be seen from Fig. 12, the debt is more or less continuously rising after 1858, both in nominal and in real terms. Note that the debt in nominal and real values is presented in logarithmic form. The initial increase in the debt was due to government borrowing abroad for the building of the Swedish national railway system. The debt-to-GDP and real per capita debt ﬁgures show similar patterns. The debt went up sharply in three periods: during World War II, during the oil crises between 1975 and 1985 and after a large economic and ﬁnancial crisis in the early 1990s.

Fiscal Statistics for Sweden 1719–2003

199

Current prices (thousands of SEK), logarithmic scale

Real, 1858 prices (thousands of SEK), logarithmic scale

10 000 000 000

10 000 000 000

100 000 000

100 000 000

1 000 000

1 000 000

10 000 1858

1878

1898

1918 1937/38 1957/58 1977/781998

10 000 1858

200 000

0 .9

150 000

0 .6

100 000

0 .3

50 000

1878

1898

1918

1937/38 1957/58 1977/78

Fig. 12.

1898

1918 1937/38 1957/58 1977/78 1998

Real per capita, 2003 prices (SEK)

As fraction of GDP 1 .2

0.0 1858

1878

1998

0 1858

1878

1898

1918 1937/38 1957/58 1977/78 1998

Central Government Debt (1858–2003).

5. ACCUMULATED DEFICITS VERSUS DEBT EVOLUTION In this section, we check the consistency of the debt data with the data of revenues and expenditures. The two sets of data are linked through the one-period budget constraint Eq. (3). This is, however, an approximation and not a true accounting identity as shown below. To examine the reliability of the data, we investigate the difference between the deﬁcit and the change in the debt. Accounting systems, such as the SNA, decompose changes in assets and liabilities into two types: (a) net ﬂow of transactions and (b) ‘‘other changes’’, that is, changes not linked to transactions, in turn divided into price changes (capital or holding gains) and volume changes (e.g. defaults). Thus we should subtract from the right-hand side of Eq. (3) a term reﬂecting ‘‘other changes’’,nt , such that the right-hand side only reﬂects the transactions part and thereby matches the deﬁcit, a transaction measure. Using ofﬁcial statistics, we must also take into account the possibility of ‘‘errors and omissions’’ in the recording of transactions and debts. We

200

KLAS FREGERT AND ROGER GUSTAFSSON

should therefore subtract left-out expenditures and revenues, eDEF , from the t right-hand side of Eq. (3) and add left-out debt items, eBt . Thus the accounting identity is, DEF t DBt DAt nt eDEF þ eBt . t

(11)

The recorded deﬁcit can be larger than the recorded change in debt because: (i) some of the deﬁcit is paid by using government ﬁnancial assets, DAt , (ii) there are reductions in government debt that arise from ‘‘other changes’’, nt , or (iii) there are left-out revenues, expenditures and debt items, et . As pointed out in Section 4, capital losses due to initial discounts of new debt are included in the budget as expenditures, and hence increase the deﬁcit, such that the deﬁcit is equal to the change in the debt at par value. Still there may be changes in the debt reﬂecting ‘‘other changes’’,nt , which today are accounted for by the National Debt Ofﬁce under the heading ‘‘debt dispositions’’. An example of holding changes is currency adjustment of debts denominated in foreign currency. In the ﬁscal year 1992/93, when the Swedish currency began to ﬂoat (19 November 1992), the market value of the foreign debt in SEK increased by 39.5 billions. Since the foreign debt is measured at the actual exchange rate since 1988, this led to an increase in the foreign debt without a corresponding expenditure. Before 1988, the foreign debt was recorded at the initial cost and there is no error. An example of volume change is the transfer of government bonds in 2001 from the social security system (outside the central government), to the National Debt Ofﬁce (inside the central government), which reduced the central government debt by 85 billions SEK. ‘‘Errors and omissions’’ of transaction ﬂows occur for two reasons. First, there are timing errors occur if revenues and expenditures are recorded on an accrual basis, that is, when the resource ﬂow takes place, as opposed to a cash basis. Since 1993, the Swedish accounting system follows an accrual basis, in line with SNA 1993 and IMF (2001). Before that, our impression is that the goal was to use a cash basis. Second, ‘‘errors and omissions’’ arise from off-budget transactions not accounted for in the government budgets. An example is the temporary wealth tax on life insurance companies 1986–1988, which was not included in the budget. Since 1997, an entry called ‘‘adjustment to cash basis’’ (kassama¨ssig korrigering) is present on the expenditure side of the budget. It is constructed as the difference between the net budget-related transaction ﬂows of the government proper’s checking account at the Riksbank and the accounting budget surplus. This correction takes care of all the ‘‘errors and

Fiscal Statistics for Sweden 1719–2003

201

omissions’’ due to the timing problem and the off-budget payments, so there are no et after 1996. It also makes the ofﬁcial deﬁcit identical to the borrowing requirement of the central government (statens la˚nebehov).38 Eq. (12) shows that it is not necessary for the budget deﬁcit to be equal to the change in debt as the difference, d t , between them is equal to, d t ¼ DEF t DBt ¼ DAt nt eDEF þ eBt t

(12)

Even with no errors, there may thus be systematic non-zero differences. On the other hand, unsystematic and small differences indicate that the data are internally consistent in the sense that there are no systematically left-out debts or transactions. Fig. 13 shows the individual differences, d t , as fraction of expenditures, while Fig. 14 shows the debt, normalized to zero at the start of each period, and the accumulated deﬁcits where their difference is the accumulated P difference, Tt¼1 d t . The differences in Fig. 13 appear to be on average zero. This can also be seen from the small differences between the debt and the accumulated deﬁcits in Fig. 14. Note that a large difference in a single year leads to a permanent increase in the accumulated differences. Thus for 1722-1809

1810-1911

0.70

0.70

0.35

0.35

0.00

0.00

-0.35

-0.35

-0.70 1722

-0.70 1734

1746

1758

1770

1782

1794

1806

1810

1824

1838

1852

1866

1880

1894

1908

1912-2003 0.70

0.35

0.00

-0.35

-0.70 1912

Fig. 13.

1924/25 1937/38 1950/51 1963/64 1976/77 1989/90

2003

The Difference, d t , between the Deﬁcit and the Change in Debt as a Fraction of Total Expenditures.

202

KLAS FREGERT AND ROGER GUSTAFSSON 1722-1809

1810-1911 700 000

40 000 30 000

500 000

20 000 300 000 10 000 100 000

0 -10 000 1722

1734

1746

Debt

1758

1770

1782

1794

1806

-100 000 1810

Accumulated Deficits

1824

1838

1852

Debt

1866

1880

1894

1908

Accumulated Deficits

1912-2003 10 000 000 000

100 000 000

1 000 000

10 000 1912 1924/251937/381950/511963/641976/771989/90 2003

Debt

Fig. 14.

Accumulated Deficits

Accumulated Deﬁcits and Debt Evolution.

certain periods, such as 1860–1900, the accumulated deﬁcits and the debt move in parallel at different levels due to a few differences in the same direction at the start. One can also note that the differences become larger farther back in time. In particular, the differences are large during the 18th century (standard deviation 30 percent of expenditures), when the deﬁcit measures also include forecasting errors, since we use budget data (as opposed to closed accounts) for this period. The standard deviation diminishes over time and is down to 4 percent for the 1938–2003 period. We use t-tests to check whether there appears to be systemically left-out items such that the change in debt would not be an unbiased proxy for the deﬁcit. Table 15 presents the results of the tests of the null hypotheses that the difference has a zero mean and is not autocorrelated of the ﬁrst order. The null hypothesis that the mean is equal to zero is not rejected at any conventional signiﬁcance level, for any period. The differences do not exhibit signiﬁcant autocorrelation with the exception of the period 1912–2003. This is due to the sub-period 1912–1937, since the period 1938–2003 does not exhibit signiﬁcant autocorrelation. This indicates that the change in debt is an unbiased proxy for the deﬁcit and that there are no systematically left-out items. It should be

Fiscal Statistics for Sweden 1719–2003

203

Table 15. Statistical Properties of the Difference, d t , between the Deﬁcit and the Change in Debt as a Percentage of Expenditures. d (%)

Number of Observations 1722–1809 1810–1911 1912–2003 1938–2003

88 102 92 66

p-Values

Mean

Std. dev.

H0:m=0

H0:r=0

0.0118 0.0010 0.0086 0.0016

0.2976 0.1456 0.0916 0.0376

0.7103 0.9459 0.3686 0.7386

0.1940 0.3104 0.0000 0.6652

Notes: The column ‘‘H0: m ¼ 0’’ gives probability values from a t-test for zero mean. The column ‘‘H0: r ¼ 0’’ gives probability values from a t-test for no ﬁrst-order autocorrelation (the correlation between d t and d t1 ).

noted that if we use the absolute values of the differences instead of the fraction of expenditure measure, we cannot reject the null hypothesis of mean zero and the null hypothesis of no correlation for any period. We conclude that both left-out items and revaluations have been decreasing over time but are on average zero from the beginning, since it is unlikely that these two sources of differences are negatively correlated. In other words, we feel conﬁdent that the consolidations cover most central government activities and that the different sources for deﬁcit and debt data that we have used are consistent. For a more detailed discussion of the possible causes of large differences in certain years, see Fregert and Gustafsson (2005).

6. THE CONSOLIDATED CENTRAL GOVERNMENT AND FISCAL SEIGNIORAGE For the purpose of joint analysis of ﬁscal and monetary policy, it is useful to look at the consolidated central government (ﬁscal plus monetary branch). In particular, it enables us to calculate seigniorage revenue in a manner, which is consistent with the general accounting principles presented in Section 2.39 The central bank budget constraint, derived from its balance sheet and income–expenditure statement, can be written as, CB DH t þ DAt þ it BCB t1 þ i t Bt1 þ OS t ¼ DBt þ DBt þ CBT t þ OU t |ﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ{zﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ ﬄ} |ﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ{zﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄﬄ} Sources

(13)

Uses

The central bank receives funds from: new high-powered money, DH t ; new government deposits at the central bank, DAt ; interest income on its

204

KLAS FREGERT AND ROGER GUSTAFSSON

holdings of domestic government bonds, it BCB t1 , and non-government and foreign bonds, it Bt1 ; and other sources, OS t .40 The central bank uses funds to: buy domestic government bonds,DBCB t ; buy domestic non-government and foreign bonds, DBt ; transfer funds to the ﬁscal branch, CBT t ; and other uses, OU t .41 Consolidation is achieved by adding the one-period budget constraints of the ﬁscal branch Eq. (3) and of the central bank Eq. (13),42 T t þ CBT t þ DBt þ DH t þ DAt þ it BCB t1 þ it Bt1 þ OS t ¼ G t þ I t þ it Bt1 þ DAt þ DBCB t þ DBt þ CBT t þ OU t

(14)

Eq. (14) can be simpliﬁed as,43 T t þ S t þ DBPublic ¼ G t þ I t þ it BPublic t t1

(15)

where BPublic ¼ Bt BCB is government debt held by the public and S t is t t seigniorage, St ¼ DH t DBt þ it Bt1 þ ðOS t OU t Þ (16) Seigniorage represents a true revenue source for the consolidated government as it can be used to ﬁnance expenditures. It consists of the ﬂow of new high-powered money, which is not used to buy domestic non-government bonds and foreign bonds, DH t DBt , interest on non-government bonds, it Bt1 , and other net inﬂow to the central bank, OS t OU t .44 Rearranging Eq. (15) gives us an expression for the consolidated budget deﬁcit, Public ðT t þ St Þ ¼ DBPublic DEF C (17) t ¼ G t þ I t þ it Bt1 t where DEF C t is the deﬁcit of the consolidated central government. We note three differences between the consolidated and the ﬁscal branch deﬁcit. First, seigniorage enters as additional revenue for the consolidated central government. Second, only interest payments to the public matter for the consolidated deﬁcit, since the interest payments from the ﬁscal branch to the central bank are an internal transaction that washes out as shown by Eq. (15). Third, central bank transfers do not matter, since they also represent internal transactions as shown by Eq. (15). Table 16 shows the source–use statement for the consolidated government corresponding to Table 4 with the budget deﬁcit below the line and seigniorage above the line as sources.

Fiscal Statistics for Sweden 1719–2003

Table 16.

205

Fiscal and Monetary Branch Sources and Uses. Sources (Inkomster)

Monetary branch (central bank) Fiscal branch

Seigniorage=net sources= St ¼ DH t DBt þit Bt1 þ ðOSt OU t Þ Taxes and other revenues, T t

Public Budget deﬁcit, DEF C t ¼ DBt Total

Uses (Utgifter)

Current expenditures, G t Investment, I t Interest on government debt, it BPublic t1 Total

For practical reasons, we use an alternative expression of seigniorage obtained from combining the deﬁnition of S t in Eq. (16) with the central bank budget constraint in Eq. (13),45 CB St ¼ DBCB (18) t DAt þ CBT t i t Bt1 Seigniorage is here expressed as the sum of ﬁscal branch net borrowing from the central bank, DBCB t DAt , and transfers from the central bank to the ﬁscal branch net of interest payments from the ﬁscal branch to the central bank, 46 CBT t it BCB t1 . We include the debt in riksga¨ldssedlar issued by the National Debt Ofﬁce during the war 1789–1792 in the debt owned by the central bank, though they were initially considered public debt, but as mentioned in Section 3.2, they quickly turned into a currency of its own. The decision in 1803 to let the Riksbank redeem 15 out of the 18 million riksdaler riksga¨lds against 10 million riksdaler specie, conﬁrmed the monetization of this debt.47 Yearly data on seigniorage, the interest on public debt, the consolidated deﬁcit and the public debt are presented in the Appendix. Fig. 15 shows that ﬁscal seigniorage has been highly volatile around a mean close to zero. Most of the variation is due to changes in ﬁscal branch borrowing from the 48 Riksbank, DBCB t , and is highly correlated with the ﬁscal branch deﬁcit. Fiscal seigniorage appears to have been used most of the time only for shortterm ﬁnance, consistent with ﬁxed exchange rate regimes for most periods. Table 17 summarizes expenditures, revenues, deﬁcits and debts according to the two deﬁnitions, all given as shares of GDP. Consolidated deﬁcits and debts show the same pattern across periods as ﬁscal branch measures, but are generally smaller. Partly this reﬂects seigniorage revenues to the government.49 Seigniorage tends on average to be small and positive during ﬂoating periods and negligible under ﬁxed exchange rate periods.

206

KLAS FREGERT AND ROGER GUSTAFSSON

0.12

0.08

0.04

0.00

-0.04

-0.08

-0.12 1722

1747

1772

1797

Fig. 15.

Table 17.

ðG þ IÞ=Y iB=Y T=Y CBT=Y iBP =Y S=Y DEF =Y DEF C =Y B=Y BP =Y

1822

1847

1872

1897

1922 1946/47 1971/72 1997

Fiscal Seigniorage as Fraction of GDP.

All Variables as Percentage Shares of GDP.

1722–1809 (%)

1810–1911 (%)

1912–2003 (%)

Float. E.R. (%)

Fixed E.R. (%)

WWI (%)

WWII (%)

7.1 0.6 7.5 0.0 0.4 0.0 0.3 0.0 27.2 17.4

6.2 0.4 6.1 0.1 0.3 0.0 0.4 0.4 10.1 8.4

23.5 2.5 22.5 0.3 2.3 0.3 3.3 3.0 40.4 36.7

13.2 1.4 12.9 0.1 1.2 0.3 1.6 1.2 29.8 21.7

11.5 1.0 11.2 0.1 0.8 0.0 1.2 1.1 22.8 19.7

9.8 0.7 8.2 0.1 0.7 0.1 2.1 2.2 18.9 18.8

24.7 1.2 15.8 0.1 1.2 1.1 10.1 9.0 46.3 40.1

Notes: Floating exchange rate periods: 1741–1776, 1789–1803, 1809–1832, 1914–1922, 1931–33, 1976–1982 (repeated devaluations) and 1992–2003; WWI=1914–1918, WWII=1940–1945.

7. CONCLUSIONS We have presented two sets of data: ﬁscal branch measures and consolidated (ﬁscal and monetary branch) measures. Both should be useful for

Fiscal Statistics for Sweden 1719–2003

207

macroeconomic research, particularly studies that focus on the interplay between ﬁscal and monetary policy. Depending on the institutional set-up, studies of causation between revenues and expenditures and the sustainability of ﬁscal policy may use either or both measures. The most salient feature of the data is the recent rise in government debt as a fraction of GDP. It passed 60 percent in the late 1970s, the previous record seen only in connection with the wars of Karl XII and World War II. The fast debt build-ups 1978–1982 and 1991–1994 dwarf the slow increase between 1858 and 1914, when the government borrowed in international markets to build a national railway system. Remarkable are also the steady repayments of war-induced debts, inherited from despotic kings, through budget surpluses between 1719 and 1756 and between 1810 and 1854 under new proto-democratic constitutions. The major break in the series occurs in 1821 when published closed accounts begin. Major changes in the presentation of the budget occurred in: 1912, when the budget was uniﬁed to show only ﬂows; 1938, when the accounts were divided into a current and capital account; 1980, when the budget was uniﬁed; and 1996, when the deﬁcit was deﬁned as the borrowing requirement. The reliability of the ﬁgures increases by the nearness to the present as indicated by the decreasing difference between the change in debt and the budget deﬁcit. For future research, we believe the construction of the current deﬁcit is the most fruitful, which also corresponds to recent efforts by IMF to redirect attention from borrowing requirements to changes in government wealth.

NOTES 1. Data on revenues, expenditures and debt are available yearly from 1692 for the United Kingdom in Mitchell (1988, 1998b) and for the United States from 1789 in Historical Statistics of the United States: Millennial Edition (2006). Mitchell (1998b) includes revenue and expenditure data for Austria (1781–), Belgium (1830–), France (1815–), the Netherlands (1845–) and Russia (1805–). In addition, data are available for a number of former French and English colonies from the ﬁrst half of the 19th century (Mitchell, 1998a, 1998c). 2. For an overview of ﬁscal and monetary policy in Sweden 1668–1931, see Fregert and Jonung (1996). 3. See Blejer and Cheasty (1991) for a general discussion on deﬁnitions and measurement problems. 4. Finland was ceded to Russia in 1809 after a war. Pomerania was occupied by France in 1807 and ceded formally in 1815. Wismar was ‘‘lent’’ to Mecklenburg

208

KLAS FREGERT AND ROGER GUSTAFSSON

in 1803 against a down payment to Sweden. Sweden had the right to repurchase Wismar after 100 years at the price of the down payment compounded at QJ;3 percent yearly interest. In reality it was a sale, since no party foresaw a return to Sweden. 5. The size of the Finnish budget varied during the years, but was usually below 10 percent of the total budget. The average for the 1722–1792 period was 8.2 percent. 6. From 1809, Konvojfonden and Krigsmanhuskassorna were included in the government proper (Statsverket), while Manufakturfonden was administered by the National Board of Trade from 1809. 7. This is the term used in Blejer and Cheasty (1991), see also below on the deﬁnition of the deﬁcit. 8. The deﬁnition of the deﬁcit as the borrowing requirement has been criticized, since it bears no direct relation to the change in the government’s net wealth position, nor its change in ﬁnancial wealth. The reason is that all expenditures affect the borrowing requirement, but only current expenditures affect net wealth, since investment expenditures are used to buy real or ﬁnancial assets. Instead it has been suggested that a better deﬁcit measure is the ‘‘current deﬁcit’’ that only includes current and interest expenditures which reduce total wealth. This measure would exclude all ﬁnancial and real investment and sales of real assets from the deﬁcit. IMF (2001) on this ground advocates a shift to the current deﬁcit, by IMF termed the operating balance. Attempts to calculate net wealth changes can be found in Eisner and Pieper (1984) and Hamilton and Flavin (1986). Against this line of reasoning, it may be argued that most government assets are not saleable and therefore do not realistically back up government borrowing. Financial market analysts and macroeconomists overwhelmingly still use the borrowing requirement and total debt to analyze the sustainability of ﬁscal policies. 9. This is not a true accounting identity, since it leaves out changes in debts and assets unrelated to transactions. With real data accounting errors must also be added. See further Section 5. 10. Financing by high-powered money does not appear explicitly since all new debt, whether sold to the public or the central bank, is included for the ﬁscal branch (the ofﬁcial budget). When the ﬁscal and monetary branches are consolidated, highpowered money appears explicitly as shown in Section 6. 11. See Jo¨rberg (1972, Vol. 1, pp. 81–83). 12. Confusingly, Riksdaler specie, literally riksdaler in silver specie, was sometimes used to denote the paper currency riksdaler riksga¨lds according to Ahlstro¨m (1972). This means that we cannot be sure which unit of account was used in the government budgets between 1789 and 1803. We ﬁnd no indication in A˚mark (1961) on this problem. From the consistent evolution of the ﬁgures before and after the period 1789–1803, we have treated the ﬁgures as being literally in riksdaler specie (and therefore needing to be translated to riksdaler riksga¨lds). After 1803 the currency in the government accounts is explicitly in riksdaler banco, the new name for riksdaler specie after 1803. (The riksdaler banco became inconvertible in 1809, but the 1.5 exchange rate to the riksdaler riksga¨lds continued.) 13. The deﬂators are available in Appendix M in Fregert and Gustafsson (2005).

Fiscal Statistics for Sweden 1719–2003

209

14. Speciﬁcally, we use A˚mark’s ‘‘unreduced price series’’ which corresponds to the nominal series in SEK as explained in Heckscher (1942, p. 22). 15. The price index 1719–1731 is an unweighted average of prices of rye, barley, wheat, hops, butter, pork, tallow, herring, salmon, ﬁrewood and bar iron in the city of Stockholm. 16. The ﬁrst general ex post account for the whole central government in Sweden was constructed in 1622 by a Dutchman, Abraham Cabeljau. General accounts were constructed almost yearly until 1677 and between 1688 and 1711. The increasing size and complexity of the government in combination with the mix of monetary and inkind payments led to the abandonment of general accounts. Attempts to reestablish them recurred through the 18th century, but without result. In 1810 the Riksdag decided to create general accounts and the ﬁrst one was published in 1821. See Stuart and Rystedt (1905, chapter 1) and A˚mark (1961, pp. 72–75). 17. See Appendix G in Fregert and Gustafsson (2005) for the exact calculations. 18. ‘‘This year’s means’’ refers to expected revenues during the ﬁscal year emanating from taxes formally leveled the same year and ‘‘last year’s means’’ refers to taxes levied previous years, but collected this year. According to A˚mark (1961, p. 99), there may be some double counting because not all ‘‘this year’s means’’ will actually be collected during the year and thus appear later as ‘‘last year’s means’’. In principle, there should be no double counting as the instructions for the construction of the budget was to estimate the likely actual collection based on previous experience. A˚mark argued that it was more likely that the revenue calculation was an overestimate than an underestimate and thus there is probably some double counting. 19. Separate data for the government budget and the sum of the National Debt Ofﬁce, the special funds and other ‘‘off-budget’’ items are available in Appendix H in Fregert and Gustafsson (2005). 20. Data on investment expenditures paid out by the National Debt Ofﬁce is obtained from the ofﬁcial records of the parliament sessions. 21. Speciﬁcally, we use Statsregleringsfonden were we can ﬁnd transfers to other funds, that is, transfers that increase assets in other funds but have no impact on current expenditures. 22. The most noticeable were appropriations to Arbetarfo¨rsa¨kringsfonden (22.4 millions). Yearly ﬁgures on investment expenditures are presented in Appendix I in Fregert and Gustafsson (2005). 23. The netting procedure is exempliﬁed and more thoroughly explained in Fregert and Gustafsson (2005). 24. Separate funds data are available in Appendix J in Fregert and Gustafsson (2005). 25. For a discussion on the Swedish budget during the 20th century, see Lane and Back (1989). 26. We have not adjusted the ﬁgures for 1923 and 1995/96 here as the user may wish to choose his own method. 27. The data on changes in standing government credits are taken from the balance sheets of the National Debt Fund. 28. The ofﬁcial consolidated debt does not correspond to the consolidated central government as deﬁned in Section 6, since the debt held by the Riksbank is not

210

KLAS FREGERT AND ROGER GUSTAFSSON

subtracted in the ofﬁcial measure. The ofﬁcial consolidated government debt constitutes the central government’s contribution to the public sector debt that is used in the Maastricht criteria. 29. Selling new callable government debt below par can in some instances be advantageous for the government through early retirement. 30. More details on valuation principles and an overview of measures of government debt used before the current three measures are available in Riksga¨ldskontoret (2002). 31. See Dahme´n (1989) on the history of Swedish government borrowing since 1789. 32. Details are given in Appendix A in Fregert and Gustafsson (2005). 33. For the exact calculations and further explanations see Appendix B in Fregert and Gustafsson (2005). 34. The initial rise in real debt between 1719 and 1722 was due to deﬂation after a temporary, war-induced increase in the price level 1715–1719 (see Jansson et al., 1991, p. 159). Thus the low real debt in 1719 must be regarded as a temporary aberration. 35. Separate data for the National Debt Ofﬁce, the government proper and the special funds are available in Fregert and Gustafsson (2005). 36. Note that the debt–GDP ratio uses GDP at factor cost, which results in larger ﬁgures than if GDP at market prices had been used. 37. Separate data for the National Debt Ofﬁce and The National Board of Trade are available in Appendix F in Fregert and Gustafsson (2005). 38. The National Debt Ofﬁce regularly reports ﬁgures according to the identity: net borrowing requirement ðDEF t Þ+debt adjustments ðnt Þ+short-term investments ðDAt Þ=change in central government debt ðDBt Þ. 39. For a related discussion see Neumann (1992, 1996) and for the case of Sweden see Gustafsson (2005). 40. Other sources of the Riksbank ðOSt Þ is equal to: the change in the capital plus other revenues (for instance capital gains) plus the increase in deposits from other than the government (or bank deposits that is included in H t ) plus the increase in other liabilities. 41. Other uses of the Riksbank ðOU t Þ is equal to: the increase in other assets plus the operational costs plus its proﬁt (not transferred to the ﬁscal branch). 42. See Walsh (1998, pp. 132–138) for a discussion on the consolidated budget constraint and the measurement of seigniorage in the closed economy. 43. We assume that all government ﬁnancial assets ðAt Þ are held at the central bank and, for simplicity, we ignore the term ‘‘price and volume changes’’ and ‘‘error and omissions’’. 44. If the central bank does not buy domestic non-government bonds or foreign bonds and if the other net inﬂow of the central bank is zero, seigniorage is equal to DH t , a common empirical measure, see for example Fischer (1982). 45. See Appendix L in Fregert and Gustafsson (2005) and Gustafsson (2005). An Public additional way to express the seigniorage is as St ¼ Gt þ I t þ it BPublic , t1 T t DBt by Eq. (18). However, following the discussion in the previous section this will not be correct due to ‘‘price and volume changes’’ and ‘‘errors and omissions’’. 46. This measure is also used by Neumann (1992) from whom we have borrowed the term fiscal seigniorage, that is, seigniorage directly used for budget purposes. This

Fiscal Statistics for Sweden 1719–2003

211

ﬂow measure should be distinguished from the extra revenue the government makes from capital gains due to unexpected inﬂation which erodes the real value of the debt. 47. A small error arises from the 3 million (=18–15) riksdaler riksga¨lds not redeemed by the Riksbank and which hence should be treated as public debt. Since we cannot associate the creation of these 3 million with any speciﬁc year, we have not corrected for this. 48. All ‘‘large’’ (over 4 percent) spikes in Fig. 14 are almost entirely due to changes in BCB t . 49. The difference DEF t DEF C t is equal to net borrowing from the central bank DBCB t DAt , that is, a part of seigniorage.

ACKNOWLEDGMENTS We are grateful for the valuable comments from Svante Hellman (The Swedish National Financial Management Authority), Jesper Hansson (The Swedish Riksbank) and seminar participants at the Swedish National Financial Management Authority and at the Department of Economics at Lund University. Klas Fregert thanks Finanspolitiska institutet for initial ﬁnancing.

REFERENCES Ahlstro¨m, G. (1972). The period 1776–1802. In: L. Jo¨rberg (Ed.), A history of prices in Sweden 1732–1914 (Vol. II). Lund: CWK Gleerup. A˚mark, K. (1961). Sveriges statsfinanser 1719–1809. (Swedish government ﬁnances 1719–1809). Stockholm: Norstedt & So¨ners fo¨rlag. Barro, R. J. (1987). Government spending, interest rates, prices, and budget deﬁcits in the united kingdom 1701–1918. Journal of Monetary Economics, 20, 221–245. Blejer, M., & Cheasty, A. (1991). The measurement of ﬁscal deﬁcits: Analytical and methodological issues. Journal of Economic Literature, 29, 1644–1678. Dahme´n, E. (Ed.) (1989). Uppla˚ning och utveckling. Riksga¨ldskontoret 1789–1989. (Borrowing and development. The National Debt Ofﬁce 1789–1989). Stockholm: Allma¨nna fo¨rlaget. Eisner, R., & Pieper, P. (1984). A new view of the federal debt and budget deﬁcits. American Economic Review, 74, 11–29. Fischer, S. (1982). Seigniorage and the case for a National money. Journal of Political Economy, 90, 295–313. Fregert, K., & Gustafsson, R. (2005). Fiscal statistics for Sweden 1719–2003. Working Paper No. 40, Department of Economics, Lund University, Lund, Sweden. Fregert, K., & Jonung, L. (1996). Inﬂation and switches between specie and paper standards in Sweden 1668–1931: A public ﬁnance interpretation. Scottish Journal of Political Economy, 43, 444–467.

212

KLAS FREGERT AND ROGER GUSTAFSSON

Gustafsson, R. (2005). Seigniorage in Sweden 1908–2003. Unpublished manuscript. Department of Economics, Lund University. Hamilton, J., & Flavin, M. (1986). On the limitations of government borrowing: A framework for empirical testing. American Economic Review, 76, 808–819. Heckscher, E. F. (1942). De svenska penning-, vikt och ma˚ttsystemen. (The Swedish monetary, weight, and length measurement systems). Stockholm: P.A. Norstedt och So¨ner. Historical Statistics of the United States: Millennial Edition. (2006). In: S. B. Carter, S. Gartner, M. R. Haines, A. L. Olmstead & R. Sutch (co-eds), Part E. Cambridge: Cambridge University Press. IMF. (2001). Government finance statistics manual. Washington, DC: International Monetary Fund. Jo¨rberg, L. (1972). A history of prices in Sweden 1732–1914 (Vol. 1). Lund: CWK Gleerup. Lane, J.-E., & Back, S. (1989). Den svenska statsbudgeten. (The Swedish Budget). Stockholm: SNS fo¨rlag. Mitchell, B. R. (1988). British historical statistics. Cambridge: Cambridge University Press. Mitchell, B. R. (1998a). International historical statistics: The Americas 1750–1993. Basingstoke: Macmillan. Mitchell, B. R. (1998b). International historical statistics: Europe 1750–1993. Basingstoke: Macmillan. Mitchell, B. R. (1998c). International historical statistics: Africa, Asia & Oceania 1750–1993. Basingstoke: Macmillan. Neumann, M. J. M. (1992). Seignorage in the United States: How much does the U.S. Government make from money production. Review-Federal Reserve Bank of St. Louis, 74, 29–40. Neumann, M. J. M. (1996). A comparative study of Seigniorage: Japan and Germany. Bank of Japan Monetary and Economic Studies, 14, 104–142. Riksga¨ldskontoret. (2002). Statsskulden-hur och av vem ska den ma¨tas och redovisas? (The government debt – how and by whom should it be measured and reported?). Regeringsrapport, Dnr 2002/104. Stuart, J., & Rystedt, M. (1905). Handbok o¨fver statsverkets jemte dertill ho¨rande fonders medelsfo¨rvaltning. (Handbook for the Central Government and its Funds). K. L. Beckmans Boktryckeri: Stockholm. Walsh, C. (1998). Monetary theory and policy. Cambridge: MIT Press.

APPENDIX. DATA SOURCES Revenues and Expenditures A˚mark, K. (1961). Sveriges statsfinanser 1719–1809 (Swedish Government Finances 1719–1809.). Stockholm: Norstedt & So¨ners fo¨rlag. Ekonomistyrningsverket. (1998–2003) yearly. Statsbudgetens utfall. Rathsman, C. N. (1855). Underda˚nigt beta¨nkande till Kongl. Maj: t anga˚ende uppgjord jemfo¨relse emellan svenska folkets skatter samt stats- och riksga¨lds-verkens utgifter vid bo¨rjan af 1809 och 1810 a˚rs

Fiscal Statistics for Sweden 1719–2003

213

riksdag samt enehanda skatter och utgifter vid 1850 a˚rs slut: afgifvet den 4 juni 1855. Stockholm. Rikets sta¨nders revisorer. [18– ]–(1866). Rikets ho¨glof. Sta¨nders a˚r y fo¨rsamlade revisorers bera¨ttelse om granskningen af riksga¨lds-kontorets tillsta˚nd och fo¨rvaltning, in Bihang till samtlige rikssta˚ndens protokoll (1866 i: Bihang till Riksdagens protokoll y 1867, samling 2, afd. 1), Stockholm. Rikets sta¨nders revisorer. (1817–1818). Tabellbilaga Litt B och Litt C till nr 9 i Bihang till samtlige rikssta˚ndens protokoll, 2:dra samlingen, Stockholm. Riksdagens revisorer. (1868–1911). Riksdagens a˚r y fo¨rsamlade revisorers bera¨ttelse om granskningen av Riksga¨ldskontorets tillsta˚nd och fo¨rvaltning, in Bihang till riksdagens protokoll y samling 2, Stockholm. Riksrevisionsverket. (1960/61–1979/80) yearly. Budgetredovisning fo¨r budgeta˚ret Riksrevisionsverket. (1980/81–1997) yearly. Statsbudgetens utfall. Riksra¨kenskapsverket. (1920–1959/60) yearly. Budgetredovisning fo¨r budgeta˚ret (year) . Statskontoret. (1821–1854) yearly. Capital-Ra¨kning till Riks-Hufvudboken fo¨r a˚r y med dertil ho¨rande tabla˚er och bilagor. Statskontoret. (1854–1911) yearly. Kapital-Konto till Riks-Hufvud-Boken fo¨r a˚r y med dertil ho¨rande tabla˚er och bilagor. Statskontoret. (1912–1919) yearly. Budgetredovisning fo¨r budgeta˚ret (year) .

Debt A˚mark, K. (1961). Sveriges statsfinanser 1719–1809 (Swedish Government Finances 1719–1809). Stockholm: Norstedt & So¨ners fo¨rlag. Rikets sta¨nders revisorer. ([18–]–1866). Rikets ho¨glof. Sta¨nders a˚r y fo¨rsamlade revisorers bera¨ttelse om granskningen af Riksga¨ldskontorets tillsta˚nd och fo¨rvaltning, in Bihang till samtlige rikssta˚ndens protokoll (1866 i: Bihang till Riksdagens protokoll y 1867, samling 2, afd. 1), Stockholm. Riksbankens statistiska avdelning. (1931). Statistiska tabeller 1668–1924 (Statistical tables 1668–1924), in Sveriges Riksbank 1668–1918–1924, part V. Riksdagens revisorer. (1868–1911). Riksdagens a˚r y fo¨rsamlade revisorers bera¨ttelse om granskningen av riksga¨ldskontorets tillsta˚nd och fo¨rvaltning, In: Bihang till riksdagens protokoll y samling 2. Stockholm.

214

KLAS FREGERT AND ROGER GUSTAFSSON

Riksga¨ldskontoret. (1920–1980/81) yearly. A˚rsbok (Yearbook). Riksga¨ldskontoret. (1982/83–1997) yearly. Statistical Yearbook. Riksga¨ldskontoret. (1982/83–1997) yearly. Annual Report.

Exchange Rates, GDP, Population and Prices Ahlstro¨m, G. (1972). The period 1776–1802. In: L. Jo¨rberg (Ed.), A History of Prices in Sweden 1732–1914 (Vol. II). Lund: CWK Gleerup. A˚mark, K. (1921). En svensk prishistorisk studie (A historical study of prices in sweden). Ekonomisk Tidsskrift, 23, 147–170. Andersson Palm, L. (2001). Livet, ka¨rleken och do¨den: Fyra uppsatser om svensk befolkningsutveckling 1300–1850 (Life, love and death: Four essays on Swedish population development, 1300–1850). Go¨teborg. Edvinsson, R. (2005). Annual estimates of Swedish GDP in 1720–1800. Unpublished manuscript. Department of Economic History, Stockholm University. Jansson, A., Andersson Palm, L., & So¨derberg, J. (1991). Dagligt bro¨d i onda tider: Priser och lo¨ner i Stockholm och Va¨stsverige 1500–1770 (Daily bread in hard times: Prices and wages in Stockholm and Western Sweden 1500–1770). Go¨teborg: Institutet fo¨r lokalhistorisk forskning. Krantz, O. (2001). Swedish Historical National Accounts 1800–1998 – Aggregated Output Series. Unpublished manuscript. Department of Economic History, Umea˚ University. Statistiska Centralbyra˚n. (2004a). Konsumentprisindex 1830–2003 (Consumer price index 1830–2003). Statistiska meddelanden, PR15 SM 0401. Statistiska centralbyra˚n. (2004b). Nationalra¨kenskaper 1998–2003 (National accounts 1998–2003). Statistiska meddelanden, NR10 SM 0401.

Seigniorage and Riksbank Figures Riksbankens statistiska avdelning. (1931). Statistiska tabeller 1668–1924, in Sveriges Riksbank 1668–1918, part V. Sveriges Riksbank. (1908–1977) yearly. A˚rsbok (yearbook). Sveriges Riksbank. (1978–2000) yearly. Statistisk a˚rsbok (Statistical yearbook). Sveriges Riksbank. (2001–2003) yearly. A˚rsredovisning (Annual report).

Year

1 021 1 023 1 005 967 914 945 1 147 1 096 916 896 823 839 847 930 950 949 855 997 945 1 689 1 506 1 471 1 326 1 103 1 174 1 767 1 471

it Bt1

70 70 70 70 70 70 70 70 70 66 65 66 65 66 66 65 66 66 67 70 72 72 130 134 134 137 156

Tt

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

143 187 190 194 089 338 413 428 248 209 151 184 197 283 303 305 216 238 257 757 619 523 716 441 487 617 841

CBT t

– – – – – – – – – – – – – – – – – – – – – – – – – – –

DEF t

53 95 115 156 105 323 197 262 262 247 263 279 285 288 288 291 296 175 245 2 41 19 260 204 180 286 213

it BPt1

Bt 7 7 7 7 7 7 7 7 6 6 6 6 6 6 5 5 5 5 5 4 4 4 4 4 5 5 4 4 5 4

802 749 703 659 442 304 256 130 976 837 634 498 241 043 859 674 481 278 078 875 744 604 504 368 189 114 934 796 004 922

DEF C t

St

– – – – – – – – – – – – – 0 0 0 0 1 2 5 7 6 8 11 16 22 23

93 122 74 30 97 163 101 162 106 114 65 45 60 104 21 82 165 237 71 10 8 978 31 161 121 158 62

30 42 111 256 78 230 166 170 225 199 263 300 290 249 332 274 197 3 239 53 98 1 024 351 166 176 14 285

BPt 6 6 6 6 6 6 6 5 5 5 5 5 5 4 4 4 4 4 3 3 3 3 3 3 3 3 2 2 2 2

651 608 563 520 303 165 024 898 744 605 472 336 154 956 772 586 394 191 991 787 656 516 416 280 153 072 979 892 774 662

215

1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748

Gt þ I t

Fiscal Statistics for Sweden 1719–2003

FISCAL VARIABLES IN SWEDEN, 1719–2003 (THOUSANDS OF SEK)

216

(Continued ) Year

1 1 1 1 1 1 1 1 2 3 2 3 3 2 4 2 2 2 2 1 1 2 2 2 3 3 2 3 2 3

298 698 800 882 567 620 645 827 447 306 453 239 033 789 251 700 417 171 736 971 878 171 391 712 065 090 533 115 949 197

it Bt1 158 158 153 168 168 173 213 136 136 173 193 231 257 273 300 336 337 329 331 335 121 105 181 408 430 430 429 480 557 258

Tt 1 2 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3 2 2 2 2 3 2 3 3 3 3 5

828 443 677 806 864 933 850 046 211 453 306 711 687 499 576 699 367 450 329 601 352 128 717 375 857 605 288 063 173 013

CBT t – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

DEF t 372 586 276 245 129 141 7 83 372 1 026 340 759 603 562 1 975 336 387 950 262 295 353 148 145 255 638 84 327 532 333 1 558

it BPt1

Bt 4 4 4 4 4 4 4 4 5 6 7 8 8 10 11 10 10 10 10 10 9 11 11 12 12 12 13 14 14 13

793 539 823 737 714 691 714 911 634 676 663 303 721 246 169 523 567 763 546 025 295 552 963 636 536 422 235 503 724 585

22 23 25 26 24 25 61 58 56 68 65 94 103 112 102 132 139 127 120 127 121 105 181 180 182 183 183 245 335 258

DEF C t

St 110 206 3 167 47 117 154 43 741 776 62 179 42 1 297 91 161 83 16 120 599 34 108 220 247 306 192 497 427 208 79

398 516 145 269 226 172 10 118 449 145 274 444 407 896 1 867 294 272 1 136 354 96 387 40 366 730 84 140 76 724 320 1 637

BPt 2 2 2 2 2 2 2 2 2 2 3 3 3 3 4 3 3 3 3 3 2 4 4 4 4 4 5 6 7 6

530 400 449 336 242 161 105 231 131 413 116 176 365 619 394 923 831 718 603 385 717 449 370 372 296 232 416 951 160 021

KLAS FREGERT AND ROGER GUSTAFSSON

1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778

Gt þ I t

401 426 303 321 822 686 913 780 863 810 420 614 532 333 587 337 953 630 020 739 862 731 721 096 606 451 859 454 837 986 646 611 751 347 047 650

1 1 1 1 1

1 1 1 1 1

270 244 243 236 242 305 349 356 350 389 235 947 952 872 953 984 871 695 795 838 988 950 035 083 136 023 015 949 914 945 963 135 306 381 044 036

3 411 3 225 3 315 3 423 4 518 3 876 3 819 3 554 3 777 3 767 5 231 13 662 9 904 7 482 5 896 5 859 6 659 6 643 5 513 6 445 7 060 7 496 6 988 9 844 8 987 9 950 11 118 10 874 10 518 17 917 8 954 12 807 14 027 21 546 33 131 32 454

– – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

260 445 231 135 455 114 442 582 436 432 12 424 899 581 7 723 644 1 463 164 1 683 303 132 790 1 185 767 1 664 1 246 524 243 471 233 9 013 13 655 5 938 1 029 8 818 960 1 231

6 6 6 6 7 7 8 8 9 10 21 22 24 33 33 36 34 32 32 33 38 38 40 42 40 36 32 30 27 27 35 41 42 33 34 35

895 446 328 559 799 672 094 567 720 363 351 237 261 507 934 937 151 922 738 976 142 901 867 349 248 854 723 098 576 984 135 074 103 285 245 476

1 1 1 1 1

270 244 243 236 242 305 349 356 350 389 235 947 952 872 953 984 871 695 795 838 988 950 035 083 136 023 015 949 914 945 903 720 785 795 439 461

7 415 119 1 114 215 153 26 27 80 214 5 162 2 839 24 2 192 1 187 1 333 2 055 807 1 724 1 143 1 085 2 040 492 813 3 053 2 822 1 799 2 414 1 470 488 6 253 1 683 1 771 147 1 795 2 247

7 675 563 230 248 241 39 468 555 516 218 7 263 1 940 605 5 531 543 129 1 891 875 2 027 1 011 295 855 1 260 2 477 4 192 3 346 1 556 1 942 1 703 8 524 7 341 3 841 1 262 9 550 2 150 1 591

6 6 6 6 7 7 8 8 9 10 16 13 16 23 22 24 19 17 18 19 22 20 22 23 23 21 20 19 19 19 18 22 22 13 13 12

695 446 328 559 799 672 094 567 720 363 120 792 215 237 591 233 511 463 963 098 105 845 243 406 041 986 269 922 619 668 578 497 479 527 576 803

217

3 3 3 3 3 3 3 3 3 3 17 13 9 14 5 6 5 7 5 5 6 7 6 7 8 9 9 9 9 25 21 17 13 11 33 32

Fiscal Statistics for Sweden 1719–2003

1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814

218

(Continued ) Gt þ I t

1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844

13 15 14 15 17 15 17 16 15 17 16 18 18 17 12 22 21 20 20 20 20 22 25 24 23 23 23 23 22 20

459 427 553 900 029 542 403 355 318 178 509 278 880 276 664 163 853 991 415 977 956 774 484 280 740 199 514 521 197 536

it Bt1 1 012 747 724 702 678 685 679 668 657 631 618 605 593 581 328 328 355 394 397 399 432 455 473 418 362 316 296 343 355 348

Tt 20 16 16 17 17 17 16 16 17 17 17 19 18 17 21 20 21 21 20 20 21 22 24 25 24 23 24 19 22 20

965 632 285 638 976 016 677 952 034 131 671 530 423 436 076 540 772 302 938 765 572 192 468 043 079 463 078 873 778 923

CBT t – – – – – – – – – – – 533 502 528 537 481 75 121 207 167 107 746 – – – – – 3 467 – –

DEF t 6 495 458 1 008 1 036 268 789 1 405 72 1 059 678 545 1 180 547 107 8 620 1 469 360 39 334 444 292 291 1 490 345 24 52 267 525 225 40

it BPt1

Bt 28 28 27 26 26 25 24 23 17 17 17 17 16 15 10 11 13 13 13 13 13 12 11 10 8 8 10 11 10 8

981 523 515 478 210 421 504 867 805 109 920 366 764 916 073 811 603 333 043 183 249 980 326 849 830 349 369 515 209 907

449 211 223 236 249 278 295 307 320 318 315 312 310 307 64 63 91 130 133 135 168 196 220 170 120 79 64 123 134 132

St 2 100 2 637 857 973 2 866 1 367 1 117 831 906 268 876 771 404 893 7 401 633 180 860 51 623 637 84 357 1 278 2 589 1 467 2 215 2 982 2 024 303

DEF C t 4 958 1 642 653 529 2 168 171 2 138 542 490 97 29 169 361 1 040 946 2 319 9 678 442 276 188 862 1 593 1 871 2 369 1 282 1 716 789 2 471 559

BPt 8 9 9 8 9 10 9 9 3 3 4 4 3 3 3 5 7 6 6 6 6 6 5 4 2 2 4 5 4 3

456 524 540 823 856 108 861 850 944 504 559 244 883 276 471 211 003 733 443 583 781 644 122 777 890 541 693 971 797 627

KLAS FREGERT AND ROGER GUSTAFSSON

Year

679 233 222 604 776 204 798 340 933 145 940 852 675 975 573 322 967 635 804 478 571 215 872 600 670 979 325 757 977 557 567 417 670 316 686 822

1 2 2 2 2 3 3 4 5 5 5 5 5 5 5 6 7 8 8 9 9

341 339 338 361 384 355 309 304 298 323 324 320 287 270 486 607 308 190 379 696 524 992 541 240 272 393 634 880 534 642 030 108 093 884 588 698

23 25 24 20 26 25 27 23 28 30 29 38 36 31 33 35 39 39 43 44 40 42 38 38 39 49 58 62 72 81 79 83 82 79 71 81

190 152 023 645 599 579 760 870 193 279 560 538 683 527 540 776 013 636 543 347 659 522 638 658 666 819 681 577 313 292 689 732 281 823 207 026

– – 907 4 922 – – – 3 150 – – 4 050 – – 1 500 – – 1 000 – – – 5 000 – 3 400 1 800 1 500 3 500 1 500 1 351 – – 500 1 250 1 300 900 1 350 1 300

170 579 370 602 439 19 347 1 624 961 1 189 654 634 5 280 18 219 16 519 16 153 9 262 11 189 13 639 19 827 17 436 25 685 10 375 14 381 13 776 4 053 2 222 1 710 802 14 907 17 408 15 544 17 182 26 478 23 717 6 194

9 9 11 12 12 11 13 12 10 13 12 17 14 23 29 49 57 52 54 67 78 91 99 110 114 118 123 124 121 130 140 176 182 212 220 230

362 823 146 472 158 408 639 893 828 929 223 206 091 645 051 788 060 410 695 693 248 004 672 663 923 616 020 302 113 697 360 356 343 734 482 542

1 2 2 2 2 3 3 4 4 4 5 5 5 5 5 5 6 7 8 9 9

130 116 97 141 190 151 125 125 125 158 158 161 138 119 343 470 181 061 076 403 239 662 170 817 933 073 310 555 236 401 804 954 955 597 442 341

422 687 1 272 7 267 1 677 758 283 3 206 317 3 979 84 4 933 183 3 564 1 613 3 680 1 099 5 254 282 2 438 4 405 1 503 2 573 610 1 696 126 1 341 3 449 5 147 9 928 5 868 181 8 710 4 010 6 966 1 340

803 115 1 568 3 166 1 044 534 446 1 388 817 5 003 4 622 4 458 5 314 16 003 14 763 19 695 11 234 5 805 13 618 17 095 17 746 23 852 10 831 15 149 13 241 7 106 2 387 714 4 047 24 593 11 815 16 459 9 634 31 100 17 956 5 796

4 4 6 7 7 6 9 8 6 9 8 13 10 20 25 46 54 45 48 61 70 83 91 104 107 111 116 117 116 126 137 173 175 210 214 222

214 807 262 720 538 920 283 669 736 969 395 510 527 213 751 620 024 596 070 254 820 149 160 131 866 742 123 376 196 585 443 494 585 008 406 663

219

22 24 24 24 25 25 27 28 26 31 33 38 41 50 49 50 46 48 54 61 59 64 47 49 49 51 52 59 65 90 91 93 92 98 86 78

Fiscal Statistics for Sweden 1719–2003

1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880

220

(Continued ) Gt þ I t

1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910

81 72 77 77 81 88 90 85 86 87 98 97 98 98 94 123 108 117 132 149 160 171 171 190 193 198 219 252 257 250

623 790 785 240 410 333 154 050 111 641 684 837 452 401 923 946 910 477 797 593 508 354 997 373 369 385 597 441 775 573

it Bt1 9 9 9 9 9 10 9 9 10 10 9 9 10 10 10 10 10 10 10 10 12 12 12 12 13 13 14 17 18 19

126 448 196 449 633 011 676 761 151 983 779 984 329 544 456 269 086 028 456 905 276 192 121 056 411 301 693 133 585 411

Tt 87 208 85 853 87 644 88 666 88 644 84 560 81 114 92 861 98 965 102 146 98 017 95 494 99 495 113 316 119 121 121 780 136 451 144 023 153 516 153 359 143 636 147 398 183 177 185 985 194 613 188 519 208 755 196 096 206 570 236 001

CBT t 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 1 2

2 2 3 4 5 5 6 8 8

185 250 600 300 300 200 – 250 – 300 300 850 750 801 480 340 700 800 000 – – 000 716 285 000 100 350 400 675 323

DEF t 2 4 2 3 1 12 18 2 4 9 10 7 7 16 10 20 18 12 7 29 34 1 13 8 18 20 67 61 25

it BPt1

Bt 356 865 263 277 100 584 716 699 703 822 146 478 536 173 222 095 155 317 263 138 148 149 775 159 166 066 186 077 115 661

234 229 228 230 247 245 246 265 259 259 265 274 278 293 287 292 290 286 319 343 354 354 352 391 412 426 468 519 554 543

902 092 013 347 210 949 109 034 054 653 413 047 811 292 647 878 750 920 724 090 277 040 511 190 243 907 102 071 173 383

8 9 8 9 9 9 9 9 9 10 9 9 10 10 10 10 9 9 10 10 12 12 12 11 13 13 14 17 18 19

533 046 922 156 437 756 464 527 938 638 477 791 118 330 290 079 913 838 304 759 152 111 008 976 352 228 622 099 585 411

St 533 181 993 879 073 406 1 530 1 720 3 685 1 916 2 335 2 981 2 583 3 858 84 1 687 1 940 3 548 16 722 59 5 736 9 256 22 388 3 532 1 870 13 091 30 570 4 903 21 944 15 616 2 2 2 4 2

DEF C t 5 1 3 2

481 836 930 609 130 13 122 16 974 3 436 6 601 5 783 12 479 9 154 6 492 727 13 824 10 558 15 687 20 256 27 137 7 052 34 760 45 323 23 217 12 832 13 978 10 003 5 104 78 346 91 734 18 368

BPt 227 223 221 225 242 241 241 260 254 253 260 268 274 288 282 287 285 282 315 340 352 351 350 389 410 425 467 519 554 543

505 243 589 897 084 244 048 875 140 036 822 845 085 858 653 907 813 778 676 185 061 190 671 390 443 107 037 071 173 384

KLAS FREGERT AND ROGER GUSTAFSSON

Year

1

1

1

235 878 019 197 554 638 720 966 003 366 690 130 679 698 994 909 977 009 415 193 202 045 400 101 607 583 475 862 834 450 930 613 544 234 425 532

19 22 22 25 30 36 44 59 59 61 64 72 34 81 77 81 83 87 86 85 82 81 91 99 97 94 91 88 91 98 145 185 226 265 288 271

140 511 697 438 840 193 056 391 480 874 216 231 288 431 286 072 372 068 626 997 439 258 650 296 881 457 928 236 144 286 514 176 727 213 303 791

235 768 223 265 239 891 234 934 356 682 406 547 618 007 817 905 862 038 834 563 759 162 648 592 260 515 647 503 620 650 638 167 656 938 691 131 718 516 761 466 763 844 720 482 726 946 755 551 895 847 991 005 1 146 952 1 290 893 1 440 346 1 849 329 2 013 633 2 248 133 2 684 105 3 096 647 3 244 617 3 521 189

6 6 6 7 8 8 8

256 399 311 058 800 760 120 – – – – 3 500 12 000 16 000 14 000 15 500 16 000 16 200 13 900 17 000 19 000 16 000 14 000 27 000 7 000 4 250 2 750 3 900 1 500 – 23 000 19 500 20 000 9 500 6 000 7 000

28 17 6 20 40 12 16 732 18 71 319 246 122 76 113 89 92 4 34 8 3 118 318 158 199 61 48 7 136 1 035 1 846 1 827 1 799 1 515 1 452 126

350 725 514 643 912 525 648 451 555 677 744 270 452 626 630 313 411 746 625 277 797 821 105 846 640 784 299 305 132 407 810 155 166 300 111 133

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 3 5 6 8 9 10 11

611 607 648 744 854 993 149 656 566 496 510 551 566 634 734 735 812 825 835 800 845 155 358 348 487 387 236 430 633 625 170 926 731 773 952 194

959 964 287 455 871 246 226 179 998 524 987 464 441 038 032 286 799 572 170 846 644 333 532 962 099 183 798 242 617 020 035 901 848 651 782 922

19 22 22 25 30 35 44 58 59 61 64 71 34 81 76 80 82 86 86 85 81 80 86 91 92 92 91 87 89 92 135 173 217 259 283 269

140 511 697 438 631 862 024 898 237 746 112 932 216 195 740 308 666 464 217 176 624 970 920 674 906 191 437 951 782 010 093 678 210 024 342 513

8 340 19 113 16 288 11 046 346 21 636 12 018 56 321 93 717 50 289 165 983 59 973 221 224 29 291 9 294 108 792 20 146 50 148 5 803 59 654 53 554 287 465 13 068 62 563 90 354 90 392 32 529 136 984 12 802 910 985 155 715 288 784 196 892 6 189 705 438 317

26 43 29 16 49 42 36 675 74 21 485 309 86 121 117 4 87 70 53 51 75 152 340 240 292 154 78 147 123 118 2 015 1 546 2 006 1 512 1 453 569

266 237 113 655 849 590 754 637 918 261 622 444 844 681 790 742 559 489 918 752 536 932 442 788 020 160 569 904 468 146 105 373 541 422 855 173

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 4 5 7 8 9 10

611 607 648 741 848 992 141 653 564 494 507 548 555 620 721 722 802 814 829 778 842 936 098 197 358 358 234 389 481 957 334 860 684 652 757 255

959 964 287 200 895 644 226 450 086 149 335 290 751 273 061 335 063 554 122 714 625 913 455 901 460 683 798 142 515 677 086 910 797 414 954 515

221

1 1 1 2 3 3 4 4 4 3

251 224 230 237 375 391 598 490 784 844 014 826 360 658 670 661 681 625 680 684 704 774 967 842 004 962 009 213 486 786 737 909 276 356 414 382

Fiscal Statistics for Sweden 1719–2003

1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1923/24 1924/25 1925/26 1926/27 1927/28 1928/29 1929/30 1930/31 1931/32 1932/33 1933/34 1934/35 1935/36 1936/37 1937/38 1938/39 1939/40 1940/41 1941/42 1942/43 1943/44 1944/45 1945/46

222

(Continued ) Year

3 4 4 4 5 7 8 8 9 10 11 12 13 14 15 16 18 20 23 27 31 33 36 41 44 52 56 65 76 90

545 052 827 999 813 229 506 843 341 237 348 729 474 600 395 608 774 221 574 484 153 956 186 024 989 025 616 828 981 958

347 614 908 799 800 516 164 610 974 843 296 269 962 219 763 275 523 163 560 465 422 816 948 809 486 330 829 865 191 446

it Bt1

Tt

CBT t

DEF t

310 310 321 315 330 342 297 339 350 432 456 569 598 735 863 871 768 792 849 885 918 075 296 679 019 944 234 696 738 133

3 605 836 4 437 798 4 953 957 4 829 105 5 805 598 7 492 224 7 793 157 8 519 693 8 891 828 10 056 572 10 676 207 12 004 033 12 589 548 13 607 010 16 590 675 17 956 848 19 769 053 20 827 171 24 156 689 27 914 920 30 290 702 31 951 424 34 636 101 38 687 114 44 177 767 50 103 063 52 446 738 58 882 620 69 672 245 90 917 610

– – – 10 000 15 000 15 000 15 000 15 000 15 000 15 000 15 000 15 000 15 000 50 000 50 000 50 000 100 000 100 000 100 000 100 000 150 000 150 000 200 000 200 000 200 000 200 000 200 000 250 000 350 000 450 000

250 74 195 475 323 65 995 648 785 598 1 113 1 279 1 469 1 678 381 527 326 86 166 354 1 631 2 930 2 647 3 816 2 631 3 667 6 205 9 392 10 697 3 724

1 1 1 2 1 2 2 3 4

856 447 157 265 521 794 767 157 145 214 252 424 587 612 561 394 056 136 038 246 823 168 803 259 802 822 942 469 664 185

it BPt1

Bt 367 737 108 960 722 087 774 074 291 485 341 660 001 821 351 178 474 128 908 791 543 560 650 953 522 089 033 714 610 021

11 11 11 12 12 12 12 13 14 15 16 18 19 20 20 19 19 19 19 19 21 25 27 30 33 37 43 52 63 68

419 486 861 072 420 329 531 586 524 477 818 361 173 660 410 573 075 270 251 622 434 033 000 959 401 209 211 675 680 759

869 857 090 926 618 381 548 656 089 181 171 740 374 080 939 850 002 222 809 132 913 906 980 881 947 105 336 454 071 943

1 1 1 1 2 2 3 3

303 268 257 261 268 283 255 309 335 419 437 500 491 624 728 734 675 745 815 859 885 981 194 594 811 693 077 505 328 245

113 834 740 370 920 690 360 324 565 482 957 048 835 204 328 519 597 644 307 987 071 686 798 523 152 230 132 662 611 746

1 703 196 183 56 529 988 63 50 581 989 792 93 795 254 306 877 305 282 757 1 349 1 263 617 40 1 588 768 1 653 2 607 1 925 6 414 7 801

BPt

DEF C t

St 516 750 147 114 244 176 480 163 251 722 802 601 652 219 333 587 134 112 393 956 855 882 448 777 937 506 492 462 698 129

1 460 313 314 488 252 1 009 1 031 583 204 388 317 1 131 2 172 1 363 772 263 13 142 990 920 483 2 369 2 786 2 343 1 853 1 961 8 854 7 526 4 222 11 087

892 100 839 179 122 159 848 078 459 969 245 683 901 194 917 533 800 477 570 424 936 196 093 440 934 991 714 445 860 712

9 8 9 9 9 10 10 11 12 12 12 14 15 16 16 16 16 15 16 15 16 19 21 24 25 27 36 44 48 61

047 762 240 679 379 152 801 689 239 181 830 193 740 821 210 138 019 993 775 913 560 609 736 250 821 985 650 128 668 215

745 807 913 609 673 890 247 988 826 411 345 271 625 757 494 374 489 390 401 711 099 935 121 367 598 794 913 532 673 363

KLAS FREGERT AND ROGER GUSTAFSSON

1946/47 1947/48 1948/49 1949/50 1950/51 1951/52 1952/53 1953/54 1954/55 1955/56 1956/57 1957/58 1958/59 1959/60 1960/61 1961/62 1962/63 1963/64 1964/65 1965/66 1966/67 1967/68 1968/69 1969/70 1970/71 1971/72 1972/73 1973/74 1974/75 1975/76

Gt þ I t

107 127 146 164 191 207 229 237 253 256 272 284 299 341 387 432 534 487 482 728 556 583 553 607 635 661 665

042 549 139 067 473 439 683 878 902 152 748 819 770 590 474 629 779 959 689 674 789 246 262 855 121 969 907

809 487 105 546 091 903 531 006 037 965 544 070 624 624 767 822 861 873 118 864 012 734 160 165 100 753 740

5 6 8 14 23 27 48 60 75 66 63 53 53 63 61 60 73 94 123 123 98 113 89 90 81 67 42

416 915 802 508 765 724 195 386 234 508 812 410 178 696 033 042 090 736 200 757 366 405 884 213 257 339 173

337 824 487 895 021 094 980 504 268 793 064 443 902 173 277 701 178 213 957 585 882 226 709 314 874 655 348

101 108 115 127 154 165 187 217 255 269 312 326 361 395 396 390 370 367 416 808 640 697 717 790 726 703 654

424 636 613 842 436 130 280 165 595 437 751 440 107 552 486 025 443 424 983 878 828 013 503 200 925 187 231

916 130 786 813 801 779 186 492 866 949 208 915 401 551 593 287 351 870 612 360 016 519 603 479 545 584 438

550 000 650 000 650 000 750 000 850 000 2 000 000 4 000 000 4 000 000 5 000 000 5 661 000 7 354 000 6 111 000 6 600 000 6 000 000 7 000 000 7 700 000 7 300 000 9 500 000 6 200 000 8 100 000 8 100 000 9 300 000 7 600 000 9 800 000 28 200 000 27 300 000 7 500 000

10 25 38 49 59 68 86 77 68 47 16 5 14 3 45 94 230 205 182 35 6 9 81 101 38 1 46

484 179 677 983 951 033 599 099 540 562 455 677 757 734 021 947 126 771 706 454 227 661 956 932 746 178 349

230 180 806 629 311 218 325 019 438 810 401 598 876 246 452 236 689 216 464 089 878 559 734 000 571 176 650

1 1 1 1 1 1 1 1 1 1

82 105 139 192 252 319 407 482 559 596 609 597 589 582 626 710 960 178 370 411 432 448 374 279 156 160 228

340 237 086 088 967 686 325 636 459 015 248 621 712 456 698 982 611 643 401 632 076 859 180 205 827 329 741

463 931 309 176 984 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000

4 5 6 11 20 22 42 54 68 58 54 45 46 57 54 50 59 87 117 108 94 110 88 88 80 67 42

079 279 873 728 010 805 787 858 832 380 634 041 474 672 285 478 780 174 271 514 764 972 094 729 683 339 173

064 391 123 720 457 432 702 086 350 362 064 943 402 173 277 701 678 713 457 085 882 226 709 314 874 655 348

108 1 839 1 454 7 682 6 777 16 360 21 475 6 327 528 40 778 11 138 1 080 9 361 17 500 19 873 12 213 76 359 126 684 10 511 25 699 1 459 13 379 52 632 6 126 29 504 7 500

530 491 017 431 437 338 277 582 918 569 000 500 500 000 000 000 500 500 500 500 000 000 000 000 000 000 000

9 26 35 40 50 48 106 69 67 4 25 4 24 21 65 105 300 81 193 54 12 10 76 94 17 3 46

805 032 944 271 269 754 666 243 667 316 769 500 223 210 146 296 476 025 488 010 184 584 094 248 246 382 349

487 239 425 022 311 218 325 019 438 810 401 598 876 246 452 236 689 216 464 089 878 441 734 000 571 176 650

1 1 1 1 1 1 1 1 1 1

73 97 128 171 223 270 378 446 517 515 537 525 507 517 582 676 996 090 292 352 378 416 345 258 156 160 228

961 790 970 852 390 861 630 110 369 375 812 115 775 995 362 995 974 260 800 587 988 017 182 477 827 329 741

639 186 281 176 984 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000

Fiscal Statistics for Sweden 1719–2003

1976/77 1977/78 1978/79 1979/80 1980/81 1981/82 1982/83 1983/84 1984/85 1985/86 1986/87 1987/88 1988/89 1989/90 1990/91 1991/92 1992/93 1993/94 1994/95 1995/96 1997 1998 1999 2000 2001 2002 2003

223

TRANSPORT CAPACITY MANAGEMENT AND TRANSATLANTIC MIGRATION, 1900–1914 Drew Keeling ABSTRACT Early twentieth century transatlantic migration was both a massive transoceanic population transfer and a complex travel business. The successful growth of this multinational commerce was based not on fare reductions, but on risk management strategies. Shipping lines provided costly carrying capacity sufficient to accommodate severely fluctuating demand for transatlantic migration, and did so in a manner which improved the reliability and quality of travel for migrants.

1. INTRODUCTION Nineteenth and early twentieth century migration between Europe and the United States was one of the most sizable, long-lived, and well-documented population transfers of all time. Historical studies of it, however, have tended to skirt around the details of the complex international transportation business which carried out this mass relocation.1 This essay examines the Research in Economic History, Volume 25, 225–283 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(07)25005-0

225

226

DREW KEELING

capacity challenges which passenger shipping lines faced in carrying migrants, and analyzes the capacity management practices they used during the ﬁrst decade and a half of the twentieth century.2 It is an outgrowth of the author’s Ph.D. dissertation (Keeling, 2005a), and relies upon some of the same sources and quantitative data sets developed there. Prior analyses of the powerful intersection of shipping and migration in the nineteenth and early twentieth centuries have contradicted assumptions that steamship lines stimulated the long-term growth of transatlantic labor ﬂows either by lowering fares or through clever advertising.3 Previous studies, however, have also produced three mutually inconsistent observations. Migrants in the steerage class, it has been claimed, were charged a higher percentage margin over cost than was levied on tourist or business travelers. Nevertheless, shipping companies steadily improved the quality of accommodations provided to migrant passengers from the 1890s onward, and without raising fares commensurately.4 Despite such enhancements, however, migration grew fastest on the routes where improvements to travel conditions were slowest.5 It also turns out (as described in Section 3) that the overall rate of ‘‘capacity utilization’’ on these early twentieth century North Atlantic passenger steamers was comparatively quite low. On average, steamship companies managed to ﬁll only about 40 percent of their passenger berths in the second and third class (passengers compartments used mainly by migrants). Making maximum commercial use of on-board space is widely recognized as a central component of success in ocean shipping.6 The particularly acute capacity challenges created by mass migration have not, however, been widely appreciated or speciﬁcally examined in prior histories of either migration or shipping. In order to more comprehensively assess the interaction between transport services and voluntary mass relocation between Europe and the United States across the North Atlantic a century ago, this article relies on extensive quantitative data gathered from both shipping and migration sources. The most important data set employed here was created by integrating Transatlantic Passenger Conference records7 with shipping reference book information.8 This enables the measurement of both passenger totals and passenger capacity by voyage, day, month, quarter, year, vessel, route, shipping line, etc., eastward and westward, for the entire 1900–1914 period. A description of the economic structure of the early twentieth century North Atlantic migration business is followed by an analysis of how shipping lines supplied transport capacity for European labor migrants traveling to – and in some instances back from – the United States. This, in turn, offers

Transport Capacity and Transatlantic Migration

227

insights into the fundamental causal interactions between the movers and the moved.

2. THE ECONOMICS OF NORTH ATLANTIC MIGRATION AND TRANSPORTATION Observers of oceanic shipping have long agreed with maritime historians that this is a ‘‘high-risk industry’’ exposed to ‘‘continual dangers’’ of ‘‘the very greatest ﬂuctuations.’’ ‘‘Throughout its history,’’ as one historian put it, shipping has never been a business for people of ‘‘timid dispositions.’’ Other commentators have concurred in regarding this business as being subject to variations ‘‘entirely without parallel in any other branch of trade,’’ and hence ‘‘as shifting and unstable as the sea’’ itself.9 To better appreciate the origins of this inherent riskiness, however, and how it inﬂuenced the strategies of passenger shipping lines, a more thorough analysis of its underlying causes and characteristics is required. In the early twentieth century, migrant trafﬁc was the largest business segment of North Atlantic passenger steamship lines, generating, on average, about half of their revenues and an even larger share of both proﬁts and variability in proﬁts.10 From the 1860s, when the migrant trade between Europe and North America shifted from sailing vessels to steamships, and across the subsequent half century until the outbreak of World War I, three principal trends stand out: Firstly, there was ongoing and signiﬁcant technological development (such as more energy efﬁcient engines and better hull designs which reinforced the scale economies of building larger and more modern vessels),11 which, secondly, had the effect of continually lowering the cost per unit of capacity.12 Thirdly, however, though these costs were in steady long-term secular decline, they were almost completely ‘‘ﬁxed’’ in the sense of being nearly totally unaffected by short-term changes in revenues. The cost of moving a steamship across the Atlantic was only very marginally affected by ﬂuctuations, from one voyage to the next, in what or how much it carried.13 These features applied across the spectrum of the various transport services offered by oceanic shipping lines. A fourth trend in late nineteenth and early twentieth century North Atlantic passenger shipping, however, is in stark contrast to contemporaneous developments in freight transport. Transatlantic freight rates fell by more than a quarter between 1881 and 1913, but fares for steerage passengers rose between these two peak migration years. This means that the doubling of annual European migration to the

228

DREW KEELING

United States over these three decades cannot be ascribed to falling steamship fares, since the fares did not fall.14 Even taking into account increases in wages for unskilled labor in the U.S. (which, of course, lay outside the inﬂuence of steamship company pricing strategy), the affordability of reaching America changed little over these many years. Already in 1881, European migrants working in the U.S. at typical unskilled wage rates15 could expect to earn the cost of the steamship ticket within about four weeks. By 1913, the payback could be expected in just under three and a half weeks.16 Available annual time series of westbound steerage passengers and fares for the years 1901–1913 are presented in Fig. 1. (99 percent of steerage

700

600 $40 500 $30

400

300

$20

200 Weighted Average Fares (lefthand scale)

$10

Annual Passengers (righthand scale)

Annual passengers on these lines ('000s)

Weighted average annual fares of these lines

$50

100

$0

0 1901

1904

1908

1911

1913

Fig. 1. Westbound Steerage Passengers and Weighted Average Fares, 1901–1913. Note: 1904, 1908, and 1911 were recession years in the U.S. (Jerome, 1926, pp. 100–119). Data sources: Lines and routes are those given in Table A.3. Passenger totals are from Table A.4. Fares are based those in Table A.4, with missing years estimated. The weighted average fare series shown in the graph was compiled by adjusting the fares of each route for differences in the lengths of the routes (see Table A.4), and then taking an overall average by weighting each line’s fares, as adjusted, by the number of passengers it carried each year.

Transport Capacity and Transatlantic Migration

229

passengers in this period were migrants and 80–90 percent of migrants were steerage passengers.)17 Annual levels of passengers and fares both increased over the period. The fares rose much more modestly than the passenger volumes did, however. Fares were also more stable than passenger levels, and there was no signiﬁcant systematic relationship between the yearto-year movements of passengers and fares. There were three main reasons why migrant passage prices did not fall secularly. The most central of these is that migrant volumes between Europe and the U.S., at least during this period, were relatively insensitive to transatlantic fares. Reductions in fares sometimes inﬂuenced the migrants’ choice of shipping line but rarely affected their decision to migrate in the ﬁrst place. By the late 1880s, already, the total costs of migrating (of which the oceanic fare made up somewhat less than half) were already lower than the amount an average European migrant working in the U.S. could expect to have saved, net of living costs, six months after the move. After three to four years of reasonably steady unskilled work in America, such a migrant could then return to Europe with, on average, roughly double the accumulated savings obtainable by staying at home, even after deducting travel costs both ways. Political and economic conditions at home, and the state of the U.S. job market, had powerful effects on migration decisions. Temporary fare drops, reducing by a month or two the working time in the United States needed to cover total migration costs from Europe, were of decidedly secondary importance.18 Because overall migrant volumes were little affected by movements in oceanic fares, especially over the short term, periodic price wars lowered passenger revenues for the industry as a whole, and this encouraged the formation of ‘‘conferences’’ (cartels) designed to inhibit rate-cutting.19 The conference arrangements thus acted as a second, albeit intermittent and incomplete, source of ticket price stability. A third and ﬁnal reason for North Atlantic fares not exhibiting a long-term downwards trend was that, unlike bags of ﬂour, passengers (not excepting migrants) could sometimes be lured from one line to another by improvements in travel conditions.20 As a consequence, shipping lines passed the cost savings from technical development on to customers in the form of more space and better amenities rather than through lower fares. But shipping lines imitated each other copiously and incessantly. As one ofﬁcial of the Cunard line put it: ‘‘We all watch what another steamship company does ... and see if we cannot go her better.’’21 Shipping lines copied shipping lines, shipbuilders copied shipbuilders, and countries copied countries in sponsoring national ﬂag carriers. Companies could therefore compete on quality, without risking the

230

DREW KEELING

‘‘throat-cutting’’ to which price competition was prone, but this by no means sufﬁced to secure permanent proﬁtability. Possibilities for growing by absorbing competitors were also limited. The same national pride that promoted new upstart lines in countries without long-standing indigenous maritime sectors also inhibited cross-border mergers. By the early twentieth century, most major European ports had but one dominant and locally favored transatlantic passenger shipping line, and foreign shipping ﬁrms were generally heavily disadvantaged in trying to set up direct operations or obtain docking rights by acquiring the locally based shipping company. In the United States, a small number of ports were served by a large number of steamship lines. The four largest, New York, Boston, Philadelphia, and Baltimore, functioned as entry points for 95 percent of European immigrants.22 A greater diversity of ports was used by migrants on the European side of the Atlantic. About one-third of transatlantic migrants to America traveled by way of Liverpool and Naples, and these two leading migration ports were served by multiple shipping lines (although after 1902, Liverpool’s trafﬁc to the U.S. was handled by only two shipping groups, both effectively British). Nearly every other European port had only one major shipping line traveling to the U.S. In most of those cases (e.g. in France, Belgium, the Netherlands, Denmark, Spain, Greece, and Austria-Hungary) there was only one such shipping line under that country’s registry.23 A number of factors lay behind this local exclusivity.24 The most important bar to outside entry of new overseas shipping lines in most European harbors was the need to obtain scarce docking privileges from local port authorities who were often connected to and gave preference to the locally dominant headquartered company.25 Other local requirements, from coaling and ship repairing to inland rail links, similarly favored a single locally dominant shipping organization. Locally based and locally supported shipping ﬁrms were further buttressed by long-standing national systems of postal and naval subsidies, by the growth of immigrant inspection regimes favoring domestic shipping ﬁrms, by direct regulatory favoritism, such as that shown by the UK Board of Trade or the German Immigration Act of 1897, and by long-term strengthening of market-sharing cartel arrangements specifying local route exclusivity.26 A range of private-sector agreements and ofﬁcial regulations, also instigated after 1900, had the further effect of protecting the major transatlantic shipping lines from coming under foreign ownership.27 Ironically, however, these extensive webs of local protectionism proved to be more pro- than anti-competitive, because while the oceanic transporters

Transport Capacity and Transatlantic Migration

231

of migrants operated under, and were favored by, separate sovereign jurisdictions, their customer markets overlapped considerably. Over half of all migrants journeying westward from Europe to the United States between 1900 and 1913 traveled ‘‘indirect,’’ meaning that their transatlantic embarkation occurred in a port outside their country of origin.28 The same ownership restrictions helping to shield European transatlantic passenger shipping lines from foreign takeovers or from competition in their home ports, also helped make the Europe-wide oceanic passenger shipping oligopoly largely impervious to cross-border consolidation and monopolization, and hence more competitive than otherwise. For all these reasons, therefore, neither price-cutting to grow the overall market, nor transnational amalgamation to widen potential slices within it, were effective long-term business strategies. More than on size maximization, or even proﬁt maximization, successful Atlantic steamship lines concentrated instead on managing the risks of their ‘‘ﬂuctuating’’ business. Shipping lines were, of course, in business to make proﬁts and pay dividends to owners in good years. To do so successfully, however, they also needed to minimize – or at least cope with – the risks of their business, particularly the risk of red ink during economic slumps.29 Cyclical variations were an absolutely critical concern because of the very high percentage of costs which were ﬁxed in the short or medium term (in the sense of being unaffected by changes in revenues). Operating proﬁts on migration trafﬁc in this period averaged roughly 30 percent of revenues. Because costs hardly moved at all with revenues, a 30 percent move up or down in ticket receipts therefore translated into a nearly 100 percent swing in operating proﬁts: basically a ‘‘double or nothing bet.’’30 Shifts in migrant passenger receipts approaching 30 percent were not at all unusual for these steamship lines, however (see Fig. A.2). The challenge, as one shipping historian has put it, of ‘‘applying successive blocks of capital’’ to the ‘‘continuous rebuilding’’ of ﬂeets, while avoiding ‘‘surplus capacity’’ in a ‘‘highly hazardous business,’’31 required considerable managerial skill and organizational capability. Along with ﬁscal prudence, political savvy, and reputation with traveling customers, matching the number of passenger berths supplied to the ﬂuctuating number demanded was a crucial determinant of success in mass passenger transport on North Atlantic steamships. Although shipping companies had little, if any, inﬂuence over their ﬂuctuating ﬂows of migrant trafﬁc, they did have considerable scope for deciding when and how to supply capacity to meet those travel demands. To see how this was accomplished, the dilemmas faced and approaches taken by the steamship lines carrying migrants across the North Atlantic need to be examined more closely.

232

DREW KEELING

3. MANAGEMENT OF PASSENGER CAPACITY: DILEMMAS AND APPROACHES Making decisions concerning passenger capacity forced transatlantic shipping companies to confront a double-edged risk. If they had insufﬁcient capacity, they could incur considerable costs trying to accommodate or mollify customers during periods of peak demand. Too much excess capacity, on the other hand, was an easy way to lose money, even without an accompanying price war, which, if it broke out, would tend to increase losses still further.32 The two ‘‘edges’’ of this capacity risk were asymmetrical. Vessel construction times being about two to three years, a deﬁciency of overall capacity in international shipping33 could be corrected in a few years. If shipping lines overbuilt, by contrast, they were then stuck with assets lasting 20–30 years.34 The resulting general tendency of shipping companies toward overcapacity rather than undercapacity35 affected passenger liners carrying migrants in several interlinked ways. This section describes how shipping lines on the early twentieth century North Atlantic used scheduled passenger capacity to try to proﬁtably handle migrants’ ﬂuctuating travel volumes. Section 4 discusses the logic behind how that capacity was priced to users of it. Section 5 then addresses the companies’ decisions as to what kind of ships to build or acquire in the ﬁrst place, and how to schedule their voyages. During peak periods of migration, transatlantic shipping companies had to scramble to accommodate the resulting crowds. If a liner did not have enough room, the shipping executives faced an unpalatable set of choices. They could transfer excess passengers to another vessel or another line, which would tend to cause uncomfortable delays for passengers,36 they could charter extra vessels, they could turn away would-be clients at ticket counters, or they could overstuff their ships. All these methods were used, at one time or another, by nearly every line on the North Atlantic, and at considerable expense to their bank balances and reputations. Keeping such instances to a minimum was a clear priority. Even so, the actual rates of capacity usage revealed by the database of voyages used here might seem, at ﬁrst glance, remarkably low. In fact, they were quite the opposite. Total carrying capacity, shown in the top curve of Fig. 2 (capacity deployment by calendar quarter), rose steadily at ﬁrst, then rather more steeply during the overheated boom which ended in 1907, and then (given the two to three year ship construction lag) stabilized from 1909 on, as companies which had overextended gradually worked down their surplus of ship space. This graph accounts (in its top three layers) for the 30 percent of

Transport Capacity and Transatlantic Migration

233

Passenger berths on transatalantic vessels, in '000s

300

In Repair Other Routes Idle 200

Unused East

Used East

100

Unused West

Used West 0

1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913

CALENDAR

QUARTERS

Capacity of vessels in repair, retrofit, or transfer between lines Capacity temporarily used outside major N. Atlantic routes Capacity of vessels idled in port Unused Voyage Capacity Eastbound USED Voyage Capacity Eastbound Unused Voyage Capacity Westbound USED Voyage Capacity Westbound

Fig. 2. Deployment of Second and Third Class Passenger Carrying Capacity of North Atlantic Passenger Liners, by Calendar Quarter, 1900–1913 (on Routes to and from New York, Boston, Philadelphia, and Baltimore). Source: Voyage Database.

capacity which, on average, was tied up in vessels being repaired, swapped out to other routes37 or idled for routine periodic maintenance (Bisset & Stephensen, 1959, p. 275) or during seasonal or cyclical slack periods. The remaining four layers, below this, track the capacity actually deployed on scheduled voyages to and from America during 1900–1913. The pattern displays a characteristic spring quarterly peak reﬂecting the seasonal high point of the westward migration, which was these companies’ leading activity. The ﬁrst and third layers from the bottom of the graph

234

DREW KEELING

represent space used by passengers, on the westbound and eastbound legs of the voyage, traveling in the second and third class sections of the ships. The solid layers, second and fourth from the bottom, represent unused passenger space, and it was considerable.38 The overall average capacity utilization rate on these voyages was only about 40 percent. This means that, on average, 60 percent of the space reserved for migrant passengers on these massive multi-million-gold-standard-dollar ships, burning hundreds of tons of coal a day, and carrying crews of several hundred, was not being used!39 By contrast, available ‘‘load factor’’ ﬁgures for airlines crossing the North Atlantic during 2001–2004 show a capacity utilization of 78 percent, or nearly double that of their maritime counterparts a century earlier. The biggest single reason for this difference is that migrant ﬂows a century ago took place with much greater cyclical variation than have recent movements of air travelers (see Fig. A.2).40 Two further reasons for the relatively low occupancy of second and third class quarters on early twentieth century North Atlantic steamships are apparent from a simple breakdown by class and direction (see Table 1 below). Firstly, as noted already, nearly every ship voyage was a roundtrip traversing of the Atlantic in both directions, although migrants made two and a half times as many crossings west as they did east.41 Secondly (with more, and more interesting exceptions, to be discussed shortly below), the ships usually steamed with the same ﬁxed allocation of capacity between the classes, even though passengers’ use of those capacities varied over time. The 40 percent rate at which migrant capacity was employed is the average over roughly 17,000 voyages, a great many of which had capacity utilization rates well above that average. Shipping lines had to manage peaks (and troughs), not just period averages. Some examples of the unevenness of North Atlantic passenger trafﬁc, and of resulting variations in capacity usage rates, are shown by Figs. 3–5. The quarterly westward peak for second and third class capacity utilization, over the period, was 86 percent during the second quarter of Table 1.

Average Capacity Utilization, Europe–USA, 1900–1913.

2nd Class (%) 3rd Class (%) Source: Voyage Database.

Westbound

Eastbound

72 50

38 23

All voyages, broken down by rate at which passenger capacity was used

Transport Capacity and Transatlantic Migration

235

> 90 % > 90 % (N=1025)

30 - 90 % Capacity Utilization Rates (percentages also shown on each section of each bar)

30 - 90 %

< 30 % (N=586)

> 90% 30-90% < 30%

< 30 % 2nd & 3rd Class ( 3,297 voyages )

1st class (on voyages where 2nd & 3rd class was filled to more than 90% of capacity) ( 871 voyages )

Fig. 3. Voyages by Extent of Capacity Utilization: Spring (March 16–June 15), Westbound. Notes: The years selected omit the recession years of 1904, 1908, 1911, and the slow recovery year of 1912. Of the 1,025 voyages, during the Spring weeks of these years, on which the 2nd and 3rd class capacity usage westbound exceeded 90% (shown in the top portion of the left-hand column), 871 also had accommodations for passengers in ﬁrst class, and the breakdown of the ﬁrst class capacity usage in the westbound direction on those 871 voyages is shown in the right-hand column. Source: Voyage Database (1900–1913) (trips between Europe and New York, Boston, Philadelphia, and Baltimore).

1907.42 That may seem to be still a rather low rate, but, being an average over several hundred voyages, it is not. Fig. 3 shows westward voyages for the spring (for 1907 and nine other years during the period): divided into three categories by extent of capacity usage, rather than shown as a single overall average. Fully one-third of those westward ship crossings used over 90 percent of second and third class passenger capacity to house passengers (see the upper left-hand portion of the graph), at a time of year when (as the right-hand part of diagram shows) most of those same voyages had a westward capacity usage in ﬁrst class of less than 30 percent.

236

DREW KEELING

> 90% All voyages, broken down by rate at which passenger capacity was used

> 90%

30-90 %

Capacity Utilization Rates

< 30%

30-90%

(percentages also shown on each section of each bar) > 90% 30-90% < 30%

< 30%

West ( 811 voyages )

East ( 788 voyages )

Fig. 4. Voyages by Extent of Capacity Utilization: First Class, Early Fall (September 1–October 15) between North Europe and USA. Note: Number of trips west differs slightly from east, because some voyages had passengers only in one direction. Source: Voyage Database (1900–1913) (trips between the main ports of North Europe (Bremen, Hamburg, Southampton, and Liverpool) and the U.S. (New York, Boston, Philadelphia, and Baltimore).

Fig. 4 goes on to show that when ﬁrst class travel peaked westward, in early autumn when many American tourists were homebound, well over one-third of voyages were more than 90 percent full in that class. Going east, however, the vast majority of the same voyages had a ﬁrst class that was at least 70 percent empty. This seasonal discrepancy was at least somewhat predictable, however. Fig. 5 shows an example of a less predictable cyclical imbalance. One of the ﬁrst effects of the recession touched off by the ‘‘ﬁnancial panic’’ of October 1907 was that unused capacity became atypically greater westward than eastward. This was the counterpart to an unusually strong net eastbound ﬂow

Transport Capacity and Transatlantic Migration

237

All voyages, broken down by rate at which passenger capacity was used

> 90% > 90%

Capacity Utilization Rates 30-90% 30-90%

(percentages also shown on each section of each bar) > 90% 30-90%

< 30%

< 30%

West (204 voyages)

East (199 voyages)

< 30%

Fig. 5. Voyages by Extent of Capacity Utilization: 2nd and 3rd Class, between Europe and USA, November and December, 1907. Note: Number of trips west differs slightly from east, because some voyages had passengers only in one direction. Source: Voyage Database (1900–1913) (trips between Europe and New York, Boston, Philadelphia, and Baltimore).

of migrants going back to Europe at the end of that year, in order to escape the ensuing U.S. recession. For 1900–1913 as a whole, westbound migrant passenger crossings during the months of November and December exceeded eastbound by 42 percent. In 1907, however, the westbound ﬂow was 15 percent below the eastbound for those two months.43 These diagrams illustrate the impediments to increasing the ‘‘low’’ 40 percent average usage rate in migrant travel by reducing voyages. Spaces in some class-directions were often mostly empty, but frequently at times when capacity utilization in some other class-directions of the same roundtrip voyage was at or near maximum. In fact, the constraints were even more severe than indicated by Figs. 3–5.

238

DREW KEELING

The many voyages with capacity utilizations in excess of 90 percent reﬂect the results of two widespread capacity management measures without which the 40 percent overall rate of usage would have been still lower. Many of those voyages with capacity usage ‘‘over-90 percent’’ were actually over 100 percent. Indeed, nearly a quarter of the voyages in the entire database of voyages had an over 100 percent capacity utilization rate in one or more of the six class-directions.44 Fig. A.3 shows that between about 1–3 percent of all westbound passengers (in ﬁrst, second, and third class) were carried above and beyond capacity ‘‘limits,’’ and illuminates how this feat was accomplished. ‘‘Adjusted passengers in excess of capacity,’’ in the diagram of Fig. A.3, reﬂect the use of two general practices for coping with high passenger levels.45 First of all, in the second class especially, where most of these instances of capacity utilization going above 100 percent occurred, shipping lines made frequent use of ‘‘temporary’’ or ‘‘portable’’ capacity (fold-down beds, sofa beds, and cots) in order to exceed ‘‘normal’’ capacity limits.46 The second practice was to assign passengers from an overbooked class to the unused berths of an underbooked class. The calculation for this in Fig. A.3 makes allowances for such temporary capacity and ‘‘upwards’’ transfers47 between classes. Taken together, the two measures are estimated to have reduced the number of passengers in excess of class capacity ‘‘limits’’ to below 1 percent of all passengers transported (the curve labeled ‘‘Adjusted Passengers in excess of capacity’’ in the graph). That was still not the end of the capacity challenge since, as the ﬁnal of the three curves (in the middle, roughly, of the graph) of Fig. A.3 shows, 5–10 percent of the total passengers traveled on voyages where there was some degree of ‘‘adjusted excess.’’ All passengers on such voyages were potentially inconvenienced in some manner by the overcrowding, even if most of them were not actually bumped out of a (double-booked) berth or given a cot in a room where all the regular bunks were already taken.48 Fig. A.4 is a seasonal (rather than cyclical) depiction of the same set of squeezing and transferring procedures.49 Precisely how companies dealt with peak passenger demands is not easily determined. Nevertheless, a wealth of evidence50 indicates that temporary capacity and class interchanges of permanent capacity were indeed both used often, and the passenger and capacity numbers make it plain that something like this set of measures must have been employed frequently. Airlines and concert halls today will, from time to time, move people up to higher-priced seats as a goodwill gesture. Turn of the twentieth century steamship companies built this sort of space and crowd regulation into the supervision of their moving hotel/bunkhouses. Capacity utilization levels were not ‘‘low’’

Transport Capacity and Transatlantic Migration

239

when one considers the wide voyage-to-voyage and class-to-class variability of passenger ﬂows, and the extensive juggling by ship designers and pursers needed just to achieve an overall result as high as 40 percent. There were probably a number of mechanisms by which shipping companies took care of the ﬁnal one-half of a percent of excess capacity described above (and shown in Figs. A.3 and A.4).51 The most likely procedure was ‘‘downwards’’ transfers. By carrying the foregoing analysis one step further, and calculating that any passengers overbooked from ﬁrst class might have been accommodated in second if space were available there, and from second class if full to any empty slot in third, the remaining excess (e.g. as shown in Fig. A.5) vanishes to near zero.52 Many ships were indeed designed explicitly with rooms being for either ‘‘ﬁrst or second’’ class passengers.53 Some vessels had actually just one uniform set of accommodations which was identiﬁed alternatively as ﬁrst or second class, on a voyage-by-voyage basis. Other ships had permanently labeled second class cabins, distinct from ﬁrst class cabins, and yet the two sets of cabins, or most of them, were identical. In this latter case, the only distinction for passengers, besides ticket price, was the dining room assigned, the menu offered, the ‘‘smoking salon’’ made available, and the promenade decks provided.54 Either way, interchangeability of accommodations between ﬁrst and second class meant, in practice, tourists paying ﬁrst class fares being housed, on rare but not completely avoidable occasions, in the same compartment of rooms as migrants paying second class fares. Even without such interchanges, passengers booked in second class were anyway a mixture averaging about 25 percent tourists and 75 percent migrants.55 Subject to one major caveat, another form of ‘‘downwards’’ transfers presented even less potential for class friction. In this other main instance of interchanging capacity, the housing of second class migrants in third class quarters, those second class migrants shared a section of compartments with third class migrants. The caveat, however, is that for this sort of ‘‘downwards transfer’’ to work smoothly, the third class quarters used to house those transferred second class passengers needed to be closed-berth accommodations.56 Being able to make that kind of capacity swap was particularly advantageous for shipping lines during westbound crossings in the early fall, when returning American tourists in ﬁrst class and summer repeat migrants in second class were both at their yearly high points, but migrants in the third class were at only a secondary seasonal peak. (Fig. A.5 shows both these quarterly peaks and the growing use of transfers from second class to closedberth third class late in the period.) Contrary to common belief, and actual

240

DREW KEELING

practice of some sailing vessels, North Atlantic steamships did not frequently interchange migrants and freight cargos.57 The proportion of capacity devoted to closed berths grew rapidly during the period. Closed berths were popular with migrants wanting more privacy (e.g. families with children, or more condition-conscious repeat migrants) and those willing to pay a modest premium for the better decks, more varied food, and reduced scrutiny upon the U.S. arrival enjoyed by arrivers to America in second class (which was also popular with summer repeat migrants who went eastwards during the late Spring, when westbound migration was highest).58 In addition, however, closed berths in third class were also desirable for shipping companies who could beneﬁt by being able to more frequently rent the same capacity out in both directions due to this double-usage. The intermediate second class was also well suited for interchanging with either ﬁrst class or closed-berth steerage, and grew markedly as a percentage of overall passenger capacity. Anecdotal accounts by passengers indicate that second class was a better travel bargain than steerage (third class), relative to the service it provided.59 Yet, as shown in Section 4, second and third class both generated a higher relative margin of revenue over cost for shipping companies than did ﬁrst class. For closed berths in second and third class, the paradox of travel bargains for passengers also generating high margins to carriers is partly explained60 by sophisticated capacity management measures. Upgrading migrant travel accommodations so that they could be interchanged (at least at the upper level) with tourist accommodations (at least at the lower level), reduced some hazards and discomforts of the crossing for migrants, but at a low incremental cost to shipping lines because those same closed berths also helped these companies reduce their ﬁnancial costs and risks, e.g. those costs and risks which resulted from seasonally and cyclically ﬂuctuating migrant ﬂows. Shifting allocation of berths between travel classes had relatively small effects on the overall rate of shipping lines’ capacity utilization (see Figs. A.3–A.5). High ﬁxed costs, however, meant greatly magniﬁed effects on bottom line proﬁts.

4. THE ‘‘EXPLOITATION’’ OF MIGRANT PASSENGERS Passenger statistics, fare data, and a sample of deck plans indicate that, although steerage passengers provided 55 percent of passenger revenues on

Transport Capacity and Transatlantic Migration

241

steamships between Europe and the United States, steerage quarters took up only about 45 percent of the space devoted to passengers on those vessels.61 This imbalance of price and cost62 corroborates previously widely held, though weakly substantiated, suspicions that migrants were in some sense systematically overcharged.63 Drechsel, for example, indicated that immigrants ‘‘provided one-third of the shipping companies’ ticket income but one-half of the actual earnings.’’64 Calculations based on deck plans and price data yield less extreme but still signiﬁcant results. On average, steerage passengers paid 40 percent more, per square foot of ship ﬂoor space made available to them, than did ﬁrst class passengers.65 The degree of ‘‘overcharging’’ would be more severe were one to take into account that polished metals and ﬁne hardwood interiors, for example, were not equally distributed between passenger classes.66 On the other hand, open-berth steerage travelers had effective use of many empty berths (Dillingham, vol. 37, p. 14). And (as shown already above), on average, over half of the berths were empty. First class rooms were usually even emptier than were bunks in steerage, but ﬁrst cabin passengers could not beneﬁt from the unused space of unoccupied rooms.67 If all unoccupied open berths in steerage and no unoccupied closed cabins are assumed to have been usable by passengers in those respective travel classes, then the ‘‘overpricing’’ of steerage amounted to only 15 percent instead of the 40 percent stated above.68 Either way, ﬁrst class passengers nevertheless clearly received, on average, better bargains for their passage money69 than did steerage class passengers. This begs the question of why shipowners should forgo proﬁt opportunities in ways that beneﬁted their ﬁrst class clientele at the expense of steeragers. In other words, why did they not cut back on low-margin (low-proﬁt) cabin capacity, and provide more high-margin steerage capacity? There are ﬁve general answers which collectively explain the seeming irrationality of ‘‘overpricing’’ steerage passages. Firstly, tourists had a greater willingness to make travel decisions based on the level of on-board comforts offered. Shipping companies tended to be skeptical that across-the-board improvements to accommodations for all steerage passengers would, in and of themselves, pay off in terms of an increased volume of steerage travelers. Nevertheless, the companies did improve average on-board conditions for migrant travelers by providing a growing percentage of them with closed berths.70 Secondly, innovations on new vessels such as new a` la carte dining rooms, ballrooms, or indoor swimming pools in ﬁrst class conferred prestige on the line, and enhanced its reputation amongst passengers in all classes.71

242

DREW KEELING

Carrying a considerable number of tourists also made it easier to attract the patronage of other tourists, and migrants. By contrast, carrying a large number of migrants was less helpful in luring additional migrants and did nothing at all to improve appeal to tourists. Less than ‘‘fully’’ compensated improvements to ﬁrst class compartments thus made economic sense as a kind of ‘‘loss leader’’ or ‘‘prestige discount.’’72 A third reason for charging migrants more per square foot (than tourists were charged) was in order to provide an offset, or a ‘‘risk premium’’ for the greater cyclical variability in steerage trafﬁc (as compared to movements of cabin passengers).73 Furthermore (fourthly), although tourist trafﬁc and migrant trafﬁc both exhibited considerable seasonal ﬂuctuation, steamship companies exacted a higher ‘‘peak season’’ differential from ﬁrst class passengers.74 The susceptibility of steerage fares to price wars compounded the relatively high cyclical risks of migration. This is the ﬁfth and ﬁnal reason for steerage passengers being in effect charged an extra ‘‘premium’’ over cost compared to ﬁrst cabin travelers. ‘‘Conferences’’ (cartels) organized as a collective defense against competitive fare-slashing put nominal price ﬂoors on all fare classes, but market-share ceilings only on steerage. The most fruitful way for steamship lines to therefore increase revenues without violating conference rules,75 and risking fare wars, was by attracting tourist passengers in ﬁrst class and tourists and migrants in second class, through increased comforts, which necessarily meant more space per passenger (Himer, 1907, p. 76). Keeping general (open-berth) steerage conditions relatively unchanged76 while offering appealing and attractively priced cabins to those migrants interested and able to pay the higher ticket fares of second class77 was frequently an effective way to increase the average revenue received per migrant.

5. INVESTMENT IN AND DEPLOYMENT OF VESSELS Over 90 percent of migrant passengers during 1900–1914 traveled on ships operated by their original owners.78 Steamship lines’ decisions regarding these vessels’ size and speed, and the mixture of passenger class capacity they provided, mostly depended on the number and market share of ﬁrst class passengers which the lines thought they could attract. This conclusion comes from an analysis which assigns every vessel in the Voyage Database

Transport Capacity and Transatlantic Migration

243

to one of the eight ‘‘groups,’’ based on whether its speed, size, and number of ﬁrst class berths were ‘‘low’’ or ‘‘high,’’ viz:79 Low speed, Low size, Low 1st class capacity Low speed, Low size, High 1st class capacity ..and so forth to High speed, High size, High 1st class capacity

If these three characteristics – speed, size, and ﬁrst cabin capacity – were not correlated, then one might expect roughly equal numbers of migrant passengers80 to have traveled on each of the eight groups of ships (e.g. somewhere in the neighborhood of twelve and a half percent on each). This was not at all the case, however. Half of migrant crossers traveled on the ‘‘High speed, High size, and High ﬁrst class’’ group, called here ‘‘Express.’’81 Another 30 percent were passengers on the ‘‘Low speed, Low size, and Low ﬁrst class’’ group, called here ‘‘Regular.’’ This clustering is nonetheless not particularly surprising: ships built to attract large numbers of ﬁrst class passengers needed to be fast and spacious, and considerable additional size was also necessary in order to hold the relatively large engines and coal bunkers required for speed. The spaciousness of the ships made it advantageous for shipping lines to use them for carrying both tourists and migrants: each group in different sections of the vessels. These oceanic leviathans had deep cavernous interiors on their lower decks where ﬁrst class passengers would not wish to enter, but which could be proﬁtably ﬁlled with migrants. The upper decks were then reserved for the upper class clientele which would not travel any other way. Rather less obviously, there were also marketing and capacity management advantages to combining migrants and tourists on the same ships. It is a common maxim of shipping chroniclers and historians that migrants were partial to large and powerful ships with as many smokestacks as possible.82 Since ship size went up steadily (nearly 70 percent overall, 1900–13) and speed was correlated with the number of smokestacks, migrants (to the extent they consciously chose the liner on which to travel) preferred the same fast large ships favored by ﬁrst class passengers. The highest capacity utilization rates on ‘‘express’’ ships, relative to other vessels, was in the third class eastbound, where the type of ship was, in at least some instances, the deliberate choice of relatively savvy repeat migrants, and in the second class (westward and eastward). Capacity usage in second class depended on the extent to which overbooked passengers could be

244

DREW KEELING

transferred ‘‘upwards’’ to the ﬁrst class (a greater possibility on ‘‘express’’ ships which had larger ﬁrst class compartments than did their smaller and slower counterparts).83 In general, shipping lines preferred the ﬁrst class express ships which brought prestige to the line while also receiving high patronage from migrants. The speed and luxury of these ships were costly investments, however. Where bulky pocketbooks to pay for rapid transatlantic crossings and generously appointed tourist accommodations were infrequent, for example on routes from the more utilitarian ports of Baltimore and Philadelphia in America, and to and from most Mediterranean ports (except for spring cruises), smaller and slower (‘‘regular’’) vessels were usually deployed instead. A ﬁxed schedule of departures and arrivals that could be counted upon was the chief beneﬁt prompting the inauguration of transatlantic steamship lines in the mid-nineteenth century, and it remained a paramount requirement thereafter. In striving to always have a ship ready to depart whenever there was customer demand for it, regardless of seasonal and cyclical ups and downs and directional imbalances, companies maintained a surplus of carrying capacity throughout the 1900–1914 period. Compared to the numbers required for maintaining their weekly, bi-weekly, or monthly services, companies kept ﬂeets about 40 percent bigger than they needed.84 Parts 2 and 3 of Table A.5 show that this surplus capacity was provided on smaller than average ships,85 and used to provide a cushion of extra vessels by which cyclical and seasonal surges in migratory travel demand were accommodated. Comments in shipping company records and by shipping historians regarding ships ‘‘idled’’ or serving as ‘‘reserve vessels’’86 generally refer to the precautionary measure of having extra ships ready to substitute for vessels lost, damaged, or hired away for wartime service. Calculations based on the 1900–1914 Voyage Database, and shown in Parts 2 and 3 of Table A.5, reveal further practices by which an identiﬁable minority of company ﬂeets were kept in ‘‘reserve’’ for handling seasonal and cyclical upsurges in passenger ﬂows. As a group, these vessels, labeled as ‘‘infrequent’’ in Table A.5, were older, smaller, and more often ‘‘non-express’’ ships than were their ‘‘frequent’’ counterparts.87 With considerable predictability, the ﬁrst three months of the calendar year marked the low ebb of migratory travel demand, and the second three months the high tide. Not surprisingly therefore, it was during the ‘‘slack’’ winter months that many, if not most, ships were rotated out of service for annual overhauls.88 ‘‘Frequent’’ and ‘‘infrequent’’ vessels both made

Transport Capacity and Transatlantic Migration

245

considerably more voyages in the second calendar quarter than in the ﬁrst quarter.89 Nevertheless, although voyages in the second quarter overall (frequent and infrequent taken together) exceeded those in the ﬁrst quarter by 34 percent, second and third class passenger trafﬁc was 86 percent higher in the second quarter, compared to the ﬁrst quarter. Shipping lines’ desires to maintain at least minimal regular schedules throughout the year (e.g. including the winter season of low passenger demand) limited those companies’ abilities to match voyage frequency with quarter-to-quarter passenger ﬂuctuations.90 ‘‘Infrequent’’ ships were disproportionately used, however, in the second quarter. Cyclical variations in migration were even more difﬁcult to match because they were less predictable. In 1908, during the most severe recession of the period, second and third class passenger trafﬁc dropped by 42 percent (vs. 1907) but ship arrivals and departures were reduced by only 12 percent, and nearly all of that voyage adjustment was made by curtailing the deployment of ‘‘infrequent vessels.’’91 Maintaining backup vessels for seasonal and cyclical peak deployment, however, helped insure companies against the risk that ﬂuctuations in travel demand posed to the integrity of schedules and to painstakingly earned reputations for steady and regular service. Such scheduling imperatives were governed more by the requirements of tourist and business travelers than by the less time-sensitive travel patterns of labor migrants, but shipping lines were well aware that most migrants, like most non-migrants, were, in one sense or another, repeat customers.92

6. CONCLUSIONS Shipping lines transporting migrants across the early twentieth century North Atlantic achieved rates of capacity utilization that were low in an absolute sense, but less so when the ﬂuctuating and unpredictable nature of the migrant passengers’ ﬂows is taken into account. Heavy ﬁxed costs magniﬁed the effect of those ﬂuctuations upon the bottom line proﬁts of the transport ﬁrms. This, in turn, exacerbated the double challenge faced by shipping company managers: how to provide passenger capacity sufﬁcient for a reliable travel schedule while minimizing unused berths on the crossings. Although migrant trafﬁc was their mainstay, North Atlantic passenger shipping ﬁrms gave higher priority to serving tourist passengers. This, however, was consistent with a rational and viable set of long-run business

246

DREW KEELING

strategies. First class tourist and business travelers were smaller in number but brought shipping companies more prestige and less volatility than did migrants. Catering to the luxury class also produced other advantages for shipping enterprises, particularly by raising the crucially important capacity utilization in the budget class sections of their steamers. A focus on attracting tourists was reﬂected in shipping lines’ scheduling, in their offering cabin class accommodations at a lower margin over cost than steerage berths, and in the design and outﬁtting of their vessels, but this helped more than it hurt their migrant business. Tourists strongly preferred larger, faster, and more elegantly appointed ships, but so did migrants. Smaller, slower, more utilitarian (and generally older) vessels were more often relegated to secondary routes, or retained as ‘‘reserve vessels.’’ In addition to the considerable scale economies and opportunities for more ﬁnely segmented accommodations which accompanied increased vessel size, larger steamships offered more scope and ﬂexibility for adding supplemental capacity during peak periods of migrant ﬂows. The more important example of tourist–migrant synergy came, however, as a result of the overlapping possibilities resulting from the ability of large, fast, ‘‘ﬂoating palaces’’ to appeal to and accommodate considerable numbers of both tourists and migrants. Because tourists and migrants had somewhat offsetting seasonal and directional patterns of travel, capacity utilization could be raised by offering the intermediate second class to a shifting mixture of both upscale migrants and budget-minded tourists, and by frequently reallocating capacity between the ﬁrst, second, and closed-berth third classes. Other capacity management measures also helped shipping companies handle ﬂuctuating migrant ﬂows while continuing to pay close attention to the needs of their more prosperous traveling clientele. Because there were considerable negative effects on shipping lines’ reputations if they were more than very rarely unable to provide sufﬁcient passenger berths during peak periods, companies erred on the side of excess rather than insufﬁcient capacity. Keeping older vessels in ‘‘reserve,’’ to be called into service for supplemental departures during upsurges in travel demand, was cheaper than chartering ships from outside parties, and less damaging to shipping lines’ reputations than canceling voyages during slumps in the volume of migrant passengers. ‘‘Temporary’’ capacity (such as fold-down beds) provided a small amount of additional ﬂexibility. Other procedures, such as those for transferring passengers between departure dates, vessels, or lines were less desirable for migrants or shipping lines, but generally worked smoothly under the conference arrangements. All these measures helped shipping lines improve capacity utilization, and minimize unhappiness among delayed or

Transport Capacity and Transatlantic Migration

247

stranded migrant passengers, by providing a passage for nearly every eligible and paying migrant to the U.S. More effective use of expensive ship capacity provided the transport companies with an extra cushion of proﬁts during prosperous times, which could then be used to help them outlast business slumps.93 Such cyclical ﬂuctuations were the central risk faced by both overseas labor migrants and their transporters. To maintain their all-important reputations, including the most fundamental requirement of providing reliable service on a regular schedule, shipping lines used a range of methods to deal with peak surges in migrant travel, and relied on long-term ﬁnancial prudence to help survive cyclical downturns. An understanding of the capacity management strategies of early twentieth century passenger steamship corporations provides explanations for the seeming inconsistencies of prior observations. Migrant trafﬁc across the North Atlantic was riskier than tourist trafﬁc, and increasing migrant volumes did not enhance overall company reputation the way increasing tourist patronage did, so companies charged migrants a premium in terms of price per space rented out. Given the vital importance of maintained schedules to their reputations, however, it was also in the shipping lines’ self-interest to improve conditions to migrants, in order to raise capacity utilization (on those relatively unaltered schedules) by interchanging space between tourists and migrants. Finally, the strategy of giving migrants closed cabins, in order to double them up with tourists, made particular sense on the major tourist routes to and from Northwestern Europe, and less so on the transit corridors from eastern and southern Europe where migration grew fastest after 1890. Nevertheless, being in the same cyclical ‘‘boat,’’ shipping lines and migrants helped more than they hindered each other. Migrants relied on steamships’ regular service to get quickly back to Europe during recessions in the United States (and, often, to migrate again to America once the job market improved there). In order to spread the risks of long-distance relocation over multiple individuals making multiple ocean crossings, migrants used family and community networks, or ‘‘chains,’’ and the resulting movement patterns helped ameliorate the cyclicality of steamship lines’ passenger revenues. Dependents ﬁnanced by prepaid tickets, or by remittances sent from prior migrants, moved in a less cyclical pattern than did ﬁrst-time, self-ﬁnanced ‘‘pioneers.’’ Eastbound ﬂows, peaking during U.S. recessions, were countercyclical.94 Complementary risk management strategies, not travel cost reductions or misleading advertising, were the principal feedback through which the growth in migration and in shipping capacity during the early twentieth century reinforced each other.

248

DREW KEELING

NOTES 1. For historiographical background, see Thistlethwaite (1960, pp. 17–25, 32), Taylor (1971, pp. 164, 308), Daniels (1990, pp. 16–29), Baines (1991, pp. 1–30), and Chiswick and Hatton (2003, pp. 69–71, 82–83). 2. As used here (unless otherwise deﬁned), ‘‘capacity’’ means beds (or ‘‘berths’’) for second and third class passengers traversing the Atlantic (of which about 95 percent were migrants (see Keeling, 2005a, p. 346)) and ‘‘capacity management’’ refers to measures taken by shipping companies and designed to help make sure that there were enough berths available on their transatlantic ocean liners to accommodate their migrant passengers, while minimizing the number of unused beds. 3. See, for instance, Hourwich (1912, pp. 96–99), Gould (1979, pp. 610–615), and Brattne and Akerman (1976, pp. 194–197). 4. See, for example, Moltmann (1989, p. 312), Nadell (1981, pp. 274, 283), and Maxtone-Graham (1992, pp. 28–29). There is also a scholarly consensus that shipping agents, regardless of the degree of afﬁliation with any particular steamship line, facilitated the travel of those who had already decided to migrate, but had no signiﬁcant inﬂuence on the decision to migrate itself. See Keeling (2007, pp. 118–122). 5. i.e. from southern Europe. 12 percent of steerage (third class) passengers arriving at New York City in 1890 came there on ships from Mediterranean ports (table in Kludas, 1986, vol. 1, p. 223). By 1913, the proportion had risen to 34 percent (NDLJ, 1913/1914, pp. 382–387). Improvements to on-board accommodations provided on these routes from southern Europe lagged behind improvements on the routes from Northern Europe, however (see Dillingham, vol. 37, p. 10; Jones, 1993, p. 76). 6. Rates of capacity utilization on passenger ships between Europe and the U.S., 1900–1914, are shown and analyzed in Section 3, and in the appendix (Figs. A.2–A.5). The general importance of capacity management in shipping is discussed in Kellogg (1904, p. 127), Bastin (1971, p. 156), and Boyce (1995, p. 36). 7. which detail for every transatlantic voyage, between Europe and the four largest U.S. ports (New York, Boston, Baltimore, Philadelphia), the shipping line, vessel, European and U.S. ports serviced, arrival and departure dates, and the passenger totals in ﬁrst, second, and third class, in both directions. Between 1900 and 1914, these four ports received over 95 percent of immigrants to the United States from Europe, nearly 90 percent of all immigrants to the U.S. from anywhere, and roughly 60 percent of all westbound transatlantic migrants (see Fig. A.1 in the appendix). The smallest of these four ports, Philadelphia, handled more than the combined total of migrants arriving at all other Atlantic ports plus Galveston and San Francisco. This data set of approximately 17,000 voyages during 1899–1914 (roughly 16,000 roundtrip voyages and 2,000 ‘‘half voyages’’) is referred to hereafter simply as the ‘‘Voyage Database’’ (see Keeling, 2005a, pp. 329–330 for further particulars and sources). 8. such as vessel size, speed, builder, passenger capacities, deployment history, ownership information and retroﬁtting details, and, where available, crew numbers, cargo capacities, coal consumption, and ﬁgures breaking down third class or steerage capacity into ‘‘closed’’ versus ‘‘open’’ berths. As used here, ‘‘closed berths’’ means on-board housing consisting of enclosed ‘‘cabins’’ for 2–8 passengers. By deﬁnition, ﬁrst and second class (synonymous with ﬁrst and second cabin) passengers always expected some form of such enclosed quarters. In the 1890s, North Atlantic lines

Transport Capacity and Transatlantic Migration

249

began providing some of their third class, or steerage, passengers with closed berths as well, instead of the older, more traditional ‘‘open-berth’’ or dormitory style accommodations. Closed berths in the steerage were typically accompanied by higher standards of on-board ventilation, sanitation, cleanliness, food quality, dining facilities, available deck space, and privacy, than were made available to passengers in open berths. Closed berths were generally popular with migrant travelers and rose over the period as a percentage of on-board passenger capacity, in two ways: (1) a growing portion of third class accommodations consisted of closed berths and (2) accommodations in second class (where, on average, 75 percent of passengers were migrants, and which were 100 percent closed berth) rose relative to third (see Dillingham, vol. 37, pp. 6–13; Gibbs, 1957, p. 25; Maxtone-Graham, 1992, pp. 5, 28–29). As a percentage of all second and third class accommodations, closed berths grew over the period as follows (Voyage Database):

1900 (%) 1914 (%)

2nd Class

3rd Class

2nd and 3rd Class

11 14

9 21

20 35

Some writers have used the term ‘‘third class’’ to mean closed berths offered in an accommodation category priced below that of second class (see, for example, Taylor, 1971, p. 156; Braynard, 1972, p. 26). This is an ambiguous choice of terminology, since other sources, most especially the Transatlantic Passenger Conferences trafﬁc reports, use ‘‘third class’’ to mean all passenger accommodations priced lower than second class. That latter approach is followed here.

9. These examples come from Stopford (1997, p. 219), de Boer (1923), Introduction, Kirkaldy (1919, pp. 205–206), Meade (1905, p. 111), and those cited in Strouse (1999, p. 476). 10. For the regular passenger lines plying between Europe and the U.S. ports of New York, Boston, Philadelphia, and Baltimore which are included in the ‘‘Voyage database’’ data set used here, the breakdown of revenues by business segment during 1900–1913 was roughly: 50% 20%

migrant passengers non-migrant passengers

25% freight 5% mail

(Figures from Keeling (2007, p. 122, footnote 26). Re proﬁts and variability in proﬁts see Section 4.)

11. See, for example, Tables A.2.A and A.2.B. 12. In ship descriptions, the unit of capacity is typically the ‘‘ton’’ (100 cubic feet). In this article, however (as explained in endnote 2), it is the passenger berth (bunk or bed). 13. Between 1900 and 1914, for the shipping lines and routes considered in this analysis, about 20 percent of costs were those of acquiring and maintaining vessel ﬂeets, 10 percent were for administration on land, and 70 percent were associated with operating the ships (coal fuel, crew wages and provisions required for the oceanic crossing, and the costs of docking and loading in port). Coal, the biggest single cost item, varied depending on the size and speed of the ship and the voyage conditions, but not the number of passengers on board (see Keeling, 2005b, pp. 9–12).

250

DREW KEELING

Concerning the high degree of ﬁxed costs in passenger shipping, see Thornton (1939, pp. 201–203), Grotewald (1914, pp. 191–196), and Murken (1922, pp. 326–327) (see also Stopford, 1997, pp. 353–358). 14. Freight rates are from Harley (1988, p. 853). Annual immigration to the U.S. from Europe for the calendar years 1881 and 1913 is best measured for 1881 by the ﬁgures in the Bureau of Statistics (p. 55), and for 1913 from the monthly immigration by race (for the ‘‘races’’ or ethnicities of European origin) from the BI Bulletins. The absence of any signiﬁcant and long-lasting trend – up or down – in North Atlantic fares, 1840–1914, is discussed in Keeling (1999b, pp. 42–43). 15. Over half of migrant passages during 1900–1914 were paid for out of U.S. earnings, mostly in the form of remittances, prepaid tickets sent to relatives in Europe, or by migrants who had already been in America paying for their own repeat crossings of the Atlantic (back to Europe, and sometimes then forth to America for a second or even third time). See Fig. A.6, and Keeling (2007, Appendix 11). 16. These estimates are based on fares from Germany and Britain where the largest volume of nineteenth century emigrants embarked. Fare levels from Italy, which became the largest single emigration source country after 1900, were slightly higher by 1910 (see Table A.4). (Volumes of U.S. immigration by source country are in Historical Statistics of the United States, pp. 105–106 (series 89–101).) Taking into account all other migration costs, in addition to the transatlantic ticket, and the costs of living in the U.S. while working there, the total working time in America needed to accumulate savings sufﬁcient to ‘‘breakeven’’ on the migratory move there declined from an average of about seven months in the 1880s to about four months on average by 1913 (see Keeling, 2007, Appendix 10). The calculations for the time needed for an unskilled worker in the U.S. to earn the nominal steamship fare from Europe to America only (neglecting living costs and all other costs) are as follows: Fares to America In 1881 In 1901 In 1913

From Britain

From Germany

Source

$23 $23 $30

$29 $31 $37

see below Table A.4 Table A.4

Comparisons, 1881–1913

From Britain

From Germany

Nominal fare increase 1881–1901 (%) ‘‘Real’’ fare increase 1881–1901 (%) Nominal fare increase 1881–1913 (%) ‘‘Real’’ fare increase 1881–1913 (%)

0 17 29 32

8 25 30 33

Weeks to earn back fare in 1881 Weeks to earn back fare in 1901 Weeks to earn back fare in 1913

3.3 3.2 2.9

4.1 3.9 3.5

Sources: Nominal fares to America, for 1881 from Bonsor (1980, p. 821) and Cecil (1967, p. 18). Real fares based on consumer price index which was down 17 percent from 1881 to 1901, and up 3 percent from 1881 to 1913 (from eh/hmit, of August 2006). Weeks to earn back fare=Fares divided by average weekly unskilled wages in the U.S. of $7 in 1881, $8.1 in 1901, and $10.5 in 1913, based on Keeling (2007, Appendix 8).

Transport Capacity and Transatlantic Migration

251

17. See Keeling (2005a, pp. 345–346). 18. The generally quite modest impacts of fare changes on migration volumes is discussed in Salz (1916, p. 849), Brattne and Akerman (1976, p. 199), and Foerster (1919, p. 47). The role of political and economic conditions in Europe, and the business cycle in the U.S. are discussed, for example, by Nugent (pp. 93–97) and in Jerome (1926). Examples of fare changes leading to temporary changes in migrants’ choice of shipping lines can be found in Cecil (1967, pp. 18–20) and Murken (1922, pp. 112, 198–301, 379). The most important fare wars of the period are mentioned in Table A.4, and described in considerable detail in Murken (pp. 264–281, 374–378, 569–574). The cost, savings, and breakeven time ﬁgures presented here are based on Keeling (2007, Appendices 8–10). 19. Such rate-cutting was, with rare exceptions, matched by competing shipping lines. This might have been a viable strategy had the lower fares led to an overproportionate expansion of the total market, but there is little evidence of this occurring either episodically or as a long-term trend. For additional details on the passenger conferences, see also Aldcroft (1968, pp. 350–351), Salz (1916, pp. 848–854), and Murken (1922, pp. 26–51, 325–360, 633–689). These sources generally agree that the conferences’ principal purpose was to protect proﬁts and market shares from the negative effects of cutthroat competition to which the high ﬁxed cost, and cyclically volatile, early twentieth century transatlantic passenger shipping business was prone. There is no evidence in this literature, or in the extensive internal conference correspondence entered as evidence in U.S. vs. HAPAG (1916), of the conferences ever being designed for, or having the effect of, signiﬁcantly inﬂuencing the volume of migration. Occasional contemporary rhetoric asserting a powerful link between fare levels and migration volumes came most noticeably from immigration restrictionists decrying the ‘‘artiﬁcial stimulation’’ of migration by steamship lines. Less subjective observers noted already then that other factors, such as the business cycle, and the availability of kinship networks, were much more powerful inﬂuences shaping decisions to migrate, or not migrate. See, for example, the discussion and citations in Keeling (1999a, 2007). See also Fig. 1. 20. and, thus, non-price competition was also available as alternative to price competition. See the observations by Thiess (1907, pp. 77, 81), Aldcroft (1968, p. 355), and Hyde (1975, p. 103). 21. Statement of Sidney Lister, Cunard Passenger Manager, U.S. vs. HAPAG (p. 1503). See also McCart (1990, p. 23), Moss and Hume (1986, pp. 84–85, 134–135), and Huldermann (1922, p. 199): ‘‘‘Imitation is the sincerest form of ﬂattery’. In the realm of shipping it has always been customary for each company to proﬁt by the experience gained and the progress made by its competitors.’’ 22. See endnote 7. 23. As shown by ‘‘Foreign Ports of Arriving Aliens, 1903’’ (U.S. National Archives, Record Group 85, Entry 10, vol. 6, p. 189). 24. Not having the advantage of an exclusive ‘‘home port’’ is also one of the reasons why, after European shipping lines became dominant on the North Atlantic during the U.S. Civil War, it proved nearly impossible for new American-based companies to successfully challenge their dominance of the market (see Keeling, 2005a, pp. 78–83).

252

DREW KEELING

25. Examples and descriptions of this process in Liverpool can be found in Milne (2000, pp. 198–207, 213–217) (for the mid-nineteenth century) and Jarvis (1996). For Germany, see Cecil (1967, pp. 21, 37–38) and Powell (1908, pp. 9, 19, 25). 26. For examples, see Murken (1922, pp. 58, 364–367, 644), ARCN (1901, pp. 188–192), Whelpley (1905, pp. 151–155, 213–225), Harcourt (1988), and Hyde (1975, pp. 137–148, 157). 27. Three important sets of examples, coming in the wake of J.P. Morgan’s ultimately failed international shipping amalgamation of 1901–1903, can be found in Sections 6 and 14–15 of the ‘‘Morgan Agreement’’ of 1901, clauses 2–4 of the ‘‘IMM-UK’’ agreement of 1903 (Murken, 1922, pp. 649–658), and point 4 of the Cunard Line’s September 1902 circular to shareholders (ARCN, 1902, p. 409). See also Vale (1984, pp. 77–80), Cecil (1967, pp. 49–50), Himer (1927, p. 76), and Die Zukunft (1912, pp. 32–34). 28. Based on Keeling (2007, pp. 129–130, footnote 52). After 1900, most of these passengers came from Austria-Hungary and Russia, countries whose relatively modest port facilities offered only limited direct steamship service to the U.S. In addition, however, for reasons of convenience, comfort, travel speed, fare level, or military service avoidance, many migrants from countries with frequent direct transatlantic travel service also chose to embark from third countries. A notable example after 1900 were Italians traveling to the U.S. by way of France. There are also documented instances (notably in the 1904–1905 ‘‘fare war’’) of migrants changing their routing preferences in response to changed conditions (see New York Times, August 25, 1904, p. 2; Murken, 1922, pp. 287–288, 207). As Nugent (1992, p. 33) put it, ‘‘by 1900 most emigrants had several options as to ports of embarkation [and] passenger lines.’’ Despite local protection, there was thus considerable competition between the various lines for westbound migrant trafﬁc, and even more so in the eastbound direction. Based on calculations from the Voyage Database, New York was the departure port for 80 percent of migrants returning to Europe. With the experience of at least one prior oceanic crossing, these eastbound return migrants could readily compare fares and schedules since nearly every line was represented in New York, and many had adjacent booking ofﬁces there. The general concept of ‘‘contestability’’ applies here (see, for example, Hanlon, 1996, pp. 35–37). None of this, it may be reiterated, is any indication of shipping lines inﬂuencing decisions to migrate or not migrate across the early twentieth century Atlantic, and thereby affecting the overall volume of migration ﬂows. 29. Regarding ‘‘growing the overall market,’’ it may be reiterated that the ‘‘market’’ (annual immigration to the U.S. from Europe) was growing rapidly anyway from 1898 up until the recession of 1908 without any sustained decline in fares. (Despite some year-to-year inconsistencies, the general upward trend of U.S. immigration is accurately reﬂected by ﬁgures in the Historical Statistics of the United States, showing a tripling of the inﬂow over this 10 year span.) The importance, to passenger shipping ﬁrms, of strategies for coping with cyclical risks is outlined in Keeling (2007), ‘‘Risks Related to Migration’’ (pp. 137–141) and ‘‘Cyclicality and Shipping Industry Networks’’ (pp. 147–150). 30. The 30 percent operating proﬁt margin is an estimate based on the annual reports of the HAPAG, North German Lloyd, Cunard, Holland America, CGT, and Anchor steamship lines, which are available, and provide varying amounts of ﬁnancial

Transport Capacity and Transatlantic Migration

253

detail, for the entire 1900–1914 period. These companies carried about half of all migrants between Europe and the USA during that period. Their annual reports show an average overall operating proﬁt margin over that period (and on all segments of their business, not just migrant trafﬁc), of 30 percent for NDL, the largest migrant carrier, 36 percent for Cunard, and 26 percent for CGT. Margins on migrant trafﬁc were higher than on tourist trafﬁc (sec Section 4) but not much higher than the overall operating margin, because migrant revenues constituted a considerably larger share of total revenues. The cyclical volatility of migrant passenger ﬂows was tempered somewhat by other business segments also handled by these lines (see endnote 10 for the breakdown by segment). Nonetheless, small ups and downs in the fraction of on-board passenger quarters occupied by paying passengers still had greatly magniﬁed impacts upon shipping companies’ ﬁnancial performance. For example, in the recession year 1908, second and third class passenger volume dropped by 41 percent versus 1907 (a 65 percent plunge in westbound trafﬁc being partially offset by a 12 percent rise eastbound). Total net proﬁts reported in these available annual reports, however, dropped by 162 percent, e.g. they went from a total of a $6.1 million proﬁt in 1907 to a $3.8 million loss in 1908. (Passenger volume ﬁgures here are from the Transatlantic Passenger Conference reports on which the Voyage Database is based.) 31. See Hyde (1975, pp. 60, 63, 157). Note, however, that the focus here is more on the management of existing capacity, i.e. ways to get more use of it, and less on longterm decisions about how much capacity to acquire by ordering new ships (for that, see, for example, Meade, 1905). Re the types of ships acquired and their deployment scheduling, see Section 5. 32. because of the insensitivity of migration movement to fare levels, as described in Section 2. Also see Stopford (1997, pp. 38–39). 33. Ships being movable, any purely local deﬁciency could be readily remedied. 34. Calculations based on the Voyage Database, which are also consistent with Stopford (1997, p. 456). 35. Discussed, for instance, in Thiess (1907, pp. 67–68) and Thornton (1939, p. 149). 36. By the early 1900s, under most jurisdictions, passengers forced to wait in-port for a vessel departure had to be housed at shipping company expense. See, for instance, Kraut (1994, pp. 47–48), Italian Emigration Act of 1901 (p. 16), Dillingham (vol. 4, p. 98), Himer (1907, p. 66), and de Boer (1923, p. 66). 37. e.g. outside the routes between Europe and New York, Boston, Philadelphia, and Baltimore covered in the Voyage Database. 38. As a clarifying example, roundtrip ‘‘capacity utilization,’’ as used in this article, means counting upwards from the bottom of Fig. 2, the subtotal of layers 1 and 3 divided by the sum of layers 1–4. 39. Though precise ﬁgures are lacking, it is clear that very little of the passenger space on these ships was ever used to carry freight, i.e. it was indeed mostly empty (see endnote 57). 40. Load factor average calculated from AEA Monthly Trafﬁc Reports. An apt, if possibly apocryphal, quip from the modern airline industry has it that ‘‘recession is when you tighten your belt, depression is when you have no belt to tighten, and when you have no trousers, you are in the airline business.’’ (Attributed to Adam Thompson, CEO of British Caledonian Airways, according to various internet sources.) As Fig. A.2 indicates, however, early twentieth century North Atlantic

254

DREW KEELING

migration was quite a bit more cyclical even than early twenty-ﬁrst century North Atlantic airline travel. (Based on that ﬁgure, if migrant passenger ﬂows during 1901–1913 had had only the degree of cyclical ﬂuctuation exhibited by airline passenger movements during 1992–2004, the rate of capacity utilization would have been about 65 percent instead of 40 percent.) European migrants in early twentieth century America worked disproportionately at temporary jobs, marginal employment, and cyclical industries. The change in the level of migrant workers, i.e. the migration itself, was still more cyclical. 41. despite the rise of back-and-forth repeat migration over the period (see Keeling, 2006, pp. 12–13, 15–16, 21–23). 42. See Fig. 2. 43. Calculations from the Voyage Database. Westbound trafﬁc at the end of 1907 would undoubtedly have been much lower still were it not for many European departures which had already committed to before the October ‘‘panic,’’ or before the news of it reached migrant source villages in Europe (Boston Evening Transcript, December 4, 1907, p. 16). 44. Those six: ﬁrst, second, and third class, west- and eastbound. The Voyage Database incorporates ship revisions and retroﬁts which not infrequently (sometimes more than once over the course of a vessel’s lifetime) reallocated ‘‘permanent’’ berths from one travel class to another. 45. Deck plans, shipping catalogues, newspaper accounts, and investigative reports conﬁrm that shipping companies planned for and made extensive use of both techniques described here (see also Murken, 1922, p. 91). 46. See deck plans and Isherwood in Sea Breezes 21 (1956, p. 105) (Olympic), 27 (1959, p. 327) (Imperator), 46 (1972, p. 87) (Principe di Piemonte). 47. From third class to second class or from second class to ﬁrst class. 48. In the process of juggling berth assignments in order to ﬁnd space for everyone on board. 49. Fig. A.4 shows that the biggest ‘‘crunch’’ came during the fall season of homebound American tourists, and for reasons explained below. 50. See, for instance, Zeitschrift der Hamburg Amerika Linie (1912, p. 65), Babcock (1931, p. 161), Brinnin (1971, p. 279), Bonsor (1980, p. 1375), Report of Conditions (1908, p. 123), Dillingham (vol. 37, pp. 31–32), and Murken (1922, p. 91). 51. There are documented instances of passengers sleeping in ‘‘hatchways’’ and being given spare berths in the crew quarters. See letter of September 26, 1904, from Mr. and Mrs. A.P. Anderson, and Mrs. Brook, passengers on Cedric arriving at New York September 12, 1904 (U.S. Bureau of Immigration ﬁles (1900–1914); U.S. National Archives, Record Group 85, Entry 9 51435/1 (box 13), and Report of Conditions, 1908, p. 130). At this scale, measurement error in the capacity numbers is also a possibility, but is neither very likely, given the large number of sources and observations, nor is it at all necessary on general principles since it is clear that transatlantic steamers could physically hold as much as double their ofﬁcial passenger capacities. A number of these ships did so when used as troop transporters in World War I. For further details, see Knox (2003, under Chapter III, ‘‘Troops Carried’’). 52. The measured excess disappears for all but a couple of dozen voyages (out of about 17,000). In other words, almost no voyages had capacity utilizations of over 100 percent for all three classes in the same direction.

Transport Capacity and Transatlantic Migration

255

53. As evidenced by the deck plans of Cedric, Caronia, Carmania, by Flayhart (2000, p. 197) re St. Louis, Isherwood Sea Breezes 10 (1950, pp. 220–221) (Oceanic), and Bonsor (p. 746), re Titanic, and in Murken (1922, p. 113). 54. See, for instance, New York Times (June 21, 1907, p. 6) and Boston Evening Transcript (April 27, 1911, p. 4). 55. See endnote 8 (beginning with ‘‘such as vessel size’’). 56. This is because second class passengers had paid for and expected to be quartered in closed berths (see endnote 8 re ‘‘open’’ versus ‘‘closed’’ berths). In other words, second class passengers transferred to third class berths needed to end up in a room very similar to that which they would have had in second class, in order to not feel inconvenienced or mistreated by the transfer. (An analysis of contemporary reports by passengers and of shipping deck plans indicates that, most of the time, closed-berth third class rooms were indeed very similar in size, shape, appointments, and ship location to second class rooms. See deck plans and New York Tribune, July 9, 1903, p. 3, NDL Annual Report, 1908, p. 4, and New York Herald, September 12, 1909, p. 12.) There are no known instances of a passenger who paid for closed rooms being housed in open-berthed steerage. On very rare occasions it may have happened, friends not wanting to be separated, etc., but even then probably not because it had to happen in order to cram everyone on board. Without known exception, all second class accommodations were closed berths whereas only a small, though rapidly growing, portion of steerage capacity had this desirable quality, on the North Atlantic in the years before World War I. Exactly how these transfers were carried out in each instance is of course not known, but it is entirely conceivable that in most cases, a portion of the closed-berth third class was simply relabeled as second for that particular ocean crossing (see the references in endnote 50). 57. For a number of reasons, but mainly because these passenger liners carried ﬁne freights in both directions (no natural offset to migrant ﬂows) and because the large cargo holds for bulk freights, on some German and Dutch liners, for instance, could not conveniently handle many passengers. Calculations suggesting freight being used as a ‘‘backhaul’’ to migrant passengers (taken in the opposite direction of a roundtrip voyage), as shown, for example, in Harley (1982, pp. 79–85), do not prove that on any given individual ship, the identical on-board space was interchangeably used for transporting freights and migrant passengers. By 1900, this was evidently a quite infrequent occurrence. Large freight carriers among the passenger lines, such as HAPAG and Holland America, also relied on extensive ﬂeets of freight-only vessels. See Keeling (2007), the section entitled ‘‘Human versus Freight Cargoes, pp. 122–129.’’ 58. See endnote 8 and Keeling (2007, pp. 128–129, footnotes 49 and 50). Concerning the relatively minimal impact of governmental regulations upon onboard conditions before World War I, see Jones (1989). 59. See, for example, Bastin (1971, pp. 61–62) and Steiner (1906, p. 41). 60. Further reasons and details are in Section 4. 61. These revenue and space percentages are for steerage/all passengers. Passenger ﬁgures used to derive these results are from the Voyage Database covering travel between Europe and New York, Boston, Baltimore and Philadelphia during 1900–1913, and described in the introduction (Part 1) of this article. See also notes 7 and 8 above. Fare ﬁgures used in this comparison are derived from Tables A.3

256

DREW KEELING

and A.4. in the appendix. The deck plans used are from 14 vessels which collectively carried about 10 percent of migrant passengers on those routes during those years (see ‘‘Vessel Deck Plans’’ in the ‘‘References’’). The sample is weighted toward larger, more luxurious ships but is still believed to be representative for purposes comparing steerage to ﬁrst class. The vessels in the deck plan sample had both a more spacious ﬁrst class than the general population of North Atlantic passenger liners and a higher percentage of closed berths in steerage. Because fare wars over the migrant trafﬁc (sharp reductions in steerage ticket prices) were relatively infrequent, they had no great effect on the long-term average fare level, and thus no major impact upon the relatively low level of space per fare paid in steerage (vs. ﬁrst class). Because the second class had a much higher capacity utilization rate, it could be (and was) offered to migrants at lower price per ‘‘ton’’ (a ‘‘ton’’ being 100 cubic feet of space on a ship), than steerage or ﬁrst class, yet provided shipping lines with a higher yield per ‘‘ton’’ than the other two classes did. 62. For any given vessel, the proportion of costs allocable to each class of travel service (ﬁrst, second, and third or steerage) was effectively correlated with the proportion of revenue-generating space provided for that class of passengers. (The costs were also generally related, though not in measurably direct proportion, to the quality of service provided to passengers). There are two parts to the reasoning behind this result. Firstly, building and powering these large ocean steamers soaked up roughly half of shipping lines’ total expenses. That estimate is based on annual ﬁnancial statements of the six major transatlantic shipping lines (cited under References) which transported over half of all passengers between Europe and the U.S. during the 1900–1914 period. These annual reports show the cost of building, maintaining, insuring, and repairing ships at about 20–25 percent of all shipping line costs. According to the voyage abstracts and annual reports of the Cunard line for those years, coal expenses were another 26 percent of total costs (coal was 36 percent of voyage costs and voyage costs 72 percent of total costs). This 26 percent ﬁgure can be considered as roughly representative for the other major lines. On the one hand, Cunard’s ships were 16 percent faster (measured in knots) and 22 percent bigger (measured in gross tons per passenger berth) than the overall Europe–USA ﬂeet of all passenger lines (based on calculations from the Voyage Database). The greater speed and the greater spaciousness of its liners both had the effect making the number of tons of coal burned per passenger mile higher on Cunard. On the other hand, the price of coal per ton was at least 10 percent higher for non-British lines than for British lines, the average distance was 25 percent further (coal usage is directly proportional to distance), and crew wages were 30 percent lower (based on a comparison of 15 crew lists of the Holland America line versus eight from Cunard and White Star, and the same 30 percent ﬁgure is conﬁrmed in ARCN (1901, p. 39), and, at Cunard, crew wages were 13 percent of total costs). The net effect is that a (rounded) 25 percent ﬁgure for Cunard’s percentage of costs taken up by coal expenditures applies as a reasonable estimate for the North Atlantic passenger shipping lines as a whole. Secondly, nearly all other voyage and operating costs were proportional to the amount of passenger capacity, and thus the amount of space, on the ships, not the numbers of passengers carried. (This conclusion derives from the analysis of the crew lists referenced here and from Keeling (2005b, pp. 9–12).) Over time, based on the deck plans referenced here, and the Voyage Database, technical

Transport Capacity and Transatlantic Migration

257

improvements and scale economies meant that each new generation of ever-larger ships provided more space per passenger in each class, and with little or no corresponding differential in fare (see Tables A.2.A and A.2.B). Thus, passengers on newer ships got better bargains for their ticket expenditures than did travelers on older vessels. But the more favorable ratio of price paid to space (and service) enjoyed in the ﬁrst class still applied to each ship, and thus to the North Atlantic crossing overall, and throughout the period (and, evidently, in earlier decades as well: see Bastin, 1971, pp. 147–151). 63. i.e. as per Nadell (1981, p. 274), Bastin (1971, pp. 60–61), Steiner (1906, pp. 35–41, 347–348), and Brandenburg (1904, pp. 175–176). 64. Drechsel (1995, p. 297) provides no proof or citation for these ﬁgures, which are, however, broadly consistent with widely found anecdotal observations (e.g. Dillingham, vol. 37, p. 37; Pitkin, p. 68) and the quantitative results presented here (see the immediately following endnote). As Braynard and Miller (1991, p. 6), put it: ‘‘First class brought the prestige, but third the income.’’ Bureau of Immigration investigators exaggerated, however, in concluding, in their ‘‘Report of Conditions’’ (1908, p. 106), that ‘‘although steerage passengers paid fully onethird of the average ﬁrst class rate, they do not receive one-third the consideration given ﬁrst-class passengers; in fact, one-tenth would be a more nearly accurate estimate.’’ Based on the data used here, the ratio of operating expense per passenger berth in steerage was about a ﬁfth of what it was for ﬁrst class. 65. This comparison measures space, not cost, per se. But, since nearly all of the shipping lines’ costs were ﬁxed once a voyage was committed to (see endnote 13), and were expended by passenger travel class – ﬁrst, second, and third – in proportion to the space assigned to each class, as shown on the deck plans (see endnote 62), those costs can be allocated between the classes based on the relative space used to accommodate each class. The deck plans used are listed in the References, the price data is estimated based on the sources in Table A.3, supplemented – for the ﬁrst class – by available ﬁgures from the steamship line brochures listed under References. The average ﬁrst class fare for 1900–1914 was calculated to be $103, and the average steerage fare $31, which means that a ﬁrst class ticket cost, on average, 3.3 times as much as a steerage ticket did. According to the measurements from the deck plan sample used here, however, the on-board ship space reserved for the ﬁrst class works out to 4.6 times as much space per passenger as was allocated for steerage. 4.6 divided by 3.3 is just about 1.4, i.e. steerage passengers paid, on average, 40 percent more per unit of space than their ﬁrst class counterparts did. Assuming that by ‘‘immigrants’’ Drechsel (1995) means ‘‘steeragers’’ and by ‘‘actual earnings’’ operating proﬁts, an estimate for 1900–1913 using Drechsel’s terminology, but consistent with the data analyzed here, is that ‘‘immigrants’’ produced about 40 percent of ticket revenues (including non-passenger revenues such as freight) and 50 percent of earnings. See endnote 10 for the estimated revenue breakdown between passenger types: migrants (e.g. in the steerage as well as the non-steerage ‘‘cabin’’ class) tourists, freight, etc. 66. i.e. the construction cost of a square foot in ﬁrst class exceeded that of a square foot in steerage. 67. Cabin passengers could not readily sprawl into unused adjoining cabins in the same way that an open-berth passenger could, for instance, place hand baggage on a nearby empty bunk.

258

DREW KEELING

68. These calculations ignore, for example, the vertical space obtainable by not using upper fold-down bunks in cabins (e.g. Warren, 1995, p. 24). If the more generous pricing of space rented to the second class, where most passengers were migrants, is taken into account, the discrepancies diminish. Based on the deck plans sampled here, steerage (third) and second class passengers as a group paid, on average, only 24 percent more per area than did ﬁrst class passengers. For usable space, as deﬁned here, the premium paid by steerage and second class (together) versus ﬁrst class, amounted to only 5 percent. 69. See the ﬁrst paragraph in this section. Another form of ‘‘exploitation’’ was the luring of passengers through misleading persuasion or advertising. For a typical contemporary account of such ‘‘unnatural’’ stimulation of migration, see Whelpley (1905, pp. 10–21) (cf. Hourwich, 1912, pp. 96–99). The consensus of later historians is that such luring was not a major factor in decisions to migrate (see, for example, Taylor, 1971, p. 105; Brattne and Akerman, 1976, pp. 191–195). 70. Re ‘‘closed berths’’ generally, see endnote 8. Cunard used Saxonia and Ivernia, the ﬁrst all-closed berth steerage vessels, on its Liverpool–Boston route, beginning in 1900 (Voyage Database). As described in Keeling (2006, pp. 14–19), the routes to and from Northern Europe where closed berths were ﬁrst widely adopted, also had the highest incidence of summer trips to Europe by repeat migrants, e.g. travel patterns seasonally congruent with summer tourism. Further quantitative evidence for this conclusion can also be found in Dillingham, vol. 3, p. 359, and in the passenger lists of the U.S. National Archives. Contemporary impressions and examples are given in Wall Street Journal (May 11, 1903, p. 2, August 19, 1904, p. 2), and Boston Evening Transcript (July 3, 1901) (see also Tedebrand, 1976, p. 226, and endnote 5). 71. See, for instance, Pedraja (1992, p. 10). 72. As conﬁrmed by Thiess (1903, p. 37), Drechsel (1995, p. 297), Ottmu¨ller (1986, p. 143), and Flayhart (2000, p. 351). 73. Revenues from migrants were nearly twice as volatile as were revenues from ﬁrst class passengers and freight transport (see Keeling, 2007, footnote 26). Re the imperative of maintaining unchanged travel schedules see, for example, Maxtone-Graham (1972, p. 44) and Grotewald (1914, p. 194). 74. See the notes to Fig. A.2, and also Keeling (2007, pp. 164–165, Appendix 6). Figures in HAPAG, Zeitschrift der Hamburg Amerika Linie also indicate a wider seasonal variation in fares for ﬁrst class than for steerage. In general, intra-class price discrimination was much more extensive in ﬁrst class (not just seasonal differential seasonal pricing, but also variations depending on vessel, size of the room, its location on the ship, etc.). 75. See Keeling (1999a, pp. 202–203) for more on the origins and effects of these conferences. 76. i.e. not improving on-board space or amenities in open-berth steerage, and lowering steerage fares only when the prescribed market share was being signiﬁcantly underachieved, and even then lowering rates only as far down as the agreed-upon ‘‘compensation’’ rate under conference rules. The typical passenger conference agreement in those days stipulated that participating steamship lines owed a compensation payment to the ‘‘pool’’ for each passenger carried in excess of their quotas, while those lines who were below-quota were to receive recompense at

Transport Capacity and Transatlantic Migration

259

that same rate from the pool. Either way companies would be throwing away money if they sold steerage passages at below the ‘‘compensation price’’ (typically set around d3–4 (equivalent to roughly $15–20)) (see Murken, 1922, pp. 32, 644–645, 661–662, 681). 77. or, on some German, Austrian, and Dutch line vessels, a special, separate and higher-priced closed-berth third class. (Open berths, on those vessels, were termed ‘‘fourth class.’’) See, for instance, Braynard (1972, p. 26). Most vessels in the deck plan sample used here (see the ‘‘References,’’ for the list of those vessels) had both closed and open berths in the steerage, which suggest that the overall trend toward closed berths did not alter the greater proﬁtability of steerage of either sort. See also the ﬁrst endnote in this section (endnote 61), re second class having a higher capacity utilization than either ﬁrst or steerage. 78. This conclusion and the immediately following analysis are based on calculations from the Voyage Database. 79. The dividing lines between ‘‘high’’ and ‘‘low’’ were set at or near the mean, median, and mode of knots (speed), gross tonnage (size), and ﬁrst class berths for each of the ﬁve ‘‘generations’’ of ships (vessels built before 1890, 1890–1899, 1900–1904, 1905–1909, and after 1910). Net tonnage could also have been used since it was closely correlated with gross tonnage during the period (see Keeling, 1999b, p. 49). ‘‘Net tonnage’’ measures the volume of revenue-generating areas on a ship, such as passenger accommodations and freight holds. ‘‘Gross tonnage’’ is the total volume of the whole vessel. 80. Here approximated by second and third class passengers. 81. 16 percent of vessels in the Voyage Database were ‘‘Express’’ and they carried 49 percent of roundtrip second and third class passengers. 82. See, for example, Guthrie (1971, p. 182), Drechsel (1995, p. 297), Maxtone-Graham (1972, pp. 30–31), and Fletcher (1913, pp. 94–95). 83. Average second class capacity utilization was 27 percent higher on express ships as compared to non-express ships. The comparable difference for third class was 24 percent. Breaking down the third class by direction of travel, eastbound express vessels had a 59 percent higher capacity utilization than non-express, westbound the difference was 14 percent (ﬁgures from Voyage Database). The technical need for the third or fourth smokestack (funnel) had gone by about 1910, but builders perpetuated the popular association of speed (and safety) with the number of funnels by including ‘‘dummy funnels’’ on ships such as the 1911 Olympic and the 1914 Vaterland (Braynard, 1972, p. 21; Isherwood, 1956 in Sea Breezes 21, p. 103). 84. Based on the Voyage Database. Re ﬁxed schedules, see endnote 73 and Broeze (1992, p. 145). 85. The ‘‘II & III Class Capacity, Infrequent vs. Frequent’’ column. 86. Examples include the NDL Annual Report (1902, p. 3), Isherwood (1960), Sea Breezes 30, p. 16 (La Touraine), and (1967) Sea Breezes 41, p. 92 (Majestic). 87. For the analysis of cyclical and seasonal vessel deployment in Table A.5, it was not feasible to use identical demarcations between ‘‘frequent’’ and ‘‘infrequent’’ vessels. (The deﬁnitions are given at the end of Table A.5.) Nonetheless, 80 percent of ‘‘infrequent’’ cyclical ships were also ‘‘infrequent’’ seasonal ships. Vessels switched at least once between different routes were mostly ‘‘infrequent.’’ They made up only a

260

DREW KEELING

small fraction of such ‘‘infrequent’’ ships, but were representative with respect to age, carrying capacity etc. 88. See Bisset and Stephensen (1959, p. 275) and McCart (1990, p. 43) re annual maintenance. Voyage Database measures of the intervals between voyages of each vessel (this is different from the interval between voyages – of several vessels – on a particular route) show that half of all interruptions of the regular sequence (usually monthly or bi-monthly) of a vessel’s voyages occurred during December through March. Some examples are also in The Times (London, November 9, 1911, p. 18). 89. See Table A.5, Part 3, ‘‘Excess of 2nd quarter voyages over 1st quarter voyages.’’ Note, however, that the column ‘‘% of vessel years with voyages in 2nd quarter but not 1st,’’ shows a nearly fourfold greater incidence (29 percent vs. 8 percent) of such ‘‘second quarter only’’ ships in the ‘‘infrequent’’ category, as compared to the ‘‘frequent’’ ships. 90. The Voyage Abstracts of Cunard, one of the more proﬁtable lines, indicates the extent of ﬁnancial sacriﬁces companies were prepared to undergo in order to maintain schedules during ‘‘slack’’ months of the year. Of 687 Cunard voyages, Liverpool to the U.S., between 1900 and 1913, over one-quarter did not cover variable (!) costs (i.e. voyage expenses exceeded voyage revenues). Nearly two-thirds (64 percent) of these loss voyages occurred during just one-third of the calendar year: the months of December through March. Loss-defying schedules were maintained cyclically as well. In the recession year of 1908, 38 percent of voyages had a loss against variable costs (voyage costs) versus 18 percent in the boom year of 1907. Calculations from Cunard Voyage Abstracts. 91. See Table A.5, Part 2, ‘‘% by which 1907 voyages exceeded 1908 voyages.’’ These voyage and passenger comparisons (2nd to 1st quarter, and 1907–1908) are measured based on the Voyage Database. 92. As shown in Fig. A.6, 55 percent of migrant crossings were made by people traveling west from Europe for the ﬁrst time. But many of those had their passage paid for, directly or indirectly, by relatives who had crossed before them (see, for example, Dillingham Report, vol. 4, p. 61). Migrants crossing under the direction of, or being funded by, someone else are designated in Fig. A.6 as ‘‘dependent’’ migrants. 93. Recall that ﬁxed costs meant that such measures had a magniﬁed effect on proﬁts. 94. See Figs. A.6 and A.7.

ACKNOWLEDGMENTS The author would like to thank Ulrich Woitek, Skip Fischer, and attendees at an October 2005 seminar presentation to the Institut fu¨r Empirische Wirtschaftsforschung at the University of Zurich, and at the conference on ‘‘The impact of maritime and migration networks on transatlantic labour migration, 18th–20th centuries’’ held at the European University Institute, Florence, November 2005, and an anonymous reader, for their comments and suggestions. Any errors remaining, however, are the author’s responsibility.

Transport Capacity and Transatlantic Migration

261

REFERENCES PRIMARY AND REFERENCE BOOK SOURCES AEA (Association of European Airlines). (2002–2004). Monthly Trafﬁc Reports. Anchor Line. (1900–1914). Annual reports and archives. Business History Centre Archives, University of Glasgow, UK. Bisset, J., & Stephensen, P. R. (1959). Tramps and ladies. New York: Criterion Books. Bollettino Dell’ Emigrazione. (1914–1915). Ministero Degli Affari Esteri. Rome: Stab. Tip. Societa` Cartiere Centrale. Bonsor, N. R. P. (1975–1980). North Atlantic seaway: An illustrated history of the passenger services linking the old world with the new. Newton Abbot: David & Charles. Boston Evening Transcript. (1901, 1907, 1911). Brandenburg, B. (1904). Imported Americans. New York: Frederick A. Stokes. Braynard, F. O., & Miller, W. H., Jr. (1991). Picture history of the Cunard line, 1840–1990. New York: Dover. [British Airways Trafﬁc Reports] British Airways, Monthly Trafﬁc and Capacity Statistics Reports. (1991–2004). Aavailable on the Internet. CGT Annual Report. (1900–1914). CGT Archives. Le Havre, France. Cunard Annual Reports. (1885–1914). Cunard Archives. University of Liverpool, Liverpool, UK. Cunard Voyage Abstracts. (1885–1914). Cunard Archives. University of Liverpool, Liverpool, UK, AC 12/1–6. De Boer, M. G. (1923). The Holland-America Line, 1873–1923. Amsterdam: L. Van Leer. Denevan, W. M. (1992). The native population of the Americas in 1492. Madison: University of Wisconsin Press. Die Zukunft. (July, 1912). Drechsel, E. (1995). Norddeutscher Lloyd, Bremen, 1857–1970: History, fleet, ships, mails. Vancouver: Cordillera. Economist. (2003). Open the skies. Economist, October 2. Fletcher, R. A. (1913). Travelling palaces: Luxury in passenger steamships. London: Pitman. Foerster, R. F. (1919). The Italian emigration of our times. Cambridge: Harvard University Press. Frankfurter Zeitung. (1908). Fry, H. (1896). The history of North Atlantic steam navigation. London: Sampson Low. Gibbs, C. R. V. (1957). Passenger liners of the western ocean: A record of the North Atlantic steam and motor passenger vessels from 1838 to the present day (2nd ed.). London: Staples Press. Guthrie, J. (1971). A history of marine engineering. London: Hutchinson. HAPAG. (1899–1914). Annual Reports. Staatsarchiv, Hamburg, Germany. [HAPAG Guide] HAPAG. (1906). Guide through Europe. Berlin. HAPAG. (1901–1912). Zeitschrift der Hamburg Amerika Linie. Himer, K. (1907). Die Hamburg-Amerika Linie im sechsten Jahrzehnt ihrer Entwicklung, 1897–1907. Berlin: Ecksteins. Himer, K. (1927). Geschichte der Hamburg-Amerika Linie (Part 2: ‘‘Albert Ballin’’). Hamburg: Gustav Petermann. Historical Statistics of the U.S. (1975). Series C89–C101. Holland America Line. (1899–1914). Annual reports. Rotterdam: Gemeentearchief.

262

DREW KEELING

Holland America Line. (1901–1914). Passagegelden. Rotterdam: Gemeentearchief. Holland America Line. (1900–1913). Aamonstering-Boeken. Rotterdam: Gemeentearchief (crew lists). Holland America Line. Voornaamste Technische Gegevens. Rotterdam: Gemeentearchief. [Holland-America Tech]. Hourwich, I. A. (1912). Immigration and labor: The economic aspects of European immigration to the United States. New York: Putnam. Huldermann, B. (1922). Albert Ballin. London: Cassel. [Isherwood] Isherwood, J. B. (1949–1987). Steamers of the past. Series in Sea Breezes. Italian Emigration Act of 1901. English Translation in U.S. National Archives, Record Group 85, 51411/26. Jordan, H. (1936). Mauretania. Bath: Patrick Stephens. Kellogg, A. P. (1904). The backwater of immigration. Charities, 12(6), 127–128. Kludas, A. (1986). Die Geschichte der deutschen Passagierschiffahrt (5 volumes). Berlin: Ernst Kabel. Kristiania Emigranten Protokollen and Statistik opgave over Udvandringen (1899–1914). Politietaten – Reisekontroll/Emigrasjon. Oslo: Riksarkivet. Liverpool Public Library Archives. (1900–1902). Crew lists of British steamers. Lufthansa Annual Reports. (1995–2004). Maritime History Archive. Memorial University, St. Johns, Newfoundland (crew lists). McCart, N. (1990). Atlantic liners of the Cunard line: From 1884 to the present day. Wellingborough, Great Britain: Patrick Stephens. McEvedy, C., & Jones, R. (1978). Atlas of world population history. Harmondsworth: Penguin. Meade, E. S. (1905). The capitalization of the international mercantile marine. In: W. Ripley (Ed.), Trusts, pools, and corporations (pp. 105–120). Boston: Ginn & Company. Menard, R. R. (1991). Migration, ethnicity, and the rise of an Atlantic economy: The repeopling of British America, 1600–1790. In: R. Vecoli & S. Sinke (Eds), A century of European migrations, 1830–1930 (pp. 58–77). Urbana: University of Illinois Press. Murken, E. (1922). Die groXen transatlantischen Linienreederei-Verba¨nde, Pools und Interessengemeinschaften bis zum Ausbruch des Weltkrieges: Ihre Entstehung, Organisation und Wirksamkeit. Jena: Gustav Fischer. NDL. (1914). Norddeutscher Lloyd Bremen. Jahrbuch 1913/1914. Bremen: Hofbuchdruckerei H.M. Hauschild. In the archives collections of the Library of the Deutsche Schiffahrtsmuseum, Bremerhaven [NDLJ, 1913–14]. NDL Annual Reports. (1899–1914). Schiffahrtsmuseum Archiv. Bremerhaven, Germany. New York Herald. (1909). New York Times. (1904–1907). New York Tribune. (1903). Salz, A. (1916). Auswanderung und Schiffahrt mit besonderer Beru¨cksichtigung der o¨sterreichischen Verha¨ltnisse. Part 2: Archiv fu¨r Sozialwissenschaften und Sozialpolitik (Vol. 42, pp. 842–884). Steiner, E. A. (1906). On the Trail of the immigrant. New York: Revell. Stopford, M. (1997). Maritime economics (2nd ed.). London: Routledge. Thiess, K. (1903). Organisation und Verbandsbildung in der Handelsschiffahrt. Meereskunde in gemeinversta¨ndlichen Vortra¨gen und Aufsa¨tzen. Herausgegeben vom Institut fu¨r Meereskunde an der Universita¨t Berlin (Vol. 1, No. 1, pp. 1–48). Berlin: Mittler und Sohn.

Transport Capacity and Transatlantic Migration

263

Thiess, K. (1907). Deutsche Schiffahrt und Schiffahrtspolitik der Gegenwart. Leipzig: Teubner. Thornton, R. H. (1939). British shipping. London: Cambridge University Press. The Times (London). (1899, 1911). Transatlantic Passenger Conferences Reports. (1899–1914). New York. [Powell] U.K. Report on Activity of the Hamburg-American Packet Company. (1908). Report by Consul Wilfred Powell. British Foreign Ofﬁce (Public Record Ofﬁce, UK), October. [BI Bulletins] U.S. Bureau of Immigration. (1909–1914). Immigration Bulletins (monthly). U.S. Bureau of Immigration ﬁles. (1900–1914). National Archives, Record Group 85. [Report of Conditions] U.S. Bureau of Immigration. (1908). Report of conditions existing in Europe and Mexico affecting emigration and immigration. National Archives, Record Group 85, 51411/1,2. U.S. Bureau of Statistics. (1889). Treasury department. Tables showing arrivals of alien passengers and immigrants in the United States from 1820 to 1888 (Table 2, p. 55). Washington, DC. U.S. Citizenship and Immigration Service. (2006). Immigration and permanent residence tables from http://www.uscis.gov/graphics/shared/statistics/data/index.htm on June 21. [ARCN] U.S. Commissioner of Navigation. (1900–1902). Annual Reports. [Dillingham] U.S.: Reports of the Immigration Commission of 1907–1911, chaired by Senator William P. Dillingham (published in 42 volumes by the 61st U.S. Congress). [Industrial Commission] U.S.: Reports of the Industrial Commission on Immigration and Education (House of Representatives document No. 184, 57th U.S. Congress, 1st session). (1901). [Passenger lists] U.S. National Archives. (1900–1914). Microﬁlmed passenger lists of arriving vessels. U.S. National Archives. (1903). Foreign ports of arriving aliens. Record Group 85, Entry 10 (Vol. 6, p. 189). U.S.: Statistical Abstracts of the United States. (1870–2003). U.S. vs. HAPAG et al. (1911–1916). Records and briefs, 239 US 466 (microﬁlm). Wall Street Journal. (1903–1904). Warren, M. D. (1995). Distinguished liners from ‘‘The Shipbuilder,’’ 1906–14. New York: Blue Riband. Whelpley, J. D. (1905). The problem of the immigrant. London: Chapman & Hall. Willcox, W. F., & Ferenczi, I. (Eds). (1929–31). International migrations. New York: National Bureau of Economic Research.

STEAMSHIP COMPANY BROCHURES Hamburg Staatsarchiv Library, Hamburg, Germany, A902. La Veloce Line. Steamship brochure dated April 29, 1901. Powderly papers. Catholic University Archive, Box 138 (Miscellaneous Correspondence), Folder: ‘‘Immigration Ofﬁce Correspondence’’, 1900–1902. Steamship brochures collection. Steamship Historical Society, Baltimore, MD. Steamship brochures. Ellis Island Archive, New York, NY. U.S. National Archives, Record Group 85. Entry 9 51411/29.

264

DREW KEELING

VESSEL DECK PLANS Vessel

Location of deck plan

Campania Carmania Cedric Cincinnati France Friesland Haverford Imperator Kaiserin Auguste Viktoria Kristianiafjord La Provence Lusitania Nieuw Amsterdam Olympic President Lincoln Rotterdam (III) Rotterdam (IV)

Merseyside Maritime Museum, Ship # 26/1-2. Glasgow Business History Centre, Ship # 366. Merseyside Maritime Museum, Liverpool, deck plans. Staatsarchiv Hamburg, HAPAG Ship Plans 4276 1/ Warren, pp. 149ff. Glasgow Business History Centre, Ship # 242. Glasgow Business History Centre, Ships #s 344, 345. Staatsarchiv Hamburg, HAPAG Ship Plans 3980 1/ Schiffahrtsmuseum Bremerhaven,VII 4 II 312,313,314 Shipbuilder, August 1913. CGT Archives, Le Havre, 9AQ 52–59. Glasgow Business History Centre, Ship # 367. Museum Maritiem, Rotterdam, deck plans. Shipbuilder, Summer 1911. Staatsarchiv Hamburg, HAPAG Ship Plans 3914 1/ Oosterwijk, B. Zes Mall Rotterdam. Rotterdam: Soeren, 1998, p. 139. Oosterwijk, op. cit., p. 275.

SECONDARY SOURCES Aldcroft, D. (1968). The mercantile marine. In: D. Aldcroft (Ed.), The development of British industry and foreign competition, 1875–1914 (pp. 326–363). London: George Allen. Babcock, F. L. (1931). Spanning the Atlantic. New York: A. A. Knopf. Baines, D. (1991). Emigration from Europe, 1815–1930. London: Macmillan. Bastin, R. (1971). Cunard and the Liverpool emigrant traffic, 1860–1900. MA thesis, University of Liverpool. Boyce, G. (1995). Information, mediation, and institutional development: The rise of large-scale enterprise in British shipping, 1870–1919. Manchester: Manchester University Press. Boyce, G. (2001). Transferring capabilities across sectoral frontiers: Shipowners entering the airline business, 1920–1970. International Journal of Maritime History, 13(1), 1–38. Brattne, B., & Akerman, S. (1976). Importance of the transport sector for mass emigration. In: H. Runblom & H. Norman (Eds), From Sweden to America: A history of the migration (pp. 176–200). Minneapolis: University of Minnesota Press. Braynard, F. (1972). World’s greatest ship: The story of the Leviathan (Vol. 1). New York: South Street Seaport Museum. Brinnin, J. (1971). The sway of the grand saloon. New York: Delacorte Press. Broeze, F. (1992). Albert Ballin, the Hamburg America Line and Hamburg: Structure and strategy in German shipping industry (1886–1914). Deutsches Schiffahrtsarchiv (15), 135–158.

Transport Capacity and Transatlantic Migration

265

Cecil, L. (1967). Albert Ballin: Business and politics in imperial Germany, 1888–1918. Princeton: Princeton University Press. Chiswick, B. R., & Hatton, T. J. (2003). International migration and the integration of labor markets. In: M. Bordo, A. M. Taylor & J. G. Williamson (Eds), Globalization in historical perspective (pp. 65–120). Chicago: National Bureau of Economic Research. Daniels, R. (1990). Coming to America: A history of immigration and ethnicity in American life. Princeton: Harper Collins. Flayhart, W. H. (2000). The American Line (1871–1902). New York: W.W. Norton. Gould, J. D. (1979). European inter-continental emigration, 1815–1914: Patterns and causes. Journal of European Economic History, 8, 593–679. Grabbe, H-J. (2001). Vor der groXen Flut: Die europa¨ische Migration in die Vereinigten Staaten von Amerika 1783–1820. Stuttgart: Franz Steiner. Grotewald, C. (1914). Die Deutsche Schiffahrt in Wirtschaft und Recht. Stuttgart: Enke. Hanlon, P. (1996). Global airlines: Competition in a transnational industry. Oxford: ButterﬁeldHeinemann. Harcourt, F. (1988). British oceanic mail contracts in the age of steam. Journal of Transport History, 9(1), 3rd series, 1–18. Harley, C. K. (1988). Ocean freight rates and productivity, 1740–1913: The primacy of mechanical invention reafﬁrmed. Journal of Economic History, 48(4), 851–876. Harley, C. K. (1982). Issues on the demand for shipping services, 1870–1913: Derived demand and problems of joint production. In: L. Fischer & E. Sager (Eds), Merchant shipping and economic development in Atlantic Canada (pp. 67–86). (Proceedings of the 5th conference of the Atlantic Canada Shipping Project, June 25–27, 1981.) Maritime History Group, Memorial University of Newfoundland. Harrison, M. J. (2003). European integration and transatlantic air services agreements: Who has the authority? American Consortium on European Union Studies, p. 7 (http://www. american.edu/aces/pages/publications.html Retrieved, August 2005). Hatton, T. J., & Williamson, J. G. (1998). The age of mass migration. New York: Oxford. Hyde, F. E. (1975). Cunard and the North Atlantic, 1840–1973: A history of shipping and financial management. London: Macmillan. Jarvis, A. (1996). Serving the ﬂoating palaces. Paper presented at a Research Day School held at Merseyside Maritime Museum, 15 June. Jerome, H. (1926). Migration and business cycles. National Bureau of Economic Research Publication No. 9. St. Albans, Vt.: Messenger. Jones, M. A. (1989). Aspects of North Atlantic migration: Steerage conditions and American law, 1819–1909. In: K. Friedland (Ed.), Maritime aspects of migration (pp. 321–331). Cologne: Bo¨hlau. Jones, M. A. (1992). American immigration (2nd ed.). Chicago: University of Chicago Press. Jones, M. A. (1993). Transatlantic steerage conditions from sail to steam, 1819–1920. In: B. F. Larsen, & H. Bender (Eds), On distant shores (M. L. Hansen immigration conference proceedings, Aalborg Denmark, 1992) (pp. 59–82). Aalborg: Danes Worldwide Archives. Keeling, D. (1999a). Transatlantic shipping cartels and migration between Europe and America, 1880–1914. Essays in Economic and Business History, 17, 195–213. Keeling, D. (1999b). The transportation revolution and transatlantic migration, 1850–1914. Research in Economic History, 19, 39–74.

266

DREW KEELING

Keeling, D. (2005a). The business of transatlantic migration between Europe and America, 1900–1914. Ph.D. dissertation, University of California, Berkeley. Re the ‘‘Voyage Database’’ compiled for this dissertation, see endnotes 7 and 8 above. Keeling, D. (2005b). The economics of migratory transport between Europe and the USA, 1900–14. University of California World History Workshop Working Paper. Keeling, D. (2006). Repeat migration between Europe and the United States, 1870–1914. Institute of European Studies Working Paper. University of California. Accessed on http://repositories.cdlib.org/ies/050411 Keeling, D. (2007). Costs, risks, and migration networks between Europe and the United States, 1900–1914. In: T. Feys, L. Fischer, S. Hoste & S. Vanfraechem (Eds), Maritime transport and migration: The connections between maritime and migration networks, Research in maritime history (Vol. 33, pp. 113–173). St Johns: International Maritime Economic History Association. Kirkaldy, A. W. (1919). British shipping: Its history, organisation and importance. London: Kegan Paul, Trench, Trubner & Co. Knox, D. W. (2003). American naval participation in the Great War (with special reference to the European theater of operations). From http://www.history.navy.mil/library/special/ american_naval_part_great_war.htm (as of June 18, 2006). Kraut, A. (1994). Silent travellers: Germs, genes, and the ‘‘immigrant menace’’. New York: Basic Books. Maxtone-Graham, J. (1972). The only way to cross. New York: MacMillan. Maxtone-Graham, J. (1992). Crossing and cruising: From the Golden Era of ocean liners to the luxury cruise ships of today. New York: Scribner. Milne, G. J. (2000). Trade and traders in mid-Victorian Liverpool. Liverpool: Liverpool University Press. Moltmann, G. (1989). Steamship transport of emigrants from Europe to the United States, 1850–1914: Social, commercial and legislative aspects. In: K. Friedland (Ed.), Maritime aspects of migration (pp. 309–320). Cologne: Bo¨hlau. Moss, M., & Hume, J. R. (1986). Shipbuilders to the World: 125 Years of Harland and Wolff, Belfast, 1861–1986. Belfast: Blackstaff Press. Nadell, P. S. (1981). The journey to America by steam: The Jews of Eastern Europe in transition. American Jewish History, 71, 269–284. Nugent, W. (1992). Crossings: The great transatlantic migrations, 1870–1914. Bloomington: Indiana University Press. Ottmu¨ller-Wetzel, B. (1986). Auswanderung u¨ber Hamburg: Die H.A.P.A.G. und die Auswanderung nach Nordamerika, 1870–1914. Ph.D. dissertation, Freie Universita¨t Berlin. Pedraja Toma´n, R. (1992). The rise and decline of U.S. merchant shipping in the twentieth century. New York: Twayne. Pitkin, T. (1975). The keepers of the gate: A history of Ellis Island. New York: NYU Press. Strouse, J. (1999). Morgan: American financier. New York: Random House. Taylor, P. (1971). The distant magnet: European migration to the U.S.A. New York: Harper & Row. Tedebrand, L. (1976). Remigration from America to Sweden. In: H. Runblom & H. Norman (Eds), From Sweden to America: A history of the migration (pp. 201–227). Minneapolis: University of Minnesota Press.

Transport Capacity and Transatlantic Migration

267

Thistlethwaite, F. (1960). Migration from Europe overseas in the nineteenth and twentieth centuries. In: R. Vecoli & S. Sinke (Eds), A century of European migrations, 1830–1930 (pp. 17–57). Urbana: University of Illinois Press (original publication 1960). Vale, V. (1984). The American peril: Challenge to Britain on the North Atlantic, 1901–04. Manchester: Manchester University Press. Voyage Database: see endnotes 7 and 8 in this article, and Keeling (2005a, pp. 329–330). Walton, G. M., & Rockoff, H. (1994). History of the American economy (7th ed.). Fort Worth: Dryden Press. Wokeck, M. S. (1999). Trade in strangers. The beginnings of mass migration to North America. University Park, PA: Pennsylvania State University Press.

APPENDIX Table A.1. U.S. Immigrants as a Percentage of U.S. Population, 1810–2005, with a Focus on Key Peaks and Troughs of Immigration. Average Annual Immigrants to U.S. (in Thousands) 1620–1699 1700–1774 1775–1815 1816–1819 1820–1825 1826–1835 1836–1846 1847 1848 1849 1850 1851 1852 1853 1854 1855–1865 1866–1873 1874–1880 1881 1882 1883 1884–1893 1894–1899 1900 1901

2 6 9 25 8 36 81 235 227 297 370 379 372 369 428 169 338 23 669 789 603 487 285 449 488

U.S. Population (Period Averages in Millions)

Average Annual Immigrants/U.S. Population (%)

4 4 6 9 10 13 18 21 22 23 23 24 25 26 27 31 40 47 52 53 54 61 72 76 78

0.05 0.1 0.1 0.3 0.1 0.3 0.5 1.1 1.0 1.3 1.6 1.6 1.5 1.4 1.6 0.5 0.9 0.5 1.3 1.5 1.1 0.8 0.4 0.6 0.6

268

DREW KEELING

Table A.1. Average Annual Immigrants to U.S. (in Thousands) 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915–1919 1920–1924 1925–1929 1930–1945 1946–1965 1966–1980 1981–1990 1991–2005

649 857 813 1,026 1,101 1,250 410 1,068 1,042 879 838 1,198 1,218 235 555 304 60 246 443 784 1,189

(Continued ) U.S. Population (Period Averages in Millions) 79 81 82 84 85 87 89 90 92 94 95 97 99 103 110 119 130 167 210 239 267

Average Annual Immigrants/U.S. Population (%) 0.8 1.1 1.0 1.2 1.3 1.4 0.5 1.2 1.1 0.9 0.9 1.2 1.2 0.2 0.5 0.3 0.05 0.1 0.2 0.3 0.4

Note: ‘‘Peak’’ years, when immigration exceeded 1% of the population, are shown in italics. ‘‘Troughs’’ (corresponding to major wars or economic recessions), are marked by asterisks. Sources and definitions: Immigration: For 1600–1819: Walton and Rockoff (1994, p. 28), McEvedy and Jones (1978, p. 286), Jones (1992, p. 19), Wokeck (1999, pp. 45, 172), Grabbe (2001, pp. 93–94), and Menard (1991, p. 61). For 1820–1970: Historical Statistics of the U.S. (1975). For 1971–2005: U.S. Citizen and Immigration Service tables. 1620–1808 immigration ﬁgures include African slaves (about 1/3 of the total). 1900–1914 immigration ﬁgures are not adjusted for undercount of cabin class and repeat migrants (would have added about 10% to the total). 1960–2005 immigration ﬁgures are adjusted for estimated illegal immigration (about 15% of the total). Population: For 1620–1789: Walton and Rockoff (1994, p. 33), McEvedy and Jones (1978, p. 286) (population for these years includes estimated numbers of aboriginals per Denevan, 1992, p. xxvii; Menard, 1991, pp. 59–60). For 1790–2005: U.S. Census (interpolated). Periods: U.S. government ﬁscal years except for ‘‘1907’’: July 1906 to December 1907; ‘‘1908’’: calendar year; and ‘‘1909’’: January–June, 1909 (for these three years the numbers of immigrants have been adjusted to show an annual rate, consistent with the rest of the table). ‘‘Trough’’ years are those dominated by major wars and major recessions. With few exceptions (1904, 1911–1912), ‘‘peak years’’ are those (ﬁscal) years of strong economic growth (e.g. per Walton & Rockoff, 1994, p. 400).

Economies of Scale in Costs and Ship Size.

Coal consumption per ship size as a function of ship size Vessel Group 1a 1b 2a 2b 3a 3b 4a 4b 5a 5b Totals

# Vessels per Group

% all II & III Class Passengers, 1900–1913

Vessel Start Date

Knots

Gross Tons (Space)

Tons (Weight) of Coal per Day per 1,000 Tons (Space)

2 5 5 6 2 7 5 7 3 3

1 2 2 6 1 5 4 7 2 2

1895 1900 1900 1901 1909 1907 1896 1900 1907 1909

14 14 15 15 17 17 19 20 23 24

6,467 11,261 9,094 13,097 10,895 20,741 11,107 17,178 20,809 39,256

14 10 12 11 16 13 30 28 34 25

45

32

Transport Capacity and Transatlantic Migration

Table A.2.A.

Notes: Vessel group: 1a: Principe di Piemonte, Rotterdam (III), Spaarndam; 1b: Haverford, Merion, Rhein, Statendam, Suevic; 2a: Bergensfjord, Czar, Germanic, Romanic, Zeeland; 2b: Ivernia, Kroonland, Noordam, Potsdam, Rijndam, Saxonia; 3a: Caledonia, Kaiser Franz Josef I; 3b: Adriatic, Baltic, Cedric, Lapland, Nieuw Amsterdam, Prinz Friedrich Wilhelm, Rotterdam; 4a: La Lorraine, La Savoie, Majestic, St Louis, St Paul; 4b: Amerika, Campania, George Washington, Kaiser Wilhelm der Grosse (K. W. d. Grosse), Kronprinz Wilhelm, Lucania, Oceanic; 5a: France, Kaiser Wilhelm II, Kronprinzessin Cecile (Kronpr’sin Cecile); 5b: Lusitania, Mauretania, Vaterland. Sources: Size, speed, service dates, and coal usage data per Isherwood, Holland-America Tech, Bonsor (1980).

269

270

Table A.2.B.

Economies of Scale in Costs and Ship Size. Technical crew levels and ship size

Line

Vessel

% all II & III Class Passengers, 1900–1913

Vessel Start Date

Knots

Tech Crew

Gross Tons (Space)

Tech Crew per 1,000 Tons (Space)

Potsdam

0.7

1900

15

105

12,606

8

Holl. Am. White Star Holl. Am.

Nieuw Amsterdam Cedric Rotterdam (IV)

0.6 1.0 0.4

1905 1903 1908

16 16 16

150 158 175

16,967 21,035 24,149

9 8 7

White Star White Star Cunard

Majestic Oceanic Caronia

0.7 1.2 0.9

1890 1899 1905

19 19 18

223 220 196

9,965 17,272 19,678

22 13 10

Cunard NDL White Star

Campania K. W. d. Grosse Olympic

1.1 1.3 0.3

1893 1897 1911

21 22 21

256 242 349

12,950 14,369 45,324

20 17 8

NDL Cunard HAPAG

Kronpr’sin Cecilie Mauretania Vaterland

0.5 0.9 0.1

1907 1907 1914

23 25 23

366 393 513

19,400 31,937 54,282

19 12 9

Totals

13

8

Notes: ‘‘Tech crew’’=technical crew=sailors, ofﬁcers, engine room workers; Holl. Am. ¼ Holland America. Sources: ARCN (1901, p. 39), Braynard (1972, p. 52), Holland-America Tech, UK crew lists, Jordan (1936, p. 319), and Steamship Historical Society brochures.

DREW KEELING

Holl. Am.

Ship Line (s)

Westbound Steerage Fares, 1901–1913.

Cunard

Anchor

From Norway

Holland America

German

Italian

Liverpool to U.S.

Mediterranean to NY

Glasgow to NY

Oslo to U.S.

Rotterdam to NY

Germany to NY

Italy to U.S.

Currency

(d)

(d)

(d)

(Kr)

(Dﬂ)

(Mark)

(Lira)

1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912

4.64 4.51 4.55 2.80 4.31 4.65 5.28 4.72 6.45 6.48 6.50 6.55

4.93 4.91 5.20 3.51 5.05 5.24 5.44 5.81

115 108 103 72 103 114 115 123 147 161 150 138

67.41

130 162 145 140 155 160 165 150 167 150 160 165

175

Route

5.21 5.61 5.61 5.94 6.01 5.94 5.80 5.72 6.19

78.00 67.46 79.72 87.41 77.15 74.39 98.08 96.88 99.11 105.46

175 176 180 205 207 207 210

271

Cunard

Transport Capacity and Transatlantic Migration

Table A.3.

Table A.3.

Route Currency 1913 (NY=New York)

Cunard

Cunard

Anchor

From Norway

Holland America

German

Italian

Liverpool to U.S.

Mediterranean to NY

Glasgow to NY

Oslo to U.S.

Rotterdam to NY

Germany to NY

Italy to U.S.

(d)

(d)

(d)

(Kr)

(Dﬂ)

(Mark)

(Lira)

6.14

4.00

162

94.76

156

210

DREW KEELING

Note re fares calculations: The Holland America fares were measured by dividing actual passenger receipts for each voyage by the number of (adult equivalent) passengers, summed over all voyages for the period. 1902 is omitted because three months of records are missing from that year. The Anchor Line fares were computed by taking the available annual ﬁgures of ‘‘Passage Money’’ and dividing by adult equivalent passengers. Cunard’s fares were derived by dividing passenger revenues in the Cunard Voyage Abstracts by the adult equivalent passenger ﬁgures, summed over all voyages. (‘‘Adult equivalent’’ ﬁgures are used because infants under one year of age traveled for free and children 1–12 years old were charged only half fare). The Norwegian fares were computed as Revenue/Total Passengers (i.e. by including children, they are understated by about 5%). The Italian and German fares are estimates based on available published fares of German lines (HAPAG and NDL) and the Italian lines (particularly NGI). Most westbound steerage passengers in this period were foreigners subject to the U.S. head tax, and fares were adjusted, as needed (see below), to reﬂect this. The head tax was $1 in 1901, $2 as of 1903, and $4 as of 1907 (Dillingham Report, 1911, Vol. 39, pp. 45, 51, 60) and, relative to the average total fare, grew from 3% in 1901 to 11% in 1913. Note re head tax: Presumably, the Cunard Voyage Abstract revenue ﬁgures (the basis for the Cunard fares shown here) were net of the U.S. head tax, because head tax is not listed in the cost columns of those accounts. According to the 1901 U.S. Industrial Commission (pp. 18, 105), however, the head tax was added to the westbound ticket price paid by non-U.S. passengers. (According to Kludas, the German lines in 1909 were an exception, and charged separately for the ticket and the head tax.) The Cunard fares in the table, and the German fares in 1909, thus reﬂect the addition of the head tax which, it is assumed, was paid by the passengers as part of the fare, but not included in the Cunard revenue ﬁgures (and in the German 1909 case, paid outside of the fare but as part of the effective transit cost). Because of rounding differences and this adjustment for the head tax, the Cunard fares shown here are higher than those shown in Keeling (1999b, pp. 64–65), by about 5% in 1901, 19% in 1913, and 12% over 1901–1913 period. The published fares of the Italian and German lines (except in 1909) are assumed to include head taxes, as are the Holland America and Anchor line and Norwegian data – monies paid by passengers (rather than net revenue to the steamship line as with Cunard). Data sources: Cunard Voyage Abstracts (1885–1914), Holland America Passagegelden, Anchor Line Archives (1900–1908), Glasgow Business History Centre, UGD 255/1/2/7, Kristiania Emigranten Protokollen and Statistik opgave over Udvandringen (1899–1914), La Veloce brochure (for 1901 Italian fares), Bollettino Dell’ Emigrazione (March 15, 1915, pp. 60–77) (Italian fares, other years). German fares for 1901, 1904–1907, 1912: Zeitschrift der HAPAG, for 1902–1903, 1909–1911: Kludas (1986, Vol. 3, pp. 331–332), for 1908: Kludas and Frankfurter Zeitung (June 9, 1908 (A), p. 4), for 1913: Kludas, Zeitschrift, Willcox and Ferenczi (1931, Vol. 2, p. 329).

272

Ship Line (s)

(Continued )

Table A.4.

Westbound Steerage Data: Annual Fares (in $) and Passengers, 1901–1913.

Cun (L) 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913

$22.5 $21.9 $22.1 $13.6 $20.9 $22.6 $25.6 $23.0 $31.3 $31.5 $31.6 $31.9 $29.8

Average

$25.2

Cun (M)

$25.3 $27.3 $27.3 $28.9 $29.2 $28.8 $28.2 $27.8 $30.1 $19.4

Anchor

Norway

Holland America

German

Italian

$24.0 $23.9 $25.3 $17.1 $24.5 $25.5 $26.5 $28.2

$30.8 $29.0 $27.6 $19.3 $27.6 $30.6 $30.8 $33.0 $39.4 $43.2 $40.2 $37.0 $43.4

$27.0

$31.0 $38.6 $34.5 $33.3 $36.9 $38.1 $39.3 $35.7 $39.8 $35.7 $38.1 $39.3 $37.2

$33.0

$31.2 $27.0 $31.9 $35.0 $30.9 $29.8 $39.2 $38.8 $39.6 $42.2 $37.9

$31.9

$33.0 $33.2 $34.0 $38.7 $39.1 $39.1 $39.6 $39.6

$37.0

Transport Capacity and Transatlantic Migration

Average Annual Fares

Annual Passengers Cun (L) 36,616 46,946 61,528 69,128 64,812 93,195 93,282 43,752 61,850

24,016 39,388 50,363 54,105 13,823 38,755

Anchor

Norway

Holland America

German

6,760 9,574 11,909 12,980 17,270 23,635 26,509 8,439 14,991

6,638 11,192 14,958 11,695 11,608 11,289 10,688 5,122 7,890

25,967 32,530 36,763 25,630 42,144 42,988 46,718 11,436 30,412

139,666 166,455 183,358 156,655 189,819 230,509 267,856 76,581 182,729

Italian

38,192 51,612 58,532 36,209 68,661 87,040 115,699 32,269 126,993

Total these lines 253,839 318,309 367,048 336,313 433,702 539,019 614,857 191,422 463,620

As % of all lines 51 48 48 49 48 49 51 55 54

273

1901 1902 1903 1904 1905 1906 1907 1908 1909

Cun (M)

Table A.4.

(Continued ) 274

Annual Passengers

1910 1911 1912 1913 Totals

Cun (L)

Cun (M)

Anchor

Norway

Holland America

German

Italian

Total these lines

70,713 50,125 58,074 68,911

41,930 20,118 31,584 37,205

18,429 11,272 11,205 16,190

9,367 6,547 5,619 5,987

35,814 22,744 33,858 49,158

176,265 105,412 175,356 253,563

111,250 71,760 100,291 159,891

463,768 287,978 415,987 590,905

818,932

351,287

189,163

118,600

436,162

2,304,224

1,058,399

5,276,767

As % of all lines 52 48 50 53

As a % of all Steerage Passengers to New York, Boston, Baltimore, and Philadelphia, 1901–1913 8

3

2

1

4

22

10

51

DREW KEELING

Data sources: Lines, routes and fares are from Table A.3 (currency conversion per HAPAG Guide, 1906). ‘‘Cun (L)’’=Cunard, Liverpool, ‘‘Cun (M)’’=Cunard, Mediterranean ports. Passenger ﬁgures, from Transatlantic Passenger Conference Reports, are steerage travelers westbound to New York, Boston, Baltimore, and Philadelphia, ports accounting for 99% of steerage arrivals to the U.S. in the period. Comments: In general, passenger ﬂows were not negatively correlated with fares, as is often assumed. One major exception to this occurred during the 1904 fare war in Britain (discussed in Keeling (2005a, Chapter 4) and is reﬂected in the Cunard (Liverpool) and Anchor fares. Overall, 1904 was a recession year and outside of Britain passenger volumes declined. Fare discounting also occurred during the 1908 recession and in 1913. Over 1901–1913, the passenger-fare correlation rates were: Cunard (L) 2%, Holland America 22%, Italian lines 45%, German lines 37%, and Norwegian 63%. Another exception, the anomalous inverse correlation of Norwegian fares and passengers, largely reﬂects a sharp drop off of Scandinavian, especially Norwegian, emigration after 1908, due, for instance, to the rapid growth of GNP and wages in Scandinavia during the period (see, for example, Hatton & Williamson, 1998, pp. 190–199). The Norwegian fares have an over 90% correlation with Cunard’s fares from Liverpool. (There is, in fact, an overlap between the two because about 20% of the passengers in the Norwegian data series were transit migrants traveling to the USA on the Cunard line.) Except for the 1904 fare war, fare differences between routes were fairly stable over the period. Persisting fare differences across routes mainly reﬂect cost differences. Voyage costs were proportional to route distance. For example, route length explains 30% of the average fare difference between the German Lines and Cunard–Liverpool (re distances, see ARCN, 1900, pp. 312–317). Other cost differences between routes are described in Keeling (2007, pp. 132–136). German and Italian fares are slightly overstated relative to those of the other lines because their published fare ﬁgures do not reﬂect unofﬁcial discounting. Seasonally, steerage fares varied hardly at all (ibid., Appendix 6, pp. 164–165). The general fare rise of 1909 offset increasing costs incurred by shipping lines, was made possible by a strengthening of market-sharing conference agreements between shipping lines (Keeling, 2005a, pp. 263–268, 370), and represented, in some cases, a rebound from fare drops during the 1908 recession.

Transport Capacity and Transatlantic Migration

Table A.5.

275

Vessel Deployment Trends.

Part 1: Secular: 1910–13 vs. 1900–1903 1910–1913 vs. 1900–1903 (%)

1910–1913 vs. 1900–1903

Annual averages Lines Routes Voyages Vessels

+31 +44 +30 +20

(24 (42 (1,160 (177

Total II & III capacity in ’000s

+68

(2,072 vs. 1,235)

Ratios Routes per line Vessels per route Voyages per vessel

+10 +17 +9

(1.7 vs. 1.6) (4.2 vs. 5.1) (6.6 vs. 6.0)

II & III Capacity per voyage

+29

(1,786 vs. 1,385)

+4

(38% vs. 36%)

II & III Passengers per II & III capacity

vs. vs. vs. vs.

19) 29) 892) 148)

Part 2: Cyclical: 1908 vs. 1907 (159 Vessels, 1,806 Voyages) Vessels classiﬁed by frequency of voyages

Ex (%)

Non-Ex (%)

47 53

1 59

99 41

Infrequent Frequent

Voyages by vessel type

% of voyages

Ex (%)

Non-Ex (%)

13 87

3 63

97 37

Infrequent Frequent

Infrequent Frequent

Vessels by vessel type

% of vessels

% by which 1907 voyages exceeded 1908 voyages

% of voyages switched between routes

II & III class capacity infrequent vs. frequent

Vessel age infrequent vs. frequent

53 4

16 1

21%

24%

Table A.5.

(Continued )

Part 3: Seasonal: April–June vs. January–March (345 Vessels, 7,758 Voyages)

Infrequent Frequent

44 56

Infrequent Frequent

Infrequent Frequent

Vessels by vessel type

% of vessels

Vessels classiﬁed by frequency of voyages

Ex (%)

Non-Ex (%)

5 34

95 66 Voyages by vessel type

% of voyages

Ex (%)

Non-Ex (%)

19 81

10 51

90 49

Excess of 2nd quarter voyages over 1st quarter voyages

% of voyages switched between routes

II & III class capacity infrequent vs. frequent

% of vessel years with voyages in 2nd quarter but not 1st quarter

39% 33%

13 1

15%

29 8

Notes and definitions: The ﬁgures in this Table A.5 exclude 167 vessels (32% of all in the Voyage Database) which, though they made 2,082 voyages (12% of the total) during 1899–1914, carried only 217 thousand 2nd and 3rd class passengers (1.1% of the total). Ex=‘‘Express’’ vessel, as deﬁned in Section 5 and Endnote 81; Non-Ex=Non-‘‘express vessels’’=all vessels other than ‘‘express’’ vessels, as deﬁned in Section 5; A voyage switched between routes means that the immediately preceding voyage of that vessel took place on a different route; ‘‘II & III capacity’’=‘‘II & III class capacity’’=Number of passenger berths in 2nd and 3rd class, westbound and eastbound. Cyclical: ‘‘Infrequent’’ vessels=Vessels making fewer than ﬁve voyages in 1907 and fewer than ﬁve voyages in 1908; ‘‘Frequent’’ vessels=Vessels making ﬁve or more voyages in 1907 and ﬁve or more voyages in 1908; Vessels with maiden voyages in 1907 or 1908 are neither ‘‘frequent’’ nor ‘‘infrequent.’’ (In this classiﬁcation, 21% of vessels, making 20% of voyages, were neither ‘‘frequent’’ or ‘‘infrequent,’’ and those vessels are not shown in Part 2 of this appendix.); The average number of voyages per year, for all vessels during 1907–1908, was 5. Seasonal: ‘‘Infrequent’’ vessels=Vessels which averaged fewer than three voyages during the months of January–June, 1900–1914; ‘‘Frequent’’ vessels=Vessels which averaged three or more voyages during the months of January–June, 1900–1914; Excluded from this calculation are vessels whose maiden voyage occurred in the 2nd quarter or whose ﬁnal voyage occurred in the 1st quarter. In this classiﬁcation, every vessel was either a frequent or an infrequent vessel; (Average voyages for all vessels was 3.5 per January–June half year, 1900–1914.). Vessel years with voyages in 2nd quarter but not 1st quarter: For any given vessel, every year in which there were no voyages in the 1st quarter (January–March) and at least one voyage in the 2nd quarter (April–June); The table shows these vessel years as a % of all vessel years with at least one voyage in the months of January–June; Excluded from this calculation are vessels whose maiden voyage occurred in the 2nd quarter or whose ﬁnal voyage occurred in the 1st quarter. Source: Voyage Database.

Transport Capacity and Transatlantic Migration

Table A.6.

Passenger Arrivals at New York from Europe, 1863–1914.

Year

Big Four Steerage

1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905

277

All Lines

Cabin

Total

19 26 27 62

7 9 11 15

26 35 38 76

181 212 230

14 18 27

57 81 59 82 121 107

16 17 15 19 24 20

73 98 74 101 145 127

212 257 213 229 293 255

47

17

64

184 179 159

23 25 26

208 204 185

148 118 135

30 33 38

178 151 173

147 178 207

15 53 56

162 231 263

143 131 94 124 166 217 230 275 313 316 393

63 53 55 56 51 53 66 85 80 83 100 105 115

198 187 146 177 233 302 310 357 413 421 508

Steerage

Cabin

Big 4/All Lines Total

Steerage (%)

Cabin (%)

Total (%)

14 13 27

65 61 54

16 18 16 30

28 29 28 31 37 33

162 194 230 257 245 240 287 241 260 330 288

27 31 28 36 41 42

59 59 54 62 64 59

30 34 31 39 44 44

100

35

135

47

48

47

64

33

97

321 441 455 388 321 281 301 372 384 314 372 445 388 365 188 259 252 192 220 300 403 437 573 642 568 774

45 51 58 59 65 55 69 79 86 106 99 105 121 122 93 97 99 90 81 107 138 128 140 160 161 184

366 492 513 447 386 336 370 451 470 420 471 550 509 487 281 355 351 282 301 407 541 566 713 802 729 958

42 39 41

46 43 44

42 40 41

53 39 36

54 48 48

53 41 38

47 48 47

14 53 53

39 49 48

55 52 49 56 55 54 53 48 49 56 51

52 58 56 56 57 66 62 62 62 59 62 65 62

56 53 52 59 57 56 55 50 51 58 53

278

DREW KEELING

Table A.6. Year

Big Four Steerage

1906 1907 1908 1909 1910 1911 1912 1913 1914 1863–1914 Totals

(Continued ) All Lines

Cabin

Total

490 529 164 365 370 233 325 447 212

133 159 133 154 167 157 162 179 111

624 688 296 519 537 390 487 626 323

7,679

2,789

10,351

Steerage

Big 4/All Lines

Cabin

Total

Steerage (%)

Cabin (%)

Total (%)

937 1,034 307 762 766 503 714 946 439

217 250 200 251 280 269 279 315 235

1,154 1,284 507 1,013 1,046 772 993 1,261 674

52 51 53 48 48 46 46 47 48

61 64 66 61 60 58 58 57 47

54 54 58 51 51 51 49 50 48

17,233

4,821

22,461

46

57

49

Notes: This table is an update and correction to Keeling (1999b), Table A.3 (pp. 60–61). Blank=not available for that year ‘‘Big Four’’=Cunard, White Star, HAPAG, NDL (not including afﬁliates Anchor, American, Red Star, etc.) 1863–1914 period percentages (Big 4/all lines) computed over years for which data available for Big 4. Between 1870 and 1914, New York was entry port for 73% of all immigrant arrivals to the US including those from non-European countries (ﬁscal year data from Statistical Abstracts of the United States, 1870–1914). Sources: 1863–1880: New York Commissioners of Emigration; 1881–1914: Keeling (1999a), Table 1, except 1892–1894: Fry (1896), 1898: Times (London) (January 26, 1899, p. 4).

U.S. -

regular traffic to four largest ports (96% of all European immigration to the U.S.)

U.S.- all other (mostly via Canada)

Other Brazil

Argentina

Canada

Fig. A.1. Westbound Transatlantic Migrants by Destination, 1900–1914. Note: Total westbound transatlantic migration, 1900–1914, was about 22.5 million. Source: Keeling (2007, Appendix 4).

Transport Capacity and Transatlantic Migration

279

Percentage change versus prior year

40%

20%

0%

1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913

-20%

Migrants

Migrant Traffic Migrant Capacity Air Passenger Traffic

-40%

Airline Passengers

Air Passenger Capacity

Migrants

-60% 1992

1997

2001

2004

Airline Passengers

Fig. A.2. Fluctuations in North Atlantic Trafﬁc and Capacity (Migrants 1901–1913 vs. Airline Passengers 1992–2004). Note: The pattern of ﬂows here would look quite similar if charted as ‘‘deviation from trend’’ instead of year-to-year percentage change. Cyclical variations in the GNP of the U.S. were greater during 1901–1913 than during 1992–2004. This, however, explains only a small portion of the difference in the variability of the trafﬁc volumes shown in this graph. Migrant and air passenger volumes also ﬂuctuated seasonally (not shown here) to a very similar degree, though with different patterns (Spring peak for migrants, Summer peak for air passengers). Turn of the twentieth century passenger steamship lines share many characteristics with the modernday passenger airlines, which were their successors on the North Atlantic (see Boyce (2001) for an insightful overview of the strategic decision-making behind the transition from sea to air travel). These transport industries were, for example, similarly concentrated. The 10 largest suppliers of budget class passenger steamship capacity between the major ports of Europe and the U.S. during 1900–1913 accounted for 81% of that capacity. In 2002, the 10 largest air carriers between the U.S. and major European destinations provided 88% of the seats ﬂown (based on Voyage Database; Harrison, 2003, p. 7). For further background, see Hanlon (1996, pp. 61–62, 201–211) and Economist (2003). The share of arriving New York passengers brought there by the four largest steamship lines is shown in Table A.7. Sources: Voyage Database, Lufthansa Annual Reports (1995–2004), and British Airways Trafﬁc Reports (1991–2004).

280

DREW KEELING 15%

4% Passengers in excess of capacity (left hand scale) ADJUSTED Passengers in excess of capacity (left hand scale) ALL Passengers on voyages with ADJUSTED excess of capacity (right hand scale)

3% 10%

2%

5% 1%

0%

0% 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912

Fig. A.3. Passengers in Excess of Capacity by Year (Westbound Passengers on Voyages where Passengers Exceeded Capacity, as a Percentage of all Westbound Passengers, 1900–1913). Notes: ‘‘Passengers in excess of capacity’’ is measured separately for ﬁrst, second, and third class. Thus, if on a voyage of 1,900 passengers, 1,600 were third class while capacity in third class was only 1,400, and ﬁrst and second class were not ﬁlled beyond capacity, and if that was the only voyage that quarter with such an excess, while total passengers on all voyages that quarter were 10,000, then ‘‘Passengers in excess of capacity’’ would be 2% (=(1,6001,400)/ 10,000). ‘‘ADJUSTED passengers in excess of capacity’’ allows for 20 passengers or 10% (whichever is greater) beyond capacity in any ‘‘ﬁlled-up’’ class AND for ‘‘upwards’’ transfers of passengers (2nd class passengers exceeding capacity given available berths in 1st, or 3rd class in 2nd) before any excess is measured. If, using the same example as above, that voyage had carried 180 passengers in second class but had 200 second class berths, then most of the 200=(1,6001,400) ‘‘overbooked’’ third class passengers would be assumed to have been accommodated without an ‘‘adjusted excess.’’ 140 could be taken in third (10% of 1,400) and 20 ‘‘uptransferred’’ to second class (200180=20). That would leave 40 (20014020) still excess of ‘‘adjusted capacity,’’ thus ‘‘ADJUSTED passengers in excess of capacity’’ would be 0.4%=(40/10,000). ‘‘All passengers on voyages with ADJUSTED excess of capacity,’’ meanwhile, would be 19% (=1,900 total on this voyage, it being the only one with an adjusted excess, divided by 10,000 on all voyages). Source: Voyage Database.

3%

15%

2%

10%

1%

5%

0%

0% 1st (Jan-Mar)

2nd (Apr-Jun)

3rd (Jul-Sep)

Passengers in excess of capacity (left hand scale)

4th (Sep-Dec)

ADJUSTED Passengers in excess of capacity (left hand scale)

ALL Passengers on voyages with ADJUSTED excess of capacity (right hand scale)

Fig. A.4. Passengers in Excess of Capacity by Quarter (Westbound Passengers on

1. Capacity Utilization vs 2. "Adjusted" passengers, on voyages where capacity was exceeded, vs 3. Probably housed in closed berth third class, on those same voyages, (trend lines 2. & 3. are shown as a % of all passengers on all voyages) 10%

100%

9% 8%

80%

7% 6%

60%

5% 4%

40%

3% 2%

20%

Passengers / Capacity

"Excess" Passengers, Housed in closed third as % ofAll Passengers on all voyages

Voyages where Passengers Exceeded Capacity, as a Percentage of all Westbound Passengers, by Calendar Quarter, 1900–1913). Definitions: See notes to Fig. A.3. Source: Voyage Database.

1% 0%

0%

1900

1901

C A L E N D A R Q U A R T E R S 1902 1903 1904 1905 1906 1907 1908

2. Adjusted Passengers in Excess of Capacity ( lefthand scale ) 3. of which probably housed in III class closed berths ( lefthand scale ) 1. Capacity Utilization ( righthand scale )

Fig. A.5.

1909

1910

1911

Definition of "Adjusted passengers in excess of capacity": see TableA.3. "probably housed" - based on amount of unused closed berths in 3rd class on the voyages carrying second class passengers in "adjusted excess"

Capacity Utilization and Closed Berths (Westbound 2nd Class from Europe to New York, Boston, Philadelphia, Baltimore by Calendar Quarter, 1900–1913). Source: Voyage Database.

282

2,000,000

Migrant crossings per year

1,750,000 1,500,000

'REPEAT' - east [30%] 'REPEAT' - west [15%] 'DEPENDENT' (west only) [20%] 'INDEPENDENT' (west only) [35%] "REPEAT"- east

1,250,000 1,000,000

"REPEAT" - west

750,000 "DEPENDENT" (west only) 500,000 250,000 0 1900

"INDEPENDENT" (west only) 1901

1902

1903

1904

1905

1906

1907

1908

1909

1910

1911

1912

1913

DREW KEELING

Fig. A.6. Migrant Crossings by Type, Calendar Years, 1900–1913 (Passenger Movements between Europe and New York, Boston, Baltimore, and Philadelphia). Definitions and sources: A repeat migrant crossing was any crossing other than the ﬁrst crossing by that migrant. Repeat crossings were both west- or eastbound, but ﬁrst crossings were always westbound only (U.S.-born migrants to Europe were a negligible quantity in this period). First crossings are divided here into dependent (those in which the relocation decision was essentially made by someone else, usually a sponsoring or accompanying close relative) and independent (all other migrant crossings). The annual levels of ‘‘dependent migrants’’ are estimated based on Keeling (2007, Appendix 11). Annual ‘‘repeat’’ migrant ﬂows, and total migrant ﬂows are calculated in Keeling (2006, Appendix 1). ‘‘Independent’’ migrant ﬂows are calculated here as total less dependent less repeat (west and east). The legend box above shows the approximate percentage breakdown between these four categories of crossings for the period as a whole.

2400

1800

600

East 400

1200

Repeat - West 1st Time - West

600

200

Annual Flows ('000s), 1904-13

Annual Flows ('000s), 1870-82

Annual average migrant flows between Europe and USA (by direction, type, and business cycle phase, in '000s), 1870-82 versus 1904-13

0 Expansion 1870-73

Contraction 1874-78

Expansion 1879-82

Expansion 1904-07

Contraction 1908

Expansion 1909-13

283

Fig. A.7. Migrant Passenger Crossings and Business Cycles. Sources for fiscal years 1870–1882: East ﬂows: steerage departures (U.S. Bureau of Statistics, 1889), total west ﬂows: ‘‘Immigrants’’ from Europe (Historical Statistics of the U.S., 1975), business cycle phase for ﬁscal years 1870–1882 per Walton and Rockoff (1994, p. 400). The breakdown of westward ﬂows into repeat-west and ﬁrst time-west ﬂows is estimated here as follows: Total repeat ﬂow during 1870–1882 estimated at 50% of eastward ﬂow (during 1900–1914, it was 46% of the eastward ﬂow), distribution of that total over the three time periods shown here is set to be equal to the same relative proportions for same phases of the business cycle for 1904–1913. As shown here, ‘‘ﬁrst time’’ for 1904–1913 includes both ‘‘dependent’’ and ‘‘independent’’ (as deﬁned in Figure A.6). Sources for calendar years 1904–1913: Flows and business cycle phases per Keeling (2006, Appendix 4). Significance of these trends: The rise in back and forth migration, through risk-reducing migrant chains, between the 1870s and the early 1900s, had the effect of reducing the cyclical ﬂuctuations for migrant carriers primarily due to the relative growth of countercyclical eastbound movements.

Transport Capacity and Transatlantic Migration

800

GLOBALIZATION AND LABOR MARKET INTEGRATION IN LATE NINETEENTH- AND EARLY TWENTIETH-CENTURY ASIA Gregg Huff and Giovanni Caggiano ABSTRACT This chapter uses new data sets to analyze labor market integration between 1882 and 1936 in an area of Asia stretching from South India to Southeastern China and encompassing the three Southeast Asian countries of Burma, Malaya, and Thailand. We find that by the late nineteenth century, globalization, of which a principal feature was the mass migration of Indians and Chinese to Southeast Asia, gave rise to both an integrated Asian labor market and a period of real wage convergence. Integration did not, however, extend beyond Asia to include core industrial countries. Asian and core areas, in contrast to globally integrated commodity markets, showed divergent trends in unskilled real wages.

Research in Economic History, Volume 25, 285–347 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(07)25006-2

285

286

GREGG HUFF AND GIOVANNI CAGGIANO

1. INTRODUCTION Beginning in the late nineteenth century, globalization swept through Asia, transforming its product and labor markets. By the 1880s steamships had largely replaced sailing vessels for transport within Asia as well as to Western markets, and shipping fares had begun to fall sharply. Also already underway was the mass migration of Indian and Chinese workers, principally from the labor-abundant areas of Madras in India and the provinces of Kwangtung (Guangdong) and Fukien (Fujian) in Southeastern China, to land-abundant but labor-scarce parts of Asia. Chief among the immigrant-receiving countries were Burma, Malaya, and Thailand (Siam) in Southeast Asia. Indian and Chinese labor inﬂows to these countries constituted the bulk of two of the three main late nineteenth- and early twentieth-century global migration movements, the other being European immigration to the New World. Immigration to Southeast Asia was almost entirely in response to its growing demand for workers which, in turn, derived from rapidly expanding demand in core industrial countries for Southeast Asian exports. Studies by Latham and Neal (1983) and by Brandt (1985, 1989) establish the development of an integrated Asian rice market beginning in the latter part of the nineteenth century (see also Myung, 2000). Furthermore, a series of articles and books by Williamson and his co-authors reveal internationally integrated commodity markets and relative factor price convergence in conjunction with pre-World War II globalization (Williamson, 2000, 2002; O’Rourke & Williamson, 1999; Hatton & Williamson, 2005). But in contrast to work on product market integration, the possible emergence of an integrated Asian labor market has attracted less attention. In part this reﬂects the lack of Asian wage data. As Harley (2000, p. 928) observes, ‘‘analysis of the low-wage periphery, which is most relevant to modern [globalization] debate, is restricted by data availability’’. This chapter makes available for the ﬁrst time the data needed to test for labor market integration over a large part of Asia. The chapter has two main aims. One is to analyze whether as part of pre-World War II globalization an integrated Asian market for unskilled labor existed to encompass Asia’s chief emigrant-sending regions of South India and Southeastern China and the principal Southeast Asian receiving countries for Indian and Chinese immigrants. Our metric for integration, following both econometric work on GDP convergence and Robertson’s recent analysis of integrated labor markets, comprises three complementary criteria: (i) that wages do not diverge from a

Globalization and Integration in Asia

287

common trend; (ii) that over time wage dispersion does not increase; and (iii) that a correction mechanism pushes wages towards an equilibrium relationship after shocks. It can be misleading, as Robertson (2000, p. 728) warns, to rely on price as a criterion for integration. Markets are integrated if adjustment mechanisms operate to correct deviations from a wage differential or ‘‘gap’’. Second, the chapter aims to compare wage trends in the area of Asia from South India to South China and including Burma, Malaya, and Thailand with an industrial core of the global economy, deﬁned as United Kingdom, United States, Germany, and France. Were unskilled labor markets in Asia and the industrial core similarly affected by globalization such that in these two parts of the world wages followed a common trend? Or, in contrast to commodity markets, was globalization in Asia and the industrial core associated with a drifting apart of unskilled real wages? We argue that by the late nineteenth century South India, Southeastern China, and the three Southeast Asian countries had become integrated and constituted a uniﬁed labor market. Furthermore, Asian evidence reveals a period of real wage convergence prior to the 1930s. But labor market integration that characterized Asia, and also obtained in the industrial core, stopped at the geographical frontiers of each of these two regions. Unlike Asia’s export of primary commodities, ﬂows of Asian labor hardly penetrated either the core industrial countries or the wider Atlantic economy. The pre-World War II labor market pattern was, instead, one of strong divergence between Asia and the world’s rapidly developing and industrializing core economies.

2. SOUTHEAST ASIAN GROWTH AND INDIAN AND CHINESE IMMIGRATION There was a fundamental difference between the Southeast Asian worlds of 1860 and of the 1880s. The earlier period pre-dated a global transport and communications revolution and the opening of the Suez Canal. Nor was there as yet the great demand for Southeast Asian primary commodities that soon materialized in the West as part of its rapid industrialization and urbanization (see Huff, 2007). In the 1870s, Malaya was still sparsely populated, largely unmapped and ‘‘land was so abundant and readily available that it had no value’’ (Gullick, 1985, p. 59). Although in Burma after the mid-nineteenth century a growing output of rice was evident, the

288

GREGG HUFF AND GIOVANNI CAGGIANO

big increases in planted acreage and production began only in the 1870s (Cheng, 1968, pp. 237–241). The Thai rice frontier was reminiscent of the United States’ wild west but lay geographically to the south where ‘‘in every direction the land was cleared of the heavy jungle grass which afforded shelter to wild elephants’’ (Johnston, 1981, p. 111). Clearance occurred mainly in the 1890s and 1900s when Thailand’s rice industry ﬁrst boomed. The main export regions in Burma, Malaya, and Thailand were not initially resource-rich areas. They became so because for them the 1880s globalization had altered the deﬁnition of resource abundance. A relevant comparison is North America where, as Harley (1980, p. 218; see also Wright, 1990) points out, globalization transformed a previously ‘‘uneconomic ‘desert’’’ of prairie into a region of rich natural resources. The same was true with the jungles and swamps of Southeast Asia, including almost all of Burma’s best rice land originally regarded as uninhabitable because of the risk of disease or because it was under the sea at high tide. For centuries there was at least some migration from India and China to Southeast Asia and during the eighteenth century migrants began to come in signiﬁcant numbers (Trocki, 1999, pp. 105–106). It might be interesting to compare these migrations and the still small migrant ﬂows of the 1860s with subsequent mass immigration to Southeast Asia. But the absence of data makes meaningful quantitative comparison impossible. Data are non-existent because prior to globalization in Southeast Asia the lack of incentives to migrate limited international immigration to a trickle which no one seems to have thought worth recording. By the mid-1880s Burma and Malaya, including the Straits ports of Singapore and Penang, were effectively under British colonial rule. Thailand, nominally independent, had quasi-colonial arrangements and a British ﬁnancial advisor. From the late nineteenth century onwards, growth in Burma, Malaya, and Thailand stemmed predominantly from an abundance of land. Rapid export expansion depended on the settlement of a moving frontier. For Southeast Asia, international trade provided a ‘‘vent’’ or outlet to utilize surplus land in the production of primary commodities which, unless exported, would not have been worth the effort of producing. Exports from Burma, Malaya, and Thailand, expressed in 1913 US dollars, increased from $104.0 million in 1880/82 to $639.6 million in 1936/38, equivalent to 3.4% annual average growth. Rice was Burma’s and Thailand’s staple export while Malaya’s staple exports were tin and, by World War I, rubber. Vent-for-surplus growth in the three countries required substantial inwards migration. A traditional, or non-export, sector provided part

Globalization and Integration in Asia

289

of the labor to plant previously uncultivated acreage with export crops (Feeny, 1982, pp. 42–43; Adas, 1974, pp. 41–57). Insofar as labor from the traditional sectors of the region’s dual economies was unavailable in sufﬁcient quantities or unwilling to join in export production, immigration from India and China supplied workers. Colonial authorities in Malaya and Burma and the government in Thailand advocated mass immigration to assist trade expansion. Burma, Malaya, and Thailand, all of which, apart from a few brief periods, allowed unrestricted migration until the 1930s, were by no means the sole world outlets for emigration from India and China. But they attracted a large and increasing proportion of all emigrants from India and China and were the dominant outlet for both streams of emigration (Table 1). Burma received chieﬂy Indian immigrants and Thailand mainly Chinese. Malaya, about equidistant between China and India, was the destination for large numbers of both Chinese and Indians. By the 1880s Madras and the Chinese provinces of Kwangtung and Fukien had long histories of hardship and periodic famine and were clearly excess labor areas (see, for example, India, 1902, pp. 27–32; India, 1923, p. 31; 1932a, p. 61; 1932b, p. 93; Kumar, 1965, pp. 104–105, 144, 161–167; Davis, 1951; Buck, 1937a, pp. 76–77, 125–128; Buck, 1937b). In 1881 comparative populations were 31 million in Madras, 37 million in Kwangtung and Fukien, and 14.3 million in the three Southeast Asian countries. At this time Madras and Kwangtung had population densities of 217 and 255 persons per square mile and Fukien a density of over 300 persons compared to a density of between 25 and 30 in the Southeast Asian countries. From 1881 to 1939 Burma, Malaya, and Thailand received over 15 million Chinese and Indian immigrants, more than the three countries’ 1881 population (Table 2). During this period, Malaya averaged immigrant inﬂows per decade of 826 persons per 1,000 resident population. Its immigration rate was easily the world’s highest and almost ﬁve times the rate for Argentina, which itself exceeded any other New World country. Immigrant inﬂows to Burma and Thailand were on a par with, or above, New World rates. Typically, immigrants to Southeast Asia intended to stay for from three to ﬁve years, and over the six decades from the 1880s to World War II in Southeast Asia immigrant retention (net as a proportion of gross immigration) of under a ﬁfth compares poorly with the United States’ two-thirds (Table 2). But in Southeast Asia new arrivals more than replaced departures and, together with greater natural increase, continuously augmented labor supply. Appendix A and Appendix B provide the entire data sets for annual immigration to and emigration from Southeast Asian countries and the New World.

290

GREGG HUFF AND GIOVANNI CAGGIANO

Table 1.

Burma, Malaya, and Thailand Measures of Indian and Chinese Immigration, 1881–1937.

Panel A: Immigration of Indians to Burma and Malaya and as a percentage of total emigration from India Emigration from India (’000 Persons)

Immigration to Burma (’000 Persons)

Immigration to Burma as a % of Indian Emigration

Immigration to Malaya (’000 Persons)

Immigration to Malaya as a % of Indian Emigration

3,006 4,288 3,292 4,570 6,060 2,755

616.1 1,260.7 2,482.9 3,050.8 3,864.6 2,402.2

20.5 29.4 75.4 66.8 63.8 87.2

159.9 216.0 443.0 908.1 881.2 384.6

5.3 5.0 13.5 19.9 14.5 14.0

1880–1890 1891–1900 1901–1910 1911–1920 1921–1930 1931–1937

Panel B: Distribution of emigrants from China and India, 1930 Chinese

Thailand Malaya Indonesia Indochina All other countries Total

Indians

’000 persons

%

1,900 1,800 1,240 700 4,360 10,000

19.0 18.0 12.4 7.0 43.6 100.0

Burma Ceylon Malaya Mauritius All other countries Total

’000 persons

%

1,300 1,133 628 281 783 4,125

31.5 27.5 15.2 6.8 19.0 100.0

Note: In Panel A for 1931–1937 immigration to Burma and Malaya adds to more than 100% of emigration from India because of different data sources for immigration and emigration. Sources: Panel A: Appendix E and Davis, Population of India, p. 99 for emigration from India. Panel B: Mukerjee (1936, Appendix A).

Indentured labor was never important in any of the three Southeast Asian countries. Indian and Chinese immigrants reached Southeast Asia either through a variety of organized systems which ﬁnanced immigration or through paying their own passage. This latter applied to an increasing, and by the twentieth century, large number of immigrants to Southeast Asia. The predominant picture is of a mobile immigrant workforce and competitive Southeast Asian labor markets. (For discussion of systems of immigration, see Huff & Caggiano, 2007; Madras, 1874, p. 75; India, Census of India, 1912, p. 26; India, Census of India, 1932b, p. 80; India, 1926–1939/ 40; India, Census of India, 1933, pp. 67–72; Sugihara, 2005; Look Lai, 2002; Mckeown, 2004.)

Globalization and Integration in Asia

Table 2.

291

Southeast Asia and New World Immigration, 1881–1939.

Panel A: Immigration to the United States, Burma, Malaya, and Thailand, 1881–1939 (millions of persons, total ﬂow per decade) 1881–1910

United States Burma Malaya Thailand Total Southeast Asia Southeast Asia as % of United States

1911–1929

1930–1939

Gross

Net

Gross

Net

Gross

Net

5.91 1.45 1.87 0.34 3.66 61.9

4.10 0.26

3.20 3.27 2.75 0.81 6.83 213.0

2.15 0.50 0.78 0.27 1.55 72.1

0.70 2.64 1.62 0.50 4.76 680.0

0.21 0.17 0.07 0.12 0.22 104.8

0.12

Panel B: Southeast Asia and New World immigration rates by decade 1881–1890 to 1931–1939 (per 1,000 mean population)

Burma Thailand Malaya United States Canada Argentina Brazil

1881–1890

1891–1900

1901–1910

1911–1920

1921–1930

1931–1939

85.3 22.4 921.9 91.6 193.4 267.4 40.2

138.4 39.6 994.5 52.5 67.1 163.8 69.8

219.7 75.9 993.5 103.8 268.4 292.9 33.2

240.9 74.3 838.9 57.2 216.3 150.1 31.9

277.2 102.1 859.7 35.3 130.4 133.2 27.4

167.8 30.8 346.0 3.6 13.6 39.7 7.3

Source: See Appendix E.

3. EMPIRICAL ANALYSIS In this section we ask two questions. First, as mass intra-Asian migration might suggest, was there in fact an integrated labor market in Asia? Integration requires that, in the absence of government intervention or other political disturbances, wages in Asia converged to some stable, long-term equilibrium relationship. Such a relationship implies the existence of a correction mechanism, not due to common external shocks, that quickly restored equilibrium whenever wages departed from it. Second, if Asia had an integrated labor market, did integration, perhaps as a consequence of trade links, extend to the industrial West? Speciﬁcally, did wages in Asia and in the industrial core of United Kingdom, United States, Germany, and France follow a common trend and signiﬁcantly affect one another so as to form an integrated global labor market? Or did separate labor markets

292

GREGG HUFF AND GIOVANNI CAGGIANO

persist despite an increasingly integrated late nineteenth- and early twentieth-century global economy?

3.1. Wage Data To answer these questions we ﬁrst collect real wage data for South India (Madras), Southeastern China (Kwangtung and Fukien), Southeast Asia, and the four core industrial countries. Asian wages comprise six series because these include, as well as Madras, Southeastern China, Burma, and Thailand, data for both Malayan Indian and Chinese wages. Data are for 1882–1936 – the period for which comparative wage series can be assembled. Asian wage data are chieﬂy, but not exclusively, from government reports and are largely new. All wages are deﬂated by separate price indexes for Madras, Southeastern China, and each of the three Southeast Asian countries to obtain real wages. For Southeast Asia, price indexes go well beyond earlier work because, rather than using a single or at most two goods, they include rice, dried ﬁsh, sugar, tea, beer and ale, kerosene, tobacco, and white and grey shirting. Index weightings are based on contemporary budget surveys (Bennison, 1928, pp. 176–181; Andrew, 1933, pp. 226–250; Malaya, 1922–1938; Creutzberg, 1979, p. 78 (budget devised by Polak); Indonesia, 1958; van Niel, 1956; Runes, 1939, pp. 19, 21). For Madras and Southeastern China we use unskilled male, and predominantly rural, wages, since emigrants from these areas of India and China to Southeast Asia were almost all unskilled, largely men, and mainly from agricultural areas. A substantial proportion of immigrants to Southeast Asia took rural jobs. Even if immigrants stayed in cities, in Southeast Asia’s vent-for-surplus economies the importance of primary production and its labor-intensive character made employment in the staple industries typically the dominant inﬂuence in setting unskilled wages. Until 1910 Chinese wages in Malaya are for tin mining as the chief source of employment and thereafter for work on rubber estates. Indian wages in Malaya are for unskilled, chieﬂy plantation labor until 1910, and then for rubber estate employment. Burma wages for 1880–1901 are for agricultural labor and subsequently for coolie labor, predominantly in rice mills. Thailand is an exception both to the use of rural wages and to a new wage series. Wage data for anywhere in pre-World War II Asia must be treated with caution and information for Thailand is fragmentary, particularly before 1900. We rely on Thai wage data collected by Feeny (1982, pp. 29, 132–133) and Ingram (1964, p. 115).

Globalization and Integration in Asia

293

Wages are for unskilled urban labor and this reﬂects the overwhelming preference of native Thais to remain cultivators and the tendency for Chinese to congregate in cities, mainly Bangkok, and engage in dock, railway, or other institutional work. No adequate basis exists to adjust wages for unemployment and none of the six wage series includes every year. Gaps in series are interpolated by applying the Kalman ﬁlter, which uses known values to give a statistically best prediction of missing observations (Harvey, 1992, pp. 143–147). Core industrial country wages are, like the data for Asian wages, for unskilled, predominantly male, workers. For each of the four core countries nominal are converted to real wages using country-speciﬁc indexes of consumer prices or the cost of living. The ten wage series are presented in Appendix C and fully discussed in terms of sources, reliability, and construction in Appendix E.

3.2. Asian and Industrial Core Labor Market Integration Since our purpose is to investigate whether wages in Asian countries moved together and how their dynamic is related to the industrial core rather than to try to account for migration patterns, we do not adjust real wages for exchange rate ﬂuctuations. Nor is any adjustment for purchasing power parity desirable since we are not attempting to compare cross-country living standards. Two points should, however, be noted. One is that across the world in 1882 unskilled wages measured in current US dollars stood at quite different levels. Wages in Southeast Asia were about three times as high as in Madras and Southeastern China while, in turn, wages in United States and United Kingdom were three or more times those in Southeast Asia (Table 3). German and French wages were, however, only about a third more than in Thailand, the highest-wage Southeast Asian country. Wage gaps of the magnitudes between Southeast Asia on the one hand and Madras and Southeastern China on the other point to an important reason for the mass migration that occurred in Asia. The differentials also suggest that had industrial countries been willing, as were Southeast Asian governments, to allow unlimited entry to Indian and Chinese workers, there might have been very much greater migration from Asia to the global core than in fact occurred. Even in the United States, historically open to immigration, the only real question was whether to restrict European immigration, something America abruptly did in 1921 (Goldin, 1994; Hatton & Williamson, 2005, pp. 148–149).

294

Table 3.

GREGG HUFF AND GIOVANNI CAGGIANO

Southeast Asia, Southeastern China, Madras, and Global Core Monthly Unskilled Wages, 1882 (US$ Current Prices).

Southeast Asia Burma Malaya Indians Malaya Chinese Thailand

Southeastern China and Madras 5.79 6.43 7.07 8.65

Madras Southeastern China

1.80 2.07

Global Core United States United Kingdom Germany France

29.52 22.92 11.80 11.41

Source: See Appendix E.

The other point to bear in mind is the contrasting implications for wage gaps of different possible ﬁndings for Asian and core wage trends. Insofar as the Asian and industrial core groups displayed common trends between 1882 and 1936 within their respective groupings, and trends between the two groupings did not diverge, this would imply an approximate maintenance of 1882 wage differentials up to the World War II. Conversely, if the Asian or the industrial groupings, at the same time as sharing a common trend with others of their own group, trended more rapidly upwards than the other grouping, there would be either wage convergence or wage divergence between these Asian and industrial core components of the world economy. Figs. 1 and 2 plot the log of unskilled real wages between 1882 and 1936 for the six Asian and four core wage series respectively. Visual inspection of the ﬁgures suggests possible convergence within Asia and among the global industrial core. For each of these two groupings, real wage variance, although readily apparent, remains clustered around the time trends drawn for all Asian and for all core wage series. There is, however, a marked divergence between Asian and core trends (Fig. 3). For Asia the trend in real wages remains almost ﬂat with a slight downwards bias. None of the four Southeast Asia series show substantial and sustained wage advance. For both Indians and Chinese in Malaya the trend is near zero. In Burma and Thailand the trend in wages through 1932 is ﬂat. Thereafter in both countries moderate upwards pressure on wage movements reﬂected the end of unconstrained immigration. In Thailand, new 1932 immigrant permit and residence fees together with scope for arbitrary ofﬁcial exclusion discouraged immigration from China. In Burma a series of anti-Indian riots similarly affected labor inﬂows from Madras. Wages in the industrial core, unlike those in Asia, trend markedly upwards. In the twentieth century, Asian and core wage divergence gathered momentum. The unmistakable impression is of separate Asian and core labor market ‘‘clubs’’. But is this picture of a non-divergence of wages, and so

Globalization and Integration in Asia

5.5

295

China

India

Burma

Thailand

Malaya Chinese

Malaya Indians

Common trend

ln real wages

5

4.5

4

3.5

3 1882 1885 1888 1891 1894 1897 1900 1903 1906 1909 1912 1915 1918 1921 1924 1927 1930 1933 1936

Fig. 1.

France Germany UK US Common trend

5.8 5.6 5.4

ln real wages

Asia Unskilled Real Wages, 1882–1936.

5.2 5 4.8 4.6 4.4 4.2 4 1882 1885 1888 1891 1894 1897 1900 1903 1906 1909 1912 1915 1918 1921 1924 1927 1930 1933 1936

Fig. 2.

Industrial Core Unskilled Real Wages, 1882–1936.

of potential market integration, within Asia and within core countries, borne out statistically? Was wage convergence between Asia and the industrial core in fact absent? To try to answer these questions, we now test econometrically. We begin by testing whether the two conditions for labor market integration – non-divergence of wage pairs and non-increasing wage dispersion – are met within Asia and between it and the industrial core. These two conditions are not sufﬁcient to establish labor market integration. But they are necessary for it.

296

GREGG HUFF AND GIOVANNI CAGGIANO 5.4

Asia common trend Core common trend

5.2

ln real wages

5

4.8

4.6

4.4

4.2

4 1882 1885 1888 1891 1894 1897 1900 1903 1906 1909 1912 1915 1918 1921 1924 1927 1930 1933 1936

Fig. 3.

Asia and Industrial Core Unskilled Real Wages, 1882–1936.

The ﬁrst of the tests requires that over the observed time span for a given set of economies real wages should not drift apart. If n labor markets are integrated, the corresponding n real wage series must satisfy the convergence hypothesis: that wage differences behave as stationary series around a constant mean and that differences between real wages in the n countries do not systematically change. To test for non-divergence in wages, we adopt a procedure recently developed by Pesaran (2007). His approach tests whether wage gap pairs are stationary and can be summarized as follows. For N economies, consider all possible N(N1)/2 possible wage gap pairs, d ij;t wi;t wj;t , for i=1, y, N1 and j=1, y, N. Countries i and j form an integrated labor market if wi;t wj;t is a stationary process and therefore contains neither a unit root nor time trend. We ﬁrst test for a unit root in all possible pairs d ij;t using augmented Dickey–Fuller regressions with an intercept and a linear trend: Dd ij;t ¼ aij þ bij ðgi gj Þt þ rij d ij;t1 þ

pij X

dij;s Dd ij;ts þ uij;t

(1)

s¼1

If the null hypothesis of a unit root is rejected, the next step is to test if d ij;t is not trended, that is, whether gi ¼ gj . If real wages converge, or rather do not diverge, the expectation is that the fraction of wage pairs for which a unit

Globalization and Integration in Asia

Table 4.

297

Asia and Industrial Core Proportion of Wage Pairs that do not Satisfy the Null Hypothesis of Convergence.

Group Number of wage series Number of pairs % of no unit root % of signiﬁcant time trend

All

Asian

Core

10 45 91% 44%

6 15 93% 7%

4 6 50% 0%

Source: See Appendix C.

root exists and the fraction of pairs for which there is a signiﬁcant time trend are close to the nominal size of the test. In other words, if countries i=1, y, N form an integrated labor market, and the non-divergence in wages hypothesis is tested at a 95% conﬁdence level, both the unit root and the time trend hypotheses should not be rejected for approximately 5% of all possible pairs wi;t wj;t . We now apply this measure of convergence to real wages series for Madras, Southeastern China, Thailand, Burma, Malaya Chinese, Malaya Indians, United Kingdom, United States, Germany, and France between 1882 and 1936 (Table 4). Estimation of Eq. (1) does not reject the null hypothesis of a unit root for 9%, and of a time trend for 44%, of all possible 45 pairs. This time trend percentage falls far outside the size requirement of 5% for a 95% conﬁdence interval. But when Eq. (1) is estimated for Asia only, the fractions are 7% and 7%, respectively, quite close to the required nominal size of 5%. These results support the claim that unskilled real wages in Asia did not drift apart and point to possible labor market integration, but indicate divergence between Asian and core country wages.1 Labor market integration requires not only comovements in real wages but also that variability between wages must not change systematically over time. Our ﬁrst test, although revealing comovements in Asian wages, does not deal with the issue of wage dispersion. A second test, also due to Pesaran (2007), is for non-divergence in wages and based on an average measure of convergence, the cross-section mean difference of wages: N1 N XX 2 ðwi;t wj;t Þ2 NðN 1Þ i¼1 j¼1 (P 2 ) N ¯t i¼1 wi;t w ¼2 N 1

D2t ¼

(2)

298

GREGG HUFF AND GIOVANNI CAGGIANO

PN where w ¯ t ¼ N 1 j¼1 wj;t . Since D2t is a measure of real wage dispersion, under the convergence hypothesis it must not be trended but stationary around a constant mean. We test the null hypothesis that D2t does not have a unit root and that it is not trended. The possibility of a unit root is rejected in all cases. Next, we examine the presence of a linear deterministic trend in D2t . For Asian and core wages treated as a single sample a trend is apparent: the t-ratio of 9.21 is much larger than the 95% critical value, 1.96. Treating Asia separately, however, gives a t-ratio of 1.63, a result well within the required conﬁdence level. Accordingly, the null hypothesis of trended D2t , and therefore the possibility that real wages in the Asian economies diverged, can be safely rejected. The two above ﬁndings on convergence of real wages in Asian labor markets are further conﬁrmed by the third and last of our three preliminary tests, this one proposed by Evans (1996). It looks at the statistical properties of the cross-country variance of real wages.2 Let wi;t be the logarithm of real wages for country i=1, y, N observed for periods t=1, y, T. The crosscountry variance at time t is given by Vt ¼

N 2 1X wi;t w¯ t N i¼1

(3)

PN with w ¯ t ¼ N 1 i¼1 wi;t . If real wages of the observed N countries converge, then the cross-country variance must be a stationary series. In other words, it must neither contain a unit root nor a time trend. To test this hypothesis, we estimate DV t ¼ a þ Zt þ rV t1 þ

p X

fi DV ti þ et

(4)

i¼1

^ and tð^ZÞ to test the null hypothesis that r=0 and construct the t-ratios, tðrÞ and Z=0, respectively. Evidence in favor of convergence requires the rejection of r=0 (unit root) but not of Z=0 (no time trend). With a ﬁnite sample such as the Asian and core wage series, critical values may differ substantially from the fractiles of the standard normal distribution. To address this possibility, we estimate the critical value, c^0:05 , for a test of size 0.05 using Monte Carlo simulations (Tables 5 and 6; for further details on these Monte Carlo techniques, see Evans, 1996, pp. 1033–1034). When Asian and core wage series are considered together, we can reject the null ^ ¼ 3:11o^c0:05 ¼ 2:15 – and also the hypothesis of a unit root – tðrÞ hypothesis that there is no time trend – tð^ZÞ ¼ 1:794^c0:05 ¼ 1:68. But we

Globalization and Integration in Asia

Table 5.

299

Asia and Industrial Core Unit Root Test for Cross-Country Variance.

Group r^ ^ t ðrÞ c^005

All

Asia

Core

0.29 3.11 2.15

0.32 3.22 2.21

0.23 2.59 2.12

Source: See Appendix C.

Table 6.

Asia and Industrial Core Time Trend Signiﬁcance for Cross-Country Variance.

Group Z^ t ð^ZÞ c^0:05

All

Asia

Core

0.002 1.79 1.68

0.000085 0.14 1.55

0.0013 1.97 1.52

Source: See Appendix C.

ﬁnd the reverse, and so in favor of convergence, when calculating Eq. (3) ^ ¼ for wages in Asia only: evidence exists against a unit root in V t – tðrÞ 3:22o^c0:05 ¼ 2:21 – and also against the presence of a time trend – tð^ZÞ ¼ 0:14o^c0:05 ¼ 1:55. The ﬁndings for this last test conﬁrm and strengthen the econometric results of the ﬁrst two tests. To summarize, we ﬁnd that Asian wage behavior was consistent with an integrated labor market; and that between the 1880s and World War II real wages in Asia diverged from those in the industrial core. Although late nineteenth- and early twentieth-century globalization gave rise to world commodity and capital markets, it did not have the same effect as between Asian and core real wages. Rather, market segmentation prevailed. Furthermore, within this framework of separation, Asia and the core each displayed characteristics of a club in which members signiﬁcantly inﬂuenced one another and moved in like direction.

3.3. Asian Labor Market Integration, Terms of Trade Shocks, and Wage Gaps The common trend followed by Asian labor markets suggests integration but does not establish it. Market integration requires the existence of a

300

GREGG HUFF AND GIOVANNI CAGGIANO

correction mechanism. Furthermore, even if wages between regions or countries are continuously pushed towards an equilibrium relationship, this may not be due to labor market forces. Hatton and Williamson (2005, p. 145) raise a similar issue in assessing for the late nineteenth century whether to attribute wage–rent convergence in Asia mainly to migration or to trade. They acknowledge: ‘‘we simply do not know whether migration or the terms of trade mattered most in the convergence, but our best guess would be the terms of trade’’. In light of this and the highly globalized post1880s world of which Madras, Southeastern China, and Southeast Asia became part, two questions arise. First, could the mechanism which made Asia seem an integrated labor market have been merely a response to common external shocks operating through the terms of trade? Second, if integration was effected through labor markets rather than shared shocks transmitted from the world’s industrial core, did real wage convergence in Asia occur? The present section attempts to answer these questions. To deal with the ﬁrst question, we sketch a simple labor market model for the Asian periphery and explicitly include the terms of trade as an external shock. The required terms of trade series did not, however, exist for three of the Asian regions and were only partially available for the other two. As a ﬁrst step, we therefore constructed new net barter terms of trade series for 1882 onwards for all ﬁve of the Asian regions or countries. The series are location speciﬁc to Madras and the Southeastern China provinces of Kwangtung and Fukien. Series all take account of the several major exports of each region or country and are weighted to reﬂect shifting export composition. All series use country-speciﬁc imports rather than, as often in previous work, making the same denominator serve for several countries (Blattman, Hwang, & Williamson, 2004, p. 31). The model’s labor market speciﬁcation, adapted from Robertson (2000, pp. 744–747), focuses on an export-dependent Asian periphery unable to inﬂuence industrial core wages and where labor market equilibrium depends on wages at home, wages in a contiguous country, and external demand shocks. Labor demand in country i (Madras or Southeastern China) responds negatively to changes in the domestic wage level and positively to lagged wages in a contiguous Southeast Asian country j. To capture the effect of external demand shocks, labor demand is assumed to be positively correlated with the terms of trade. Improvement in the terms of trade at time t reﬂects an increase in industrial core demand for exports from country i. Labor demand in Madras and Southeastern China is thus given by: Ldi;t ¼ a0 þ a1 wj;t1 a2 wi;t fwi;t1 þ a3 TOT i;t (5)

Globalization and Integration in Asia

301

where f measures demand responsiveness to changes in domestic wages, and so movements along the labor demand curve, and TOT i;t represents external demand shocks measured by the terms of trade, and hence shifts in the labor demand curve. Since workers’ decisions in Madras and Southeastern China (country i) include the possibility of migrating to some Southeast Asian country j, both the wage level and demand conditions in Southeast Asia enter labor supply: Lsi;t ¼ b0 b1 wj;t þ b2 wi;t jwi;t1 b3 TOT j;t (6) where, as before, f represents movements along the labor supply curve and TOT j;t shifts in it caused by external demand shocks. The coefﬁcients a1 and b1 account for the expenses of transport and ﬁnance, compensation for the psychic costs of migration, and a higher recipient country wage to enable emigrants to remit home. These migration-related costs, discussed below and well known in the literature to create a wage gap, prevented wage equalization between Madras and Southeastern China on the one hand and Southeast Asia on the other (see Williamson, 1988, pp. 433–435 for an overview of the concept of wage gaps). Accordingly, international labor market equilibrium is deﬁned as convergence in the marginal product of labor in country i towards the marginal product of labor in the Southeast Asian country j plus a wage differential. Equating labor demand and supply gives the equilibrium condition: a0 þ a1 wj;t1 a2 wi;t fwi;t1 þ a3 TOT i;t ¼ b0 b1 wj;t þ b2 wi;t jwi;t1 þ b3 TOT j;t ð7Þ Solving for the Southeast Asian wage,wj;t : wj;t ¼

b0 a0 b2 þ a2 þ wi;t b1 b1 b j þ a2 f a1 2 wi;t1 wj;t1 b1 b1 b a3 þ 3 TOT j;t TOT i;t b1 b1

ð8Þ

Simplifying notation gives: wj;t ¼ d0 þ d1 wj;t1 þ g1 wi;t þ g2 wi;t1 þ l1 TOT i;t l2 TOT j;t

(9)

302

GREGG HUFF AND GIOVANNI CAGGIANO

Subtracting wj;t1 from both sides and assuming long-run homogeneity between wi;t and wj;t (which implies that d1 þ g1 þ g2 ¼ 1)3 gives: Dwj;t ¼ d0 þ g1 Dwi;t þ s1 wj wi t1 þ l1 TOT j;t l2 TOT i;t (10) Eq. (10) provides an empirical model to test for labor market integration. We adopt a two-stage testing procedure. Initially, both countries are assumed to be unaffected by external shocks: l1 ¼ l2 ¼ 0. Labor market integration requires that wages in country i and in country j respond to the same shock, which implies that g1 must be positive and signiﬁcant, and, furthermore, that an error correction mechanism operates such that wages revert to their long-run equilibrium, that is, s1 o0. We pool data for pairs of migrant sending and receiving regions: Madras and Burma, Southeastern China and Malaya Chinese, Madras and Malaya Indians, and Southeastern China and Thailand. Although data are differenced, the regression speciﬁcation includes ﬁxed effects. Their signiﬁcance is conﬁrmed by Lagrange multiplier tests for redundant ﬁxed effects. Results are summarized in Table 7.4 It shows that wages in receiving countries – Thailand, Burma, and Malaya – and wages in sending regions – Madras and Southeastern China – respond to the same shock. The estimated elasticity is 0.26 and signiﬁcant at the 5% level (standard errors are robust to heteroskedasticity). There is a strong reversion to the equilibrium wage gap: the error correction coefﬁcient is 0.32 and signiﬁcant at any level. Following Boyer and Hatton (1994, p. 96), we estimate the speed of convergence as ð1 s^ 1 Þ=s^ 1 . From the results in Table 7, the predicted lag between an initial shock and return to equilibrium is about two and a half years. Tests of the hypothesis of different convergence speeds suggest a slightly positive difference and that Chinese migration pairs converge faster to equilibrium than Indian. But these results fall short of statistical signiﬁcance.

Table 7. Wage Relationships between Southeast Asian Receiving Countries and Madras and Southeastern China Sending Regions, 1882–1936. Estimated Coefﬁcient

Std. Error

t-Statistic

p-Value

0.264696 0.300490

0.108203 0.080588

2.446291 3.728727

0.0153 0.0002

Wage shock Error correction Source: See Appendix C.

Globalization and Integration in Asia

303

The ﬁrst stage of testing meets the criteria of our metric of labor market integration. But it does not rule out the possibility that like comovements in Asian labor markets arose from shared terms of trade shocks.5 Recalling the Asian periphery’s high dependence on industrial core demand, it is possible that correlation and reversion to the equilibrium wage gap resulted from exogenous demand shocks manifested through the terms of trade. If the terms of trade were the determining consideration, the existence of the error correction mechanism revealed by ﬁrst-stage testing would be driven by an omitted variable bias attributable to unaccounted-for trade-related shocks. To investigate this possibly, we ﬁrst construct for 1882–1936 terms of trade series for the receiving countries of Thailand, Burma, and Malaya and for the sending regions of Madras and Southeastern China (Appendix D). The hypothesis that comovements in wages arose not because of genuine labor market integration but were due to common external terms of trade shocks implies that l1 and l2 are both signiﬁcant and that l1=l2. To test whether these conditions are satisﬁed, we relax the assumption that l1=l2=0 and re-estimate Eq. (10) (Table 8). Inclusion of terms of trade shocks leaves the previous results virtually unchanged: wages respond symmetrically to terms of trade movements, since the hypothesis that l1=l2 cannot be rejected. The terms of trade enter the estimation equation insigniﬁcantly at the 5% level. Moreover, g^ 1 and s^ 1 remain strongly signiﬁcant and do not differ in magnitude after the second-stage inclusion of the terms of trade. There is no incorrect impression of integration because of shared terms of trade shocks. Asian labor markets were genuinely integrated. The second question of real wage convergence is not the persistence of wage gaps between the sending areas of Madras and Southeastern China on the one hand and Southeast Asian receiving countries on the other. In most settings the norm is a continuance of (often substantially) higher Table 8. Wage and Terms of Trade Relationships between Southeast Asian Receiving Countries and Madras and Southeastern China Sending Regions, 1882–1936.

Wage shock Error correction ToT receiving ToT sending

Estimated Coefﬁcient

Std. Error

t-Statistic

p-Value

0.267342 0.319264 0.064928 0.109927

0.105422 0.076260 0.059301 0.063064

2.535923 4.186515 1.094874 1.743120

0.0120 0.0000 0.2749 0.0828

Source: See Appendix C and Appendix D.

304

GREGG HUFF AND GIOVANNI CAGGIANO

Table 9. Madras, Southeastern China and Southeast Asia Comparative US$ Wages at 1913 Prices, 1880–1882 to 1931–1933.

1880–1882 1900–1902 1911–1913 1925–1927 1931–1933

Madras

Southeastern China

Burma

Malaya Indians

Malaya Chinese

Thailand

3.68 2.02 2.37 2.44d 2.55

3.77a – 2.29 3.15 1.10

9.01 5.66 4.87b 5.79 8.43

– 6.19 5.81 5.66 4.84

10.91 9.43 9.88 6.12d 3.34

– 5.92 7.10c 7.09 8.55

Source: See Appendix E. a Refers to 1882/83. b 1911 and 1913 only. c 1912 only. d 1926 only.

real wages in receiving than sending areas. After all, the elimination of wage gaps would negate the main incentive to migrate. The issue is, rather, whether in Asia during the pre-World War II decades the forces of labor market integration were sufﬁciently strong to reduce real wage differentials. To confront this question of whether Asia moved towards real wage convergence, we compare between the 1880s and 1930s the six series of Asian wages expressed in 1913 US$. Where possible, wages are averaged over three years (Table 9). Two patterns are evident in the table. One is a narrowing of the Asian real wage gap by the latter 1920s; the other, divergence in the 1930s. Real wages in Madras/Southeastern China remained at about a third of the level in Southeast Asia until World War I but by 1925–1927 rose to 45.3% of destination wages. Convergence in the 1920s is observable for all sending and destination pairs and occurred mainly through a reduction in Southeast Asian wages towards sending area levels. The initial wage ratio between receiving and sending areas of about three is closely comparable to that suggested as likely in 1873–1883 for Thailand and China (Williamson, 2000, Table 1.1; Hatton & Williamson, 2005, p. 137). Our ﬁnding of real wage convergence also shows some similarity to the identiﬁcation of pre-World War II relative factor price convergence both for Asia and for Atlantic economies, although the timing differs in beginning after 1913 rather than ending there (O’Rourke & Williamson, 1999; Williamson, 2002; Hatton & Williamson, 2005). The only real constraints on mass immigration within Asia since the onset of globalization probably explain the 1930s real wage divergence

Globalization and Integration in Asia

305

apparent in Table 9. Impediments in Thailand and Burma to immigration have already been noted. In Malaya, the August 1930 imposition of quotas drastically limited immigration from China and helped to avoid even larger falls in Malayan wages. At the same time, the complete collapse of Kwangtung’s silk industry decimated that province’s economy and led to severe social dislocation. Silk production had been at the centre of economic life in Kwangtung and in 1925 80% of the banks in Canton (Guangzhou) were said to be ﬁnanced by Shunde silk capital (Howard & Buswell, 1925, p. 16). During the early 1930s most Kwangtung mulberry plantations were abandoned; three quarters of silk ﬁlatures had closed by 1934 and some 200,000 silk-reelers lost their jobs. A two-thirds fall in silk output and low prices left the value of Kwangtung’s silk exports below their level in 1875 (Lin, 1997, p. 86). Many of those who had worked in the silk industry tried to emigrate, including women who began to come to Malaya in large numbers for the ﬁrst time (Blythe, 1947; Purcell, 1967). Other Kwangtung women formerly engaged in silk production and remaining in the province – both the tzu-shu nu¨ (zishu nu¨ ) who had taken celibacy vows and the pu lo-chi (bu luojia) who were separated from their husband but were expected to support him, his concubine, and their children as well as her in-laws – now sought refuge in local spinsters’ houses and vegetarian halls (Topley, 1975, pp. 82–86). Pre-World War II ratios of Southeast Asia to Madras/Southeastern China wages of between about two to a little over three are comparable in size to gaps elsewhere. O’Rourke and Williamson (1999, p. 127) report, for example, that between the 1870s and 1910–1913 Italian real wages rose from 38% to 46% of wages in France, Germany, the United States, and Argentina. Even at the end of the Atlantic economy’s transition to mass migration the ratio of wages in labor-scarce regions abroad to those in Europe ranged from 1.7 in Britain to 3.7 for Norway (Hatton & Williamson, 2005, p. 136). Four main explanations account for the Southeast Asia and Madras/ Southeastern China wage gap. Of these, shipping fares between sending and destination areas are almost certainly the least important. It is not far from Madras or Southeastern China to Southeast Asia and shipping passage was not expensive. Immigrant fares averaged, apart from the 1930s when shipping companies dramatically raised rates to try to make up for lost business due to immigration restrictions, between a half and three weeks’ wages in Southeast Asia. Over a typical immigrant sojourn of four years in Southeast Asia return shipping fares worked out to about 0.5–3.0% of expected immigrant earnings (Huff & Caggiano, 2007, p. 46).

306

GREGG HUFF AND GIOVANNI CAGGIANO

A second consideration, compensation for emigrants’ psychic costs of re-locating in Southeast Asia, while probably signiﬁcant, is not easily quantiﬁable. These costs are likely to have been greater for Indians than Chinese. Indians were, as often observed, not always easily prised away from home. But emigration was a way of life for many in Southeastern China. Some Chinese, like a Maritime Customs Report’s assessment of the inhabitants of the area around Swatow in Kwangtung, could even be described as ‘‘of a roving disposition, not averse to leaving their homes for foreign parts in quest of fortune’’ (China, Imperial Maritime Customs, 1902–11, vol. II, Southern and Frontier Reports, p. 130). Third, emigrants had to meet the relocation expenses additional to shipping fares of moving to Southeast Asia and allow for a margin to cover subsistence costs while looking for work. For Chinese a system of lodging houses developed to ﬁnance both these expenditures and shipping fares as well as serving as labor exchanges for newly arrived immigrants (Huff, 1994, pp. 155–157; Sugihara, 2005). Costs for Indians going to Burma were probably less than for Chinese emigrating to Malaya and Thailand because of Burma’s comparative nearness and because of a maistry (experienced Indian worker acting as a labor recruiter) system such that Indians often traveled as work gangs with others from their home settlement or nearby villages. The fourth, and by all accounts the most important single component of wage gaps, was the almost universal stipulation among immigrants of highenough earnings in Southeast Asia to permit both substantial savings and remittances home. The share of immigrant earnings remitted can be no more than estimated but a likely ﬁgure is 30% (Huff & Caggiano, 2007, p. 44). Throughout the emigrant areas of Southeastern China, whole villages relied on remittances from abroad and otherwise were not economically viable. Often, the only future for young, aspiring Chinese and for their relations remaining in China appeared to be in Southeast Asia (Chen, 1939, pp. 60–72; Freedman, 1957, pp. 16–17). The demographic behavior of parents in China may not have included having ‘‘surplus’’ children to be able to ‘‘vent’’ them as emigrants. But it is not too much to say that emigrating children served as an important form of social insurance.

4. CONCLUDING REMARKS In recent years Williamson has often emphasized the urgency of ‘‘W. A. Lewis’s grand Third World research agenda’’. It encompasses an analysis of

Globalization and Integration in Asia

307

big questions: the economic forces of globalization that fundamentally reshaped the world economic order between the 1870s and 1930s and, especially, how change affected the global periphery of Latin America, Africa, and Asia (Williamson, 2002, p. 82; 2000, pp. 14–15, 40–42; 2006, p. 37; Hadass & Williamson, 2003, p. 635). Despite this highlighting of large issues and desirability of truly global economic history, Lewis’s agenda has attracted relatively few economic historians or economists and much remains to be done. This chapter has provided a signiﬁcant chunk of the data needed to address the Lewisian agenda for that part of Asia extending from South India to Southeastern China and including much of Southeast Asia. We demonstrate that as between these parts of Asia and the world’s leading industrial countries real wage divergence obtained between the 1880s and 1936. The ﬁnding conﬁrms for these areas a conclusion that Williamson (2006, p. 61) reaches for the world as a whole over a similar historical period: that there was powerful absolute factor price divergence between core and periphery. The forces of industrialization and technical change that before World War II transformed the United States and European core reverberated in Asia principally through a demand for primary commodities and technology embodied in manufactured goods that Asian countries imported in return. Between the 1880s and 1930s, India and China were, by any reckoning, areas with large labor surpluses. But labor from these parts of the economic periphery was effectively prevented from emigrating to the global industrial core. Workers in the core were, as Lewis (1978a, p. 192; 1978b, pp. 19–20) argued, fully aware that mass immigration from India or China would greatly drive down wages. Instead, Indians and Chinese migrated en masse to Southeast Asia. There is no reason to suppose that a single answer exists for whether, in the global periphery, labor markets were genuinely integrated or if apparent integration might be merely an artefact of like comovements in response to exogenously determined terms of trade. This chapter has shown that genuine labor market integration existed for South India, Southeastern China, and Southeast Asia. Lewis’s grand Third World research agenda is, as the phrase suggests, large. This chapter has addressed only part of the agenda and not the famous Lewisian hypothesis of immigrant-augmented elastic labor supplies. As shown elsewhere, however, long-term unskilled real wages in the Southeast Asian countries considered in this chapter bear out the elastic labor argument of Lewis (Huff & Caggiano, 2007). The implications for Southeast Asian economic development of both of an integrated Asian labor market and immigrant-augmented elastic labor are considerable.

308

GREGG HUFF AND GIOVANNI CAGGIANO

As more economists and economic historians are drawn to the comparative analytical study of Asia, these ﬁndings will form central building blocks in helping to answer the questions that Lewis posed.

NOTES 1. Note that a result of 7% is in effect 5% because of the relatively small sample size. We test six series and so 15 pairs. If one pair is rejected the fraction is 1/15 which is approximately 7%. 2. Here we again test all ten series, in effect treating the Malaya Indian and Malaya Chinese series as two separate countries. 3. On this point, see Hendry and Ericsson (1991, p. 21). 4. The model speciﬁcation in Eq. (10) may imply that wages in sending and receiving countries are cointegrated. However, cointegration implies that wages are integrated of order one. We test for a unit root in wages series by using the Im, Pesaran, and Shin (2003) test and reject the hypothesis at any signiﬁcance level. 5. Suspicion of the likelihood of this possibility is, however, aroused by the ﬁnding that in the periphery between 1870 and 1913 the terms of trade rose everywhere except in land-scarce East Asia, that is to say areas like Madras, Kwangtung, and Fukien. For this analysis of the terms of trade, see Hadass and Williamson (2001, p. 18). The same observation is omitted from the published version (2003, p. 639) of this working paper.

ACKNOWLEDGMENTS We owe a debt of gratitude to Campbell Leith, Ramon Myers, Ulrich Woitek, anonymous referees and the Editor all of whose many suggestions greatly improved the chapter. Thanks go to George Grantham, Debin Ma and Atchi Reddy who gave advice and provided data. Huff gratefully acknowledges grants from the Carnegie Trust, Scotland, the British Academy, the East Asia National Resource Center, Stanford University, and the Royal Economic Society which helped to ﬁnance data collection. A Leverhulme Research Fellowship provided the time and space which allowed the chapter to be written and Huff acknowledges with thanks this generous help.

REFERENCES Adas, M. (1974). The Burma delta. Madison: University of Wisconsin Press. Andrew, E. J. L. (1933). Indian labour in Rangoon. London: Oxford University Press.

Globalization and Integration in Asia

309

Andrus, J. R. (1948). Burmese economic life. Stanford: Stanford University Press. Baker, C. J. (1984). An Indian rural economy 1880–1955. Oxford: Clarendon Press. Bauer, P. T. (1948). The rubber industry. London: Longmans, Green. Baxter, J. (1941). Report on Indian immigration. Rangoon: Superintendent Government Printing and Stationary. Becher, H. M. (1892–93). Mining in the Malay Peninsula. Transactions of the Institution of Mining and Metallurgy, 1, 78–102. Bennison, J. J. (1928). Report of an enquiry into the standard and cost of living of the working classes in Rangoon. Rangoon: Superintendent Government Printing and Stationary. Bhatia, B. M. (1969). Terms of trade and economic development. Indian Economic Journal, 16, 424–426. Blattman, C., Hwang, J., & Williamson, J. G. (2004). The impact of the terms of trade on economic development in the periphery, 1870–1939: Volatility and secular change. NBER Working Paper 10600. National Bureau of Economic Research, Cambridge. Blythe, W. L. (1938). Methods and conditions of employment of Chinese labour in the Federated Malay States. Kuala Lumpur: Government Printer. Blythe, W. L. (1947). Historical sketch of Chinese labour in Malaya. Journal of the Malayan Branch of the Royal Asiatic Society, 20(1), 64–114. Boyer, G. R., & Hatton, T. J. (1994). Regional labour market integration in England and Wales, 1850–1913. In: G. Grantham & M. McKinnon (Eds), Labour market evolution: The economic history of market integration, wage flexibility and the employment relation (pp. 84–106). London: Routledge. Brandt, L. (1985). Chinese agricultural and the international economy, 1870s–1930s: A reassessment. Explorations in Economic History, 22(2), 168–193. Brandt, L. (1989). Commercialization and agricultural development: Central and Eastern China 1870–1937. Cambridge: Cambridge University Press. Brazil. (1960). O Brasil em numeros apendice do anuario estatistico do Brasil – 1960. Rio de Janeiro: IBGE-Conselho Nacional de Estatistica. Bry, G. (1960). Wages in Germany, 1871–1945. Princeton: Princeton University Press. Buck, J. L. (1937a). Land utilization in China. London: Humphrey Milford. Buck, J. L. (1937b). Land utilization in China, statistics. London: Humphrey Milford. Burma. (1868/69–1935/36). Report on the administration of Burma, annual series, 1868–691935–36. Rangoon: Ofﬁce of the Superintendent, Government Printing and Stationery. Burma. (1897–1939). Report on the working of the Indian Factories Act in Burma, annual series, 1897–1939. Rangoon: Ofﬁce of the Superintendent, Government Printing and Stationery. Burma. (1912–1913, 1922–1923). Office of the Chief Collector of Customs, Annual statement of the sea-borne trade and navigation of Burma, annual series, 1912–1913, 1922–1923. Rangoon: Ofﬁce of the Superintendent, Government Printing and Stationery. Burma. (1917). Quinquennial wages census in urban and rural areas in Burma, 1916. Rangoon: Superintendent, Government Printing and Stationery. Carter, S. B., et al. (2006). Historical statistics of the United States: Governance and international relations (Vol. 5, Part E). New York: Cambridge University Press. Chang, K.-N. (1958). The inflationary spiral: The experience of China, 1939–1950. London: MIT Press. Chang, L. L. (1932). Farm prices in Wuchin, Kiangsu, China. Chinese Economic Journal, 10, 449–512.

310

GREGG HUFF AND GIOVANNI CAGGIANO

Chang, L. L. (1933). Farm prices in Wuchin, Kiangsu (in Chinese). Nanking University College of Agriculture and Forestry Bulletin, 8 (new series), 1–64. Chen, T. (1923). Chinese migrations, with special reference to labor conditions. Washington, DC: US Department of Labor. Chen, T. (1939). Emigrant communities in South China. London: Oxford University Press. Cheng, S.-H. (1968). The rice industry of Burma, 1852–1940. Kuala Lumpur: University of Malaya Press. Cheng, Y.-K. (1956). Foreign trade and industrial development of China. Washington, DC: University Press of Washington. China, Imperial Maritime Customs. (1879–1939). Returns of trade and trade reports (variously titled – annual series). Shanghai. China, Imperial Maritime Customs. (1882–91 to 1922–31). Decennial reports on the trade, industries, etc of the ports open to foreign commerce (variously titled – decennial series). Shanghai. Chou, S.-H. (1963). The Chinese inflation 1937–1949. New York: Columbia University Press. Commodity Research Bureau. (1939). Commodity year book 1939. New York: Commodity Research Bureau. Creutzberg, P. (1979). Changing economy of Indonesia: National income (Vol. 5). The Hague: Martinus Nijhoff. David, P. A., & Solar, P. (1977). A bicentenary contribution to the history of the cost of living in America. Research in Economic History, 2, 1–80. Davis, K. (1951). The population of India and Pakistan. Princeton: Princeton University Press. Diaz Alejandro, C. F. (1970). Essays on the economic history of the Argentine Republic. New Haven: Yale University Press. Dodge, N. N. (1980). Population estimates for the Malay Peninsula in the nineteenth century, with special reference to the East Coast States. Population Studies, 34(3), 437–475. Doyle, P. (1879). Tin mining in Larut. London: E. & F. N. Spon. Drabble, J. H. (1973). Rubber in Malaya, 1876–1922. Kuala Lumpur: Oxford University Press. Drabble, J. H. (1991). Malayan rubber: The inter-war years. London: Macmillan. van der Eng, P. (1993). The silver standard and Asia’s integration into the world economy. Working paper No.175. Department of Economic History, Australian National University, Canberra. Evans, P. (1996). Using cross-country variances to evaluate growth rates. Journal of Economic Dynamics and Control, 20, 1027–1049. Federated Malay States. (1902). Census of population, 1901 by George Thompson Hare. Kuala Lumpur. Feeny, D. (1982). The political economy of productivity: Thai agricultural development 1880–1975. Vancouver: University of British Columbia Press. Feeny, D. (1983). Extensive versus intensive agricultural development: Induced public investment in Southeast Asia, 1900–1940. Journal of Economic History, 43, 687–704. Fenichel, A. H., & Huff, W. G. (1971). The impact of colonialism on Burmese economic development. Montreal: McGill University Centre for Developing Area Studies. Figart, D. M. (1925). The plantation rubber industry in the Middle East. Washington, DC: U.S. Department of Commerce. France. (1887). Ministe`re de l’Agriculture, Statiistique agricole de la France: Re´sultats ge´ne´raux de l’enqueˆte de´cennale de 1882. Nancy: Ministe`re de l’Agriculture.

Globalization and Integration in Asia

311

Freedman, M. (1957). Chinese family and marriage in Singapore. London: HMSO. Freedman, M. (1958). Lineage organization in Southeastern China. London School of Economics Monographs on Social Anthropology No. 18. London: Athlone Press. Gamble, S. D. (1943). Daily wages of unskilled Chinese laborers, 1807–1902. Far Eastern Quarterly, 3(1), 41–73. Goldin, C. (1994). The political economy of immigration restriction in the United States, 1890–1921. In: C. Goldin & G. D. Libecap (Eds), The regulated economy: A historical approach to political economy (pp. 223–257). Chicago: University of Chicago Press. Gullick, J. M. (1985). The entrepreneur in late 19th century Malay peasant society. Journal of the Malaysian Branch of the Royal Asiatic Society, 58, 59–70. Hadass, Y. S., & Williamson, J. G. (2001). Terms of trade shocks and economic performance 1870–1940: Prebisch and Singer revisited. NBER Working Paper 8188. National Bureau of Economic Research, Cambridge. Hadass, Y. S., & Williamson, J. G. (2003). Terms of trade shocks and economic performance 1870–1940: Prebisch and Singer revisited. Economic Development and Cultural Change, 51(3), 629–656. Harley, C. K. (1980). Transportation, the world wheat trade, and the Kuznets cycle, 1850–1913. Explorations in Economic History, 17, 218–250. Harley, C. K. (2000). A review of O’Rourke and Williamson’s Globalization and history: The evolution of a nineteenth century Atlantic economy. Journal of Economic Literature, 38, 926–935. Harvey, A. C. (1992). Forecasting, structural time series models and the Kalman filter. Cambridge: Cambridge University Press. Hatton, T. J., & Williamson, J. G. (2005). Global migration and the world economy: Two centuries of policy and performance. Cambridge: MIT Press. Hendry, D. F., & Ericsson, N. R. (1991). An econometric analysis of U.K. money demand in monetary trends in the United States and United Kingdom by Milton Friedman and Anna J. Schwartz. American Economic Review, 81, 8–38. Hlaing, U. A. (1964a). Trends of economic growth and income distribution in Burma, 1870–1940. Journal of the Burma Research Society, 47, 89–148. Hlaing, U. A. (1964b). A study of the economic development of Burma, 1870–1940. Rangoon: Department of Economics, University of Rangoon. Howard, C. W., & Buswell, K. P. (1925). A survey of the silk industry of South China. Agricultural Bulletin No. 12, Ling Nan Agricultural College, Canton Christian College. Hong Kong: Commercial Press. Hsiao Liang-lin. (1974). China’s foreign trade statistics, 1864–1949. Cambridge: Harvard University Press. Huff, G. (1989). Barter, money, credit and Singapore’s international rice trade, 1870–1939. Explorations in Economic History, 26, 161–189; Editor’s Note 27 (1990). Huff, G. (1994). The economic growth of Singapore: Trade and development in the twentieth century. Cambridge: Cambridge University Press. Huff, G. (2002). Boom-or-bust commodities and industrialization in pre-world war II Malaya. Journal of Economic History, 62, 1074–1115. Huff, G. (2007). Globalization, natural resources and foreign investment: A view from the resource rich tropics. Oxford Economic Papers, 59, p. i127–i155. Huff, G., & Caggiano, G. (2007). Globalization, immigration and Lewisian elastic labor in pre-world war II Southeast Asia. Journal of Economic History, 67, 33–68.

312

GREGG HUFF AND GIOVANNI CAGGIANO

Hunt, E. H. (1973). Regional wage variations in Britain 1850–1914. Oxford: Clarendon Press. Im, K. M., Pesaran, M. H., & Shin, Y. (2003). Testing for unit roots in heterogeneous panels. Journal of Econometrics, 115, 53–74. India. (1926–1939/40). Report of the Agent of the Government of India in Malaya. Annual series. Calcutta: Government of India Central Publication Branch. India, Census of India. (1902). Census of India, 1901, Vol. XV. Madras. Part I. Report by W. Francis and Part II. Madras: Superintendent, Government Press. India, Census of India. (1912). Census of India, 1911, Vol. XII. Madras. Part I. Report by J. Chartres Molony. Madras: Superintendent, Government Press. India, Census of India. (1922). Census of India, 1921, Vol. XIII. Madras. Part I. Report by G. T. Boag and Part II. Madras: Superintendent, Government Press. India, Census of India. (1923). Census of India, 1921, Vol. X. Burma. Part I. Report by S. G. Grantham. Rangoon: Superintendent, Government Printing. India, Census of India. (1932a). Census of India, 1931, Vol. XI. Burma. Part I. Report by J. J. Bennison. Rangoon: Superintendent, Government Printing and Stationary. India, Census of India. (1932b). Census of India, 1931 Vol. XIV. Madras. Part I. Report by M. W. M. Yeatts and Part II. Madras: Superintendent, Government Press. India, Census of India. (1933). Census of India, 1931, Vol. I. India. Part I. Report by J. H. Hutton. Delhi: Manager of Publications. India, Department of Commercial Intelligence and Statistics. (1933). Index numbers of Indian prices 1861–1931. Delhi: Manager of Publications. India, Director-General of Commercial Intelligence. (1902–1923). Prices and wages in India. Annual series. Calcutta: Superintendent of Government Printing. Indonesia, Central Bureau of Statistics. (1958). The living conditions of municipally employed coolies in Batavia in 1937. In: Wertheim, W. F. (Ed. and translator), The Indonesian Town: Studies in Urban Sociology (pp. 85–224). The Hague: van Hoeve. Ingram, J. C. (1964). Thailand’s rice trade and the allocation of resources. In: C. D. Cowan (Ed.), The economic development of South-East Asia (pp. 102–126). London: George Allen & Unwin. Ingram, J. C. (1971). Economic change in Thailand 1850–1970. Stanford: Stanford University Press. International Tin Research and Development Council. (1939). Statistical year book 1938. The Hague: Statistical Ofﬁce, International Tin Research and Development Council. International Tin Study Group. (1953). Statistical year book 1952. London: International Tin Council. Jackson, R. N. (1961). Immigrant labour and the development of Malaya, 1786–1920. Kuala Lumpur: Government Press, Federation of Malaya. Johnston, D. B. (1981). Rice cultivation in Thailand: The development of an export economy by indigenous capital and labor. Modern Asian Studies, 15, 107–126. Kaur, A. (1980). The Impact of railroads on the Malayan Economy, 1874–1941. Journal of Asian Studies, 39, 693–710. Kumar, D. (1965). Land and caste in south India. Cambridge: Cambridge University Press. Kumar, D. (1983). South India. In: D. Kumar (Ed.), The Cambridge economic history of India: c. 1757-c. 1970 (Vol. 2, pp. 207–241). Cambridge: Cambridge University Press. Kuznets, S., & Rubin, E. (1954). Immigration and the foreign born. New York: National Bureau of Economic Research. Kwangtung Government, China. (1928). Reports of statistics: Wage Indexes of labourers in Canton, China 1912–1927 (Vol. III). Canton: Che Wo Printing.

Globalization and Integration in Asia

313

Latham, A. J. H., & Neal, L. (1983). The international market in rice and wheat, 1868–1914. Economic History Review, 36, 260–280. Lebergott, S. (1964). Manpower in economic growth: The American record since 1800. New York: McGraw-Hill. Lewis, A. B., & Wang, L. (1936). Farm prices in Wuchin, Kiangsu. Economic Facts, 2, 73–91. Lewis, W. A. (1969). Aspects of tropical trade, 1883–1965. Stockholm: Almqvist and Wicksell. Lewis, W. A. (1978a). Growth and fluctuations 1870–1913. London: George Allen & Unwin. Lewis, W. A. (1978b). The evolution of the international economic order. Princeton: Princeton University Press. Lieu, D. K. (1941). The silk industry of China. Shanghai: Kelly and Walsh. Lin, A. H. Y. (1997). The rural economy of Guangdong, 1870–1937. London: Macmillan. Look Lai, W. (2002). Asian contract and free migrations to the Americas. In: D. Eltis (Ed.), Coerced and free migrations: Global perspectives (pp. 229–258). Stanford: Stanford University Press. van Luijk, E. W., & van Ours, J. C. (2001). The effects of government policy on drug use: Java, 1875–1904. Journal of Economic History, 61, 1–18. Lyons, T. P. (2003). China maritime customs and China’s trade statistics 1859–1948. Trumamsburg, NY: Willow Creek Press. Madras. (1874). Report of the census of the Madras Presidency, 1871 by W. R. Cornish. Madras: Government Gazette Press. Madras. (1911–1941). Wage censuses [variously published and known as the Quinquennial wage censuses]. Madras: Government of Madras. Madras. (1950). Monthly digest of economics and statistics Madras State (Vol. 1, No. 1 (April)). Madras: Economic Advisor to the Government of Madras. Malaya. (1911). Federated Malay States, Review of the census operations and results, 1911, by A. M. Poutney. London: HMSO. Malaya. (1921). The census of British Malaya, 1921, by J. E. Nathan. London: Waterlow and Sons. Malaya. (1921–1928). Return of foreign imports and exports (Annual series). Singapore: Government Printing Ofﬁce. Malaya. (1922–1938). Average prices, declared trade values, exchange, currency and cost of living (Annual series). Singapore: Government Printing Ofﬁce. Malaya. (1930–1937). Foreign imports and exports. Annual series. Singapore: Government Printing Ofﬁce. Malaya. (1932). British Malaya, A report on the 1931 census, by C. A. Vlieland. London: Crown Agents. Malaya. (1939/1940). Annual report of the labour department, Malaya, 1938 and 1939. Kuala Lumpur: Federated Malay States Government Press. Malaya. (1949). Malaya, A report on the 1947 census of population, by M. V. Del Tufo. London: Crown Agents. McAlpin, M. (1983). Price movements and economic activity (1860–1947). In: D. Kumar (Ed.), The Cambridge economic history of India: c. 1757-c. 1970 (Vol. 2, pp. 878–904). Cambridge: Cambridge University Press. McFadyean, A, Sir. (1944). The history of rubber regulation 1934–43. London: George Allen & Unwin. McKeown, A. (2004). Global migration, 1846–1940. Journal of World History, 15, 155–189.

314

GREGG HUFF AND GIOVANNI CAGGIANO

Mitchell, B. R. (1988). British historical statistics. Cambridge: Cambridge University Press. Mitchell, B. R., & Dean, P. (1962). Abstract of British historical statistics. Cambridge: Cambridge University Press. Mukerjee, R. (1936). Migrant Asia. Rome: Tipograﬁa I. Failli. Myers, R. H., & Wang, Y. (2002). Economic developments, 1644–1800. In: W. J. Peterson (Ed.), The Cambridge economic history of China: The Ch’ing Empire to 1800 (Vol. 9, Part 1, pp. 563–645). Cambridge: Cambridge University Press. Myung, S. C. (2000). Integration and segmentation in international markets for rice and wheat, 1877–1994. Korean Economic Review, 16, 107–123. van Niel, R. (1956). Living conditions of plantation workers in 1939–1940: Final report of the coolie budget commission. Ithaca: Cornell Modern Indonesia Project. O’Rourke, K. J., & Williamson, J. G. (1999). Globalization and history: The evolution of a nineteenth-century Atlantic economy. Cambridge: MIT Press. Owen, F. (1897–98). Mining in Perak. Transactions of the Institution of Mining and Metallurgy, 6, 51–84. Page, A. J. (1931). Report on wages in rice mills in Burma. Rangoon: Superintendent, Government Printing and Stationary. Parmer, J. N. (1960). Colonial labor policy and administration. Locust Valley, NY: J.J. Augustin. Perkins, D. H. (1969). Agricultural development in China, 1368–1968. Chicago: Aldine. Pesaran, M. H. (2007). A pair-wise approach to testing for output and growth convergence. Journal of Econometrics, 138, 312–355. Planters’ Association of Malaya. (1922). Report of the special committee on Chinese labour. Kuala Lumpur: Planters’ Association of Malaya. Purcell, V. (1967). The Chinese in Malaya. London: Oxford University Press. Rawski, T. G. (1989). Economic growth in prewar China. Berkeley: University of California Press. Robertson, R. (2000). Wage shocks and North American labor-market integration. American Economic Review, 90, 742–764. Runes, I. T. (1939). General standards of living and wages of workers in the Philippine sugar industry. Manila: Philippine Council Institute of Paciﬁc Relations. Saito, T., & Lee, K. K. (1999). Statistics on the Burmese economy. Singapore: Institute of Southeast Asian Studies. Sandhu, K. S. (1969). Indians in Malaya: Some aspects of their immigration and settlement (1786–1957). London: Cambridge University Press. Saw, S.-H. (1970). Singapore population in transition. Philadelphia: University of Pennsylvania Press. Shein, M. (1964). Burma’s transport and foreign trade (1885–1914). Rangoon: Department of Economics, University of Rangoon. Simiand, F. (1932). Le Salaire: L’e´volution sociale et la monnaie. Paris: Librairie Felix Alcan. ˆ de la vie a` Paris de 1840 a` 1954. Paris: Librairie Armand Colin. Singer-Ke´rel, J. (1961). Le cout Skinner, G. W. (1957). Chinese society in Thailand: An analytical history. Ithaca: Cornell University Press. Soliva, R. (1931). An economic view of rubber planting. Singapore: Kelly & Walsh. Sompop, M. (1989). Economic development of Thailand, 1850–1950: Response to the challenge of the world economy. Groningen: University of Groningen. Straits Settlements. (1881–1938). Annual departmental reports of the Straits Settlements (SSAR) (Annual series). Singapore: Government Printing Ofﬁce.

Globalization and Integration in Asia

315

Straits Settlements. (1891). Report of the Commission appointed to enquire into the state of labour in the Straits Settlements and protected native states, 1890. Singapore: Singapore Government Printing Press. Straits Settlements. (1921). Commission on profiteering in Straits Settlements Legislative Council Proceedings, 1920 (C202–C277). Singapore: Government Printing Ofﬁce. Sugihara, K. (2005). Patterns of Chinese immigration to Southeast Asia, 1869–1939. In: K. Sugihara (Ed.), Japan, China, and the growth of the Asian international economy, 1850–1949 (pp. 244–274). Oxford: Oxford University Press. Thailand, Department of Commerce and Statistics. (1920, 1939–40). Statistical year book of Thailand. Annual series. Bangkok: Department of Secretary General of the Council. Thoburn, J. T. (1977). Primary commodity exports and economic development. London: Wiley. T’ien-p’ei, M., & Gamble, S. D. (1926). Prices, wages and the standard of living in Peking 1900–1924, Supplement to the Chinese Social and Political Science Review. Peking: Peking Express Press. Topley, M. (1975). Marriage resistance in rural Kwangtung. In: M. Wolfe & R. Witke (Eds), Women in Chinese society (pp. 67–88). Stanford: Stanford University Press. Trocki, C. A. (1999). Political structures in the nineteenth and early twentieth centuries. In: N. Tarling (Ed.), The Cambridge history of Southeast Asia: From c. 1800 to the 1930s (Vol. 3, pp. 75–126). Cambridge: Cambridge University Press. United Kingdom. (1928). Royal Commission on agriculture in India (Vol. 3): Evidence taken in the Madras Presidency (1927); Vol. 4: Appendix to the Report. London: HMSO. United Kingdom. (1931). Royal Commission on labour in India. Evidence: Madras Presidency and Coorg (Vol. 7, parts 1 and 2). London: HMSO. United States, Department of Commerce. (1970). Historical statistics of the United States (Part 1). Washington, DC: US Department of Commerce. Urquhart, M. C., & Buckley, K. A. H. (1965). Historical statistics of Canada. Cambridge: Cambridge University Press. Wang, Y.-C. (1972). The secular trend of prices during the Ch’ing period (1644–1911). Journal of the Institute of Chinese Studies of the Chinese University of Hong Kong, 5, 347–371. Whittlesey, C. R. (1931). Government control of crude rubber. Princeton: Princeton University Press. Williamson, J. G. (1988). Migration and urbanization. In: H. Chenery & T. N. Srinivasan (Eds), Handbook of development economics (Vol. 1, pp. 426–465). Amsterdam: North Holland. Williamson, J. G. (1995). The evolution of global labor markets since 1830: Background evidence and hypotheses. Explorations in Economic History, 32, 141–196. Williamson, J. G. (1998). Real wages and relative factor prices in the third world 1820–1940: Asia. Working Paper 1844, Harvard Institute of Economic Research, Cambridge. Williamson, J. G. (2000). Globalization, factor prices and living standards in Asia before 1940. In: A. J. H. Latham & H. Kawakatsu (Eds), Asian Pacific dynamism 1550–2000 (pp. 14–45). London: Routledge. Williamson, J. G. (2002). Land, labor, and globalization in the third world, 1870–1940. Journal of Economic History, 62, 55–85. Williamson, J. G. (2006). Globalization and the poor periphery before 1950. Cambridge: MIT Press. Wilson, C. M. (1983). Thailand: A handbook of historical statistics. Boston: G. K. Hall. Wong, L. K. (1965). The Malayan tin industry to 1914. Tucson: University of Arizona Press. Wright, G. (1990). The origins of American industrial success, 1879–1940. American Economic Review, 80, 651–668.

316

GREGG HUFF AND GIOVANNI CAGGIANO

APPENDIX A. SOUTHEAST ASIA AND NEW WORLD IMMIGRATION AND EMIGRATION, 1880–1939 Malaya Chinese Immigrants Examined at Singapore 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913

89,801 101,009 109,136 106,748 111,456 144,517 167,906 164,300 150,809 127,936 126,088 134,448 213,717 153,954 190,901 175,718 114,978 133,558 149,697 200,947 178,778 207,156 220,321 204,796 173,131 176,587 227,342 153,452 151,752 216,321 269,854 251,644 240,979

Malaya Chinese Departures

Malaya Net Chinese Immigration

Malaya Indian Immigrants

Malaya Indian Departures

5,053 6,807 9,937 10,605 16,081 21,510 20,308 17,202 20,813 18,206 18,473 30,182 18,421 18,220 14,956 16,005 20,150 20,599 19,026 19,920 38,529 28,259 20,242 22,030 30,701 39,539 52,041 62,130 54,522 49,817 83,723 108,471 106,928 118,583

3,812 5,269 5,947 9,041 10,749 13,417 18,105 12,596 13,190 14,099 15,276 23,912 17,722 14,044 13,537 12,360 12,977 14,280 11,500 19,766 11,251 16,204 18,183 17,832 19,550 19,754 21,879 30,522 30,920 31,374 39,080 48,103 63,885 70,090

Globalization and Integration in Asia

317

APPENDIX A. (Continued ) Malaya Chinese Immigrants Examined at Singapore 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939

147,150 95,735 183,399 155,167 58,421 70,912 126,077 191,043 132,886 159,019 181,430 214,692 348,593 359,262 295,700 293,167 242,149 79,025 33,534 27,796 98,864 141,892 149,517 246,371 98,863

Malaya Net Indian Immigration 1880 1881 1882

1,241 1,538 3,990

Malaya Chinese Departures

61,630 41,282 35,585 37,590 68,383 98,986 96,869 78,121 87,749 77,920 120,308 155,198 149,354 139,967 167,903 213,992 282,779 86,555 68,129 69,025 80,578 66,502 54,603

Malaya Total Chinese and Indian Immigrants 96,608 110,946

Malaya Indian Immigrants

Malaya Indian Departures

51,217 75,323 95,566 90,077 65,291 101,433 95,220 45,673 58,674 49,502 55,526 90,708 174,795 157,626 63,755 114,597 70,317 20,374 18,637 20,242 89,584 66,350 45,706 123,732 44,839 23,961

63,073 50,320 54,479 57,583 52,132 46,767 55,481 61,551 45,733 42,778 37,326 43,144 65,786 93,269 91,430 76,854 152,231 103,090 85,051 32,738 28,407 38,869 40,557 45,167 76,199 42,724

Thailand Chinese Arrivals

Thailand Chinese Departures

Thailand Net Chinese Immigration

17,300

9,300

Malaya Net Chinese Immigration

121,769 113,885 22,836 33,322 57,694 92,057 36,017 80,898 93,681 136,772 228,285 204,064 146,346 153,200 74,246 134,967 249,245 58,759 30,735 72,867 68,939 179,869 44,260

8,000

318

GREGG HUFF AND GIOVANNI CAGGIANO

APPENDIX A. (Continued )

1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917

Malaya Net Indian Immigration

Malaya Total Chinese and Indian Immigrants

Thailand Chinese Arrivals

Thailand Chinese Departures

Thailand Net Chinese Immigration

1,564 5,332 8,093 2,203 4,606 7,623 4,107 3,197 6,270 699 4,176 1,419 3,645 7,173 6,319 7,526 154 27,278 12,055 2,059 4,198 11,151 19,785 30,162 31,608 23,602 18,443 44,643 60,368 43,043 48,493 11,856 25,003 41,087 32,494

119,741 122,829 132,966 164,825 185,108 185,113 169,015 146,409 156,270 152,869 231,937 168,910 206,906 195,868 135,577 152,584 169,617 239,476 207,037 227,398 242,351 235,497 212,670 228,628 289,472 207,974 201,569 300,044 378,325 358,572 359,562 198,367 171,058 278,965 245,244

18,000 13,100 13,900 14,200 15,000 15,700 18,300 18,900 16,000 17,100 27,700 33,800 29,000 27,800 31,000 33,600 33,700 27,300 30,400 36,500 54,500 44,000 45,800 68,000 90,300 61,600 66,800 80,800 76,700 72,800 73,300 60,100 69,200 53,400 39,400

9,900 8,400 7,800 7,900 9,200 7,900 10,100 10,400 9,100 9,400 11,200 16,100 17,300 18,200 18,600 19,100 20,700 19,000 19,300 18,800 29,900 23,700 30,000 38,900 53,000 49,200 57,400 73,000 63,900 60,500 57,200 56,800 47,100 40,300 36,700

8,100 4,700 6,100 6,300 5,800 7,800 8,200 8,500 6,900 7,700 16,500 17,700 11,700 9,600 12,400 14,500 13,000 8,300 11,100 17,700 24,600 20,300 15,800 29,100 37,300 12,400 9,400 7,800 12,800 12,300 16,100 3,300 22,100 13,100 2,700

Globalization and Integration in Asia

319

APPENDIX A. (Continued )

1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939

Malaya Net Indian Immigration

Malaya Total Chinese and Indian Immigrants

Thailand Chinese Arrivals

Thailand Chinese Departures

Thailand Net Chinese Immigration

13,159 54,666 39,739 15,878 12,941 6,724 18,200 47,564 109,009 64,357 27,675 37,743 81,914 82,356 66,414 12,496 61,177 27,481 5,149 78,565 31,360 18,763

123,712 172,345 221,297 236,716 191,560 208,521 236,956 305,400 523,388 516,888 359,455 407,764 312,466 99,399 52,171 48,038 188,448 208,242 195,223 370,103 143,702 23,961

67,900 65,700 70,400 76,500 95,400 115,000 92,700 95,500 106,400 154,600 101,100 134,100 86,400 74,800 59,500 25,700 27,000 45,200 48,900 60,000 33,500 25,100

37,000 43,400 36,800 46,900 65,200 66,400 66,100 60,600 73,700 76,900 72,800 68,200 62400 42,400 44,100 32,600 31,100 36,500 28,000 22,000 30,000 18,800

30,900 22,300 33,600 29,600 30,200 48,600 26,600 34,900 32,700 77,700 28,300 65,900 24,000 32,400 15,400 6,900 4,100 8,700 20,900 38,000 3,500 6,300

United States Immigrants

United States Departures

Burma Immigrants 1880 1881 1882 1883 1884 1885 1886 1887

56,100 78,700 66,200

Burma Departures

50,600 55,400 59,500

Burma Net Immigration

5,500 23,300 6,700

669,431 788,992 603,322 518,592 395,346 334,203 490,109

320

GREGG HUFF AND GIOVANNI CAGGIANO

APPENDIX A. (Continued )

1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924

Burma Immigrants

Burma Departures

Burma Net Immigration

United States Immigrants

United States Departures

86,700 194,900 133,500 151,200 123,400 129,100 119,500 – 134,600 123,400 149,200 167,000 163,300 154,600 142,800 180,200 182,700 238,500 360,500 271,100 319,200 302,200 331,100 368,300 327,500 380,200 268,400 338,800 258,800 223,100 259,900 284,700 341,100 331,900 360,000 382,700 388,200

69,500 163,000 98,400 112,900 116,600 58,300 129,900 – 86,900 91,600 106,700 105,700 120,500 114,200 135,000 139,700 125,200 175,700 319,800 267,600 301,000 301,900 298,600 311,500 331,500 355,300 146,200 249,000 252,300 237,100 234,200 219,000 247,900 303,800 310,300 295,300 315,800

17,200 31,900 35,100 38,300 6,800 70,800 9,600 – 47,700 31,800 42,500 61,300 42,800 40,400 7,800 40,500 57,500 62,800 40,700 3,500 18,200 300 32,500 56,800 4,000 24,900 122,200 89,800 6,500 14,000 25,700 65,700 93,200 28,100 49,700 87,400 72,400

546,889 444,427 455,302 560,319 579,663 439,730 285,631 258,536 343,267 230,832 229,299 311,715 448,572 487,918 648,743 857,046 812,870 1,026,499 1,100,735 1,285,349 782,870 751,786 1,041,570 878,587 838,172 1,197,892 1,218,480 326,700 298,826 295,403 110,618 141,132 430,001 805,228 309,556 522,919 706,896

395,000 226,000 202,000 296,000 333,000 308,000 303,000 204,000 130,000 66,000 95,000 124,000 288,000 248,000 199,000 81,000 77,000

Globalization and Integration in Asia

321

APPENDIX A. (Continued )

1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939

Burma Immigrants

Burma Departures

Burma Net Immigration

372,700 408,400 428,300 418,600 405,300 368,500 319,600 334,200 263,800 279,100 296,600 269,200 271,200 240,500

350,900 342,500 361,200 333,000 371,800 399,200 367,100 288,400 252,200 226,600 234,200 221,600 232,300 253,400

21,800 65,900 67,100 85,600 33,500 30,700 47,500 45,800 11,600 52,500 62,400 47,600 38,900 12,900

United States Net Immigration 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897

United States Immigrants 294,314 304,488 335,175 307,255 279,678 241,700 97,139 35,576 23,068 29,470 34,956 36,329 50,244 67,895 82,998

United States Departures 93,000 77,000 73,000 78,000 69,000 51,000 62,000 103,000 80,000 40,000 39,000 36,000 27,000 25,000 27,000

Canada Immigrants

Brazil Immigrants

38,505 47,991 112,458 133,624 103,824 79,169 69,152 84,526 88,766 91,600 75,067 82,165 30,996 29,633 20,829 18,790 16,835 21,716

30,355 11,548 29,589 34,015 23,574 34,724 32,650 54,932 132,070 65,165 106,819 215,239 85,906 132,589 60,182 164,831 157,423 144,866

322

GREGG HUFF AND GIOVANNI CAGGIANO

APPENDIX A. (Continued ) United States Net Immigration 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934

387,870 525,786 839,570 582,587 505,172 889,892 915,480 122,700 168,826 229,403 15,618 17,132 142,001 557,228 110,556 441,919 629,896 201,314 227,488 262,175 229,255 210,678 190,700 35,139 67,424 56,932 10,530

Canada Immigrants

Brazil Immigrants

31,900 44,543 41,681 55,747 89,102 138,660 131,252 141,465 211,653 272,409 143,326 173,694 286,839 331,288 375,756 400,870 150,484 36,665 55,914 72,910 41,845 107,698 138,824 91,728 64,224 133,729 124,164 84,907 135,982 158,886 166,783 164,993 104,806 27,530 20,591 14,382 12,476

76,862 53,610 37,807 83,116 50,472 32,941 44,706 68,488 72,332 57,919 90,536 84,090 86,751 133,575 177,887 190,343 79,232 30,333 31,245 30,277 19,793 36,027 69,041 58,476 65,007 84,549 96,052 82,547 118,686 97,974 78,128 96,186 62,610 27,465 31,494 46,081 46,027

Globalization and Integration in Asia

323

APPENDIX A. (Continued )

1935 1936 1937 1938 1939

United States Net Immigration

Canada Immigrants

Brazil Immigrants

4,044 329 23,244 42,895 55,998

11,277 11,643 15,101 17,244 16,994

29,585 12,773 34,677 19,388 22,668

Argentina, 1881–90 to 1936–40 (’000 persons)

1881–1890 1891–1900 1901–1910 1911–1920 1921–1925 1926–1930 1931–1935 1936–1940 Sources: See Appendix E.

Immigrants

Departures

Net Immigration

841 648 1,764 1,205 708 690 331 135

203 328 644 936 255 286 204 58

638 320 1,120 269 452 404 127 77

324

APPENDIX B. SOUTHEAST ASIA AND NEW WORLD IMMIGRATION PER 1000 MEAN POPULATION, 1880–1939 Malaya Indians Immigrants

Malaya Total Chinese and Indians

57.7 65.0 70.2 68.6 71.7 92.9 108.0 105.7 97.0 82.3 69.3 73.9 117.4 84.6 104.9 96.5 63.2 73.4 82.3 110.4

3.2 4.4 6.4 6.8 10.3 13.8 13.1 11.1 13.4 11.7 11.9 16.6 10.1 10.0 8.2 8.8 11.1 11.3 10.5 10.9 21.2

62.1 71.3 77.0 79.0 85.5 106.0 119.0 119.0 108.7 94.2 85.9 84.0 127.4 92.8 113.7 107.6 74.5 83.8 93.2 131.6

Thailand

2.7 2.8 2.0 2.2 2.2 2.3 2.4 2.8 2.9 2.3 2.4 4.0 4.8 4.1 4.0 4.4 4.8 4.8 3.9

Burma

7.8 10.9 9.2 12.0 27.0 18.5 16.6 13.6 14.2 13.1 14.8 13.6 16.4 18.3 17.9

United States

Canada

Brazil

11.7 13.8 10.5 9.1 6.9 5.8 8.6 9.5 7.8 7.9 8.0 8.3 6.3 4.1 3.7 4.9 3.3 3.3 4.4 6.4

8.4 10.5 24.5 29.2 22.7 17.3 15.1 18.4 19.4 20.0 16.4 16.3 6.1 5.9 4.1 3.7 3.3 4.3 6.3 8.8 8.2

2.3 0.9 2.3 2.6 1.8 2.7 2.5 4.2 10.1 5.0 8.2 13.3 5.3 8.2 3.7 10.2 9.7 8.9 4.7 3.3 2.3

GREGG HUFF AND GIOVANNI CAGGIANO

1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900

Malaya Chinese Examined at Singapore

11.9 8.5 9.3 13.0 16.7 22.0 26.2 23.0 21.0 35.4 36.3 35.8 39.7 17.1 25.2 32.0 30.1 21.8 33.9 31.9 11.9 15.3 12.9 14.5 23.6 45.5 41.1 16.6 29.9

87.4 96.0 102.3 99.4 89.8 96.5 122.2 87.8 85.1 126.7 126.6 120.0 120.3 66.4 57.2 93.3 82.0 41.4 57.7 74.0 61.7 49.9 54.3 61.7 79.6 136.4 134.7 93.7 106.3

3.9 4.7 7.0 5.6 5.9 8.7 11.6 7.9 8.6 10.4 8.8 8.3 8.4 6.9 7.9 6.1 4.5 7.8 7.5 8.1 7.4 9.2 11.1 9.0 9.2 10.3 14.9 9.8 12.9

13.7 12.6 15.9 16.2 21.1 31.9 24.0 28.2 26.7 29.3 29.1 25.9 30.0 21.2 26.8 20.4 17.6 20.5 22.5 26.9 23.8 25.8 27.5 27.8 26.7 29.3 30.7 30.0 29.1

5.8 7.7 10.1 9.6 12.1 13.0 15.2 9.2 8.9 12.3 0.9 0.8 1.2 1.2 0.3 0.3 0.3 0.1 0.1 0.4 6.9 2.7 4.5 6.1 2.5 2.6 2.9 2.6 2.4

9.1 14.5 22.6 21.4 23.1 34.5 44.5 23.4 28.4 46.8 41.9 47.5 50.6 19.0 4.6 7.1 9.2 5.3 13.6 17.5 9.7 6.8 14.2 13.2 9.0 14.4 16.8 17.7 17.5

4.1 2.5 1.6 2.2 3.4 3.6 2.9 4.5 4.2 4.3 5.3 7.1 7.6 3.2 1.2 1.3 1.2 0.8 1.4 2.8 1.9 2.1 2.8 3.1 2.7 3.9 3.2 2.5 3.1

325

75.5 87.5 93.0 86.5 73.1 74.6 96.0 64.8 64.1 91.4 90.3 84.2 80.6 49.2 32.0 61.4 51.9 19.5 23.7 42.2 49.8 34.6 41.4 47.3 55.9 90.8 93.6 77.1 76.4

Globalization and Integration in Asia

1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929

326

APPENDIX B. (Continued ) Malaya Indians Immigrants

Malaya Total Chinese and Indians

Thailand

Burma

United States

Canada

Brazil

63.1 16.9 7.2 5.9 21.1 30.3 32.0 52.7 21.1

18.3 4.4 4.0 4.3 19.2 14.2 9.8 26.5 9.6

81.4 21.3 11.2 10.3 40.3 44.5 41.8 79.1 30.7

8.3 5.8 4.6 2.1 2.1 3.5 3.8 4.6 2.6 1.9

26.4 20.3 21.2 16.8 17.7 18.8 17.1 17.2 15.3

2.1 0.8 0.3 0.2 0.2 0.3 0.3 0.4 0.5 0.7

11.1 2.5 1.9 1.3 1.2 1.0 1.1 1.4 1.6 1.6

2.0 0.7 0.8 1.2 1.2 0.8 0.3 0.9 0.5 0.6

Source: See Appendix E.

GREGG HUFF AND GIOVANNI CAGGIANO

1930 1931 1932 1933 1934 1935 1936 1937 1938 1939

Malaya Chinese Examined at Singapore

SE China 100.0 111.1 112.8 107.2 103.1 117.0 116.6 130.1 125.3 127.2 119.7 113.6 108.7 104.5 100.8 97.4 94.0 90.5 86.6 113.3 97.5 83.2

Burma

100.0 94.3 93.8 96.3 95.8 102.8 97.2 91.5 63.3 57.3 62.8 69.1 76.3 74.6 59.4 55.8 64.4 57.0 55.0 64.9 79.8 81.1

100.0 100.3 100.7 91.2 84.7 94.6 88.3 87.9 98.7 88.2 80.9 91.4 92.5 89.3 84.9 74.9 85.9 85.0 77.3 80.7 85.9 70.1

Thailand 100.0 94.5 106.6 83.5 78.6 83.2 84.6 82.5 81.8 67.4 53.4 51.4 49.1 46.6 43.7 42.2 40.1 49.7 55.9 62.1 66.2 61.5

Malaya Chinese

Malaya Indians

France

100.0 100.5 100.6 100.6 100.3 99.5 97.8 94.2 86.7 70.9 75.5 106.8 104.7 104.7 106.8 129.9 110.6 189.3 141.6 117.7 103.6 90.9

100.0 100.5 100.6 100.6 100.3 99.5 97.8 94.2 86.7 75.5 65.0 53.4 51.6 50.6 72.6 41.6 57.4 70.8 84.2 78.7 74.3 70.5

100.0 99.5 102.0 104.3 106.4 106.9 108.6 108.0 109.3 107.8 105.8 113.3 111.6 113.6 117.5 117.2 114.6 118.4 117.4 122.4 122.4 118.7

Germany 100.0 100.0 104.2 112.4 115.7 123.7 128.0 130.3 134.1 130.7 132.4 134.1 136.0 135.3 144.8 138.4 142.0 142.0 149.7 147.7 150.1 159.5

United Kingdom 100.0 101.4 107.1 111.9 114.2 119.7 118.8 116.7 120.3 120.3 127.3 126.1 136.1 135.5 137.7 138.3 136.7 131.2 125.9 136.1 138.1 138.1

United States 100.0 105.8 110.2 110.3 114.2 115.8 121.1 124.3 126.6 126.6 126.6 127.8 128.2 133.7 131.7 133.0 135.0 137.0 138.6 141.2 143.7 145.3

327

1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903

Madras

Globalization and Integration in Asia

APPENDIX C. ASIA AND GLOBAL CORE UNSKILLED REAL WAGES, 1882–1939 (1882=100)

328

APPENDIX C. (Continued ) SE China

85.1 87.5 75.5 70.9 77.1 90.2 83.7 77.5 98.9 100.0 100.4 98.5 94.6 94.8 93.5 93.3 91.6 89.6 96.0 94.3 94.2 94.6 99.9 110.4

64.1 54.3 51.6 48.3 58.4 58.3 58.1 77.9 57.9 78.1 58.1 58.1 82.3 58.3 37.1 58.5 58.2 74.2 58.2 58.3 58.3 58.2 71.9 58.2

Burma

79.1 76.0 73.3 69.7 71.3 77.4 81.6 73.7 70.3 66.2 65.5 64.0 61.5 57.9 53.2 43.6 43.4 52.1 88.5 73.1 68.5 82.1 68.0 73.0

Thailand

57.1 52.8 53.5 54.2 54.8 55.4 56.0 56.6 57.2 58.8 58.9 58.0 69.9 63.1 44.5 29.3 34.7 46.9 50.9 50.0 54.2 48.3 47.7 49.5

Malaya Chinese

Malaya Indians

France

72.3 86.1 91.1 79.5 80.6 95.4 107.0 151.1 126.6 146.4 75.8 69.2 85.6 106.3 89.3 77.4 48.9 98.4 84.9 89.7 87.9 91.2 88.7 93.3

66.5 61.6 55.1 45.9 50.6 51.7 50.8 76.9 72.6 72.4 70.4 70.2 91.0 85.1 69.6 29.7 26.2 53.8 55.0 56.3 47.8 65.9 75.8 77.2

121.7 119.5 122.3 124.3 125.9 127.2 122.9 122.0 118.9 120.4 112.9 92.3 88.1 90.0 85.3 98.2 79.7 116.6 125.7 127.6 124.9 120.3 113.3 106.9

Germany

United Kingdom

162.1 165.1 161.9 172.5 178.8 180.9 186.9 184.9 179.3 182.9 180.1 177.2 174.4 171.6 168.8 166.0 163.3 160.5 157.7 155.0 152.2 199.7 213.8 213.8

136.1 132.3 123.7 119.1 130.5 128.7 122.1 121.2 116.1 120.2 104.3 89.9 83.6 84.8 96.1 108.5 115.4 184.0 158.5 139.9 129.9 144.7 161.9 167.2

United States 144.9 148.1 152.8 152.2 147.7 152.3 150.9 147.4 146.2 150.4 150.0 150.4 172.3 182.4 195.0 197.6 195.4 181.3 178.5 192.6 199.0 192.4 193.8 201.0

GREGG HUFF AND GIOVANNI CAGGIANO

1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927

Madras

108.0 83.7 78.3 73.0 71.1 91.0 91.3 86.3 88.6

60.9 58.3 57.8 107.8 57.8 58.3 58.3 58.0 88.4

62.8 84.5 98.6 154.3 135.3 118.0 122.6 111.6 115.3 112.3 114.8 106.8

50.7 50.0 56.4 60.5 67.6 73.8 79.5 74.3 74.5 69.5 71.0 71.6

93.3 88.8 66.3 58.3 42.2 106.3 140.5 131.6 140.6 172.6 101.9

84.0 91.8 55.5 70.4 54.3 83.8 103.3 93.6 100.3 104.8 96.9 113.2

110.9 118.7 124.5 123.1 129.7 139.6 141.8 151.4 135.2 152.7 138.6 133.3

225.0 237.6 250.9 254.2 225.3 207.7 202.6 199.3 197.7 197.6 196.0 197.4

170.0 177.4 210.3 245.8 250.5 249.1 242.2 242.8 231.2 208.8 246.7

204.9 210.2 212.0 223.6 217.8 229.7 265.0 267.7 268.1 294.2 307.7 317.5

Globalization and Integration in Asia

1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939

Source: See Appendix E.

329

330

GREGG HUFF AND GIOVANNI CAGGIANO

APPENDIX D. TERMS OF TRADE SOUTHEAST ASIA, MADRAS, AND SOUTHEASTERN CHINA, 1882–1936 (1882=100)

1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918

Burma

Malaya

Thailand

Madras

SE China

100.0 120.6 121.3 120.2 144.6 141.6 140.3 153.7 143.2 145.5 160.3 169.4 158.0 163.5 164.6 182.9 193.6 191.3 186.9 174.4 174.0 184.3 195.8 179.4 188.4 200.9 217.0 216.0 202.8 200.3 216.3 225.5 198.5 201.7 159.6 142.5 133.8

100.0 90.0 93.5 90.6 99.7 115.6 122.8 94.5 92.9 93.8 95.7 88.5 74.0 69.7 54.0 60.2 61.8 93.6 103.3 95.4 98.1 96.2 94.9 106.7 131.6 125.3 105.1 107.0 120.6 156.0 153.1 147.3 121.8 129.0 109.8 102.5 77.9

100.0 105.3 94.7 103.1 121.3 115.5 110.9 109.7 122.5 128.6 145.6 141.6 153.6 113.2 113.7 106.3 129.7 147.1 143.1 137.0 142.0 153.7 146.6 150.8 145.1 147.9 139.1 122.3 128.7 147.2 163.7 123.0 116.0 131.0 140.1 136.1 205.2

100.0 104.8 107.8 110.0 104.7 101.4 94.9 99.5 101.1 105.5 106.3 109.3 110.0 107.5 104.8 115.1 118.0 102.0 104.5 114.7 112.9 103.3 107.8 104.6 105.3 122.4 125.2 112.3 119.1 119.1 115.2 128.4 98.5 57.4 58.9 56.6 68.8

100.0 103.7 94.0 90.6 84.3 81.4 80.7 77.9 86.1 91.2 93.9 82.2 58.4 57.6 56.3 56.8 60.1 71.5 57.5 52.6 63.9 59.3 55.5 62.2 65.5 66.0 49.1 51.1 48.5 47.9 43.9 48.7 45.5 41.1 46.0 44.7 39.4

Globalization and Integration in Asia

331

APPENDIX D. (Continued )

1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936

Burma

Malaya

Thailand

Madras

SE China

144.2 157.7 186.1 202.9 192.4 189.6 210.6 234.9 224.9 194.9 212.8 231.0 234.0 244.1 208.1 199.8 216.3 200.5

49.9 57.8 43.3 43.7 61.3 59.5 102.8 87.5 79.5 54.3 50.7 38.4 36.9 41.9 61.3 79.2 71.2 70.0

350.8 144.5 146.4 141.4 159.4 177.9 188.7 195.6 182.0 173.7 175.5 141.2 88.7 87.8 92.2 97.5 108.4 111.8

81.2 85.7 60.4 85.2 94.4 89.5 86.9 75.9 72.5 89.8 93.3 78.4 69.0 63.8 68.2 71.0 77.6 76.9

39.8 35.5 39.3 56.0 58.5 50.5 50.2 51.5 45.4 45.3 44.3 36.8 29.6 24.2 23.9 21.8 25.6 26.4

Source: See Appendix E.

APPENDIX E. DATA SOURCES AND METHODS Immigration Governments in each of the three Southeast Asian countries kept records for Asian arrivals and departures and these correspond closely to migration. Malayan data alone distinguish between Indians and Chinese. Statistics for Burma relate almost entirely to Indians and those for Thailand to Chinese. Sources for New World data are for population as well as immigration. Burma: For Burma data relating to annual immigration and emigration were published from 1885 onwards. Information derives from the records of the Port Health Department. Figures for Rangoon, which normally handled two-thirds to three-quarters of those traveling by ship, represent an actual count and include infants and persons without a ticket. For other Burmese

332

GREGG HUFF AND GIOVANNI CAGGIANO

ports, ﬁgures were obtained from the shipping companies and are for tickets sold. Immigration data appear to be approximately accurate but ﬁgures for emigration probably give no more than a broad indication of trends (Baxter, 1941, pp. 10–14, 121; India, 1932a, part I, pp. 19–20; Cheng, 1968, pp. 263–264). Immigration statistics do not include ambulatory arrivals or departures. These were mainly Bengali immigrants from the area around Chittagong. Typically each year in the inter-war period some 40,000 Chittagongians walked across the East Bengal (now Bangladesh)–Burma border for work related to the rice harvest, principally in the Arakan district of the Burmese province of Akyab and most returned home after the end of the harvest (Baxter, 1941, p. 50). Data sources: Cheng (1968, pp. 262–263); Baxter (1941, p. 121); Fenichel and Huff (1971, pp. 41–42). Malaya: Annual data for Indian immigration and emigration exist from 1880 onwards and are accurate because of the Malayan government’s role in bringing Indian workers to Malaya. Unskilled laborers from the subcontinent constituted the great bulk of the Malayan trafﬁc, but the published data also include an unknown number of other Indians such as merchants traveling between the two countries. Until the 1930s, when demand for labor on rubber estates declined sharply, non-laborers were a small proportion of the Indian totals (Sandhu, 1969, pp. 95–125). Data source: Saw (1970, p. 52). Almost all Chinese immigrants to Malaya ﬁrst landed at Singapore. Beginning in 1881, records of Chinese examined at the port by its ofﬁcials, by health ofﬁcers, or by the Chinese Protectorate (a government department set up to safeguard Chinese welfare) provide a reliable measure of annual immigrant inﬂows. But no statistics for Chinese emigration were kept before 1916, and until 1930 include only Chinese deck passengers departing from Singapore. Beginning in 1931 data are for deck passengers leaving all Malayan ports (in effect Penang as well as Singapore) and suggest understatement in the 1916–1930 departure ﬁgures. For 1911–1915 Chinese emigration from Malaya was estimated as 400,000 (Malaya, 1932, p. 113). Data for 1930–1939 refer to 1930–1938 only. Data sources: Straits Settlements (1881–1938); (from this source see years 1881–1911 immigration reports; Secretary for Chinese Affairs for 1930–1938; Progress of the People of the Straits Settlements for 1934–1938); Malaya (1921, p. 21, 1932, p. 113). Thailand: Comprehensive immigration data for Thailand ﬁrst become available in 1882, when the great majority of passengers from China began traveling on steamers under European ﬂags. Utilizing statistics for Chinese

Globalization and Integration in Asia

333

emigrants traveling in non-China ships for each of Southeastern China’s emigrant ports, together with records at the port of Bangkok which date from the late 1880s, G. William Skinner compiled ﬁgures for immigration. For annual Chinese arrivals in Thailand, Skinner estimated the maximum probable error as 10% for 1882–1892, 7.5% for 1893–1905, and 5% for 1906–1917. For departures in each period, however, he speciﬁed nearly twice these margins of error. Data for 1881–1910 refer to 1882–1910 only. Data sources: Skinner (1957, pp. 61, 173); Sompop (1989, pp. 207–208). United States: United States, Department of Commerce (1970, part 1, pp. 8, 105–106) and Kuznets and Rubin (1954, pp. 94–96). In Table 4 the 1881–1910 net immigration ﬁgure is for 1880–1910 as estimated by Kuznets and Rubin (1954, p. 94). Canada: Urquhart and Buckley (1965, pp. 14, 23). Argentina: Diaz Alejandro (1970, pp. 421, 424). In Table 4 the 1931–1939 ﬁgures are for 1931–1940. Brazil: Brazil (1960, pp. 5, 12).

Wages and Prices Wages: For the three Southeast Asian countries data in the sources used is typically for daily unskilled wages but for long periods for India and China wages are stated monthly or yearly. The usual caveats for unskilled wages in an underdeveloped area apply. These include the possibilities of: underemployment; greater variations in days employed than daily wages over periods of even four or ﬁve years; payment in kind; and that some ﬁgures give a range of wages without accompanying information on the distribution of wages within that range (here we use the mid-point). However, no suggestion of systematic bias exists. Nominal wages are not available for every year for any of the countries or areas we consider. At the beginning of the discussion of wage data for each country or area we state the years for which we have wage observations. Burma: Data for 1873–1911 and 1918–1939. Wages for Burma derive from government reports. From the 1860s to 1901, Burma’s government collected average daily wages for unskilled male workers in Lower Burma in regular employment. Government ﬁgures record a simple average of several locations in Lower Burma including Rangoon. Between 1871 and 1901, Lower Burma’s population grew almost threefold from 2.0 million to 5.6 million (measured in terms of 1872 census boundaries). The government’s reported average wage ﬁgure has been used because it reﬂects the overall

334

GREGG HUFF AND GIOVANNI CAGGIANO

wage that potential immigrants might anticipate and because with such ﬂuid demography and heavy immigration, labor was likely readily to respond to wage differentials between areas. Under these circumstances a population-weighted average might largely reﬂect ex post not ex ante wage opportunities. The 1880–1901 wage statistics relate mainly to immigrant Indian wages. Although some Burmese had always worked as agricultural laborers, no signiﬁcant class of such individuals existed until the turn of the century. By 1902, and especially after 1910 with the clear emergence of a class of Burmese agricultural laborers, wage statistics reﬂect a uniﬁed labor market in Burma (Hlaing, 1964a, pp. 122–123; Adas, 1974, pp. 128–129; Baxter, 1941, pp. 36, 39, 42, 66–67, 90–92). For 1902–1911 wages are from statutory annual reports relating to the Indian Factories Acts and are the average of daily wages for rice mill coolies in the ﬁve locations of Akyab, Rangoon Town, Hanthawaddy, Basseim, and Amherst. For 1918–1939 wages are an average of minimum and maximum coolie wages in Burma’s factories. These were predominantly rice mills. The nominal wage for 1882 is for that year, from Burma (1868/69–1935/36 (report for 1882)), and is the unskilled (coolie) wage per month. Data sources: Burma (1868/69–1935/36 (years 1868–1901)); Burma (1897–1939); Burma (1917); Page (1931, pp. 11–51). Malaya: Indian workers data for 1890, 1893–1895, 1897, 1907, 1909–1938. Chinese workers data for 1875–1879, 1891–1893, 1896–1899, 1904, 1906–1908, 1910–1922, 1924, 1926, 1929–1934, 1937–1938. The Malayan labor market consisted very largely of Chinese and Indians. Comparatively few Malays worked for wages, preferring to concentrate either in the traditional activities of ﬁshing and farming or to grow rubber on their own smallholdings. From the 1870s to about 1910, tin mining remained the largest single source of Chinese employment and a principal inﬂuence on immigration to Malaya. By the 1910s, however, tin mining was more than counterbalanced by the expansion of rubber cultivation, which drew large numbers of Chinese and, for the ﬁrst time, Indians to Malaya. Job overlaps, including the many Chinese working on rubber estates, and considerable labor mobility allow one to speak, if not of a common Malayan wage, of wage movements ﬂuctuating around the level obtaining for unskilled Indian rubber estate workers (Bauer, 1948, p. 21; see also Whittlesey, 1931, pp. 87–91, 117 on labor shortages and the mobility of labor in response to rising wages). The wages of Indian rubber estate workers served as a benchmark for all workers in Malaya (Malaya, 1939/1940 from which see the report for 1939, p. 39). It was well known, however, that in boom years Chinese exacted high

Globalization and Integration in Asia

335

wages, while in a bust Chinese wages fell appreciably more than Indian (Bauer, 1948, p. 219). To reﬂect these differences we construct separate series for Indian and Chinese wages. Overlaps for wages paid to Chinese in the tin and rubber industries exist for 1912, 1913, and 1922. Comparison for these years shows that Chinese wages in the rubber industry were 0.78 of those in tin mining and tin and rubber wage series were linked on this basis. During the 1920s tin mining became an increasingly capital-intensive, European-dominated industry that employed relatively few Chinese. Insofar as wages for Chinese in mining exist for this later period, they cannot meaningfully be compared with earlier mining wages or rubber industry wages and have not been used. For 1882–1889 Indian wages are traced back using Malaya Chinese data and on the basis of the subsequent relationship between Indian and Chinese wages. For Indians the nominal wage for 1882 is for 1890 and for Chinese an average of wages for tin mining workers for 1878 and 1879. The Indian wage is from Straits Settlements (1891, p. 46) and the Chinese from Jackson (1961, pp. 41, 154) and Doyle (1879, p. 29). Data sources: Indians – Straits Settlements (1891, p. 46); Kaur (1980, p. 698); Owen (1897–1898, p. 84); Thoburn (1977, pp. 285–286). Chinese – Jackson (1961, pp. 41, 154); Doyle (1879, p. 29); Becher (1892–1893, p. 101); Owen (1897–1898, p. 67); Wong (1965, pp. 100, 175, 206, 219); Chen (1923, pp. 89, 94); Planters’ Association (1922, appx. IV); Figart (1925, p. 179); Soliva (1931, p. 28); Drabble (1991, p. 40); Bauer (1948, pp. 219, 232–243); Blythe (1938) indicates Chinese estate wages of $11.40 in 1936 (p. 27) and $16.80 in 1937 (pp. 33, 35). It is clear that Chinese estate wages were cut at the end of 1937 and, with the emergence of heavy Chinese unemployment, fell sharply in 1938 (Parmer, 1960, p. 245; Bauer, 1948, p. 241). The 1938 wage is based on that year’s Labour Department Report, which put Chinese estate wages at about 20% above Indian (Malaya, 1939 and 1940, from which see the report for 1938, p. 40). Thailand: Data for 1889–1890, 1896, 1898, 1901–1902, 1905, 1912–1939. Even in the 1950s good land was still available in Thailand’s fertile Central Plain. The existence of surplus land and, until at least 1929, higher earnings for commercial farmers than coolie employment, encouraged the Thai to continue to concentrate on cultivation of the land (Sompop, 1989, pp. 167–68. From 1910 to 1929, however, small farmers with about two hectares of land did not earn more than coolie labor due to low rice prices). Chinese did not plant rice in competition with the Thai and performed almost all wage labor outside agriculture (Ingram, 1971, pp. 43, 56–57).

336

GREGG HUFF AND GIOVANNI CAGGIANO

In Thailand rubber cultivation ﬁrst assumed importance towards the end of World War I and exports from 1923 onwards. Chinese had a major role in the rubber industry as laborers, as the Thai were not responsive to high wages. During the inter-war period, rubber production engaged some 50,000–60,000 tappers, largely Chinese, and constituted a substantial part of Chinese employment. Chinese tappers were typically paid on a share basis of 50% of the selling price of ﬁnished sheets of rubber, an arrangement which involved the tapper processing the rubber collected (Ingram, 1971, pp. 103–104. For prices of rubber exports from Thailand, see Sompop, 1989, p. 217). Wage data, assembled by Feeny and by Ingram from a variety of sources, are for unskilled labor. Improved communications, especially railway construction, increasingly facilitated the movement of labor in Thailand (Skinner, 1957, pp. 198–199; Sompop, 1989, pp. 17, 176–178). Feeny (1983, p. 697, 1982, p. 163) convincingly argues that an approximate equality of real urban and rural wages resulted from a combination of this transport availability, labor mobility, and the movement of workers from Bangkok to public works projects in the Central Plain and beyond. Until 1900 wage series, Feeny (1982, p. 29) stresses, are ‘‘based on fragmentary evidence’’. For 1882–1888 Thai wages are traced back using the Thailand wage series in Williamson. Data sources: Feeny (1982, pp. 132–133); Ingram (1964, p. 115); Williamson (1998, appendix). The 1882 nominal wage is for 1889 and from Ingram (1964, p. 115). It is the Bangkok unskilled daily wage assuming 24 days employment per month. Southeast Asia Prices: No consumer price index covering 1880–1939 exists for Burma, Malaya, or Thailand. Typically for these and other Southeast Asian countries rice, and sometimes also textile, prices have been used as a deﬂator (e.g. Hlaing, 1964a, p. 121; van Luijk & van Ours, 2001, pp. 8–9). We construct an eight-commodity price index to provide a more representative measure than hitherto available of the cost of living for unskilled workers in the three Southeast Asian countries. The index consists of: rice (.58), dried ﬁsh (.06), sugar (.05), tea (.03), beer and ale (.12), kerosene (.04), tobacco (.03), and white and grey shirting (.09). For 1880–1884 data are available only for rice, dried ﬁsh, sugar, and shirting. For these years we weight all commodities as above except rice. Its share is increased to stand for the unavailable data. Until 1919 data for the index are from unit values derived from annual trade returns and beginning in the 1920s also include some wholesale prices. Reliance on trade prices is not ideal but for much of the period 1880–1939 affords the only consistent series possible. Trade prices are

Globalization and Integration in Asia

337

almost certainly a good reﬂection of equivalent movements in wholesale and even retail prices, because internal and external markets in all three countries were exceptionally open and competitive. Trade restrictions were either non-existent or, in the few instances they did apply, minimal until the 1930s, when Malaya and Burma attempted to limit imports of Japanese manufactures, mainly textiles. The rice market in Southeast Asia was, even by the 1880s, well integrated but never perfectly so. Deviations from long-run equilibrium rice prices had signiﬁcance for real wages in Southeast Asia and so potentially also for immigration. To take the fullest possible account of ﬂuctuations we use country-speciﬁc rice prices for the three Southeast Asian countries. For other items, price data from one Southeast Asian country represent prices in the other two. In the case of sugar and dried ﬁsh this is acceptable because Singapore, which served as an entrepot for much of Southeast Asia, traded extensively with Thailand and Burma. Both countries obtained sugar via Singapore. It was an outlet for Burma’s rice and bought large quantities of rice and dried ﬁsh from Thailand. Both foods were consumed in Malaya while, in exchange for rice, ﬁsh was shipped to Burma and textiles to Thailand (Huff, 1994, pp. 54–55, 102–106). Almost all textiles were imported in the absence of signiﬁcant manufacture in Southeast Asia. Other goods in the index were also internationally traded and obtained in Southeast Asia at world prices. For all three countries the United Kingdom was the main source of manufactured goods. Japan’s growing role as a lowcost supplier of manufactures to Southeast Asia in the inter-war years, especially between 1930 and 1934, is reﬂected in the index’s use of trade prices of white and grey shirting. Weights in the index favor essential consumption and are based on a composite of contemporary budget surveys for unskilled urban and plantation workers (Bennison, 1928, pp. 176–181; Andrew, 1933, pp. 226–250; Malaya, 1922–1938; Creutzberg, 1979, p. 78 (budget devised by Polak); Indonesia, Central Bureau of Statistics, 1958; van Niel, 1956; Runes, 1939, pp. 19, 21). Food accounted for 73% of the spending of ﬁeld and factory laborers living on plantations in Java in 1939 (van Niel, 1956, p. 78). Our index uses unchanged weights and in it food accounts for 72% and rice for 58% of total expenditure. A 1937 survey of municipally employed workers in Batavia found that food took 60% of expenditure of two-to-ﬁve-person households with a household gross daily wage of 30 cents (US$ 0.54 or 1s 8d.). Such a wage was effectively for unskilled work and the one received by half of all households surveyed. For these households food was, however, probably a smaller proportion of spending than for immigrant workers in

338

GREGG HUFF AND GIOVANNI CAGGIANO

Burma, Malaya, and Thailand, since in Batavia 15% of spending was for rent and 3% for school fees. Uncooked and prepared rice accounted for 70% of food expenditure in Batavia (Indonesia, Central Bureau of Statistics, 1958, pp. 126–147, 220–223). In the three Southeast Asian countries, rice, although apparently often two-thirds or more of gross daily calorie intake of unskilled workers, generally made up (as in Batavia) no more than 35% or 45% of total expenditure (Bennison, 1928, pp. 28–29, 37; see also van Niel, 1956, p. 79). Accordingly, in our index rice must serve as a proxy for a number of other food purchases. It does so reasonably well. According to a 1920 commission, ‘‘The position which rice occupies in the economies of this part of the world is not merely that of an article of food, it really represents the standard of value, the ‘pecunia’ of the East y shopkeepers considered any rise in the price of rice to be a good and sufﬁcient reason for advancing the price of every commodity they sold’’ (Straits Settlements, 1921, p. C273 and for discussion of the ‘‘moneyness’’ of rice, Huff, 1989). Contemporary budget information indicates limited expenditure on protein. Dried ﬁsh, consumed throughout the three Southeast Asian countries and weighted 6%, stands for such expenditure. Tea (3%) was a ubiquitous consumption item, while the 5% weight for sugar represents its use not just in cups of tea but in confectionery and cooking. Among non-food items we weight beer and ale as 12% of total expenditure and in this are persuaded by Bennison’s survey data and his observation that men living in bad housing, working long hours, and without home life naturally spend large amounts on alcoholic drink. Some of this was local production, for example toddy or Mandalay (Burma) or Tiger (Malaya) beer, but imports, both of beer and alcohol, made up a considerable amount of consumption (Bennison, 1928, pp. 32–33). The beer and ale component of our index may affect prices by including only imports and in this regard tilt the index too far in the direction of urban consumption. However, beer and ale in our index ﬁgure less prominently than in Bennison’s where for Tamils, Telugus (from the Vizagapatam district of Madras) and Uriyas (from Madras’s Ganjam district) alcohol is more than 25% of expenditure. Textiles are the index’s other main non-food component and represented by the equally weighted price of white and grey shirting imports to Thailand. Cotton was the predominant imported textile material used in Southeast Asia. Our index’s 9% overall weighting is because clothing and bedding were made of similar materials and because cotton goods must be taken to represent all other textile materials (Bennison, 1928, p. 68). Kerosene (4%) was important as fuel and in cooking, including the cooked food bought

Globalization and Integration in Asia

339

from hawkers. Tobacco (3%) is raw tobacco except for 1914–1919 when it is cigarettes. The weighting takes into account both imported cigarettes and locally manufactured smoking materials such as cheroots and the two-for-a penny cigarettes which in Malaya gained popularity in the inter-war period and especially so at times of economic downturn. We have no data for rent but that omission is not so serious as might be imagined, since immigrants tended to club together in barrack housing. It seems likely that for unskilled workers in the three countries drink to some extent substituted for rent in the sense of helping to make up for the poor living conditions associated with low rents. Data sources: Rice: Burma: 1880–1931, Rangoon export price of ngatsain rice taken from India, Department of Commercial Intelligence and Statistics (1933, p. 10). Ngatsain grain is a group of rice classiﬁed as bold, deﬁned as a grain broad in proportion to its length. It constituted the bulk of rice exported from Rangoon and Bassein and was known everywhere as ‘‘Burma Rice’’. Cheng (1968, pp. 37–38), 1932–1939: Saito and Lee (1999, p. 98). Malaya: Huff (1994, pp. 373–381) and for 1928–1930: Malaya (1930–1937 (issue for 1930)). Thailand: Feeny (1982, pp. 127–128). Dried ﬁsh and sugar: Huff (1994, pp. 373–381) and for 1928–1932: Malaya (1930–1937 (issues for1930 and 1932)). Tea, beer and ale, kerosene, tobacco: 1880–1919: Shein (1964, pp. 223–233), Burma (1912–1913, 1922–1923), Malaya (1922–1938 and speciﬁcally issues for 1926, pp. 24–25; 1930, pp. 23–24; 1935, pp. 35–36; 1939, pp. 35–36). White and grey shirting: Ingram (1964, pp. 123–124). Data are not available for white shirting for 1886–1888, and 1890–1894 and for grey shirting in these years and also 1889. Where data for Thailand are not available we use the price of grey shirting imports at Calcutta taken from India, Department of Commercial Intelligence and Statistics (1933, p. 9). India Wages: Data for 1873–1907, 1911, 1916, 1918, 1921, 1926, 1928, 1931, 1936. Immigrants from India to Southeast Asia were very largely unskilled working age males previously engaged in agriculture, usually as laborers (Baxter, 1941, p. 47). Typically at least four-ﬁfths and generally an even higher proportion of Indian immigrants to Malaya came from South India. Most were low caste Tamils from the Madras Presidency (Sandhu, 1969, pp. 159–162). Large numbers of immigrants to Burma were from Madras but Bengal was also a signiﬁcant source of labor. However, since most Bengali immigrants traveled on foot to Burma and do not appear in Burma’s immigration statistics, we use Madras wages only in analyzing immigration into Burma. Data sources: For 1873–1907 the Government of India published as India, Director-General of Commercial Intelligence (1902–1923) (see 1902,

340

GREGG HUFF AND GIOVANNI CAGGIANO

pp. 264–283, 1908, pp. 174–191) average monthly wage for agricultural laborers in seven districts in Madras (Ganjam, Vizagapatam, Bellary, Tanjore, Tirunelveli, Salem, and Coimbatore). Together these districts accounted for some two-ﬁfths of the 1901 Madras population of 38.2 million. The seven districts do not cover all parts of the Presidency from which immigrants left but were typical of emigrant areas. For a map of the districts from which South Indian immigrants to Malaya originated, see Sandhu (1969, p. 164). In all but one year the Madras statistics specify a single wage rather than the range of wages often given in wage data for districts elsewhere in India. We weight nominal Madras wages by the 1901 population share in each of the seven districts to measure average agricultural wages in the Presidency. For 1911 and 1916 wages are the population-weighted average of ﬁve districts (Coimbatore, Madurai, Tanjore, Salem, and Tiruchirapalli) which together accounted for 24.5% of the population of Madras. Data are from Madras (1911–1941, wage censuses for 1911 and 1916) and United Kingdom (1931, vol. 7, part 1, p. 301). The 1921 wage is calculated on the same basis as wages for 1882–1907 using data from the 1921 Wage Census for other agricultural laborers. Madras (1911–1941, census for 1921, pp. 16, 18). For 1918 and 1928 wages are from United Kingdom (1931, vol. 7, part 1, p. 296). For 1926, 1931 and 1936 wages are based on the wage censuses for those years and the average wage for ﬁeld laborers. Madras (1911–1941, censuses for 1931, p. 2 and 1936, p. 2). The 1926 wage differs somewhat from, but is consistent with, the average wage of Madras agricultural laborers in United Kingdom (1931, vol. 7, part 1, p. 4; see also vol. 7, part 2, p. 2; 1928, vol. 3, p. 314). The nominal wage for 1882 is for that year and from India, DirectorGeneral of Commercial Intelligence (1902–1923) from which see the report for1902. It is the average of monthly wages for agricultural laborers in seven districts in Madras India Prices: The price index used to express nominal as real wages is a weighted average of the Madras retail price of the four main foodgrains. These are rice and three less expensive coarse grains, namely jawar (cholum), bajra (cambu), and ragi. Our index uses the four grains, weighted by the average acreage in Madras of each crop in 1898/99–1900/01, in preference to rice only because its consumption was by no means universal, especially among poor classes. Kumar (1983, p. 235) observes that around the turn of the century a sign of workers being better off in some parts of the Presidency was that they could afford to eat rice. In times of famine or

Globalization and Integration in Asia

341

distress the price of coarse grains rose disproportionately to rice, and would cause real wages to move differently than suggested by rice prices alone. For discussion of prices in India, see also McAlpin (1983). Data source: Madras (1950, pp. 59–60). China Wages and Prices: Data for 1875, 1877–1878, 1880–1892, 1900–1935. Chinese immigrants to Thailand and Malaya came overwhelmingly from Southeastern China. Almost all originated from the two coastal provinces of Kwangtung and that part of Fukien centered around the port of Amoy. Other emigrants were mainly from the island of Hainan south of Kwangtung and the province of Kwangsi bordering on Kwangtung to the east (Skinner, 1957, p. 35; Malaya, 1932; Chen, 1939, pp. 261–270). Emigrant areas of Kwangtung and Fukien correspond closely to J. L. Buck’s double-rice cropping area in the two provinces. Also included in this zone are parts of the neighboring provinces of Kwangsi and Kiangsi. (Buck, 1937b, p. 10 and compare with the maps in Skinner, 1957, pp. 34–36. Rice was not, however, uniformly important throughout the doublecropping rice zone. Some rural areas around Canton and Swatow, two of the main Kwangtung emigrant ports, were deﬁcient in rice. These two ports and the ports of Amoy and Foochow in Fukien were major inlets for rice imports from abroad and so helped to link China to the world rice market. Freedman, 1958, pp. 9–10; see also, Brandt, 1989, pp. 16–20.) Wage data for Southeastern China are notoriously sparse. As well as Buck’s well-known wage series, we utilize ﬁve further series to represent wage movements in Southeastern China. Where possible we average wage series to try to ensure as representative an index as possible. The exception to this averaging is the wage series for Peking unskilled labor. Buck compiled money wages for a year’s farm labor for seven counties (hsien) in the three provinces of Fukien, Kwangtung, and Kwangsi. For ﬁve of the seven counties and for each of the three provinces, the data cover all but a few years during 1906–1933 and taken together extend to the entire period. The seven districts include some 273,900 households. We weight Buck’s data by the number of households in each hsien. As for wages, we average price indexes if possible. Evidence suggests, however, that at least for most of the period of our study, and even when price information is not abundant, differences in prices were probably not too great because strong regional links forged through a network of small markets allowed national price movements within China. (Rawski draws on the evidence of Brandt (1985) of strong interregional price links and cites an unpublished study by Schram as well as prices for a number of commodities and services including farm labor, draft animals, and rural land. Schram

342

GREGG HUFF AND GIOVANNI CAGGIANO

found price changes passed along to numerous minor markets throughout China. As a result price movements parallel to national price averages occurred in most localities at most times. Rawski, 1989, pp. 295, 325; Myers & Wang, 2002, pp. 580–591, 612.) Buck’s wages are deﬂated by an average of three alternative price indexes: (i) a price index from Chang (1932, 1933 and available in Buck, 1937b, p. 151) (ii) an index from Buck (1937b, p. 151) for retail prices paid by farmers for commodities used in living and production. It is the average of seven (but for 1907–1911 between three and six) localities in the double-cropping rice region; (iii) Brandt’s price index for nonagricultural goods. For 1906–1912 his index is for handicraft cloth, yarn, coal, and sugar and for 1913–1936 includes cotton cloth, yarn, kerosene, coal, sugar, cigarettes, groundnut oil, iron, steel, and tin (Brandt, 1989, pp. 103–104). In index (i) for 1906–1909 data for retail prices paid by farmers are not available and ﬁgures are for prices received by farmers linked in 1910 to the index for prices paid. The other ﬁve wage series used are for the daily wages of Canton porters 1882–1891 (China, Imperial Maritime Customs, 1882–91 to 1922–31, report for 1882–1891, p. 562), which we convert to real wages by constructing a weighted price index for Canton prices of rice, tea, salt, oil, and ﬁrewood given on p. 561; two series for daily Peking unskilled real wages, for 1865–1900 from Gamble (1943, p. 72), and for 1900–1925 from T’ien-p’ei and Gamble (1926, p. 106); wage and price indexes for 1912–1927 for Canton laborers (Kwangtung Government, China, 1928); and a series for wages of farm year labor for 1910–1935 in Wuchin, Kiangsu (Lewis & Wang, 1936, p. 86). These ﬁve series and Buck’s wage data are used as follows. The average growth rate for 1882–1891 for Canton porters wages was calculated and this series extrapolated to 1881 and 1892 to obtain an overlap with Peking unskilled labor. We then calculated average growth for Peking unskilled labor between 1881 and 1892. Information on percentage wage changes from Canton porters was used to adjust growth factors of Peking unskilled labor such that the dynamics of the indexes matched, but the level in 1892 corresponded to the pre-1882 Peking series. This yielded an interpolated wage series for 1882–1906 that accounts both for the levels information from Peking unskilled labor and the dynamics of Canton porters wages. For 1906–1910 we used the index based on Buck, Chang, and Brandt. In addition to this index, we utilized for 1910–1912 the 1910–1935 series for farm wages. For 1912–1925 our index is an average of three series, namely the Buck, Chang, and Brandt index, Canton wages, and the 1910–1935 farm wages. For 1925–1927 our index includes Canton laborers, Buck, Chang,

Globalization and Integration in Asia

343

and Brandt, and the farm wages, and for 1928–1933 the last two of these. Throughout we adjust the level of all the averages (possible because of overlaps) and splice them. Values for 1934 and 1935 are obtained by using the growth rates of the farm wages for these years. Price data are as cited above and see additionally Wang (1972, pp. 357–358), Chou (1963, p. 243), Chang (1958, p. 371). The nominal wage for 1882 is for that year and from China, Imperial Maritime Customs (1882–91 to 1922–31, report for 1882–1891, p. 562). It is the daily wage of Canton porters assuming 24 days of employment per month. United States: Wages are rates paid for common or unskilled labor deﬂated by an index of consumer prices, both from David and Solar (1977, pp. 16, 59). The nominal wage for 1882 is for 1880 and from Lebergott (1964, p. 541). The wage is the average daily earnings for a common laborer assuming 24 days employment per month. United Kingdom: For 1880–1914 wages are agricultural wages for England, Wales, and Scotland and for 1920–1938 for England and Wales only. Data for 1915–1919 are estimated by applying growth rates in the Williamson wage series which is for adult males in manufacturing. Nominal wages are deﬂated using the Saurbeck-Statist price index. Sources are: Mitchell and Dean (1962, pp. 350–351, 474–475) and Williamson (1995, pp. 165–166). The nominal wage for 1882 is for 1886 and from Hunt (1973, p. 70). The wage is a laborer’s wage for a nine-hour day in the Midlands assuming 24 days employment per month. Germany: Wages are for unskilled building workers. For 1880–1913 these are an average of wages in the three cities of Berlin, Nuremberg, and Rostock and for 1924–1939 for all cities. Nominal wages are deﬂated by a cost of living index. Data are from Bry (1960, pp. 325–326, 335–337). The nominal wage for 1882 is for that year and from Bry (1960, p. 339). The wage is the average weekly wage for unskilled building workers in Berlin, Nuremberg, and Rostock assuming four weeks work per month. France: Data are an index of real wages for workers (ouvriers) in Singer-Ke´rel (1961, pp. 540–541). For discussion of the 213-commodity price index used as a deﬂator and the choice of base year in the index, see pp. 84, 276–283. The 1882 nominal wage is for that year and taken from France (1887, pp. 382, 395) and Simiand (1932, p. 23). The wage is the daily wage for male agricultural laborers, assumes 24 days employment per month, and is the weighted average of the winter, or outside harvest, wage (eight months) and the summer wage (four months).

344

GREGG HUFF AND GIOVANNI CAGGIANO

Terms of Trade The terms of trade are an index of the price of exports divided by an import price index with both indexes weighted to reﬂect commodity shares in trade. Recorded ﬁgures for imports and exports of individual countries or regions are used where possible but it has been necessary also to use world prices. Checks showed that country-speciﬁc and world prices are nearly identical. Burma: For 1886–1915 terms of trade are from Shein (1964, pp. 223, 232) and include 14 main exports and 53 imports weighted by the proportion of each item in a base year of 1890–1892. Re-weighting for 1911–1912 as a base yielded almost unchanged indexes for both exports and imports (pp. 211–213). Import prices for 1882–1885 are a weighted average of textile prices (0.80) from the Saurbeck-Statist index and Lewis’s index of the price of manufactures (0.20), and for 1916–1936 a weighted average of the price of consumer goods imported to Burma, taken from Hlaing (1964a, pp. 147–148) and, to reﬂect imports of industrial goods, Lewis’s index of manufactures linked to US industrial prices at the 1913 overlap for the missing years of 1914–1920. The consumer goods price index from Hlaing consists of foodstuffs, vegetable oil, sessamum, salt, soap, cotton yarns, grey, white and colored cotton piece goods, silk, and woolen piece goods. Sources are Hlaing (1964a, pp. 147–148); Textiles: Mitchell and Dean (1962, pp. 474–475); Industrial goods: Lewis (1969, pp. 49–50) and United States (1970, part 1, p. 199). For exports in years when Shein’s index does not exist, export prices are a weighted average of four commodities of which rice is the most important. The four commodities are rice: 1882–1885 (1.0); 1915–1921 (0.70); 1922–1929 (0.65); 1930–1936 (0.60); teak: 1915–1921 (0.15); 1922–1929 (0.10);1930–1936 (0.05); tin: 1915–1921 (0.03); 1922–1929 (0.04); 1930–1936 (0.06); petroleum: 1915–1921 (0.12); 1922–1929 (0.21); 1930–1936 (0.29). Sources for exports are, rice: Until 1931 data are for the Rangoon export price of ngatsain rice from India, Department of Commercial Intelligence and Statistics (1933, p. 10) and thereafter the price of all Burma’s rice exports (in practice almost all via Rangoon) from Saito and Lee (1999, p. 98). Teak: Wilson (1983, pp. 212–217); Petroleum: Huff (1994, pp. 372–378). Tin: International Tin Research and Development Council (1939, p. 52); International Tin Study Group (1953, p. 256). Export and import index weighting are based on Shein (1964, pp. 212–217) and Hlaing (1964a, pp. 110, 112).

Globalization and Integration in Asia

345

Malaya: Imports are a weighted average of rice (0.50), textiles (0.20), and industrial goods (0.30). Exports are a weighted average of tin 1882–1910 (1.0); 1911–13 (0.85); 1914–1936 (0.254) and rubber 1911–13 (0.15); 1914–1936 (0.746). For discussion of export weights see Huff (2002, pp. 1093–1094). Sources are tin: International Tin Research and Development Council (1939, p. 52), International Tin Study Group (1953, p. 256). Rubber: Drabble (1973, p. 213); 1896–1939: McFadyean (1944, p. 239). For 1907–1910 prices are for ﬁrst grade plantation crepe and thereafter for London average standard quality. The two qualities are closely comparable. Rice: Singapore imported most of its rice from Thailand and data are the Thailand export price of rice from Feeny (1982, pp. 127–128). Textiles: the Saurbeck-Statist index for textile ﬁbers was preferred because it offers better coverage given the wide variety of Malaya’s textile imports. Mitchell and Dean (1962, pp. 474–475). Industrial goods: Lewis (1969, pp. 49–50), United States (1970, part 1, p. 199). Thailand: Exports are a weighted average of rice, teak, tin, and rubber. Export weightings are, rice: 1882–1895 (1.0); 1896–1910: (0.80); 1911–1921 (0.79); 1922–1934 (0.75); 1935–1936 (0.66); teak: 1896–1910 (0.10); 1911–1921 (0.06); 1922–1934 (0.06); 1935–1936 (0.05); tin: 1896–1910 (0.10); 1911–1921 (0.15); 1922–1934 (0.14); 1935–1936 (0.17); rubber: 1922–1934 (0.05); 1935–1936 (0.12). Export weights are based on Ingram (1971, p. 94). Imports are a weighted average of the price of white and grey shirting (weighted equally) and industrial goods. Import weightings are 0.70 shirting and 0.30 industrials. Sources are, rice: Feeny (1982, pp. 127–128. Teak, tin, and rubber: Wilson (1983, pp. 213–217). Imports: white and grey shirting, Ingram, 1964, pp. 123–24. Data are not available for white shirting for 1886–1888 and 1890–1894 and for grey shirting these years and also 1889. Where data are not available the index is linked to the Saurbeck-Statist price index for textile ﬁbers from Mitchell and Dean (1962, pp. 474–475). Industrial goods: 1875–1912, Lewis manufactures from Lewis (1969, pp. 49–50) linked with 1913 overlap to US industrial commodities in US Department of Commerce (1970, part 1, p. 199). Madras: Export prices are the weighted average of sugar (0.04), tea (0.09), hides (0.30), goat skins (0.07), sheep skins (0.04), castor oil (0.21), and raw cotton (0.25). Madras weights are based on discussion and data in Baker (1984, p. 110). Data are from India, Department of Commercial Intelligence and Statistics (1933, pp. 10–13). Import prices, and for 1932–1936 export prices, are for India as a whole and from Bhatia (1969, pp. 424–426).

346

GREGG HUFF AND GIOVANNI CAGGIANO

Southeastern China: Silk and silk fabrics dominated exports from Kwangtung and tea those of Fukien. The export price index for Southeastern China includes these three goods and is weighted according to relative shares of the two provinces in total exports. Weights are, 1875–1888: raw silk 0.50; silk fabric 0.16; and tea 0.34; 1889–1900: raw silk 0.56; silk fabric 0.19; and tea 0.25; 1901–1925: raw silk 0.64; silk fabric 0.20; and tea 0.16; 1925–1936: raw silk 0.60; silk fabric 0.20; and tea 0.20. For discussion and data on export composition in the two provinces, see Lin (1997, pp. 63–88), Lyons (2003, pp. 121–152), and China, Imperial Maritime Customs (1879–1939). Sources are, China raw silk and silk fabric exports: Lieu (1941, p. 265). Tea: Lyons (2003), CD spreadsheet; 1890–1938 and the New York price of Formosa tea in Commodity Research Bureau (1939, p. 348). Import prices are for China as a whole and from Hsiao (1974, pp. 273–275). These statistics are Nankai’s index numbers originally published in 1937 but for 1870–1903 incorporate the corrections made by Hou Chi-ming to take account of the change in ofﬁcial statistics after 1903 from the use of market prices to c.i.f. for imports and f.o.b. for exports. From 1904 onwards they are identical to the statistics in Cheng (1956, pp. 258–259).

Population and Population Density Burma: Figures for Lower Burma refer to the 1872 census area. The ﬁgure for 1938 refers to 1941 (Hlaing, 1964b, p. 13). Malaya: The 1881 population is estimated by assuming that population grew from 1881–1891 at the same rate as in 1891–1901. For 1891 and 1901 ﬁgures are estimated for the Unfederated Malay States (UMS) only. Estimation is on the basis of the 1911 census ﬁgure of a UMS population of 899,968 and backwards extrapolation assuming that during both decades the population grew at 0.65% per annum. A basis for this assumed rate of UMS population growth is Dodge (1980, pp. 457–474). Data for 1891–1911 are from Federated Malay States (1902) and Malaya (1911, pp. 18, 95, 1921, p. 18). For 1921 onwards data are from Malaya (1949, p. 39). The 1938 population ﬁgure is an estimate and assumes proportional population growth between 1931 and the 1947 census ﬁgure of 5,848,910. Thailand: For 1881–1901 ﬁgures refer to 1880, 1890, and 1900 (Skinner, 1957, p. 79). Subsequent ﬁgures are from Thailand (1920, 1939–1940) and refer to the Yearbook for 1937–1938 and 1939–1940, p. 46 which are the census returns for 1919, 1929, and 1937.

Globalization and Integration in Asia

347

Madras: Kumar (1965, pp. 120–121) citing Madras Census, 1881; India, Census of India (1922, Part I Report, p. 9); India, Census of India (1932b, Part II Tables, p. 4). Kwangtung and Fukien: Figures for 1881 refer to 1873; for 1891 to 1893; for 1911 to 1913; and for 1931 to 1933. For 1901 and 1921 ﬁgures are estimated by simple interpolation from published ﬁgures for 1893 and 1913 and 1913 and 1933. The 1938 ﬁgures are 1953 populations (Perkins, 1969, p. 212). Area Burma: Andrus (1948, pp. 24–25). Lower Burma consisted of the four southern administrative divisions of Arakan, Irrawaddy, Pegu, and Tenasserim. Malaya: Malaya (1921, p. vi). Thailand: Ingram (1971, p. 7). Madras: India, Census of India (1922, Part II Tables, p. 2). Kwangtung and Fukien: Perkins (1969, p. 219). Exchange Rates van der Eng (1993, p. 28); Carter et al. (2006, vol. 5, p. 5–565); Mitchell (1988, p. 702).

Choice in Economic Contexts, Volume 25: Ethnographic and Theo Enquiries (Research in Economic Anthropology) (Research in Economic Anthropology)

Research in the History of Economic Thought and Methodology: Documents From the History of Economic Thought

Advances in Agricultural Economic History, Vol. 2 (Advances in Agricultural Economic History) (Advances in Agricultural Economic History)

Research in the History of Economic Thought and Methodology, Volume 25B: Documents From the History of Economic Thought (Research in the History of Economic ... History of Economic Thought and Methodology)

Research in the History of Economic Thought and Methodology: A Research Annual, Vol 27A (Research in the History of Economic Thought & Methodology)

Research in the History of Economic Thought and Methodology (Vol 25, part 3)

Research in the History of Economic Thought and Methodology, Volume 25A: A Research Annual (Research in the History of Economic Thought and Methodology) ... History of Economic Thought and Methodology)

Research in Economic History, Volume 25 (Research in Economic History)