Research in Economic History, Volume 22, Volume 22

CONTENTS LIST OF CONTRIBUTORS

vii

INTRODUCTION Alexander J. Field

ix

FROM FORAGING TO FARMING: THE SO-CALLED “NEOLITHIC REVOLUTION” Frederic L. Pryor

1

THE PRICE HISTORY OF ENGLISH AGRICULTURE, 1209–1914 Gregory Clark

41

THE GROWTH OF WORLD AGRICULTURAL PRODUCTION, 1800–1938 Giovanni Federico

125

THE GREAT DEPRESSION AS A CREDIT BOOM GONE WRONG Barry Eichengreen and Kris J. Mitchener

183

THE LENGTH AND THE DEPTH OF THE GREAT DEPRESSION: AN INTERNATIONAL COMPARISON Jakob B. Madsen

239

THE DECLINE AND RISE OF INTERSTATE MIGRATION IN THE UNITED STATES: EVIDENCE FROM THE IPUMS, 1850–1990 Joshua L. Rosenbloom and William A. Sundstrom

289

v

LIST OF CONTRIBUTORS Gregory Clark

Department of Economics, University of California, Davis

Barry Eichengreen

Department of Economics, University of California at Berkeley

Giovanni Federico

Department of History and Civilization, European University Institute

Jakob B. Madsen

Institute of Economics, University of Copenhagen

Kris J. Mitchener

Department of Economics, Santa Clara University

Frederic L. Pryor

Department of Economics, Swarthmore College

Joshua L. Rosenbloom

Department of Economics, University of Kansas

William A. Sundstrom

Department of Economics, Santa Clara University

vii

INTRODUCTION Volume 22 of Research in Economic History contains six papers. Three are on agriculture and two on macro issues related to the Great Depression. A concluding paper examines trends in interstate migration in the United States. Fred Pryor begins the volume with a provocative exploration of the degree to which the Neolithic revolution was in fact revolutionary. Pryor argues for a much less sharp break with the past than has been commonly asserted. He maintains, in particular, that hunter-gatherer methods of procuring subsistence persisted alongside a continuum of agricultural practices. His evidence is drawn largely from records of surviving hunter-gatherer societies. Moving forward 10 millenia, Gregory Clark provides details of his construction of an annual price series for English net agricultural output from 1209 to 1914. Clark incorporates fresh archival material with existing published series, using consistent methods to build and aggregate 26 component series. In the third paper on farming, Giovanni Federico estimates world agricultural production from 1800 to 1938. He concludes that output grew more rapidly than population, and did so on all continents, although more rapidly in countries of Western Settlement and in Eastern Europe than in Asia or in Western Europe. Federico also finds that output grew faster before World War I than in the inter-war years, and exhibited over time an increase in the share of livestock products. Continuing into the twentieth century, we have two papers on the Great Depression. First, Barry Eichengreen and Kris Mitchener explore the degree to which the seeds of economic downturn were sown during the 1920s, particularly through “excessive” credit creation. The authors develop quantitative measures of credit expansion and ask how well these indicators account for “uneveness” in the twenties expansion as well as the depth and severity of the depresion in individual counties. They complement this macro analysis with sectoral studies of real estate, consumer durables, and high-tech sectors. Jakob Madsen’s contribution is also based on an examination of Depression macro history in a number of countries. But his focus is on output and labor rather than credit markets. He explores the perrenial questions of how sticky were wages and prices and whether such stickiness played a significant causal role in the rise in unemployment. Contrary to many models that assume or assert that prices are inherently more flexible than nominal wages, Madsen finds the reverse: prices ix

x

adjusted slowly to changes in nominal wages, and this stickiness played a role in propagating economic depression. Finally, Josh Rosenbloom and Bill Sundstrom explore changing rates of interstate migration by examining individual-level data from population censuses available in the Integrated Public Use Microdata Series (IPUMS). Their central finding is that propensities to migrate within the United States have traced out a U-shaped pattern, tending to fall between 1850 and 1900 and then, during the twentieth century, rising until around 1970. Potential contributors, particularly younger authors, are encouraged to consider Research in Economic History as an outlet for their work. We look forward to continuing to publish innovative, well written, and carefully considered contributions to economic history, providing a niche which complements other specialized field journals such as the Journal of Economic History, Explorations in Economic History, and the Economic History Review. We can, for example, to a greater degree than these publications, publish data rich contributions and those which run to somewhat greater length. Prospective contributors are urged to contact the editor for information on submission requirements. Alexander J. Field Series Editor

FROM FORAGING TO FARMING: THE SO-CALLED “NEOLITHIC REVOLUTION” Frederic L. Pryor ABSTRACT This essay provides evidence that the invention of agriculture was not a dramatic technological advance in the history of humankind and that agriculture was quite consistent with nomadic hunting and gathering. The available clues also suggest that exact origins of agriculture do not seem important. Rather, the crucial question is why certain societies dramatically increased their dependency on agriculture for subsistence two to ten millennia ago. Unfortunately, most of the major theories purporting to explain the neolithic revolution – either the origins or the spread of agriculture – are either untestable or inconsistent with the available evidence. What is at stake for economic historians is to rethink the process of the adoption of agriculture using a multi-causal approach.

INTRODUCTION Humans (genus Homo) have been on earth in one form or another for several million years and anatomically modern man – the kind of person who wouldn’t look out of place shopping at the local Wal-Mart – about 100,000 years. We learn

Research in Economic History Research in Economic History, Volume 22, 1–39 Copyright © 2004 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(04)22001-8

1

2

FREDERIC L. PRYOR

from archaeologists, however, that “agriculture” suddenly appeared only about 10,000 years ago and that, during the next 8,000 years, became the primary source of food for the vast majority of human societies on the planet. This rapid change – called the “neolithic or agricultural revolution” – raises many questions. In this essay I present evidence suggesting that agriculture, at least as a minor source of food, probably existed long before 10,000 years ago and that domestication of plants and animals did not represent a dramatic technological advance. This argument means that the important question is why agriculture spread in the last ten millennia; and that we should not waste our time exploring how or why it was invented. The first step is to define “agriculture,” a conceptual problem that is more complicated than it first appears. I then examine three major approaches for explaining the invention and/or spread of agriculture that have received particular attention in the last quarter century – geographical, social/political, and demographic – and show that the available evidence does not offer convincing support for any of them. Finally, I argue that the spread of agriculture was a process with quite different causes in different places so that it is impossible to specify its necessary and sufficient conditions. A more promising approach is to examine the costs, benefits, and risks of agriculture in specific situations with particular attention to a positive feedback process involving demographic growth and sedentarism. Before plunging into the analysis, we must consider what kind of evidence is available. The archaeological record is skimpy and rests primarily on finding domesticated seeds or agricultural implements at particular sites which can be dated. Comparison of such findings at various sites allows a few inferences about the process of the adoption or spread of agriculture, particularly when this can be correlated with changes in climate. But such physical evidence seldom reaches back to the period when these societies relied completely on foraging (hunting, gathering, or fishing). It also does not allow us to learn many correlates of reliance on agriculture to these societies’ social or political characteristics or to their property relations or distribution mechanisms. While I bring certain relevant archaeological evidence into the discussion, particularly regarding climate change, I find it necessary to pursue the topic differently. More specifically, I rely on an examination of foraging societies described by Westerners from the 16 to 20th centuries. Given the well-known dangers of generalizing from cross-section evidence from the relatively recent present about events occurring over a long period of time in the far-distant past, this means that I cannot draw any definite conclusions. Rather, I can only make a plausible case. Nevertheless, my approach can be justified in several ways. There is little to suggest that endogenous causal mechanisms underlying important economic changes in these foraging societies in the distant past were much different than in

From Foraging to Farming

3

the more recent present. This means, however, that although most of the societies in my sample were relatively isolated from the West, we must still be on the lookout for different exogenous forces of change that might have influenced the spread of agriculture. Nevertheless, the archaeological evidence of certain foraging societies such as the !Kung San (Ju/’hansi) of the Kalahari desert shows relatively little change in the society over the past three or four centuries and the rock drawings depicting their life centuries ago show activities quite similar to those existing in 1950. Some of the societies in my sample have made technological advances in the past millennia, but there is no evidence that their economic systems were significantly different than in the recent past.1 The societies used in this study come from the “Standard Cross-Cultural Sample” (SCCS), a group of 186 preindustrial societies from all over the world (Barry & Schlegel, 1980). Each of these is pinpointed in location and time (mostly in the last 400 years) and represents a different “cultural area.” Most of the important data I obtained by consulting more than 500 ethnographic sources; other data series come from the 1,700 different series collected by Divale (2001) for the SCCS.

Agriculture and its Distant Relatives Determining whether a society practices “agriculture” raises some problems. Agriculture involves both modifying the environment and manipulating the genetic material of plants or animals (i.e. domestication) to increase the marginal labor productivity of obtaining food. For plant production agriculture involves several distinct tasks: preparing the land and planting; certain nurturing activities such as fertilizing, irrigating, weeding and warding off predators; and, finally, harvesting and the selection of seeds to store for next year. For meat production, agriculture involves breeding, feeding, and protecting the animals. To decide where the line should be drawn between agriculture and related subsistence activities, some distinctions need to be made. Proto-Plant-Production Proto-plant-production2 places a greater emphasis on managing the environment for plant production, rather than nurturing the crops or deliberate manipulation of the genetic materials of the plants. As a result, it also requires only a few of the tasks noted above. Several types of proto-agriculture deserve mention:
Fire-stick agriculture. Certain tribes in Australia, North America, and elsewhere (Keeley, 1995; Mulvaney & Kammings, 1999, pp. 60–62; O’Dea, 1992) set fire to fields to encourage the growth of new plants by burning away the underbrush

4









FREDERIC L. PRYOR

and the coarse tops of the old grass. The ash from the burning also served as a fertilizer. In some cases, the purpose was merely to raise hunting productivity by attracting grazing animals to the young grass; in other cases, to supply seed and nut crops that would be later gathered for subsistence. Some societies (for instance, Shoshonians in the Great Basin) took this practice a step further by planting wild seeds in the burnt field and later harvesting the crop (Keeley, 1995). Tending tubers. In northern Australia the long yam was a dry season staple. When harvesting it, certain groups would leave the top of the tuber attached to the tendril of the vine to insure that it would grow again in the following year (O’Dea, 1992). Watering fields. Certain foraging societies, such as the Owens Valley Paiute of California, flooded fields containing wild root crops to encourage their growth (Keeley, 1995). Soil aeration. Some Australian aboriginal tribes engaged in extensive turning of the soil to collect edible roots and tubers, a practice which encouraged their growth in the next year (Mulvaney & Kammings, 1999, p. 87). Semi-sowing. In various parts of the world some societies have broadcast wild seeds on fresh alluvial silts (Watson, 1995). The Menomini Indians of Wisconsin purposely allowed some of the wild rice they harvested to fall back into the water to insure a crop in the next year (Cohen, 1977a, p. 21). The Kumeyaay tribe on the California-Mexico border transplanted various plants and trees (Smith, 1995).

Quantitative analyses of the various practices of proto-agriculture are rare, with the exception of the work of Keeley (1995, 1999). From a regression analysis of the practice of burning fields among various groups of Shoshone and Paiute of the Great Basin in the United States, Keeley concludes (1995, p. 271): “The most important variables for explaining the intensification of plant exploitation, including cultivation of wild plants, are, in order of decreasing importance, ecological “latitude,” precipitation, and population pressure. Population pressure plays only a minor, but still essential, role in intensifying plant exploitation. Social demand and social complexity are completely irrelevant.” By “social demand” he refers to the need for a food surplus to support higher social classes, a high ceremonial intensity, or give-away feasts. Most early proto-agriculturalists seemed to understand well what they were doing. According to some anthropologists, we should no longer consider gatherers and hunters as a ragged and scruffy band of nomads. Kent V. Flannery (1968) declares that, instead, “they appear as a practiced and ingenious team of lay botanists who know how to wring the most out of a superficially bleak environment. . . . We know of no human group on earth so primitive that they are ignorant of the

From Foraging to Farming

5

connection between plants and the seeds from which they grow and this is particularly true of groups . . . [utilizing] seasonal plant resources.” Other evidence along these lines is discussed by Cohen (1977a, pp. 18–22). But we will never know how the original connection between seeds and plant growth came about – a seminal insight in the true sense of the term – for this moment is lost in the shrouds of history. We can speculate (as have Rindos, 1984; Tudge, 1998) that proto-plantproduction was probably around for tens of thousands of years before the so-called “agricultural revolution” of 10,000 years ago and, perhaps, began in the middle Paleolithic. Unfortunately, neither the sites of proto-plant-production nor nomadic-plant-production (see below) leave sufficient archaeological traces to support directly this conjecture. It seems likely that proto-plant-production preceded full scale agriculture, because it allowed the accumulation of knowledge about plant production; but this conjecture is far from proven. Domestication and Full-Scale Plant-Production Aside from most – if not all – of the agricultural activities enumerated above, full-scale agriculture is supposed to include the development of domesticated plants and animals, that is, flora and fauna with a genetic makeup different from their “original” states so that they are more useful for humans. For seed bearing plants, for instance, domestication can mean larger seeds, compaction of seeds on the plant, less scattering of seeds before harvesting, thinner seed coats, seeds which more quickly sprout or are more resistant to drought or to the impact of weeds, and so forth. For tubers domestication might mean, for instance, larger edible parts, earlier maturation, low content of poisonous or acrid substances, or higher sugar content (L´eon, 1977). In recent years, experimental archaeologists and others have arrived at two important insights. First, domestication did not need to be a conscious undertaking and could have been an accidental outcome of various activities such as harvesting (Anderson, 1991, 1999; Rindos, 1984). Second, in the words of Jack Harlan (1999), “Domestication of most cereals is relatively simple and straightforward.” As evidence, he cited various experiments where certain grasses were fully domesticated in one year; and certain rices, in three. For domestication to occur, a genetic variation among wild plants and a culling of those with certain characteristics is necessary. MacNeish (1991, p. 9) discusses an hypothesis advanced by various scholars that most domesticated plants are “open-habitat plants.” In disturbed soils (for example, near human habitation), either they more readily mutate or mutant strains are more likely to flourish. This biological feature would have also aided domestication. Further, domestication is not an either/or proposition, but rather a matter of degree. Rindos (1984) also argues that domestication must really be considered a coevolutionary process between plants/animals and humans. In any case, domestication was not an exclusive

6

FREDERIC L. PRYOR

feature of full-scale agriculture but could occur with both proto-plant-production and nomadic-plant-production (see below). Some Conclusions The evidence discussed above suggests that the invention of agriculture was not difficult. The discovery by humans of the seed mechanism for plant growth, by way of contrast, was a critical technological breakthrough, and it undoubtedly occurred far earlier than 10,000 years ago. But we will never know when or where it took place. We also will not know if proto-plant-production was a straightforward development from this initial insight about seeds because the various techniques used in this subsistence activity might have been discovered separately and accidentally. For the rest of this essay, I focus primarily on full-scale agriculture. Nevertheless, one paradox deserves immediate note because it provides a theme for further analysis: While it is likely that most forms of proto-plant-production increased labor productivity in obtaining food and that, because of seasonality factors, the same can probably be said of nomadic-plant-production, this is not necessarily true of full-scale agriculture. Rindos (1984, p. 94) cites evidence that extensive natural stands of wild wheat in the Near East were quite as productive as many of the older varieties of domesticated wheat. Cohen (1977b) and Hayden (1993, p. 220) argue that those early societies relying primarily on agriculture were less well-nourished than their foraging cousins and just as prone, or more so, to famine (more evidence on this below). Time-budget data also show that agriculturalists often worked much longer for their food than foragers.3 So why should a society increase its reliance on agriculture for subsistence?

Hypothesized Environmental/Geographical Prerequisites of Agriculture Ecological and climatic conditions have received considerable attention as explanations for the origins or spread of agriculture. The evidence presented below suggests that their importance is probably much less than commonly believed. I first look at some simple indicators of environmental conditions which are believed favorable to agriculture and then explore some more subtle arguments about biological/geographic determinants. Simple Environmental Indicators for the Presence of Agriculture Students of prehistory fall into two opposing camps with regard to the impact of environmental conditions on the origins or spread of agriculture. Some archaeologists and anthropologists claim that agriculture was more likely to develop in, or spread to, rich environments with abundant per capita plant

From Foraging to Farming

7

resources (see, for instance, various papers in Price, Gebauer (Eds), 1995) and they offer several interesting arguments: Needy and miserable societies were not inventive, so only a society living in a rich environment would have the leisure to experiment with plant production. Further, the benefit/cost ratios of initial agriculture were low, so agriculture could only have been worthwhile where climatic conditions enhanced its benefits (Hayden, 1995); or where there was great diversity of plant life; or where plant production increased the productivity of foraging, for example, by providing poisons used for fishing (Sauer, 1969, p. 24). Finally, societies situated in rich environments were more likely to press for ever-increasing food surpluses to finance religious, social, cultural, and political aims (Hayden, 1990). The first two arguments rest on the assumption that the invention of agriculture was extremely difficult, which, as I argue above when discussing proto-plant production, seems unlikely. I deal with the third hypothesis in the next section, and raise some doubts about it as well. The other group of scholars claims that agriculture was more likely to develop or spread in relatively poor environments – the marginal areas – because the stress on land resources was higher (see, for example, Binford, 1968 or Cohen, 1977a). A telling example is provided by Keeley (1995, p. 261). In his discussion of the diffusion of the maize-bean agricultural complex, he points to several societies which adopted agriculture that lived in poor ecological zones, such as Death Valley. At the same time their neighbors who lived in much richer areas – in this case, the Owens Valley Paiute (who were also economically more developed according to the Carneiro scale) – were content to practice only proto-agriculture and to rely primarily on gathered foodstuffs. To test statistically the validity of either approach requires us to develop an indicator defining a “rich” or “poor” environment and to determine what conditions for agriculture – climate, soil, and terrain – deserve attention. In part, these difficulties can be overcome by examining many series of indicators, each scaled according to their favorability to agriculture. Another problem is to select the proper sample of societies for examining possible correlations between the importance of agricultural products in their diet and the various environmental indicators. Even under extreme conditions certain kinds of agriculture can be practiced: irrigation can offset extreme aridity and one can herd reindeer in very cold climates. To deal with this latter difficulty I have eliminated from the SCCS all societies with less than 200 millimeters of rainfall per year or with an effective temperature (a measure of average temperature that also reflects the length of the growing season; see Kelly, 1995, p. 66 ff.) of less than 11 ◦ C. A final difficulty arises because agriculture might not be necessary for a society which obtains a predominant part of their diet by fishing and the hunting of sea mammals. So I have also removed from the sample all societies with more than a 35% reliance for subsistence on aquatic resources. This left 136

8

FREDERIC L. PRYOR

societies in the SCCS sample, which were then arranged according to their reliance on agriculture (plant production and/or animal husbandry) for their subsistence. Table 1 presents some environmental indicators of the suitability of agriculture. I use three rainfall indicators: (a) precipitation; (b) evapotranspiration (which, roughly speaking, is the water available to plants for maintenance and growth, i.e., rainfall minus the water runoff and the water that percolates into the earth); and (c) the rainfall seasonality, which is the ratio of potential to actual transpiration, where potential evapotranspiration is the maximum amount of water which plants in the area can use for maintenance and growth without becoming waterlogged. The actual rainfall experienced in a region is of less relevance to agriculture than the water available for agriculture, as measured by evapotranspiration, but estimates of the latter requires knowledge of soil conditions and plant cover. Fortunately, climatologists have taken pains to make such estimates. Moving from societies with a total reliance on foraging to total reliance on agriculture, Table 1 reveals that the amounts of precipitation and evapotranspiration show a rise and then fall as dependency on agriculture increases, while rainfall seasonality shows a fall and then a rise. No linear relationship in either direction is apparent, which is not unexpected, given the conflicting arguments about the positive or negative impact of a good climate on agriculture.4 The average annual temperature is of less relevance to agriculture than the effective temperature, which can be calculated from data on the average temperatures in the hottest and coldest months.5 Both temperature series in Table 1 provide little support for those arguing for the importance of environment, since they show no linear relationship with reliance on agriculture for subsistence. Moreover, for three other environmental variables – slope of terrain, soil suitability, and climate – no dramatic differences between the societies with different degrees of reliance on agricultural products for subsistence can also be seen since most of the results are not significantly different from the means of the entire sample. The same can be said for the two combined indicators of these three variables. The message seems clear: Although the various environmental variables certainly had some impact on the importance of agriculture to subsistence of the preindustrial societies included in the SCCS, they do not appear to have been the key constraints to the adoption or spread of agriculture. The exceptions to this generalization were usually societies in extremely arid or cold conditions, which are not included in the sample. The visual impressions gained from the table are confirmed when we attempt to tease out causal relations using a simple regression analysis.6 Change in Environmental Conditions A number of scholars have argued the case for the impact of the environment on the origins or adoption of agriculture with an interesting twist. It is not the

Percentage of subsistence from agriculture (plant production and animal husbandry) Number of societies

0–5%

5–25%

25–55%

55–75%

75–95%

95–100%

Entire sample

14

6

9

45

45

16

135

Rainfall indicators (precipitation and evapotranspiration in millimeters) Annual precipitation 1063 1455 Annual evapotranspiration 754 1205 Rainfall seasonality: index 1.76 1.11

1808 1105 1.07

1755 1096 1.42

1551 953 1.45

1117 824 1.59

1530 986 1.45

Temperature indicators (both in centigrade) Average temperature Effective temperature Ecological suitability for crops Slope of terrain (4 = steeply dissected; 8 = level to gently undulating) Soil suitability (0 = very poor, 8 = very good) Climate (for major crops) suitability (0 = impossible for growing, 8 = very good) Agricultural potential Sum of terrain, soil, and climate suitability (4 = extremely poor; 24 = best possible for agriculture) Lowest value of rating for terrain, soil, and climate suitability (0 = lowest; 8 = best for agriculture)

17.4 16.2

22.3 18.6

22.2 20.2

22.9 20.2

21.1 18.6

21.5 18.3

21.5 18.9

6.8

8.0

6.9

6.9

6.5

6.5

6.7

3.9 6.3

3.0 7.2

4.0 7.6

4.4 7.0

4.6 6.7

4.3 6.6

4.3 6.8

16.9

18.2

18.4

17.7

17.5

17.4

17.6

3.7

3.0

3.9

4.2

4.3

4.2

4.1

From Foraging to Farming

Table 1. Conditions for Agriculture in a Selected Sample of Tribal and Peasant Societies.

Note: These indicators are unweighted averages and are discussed in the text. Rainfall seasonality is potential evapotranspiration divided by actual evapotranspiration.The sample excludes all societies relying for 35% or more of their subsistence on fishing or sea mammals, as well as societies in extreme climatic conditions. Sources and methods are presented in the Appendix.

9

10

FREDERIC L. PRYOR

environment per se, but, rather, a change in the environment that encourages agriculture, by creating stress on land resources. In the 1920s and 1930s, V. Gordon Childe popularized the notion that desiccation in the Near East during a period of several hundred years forced the inhabitants onto oases, and the resulting resource stress encouraged agriculture, particularly animal husbandry. This hypothesis has been vigorously disputed, but recently Bar-Yosef and Meadow (1995) have argued that the cold, dry spell in the Near East in the period from roughly 10,800 to 10,300 years ago (Younger Dryas period) led to contraction of vegetation belts and smaller yields from natural stands of plants. This change, combined with population pressure and several other factors forced an increasing reliance on agriculture and the emergence of farming communities. In several other areas and periods, the reliance of agriculture began shortly after similar dry spells (Hayden, 1995). But some contrary archaeological evidence is also at hand. For instance, Price, Gebauer, Keeley (1995) point out that agriculture spread into Europe north of the Alps when the general climate became warmer, wetter, and less variable than it is today. But favorable climatic conditions may have only aided the spread – but not the origin – of agriculture, so the question about ultimate causality is still open.7 Other aspects of this theme are taken up below in greater depth when I focus on resource stress as a causal factor. At this point we need to turn to more subtle environmental effects. Biological/Geographical Limitations on Agriculture Jared Diamond (1999) presents a different and quite interesting perspective on the impact of the environment. He argues that there are only a limited number of edible and productive plant and animal species for humans and, moreover, these sources of human nourishment have a limited geographical distribution. As a result, agriculture began in just a few places and then diffused outwardly to other locations, aided by the fact that agriculturalists were militarily more powerful than foragers and forced agriculture upon many of the hunting and gathering groups, whose territories they conquered or dominated. Two objections can be raised against this intriguing argument. First, although Diamond is undoubtedly correct that high labor productivity can be expected from cultivation of only a very limited number of food plants, this does not mean that agricultural societies could not have arisen from the cultivation of “inferior” crops. For instance, as Diamond himself notes (1997, p. 100), in the eastern part of the United States agriculture began somewhere around 4500 years ago with the cultivation of such plants as sunflower and goosefoot; and only later did some of these societies adopt maize, which originated from Mesoamerica, as a major staple. Thus, “superior” crops were not necessary for a society to start on the path toward full subsistence dependence on agricultural foodstuffs.

From Foraging to Farming

11

Second, as noted above, the line between proto-agriculture and full-scale agriculture is narrow and despite the obvious diffusion of certain agricultural practices, agriculture could have been independently invented in many more centers than we now know. As shown above, societies with a limited amount of agriculture existed in quite different environments and utilized quite different native plants. Certainly the small number of useful and productive plant species that could be domesticated did not seem to have acted as a constraint on gathering; and many of these gathered plants could have been later cultivated as well. Nevertheless, the low productivity of these plants might have limited the full-scale adoption of agriculture, as well as the development of a greater division of labor, higher complexity, and military success, all of which depend on a food surplus that can support non-agriculturalists. Some Conclusions From comparisons of gross environmental conditions between societies with different degrees of dependency on agriculture for subsistence, it does not appear as though temperature or rainfall, soil conditions, or land slope played a very important role in the origin or spread of agriculture. By way of contrast, in certain situations the change in the climate might have stimulated the spread of agriculture, but many cases can be cited where humans did not adapt their subsistence activities to changing climatic conditions and, as a result, disappeared,8 so that a changing climate was neither a necessary nor sufficient condition for the origin or spread of agriculture. The biological/geographic considerations raised by Diamond appear to tell us little about the origins of agriculture, but they may help us understand its spread.

Hypothesized Cultural/Social/Political Conditions as Causes/Prerequisites/Incentives for Agriculture At the present time most anthropologists and archaeologists favor the hypothesis connecting the origins or spread of agriculture to various social, political, or cultural factors that require the production of a food surplus. Unlike the environmental variables, however, a correlation between such factors and reliance on agriculture for subsistence does not necessarily indicate causation, since the direction of causality is not be clear. Moreover, if a correlation exists, the existence of many important exceptions suggests that the “causal factor” under examination might be neither a necessary nor sufficient condition for agriculture. Fortunately, the lack of a correlation suggests that the variable under examination did not play a causal role; and, in the analysis below, I find little evidence that any of a variety of social/political/ cultural factors were important factors in the origin or spread of agriculture.

12

FREDERIC L. PRYOR

A Cultural Factor: Changes in Mentality For over a century anthropologists and archaeologists have offered a variety of hypotheses based on the notion that the cultures of foraging and agricultural societies are very different and that a change of mentality, arising for unspecified reasons, must have played a crucial role in the origins or spread of agriculture. Understandably, the supporting evidence has usually been very scanty; but two exceptions must be noted. Jacques Cauvin (2000) presents archaeological data to suggest that religious practices (and ideas) in the Near East changed considerably immediately before the rise of agriculture and triggered the shift in subsistence practices. Using a comparison of data from forty archaeological sites in the Near East, however, Fuller and Grandjean (2001) reach the opposite conclusion and argue that an agricultural food surplus preceded, rather than followed, a change in religious ideas. Unfortunately, I have been unable to find systematic archaeological evidence on this matter from archaeological sites in other areas; and the various SCCS series also do not seem to lend themselves easily to a direct test of this hypotheses about the impact of changing ideas. One highly indirect test, however, seems possible. Since societies with higher levels of complexity have a much more developed division of labor (Carneiro, 1970) and since this complexity is highly correlated with an increasing reliance on agriculture (line 12 in Table 2), we might ask whether a specialized priesthood arose before or after the emergence of agriculture. A data series for all of the SCCS societies to test this idea is not available. For societies with less than a 45% reliance on agriculture for subsistence, however, I have coded a proxy variable focusing on medical practitioners, which is a specialty relying on “intangible capital” to help make a living (Pryor, 2003b). More specifically, my variable designates whether medical practitioners or curers receive a fee for their services (1 = no fee; 2 = a small fee; 3 = an economically important fee). The results (line 3 in Table 2) show that such fees existed in many foraging societies, although these curers were usually paid in foraged foods and handicrafts, rather than agricultural goods or money. Since medical practices were closely connected with religious practices in most of these societies, it seems plausible, therefore, that semi-professional priests existed in many societies which relied completely on foraging for their food and that Cauvin’s hypothesis linking agriculture with the rise of a priesthood or intricate religious practices probably does not account for the emergence or spread of agriculture. Social Obligations A number of anthropologists and archaeologists argue that mounting social ceremonial obligations arising in more complex preindustrial societies require a food surplus which can only be obtained by agricultural production.9 More specifically,

Percentage of subsistence from agriculture My codes 1. Number of societies in sample 2. Curers receive a fee for their services (1–3) 3. Presence of potlatch (1–3) 4. Wealth differences (1–3) 5. Social differentiation (1–3) 6. Slavery presence (1–3) (at or previous to pinpointed date) 7. Taxes or levies by political leaders (0–1) 8. Political centralization (0–4) 9. Territoriality (0–5) 10. Private property in land (1–4) Codes by others 11. Number of societies in sample 12. Carneiro complexity scale (0–600) 13. Existence of corporate descent groups (0–1) 14. Social stratification (1–5) 15 Political differentiation (1–4)

0–5%

5–25%

25–45%

45–75%

14 2.1 1.1 1.6 1.1 1.2

6 2.2 1.0 1.4 1.4 1.1

9 2.7 1.1 1.7 1.4 1.4

NOT CODED

0.1 0.6 2.4 1.4

0.2 0.7 1.5 1.0

0.2 1.1 2.2 1.9

14 18 0.2 1.1 1.4

6 26 0.2 1.5 1.7

9 75 0.6 1.3 1.6

75–95%

95–100%

Whole sample 29 2.3 1.1 1.6 1.3 1.3

From Foraging to Farming

Table 2. Unweighted Averages of Social and Political Variables Allegedly Related to Agriculture.

0.1 0.8 2.2 1.5 44 168 0.6 2.3 2.0

40 297 0.8 3.1 2.6

15 415 0.7 3.9 2.9

128 206 0.6 2.5 2.2

Note: The numbers in parentheses are the range of my codes for the particular variable, with the high value indicating an hypothesized favorability for agriculture. The sample excludes all societies in extreme climatic conditions, those relying for 35% of more of their subsistence on fishing or sea mammals, and all nomadic herding societies. Sources of data and scale values are presented in the appendix. The political differentiation indicator (line 15) is an unweighted average of seven different indicators. The meanings of the various indicators are explained in the text. Data sources and details are described in the Appendix.

13

14

FREDERIC L. PRYOR

a socially and politically homogeneous band with few lavish ceremonies or differentiated subgroups, has little need to develop a food surplus except to tide it over a lean period. If, however, people need food not just for the subsistence of their families but also to fulfill new social obligations, cement social alliances, finance ceremonies or give-away feasts, support retainers, maintain prestige or class status, or validate high political position as the “big men,” then a more intensive method of food production than foraging might be needed. For instance, Barbara Bender (1978, p. 206) argues that “the enquiry into agricultural origins is not, therefore, about intensification per se, nor about increased productivity, but about increased production and about why increased demands are made on the economy.” Bender’s approach, however, raises two problems: First, it is unclear to me why these increased needs could not have been met with foraged foodstuffs, especially since anthropologists argue that in many foraging societies, people work relatively short hours (e.g. Sahlins, 1972). Second, it is difficult to find any direct way to test her hypothesis and we must rely on indirect proxies. At one point, Bender (1975, p. 8) notes that agricultural societies tended to be organized on a more permanent corporate basis than foraging societies. If this means that they were also more likely to feature corporate kin group structures,10 then her idea is testable. The data (line 13 in Table 2), however, reveal that corporate kin structures occurred in some societies with little reliance on agriculture (although they were not common), so that this type of social arrangement was not a necessary condition for agriculture. Although they appeared in most – but by no means all – of the societies with a heavier reliance on agriculture, such structures were also not a sufficient condition for this type of subsistence production. Although Bender’s presumed causal relationship cannot be tested in a conclusive fashion with the available evidence, it does not appear promising. Competition for Status Brian Hayden has taken Bender’s rather vague ideas and made them more rigorous. He argues (1990) that “socioeconomic inequalities and competition among complex, economically specialized hunters/gatherers toward the end of the Pleistocene [20,000–11,500 years ago] as well as in the Holocene [11,500 years ago to the present]” is the key. More specifically, “the competitive and feasting aspects of economic rivalry among these complex hunters/gatherers [was] the driving force behind food production [agriculture].” Further, competition over certain food resources is destructive of these resources and, hence, maladaptive, especially if the group lives primarily off of animals which have limited offspring and long maturation periods. Overexploitation of cultivated grasses and legumes is less likely.11 “The eclipse of rigid egalitarianism and sharing that was brought about by the emergence of economic competition (made possible by the effective

From Foraging to Farming

15

exploitation of highly productive r-selected [plant and fish] resources) is possibly the most important development in cultural evolution in the last 2 million years. It can be linked to the emergence of food production, hierarchical societies, craft specialization, slavery, intensive warfare, and many other important cultural traits.” Hayden speculates that the first domesticates were not staple crops, but rather crops with desirable qualities for feasting, for instance, dog meat, specialty roots, or grain for beer. The available evidence on such matters is slim. It is, however, possible to test his general hypothesis in other ways, and three variables in particular deserve our attention.
Feasting and potlatches. Given the ethnographic sources available to me, I found it impossible to code for the importance of feasting in many societies, nor was I able to find any series for feasting compiled by others. Nevertheless, I was able to code for the presence of potlatches (feasts in which property is given away or destroyed) for the SCCS societies relying on agriculture for less than 45% of their subsistence. The results are presented in line 3 of Table 2 (1 = absent; 2 = present but property given away and not destroyed; 3 = property either given away or destroyed). Among this sample of 57 SCCS societies, only eight featured potlatches and, of these, six (all on or near the northwest Pacific coast of the North America) are not included in the table because they rely for 35% of more of their subsistence on fishing or sea mammals. The other two societies with potlatches were the Omaha of Nebraska and the Gros Ventre of Wyoming, both of which featured only an attenuated form of this ceremony. Potlatches did not occur in most of the societies in the sample and such a ceremony does not seem a likely cause for the origins of agriculture. My research on agricultural societies (Pryor, 2003a) suggests that such give-away ceremonies to achieve or maintain social status were also only infrequently present in most societies with a heavier reliance on agriculture.
Social differentiation. My codes for wealth differentiation (line 4 of Table 2) (1 = relatively egalitarian, 2 = some differences in wealth; 3 = considerable wealth differences) are, of course, impressionistic. Nevertheless, the results suggest a certain increase in wealth inequality when a society came to rely on agriculture for more than 25% of its subsistence, but with many exceptions. My codes for social differentiation (line 5 of Table 2) (1 = egalitarian; 2 = social ranks which are not inherited or fixed; 3 = classes with inherited rank) exclude the presence of slavery. These data reveal an irregular pattern for societies with less than 45% reliance on agriculture for subsistence and also do not appear to support the link between social differentiation and agriculture. Finally, my data on slavery either at or before the pinpointed date (line 6 of Table 2) (1 = no slavery; 2 = slaves present, but not extensive, and held primarily by elite;

16

FREDERIC L. PRYOR

3 = more extensive holding of slaves) also show no particular trend, at least at these early stages of agriculture. The codes of Murdock, Provost (1973) (line 14 in Table 2) are consistent with my results in revealing little trend toward social stratification until the society reached a reliance on agriculture for more than 45% of its subsistence.12 Social stratification might have either accompanied or provided an incentive for a greater reliance on agriculture, especially since the food production of subordinates is easier to monitor than in foraging, but the existence of social stratification in some foraging societies and the lack of such stratification in societies more heavily reliant on agriculture suggest that stratification was neither a necessary nor sufficient condition for agricultural production. It is, of course, possible that social differentiation did not appear in those societies with little reliance on agriculture because such a social structure itself spurred a society to move rapidly toward agriculture. This possibility can be investigated by exploring those societies which rely on agriculture for more than 75% of their subsistence and examining whether any can be considered egalitarian (that is, a Murdock-Provost code = 1). In fact, a number can be found: Some were nomadic herding societies,13 but about 8% of the societies in this sample of agricultural societies resided in fixed communities and did not have significant social-class inequalities.14 Thus social differentiation or the rituals of status validation did not appear a necessary precondition for full-scale agriculture.
Political differentiation. The social competition hypothesis can be recast in political terms by focusing on political differentiation and the emergence of a political elite. In line 7 of Table 2, 0 = no tax or tribute collected by the political leader; 1 = a portion of the food appropriated or given to the leader. Line 8 is a composite variable equally weighting four aspects of political differentiation: the political leader’s relative wealth, his (always a male) perceived power, the manner in which he was selected, and whether he received some type of taxes (line 7). The tax variable reveals no significant trend among the societies with less than a 45% reliance on agriculture. The composite variable shows a slight jump in political centralization after societies relied on agriculture for more than 25% of subsistence. The political differentiation variable in line 15 manifests an upward trend only among those societies with more than 45% of their diet coming from agricultural products. For the political centralization variable many exceptions to the trend occur. More specifically, about one in seven of the non-nomadic herding societies relying more than 75% on agriculture for subsistence had a very low political centralization (less than a code of 1.5) on a scale from 1 to 4.15 This means that political centralization was neither a necessary nor a sufficient condition for the emergence of agriculture.

From Foraging to Farming

17

In brief, various aspects of political specialization generally did not emerge until the society had a reliance of at least 45% on agricultural products. Hence, political centralization and the need for a food surplus above subsistence needs that it allegedly created cannot account for either the origins of agriculture nor its spread. Further, a heavy reliance on agriculture occurred in societies without such centralization. The Existence of a System of Private Property During the industrial revolution profit-seeking entrepreneurs would not have invested in buildings and equipment where their ownership was not respected by other individuals or by such institutions as the government or the church. Some make the same type of argument for the neolithic revolution. Most foraging societies recognized a group “ownership” of a particular territory, which they were willing to defend by force (Pryor, 2003b). Many even recognized private property in land: a band of the foraging Vedda of Sri Lanka, for instance, recognized individual ownership (in the sense of exclusive rights of use) of hills and trees within the band’s territory. This was not unusual; other foraging societies in my sample also recognized private ownership of particular fruit trees or fishing spots. Many Western economists argue that a “change in property rights” is the key explanation for the rise of agriculture. For instance, according to Douglass C. North and Robert Paul Thomas (1977), “[I]t is the incentive change resulting from exclusive property rights that will inevitably create agriculture . . .” They also argue that such exclusive rights prevent over usage of the property – a phenomenon now labeled “the tragedy of the commons.” Some vivid examples illustrating the North and Thomas hypothesis can also be easily found. For instance, the !Kung San in the Kalahari desert in Namibia had a custom labeled “demand sharing” by anthropologists: if one person asked another for a particular article, the latter felt obligated to give it in order not to appear stingy. This transfer was not necessarily a gift, since it conferred to the giver the right to request something from the receiver in the future. A related phenomenon was “demand reciprocity,” in which one person gave something to another person and then demanded a particular article in return. Demand sharing and/or demand reciprocity were found in a considerable number of foraging societies (Pryor, 2003b). Some anthropologists have argued that the custom of demand-sharing discouraged the !Kung San from working long hours because any accumulated surplus of goods would, sooner or later, have to be given away. According to Woodburn (1998), some !Kung San, who attempted to set up farms, were unable to prevent kin from coming to eat up all of the harvested grain, including the seed stock; the Hazda of Tanzania also had the same experiences.

18

FREDERIC L. PRYOR

The evidence against the North-Thomas hypothesis is, however, strong and, according to my data, it seems unlikely that exclusive usufruct rights were either a necessary or sufficient condition for agriculture. In the SCCS sample, all of the foraging societies had such exclusive rights, at least for some articles foraged, produced, or otherwise acquired by individuals or nuclear families (Pryor, 2003b). Further, I have not yet found any society, no matter how primitive, outside the SCCS sample without some types of exclusive property rights either.16 The big question is where these foraging societies drew the line between mine and thine. Several types of evidence deserve note:
Co-existence. A society could have both agriculture and demand sharing. For instance, two societies in Brazil, the Timbira and Tupinamb´a, featured both (Crocker, 1990, pp. 178, 184; d’Evreux, 1864). In both, the fields were cleared by communal groups, but they considered the individual manioc gardens and their produce as the private property of the individuals who tended them, even though food and other goods were subject to demand sharing.
Food inventories. In high latitude regions where agricultural products are seasonally harvested, many foraging societies also had private food storage, especially where foodstuffs were gathered only in particular seasons or animals were hunted only in certain months. In such cases, at least for societies in the SCCS sample, all such food caches were considered private property and could not be touched by others except in extreme circumstances.17
Change. The line between mine and thine could change when external conditions changed. For instance, when the !Kung San came in contact with Western market, they became more individualistic and acquisitive within several decades. Instead of pitching their huts in a circle, with their possessions in full view of everyone, they began to build their huts further from each other, either in a row or scattered, and they stored precious goods in locked trunks where they could not be seen by others (Yellen, 1998). Change in property rights in other societies could occur as a result of changes in group composition. More specifically, because changes in band membership were one way of adapting to fluctuations in weather conditions and were a regular occurrence, it would not have been difficult for those wishing to define property rights in a different way from those in their own band to form separate bands. These three examples offer contrary evidence to the hypothesis that private property (exclusive use-rights) was a necessary or sufficient condition for the origin or spread of agriculture. Additional negative evidence is provided in Lines 9 and 10 in Table 2, which show an irregular pattern of territoriality and private ownership of land, rather than a clear-cut one-step jump. I also attempted a broader test of the

From Foraging to Farming

19

North-Thomas hypothesis by defining the economic systems of foragers in terms of ten aspects of property relations and the distribution of goods (Maxwell et al., 2002; Pryor, 2003b), and then seeing whether societies with one type of economic system were more likely to make the transition to agriculture than another. No such relationship could be found. Given space limitations, I cannot present at this point the lengthy statistical details of this analysis. Other rights normally associated with private property, such as the right to sell or destroy property might have co-evolved with the development of agriculture, but they seem irrelevant to this analysis. A Brief Summary In some cases a variety of cultural, social, and political factors might have required a food surplus over and above that which was necessary for bare subsistence. Further, such a surplus might have been more easily obtained by agricultural production than by foraging. Nevertheless, the hypothesis that such cultural, social, and political factors created demands on production in the society that triggered the emergence of agriculture does not receive support from my data. On the other hand, after the point where 25–45% of subsistence came from agriculture, certain correlations of particular cultural, social, and political factors and greater reliance on agriculture can be found. It is, of course, difficult to determine whether such non-economic factors caused the spread of agriculture, simply occurred simultaneously, or were the effect of a greater reliance on agriculture. Nevertheless, a sufficient number of exceptions to these correlations can be identified to suggest that these various proposed causal factors were certainly neither necessary nor sufficient conditions for a heavy reliance on agriculture for subsistence. Rather, they either facilitated the adoption of agriculture or were themselves brought about by the increasing importance of agriculture in the society. The evidence presented above also suggests that in most foraging societies, exclusive usufruct rights to certain patches of land and the products therefrom or to food inventories were not unusual. I provided examples where agriculture and demand-sharing coexisted and pointed to evidence showing that the transition to agriculture was not dependent on a particular economic system, defined by cluster of property right and distribution mechanisms. I also noted that under certain conditions in particular societies, attitudes toward property rights were quite malleable. Finally, I pointed to evidence that the economic system, defined by various types of property arrangements, also did not appear to have played an important role in the origin or spread of agriculture. In brief, it does not seem likely that the lack of private property rights at a particular time or place served as a serious constraint on the development of agriculture.

20

FREDERIC L. PRYOR

Hypothesized Impact of Land Resource Stress For several decades in the second half of the twentieth century, the hypothesis that the origins and/or spread of agriculture lay in the adaptation of societies facing a higher population density and diminishing returns of hunting and gathering gained considerable acceptance (see particularly Cohen, 1977a, b). According to this approach, originally a foraging band could avoid diminishing returns by fissioning, whereby subgroups would hive off and move into unoccupied lands and such a process gradually led to a spread of population over much of the globe. At some point within the last 10,000 years, however, most foraging areas were peopled. To offset diminishing returns, hunters and gatherers could either work harder, rely on less appetizing foods, turn to fishing (including the hunting of sea mammals) or finally, increase their reliance on agriculture for subsistence. By the 1990s, this resource stress hypothesis was distinctly a minority viewpoint among anthropologists and archaeologists.18 To gain a broader perspective on these issues, I discuss in this and the following section two related phenomena: evidence for a stress on land resources and sedentarism. Resource Stress: Potential Food Resources In order to focus more sharply on population density, settlement permanence, and stress on terrestrial resources, I omit in the calculations in Table 3 all SCCS societies which can be classified as nomadic herding or which relied for more than 15% of their subsistence on fishing. Population density estimates in line 4 of Table 3 are rough, particularly for nomadic (but non-herding) societies, but they suggest that while population density was noticeable higher when reliance on agriculture increased from 25 to 45%, they were higher by an order of magnitude thereafter. Nevertheless, only one of the societies19 in Table 3 which relied less than 45% on agriculture had a population density of more than 0.23 people per square kilometer, while the population densities of societies relying more than 75% on agriculture varied considerably and in some cases were quite low.20 These results suggest that high population density was a sufficient but not necessary condition for the origin and spread of agriculture. Underlying such results is the oft-discussed fact that nomadic foragers found it disadvantageous to have large families, given the difficulty of carrying many small children from place to place. Whether consciously or not, such societies took measures taken to space children and limit family size, such as sexual abstinence at certain times (for instance, before a hunt or before certain ceremonies), abortion, infanticide, or long nursing periods and the resulting lactational amenorrhea, even after they began to practice some agriculture for subsistence and follow a less

From Foraging to Farming

Table 3. Demographic and Consumption Indicators. Reliance on agriculture for subsistence 1. Number of societies in sample Demographic indicators 2. Fixity of communities 3. Population density (people per square km) Consumption indicators 4. Natural biomass production (grams/square meter/year) 5. Available consumable biomass (grams/square meter/year) 6. Consumable biomass per person per day (an index)

0–5% 10

5–25% 4

1.0 0.10

1.8 0.14

966 12.7 236.6

2923 47.0 626.1

25–45% 2

45–75% 27

75–95% 40

95–100% 17

Total sample 100

4.5 0.69

5.0 15.43

5.8 69.99

5.8 98.41

4.9 48.71

1887 19.9 51.5

2378 32.8 3.8

2000 27.8 0.7

1569 20.9 0.3

1960 27.0 1.00

Note: These averages exclude nomadic herding societies and also those societies relying more than 15% on fishing and water mammals for subsistence. For fixity of residence (line 3), 1 = migratory; 2 = semi-nomadic (fixed for several months, then migratory); 3 = rotating among 2 or more fixed residences; 4 = semi-sedentary (fixed core, then migratory for a few months); 5 = impermanent (periodically moved)’ 6 = permanent. Line 6 is calculated from lines 3 and 5 of this table. Sources of data are presented in the Appendix.

21

22

FREDERIC L. PRYOR

nomadic way of life. Therefore, at least during the early stages of agriculture the stress on land resources, as proxied by population density, appeared to have still remained generally low. A more exact determination of a stress on land resources requires calculation of the per capita consumable natural biomass on the territory of the society, an estimate which gives rise to considerable difficulties. It is first necessary to determine the total biomass produced at the specified location of the various SCCS societies. For this purpose, I used Rosenzweig’s (1968) formula relating evapotranspiration to natural biomass production, Since his formula appears to have a large margin of error, however, we must proceed cautiously.21 Line 4 of Table 3 shows that the average SCCS society (non-fishing, non-nomadic herding) lived in an environment producing 1960 grams of plant growth per square meter per year. Only a small fraction of this biomass was consumable, however, and the fraction depended on the relative humidity (based in turn on rainfall and temperature) of the environment. In line 5, I use the formulae devised by Kelly (1983) that take such factors into account in estimating the amount of above-ground plant biomass edible by humans. Even if the natural biomass estimates are correct, these estimates of available consumable biomass are also subject to considerable error.22 The final step, shown in line 6, is to divide total consumable biomass by the population density for each society to obtain the consumable biomass per unit of area per person. The result is the biomass potentially available per day for a member of the society in a given area, which is my measure of land stress. Since the measure of natural biomass production is terrestrial, it makes sense to omit societies relying 15% or more on fishing for subsistence. For those societies relying on agriculture for less than 45% of their subsistence, the results are quite jagged because the sample of such societies is relatively small and, in addition, three approximations are involved in making the calculation. Therefore, we can only focus on gross differences.23 The results suggest a high per capita availability of consumable natural biomass for those societies with 25% or less reliance on agriculture for subsistence, considerably less for societies relying 25–75% on agriculture, and very much less for those societies with more than 75% reliance on agriculture. With one exception, all of the societies relying 25% or less on agriculture were in environments with an abundance of available consumable biomass.24 On the other hand, a number of societies with high available consumable biomass per capita had more than a 75% reliance on agriculture for subsistence.25 Such results do not tell us whether the low per capita consumable biomass was a cause or an effect of heavier reliance on agriculture. But they suggest: (i) that a low per capita available biomass was a sufficient but not necessary condition for a

From Foraging to Farming

23

heavy subsistence reliance on agriculture; and (ii) that a high per capita biomass was not a sufficient cause for a low reliance on agriculture. Such conclusions generally parallel the results discussed above about population density. But these arguments raise some difficulties, which can be seen from a common criticism of the resource-stress argument, namely that the (average) productivity of foraging was generally much higher than that of early agriculture. This criticism is, of course, is irrelevant because it is the marginal, not the average, productivity which is important. The marginal productivity of gathering might have been very low, even while the average was high.26 The calculations in Table 3 are also of average, not marginal, quantities, although changes in the averages as we move from societies with less to greater reliance on agriculture present some important clues about the marginal productivities. In brief, my calculations provide suggestive, but not conclusive, arguments about the relationship between land stress and the origin/spread of agriculture. Suitably modified, I believe that the resource stress argument has merit. In working through its implications, several important factors underlying the origin and spread of agriculture are brought to light:
Benefits and costs of agriculture. The fact that some amount of agriculture was practiced by some preindustrial societies apparently not suffering from a stress on land resources suggests that we look upon agriculture as one element of a “subsistence portfolio.” This means that we should focus attention on attempts to trade off risk, work effort, and return. Too heavy a reliance on one particular food or technique for obtaining food was risky if there were fluctuations in productivity due to natural causes; so that a variety of foods and techniques for obtaining food – including agriculture – lowered the risk of starvation. Furthermore, if the correlation from year to year between productivities of foraged and cultivated plant foods was low, societies would be rational to engage in agriculture, even if its marginal productivity were considerably lower than foraging, in order to lower famine risks. This benefit-cost calculation can be clearly seen when the external conditions change, and can either raise or lower the relative marginal labor productivities of foraging and agriculture. A well-known example is provided by the Native Americans of the Great Plains, who relied primarily on agriculture until they obtained horses from the West. At this point, they greatly reduced their reliance on agriculture and turned instead to buffalo hunting – which had suddenly become much more productive – for a major part of their subsistence. Another example is provided by the Moriori of the Chatham Islands (Diamond, 1999, pp. 53–54). These islands were originally colonized by the Maori, who relied in part on farming. In the succeeding centuries, however, the Moriori abandoned agriculture

24

FREDERIC L. PRYOR

for hunting and gathering which, given their much lower population density and the lush environment, provided them with a stable source of subsistence without as much labor as farming. On the other hand, it is quite possible that dessication in the Near East 10,000 years ago lowered the marginal productivity of foraging and, in the absence of measures to reduce the population, a greater emphasis on agriculture for subsistence was the only way to avoid starvation.
Labor utilization. Foraging society had members who were unable to engage in much hunting and gathering, for instance, women nursing babies, elderly without sufficient stamina for long hikes, or children. If such societies remained in any spot for several months, then these members would have been able to produce some food for their families through agriculture. A benefit-cost analysis might show that such a use of their time was quite worthwhile because their marginal productivity, while not high, was still greater than zero. The resource stress argument has other implications for more advanced societies.27 For our purposes, however, the data presented above suggest – but do not prove – that a stress on land resources, is a sufficient but not necessary condition for a shift from foraging to a greater reliance on agriculture (or fishing) for subsistence.

Agriculture and Sedentarism Agricultural production does not require sedentarism and nomadic societies could have carried out plant-production in three ways. Either they could have planted and harvested fields during periods in which they remained in one camp for several months; or they could have planted the fields and then periodically have returned to tend them; or, like the Sirion´o of eastern Bolivia (Holmberg, 1950), they could have planted gardens of tubers and maize during one season and then have returned many months later to harvest that part of the crop which survived. Since patterns of nomadism and frequency of moves differed considerably among foraging societies, depending in part on seasonality patterns and the richness of the environment (Kelly, 1983, 1995), the types of nomadic-plant-production also differed. In brief, although sedentarism is helpful for agriculture, it is neither a necessary nor, a sufficient condition for it.28 Nevertheless, sedantarism, population density, and a heavy reliance on agriculture have some important interconnections which deserve to be dissected. If the society relied for subsistence primarily on large game, which reproduce slowly, then nomadism would have been necessary to prevent the over-hunting of any one region (Carneiro, 1969). The more a society relied on gathering and/or fishing for subsistence, the less necessary nomadism became, especially since the

From Foraging to Farming

25

more rapid reproduction of most plants and fishes means a more distant point of diminishing returns. Moreover, if the society lived in areas of marked seasonality, then by the time that it had gathered in one place and moved on, the season of plenty would be over and the move would be – literally – fruitless. Such considerations suggest that semi-sedentary or settled foraging societies were probably not unusual in the past, assuming that: (i) the population of the village remained low or the area was relatively rich; and (ii) the domesticated plants were slow growing and required a certain attention (using various proto-agricultural techniques). For instance, archaeologists have found small pre-agricultural village sites in Central Europe dating back to 25,000–30,000 years ago (Cauvin, 2000). In the Near East in the period from 10,000 to 12,500 years ago, many of the Natufians also lived in pre-agricultural villages, subsisting by gathered wild grains which, in a later period, they domesticated. On average, as shown of line 3 in Table 3, the SCCS societies with up to 25% reliance on agriculture ranged between fully nomadic and semi-nomadic. The sample has no examples of foraging societies, such as the early Natufians or Owens Valley Paiute in the 19th century, who lived in settled communities but had no agriculture. Beyond the point of 25% reliance on agriculture, however, settlement fixity increased dramatically in a two-step pattern, although the sample is too small for societies with 25–45% reliance on agriculture for us to have much confidence in this part of the table. It is noteworthy that several societies with 45–75% reliance are classified as semi-nomadic or rotating between two or more mixed residences.29 A high degrees of sedentarism provided the conditions for several positive feedback loops that seem likely to have encouraged an increasing reliance on agriculture. For instance, in settled communities, it was less necessary to limit family size, and an imperceptible change in the population growth rate can have had serious long-term consequences over a millennium. Moreover, if the society engaged in a small amount of agriculture when it became settled, the children (who were useless in hunting or a nuisance in gathering) could be employed tending the crops or watching the herd. Since people did not need carried all their goods around, sedentarism also meant that it was easier to accumulate wealth and self-reinforcing differences in wealth or status validation rituals could and did arise (as shown in my regression results discussed above in the section on social differentiation). Finally, members of sedentary communities could have also engaged in more systematic efforts to domesticate particular plants and animals and to develop new tools and technologies for food procurement, production, and processing. Nondemographic positive feedback loops encouraging greater reliance on agriculture can also be specified in the social and political sphere (Smith, 1973). Given the high correlation of sedentarism and reliance on agriculture for more than 25% of

26

FREDERIC L. PRYOR

subsistence, it may be more fruitful to attack the problem of the spread of agriculture by asking two questions in future research: Why (and where) did particular societies not more fully adopt agriculture when then relied from 5 to 45% on agricultural for subsistence and were semi-sedentary. And how were they able to continue the family limitation practices, which they had used in their more nomadic phase, when their residence became more fixed? Unfortunately, these must be left for analysis by others. In brief, although permanent settlements in a single place were neither a necessary nor a sufficient condition for heavy reliance on plant agriculture for subsistence, certain mechanisms within fixed communities could have easily led to a heavier stress on land resources and a greater need for food surpluses, both of which could have resulted in an increased dependence on agriculture (or fishing, where that was possible). Of all the variables under examination in this essay, sedentarism seems the most closely related to the spread of agriculture, but the larger question remains of why sedentarism itself emerged.

Summary and Conclusions My conclusions regarding the origins of agriculture can be easily summarized. Domestication of plants and nomadic agriculture probably developed in all but the most hostile environments. Furthermore, most foraging societies undoubtedly saw agriculture as just one of a variety of methods of subsistence. The small extent to which some foraging societies resorted to deliberate food production for subsistence depended on such factors as climate, the level of technology, the correlation between agricultural and gathering yields, the availability of particular plants and animals, the labor on hand that could not be used in foraging, and similar factors affecting the ratio of benefits to costs. In sum, the origin of agriculture was not a dramatic event, since it was probably independently invented in widely scattered places with different climates and environments. The search for its origins or for particular cradles of agriculture does not seem worthwhile, an idea argued by many others (e.g. Harlan, 1988). If the positive feedback mechanisms discussed above that led to a much greater reliance on agriculture were not set in motion, a foraging society where agriculture had become a small part of its subsistence repertoire might remain at the same stage of economic development for thousands of years. In contrast to the origins of agriculture, the factors underlying an increasing dependence on agriculture for subsistence, especially in such a brief historical period, were much different and more complicated. Up to now I have looked for correlations between the relative importance of agriculture in a society’s way of

From Foraging to Farming

27

life and a number of possible causal variables. In some cases, I found only a low correlation, which means that such a variable can be rejected as a major cause. Many of the alleged causal variables show a rough simultaneity with reliance on agriculture, which makes our task of causal analysis more difficult. Was increased population density a response to agriculture, or its cause? Or is this a case of mutual causation? The traditional statistical methods used by economists to handle such difficulties will not work here.30 Instead, I have looked at exceptions to the various correlations between suspected causal variables and dependence on agriculture to determine whether a given variable was either necessary or sufficient for the spread of agriculture. The most promising results suggest that a rise in population density, a decline in available consumable biomass, and a greater fixity in the residence pattern of the community were sufficient, although not necessary, conditions for greater reliance on agriculture. In particular, greater sedentarism often set up positive feedback loops (notably through population growth and greater social differentiation) for more dependence on agricultural production for food. Nevertheless, other causes for greater reliance on agriculture cannot be excluded. For some societies, environmental or climatic changes led to a shift toward agriculture. For other societies, political centralization might have reached the point where it was possible for group leaders to force members to farm and hand over some of their crop. And finally, for still other societies, social differentiation, combined with competitive feasting and potlatches to validate social status, might have triggered a shift toward more agriculture because food production above subsistence needs were required. Since the cross-cultural data reveal only sufficient, not necessary, conditions for the spread of agriculture, I see no way to support one of these general hypotheses over another. It is far from obvious that a single cause underlay the spread of agriculture in different times, places, and environments (see MacNeish, 1991 for example). Since greater sedentarism set in motion positive-feedback mechanisms that in many cases promoted greater dependency on agriculture, it seems more fruitful to explore why certain societies did not become more agricultural. But now the key methodological question must be asked: What is gained from the generalizations from my cross-section results and is it legitimate to generalize from them to events over time? The cross-cultural results provide a plausible, but not air-tight case, about the spread of agriculture; Nevertheless, if some societies within historic time were heavily reliant on agriculture but did not have property X, then the burden of proof that X was a key variable for the origins of agriculture 10,000 years ago lies on those making the argument. More specifically, they must demonstrate that conditions 10,000 years ago were sufficiently different from those observable in the SCCS sample to justify the

28

FREDERIC L. PRYOR

supposition that causal mechanisms were different. Unfortunately, it seems very difficult – if not impossible – to make this argument, since the archaeological evidence seems too ambiguous and scattered to provide conclusive proof about most of the cultural, social, political, or demographic variables discussed above that might have caused a society’s increasing reliance on agriculture. But one conclusion is very clear, namely, that the shifts of production into industry and agriculture differed in several major ways that make the analogy between the two extremely misleading. The industrial revolution required major changes in technology with powered machinery replacing hand tools, the factory supplanting the home workshop, and the scale of production (and the corresponding division of labor) increasing dramatically. By way of contrast, the “agricultural revolution” represented no sharp break in technology but emerged as part of an incremental historical process. Initially, the industrial revolution required a particular pattern of property rights, at least where the production was carried out by private individuals, while the “agricultural revolution” was implemented under a variety of property arrangements. Initially, the industrial revolution was primarily a deliberate activity undertaken for monetary profit, rather than an adjustment to demographic, social, or political changes that influenced the risk/return of various subsistence activities. Finally, the industrial revolution occurred relatively quickly – in the course of a century or two – in most societies, while the shift to a heavy reliance on agriculture probably took millennia, to judge from the many foraging societies in historic time.31 In many ways, “revolution” is an inappropriate concept for understanding either how societies first began to practice agriculture on a small scale or how it eventually came to be their primary source of food.

NOTES 1. Two additional objections can be raised to my approach that are not met by these arguments. First, my sample may be biased because it includes “unsuccessful” societies which, for one reason or another, did not advance to become full-fledged agricultural or industrial societies. Further, many of these societies, which managed to survive in their “original state” until they could be visited by Western travelers or anthropologists, were located in isolated or extreme environments, which may not reflect the habitat of most humans 10,000 years ago. The importance of these a priori objections, however, can only be judged when the results of the analysis are presented. 2. Proto-animal-husbandry has received much less attention, although certain practices among reindeer hunters who followed the herd, culled it, and employed other techniques to manage it (Bender, 1975, p. 2) might be so classified. When most authors speak of protoagriculture, they really mean proto-plant-production. Not many parallels can be drawn between proto-agriculture and proto-industrialization.

From Foraging to Farming

29

3. Time budgets for a number of societies are presented by Hayden (1993, p. 139) and Kelly (1995, p. 20). 4. The non-linear relationship could mean that societies living in environments with abundant rainfall have the fortunate choice of relying on both foraging and agriculture, while those in less favorable environments are forced to rely on one or the other, but not both, depending on the particular conditions. 5. Although those societies with 0–5% agricultural subsistence had somewhat lower average and effective temperatures than the other societies in the sample, the temperature differences between the societies in the sample and that of the sample mean are not, with one exception, statistically significant. 6. More specifically, neither annual evapotranspiration nor effective temperature provided any statistically significant explanation for percentage reliance on agriculture (plant production and animal husbandry), either together or when combined with the measurements of climate, slope, and soil variables. The summed agricultural potential variable proved no better; and while the minimum indicator variable was positively and significantly correlated with the agricultural subsistence variable, it only explained 10% of the variance. Similar experiments with a subsample of societies with less than 45% reliance on agriculture for subsistence yielded roughly similar results. A variety of other experiments were attempted, but with no more success. In a previous study (Pryor, 1986) I used many of the same variables, but with less restrictive criteria for eliminating societies living in extreme environmental conditions. This exercise showed a very weak but positive relation between environmental conditions and the presence of agriculture. 7. Another open question is why, if climatic change underlies the origins of agriculture, did agriculture not develop during the Pleistocene which experienced many climatic change. Richardson et al. (2001) present evidence that during the late Pleistocene, climates were not just cold, dry, and low in atmospheric carbon dioxide (necessary for photosynthesis), but, most importantly, their changes were too variable and abrupt to serve as an impetus to adopt agriculture as a major subsistence activity, while, during the early Holocene period (roughly 11,500 years ago), climates were relatively warmer, wetter, and richer in atmospheric carbon dioxide and their fluctuations were more damped and less abrupt. 8. An oft-cited example is the Viking colony on Greenland around 1000 A.D., which continued its practice of a Scandinavian-type agriculture after the onset the “Little Ice Age.” This type of farming, however, was increasingly unsuitable as the climate became progressively colder and around the middle of the 15th century the colony disappeared, at the same time that the neighboring Inuit communities, which relied for subsistence primarily on fishing and hunting, flourished. 9. This literature is summarized in a particularly insightful way by Hayden (1990). 10. That is, descendants of a common ancestor who, as a group, have certain rights (for instance, to land), certain obligations (ceremonial or social), and certain restrictions (for instance, on eligible marriage partners). 11. This distinction is between what biologists call K-selected and r-selected species. The former reproduce slowly and devote considerable efforts to raise offspring to maturity (for instance, large animals); the latter reproduce quickly and leave their offspring to fend for themselves (for instance, fish). 12. More insight into this relationship between social differentiation and agriculture can be gained by exploring the determinants of social differentiation for societies relying on

30

FREDERIC L. PRYOR

agriculture for less than 45% of their subsistence. A regression calculation shows that the degree of social differentiation was significantly and positively related to its residential fixity and also to the relative importance of fishing in its subsistence. If societies relying on fish and sea mammals for more than 35% of their subsistence (for instance, the societies on the northwest Pacific coast of America) are removed from the sample and the regression recalculated, such relationships disappear. 13. These included the Pastoral Fulani of Niger, the Masai of Tanzania, and the Toda of India. 14. These included the Nyakyusa of Tanzania, the Popoluca of Mexico, the Santal of India, the Yanomamo of Venezuela, and the Zuni of Arizona. 15. These included the Aymara of Peru, the Iban of Borneo, the Ibo of Nigeria, the Kikuyu of Kenya, the Luguru of Tanzania, the Pentacost Islanders of the South Pacific, the Rhade of Vietnam, the Songhai of Mali, and the Teda of Chad. For the Aymara, however, I disagree of the code. 16. Even certain non-human primates have a property instinct. For instance, in the Yerkes Laboratory – a large outdoor reserve – chimpanzees earned tokens for various tasks which they could later exchange for food. They began to hoard the tokens and became “hysterical” when other animals came near their cache (Dare, 1974). Sigg, Falett (1985) also report some interesting experimental data using hamadryas baboons. Many different animal species also exhibit territoriality (Pryor, 2003c). 17. For instance, among the Montagnais of Canada, a starving passerby could use such inventories. 18. Critics advance four arguments against the resource stress hypothesis:


Unaccepted analogy. Some of the evidence for the hypothesis was based on an analogy with the “broad spectrum revolution” which occurred 10,000–30,000 years ago and represented a shift from the consumption of large, placid animals and easily available plant life to small animals, dangerous animals difficult to kill, rodents, frogs, snails, crustaceans, other products from the sea, and particular plant foods, all of which required more work hours to obtain or to process. Although such dietary changes suggest that marginal returns from foraging were diminishing (aggravated by the extinction of certain megafauna), many researchers argued that this analogy did not hold for the origins or spread of agriculture.
Direct evidence. Several kinds of conflicting evidence are available. If marginal returns in foraging had diminished, people should have become shorter and more malnourished. However, skeletons at these sites where agriculture later developed do not always show such symptoms of increasingly poor nutrition (as summarized by Roosevelt, 1984). By way of contrast, Keeley (1995, 1999) supplies positive evidence, showing that proto-agriculture among a number of Shoshone and Paiute groups was directly related to resource stress. His direct evidence, however, has been strongly disputed by others (see, for instance, Hayden, 1995) and, moreover, is drawn only from pure foragers.
Measurement. Population density by itself does not indicate a stress on natural plant and animal resources. It is necessary to take into account the natural productivity of the land as well, a defect that several anthropologists (notably Kelly, 1983, 1995; and Keeley, 1988) have tried to repair. As I note below, however, such exercises are fraught with peril.

From Foraging to Farming

31


Relative productivities. Some point out that the average labor productivity of foraging was much higher than in agriculture, so that land stress could not have been involved. But the key point is the marginal productivity and the point of diminishing returns in foraging is usually reached much sooner than in agriculture. Average productivities are quite irrelevant for the analysis. 19. Murdock and Wilson (1973), whose population density data I use, rate the Timbira of Brazil as having a population density of 1.16 people per km2 . My reading of the ethnographic materials for this society suggest that this is much too high. 20. These include the Lepcha of Sikkim, the Lolo of China, the Teda of Chad, the Yanomamo of Venezeula, and the Zuni of Arizona. 21. Rosenzweig’s regression calculation relating net terrestrial biomass production to environmental variables uses only 48 locations, all but two of which are in the United States or Canada. Two other formulae are also available: the Lieth-Box formula (cited by Sharpe, 1975), which also uses evapotranspiration as the single explanatory variable; and my formula, which is based on Lieth’s (1975) sample of 45 different locations over the planet and which follows a procedure that he suggests for using average annual temperature and rainfall data instead of evapotranspiration. Calculated for the 186 SCCS societies, the average difference of the latter two series from the Rosenzweig calculation is about 50%; these two other formulae also yield a somewhat lower mean. There is, however, no systematic difference in the estimates for societies with different reliance on agricultural production for subsistence. For this essay I use the Rosenzweig formula only because the adjustment that Kelly (1983, 1995) made to calculate the consumable biomass from the actual biomass production used the Rosenzweig formula. 22. Kelly (1983, 1995) used somewhat different estimates of evapotranspiration, temperature, and precipitation than I do, so that our estimates of natural biomass production and the classification of environments (humid and arid) also vary. This, in turn, raises problems in using his formulae for societies whose environments are on the edge between these two classifications. After a useful communication with him about these matters, I reclassified the environment of several of these societies to eliminate extreme cases because their environment was close to the critical borderline. 23. I tried four tests of the meaningfulness of these estimates of consumable biomass per person (the inverse of land resource stress) by correlating other variables that might be related to land resource stress. (a) For primarily foraging societies a stress on land resources might be revealed by a higher ratio of gathering to hunting. This would arise because eating animal flesh is an inefficient way of obtaining nourishment from the environment since animals consume considerable biomass which is not contained in their meat. Unfortunately, after eliminating societies with less than 25% reliance on fishing and agriculture (plant production and animal husbandry) and two societies with data problems, I was left with only 11 societies for the investigation. The regression coefficient has the right sign, but the results are not statistically significant. (b) Stress on land resources might be related, as Rosenberg (1990) and Hayden (1995) have suggested, to the existence of territoriality and/or private ownership of land. Unfortunately, only SCCS societies with less than 45% reliance on agriculture have been coded for territoriality and private property (Pryor, 2002b). With regard to territoriality, the predicted relation is found and it is statistically significant. The

32

FREDERIC L. PRYOR results, however, seem to be driven by several societies with particularly high estimates of consumable biomass per person; when these are removed, the relationship is no longer statistically significant. As for private ownership of land (or trees), the regressions do not reveal statistically significant relationships. (c) Another possible indicator of stress on land resources, some anthropologists have argued, is famine. This seems dubious, since famine could be the result of weather conditions that are unrelated to land stress and that could affect gathering as well as agriculture. Nevertheless, I dutifully tried to test this proposition. For the sample as a whole, consumable biomass per person was inversely and significantly related to the presence and frequency of famine, in part, I suspect, because agriculture is much more specialized than gathering and, hence, more prone to general crop failure. For foraging societies with less than 35% reliance on fishing, I found no relationship. The famine indicator was the average of four indicators of famine in Dirks (1993) and Ember and Ember (1992), reproduced as Variables 1265, 1267, 1683, and 1685 by Divale (2001). (d) Some anthropologists have suggested that frequency of external warfare is related to land stress, an hypothesis that overlooks other incentives for warfare such as booty, slaves, or wives. I explored the relationship of reliance on agriculture to frequency of external warfare (from Ember & Ember, 1992, and reported as Variable 1650 by Divale, 2001) but found little evidence for such a relationship.

It might be added that this measure of consumable biomass tells us nothing about the available biodiversity. 24. The exception was the Aranda of Australia, who registered 5.4 on the index. They relied considerably on tubers and plant roots, which are not included in the estimate of the biomass. 25. Well-known examples are a number of Indians in American Great Plains which, before contact with the West, heavily relied on agriculture, although they lived in a relatively rich and sparsely populated area (but see below). The SCCS sample includes other similar examples, for instance, the Barusho of Kashmir, the Yanomamo of Venezuela, the Zuni of Arizona, and the Banen of Cameroon. 26. It is sometimes claimed that diminishing returns in gathering is considerably greater than in agriculture, but I have never seen any empirical evidence on the matter. 27. For instance, it seems likely that at some point a rising population density would lead to a shift in both the pattern of crops planted and the pattern of land use. SCCS societies that relied less than 75% on agriculture primarily planted crops that did not require a plow to obtain high productivity, such as millet, sorghum, maize, dry rice, or root crops; thereafter, there is a marked shift toward crops requiring extensive land and the use of a plow. For the percentage of land used in agriculture, two critical points appear, one after 25% reliance on agriculture, the other after 75% reliance (where double cropping became more common). 28. I should add that it is not even necessary to be a human being to engage in engage in agriculture, since it is also found in the animal world. H¨olldobler and Wilson (1990, pp. 527–529; 604–608) discuss how certain species of ants herd aphids to obtain the honeydew they secrete, while other species cultivate fungus farms, which they then consume. 29. The most striking exception to the correlation between settlement fixity and reliance on agriculture was the Teda, a Saharan society in Chad rated as semi-nomadic. It relied for over 75% of its subsistence on agriculture (including considerable animal husbandry),

From Foraging to Farming

33

but its environmental circumstances were quite special. An additional 4 societies with 45–65% reliance on agriculture are rated as semi-nomadic or rotating among two or more fixed residences. These include the Cayua of Brazil, the Havasupai of Arizona, the Lozi of Zambia, and the Papago of Mexico. 30. One traditional method is the use of simultaneous equation models, but this is not possible for the present problem. First, the data are not sufficiently precise to allow such a technique to be meaningfully employed. Second, the factors influencing the alleged causal variable are also difficult to specify and raise additional simultaneity problems. Third, the determination of inflection points in the various series to determine causation raises other unsolvable problems. 31. Although the industrial revolution first occurred in the temperate zone of the northern hemisphere, it is difficult to discuss climatic prerequisites for manufacturing, at least with a straight face. In contrast, agriculture seems to have emerged in a wide variety of climates.

ACKNOWLEDGMENTS I am grateful to Bruce Grant, Steven Piker, Zora Pryor, David Smith, Michael Speirs, Victoria Wilson-Schwartz, several anonymous referees, and members of the Tri-College Brown Bag Lunch Group for helpful comments on a previous draft of this essay.

REFERENCES Anderson, P. C. (1991). Harvesting of wild cereals during the natufian as seen from experimental cultivation and harvest of wild einkorn wheat and microwear analysis of stone tools. In: O. Bar-Yosef & F. K. Valla (Eds) (pp. 527–556). Anderson, P. C. (Ed.) (1999). Introduction. 1–6. Bailey, H. (1960). A method of determining warmth and temperateness of climate. Geografiska Annaler, 45, 1–16. Barry, H., III, & Schlegel, A. (Eds) (1980). Cross-cultural samples and codes. Pittsburgh: University of Pittsburgh Press. Bar-Yosef, O., & Meadow, R. H. (1995). The origins of agriculture in the Near East. In: T. D. Price & A. B. Gebauer (Eds) (pp. 39–94). Bender, B. (1975). Farming in prehistory: From Hunter-Gatherer to food-producer. New York: St. Martin’s Press. Bender, B. (1978). Gatherer-Hunter to farmer: A social perspective. World Archaeology, 10(2), 204–222. Binford, L. R. (1968). Post-pleistocene adaptions. In: S. R. Binford & L. R. Binford (Eds), New Perspectives in Archaeology (pp. 313–341). Chicago: Aldine. Carneiro, R. L. (1969). The transition from hunting to horticulture in the Amazon Basin. In: 8th International Congress of Anthropological and Ethnological Sciences: Proceedings (Vol. 3, pp. 244–248). Tokyo: Science Council of Japan.

34

FREDERIC L. PRYOR

Carneiro, R. L. (1970). Scale analysis, evolutionary sequences, and the rating of cultures. In: R. Naroll & R. Cohen (Eds), A Handbook of Method in Cultural Anthropology (pp. 834–873). Garden City, NY: Natural History Press. Cauvin, J. (2000). The birth of the gods and the origins of agriculture. New York: Cambridge University Press. Cohen, M. N. (1977a). The food crisis in prehistory: Overpopulation and the origins of agriculture. New Haven: Yale University Press. Cohen, M. N. (1977b). Population pressure and the origins of agriculture. In: C. A. Reed (Ed.), Origins of Agriculture (pp. 135–177). The Hague: Mouton Publisher. Crocker, W. H. (1990). The Canela (Eastern Timbira) I: An ethnographic introduction. In: Smithsonian Contributions to Anthropology (Vol. 33). Washington, DC: Smithsonian Institution Press. Dare, R. (1974). The ecology and evolution of food sharing. California Anthropologist, 2, 13–25. Diamond, J. (1999). Guns, germs, and steel: The fates of human societies. New York: W. W. Norton. Dirks, R. (1993). Starvation and famine: Cross-cultural and some hypothesis tests. Cross-Cultural Research, 27, 28–69. Divale, W. (2001). Pre-coded variables for the standard cross-cultural sample. World Cultures, Spring Edition, compact disc. Ember, C. R., & Ember, M. (1992). Codebook for ‘warfare, aggression, and resource problems: Cross-cultural codes’. Behavior Science Research, 28, 169–185. d’Evreux, Y. (1864). Voyage dans le nord du Br´esil fait durant les ann´ees 1613 et 1614. New Haven: Human Relations Area File SO–09. Flannery, K. V. (1968). Origin and ecological effect of early domestication in Iran and the Near East. In: P. J. Ucko & G. W. Dimbleby (Eds), The Domestication and Exploitation of Plants and Animals (pp. 73–100). London: Duckworth. Food and Agricultural Organization/UNESCO (1971–1978). Soil maps of the world. Paris: UNESCO. Fuller, J. E., & Grandjean, B. D. (2001). Economy and religion in the neolithic revolution: Material structure and the proto-religious ethnic. Cross-Cultural Research, 35, 370–399. Harlan, J. R. (1988). Plant domestication: Diffuse origins and diffusion. In: C. Barigozzi (Ed.), The Origin and Domestication of Cultivated Plants (pp. 21–35). Amsterdam: Elsevier. Harlan, J. R. (1999). Harvesting of wild-grass seed and implications for domestication. In: P. C. Anderson (Ed.) (pp. 1–6). Hayden, B. (1990). Nimrods, piscators, pluckers and planters: The emergence of food production. Journal of Anthropological Archaeology, 9, 31–69. Hayden, B. (1993). Archaeology: The science of once and future things. New York: W. H. Freeman. Hayden, B. (1995). A new overview of domestication. In: T. D. Price & A. B. Gebauer (Eds) (pp. 273–297). H¨olldobler, B., & Wilson, E. O. (1990). The ants. Cambridge, MA: Harvard University Press. Holmberg, A. R. (1950). Nomads of the long bow: The Sirion´o of Eastern Bolivia. New York: Natural History Press. Keeley, L. H. (1995). Protoagricultural practices among hunter-gatherers. In: T. D. Price & A. B. Gebauer (Eds) (pp. 243–273). Keeley, L. H. (1999). Use of plant foods among hunter-gatherers: A cross-cultural survey. In: P. C. Anderson (Ed.) (pp. 6–23). Kelly, R. L. (1983). Hunter-gatherer mobility strategies. Journal of Anthropological Research, 39, 277–306. Kelly, R. L. (1995). The foraging spectrum: Diversity in hunter-gatherer lifeways. Washington, DC: Smithsonian Institution Press.

From Foraging to Farming

35

L´eon, J. (1977). Origin, evolution and early dispersal of root and tuber crops. In: J. Cock, J. MacIntyre & M. Graham (Eds), Proceedings of the Fourth Symposium of the International Society for Tropical Root Crops (pp. 20–36). Ottawa: International Development Research Center. Lieth, H. (1975). Modeling the primary productivity of the world. In: H. Lieth & R. H. Whittaker (Eds) (pp. 237–263). MacNeish, R. S. (1991). The origins of agriculture and settled life. Norman: University of Oklahoma Press. Maxwell, B. A., Pryor, F. L., & Smith, C. (2002). Cluster analysis in cross-cultural research. World Cultures, 13, 22–39. Mulvaney, J., & Kamminga, J. (1999). Prehistory of Australia. Washington, DC: Smithsonian Institution Press. Murdock, G. P., & Provost, C. (1973). Measurement of cultural complexity. Ethnology, 12, 379–392. Murdock, G. P., & Wilson, S. (1972). Settlement patterns and community organizations: Cross-cultural codes 3. Ethnology, 11, 254–295. North, D. C., & Thomas, R. P. (1977). The first economic revolution. The Economic History Review, 30(2nd Series), 229–241. O’Dea, K. (1992). Traditional diet and food preferences of Australian Aboriginal hunter-gatherers. In: A. Whiten & E. M. Widdowson (Eds), Foraging Strategies and the Natural Diet of Monkeys, Apes, and Humans (pp. 73–80). Oxford: Clarendon Press. Papadakis, J. (1966). Climates of the world and their agricultural potential. Buenos Aires: Papadakis. Price, T. D., & Gebauer, A. B. (Eds.) (1995). Last hunters, first farmers. Santa Fe: School of American Research Press. Price, T. D., Gebauer, A. B., & Keeley, L. H. (1995). The spread of farming into Europe north of the Alps. In: Price & Gebauer (Eds) (pp. 95–126). Pryor, F. L. (1985). The invention of the plow. Comparative Studies in Society and History, 27, 727–743. Pryor, F. L. (1986). The adoption of agriculture: Some theoretical and empirical evidence. American Anthropologist, 88, 879–897. Pryor, F. L. (2003a). Agricultural economic systems. Submitted for publication. Pryor, F. L. (2003b). Economic Systems of Foragers. Cross-Cultural Research, 37, 393–427. Pryor, F. L. (2003c). What Does It Mean to Be Human? A Comparison of Primate Economies. Journal of Bioeconomics, 5, 97–146. Richardson, P. J., Boyd, R., & Bettinger, R. L. (2001). Was agriculture impossible during the pleistocene but mandatory during the holocene? A climate change hypothesis. American Antiquity, 66(2), 387–413. Rindos, D. (1984). The origins of agriculture: An evolutionary perspective. Orlando, FL: Harcourt, Brace, Jovanovich. Roosevelt, A. C. (1984). Population, health, and the evolution of subsistence: Conclusions from the conference. In: M. N. Cohen & G. J. Armelagos (Eds), Paleopathology at the Origins of Agriculture (pp. 559–581). Orlando, FL: Academic Press. Rosenberg, M. (1990). The mother of invention: Evolutionary theory, territoriality, and the origins of agriculture. American Anthropologist, 92, 399–415. Rosenzweig, M. L. (1968). Net primary productivity of terrestrial communities: Predictions from climatology. The American Naturalist, 102, 67–74. Sahlins, M. D. (1972). Stone age economics. Chicago: Aldine-Atherton. Sauer, C. O. (1969). Seeds, spades, hearths, and herds: The domestication of animals and foodstuffs (2nd ed.). Cambridge, MA: MIT Press.

36

FREDERIC L. PRYOR

Sharpe, D. (1975). Methods of assessing the primary productivity of regions. In: H. Lieth & R. H. Whittaker (Eds) (pp. 147–160). Sigg, H., & Falett, J. (1985). Experiments on respect of possession in Hamadryas Baboons (Papio hamadryas). Animal Behavior, 33, 978–984. Smith, B. (1995). The emergence of agriculture. New York: Scientific American Library. Smith, P. E. L. (1973). The consequences of food production. In: Module in Anthropology (Vol. 31). Reading, MA: Addison-Wesley. Thornthwaite, C. W., & Associates, Laboratory of Climatology (1964–1968) (8 Vols). Average climatic water balance data of the continents. Centerton, NJ: C. W. Thornthwaite. Tuden, A., & Marshall, C. (1972). Political organization: Cross-cultural codes 4. Ethnology, 11, 436–464. Tudge, C. (1998). Neanderthals, bandits and farmers: How agriculture really began. New Haven: Yale University Press. Watson, P. J. (1995). Explaining the transition to agriculture. In: T. D. Price & A. B. Gebauer (Eds) (pp. 21–37). Woodburn, J. (1998). Egalitarian societies. In: J. M. Gowdy (Ed.) (pp. 87–110). World Cultures (2001). See Divale (2001). Yellen, J. T. (1998). The transformation of the Kalahari !Kung. In: J. M. Gowdy (Ed.) (pp. 223–235).

APPENDIX Sources of Data for Table 1 The data for the three rainfall series come from a database of roughly 8,700 weather stations over the globe reported in the eight volumes of C. W. Thornthwaite Associates, Laboratory of Climatology (1964–1968). With the aid of Nicole Perez, I coded the data from the weather stations closest to the society in question (and at roughly the same altitude); in cases, where several weather stations were roughly the same distance away, the data were averaged. For seven-eighths of the sample, the average distance of the society from the closest weather station was 47 miles; for the remaining eighth of the sample, which were composed primarily of nomadic societies, the distance from the weather station ranged from 140 miles upward. My precipitation data vary somewhat from unpublished data gathered by John W. M. Whiting (reported by Divale, 2001, variable 189). Unfortunately, Divale does not provide any information on the sources and methods employed by Whiting; nevertheless, his data lead to the same general conclusions as mine. Rainfall seasonality is the ratio of potential evapotranspiration, divided by the actual evapotranspiration. Temperature data come from Whiting (reported by Divale, 2001, variables 186 through 188). Effective temperature is calculated from data on the average temperature in the hottest and coldest months and also reflects the length of the

From Foraging to Farming

37

growing season. This statistic is discussed by Robert Kelly (1995, p. 66 ff.) and is based on a formula developed by H. Bailey (1960). Slope and soil data come from maps presented by FAO/UNESCO (1971–1978), adjusted in a manner described by Pryor (1986). The suitability of climate to major crops is based on data from Papadakis (1966, 1970), adjusted in a manner described by Pryor (1986). All three variables are also reported by Divale (2001, variables 921, 922, 924, and 926). The composite variables are also presented in Pryor (1986) and reported in Divale (2000, variables 921 and 928).

Sources and Meaning of Data for Table 2 My data are based on codes derived from a variety of ethnographic sources discussed in Pryor (2002b). The detailed series are presented on my website in a Microsoft Excel file: http://www.swarthmore.edu/SocSci/Economics/fpryor1/. The notes below concern only the codes of others. I present the line number in Table 2, the label of the variable, the original source, and then the number of the variable (preceded by V) as listed in the special CD included in World Cultures (2001). Line 10. Carnaro complexity scale (1970), modified and extended to the SCCS sample in Pryor (2002b). Line 11. Existence of corporate descent groups, Murdock, Wilson (1972); V70. I have recoded this series so that 0 = no corporate kinship groups (coded 5 by Murdock, Wilson); 1 = membership in any type of corporate kinship group (all the remaining Murdock, Wilson codes). Line 12. Social stratification; Murdock, Provost (1973); V158. 1 = egalitarian; 2 = hereditary slavery; 3 = two classes, no castes/slavery; 4 = two classes, castes/slavery; 5 = three social classes or castes, with or without slavery. Line 13. Political differentiation. This is an unweighted average of seven political variables, all of which led to roughly the same conclusions when presented in the format of Table 2. The first step was to recode one of the variables (the leadership variable) so that it formed a linear scale. The second step was to linearly transform all of the variables onto a scale running from 1 to 4. And the final step was to calculate the unweighted average. The seven variables were: (i) Community leadership. Murdock, Wilson (1972); V76. 1 = none; 2 = local council; 3 = single leader plus council; 4 = dual/plural headmen; 5 = single local leader; 6 = single local leader and subordinates; 7 = higher level of political authority.

38

FREDERIC L. PRYOR

(ii) Levels of sovereignty. Tuden, Marshall (1972); V83. 1 = stateless society; 2 = sovereignty, first political hierarchical level up from village; 3 = sovereignty, second hierarchical level up; 4 = sovereignty third or higher hierarchical level. (iii) Higher political organization. Tuden, Marshall (1972); V84. 1 = absent; 2 = groups sworn to peace; 3 = alliances; 4 = confederation; 5 = international organization. (iv) Executive concentration. Tuden, Marshall (1972); V85. 1 = absent; 2 = council; 3 = executive and council; 4 = plural executive; 5 = single leader. (v) Specializing of policing. Tuden, Marshall (1972); V90. 1 = not specialized; 2 = incipient specialization; 3 = retainers of chiefs; 4 = military; 5 = specialized. (vi) Administrative hierarchy. Tuden, Marshall (1972); V91. 1 = absent; 2 = popular assemblies; 3 = heads of kin groups; 4 = heads of decentralized territorial divisions; 5 = heads of centralized territorial divisions; 6 = part of a centralized political system. (vii) Frequency of external war. Ember, Ember (1992); V1650. 1 = warfare rare or absent; 5 = warfare seems to occur once every 3–10 years; 10 = warfare seems to occur at least every 2 years; 14 = warfare seems to occur every year, but usually during a particular season. The code ends at 17 = warfare seems to occur almost constantly and at any time of year. The values between those supplied here are intermediate values in the specified period.

Sources of Data for Table 3 Each line contains the original source of the data and then, for data contained in Divale (2001), the number of the variable preceded by “V.” Line 2. Data on the fixity of community come from Murdock, Wilson (1973), V61. My own ratings for this variable for societies with less than 45% reliance on agriculture for subsistence are practically the same. Line 3. For 68 societies in the SCCS, the data on population density come from Pryor (1985); they are supplemented by data from Murdock, Wilson (1973), V1130. I must emphasize that these estimates are very rough. To calculate the averages, assumptions also had to be made about the average densities at the end points of the scale: I assumed 0.1 for the density of all societies with less than 0.2 persons per square mile; I also assumed 650 for all societies with a density greater than 500 people per square mile. Line 4. The estimation is discussed in the text.

From Foraging to Farming

39

Line 5. Following a procedure derived by Kelly (1983; 1995, p. 121) I calculated consumable biomass (also called “primary biomass” or “secondary productivity” by others) from data on the natural biomass production, effective temperature and precipitation. Effective temperature, which measures the effective solar radiation as well as its annual distribution, was calculated using a formula discussed by Kelly (1995, p. 66) and employs data on the average temperature in the hottest and coldest months (from John Whiting, unpublished; V187 and V188). The precipitation data come from Thornthwaite (1964–1968). In a correspondence concerning his chart (1983), Kelly informed me that the axis of the consumable biomass should have been labeled grams/meters2 , rather than kilograms/meters2 , so that the consumable biomass would be a fraction of natural biomass production. This means, however, that the formulae he presents are in milligrams/meters2 , rather than grams/meters2 . I have made the necessary corrections in the table. Line 6. This is discussed in the table note.

THE PRICE HISTORY OF ENGLISH AGRICULTURE, 1209–1914 Gregory Clark ABSTRACT The paper constructs an annual price series for English net agricultural output in the years 1209–1914 using 26 component series: wheat, barley, oats, rye, peas, beans, potatoes, hops, straw, mustard seed, saffron, hay, beef, mutton, pork, bacon, tallow, eggs, milk, cheese, butter, wool, firewood, timber, cider, and honey. I also construct sub-series for arable, pasture and wood products. The main innovation is in using a consistent method to form series from existing published sources. But fresh archival data is also incorporated. The implications of the movements of these series for agrarian history are explored.

INTRODUCTION Despite the considerable research conducted on the price history of England since James Thorold Rogers’ classic nineteenth century work there is no aggregate series available on the movement of agricultural prices over these years. Lord Beveridge gave price series for a large number of commodities in the years 1540–1830, but no overall aggregate. The extensive volumes of the Agrarian History of England and Wales all contain considerable price evidence. But the years 1200–1914 are covered by six different volumes all of which treated price information differently. The medieval volumes have individual price series constructed by David Farmer, but again no aggregate series. The volumes for Research in Economic History Research in Economic History, Volume 22, 41–123 © 2004 Published by Elsevier Ltd. ISSN: 0363-3268/doi:10.1016/S0363-3268(04)22002-X

41

42

GREGORY CLARK

1500–1640 and 1640–1750 both give overall price indices constructed by Peter Bowden. But because of the untimely death of A. H. John, the 1750–1850 volume gave only price series for individual commodities. The volume for 1850–1914 has a general series constructed by Bethany Afton and Michael Turner starting in 1867. Thus there is no continuous series available for most of the years 1209–1914, which include those of the reputed agricultural revolution. This paper constructs such an index that combines the available published price series and collections with archival sources. The main published sources of prices before 1750 have been Thorold Rogers (1866, 1882, 1888, 1902), Beveridge (1939), and the series compiled by David Farmer for the Agrarian History of England and Wales, Volumes 2 and 3 (Farmer, 1988, 1991b). The main sources thereafter have been the series published in the Agrarian History of England and Wales Volumes 6 and 7 (John, 1989), Afton and Turner (2000), and the Board of Trade Report of 1903 (Parliamentary Papers, 1903). But these sources have been supplemented with prices from printed churchwarden’s and town treasurer’s accounts, and from the printed records of the Carpenters Company (Marsh, 1915–1939). Beveridge also accumulated much unpublished material for the projected second volume of his Price History which would deal with the manorial era. Most of this unpublished price material, now at the Robbins Library at LSE, has been incorporated. For the most important commodity in the years before 1500, wheat, the underlying data from the Farmer series was obtained from the Farmer Archive at the University of Saskatchewan Library, Saskatoon, Saskatchewan, and combined with other sources such as Thorold Rogers, and Beveridge’s Exeter prices to form a new series. Prices for hay, firewood and timber were also collected for the years after 1780 from a variety of archival sources, including the King’s College, Cambridge, Mundum Books.

THE METHOD OF CONSTRUCTION The index aimed for here is of the price of the net output of products of the agricultural sector of the economy. Thus products such as oats and hay that were used, for example, as animal fodder within the farm sector will get less weight than they would if the index was weighted by gross output. The price index was formed as a geometric index of the prices of each component, with the assumed output shares of each commodity used as weights. That is, if pit is the price index for each commodity i in year t, and ␣i is the output share of commodity i, then the overall price level in each year, pt is calculated as, p t = p ␣it i i

The Price History of English Agriculture, 1209–1914

43

Thus it assumes constant shares in the value of output for each item as relative prices change from year to year. This means that if the relative price of an item such as wheat increases in a given year then it assumed to be because the relative supply of wheat by the farm sector has dropped proportionately. This assumption seems particularly appropriate for the years before the repeal of the Corn Laws in 1846 when farm imports were limited. After 1846, when imports increasingly dominated, the logic of weighting in this way is no longer compelling, but for consistency the geometric index is employed throughout. A more common way of constructing price indices is to use an arithmetic weighted index of the form
pt = ␣i p it i

where the ␣i are again the output shares of commodity. This index assumes that even when the relative price of an item increased in a given year its relative output was unchanged. An arithmetic weighted index thus will show more year to year fluctuations in prices. The component price series have been constructed in a standard fashion. Prices from different sources were combined into a single series by running regressions of the form


ln(p ikt ) = ␣t D t + ␤k DUNITk + ␪i DLOCi + ␧ikt t

k

i

where the Dt ’s are a set of indicator series for each year, the DUNITk ’s are indicators for the unit of measurement and the DLOCi ’s are indicators for the location or the quality of the output. The logarithmic form was chosen to allow for consistent proportional differences in the level of prices across different price series as a result of measurement or quality differences. The reason for doing this is that even with agricultural commodities there could be considerable quality differences. Thus the average price of butter, in pence per pound, in the years 1815–1827 from five different sources utilized was as follows: Irish Imports, 4.87, Bethlem Hospital, 12.87, Greenwich Hospital, 9.52, Lord Steward (King’s household), 19.87, Navy Victualling, 9.43. The king, unsurprisingly, got better butter than did the inmates of the insane asylum (though the inmates did better than sailors in the navy)! And butter imported by the barrel from Ireland was a lot cheaper than butter bought by institutions by the pound. In earlier years, prices typically come from accounts that ran from Michaelmas (September 29) to Michaelmas.1 Where the date within the year is not given these prices have been attributed to the year of the following January since most of the account falls in that calendar year.2 The weights of the different products changed over time, in the way discussed below. Thus the overall index was composed by splicing together the various sub-period indices, generally using a five year

44

GREGORY CLARK

overlap to establish the relative levels. Appendix Table A.5, lists the various price sources and the numbers of place-years of price data they contribute.

OUTPUT WEIGHTS The shares of commodities in net output in English agriculture changed over time. Since the price of individual commodities moved in different ways, the weighting thus matters to the movement of the overall price index. There are a number of sources on weights for the late nineteenth century, after agricultural statistics were established in 1866.3 Before 1866 the weightings have to be inferred indirectly. Since before the late eighteenth century there were few imports or exports of food products I infer some of the weightings of farm output from the consumption patterns of workers in mid-nineteenth century England. Henry Rew, for example, estimates that in 1892 the British population consumed 12 lbs. of cheese, 15 lbs. of butter and 15 gallons of milk per capita (Rew, 1892, p. 272). This implies relative weights for cheese, butter and milk for the years before 1866 of 0.2, 0.39 and 0.41 based on relative prices in the nineteenth century. Rew (1904) also reports on late nineteenth century studies of meat consumption relative to dairy products. One important consideration is the much greater importance of wood products before 1760. In the years after 1810, as witnessed by charity farmland holdings, timber was a small share of output from English agriculture: construction timber was mainly imported. Further coal had replaced firewood as the main source of heat and energy. But before the great rise of coal output and of timber imports in the eighteenth century, all the wood for construction and most of the energy for heating homes, making bricks, and smelting iron was produced domestically. Wood output must have been a much greater share of agricultural production. I estimate, for the reasons enumerated below in the section on wood and fuel, that 10% of output from English agriculture in the years before 1760 was wood. Since wood land required little labor input, the gross output per acre had to be much higher on arable and pasture land to equalize rental values between woodland and arable and pasture. This implies that as much as 15–20% of the land area before 1760 was devoted to wood production, though much of this would be timber and fuel in hedgerows. Because the computation is complex the weights were first constructed as sub-weights for a series of largely arable products (grains, straw, potatoes, hops, mustard seed, saffron), for a series of largely pastoral products (meat, dairy, tallow, wool, hay and eggs), and for wood (faggots, bark, timber). For the arable component of the index the weights employed by sub-period are given in Table 1. For the years between 1540 and 1867 these weights are largely guesses. For the medieval period we actually get quite reasonable information on relative outputs

The Price History of English Agriculture, 1209–1914

45

Table 1. Weights of the Arable Components on the Index by Period. 1209–1550 1551–1760 1760–1850 1851–1882 1883–1914 Price 1860–1869 Wheat Rye Barley Oats Peas Beans Potatoes Hops Straw Mustard Seed Saffron

0.530 0.000 0.244 0.108 0.049 0.000 0.000 0.000 0.049 0.010 0.010

0.450 0.018 0.234 0.126 0.018 0.018 0.000 0.027 0.100 0.010 0.000

0.418 0.018 0.212 0.109 0.018 0.018 0.051 0.027 0.128 0.000 0.000

0.406 0.000 0.206 0.107 0.016 0.016 0.115 0.030 0.103 0.000 0.000

0.219 0.000 0.219 0.193 0.016 0.016 0.184 0.051 0.102 0.000 0.000

6.269 s./bu. – 4.379 s./bu. 2.892 s./bu. 4.768 s./bu. 5.087 s./bu. 6.039 s./cwt. 137.52 s./cwt. 21.65 s./ton – –

Sources: Post-1851, Fletcher (1961) and Turner (2000, pp. 283–295, 1551–1851) weights were obtained by keeping the grain and straw proportions similar to those post 1851, but reducing the potato and hops output. For 1200–1550 weights were obtained from the production figures derived from manorial accounts in Clark (1991).

from manorial accounts, though there are important uncertainties about how much of gross output was consumed as fodder within the farm sector. Table 2 gives the weights used within the “meat” component of the index, which also includes tallow. Table 3 shows the weights used to combine “meat,” dairy, wool, eggs and hay into a general pasture index. Here the weights were kept largely unchanged over time, except that the share of wool in pasture output was set at 10% before 1883, and 7% after. The final index was a combination of arable, pasture and wood with the weights here given in Table 4. The weighting before 1867 is again speculative. But since after the end of free trade initially the bulk of food imports was of arable products the assumption has been that before 1850 arable was typically about 10% more of output than pasture. Finally in earlier years, before extensive timber imports, the development of the coal industry, and Table 2. Weights of the Components of the “Meat” Index by Period. Period

Beef

Mutton

Pork

Bacon

Tallow

Eggs

1209–1549 1550–1849 1850–1882 1883–1914

0.00 0.30 0.29 0.31

0.00 0.30 0.24 0.19

0.00 0.20 0.24 0.35

0.00 0.05 0.08 0.00

0.67 0.10 0.10 0.10

0.33 0.05 0.05 0.05

Prices 1860–1809 6.675 d./lb. 6.346 d./lb. 6.521 d./lb. 16.313 d./lb. 5.071 d./lb. 8.352 d./doz. Sources: Post-1851, Fletcher (1961) and Turner (2000, pp. 283–295), Pre-1851 Clark (1991).

46

GREGORY CLARK

Table 3. Weights of the Components of the Pasture Index by Period. Period

Meat

Dairy

Wool

Hay

1209–1759 1760–1849 1850–1882 1883–1914

0.49 0.53 0.57 0.50

0.22 0.22 0.23 0.28

0.24 0.15 0.10 0.07

0.05 0.10 0.10 0.15

–

–

21.39 d./lb

96.1 s./ton

Prices, 1860–1869

Sources: Post-1849, Fletcher (1961) and Turner (2000, pp. 283–295). Pre-1850, Clark (1991).

Table 4. Weights of the Components of Farm Output Index by Period. Period 1209–1539 1540–1759 1700–1759 1760–1849 1850–1882 1883–1914

Arable

Pasture

Wood

Cider/Honey

0.60 0.60 0.50 0.54 0.50 0.32

0.30 0.32 0.40 0.44 0.50 0.68

0.08 0.08 0.10 0.02 0.00 0.00

0.02 0.00 0.00 0.00 0.00 0.00

Sources: Post-1849, Fletcher (1961) and Turner (2000, pp. 283–295). Pre-1850, see text and Clark (1991).

the widespread use of brick for building purposes it is assumed that agriculture had to supply both the needs for energy and for construction materials in the forms of firewood and timber for building.

ARABLE PRODUCT PRICES Appendix Table A.1 gives the detailed price series for eleven arable products, counting saffron as arable. Hops prices are quoted net of the excise tax levied between the 1660s and 1860s.4 In constructing an index of arable prices overall where individual price quotes were missing the index was interpolated using the prices of the other products. Within the arable sector relative prices changed significantly over the years. Figure 1 displays, as a 10 year average, the prices of barley, peas and oats, relative to wheat. After 1810–1819, with increasing imports of grains into Britain, from Ireland and from Europe and later the Americas, the relative prices of all other grains increased steadily relative to wheat. From 1200 to 1800 the general trend is a modest decline in other grain prices relative to wheat.

The Price History of English Agriculture, 1209–1914

47

Fig. 1. Barley, Peas and Oats Prices Relative to Wheat, 1209–1909. Source: Appendix Table A.1.

Campbell et al. (1993) report the kcal content per bushel of various grains if they were completely milled. These imply, for our bushel measure, that a bushel of barley had 82% of the calorie content of a bushel of wheat, while a bushel of oats had 74% of the calorie content.5 This in turn implies that throughout the preindustrial period both barley and oats were a much cheaper source of food energy than wheat. In the years before 1350 oats cost only 54% as much per calorie as wheat. From 1350 to 1550 the cost was only 47% in calorie terms. Only in the late nineteenth century did the relative costs of barley and oats approach their relative calorie content. The relatively high prices of wheat calories in the pre-industrial period seems to reflect just the constraints of arable rotations in Europe where it was impossible to produce a wheat crop more than once every three or four years on arable land, despite the strong preference for wheat in consumption. Thus the poor ate porridge and the rich wheat bread.

PASTURE PRICES Appendix Table A.2 gives the detailed price series for eleven pasture products (where I have including eggs in pasture). In forming these series into an overall pasture series the butter, cheese and milk/cream series were first formed into a

48

GREGORY CLARK

dairy series (weighting each series equally), and beef, mutton, pork, bacon, tallow and eggs were formed into a meat series (weighting as in Table 2). Before 1600 it is very hard to find meat prices by the pound as most meat was sold by the animal, or the joint. Peter Bowden uses these animal prices to estimate the movement of meat prices (Bowden, 1967). But an unstated assumption in that case is that the weights of animals did not change over time. Here I prefer to extend the meat price series in earlier years by using the prices of tallow, an animal product sold by the pound, and of eggs. Appendix Table A.4 shows these meat and dairy series, in each case set to average 100 in 1860–1869. As with arable outputs the relative prices of the different pastoral outputs were changing over time. Figure 2 shows pasture prices relative to the cost of hay, which indexes the cost of fodder, a major cost in the production of animal products, with 1860–1869 set to 1. Interestingly there is little trend in this series all the way from 1600–1609 to 1910–1914. A pound of meat or a gallon of milk cost the same relative to a ton of hay in the early twentieth century as in the early seventeenth century. If there had been major gains through animal breeding in the efficiency with which animals converted fodder into animal products then we would expect to see these ratios fall over time. Thus the price series here suggest little productivity growth in the pastoral sector in the years 1600–1914. Before the sixteenth century, however, pasture outputs were relatively expensive compared

Fig. 2. Pasture Output Prices relative to Hay Prices, 1300–1900. Source: Appendix Table A.3.

The Price History of English Agriculture, 1209–1914

49

Fig. 3. Arable Prices Relative to Pasture (1860–1869 = 100).

to hay, suggesting possible improvements in livestock breeding and husbandry practices in these years. Thus the relative price of pastoral products compared to hay was between two and three times as great in 1300–1399 as in 1600–1699. This implies that sometime in the sixteenth century there must have been significant gains in the efficiency of the production of pastoral products. Figure 3 shows the ratio of arable to pasture prices by decade. From 1209 to 1500 there is little change in this ratio. But from 1500 to 1600 arable prices rose much more than pastoral prices. Since the inputs for arable and pasture were not that dissimilar it implies that productivity growth was much greater in the pastoral sector in these years. After 1600 the ratio of arable to pastoral prices changed little until the 1810–1819 decade. Again this implies little change in the relative productivity of the arable and pastoral sectors in these years. After the 1810s the prices of arable products fall steadily relative to those of pasture. This fall pre-dates the well known decline in arable relative to pasture prices of the late nineteenth century.6 The source of this longer term decline in relative arable prices is at least in part the increasing importance of imports of foodstuffs and raw materials. Before refrigerated ships arable products were more easily transportable so imports reduced arable prices more. Also Britain seems to have had a comparative advantage in the production of pastoral products. Thus as imports increased and arable prices fell relative to pasture prices English agriculture became increasingly focused on pastoral production. But these price trends may also reflect in part differential

50

GREGORY CLARK

productivity gains in the arable and pastoral sector of the farm economy, with the arable gains greater. For a lot of land in England can be used for either arable or pasture, so if there had not been some gains in relative arable productivity an even more complete conversion to pasture crops would have occurred.

WOOD PRODUCTS The wood index was constructed as an equally weighted index of firewood (faggots, bavins, billets, etc. as well as turf) and of construction wood. Construction timber was sold by the cubic foot, the load, and the ton. The load was sometimes 50 cubic feet of wood, and sometimes 40. The construction timber series was constructed as the price per cubic foot of timber. The principle woods mentioned are oak, ash and elm. Oak was more expensive than ash or elm and this is controlled for in the regression estimates. Material described as “timber” was used though this may also have covered planks and other more manufactured outputs. Because of a scarcity of prices in some years “deals” prices were included even though these are manufactured timber products. Firewood was sold in various forms. The price here is given as the average price in shillings per “hundred” bundles of faggots, where “hundred” probably means 120. Appendix Table A.3 gives the price series each type of wood, and for wood overall as an index. Now that we have price series for timber and firewood we can consider the likely share of output from wood production in pre-industrial England. By the 1860s coal consumption per person in England and Wales was nearly 3.5 tons per year. Most of this was not consumed for domestic heating, but coal consumption per capita for domestic purposes was at least 0.73 tons (Church, 1986, p. 19) reports this figure for 1855. Coal used for domestic consumption is estimated for Britain in 1700 at as low as 0.2 tons per capita (Hatcher, 1983, pp. 68, 409). Incomes were lower in 1700, as Clark (2004) shows. But if we assume an income elasticity of demand for fuel of 1, and that the relative price of fuel and other goods did not change then consumers in 1700 would be expected to consume at least the equivalent of 0.5 tons of coal per capita. Thus they would have to consume the equivalent of 0.3 tons of coal in the form of wood and turf. Assuming that a pound of coal contains 12,000 Btu. and a pound of wood (dry weight) 8,600 Btu., and also that a cubic foot of firewood weighs 29.4 pounds, this implies that agriculture in 1700 should have been producing 2.3 million tons of firewood (dry weight), the equivalent of 175 million cubic feet of firewood.7 Wood was also used for fuel in brick making, iron and steel, salt making and pottery making. Iron production in England in the early eighteenth century was

The Price History of English Agriculture, 1209–1914

51

a very modest 17,000 tons annually. Yet each ton seems to have required about 1,800 cubic feet of wood. That implies that 30.6 m. cubic feet of firewood were used for iron production each year in 1700. Thus in total we expect the agricultural sector would have had to produce at least 200 million cubic feet of firewood per year circa 1700. The construction industry in the 1860s imported annually into England the equivalent of 6.9 cubic feet of timber per person.8 Again with an income elasticity of 1 the average person in 1700 would consume nearly 5 cubic feet of construction timber. That would imply that agriculture produced 27 million cubic feet of construction timber circa 1700. To estimate how many acres of land this required and the value of this output is difficult because of the irregularity of measures of firewood. This, as noted, was often sold by the hundred faggots (which actually equaled 120) where the size of the bundle and its density is unknown. Alternately it was sold as the load of faggots or billets, the load of firewood, or occasionally by a recognized volume measure such as the cord. Thus to estimate the amount of land required and the cost per unit of energy for firewood I instead use modern estimates of the productivity of coppiced woodland in England as a basis for the estimate. The average reported annual yield of coppiced wood of different species in recent years, including an allowance for the small branches which would be bundled into faggots, is 1.27 tons per acre of dried wood, or 92.5 cubic feet of wood.9 To produce 200 million cubic feet of firewood annually through coppices in 1700 would thus require 2.16 million acres of land devoted to firewood and iron production (330,000 acres for iron production alone). As a check on this method we can also consider the cost of a Btu derived from firewood compared to a Btu derived from coal. In 1700–1749 the average retail cost per million Btu for coal was 9.03 d. The output of firewood per acre Q, measured in cubic feet, must be such that Rent = pQ − wL − rK ⇒

Q=

Rent + wL + rK p

The average rental of land (including tithe and taxes) was about 9 shillings per year (Clark, 2002). I assume, somewhat arbitrarily, that 25% of the cost of firewood was in cutting, binding and carting. Since firewood was cut only once every 12 years or more, there was also a considerable capital cost involved in production of such fuel. The rent had to be paid long in advance of receipts. When all this is factored in the value of firewood per acre would have to average 14 shillings per year to produce a net rent of 9 shillings. This implies that a cubic foot

52

GREGORY CLARK

Fig. 4. Coal and Firewood Prices (d. per m. btu).

of firewood sold for 1.82 d in 1700–1749. Consequently firewood cost 0.062 pence per dry pound, or 7.18 d. per million Btu. This implies that energy from wood and coal had about the same cost per btu before 1700, but that after 1700 at the time when coal was increasing greatly in use, its cost per btu was also rising relative to wood. Figure 4 shows, for comparison, the price per million Btu from of coal compared to the price per million Btu from firewood by decade from 1480–1489 to 1860–1869. In part the strange price behavior of wood relative to coal after 1700 may be explained by the fact that coal and firewood sold in different markets: coal in cities such as London where people were concentrated and transport was cheap, firewood in the smaller towns and the countryside close to the sources of firewood where transport costs were minimized. Oliver Rackham estimates that firewood was rarely consumed more than 20 miles from where it was produced.10 The cost of firewood in London may have been much greater than I estimate here where I assumed cutting and carting costs were only 25% of the final delivered cost. A final check on the reasonableness of these production and acreage estimates comes from the price of firewood per cord at Harting in Sussex in the early eighteenth century, one of the few occasions where the price was given by a clear volume. Assuming a cord contained 80 cubic feet of actual timber (as in modern estimates), the average price of 8.4 shillings per cord implies a price of 1.26 d. per cubic foot, which is even lower than my rent based estimate of 1.82 d.

The Price History of English Agriculture, 1209–1914

53

Fig. 5. Wood Prices Relative to all Farm Prices, 1260–1860.

Fig. 6. The Bowden Index Compared to the New Index. Note: The two indices are set to equality in the decade 1741–1950. Source: Appendix Table A.3. Bowden (1967), Bowden (1985).

54

GREGORY CLARK

Figure 5 below just shows an index of coal, firewood and construction wood prices relative to the general price index for farm output from 1500 to 1829 as a way of comparing trends in costs. Firewood and coal were both relatively cheap compared to all farm output in the late sixteenth and early seventeenth centuries, and both got relatively expensive in the early eighteenth century. Construction timber was more expensive per cubic foot than firewood, and production rates must have been much lower. Here I can again estimate the number of acres required by comparison of the price per cubic foot circa 1700 (about 11.5 d.) with the annual rent and tithe of land. Since construction timber was cut only once every 25 years or more, the capital cost involved in production of such wood was greater. But cutting and carting costs were said to be somewhat lower. Assuming that cutting and carting costs were 20% of the final cost of timber, this again implies that the net value of output per acre would have to be about 14 shillings to produce a net rent of 9 shillings. This implies output would be only Table 5. The Bowden Farm Price Index Relative to the New Index by Decade. Decade 1500 1510 1520 1530 1540 1550 1560 1570 1580 1590 1600 1610 1620 1630 1640 1650 1660 1670 1680 1690 1700 1710 1720 1730 1740

Bowden/Clark (Set Equal to 1.00 in 1740–1749) 0.78 0.80 0.77 0.79 0.83 0.80 0.84 0.90 0.92 0.84 0.84 0.85 0.84 0.85 0.83 0.89 0.85 0.86 0.92 0.89 0.97 0.96 0.95 0.99 1.00

The Price History of English Agriculture, 1209–1914

55

14.7 cubic feet per acre. Thus the 27 million cubic feet of construction timber would have required an additional 1.84 million acres of woodland. Adding firewood and construction timber we find that at least an implied 4 million acres of land was devoted to wood and fuel production circa 1700, or 15% of the farmland area of England and Wales. About half the output was for each type of product. Thus I weight the wood index equally for firewood and construction timber.

THE OVERALL INDEX The last column of Appendix Table A.4 reports the overall annual price index. Figure 6 shows how this index compares with the Bowden indices for the years 1500–1749. Even though the series have important common components there is a fair amount of divergence from that series by year and by longer periods. In particular prices on the Bowden series rise by about 20% more between 1500 and 1750 than they do on this new series. Table 5 shows the Bowden index by decade relative to the new index. As can be seen deviations by decade are significant for the sixteenth century. These deviations imply that attempts to measure

Fig. 7. Price Indices for 1867–1914. Sources: Appendix Table A.3. Afton and Turner (2000, pp. 1910–1911).

56

GREGORY CLARK

productivity growth in English agriculture using the ratio of input prices (rents, wages and returns on capital) to output prices would show much less productivity growth from 1500 to 1750 on the Bowden series than on the new series constructed here. For the other period where we have other series to compare with this new series, the fit is better. The new series corresponds reasonably well with series constructed by Bellerby, and by Afton and Turner for 1867–1914. Figure 7 shows these other two series relative to the new one for these years. But even though here all three series use substantially the same data, differences in the weighting of the component series, the treatment of the data, and the splicing of individual series, still lead to some deviations even within this relatively short period. Thus prices decline about 10% more on my new series between the 1860s and the 1890s than on the Afton and Turner or Bellerby series.

NOTES 1. Thought the most popular account year was the one beginning in Michaelmas, there were many variations, and where the date was not indicated the price was placed in the calendar year that the majority of the account lay within. 2. By contrast Beveridge and Thorold Rogers both date prices according to the calendar year of the beginning of the account year. 3. Fletcher and Turner discuss these weightings. 4. The details of the tax were obtained from Dowell (1884). 5. Campbell et al. (1993), p. 41. 6. See Turner (2000), pp. 302–305. 7. On the weight of wood per cubic foot see Rollinson and Evans (1987). 8. This is assuming that U.K. imports were distributed according to population except that the Irish consumed half as much per head because of lower incomes. 9. See Begley and Coates (1961), Evans (1984), Rollinson and Evans (1987). Hammersley (1973), pp. 604–605, notes that woodland can produce “up to 100 cubic feet per year.” 10. Rackham (1980), p. 144.

ACKNOWLEDGMENTS This research was supported by NSF grants #SES 91–22191 and #SES 02–41376. I am grateful to my colleagues Peter Lindert and Alan Olmstead for comments, and to David Jacks and Derek Stimel for research assistance. John Munro very kindly shared data he had inputted from the Beveridge Collection with me. I am

The Price History of English Agriculture, 1209–1914

57

grateful also to the archivists at the University of Saskatchewan Library, and the Robbins Library at LSE for their help in accessing the Farmer and Beveridge archives.

REFERENCES Archival Cheshire Record Office. DTW/2343/1/3/1 (Hay, 1810–1820). Dorset Record Office. PE/SYM/IN/6 (Suet, Butter, Cheese, 1750–1763). Essex Record Office. Gidea Hall D/DBe A1-A13 (Hay, 1748–1750, 1753–1755, 1769–1779, 1782, 1791–1800), Bentall and Linnett Families D/DE A10 (Hay, 1804, 1808, 1815, 1819), Bassom Mss. D/DU 84/18–20 (Hay, 1832–1836, 1847–1850), Watkinson of Black Notley D/DU 224/1 (Hay, 1804–1810, 1816). Guildhall Library, London. Brewer’s Company, Pratt’s Almshouse, Accounts. Mss. 5491/1–3, 5473/1–5 (firewood). King’s College Library, Cambridge. King’s College Bursar, Mundum Books, 1786–1866 (hay, timber, firewood). Lancashire Record Office. DDF/37 (Hay, 1782, 1789), DDHe/62/25–29, 37a (Hay, 1820–1823, 1825, 1847, 1854–1866), PR 2851/5/11–12 (Hay, 1841, 1843, 1847–1848, 1852–1859, 1861–1866). London School of Economics, Robbins Library. Beveridge Papers. Bath Town Accounts, 1568–1664 (I10). Battle Abbey, 1309–1533 (H5-H11). Cambridge, King’s College, 1501–1559 (G2). Cambridge, King’s Hall, 1344–1543 (G2). Cambridge, Trinity College, 1547–1718 (G2). Canterbury Cathedral, 1274–1385 (D2). Croyland Abbey, 1253–1322 (G12). Dover Almshouse, 1629–1690 (H13). Durham, 1278–1540 (C4-C5). Hinderclay Manor, 1262–1404 (G14). Hull Grain Prices, 1708–1799 (I14). Norwich Cathedral, 1273–1536 (G5). Norwich, 1552–1603 (I15). Nottingham Guildhall Records, 1485–1787 (I15). Pelham Papers, Waldron, Brightling and Bibleham Iron Works, 1628–1715 (H12). Sion Monestary, 1441–1539 (I4). Westminster Abbey, 1293–1540 (P1-P7). Winchester Estates, 1208–1453 (A40-A42). York City, 1584–1763 (I15). University of Saskatchewan Library, Saskatoon, Saskatchewan. Farmer Papers (Wheat Prices 1209–1499). Stafford Record Office. D260/M/E/160–1, 169, 207 (Hay, 1850–1851, 1861–1862, 1865–1874, 1876–1879, 1881).

Printed Afton, B., & Turner, M. (2000). The statistical base of agricultural performance in England and Wales, 1850–1914. In: E. J. T. Collins (Ed.), The Agrarian History of England and Wales (Vol. VII 1850–1914, Part II, pp. 1,757–2,140). Cambridge: Cambridge University Press. Atkins, P. J. (1977). The milk trade of London, c. 1790–1914. Ph.D. thesis, University of Cambridge.

58

GREGORY CLARK

Bailey, F. A. (1953). The churchwarden’s accounts of prescot, Lancashire, 1523–1607 (Vol. 104). Preston: Record Society for the Publication of Original Documents Relating to Lancashire and Cheshire. Publications. Barmby, J. (1888). Churchwardens’ accounts of Pittington and other parishes in the diocese of Durham, 1580–1700. Publications of the Surtees Society (Vol. 84). Durham: Andrews and Company. Barmby, J. (1896). Memorials of St Gile’s, Durham. Publications of the Surtees Society (Vol. 95). Durham: Andrews and Company. Begley, C. D., & Coates, A. E. (1961). Estimating yield of hardwood coppice for pulpwood growing. Forestry Commission Report on Forest Research 1959/1960 (pp. 189–196). London: Her Majesty’s Stationary Office. Beveridge, W. H. (1929). A statistical crime of the seventeenth century. Journal of Economic and Business History. Beveridge, W. H. (1939). Prices and wages in England (Vol. 1): The mercantilist era. London: Longmans and Green. Botelho, L. A. (1999). Churchwardens’ accounts of Cratfield, 1640–1660 (Vol. 42). Woodbridge: Boydell & Brewer, Suffolk Records Society. Boulton, J. (2000). Food prices and the standard of living in London in the ‘century of revolution,’ 1580–1700. Economic History Review, 53(3), 455–492. Bowden, P. J. (1967). Statistical appendix. In: J. Thirsk (Ed.), The Agrarian History of England and Wales (Vol. IV, pp. 814–870). Cambridge: Cambridge University Press. Bowden, P. J. (1985). Statistical appendix. In: J. Thirsk (Ed.), The Agrarian History of England and Wales (Vol. VII, pp. 827–902). Cambridge: Cambridge University Press. Brinkworth, E. R. C. (1964). South Newington churchwardens’ accounts, 1553–1684 (Vol. 6). Banbury, Oxford: Banbury Historical Society. Records publications. Burgess, C. (2000). The pre-reformation records of all Saints’, Bristol (Vol. 2). Stroud: Bristol Record Society, Publications (Vol. 53). Campbell, B., Galloway, J., Keene, D., & Murphy, M. (1993). A medieval capital and its grain supply: Agrarian production and distribution in the London region c. 1300. Historical Geography Research Series, #30 (London, Institute of British Geographers). Clark, G. (1991). Labour productivity in english agriculture, 1300–1860. In: B. M. S. Campbell & M. Overton (Eds), Agricultural Productivity in the European Past (pp. 211–235). Manchester: Manchester University Press. Clark, G. (2002, December). Land rental values and the agrarian economy: England and Wales, 1500–1912. European Review of Economic History, 6(3), 281–309. Clark, G. (2004). The condition of the working-class in England, 1200–2000: Magna Carta to Tony Blair. Working Paper, UC-Davis. Church, R. (1986). The history of the British coal industry (Vol. 3), 1830–1913. Oxford: Clarendon Press. Coleman, M. C. (1984). Coleman-in-the-Isle: A study of an ecclesiastical manor in the thirteenth and fourteenth centuries. Woodbridge, Suffolk: Boydell Press. Doree, S. G. (1994). The early churchwardens’ accounts of Bishops Stortford, 1431–1558. Hertfordshire record publications; Vol. 10. Hitchin, Hertfordshire: Hertfordshire Record Society. Dowell, S. (1884). A history of taxation and taxes in England (Vol. 2). London: Longmans Green. Erskine, A. M. (1981). The accounts of the fabric of exeter Cathedral, 1279–1353. Part 1: 1279–1326. Torquay: Devon and Cornwall Record Society, New Series (Vol. 24).

The Price History of English Agriculture, 1209–1914

59

Erskine, A. M. (1983). The accounts of the fabric of exeter Cathedral, 1279–1353. Part 2: 1328–1353. Torquay: Devon and Cornwall Record Society, New Series (Vol. 26). Evans, J. (1984). Silviculture of Broadleaved Woodland. Forestry Commission Bulletin 62. London: Her Majesty’s Stationary Office. Farmer, D. L. (1988). Prices and wages. In: H. E. Hallam (Ed.), The Agrarian History of England and Wales (Vol. II, 1042–1350). Cambridge: Cambridge University Press. Farmer, D. L. (1991a). Marketing the produce of the countryside, 1200–1500. In: E. Miller (Ed.), The Agrarian History of England and Wales, 1348–1500 (Vol. III, pp. 324–430). Cambridge: Cambridge University Press. Farmer, D. L. (1991b). Prices and Wages, 1350–1500. In: E. Miller (Ed.), The Agrarian History of England and Wales, 1348–1500 (Vol. III, pp. 431–525). Cambridge: Cambridge University Press. Finberg, H. P. R. (1951). Tavistock abbey: A study in the social and economic history of Devon. Cambridge: Cambridge University Press. Fletcher, T. W. (1961). The great depression of english agriculture, 1873–1896. Economic History Review, 13, 417–432. Gayer, A. D., Rostow, W., & Schwartz, A. J. (1953). The growth and fluctuations of the British economy, 1790–1850. Oxford. Gras, N. S. B. (1915). The evolution of the english corn market. Cambridge, MA: Harvard University Press. Hammersley, G. (1973). The charcoal iron industry and its fuel, 1540–1750. Economic History Review, 24, 593–613. Hill, J. W. F. (1956). Tudor and Stuart Lincoln. Cambridge: Cambridge University Press. Hill, J. W. F. (1966). Georgian Lincoln. Cambridge: Cambridge University Press. John, A. H. (1989). Statistical appendix. In: G. E. Mingay (Ed.), Agrarian History of England and Wales, 1750–1850 (Vol. VI, pp. 1156–1177). Cambridge: Cambridge University Press. Marsh, B. (1913–1939). Records of the worshipful company of carpenters. Vol. 2, Warden’s Account Book, 1438–1516. Vol. 4, Warden’s Account Book, 1546–1571. Oxford: Oxford University Press. Maslen, M. M. (1993). Woodstock Chamberlains’ accounts, 1609–1650. Oxfordshire Record Society (Vol. 58). Oxford: Oxfordshire Record Society. Mellows, W. T. (1939). Peterborough local administration; parochial government before the reformation. Churchwardens’ accounts, 1467–1573. Northamptonshire Record Society, Publications Vol. 9. Kettering, Printed for the Northamptonshire Record Society by the Northamptonshire Printing and Publishing Company. Mercer, F. R. (1928). Churchwardens’ accounts at Betrysden, 1515–1573. Kent Archaeological Society. Records Branch. Kent records (Vol. V, p. 3). Ashford, Printed for the Records Branch by Headley brothers. Mitchell, B. R., & Deane, P. (1971). Abstract of British historical statistics. Cambridge: Cambridge University Press. Northeast, P. (1982). Boxford churchwardens’ accounts, 1530–1561. Suffolk Records Society (Vol. 23). Woodbridge, Suffolk: Boydell Press for the Suffolk Records Society. Parliamentary Papers (1903). Report on wholesale and retail prices in the U.K. in 1902 with comparative statistical tables for a series of years (Vol. LXVIII). Plomer, H. R. (1915). The churchwardens’ accounts of St. Nicholas, Strood (B. M. add. ms. 36,937). Kent Archaeological Society. Records Branch. Kent records (Vol. V). Rackham, O. (1980). Ancient woodland: Its history, vegetation and uses in England. London: Edward Arnold.

60

GREGORY CLARK

Rappaport, S. (1989). Worlds within worlds: Structures of life in sixteenth-century London. Cambridge: Cambridge University Press. Rew, R. H. (1892). An inquiry into the statistics of the production and consumption of milk and milk products in Great Britain. Journal of the Royal Statistical Society, 55(2), 244–286. Rew, R. H. (1904). Observations on the production and consumption of meat and dairy products. Journal of the Royal Statistical Society, 67(3), 413–427. Rollinson, T. J. D., & Evans, J. (1987). The yield of sweet chestnut coppice. Forestry Commission Bulletin, 64. London: Her Majesty’s Stationary Office. Sauerbeck, A. (1886). Prices of commodities and precious metals. Journal of the Statistical Society of London, 49(3), 581–648. Stallard, A. D. (1922). The transcript of the churchwardens’ accounts of the parish of tilney all Saints, Norfolk, 1443–1589. London: Mitchell Hughes and Clarke. Swayne, H. J. F. (1896). Churchwardens’ accounts of S. Edmund & S. Thomas, Sarum, 1443–1702, with other documents. Wilts Record Society, Publications. Salisbury, Bennett Brothers. Thorold Rogers, J. E. (1866). A history of agriculture and prices in England (Vol. 2). Oxford: Clarendon Press. Thorold Rogers, J. E. (1882). A history of agriculture and prices in England (Vol. 3). Oxford: Clarendon Press. Thorold Rogers, J. E. (1888). A history of agriculture and prices in England (Vol. 6). Oxford: Clarendon Press. Thorold Rogers, J. E. (1902). A History of agriculture and prices in England (Vol. 7, Part 1). Oxford: Clarendon Press. Tooke, T., & Newmarsh, W. (1857). A history of prices and of the state of the circulation from 1792 to 1856. New York: Adelphi. Turner, M. (2000). Agricultural output, income, and productivity. In: E. J. T. Collins (Ed.), The Agrarian History of England and Wales (Vol. VII 1850–1914, Part I, pp. 224–320). Cambridge: Cambridge University Press. Wardle, F. D. (1923). The accounts of the chamberlains of the City of Bath, 1568–1602. Somerset Record Society (Vol. 38). London: Butler and Tanner. Weatherill, L. (1990). The account book of Richard Latham, 1724–1767. Oxford: Published for the British Academy by Oxford University Press. Weaver, F. W., & Clark, G. N. (1925). Churchwardens’ accounts of Marston, Spelsbury, and Pyrton. Oxfordshire Record Society, record series Vol. 6. Oxford: Issued for the Society. Woodward, D. (1995). Men at work: Labourers and building craftsmen in the towns of Northern England, 1450–1750. Cambridge: Cambridge University Press.

The Price History of English Agriculture, 1209–1914

61

APPENDIX Table A.1. Arable Prices. Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253

0.29

0.21

0.10

0.22

0.36 0.28

0.24 0.18

0.12 0.11

0.35 0.25

0.24

0.16

0.12

0.20

0.30 0.46 0.52 0.52 0.39 0.54

0.18

0.12

0.33

0.38 0.34 0.24 0.33

0.18 0.18 0.14 0.16

0.43 0.40 0.34 0.39

0.22 0.29 0.61 0.50 0.58

0.17 0.39 0.36 0.36

0.10 0.18 0.19 0.20

0.22 0.37 0.72 0.53

0.46 0.43 0.38

0.32 0.27

0.18 0.18

0.46 0.40

0.38 0.38 0.42

0.21 0.30

0.16 0.22

0.30 0.39

0.25 0.35 0.65 0.62 0.34

0.16 0.22 0.42 0.47 0.22

0.12 0.16 0.25 0.26 0.15

0.25 0.29 0.62 0.69 0.27

0.35 0.36 0.54

0.33 0.24 0.43

0.19 0.27

0.46 0.33 0.60

62

GREGORY CLARK

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302

0.35 0.31

0.21 0.19

0.14 0.14

0.29 0.29

0.75 0.85

0.52 0.54

0.26 0.29

0.72 0.75

0.32

0.22

0.44

0.37 0.31

0.20 0.17

0.44 0.36

0.27 0.39 0.34 0.45 0.61 0.46 0.51 0.54 0.40 0.54 0.43 0.37 0.40 0.37 0.39 0.51 0.50 0.35 0.44 0.37 0.23 0.27 0.37 0.46 0.45 0.45 0.54 0.61 0.67 0.41 0.50 0.57 0.44 0.37 0.37

0.17 0.27 0.27 0.22 0.29 0.26 0.29 0.29 0.27 0.32 0.26 0.24 0.22 0.22 0.28 0.26 0.26 0.22 0.27 0.24 0.17 0.19 0.24 0.29 0.25 0.28 0.26 0.32 0.26 0.19 0.31 0.31 0.26 0.21 0.21

0.28 0.53 0.60 0.62 0.83 0.67 0.75 0.64 0.44 0.79 0.60 0.42 0.49 0.48 0.61 0.57 0.72 0.46 0.51 0.48 0.26 0.30 0.47 0.63 0.63 0.46 0.70 1.07 0.88 0.63 0.67 0.62 0.51 0.39 0.39

0.45 0.47 0.51 0.44 0.42 0.47 0.46 0.52 0.40 0.54 0.61 0.85 0.85 0.61 0.77 0.80 0.63 0.81 0.57 0.52 0.63 0.61 0.77 0.77 0.76 0.52 0.65 0.51 0.32 0.36 0.56 0.74 0.65 0.69 0.96 1.01 0.98 0.61 0.75 0.62 0.62 0.59 0.58

11.09

7.37

6.00

5.00 4.39

8.00

1.24

The Price History of English Agriculture, 1209–1914

63

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351

0.53 0.46 0.64 0.60 0.52 0.65 0.82 0.88 0.93 0.58 0.57 0.63 0.79 1.75 1.72 0.90 0.48 0.59 0.71 1.47 1.04 0.78 0.86 0.59 0.45 0.54 0.73 0.74 0.81 0.90 0.61 0.50 0.51 0.57 0.51 0.43 0.39 0.68 0.46 0.52 0.48 0.68 0.44 0.49 0.82 0.77 0.49 0.71 0.97

0.27 0.34 0.43 0.45 0.40 0.44 0.50 0.58 0.53 0.43 0.41 0.45 0.54 1.05 1.12 0.62 0.33 0.39 0.52 0.87 0.64 0.43 0.59 0.42 0.32 0.33 0.37 0.44 0.50 0.60 0.40 0.34 0.35 0.39 0.35 0.26 0.20 0.39 0.33 0.36 0.36 0.44 0.34 0.36 0.51 0.52 0.26 0.45 0.70

0.22 0.18 0.24 0.24 0.23 0.25 0.25 0.34 0.35 0.25 0.28 0.28 0.32 0.79 0.47 0.31 0.22 0.22 0.33 0.39 0.32 0.24 0.35 0.25 0.25 0.22 0.27 0.25 0.31 0.37 0.26 0.22 0.22 0.22 0.20 0.18 0.13 0.19 0.22 0.22 0.22 0.27 0.21 0.24 0.28 0.29 0.16 0.29 0.46

0.40 0.44 0.58 0.63 0.53 0.48 0.68 0.84 0.68 0.53 0.48 0.55 0.56 1.61 1.50 0.83 0.36 0.48 0.48 0.95 0.82 0.50 0.72 0.55 0.54 0.47 0.71 0.58 0.85 0.93 0.65 0.48 0.44 0.44 0.34 0.27 0.53 0.44 0.40 0.42 0.55 0.39 0.44 0.65 0.61 0.34 0.53 0.78

1.32 1.39 1.24

6.00 5.00

1.61 1.83 1.53 1.32

4.95 1.39 3.33

4.67 4.98 6.18 5.42 3.54 2.41 1.53 1.53

1.32 1.37 0.75 1.24

3.08

1.86

4.00

1.45 1.96

18.00

3.33 6.32

64

GREGORY CLARK

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400

1.30 0.63 0.53 0.69 0.74 0.77 0.84 0.71 0.77 0.74 0.59 0.86 1.08 0.85 0.65 0.64 0.89 0.83 1.82 0.86 0.74 0.93 0.67 1.06 0.98 0.53 0.47 0.46 0.71 0.74 0.62 0.60 0.59 0.51 0.64 0.53 0.43 0.41 0.71 1.05 0.61 0.39 0.45 0.46 0.53 0.75 0.65 0.60 0.67

0.80 0.61 0.70 0.32 0.50 0.56 0.58 0.67 0.59 0.69 0.81 0.77 0.74 0.71 0.66 0.72 0.57 0.60 1.12 0.76 0.51 0.58 0.45 0.60 0.64 0.47 0.37 0.32 0.39 0.41 0.39 0.39 0.46 0.39 0.40 0.36 0.29 0.27 0.40 0.68 0.41 0.32 0.33 0.40 0.39 0.46 0.54 0.40 0.44

0.48 0.44 0.25 0.26 0.27 0.37 0.35 0.29 0.25 0.31 0.34 0.36 0.33 0.29 0.28 0.32 0.33 0.34 0.50 0.36 0.26 0.30 0.26 0.35 0.36 0.25 0.22 0.22 0.24 0.27 0.24 0.23 0.29 0.26 0.28 0.19 0.18 0.19 0.26 0.38 0.25 0.21 0.23 0.24 0.23 0.28 0.35 0.24 0.26

0.85 0.82 0.40 0.43 0.55 0.89 0.69 0.54 0.48 0.65 0.67 0.82 0.69 0.54 0.48 0.69 0.71 0.74 1.24 0.72 0.44 0.60 0.55 0.70 0.81 0.49 0.46 0.44 0.46 0.47 0.48 0.43 0.57 0.53 0.69 0.37 0.37 0.40 0.50 0.84 0.44 0.43 0.42 0.48 0.53 0.60 0.61 0.55 0.51

2.05 2.25 2.48 2.26

19.00

2.56 2.83 13.33 1.83 1.97 2.23 3.94 2.96 2.48 3.41 3.10 6.00 1.50 3.74 5.60 2.96 9.29 1.99 3.64 2.34

0.75 0.89 3.74 1.75 2.05 0.43 1.58 3.09 1.24 3.39 2.90 2.99 4.49 2.95

12.00 20.00 20.00 20.00 17.49 18.00 16.00 16.00 16.00 16.00 13.33 13.33 13.33 13.33 13.33 14.60 13.82 6.67 14.00 12.59 14.00 12.00 10.32 14.00 16.00 19.60 18.47

The Price History of English Agriculture, 1209–1914

65

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448

0.84 1.06 0.72 0.54 0.51 0.43 0.49 0.69 0.80 0.95 0.68 0.58 0.55 0.51 0.52 0.74 0.97 0.66 0.82 0.54 0.75 0.59 0.47 0.50 0.57 0.51 0.46 0.47 0.84 0.83 0.68 0.48 0.77 0.58 0.54 0.66 0.62 1.14 1.35 0.87 0.43 0.47 0.48 0.47 0.43 0.66 0.69 0.60

0.45 0.66 0.51 0.42 0.37 0.34 0.32 0.39 0.47 0.61 0.48 0.39 0.34 0.37 0.39 0.50 0.56 0.39 0.37 0.43 0.39 0.34 0.40 0.35 0.34 0.36 0.34 0.30 0.44 0.46 0.34 0.32 0.40 0.44 0.33 0.27 0.29 0.44 0.78 0.51 0.27 0.27 0.32 0.28 0.27 0.27 0.30 0.31

0.26 0.30 0.29 0.24 0.21 0.22 0.21 0.24 0.34 0.32 0.31 0.24 0.25 0.27 0.25 0.29 0.28 0.25 0.22 0.29 0.22 0.21 0.22 0.21 0.24 0.19 0.21 0.22 0.32 0.29 0.23 0.19 0.23 0.24 0.19 0.19 0.20 0.31 0.35 0.22 0.16 0.16 0.27 0.21 0.19 0.22 0.21 0.19

0.54 0.64 0.68 0.41 0.41 0.47 0.42 0.48 0.53 0.76 0.64 0.50 0.42 0.50 0.56 0.62 0.57 0.69 0.29 0.56 0.55 0.39 0.39 0.56 0.42 0.46 0.37 0.37 0.46 0.97 0.46 0.69 0.70 0.56 0.49 0.49 0.46 0.57 1.05 0.68 0.43 0.34 0.49 0.43 0.35 0.49 0.49 0.36

2.94 2.82 2.96 3.17 2.47 2.96 2.99 3.03 3.34 2.89 2.66 3.41 3.49 2.77 3.45 3.56 3.83 2.25 2.31 3.08 2.76 3.21 2.64 2.21 2.58 2.74 2.85 2.26 2.85 3.56 2.82 2.76 3.78 2.54 2.61 3.14 2.69 2.56 3.49 3.37 2.92 2.33 2.82 2.40 3.63

1.13

18.00 18.00 18.00 15.16

2.50 2.67 1.58 1.83 1.67

14.69 13.73 14.00 14.00 13.49 14.00 13.66 12.00

1.15 1.33 0.67 1.50 1.17 1.00 1.53 0.83 1.17 1.00

1.00 0.92

2.50 0.67

14.00 10.16 17.33 17.00 14.00 11.83 10.00 11.45 13.86 14.00

9.49 13.64 13.33

16.00

1.63

14.67

1.50

1.02

2.00

13.33 16.24 9.33 13.33 9.20 11.55

66

GREGORY CLARK

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496

0.58 0.62 0.71 0.73 0.62 0.52 0.49 0.60 0.61 0.76 0.73 0.62 0.88 0.90 0.50 0.42 0.51 0.52 0.66 0.70 0.72 0.73 0.78 0.69 0.52 0.50 0.54 0.66 0.64 0.82 0.87 0.65 0.69 0.93 1.19 0.84 0.63 0.58 0.72 0.71 0.71 0.70 0.74 0.75 0.62 0.59 0.56 0.50

0.33 0.31 0.33 0.31 0.32 0.39 0.32 0.33 0.29 0.30 0.28 0.34 0.43 0.39 0.37 0.24 0.49 0.41 0.37 0.27 0.31 0.29 0.34 0.32 0.34 0.40 0.31 0.29 0.31 0.31 0.27 0.34 0.33 0.58 0.59 0.46 0.38 0.29 0.38 0.33 0.41 0.33 0.38 0.37 0.39 0.29 0.27 0.26

0.18 0.18 0.19 0.18 0.22 0.23 0.18 0.20 0.19 0.19 0.23 0.24 0.22 0.23 0.15 0.15 0.22 0.19 0.20 0.19 0.18 0.19 0.19 0.22 0.23 0.22 0.15 0.15 0.19 0.16 0.21 0.16 0.18 0.27 0.22 0.17 0.17 0.16 0.19 0.17 0.22 0.19 0.19 0.22 0.16 0.17 0.17 0.21

0.40 0.33 0.44 0.32 0.33 0.49 0.39 0.29 0.46 0.34 0.34 0.48 0.54 0.39 0.39 0.39 0.36 0.78 0.34 0.44 0.36 0.39 0.58 0.54 0.42 0.36 0.39 0.39 0.49 0.49 0.54 0.58 0.85 0.92 0.78 0.59 0.52 0.51 0.33 0.36 0.40 0.58 0.43 0.39 0.32 0.34

2.15 2.80 3.01 2.64 3.31 3.23 3.06 5.35 2.33 3.06 3.04 2.48 4.03 1.88 5.14 2.83 2.16 2.01 2.69 2.97 2.29 2.69 2.21 2.69 2.69 2.69 2.69 2.37

1.05

1.83

1.42 1.41 1.47 1.25

1.35

4.00

0.92 0.71

2.47 2.13 2.34 2.75 2.89 1.77

1.67

13.96 12.00 8.33 5.85 6.67 11.54 10.41 15.32 13.06 11.15 11.48 10.92 10.00 13.33 12.67 14.24 14.98 12.00 9.09 12.70 13.82 10.67 13.33 16.67 11.05 17.82 13.33 13.27 13.33 9.25 20.57 11.16 21.77 8.16 11.25

2.00 14.93

2.14 2.77 2.38 2.31 2.24 2.14 2.14 2.76 3.12

0.91

0.77 1.25

14.60 18.47 14.60 17.52 8.72 14.97 18.21

The Price History of English Agriculture, 1209–1914

67

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544

0.76 0.67 0.67 0.58 0.92 0.97 0.85 0.88 0.71 0.68 0.71 0.75 0.39 0.45 0.53 0.75 1.11 0.71 0.66 0.85 0.58 0.79 0.74 1.02 1.13 0.89 0.84 0.81 0.63 0.63 0.82 1.87 1.12 1.16 1.25 1.17 1.12 0.88 1.12 1.24 1.12 0.86 0.82 0.85 1.05 1.03 1.09 1.31

0.28 0.30 0.32 0.31 0.31 0.39 0.32 0.38 0.41 0.38 0.27 0.32 0.40 0.33 0.24 0.26 0.60 0.57 0.31 0.49 0.42 0.43 0.39 0.45 0.61 0.47 0.33 0.30 0.33 0.34 0.50 0.79 0.77 0.50 0.44 0.49 0.59 0.49 0.42 0.56 0.40 0.36 0.43 0.54 0.56 0.48 0.66 0.39

0.20 0.26 0.26 0.20 0.23 0.21 0.25 0.23 0.27 0.21 0.22 0.23 0.21 0.20 0.22 0.22

0.27 0.23 0.25 0.29 0.35 0.31 0.25 0.27 0.38 0.27 0.32 0.50 0.25 0.32 0.37 0.41 0.39 0.34 0.42 0.37 0.40 0.32 0.36 0.32 0.41 0.34 0.37 0.37

0.33 0.34 0.61 0.36 0.53 0.55 0.34 0.45 0.40 0.43 0.42 0.34 0.35 0.47 0.66 0.49 0.52 0.66 0.68 0.84 1.12 0.69 0.62 0.70 0.70 0.79 1.06 0.66 0.51 0.69 0.70 0.70 0.67 0.55

0.69 1.08 0.71

0.30 0.27 0.28 0.38 0.38 0.35 0.31 0.25 0.25

0.38 0.73 0.45 0.32 0.40 0.29 0.46 0.58 0.78 0.62 0.42 0.37 0.36 0.51 0.61 0.43 0.59 0.42 0.46 0.49 0.46 0.48 0.60 0.58 0.53

24.48

18.25 15.82 21.96 23.98

2.58 2.82 3.71 2.48 2.54 2.92 3.10 3.20 2.94 2.32 2.03 2.45 3.16 2.20 3.29 2.46 2.74 2.57 2.76 2.78 3.31 3.37 3.90 2.42 3.96 2.93 2.20 2.31 2.74 2.12 3.22 3.43 2.25 2.05 2.39 2.82 4.14 2.88 2.48 3.08 3.23 2.72 3.34 2.79 2.72 3.27 2.36 2.82

0.75 1.00 2.36

0.92 1.25 1.70

1.75 2.00 1.49 1.50 1.65 3.11 2.44

2.67 3.27 1.00 1.29 0.83 1.63 3.33 2.75

21.33 17.82 11.55 11.63 8.54 10.38 6.71 9.42 17.85 18.47 26.40 18.67 10.60 11.15 13.75 13.33 14.67 12.53 11.53 13.99 11.76 16.00 14.17 14.60 11.55 17.33 34.67 16.65 16.00 18.43 19.60 15.49 8.33 45.21 19.31 12.00 18.62 32.00 16.74 16.65 22.56 13.33

1.00 2.00

14.93 12.00

68

GREGORY CLARK

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592

1.56 2.08 0.89 0.75 1.22 1.91 2.47 1.89 1.38 1.34 1.85 2.79 3.15 1.19 1.50 1.55 1.93 1.95 1.94 2.33 1.51 1.57 1.63 1.49 1.61 1.61 1.35 1.79 1.97 2.78 1.95 2.07 2.28 1.98 2.03 1.65 2.40 2.29 2.06 1.92 2.35 3.28 3.14 2.00 2.23 2.79 2.57 2.01

0.45 0.96 0.56 0.43 0.65 0.56 0.93 0.69 0.69 0.77 1.05 2.23 1.91 0.73 1.25 0.94 0.75 1.09 0.99 0.94 0.83 0.89 1.47 1.15 0.79 0.71 0.93 0.94 1.47 0.94 1.01 1.20 1.26 0.96 0.97 1.21 1.11 1.04 1.08 1.05 1.51 1.20 1.16 1.05 1.46 1.87 1.11

0.44 0.56 0.43 0.35 0.43 0.97 0.74 0.68 0.67 0.82 0.83 0.90 1.66 0.57 0.60 0.59 0.80 0.88 0.78 0.74 0.54 0.83 0.74 0.83 0.80 0.67 0.66 0.70 0.74 0.87 0.71 0.72 0.78 0.76 0.60 0.84 0.67 0.74 0.83 0.85 0.81 0.98 1.34 1.05 1.15 1.31 1.59 1.17

0.87 0.99 0.61 0.57 0.71 1.70 1.23 1.38 0.82 1.35 1.54 2.57 3.35 1.14 0.87 1.13 1.29 2.23 2.08 1.47 0.95 1.22 1.31 1.55 1.35 1.11 1.07 1.09 1.88 2.03 1.65 1.27 1.35 1.40 1.41 1.54 1.36 1.64 1.57 1.63 1.67 2.40 2.49 1.61 1.91 2.20 3.23 1.99

0.65 0.69 0.51 0.49 0.53 1.05 1.21 1.03 0.85 0.71 0.84 1.69 2.42 0.62 0.78 0.84 1.32 1.28 1.37 1.17 0.68 1.06 1.00 1.07 1.09 0.76 0.73 0.70 1.20 1.53 1.15 0.89 1.09 1.04 0.85 1.07 0.99 1.18 1.12 1.19 1.10 1.68 1.81 1.16 1.29 1.54 2.22 1.44

25.78 31.25 42.09 47.19

22.01 26.48 36.90 24.11 25.26 31.12 40.23 42.81 38.60 85.66 38.34 24.09 53.37 64.44 60.41 80.78 54.63 63.93 54.29 57.10 40.28 42.63 43.99 37.49 37.47 81.17 57.16 54.13 44.11 43.79 46.57 57.26 49.65 42.64 54.98 85.45 35.28 65.78

2.98 2.87 3.46 3.38 3.47 4.50 5.17 6.59 5.32 5.86 7.36 4.11 7.57 5.75 4.85 5.50 6.32 8.32 13.62 9.95 6.23 6.55 7.53 7.83 7.84 5.81 5.64 6.43 7.33 6.74 6.00 6.11 7.06 6.51 6.94 7.92 7.97 8.14 7.75 7.51 7.36 7.45 7.79 6.26 5.48 7.60 10.02 8.45

2.65 1.63 1.67 3.00 2.74 2.58 2.00 1.75 3.00 2.67 4.08 2.00 1.17 2.20 3.08 7.00 4.00 3.87 2.00 2.50 3.33 3.61 2.17 2.08 3.33 4.00 5.00 4.50 4.67 4.00 2.75 3.00

2.75 5.00 2.75 3.00 2.00 3.75 3.38 7.26 6.00

10.00 17.33 26.67 26.67 28.44 26.67 29.33 24.00 24.00 19.32 22.31 21.83 18.67 21.33 28.34 48.00 42.67 32.00 28.28 27.36 32.00 26.67 29.33 22.63

The Price History of English Agriculture, 1209–1914

69

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640

2.32 3.37 4.21 4.77 5.83 4.79 2.97 3.20 3.20 2.89 3.05 3.01 3.26 3.06 3.34 4.81 4.58 3.49 3.77 3.98 4.56 4.57 3.93 3.89 4.63 4.47 3.55 2.98 3.52 4.80 4.95 4.23 4.70 4.43 3.66 3.31 3.75 5.01 5.75 4.92 4.77 4.90 4.83 4.76 4.88 5.09 4.12 3.74

3.12 3.12 3.47 3.47 3.22 3.15 3.25

4.69 4.70 5.10 4.69 2.71 4.70 5.09

1.13 1.26 2.00 2.11 2.90 2.11 1.57 1.74 1.96 1.42 1.19 1.24 1.62 1.57 1.52 1.95 2.14 1.81 1.65 2.06 2.25 2.28 2.22 2.16 1.86 1.82 1.57 1.36 1.47 2.37 2.36 1.86 1.97 2.13 1.54 1.63 2.17 2.70 2.88 2.33 2.34 2.62 2.36 2.37 3.15 3.25 2.26 2.00

0.88 0.92 1.61 1.56 1.84 1.47 1.14 0.96 1.78 0.83 0.86 1.07 1.13 1.16 1.14 1.17 1.37 1.28 1.31 1.78 1.52 1.65 1.70 2.02 1.30 1.27 1.22 1.24 1.12 1.35 1.25 1.25 1.31 1.65 0.97 1.27 1.37 1.84 2.06 1.35 1.42 1.62 1.50 1.61 1.90 1.91 1.42 1.14

1.69 1.75 2.74 2.38 3.44 3.07 1.91 1.99 2.78 2.76 1.93 1.98 1.78 2.06 2.14 2.40 2.79 2.76 2.28 3.01 3.00 2.98 2.89 3.61 2.48 1.92 2.03 2.04 2.24 3.07 2.54 2.21 2.52 2.80 1.76 2.29 2.95 3.55 4.56 2.58 2.50 3.23 3.07 3.42 4.33 4.44 2.95 2.06

1.19 1.30 1.95 1.82 2.77 2.26 1.27 1.21 1.88 2.00 1.30 1.36 1.36 1.60 1.25 1.54 2.17 1.74 1.57 2.53 2.71 2.68 2.23 1.72 1.32 1.49 1.70 1.76 2.48 1.87 1.74 1.85 1.96 1.53 1.92 2.62 2.28 2.35 2.14 2.98 2.98 4.13

90.50 45.70 40.54 62.50 42.91 46.65 41.69 51.52 54.41 99.08 169.39 104.80 90.31 95.40 89.58 128.86 66.54 58.04 29.86 38.23 57.71 50.91 76.77 67.10 76.39 75.08 111.34 97.06 123.19 101.80 94.04 48.03 39.97 44.19 68.04 98.05 165.95 65.71 59.98 68.88 102.33 101.94 100.63 98.56 98.08 88.98 60.57 58.63

8.98 10.74 13.29 10.40 11.47 10.55 8.25 10.56 12.17 9.73 8.86 9.96 8.62 15.21 14.90 12.80 11.56 13.86 11.75 20.60 18.24 12.00 13.99 15.28 12.63 12.37 14.70 14.51 15.96 16.25 14.84 13.83 14.46 13.70 13.70 13.79 14.91 18.49 17.17 14.34 14.86 16.07 16.09 17.17 17.13 20.73 14.70 11.75

3.00 3.19 3.50 3.78 5.49 4.04 5.68 3.50 5.02 4.04 3.35 4.08 4.08 4.04 4.67 5.81 4.08 4.08 4.92 10.55 5.82 5.69 4.40 4.85 4.85 4.85 4.85 4.85 8.53 5.26 4.85 4.76 5.32 7.64 6.96 6.71 8.24 10.78 8.27 9.69 9.69 9.69 9.69 8.04 7.63 8.81 7.78 8.04

70

GREGORY CLARK

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688

4.45 4.04 4.30 4.06 4.04 4.51 6.33 7.31 6.79 6.49 5.98 4.94 3.64 2.79 2.71 4.17 4.44 5.63 5.93 5.42 6.19 6.73 4.93 4.80 4.14 3.35 3.08 3.30 4.36 3.93 4.01 3.77 4.11 5.93 5.21 3.47 3.73 4.77 5.02 3.99 4.40 4.12 4.00 4.14 4.36 3.41 3.48 2.93

3.39 3.91 3.91 3.47 6.25 4.17

3.12 1.74 1.96 2.21 3.12 3.12

3.65 4.86 3.39 2.34 3.47 2.60 2.34 2.08 4.17 3.47 2.60 3.43 2.60 3.65 2.08 2.08

2.19 2.10 1.79 1.89 2.11 2.30 3.09 3.48 3.16 2.67 2.30 2.19 2.17 1.57 1.59 2.01 2.54 2.50 2.81 2.51 2.85 3.01 2.55 2.14 2.00 1.76 1.77 1.77 2.20 2.07 1.80 1.71 1.97 2.72 2.34 1.84 2.25 2.16 2.05 1.76 2.06 2.42 2.05 2.15 2.14 2.21 1.92 1.71

1.47 1.38 1.62 1.16 1.26 1.54 1.79 2.55 2.01 2.41 1.42 1.80 2.05 1.27 1.13 1.31 1.59 1.63 2.11 1.55 1.50 1.95 1.67 1.52 1.47 1.46 1.15 1.37 1.28 1.34 1.44 1.19 1.30 1.43 1.52 1.33 1.49 1.43 1.38 1.28 1.50 1.63 1.43 1.56 1.60 1.57 1.33 1.28

2.81 2.13 2.66 1.77 2.54 2.42 3.09 4.35 4.58 5.25 3.01 3.74 4.13 3.01 2.40 2.72 3.44 3.24 3.29 3.21 2.96 3.43 3.14 2.99 2.14 2.76 2.11 2.19 2.33 2.68 3.13 2.19 2.87 3.49 2.77 2.60 3.09 2.76 2.42 2.74 2.60 3.12 3.21 2.64 3.95 2.90 2.03 2.78

3.34 1.85 1.76 2.28 2.04 1.56 3.15 3.07 2.68 2.38 2.02 4.00 1.67 1.67 2.24 2.08 2.76 2.38 2.46 1.63 2.27 2.25 2.39 1.64 1.99 2.15 1.81 1.95 2.29 2.33 1.92 1.99 2.78 2.31 2.00 2.29 2.50 1.87 2.10 2.08 2.76 2.24 2.47 3.29 2.68 2.37 2.38

81.44 107.15 196.72 85.61 71.29 49.70 63.19 102.06 101.95 179.70 46.05 114.99 72.62 117.30 136.30 134.00 177.90 105.17 104.38 71.75 77.12 62.53 64.58 104.04 86.20 135.91 78.19 76.04 69.44 75.80 68.30 71.85 144.42 112.00 104.24 122.37 75.41 60.60 54.06 43.64 99.38 112.00 89.18 110.05 120.90 134.97 196.85 108.82

19.10 16.09 14.57 17.29 16.96 13.54 17.38 21.22 17.46 21.21 14.07 15.10 16.99 14.80 16.68 19.62 19.99 15.74 12.95 15.62 15.05 15.72 15.69 14.53 16.01 21.24 16.75 11.04 14.86 13.91 13.98 13.78 15.91 15.40 16.04 12.63 17.33 15.26 16.47 15.33 15.66 15.86 14.78 15.36 24.80 16.51 12.64 15.62

8.53 8.53 7.50

The Price History of English Agriculture, 1209–1914

71

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 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

2.85 3.27 3.11 4.45 5.44 5.32 4.34 5.27 5.93 6.69 5.65 4.24 3.38 2.93 2.88 3.72 3.10 2.89 2.97 3.89 6.63 6.92 4.77 4.13 4.39 5.22 3.69 4.38 4.17 3.66 3.30 3.98 3.72 3.49 3.72 3.55 4.28 4.81 4.32 5.84 4.94 3.62 3.18 2.57 2.90 3.62 4.27 4.12

2.34 2.34 2.34 2.80 3.12 3.12 2.86 2.60 4.17 4.56 3.12 2.86 2.08 1.96 1.96

2.08 2.88 5.21 3.86 3.50 2.84 3.33 3.38 2.82 2.64 2.76 2.47 2.29 2.68 2.27 2.15 2.76 2.84 3.21 3.39 3.73 4.44 3.53 2.49 2.30 2.11 2.19 2.78 3.30 2.90

1.50 1.40 1.33 1.76 2.56 2.18 1.98 2.12 2.30 2.70 2.81 2.11 1.80 1.85 1.67 1.78 1.63 1.92 2.02 2.27 2.65 2.85 2.57 2.17 2.00 2.35 2.41 2.16 2.04 1.90 2.23 2.65 2.28 1.83 1.90 2.34 2.28 2.29 2.48 3.22 2.88 2.11 2.13 2.01 1.90 1.88 1.92 2.05

1.11 1.26 1.23 1.42 1.69 1.54 1.42 1.56 1.31 1.70 1.53 1.68 1.32 1.42 1.33 1.35 1.42 1.64 1.28 1.60 1.59 1.56 1.55 1.64 1.46 1.51 1.42 1.41 1.37 1.35 1.42 1.62 1.40 1.25 1.28 1.63 1.42 1.37 1.37 1.66 1.81 1.43 1.34 1.36 1.27 1.33 1.46 1.51

1.97 2.07 1.79 2.21 3.31 3.32 2.82 3.21 3.16 3.60 3.53 3.49 2.24 2.64 2.17 2.42 2.36 3.04 2.40 2.23 3.41 3.95 3.68 3.37 2.88 2.98 3.20 2.71 2.50 2.45 2.49 3.66 2.43 2.33 2.53 3.31 3.33 2.71 3.05 2.99 3.03 2.43 2.30 2.28 2.02 2.30 2.41 2.40

1.65 1.69 1.78 2.12 2.64 2.55 2.75 3.02 2.60 2.93 2.87 2.62 1.64 1.94 1.94 2.02 1.96 2.09 2.14 2.51 2.16 2.41 2.44 2.29 2.01 2.51 2.76 2.41 2.17 2.09 2.22 2.74 2.23 1.86 2.11 2.52 2.42 2.50 2.20 2.66 2.73 2.39 2.38 2.24 2.19 2.21 2.24 2.12

2.52

66.02 45.79 42.33 69.11 70.50 77.34 112.44 179.54 203.59 240.82 208.85 115.90 68.34 113.56 88.31 72.81 147.66 110.63 109.04 98.79 97.79 143.93 120.34 71.47 89.11 147.43 167.40 146.74 138.32 91.26 154.67 93.28 75.72 50.97 89.35 101.71 103.75 136.69 69.63 65.84 63.04 57.80 83.98 140.63 163.94 122.25 114.26 109.76

12.65 14.94 14.78 15.23 15.37 15.37 13.82 15.08 15.22 14.97 22.83 16.59 14.04 14.75 16.38 18.51 14.81 17.04 14.56 16.10 13.42 15.36 16.94 14.72 14.08 16.21 14.45 14.45 15.25 14.99 15.25 17.41 17.04 14.16 13.75 15.86 15.50 17.70 18.66 15.53 16.45 16.45 17.59 18.95 10.94 13.70 11.02 18.55

72

GREGORY CLARK

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 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 1779 1780 1781 1782 1783 1784

3.52 3.31 3.67 5.04 5.10 3.26 2.68 2.59 2.92 3.61 3.43 3.61 3.58 3.49 3.92 4.41 4.39 3.76 3.45 4.30 6.13 4.91 3.65 3.34 3.24 3.80 4.06 4.80 5.24 4.98 6.39 6.01 4.91 4.72 5.85 6.37 6.48 6.58 6.21 4.75 5.41 5.31 4.17 4.35 5.71 6.10 6.55 6.20

2.68 2.52 2.56 3.97 3.27 2.29 1.91 1.81 2.44 2.65 2.44 2.60 2.55 2.34 2.35 2.78 2.81 2.61 2.47 3.26 4.47 3.02 2.14 2.19 2.23 3.48 3.37 3.48 3.72 4.01 3.84 3.44 3.36 3.70 4.87 5.06 4.07 4.45 4.25 3.51 3.58 3.33 2.84 2.70 3.53 3.61 4.27 4.23

2.22 2.11 2.01 2.44 2.35 2.38 1.97 1.81 1.60 1.59 1.58 1.86 2.07 1.86 1.99 2.05 2.20 2.13 1.85 2.14 3.03 2.71 1.98 1.98 1.76 2.12 2.82 2.71 2.67 2.76 3.09 2.67 2.13 2.23 2.95 2.99 3.36 3.46 3.22 2.63 2.39 2.63 2.36 2.14 2.08 2.69 3.62 3.46

1.62 1.53 1.33 1.77 1.80 1.57 1.47 1.25 1.47 1.32 1.19 1.28 1.72 1.55 1.59 1.62 1.47 1.59 1.48 1.64 1.98 1.88 1.36 1.42 1.40 1.59 2.04 1.77 1.80 2.04 1.96 1.81 1.63 1.70 2.03 1.98 2.04 2.11 2.05 1.88 1.83 1.74 1.74 1.66 1.69 1.76 2.43 2.25

2.64 2.94 2.37 2.95 3.73 2.53 2.33 1.89 2.04 2.17 1.99 2.03 2.28 2.10 2.28 2.38 2.58 2.81 2.21 2.49 3.73 3.20 2.83 2.28 2.46 2.39 3.32 2.62 2.81 3.38 3.43 3.29 2.77 2.88 3.50 3.91 3.95 3.86 3.72 3.57 3.60 3.84 3.52 3.15 3.41 3.12 4.09 3.73

2.61 2.58 2.31 2.66 3.01 2.56 2.06 1.91 1.88 1.88 1.84 2.02 2.27 2.33 2.60 2.51 2.36 2.45 2.24 2.53 3.32 2.97 2.25 2.33 2.31 2.56 3.25 3.02 2.93 3.31 3.04 2.70 2.89 2.75 3.16 3.52 3.55 4.00 3.35 3.14 3.29 3.03 2.78 2.60 2.70 2.86 4.27 3.91

2.52

2.75 3.29 4.36 3.01 3.03 3.51 3.31 3.01 1.88 1.88 2.24 2.59 2.14 2.50 2.24 2.06 2.34 2.61 2.14 2.34 3.71 2.92 3.28 2.54

144.46 88.23 56.55 74.39 70.37 63.30 63.23 83.99 125.25 150.11 99.43 97.15 107.02 119.11 92.03 92.75 83.43 71.11 55.31 57.01 69.18 69.28 97.40 135.25 80.90 65.81 70.07 61.71 173.04 122.63 78.89 114.33 93.46 147.54 121.87 170.55 148.53 145.98 97.31 105.97 89.27 97.82 62.11 61.29 62.65 63.08 131.16 113.21

17.30 20.26 14.80 15.82 17.09 18.21 16.84 13.72 13.03 15.13 12.88 12.42 15.56 14.16 15.05 13.90 20.44 20.71 16.88 18.26 21.14 18.10 18.51 20.55 16.84 16.49 24.01 22.96 26.62 24.27 18.49 18.95 21.29 21.95 19.93 23.50 29.79 24.06 30.00 30.37 33.86 21.52 22.71 24.41 23.81 23.06 22.48 34.79

The Price History of English Agriculture, 1209–1914

73

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 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 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832

5.62 4.99 5.22 5.62 6.31 6.64 5.97 5.49 5.98 6.24 8.05 9.74 6.73 6.56 7.84 12.86 14.84 8.22 7.15 7.25 10.49 9.56 9.47 9.39 11.64 13.05 11.73 14.38 13.46 9.68 8.14 8.89 12.16 10.82 8.84 8.26 7.21 5.63 6.23 7.57 8.49 7.19 6.73 8.26 7.51 8.22 7.71 7.00

3.64 3.45 3.66 3.48 3.83 4.15 4.13 3.72 4.52 4.88 6.00 6.03 4.16 4.04 4.56 7.11 9.87 5.38 4.62 4.65 6.78 6.04 6.26 7.23 7.09 7.43 5.66 9.18 9.59 5.97 5.08 5.09 6.95 6.84 6.98 5.20 4.10 2.83 3.98 5.18 5.28 5.14 5.02 4.27 4.35 4.48 5.00 4.32

3.01 3.09 2.83 2.80 2.74 3.01 3.15 3.43 3.95 4.10 4.19 4.34 3.61 3.69 4.13 6.92 8.69 4.55 3.26 3.72 5.94 5.10 5.24 5.26 5.70 5.83 5.12 8.09 7.09 4.53 3.67 4.11 5.98 6.53 5.55 4.10 3.15 2.65 3.82 4.41 4.85 4.16 4.56 3.98 3.94 3.95 4.61 4.01

2.07 2.30 2.05 1.99 1.90 2.32 2.26 2.23 2.61 2.83 3.11 2.79 2.21 2.42 3.23 4.44 4.49 2.47 2.61 2.94 3.44 3.34 3.44 4.04 3.81 3.47 3.34 5.39 4.67 3.11 2.86 3.29 3.93 3.93 3.42 2.93 2.36 2.19 2.78 3.01 3.11 3.23 3.42 2.73 2.76 2.96 3.07 2.48

3.68 3.93 4.00 3.47 3.31 3.67 3.70 3.92 4.72 5.56 6.26 5.39 3.79 3.64 4.73 7.86 7.92 4.55 4.57 4.72 5.42 5.54 7.38 10.90 9.80 6.89 5.81 8.31 9.91 6.54 4.97 4.44 6.33 5.91 6.68 5.11 3.84 3.14 3.73 4.37 5.69 5.69 6.04 4.69 4.22 4.52 4.86 4.54

3.79 4.34 4.01 3.14 3.02 3.82 3.55 3.95 4.18 4.21 5.29 4.56 3.10 3.42 5.14 7.43 7.84 4.17 3.87 4.46 5.29 5.26 5.56 6.66 7.12 6.75 5.83 8.49 9.74 5.75 4.57 3.95 5.77 7.24 7.42 5.41 4.55 3.71 3.66 4.39 5.31 5.73 6.78 5.15 4.84 4.75 4.92 4.57

2.03 2.46 2.40 1.80 2.21 1.93 3.02 2.12 1.79 2.61 4.13 2.52 2.45 2.24 2.96 6.38 4.73 2.52 2.99 2.86 2.24 2.38 2.80 3.36 3.22 3.15 3.01 4.24 5.20 3.30 3.30 3.48 3.97 4.04 4.18 3.57 3.46 4.47 2.96 3.59 3.84 4.20 3.97 2.77 3.43 3.79 3.59 2.39

111.40 104.52 128.11 192.12 152.54 127.57 109.48 117.52 109.88 217.55 100.30 117.27 124.24 124.48 279.94 374.67 322.40 91.09 171.17 108.30 108.54 155.82 116.19 95.43 90.20 124.85 173.13 150.15 262.79 198.55 186.35 220.02 340.82 305.13 91.43 78.62 61.07 73.26 138.52 156.85 370.10 152.04 97.40 93.87 141.86 236.16 117.30 146.70

23.53 25.43 25.21 25.99 29.19 22.35 22.57 25.82 29.76 49.99 44.25 38.38 23.38 38.00 43.50 49.00 38.12 50.17 34.25 53.62 43.15 52.87 42.38 39.75 57.67 70.75 56.42 40.83 43.76 35.75 44.04 37.37 50.50 58.17 31.00 33.50 40.33 44.75 51.80 47.50 42.75 41.80 33.12 35.38 47.50 34.00 37.75

74

GREGORY CLARK

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 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

6.53 5.35 4.77 5.88 6.77 7.83 8.57 8.04 7.80 6.94 6.07 6.21 6.16 6.63 8.46 6.12 5.37 4.88 4.67 4.94 6.46 8.78 9.05 8.39 6.83 5.36 5.30 6.46 6.71 6.72 5.43 4.87 5.07 6.05 7.81 7.73 5.84 5.69 6.87 6.91 7.11 6.76 5.48 5.60 6.88 5.63 5.31 5.38

4.11 4.09 3.79 4.17 4.34 4.39 5.25 4.62 4.59 4.12 3.82 4.24 4.06 4.38 6.12 3.80 3.21 2.91

3.33 3.52 3.63 3.98 3.68 3.81 4.79 4.42 3.98 3.33 3.58 4.08 3.84 3.96 5.36 3.82 3.36 2.91 3.08 3.55 4.12 4.48 4.32 5.10 5.23 4.31 4.16 4.55 4.49 4.36 4.21 3.72 3.70 4.65 4.97 5.34 4.90 4.30 4.50 4.64 5.02 5.59 4.78 4.37 4.93 4.99 4.22 4.11

2.23 2.54 2.67 2.80 2.80 2.72 3.14 3.11 2.72 2.33 2.22 2.50 2.73 2.87 3.48 2.49 2.12 1.99 2.25 2.31 2.55 3.38 3.32 3.05 3.03 2.97 2.81 2.96 2.88 2.74 2.57 2.44 2.65 2.98 3.15 3.41 3.15 2.77 3.05 2.81 3.08 3.50 3.48 3.18 3.14 2.95 2.64 2.80

4.30 4.47 4.34 4.36 4.60 4.50 4.72 5.28 4.81 4.19 3.75 3.75 4.31 4.58 5.45 5.09 4.14 3.36 3.19 3.58 4.51 5.46 5.39 4.97 4.71 4.82 4.53 4.69 4.28 4.35 4.10 4.11 4.29 4.39 4.80 5.18 4.53 4.37 4.59 4.55 4.55 4.86 4.84 4.57 4.49 4.13 4.04 4.26

4.12 4.26 4.51 4.56 5.11 4.50 4.89 5.39 5.08 4.53 3.58 4.13 4.69 4.80 6.01 4.76 4.04 3.68 3.58 3.80 3.77 5.62 5.54 5.15 5.23 5.04 5.05 5.16 5.01 4.39 4.38 4.61 5.02 5.48 5.34 5.81 5.34 5.24 5.18 4.91 5.17 5.75 5.51 4.91 4.44 4.86 4.71 4.96

3.30 4.59 3.72 3.51 4.80 3.11 5.98 4.64 4.06 4.20 2.70 4.00 3.47 5.11 10.18 4.06 7.30 4.72 5.11 5.11 4.20 4.12 5.27 4.66 5.90 5.04 4.87 6.27 6.99 6.43 5.55 4.76 5.05 5.68 7.08 6.48 6.12 6.96 6.36 6.82 7.11 6.56 6.04 7.15 7.34 6.84 7.37 7.27

105.05 117.30 91.58 78.10 78.10 62.18 168.75 102.60 165.69 147.92 86.68 68.92 91.58 220.20 152.82 270.42 130.78 315.74 206.92 179.44 108.59 109.52 94.17 92.37 160.25 166.00 156.54 131.91 162.11 138.42 164.76 136.43 74.05 84.74 83.25 101.64 124.25 122.02 158.05 115.06 113.32 116.30 92.94 115.31 114.31

30.50 31.25 39.50 32.00 39.00 42.00 39.50 38.00 41.25 37.75 49.50 31.25 37.25 32.50 32.25 27.00 31.50 24.75 25.25 26.00 29.50 35.50 28.25 26.00 27.25 27.50 26.50 32.25 34.00 36.00 31.00 27.50 30.00

The Price History of English Agriculture, 1209–1914

75

Table A.1. (Continued ) Year Wheat Rye Barley Oats Peas Beans Potato Hops Net Straw Mustard Saffron (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./bu.) (s./cwt.) Tax (s./cwt.) (s./load) Seed (s./bu.) (s./lb) 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 1914

5.50 5.47 5.04 4.32 3.98 3.76 3.94 3.86 3.61 3.87 4.49 3.67 3.19 2.77 2.80 3.17 3.66 4.12 3.11 3.26 3.24 3.41 3.24 3.44 3.60 3.43 3.71 3.88 4.48 3.84 3.84 4.21 3.84 4.23

3.97 3.87 3.96 3.81 3.74 3.30 3.15 3.45 3.21 3.57 3.51 3.25 3.18 3.04 2.73 2.85 2.92 3.37 3.18 3.10 3.13 3.19 2.82 2.78 3.02 3.00 3.12 3.21 3.33 2.87 3.38 3.81 3.38 3.37

2.64 2.65 2.60 2.46 2.50 2.30 1.97 2.03 2.15 2.25 2.42 2.40 2.27 2.07 1.76 1.79 2.05 2.23 2.06 2.13 2.23 2.45 2.08 1.98 2.10 2.22 2.28 2.16 2.29 2.10 2.28 2.61 2.31 2.54

4.23 4.14 4.13 3.62 3.52 3.27 3.06 3.05 3.43 3.44 3.73 3.61 3.31 2.98 3.00 2.96 2.86 3.31 3.42 3.63 3.83 3.81

4.82 4.73 4.62 3.98 3.75 3.79 3.87 3.72 3.81 3.61 4.00 3.75 3.47 3.11 3.19 3.28 3.27 3.55 3.72 3.79 4.10 4.13 4.10 3.77 4.11 4.44 3.19 4.35 4.24 4.46 4.44 4.56 4.49 4.24

6.71 7.26 7.18 6.80 6.54 6.23 6.96 6.50 7.05 6.54 7.49 6.26 6.25 6.92 6.49 6.27 5.50 5.65 5.61 4.84 5.14 4.94 6.50 5.58 8.77 7.97 6.57 6.39 7.52 8.08 9.15 6.85 6.27 10.54

117.05 230.37 208.50 156.31 93.19 72.32 73.31 91.95 88.97 116.05 124.75 127.24 138.67 101.64 73.81 71.07 79.27 104.87 111.58 99.65 98.91 103.63

76

GREGORY CLARK

Table A.2. Pastoral Product Prices. Year

1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256

Hay (s/ton)

Cheese (d./lb.)

Butter (d./lb.)

Milk (d./gal.)

Beef (d./lb.)

Mutton (d./lb.)

Pork (d./lb.)

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

0.35

1.39

2.83

0.36 0.39

1.22

2.73 2.28

0.38

2.73

0.33

2.93

0.35 0.38 0.36 0.35

2.68 2.63 2.88 2.63

0.37 0.46 0.39 0.36

1.61 1.82

3.22 3.07 2.78 2.73

0.36 0.42

1.61

4.11 4.26

0.44 0.46

0.95

3.97 3.92

0.41 0.44 0.54 0.41 0.37

1.31 1.43 1.47 3.25

0.51 0.43 0.45 0.48

1.18 1.43 1.26 1.40

4.11 4.37 4.71 3.77 3.36

3.72 3.72 3.17 3.36 4.66

Eggs (d./doz.)

The Price History of English Agriculture, 1209–1914

77

Table A.2. (Continued ) Year

1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304

Hay (s/ton)

2.02

Cheese (d./lb.)

4.45 4.63 3.84 8.67

Milk (d./gal.)

0.48 0.48

0.55 0.53 0.56 0.45 0.46 0.60

3.21

Butter (d./lb.)

0.55 0.56 0.53 0.56 0.52 0.51 0.69 0.55 0.57 0.56 0.55 0.62 0.53 0.55 0.51 0.50 0.49 0.45 0.49 0.51 0.51 0.52 0.45 0.52 0.54 0.45 0.50 0.52 0.44 0.54 0.43 0.61 0.58 0.55 0.48 0.54 0.45

0.82 0.94 0.84 0.80 0.84 0.90 1.03 0.84

Mutton (d./lb.)

Pork (d./lb.)

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

1.39 1.68 2.79

3.72 3.87 4.66

1.68 1.54 2.85 1.71 2.46 2.45 2.56 3.00 2.49 3.00 2.53 3.76 2.40 2.50

0.88 0.82

1.05 1.10 1.09 0.85 0.81 0.94 0.82 0.82 0.81 0.86 0.80 0.81 0.91 0.79 1.04 0.74 1.12 0.75 0.75 0.82 0.85 0.79 0.73 0.74 0.98 0.85 0.79 0.75 0.64 0.70

Beef (d./lb.)

1.51 1.69 1.25 1.38

3.00 2.79 2.63 3.38 2.47 2.44 2.00 2.57 2.10 2.08 1.85 2.12 2.09 2.34 2.55 2.74 2.40 2.39 2.19 2.09 3.30 2.66 2.30 2.16 1.60

6.14 4.47 6.14 4.47 4.42 4.91 5.10 5.46 5.15 5.01 4.96 5.15 5.80 6.79 6.50 8.08 7.93 6.55 7.04 5.90 5.75 5.70 5.70 6.19 6.35 6.40 6.55 6.94 6.50 5.66 6.14 5.51 3.77 5.05 4.86 5.70 7.30 6.65 6.19 5.51 5.56 5.70

Eggs (d./doz.)

0.35 0.31 0.26 0.26 0.61 0.31 0.31 0.36 0.41 0.47 0.34 0.42 0.39 0.42 0.41 0.41 0.42 0.36 0.44 0.49 0.38 0.33 0.41 0.42 0.37 0.33 0.35 0.39 0.33 0.39 0.35 0.39 0.36 0.35 0.37 0.39 0.40 0.39 0.39 0.41 0.36 0.41

78

GREGORY CLARK

Table A.2. (Continued ) Year

1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352

Hay (s/ton)

4.46 5.53 4.81 4.62 4.79 4.44

3.25 4.23 2.51 1.70 1.34 2.67 4.56 5.02 14.02 4.91 3.84 5.46 9.07 4.57 6.23 3.61 1.43 0.95 9.56 4.35 3.56 2.15 4.99 4.22 4.78 3.18 4.89 2.92 4.62

Cheese (d./lb.)

Butter (d./lb.)

Milk (d./gal.)

0.48 0.56 0.47 0.53 0.55 0.73 0.60 0.59 0.58 0.61 0.61 0.69 0.68 0.62 0.51 0.64 0.67 0.60

0.65 0.91 0.95 0.84 1.00 1.22 1.29 1.03 1.17 0.82 1.18 1.31 1.18 1.13 0.95 1.10 1.15 1.40 1.44 1.05 0.94 1.14 0.72 1.24 1.06 0.94 0.92 0.88 0.91 1.05 1.10 1.09 0.96 1.05 0.84 0.96 1.02 0.87 0.86 1.08 1.36 1.13 1.22 1.08 1.10 0.94 0.94 0.82

1.38 1.85 1.51 1.42 1.48 1.41 1.41 1.41 1.38 1.44 1.43 2.12 1.74

0.48 0.56 0.73 0.50 0.52 0.57 0.57 0.63 0.50 0.49 0.53 0.57 0.56 0.49 0.50 0.43 0.54 0.43 0.44 0.41 0.44 0.41 0.51 0.53 0.48 0.37 0.44 0.44 0.74

1.43 1.43

1.77 1.59 1.71

1.85 1.85

1.67 1.59 1.55 1.58

1.50

1.45 1.13 0.86

Beef (d./lb.)

Mutton (d./lb.)

Pork (d./lb.)

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

Eggs (d./doz.)

2.17 2.81 1.86 2.38 3.07 3.19 2.99 3.15 2.52 2.36 2.72 3.33 4.03 4.17 3.32 2.91 2.31 2.93 3.45 3.66 3.49 3.05 2.67 2.43 2.54 2.59 2.83 2.62 2.67 2.49 2.49 2.58 2.40 2.26 2.04 2.28 2.23 2.03 2.22 2.32 2.58 2.38 2.78 2.43 2.71 2.36 2.61 2.54

6.50 7.04 7.49 7.88 8.03 7.39 6.35 5.46 5.85 6.70 6.60 6.31 6.65 7.25 7.13 8.58 9.43 7.44 7.44 7.74 8.13 6.99 7.35 6.74 6.19 6.05 6.35 6.14 5.41 4.37 4.66 5.70 4.91 3.82 4.21 4.21 4.26 5.15 4.21 5.41 5.05 5.51 5.05 5.15 3.07 3.87 4.06 4.31

0.43 0.41 0.64 0.16 0.48 0.60 0.46 0.46 0.45 0.51 0.49 0.49 0.69 0.43 0.39 0.45 0.46 0.53 0.54 0.45 0.52 0.52 0.43 0.61 0.45 0.47 0.45 0.42 0.46 0.44 0.43 0.44 0.41 0.40 0.37 0.39 0.47 0.45 0.45 0.42 0.38 0.39 0.41 0.50 0.49 0.45 0.44 0.50

The Price History of English Agriculture, 1209–1914

79

Table A.2. (Continued ) Year

Hay (s/ton)

1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400

5.02 4.24 3.31 5.38 4.77 2.90 4.67 4.18 5.53 2.42 3.50 3.79 4.65 3.23 2.43 6.07 4.80 7.46 5.10 4.91 7.49 4.39 4.75 7.23 5.88 6.20 4.38 4.66 3.49 4.66 4.57 4.53 3.85 5.19 3.86 4.81 4.63 5.04 4.65 5.81 4.13 6.30 4.55

Cheese (d./lb.)

Butter (d./lb.)

0.60 0.54 0.52 0.56 0.63 0.65 0.53 0.58 0.70 0.58 0.45 0.49 0.54 0.54

1.01

0.54 0.45 0.48 0.46 0.50 0.44 0.47 0.54 0.40 0.46 0.50 0.45 0.45 0.50 0.44 0.50 0.50 0.45 0.48 0.48 0.50 0.40 0.50 0.46 0.50 0.48 0.56 0.47 0.54 0.56 0.49 0.51 0.46

1.13 1.14 1.05 1.14 1.03 1.30 1.14 0.94 1.22 1.17 1.19 1.20 1.17 1.21 1.23 1.62 0.91 1.16 0.95 1.04 1.20 1.42 1.12 1.08 1.08 1.08 1.07 1.12 1.20 1.06 1.27 1.02 1.30 1.16 0.82 1.08 0.94 1.18 1.08 1.02 1.34 1.26 1.07 1.03 1.07 1.06

Milk (d./gal.)

1.34 1.51 1.68 1.58 1.53 1.52 1.29 1.40 1.28 1.22 1.29

1.29 1.29 1.17 1.29 1.33 1.51

1.06 1.06 1.23 1.06 1.30 1.07

Beef (d./lb.)

Mutton (d./lb.)

Pork (d./lb.)

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

Eggs (d./doz.)

3.03 3.04 2.78 2.65 2.73 2.96 2.83 2.86 2.86 2.78 2.74 2.69 3.02 2.77 2.82 2.83 2.80 3.19 2.97

3.97 4.11 4.57 4.16 4.47 4.62 3.62 4.81 4.26 4.76 4.16 4.31 5.61 5.61 6.45 6.31 5.46 6.35 6.24 6.65 6.84 7.39 7.25 7.09 8.13 6.94 6.70 6.40 4.91 4.81 5.51 5.41 6.19 4.71 4.52 4.16 4.31 4.37 3.58 4.62 5.20 5.01 4.52 4.96 4.91 5.61 5.31 5.01

0.70 0.47 0.53 0.46

2.96 3.04 3.10 2.96 2.84 2.47 2.45 2.96 2.96 2.45 3.09 2.32 2.26 2.96 2.96 2.96 3.05 2.96 2.07 2.48 2.63 3.06 2.44 1.48 2.96 2.96 2.96 2.96

0.51 0.46 0.52 0.49 0.42 0.52 0.40 0.46 0.45 0.54 0.54 0.52 0.52 0.67 0.52 0.52 0.52 0.62 0.52 0.57 0.52 0.52 0.52 0.52 0.52 0.52 0.52 0.52 0.57 0.53 0.52 0.52 0.52 0.57 0.53 0.57 0.60 0.57 0.52 0.60 0.54

80

GREGORY CLARK

Table A.2. (Continued ) Year

Hay (s/ton)

1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448

5.07 4.78 5.06 3.19 7.08 5.13 5.52 5.95 4.80 4.55 5.19 5.67 4.99 5.35 4.91 5.49 4.17 7.61 4.94 4.64 5.95 5.66 5.93 6.22 6.08 5.27 5.76 6.85 5.73 6.65 6.12 6.13 5.89 5.43 5.34 5.49 6.50 5.96 4.93 6.30 6.38 8.86 5.53 6.49 6.73 6.31

Cheese (d./lb.)

Butter (d./lb.)

0.44 0.48 0.53 0.42 0.52 0.44 0.49 0.45 0.49 0.47 0.47 0.47 0.55 0.52 0.53 0.53 0.43 0.49 0.48 0.48 0.40 0.49 0.45 0.41 0.59 0.51 0.48 0.45 0.58 0.55 0.54 0.44 0.58 0.50 0.60 0.60

0.97 1.15 1.46 0.77 0.97 0.87 0.93

0.61 0.60 0.51 0.45 0.52 0.52 0.50 0.44 0.47 0.40

Milk (d./gal.)

1.06

1.07 1.06

1.00 1.06 1.30 1.06 1.06

1.07 1.15 1.07

1.11 1.06 1.06

1.24

1.52

0.96 0.96

1.07

0.42

1.07 1.07

1.07

1.07 1.24

1.06 1.06

Beef (d./lb.)

Mutton (d./lb.)

Pork (d./lb.)

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

Eggs (d./doz.)

2.80 3.02 3.24 2.62

4.91 5.10 5.46 6.60 5.05 5.01 6.14 6.00 5.66 6.99 6.50 5.80 6.55 5.31 4.91 5.20 4.62 4.01 4.06 4.37 6.19 4.52 3.92 4.21 4.71 4.01 4.42 4.31 5.25 6.70 5.51 5.36 5.41 4.21 4.66 5.05 4.16 4.11 5.46 4.96 5.01 4.91 4.26 4.57 3.72 4.26 5.41 4.31

0.55 0.51 0.55 0.51 0.55 0.51 0.55 0.51 0.51 0.51 0.54 0.43 0.52 0.55 0.55 0.55 0.52 0.54 0.56 0.49 0.41 0.46 0.51 0.51 0.55 0.56 0.55 0.53 0.55 0.45 0.56 0.53 0.56 0.67 0.56 0.65 0.55 0.56 0.62 0.65 0.62 0.62 0.46 0.65 0.60 0.45 0.41 0.55

3.27 3.15 2.93 3.13 3.21 3.28 3.16 3.20 3.20 3.23 3.21 3.16 3.04 3.22 3.06 2.99 3.11 3.20 3.06 3.25 3.22 3.33 2.98 3.15 3.16 3.09 3.04 3.13 3.26 3.70 3.41 2.31 3.50 2.98 2.53 2.66 2.58 2.55 2.83 3.04 2.60 2.68 3.00

The Price History of English Agriculture, 1209–1914

81

Table A.2. (Continued ) Year

Hay (s/ton)

1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496

5.72 5.93 7.71 7.52 6.35 5.41 6.71 7.88 5.54 5.98 5.36 8.69 5.63 6.01 4.10 4.48 7.58 5.09 5.42 9.22 6.93 9.18 5.41 4.30 5.57 4.81 4.99 6.33 6.09 8.22 8.87 4.64 7.09 5.63 5.36 4.75 4.31 4.21 8.76 3.00 5.89 6.32 7.65 4.45 4.28 4.92 5.80 8.13

Cheese (d./lb.)

Butter (d./lb.)

Milk (d./gal.)

0.48 0.53 0.41 0.48 0.57 0.43 0.50

1.24

1.06

1.07

1.06

1.23 1.24 1.18 1.24

1.06

0.42 0.46 0.46 0.50 0.47 0.39 0.37 0.42 0.43

1.06

1.07 1.24 1.07 1.23 1.00 1.00

0.42 0.42 0.42 0.43 0.47 0.44 0.40 0.40 0.42 0.38 0.43 0.43 0.42 0.43 0.45 0.45 0.51 0.52 0.52 0.57 0.45 0.47 0.45 0.44 0.42 0.43 0.41

1.00

0.51

0.91

1.41 1.15 1.07 1.07 1.15

1.22 1.23 1.20 1.10 1.18 1.18 1.21

1.21 1.16 1.13 1.51

Beef (d./lb.)

Mutton (d./lb.)

Pork (d./lb.)

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

Eggs (d./doz.)

2.71 2.40 3.41 1.91 2.74 3.22 2.70 2.47 2.44 2.48 2.44 2.46 2.65 2.47 2.56 2.89 2.62 2.82 2.58 2.75 2.63 2.72 2.62 2.82 2.93 2.57 2.86 2.55 2.41 2.18 2.16

4.37 3.82 3.56 3.11 2.49 3.23 3.21 2.89 3.23 3.43 3.40 3.08 3.21 3.74 3.89 4.96 5.31 5.11 5.18 4.10 3.50 3.79 3.70 3.52 3.57 3.43 3.13 3.01 2.91 4.01 3.15 3.49 4.28 5.52 4.13 5.43 5.56 4.25 3.87 3.66 3.90 3.59 3.08 3.88 3.43 3.83 3.30 3.94

0.62 0.60 0.57 0.61 0.57 0.35 0.62 0.63 0.52

2.38 2.62 1.14 2.29 2.96 3.76

1.50 0.77

2.79 2.59 2.53 2.01 2.80 1.97 2.15

0.35 0.56 0.56 0.60 0.60 0.60 0.56 0.60 0.60 0.60 0.60 0.56 0.50 0.60 0.60 0.60 0.60 0.73 0.60 0.60 0.61 0.60 0.52

0.54

0.66 0.66 0.67 0.63 0.52

82

GREGORY CLARK

Table A.2. (Continued ) Year

Hay (s/ton)

1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544

7.08 7.24 9.13 5.60 5.30 5.51 5.85 6.90 7.21 5.31 5.72 5.30 6.43 7.51 6.27 5.92 6.78 8.26 6.28 6.44 8.59 7.57 8.29 5.71 5.06 5.48 6.21 6.49 6.67 6.09 7.34 5.14 6.13 7.06 7.05 6.11 7.06 6.53 7.89 8.12 5.31 6.01 6.26 8.94 8.48 8.33 7.62

Cheese (d./lb.)

0.42 0.38 0.42 0.43 0.54 0.38 0.45 0.41 0.45 0.45 0.43 0.44 0.44

Butter (d./lb.)

Milk (d./gal.)

1.24 1.24 1.43 1.18 1.55 1.19 1.14

1.55 1.30 0.77 0.77 0.77 1.28 0.77 0.77 0.77 0.83 1.22 0.83 0.77 1.04 0.77 0.77 0.77 1.06 0.77 0.83 0.77 0.77 1.35 1.12 1.19 1.28 1.12 0.96 1.07 0.61 1.10 1.34

1.08 1.07 1.03 1.01 1.22 1.35

0.71

1.35 1.35

0.47 0.47 1.25 0.56 0.82 0.78 0.53

1.35 1.35 1.03 1.03 1.07 0.95 1.31

0.57 0.62 0.88 0.53 0.51 0.60 0.48

0.83 0.92 0.50 0.50

1.29 1.03 1.08 1.08 1.10 1.11 1.00 1.06 0.95 0.94

1.03 1.61 1.17 0.89

1.00 1.69 1.45 1.15 1.06 0.96 1.55 1.07 1.25 1.44 1.15

Beef (d./lb.)

Mutton (d./lb.)

Pork (d./lb.)

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

1.97 2.75 2.83 2.57 2.05

3.93 3.19 3.78 2.99 3.44 3.57 3.27 3.63 3.67 3.12 3.04 3.41 3.79 4.01 4.09 3.86 4.09 4.01 4.47 5.06 4.83 5.29 4.24 4.72 4.39 3.86 4.58 3.52 3.60 4.47 4.72 4.13 4.55 3.82 3.27 4.36 4.91 5.48 5.10 4.55 5.63 4.39 4.05 4.20 7.13 5.82

2.54 2.59 2.62 2.62 2.51 2.25 2.36 1.97 1.96 1.97 1.97 2.15 1.97 2.68 2.34 2.34 2.34 2.03 2.30 2.33 2.30 2.30 2.58 2.30 2.30 2.30 2.30 2.00 2.87 2.56 3.07 2.60 1.96 2.37

1.96

5.63

Eggs (d./doz.)

0.60 0.80 0.73 0.69 0.58 0.66 0.70 0.95 0.87 0.98 0.73 0.61 0.52 0.63 0.82 0.77 0.87 0.62 0.62 0.62 0.62 0.62

0.88

2.48

1.44 1.93 0.58 0.58

The Price History of English Agriculture, 1209–1914

83

Table A.2. (Continued ) Year

Hay (s/ton)

1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592

7.92 7.55 7.29 7.30 11.86 12.26 14.71 11.55 14.15 11.65 20.77 19.03 12.05 10.40 14.96 16.66 20.26 18.44 20.06 15.98 17.78 21.89 23.91 20.19 15.38 17.98 15.13 17.26 17.11 23.04 19.55 19.16 20.64 17.79 21.63 16.35 20.34 21.87 27.38 15.93 14.05 22.69 15.75 14.25 32.17 24.95 20.64

Cheese (d./lb.)

1.18 1.10 1.77

Butter (d./lb.)

2.49 2.13 2.59

3.89

1.69 1.72 1.75

1.81 1.81 1.37 1.58 1.53 1.55 1.64 1.40 1.56 1.48 1.32 1.95 1.53 1.75 2.33 1.71 1.50 1.62 1.64 1.68 1.65 1.45 1.72 2.16 1.63 1.80 2.33 2.62

2.89 4.79 3.13 3.13 3.81 3.38 2.90 2.78 2.52 2.72 2.91 3.31 3.13 3.57 1.92 2.39 3.62 3.78 3.29 2.62 3.14 2.99 3.51 3.39 3.16 3.26 4.05 3.52 3.50 4.55 4.06

Milk (d./gal.) 1.15 1.72 1.34 1.71 1.91 2.05 2.14 2.49 2.13 2.30 3.34 2.68 2.97 3.00 3.25 3.07 3.09 3.61 3.52 3.83 3.21 3.08 3.25 3.12 2.32 2.09 2.30 2.40 2.77 3.00 2.32 2.42 2.46 2.68 2.53 3.05 3.23 3.15 2.74 2.37 2.74 2.82 2.60 2.42 2.74 2.57 2.30

Beef (d./lb.)

0.91 1.03 1.30

Mutton (d./lb.)

Pork (d./lb.)

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

2.93 1.96

5.29 6.75 4.72 4.51

3.14 2.75

1.29 1.25 1.21 1.29

4.12 3.46 4.01 5.52 5.13

6.30 10.54 10.04 7.21 4.64 6.22 5.99 9.84 10.20 4.69 7.08 5.75 8.26 5.63 6.45 5.99 8.22

1.71 1.25 1.21 1.23 1.22 1.45 1.38 1.41 1.43 1.46 1.43 1.40 1.34 1.41 1.31 1.38 1.41 1.72 1.78 1.83 1.79 1.78 2.04 2.12

4.51 4.30 4.21 4.38 4.43 4.68 4.68 5.04 5.02 4.86 4.39 4.94 4.16 5.41 4.65 4.98 6.02 5.16 5.20 5.74 6.41 5.97 6.10 6.84 6.38

10.23 8.66 8.73 7.10 8.77 7.38 8.22 8.48 8.84 9.42 9.53 9.84 10.50 9.64 8.77 9.93 8.77 7.91 8.03 6.72 7.02 8.14 9.38 11.66 12.08

4.35 1.14

7.37 4.15

Eggs (d./doz.)

1.75 1.60 1.46 1.69 1.75 2.91 2.33 2.04 2.91 2.48 2.91 2.65 2.45 3.60 2.36 2.48 2.91 2.61 3.31 2.51 2.18 2.70 2.61 2.33 3.78 2.52 3.52 2.25 2.65 2.19 2.68 1.94 2.11 1.88 1.92 2.36 2.41 2.23 3.40 2.37 2.30 2.38 2.50 3.11

84

GREGORY CLARK

Table A.2. (Continued ) Year

Hay (s/ton)

1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640

27.50 19.98 28.15 26.69 24.14 18.75 22.23 33.90 34.87 29.74 26.88 27.80 35.57 27.91 33.12 37.18 37.84 34.45 31.21 34.61 32.09 31.00 41.21 42.11 35.42 34.96 32.34 41.87 37.85 38.23 38.87 33.89 30.95 33.34 29.33 30.58 41.67 32.31 28.22 34.78 47.08 61.36 42.05 57.41 41.10 34.71 40.25

Cheese (d./lb.)

2.02 3.03 2.74 2.25 2.33 2.33 2.33 2.42 2.06 2.53 2.02 1.87 2.16 2.16 2.29 2.26 2.42 2.42 2.29 2.33 2.75 2.33 2.20 2.65 2.65 2.65 2.53 2.42 2.29 2.33 2.33 2.42 2.21 2.29 2.42 2.65 2.42 2.33 2.33 2.42 2.99 2.83 2.92 3.03 2.75

Butter (d./lb.)

Milk (d./gal.)

Beef (d./lb.)

3.55 3.53 3.62 3.65 4.50 3.94 4.11 3.90 4.31 4.54 4.08 4.13 4.18 4.54 4.47 4.72 4.79 4.32 4.63 4.87 4.80 4.83 4.81 4.83 4.71 4.61 4.50 4.89 4.60 4.68 4.72 4.73 4.27 4.30 4.31 4.59 4.56 4.97 4.59 4.78 4.82 4.93 4.94 5.47 5.77 5.82 5.64 4.76

2.54 2.67 2.75 2.99 2.90 3.11 2.99 3.00 2.92 2.87 3.07 2.95 2.68 2.97 3.21 3.29 3.06 3.54 3.83 3.71 3.96 3.24 3.59 3.42 3.42 3.42 3.42 3.42 3.43 3.42

1.92 1.80 1.91 1.82 1.99 2.15 2.07 2.20 2.01 1.95 1.78 1.82 2.07 1.91 2.00 2.14 2.20 2.06 2.10 2.11 2.19 2.36 2.51 2.54 2.54 2.44 2.52 2.46 2.34 2.18 2.15 2.35 2.53 2.48 2.50 2.46 2.46 2.39 2.38 2.43 2.42 2.61 2.78 2.61 2.66 2.78 2.75 2.68

3.96 3.42 4.60 3.42 3.42 3.42 3.42

3.42 3.42 3.62 5.11 3.49

Mutton (d./lb.)

2.05 2.28 2.28 2.28 2.28 2.28 2.28 2.28 2.28 2.28 2.28 2.39 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.63 2.73 2.73 2.73

Pork (d./lb.)

2.95

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

Eggs (d./doz.)

6.72 6.21 6.17 6.64 6.69 7.18 7.28 6.88 7.91 7.29 6.52 6.72 6.88 7.29 7.22 7.26 7.06 7.12 8.01 7.44 7.24 7.22 7.40 7.45 6.94 7.09 7.49 7.37 7.23 6.48 6.00 6.85 7.06 7.17 7.32 7.40 7.47 7.12 7.22 7.17 7.30 7.96 8.22 7.63 7.45 7.56 7.47 7.04

12.52 12.90 12.97 12.19 12.16 10.46 10.84 10.35 11.11 11.93 13.87 15.45 13.91 13.58 13.35 13.66 12.52 11.15 10.16 11.24 10.73 11.62 12.90 13.91 15.75 15.01 16.06 15.10 12.67 11.85 10.01 11.85 12.23 14.17 14.13 15.64 15.26

2.35 2.41 2.35 3.48 3.52 3.74 2.51 3.27 3.51 2.82 3.06 2.22 2.42 2.91 2.80 3.25 3.07 2.64 3.13 3.76 3.09 0.97 3.00 3.29 2.88 2.94 3.27 2.94 2.85 3.00 2.35 3.28 3.17

14.74 15.18 15.18 15.05 15.94 16.06 15.90 15.71 15.10 13.91

2.58 2.84 2.75 3.13 5.12 3.73 3.38 2.94 3.41 3.68 3.55 3.45 2.90 3.18

The Price History of English Agriculture, 1209–1914

85

Table A.2. (Continued ) Year

Hay (s/ton)

1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688

40.48 44.74 57.43 36.94 37.38 36.69 39.67 49.47 60.94 72.72 44.22 43.19 55.26 44.74 38.19 41.82 41.93 40.39 43.49 45.03 38.72 38.19 52.94 41.17 46.81 45.26 35.60 34.30 36.46 31.22 45.14 42.92 47.46 40.72 38.70 44.49 46.16 37.63 35.78 49.44 46.50 39.61 38.87 44.59 48.66 40.39 42.31 35.18

Cheese (d./lb.)

2.50 2.39 2.62 2.58 2.50 4.67 3.19 2.49 4.08 1.65 1.98 1.96 2.22 2.45 2.45 2.27 2.60 2.22 2.30 2.12 2.59 2.59 2.33 2.49 3.21 2.57 2.15 2.92 2.33 2.37 2.40 2.76 2.70 2.92 2.08 2.26 2.20 2.04 2.04 1.76 1.98 1.89 2.03 1.75 1.87 1.75

Butter (d./lb.)

Milk (d./gal.)

Beef (d./lb.)

Mutton (d./lb.)

4.99 4.88 5.06 5.01 5.25 5.43 5.91 6.58 7.70 7.52 6.09 5.78 5.35 5.42 4.81 5.66 5.91 6.30 6.70 5.97 5.83 5.67 5.61 5.84 6.19 6.09 6.09 6.30 6.00 6.07 5.88 5.80 5.85 6.06 6.02 5.69 5.75 5.81 5.87 5.75 5.75 5.84 5.70 5.22 5.86 5.26 5.30 5.68

4.21 3.75 3.42 4.89 4.46 4.89

2.68 2.61 2.60 2.61 2.59 2.64 2.93 3.10 3.43 3.31 2.83 2.80 2.84 2.56 2.42 2.43 2.67 2.74 3.03 3.04 2.91 2.93 2.93 2.98 3.09 2.84 2.77 2.72 2.65 2.76 2.72 2.70 2.82 2.76 2.91 2.75 2.78 2.81 2.86 2.73 2.88 2.92 2.70 2.92 2.97 2.76 2.76 2.83

2.73

5.59 3.26 6.13

6.09 6.09 6.10 5.92 6.48 5.29 6.09 6.11 5.36 5.11 5.72 5.92 6.41 6.74 7.00 6.55

6.13

9.19 9.19 9.19 6.55 6.55 7.33

Pork (d./lb.)

2.82 2.55

2.73 2.73 2.73 3.80 4.09 3.86 3.41 3.17 2.97 2.97 2.97 3.32 3.20 3.26 3.26 3.09 3.09 3.23 3.41 3.15 3.29 3.32 3.15 3.23 3.06 3.06 2.91 3.20 3.06 3.12 3.12 3.12 3.06 3.06 3.06 3.06 3.12 3.23 3.41 3.26 3.00 2.97

Bacon (d./lb.)

8.00 9.00

2.71 3.91 2.90 2.91 3.11

1.88 1.88 1.92 3.01 3.24 3.24 2.92 2.64 3.41 2.45 3.00 3.00

12.00 12.00 10.67 10.00 10.73 10.00 10.00 10.00

2.51 2.51 2.51 2.51 2.51 2.51

9.00 9.00 9.00 9.00 9.00 9.00

2.51 2.85 2.51 2.75 2.51 2.64 2.51 2.63 2.89 2.95 2.52 2.47

9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 10.00 10.00 8.50 8.50

Tallow (d./lb.)

Wool (d./lb.)

Eggs (d./doz.)

7.47 9.08 6.89 7.48 6.72 7.41 8.74 8.66 8.95 8.70 8.47 9.02 8.84 7.21 6.91 6.97 7.43 8.42 8.34 8.70 9.09 8.86 9.71 8.89 8.92 8.93 8.57 7.69 6.55 7.59 7.54 6.72 8.15 8.03 7.85 7.41 8.05 8.01 7.28 7.28 7.63 7.44 7.32 7.41 8.12 7.95 6.68 6.98

13.05 13.32

2.94 3.09 3.21 3.17 2.77 2.62

12.75

16.02 18.64 16.02 12.97 11.88 8.46

8.92 9.10 9.18 8.55 7.57 10.71 9.99 12.06 10.09 8.01

8.01 13.95 8.37 15.04 8.64 7.39 7.74 6.57 12.97 5.31 7.02 10.09 8.28

8.10

6.04

3.52 3.52 3.29

2.85 3.49 3.24 3.57 3.55 4.08 3.32 3.53 3.52 3.75 3.87 3.70 3.24 3.44 3.70 3.82 3.57 3.76 3.49 3.54 3.54 3.54 3.87 3.54 4.58 4.18 4.26 4.26 4.06

86

GREGORY CLARK

Table A.2. (Continued ) Year

Hay (s/ton)

Cheese (d./lb.)

Butter (d./lb.)

Milk (d./gal.)

Beef (d./lb.)

Mutton (d./lb.)

Pork (d./lb.)

1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 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

39.22 40.94 43.04 47.53 38.94 35.14 42.30 47.26 43.54 44.66 46.29 45.85 44.40 45.74 48.35 35.13 37.87 40.56 44.53 38.17 35.23 35.41 38.52 38.03 47.84 45.11 41.35 52.54 39.00 42.78 46.26 51.17 40.22 32.63 35.93 45.81 44.69 39.83 40.24 44.04 40.65 34.79 46.42 57.80 49.10 37.68 38.07 39.69

2.06 2.18 2.28 2.13 2.25 2.32 2.29 2.27 2.27 2.49 2.46 1.75 1.80 1.97 1.93 1.89 1.68 1.98 1.82 1.74 1.87 2.07 2.77 2.00 1.83 1.76 1.70 1.85 1.79 1.76 1.82 1.86 1.96 1.92 1.82 1.81 1.97 1.92 1.96 1.90 2.09 1.99 1.80 1.79 1.71 1.69 1.67 1.69

5.48 5.11 5.30 5.52 6.09 6.07 5.95 6.08 5.91 5.93 5.83 5.51 5.47 5.75 5.61 5.01 4.80 5.75 5.24 4.96 5.31 5.53 6.96 5.53 5.20 4.59 4.14 4.66 4.56 4.78 4.71 4.84 4.92 4.89 4.73 4.93 5.29 5.40 5.53 5.33 5.75 5.81 5.78 5.51 5.19 5.20 5.26 5.24

6.55

2.74 2.65 2.66 2.63 2.90 2.85 3.03 3.02 2.97 3.00 2.97 2.82 2.94 2.88 2.84 2.72 2.72 2.74 2.70 2.77 2.78 2.81 2.88 2.88 2.93 2.95 2.96 2.94 3.05 2.90 2.83 3.11 2.81 2.67 2.62 2.86 2.80 2.89 2.83 2.92 3.11 3.10 2.98 2.92 2.68 2.61 2.54 2.80

2.97 2.91 2.85 2.88 3.15 3.17 3.41 3.35 2.97 3.15 3.15 3.03 3.06 3.09 2.79 2.76 2.64 2.61 2.49 2.64 2.64 2.76 2.94 2.91 2.73 2.55 2.79 2.79 3.00 2.94 2.82 2.82 2.79 2.76 2.70 2.76 2.79 2.91 2.85 3.03 3.03 3.06 3.00 2.79 2.64 2.52 2.49 2.67

2.60 2.87 2.62 2.59 2.83 3.29 3.45 3.34 3.41 3.62 3.60 3.66 3.54 3.35 3.03 2.85 2.90 2.83 2.77 2.96 3.32 4.35 4.29 3.26 3.18 2.90 2.93 3.11 3.08 3.07 2.91 3.27 3.55 2.93 2.55 2.53 3.49 3.49 3.22 3.08 3.70 3.06 2.85 2.52 2.62 2.50 2.41 2.58

6.55 6.55 6.55 6.55 6.55 6.55 6.09 6.09 6.55 6.55 6.55 7.33 5.86 5.86 6.17 5.86 5.86 5.65 5.65 5.86 5.86 5.65 5.86 5.86 5.86 5.86 5.86 5.86 6.12 6.40 6.40 6.40 6.40 6.46 6.38 6.36 6.38 6.35 6.36 6.35 5.88 5.88 5.88 5.88 6.10 6.10

Bacon (d./lb.)

9.00 10.00 9.73 10.00 12.00 12.00 12.00

10.00 10.00 9.78 8.50 8.50 8.50 8.50 8.50 9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 8.00 8.00 8.50 8.50 8.50 8.50 8.50 8.75 8.75 8.75 8.75 9.00 9.00 9.00 9.23 8.27 9.41 9.16

Tallow (d./lb.) 7.47 7.41 6.99 7.37 7.61 8.30 8.56 8.24 6.95 7.91 7.75 8.18 8.16 8.51 7.67 7.28 7.48 7.50 7.30 6.83 7.42 7.85 7.82 7.29 7.92 7.28 7.35 7.75 8.28 8.15 7.99 7.70 7.88 8.18 7.47 7.69 6.97 7.34 7.64 7.70 7.64 8.25 8.32 8.05 7.79 6.24 7.04 7.05

Wool (d./lb.)

5.94 7.21 7.92 8.24 8.24 9.09 11.32 9.90 10.89 10.27 9.45 9.72 9.81 8.55 9.81 7.89 6.88 6.66 5.67 6.39 5.58 6.06 7.11 7.59 7.29 7.21 8.53 10.20 9.53 8.70 8.22 7.28 7.26 6.61 6.10 6.51 6.43 6.81 7.32 7.05 7.79 7.48 7.56 6.95 6.38 5.75

Eggs (d./doz.) 4.36 3.06 4.13 4.13 3.91 4.44 4.86 4.25 4.27 5.30 5.30 5.30 4.56 4.46 4.07 4.01 4.24 4.42 4.42 4.42 4.25 4.25 5.56 4.42 4.37 3.93 4.55 4.88 4.73 4.86 4.71 4.82 4.61 4.61 4.61 4.61 4.54 4.54 4.54 4.54 4.54 4.61 4.61 4.72 4.72 4.72 4.72 4.72

The Price History of English Agriculture, 1209–1914

87

Table A.2. (Continued ) Year

Hay (s/ton)

Cheese (d./lb.)

Butter (d./lb.)

Milk (d./gal.)

Beef (d./lb.)

Mutton (d./lb.)

Pork (d./lb.)

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

Eggs (d./doz.)

1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 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 1779 1780 1781 1782 1783 1784

45.74 49.03 42.32 65.84 51.29 65.84 47.03 44.18 41.15 39.19 39.30 35.27 40.11 48.00 40.76 40.76 50.07 60.43 47.71 39.19 47.83 54.23 48.56 51.46 47.23 48.06 51.29 51.07 52.97 55.26 43.57 42.90 44.86 45.59 49.25 65.40 72.24 57.15 57.30 73.87 72.15 58.61 52.61 59.10 63.04 62.59 70.08 77.91

1.81 1.77 1.87 1.88 2.20 2.26 1.84 1.73 1.62 1.82 1.90 1.94 1.98 1.91 1.81 1.82 1.89 1.87 1.92 1.94 1.96 1.94 2.03 1.93 1.90 1.90 1.92 1.88 2.03 2.00 2.06 1.98 2.03 2.10 2.14 2.41 2.41 2.36 2.32 2.36 2.38 2.45 2.43 2.32 2.16 2.11 2.25 2.27

5.54 5.49 5.37 5.66 6.70 6.75 6.05 5.48 5.52 5.99 6.50 6.49 6.46 6.25 5.82 6.11 6.22 6.77 6.75 6.65 6.63 6.93 6.92 6.61 6.82 6.88 6.56 6.38 6.74 6.54 6.57 6.51 6.53 6.65 6.80 6.98 7.13 6.98 6.51 6.78 7.15 7.70 7.70 7.17 6.76 6.80 6.92 7.17

6.10 6.10 6.10 6.10 6.10 6.02 5.88 5.76 5.70 5.79 6.32 6.05 5.76 5.88 5.88 5.88 5.88 5.88 5.88 5.88 5.88 5.88 5.88 5.89 6.11 5.88 5.88 5.88 5.62 6.00 6.57 6.57 6.57 6.57 6.39 7.46 7.46 7.79 7.79 7.79 8.10 8.39 8.51 8.62 8.62 8.62 8.43 8.23

2.74 2.79 2.81 3.03 3.42 3.28 3.06 2.84 2.73 2.80 2.96 3.06 2.89 2.85 2.93 2.93 2.91 3.13 3.06 3.06 3.15 3.17 3.15 3.00 3.03 2.99 2.95 3.09 3.03 3.33 3.69 3.68 3.59 3.52 3.65 3.84 3.78 3.72 3.69 3.70 3.73 3.68 3.77 3.78 3.73 3.73 3.84 3.93

2.64 2.67 2.64 2.79 2.97 2.88 2.85 2.73 2.70 2.73 2.67 2.76 2.73 2.52 2.62 2.62 2.59 2.96 2.70 2.73 2.90 3.24 3.18 2.93 2.90 2.76 2.84 3.07 3.18 3.36 3.69 3.47 3.36 3.47 3.64 3.81 3.87 3.92 3.85 3.69 3.69 3.87 3.87 3.64 3.64 3.29 3.70 3.75

2.66 2.96 2.88 3.07 3.54 3.51 2.98 2.68 2.63 2.74 2.83 2.82 2.92 2.78 2.97 3.04 2.91 3.15 3.39 3.30 3.46 3.71 3.74 3.09 3.33 3.58 3.58 3.58 3.93 4.02 3.95 4.31 3.87 4.09 4.35 4.59 4.82 4.16 4.50 4.41 4.44 4.56 4.06 4.07 3.97 3.96 4.10 4.31

9.23 9.00 8.43 8.63 9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 7.99 8.63 9.00 9.00 9.00 9.00 8.63 9.23 9.79 9.79 8.63 7.54 8.14 8.39 9.23 9.78 10.32 10.13 10.13 9.32 10.13 10.87 10.32 10.13 9.74 9.74 9.74 10.00 10.00 10.00 10.00 10.00 10.00 10.00

8.32 6.35 7.51 7.43 7.96 8.16 7.88 8.02 7.88 7.57 7.60 7.64 7.57 7.69 7.10 6.59 7.01 7.66 7.98 7.95 7.92 7.70 7.80 8.06 8.16 7.44 7.28 7.39 7.43 7.72 7.91 7.77 7.77 7.71 7.62 7.95 7.84 7.60 7.65 7.78 7.77 7.76 8.65 8.45 8.14 8.16 7.57 8.14

5.88 5.52 5.32 5.53 5.48 5.81 7.29 7.82 7.18 7.12 6.99 7.24 7.66 8.02 8.26 8.85 6.17 6.46 6.22 6.27 7.88 8.85 8.03 7.75 6.70 6.55 7.77 8.03 7.85 8.35 7.69 6.53 6.32 6.27 6.95 6.53 6.83 7.40 7.71 7.58 7.19 6.35 5.76 6.29 5.80 5.21 6.06 6.97

4.72 4.54 4.32 4.32 4.46 4.46 4.27 4.41 4.41 4.41 4.37 4.82 4.82 4.82 4.82 4.82 4.82 4.82 4.05 4.82 4.82 4.82 4.82 4.82 4.61 4.82 4.38 4.61 4.72 4.72 5.11 5.11 5.05 4.80 5.11 5.25 5.34 5.66 5.36 6.56 5.75 5.56 6.48 6.48 5.94 6.48 5.70 6.01

88

GREGORY CLARK

Table A.2. (Continued ) Year

Hay (s/ton)

Cheese (d./lb.)

Butter (d./lb.)

Milk (d./gal.)

Beef (d./lb.)

Mutton (d./lb.)

Pork (d./lb.)

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

Eggs (d./doz.)

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 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832

83.85 79.13 60.00 64.50 72.69 62.03 56.42 70.93 78.80 79.86 82.73 98.69 83.30 60.07 78.00 104.78 103.65 89.98 102.60 96.02 86.04 81.81 90.02 103.10 102.97 121.85 152.78 115.14 95.06 97.84 99.94 93.46 88.08 111.70 121.09 81.27 66.18 73.78 92.19 117.97 105.54 99.94 114.27 88.35 71.06 70.69 73.38 87.48

2.22 2.26 2.28 2.33 2.26 2.27 2.43 2.40 2.56 2.49 2.47 2.74 3.02 3.17 3.10 3.42 3.98 4.11 3.94 4.10 4.15 4.02 4.08 4.22 4.24 4.40 4.47 4.67 4.28 4.64 4.45 4.20 3.62 4.04 4.54 4.22 3.83 3.41 3.35 3.64 4.52 4.20 4.02 4.29 3.87 3.51 3.83 4.05

6.89 7.19 7.09 6.83 6.53 7.17 7.19 7.23 7.62 7.92 8.46 8.56 9.94 10.20 10.80 12.05 11.56 10.58 10.98 11.22 11.55 11.34 12.18 13.11 13.21 13.47 15.10 15.19 14.47 14.95 15.37 11.39 10.85 13.71 12.58 11.11 10.71 9.55 9.45 10.55 11.63 11.59 11.01 10.14 9.72 9.73 10.73 9.83

8.59 9.10 9.10 9.10 9.10 9.10 9.10 9.10 9.10 9.10 9.10 9.96 12.62 12.62 12.62 12.62 12.62 12.62 12.62 13.45 14.45 15.22 15.97 16.54 16.81 17.97 19.07 19.34 20.27 19.34 20.04 19.38 20.03 20.94 20.94 20.94 20.94 20.26 19.54 19.54 19.54 19.54 19.54 19.54 19.12 19.12 18.69 18.11

4.20 4.13 4.36 4.40 4.43 4.36 4.45 4.66 4.61 4.58 4.94 5.68 6.48 6.31 6.18 7.05 7.91 8.36 7.88 7.90 7.92 7.97 7.51 7.58 7.89 8.45 8.70 8.88 9.28 9.56 8.49 7.18 6.41 5.86 7.68 7.96 6.97 5.51 5.23 6.02 6.73 7.15 7.03 6.77 6.52 5.62 6.00 5.39

4.09 3.98 4.15 4.04 4.04 4.04 4.04 4.21 4.32 4.27 4.38 4.67 5.67 5.69 5.56 6.32 7.20 7.35 7.16 6.83 6.65 6.42 6.77 7.05 6.94 7.28 7.50 7.85 8.42 8.87 7.43 6.26 6.14 6.22 7.51 7.47 6.60 4.55 4.66 5.35 6.03 6.60 6.37 6.14 5.63 5.18 5.57 5.68

4.52 4.31 4.36 4.90 4.56 5.05 5.61 5.24 5.47 5.58 5.64 7.23 7.45 6.32 5.65 6.93 9.95 8.60 7.45 6.78 6.78 7.24 7.08 7.52 8.49 8.88 8.72 8.75 10.75 10.31 8.14 6.20 7.03 7.57 8.67 7.30 5.88 5.24 4.87 6.59 7.69 6.73 6.38 7.12 6.54 5.56 5.40 6.30

10.00 10.00 10.00 10.00 9.50 9.00 9.00 9.00 9.92 10.00 10.00 10.83 11.00 10.33 9.83 11.58 15.21 15.67 13.17 11.92 12.00 12.08 12.00 12.04 13.00 13.00 13.00 13.00 13.44 17.08 15.73 11.85 12.00 14.27 15.67 14.75 13.00 10.80 10.36 11.80 13.25 14.00 14.00 12.52 13.64 10.54 11.18

7.61 6.40 8.12 8.37 7.50 7.53 7.37 7.10 7.14 7.16 8.05 9.37 9.47 8.57 8.70 10.22 10.94 11.03 11.25 11.38 10.63 10.78 9.81 12.09 14.45 11.31 10.78 12.09 15.07 15.93 12.83 9.58 9.36 13.19 11.88 11.70 9.80 8.14 8.32 7.89 8.65 8.38 8.74 8.61 8.62 7.92 8.82 8.39

6.46 6.79 8.66 8.68 8.68 9.13 8.95 11.38 8.88 9.16 10.04 10.44 10.13 9.70 12.44 12.28 13.64 14.16 14.07 15.55 17.27 16.05 14.31 13.18 17.52 16.85 12.53 14.23 16.10 20.13 20.80 13.91 16.33 21.72 15.68 15.16 12.77 11.80 12.28 13.13 15.39 9.37 9.18 10.01 8.00 9.64 11.58 12.28

6.44 5.89 5.89 5.89 5.89 5.89 5.89 5.89 5.89 5.89 6.43 6.19 6.68 6.19 7.17 7.56 8.69 8.79 8.55 7.66 8.60 7.86 8.10 8.87 9.09 8.74 8.79 8.99 13.95 14.24 12.03 10.29 9.58 9.87 11.74 9.87 9.33 7.91 8.25 7.78 8.35 8.99 8.28 7.64 7.96 8.08

The Price History of English Agriculture, 1209–1914

89

Table A.2. (Continued ) Year

Hay (s/ton)

Cheese (d./lb.)

Butter (d./lb.)

Milk (d./gal.)

Beef (d./lb.)

Mutton (d./lb.)

Pork (d./lb.)

1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 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

69.80 81.41 82.87 80.74 83.73 95.79 87.26 76.17 87.51 75.56 74.91 72.39 90.67 72.64 65.64 64.75 61.33 57.32 69.36 61.19 79.89 84.32 86.99 83.88 64.90 64.21 72.51 102.40 96.97 71.51 63.19 65.92 89.12 78.40 76.04 78.54 101.26 80.33 111.41 66.36 61.84 42.70

4.03 3.63 3.50 4.02 4.02 4.04 4.01 4.07 4.07 3.87 3.56 3.27 3.35 4.06 4.11 4.05 3.91 3.77 3.31 3.22 3.50 4.09 3.91 4.30 4.10 4.03 4.09 4.56 3.94 3.82 3.95 3.88 4.18 4.08 3.66 4.03 4.06 4.10 4.04 3.94 4.04 3.63 3.76 3.60 4.09 3.54 3.08 3.87

9.38 9.01 9.54 10.91 11.59 10.98 11.41 11.72 11.57 10.75 9.82 9.88 10.54 10.92 11.13 10.21 9.72 8.96 9.16 9.13 9.84 11.72 11.40 12.27 12.39 11.63 11.60 12.25 11.80 11.50 11.43 11.76 12.47 13.00 12.29 11.79 13.87 13.25 14.05 13.55 13.40 14.20 14.03 14.53 14.57 13.85 12.20 12.34

18.11 16.71 16.71 16.71 16.71 16.71 16.71 16.71 16.71 16.71 16.71 16.71 16.71 15.32 15.32 13.93 13.93 13.93 11.14 9.75 11.14 10.45 13.57 14.02 13.97 13.26 12.91 12.74 13.97 14.15 13.43 13.79 16.92 17.68 16.26 15.50 15.11 11.99 12.17 12.85 13.73 15.35 15.35 15.35 15.71 15.71 15.28 13.40

5.12 5.12 4.72 5.39 5.39 5.65 5.39 5.66 6.20 5.66 4.58 4.58 4.31 5.44 5.90 6.00 5.02 4.77 4.59 4.71 5.77 6.39 6.62 6.31 6.40 6.16 6.25 6.49 6.37 6.04 6.46 6.82 7.00 7.18 6.80 6.53 7.07 7.22 7.89 7.81 8.73 8.13 8.22 8.16 7.40 7.44 7.00 7.35

5.22 5.34 4.43 4.77 5.22 5.22 5.22 5.11 5.68 5.22 4.31 4.31 4.54 5.98 6.26 5.84 4.74 4.61 4.74 4.74 5.84 5.84 5.78 5.78 5.78 5.50 5.84 6.26 6.26 5.98 5.98 6.39 7.22 6.88 6.26 5.84 6.39 6.53 7.36 7.63 7.84 6.60 7.36 7.22 6.74 6.88 6.53 6.74

6.01 5.03 4.82 5.99 6.44 6.97 6.56 6.22 6.59 5.50 4.46 5.15 5.47 5.62 6.75 6.44 6.33 5.57 5.04 5.04 5.58 6.14 6.12 6.50 6.78 6.08 6.07 6.79 6.80 6.62 6.34 6.46 7.00 6.74 5.64 5.71 7.11 7.89 6.39 6.62 6.76 6.72 7.34 7.57 6.64 6.54 6.30 6.96

87.78 83.03 73.47 65.57 67.27

Bacon (d./lb.)

12.51 12.18 13.17 11.85

11.85

15.80 14.48 13.60 13.83 13.17 13.17 15.80 15.80 15.80 15.80 15.80 15.80 15.80 15.80 15.80 15.80 17.12 17.12 17.12 16.46 15.80 15.80

Tallow (d./lb.)

Wool (d./lb.)

Eggs (d./doz.)

8.67 8.53 8.02 8.12 7.94 8.89 8.60 9.00 8.53 8.42 7.64 7.32 7.09 7.21 8.70 8.54 7.36 6.83

13.33 17.60 16.83 18.39 14.23 15.34 16.37 14.24 11.61 10.63 10.55 12.69 14.36 12.79 11.97 10.98 10.15 11.06 12.50 13.62 16.00 15.50 13.00 16.00 20.50 15.62 18.62 20.12 19.50 20.50 22.62 27.38 25.75 23.50 18.88 17.50 18.12 16.75 21.38 25.62 24.50 20.75 19.75 17.75 16.25 15.00 12.50 15.12

7.89 7.89 8.30 7.47 8.24 8.43 8.44 8.04 8.01 8.21 8.07 7.76 8.26 9.45 8.85 9.37 10.82 11.79 11.86 11.87 11.47 11.55 10.93 9.37 8.34 9.93

7.53 7.95 7.85 8.95 9.25 9.57 8.65 9.36 9.51 8.71 7.80 6.91 7.69 9.18 9.82 8.69 9.69 9.90 9.47 8.78 7.53 7.32 7.47 8.65 8.39 7.82 6.74 5.95 6.41

90

GREGORY CLARK

Table A.2. (Continued ) Year

Hay (s/ton)

Cheese (d./lb.)

Butter (d./lb.)

Milk (d./gal.)

Beef (d./lb.)

Mutton (d./lb.)

Pork (d./lb.)

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 1914

71.12 87.26 62.40 65.81 59.96 66.79 69.47 85.31 75.56 57.52 56.55 76.71 110.69 101.56 66.99 63.43 67.29 65.34 62.74 69.51 78.82 84.37 73.12 68.25 66.79 63.72 65.66 60.25 61.79 61.79 77.24 88.83 83.04 59.86

3.54 3.59 3.70 3.59 3.87 3.16 3.54 3.32 3.21 3.10 3.05 2.99 2.89 2.89 2.56 2.18 2.77 2.34 2.89 2.89 2.77 2.94

14.27 13.87 13.67 13.51 12.35 12.27 12.66 12.84 12.84 12.10 12.70 12.83 11.95 11.12 10.61 10.71 10.61 10.32 10.90 11.30 11.30 11.23 12.06 12.85 12.32

13.40 13.40 13.84 12.67 12.44 12.00 11.88 11.85 11.28 11.21 11.14 12.17 12.35 12.06 10.89 9.79 9.91 9.74 9.94 10.26 10.92 11.19

7.16 7.60 7.94 7.55 6.89 6.39 6.09 6.11 6.20 6.03 6.67 6.28 6.44 6.08 5.89 5.29 5.51 5.31 5.79 6.22 5.91 6.37 6.04 5.91 5.97 5.91 5.97 5.97 6.24 6.44 6.04 6.98 6.85 7.25

7.15 7.49 7.63 6.60 5.84 6.25 5.25 5.88 6.25 5.63 5.25 5.25 5.25 5.25 5.50 4.88 5.13 4.63 5.13 5.63 5.50 5.50 5.88 6.25 6.38 6.63 6.75 6.50 5.75 6.50 6.13

7.38 7.03 6.72 6.18 5.75 5.31 5.03 4.86 5.25 4.84 4.71 5.59 6.07 5.58 4.50 4.16 5.21 5.46

13.37 13.11 13.11 12.58 13.90 14.16

8.51 8.84 8.98 9.02 8.88 9.24 9.71 9.70 9.31

5.25 5.47 5.38

Bacon (d./lb.)

Tallow (d./lb.)

Wool (d./lb.)

Eggs (d./doz.)

7.29 8.12 7.90 6.97 5.82 5.09 4.98 5.48 5.32 4.90 4.98 5.16 6.38 5.42 4.84 4.03 3.58 4.05 4.35 5.08 5.61 6.67 5.47 4.91 5.00 5.56 6.42 5.81 5.88 6.54 6.40 6.44 6.53 6.33

12.38 11.25 10.00 10.00 9.88 10.00 10.50 10.38 11.00 11.00 9.75 8.75 10.25 10.12 12.00 11.50 9.62 8.75 8.25 7.88 6.88 6.25 7.25 10.12 12.38 13.38 12.25 8.50 9.00 9.88 10.00 10.50 12.38 12.62

8.60 9.23 10.65 9.74 9.08 8.99 9.07 8.66 8.83 8.72 9.67 9.20 9.58 9.46 8.94 9.07 8.63 9.08 8.82 9.59 9.68 9.26 9.60 10.02 10.34 11.35 11.28 11.02 11.36 11.23 11.66 12.00 12.21 13.10

The Price History of English Agriculture, 1209–1914

91

Table A.3. Firewood, Timber and Miscellaneous Product Prices. Year 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252

Firewood (s./c.)

Timber (s/ft3 )

Cider (d./gallon)

Honey (d./gallon)

0.31

0.40

0.35 1.00 0.49 0.40

0.36

0.43 0.42

0.45 0.51 0.51 0.32

0.38

92

GREGORY CLARK

Table A.3. (Continued ) Year 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296

Firewood (s./c.)

Timber (s/ft3 )

Cider (d./gallon)

Honey (d./gallon)

0.30 0.39 0.30 6.06 4.69

0.31 0.38

0.37 0.39 0.40 0.38 0.34 0.31 7.25 7.06

7.12 7.68 6.04 7.27 10.40 8.07 8.23 5.17 5.24 7.08 7.24 7.06 7.30 7.49 7.29 8.41 8.05 8.88 7.75 7.78

0.72 0.65 0.57 0.50 0.55 0.55 0.61 0.71 0.61 0.61 0.58 0.67 0.61 0.62 0.55 0.58 0.66 0.69 0.55 0.57 0.54 0.45

2.00

1.25 2.00

1.08 1.11 1.22 1.06 0.84 1.13 1.05 0.95

1.82 1.65

The Price History of English Agriculture, 1209–1914

93

Table A.3. (Continued ) Year

Firewood (s./c.)

1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340

6.70 8.17 7.10 7.62 9.24 8.53 8.07 6.37 9.31 9.60 9.49 9.78 12.30 9.57 10.60 11.04 8.11 9.91 8.47 8.87 10.99 11.08 10.31 6.54 9.63 11.87 11.02 14.47 12.42 10.66 10.37 10.61 11.47 13.69 11.31 11.74 10.60 10.00 11.60

Timber (s/ft3 )

Cider (d./gallon)

Honey (d./gallon)

0.61 0.62

1.04

0.71 0.86 0.80 0.65 0.65 0.60 0.70 0.72 0.70 0.72 0.79

1.20

1.56 1.25 1.27 1.00 1.39

0.71 0.72 0.71 0.83 0.94 0.58 0.71 0.75 0.78 0.84 0.80 0.78

0.72 0.90 1.36 0.61 0.42 0.56 0.89 0.58 0.57 0.33 0.58 0.51

1.28 2.32 1.30 1.30 1.15 1.11 1.04 1.11 1.28 1.16 1.11 1.32 1.11 1.37

0.93 2.00 1.56

94

GREGORY CLARK

Table A.3. (Continued ) Year

Firewood (s./c.)

1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384

11.02 11.55 11.27 11.02 9.39 12.54 11.02 25.76 7.64 12.73 10.71 7.13 12.77 14.79 11.31 12.36 10.49 11.74 12.97 11.95 13.07 11.68 11.42 12.26 11.20 11.64 13.22 12.20 11.14 13.01 12.71 10.59 10.67 12.66 13.61 12.69 13.59 12.94 10.04 10.97 10.61 12.26 12.67 13.09

Timber (s/ft3 )

Cider (d./gallon)

Honey (d./gallon)

0.46 0.62 0.42 0.49 0.48 0.56 0.80 0.67 0.63 0.56 0.87 0.59 0.35 0.73 0.61 0.64 0.55 0.52 0.77

1.11 1.66 2.04

1.86 1.49 2.23 2.66

1.05 0.53 0.64 0.87

1.15 0.80 0.59 0.66

1.58

2.69 2.00 3.72

0.77 2.69 0.64 2.37 0.74 0.78 0.50 1.86 0.54

The Price History of English Agriculture, 1209–1914

95

Table A.3. (Continued ) Year

Firewood (s./c.)

1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428

14.38 11.57 10.42 11.76 13.34 12.36 12.80 11.47 9.64 10.71 9.61 9.44 9.63 9.59 9.33 8.51 9.59 9.33 9.46 9.73 9.94 9.67 9.64 8.75 9.34 9.34 8.48 9.01 8.77 8.79 9.01 9.65 8.54 8.10 8.83 9.22 8.63 9.26 8.88 9.03 9.02 9.75 9.70 9.66

Timber (s/ft3 )

Cider (d./gallon)

Honey (d./gallon)

0.56 0.59

1.77

0.58

1.86 2.30

0.61

1.67

0.61

2.69 2.53 2.53 2.75 2.31 2.53 2.53 2.04 2.33

0.73 0.73

0.56 0.56 0.44

1.11 0.54 0.44 0.67

0.56

2.47 2.03 2.17 2.44 2.46 2.27 2.43 1.94 2.04 2.05 2.67 2.22 2.08 1.58 1.70 1.66

0.56 2.66 1.86 1.78 2.16

96

GREGORY CLARK

Table A.3. (Continued ) Year

Firewood (s./c.)

1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472

9.83 8.59 8.53 8.98 8.26 8.13 8.07 8.23 8.36 9.83 9.69 9.11 7.97 8.29 8.70 8.54 8.06 8.37 8.98 6.75 9.66 6.40 8.24 7.89 8.77 8.58 8.25 9.44 7.00 7.67 7.50 8.36 9.44 8.99

7.67 8.22 8.16 7.92 7.65 8.49 7.66 8.12

Timber (s/ft3 )

Cider (d./gallon)

Honey (d./gallon) 2.00 1.77 1.81 2.24 2.03 2.23 1.43

0.89 0.37 0.84

0.74

2.22

0.67

1.81

0.52

1.79

0.44

0.53

0.33 0.84

1.79 1.89

2.39 2.20 2.31 2.32

0.74 1.00 0.74

2.05 1.96 2.14 1.82 1.64 2.02 2.38 1.85 1.50 2.15 2.66 2.03 1.76 2.32 2.01 2.10 3.00 1.90 1.90 2.36 1.11 2.14 2.19 1.99 1.99 2.66 2.41

The Price History of English Agriculture, 1209–1914

97

Table A.3. (Continued ) Year

Firewood (s./c.)

1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516

7.85 7.95 8.48 7.74 8.04 7.14 6.24 6.48 6.62 6.28 7.47 6.43 3.93 4.19 4.25 7.00 6.78 5.73 5.43 7.55 6.00 5.49 8.35 7.06 6.34 7.90 7.92 7.82 7.48 8.08 7.45 7.43 7.27 9.71 6.91 6.95 7.14 7.19 8.20 7.87 7.58 7.77 7.38

Timber (s/ft3 )

Cider (d./gallon)

Honey (d./gallon)

1.79

0.40

3.53 2.56 2.60 2.37 2.30 2.33 2.49 2.44 2.33 2.72

0.48

0.89 0.67 1.47 0.56 1.79 1.79 2.25

1.63 1.79 1.68 2.14 1.79

1.79 1.65 1.14 1.37 1.79 1.08 1.41 1.24 1.25 1.47 1.65 1.79 1.58

2.57 2.58 2.83 2.33 2.18 2.41 2.49 2.64 2.54 2.31 2.36 2.34 2.36 1.50 2.66 2.36 2.33

2.46

3.33 2.62 2.73

98

GREGORY CLARK

Table A.3. (Continued ) Year

Firewood (s./c.)

Timber (s/ft3 )

1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560

7.73 7.90 7.84 8.03 7.89 8.02 8.12 7.80 7.96 7.88 8.75 7.65 7.90 7.65 8.83 7.45 7.76 7.51 7.65 7.66 7.57 7.85 8.03 7.81 8.90 8.06 9.45 8.52 8.73 8.41 6.03 8.97 9.96 11.39 9.30 10.49 11.63 12.64 12.84 11.49 10.49 11.04 14.08 12.34

1.69 1.40 1.47 0.67 1.66 0.75

1.71 1.45 0.95 1.52 0.83

2.62 2.73

0.84 1.34 1.89 2.64 2.36 2.25 2.10 1.98 1.09 2.25 3.60 3.67 3.29

5.16 2.75 1.22

Cider (d./gallon)

Honey (d./gallon)

3.33 3.76 3.28 1.79 1.86 3.01 3.23 2.48 1.97

2.62 3.47 4.21 3.19 3.13 3.39 2.87 2.60

The Price History of English Agriculture, 1209–1914

99

Table A.3. (Continued ) Year

Firewood (s./c.)

Timber (s/ft3 )

1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604

17.30 9.76 11.80 11.87 10.29 13.75 11.84 11.90 16.09 16.94 13.67 14.05 12.90 11.85 12.30 11.81 14.43 14.45 12.81 13.96 14.83 15.21 13.29 14.15 15.45 16.54 15.20 15.68 15.04 15.99 15.49 15.71 15.51 18.47 15.62 18.66 18.21 18.58 18.96 18.62 17.82 20.05 19.89 19.72

3.12 3.50 3.38 3.53 3.08 3.00 3.33 2.75 3.29 3.28 3.50 3.53 2.99 2.87 2.91 2.55 4.26 2.58 3.21 2.72 2.78 3.15 2.27 2.89 3.61 2.96 3.26 2.85 1.98 3.40 3.01 3.31 3.06 3.23 3.23 3.31 3.35 2.82 3.09 2.87 4.45 4.04 4.20

Cider (d./gallon)

Honey (d./gallon)

100

GREGORY CLARK

Table A.3. (Continued ) Year

Firewood (s./c.)

Timber (s/ft3 )

1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648

21.78 20.97 23.89 23.65 24.21 24.16 24.33 25.17 20.67 23.11 26.59 22.70 22.50 22.79 22.75 23.53 23.90 24.58 25.29 21.34 24.39 25.11 25.07 28.88 29.19 26.01 27.09 24.09 26.09 27.18 25.65 27.01 26.30 29.95 29.43 31.02 32.33 29.58 29.24 34.67 28.89 28.77 35.90 34.59

4.29 4.73 5.35 6.02 5.94 6.69 6.01 5.76 6.92 5.99 5.82 7.05 6.07 6.89 7.31 6.71 4.91 5.81 6.18 6.19 5.96 6.86 6.64 7.75 7.94 3.11 8.00 7.65 5.57 7.45 9.44 7.79 7.96 8.50 8.23 9.03

8.98 8.46 6.85 6.66

Cider (d./gallon)

Honey (d./gallon)

The Price History of English Agriculture, 1209–1914

101

Table A.3. (Continued ) Year

Firewood (s./c.)

Timber (s/ft3 )

1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692

31.91 20.30 28.48 28.41 27.96 26.97 27.16 31.45 27.12 25.26 29.80 27.00 32.09 32.79 30.73 28.47 34.39 33.05 33.43 35.11 29.16 33.06 34.58 34.69 33.71 36.56 34.19 32.66 35.50 26.63 32.37 34.68 35.57 31.39 33.93 34.18 32.91 35.80 36.11 37.14 33.79 34.00 33.03 33.63

8.23 5.28 6.82 6.11 9.39 7.88 9.55 13.65 9.50 10.38 8.92 10.95 11.67 10.68 13.65 10.61 10.72 9.90 11.45 8.60 4.22 4.66 12.27 13.60 11.36 9.89 10.17 9.95 10.24 9.06 9.88 11.67 13.65 12.97 10.17 9.53 8.42 7.14 10.24 11.40

Cider (d./gallon)

Honey (d./gallon)

102

GREGORY CLARK

Table A.3. (Continued ) Year

Firewood (s./c.)

Timber (s/ft3 )

1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 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

33.51 33.76 33.61 35.93 35.21 37.94 34.25 34.00 32.13 33.06 36.19 33.41 34.14 29.58 30.37 32.05 32.20 35.09 33.75 35.07 35.15 34.02 35.72 32.36 31.65 30.75 32.05 31.47 29.37 31.59 31.43 30.88 31.59 33.66 33.44 32.91 33.77 33.43 33.04 32.66 32.06 33.83 29.05 32.91

10.80 10.24 10.24 10.24 10.24 10.78 10.24 16.64 10.24 6.93 8.96 7.89 9.50 9.50 10.32 9.50 8.91 12.07 9.29 10.45 10.14 7.98 9.00 8.64 9.12 8.85 7.99 9.69 8.68 8.72 8.57 10.24 7.37 9.23 8.69 8.67 6.44 8.41 8.29 7.78 7.78

Cider (d./gallon)

Honey (d./gallon)

The Price History of English Agriculture, 1209–1914

103

Table A.3. (Continued ) Year

Firewood (s./c.)

Timber (s/ft3 )

1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 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 1779 1780

31.06 31.06 31.06 32.15 32.63 32.63 31.16 29.78 33.22 34.08 29.17 29.53 30.46 31.62 28.20 34.43 32.15 37.58 29.53 26.95 32.71 30.46 29.22 31.65 27.14 26.87 29.38 29.13 37.34 28.49 26.87 32.11 29.38 30.76 29.72 32.17 36.25 36.51 37.48 37.77 37.77 38.50 38.13 29.85

7.78 8.26 7.41 8.05 7.41 7.31 7.83 7.31 7.83 7.31 8.02 8.76 7.76 8.55 8.27 9.53 9.77 8.41 9.28 7.90 8.88 9.19 8.97 8.74 8.19 9.42 9.14 9.38 9.32 10.17 10.18 9.06 9.76 10.25 9.67 10.25 10.88 9.10 10.24 10.24 9.69 10.24 8.53 8.53

Cider (d./gallon)

Honey (d./gallon)

104

GREGORY CLARK

Table A.3. (Continued ) Year

Firewood (s./c.)

Timber (s/ft3 )

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 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824

29.85 28.37 29.85 28.38 28.67 30.35 28.67 31.04 28.67 29.45 29.45 27.17 26.84 27.17 27.17 27.17 30.00 30.59 33.17 31.03 31.03 30.59 33.40 34.41 34.81 35.96 36.75 36.64 36.64 37.12 37.17 42.12 44.59 44.59 44.77 42.12 41.13 40.79 39.56 40.79 38.47 37.51 38.09 40.23

9.08 12.02 9.87 11.03 9.95 8.75 8.65 9.23 9.16 9.72 10.72 10.10 11.32 11.96 12.98 13.32 13.95 17.27 18.42 17.68 17.80 16.51 20.14 20.22 21.09 22.35 24.38 30.87 30.65 35.91 28.73 26.46 27.67 28.87 24.76 19.63 18.82 23.63 18.62 17.01 18.17 18.52 21.15 17.48

Cider (d./gallon)

Honey (d./gallon)

The Price History of English Agriculture, 1209–1914

105

Table A.3. (Continued ) Year

Firewood (s./c.)

Timber (s/ft3 )

1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 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

40.23 40.23 39.61 40.23 40.23 40.23 40.23

16.76 17.80 14.30 14.82 15.11 13.73 13.53 13.83 13.54 16.63 15.75 15.28 17.30 17.11 18.62 17.43 15.47 14.13 14.37 14.91 13.32 10.91 8.75 7.13 7.82 7.17 7.01 7.59 9.59 11.53 10.34 10.78 9.15 8.50 9.02 9.67 9.94 9.94 9.41 12.59 10.71 7.98 7.58 7.98 8.50

45.80 40.48 38.95 40.48 38.95 40.48 40.48 40.48 40.48 40.48 40.48 42.07 42.07 40.48 40.48 40.48 40.48 40.48

40.48 40.48 40.48 40.48 40.48

Cider (d./gallon)

Honey (d./gallon)

106

GREGORY CLARK

Table A.4. Price Indices (1860–1869 = 100). Year

Arable

Meat

Dairy

Wool

1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252

4.8

6.7

5.9

13.2

7.9

5.7

5.8 4.7

5.9

6.1 6.5

12.8 10.6

7.6 7.1

6.4 5.5

4.3

6.3

12.8

7.8

5.3

5.0 6.9 8.2 8.0 6.1 7.9

5.5

13.7

7.5

5.8 6.3 6.1 5.9

12.5 12.3 13.4 12.3

7.4 7.6 7.8 7.4

5.8 7.2 8.0 8.0 6.7 7.8

6.2 7.7 6.5 6.0

15.0 14.4 13.0 12.8

8.4 9.2 9.2 7.6

4.5 5.7 9.1 8.6 8.3

7.8

6.1 7.1

19.2 19.9

9.4 10.4

8.0 7.7

4.6

7.4 7.7

18.5 18.3

7.4 10.3

6.7 6.5 8.2

3.3 4.6 8.8 8.2 8.7

7.2 6.5 5.7 6.0 7.2

7.8 8.8

Pasture

Wood Index

Farm Index

4.4 5.8 9.9 10.0 5.6

6.3 6.9 7.1 15.6

6.9 7.5 9.1 6.9 6.3

19.2 20.4 22.0 17.6 15.7

10.1 9.0 10.0 9.0 12.9

5.9 6.8 10.0 9.7 7.5

6.7 6.1

5.7

7.6

17.4

8.2

7.9 6.8

The Price History of English Agriculture, 1209–1914

107

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296

9.2 5.6 5.2

6.9 6.0 6.7

6.4 6.8 7.2

11.2 12.1

6.7 8.1 13.4 5.1 4.5 3.7 3.7

7.2 7.2

17.4 14.8 15.7 21.8 17.4 18.1 21.8

8.7 7.9 8.7 11.2 8.8 9.6 13.7 6.1 7.0 7.8 7.1 15.6 11.4 9.9 14.0 10.7 12.5 11.9 13.1 13.6 12.1 14.6 13.7 16.6 13.4 14.5 13.8 14.4 14.7 13.9 14.4 12.6 12.8 12.2 13.6 12.2 12.8 12.4 12.5 12.4 12.8 13.2 12.3 12.2

6.7 7.0 7.6 7.4 6.3 7.9 7.2 7.8 6.5 8.9 9.1 11.1 12.5 9.9 11.5 11.7 9.5 12.1 9.3 8.3 9.4 9.0 10.5 11.1 11.2 8.2 9.8 8.4 5.6 6.3 8.7 10.9 10.0 10.2 12.6 13.8 13.6

9.7 6.8 11.0 7.8 10.4 9.1 11.2 13.0 10.4 12.5 10.9 14.6 10.8 11.0 9.9 11.9 12.1 12.1 13.2 10.2 10.8 9.6 10.9 9.1 9.3 8.9 9.2 9.6 10.0 10.9 11.2 10.2

8.5 8.9 8.7 7.6 7.9 9.3 10.8 8.6 8.4 8.7 8.6 7.7 7.6 10.8 9.8 10.0 8.3 8.7 9.0 8.9 8.5 8.2 8.1 8.3 7.6 8.0 8.6 8.0 9.3 8.3 8.4 8.6 8.3 8.8 9.3 8.3

28.7 20.9 28.7 20.9 20.7 22.9 23.9 25.5 24.1 23.4 23.2 24.1 27.1 31.8 30.4 37.8 37.1 30.6 32.9 27.6 26.9 26.7 26.7 28.9 29.7 29.9 30.6 32.4 30.4 26.4 28.7 25.8 17.6 23.6

Wood Index

15.4 11.9

18.4 17.9

18.1 19.5 15.3 18.5 26.4 20.5 20.9 13.1 13.3 18.0 18.4 17.9 18.5 19.0 18.5 21.4 20.4 22.6 19.7 19.8

Farm Index 8.9 6.4 6.2 9.9 10.1 10.7 12.1 6.6 7.1 7.7 7.4 8.6 9.0 8.0 9.5 7.7 9.8 10.1 11.8 13.0 10.7 12.4 12.2 11.5 12.5 10.6 10.1 11.3 10.8 11.7 11.8 11.2 9.6 10.6 9.9 7.6 8.4 9.9 11.4 10.8 11.2 12.9 13.0 12.8

108

GREGORY CLARK

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340

9.2 11.3 10.7 9.7 8.7 8.6 7.7 7.5 9.8 9.5 8.6 9.9 11.5 12.9 12.8 9.3 9.2 10.0 11.7 23.0 22.1 12.8 7.7 9.0 11.0 18.4 14.2 10.6 12.8 9.2 7.7 8.2 10.0 10.4 11.9 13.4 9.5 7.8 8.2 8.8 8.0 6.7 5.7 9.4

10.4 9.9 9.7 13.0 11.3 10.4 9.6 8.1 10.2 11.9 10.5 7.7 13.4 14.7 12.9 13.4 11.4 11.4 12.4 14.2 18.0 15.7 13.1 12.6 10.9 13.4 14.9 14.7 14.9 13.6 11.7 12.3 11.5 11.8 12.4 11.5 11.9 11.3 11.2 11.6 10.8 10.3 9.3 10.2

8.6 8.1 10.5 9.2 10.5 10.6 9.7 9.6 9.5 12.1 10.8 10.8 11.5 12.9 12.7 12.0 12.1 11.4 12.8 15.5 13.9 11.3 11.2 12.4 12.9 13.1 16.1 12.6 12.8 14.2 8.6 11.4 12.9 12.5 10.8 9.6 9.5 10.2 10.6 11.9 9.5 11.2 8.7 11.6

22.7 26.7 34.1 31.1 28.9 25.8 26.0 26.7 30.4 32.9 35.0 36.8 37.5 34.6 29.7 25.5 27.4 31.3 30.8 29.5 31.1 33.9 33.4 40.1 44.1 34.8 34.8 36.2 38.0 32.7 34.3 31.5 28.9 28.3 29.7 28.7 25.3 20.4 21.8 26.7 22.9 17.9 19.7 19.7

12.4 12.4 13.9 15.3 14.4 13.5 12.6 12.2 13.5 15.8 14.7 12.6 17.2 18.0 16.2 15.7 14.8 15.0 16.1 17.8 19.3 17.8 15.8 16.3 15.4 16.7 19.2 18.0 18.5 18.4 14.7 15.7 15.1 15.4 15.8 14.2 14.0 13.4 13.4 14.5 12.3 11.5 11.3 13.3

17.0 20.7 18.0 19.4 23.5 21.7 20.5 16.2 23.7 24.4 24.1 24.8

10.1 11.8 11.8 11.5 10.7 10.4 9.5 9.1 11.1 11.5 10.6 11.0 13.5 14.6 14.1 11.6 11.3 11.9 13.3 20.2 20.5 14.2 10.2 11.5 12.8 17.3 15.3 12.8 14.3 12.0 10.0 10.8 11.9 12.3 13.4 13.6 11.0 9.9 10.5 11.0 9.7 8.5 7.9 11.0

31.2 24.3 26.9 28.0 20.6 25.2 21.5 22.5 27.9 28.1 26.2 16.6 24.5 30.2 28.0 36.7 31.5 27.1 26.3 26.9 29.1 34.8 28.7 29.8 26.9 25.4 29.5

The Price History of English Agriculture, 1209–1914

109

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384

7.7 8.2 7.9 10.1 7.4 7.8 11.6 11.3 7.0 10.3 14.9 17.4 11.9 10.3 9.6 11.1 12.2 13.0 11.6 11.3 12.1 11.6 13.8 14.8 12.8 11.0 11.6 12.7 12.6 22.6 13.5 10.8 12.7 10.1 14.0 14.0 9.1 7.9 7.6 9.9 10.3 9.3 9.0 9.8

10.7 9.9 10.5 10.6 11.0 10.5 11.9 11.6 12.3 10.9 11.6 11.9 15.0 13.1 12.9 11.9 10.6 11.1 12.9 12.6 13.0 12.6 10.6 11.7 13.5 11.7 12.4 12.4 13.1 14.2 13.4 11.5 14.6 13.6 13.7 13.3 13.8 11.8 12.1 13.3 13.3 11.7 13.7 11.3

8.2 7.7 7.7 8.5 9.2 12.0 10.1 8.8 8.2 10.7 10.6 10.0 10.2 11.1 12.0 11.6 12.7 11.1 9.9 13.7 13.3 10.3 11.1 10.1 11.9 11.5 11.8 11.9 11.1 13.0 9.5 11.4 10.3 10.9 11.2 11.1 11.3 11.2 10.0 10.0 9.8 9.0 9.6 10.3

19.9 24.1 19.7 25.3 23.6 25.8 23.6 24.1 14.4 18.1 19.0 20.1 18.5 19.2 21.3 19.5 20.9 21.6 16.9 22.5 19.9 22.3 19.5 20.1 26.2 26.2 30.2 29.5 25.5 29.7 29.2 31.1 32.0 34.6 33.9 33.1 38.0 32.4 31.3 29.9 22.9 22.5 25.8 25.3

12.2 11.9 11.7 12.8 13.3 14.0 13.9 13.2 12.1 12.6 13.0 13.5 14.9 14.3 14.6 13.9 13.5 13.2 13.4 15.1 15.0 14.2 12.8 13.2 15.9 14.7 15.8 15.6 15.0 17.2 14.9 15.2 16.9 16.7 17.1 16.4 17.3 15.7 15.0 15.7 14.9 13.5 15.4 13.9

28.0 29.3 28.6 28.0 23.8 31.8 28.0 65.4 19.4 32.3 27.2 18.1 32.4 37.6 28.7 31.4 26.6 29.8 32.9 30.4 33.2 29.7 29.0 31.1 28.4 29.5 33.6 31.0 28.3 33.0 32.3 26.9 27.1 32.1 34.6 32.2 34.5 32.9 25.5 27.9 26.9 31.1 32.2 33.2

9.5 10.0 9.6 11.3 9.4 10.1 12.5 13.1 8.8 11.7 14.4 15.4 13.1 12.0 11.6 12.5 13.0 13.4 12.7 13.0 13.7 13.0 14.0 14.5 14.1 12.7 13.5 14.2 13.9 20.0 14.3 12.4 14.2 12.5 15.7 15.2 12.1 10.7 10.1 12.1 12.2 11.1 11.4 11.6

110

GREGORY CLARK

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428

8.6 9.8 8.1 7.0 6.7 10.0 14.7 9.3 7.0 7.6 8.2 8.6 11.0 11.2 9.3 10.0 11.1 14.0 11.3 9.0 8.3 7.7 7.8 10.0 11.8 13.7 11.0 9.3 8.7 8.5 9.0 11.4 12.7 10.1 10.4 9.7 10.0 8.5 8.3 8.2 8.8 8.2 7.6 7.5

11.1 13.3 13.3 13.8 13.8 13.3 10.5 11.8 12.7 13.8 12.1 8.8 13.8 13.3 14.0 13.5 13.1 13.5 14.5 12.2 11.9 14.2 14.2 13.2 13.7 14.0 14.4 13.1 14.1 14.3 14.4 14.4 14.0 13.8 14.5 13.4 12.5 13.3 14.0 13.5 14.5 14.4 14.7 13.5

7.6 9.8 10.0 9.8 7.6 9.9 8.7 10.3 10.3 10.3 11.6 10.5 9.8 9.5 10.4 9.4 8.0 9.6 14.4 7.7 9.1 8.2 8.4 10.1 9.9 9.2 8.6 8.6 10.7 10.4 10.4 10.2 10.0 8.9 8.1 9.6 10.2 9.9 8.9 8.1 8.9 10.7 9.9 9.9

28.9 22.0 21.1 19.5 20.1 20.4 16.7 21.6 24.3 23.4 21.1 23.2 22.9 26.2 24.8 23.4 22.9 23.9 25.5 30.8 23.6 23.4 28.7 28.0 26.4 32.7 30.4 27.1 30.6 24.8 22.9 24.3 21.6 18.8 19.0 20.4 28.9 21.1 18.3 19.7 22.0 18.8 20.7 20.1

13.4 14.4 14.4 14.2 13.6 14.1 11.8 13.6 14.6 15.0 14.2 12.1 14.9 14.8 15.6 14.6 13.9 14.8 17.0 13.9 13.8 14.6 15.5 15.4 15.6 16.0 15.7 14.7 16.5 15.6 15.4 15.5 14.9 13.7 14.2 14.2 15.0 14.5 14.1 13.7 14.9 14.9 15.0 14.4

36.5 29.4 26.5 29.9 33.9 31.4 32.5 29.1 24.5 27.2 24.4 24.0 24.5 24.4 23.7 21.6 24.4 23.7 24.0 24.7 25.2 24.6 24.5 22.2 23.7 23.7 21.5 22.9 22.3 22.3 22.9 24.5 21.7 20.6 22.4 23.4 21.9 23.5 22.5 22.9 22.9 24.8 24.6 24.5

10.7 11.7 10.3 9.6 9.2 11.7 14.3 11.2 9.4 10.1 10.3 10.1 12.5 12.6 11.5 11.7 12.4 14.4 13.0 10.8 10.3 10.0 10.3 11.8 13.2 14.6 12.6 11.1 11.1 10.7 11.1 12.9 13.5 11.4 11.7 11.4 11.6 10.5 10.3 10.2 10.9 10.4 10.0 9.8

The Price History of English Agriculture, 1209–1914

111

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472

11.5 11.9 9.3 7.9 10.8 9.6 8.2 8.4 8.5 13.3 17.5 11.8 6.9 7.3 8.3 7.6 6.8 8.6 8.9 8.6 8.2 8.4 9.3 9.0 8.7 8.8 7.8 8.8 8.3 9.4 9.2 8.8 11.0 11.4 7.7 6.3 9.5 8.9 9.1 8.6 9.0 9.2 9.9 9.3

14.2 13.3 14.1 13.7 14.2 15.5 15.9 15.8 11.5 15.3 14.2 12.9 13.2 12.9 11.7 13.9 14.3 11.7 11.6 13.7 13.3 12.2 15.2 10.5 13.1 12.4 13.3 12.7 11.7 12.0 11.7 11.8 10.9 12.1 10.3 13.4 12.9 13.6 12.8 13.0 12.9 13.3 12.9 13.6

9.8 11.7 6.6 9.1 11.4 10.8 13.7 12.3

24.5 31.3 25.8 25.1 25.3 19.7 21.8 23.6 19.5 19.2 25.5 23.2 23.4 22.9 19.9 21.3 17.4 19.9 25.3 20.1 20.4 17.9 16.6 14.5 11.6 15.1 15.0 13.5 15.1 16.0 15.9 14.4 15.0 17.5 18.2 23.2 24.8 23.9 24.2 19.2 16.4 17.7 17.3 16.4

15.5 16.4 14.4 15.0 16.1 15.9 17.0 16.8 13.1 15.7 16.8 14.9 14.9 14.5 13.3 15.6 14.7 13.2 14.3 14.6 14.8 13.3 14.5 12.0 12.8 13.1 13.6 13.2 11.4 12.0 12.3 12.9 11.8 13.1 10.5 14.1 14.7 13.1 14.2 12.9 12.3 12.3 12.5 12.2

25.0 21.8 21.6 22.8 21.0 20.6 20.5 20.9 21.2 25.0 24.6 23.1 20.2 21.1 16.7 21.7 20.5 21.2 23.6 17.1 24.5 16.2 20.9 20.0 22.3 21.8 21.0 22.7 17.8 19.7 19.1 21.2 15.1 22.8

13.0 13.3 11.1 10.3 12.5 11.5 10.8 10.8 10.1 14.3 17.0 13.0 9.2 9.6 9.9 10.0 9.2 10.3 10.8 10.4 10.4 10.0 11.1 10.1 10.3 10.5 9.7 10.4 9.5 10.3 10.4 10.3 11.3 12.2 8.9 8.6 11.1 10.5 10.8 10.2 10.4 10.6 11.0 10.5

11.4 13.7

9.8

11.4 9.1 10.2 10.1 11.4 10.1 9.5 10.2 11.2 11.0 10.3 13.2 6.5 7.5 9.1 10.8 9.1 9.6 5.2 8.8 9.1 8.8 6.5 6.5 10.8 9.4

19.5 20.9 20.7 20.1 19.4 21.1 19.8 20.9

112

GREGORY CLARK

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516

7.9 8.4 7.8 8.3 8.8 9.3 9.7 8.8 9.0 12.8 14.1 11.2 9.2 7.7 9.5 8.8 9.7 8.9 9.8 9.8 8.9 7.9 7.8 7.6 9.0 9.0 9.7 8.2 10.6 11.3 10.0 11.1 10.6 9.4 8.8 9.5 7.5 7.3 7.5 9.0 15.1 11.1 9.4 11.7

13.6 12.0 13.7 12.7 12.2 11.4 12.1 10.2 12.1 13.0 7.4 11.0 13.3 18.1

8.5 8.5 8.7 8.3 8.5 9.1 8.5 8.5 9.0 10.5 11.3 11.2 11.2 12.5 10.8 10.4 11.0 11.0 8.6 10.9 9.8

16.7 16.0 14.6 14.1 13.6 18.7 14.7 16.3 20.0 25.8 19.3 25.4 26.0 19.9 18.1 17.1 18.2 16.8 14.4 18.1 16.0 17.9 15.4 18.4 18.4 14.9 17.7 14.0 16.1 16.7 15.3 17.0 17.2 14.6 14.2 15.9 17.7 18.8 19.1 18.1 19.1 18.8 20.9 23.7

13.2 12.3 12.9 12.3 12.0 13.0 12.5 10.0 13.1 15.1 10.9 13.9 15.4 17.0 12.2 10.2 14.3 13.5 12.3 12.7 12.2 13.4 12.5 11.8 12.9 13.7 14.5 12.9 12.6 12.7 12.7 12.7 12.6 14.2 14.1 13.7 13.2 12.7 11.4 11.6 12.2 14.5 14.2 13.8

20.1 19.9 20.2 21.5 19.7 20.4 18.1 15.8 16.5 16.8 16.2 19.0 16.3 12.6 13.5 13.3 20.0 17.2 14.5 15.2 19.2 16.7 15.4 22.2 18.7 16.1 20.0 20.1 18.9 18.9 17.5 17.7 17.5 19.0 19.5 17.0 16.3 16.7 17.4 19.3 19.5 19.6 19.1 18.8

9.7 9.9 9.6 9.9 10.0 10.7 10.8 9.5 10.3 13.4 12.9 12.1 10.9 10.0 10.3 9.3 11.3 10.4 10.6 10.7 10.2 9.6 9.2 9.3 10.4 10.5 11.3 9.8 11.4 11.9 11.0 11.7 11.4 11.0 10.5 10.9 9.2 9.0 9.0 10.1 14.1 12.3 11.0 12.5

7.8 13.4 12.5 12.2 10.9 11.2 14.0 10.9 10.8 9.3 13.4 15.0 13.6 11.5 10.4 13.0 13.5 12.1 15.1 14.2 13.7 12.9 10.8 10.1 10.9 11.8 12.3 12.1 13.3

11.6 15.0 10.9 8.8 8.9 10.3 12.6 8.3 6.7 8.2 10.4 12.4 10.0 8.9 10.2 6.7 6.7 6.7 13.4 10.8 7.2

The Price History of English Agriculture, 1209–1914

113

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560

9.4 11.4 10.7 12.7 15.3 12.1 10.4 10.2 9.7 9.2 12.3 21.5 14.5 13.4 14.1 14.3 15.0 12.6 13.3 15.1 13.5 11.3 11.9 12.2 14.2 13.1 14.7 14.1 16.4 23.2 13.5 11.4 16.4 22.6 28.1 23.3 19.5 20.4 26.6 36.9 43.5 17.8 22.2 22.3

12.1 12.1 12.1 9.4 23.0 23.3 11.4 23.0 23.0 25.8 23.0 16.1 23.0 23.0 20.0

9.4 9.4 15.3 11.6 13.6 12.3 12.0 8.3 12.1 9.3 12.4 13.8 10.7 11.4 10.3 10.2 16.0 13.6 11.7 10.9 12.3 13.3 10.8 11.1 12.8 10.2

22.6 24.7 19.8 22.1 20.5 18.1 21.4 16.5 16.8 20.9 22.1 19.3 21.3 17.9 15.3 20.4 22.9 25.6 23.8 21.3 26.3 20.5 18.9 19.7 33.4 27.2

13.9 14.4 15.1 12.9 20.0 19.0 14.0 17.1 18.7 19.7 20.0 16.8 19.1 18.7 16.5 16.7 17.6 15.7 17.9 13.4 14.6 12.8 13.7 13.5 19.9 16.4 19.7 13.2 15.7 17.4 16.5 20.2 23.8 26.9 36.7 32.7 30.0 27.5 36.3 25.8 38.8 31.8 25.3 43.1

19.5 19.0 18.4 19.0 24.5 14.6 20.0 16.6 15.1 20.0 22.2 19.4 18.8 15.9 25.7 15.0 19.7 19.1 19.4 25.8 25.5 19.9 20.4 19.8 16.9 18.5 20.7 28.7 23.3 23.9 18.4 30.6 24.8 26.6 19.0 26.2 32.9 35.0 34.1 33.0 31.0 35.8 34.1 23.8

10.9 12.5 12.2 12.9 16.6 13.6 11.7 12.1 12.0 12.1 14.4 19.3 15.7 14.7 15.1 14.7 15.8 13.6 14.7 14.8 14.2 12.0 12.6 12.8 15.5 14.0 16.1 14.3 16.2 20.8 14.4 14.3 18.5 23.6 28.8 25.4 22.6 22.8 29.1 32.1 39.9 21.9 23.3 26.6

15.6 12.3 18.0 10.7 10.8 9.4 11.4

9.4 14.1 12.9 15.5 19.1 21.6 25.2 42.0 33.6 29.4 42.0 24.0 42.0 26.1 23.9 51.9

10.2 15.3 11.9 19.7 19.0 24.3 17.3 20.1 17.2 26.8 27.3 21.7 24.0 28.0 29.8 34.4

26.3 24.7 31.6 22.1 21.1 29.4 49.3 47.0 33.7 21.7 29.1 28.0 46.0 47.7 21.9 33.1

114

GREGORY CLARK

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604

25.9 26.2 29.3 29.9 21.7 22.7 23.9 26.7 25.9 22.2 19.8 24.7 26.8 35.7 25.8 26.7 30.1 28.1 25.7 25.5 30.9 30.0 28.2 28.0 30.2 40.1 39.0 28.8 30.2 38.7 42.2 30.9 31.8 39.1 54.5 56.7 70.6 57.1 39.5 41.9 48.4 38.5 37.2 38.4

34.1 23.1 22.6 23.0 25.4 25.1 25.5 26.0 28.3 23.0 24.0 23.4 24.1 27.1 25.7 26.4 26.4 25.5 26.6 24.1 25.6 25.0 24.4 27.2 26.4 29.6 32.5 32.9 31.7 31.9 36.0 36.8 34.3 32.3 33.3 34.4 36.5 39.3 37.0 38.8 38.6 34.5 34.1 34.0

30.2 31.3 33.3 33.9 27.8 28.4 25.9 27.4 25.3 24.1 23.8 26.4 22.3 31.8 27.2 26.3 31.7 28.8 27.5 27.6 29.9 29.5 30.6 28.2 27.0 29.4 33.5 28.6 28.6 35.6 34.7 22.3 29.7 31.4 35.9 31.7 38.2 34.6 35.3 34.7 35.7 36.5 34.5 36.1

26.9 38.6 26.3 30.1 28.0 38.4

32.5 29.8 27.2 28.5 27.9 30.3 30.3 32.6 31.9 28.2 27.3 29.0 27.4 32.0 30.9 31.2 33.0 31.9 32.2 31.6 32.0 31.1 32.1 32.6 30.2 32.5 34.3 33.1 33.5 38.0 41.7 38.2 40.2 39.2 41.8 40.6 43.3 41.8 41.5 42.7 43.7 42.0 42.5 44.1

40.8 29.0 32.5 33.1 28.7 34.6 32.4 30.5 39.6 40.9 36.3 37.0 33.1 30.8 31.8 29.6 40.1 34.0 33.7 33.8 35.4 37.6 30.8 34.8 39.7 38.9 38.0 37.1 31.9 39.9 37.5 39.0 37.7 43.2 38.6 43.8 43.3 41.4 43.3 41.7 46.9 49.1 49.5 49.6

28.1 26.8 28.2 29.0 23.3 25.0 25.6 28.0 27.8 24.5 22.4 26.1 26.8 33.3 27.0 27.5 30.9 28.9 27.4 27.1 30.8 30.1 28.8 29.1 30.1 36.7 36.5 29.9 30.6 37.6 40.7 32.8 33.8 38.5 47.8 49.0 57.2 49.5 39.4 41.1 45.7 39.3 38.7 39.9

47.8 40.5 40.8 33.2 41.0 34.5 38.4 39.7 41.3 44.0 44.6 46.0 49.1 45.1 41.0 46.4 41.0 37.0 37.5 31.4 32.8 38.1 43.8 54.5 56.5 58.5 60.3 60.7 57.0 56.8 48.9 50.7 48.4 51.9 55.8 64.8 72.2

The Price History of English Agriculture, 1209–1914

115

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648

42.1 43.4 44.1 56.1 56.5 47.9 47.1 57.8 60.7 59.3 56.3 57.7 54.9 52.2 46.9 41.6 46.5 62.1 60.8 52.1 56.1 57.4 45.5 46.5 54.9 69.2 76.4 61.5 61.8 66.7 63.8 63.8 71.8 75.7 56.9 49.3 60.7 55.2 56.1 52.6 54.5 58.2 78.0 93.7

36.2 35.7 36.3 37.6 37.7 36.5 38.0 37.8 38.6 38.2 41.6 42.0 41.2 40.8 42.0 41.2 40.2 38.3 37.1 40.1 41.5 41.2 41.1 41.2 41.2 40.6 41.6 41.3 41.2 42.8 44.5 43.8 44.6 45.4 44.7 43.9 44.3 45.7 43.8 43.0 43.6 43.6 51.2 48.2

32.6 33.7 36.6 37.2 37.1 37.6 40.3 40.6 40.6 38.2 41.8 39.0 38.0 40.1 39.8 40.9 39.4 39.1 42.3 40.6 37.4 41.4 40.5 38.2 42.7 41.1 38.8 38.9 42.5 43.8 42.7 43.4 45.0 51.6 52.0 37.1 41.4 41.2 39.9 46.6 44.5 47.3 55.6 72.7

65.0 63.5 62.4 63.9 58.5 52.1 47.5 52.6 50.2 54.3 60.3 65.0 73.6 70.2 75.1 70.6 59.2 55.4 46.8 55.4 57.2 66.3 66.1 73.1 71.3

43.8 43.2 44.5 46.0 45.0 43.0 43.3 44.7 44.4 44.4 49.2 49.5 49.8 49.5 50.8 50.5 47.2 45.4 43.6 46.5 46.5 49.3 48.7 49.4 50.6 50.3 49.3 49.1 50.5 52.6 54.7 53.6 55.2 56.3 54.9 50.0 50.6 51.8 44.0 50.7 44.4 45.5 60.8 65.7

52.9 53.3 60.6 62.6 63.3 65.8 63.8 64.3 59.9 61.5 66.9 64.2 60.7 63.9 65.1 64.7 58.9 63.5 66.0 59.0 63.7 68.0 67.2 77.8 79.0 53.5 75.3 68.6 65.1 73.7 76.8 74.5 73.8 82.2 80.4 85.9 86.3 81.3 80.7 90.4 81.7 79.9 86.3 83.4

42.4 43.0 44.3 52.0 52.0 46.4 45.9 52.7 53.9 53.2 53.5 54.3 52.4 51.0 48.1 44.6 46.5 55.3 54.1 49.7 52.3 54.3 46.8 48.2 53.8 60.1 65.3 56.6 57.1 61.1 60.3 59.8 64.9 68.0 56.5 50.5 57.7 54.6 52.4 53.1 51.7 54.2 71.2 81.5

68.9 71.0 71.0 70.3 74.5 75.1 74.4 73.5 70.6 65.0 61.0 62.3 59.6

74.9 87.2

116

GREGORY CLARK

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692

84.2 81.7 66.4 64.7 57.7 43.0 42.4 57.2 64.9 69.6 74.0 67.2 73.1 79.5 65.4 61.4 55.5 50.7 45.6 46.1 56.4 53.6 53.3 49.2 55.6 72.1 65.1 49.0 55.4 59.6 59.6 51.0 58.5 60.1 54.9 57.6 62.8 54.0 49.9 45.3 40.0 43.6 41.2 55.0

50.0 55.5 55.1 53.7 51.5 45.9 39.8 39.9 41.7 49.6 49.8 51.8 51.9 49.5 49.2 52.3 50.4 49.5 49.7 47.8 45.6 46.7 45.6 44.4 45.7 46.8 46.8 46.1 46.6 47.8 45.9 46.1 46.3 47.0 45.4 48.2 50.9 49.2 44.9 45.1 45.9 45.2 44.7 44.8

64.3 54.6 65.4 36.7 46.3 42.8 43.0 49.0 50.0 49.7 54.8 48.4 48.6 47.5 50.1 52.3 48.2 50.6 55.9 51.0 46.5 52.9 49.0 50.4 51.6 55.4 53.3 63.9 54.3 56.8 56.4 48.1 53.8 50.2 54.0 51.7 55.0 44.1 45.2 47.3 47.1 52.5 48.3 47.8

74.9

63.4 57.8 62.5 53.2 51.1 45.6 41.9 45.6 47.0 51.2 51.7 49.9 53.9 51.5 55.3 54.4 50.0 50.8 51.9 48.5 53.2 48.6 48.1 54.9 49.3 48.4 48.4 48.4 55.4 45.2 50.2 46.4 51.9 48.7 49.4 50.5 51.9 47.4 45.9 43.4 43.9 44.6 45.8 46.8

84.8 54.1 73.9 71.1 81.2 74.7 81.6 88.3 90.1 76.2 87.6 78.0 93.6 97.0 90.2 93.0 97.1 94.8 93.1 100.9 81.1 69.5 74.0 102.5 104.0 103.4 94.4 93.0 97.7 80.1 92.1 95.0 94.2 89.2 99.3 105.1 100.8 98.3 96.7 94.6 90.2 84.4 93.3 97.9

75.5 69.5 64.2 60.0 55.8 44.7 43.5 54.1 59.1 62.4 65.7 60.7 66.4 69.2 62.3 59.7 54.9 52.0 49.0 48.6 55.7 51.9 51.8 52.7 55.0 64.3 59.9 50.1 56.6 54.8 57.2 50.8 57.2 56.8 54.4 56.7 60.1 53.2 50.1 46.3 42.9 45.2 44.3 53.6

60.6 55.5 39.6

41.7 42.5 42.9 40.0 35.4 50.1 46.7 56.4 47.2 37.4

37.4 65.2 39.1 70.3 40.4 34.5 36.2 30.7 60.6 24.8 32.8 47.2 38.7

37.9

28.2 27.8 33.7 37.0

The Price History of English Agriculture, 1209–1914

117

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 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

68.2 64.4 56.6 65.2 69.2 79.6 75.4 59.4 47.4 46.2 44.3 51.1 46.1 47.8 46.8 57.0 75.1 79.3 65.3 57.2 57.1 66.3 56.0 58.4 56.2 51.2 51.8 60.5 54.5 47.8 50.8 55.4 58.9 63.1 60.4 74.9 68.2 52.5 50.0 45.7 44.9 51.0 55.2 57.6

48.5 51.3 54.4 53.0 49.6 52.8 52.1 51.3 51.3 50.7 46.9 45.2 45.0 44.9 43.7 45.1 46.6 50.5 52.3 48.3 48.0 45.4 47.2 48.3 50.1 49.0 47.1 49.3 48.7 46.3 43.8 45.5 47.6 49.1 48.0 48.9 51.7 50.5 49.0 46.4 44.6 41.6 41.8 44.6

50.1 50.8 50.2 50.5 48.6 50.2 51.1 44.7 45.0 49.3 44.7 42.8 41.3 45.5 43.0 41.0 43.0 45.6 54.2 44.6 42.9 40.6 38.8 41.5 40.7 41.2 42.0 43.4 44.3 44.0 42.7 43.3 45.5 45.3 46.1 45.0 47.6 47.0 44.2 43.5 42.0 41.9 42.3 42.4

38.5 38.5 42.5 52.9 46.3 50.9 48.0 44.2 45.4 45.9 40.0 45.9

49.3 50.6 53.6 56.2 52.0 55.4 54.6 51.5 51.8 52.8 48.2 47.7 45.6 46.1 43.7 43.2 42.5 46.1 47.4 44.5 46.2 45.0 44.8 46.4 48.3 50.2 48.8 49.5 48.2 45.1 43.8 44.2 44.8 45.9 45.5 46.5 49.0 47.5 48.0 46.6 45.1 42.1 41.5 41.9

95.9 94.7 94.4 98.7 97.4 104.2 95.6 111.9 93.9 93.4 87.1 90.0 87.5 84.6 85.5 89.2 92.0 94.8 90.4 102.7 94.2 95.8 98.0 84.8 86.9 84.1 88.0 86.1 79.5 89.0 85.5 84.6 85.4 94.5 84.4 88.8 91.5 89.1 88.3 79.4 85.7 88.4 78.2 85.0

62.1 60.3 56.6 62.9 63.7 71.2 68.1 57.5 50.6 50.3 47.2 50.6 47.1 48.0 46.5 51.4 58.7 62.5 57.1 52.8 53.1 56.7 52.1 53.2 53.2 51.4 51.3 55.6 51.8 47.8 48.6 50.8 52.7 55.6 53.7 60.6 59.2 51.2 50.1 46.8 46.2 48.0 49.1 50.7

36.9 32.2 31.1 26.5 29.9 26.1 28.3 33.2 35.5 34.1 33.7 39.9 47.7 44.5 40.7 38.4 34.1 34.0 30.9 28.5 30.4 30.0 31.9 34.2 32.9 36.4 35.0 35.3 32.5 29.8 26.9

118

GREGORY CLARK

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 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 1779 1780

55.4 53.0 51.4 66.3 66.9 53.1 45.3 42.0 44.5 49.5 46.1 49.4 54.0 51.1 53.9 58.2 60.4 56.3 50.2 59.8 80.3 68.4 53.1 52.2 48.9 56.3 66.9 67.7 73.7 73.2 78.7 74.0 63.0 64.0 75.9 82.3 86.1 86.1 83.1 70.1 73.6 69.8 60.4 60.9

45.5 44.9 45.5 47.9 52.7 51.7 48.5 46.0 45.0 45.7 46.4 47.6 46.9 45.3 46.0 45.8 45.7 50.0 49.0 49.3 50.9 53.5 53.4 49.6 50.0 49.0 49.0 51.2 52.8 55.9 59.9 59.6 57.2 57.6 60.2 63.4 63.8 61.8 61.9 61.8 61.6 62.5 62.8 61.6

44.2 43.8 44.2 45.1 50.3 50.7 45.2 42.6 41.6 44.8 48.0 47.2 47.1 46.3 44.8 45.3 46.0 47.1 47.7 47.5 48.0 48.2 48.8 47.4 48.1 47.3 47.2 46.4 47.8 48.1 50.1 49.3 49.7 50.6 50.8 56.2 56.6 56.6 55.0 56.0 57.9 60.7 60.8 58.7

27.5 25.8 24.9 25.8 25.6 27.2 34.1 36.5 33.6 33.3 32.7 33.8 35.8 37.5 38.6 41.4 28.8 30.2 29.1 29.3 36.9 41.4 37.5 36.2 31.3 30.6 36.3 37.6 36.7 39.1 35.9 30.5 29.5 29.3 32.5 30.5 31.9 34.6 36.1 35.4 33.6 29.7 26.9 29.4

43.2 42.3 42.0 44.7 47.3 48.2 47.3 46.1 44.3 45.1 46.0 46.5 47.1 47.1 47.0 47.8 44.4 47.6 46.3 46.0 50.0 53.0 51.6 48.6 47.5 46.8 48.3 49.5 50.6 53.0 53.5 51.7 50.7 51.3 53.6 57.4 58.7 57.0 57.1 58.6 58.3 57.1 55.8 56.2

81.8 83.4 80.5 84.6 83.1 82.8 82.1 77.9 85.7 85.2 79.2 82.2 80.6 85.4 78.2 93.7 90.3 95.3 83.8 74.8 88.5 85.3 82.3 86.1 76.0 79.1 83.1 83.4 98.2 84.4 81.2 88.0 85.0 89.1 85.4 91.8 101.4 96.0 101.6 102.1 100.2 103.4 96.7 82.1

50.2 48.8 47.7 55.8 57.2 51.4 47.1 44.6 45.6 48.4 46.8 48.9 51.2 50.1 51.0 54.3 53.5 53.4 49.2 53.0 64.6 60.8 52.8 50.9 48.5 52.0 58.0 59.0 62.6 63.4 66.1 63.2 57.4 58.2 65.0 70.1 72.7 71.7 70.4 65.0 66.6 64.1 58.6 58.9

The Price History of English Agriculture, 1209–1914

119

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

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 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824

70.2 74.8 88.1 88.1 76.2 77.5 75.4 76.0 80.5 82.5 80.9 79.6 88.8 95.9 109.6 119.5 92.4 86.7 110.5 171.4 191.8 107.4 100.5 99.2 137.7 123.3 131.5 131.0 143.0 156.7 148.4 188.9 174.2 125.4 106.7 118.6 152.5 153.0 132.5 106.5 91.9 81.6 96.1 113.1

60.5 59.0 61.0 63.1 66.6 63.3 67.4 68.9 67.0 67.8 69.6 70.2 71.5 71.4 75.4 86.2 96.6 90.9 88.3 102.9 123.0 122.8 115.2 110.6 109.5 109.8 107.8 114.2 121.2 122.8 124.0 128.5 146.1 151.5 126.8 101.9 99.5 103.8 123.3 118.7 101.7 79.0 77.3 89.4

56.2 55.8 56.9 57.4 57.0 59.3 59.2 58.8 57.4 59.3 60.7 60.6 63.0 63.3 64.5 69.1 79.0 79.8 80.4 87.9 93.5 91.7 91.6 95.5 99.0 99.1 103.6 108.7 109.8 114.3 121.8 124.4 120.7 123.5 124.0 108.8 102.9 117.1 118.3 110.7 105.9 97.0 95.0 101.3

27.1 24.3 28.3 32.6 30.2 31.7 40.5 40.6 40.6 42.7 41.9 53.2 41.5 42.8 47.0 48.8 47.4 45.3 58.2 57.4 63.7 66.2 65.8 72.7 80.7 75.0 66.9 61.6 81.9 78.8 58.6 66.5 75.3 94.1 97.3 65.0 76.3 101.5 73.3 70.9 59.7 55.2 57.4 61.4

54.8 53.1 56.2 59.2 60.5 59.6 62.1 63.3 62.7 63.1 63.4 67.6 67.0 67.4 70.9 79.1 84.7 79.1 83.1 94.4 106.8 105.4 103.1 102.6 103.4 101.9 101.1 105.5 113.9 117.0 116.6 118.3 125.7 133.5 122.5 99.1 98.4 110.7 116.7 107.7 93.9 80.5 81.5 92.4

83.8 89.0 86.2 86.5 84.1 83.8 80.3 86.5 81.9 85.0 87.8 81.6 84.1 86.3 88.7 89.5 97.1 105.6 113.9 107.4 107.7 104.0 117.9 120.4 123.0 128.2 133.9 144.5 144.2 153.3 142.5 150.6 158.8 161.1 153.5 136.4 132.4 142.0 128.5 127.3 125.1 123.8 130.7 127.3

62.9 64.3 71.9 73.6 68.7 68.9 69.0 70.0 71.9 73.1 72.5 73.8 78.1 81.6 89.7 98.7 88.7 83.3 97.1 130.1 145.9 106.0 101.6 100.6 120.6 113.0 116.7 118.9 128.9 137.1 132.8 152.4 150.0 129.0 113.7 109.4 124.9 132.0 124.7 107.0 93.0 81.4 89.5 103.3

120

GREGORY CLARK

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868

124.5 109.3 107.5 104.5 103.1 114.4 108.8 98.4 89.1 85.6 84.4 91.3 99.5 104.1 118.9 111.2 108.8 95.2 92.3 94.1 95.0 96.6 125.2 87.6 88.2 74.3 76.9 79.9 94.8 114.2 114.5 109.4 104.3 88.7 86.3 102.4 105.2 103.3 89.9 81.5 84.6 99.9 117.3 116.9

101.2 103.5 100.9 99.5 94.8 83.4 87.9 89.0 84.7 81.4 73.6 82.3 85.5 89.8 86.6 88.1 94.4 84.7 70.5 72.7 73.2 86.5 96.2 93.2 80.6 75.9 74.5 75.5 88.2 93.2 95.6 96.0 98.1 91.7 94.7 100.5 99.1 94.5 94.3 99.0 106.9 106.6 96.9 96.1

112.5 109.6 106.2 105.5 99.8 96.7 102.0 98.8 97.1 90.0 90.7 99.3 101.4 99.7 100.7 102.1 101.7 97.6 92.1 89.7 92.4 96.8 97.8 91.6 89.1 85.2 76.3 72.2 79.6 86.9 93.9 100.4 99.1 94.8 94.3 99.1 96.2 94.8 94.0 95.2 106.6 108.7 100.1 100.3

71.9 43.8 42.9 46.8 37.4 45.1 54.2 57.4 62.3 82.3 78.7 86.0 66.5 71.7 76.5 66.6 54.3 49.7 49.3 59.3 67.1 59.8 55.9 51.3 47.5 51.7 58.4 63.7 74.8 72.5 60.8 74.8 95.8 73.0 87.1 94.1 91.2 95.8 105.8 128.0 120.4 109.9 88.3 81.8

102.2 94.9 94.0 92.0 83.8 79.9 85.9 88.0 84.6 86.2 81.5 89.1 88.3 92.5 91.0 89.0 90.6 82.5 73.7 76.4 80.3 85.3 88.5 84.6 76.6 74.6 74.2 72.2 85.8 90.5 92.1 95.3 96.1 89.0 93.3 101.9 99.5 94.1 93.5 98.7 108.4 106.4 96.5 95.6

125.5 128.0 117.8 120.4 121.2 117.4 116.9 120.5 119.7 136.5 123.4 119.1 127.3 123.7 130.5 127.7 122.7 119.0 119.7 121.2 119.7 112.0 101.4 94.7 97.7 94.9 94.2 79.5 100.4 120.8 108.2 112.9 95.8 89.0 102.5 104.9 105.8 105.8 104.0 131.9 112.2 83.5 79.4 83.5

113.7 102.6 101.1 98.7 94.1 97.4 97.8 93.7 87.2 86.3 83.4 90.5 94.5 98.8 105.5 100.7 100.2 89.4 83.7 85.9 88.3 91.3 106.6 86.1 82.7 74.8 75.9 76.2 90.5 102.1 103.0 102.5 100.3 89.1 90.1 102.5 102.6 98.9 92.1 90.3 96.2 103.1 106.4 105.8

The Price History of English Agriculture, 1209–1914

121

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

Wood Index

Farm Index

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 1906 1907 1908 1909 1910 1911 1912

99.0 94.6 105.2 106.6 111.8 114.0 97.7 97.1 109.9 98.2 91.9 92.6 91.3 93.2 91.5 83.6 78.9 72.3 71.3 72.9 72.5 76.4 82.9 74.6 70.8 66.1 60.5 62.9 66.3 73.9 66.9 65.7 67.3 69.6 67.8 65.6 75.4 74.2 73.0 74.2 81.3 74.8 81.5 84.2

106.0 109.7 110.0 110.9 116.0 108.7 117.0 117.1 105.8 103.6 97.3 104.7 108.1 111.2 111.6 101.5 91.7 86.8 80.7 82.2 85.9 80.1 81.4 85.2 90.6 84.7 77.1 69.3 76.0 76.0 73.0 84.2 84.5 87.3 86.2 87.0 88.5 91.3 93.6 91.1 88.7 95.2 90.7 103.0

105.0 101.5 103.5 103.2 106.0 108.3 109.2 108.8 114.5 107.3 97.3 100.9 102.8 102.3 104.0 99.6 98.4 90.7 94.8 93.3 90.7 87.7 88.5 90.8 88.1 85.3 78.0 71.6 77.6 72.3 79.5 81.3 81.9 84.0 83.0 88.4 84.8 75.6 78.6 85.7 85.0 84.4 84.4 90.9

84.7 78.3 100.0 119.8 114.5 97.0 92.3 83.0 76.0 70.1 58.4 70.7 57.9 52.6 46.8 46.8 46.2 46.8 49.1 48.5 51.4 51.4 45.6 40.9 47.9 47.3 56.1 53.8 45.0 40.9 38.6 36.8 32.2 29.2 33.9 47.3 57.9 62.6 57.3 39.7 42.1 46.2 46.8 49.1

105.2 103.2 109.9 106.7 108.9 99.9 109.0 110.5 104.0 99.2 90.9 97.6 98.4 101.0 96.8 92.0 85.8 83.0 81.8 84.7 84.7 77.8 77.7 83.2 90.9 85.9 76.0 69.6 74.3 72.0 71.7 78.4 79.5 81.7 80.0 82.9 83.5 82.0 83.8 81.5 80.8 84.2 85.0 94.7

89.0

102.2 100.6 108.7 107.0 110.1 104.5 105.5 106.3 106.1 99.1 91.4 96.2 96.3 98.7 95.3 89.4 83.8 79.6 78.5 80.9 80.8 77.5 79.5 80.6 84.1 79.2 70.8 67.5 71.8 72.8 70.3 74.3 75.6 77.8 76.1 77.1 81.0 79.6 80.4 79.3 81.2 81.2 84.1 91.5

122

GREGORY CLARK

Table A.4. (Continued ) Year

Arable

Meat

Dairy

Wool

Pasture

1913 1914

76.0 87.9

101.7 106.9

91.6 94.0

57.9 59.0

94.5 93.0

Wood Index

Farm Index 88.4 91.6

Table A.5. The Source of Price Data. Source Archival sources Beveridge Mss, LSE

Cheshire record office Dorset record office Essex record office Farmer Mss Guildhall library Kings college, Cambridge Lancashire record office Staffordshire record office Printed sources Afton and Turner (2000)

Atkins (1977) Bailey (1953) Barmby (1888, 1896) Beveridge (1929) Beveridge (1939)

Botelho (1999) Boulton (2000) Bowden (1967, 1985) Brinkworth (1964) Burgess (2000) Campbell et al. (1993)

Commodity

Wheat (387), Rye (84), Barley (107), Oats (85), Peas (62), Beans (83), Straw (49), Saffron (53), Hay (258), Fats (918), Milk (85), Butter (92), Cheese (10), Wool (13), Firewood (246), Timber (260), Cider (221), Honey (126) Hay (11) Suet (8), Butter (22), Cheese (7) Hay (46) Wheat (8,501) Firewood (52) Hay (28), Firewood (5), Timber (47) Hay (43) Hay (19) Wheat (64), Barley (64), Oats (64), Beans (80), Potatoes (61), Beef (65), Mutton (66), Pork (52), Tallow (60), Eggs (96), Milk (17), Butter (61), Cheese (25), Wool (69) Hay (60) Grease (2) Hay (33), Grease (3), Wool (34) Wheat (476) Wheat (187), Barley (378), Oats (280), Peas (210), Beans (280), Straw (458), Potatoes (70), Hops (742), Hay (317), Beef (471), Mutton (281), Pork (415), Bacon (210), Fats (645), Eggs (304), Milk (297), Cream (160), Milk (613), Cheese (711), Firewood (392), Timber (116) Straw (5) Beef (109), Suet (88), Eggs (90), Cream (63) Wool (230) Barley (18) Wheat (9), Saffron (30) Wheat (35)

The Price History of English Agriculture, 1209–1914

123

Table A.5. (Continued ) Source Coleman (1984) Doree (1994) Erskine (1981, 1983) Farmer (1988, 1991a, b) Finberg (1951) Gayer, Rostow, Schwartz (1953) Gentleman’s Magazine, 1797–1865 Gras (1915) Hill (1956, 1966) John (1989)

Marsh (1913–1939) Maslen (1993) Mercer (1928) Mitchell and Deane (1971) Mellows (1939) Northeast (1982) Parliamentary Papers (1903)

Plomer (1915) Rappaport (1989) Sauerbeck (1886) Stallard (1922) Swayne (1896) Thorold Rogers (1866, 1882, 1888, 1902)

Tooke and Newmarsh (1857) Wardle (1923) Weatherill (1990) Weaver and Clark (1925) Woodward (1995)

Commodity Wheat (19) Grease (11), Timber (4) Straw (13), Hay (24), Fats (18) Barley (270), Oats (263), Peas (263), Cheese (242), Wool (268) Wheat (61), Butter (74), Cheese (61), Wool (18) Pork (61), Tallow (61) Straw (70), Hay (70), Beef (11), Pork (38), Tallow (83) Wheat (84), Barley (11), Oats (3) Wheat (256), Rye (130), Barley (145), Oats (206), Peas (132), Beans (184) Wheat (80), Rye (109), Barley (119), Oats (80), Peas (89), Beans (31), Potatoes (50), Beef (162), Butter (70), Cheese (35), Wool (280) Saffron (34), Beef (3), Suet (4), Milk (19), Butter (16), Timber (21) Straw (4), Fats (2), Butter (7) Straw (2) Wheat (221), Wool (69) Barley (6) Straw (4) Potatoes (48), Hops (104), Beef (83), Mutton (125), Bacon (29), Fats (78), Eggs (41), Milk (132), Butter (230), Cheese (88) Timber (5) Cream (104) Timber (24) Straw (2), Grease (2) Straw (3), Timber (8) Wheat (2,321), Rye (108), Barley (699), Oats (421), Peas (441), Beans (484), Straw (642), Potatoes (48), Hops (28), Mustard Seed (207), Saffron (136), Hay (890), Beef (214), Fats (622), Eggs (394), Milk (28), Cream (66), Butter (419), Cheese (58), Wool (161), Firewood (960), Timber (257), Cider (115), Honey (146) Timber (75) Tallow (2), Firewood (22) Fats (5) Barley (2) Beef (2)

THE GROWTH OF WORLD AGRICULTURAL PRODUCTION, 1800–1938 Giovanni Federico ABSTRACT World population has increased six-fold in the last two centuries, and thus agricultural production must have grown as well. The last fifty years of this increase are covered by the Food and Agriculture Organization (FAO) production series. This article aims to push our quantitative knowledge back in time as far as possible. It reviews the scattered evidence on agricultural production in the first half of the 19th century, estimates a yearly series of output for the main countries since 1870, and puts forward some guesstimates on trends in the rest of the world. In the long run, agricultural production has increased more than population. Growth has affected all continents, even if it has been decidedly faster in both the countries of Western Settlement and in Eastern Europe, than in Asia or in Western Europe. It was faster before World War I, a veritable golden age for world agriculture, than in the inter-war years. The composition of production has changed as well, with an increase in the share of livestock products.

Research in Economic History Research in Economic History, Volume 22, 125–181 © 2004 Published by Elsevier Ltd. ISSN: 0363-3268/doi:10.1016/S0363-3268(04)22003-1

125

126

GIOVANNI FEDERICO

1. INTRODUCTION: WHY SHOULD WE CARE ABOUT AGRICULTURE? D. Gale Johnson reminded the audience in his 1999 Presidential Address to the American Economic Association that “people today have more adequate nutrition than ever before and have acquired that nutrition at the lowest cost in all human history, while the world has more people than ever before – not by a little but by a lot” (Johnson, 2000, p. 1). Nowadays, world population exceeds six billion people and, in theory, each of them could consume 2800 calories per day – a more than adequate intake.1 This average conceals wide disparities among the continents and malnutrition is still widespread, especially in Sub-Saharan Africa, where the official average daily availability is about 2200 calories. However, true starvation is rare, and is almost always caused by wars and political events, which disrupt agriculture and trade in agricultural products, and make food relief efforts too dangerous. Two hundred years ago, world population was a mere one billion, and its average caloric consumption was undoubtedly lower – possibly as low as 1800 calories in France or 2200 in the United Kingdom, the two most advanced countries in Europe.2 Throughout the world, there was a real risk of starvation, especially for poor and destitute people, and terrible famines hit several countries in the 19th century (e.g. Ireland, Finland, India, and so on). Thus, there must have been a huge increase in world agricultural production. Indeed, according to the latest FAO estimates, world gross output increased by 60% from 1938 to the late 1950s, and more than doubled from then to 2001.3 Output must also have increased in the previous one hundred and fifty years, but the extent of this growth is still poorly known. Before 1870, the statistical evidence is scarce. Historians have tried to deduce the performance of agriculture from that of the overall economy: agricultural production is assumed to have grown fast in the early starters (notably, the U.K., but also the U.S.), and to have remained stagnant in the late-comers, such as Italy or Russia. The evidence on the period after 1870 is more abundant, but it does not seem to attract much attention among historians. For instance, agriculture is barely mentioned in popular textbooks on 19th and 20th century modern economic growth, such as those by Rosenberg-Birdzell (1986), Cameron (1989) and Landes (1998). Agriculture does not directly feature in the recent literature on 19th century globalization (Williamson & O’Rourke, 1999) either. Their general framework, however with its strong stress on factor endowments and migration flows, implies different rates of growth in agricultural production comparing the New World (North America, South America and Oceania) with the Old World (Europe). The combination of abundant land and immigrant labor must have caused production to grow faster in the countries of Western Settlement than in Europe, where the

The Growth of World Agricultural Production, 1800–1938

127

land endowment was roughly constant, and the labor force was not increasing fast. The fall in freight rates made it possible to feed Europeans with the production of Western Settlement countries. Agriculture regains a central (and negative) role in interpretations of economic trends after the Great War. In fact, overproduction in the 1920s and the fall in agricultural prices are routinely listed among the causes of the Great Crisis.4 One can sum up the conventional wisdom in five stylized facts: (1) agricultural production grew in the long run, at least as much as population and probably more; (2) this growth was slow in the first half of the 19th century, accelerated in the second half of the century and at the beginning of the 20th, only to slow down again after World War I; (3) the growth was faster in Western Settlement countries than in the long-settled areas of Europe and Asia, where it was faster in the “advanced” countries than in the “peripheries”; (4) before 1913, the integration of world markets caused prices to converge, so that prices rose in land-abundant exporting countries and fell in land-scarce European countries (when not artificially propped up by duties); (5) prices in the 1920s and 1930s were low and not profitable. This article aims to test these statements, focusing on the first three.5 After a brief methodological discussion in Section 2, Section 3 reviews the evidence on agricultural growth, mainly in Europe, during the first seventy years of the 19th century. Section 4 deals with the period from 1870 to 1938, on the basis of a new series of “world” production, which covers the whole of Europe (except for Norway and some Balkan countries), North America and Oceania, and substantial parts of Asia and South America.6 Section 5 discusses the reliability of this series and the possible biases from errors in the country data or in the aggregation procedure. Section 6 presents the available evidence on production trends in other countries (including China), while Section 7 puts forward some guesstimates about total world output. Finally, Section 8 deals with the change in the composition of agricultural production. Section 9 concludes.

2. SOURCES AND METHODS Agricultural production can be measured either by gross saleable production or GSP (often referred to as “gross output” or “final product”) or by Value Added (or GDP).7 The former is defined as the total market value of all products, net of re-uses within agriculture itself of seed and feed, but inclusive of farmers’ domestic consumption, while Value Added is the GSP net of the cost of inputs purchased from outside the sector. It is worthwhile computing both series, as they measure two different aspects of agricultural performance. The gross output measures the capability of agriculture to provide food, clothing, and heating, while

128

GIOVANNI FEDERICO

Value Added measures its capability to create income. Furthermore, the ratio of Value Added to Gross output is a simple proxy for the diffusion of “modern” agricultural techniques which require the purchase of industrial output (fertilizers, fuel, industrial feedstuffs, etc.). It is likely to have declined in the long run – a sixth stylized fact to test. In recent years, economic historians have worked hard to estimate national accounts and series of agricultural production. It has been possible to find yearly series for twenty-five countries (at their 1913 boundaries). In some cases, the source provides both Gross Output and Value Added, in others only one series. Some of these series extend back in time to the first half of the 19th century (as early as 1800 for Sweden), while the majority start in the 1850s or 1860s, and five start after 1870. The series for some key European countries (Russia, Germany, France, etc.) do not cover the war-time years because during the period of hostilities these countries ceased to publish statistics. With some plausible guesswork, it has been possible to build twin series of Gross Output and Value Added for all twenty-five countries from 1870 to 1913 and from 1920 to 1938.8 They refer to agriculture only, not to the primary sector as a whole, as the data on production in forestry, fishing and hunting are not available for some key countries, such as the United States, France, and the United Kingdom. However, the differences between agriculture and the primary sector are very small: the omitted activities account for more than a tenth of the production of the primary sector only in Sweden and Finland.9 “World” indices of Gross Output and Value Added are obtained by weighting the country series with their respective shares of production in 1913. This year has been chosen for sound historical reasons (it marks the end of a long period of expansion of the world economy) and for more mundane ones. It seems advisable to select a late date, because the accuracy of the data tends to increase through time, but the choice of any post-war date (e.g. 1938) would amplify the effect of any error in boundary adjustments. The value of production in 1913, measured by sources in national currencies, is converted into British pounds at the market exchange rates.10

3. THE GROWTH OF AGRICULTURAL PRODUCTION IN THE FIRST HALF OF THE 19TH CENTURY The statistical evidence on agricultural production in the first half of the 19th century (Table 1) is incomplete and, in all likelihood, less accurate and reliable than for later periods.

The Growth of World Agricultural Production, 1800–1938

129

Table 1. Rate of Growth of Agricultural Production and Population Before 1870. Country

Australia Austria Belgium Denmark France (a) France (b) England (a) England (b) England (c) England (d) Egypt Germany (a) Germany (b) Germany (c) Germany (d) Indonesia Netherlands (a) Netherlands (b) Greece Poland Portugal Spain (a) Spain (b) Sweden United States

Production

Population

Period

Rate

Period

Rate

1828–1870 1830–1870 1812–1870 1818–1870 1803–1812/1870 1821–1870 1800–1870 1800–1830 1800–1850 1800–1809/1870–1879 1821/1872–1878 1800–1810/1866–1870 1816–1849 1800–1810/1846–1850 1850–1870 1815–1817/1869–1871 1808–1870 1851–1870 1848–1870 1809–1870 1848–1870 1800–1870 1850–1870 1800–1870 1800–1870

8.42 0.57 0.64 1.31 0.90 1.12 1.10 1.18 1.00 0.93 5.19 1.50 2.61 1.60 1.49 1.43 1.10 1.40 2.72 2.65 –0.79 0.57 0.70 1.44 2.91

1828–1870 1840–1870 1816–1866 1801–1870 1806–1866 1821–1866 1801–1871 1801–1831 1801–1851 1801–1871 1821/1872–1878 1817–1870 1817–1850 1817–1850 1850–1870 1820–1870 1808–1870 1851–1870 1850–1870

7.97 0.63 0.30 0.95 0.41 0.50 1.34 1.18 1.40 1.34 1.54 0.91 1.02 1.02 0.72 0.96 0.83 0.75 2.00 NA 0.53 0.62 0.36 0.82 2.88

1841–1878 1800–1870 1857–1877 1800–1870 1800–1870

Note: All data computed as geometric interpolations between three-years moving averages (if not otherwise indicated). Sources: Population data: Mitchell (1998a, b, c, Tables A1 and A5). Production data: Australia: ButlinSinclair (1986); Austria: Kausel (1979, Table 1a); Belgium: Goosens (1992, p.155); Denmark: Hansen (1974, Table 4); Egypt: O’Brien (1968, Table 7); England (and Wales) (a) Deane and Cole (1968 Table 38); (b) Crafts (1985, Table 2.10); (c) Allen (1999, p. 215); (d) Clark (2002, Table 5) England and Wales France: (a) Toutain (1961), (b) Levy-Leboyer (1968); Germany: (a) Helling (1965), (b) Tilly (1978), (c) Franz (1976, Tables 16 and 17); (d) Hoffmann (1965, ii Table 64); Greece: Petmezas (1999) and personal communication; Indonesia (Java): Van Zanden (2003) and personal communication; Netherlands: (a) Van Zanden (2000), (b) Knibbe (1994); Poland (Kingdom) Kostrowicka (1984, Table1); Portugal: Lains-Silveira Sousa (1998); Spain: (a) Gutierrez Brigas (2000, quadro VI.1), (b) Prados (2000); Sweden: Schon (1995, Table J1); United States: Weiss (1994, Table 1.6).

130

GIOVANNI FEDERICO

The results tally only partially with the conventional wisdom. First, the performance is better than often assumed. Total production rose in all countries except Portugal, and, in nine cases out of fifteen, it grew substantially faster than population.11 Second, the country ranking differs quite markedly from a priori expectations. The most striking result is the boom in Egypt, which, however, as warned by Hansen and Whattleworth (1978, p. 458), seems too good to be true. At the other end of the range, the fall in production per capita in England is also striking. It contrasts not only with the country’s reputation as a beacon for technical progress (Deane, 1967; Overton, 1996), but also with the likely increase in consumption per capita during the Industrial revolution, when imports of agricultural products were negligible. There is no easy solution to this “food puzzle” (Clark, Huberman & Lindert, 1995) but the fact that production growth was not impressive seems now well-established (Allen, 1994; Clark, 2002). As expected, production grew very fast in the countries of Western Settlement (a 3% increase over 70 years corresponds to an eight-fold growth). However, the achievement is less impressive than it might seem: the increase barely exceeded population growth, both in Australia and in the United States.12 In contrast, according to these estimates, European performance was surprisingly good. Production per capita increased in all countries, except Austria and Portugal, and, in some cases, quite fast – up to 0.7% per year. Scattered evidence points to an increase in output also in other countries, such as Austria before 1830, Hungary, and Russia.13 However, the relative prices of agricultural products rose quite substantially, especially during the “hungry Forties,” and heights, which, ceteris paribus depend on food consumption, were falling or stagnant in the first half of the century in the United States and in several European countries.14 These facts cast some doubt on the reliability of the figures in Table 1, which should be considered an upper bound on the true rate of growth. The world outside the “Atlantic economy” (with the exception of Java) is, statistically speaking, terra incognita. Maddison opines that, in Togukawa Japan, agricultural production grew faster than the population – i.e. by 20% from 1820 to 1870.15 In China, production may have grown slightly less than population, which rose from about 340 million in 1800, to 410 in 1840, to plunge to 360 million in 1870 because of the Tai’ping rebellion.16 The total population of the Third World countries, including China, increased at about 0.3–0.4% yearly in the first half of the 19th century – i.e. by a quarter or by a third (the data are extremely uncertain).17 If production had been stagnant, consumption per capita would have fallen by the same amount. Such a fall is unlikely. Caloric consumption at the beginning of the century was quite low – perhaps less than 2000 calories per day per capita in Asian countries, such as Japan and Java (Van Zanden, 2003). Furthermore, in most countries, land was still quite abundant, and thus there was ample scope for

The Growth of World Agricultural Production, 1800–1938

131

production growth even without technical progress. In other words, the best, or least bad, guess, suggests that agricultural production in the LDCs must have risen, possibly as much as their population. As said previously, production per capita in “advanced” countries was rising. Thus, one can, very tentatively, conclude that, in the first seventy years of the 19th century, world output per capita did not fall and may have increased.

4. LONG-TERM GROWTH AND POLITICAL SHOCKS, 1870–1938 The yearly series confirm the conventional wisdom about long-term growth.18 From 1870 to 1938, “world” gross output increased by 2.5 times (1.31% yearly) and “world” GDP by 2.2 times, at 1.18% per annum (Fig. 1). As expected, the growth was faster before 1913 than afterwards, and there is some (weak) evidence of a slowdown during the so-called Great Depression.19 The data also confirm the received wisdom about the effects of modernization of agriculture. Purchases outside the sector absorbed 8.5% of total GSP in the 1870s, 11% on the eve of World War I and, after a fall caused by the war itself, more than 15% in the late 1930s. Most of these sums were spent to purchase fertilizers, as the use of tractors and other machinery was to spread massively only

Fig. 1. Agricultural Production.

132

GIOVANNI FEDERICO

Table 2. Growth in Agricultural Production, by Area and Period. Gross Output

Value Added

1870–1938 1870–1913 1913–1938 1870–1938 1870–1913 1913–1938 Europe North Western Europe Southern Europe Eastern Europe Asia South America Western Settlement World

1.19 0.97 0.88 1.67 0.97 3.80 1.37 1.31

1.36 1.02 0.97 2.13 1.11 4.43 2.20 1.56

0.76a 1.50 0.96 0.36a 0.58 3.05 0.74 0.67

1.05 0.74 0.84 1.61 0.96 3.89 1.22 1.18

1.30 0.90 0.96 2.09 1.18 4.86 1.92 1.48

−0.12a 1.41 0.73 0.16a 0.56 3.07 0.62 0.38a

Source: Statistical Appendix Table A.1. a Not significantly different from zero.

after World War II (Federico, forthcoming). Thus, this statistical reconstruction by and large buttresses the conventional wisdom. However, there are also substantial divergences in long-term trends by country/area performance (Table 2) and in short-term changes during the interwar period (Figs 2 and 3). Before 1913, the growth in agricultural output was slower than expected in the countries of Western Settlement (with the remarkable exception of Argentina) and faster in Eastern Europe. Agricultural production in the rest of Europe and in

Fig. 2. Agricultural Output, by Continent.

The Growth of World Agricultural Production, 1800–1938

133

Fig. 3. Agricultural Output, Europe.

Asia grew as well, even though less than in the countries of Western Settlement or in Russia. However, performance widely differed between countries in the same area (Statistical Appendix Table D.2). India dragged down the Asian aggregate rate in spite of the high growth in Indonesia and Japan. In Northwestern Europe, the good performance of Germany and Denmark contrasts with the lackluster growth in France, the Netherlands and Belgium, and the stagnation in the United Kingdom. Greece outshone the two other Mediterranean countries, with a growth rate that was twice that of Italy and 4.5 times that of Spain. These differences reflect different combinations of growth in inputs (extensive growth) and in their productivity (intensive growth). At one end of the range, Argentina was the prototype of extensive growth, featuring an exceedingly fast population growth, an almost infinite supply of land and, at least in the 1900s, declining productivity.20 In some European countries, such as France, Ireland, and the United Kingdom, Total Factor Productivity grew more than output, and the quantity of inputs (especially labor) declined.21 All other countries fall somewhere between these extremes. For instance, in the United States, from 1870 to 1900 inputs roughly doubled, while output increased by 135%: Total Factor Productivity thus accounted for about a fifth of production growth (Craig-Weiss, 2000). The period to 1913 not only shows a growth in production, but also quite favorable price trends. At the very least, the real prices of agricultural products remained constant or rose, as in the United States, while the terms of trade (relative to manufacturers only) increased in almost all countries. As expected,

134

GIOVANNI FEDERICO

there is some evidence of price convergence between the land-abundant New World and the land-scarce Old World, but it is quite weak. In fact, the range of country cases is quite wide. However, this combination of growing production and (probably) rising prices singles out the period to 1913 as a golden age for agriculture, at least in the Atlantic economy. The outbreak of the war changed the situation. As already said, it is impossible to calculate the “world” indices during war-time years, but it is possible to compute series for some areas (Table 3), and there are independent estimates of production (especially of cereals) for almost all the missing countries. Assuming that these estimates are reliable enough, and that cereal output is a good proxy for the whole of agricultural production, it is possible to estimate that the “world” gross output in 1915–1918 was about 8% lower than in 1913.22 This overall decline is the outcome of widely different country trends. Asia was relatively unaffected by war, and, in fact, in 1915–1918, its production continued to rise exactly at the pre-war rate. Production stagnated in neutral European countries and in overseas countries. The increase in freights and the embargo on Germany disrupted their traditional exports flows, even though cereals were no longer subject to Russian and Romanian competition after the closure of the Dardanelles. In all the belligerent European countries production fell. The mobilization drained men and horses from the fields and the conversion of chemical plants to the production of explosives drastically curtailed the supply of fertilizers. This shortage may account for the poorer performance in “modern” countries, such as France or Germany, as compared with Italy or Russia. Table 3. Gross Output 1915–1918 (1913 = 100). Indices Asia Southern America Western Settlement European Neutral countriesa Italy Austria

106.6 96.4 102.8 99.6 87.6 65.4b

Other Sources United Kingdom France Germany Russia Hungary

(a)

(b)

(c)

114.5 68.1 67.3

96.8 66.8 67.5 79.0

99.2 80.5 62.2 74.9 79.8

(e)

81,1b

Sources: Indices: Statistical Appendix Table D.1; (a) League of Nations (1943) (cereals and potatoes); (b) Dessirer (1928) (cereals); (c) United Kingdom: estimate of the author (The figure is obtained by weighting the 1915–1918 average gross output of cereals, potatoes, milk and meat (Mitchell, 1988) with the shares of these products on gross output in 1911–1913 (Ojala, 1958, pp. 208–209), France Hautcoeur (forthcoming); Germany Holtfrerich (1986, Table 33) (cereals); Russia: Adamets (1997, Table 2) (cereals) and Hungary: Schultze (forthcoming) (cereals); (e) Harrison-Gattrell (1993, Table 12). a Denmark, Greece, the Netherlands, Portugal, Spain, Sweden, Switzerland. b 1915–1917 only.

The Growth of World Agricultural Production, 1800–1938

135

The post-war recovery was decidedly slow. In 1920–1922, “world” output was still about 8–9% below the pre-war level.23 Actually, production exceeded the 1913 level in the majority of countries, including the United States, but “world” recovery was hampered by failure in three major countries, Austria-Hungary, Germany, and Russia, which accounted for about a quarter of “world” output in 1913. In the former Central Empires, production stagnated around its war-time level, while in Russia, where the civil war was raging, it collapsed to (perhaps) half the pre-war level in 1920–1921. As late as 1927–1929, “world” production was only 10% higher than in 1913, and European production was only 5% higher. Thus, looking at aggregate production figures, there is little evidence of the alleged overproduction in the 1920s. In-fact, the growth in “world” production barely matched the increase in population (from 1913 to 1930, by 11% in the world, and by 13% in the 25 countries). Nor did trends in prices confirm the conventional wisdom. Indeed, prices fell in the early 1920s, but, in most countries, they returned quite quickly to their pre-war peaks (and, in a handful of countries, terms of trade actually exceeded the 1913 level). During the Great Depression, prices fell drastically (by 25–30% in most countries), while production remained constant. The three-year moving averages (a rough measure to smooth the effect of crop fluctuations) only decreased in 1931, by less than 1%, which was exclusively because of the collectivization disaster in the Soviet Union.24 On the eve of World War II, “world” production was 3–5% higher than in 1927–1929. Gross output grew even more (by 8–9%) according to the estimates of the League of Nations.25 The combined effect of World War I, the Great Crisis and collectivization in the Soviet Union account for the difference in growth rates before and after the war. In the inter-war years, the growth rate of agricultural production matched or exceeded the pre-war rate only in Northwestern and Southern Europe. Elsewhere, it fell drastically, plummeting to zero in Eastern Europe. The slowdown can be measured by computing the level which production would have attained had it gone on growing as quickly as it had done in 1870–1913 (Table 4).

Table 4. Counterfactual Production Estimates in Interwar Years (Actual Production = 100). World GSP GDP, by Northwestern Southern Eastern Asia South Regions of GDP Area Europe Europe Europe Europe America Western Settlement 1920 1929 1938

130 112 125

127 109 121

145 114 124

133 110 108

108 98 117

198 132 158

109 103 115

128 143 242

125 124 140

136

GIOVANNI FEDERICO

The 1920 “counterfactual” production would have been 30% higher in the “world”, and almost two times higher in Eastern Europe. The recovery of the 1920s was “sufficient” to return to the steady state growth path only in Asia and Southern Europe, while the gap between actual and potential output was still about 10% for “world” production (and 30% for Eastern Europe). It widened again as a consequence of the stagnation during the Great Crisis. In no area was the 1938 “counterfactual” output close to the actual one.26 Clearly, the “counterfactual” output is a purely statistical artifact. Even without wars, the pre-1913 growth rate could not have been sustained. The supply of new land to be settled was dwindling in most Western Settlement countries and the workforce started to fall in all “advanced” countries. In fact, the growth rate of Total Factor Productivity and its contribution to output growth were decidedly higher after World War I than before it. It is impossible to know whether technical progress could have been faster, even without the adverse shocks of wars and economic crisis.

5. CAVEATS: SHALL WE BELIEVE THESE NUMBERS? The reconstruction of historical national accounts is not an exact science. Its results are always uncertain and, at times, are positively controversial. In the 1960s, Nakamura argued that the data available then grossly overestimated the growth of Japanese agricultural production before 1913. After a very lively controversy, his views were accepted and the quasi-official series were revised downwards, although less than he had advocated.27 In other cases, such as the Soviet Union, the issue is still open. The official production figures have been revised many times, and most Western scholars suspect that they have been “cooked” to extol the successes of Stalinist planning.28 Consequently, they have suggested alternative estimates: Fig. 4 reproduces two series by Wheatcroft and Allen and compares them to the Soviet figures in their latest version.29 According to the official data, gross output exceeded the pre-war peak already in 1924 and never fell below it afterwards. According to Wheatcroft, production barely recovered the pre-war level in 1929, before plunging to three quarters of the 1913 level during the collectivization crisis. The series by Allen, which has been used to compute the overall index, is midway between these two extremes. Table 5 compares the base-line estimates (those used to compute the index) with all the alternative ones that the author is aware of. In about half the cases, the difference is so small as to be negligible, while, in the others, the alternative series grows faster than the base-line one. India is

The Growth of World Agricultural Production, 1800–1938

137

Fig. 4. Alternative Estimates of Soviet Gross Output (1913 = 100).

arguably the most important case, because of the size of the difference and the importance of the country, the second largest among the twenty-five (Table 6). According to the official statistics, in the first half of the 20th century, yields of main food-crops fell, acreage grew slowly, and per capita consumption declined. Table 5. Alternative Estimates of Production Growth by Country. Country Argentina Austria Canada France India Italy Netherlands Sweden

Period 1900–1938 1871–1913 1971–1927 1820–1913 1900–1938 1870–1913 1851–1913 1861–1931

Base 3.15 1.44 2.77 0.72 0.45 1.14 0.60 1.07

Alternative (a)

Alternative (b)

2.94a 1.39a 2.74a 0.93** 0.90** 0.85* 0.90** 1.25a

0.77**

Sources: “Base” series: Appendix B; “alternative” Austria: Kausel (1979, Table 1a), Canada: McInnis (1986, Table 14 A.2), France: Levy-Leboyer (1968); Netherlands: Knibbe (1994); India (a) Heston (1984) (b) Maddison (1985, Table 4); Sweden: Lindhal (1937, Table 1); Italy: Ercolani (1969, Table XIII.1.1.4). a Not significant; different from the “base.” ∗ Significantly different from the “base” series at 10%. ∗∗ Significantly different from the “base” series at 1%.

138

GIOVANNI FEDERICO

Table 6. Shares in “World” Agricultural Production. (a) Argentina Australia Austria Hungary Belgium Canada Chile Denmark Finland France Germany Greece India Indonesia Japan Italy Netherlands Portugal Russia Spain Sweden Switzerland U.K. USA Uruguay Correlation

2.3 2.2 3.1 3.7 0.6 1.9 0.3 0.6 0.3 9.2 9.3 0.3 15.1 1.9 2.9 5.8 0.5 0.5 12.9 2.4 0.6 0.6 2.4 20.6 0.2

(b)

(c)

(d)

(e)

(f)

2.3 2.4 2.9 3.5 0.8 1.9 0.3 2.0 0.3 9.1 8.9 0.3 13.8 1.7 2.9 5.5 0.9 0.4 11.9 2.3 0.7 0.5 3.1 21.5 0.2

2.0 2.2 3.1 3.7 0.6 1.7 0.2 0.6 0.3 9.3 9.1 0.3 16.6 1.9 2.9 5.6 0.6 0.6 11.5 2.5 0.7 0.6 2.5 20.7 0.2

2.4 2.4 2.6 3.2 0.7 2.0 0.3 0.7 0.3 7.9 8.0 0.3 16.2 2.0 3.1 5.0 0.6 0.5 13.9 2.1 0.6 0.5 2.6 22.1 0.2

3.3 2.2 3.0 3.7 0.5 2.2 0.4 0.4 0.3 6.4 12.6 0.2 14.5 3.0 2.3 4.3 1.5 0.3 14.3 2.0 0.9 0.5 3.7 16.9 0.5

2.3 1.6 3.7 5.3 0.6 1.5

0.995

0.997

0.995

0.969

0.6 0.4 8.7 10.4 0.5

6.6 7.2 0.7 0.7 26.9 3.0 0.7 2.3 16.4 0.865

Sources: See text.

This fall is controversial. Sivasubramonian (2000), in his base-line estimate, endorses the official production statistics, while other scholars deem a decline in consumption implausible. Heston, in his own estimate of Indian GDP (alternative a), revises the production data under the assumption that yields had remained constant from the beginning of the century to the early 1950s.30 The two series thus imply quite different assessments of the performance of Indian agriculture, with far-reaching implications for the economic history of the country during the last period of British domination. But the choice of one of them would not substantially affect the analysis of “world” and area trends. Substituting the Sivasubramonian series for Heston’s in 1900–1938 would increase the Asian growth rate from 0.74 to 0.94% per year (causing production in 1938 to be 8%

The Growth of World Agricultural Production, 1800–1938

139

higher) and the “world” rate by 0.02 points. Errors in country series must be huge to affect the “world” index. For instance, a 100% mistake in the American series leads to only 0.2 mistake in the “world” series in 1870–1913, and the error would be proportionally greater for area series, but the “world” indices could be seriously biased only if several country series were in error, and all in the same direction. This coincidence cannot be ruled out, but it seems quite implausible. Mistakes in the weighting procedure are potentially more serious than those in the country series. A wrong set of country shares might bias the index upwards (downward) if fast-growing countries are given a too high (low) weight. This can happen either because 1913 production in those countries was unusually high (low) or because 1913 market exchange rates overvalued (undervalued) the real purchasing power of the country’s currency. Although agricultural products are highly tradable, duties, quotas, and other trade barriers hampered trade. O’Brien and Prados estimate that, in 1911, the market exchange rate overvalued the “agricultural” Italian lira by 16% and the German mark by 10%.31 The effect of these potential biases can be explored by computing the “world” indices with different weights (Table 6). The two first columns on the left reproduce the “basic” country shares (column a for “world” value added and column b for gross output). Column c takes the short-term fluctuations into account by replacing gross output in 1913 with an estimate for 1909–1913.32 The three other columns use different methods for converting the 1913 output into a common monetary unit. The shares in column d are computed by simply reducing the value of the output of the “protectionist” countries (Austria-Hungary, Italy, France, Germany, Spain, Portugal and Sweden) by a fifth. Column e uses the author’s estimate of the agricultural gross output for some 50 countries in 1913, which uses a standard set of international prices.33 Column f is calculated with the exchange rate implicit in Prados’s recent estimates of national income in purchasing power parity in 1913.34 As shown in the bottom row, in three cases out of four, the coefficients of correlation between the basic set of weights (column a) and the alternative ones are extremely high and thus the long-run growth rates are almost identical.35 The last set of weights (column f) differs from the basic ones: as expected, the value of output is higher in “underdeveloped” countries, such as Russia. However, the long-term growth rate of “world” output comes out to be very close to the basic one (1.28%, instead of 1.33% for the same countries) and also the short term differences are relatively small (cf. Fig. 5). In short, this section shows that one can trust the overall reliability of the “world” (and area) indices in spite of errors in some country series and possibly in the weighting procedure.

140

GIOVANNI FEDERICO

Fig. 5. Indexes of Output, with Alternative Weighting Schemes.

6. EXTENSIONS: THE “OTHER” COUNTRIES What happened in the rest of the world? Did agricultural production increase as much as in the twenty five “core” countries? Table 7 provides a partial answer. It reports the evidence on the growth of agricultural production in a dozen other countries, which have been omitted from the base series, because they do not cover the whole period 1870–1938 and/or refer only to benchmark years. By and large, these additional data confirm the previous results: production increased in the long-run in almost all countries, and it grew faster before rather than after World War I. Unfortunately, none of these countries was really important from a worldwide perspective. Their cumulated gross output in 1913 was about 6–7% of the “world” total.36 It would be much more important to know something about China, which in 1913 accounted for a quarter of world population and produced about 20% more than the United States. Indeed, there are several estimates, but, unfortunately, there is no consensus.37 Perkins, in his classic book on Chinese agriculture, surmises that agricultural output increased more or less as much as the population from 1850 to 1957 (i.e. at about 0.5% per year). Feuerwerker, in his authoritative survey of Chinese economic history, endorses Perkins’ view, which is deemed too optimistic by Chao, who implicitly suggests a growth of around 0.4% from 1882 to 1950.

The Growth of World Agricultural Production, 1800–1938

141

Table 7. Rate of Growth in Agricultural Production, “Other” Countries. 1870–1913 Bulgaria Montenegro Serbia Egypt (a) Egypt (b) Palestine Taiwan Korea Philippines Thailand Burma Mexico (a) Mexico (b) Brazil South Africa New Zealand

1.14 2.12 1.18 2.19 2.23 −0.91 7.7 1.32 0.14 2.92 3.35 2.31 3.94

1913–1938

0.94 1.15 7.39 2.85 2.76 1.11 2.20 −0.16 −0.27 2.02 3.15 2.55 1.61

Sources: Bulgaria (1865–1873 to 1911–1914), Montenegro (1873 to 1911–1912) and Serbia (1873–1875 to 1911–1912): Palairet (1997, Tables 7.1, 8.2 and 10.2) (total output); Egypt: (a) (1872–1878 to 1910–1914 and 1910–1914 to 1935–1939) O’Brien (1968, Table 10) (gross output for eight major crops), (b) (1887 to 1911–1913 and 1911–1913 to 1936–1938): Hansen-Whattleworth (1978) (production); Palestine (1921–1923 to 1936–1939): Metzler (1998, Table A.11) (gross output); Taiwan: (1887 to 1911–1913 and 1911–1913 to 1936–1938) and Korea (1911–1913 to 1936–1939): Mizoguchi-Umemura (1988, Tables 5 and 7) (NDP at factor costs), Philippines: (1902–1918 and 1918–1938): Crisostomo-Barker (1979, Table 5.1); Thailand (1870–1913 and 1913–1938): Manarungsan (1989, Table c.3) (GDP at market prices); Burma (1901–1902 to 1911–1912 and 1911–1912 to 1938–1939): Saito-Kin (1999, Table IX-2) (NDP at factor costs); Mexico (a) (1900–1902 to 1911–1913) Carr (1973, Table 1) (“total output”), (b) (1900–1910 and 1910–1940): Reynolds (1970, Table 3.2) (“production”); Brazil (1901–1911 and 1911–1941): Merrick-Graham (1979, Table II.3); South Africa (1911–1913 to 1936–1938): Union of South Africa (1960, Table I-27) (“physical output”); New Zealand (1900–1910; 1910 to 1936–1938): Bloomfield (1984) (gross output Table v.3 deflated with wholesale prices IX.13 and IX.14).

Rawski disagrees. He argues that labor productivity must have grown as much as real wages. If this were the case, agricultural output must have grown much faster than Perkins assumed – by 1.4–1.7% per year, from 1914/18 to the early 1930s. Rawski’s argument has not convinced prominent Western scholars, such as Wiens and A. Maddison, who, in his latest book, reinstates Perkins’ view. Output grew slightly slower than population from 1890 to 1913, and slightly faster from 1913 to 1933. On the other hand, some years before, the Chinese scholar Wang Yu-ru, apparently oblivious to the Western debate, had put forward a figure (a growth

142

GIOVANNI FEDERICO

rate of 1.2% from 1887 to 1928) which is only marginally lower than Rawski’s “preferred” estimate. The end of the debate is not in sight, but there is no doubt that total production grew substantially, as the population increased from about 360 million in 1870 to about 500 in 1933 – i.e. by 40% (Maddison, 1998, Table D.1). As far as the author knows, there are no data, even tentative ones, on agricultural production in all the other countries, including large areas of Asia and almost the whole of Africa.38 Trends in agricultural production can be inferred from the available, very tentative, estimates of change in GDP per capita. Reynolds (1985) argued that, by 1870, “intensive growth” (i.e. the increase in GDP per capita) had already started or was about to start all over the world. His statement is buttressed by some recent guesstimates by Maddison. He surmises that, from 1870 to 1950, the average GDP per capita in the “rest of the world” (including China) grew by a half.39 Such an increase must have augmented the demand for food, which had to be satisfied by local production, as imports from the twenty-five “core” countries were very small or negligible. A (conservative) back-of-the-envelope estimate suggests that per capita production of foodstuffs may have risen by a quarter.40 On top of this, exports of agricultural products from most Third World countries grew quite substantially. Thus, if Maddison is right, per capita agricultural production in the “rest of the world” must have grown by at least by 25% from 1870 to 1938.

7. EXTENSIONS: AN ESTIMATE OF TOTAL WORLD OUTPUT The rate of change in total world output can be estimated as an average of the growth rates for the “core” twenty-five countries and for the “rest of the world,” weighted with their respective share of output in 1913. Unfortunately, the latter are not available. One can proxy them with the proportion of output in 1970, or with the share of acreage (arable and tree-crops) in the late 1940s, or with the percentage of the population in 1913. The “rest of the world” accounted for about a third, two fifths and 45% of the total respectively.41 Clearly, none of these figures is an exact proxy for their share of gross output, and it is difficult to assess a priori whether they underestimate or overestimate the actual share. Thus, Table 8 assumes that the “rest of the world” accounted for 45% (column a) or 35% (column b) of world gross output. It also assumes (conservatively) that its production per capita remained constant.42 Needless to say, the estimate is highly tentative. However, it confirms that the growth in total production was substantial, and that it was decidedly faster before 1913 than after. The growth in production per capita was not spectacular, nor was it negligible, either, especially in the period before the war. Furthermore, if

The Growth of World Agricultural Production, 1800–1938

143

Table 8. Growth in World Gross Output. 25 Countries

Rest of The World

Total Gross Output (a)

(b)

Total 1870–1913 1913–1938 1870–1938

1.54 0.71 1.24

0.58 0.73 0.64

1.06 0.72 0.94

1.17 0.72 1.01

Per capita 1870–1913 1913–1938 1870–1938

0.55 −0.08 0.32

0.00 0.00 0.00

0.26 −0.05 0.15

0.38 −0.05 0.22

Source: Statistical Appendix Table D.1.

Reynolds and Maddison are right, the estimate of Table 8 should be considered as a lower bound, with an upper bound around 0.20 –0.30% per year. If this latter figure were true, there would be very little difference between the performance before and after World War II. Even in the lower, more conservative, version, the period would mark a clear discontinuity from the previous historical experience. Maddison surmises that world GDP per capita (and thus also agricultural output) grew at about 0.05% per year from 1000 to 1820 – i.e. by a half.43 This estimate seems too optimistic. In fact, according to Allen (2000, Table 7) agricultural production per capita decreased in all the major European countries from 1400 to 1800. It is unlikely that it had increased in Europe before 1400, or in the rest of the world, sufficiently to compensate for this loss and to achieve the long-run growth rate suggested by Maddison. It seems more likely that agricultural production per capita had remained roughly constant in pre-industrial times, albeit with wide fluctuations.

8. EXTENSION: THE CHANGES IN COMPOSITION It is likely that the demand for agricultural products changed in the long run for at least two reasons. First, industrialization must have increased the demand for raw materials, and thus their share of total agricultural production, because artificial substitutes were not available before the 1920s (and their production boomed only after World War II). Second, the rise in income per capita must have increased the demand, and thus the share, of high income-elastic goods. However, the definition of the latter varied a lot by area: meat and dairy products were “luxury” goods in

144

GIOVANNI FEDERICO

Table 9. Share of Raw Materials on Total Gross Output.

Australia Belgium USA France Italy Russia Japan U.K. Spain

1800

Ca 1850

6.1

14.5 15.8 10.1

Ca1880

1910

Ca 1938

58.6 22.4 14.0 11.6 10.1 12.0 9.8 7.8 2.3

53.7 28.3 16.6 7.5 10.5 9.6 8.9 6.5 3.3

47.8 14.4 7.4 8.5 10.9 3.9 3.7

Sources: Australia (“pastoral” 1879–1881, 1911–1913 and 1936–1938): Butlin (1962); Belgium: Blomme (1993, Table 1); France (textile materials, tobacco and timber in 1845–1854, 1875–1884, 1905–1914 and 1935–1938): Toutain (1961, Tables 76, 76 bis and 77); Italy (1891, 1911 and 1938): Federico (2000); Russia (1879–1881 and 1911–1913, “industrial crops”): author’s estimate (cf. Appendix B); Japan (cocoons, 1879–1881, 1911–1913 and 1936–1938): Okhawa-Shinohara (1979, Table A16); Spain (raw materials, circa 1890, 1909–1913 and 1929–1933): Prados (1993, Table 1); United Kingdom (1879–1881, 1911–1913): Afton-Turner (2000, Table 38.8) and (1935–1939): Ojala (1952, pp. 208–209); United States (textile raw materials and tobacco) 1800 and 1850: Towne-Rasmussen (1960, Table 6), 1879–1881, 1911–1913 and 1935–1937: Strauss-Bean (1940, Tables 10 and 27).

Asia and Southern Europe, while they were almost the staple diet in North-Western Europe, where the real luxuries were fruit and vegetables. Unfortunately, testing these hypotheses is very difficult. Only a few sources provide data by product, even if they estimate total production. Table 9 shows the available data on the share of raw materials. These data are not accurate. The Australian data refer to “pastoral” production, inclusive of mutton, and thus overvalue the share of raw materials. Other country data omit some products (notably wood from tree crops), and thus undervalue the share, even if the bias is not likely to exceed a few percentage points. In spite of these biases, the story is clear: the share of raw materials was low in all countries except Australia and, contrary to expectations, it did not increase over time – either decreasing (as in France or the U.K.) or fluctuating without a clear trend (as in the U.S.). In most countries, one or two goods (wool in Australia and the U.K., cotton in the U.S., cocoons in Japan and Italy) accounted for most of the aggregate “raw materials.” The output of these “core” products was deeply affected by the state of the world market, especially by competition from other countries, which was almost never fettered by protection. For instance, the production of British wool remained constant (and thus fell as a share of total output) because of Australian competition.

The Growth of World Agricultural Production, 1800–1938

145

Table 10. Share of Livestock Products in Gross Output.

1870–1872 1889–1891 1911–1913 1920–1922 1936–1938

(a)

(b)

(c)

(d)

(e)

38.3 41.6 43.4 44.1 44.7

54.5 51.7 48.7 49.2 49.8

32.6 36.6 40.0 40.9 41.2

38.3 40.1 42.1 41.9 43.4

37.3 40.9 44.2 43.2 45.0

Note: (a) Share of livestock products in total gross output; (b) Share of livestock products in the gross output of land abundant countries (Australia, Argentina, Canada, Russia, Uruguay and USA); (c) Share of livestock products in the gross output of other countries; (d) Counterfactual estimate assuming constant share of livestock by country at its 1870–1872 level; (e) Share of land-abundant countries in total “world” gross output of livestock products.

Unfortunately, the data are too scarce to draw any meaningful inference on world trends. It is possible to be somewhat more precise about the distribution of gross output between crops and livestock products (Table 10).44 As column a shows, the share of livestock products in gross output of the twenty-five “core” countries grew substantially, especially before World War I. The share of these countries in world totals was rising (Table 8), and livestock products accounted for a lower share in the “rest of the world” than in the “core” countries. In 1913, they accounted for about a quarter of gross output in a group of twenty-five other countries, including China, Mexico and Turkey (Appendix Table A.6). Extending (somewhat arbitrarily) this figure to the whole “rest of the world” for all years, it is possible to estimate that the share of livestock products in world gross output grew from about 30% in 1870 to about 35% in 1913, and remained almost stable thereafter. Relative prices of livestock products increased substantially before 1913 and remained roughly constant in interwar years, albeit with substantial fluctuations.45 A contemporary increase in prices and production strongly suggests a growing demand, not matched by an increase in (relative) productivity. How was the growing demand for livestock products satisfied? Traditional livestock-raising was quite a land-intensive activity, and thus one would expect that it accounted for a greater share in land-abundant countries (column b) than in the others (column c). Indeed, this was the case at the beginning of the period: in 1870–1872, livestock products accounted for 96% of Argentinian gross output and for a mere 17% of Indian output. Since then, their share declined in all land-abundant countries except the United States, and rose in 15 out of the 19 land-scarce countries (the main exception being Indonesia).

146

GIOVANNI FEDERICO

This convergence is by no means surprising, given the underlying change in factor endowment. However, this change in the country composition of output only accounts for a fifth of the increase in the “world” share of livestock products, as shown by a comparison of columns d and a. The rest is accounted for by the growth in the share of land abundant countries on the “world” output of livestock products (column e). The population and incomes in these countries was growing faster than in the rest of the “world” and these countries also supplied increasing quantities of livestock products to (land-scarce) Europe.

9. CONCLUSIONS The results of this paper can be summed up in five statements:
agricultural output increased from the beginning of the nineteenth century, and the growth accelerated over the century, peaking on the eve of World War I. It was a veritable “golden age” for world agriculture, as relative prices were rising or constant.
the War and the Great Crisis hit agriculture quite hard, and growth in the interwar years never reached the pre-war pace. However, prices did not rise, even if they did not fall as catastrophically as has sometimes been argued.
The growth affected all areas, even if rates of increase were decidedly greater in the countries of Western Settlement and in Eastern Europe than in Asia and Western Europe.
in the long run, the increase in output exceeded that of population by a substantial margin especially in the Atlantic economy – but probably throughout the world.
the production of livestock products increased more than the total, probably as a result of changes from the demand side. These results answer, at least to some extent, the questions raised at the beginning of this paper. But there is much work to be done. The main priority is to add further countries to the sample, and to extend the existing series back in time. Even imprecise estimates are better than total ignorance. It would also be useful to revise several country estimates, even if, as argued in section 5, none of them would affect the world total that much. In fact, accurate country series are essential in assessing country performance. Last but surely not least, all this statistical ground-work is only preliminary for tackling the real big issues: how was this growth achieved? What was the contribution of productivity growth and technical progress? How much did agricultural performance foster or hamper modern economic growth?

The Growth of World Agricultural Production, 1800–1938

147

NOTES 1. Population from Maddison (2001), calories from FAO (http://www.fao.org). 2. Fogel (1997, p. 450). The long-run growth in caloric availability is shown also by the rise in heights. 3. The first figure is estimated from FAO, Yearbook, various years. It excludes the Communist countries, and thus may overvalue actual growth. The data for 1961–2000 are taken from the FAO website (http://www.fao.org). 4. The role of agricultural crisis was first highlighted by Arndt (1963, p. 10). Cf. for instance Feinstein et al. (1997, pp. 78–80) or James (2001, pp. 112–113). 5. Price trends will be dealt with succinctly, on the basis of the discussion in Federico, forthcoming, ch. 3.3 6. In the following, the word “world” is written between brackets when it refers to the 25 countries covered in the index and without brackets when it refers to all countries. 7. Cf., Rao (1993, pp. 12–14). In the following, the words “output” and “gross output” will be used for GDP and GSP respectively, while “production” refers to both. 8. For a detailed description of the data, sources, and methods, see Appendix B. The missing (and interpolated) years are 1870–1873 for Japan, 1870–1874 for Argentina, 1870–1879 for Belgium and Indonesia, 1870–1871 and 1873–1881 and 1883 for India, 1920–1924 for Germany and the Soviet Union. When necessary, gross output (value added) is estimated starting from value added (gross output) with information provided by the source itself or with VA/GSP ratios for similar countries. Some series adopt slightly different concepts (e.g. the net instead of gross domestic products), and these differences are taken into account whenever possible. Boundaries are adjusted to those existing in 1913 with data on output or, when the latter are not available, on agricultural acreage. In this case, it is implicitly assumed that the production per acre was similar throughout the whole country. 9. The omission of forestry, fishing, and hunting reduces the bias in the series for countries of Western Settlement arising from the omission of the output by native population. Their contribution to agriculture was minimal, while they accounted for a sizeable, even if fast shrinking, share of the total primary output in the USA (Mancall-Weiss, 1999) and Australia (Butlin-Sinclair, 1986) in the 18th and early 19th century. 10. Exchange rates from League of Nations 1913–1925. The effect of alternative methods of conversion (wheat units and PPP-adjusted exchange rates etc.) is explored in section five. 11. The extent of the fall in Portuguese production depends a lot on the starting point. Omitting 1848 (an exceptionally good year) the rate of decline would halve to –0.36% per year. 12. At least for the United States, the coincidence is not entirely casual: before 1840 the output of most goods is calculated by assuming constant per capita consumption at the 1840 level, and adding net exports (Towne & Rasmussen, 1960, p. 264). 13. For Austria, Good (1984, Tables 11 and 22) reports growth rates for crops (1789–1841) of 1% per year and for livestock (1818–1850) of 0.6% per year. Komlos (1983, pp. 52–89) argues that in Hungary, production grew in the whole period from the 1830s to the 1860s (with no noticeable effect of the emancipation of serfs in 1848), and that the output of grain rose faster than the population. According to Khromov (quoted by Mitchell (1998c, p. 315), the output of grain in European Russia increased by 40% between 1800–1813 and 1857–1861. Cf., also, on Spain in the first half of the 19th century,

148

GIOVANNI FEDERICO

the debate between Prados de la Escosura (1989) and Simpson (1989a, b), who suggests a 0.65% yearly growth for the whole century. 14. Cf., on prices, the analysis in Federico, forthcoming, chap. 3.3; for the fall in heights (or “early industrialization puzzle”) Steckel (1995, 1998), Komlos (1998), Steckel and Floud (1997), Baten (2000). 15. Maddison (2001, Table B-17) and also Yamamura-Hanley (1977, pp. 70–74). 16. Richardson (1999, p. 20) and population data from Maddison (1998, Table D-1). 17. From 1800 to 1850, the population of Asia, Africa, and South America rose from 750 to 925 million people according to Biraben (1979), or from 700 to 880 million according to McEvedy-Jones (1978) – corresponding to growth rates of 0.42% and 0. 46% respectively. According to Maddison (2001, Table B-10), from 1820 to 1870, the population of the overseas LDCs increased from 805 to 895 millions – i.e. at 0.21% yearly only (for the consequences of the Chinese disaster). In the same years, the population of Eastern Europe increased from 95 to 145 millions (0.85% yearly). Needless to say, all these figures are highly tentative and give us only a rough order of magnitude. 18. Statistical Appendix Table 1. Unless otherwise specified, the growth rates are calculated with a linear regression (adjusted to take into account the autocorrelation of residuals if necessary). 19. A dummy for 1879–1896 is negative and significant in the time trend regressions for the whole world, North-Western and Southern Europe, while it is not significant in Eastern Europe, South America and countries of Western settlement. 20. From 1900–1904 to 1910–1914 the agricultural workforce increased by 40%, land by almost 50% and Total Factor Productivity fell by almost 20% (Diaz Alejandro, 1970, Table C.3.2). The total population of the country soared from 1.8 million in 1870 to 7.6 in 1913 (Mitchell, 1998b). 21. Cf., for France, Grantham (1996, Tables 5 and 6), for Ireland O’Grada (1993, Table 30), and for the United Kingdom, Turner (2000, Table 3.33). Cf., for further cases and a more detailed analysis, Federico, forthcoming. 22. It is assumed that the gross output was three quarters of the 1913 level in Finland and two thirds in Belgium. Production of meat and livestock products may have fallen more than cereal output and animal stock (League of Nations, 1943). 23. This slow recovery contrasts with the experience after World War II. In 1948–1952, output exceeded pre-war levels by 7% in Europe, 41% in North America, 11% in Oceania, 26% in Latin America, 5% in the “Far East” (i.e. Asia) and by 20% in the “world”, which includes Africa and the Near East, but not the Socialist countries. Factoring them in would probably reduce the overall increase. In fact, according to Davies (1998, pp. 64–69), the Soviet production returned to pre-war levels only after 1950, and probably the Chinese even later. 24. If Soviet output had remained constant at the 1929 level, “world” output would have risen until 1933, and then it would have fluctuated until 1939. 25. League of Nations (various years). The estimate takes into account the most important commodities only, but covers more countries. The same source reports an index for crops only, starting in 1920, which can be compared with the implicit “world” index for crops only. In 1920–1922, the two indices are very similar (92.8 for the League of Nations instead of 91.5) while the Leagues of Nations index grows decidedly more in the 1920s (in 1927–1929, it reaches 121.4 instead of 111.4) and in the 1930s (136.5 instead of 116.3).

The Growth of World Agricultural Production, 1800–1938

149

26. It is possible to calculate the “losses” from the Great Crisis under the assumption that production had been growing as fast as in the 1920s. The counterfactual “world” 1938 production would have been about a quarter greater than the actual one. 27. Cf., Nakamura (1966) and the short survey by MacPherson (1987, p. 53). 28. Wheatcroft-Davies (1994a, b). Allen (2002) is less critical. He remarks that the archival sources, recently made available, do not prove the allegations. The lack of “corrections” by the Moscow statistical offices, however, does not rule out the “cooking” of the figures by farm or district managers at the local level, in order to fulfill their plan targets and to please their Moscow bosses. 29. Cf. Clarke-Matko (1984, Table 5). In all three cases, the rate of change in 1920–1938 is not significantly different from zero. 30. Heston’s skepticism is fully supported by Pray (1984), who remarks that official figures imply a 40% fall in per capita consumption in Bengal. Maddison (1985) and McAlpin (1983) admit that the official statistics may be wrong, but do not fully endorse Heston’s alternative hypothesis. In contrast, Blyn (1966, pp. 150 ff) and Mishra (1983) trust the official figures. Cf., for the whole debate, Roy (2000, pp. 52–55). 31. Cf., O’Brien-Prados (1992, Table 2). The rates for French Francs and the Spanish peseta coincide almost perfectly with the PPP. 32. The latter is obtained for each country as the 1913 value times the ratio of output in that year to the 1909–1913 average. The result would be unbiased if relative prices of agricultural products had not changed. 33. Cf. Appendix A. 34. Prados (2000). The shares are not exactly comparable to those of the other columns of Table 6 because he omits four countries (Chile, India, Indonesia and Switzerland). 35. The long-run growth rate is 1.18% for the basic series (column a), 1.15% for adjusted 1909–1913 output (column c), 1.21% for “protectionist” (column d) and 1.24% for “agricultural” PPPs (columne). None of these differences is significant even at the 10% level. 36. Cf., Appendix A. The missing Brazilian output is crudely estimated according to its agricultural workforce (Mitchell, 1998b). 37. Cf., Perkins (1969, Table D.32 – he puts forward a range from 0.24 to 0.64% – and 0.5% is his “preferred” estimate), Feuerwerker (1980, p. 6 and 1983, p. 63), Chao (1986, p. 216) (multiplying his estimates of consumption for the population estimates by Maddison (1998, Table D1), Rawski (1989, pp. 322–328 and Table 6.11), Wiens (1997, pp. 65–71), Maddison (1998, Tables C.1 and D.1) and Wang (1992, Table 4.1). Cf., also, on the “optimist” side, Brandt (1989, pp.132–133 and 1997, pp. 289–292) and the survey by Richardson (1999, pp. 31–39). 38. Production is said to have increased in Syria from the 1830s to World War I (Schilcher, 1991 p. 173), and in East Africa in the interwar years (Mosley, 1983, p. 121) but not in Macedonia (Akarli, 2000, pp.127–129). 39. Calculation by the author from data in Maddison (2001, Tables A-2, A-3, B-10 and B-18). According to his estimates, the Chinese GDP per capita declined by almost a fifth. Thus, the GDP of the “rest of the world” excluding China increased by 120%. The “rest of the world” includes all Africa, Asia (without India, Indonesia and Japan) and Latin America (without Argentina, Chile, and Uruguay). Unfortunately, Maddison does not provide enough data to compute the GDP per capita of Balkan countries. 40. It is assumed that prices increased by 20% from 1870 to 1938 – i.e. by 0.30% per year (cf., Federico, forthcoming), that income elasticity was 0.6 and price elasticity was –0.2.

150

GIOVANNI FEDERICO

41. Cf., Rao (1993, Table 5.4) for the output, FAO (1952) for the acreage (55% for meadows and pasture) and Appendix C for the population. The acreage of the twenty-five countries at their 1913 borders is proxied by that of the “corresponding” countries in the 1940s. For instance, it includes Yugoslavia, which included a sizeable part of the defunct Austro-Hungarian Empire, net of pre-1913 Serbia (from Institute Internationale d’Agriculture, 1909a, 1921). 42. The rates for the 25 countries differ from those of Table 2 because they are calculated as geometric interpolation. 43. Cf., Maddison (2001, Table B-22) and Maddison’s estimates are discussed by Federico (2002), while Bairoch (1999, pp.130–134) provides additional references and discussion on the growth in the very long run, from pre-history to 1800. 44. Cf., Statistical Appendix Table I and Appendix B for the sources and methods. Some of the shares have been obtained as a linear interpolation from benchmark years, and thus they are bound to be less volatile than in reality. 45. Cf., Federico (forthcoming, Table 3.7). The data refer to a dozen “advanced” countries. 46. The exact figures are 72% for Italy in 1911 (Federico, 1992), 69% for Belgium in 1913 (Blomme, 1992a, b), 69% in the United States in 1900 (Towne & Rasmussen, 1960) and 72% in China in 1914–1918 (Perkins, 1969) – the last figure being an upper bound because the gross output omits some minor products. 47. Most of the data are from Mitchell, while the production of textile fibres (flax, hemp and cotton) and tobacco is the 1909–1913 average from Institute Internationale d’Agriculture (1909a, 1921). The production of cocoons is estimated from that of silk (Federico, 1997, Table A VI) assuming a 12:1 yield. The information from these sources are supplemented or substituted with figures from Sandgruber (1978, Table 135) for Austria, Blomme (1992a, b) for Belgium, Petmezas (1999) for Greece, Lains-Silveira Sousa (1998) for Portugal, Federico (1992), adjusted to 1913 for Italy, U.S. Bureau of the Census (1975) for the United States, Perkins (1969, Appendix D) for China, McCarthy (1982, sections 14 and 15) for the Ottoman Empire and Manarungsan (1989, Tables A.3, A.5 and 3.2) for Thailand. 48. Mitchell reports figures for the 1913 gross production of livestock products in Finland, Canada, Australia (milk and wool only) and Japan (meat only). Additional data are taken from country sources for the United States (U.S. Bureau of the Census, 1975), Italy (Federico, 1992), Germany (Hoffmann, 1965), Belgium (Blomme, 1992a, b), the Netherlands (Knibbe, 1994), India (Sivasubramonian, 2000), Denmark (Jensen, 1937), Austria (Sandgruber, 1978), Portugal (Lains and Silveira Sousa, 1998), the United Kingdom (Mitchell, 1988) and China (Perkins, 1969, Appendix D), assuming a dead weight of 150 kg. for cattle, 80 for pigs and 10 for sheep). The Hungarian productivity is assumed to have been equal to the Austrian one for meat and four fifths its level for milk. The data for the 1930s and 1950s are taken from Mitchell (1998 a, b and c; Institute Internationale d’Agriculture (1939–1940), and FAO Yearbook (1956, Tables 72A and 77). 49. The series omits the output of Western Australia (Butlin-Sinclair, 1986, Table 6), which is, however, included in the total GDP of Table 1 (p. 129). On the other hand, it includes mining, other than gold mining in South Australia (p. 137). The first omission is corrected by adding 70% of the Western Australian GDP. After 1900/01, the data are calculated as a simple average of two consecutive fiscal years. 50. Butlin’s definition of GSP differs from the standard one. Thus, the figures are calculated ex-novo as the value of “gross output” less the expenditure for seed (Tables 49

The Growth of World Agricultural Production, 1800–1938

151

and 50) and for fodder (Tables 53 and 54 for “agriculture,” 68 and 69 for “dairying” and 40 for “pastoral.” There are no data for the fodder expenses in the “pastoral” sector before 1900 (i.e. in Table 39). The omission is not corrected, as it seems more likely that Butlin reckoned them to be negligible than that he simply forgot to estimate the item altogether. Finally, the aggregate GSP at current prices have been deflated with the price indices of Table 267 in order to get a series at constant (1910–1911) prices. The data for VA at constant (1910–1911) prices are taken from Table 269, those at current prices from Tables 41 (“pastoral”) 53 and 54 (“agriculture”), and 68 and 69 (“dairying, forestry, fisheries”). The VA of forestry and fishing is deducted from the total of Tables 68 and 69 by assuming that its share on VA was the same on the GSP. 51. The missing Hungarian output in 1921–1923 is interpolated with an average of the Austrian and Czechoslovak figures for the same years. Output in Austria and Hungary in 1938 is crudely estimated by extrapolating the 1937 production with wheat output (Mitchell, 1998a). Finally, the agricultural VA in Yugoslavia is computed as the total one (Table XIII) times a linear interpolation of the share of agriculture in total GDP in 1910, 1931 and 1953 from tab. XVIII. 52. McInnis’ index is preferred to the original constructed by Urquhart (1993, p. 24, Table 1.6), which also includes non-agricultural sectors. 53. Population figures for Java and Madura have been provided by Van der Eng, while, following his method (1996, p. 271), the population of the Other Islands is assumed to have grown at 1.5% per year. 54. In both cases, the differences between the ISTAT estimate and the new one at benchmark years is minimal. In 1911, the new estimate is 1.4% lower than the ISTAT one, while the share of livestock products is 69.2% instead of 68.8% according to the ISTAT. 55. The use of the value of stock as a proxy for output may undervalue the growth in production if the increase in productivity has not been fully translated in the price of animals. On the other hand, Wheatcroft (1990, pp. 90–91) argues that Gregory’s figures overstate the growth of stock – and these two biases might compensate. 56. It is assumed that 60% of the meat was produced from cattle, 25% from pigs and 15% from sheep and that the cow milk accounted for 85% of the total (Falkus, 1968, Table 7). It is also assumed that there was no increase in productivity per head from 1870 to 1885. 57. Wheatcroft-Davies (1994b) report somewhat different data on production in 1913. Using their estimates would not change the long-term growth rate of gross output, but it would yield an implausibly high share of livestock (up to 80% in 1891). 58. Allen’s index refers to the Soviet Union at 1939 boundaries. Its use for Russia at 1913 boundaries is bound to bias the overall trend as the lost areas (mainly Poland) did not experience the dramatic fall and recovery in the 1930s. 59. The 1920 estimate (54) is substantially lower than the official Fig. (64), reported by Clarke-Matko (1984, Table 5). As Adamets (1997) points out, data are extremely uncertain, and estimates range from 25 to 75% of pre-war level. 60. Lewis has computed his index by splicing the annual production index by Drescher (1955) upon Ojala’s (1952) multi-year averages (Lewis, p. 259) and by extrapolating back to 1852 with assumptions on per capita consumption. The Drescher series (called “economic index of production”) is a weighted average of twelve product series, including feedstuffs such as turnips and mangolds. In a comment, Fletcher (1955) argues that Drescher does not follow the standard definition of GSP and the index rises more than an (apparently comparable) index from Ojala, because of the fast rise in livestock output.

152

GIOVANNI FEDERICO

61. The figure is obtained by comparing Feinstein’s Tables 8 and 54 (column1), which refers to Great Britain at 1913 boundaries. In 1911, Eire accounted for 18% of the ploughland, 31.5% of the meadows and for 28% of the whole agricultural acreage of the United Kingdom (Institute Internationale d’agriculture (1909 a` 1921) Table 4). 62. The shares are from O’Grada (1991); the underlying data from Mitchell (1998c, Tables C2 C6 C7 C8). 63. These series might include some purchases of imported stuff and of feed of industrial origin, and this could cause a small undervaluation of GDP. The shares of livestock on gross output is obtained first dividing the “home consumption” between crops and livestock products according to the respective shares of the sum of the two categories, and then deducting “seed” from the value of crops and feed and livestock from the value of livestock products. 64. The two latter items are simply omitted. Rents belong to the dwellings, while subsidies are negative taxation – i.e. impinge on the difference between figures at market price and at factor costs. 65. Maddison (1998, Table D-1) for China, Visaria and Visaria (1983, Table 5.7) (Davis & Gujaral estimates) and Sivasubramonian (2000, Table 6.9) for India, Institute Internationale d’Agriculture 1909/13 and 1925 for the Soviet Union in 1920, Wheatcroft-Davies (1994a), Table 1 for the Soviet Union in 1938 and 1938 and from personal communications by S. Petmezas for Greece and P. Van der Eng for Indonesia.

ACKNOWLEDGMENTS The author thanks B. Allen, T. David, S. Fenoaltea, P. Lains, D. Ma, S. Pamuk, S. Petmezas, L. Prados, M. S. Schultze, A. Taylor, P. Van der Eng, J. L. Van Zanden and J. Williamson for having provided highly useful information and shared with me the results of their research before publication, and the participants in seminars at UC-Los Angeles and UC-Davis, and to the Fourth World Cliometrics Conference (Montreal 5–9 July 2000) for their comments on earlier versions of the paper (published as Working paper n103 of the Agricultural History Center. University of California at Davis). The remaining errors are mine. The data are available at http://www.iue.it/HEC/People/Faculty/Profiles/Federico.shtml.

REFERENCES Adamets, S. (1997). ‘A’ l’origine de la diversit´e des measures de la famine sovietique: La statistique des prix, des r´ecoltes et de la consommation. Cahiers du Monde Russe, 38, 559–586. Afton, B., & Turner, M. (2000). The statistical base of agricultural performance in England and Wales, 1850–1914. In: E. J. T. Collins (Ed.), The Agrarian History of England and Wales, Vol. VII 1850–1914. Cambridge: Cambridge University Press. Akarli, A. (2000). Growth and retardation in Ottoman Macedonia, 1880–1910. In: S. Pamuk & J. Williamson (Eds), The Mediterranean Response to Globalisation Before 1950. London: Routledge.

The Growth of World Agricultural Production, 1800–1938

153

Allen, R. (1994). Agriculture during the industrial revolution. In: D. McCloskey & R. Floud (Eds), The Economic History of Britain since 1700 (Vol. 7, 2nd ed.). Cambridge: Cambridge University Press. Allen, R. (1999). Tracking the agricultural revolution in England. Economic History Review, 52, 209–235. Allen, R. (2002). Farm to factory: A reinterpretation of the Soviet industrial revolution. Unpublished manuscript. Arndt, H. W. (1963). The economic lessons of the 1930s (2nd ed.). London: Frank Cass. Bairoch, P. (1999). L’agriculture des pays d´evelopp´es. 1800 a` nos jours. Paris: Economica. Baten, J. (2000). Heights and real wages in the 18th and 19th centuries: An international overview. Historische Anthropometrie, special issue of the Jahrbuch fur Wirtschaftsgeschichte, 62–76. B´ertola, L. (1998). El PBI Uruguayo 1870–1936 y Otras Estimaciones. Montevideo: Facultad de Ciencias Sociales. Biraben, J. N. (1979). Essai sur le Nombre des Hommes. Population, 34, 13–24. Blomme, J. (1992a). The economic development of Belgian agriculture, 1880–1980: A qualitative and quantitative analysis. Leuwen: Leuwen University Press. Blomme, J. (1992b). Produktie, produktiefactor en produktivitiet: De Belgische Landbouw in de 19de Eeuw. Belgisch Tijdschrift voor Nieuwste Geschiedenis, 25, 103–114. Bloomfield, B. T. (1984). New Zealand: A handbook of historical statistics. Boston: Hall. Blyn, G. (1966). Agricultural trends in India, 1891–1947: Output, availability, and productivity. Philadelphia: University of Pennsylvania Press. Brandt, L. (1989). Commercialization and agricultural development. Central and Eastern China, 1870–1937. New York: Cambridge University Press. Brandt, L. (1997). Reflections on China’s late 19th and early 20th century economy. China Quartely, 150, 282–307. Braun, J. M., Braun, I., Briones, J. D., Luders, R., & Wagner, G. (2000). Economia Chilena 1810–1995: Estadisticas Historicas. Documentos de trabajo n.187 Pontificia Universidad catolica de Chile Instituto de Economia available at http://www.volcan.facea.puc.cl/economia/ publicaciones/documentos trabajo.htm. Brigas Gutierrez, M. A. (2000). La productividad de los factoers en la agricultura Espanola (1735–1935). Madrid: Banco de Espana Estudios de Historia Economia n 39. Butlin, N. G. (1962). Australian domestic product, investment and foreign borrowing, 1861–1938/1939. Cambridge: Cambridge University Press. Butlin, N. G. (1986). Contours of the Australian economy. Australian Economic History Review, 26, 96–125. Butlin, N. G., & Sinclair, W. A. (1986). Australian gross domestic product, 1788–1860: Estimates, sources and methods. Australian Economic History Review, 26, 126–147. Cameron, R. (1989). A concise economic history of the world. Oxford: Oxford University Press. Canada Handbook (1960). Handbook of agricultural statistics 1960, supplement. Ottawa: Dominion Central Bureau of Statistics. Carr, R. (1973). La Riforma Agraria Messicana. In: E. L. Jones & S. Woolf (Eds), Agricoltura E Sviluppo Economico. Torino: Einaudi. Chao, K. (1986). Man and land in Chinese economic history. Stanford: Stanford University Press. Clark, C. (1977). Population growth and land use (2nd ed.). London: Macmillan. Clark, G. (2002). The agricultural revolution and the industrial revolution: England 1500–1992. Paper avialable at www.econ.ucdavis.edu/faculty/Gclark/papers/pod2002.pdf.

154

GIOVANNI FEDERICO

Clark, G., Huberman, M., & Lindert, P. (1995). A British food puzzle. Economic History Review, 48, 215–237. Clarke, R., & Matko, D. (1984). Soviet economic facts, 1917–1981 (3rd ed.). New York: St Martins Press. Cortes-Conde, R. (1997). La Economia Argentina en el Largo Plazo: Ensayos de historia economica. Buenos Ayres: Editorial Sudamericana-Universidad de San Andres. Crafts, N. (1985). British economic growth during the industrial revolution. Oxford: Oxford University Press. Craig, L. A., & Weiss, T. (2000). Hours at work and total factor productivity in 19th century U.S. agriculture. Advances in Agricultural Economic History, 1, 1–30. Crisostomo, D. C., & Barker, R. (1979). Agricultural growth in the Philippines, 1918–1971. In: Y. Hayami, V. Ruttan & H. Southworth (Eds), Agricultural Growth in Japan, Taiwan, Korea and the Philippines. Honolulu: University Press of Hawai. Davies, R. W. (1998). Soviet economic development from Lenin to Khrushchev. Cambridge: Cambridge University Press. Diaz Alejandro, C. (1970). Essays on the economic history of the Argentine Republic. New Haven and London: Yale University Press. Deane, P. (1967). The first industrial revolution. Cambridge: Cambridge University Press. Deane, P., & Cole, W. A. (1962). British economic growth 1688–1959. Cambridge: Cambridge University Press. Dessirer, J. (1928). Indices compar´es de la production industrielle et de la production agricole en divers pays de 1870 a 1928. Bulletin de la Statistique Generale de la France, 18, 65–109. Drescher, L. (1955). The development of agricultural production in Great Britain and Ireland from the early 19th century. The Manchester School of Economic and Social Studies, 23, 153–175. Eckstein, A. (1955). National income and capital formation in Hungary 1900–1950. In: S. Kuznets (Ed.), Income and Wealth. London: International Association for Research in Income and Wealth. Ercolani, P. (1969). Documentazione statistica di base. In: G. Fu`a Milano & F. Angeli (Eds), Lo Sviluppo Economico in Italia (Vol. 3). Falkus, M. E. (1968). Russia’s national income, 1913: A revaluation. Economica, 35, 52–73. Food and Agricultural Organization (various years). Yearbook of food and agricultural statistics (later production yearbook). Washington: Food and Agriculture Organization. Federico, G. (1992). Il valore aggiunto dell’agricoltura. In: G. M. Rey (Ed.), I Conti Economici Dell’italia. 3. Una Stima del Valore Aggiunto Per Rami di Attivit`a per il 1911. Collana storica della Banca d’Italia. Serie Statistiche, 1. 2. Bari: Laterza. Federico, G. (1997). An economic history of the silk industry. Cambridge: Cambridge University Press. Federico, G. (2000). Una stima del valore aggiunto in agricoltura. In: G. M. Rey (Ed.), I Conti Economici dell’Italia. 3. Il Valore Aggiunto per il 1891, 1938 e 195. Collana storica della Banca d’Italia. Serie Statistiche, 3.1. Bari: Laterza. Federico, G. (2002). The world economy 0–2000 AD: A review article. European Review of Economic History, 6, 111–120. Federico, G. (2003). Le nuove stime della produzione agricola Italiana, 1860–1910: Primi risultati ed implicazioni. Rivista di storia economica n.s., 19, 359–381. Federico, G. (forthcoming). An economic history of agriculture, 1800–2000. Princeton: Princeton University Press. Feinstein, C. H. (1972). National income, expenditure and output of the U.K., 1855–1965. Cambridge: Cambridge University Press.

The Growth of World Agricultural Production, 1800–1938

155

Feinstein, C. H., Temin, P., & Toniolo, G. (1997). The European economy between the wars. Oxford: Oxford University Press. Feuerwerker, A. (1980). Economic trends in the late Ch’ing Empire, 1870–1911. In: J. Fairbank & Kwang-Ching Liu (Eds), The Cambridge History of China (Vol. 11, Part II). Cambridge: Cambridge University Press. Feuerwerker, A. (1983). Economic trends 1912–1949. In: J. Fairbank (Ed.), The Cambridge History of China (Vol. 12, Part I). Cambridge: Cambridge University Press. Fletcher, T. W. (1955). Drescher’s index: A comment. The Manchester School of Economic and Social Studies, 23, 176–183. Fogel, R. W. (1997). New findings on secular trends in nutrition and mortality: Some implications for population theory. In: M. Rosenzweig & O. Stark (Eds), Handbook of Population and Family Economics. Amsterdam: Elsevier. Franz, G. (1976). Landwirtschaft 1800–1850. In: H. Aubin & W. Zorn (Eds), Handbuch der Deutschen Wirtschafts-und Sozialgeschichte (Vol. 2). Stuttgart: Ernst Klett. Gadisseur, J. (1973). Contribution a l’etude de la production agricole en Belgique de 1846 a 1913. Revue Belge d’Histoire Contemporaine, 4, 1–48. Goldsmith, R. W. (1961). The economic growth of Tsarist Russia, 1860–1913. Economic Development and Cultural Change, 9, 441–475. Good, D. (1984). The economic rise of the Habsburg empire, 1750–1914. Berkeley: University of California Press. Goosens, M. (1992). The economic development of Belgian agriculture, 1812–1846: A regional perspective. Leuwen: Leuwen University Press. Grantham, G. (1996). The French agricultural capital stock, 1789–1914. Research in Economic History, 16, 39–84. Gregory, P. (1982). Russian national income 1885–1913. Cambridge: Cambridge University Press. Haig, B. (2001). New estimates of Australian GDP: 1861–1948/1949. Australian Economic History Review, 41, 1–34. Hansen, S. A. (1974). Økonomisk vækst i Danmark. Copenhagen: Universitet Institutut for økonomisk historie publikation, n. 6. Hansen, B., & Wattleworth, M. (1978). Agricultural output and consumption of basic foods in Egypt, 1886/1887–1967/1968. International Journal of Middle East Studies, 9, 449–469. Harrison, M., & Gattrell, P. (1993). The Russian and Soviet economies in two world wars: A comparative view. Economic History Review, 46, 425–452. Hautcoeur, P.-C. (forthcoming). Was the Great War a watershed? The economics of WWI in France. In: S. Broadberry & M. Harrison (Eds), The Economic of WWI. Helling, G. (1965). Berechnung eines Index der Agrarproducktion in Deutschland im 19 Jahrhundert. Jahrbuch fur Wirtschaftsgeschichte, 125–150. Heston, A. (1983). National income. In: D. Kumar & M. Desai (Eds), The Cambridge Economic History of India. II. c. 1757– c. 1970. Cambridge: Cambridge University Press. Hjerrpe, R. (1989). The Finnish economy, 1860–1985. Growth and structural change. Helsinki: Bank of Finland. Hoffmann, W. (1965). Das Wachstum der Deutschen Wirtschaft Seit der Mitte des 19. Jahrhunderts. Berlin: Springer. Holtfrerich, C. L. (1986). The German inflation, 1914–23. Berlin-New York: de Gruyter. IEERAL (1986). Estadisticas de la evolucion economica de Argentina, 1913–1984. Estudios, No. 39 (July/September 1986), 101–184.

156

GIOVANNI FEDERICO

Institute Internationale d’agriculture (various years). Annuarie international de statistique agricole, Rome: Institute Internationale d’Agriculture (later international yearbook of agricultural statistics). Rome: International Institute of Agriculture. ISTAT (1957). Indagine Statistica sullo Sviluppo del Reddito Nazionale dell’Italia dal 1861 al 1956. Annali di Statistica, Serie 8, 9, Rome: ISTAT. ISTAT (1958). Sommario di statistiche storiche. Roma: ISTAT. James, H. (2001). The end of globalization. Cambridge, MA: Harvard University Press. Jensen, E. (1937). Danish agriculture. Its economic development. Copenhagen: J. H. Schultz Forlag. Johansen, H. (1985). Dansk historical statistics, 1814–1980. Copenhagen: Glydenal. Johnson, D. G. (2000). Population, food, and knowledge. American Economic Review, 90, 1–14. ¨ Kausel, A. (1979). Osterreisches Volkseinkommen 1830 bei 1913. In: Geschichte und Ergebnissse der ¨ ¨ Zentralen Amtlichen Statistik in Osterreich 1829–1979. Vienna: Osterreichisches Statistiches Zentralamt. Kausel, A. et al. (1965). Osterreichs Volkseinkommen 1913 bis 1963. In: Monatsberichte des Osterichischen Institutes f¨ur Wirstschaftsorchung. 14 Sonderheft. Knibbe, M. (1994). Agriculture in the Netherlands. Amsterdam: NEHA. Komlos, J. (1983). The Habsburg monarchy as a custom union. Princeton: Princeton University Press. Komlos, J. (1998). Shrinking in a growing economy? The mystery of physical stature during the industrial revolution. Journal of Economic History, 58, 778–795. Kostrowicka, I. (1984). Changes in agricultural productivity in the kingdom of Poland in the XIX and early XX century. Journal of European Economic History, 13, 75–97. Lains, P., & Silveira Sousa, P. (1998). Estatistica e produc¸a˜ o agr´ıcola em Portugal, 1848–1914. Anal´ıse Social, 33, 935–968. Landes D. (1998). The wealth and poverty of nations. New York: Norton. League of Nations (1926). Memorandum on currencies and central banks, 1913–1925. Geneva: League of Nations. League of Nations, Economic, Financial and Transit Department (1943). [II. Economic and financial II. A.7] Agricultural Production in Continental Europe during the 1914–1918 War and the Reconstruction Period. Geneva: League of Nations. League of Nations, Economic and Financial section (various years). Memorandum on production and trade[ . . . ], later review of world production, later world production and prices. Geneva. Levy-Leboyer, M. (1968). La croissance economique en France au XIXe si`ecle. R´esultats pr´eliminaires. Annales, 23, 788–807. Lindahl, E. et al. (1937). National income of Sweden 1861–1930. London: King. MacPherson, W. J. (1987). The economic development of Japan c. 1868–1941. London and Basingstoke: MacMillan. Maddison, A. (1985). Alternative estimates of the real product of India, 1900–1946. The Indian Economic and Social History Review, 22, 201–210. Maddison, A. (1991). Dynamic forces in capitalist development. Oxford: Oxford University Press. Maddison, A. (1995). Monitoring the world economy. Paris: OECD. Maddison, A. (1998). Chinese economic performance in the long run. Paris: OECD. Maddison, A. (2001). The world economy. A millennial perspective. Paris: OECD. Manarungsan, S. (1989). Economic development of Thailand, 1850–1950. Response to the challenge of the world economy. Bangkok, Chulalongkorn University. Institute of Asian Studies, Monograph No. 042. Mancall, P. C., & Weiss, T. (1999). Was economic growth likely in Colonial British North America? Journal of Economic History, 59, 17–40.

The Growth of World Agricultural Production, 1800–1938

157

McAlpin, M. (1983). Price movements and fluctuations in economic activity. In: D. Kumar & M. Desai (Eds), The Cambridge Economic History of India. II. c. 1757–c. 1970. Cambridge: Cambridge University Press. McCarthy, J. (1982). The Arab World, Turkey and the Balkans: A handbook of historical statistics. Boston: Hall. McInnis, M. (1986). Output and productivity in Canadian agriculture, 1870–1871 to 1926–1927. In: S. Engerman & R. E. Gallman (Eds), Long Term Factors in American Economic Growth. Chicago: University of Chicago Press. Merrick, T. W., & Graham, D. H. (1979). Population and economic development in Brazil, 1800 to the present. Baltimore and London: John Hopkins University Press. Metzler, J. (1998). The divided economy of mandatary Palestine. Cambridge: Cambridge University Press. Mishra, S. C. (1983). On the reliability of pre-independence agricultural statistics in Bombay and Punjab. The Indian Economic and Social History Review, 20, 171–190. Mitchell, B. R. (1988). British historical statistics. Cambridge: Cambridge University Press. Mitchell, B. R. (1998a). International historical statistics: Africa, Asia, Oceania, 1750–1993 (3rd ed.). London: MacMillan. Mitchell, B. R. (1998b). International historical statistics: The Americas, 1750–1993 (4th ed.). London: MacMillan. Mitchell, B. R. (1998c). International historical statistics: Europe, 1750–1993 (4th ed.). London: MacMillan. Mizoguchi, T., & Umemura, M. (1988). Basic economic statistics of former Japanese colonies, 1895–1938. Tokyo: Keizai Shinposha. Moore, W. E. (1945). Economic demography of Eastern and Southern Europe. Geneva: League of Nations. Mosley, P. (1983). The settler economies. Studies in the economic history of Kenya and Southern Rhodesia, 1900–1983. Cambridge: Cambridge University Press. Nakamura, J. L. (1966). Agricultural production and the economic development of Japan, 1873–1922. Princeton: Princeton University Press. O’Brien, P. (1968). The long-term growth of agricultural production in Egypt: 1821–1962. In: P. M. Holt (Ed.), Political and Social Change in Modern Egypt. London. O’Brien, P., & Prados de la Escosura, L. (1992). Agricultural productivity and European industrialization. Economic History Review, 51, 514–536. O’Grada, C. (1991). Irish agriculture. North and South since 1900. In: B. M. Campbell & M. Overton (Eds), Land, Labour and Livestock: Historical Studies in European Agricultural Productivity. Manchester: Manchester University Press. O’Grada, C. (1993). Ireland before and after the Famine (2nd ed.). Manchester: Manchester University Press. Ojala, E. M. (1952). Agriculture and economic progress. Oxford: Oxford University Press. Okhawa, K., & Shinohara, M. (1979). Patterns of Japanese economic development. New Haven and London: Yale University Press. Overton, M. (1996). Agricultural revolution in England. Cambridge: Cambridge University Press. Paish, G. (1913–1914). On prices of commodities in 1914a. Journal of the Royal Statistical Society, 77, 565–570. Palairet, M. (1997). The Balkan economies. Cambridge: Cambridge University Press. Perkins, D. (1969). Agricultural development in China. Edinburgh: Edinburgh University Press.

158

GIOVANNI FEDERICO

Petmezas, S. (1999). Revising the estimates for the agricultural output of Greece (1833–1939). Third conference of the European Historical Economics Society. Lisbon, October 29–30, 1999. Prados de la Escosura, L. (1989). La estimacion indirecta de la produccion agraria en el siglo XIX: Replica a Simpson. Revista de Historia Economica, 7, 703–717. Prados de la Escosura, L. (1993). Spain’s gross domestic product, 1850–1990: A new series. Madrid: Ministerio de Economia Y Hacienda. Documentos de Trabajo D-93002. Prados de la Escosura, L. (2000). International comparisons of real product, 1820–1990: An alternative data-set. Explorations in Economic History, 37, 1–41. Prados de la Escosura, L. (2004). El progreso economico de Espana (1850–2000). Bilbao: Fundacion BBVA. Pray, C. E. (1984). Accuracy of official agricultural statistics and the sources of growth in the Punjab, 1907–1947. The Indian Economic and Social History Review, 21, 312–333. Pryor, F. et al. (1971). Czechoslovak aggregate production in the interwar period. Review of Income and Wealth, 1, 35–60. Rao, P. (1993). Intercountry comparisons of agricultural output and productivity. Rome: FAO Economic and social development paper 112. Rosenberg, N., & Birdzell, L. (1986). How the west grew rich. New York: Basic Books. Rawski, T. (1989). Economic growth in pre-war China. Berkeley: University of California Press. Reynolds, C. W. (1970). The Mexican economy. Twentieth-century structure and growth. New Haven and London: Yale University Press. Reynolds, L. G. (1985). Economic growth in the third world: An introduction. New Haven and London: Yale University Press. Richardson, P. (1999). Economic change in China, c. 1800–1950. Cambridge: Cambridge University Press. Ritzmann-Blickenstorfer, T., & David, T. (no date). Un image statistique du developpment economique en suisse. Personnes actives et produit interieur brut par branches et Cantons, 1890–1965. Lausanne: Universit´e de Lausanne, Mimeo. Roy, T. (2000). The economic history of India, 1857–1947. Oxford: Oxford University Press. Saito, T., & Kin, K. L. (1999). Statistics on the Burmese economy. Singapore: The 19th and 20th Century Institute of Southeast Asian Studies. Sandgruber, R. (1978). Osterreisches Agrarstatistik, 1750–1918. Wien: Verlag fur Geschichte und Politik. Schilcher, L. (1991). The grain economy of late Ottoman Syria and the issue of large scale commercialisation. In: C. Keyder & F. Tabak (Eds), Landholding and Commercial Agriculture in the Middle East. New York: State University of New York Press. Schon, L. (1995). Jordbruk med Bin¨aringar, 1800–1980. Lund: Skrifter Utgivna av EkonomiskHistorika F¨oreningen i Lund, Vol. XXIV. Schultze, M.-S. (2000). Pattern of growth and stagnation in the late nineteenth Habsburg economy. European Review of Economic History, 4, 311–340. Schultze, M.-S. (forthcoming). Austria-Hungary’s economy in WWI. In: S. Broadberry & M. Harrison (Eds), The Economics of WWI. Simpson, J. (1989a). La produccion agraria y el consumo Espanol en el Siglo XIX. Revista de Historia Economica, 7, 355–388. Simpson, J. (1989b). Una Respuesta al Profesor Leandro Prados de la Escosura. Revista de Historia Economica, 7, 719–723. Sivasubramonian, S. (2000). The national income of India in the twentieth century. New Delhi: Oxford University Press.

The Growth of World Agricultural Production, 1800–1938

159

Statistics Canada (1983). Historical statistics of Canada (2nd ed.). F. H. Leacy (Ed.). Ottawa. Steckel, R. H. (1995). Stature and the standard of living. Journal of Economic Literature, 33, 1903–1940. Steckel, R. H. (1998). Strategic ideas in the rise of anthropometric history and their implications for interdisciplinary research. Journal of Economic History, 58, 803–821. Steckel, R., & Floud, R. (1997). Conclusions. In: R. Steckel & R. Floud (Eds), Health and Welfare during Industrialization. Chicago: University of Chicago Press for NBER. Stillson, R. T. (1971). The financing of Malayan rubber, 1905–1923. Economic History Review, 24, 589–598. Strauss, F., & Bean, L. H. (1940). Gross farm income and indices of farm production and prices in the U.S. 1869–1937. U.S. Department of Agriculture Technical Bulletin, No. 703. Washington: Government Printing Office. Tilly, R. H. (1978). Capital formation in Germany in the 19th century. In: P. Mathias & M. M. Postan (Eds), The Cambridge Economic History of Europe (Vol. VII, Part 1). Cambridge: Cambridge University Press. Toutain, J. C. (1961). Le produit de l’agriculture Franc¸aise de 1700 a 1958 [Histoire quantitative de l’economie franc¸aise (1) and (2)]. Cahiers de l’Institut de Science Economique Appliquee. Serie AF 1 and 2. Paris: ISEA. Toutain, J. C. (1997). La Croissance Franc¸aise 1789–1990. Nouvelles estimations. Economies et soci´et´es. Cahiers de l’ISMEA. Serie Histoire quantitative de l’economie francaise. Serie HEQ No. 1. Paris: PUF. Towne, M. W., & Rasmussen, W. D. (1960). Farm gross product and gross investment in the 19th century. In: Trends in The American Economy in the 19th Century. Studies in Income and Wealth (Vol. 24). Princeton: Princeton University Press. Turner, M. (1996). After the famine. Irish agriculture 1850–1914. Cambridge: Cambridge University Press. Turner, M. (2000). Agricultural output, income and productivity. In: E. J. T. Collins (Ed.), The Agrarian History of England and Wales Vol. VII 1850–1914. Cambridge: Cambridge University Press. Union of South Africa (1960). Union statistics for fifty years. Jubilee issue, 1910–1960. Pretoria: Bureau of Census and Statistics. United Nations (various years). United Nations demographic yearbook. Geneva: United Nations. Urquhart, M. C. (1993). Gross national product, Canada 1870–1926: The derivation of the estimates. Kingston and Montreal: McGill and Queens University Presses. U.S. Bureau of the Census (1975). Historical statistics of the U.S.: Colonial times to 1975. Washington: Government Printing Office. Van der Eng, P. (1996). Agricultural growth in Indonesia. London and Basingstoke: MacMillan. Van Zanden, J. L. (1988). The first green revolution: The growth of production and productivity in European agriculture, 1870–1914. Research Memorandum, 42, Amsterdam: Vrije Universiteit Facultit der Economische Wetenschappen en Econometrie. Van Zanden, J. L. (2000). Estimates of GNP. Available at http://www.nationalaccounts.niwi.knaw.nl. Van Zanden, J. L. (2003). Rich and poor before the industrial revolution: A comparison between Java and the Netherlands at the beginning of the 19th century. Explorations in Economic History, 40, 1–23. Vinsky, I. (1961). National product and fixed assets in the territory of Yugoslavia 1961. In: P. Deane (Ed.), Studies in Social and Financial Accounting. Income and Wealth Series (Vol. 9). London.

160

GIOVANNI FEDERICO

Visaria, L., & Visaria, P. (1983). Population (1757–1947). In: D. Kumar & M. Desai (Eds), The Cambridge Economic History of India. II. c. 1757–c. 1970. Cambridge: Cambridge University Press. Waizner, E. (1928). Das Volkseinkommen Alt-Osterreichs und seine Vertellung auf die Nachfolgestaaten. Metron, 7, 97–179. Wang, Y. (1992). Economic development in China between the two world wars (1920–1936). In: T. Wright (Ed.), The Chinese Economy in the Early Twentieth Century. Recent Chinese Studies. New York: St Martin’s 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. Wheatcroft, S. G. (1990). Agriculture. In: R. W. Davies (Ed.), From Tsarism to the New Economic Policy. London and Basingstoke: MacMillan. Wheatcroft, S. G., & Davies, R. W. (1994a). The crooked mirror of Soviet economic statistics. In: R. W. Davies, M. Harrison & S. G. Wheatcroft (Eds), The Economic Transformation of the Soviet Union, 1913–1945. Cambridge: Cambridge University Press. Wheatcroft, S. G., & Davies, R. W. (1994b). Agriculture. In: R. W. Davies, M. Harrison & S. G. Wheatcroft (Eds), The Economic Transformation of the Soviet Union, 1913–1945. Cambridge: Cambridge University Press. Williamson, J., & O’Rourke, K. (1999). Globalization and history. Cambridge, MA: MIT Press.

APPENDIX A The Estimate of “PPP-adjusted” Agricultural Production in 1913 The PPP-adjusted production in 1913 is computed for forty-nine countries, the twenty-five of the sample and twenty-four others, including China (cf. the full list in Table A.6). The computation follows the three-step usual procedure: (1) estimate total production; (2) deduct seed and feed; (3) multiply by “world” prices to obtain gross output; and (4) deduct expenditures on purchased materials to get Value Added. (1) Production is computed taking twenty-three products into account: wheat, rye, barley, maize, rice, cassava, sugar-beet, cane sugar, potatoes, sweet potatoes, tobacco, cotton, wine, olive oil, citrus fruit, flax, hemp, tea, rubber, meat, milk, wool and cocoons. This list seems fairly complete for temperate agriculture. The main omissions are pulses, vegetables, wood, fruit, and poultry. In all cases where a comparison is possible, the included products accounted for about 70% of the total gross output.46 In contrast, the coverage of tropical agriculture is decidedly poor, as the list omits vegetable oils, coffee, cocoa, sorghum, etc. In any event the distortion is relatively small, because (unfortunately), the sample includes only one tropical country, Indonesia. The production data are taken from Mitchell’s well-known statistical compilations (Mitchell, 1998a, b, c), supplemented by the yearbooks of the Institute

The Growth of World Agricultural Production, 1800–1938

161

Table A.1. Percentage Change in Output Per Animal, 1910–1913 to 1936–1939.

Italy USA Belgium Germany Netherlands U.K. Australia New Zealand India

Beef

Pork

20.6 −1.6 11.4 19.5 1.1

5.6 34.7 −13.8 22.9 2.9

Mutton

Milk

13.2 23.2

49.6 38.3 20.8 1.9 25.6 −35.9

−11.6

Wool

25.6 5.0

16.6 9.0 5.1

Sources: Italy: ISTAT (1958, pp. 114 and 116–117), Belgium: Blomme (1992a, b, Statistical Appendix, Tables 7, 14–15, 29, 36–37; Netherlands output: Knibbe (1994, Table III), stock: Mitchell (1998c, Table C5); Germany output: from Hoffmann (1965, ii Tables 54 and 55), stock: Mitchell (1998c, Table C5); the United States stock: U.S. Bureau of the Census (1975, series K564, K566 and K568), output: (U.S. Bureau of the Census, series K584, K587, K590, K593 and K597), and output of wool: Strauss and Bean (1940, Table 47); United Kingdom cattle stock: Mitchell (1998c, Table C5), output of milk: Mitchell (1988, Agriculture Table 9); New Zealand and Australia: Mitchell (1998a, Tables C11, C 13 and C15); India (milk cows) Sivasubramonian (2000, Table 3.8 and Appendix Table 3(h)).

Internationale d’agriculture and country sources whenever available. The coverage is almost complete for crops, but rather poor for livestock products.47 In most countries, yearly series for livestock products are available only from the late 1930s, if not from the 1940s or 1950s.48 Yet animal products are too important to be neglected. Thus, the production of “missing” countries is estimated multiplying the number of animals (from Mitchell) around 1910 for an estimate of the output of meat, milk and wool per animal. This latter is obtained extrapolating backwards the earliest productivity figures available – usually for the 1930s, and sometimes for the 1950s. The available evidence on productivity growth is reported in Table A.1. It is assumed that, from 1913 to the 1930s, the productivity per head of stock rose by 10% for meat and by 15% for milk and wool in the “advanced” countries (Western Europe, Canada, Argentina, South Africa and Japan), and that it remained constant elsewhere. (2) The use of cereals and potatoes for seed and feed is estimated as a fixed proportion of gross output. The available data on this proportion are reported in Table A.2. The figures reflect differences in agricultural technology (sowing by hand uses more seed), in diet, levels of income and factor endowment. For instance, potatoes, as a labor-intensive and land-saving crop, were not used for animal feed in the United States. In the more advanced countries, the seed/crop ratio was lower, but a

162

GIOVANNI FEDERICO

Table A.2. Percentage of Total Output Used for Seed and Feed, Various Countries ca. 1910. U.K. Wheat Barley Rye Maize Potatoes Rice All cereals

France

20 16

30

43

Italy

Russia

Ireland

Spain

Belgium

USA

14 67 21 43 37 3

13 10 15

6 8

7 72 24

20

42

14 66 23 57 24

14 49 73 83 20

41

36

Sources: U.K. (1904–10): Ojala (1952, Table I, II and V); Italy (1911): Federico (1992); Russia: Gregory (1982, Table D.1); Ireland (1912): Turner (1996, pp.98–99); Belgium (1919–22): Blomme (1992a, b, Tables 3–4, Statistical Appendix); USA (1913): Strauss-Bean (1940, pp. 34–41); Spain (“until 1929”): Prados (1993, Table A.1); France (1905–14): Toutain (1961, Tables 79 and 82).

higher proportion of available cereals (especially of maize) was given to animals. The assumed percentages vary according to the area and the level of development (Table A.3). (3) The concept of “world” price is quite elusive. No single market place can claim to be really representative of the world, even if London is a strong candidate, and, moreover, no source provides quotations for all the twenty-three commodities in the same market. Thus, the set of “world” prices in 1913 has to be pieced together from different sources, notably the yearbook of the Institute Internationale

Table A.3. Percentage of Total Output Used for Seed and Feed, Estimates.

Wheat Rye Barley Maize Potatoes Rice

WS

W. Europe

S. Europe

E. Europe

Asia

S.Amer.

Africa

15 25 50 15 20 4

10 20 20 50 50 4

15 20 70 50 30 4

15 15 15 35 20 4

15 15 15 10 20 4

15 15 15 10 20 4

15 15 15 10 20 4

Note: WS (Western Settlement): Australia, Canada, Uruguay, South Africa, New Zealand; W Europe: Austria, Denmark, France, Germany, Netherlands Sweden, Norway and Switzerland; Southern Europe: Greece, Portugal, Algeria, Tunisia, Morocco, Egypt, and Cyprus. E Europe: Hungary, Russia, Finland, Serbia, Bulgaria and Romania. Asia: India, Indonesia Japan, China, Indochina, Korea, Philippines Taiwan and Thailand. S.Amer. (South America): Argentina, Chile and Mexico. Africa: Madagascar, Sierra Leone, and Zimbabwe.

Free Trade Countries

Protectionist Countries.

U.K. U.K. Ireland USA USA USA Indonesia Neth. Neth. Russia Canada Canada Denmark Belg. Belg. Argentina Australia India Italy Germany Austria Austria France (a) (b) (a) (b) (c) (a) (b) (a) (b) (a) (b) (a) (b) Wheat Rye Barley Maize Potatoes Sugarbeet Sugar Rice Cassava Sweet potatoes Tobacco Flax Hemp Cotton wine (hl) Olive oil Citrus fruit Tea Rubber Beefb Porkb Muttonb Vealb Milk Greasy Woola Cocoon

1

1 1 0.87 0.86 1.03 0.99 0.74 0.76 0.61

1 0.98 0.71 0.90 0.87 0.19

1 1 0.95 0.71 0.91 0.73 0.77

1.70 1.29

2.40 0.87 0.77

0.48

0.98 1.00 0.14 0.14

1.15 1.09

1 1 1 0.82 0.70 0.77 0.99 0.79 0.69 0.62 0.34 0.14 1.11

1 0.77 0.65 0.81

0.53

1 0.89 0.96 0.35

9.82 8.98

9.59

1 0.97

12.68 6.44 6.44

10.32 10.25 9.82

1 1 1 0.86 0.85 1.00 0.98 0.64 0.36 0.12

8.04

9.29

7.76 13.25 61.44 8.96 8.37 10.06 10.08 8.53 10.24 10.02 9.44

6.07 30.06 6.78 6.41 5.91 1.24 1.00

4.02

10.75 7.62 7.88 0.67

9.75 6.96 5.40

7.68 7.82 6.20 7.24

9.05 7.60 6.78

9.24 8.07

8.01 2.74 2.19

7.08 7.91 7.04 6.95

1

1 0.77 0.76 0.62 0.38 0.09 4.81 0.82 1.72

3.38 5.41 1.03 8.32 5.07 0.52 31.03 5.87 5.80 8.40 7.79 0.55 7.74 11.00

1 0.81 0.85 0.55

1 0.80 0.72 0.70

1 0.84 0.75 0.82 0.23 0.23

0.88

1 0.70 0.79

1.11 2.22 6.04 4.60

5.57

8.25 7.00 8.71 10.20

7.33 5.41 10.31

The Growth of World Agricultural Production, 1800–1938

Table A.4. Relative Prices, by Country.

0.73

13.90 16.09

163

Sources: U.K. (a) Paish (1913–1914, pp. 556–570) except rubber from Stillson (1971, Table 1); USA a) U.S. Bureau of the Census (1975, series K 504, 508, 516, 528, 534, 537, 540, 556, 560, 563, 585, 591, 594 and 605) (b) Strauss-Bean (1940, Tables 13, 15, 18, 19, 21, 22, 25, 27, 28, 30, 36, 43, 47, 48, and 54); Italy: ISTAT (1958, pp. 173–181); Indonesia: personal communication by P. Van der Eng; Belgium a) Blomme (1992a, b, Statistical Appendix, Table 26); Netherlands (a) Knibbe (1994, Tables I.2 and I.3); Austria: (b) Waizner (1928, Table I); Canada b) Historical Statistics (1983, series M 228–233). All other data from Institute Internationale d’agriculture (1913–1914, Tables 619–736). a Greasy wool. b Dressed weight.

164

GIOVANNI FEDERICO

Table A.5. Relative Prices, Averages. (a) Wheat Rye Barley Maize Potatoes Sugarbeet Sugar Rice Cassava Sweet potatoes Tobacco Flax Hemp Cotton Wine (hl) Olive oil Citrus fruit Tea Rubber Beefb Porkb Muttonb Vealb Milk Woola Cocoons

0.83 0.89 0.71 0.55 0.15 1.55 0.98 0.14 0.79 10.49 8.38 6.44 9.22 7.76 9.66 45.75 7.15 6.40 7.51 9.43 0.91 13.90 16.09

(b)

(c)

(d)

1 0.78 0.77 0.67 0.31 0.16 2.27 0.82

1 0.82 0.86 0.70 0.49 0.16 1.86 0.95 0.14 0.79 7.08 7.60 4.81 8.29 1.03 6.41 0.52 9.66 40.85 7.85 7.82 7.26 8.50 0.84 7.74 13.55

1 0.89 0.92 0.76 0.51 0.18 2.15 1.00 0.14 0.79 7.32 8.20 5.60 8.70 1.34 7.17 0.67 9.66 43.95 8.31 8.14 7.74 10.19 0.87 11.98 15.20

1.97 3.99 5.49 1.03 5.07 0.52 31.03 9.29 8.32 9.76 12.26 0.55 7.74 14.30

Van Zanden 1 0.80 0.80 0.35

0.84

7.00

1.30 4.60

6 5.5

0.50 10

Coeff. 1 0.9 0.9 0.75 0.50 0.15 2 1 0.15 0.5 10 8 5 9 1.3 7 0.7 6 40 9 8 8 9 1 13 14

Source: See text. a Greasy wool. b Dressed weight.

d’Agriculture. They provide twenty-three sets of prices for sixteen countries, which in Table A.4 are normalized to the price of wheat (rice in Indonesia). Table A.5 sums up the data of the previous table in a compact form. The column “Van Zanden” shows the set of prices used by the author in his estimate of productivity growth in Europe (Van Zanden, 1988, Table 1). Columns a and b show averages for free-trade and protectionist countries respectively (Table A.4 (i) and (ii)). Column c is the average of all the sixteen countries, while column d takes into account, quite crudely, the effect of protection on wheat by increasing all prices by 30% in the protectionist countries.

The Growth of World Agricultural Production, 1800–1938

165

Table A.6. Estimates of Gross Output and Value Added, in 1913 in Wheat Units. Gross Output

Value Added

Argentina Australia Austria Hungary Belgium Canada Chile Denmark Finland France Germany Greece India Indonesia Japan Italy Netherlands Portugal Russia Spain Sweden Switzerland U.K. USA Uruguay

22805 16518 19869 23878 5151 15721 2752 7978 2153 44063 82962 1562 92144 18212 16040 28123 17169 1724 90877 12628 6604 3468 25506 151743 3885

19689 13502 18372 22079 3265 13573 2615 4825 1790 38775 75923 1437 87863 18032 13834 26076 9270 1586 86333 11875 6604 3295 17152 127031 3691

Total

713535

604669

Gross Output

Value Added

Serbia Bulgaria Norway Romania Cyprus China Indochina Korea Burma Philippines Thailand Taiwan Turkey Algeria Egypt Madagascar Morocco Sierra Leone South Africa Tunisia Zimbabwe Fiji New Zealand Mexico Cuba

1745 3183 1772 6265 297 183410 6896 2971 7842 2583 3375 1665 15320 6721 5919 1622 1113 152 3412 2559 135 230 3757 4785 5612

1658 3024 1684 5952 282 174240 6551 2822 7450 2453 3207 1582 14554 6385 5623 1541 1057 144 3241 2431 128 219 3569 4545 5331

Total

273339

259672

Source: See text.

The prices used to calculate the value of output (“coefficients”) are, in most case, those of column d suitably rounded. There are exceptions, such as tea and beef. The former is inspired by the relative price in Indonesia, while the coefficient for beef is higher than the country averages because this latter is affected by very low prices in Argentina and because the total output includes veal, which cost more than beef. (iv) Finally, the Value Added in wheat units for each country is computed by multiplying the gross output by the VA/GSP ratio in 1913 according to the national estimates.

166

GIOVANNI FEDERICO

APPENDIX B THE COUNTRY SERIES: SOURCES AND METHODS Argentina The main source is Cortes-Conde (1997, quadro A.1), who provides yearly data from 1875 to 1935 on the GDP of crops and livestock (including fisheries). The two series are combined in an index of agricultural output by weighting with the livestock/crops shares in 1913 from Diaz Alejandro (1970, Table 19). The total VA series is extrapolated forward to 1939 with the estimates from the Banco Central de Argentina (Diaz Alejandro, 1970, Table 17) and backwards to 1870–1875 according to the rate of growth of the cattle stock from 1875 to 1882 (Mitchell, 1998b, Table c5). Livestock products accounted for more than 90% of output in 1875. The 1913 GDP at current prices is estimated by deflating the figure by Diaz-Alejandro with the index of agricultural prices from IEERAL (1986, Table 10). The gross output is computed by dividing the GDP by the VA/GSP series for Canada. The share of livestock for 1875–1935 is obtained as a by-product of the estimation of production. The share is assumed constant in 1870–1874, while the share in 1935–1938 is calculated by extrapolating the 1920–1935 downward trend. Australia The series for GDP are obtained by joining together the series by Butlin-Sinclair (1986, Table 1) and Haig (2001). The former provide figures at current prices for 1828–1860, the latter at constant prices for 1861–1938. The Butlin-Sinclair figures are converted into constant prices with the implicit GDP deflator from Butlin (1986, Table 8).49 The two series are linked together by assuming that, from 1860 to 1861, prices fell by 1% as much as in the United Kingdom. The gross output is then computed multiplying Haig’s data by the GDP/GSP ratio from Butlin (1962).50 Finally, the estimates for 1913 are converted into current price using the price series from Butlin (1962, Table 267). The share of livestock products is taken from Butlin, as a sum of “dairying” and “pastoral.” Austria-Hungary All the data for pre-1913 Austria-Hungary are taken from the recent estimates of a new set of national accounts by M. S. Schultze (2000). The series for “Austria” and

The Growth of World Agricultural Production, 1800–1938

167

“Hungary” (at 1913 boundaries) after the war are obtained as a weighted average of indices of VA of four successor states, Austria, Hungary, Yugoslavia and Czechoslovakia at their 1919 boundaries. The yearly data are taken from Kausel et al. (1965, p. 37) for Austria, Pryor et al. (1971, Table 3) for Czechoslovakia, Eckstein (1955, Tables 1 and 2) for Hungary and Vinsky (1961) for Yugoslavia.51 The weights for Austria are taken from Waizner (1928, Table III): in 1911–1913 (post-1919) Austria accounted for 20.4% of (pre-1913) Austrian agricultural VA, Czechoslovakia for 47.5%, and the territories then transferred to Yugoslavia, Poland, Italy and Romania for 5.2, 19.5, 5.2, and 2.6% respectively. As no regional series for the last three countries are available, the index is calculated as a weighted average of the series for Austria (weight 0.281), Czechoslovakia (weight 0.647) and Yugoslavia (weight 0.071) only. There is no comparable source on regional output for (pre-1913) Hungary. However, (post-1919) Hungary accounted for 45.8% of the combined output of Yugoslavia and Hungary in 1935–1939 (Moore, 1945, Table 5) and for 47% of total agricultural land (arable and tree-crops) in 1925–1926 (Institute International d’Agriculture, 1925–26). The index for (pre-1913) Hungary is thus calculated as a weighted average of the indices for Hungary (weight 0.45) and Yugoslavia (weight 0.55). All the estimates quoted so far data refer to Value Added. The gross output has to be calculated by multiplying the VA by the inverse of the VA/GSP ratio. According to Waizner (1928, Table III), the VA accounted for 97.5% of GSP in Austria in 1913, while Komlos (1983, Table D7), suggests a constant 93% ratio for Hungary for the whole period 1885–1913. Neither figure is really plausible. The figure for Austria seems too high, while Komlos’ assumption of a constant ratio contrasts with the downward trend in all other European countries. It is thus assumed that the VA/GSP ratio fell from 0.95 in the 1850s to 0.90 in the 1950s. These are the Portuguese figures, and are quite close to the Italian ones, a country with a similar level of development. Finally, the figures on the composition of gross output before 1913 have been kindly provided by M. Schultze. The shares of livestock products for the inter-war period are assumed to have remained constant at the 1904–1913 level.

Belgium The main source is the very detailed reconstruction by Blomme (1992a, b). He provides a series for “agricultural output” (i.e. gross output, as explained on p. 22) and Value Added since 1877 (with a break in 1914–1918). The former are both at current (Tables 22 and 42) and at constant prices (“volume indices” of Tables 57 and 58), and are divided also by major categories of products (arable farming

168

GIOVANNI FEDERICO

livestock and horticulture). In contrast, the data for Value Added are available only at current prices, and the series at constant prices is calculated by double deflating the gross output data with the indices of prices of output (Tables 46 and 47) and inputs (Tables 55 and 56). The 1880–1913 and 1919–1939 series are then linked together by taking the changes in gross output (a 32% fall from 1913 to 1919) and in the VA/GSP (an increase from 0.634 to 0.842 in the same years). The Blomme series are then extrapolated backwards to 1866 superimposing the yearly fluctuations of the old index of agricultural production by Gadisseur (1973, Table 5) to the revised estimates of the growth rate from 1846 to 1878 by Goosens (1992, Table 35).

Canada Urquhart (1993) provides a series of agricultural output (“farm revenue,” Table 1.9) and GNP of agriculture at current prices for the period 1870–1926 (Table 1.1), which are deflated with the implicit price index of agricultural output from McInnis (1986, Table 14.A.2).52 The Value Added from 1926 to 1938 is estimated by extrapolating Urquhart’s figures with an index computed with the data from Historical Statistics of Canada (deflating the GDP at current prices of Table F56–F58 with the index of wholesale price index of Table F49). The source does not report data for the gross output, which is estimated assuming that the ratio VA/GSP had been declining after 1926 at the same rate as before. The share of livestock in gross output is taken from McInnis (1986, Table 14.A.1) until 1926, and from Canada Handbook (Tables 21 and 22) thereafter.

Chile All the data are from the reconstruction of Chilean national accounts of the working group in the Pontificia Universidad of Santiago (Braun et al., 2000). The share of livestock and the GSP are from personal communication by I. Briones.

Denmark The agricultural GDP (from 1818) is taken from Hansen (1974, Table 4). The gross output is estimated by dividing by the Dutch VA/GSP ratio from Knibbe (1994 Erratum). The share of livestock on gross output at current prices is taken from Johansen (1985, Table 2.11).

The Growth of World Agricultural Production, 1800–1938

169

Finland The source for all the data is the book by Hjerrpe (1989). The GDP of agriculture at current prices is from Table 4 and at constant prices (“index volume”) from Table 6. The gross output is calculated assuming that the VA/GSP ratio moved as the Swedish one. The share of livestock is taken from the same source, Table 8.

France All the data are from Toutain (1997). He reports a yearly index from 1815 onwards for the GSP (series V1), series at current prices for both the GDP (series V6) and the GSP (series V10) and an index of agricultural prices (series V5). The figures for inter-war years are reduced by 1.5%, the additional acreage gained by France with the acquisition of Alsace-Lorraine. The data for the share of livestock are from Toutain (1961, Tables 76, 76 bis and 77).

Germany The main source is Hoffmann (1965), who provides series of gross output (ii Table 58) and VA (ii Table 64) both at current and constant (1913) prices. The series need two adjustments. First, the data for 1920–1924 are missing, and thus they are estimated by extrapolating the 1925 production backwards with a production index. The latter is obtained as a weighted average of indices of the gross output of crops, meat and “other livestock products” (i.e. milk), using the shares of GSP in 1925–1927 as weights. The index for crops is computed by multiplying the gross output of wheat, rye, and potatoes (divided by half) from Mitchell (1998c, Table C2) by the price ratios in Italy in the same years (ISTA, 1958), normalized to wheat. The indices for meat and “other livestock products” are calculated as the number of animals in 1920–1924 (Mitchell, 1998c, Table C5) times their average productivity in 1925–1927 (production from Hoffmann, 1965, ii Table 55, stock from Mitchell, 1998c, Table C5). Second, the Hoffman data are at current borders, and thus they omit the production of the areas lost to Poland after World War I – some 15% of its pre-war acreage in arable and tree-crops (Institute Internationale d’Agriculture, 1909 a` 1921, Table 4). The Hoffman figures for 1925–1938 are thus increased by the same amount. It is thus implicitly assumed that the production of the lost areas moved in parallel to that of the rest of the country.

170

GIOVANNI FEDERICO

Greece Greek agricultural production has recently been re-estimated by Petmezas (1999 and personal communication). He provides a series of the gross output of agriculture from 1848 and on the share of livestock (Table 7). The estimate of GDP is obtained by assuming the same trend in the VA/GSP ratio as in Portugal. Greece changed its boundaries many times in the period under consideration: the agricultural production is adjusted to 1913 boundaries, according to the total acreage of the country from Petmezas (1999, Table 7).

India The series is obtained by linking the estimates by Heston (1983, Table 4.3.A) for the period to 1899 and by Sivasubramonian (2000, Table 6.10) for the years 1900–1938. Some missing years in the 1870s have been interpolated according to the population (Heston, 1983, Table 4.1). Sivasubramoninan data refer to the Value Added: the gross output is computed adding the figures for “repairs and maintenance” and “marketing costs” (which includes expenditure in fertilizers) from Table 3.7. Both Heston and Sivasubramonian report the production of crops and livestock separately, so it is possible to calculate the relative share of gross output. The implicit level of the two series, when overlapping, differs quite substantially and they cannot be spliced. Thus, for production data, the estimates by Sivasubramonian have been extrapolated backwards to 1870 with the trend from Heston.

Indonesia Van der Eng (1996, Table A.4) provides figures of total GDP at constant (1960) prices, also divided by major items (“food crops”, “animal husbandry”, “cash crops”, “estate crops”) for the period 1880–1939. The author has kindly communicated his estimate of GDP in 1913 at current prices, which is raised by 5% to take some missing items such as fruit, vegetables and poultry into account (Van der Eng, 1996, p. 361). Gross output in 1913 is assumed to have been 1% higher than Value Added, as the expenditures outside the agricultural sector were minimal (Van der Eng, 1996, pp. 256–257). Finally, the two series have been extrapolated backwards to 1870 with the population growth.53

The Growth of World Agricultural Production, 1800–1938

171

Italy The standard reconstruction of Italy’s national accounts by sector of origin at constant (1938) prices is Ercolani (1969, Table 13.1.1). He builds on the previous work by the Italian Central Statistical Bureau (ISTA, 1957), which estimated GDP and GSP at current and constant prices. The series for the period to 1913 has long been controversial, and Federico (2003) provides an alternative estimate of gross output at current borders. It is possible to calculate a series of gross output and VA at 1911 boundaries by interpolating and extrapolating the benchmark estimates for 1891 and 1911 of the VA/GSP ratios (Federico, 2000) and of the ratio current/1951 borders (ISTA, 1957). The VA series after 1913 are obtained from Ercolani, by deducting forestry and fishing according to the proportion of the original ISTAT (1957, Tables 8 and 9) estimates. The gross output is calculated dividing this Ercolani series by the VA/GSP ratio from ISTAT. The original ISTAT publication is also the source of the data on the monetary value of GSP and of VA in 1913, and of the yearly figures of the share of livestock products in gross output.54

Japan All the data are taken from Okhawa-Shinohara (1979, Tables A16 and A17). It reproduces the estimates of the LTES (Long Term Economic statistics) project. The missing data for 1870–1873 are computed by extrapolating backwards the 1874 production according to population growth (Maddison, 1995, Table A-3a).

Netherlands All series for the Netherlands (GSP, VA and share of livestock on output) are a combination of two estimates by Van Zanden (2000) for the period 1807–1913 and Knibbe (1994, Erratum) for the period 1914–1938.

Portugal The source of the data is an article by Lains – Silveira Sousa (1998), supplemented by personal communication from the authors on the period 1913–1939. They estimate a Laspeyres index of agricultural GSP with the nine most important products

172

GIOVANNI FEDERICO

(Table A.2). The corresponding series of GDP is obtained by assuming that the VA/GSP ratio fell linearly from 0.95 in 1848 to 0.90 in 1960 (1998 fn. 40). The final step is the calculation of the value of gross output and GDP in 1913 by extrapolating the figures for 1900–1909 (Table 4) and by adding 13.4%, the share of omitted products in the same years (p. 956). The share of livestock products is calculated interpolating Lains’ estimates for 1861–1870, 1900–1909 and 1935–1936.

Russia No single GDP or GSP series is available for the whole period. Thus, a new series has to be estimated, with different procedures for Russia (to 1913) and the Soviet Union. The literature on agricultural production is quite abundant, but sometimes confusing, if not positively misleading. The standard work on Imperial Russian national accounts is the book by Gregory (1982). Unfortunately, he does not report data on Value Added by sector, even if Table 3.6 proves that he has estimated them, at least for some years. Thus, following Gregory’s suggestions (1982, p. 73), agricultural GSP is computed as a weighted average of three series, the index of the production of food crops by Gregory (1982, Table D.1, series G2), the series of the production of technical crops by Goldsmith (1961, Table 3) and the value of livestock herds by Gregory (1982, Table H.1 B).55 Then, the GSP figures are extrapolated backward to 1870 separately for crops, industrial crops and livestock – respectively, with the index of the production of “major grain and potatoes” and of “technical crops” from Goldsmith (1960, Table 1) and with the number of animals from Mitchell (1998c, Table C5).56 The weights are calculated from the data on the value of GSP in 1913 by type (food crops, industrial crops and livestock) from Falkus (1968).57 As stated in the text, the estimation of trends in production during the Soviet period is a very difficult and sensitive issue. Here, we use the most recent work by Allen (2002 and personal communication), who provides a series of gross output from 1924 to 1939 at interwar borders linked to 1913.58 The Allen series is extrapolated back to 1920 with the official figures, the only available data for 1920–1923.59 The gross output series are then converted into VA by assuming that the VA/GSP ratio has declined from 0.97 to 0.95 in 1913 and in 1920, to 0.94 in 1932 and to 0.90 to 1939. Finally, the share of livestock from 1870 to 1913 is obtained by extrapolating the 1913 shares backwards to 1870 with the Gregory/Goldsmith index and forward to 1938 with an index of livestock production obtained splicing together the official data for 1920–1927 and the figures for 1928–1938 by Wheatcroft-Davies (1994b).

The Growth of World Agricultural Production, 1800–1938

173

Spain L. Prados has been working on the reconstruction of national accounts for many years. He has provided his most recent estimates at constant and current prices for GDP and gross output (Prados, 2004). The share of livestock is estimated interpolating the shares from Prados (1993, Table 1).

Sweden The figures are taken from Schon (1995) – the gross output from Table J6 and the value added from Table J1. The share of livestock until 1931 is from Lindahl et al. (1937, Table 2), and thereafter it is assumed as constant.

Switzerland The data are taken from Ritzmann-Blickenstorfer and David (undated) and personal communication. The GDP is the sum of agriculture and horticulture. The gross output is computed assuming that the VA/GSP ratio fell as much as in France.

The United Kingdom The standard reconstruction of British historical national accounts, by Feinstein (1972) provides an index number (1913 = 100) of GDP for agriculture, forestry and fishing at constant prices at current boundaries (1972, Table 8.1). For the years 1855–1913, Feinstein quotes as his source a mimeo by Lewis, who later published a series of GDP at 1907 prices (Lewis, 1979, Table A.3).60 Quite strangely, the two series are perfectly identical from 1855 to 1912, and then diverge sharply in the last year: according to Lewis, agricultural production fell by 5% from 1912 to 1913, while, according to Feinstein, it remained constant. This latter trend seems more plausible – as the production of cereals and potatoes increased by 10–20%, that of milk remained stable and only the production of meat fell, albeit by a mere 3.6%. Thus, the index will use Feinstein’s figures. After 1920, Feinstein uses “official statistics,” and the series excludes Eire, which became independent in 1921. In 1920, Southern Ireland accounted for about 23% of all-U.K. agricultural output.61 An index of the United Kingdom at 1913 boundaries is obtained as a weighted average of Feinstein’s data for Great Britain (at 1921 boundaries) and Drescher’s (1955) ones for Eire. The latter series stops in 1930: the figures for 1931–1938

174

GIOVANNI FEDERICO

are estimated by extrapolating the 1930 level with indices of the physical output of crops (an average of wheat, barley, oats and potatoes) and livestock (butter), assuming that livestock accounted for 78% of total output.62 The GSP at constant prices is then obtained by dividing the GDP series by the VA/GSP ratio from Ojala (1952, pp. 208–209). The figures for 1913 are calculated adjusting the Ojala (1952) estimate of gross output and GDP for 1911–1913. The share of livestock is also taken (with interpolation) from Ojala. The alternative series by Turner (2000, Table 38.8), which stops in 1914, yields a somewhat lower share, but the trend is very similar.

United States The official data of national accounts, published in Historical Statistics of the United States, start in 1910 (U.S. Bureau of the Census, 1975). The gross output for crops and livestock is the sum of cash receipts (series K266 and K267) and home consumption (K269), net of the intra-sectoral expenditures for feed (K273), livestock (K274) and seed (K275).63 The total revenues (and hence the implicit GSP) thus differ from the “realized gross farm income” (K264), which includes subsidies after 1931 (K268) and rent of farm dwellings (K270).64 Then, the series of GDP is computed by deducting from the gross output the expenditures for fertilizers (K276), repairs (K277) and miscellaneous items (K280). Both gross output and GDP are transformed into constant (1913) prices by double deflating the indices of prices received (separate for crops and livestock K345-K346) and paid by farmers (K348). Both series are then extrapolated backwards to 1869. Gross output is extrapolated according to the Fisher index of total output by Strauss-Bean (1940, Table 61). The GDP is computed by multiplying the result by a series of the VA/GSP ratio obtained interpolating the benchmark figures from Towne-Rasmussen (1960) for 1860, 1880, 1890 and 1900 and from U.S. Bureau of the Census (1975) for 1910. The share of livestock products from 1869 to 1909 is also obtained with linear interpolation, using the same sources.

Uruguay All the data are taken from B´ertola (1998). The GSP is a weighted average of the two indices of “volumen fisico” for crops and livestock, using the currentprice value of gross output from Tables 3 and 4 as weights. The VA/GSP ratio is

The Growth of World Agricultural Production, 1800–1938

175

assumed, as for Argentina, equal to that of Canada. The data for 1937 and 1938 are interpolated with the Value Added for the whole economy.

APPENDIX C World Population The population data for the twenty-five countries in the sample (at current boundaries) are taken from Mitchell (1998a, b and c), McEvedy-Jones (1978), United Nations (1952) for 1920 and 1938, Institute Internationale d’agriculture 1939–1940 for 1937, Maddison (1991, Tables B2 and B3), Maddison (1995, Table A.3) and some additional country sources.65 When necessary, figures have been obtained by linear interpolation. There are several estimates of the world population at different dates, which are reported for the reader’s ease in Table C.1. As one can see, they broadly agree, even if many figures are pure guesstimates. The population data (Table C.2) are thus taken from Maddison for 1870 and 1913, the United Nations for 1920 and the Institute Internationale d’Agriculture for 1938 (the 1937 figure is increased by 1.5% to take account of the natural increase of population).

176

Table C.1. Estimates of World Population (Millions). 1850 Biraben Europe North America South Central America Africa Asia Oceania Europe and Western Offshoots Total: World

288 25 34 102 790 2

1870

Biraben

279 34 25 81 781 1

422 90 75 138 903 6

40 91 765

324 57 34 93 817 2

Clark 411 81 63 122 985 6

1913 Mc Evedy Maddison 415 95 50 110 946 7

81 125 978

375 1.241

Mc Evedy

1.201

513 140 81 140 1.107 10

1.270 1930

UN 531 135 109 155 1.047 10

532 135 109 157 1.141 10

Clark 487 117 91 140 1.072 9

UN 485 117 92 141 971 9

608 1.326

1.634

1937 Clark

1920

IIA 557 159 104 168 1.138 11

1.668

1.622

1940

1.791

1.916

1.813

1950

UN

Clark

551 146 131 172 1.202 11

573 146 131 176 1.233 11

Biraben 575 166 164 219 1.393 13

UN 547 172 167 221 1.402 13

Mc Evedy Maddison 576 191 134 205 1.389 14

166 228 1.382 749

1.990

1.987

2.084

2.137

2.214

2.270

2.530

2.522

2.509

2.525

Sources: Biraben (1979), McEvedy-Jones (1978), Clark (1977), United Nations 1920–1940 (1952, Table 1A) (average of maximum and minimum estimates), 1950 UN demographic yearbook 1999; Maddison (2001, Table A-c).

GIOVANNI FEDERICO

Total: World

1900

Mc Evedy Maddison Mc Evedy

1925

Europe North America South Central America Africa Asia Oceania Europe and Western Offshoots

1875

The Growth of World Agricultural Production, 1800–1938

177

Table C.2. Population Estimates (Millions).

1870 1913 1920 1930 1938

Sample

World

%

643 985 986 1111 1202

1270 1791 1813 1987 2169

50.6 55.0 54.4 55.9 55.4

178

STATISTICAL APPENDIX Table D.1. Series (1913 = 100). Output

Output Livestock

Crops

GDP, 1913 = 100 Europe

Northwestern Europe

Southern Europe

Eastern Europe

Asia

South America

Regions of Western Settlement

53.0 52.2 53.6 53.2 56.8 56.6 55.6 58.4 59.7 57.6 59.8 60.2 62.9 63.7 64.9 65.4 65.1 67.5 68.6 66.7 69.8 66.9 70.3

51.5 50.9 52.1 51.8 55.1 55.1 54.2 56.7 57.9 56.0 58.2 58.6 61.1 62.1 63.1 63.6 63.4 65.7 66.7 65.1 68.0 65.6 68.9

44.8 44.9 46.3 46.4 48.6 50.4 50.2 51.4 52.6 51.1 53.4 53.0 54.7 57.1 58.5 58.7 59.4 60.0 61.8 62.6 63.5 64.0 64.4

55.3 54.7 56.0 55.4 58.8 58.9 57.5 60.1 61.0 59.1 61.3 61.9 64.6 65.2 66.1 66.9 66.5 69.1 70.0 68.0 71.4 68.3 72.0

58.4 56.0 57.6 56.5 62.9 61.8 58.4 61.6 63.0 57.9 60.4 62.4 64.1 65.1 65.9 65.2 65.0 67.8 69.0 64.7 68.1 65.2 69.6

70.3 67.9 70.3 66.5 77.5 77.8 70.6 71.6 73.6 65.5 70.3 71.2 73.0 75.5 76.2 77.0 76.7 76.5 77.4 75.8 79.0 76.3 79.8

62.9 62.7 66.0 67.2 67.0 67.1 66.5 70.4 71.4 68.4 72.2 72.4 73.1 73.6 72.6 72.1 74.2 74.6 75.6 72.4 73.4 76.0 79.3

41.7 38.4 38.3 39.3 43.3 39.7 39.8 45.2 46.3 43.8 42.9 47.0 49.1 48.6 50.2 47.7 46.6 54.0 55.6 47.8 52.5 46.7 52.8

64.9 65.8 66.8 67.0 67.1 67.5 67.4 67.7 67.4 68.5 68.7 69.1 73.0 72.9 73.0 77.5 75.5 80.2 80.8 77.5 84.3 74.0 82.2

13.4 14.2 15.0 16.1 15.6 15.4 16.3 16.6 16.9 18.1 18.8 19.2 22.3 23.8 24.8 26.0 26.8 28.0 30.0 27.1 28.5 32.5 36.5

34.1 34.8 36.3 36.9 37.4 38.8 41.8 45.6 47.6 49.3 52.3 49.5 53.2 54.4 57.0 57.1 57.7 58.0 59.1 63.1 62.7 65.4 63.0

GIOVANNI FEDERICO

1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892

GDP

71.2 72.9 73.9 74.2 76.0 80.7 79.6 81.8 80.8 83.6 85.0 86.1 86.2 88.7 88.4 91.1 94.3 93.7 94.4 98.6 100.0

65.9 68.1 70.1 72.9 73.5 76.4 78.2 79.0 80.3 80.6 81.1 83.3 85.2 87.6 87.6 90.0 91.7 93.0 95.0 96.8 100.0

74.2 75.9 76.3 75.4 78.5 83.3 81.3 83.7 82.6 85.6 86.8 87.4 87.3 90.4 89.6 91.8 96.0 94.9 95.9 99.0 100.0

74.5 75.7 76.2 77.8 72.9 79.3 79.8 82.4 79.7 84.4 84.5 85.8 85.9 86.3 89.2 90.9 92.8 90.7 89.9 95.7 100.0

82.9 83.9 83.6 86.8 81.9 87.0 90.1 94.2 89.7 88.8 89.9 94.2 93.3 91.6 95.6 98.0 97.7 92.3 95.5 97.8 100.0

85.3 88.7 93.0 94.9

87.8 92.1 96.9 98.5

88.2 93.5 97.4 100.8

88.6 92.8 97.6 98.7

75.5 75.3 81.4 84.9

80.4 82.3 86.4 86.4

77.7 79.1 79.7 76.5 79.6 82.2 80.4 82.3 88.1 87.1 86.8 87.9 88.6 92.1 93.8 95.4 97.3 90.7 97.0 92.8 100.0 92.4 88.9 94.7 94.2 95.4 92.9 97.9 96.1 101.8 105.9

62.8 64.0 65.7 67.5 58.9 68.6 67.0 68.2 63.6 77.8 76.9 74.6 75.7 77.1 79.4 80.2 84.7 88.8 79.9 94.4 100.0

59.3 57.1 66.0 73.4

83.9 86.4 83.5 74.6 91.2 94.2 84.3 87.8 86.9 91.9 94.6 93.9 89.9 96.2 90.1 93.4 105.3 104.6 103.6 103.2 100.0 105.2 107.7 112.4 110.7 94.5 112.8 98.9 108.2 111.9 106.7

39.0 45.9 51.5 49.1 43.7 44.0 52.8 48.6 56.3 53.6 67.7 75.9 73.9 74.3 70.2 88.5 85.2 80.2 69.8 101.9 100.0 90.1 106.0 93.5 66.9 108.5 105.5 111.3 111.6 112.7 120.1

61.9 64.2 68.2 72.3 77.3 80.3 81.5 81.8 82.3 80.2 84.7 85.9 87.9 91.9 87.6 90.0 89.4 90.8 95.3 99.1 100.0 95.6 105.8 104.9 97.0 103.1 105.5 94.3 100.3 101.4 105.6

179

72.9 74.7 75.5 75.6 77.7 82.5 81.1 83.3 81.8 84.7 86.5 87.4 87.1 89.7 88.9 91.1 94.3 93.5 94.1 98.1 100.0

The Growth of World Agricultural Production, 1800–1938

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

180

Table D.1. (Continued ) Output

Output Livestock

Crops

GDP, 1913 = 100 Europe

Northwestern Europe

Southern Europe

Eastern Europe

Asia

South America

Regions of Western Settlement

98.6 102.7 103.3 107.8 108.8 112.4 109.2 110.8 109.9 112.8 111.1 109.8 110.6 114.9 116.6

102.5 106.4 107.1 110.8 113.1 116.9 113.4 114.9 115.1 116.9 116.2 115.0 114.8 121.0 122.9

105.2 108.4 112.2 114.9 117.3 121.5 119.6 120.3 117.7 117.9 117.7 117.1 118.8 121.2 129.3

101.8 106.4 106.4 109.8 111.6 115.9 112.6 113.4 114.0 115.7 114.5 113.2 113.4 118.8 120.4

87.0 95.7 94.6 100.6 103.3 108.4 104.1 104.8 102.6 106.5 106.5 107.3 102.7 111.6 112.6

90.1 93.0 88.8 98.2 101.6 104.9 102.8 107.5 105.6 114.3 114.4 110.4 112.5 108.1 116.0

102.2 111.5 108.0 108.5 107.0 117.2 104.2 109.5 120.2 109.5 111.0 115.1 94.2 107.2 106.4

76.4 91.8 95.7 99.8 103.5 108.8 105.7 99.3 90.9 95.5 94.8 100.0 94.5 117.9 111.2

109.9 109.6 110.1 110.7 113.0 115.1 117.3 114.3 115.8 118.0 113.3 114.2 122.4 121.1 114.3

144.1 125.1 146.7 153.1 163.2 162.6 141.0 159.7 155.5 148.1 167.1 179.2 169.7 191.3 178.4

112.5 111.0 114.7 119.1 115.5 117.5 112.2 119.5 118.8 121.0 117.0 110.4 116.2 114.1 123.3

Sources: See text and Appendix B. North-Western Europe the United Kingdom, France, Sweden, Denmark, Belgium, the Netherlands, Germany, Finland, Switzerland; Southern Europe Italy, Greece, Spain, Portugal; Eastern Europe Austria, Hungary and Russia; Asia Japan, India, Indonesia; Western Settlement Canada, Australia and USA; South America: Argentina, Uruguay and Chile.

GIOVANNI FEDERICO

1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938

GDP

The Growth of World Agricultural Production, 1800–1938

181

Table D.2. Rates of Change in GDP, by Country.

Argentina Australia Austria Hungary Belgium Canada Chile Denmark Finland France Germany Greece India Indonesia Italy Japan Netherlands Portugal Russia Spain Sweden Switzerland U.K. USA Uruguay

1870–1938

1870–1913

1913–1938

Column Difference

4.41 2.83 1.09 1.46 0.62 2.00 1.86 1.87 1.26 0.58 0.91 1.53 0.73 1.97 0.86 1.60 1.31 0.87 1.79 0.69 1.03 0.72 0.58 1.12 3.16

6.07 3.36 1.44 2.26 0.76 2.86 1.56 1.62 1.56 0.62 1.56 2.12 0.96 1.79 1.14 1.73 0.65 0.54 2.24 0.46 0.96 0.70 0.00 1.70 2.91

2.89 2.31 1.52 0.07 0.02 −1.06 1.88 3.24 1.89 0.90 0.02 3.56 0.31 1.92 0.58 0.75 2.47 3.17 0.02 −0.06 1.49 0.83 1.52 0.56 5.25

*** *** * *** *** *** a * a a *** *** *** a *** ** *** *** *** *** a a ** ** ***

Note: Column Difference: test of the difference between the growth rates in 1870–1913 and 1913–1938. a Not significant. ∗ Significantly different from zero at 10%. ∗∗ Significantly different from zero at 5%. ∗∗∗ Significantly different from zero at 1%.

THE GREAT DEPRESSION AS A CREDIT BOOM GONE WRONG Barry Eichengreen and Kris J. Mitchener ABSTRACT The experience of the 1990s renewed economists’ interest in the role of credit in macroeconomic fluctuations. The locus classicus of the credit-boom view of economic cycles is the expansion of the 1920s and the Great Depression. In this paper we ask how well quantitative measures of the credit boom phenomenon can explain the uneven expansion of the 1920s and the slump of the 1930s. We complement this macroeconomic analysis with three sectoral studies that shed further light on the explanatory power of the credit boom interpretation: the property market, consumer durables industries, and high-tech sectors. We conclude that the credit boom view provides a useful perspective on both the boom of the 1920s and the subsequent slump. In particular, it directs attention to the role played by the structure of the financial sector and the interaction of finance and innovation. The credit boom and its ultimate impact were especially pronounced where the organization and history of the financial sector led intermediaries to compete aggressively in providing credit. And the impact on financial markets and the economy was particularly evident in countries that saw the development of new network technologies with commercial potential that in practice took considerable time to be realized. In addition, the structure and management of the monetary regime mattered importantly. The procyclical character of the foreign exchange component of global international reserves and the

Research in Economic History Research in Economic History, Volume 22, 183–237 Copyright © 2004 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(04)22004-3

183

184

BARRY EICHENGREEN AND KRIS J. MITCHENER

failure of domestic monetary authorities to use stable policy rules to guide the more discretionary approach to monetary management that replaced the more rigid rules-based gold standard of the earlier era are key for explaining the developments in credit markets that helped to set the stage for the Great Depression.

1. INTRODUCTION The experience of the 1990s, especially though not exclusively in the United States, renewed economists’ interest in the role of credit in macroeconomic fluctuations.1 Among the insights of this view is that not just money but also credit matters for macroeconomic and financial conditions. Not just banks but also non-bank financial intermediaries and securities markets play an important role in the provision of credit to households and firms. Not just macroeconomic policy but also the structure, regulation and response of the financial system shape the development of financial conditions and thereby macroeconomic dynamics. The policy implication drawn by some is that central banks should not simply set monetary policy with an eye toward inflation; they should also attend to conditions in credit markets and contemplate preemptive action to prevent the development of excesses that threaten economic stability even if there is no sign of inflationary pressure. Economists at the Bank for International Settlements (BIS) have been forceful proponents of this position, which for want of a better label is referred to as the BIS view.2 A capsule account of the role of credit in macroeconomic cycles, as informed by the experience of the 1990s, would go something like this. There is first an upswing in economic activity. As the economy expands, banks and financial markets provide an expanding volume of credit to finance the growth of both consumption and investment, particularly where regulation is lax and competition among bank and non-bank financial intermediaries is intense. Whether because the exchange rate is pegged or for other reasons such as a positive supply shock, upward pressure on wholesale and retail prices is subdued. Hence, the central bank has no obvious reason to tighten and stem the growth of money and credit, leading to a further expansion of output and further increase in credit. Higher property and securities prices encourage investment activity, especially in interest-sensitive activities like construction. But, as lending expands, increasingly risky investments are underwritten. The demand for risky investments rises with the supply, since, in the prevailing environment of stable prices, nominal interest rates and therefore yields on safe assets are low. In search of yield, investors dabble increasingly in risky investments. Their appetite for risk is stronger still to the extent that these trends coincide with the development of new technologies,

The Great Depression as a Credit Boom Gone Wrong

185

in particular network technologies of promising but uncertain commercial potential. Eventually, all this construction and investment activity, together with the wealth effect on consumption, produces signs of inflationary pressure, causing the central bank to tighten. The financial bubble is pricked and, as asset prices decline, the economy is left with an overhang of ill-designed, non-viable investment projects, distressed banks, and heavily indebted households and firms, aggravating the subsequent downturn. No single policy implication necessarily flows from this story, but some readers will conclude that the monetary authorities should respond preemptively to the rise in asset prices. Central banks should not be misled, in this view, by the disconnect between asset price inflation and consumer price inflation. They should respond to the inflation of asset prices by reining in credit and preventing the expansion from taking a form that ultimately renders subsequent difficulties more severe. This tale from the 1990s has obvious appeal for historians of the 1920s. The 1920s was a decade of expansion, reflecting recovery from World War I, new information and communications technologies like radio, and new processes like motor vehicle production using assembly-line methods. Accounts of the ‘twenties in the United States (such as Kindleberger, 1973) emphasize the ready availability of credit, reflecting the ample gold reserves accumulated by the country during World War I, the stance of Federal Reserve policies, and financial innovations ranging from the development of the modern investment trust to consumer credit tied to purchases of durable goods like automobiles. Credit fueled a real estate boom in 1925, a Wall Street boom in 1928–1929, and a consumer durables spending spree spanning the second half of the 1920s. That these booms developed under the fixed exchange rates of the gold standard meant that they generated little inflationary pressure at home and that their effects were transmitted to the rest of the world. Absent overt signs of inflation, the Fed had no reason to raise the official short-term rate. Eventually, however, the Fed and other central banks grew increasingly restive over what they perceived as speculative excesses in financial markets and a growing incidence of malfeasance and graft, evident in the activities of Charles Ponzi in Florida, Clarence Hatry in London, and Ivar Kreuger in Stockholm. This concern with the effects of asset-price inflation on the economy led them finally to tighten. Banks passed along the higher cost of additional reserves to their borrowers, and, in the U.S. case, they further felt direct pressure to limit their lending to securities market participants. By this time, positions – stock market positions in particular – were highly leveraged; as a result, borrowers experienced severe financial strain when credit tightened, leading them to compress their spending, and consumption and investment turned down. Ultimately, the resulting

186

BARRY EICHENGREEN AND KRIS J. MITCHENER

deflation became sufficiently severe to threaten the stability of the financial system and the economy more generally.3 While this credit boom interpretation has multiple precursors in the qualitative literature on the Depression, its validity and explanatory power have not been assessed in a systematic, quantitative way. Doing so is our goal in this paper. We ask how well quantitative measures of the credit boom phenomenon can explain the uneven expansion of the 1920s and the slump of the 1930s. In Section 2 we consider scholarly precursors to the modern credit boom view, such as Georgist theory, the Austrian School, the Minsky-Kindleberger financial-instability thesis, and the literature which attributes the Great Depression to a credit-fueled stock market bubble. Sections 3 and 4 construct quantitative indicators of the development of the credit boom for sixteen countries and ask whether the height of the boom was positively associated with the depth of the subsequent slump. In Section 5 we complement this macroeconomic analysis with three sectoral studies that shed further light on the explanatory power of the credit boom interpretation: the property market (where recent experience suggests that credit-boom dynamics should have been particularly apparent), consumer durables industries (where financial innovation played a particularly important role in the 1920s), and high-tech sectors (where authors like Perez, 2002, suggest that the imprint of the credit boom should have been especially pronounced). Obviously, the parallels with the 1990s are never far from our minds. In Section 6 we examine the hypothesis, echoing the early Austrian school and advanced recently by The Economist (2002), that credit booms have become more of a problem as the world has moved from the gold standard to more discretionary and elastic monetary regimes. Section 7 summarizes our findings and their implications for modern debates. We find that the credit boom view provides a useful perspective on both the boom of the 1920s and the subsequent slump. In particular, it directs attention to the role played by the structure of the financial sector and the interaction of finance and innovation. The credit boom and its ultimate impact were especially pronounced where the organization and history of the financial sector led intermediaries to compete aggressively in providing credit. And the impact on financial markets and the economy was particularly evident in countries that saw the development of new network technologies with commercial potential that in practice took considerable time to be realized. In addition, the structure and management of the monetary regime mattered importantly. The procyclical character of the foreign exchange component of global international reserves and the failure of domestic monetary authorities to use stable policy rules to guide the more discretionary approach to monetary management that replaced the more rigid, rules-based gold standard of the earlier era are key for explaining the developments in credit markets that helped to set the stage for the Great Depression.

The Great Depression as a Credit Boom Gone Wrong

187

This particular constellation of monetary, financial and technological factors was what allowed the credit boom of the 1920s to develop as it did. We would be prepared to make similar arguments about the macroeconomic cycle of the 1990s. To be clear, we are not necessarily advocating a “credit-centric” interpretation of the Great Depression. Throughout, we attempt to maintain a posture of studied agnosticism regarding its merits, emphasizing the conceptual and methodological obstacles that stand in the way of testing it, including data limitations and problems of observational equivalence with alternative interpretations of the Depression. Indeed, we have vested interests, based on our own prior writings, in the literatures emphasizing other factors in the Depression.4 But the Great Depression was a multi-faceted event that is unlikely to be adequately accounted for by any monocausal explanation. The role of credit should be taken seriously, even by those convinced of the importance of other factors. In this paper we provide an agnostic’s guide to the literature and evidence.

2. SCHOLARLY PRECURSORS The BIS view has several significant precursors in the literature. To the extent that the boom of the ‘twenties and other similar episodes manifested themselves in rising property prices, the credit-boom view was anticipated in the work of Henry George (1879). The Georgist view acknowledged the role of credit in fueling speculation and argues in particular that “speculative advances in land values” are central to causing business cycles. Rising rents induce speculators to purchase land for capital gains rather than for current use, which in turn causes site values to rise in dramatic fashion, setting off further rounds of speculation that eventually erode the profits of firms by increasing mortgage costs and rents. Eventually, these burdens reduce new investment and aggregate demand and bring forth a recession. In effect, the high price of land acts to “lock out labor and capital by landowners” (George, 1879, p. 270). Only as this cycle is unwound and land prices and rents fall does investment pick up again, allowing the economy to recover toward full employment. The Georgist view differs in terms of the timing of the rise of speculation and the decline of investment and economic activity; George and his followers saw the latter as starting to decline even while the property boom was still underway (albeit in its late states), whereas the other views emphasized in the text see investment and demand generally as declining only after the bubble bursts. Another significant precursor is the Austrian interpretation of the cyclical fluctuations, which both anticipated and was informed by the events of the late 1920s. The Austrian view, with roots in the work of Ludwig von Mises (1924)

188

BARRY EICHENGREEN AND KRIS J. MITCHENER

and Friedrich von Hayek (1925), focused on the divergence between the market rate of interest and the natural rate of interest.5 When the market rate fell below the natural rate, Mises and Hayek argued, prices rose and investment boomed.6 The source of that divergence, according to Mises, was the banking system, freed from the disciplining influence of the classical gold standard. Excessive credit creation by banks, both central and commercial, encouraged asset price inflation, fueling consumption and investment. The longer that asset-price inflation was allowed to run, the greater were the depletion of the stock of sound investment projects and the accumulated financial excesses.7 Moreover, the more severe became the subsequent downturn. The credit boom thus contained the seeds of the subsequent crisis. The policy implication was that countries should avoid monetary arrangements that allowed significant divergences between the market and natural rates of interest (in particular, a gold standard of the rigid prewar variety was preferable to the more malleable interwar vintage) and that they should allow the downturn to proceed in order to purge unviable firms and investment projects from the economy, thereby clearing the way for sustainable recovery.8 The definitive application of the Austrian model to the Great Depression was by Lionel Robbins (1934) in a book largely responsible for popularizing the name now attached to this episode.9 Robbins attributed the Depression of the 1930s to the unsustainable credit expansion of the 1920s. Blame for that credit expansion he in turn laid at the doorstep of the Federal Reserve System, which had kept interest rates below the natural rate for too long in the hope that low rates might help Britain surmount its balance-of-payments problems and thereby solidify the reconstructed gold exchange standard, and ultimately on the doorstep of the new gold standard itself, which gave central banks more leeway to manipulate policy. This divergence between market and natural rates stimulated the provision of bank credit, allowing the development of financial excesses which, when revealed, led unavoidably to the downturn, the financial crisis, and the Depression.10 Central banks were misled into inaction by the tendency for the credit boom to stimulate not just aggregate demand but also aggregate supply (through increased production of consumption goods and growing investment in capacity). But, according to Robbins, the quality of much of that additional capacity was inferior. The credit boom had “the qualitative effect of providing a favourable atmosphere for the fraudulent operations of sharks and swindlers,” which meant that neither the expansion of supply nor the high level of asset prices was sustainable and only set the stage for a disruptive crisis.11 Moving from diagnosis to prescription, Robbins recommended against monetary and fiscal measures to counter the downward spiral, insisting that the economy needed to be cleansed of financial and non-financial excesses to set the stage for a sustainable recovery.

The Great Depression as a Credit Boom Gone Wrong

Fig. 1. Stock Prices: Radio Versus the Internet. Source: Global Financial Database and CRISP. 189

190

BARRY EICHENGREEN AND KRIS J. MITCHENER

Another related literature attributes the Great Depression to a bubble in the stock market.12 Galbraith (1972) describes how what he characterizes as a bubble developed in response to the combination of financial innovation and ample credit in an unregulated financial environment. Investor enthusiasm was grounded in the promise of new information and communications technologies; Radio Corporation of America (RCA) was the 1920s equivalent of America Online (AOL). (Figure 1, which superimposes the price of AOL shares in the 1990s on the price of RCA shares in the 1920s, illustrates the parallel.) In addition, however, the enthusiasm of investors was importantly fueled by financial innovation and ample supplies of credit. The 1920s saw the spread from Britain to America of the investment trust, an entity that had existed in England for half a century, but now in a variant that allowed the manager of the trust to buy stocks on margin, raising the fund’s leverage. This anticipates a theme we develop later in the paper – that the consequences of credit expansion and the extent of the boom thereby induced may depend on the structure and regulation of the financial sector. Individual investors were similarly permitted to purchase shares for 10% down, borrowing from their brokers who in turn borrowed from the banks.13 Capital gains on the representative portfolio of nearly 30% in calendar year 1927 and more than 30% in calendar year 1928 encouraged the belief that stocks could only go up.14 Share prices and dividends had broadly moved in tandem through the first quarter of 1928. They diverged thereafter, in response (it has been suggested) to the Fed’s cut in interest rates late in the preceding year (see Fig. 2).15 There are any number of explanations for what happened next, from investment guru Roger Babson’s warning at the National Business Conference that “sooner or later a crash is coming,” to the credit squeeze, to the business deceleration, to protectionist rumblings in the Congress. Whatever the cause, the Great Crash bequeathed a legacy of problems for banks, corporations, and households, which had assumed heavy debt loads and packed their portfolios full of now poorly performing assets. Some policy makers concluded from this experience that central banks should take it upon themselves to deter excessive speculation.16 Finally, there is the Minsky (1986) and Kindleberger (1978) literature on booms, panics and crashes. These authors emphasize asymmetric information and agency problems in financial markets. Among the implications of asymmetric information to which they point are endogenous credit cycles and the fragility of financial systems. Minsky’s emphasis is on the analytics of financial fragility, although he is inspired by the experience of the Great Depression and the Keynesian theorizing to which it gave rise. Kindleberger’s emphasis is on the theory’s applicability to particular historical episodes. Many of these precursors were inspired by and/or attempted to apply their framework to the Great Depression. However, few if any of them analyze the role

The Great Depression as a Credit Boom Gone Wrong

191

Fig. 2. Divergence in U.S. Dividends & Stock Prices. Source: Data are from Rappaport and White (1993, pp. 571–572).

of credit in this macroeconomic cycle in a rigorous quantitative way. It is to this task – and the associated problems – that we now turn.

3. QUANTITATIVE ANALYSIS AND ITS LIMITATIONS A particular difficulty for attempts to analyze the run-up to the Great Depression as a credit boom is that we have only limited historical information on credit itself for this period for a significant number of countries. An ideal measure would include not only loans by financial institutions and corporate stocks and bonds but also consumer credit, mortgages, and trade credit, and other private credit. Goldsmith (1985) estimates this aggregate (total private credit, as the sum of loans by financial institutions, consumer credit, mortgages, corporate stocks and bonds, trade credit and other private credit) for two benchmark years, 1913 and 1929, for a total of nine countries. We display his figures, excluding claims against financial institutions (including currency and deposits) and government debt, on which Goldsmith also includes data, in Table 1. As Table 1 shows, seven of the nine countries experienced increases in total private credit, so measured, between 1913 and 1929, and in four countries (Japan,

192

BARRY EICHENGREEN AND KRIS J. MITCHENER

Table 1. Ratio of Private Credit to GDP.

Belgium Denmark France Germany Japan Norway Switzerland United Kingdom United States

1913

1929

2.34 3.41 2.79 2.59 1.88 2.19 4.30 2.65 2.43

2.51 3.72 2.57 1.64 3.39 3.08 4.49 4.60 4.08

Notes: Computations based on data from Raymond Goldsmith (1985). The years provided by Goldsmith vary slightly for some countries: U.S. (1912); Norway (1930); Japan (1930); and U.K. (1927). Total private credit is the sum of the following national balance sheet items: mortgages, consumer credit, loans by banks and other financial institutions, corporate stocks and bonds, trade credit, and other private credit. The figures exclude government debt.

the U.K., the U.S., and Norway) this increase was quite pronounced. Germany and France are the only two countries where credit declined over the period. That Germany experienced a sharp reduction in the stock of outstanding credit is not surprising, given the massive destruction of credit wrought by the 1923 hyperinflation. France also experienced sharp inflation in the early-to-mid-1920s, consistent with this interpretation.17 Clearly, the problem with using Goldsmith’s estimates is that the second half of the 1920s is confounded with immediate post-World War I disruptions, not to mention the effects of the war itself. We suspect that if a 1925–1926 benchmark were available, the evidence of a credit boom in the second half of the 1920s would be more uniform and clearly evident. Unfortunately, the information needed to construct Goldsmith-like estimates annually for the period 1926–1929, much less surrounding years, is not available for most of these countries. An alternative is to analyze information on money rather than credit. Money has the advantage of being available for a larger sample of countries and for a continuous period of years. The corresponding disadvantage is that money and credit are not precisely the same. Bank liabilities are not the only way that credit to households and firms is financed; firms, for example, can also obtain credit through securities markets. To be sure, our period is one when banks were more important, relatively speaking, as a source of credit – securities markets in most countries not having gained the depth and liquidity they were to acquire subsequently. Thus, it may do relatively little violence to reality to use M2 (scaled by nominal GDP) as our measure of credit. Still, the problem is not a negligible one. For the nine countries on which we have data on the growth of

The Great Depression as a Credit Boom Gone Wrong

193

both money and credit over the period 1913–1929, the M2/GDP and Credit/GDP ratios have a correlation of 0.70. In what follows we measure the credit component of the boom using positive deviations from trend in the M2/GNP for a sample of 16 countries at an annual periodicity starting in 1920. The countries are Argentina, Australia, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, the Netherlands, Norway, Spain, Sweden, the United Kingdom and the United States. While the sample was admittedly selected on the basis of data availability (not just for M2 but also for the ancillary variables utilized below), the result is essentially the universe of post-World War II OECD economies. The credit boom story as told by Kindleberger and Minsky is essentially a story about the now advanced (industrial) economies; from this point of view, we have precisely the appropriate sample. The one “ringer” is Argentina. In what follows we conduct sensitivity analysis to see whether any of our results depend on its inclusion. They do not. M2/GDP has other labels, of course, such as Cambridge k and the inverse of the velocity of circulation. The literature on velocity (e.g. Bordo & Jonung, 1987) has shown that this variable can trend downward (as it did in many countries before World War II) or upward (as it did subsequently) over a period of a decades, reflecting secular developments in the financial system. Similarly, in periods like the 1920s, when money supplies were tied, albeit loosely, to stocks of monetary gold, the M2/GNP ratio may trend upward or downward depending on whether global gold supplies are growing faster or slower than output.18 For both reasons, distinguishing a credit boom cum monetary expansion from secular movements in velocity thus requires detrending the latter. We therefore fit a linear trend on data through 1930 and focus on the residuals. This allows us to analyze cumulative processes – that is, the cumulative deviation of credit from its baseline or trend level – as opposed to credit conditions in a particular year, which would be the focus if we simply considered its rate of growth in that year.19 Figure 3 shows the individual country experiences. There we see a downward trend of the M2/GNP ratio in the 1920s in half the countries, not obviously consistent with the existence of a credit boom.20 But what is relevant to our argument is not the trend but the deviations around it. Interestingly, M2/GNP is almost exactly on trend in the vast majority of our countries in 1928, which we take as the height of the ostensible credit boom on the basis of qualitative accounts. Only in a small number of countries (Argentina and perhaps France and Japan) was credit notably above trend in this year. Thus, if we are to succeed in developing systematic, quantitative evidence of the credit boom phenomenon, it will be necessary to consider other aspects. As explained in the preceding section, the literature on credit booms is concerned

194

BARRY EICHENGREEN AND KRIS J. MITCHENER

Fig. 3. Money to GDP (%) and Trend.

with both the growth of credit and its effects. A significant expansion of the supply of credit is not sufficient by itself to constitute a boom of the sort that was of concern to the Austrians or which today attracts the attention of economists like Borio and Lowe (2002). What is critical in their view is that the growth of credit should be associated with a rise in asset prices and an acceleration in rates of fixed investment relative to trend. In the view of these authors, it is this confluence of factors that might be said to comprise the distinction between credit boom and credit growth. Whether credit booms and credit growth have significantly different implications for the subsequent development of the economy is of course what determines whether this distinction has substance.21 Contemporaries focused on the impact of accommodating credit conditions on asset prices, and equity valuations in particular. These are shown in Fig. 4, normalized by the overall level of prices, again relative to trend over the period through 1930. Equity valuations rise relative to trend in the late 1920 in the majority of our countries. Although the U.S. stock market boom is the best known, these data suggest the existence of similar fluctuations in a number of other countries (as emphasized by, inter alia, Kindelberger, 1976).22

The Great Depression as a Credit Boom Gone Wrong

195

Fig. 4. Stock Index Adjusted by CPI and Trend.

Contemporaries also saw the credit boom as stimulating investment, both directly, by making external funding more freely available and reducing collateral constraints, and indirectly, by raising Tobin’s q and the incentive to invest. Investment fluctuations are shown in Fig. 5. Although those movements are dominated by the collapse in the 1930s, as a result of which fluctuations around trend in the second half of the 1920s hardly stand out, it is still evident that a number of countries experienced surges in investment in the 1920s. There are a few exceptions worth noting. For example, France experiences an investment boom in the late 1920s, which extends through 1930, reflecting its relatively late postwar stabilization in 1926, and the surge of investment initiated with the end of the post-stabilization recession in 1927 (sustained by the large amounts of financial capital that flowed back to the country as the strong franc came to be seen as a safe haven).23 In order to more systematically draw out the connections between equity valuations and investment, Table 2 reports some simple investment equations (run in double log form), where the investment ratio is regressed on log q (equity prices deflated by wholesale prices, contemporaneous or lagged), lagged output growth (the accelerator term), and a lagged dependent variable (investment tending to

196

BARRY EICHENGREEN AND KRIS J. MITCHENER

Fig. 5. Investment to GDP (%) and Trend.

be serially correlated because projects take time to complete and are less likely to be abandoned once underway). A doubling of q like that which occurred between 1926 and 1929 in the United States results, according to these equations, in an 18% increase in investment, and the collapse in share prices that occurred subsequently would have had an even larger negative effect.24 We can ask which of these aspects of the credit boom phenomenon, if any, has explanatory power for the output collapse that followed. Figure 6 juxtaposes the deviation of each of these three variables relative to trend circa 1928 – which we take to be the peak of the boom – with the subsequent collapse in GDP from 1929 to 1932. (To be clear, it is the fall in output that is displayed on the vertical axis; the larger the fall, the greater the value.) We address the problem of endogeneity by lagging our credit indicators when considering their association with subsequent business cycle movements. While this procedure is subject to Tobin’s post hoc, ergo propter hoc critique, we are not convinced that his critique is compelling in our context.25 Of these three variables, only equity prices are strongly related to subsequent output movements.26 The fact that deviations of M2/GNP from trend do not explain much of subsequent cross-country differences in the change in output follows from

Pooled q q−1 (I/Y)−1 (GDP growth)−1 Constant Number of obs. R2

0.18 (2.64) −0.01 (0.06) 0.72 (21.25) 0.01 (2.70) −1.33 (7.09) 298 0.67

Country Fixed Effects – 0.15 (4.34) 0.71 (21.00) 0.01 (3.08) −1.25 (6.78) 298 0.49

0.15 (2.25) 0.08 (1.12) 0.55 (13.42) 0.01 (2.58) −1.88 (8.79) 298 0.66

– 0.20 (4.98) 0.54 (13.18) 0.01 (2.96) −1.79 (8.57) 298 0.66

Country & Time Fixed Effects 0.20 (2.45) 0.03 (0.43) 0.58 (13.04) 0.01 (0.24) −1.71 (7.32) 298 0.73

– 0.21 (4.72) 0.56 (12.68) 0.01 (0.75) −1.55 (6.42) 298 0.72

Random Effects 0.18 (2.64) −0.01 (0.06) 0.72 (21.25) 0.01 (2.70) −1.33 (7.09) 298 0.72

– 0.15 (4.34) 0.71 (21.00) 0.01 (3.08) −1.25 (6.78) 298 0.71

The Great Depression as a Credit Boom Gone Wrong

Table 2. Tobin’s q and Investment (Double Log Regression, with Investment Ratio as the Dependant Variable).

Note: t-Statistics in parentheses. Source: See text.

197

198

BARRY EICHENGREEN AND KRIS J. MITCHENER

Fig. 6. Credit Boom Components and Subsequent Output Fall.

the fact that cross-country differences in M2/GNP relative to trend circa 1928 are so small (as noted in our discussion of Fig. 3). That deviations of investment from trend circa 1928 do not explain much of subsequent output movements is a generalization of Temin’s (1976) point for the United States.27

The Great Depression as a Credit Boom Gone Wrong

Fig. 6.

199

(Continued )

4. A COMPOSITE INDICATOR If we are prepared to be more courageous, we can combine these three dimensions of the credit boom phenomenon into a composite indicator similar to that utilized by Borio and Lowe (2002). The simplest approach is to weight the three components equally.28 The result is shown in Fig. 7, with the composite indicator averaged across countries. The idea is to capture not just the availability of credit to the private sector but also its interaction with asset prices and investment. The motivation is that the same increase in domestic credit may have different effects depending on the structure of the economy that amplifies or muffles its impact. The composite indicator thus seeks to capture both the impulse and its amplification by measuring not only the growth of credit but also its impact on asset prices and investment demand. Whether the composite indicator has more explanatory power than simpler alternatives is an empirical question. To be clear, we are not necessarily advocating the utility of this measure, but we are interested in exploring its explanatory power and implications. Figure 7 highlights the credit boom of the immediate post-World War I period, when interest rates were pegged at low levels but domestic demand was freed of wartime controls, allowing the volume of credit to be essentially demand determined and setting off a wave of merger-and-acquisitions activity and a surge of plant and equipment investment. This boom was then reined in by interest rate

200

BARRY EICHENGREEN AND KRIS J. MITCHENER

Fig. 7. Average Composite Indicator Using Equal Weights.

hikes starting in 1920 (see Lewis, 1949). Lax credit conditions reemerged in the second half of the 1920s (as emphasized by Kindleberger, 1973), peaking in 1928. In Fig. 7, the late-1920s boom does not appear particularly pronounced relative to that at the beginning of the decade. Figure 8 shows the composite indicator by country. Consistent with the interpretation of the immediate postwar credit boom in terms of the difficulty of curtailing wartime budget deficits and decontrolling interest rates, there is less evidence of the immediate postwar credit boom outside the main theaters of the war.29 Turning to the second half of the 1920s, we see evidence of France’s credit-induced recession in 1926, the year of the Poincar´e stabilization. We observe the relatively early end of the credit boom in Germany, reflecting the Reichsbank’s effort to discourage foreign borrowing in 1926 (by, among other things, allowing the Reichsmark to fluctuate more freely within the gold points, thereby introducing a foreign-investment-repelling element of exchange risk into the market) and then to put a damper on stock market speculation in the first half of 1927 (McNeill, 1986; Voth, 2002).30 Evidently, the pegged exchange rates of the interwar gold standard, while transmitting credit conditions across countries, also left room for distinctive national experiences.31 Figure 9 shows that the height of the credit boom, measured by the percentage deviation of the composite indicator from trend in 1928, varied across countries, and that its height at that date significantly predicts the severity of the subsequent

201

Fig. 8. Composite Indicator Using Equal Weights.

The Great Depression as a Credit Boom Gone Wrong

202

BARRY EICHENGREEN AND KRIS J. MITCHENER

Fig. 9. Credit Boom and Economic Slump.

downturn, here measured through 1932. We see, qua Robbins (1934), that the credit boom of the late 1920s was led by the North Americans – consistent with the U.S.-centered nature of the dominant interpretation of the financial boom and bust – with France and Italy not far behind. The Fed cut interest rates in 1927, partly in response to the motor-vehicle induced slowdown in the U.S. economy, as Henry Ford shut down his assembly lines to retool for the Model A (Kindleberger, 1973), partly to address the problems of Western farmers suffering the effects of chronically depressed agricultural prices (Noyes, 1938), and partly to relieve the pressure on sterling and other weak European currencies (Clarke, 1967).32 A more limited credit boom is also said to have developed in the United States in 1925 (Kindleberger, 1973), although this is hardly evident in our calculations (see Fig. 8). The emphasis placed by these earlier authors on credit conditions in this period derives from the upsurge in residential construction, mainly in Florida but to an extent in other parts of the country as well.33 This earlier credit boom may have similarly had roots in interest rate cuts taken by the Fed in 1924–1925 to help Britain back onto the gold standard.34 Whatever the motivations for the policy, there is little reason to doubt that monetary ease lay behind the property boom. In the words of Vanderblue (1927a, p. 116), [t]he relatively low yield on high-grade investments made it possible to tempt investors into purchasing real estate bonds . . . secured by new structures located in the boom territory.

The Great Depression as a Credit Boom Gone Wrong

203

But the 1925 boom was relatively minor and short-lived compared to what came after; this comes out clearly from our Fig. 8, if not from narrative accounts of the period. By 1927 investment in real structures had declined by 6% from its 1925 peak. Real investment declined from its peak because of the big decline in detachedstructures investment after 1925. Nevertheless, a frenzy of apartment building followed the detached dwelling boom (peaking in 1927), and a building spree in nonresidential structures continued through 1929 (Field, 1992). This clearly poses difficulties for the credit-boom interpretation of the post-1928 slump. Even if credit fueled the residential real estate boom in the United States, the timing of the latter is wrong for explaining the onset of the Great Depression, unless one is prepared to argue that the fall in investment in structures worked through to the rest of the economy with an unusually long lag.35 As investment in structures declined, the Fed cut rates. U.S. bank reserves grew faster in the second half of 1927 than in any other semester of the 1920s.36 This supports the notion that the ready availability of credit to the American economy was a factor shaping the expansion of the later 1920s. Moreover, that expansion was heavily driven by spending on consumer durables purchased on the installment plan (Olney, 1990), using credit provided mainly by nonbank lenders (finance companies, which had developed previously to finance purchases of income-earning durable goods like sewing machines and pianos but acquired new importance on the American scene when in the 1920s the major automobile producers established divisions and subsidiaries designed to finance purchases of their own durable goods), and by purchases of financial assets, financed with bank credit funneled to investors through their brokers (White, 1990b).37 The consequences showed up not just in the stock market, but in the burgeoning automobile industry, the leading sector of the 1920s, and in the commercial property market, which boomed in virtually every American city. It is no coincidence, for example, that the late 1920s was the occasion for the appearance of the modern high-rise, when the skylines of many American cities were defined. While the Florida real estate boom attracted more attention, given the sensational nature of some of the frauds and the colorful character of the individuals involved, the urban building boom that followed later in the decade is temporally more consistent with the evolution of the composite indicator.38 In France, another country where there is evidence of a credit boom, capital inflows lubricated the operation of French capital markets starting in 1927, as the flight capital of the prior period was repatriated following the Poincar´e stabilization. In the second half of 1926, this capital influx drove up the value of the franc. By the end of the year, the Bank of France and the politicians grew worried that further real appreciation would create hardships for French industry, and they pegged the currency (a policy given legal status in 1928 when gold convertibility

204

BARRY EICHENGREEN AND KRIS J. MITCHENER

was officially reestablished at the new, lower value of the franc). Nominal interest rates came down, and with the price level now stable (tied as it was to prices in the rest of the world), lower yields encouraged a movement into riskier investments (Eichengreen & Wyplosz, 1988). Thus, one of the mechanisms that might have dampened the subsequent investment boom, namely a real appreciation, was effectively disabled. Less has been written about the credit boom in Italy.39 Prior to the reintroduction of the gold standard at the end of 1927, the big universal banks could already count on discount window access at the Bank of Italy. Thereafter, capital inflows resulting from the placement of a succession of foreign loans underwrote the continued expansion of credit. In addition, the central bank continued to follow an accommodating policy in view of its concerns with the financial condition of the three largest banks, something it could afford to do to the extent that it possessed reserves in excess of those required to back currency in circulation (Fratianni & Spinelli, 1997, p. 151). The credit boom was less pronounced – though echoes were still audible – in Argentina, Australia, Belgium, Canada, Germany, and Norway.40 It was all but absent in the United Kingdom, where starting in 1927 the Bank of England was forced to maintain restrictive credit conditions to support an increasingly overvalued currency, and in Denmark, another country that brought its currency back to par, which traded heavily with Britain, and which was tightly integrated into the London market.41 Boom turned to bust in 1929. The Fed, concerned that the high level of the stock market was diverting resources from more productive uses and heightening financial fragility, began raising its discount rate in 1928; higher U.S. rates in turn curtailed capital flows to Europe and Latin America, forcing central banks there to tighten to prevent their currencies from weakening.42 Overall, this analysis points to the existence of a short but sharp credit boom in the second half of the 1920s, peaking in 1928 and most prominent in the United States. Countries with close economic ties to the U.S., such as Canada, had the greatest tendency to share in these conditions (Green & Sparks, 1988). In contrast, countries with chronic exchange rate problems, notably Britain, did not share the same conditions because they did not share the same policies, their central banks being forced to put a damper on money and credit growth in order to defend weak currencies. A few countries where economic conditions were special – France because of the relatively late date of its postwar stabilization, Spain by virtue of never joining the interwar gold standard – display different time profiles, which is itself evidence of the tendency for an international financial system organized around the pegged exchange rates of the gold exchange standard to transmit these lax credit conditions to the rest of the world.

The Great Depression as a Credit Boom Gone Wrong

205

Figure 9 also shows the bivariate regression line summarizing the relationship between the height of the credit boom circa 1928 and the magnitude of the subsequent output collapse, accompanied by regression coefficients and t-statistics. The relationship is significantly positive (at the 90% level). It retains its significance when we control for other national characteristics that also shape countries’ susceptibility to recessionary forces – for example, their openness, their trade balances, and their dependence on international capital flows.43 The variation around the regression line reminds us that the magnitude of the credit boom, so measured, was by no means the sole determinant of the severity of the subsequent slump. The downturn in the United States was significantly more severe than the magnitude of the credit boom alone would lead one to predict, particularly when the downturn is measured through 1932. This plausibly reflects policy-related shocks: the ratcheting up of interest rates to support the dollar after sterling’s depreciation in September 1931 and the country’s deepening bankingsector distress. Canada, while an outlier in the same direction, does somewhat better relative to the magnitude of its credit boom in the immediately preceding period. This may reflect that its banking system was more widely branched and that the commercial banks had been prevented from making mortgage loans in the 1920s (foreclosing one channel through which the credit boom might eventually lead to financial distress). The contrast is all the more striking given Canada’s dependence on wheat exports and the droughts that swept the Prairies. On the other hand, the country was relatively slow to make up this lost ground in subsequent years. Australia does poorly relative to expectations (that is to say, relative to the regression line).44 Japan is an outlier in the other direction: having suffered a series of economic difficulties in the 1920s and not going back onto the gold standard until 1930, it did not have far to fall when the Depression struck. Italy is also below the line. The Bank of Italy extended large amounts of secret last-resort lending to the three large ailing universal banks under cover of disguised exchange controls, supporting both the financial system and the economy. These observations – and specifically the low value of the R-squared – give us an opportunity to clarify what we are and are not prepared to claim for this analysis.45 We do not wish to be misunderstood as arguing that the height scaled by the credit boom circa 1928 provides a complete explanation for the Great Depression, or that it provides a superior explanation to popular alternatives like post-1929 policy mistakes or the constraints of the international monetary system. Readers familiar with our own previous work on the role of the gold standard and bank failures (respectively) in propagating the Depression will have anticipated this point. Here the point is evident in the fact that the credit boom indicator explains less than a third of the cross-country variation in the post-1929 slump in economic activity. In addition, there is the fact, already emphasized, that

206

BARRY EICHENGREEN AND KRIS J. MITCHENER

the components of the composite indicator are not really distinguishable from proxies that might be used to test the effects of alternatives like the monetary, stock market bubble, and over-investment interpretations of the slump. Thus, if we are going to convince the reader that the credit boom interpretation is a useful supplement to these better-known interpretations of the onset of the Great Depression, this simple quantitative analysis will need to be supplemented with qualitative evidence pointing in the same direction. We turn to this qualitative evidence in the next section. We conducted a variety of sensitivity analyses to give these measures a run for their money. For example, we considered only the fall in output through 1930 or 1931. Shortening the period over which the dependent variable is measured from 1929–1932 to 1929–1931 has essentially no effect; the t-statistic on the composite indicator changes from 2.36 to 2.49, and the R-squared of the regression is now 0.31 instead of 0.28. When we shorten the period covered by the dependent variable to 1929–1930, however, the t-statistic on the composite indicator drops to 1.64 (just on the margin of significance at the 90% level of confidence), and the R-squared falls to 0.16. There is a sense in which supporters of the credit-boom interpretation can take heart even from this negative result. Those who would emphasize the preeminence of policy mistakes (failure to act as a lender of last resort resulting in widespread bank failures, for example) would presumably argue that even if the credit boom indicator had explanatory power for the initial phase of the downturn, it can explain little of the subsequent cross-country variation in its depth and duration, which is primarily attributable to these other factors. In fact, we do not find that the shorter the period, the greater the explanatory power of the credit-boom thesis; the actual story is more complex. We also experimented with a variety of alternative weighting schemes for the components of the composite indicator. One possibility is to weight the three ratios by their respective signal-to-noise ratios – that is, by the ratio of the share of subsequent crises successfully predicted by data through 1928 to the share of false positives, where the signaling threshold is set to maximize this ratio.46 We are suspicious of this procedure insofar as it uses information on post-1928 developments (on whether a country had a banking or currency crisis) which are plausibly correlated with the magnitude of the fall in output to derive the weights used to construct the composite used for forecasting the fall in output. For what it is worth, this variant of the composite actually performs less well; the t-statistics on the composite and the R-squared of the regression are lower than when we use unweighted averages of the three components, regardless of the period covered by the dependent variable. We then looked to see whether there was any evidence of nonlinear effects of the credit boom indicators. Borio and Lowe (2002) suggest that credit booms are

The Great Depression as a Credit Boom Gone Wrong

207

likely to have larger effects when the various indicators exceed typical levels by a relatively large margin (a “critical threshold”), and when several components exceed those typical levels simultaneously (when a high level of the composite indicator reflects substantial contributions from several components and not just one). A first test simply added squared values of the composite indicator as a second independent variable; these never entered with coefficients significantly different from zero or significantly enhanced the overall explanatory power of the regressions. We obtained more interesting results when we added to the regression equation displayed in Fig. 9 interaction terms involving the individual components, setting the value of those components to zero when they were below trend. Thus, the interaction terms capture additional effects in above-trend “credit boom periods” only. When we added two-way interactions of credit with equity prices and credit with fixed investment, the coefficient on the composite remained essentially unchanged (the slope coefficient fell slightly to 1.13, and the t-statistic fell marginally to 2.33). In addition, the two-way interaction of credit and the stock market entered with a coefficient that was significantly greater than zero at the 95% confidence level, while the coefficient on the two-way interaction of credit and investment entered with a coefficient that was significantly less than zero at the 95% level. This suggests that the credit expansion in the 1920s had the largest impact on the slump of the 1930s in countries where it was mainly associated with a stock market boom, while it had the smallest adverse effect where it was mainly associated with fixed investment. Neither of these additional interaction effects was large enough to reverse the dominance of the composite indicator in any of our sample countries. But the additional effects do suggest that whether the credit expansion of the later twenties was mainly associated with an equity run-up or a fixed-investment surge did significantly shape its implications for the severity of the subsequent downturn. We then added the two other interaction terms (the two-way interaction of the stock market and investment, and the interaction of all three components of the composite index), but neither of the additional coefficients differed significantly from zero. The other effects were essentially changed.47 Some readers will worry about the combination of more and less developed countries in our sample and question whether the experience of the less developed countries speaks to the issues at hand. Eyeballing Fig. 9 is sufficient to confirm that leaving out Argentina and the low-income European countries (Spain, Italy) does not weaken the relationship between the height of the credit boom circa 1928 and the magnitude of the output fall thereafter; if anything the opposite is true. If we use weighted least squares (weighting the observations by per capita income), to more systematically reduce the weight on low income countries, the results are in fact strengthened; the fit of the equation is significantly improved. The

208

BARRY EICHENGREEN AND KRIS J. MITCHENER

same conclusion follows if we instead leave out the non-European and non-North American countries (Australia and Japan). To summarize, the aggregate evidence provides some support for a MinskyKindleberger-Robbins-style interpretation of the Great Depression as a credit boom gone wrong. But the aggregate evidence has limitations. Given the important role of equity price deviations in the composite index, it is hard to distinguish the credit-boom and stock-market-bubble interpretations of the slump. And the preceding analysis tells us little about the precise circumstances where credit boom effects were particularly pronounced or the channels through which they were transmitted. For this, it is necessary to consider other evidence.

5. SECTORAL EVIDENCE One way of shedding light on these questions is by looking more closely at the behavior of specific credit-sensitive sectors and activities, such as construction, consumer durables, and high tech. Doing so points us to two important conditioning factors. One is the structure and performance of the financial sector. We find that the credit boom and its impact were particularly pronounced where the organization and history of the financial sector led intermediaries to compete aggressively in providing credit. The other is the technological environment. We find that the credit cycle, as defined here, was particularly pronounced when accommodating finance coincided with the development of new network technologies with significant longterm commercial promise but uncertain immediate potential (such as radio in the 1920s and the Internet in the 1990s).

5.1. The Construction Sector As Figs 10 and 11 show, investment in structures, especially private residential fixed investment, rose sharply in the 1920s, not just in the United States but also in Canada, Finland, Sweden, the Netherlands, and the U.K. The availability of credit played an important role in this response. But so too did indoor plumbing, electrification, the diffusion of the automobile, and the end of World War I. The war had destroyed millions of structures and affected demographic conditions in ways that stoked the demand for housing (it led to unusually high family formation in the 1920s, for example). In turn, the cessation of the war stabilized the investment environment (or at least set the stage for doing so). But there were significant differences across countries in the size and timing of their construction booms that cannot be explained by these factors. Australia,

The Great Depression as a Credit Boom Gone Wrong Fig. 10. Newly Constructed Dwellings. Source: Authors’ calculations based on League of Nations, Urban and Rural Housing (1939) and League of Nations, World Economic Survey 1934–1935, (1936). 209

210 BARRY EICHENGREEN AND KRIS J. MITCHENER

Fig. 11. Private Capital Formation, Structures. Source: Authors’ calculations based on Aukrust and Bjerke (1959), Balboa and Fracchia (1959), Maddock and McLean (1987), Kuznets (1946), Urquhart and Buckley (1965), Istituto Centrale di Statistica (1986), Mitchell (1988), National Accounts of the Netherlands (2004), League of Nations (1939).

The Great Depression as a Credit Boom Gone Wrong

211

Canada, and the United States all experienced residential housing booms of varying degrees of intensity but had suffered no direct damage from the war. This points to the importance of credit market developments and in turn to differences in the structure and operation of the financial sector. Countries differed in terms of the institutions that were primarily used to finance mortgages (savings and loan associations and building societies in the U.S. and U.K.; savings banks in Australia; private mortgage banks in Belgium, the Netherlands, and Canada; the Credit Foncier in France; and cooperative mortgage societies in Scandinavia). They also differed in the development of secondary markets, as shown in Table 3. One Table 3. Residential Real Estate Market. Average Mortgage (as % of Home Value)

Return to Gold Standard

60 75 60 50 60 65 60 55 60 na 55

1925 1925 1927 1926 1928 1925 1928 1924 na 1925 1926

Second Mortgage Market

Usual Length of Mortgage

BE

No

5–20 years

UK DE

Yes

FI FR NE NO SW

Yes No Not active Yes Yes

US AU CA

Yes No Yes

BE UK DE FI FR NE NO SW US AU CA

20 years

9–30 years

5 or 10 years/renewable

Source: League of Nations (1939).

11 years

End of Rent Control

1928 phase out begins; re-enacted 1934 1923 phase out begins 1929(except Copenhagen, 1935) 1923 1929 laws partially phase out 1927 1931 1922 None, only regional in nature None None Primary Source for Borrowed Funds

Antwerp Mortgage Bank; Land Credit Bank of Belgium Building societies, insurance co.’s Credit associations; mortgage associations savings banks, insurance co.’s Credit Foncier de France Mortgage banks Insurance companies & savings banks Mortgage banks; Urban Mortgage Bank of the Kingdom of Sweden S&Ls Savings banks, building societies Loan & Trust Co.’s, Insurance Co.’s

212

BARRY EICHENGREEN AND KRIS J. MITCHENER

conjecture based on Table 3 is that banks more aggressively financed investment in residential housing in countries where the financial system was more intensely competitive. In the U.S., where banks and Savings & Loans were already failing at significant rates in the 1920s, financial institutions competed aggressively for high-yielding construction loans. In Australia, in contrast, where there had been prior consolidation of the financial industry, there was less of a tendency for banks to gamble for survival, and the magnitude of the construction boom was less.48 These differences in behavior in the upswing had important implications for the subsequent depression. Although the slump was severe in both Australia and Canada, in neither case was it compounded by a U.S. style banking crisis. The Australian banking system escaped the 1930s with only three bank suspensions despite a sharp decline in output. And Canada’s 11 commercial banks remained in operation throughout the period. The resilience of their banking in the slump is commonly attributed, at least in part, to more conservative behavior during the upswing.49 There may have also been a role for accumulated experience in these differences. As noted above, Australia had experienced an earlier housing boom in the 1880s, fueled by rapid increases in mortgage lending by savings banks. Bank credit as a share of GDP doubled between 1880 and 1890. The majority of the increase went into residential construction, the 1880s being a period of rapid urbanization and population growth. In the early 1890s, when this boom turned to bust, 13 of the country’s 23 banks failed or were forced to suspend operations. The U.S. also had credit booms in the nineteenth century, but none as dramatic as this earlier Australian episode. None of these nineteenth century cycles had resulted in the failure of more than half of the country’s financial institutions. This earlier experience is said to have rendered Australian savings banks more cautious during the next credit boom, that of the 1920s. As Schedvin (1970, p. 80) puts it, “Even after nearly 40 years the effect of the events of ’93 coloured in no small way the banks’ reaction to the depression.” In contrast to the 1890s, even as credit expanded rapidly at the end of the 1920s (Fig. 10), savings banks raised their capital ratios, limited their exposure to property, kept the maturity of loans relatively short, and held a relatively high share of government securities (Kent & D’Arcy, 2002). Meanwhile, in the U.S., S&Ls and other intermediaries fueled an orgy of construction that left the landscape littered with vacant apartment buildings, and with subdivisions that were prematurely divided and remained undeveloped for years (Field, 1992). Mortgage debt more than tripled from $8 billion in 1919 to $27 billion in 1929. Realtors and developers often sat on the boards of S&Ls, influencing the operation and real estate lending of these intermediaries. This conflict of interest may have led lenders to make loans of lower quality and higher risk. Moreover, new and complementary sources of credit further fueled

The Great Depression as a Credit Boom Gone Wrong

213

the boom. In 1913, regulators removed restrictions that had previously prevented national banks from holding real estate mortgages. And the growth of auto ownership (made easier in part through installment plans offered by auto finance companies, described below) accelerated the pace and extent of land subdivision and encouraged speculation on city edges and recently converted farmland. To be sure, the structure and regulation of finance was not the only conditioning factor. Governments also put in place (positive and negative) incentives for residential housing construction by the private sector. Most European countries imposed rent controls at the beginning of the war and kept them in place for some years following its conclusion.50 The behavior of labor costs also was important.51 But, then as now, the cyclical behavior of the construction industry cannot be understood without reference to the structure and regulation of finance.

5.2. Consumer Durables Consumer durables further illustrate how the structure of the financial sector shaped the credit boom of the 1920s. To be sure, rising household incomes supported the growth of consumption, but financial institutions aggressively competing to supply households with credit allowed consumer spending to rise even faster than personal income. The most prominent case is the United States, where consumer debt as a percentage of personal income doubled from 4.5% in 1918–1920 to more than 9% in 1929 (Olney, 1991).52 The only other country that appears to have come close is Canada, where proximity to the U.S. market heightened the power of example and made it relatively easy for U.S. financial firms to set up operations north of the border. By 1928 there were as many as 1300 finance companies operating in Canada (which is only slightly smaller than the comparable number for the United States – see below). Scattered evidence suggests that the rate of growth of the number of installment contracts in the 1920s was also rapid in a number of our other sample countries. But in other countries the process started from a lower base. Hence, consumer credit and household debt played a less important role in the macroeconomic upswing and eventual collapse in these other countries.53 Recall that the credit boom as measured in Section 2 above was most pronounced in the U.S. and Canada. That financial institutions providing consumer credit had penetrated these economies extensively suggests that they played a role in amplifying the credit boom in both North American countries. None of these practices was entirely new. In both the United States and Britain they were already widely commented upon in the first half of the nineteenth century.54 Singer had sold sewing machines on credit in both the U.S. and Britain

214

BARRY EICHENGREEN AND KRIS J. MITCHENER

from the 1850s. Pianos, household furniture and even books were financed using installment credit in subsequent decades. But it was the advent of assembly-line methods for the production of automobiles and the development of a mass market in motor vehicles that resulted in the rapid growth of installment credit. The General Motors Corporation established the General Motors Acceptance Corporation (GMAC) in 1919 to finance sales of its cars. GMAC having demonstrated the advantages of this mechanism, other producers followed suit, along with a large number of independent (non-producer affiliated) finance companies. By 1925 there were more than 1500 finance companies operating in the United States.55 By 1927 nearly two-thirds of new cars in the U.S. were purchased on installment terms.56 Olney (1991) shows that installment credit was of comparable importance for purchases of a variety of household appliances. While the growth of installment purchase was global, commentators were unanimous in arguing that the phenomenon was most advanced in the United States. An indication of this fact is the role of U.S. financial institutions in the development of analogous mechanisms in other countries. Almost immediately following its establishment, GMAC branched into Canada, where General Motors and other U.S. producers dominated the motor vehicle market.57 GMAC was active in the U.K. in the 1920s, prompting the development of indigenous competitors such as the United Motor Finance Corporation Ltd.58 GMAC similarly established branches in Antwerp, Berlin, and Copenhagen (Crick, 1929, p. 22). Inspired by this example, the Italian Automobile Club attempted to establish a finance company to promote sales of cars. Other American companies also participated in the development abroad of institutions of consumer credit, reflecting the relatively advanced state of installment lending in the U.S. and the country’s role as a capital exporter. An important step in the diffusion of installment credit in the U.K. came in 1919 when Continental Guaranty of America created a British subsidiary, United Dominions Trust, to handle credit sales for motor cars. In the 1920s, the Commercial Investment Trust Company, the second largest American finance company, purchased subsidiaries in Germany, France and Scandinavia while operating its own offices in Argentina, Brazil and Cuba. Installment credit spread to the Netherlands partly through the creation of the N.V. Hollandsche Disconteerings-Bank in 1925, formed with American capital participation. Other capital and commodity exporters emulated the practice. Thus, once installment finance companies sprang up in Switzerland in the 1920s, they quickly opened offices in Germany to finance purchases of Swiss products. Installment business similarly gained importance in Australia in the 1920s, especially in financing purchases of imported goods, including motor vehicles.

The Great Depression as a Credit Boom Gone Wrong

215

By 1928 at least 70 companies were engaged in the business in the State of Victoria alone. Specialized finance companies issued cash orders to be paid off in installments that were accepted at leading shops in payment for virtually anything except food. (They were even used to pay for dental care on the installment plan.) But the Australian authorities, their views still colored by the crisis of the 1890s (see Section 4.1), began to worry about the over-extension of credit as early as the summer of 1927. In response, they applied direct pressure to curtail the extension of installment credit for the purchase of imported cars. In parallel with the conservative approach of Australian banks to financing construction activity, this constrained the role of credit in fueling the upswing, and in turn limited the extent of financial distress in the subsequent slump. Elsewhere, the growth of installment credit was rapid. Scott (2003) shows that the number of hire-purchase agreements outstanding in Britain nearly tripled from 6 million in 1924 to 16 million in 1928. Crick (1929, p. 6) estimates that 80% of pianos and gramophones, 50–60% of motor cars, 70% of sewing machines and 50% of furniture sold in the U.K. in the 1920s were subject to installment agreements.59 He notes that down-payment terms were lower than in the United States and that the terms on which installment credit was extended grew increasingly liberal over the period, consistent with other observations about financial behavior in the late stages of a credit boom. However, all this growth began from a much lower base than in the United States. Even at the end of the 1920s, installment credit was still too small to significantly affect the macroeconomic aggregates. Figures in Scott (2003) suggest that installment credit financed only about 2% of British retail sales in this period. Comparable ratios for other European countries were almost surely lower. In the U.S., in contrast, nearly 9% of consumer spending in the 1920s was on durable goods.60 Of all spending on durables, 60% was on big-ticket items (major durable goods), and some 70% of that in turn was financed with consumer credit. In addition, a substantial fraction of minor durable goods were purchased under the installment plan.61 We see here the higher base from which installment credit in the United States expanded in the 1920s and thus the more significant macroeconomic consequences of its growth. As in construction, part of the explanation for these exceptional features of U.S. experience may have been the intensely competitive nature of the financial sector, including ease of entry. By the late 1920s, 1500 finance companies competed with commercial banks for a toehold in the market. To be sure, other factors played a role as well; for example, relatively high household incomes and an egalitarian distribution of income meant that there was a large pool of households in a position to purchase big-ticket items like automobiles, vacuum cleaners, audio equipment, and kitchen appliances. European motor-car producers continued to concentrate

216

BARRY EICHENGREEN AND KRIS J. MITCHENER

on high-end vehicles, partly reflecting their slowness in adopting mass-production methods, but partly because they perceived more limited working-class demand.62 Scott suggests that the household equipment component of consumer spending was less important in the U.K. than the U.S. because British husbands somehow failed to appreciate their wives’ need for a Hoover! More plausibly the explanation lies in lower living standards and greater income inequality. Thus, the structure and response of the financial sector seems to have played an important role in transmitting the credit boom of the 1920s, although financial structure was not the only factor shaping the differential response of different countries.

5.3. High Tech The end of World War I and the restoration of price stability restored investor as well as consumer confidence. The most prominent aspect of this trend was investor enthusiasm for the commercial potential and profitability of newly developed, technologically sophisticated products and processes (including but not limited to consumer durables). A famous case in point is radio, as noted in Section 1 above. Radio was the 1920s analog to the Internet, right down to the use of the medium to trumpet the promise of investment in that same medium. RCA was the market leader into which investors scrambled in anticipation of capital gains (the price of RCA stock rose from 12 in 1921 to a high of 549 in 1929 – some 73 times earnings – despite the fact that the company paid no dividends anytime in the period). But while radio was the most dramatic case in point, technological dynamism was not limited to this one sector. The 1920s was also the age of automobiles and mass production – the years following Henry Ford’s development of the assembly line and the decade of the Model T.63 It was a decade that featured technological breakthroughs in the use of electrical machinery and the production of synthetic chemicals. Along with RCA, the high-tech stars of the period included Westinghouse, General Electric, AT&T, and Montgomery Ward (the Walmart of the time, whose attractions to investors resided in the innovative nature of its retail network). Field (2003) suggests that the technological advances associated with the activities of these companies had as a corollary a significant acceleration in the rate of total factor productivity growth, which stoked the enthusiasm of investors.64 Firms captured this investor enthusiasm in rising stock prices. Our index of high tech stocks (sectors such as communications, electrical equipment and appliances, inorganic chemicals and transportation, and firms such as Dupont, Maytag, General Electric, Westinghouse, Chrysler and GM) rose by over 200% between 1926 and 1929 (see Figs 12 and 13).65 What these technological advances did not

The Great Depression as a Credit Boom Gone Wrong Fig. 12. Market Capitalization for Aggregate Technology Index. Source: Authors’ computations based on CRISP data for the 10 high-technology industries shown in Fig. 13.

217

218

BARRY EICHENGREEN AND KRIS J. MITCHENER

Fig. 13. Market Capitalization for Various Technology Industries of the 1920s.

uniformly translate into, however, was short-run profitability. With benefit of hindsight, we now know that the enthusiasm of investors for the commercial potential of these new technologies, above all radio, was premature, although not wholly unwarranted. Networks require an installed base in order to be commercially viable, and radio in particular required a significant installed base before the industry became profitable. The number of U.S. households with radio sets rose rapidly but in 1928 was still less than a third of 1939 levels. A profitable market for advertising presupposed the existence of broadcast networks, which only began

The Great Depression as a Credit Boom Gone Wrong

219

Fig. 13. (Continued )

to develop with the establishment of the National Broadcast Company in 1926 (initially as a network of 19 stations). For all these reasons, commercial viability took time. That investors overestimated the speed with which profitability would ensue may not have been entirely unrelated to prevailing credit conditions: the low level of interest rates prevailing in the United States and the ample availability of brokers loans, reflecting the liquidity of the financial system, may have encouraged investors to reach for riskier investments. Perez (2002) generalizes the point, arguing that the emergence of new network technologies regularly causes stock markets to overreact. In her view, securities markets regularly respond to the emergence of a new network technology with a boom and bust cycle which must be completed before glimmers of profitability and commercial viability finally become visible. She tells this same story of the canal boom of the 1820s, the railroad boom of the 1840s, electrification in the 1890s, the age of radio, automobiles and mass production in the 1920s, and the information and communications boom of the 1990s. Each of these innovations involved the deployment of a network technology, necessarily implying a lag of uncertain length between initial installation and eventual profitability. In each case, in her account, market participants overestimated the speed of deployment and adaptation, causing securities markets to overshoot. Importantly from the present point of view, Perez emphasizes the role of the financial system, and of accommodating credit conditions in particular, in fueling speculative activity.66 In each case, she argues, the availability of credit was enhanced by financial innovation, which provided channels for liquidity to flow to technologically dynamic sectors. In the 1920s the innovations in question included the new techniques for marketing securities to individual investors and the spread of the investment trust. In Perez’s view, the infusion of liquidity into the markets leads to easy capital gains, which in turn encourage “ethical softening”

220

BARRY EICHENGREEN AND KRIS J. MITCHENER

in the frenzy phase, followed by the inevitable fall.67 The long-run productive potential of the economy is enhanced by the investment consequences of all this financial activity, but an extended period of capital losses and consolidation still must intervene before a positive impact on profitability is felt.68 Thus, in addition to reinforcing the emphasis we place elsewhere in this paper on financial structure, this sectoral study highlights the important role played by the interaction of finance with innovation.

6. CREDIT BOOMS AND THE GOLD STANDARD How a pronounced credit boom could develop under the gold standard is not obvious. In principle, the gold standard did not provide an elastic currency at the global level, which should have worked to limit the amplitude of the credit boom.69 For the world as a whole, supplies of money and credit should have been tied down by supplies of monetary gold, which were inelastic in the short run.70 Hence, when a credit boom got underway, it was not accommodated by increased supplies of money and credit. Higher interest rates would tend to dampen investment and choke off borrowing by stock-market speculators. The implication was that credit booms should have been less pronounced under the gold standard. This was the contemporary conclusion of Mises and Hayek. Today, in contrast, “the external constraint on credit imposed by the gold standard has gone.” “Central banks now virtually ignore the pace of credit expansion so long as inflation is under control. As a result, the ‘elasticity’ of private credit creation has increased significantly” (The Economist, 2002, p. 23). How to characterize the 1920s from this point of view is not clear. There was a rise in the importance of foreign exchange reserves relative to gold, compared to the prewar era, imparting more elasticity to global supplies of money and credit. During the boom period (1924–1928) the share of foreign exchange in the total reserves (gold plus foreign exchange) of the 24 central banks considered by Nurkse (1944) rose from 27% to 42% (before falling back slightly to 37% in 1929). It then collapsed to 19% in 1931 and 8% in 1932. This lent a procyclical elasticity to money and credit under the hybrid interwar gold-exchange standard. It is one reason why the elasticity of credit creation could have been higher than suggested by textbook models of the gold standard system. Of course, this critique of the interwar gold standard may equally constitute a critique of the prewar gold standard. Central banks also held foreign exchange reserves before 1913. But the practice was not as widespread as in the 1920s. And we are aware of no other period in which the share of foreign exchange reserves fell as sharply as in 1929–1932, effectively destroying a third of the global monetary base.

The Great Depression as a Credit Boom Gone Wrong

221

Before 1914, central banks holding excess gold reserves were also able to manipulate the money multiplier by, inter alia, altering their discount rates. Bloomfield (1959) emphasized the tendency for the rates of discount of the major central banks to move together over the cycle, as if some such reaction was occurring on a global scale. Authors like Cairncross (1953) and Ford (1962) suggest that changes in money and credit conditions occurring in response to investment fluctuations were an important source of global business cycle fluctuations under the pre-1914 gold standard. Portraits of the consequences by authors like Kindleberger (1978) do not suggest that credit booms were less pronounced in the gold-standard years than under subsequent monetary regimes. The preceding discussion focuses on the gold standard as a global monetary regime, an appropriate view if the credit boom of the 1920s is seen as a global phenomenon. Alternatively, we can consider the gold standard’s operation at the country level and ask whether it would have worked to restrain or encourage a credit boom in a particular country. We are not aware of much satisfactory theoretical analysis of the connections between the exchange rate regime and endogenous credit dynamics. The Asian crisis and other recent episodes in which pegged rates have collapsed have encouraged the view that pegged rates encourage credit booms. Under pegged rates, animal spirits that drive up the stock market and investment, in turn raising interest rates, will encourage capital inflows from abroad, augmenting supplies of money and credit. When the exchange rate is pegged, there is little perceived exchange risk to deter interest arbitrage and no tendency for the currency to appreciate and tamp down the investment boom.71 Even if the supply of credit is fixed at the global level, it is elastic from the point of view of the individual country. Thus, credit booms concentrated in individual countries (or groups of countries) may be even more pronounced under fixed than flexible exchange rates. This view is informed by the experience of Scandinavian countries in the late 1980s and by the experience of Asian countries in the 1990s, when large capital inflows sustained pronounced credit booms, setting the stage for an even more painful subsequent fall (Goldstein, 1998; McKinnon & Pill, 1997). This view assumes that the boom does not originate in increases in domestic credit but rather from other sources like irrational investor exuberance and that it is then accommodated by capital inflows. If, on the other hand, the source of the boom is excessive domestic credit creation, then this will lead to balance-of-payments deficits and capital outflows that, if left uncorrected, may jeopardize the pegged exchange rate. In this situation, the pegged rate is a restraint rather than a contributor to the credit boom. From this point of view, whether pegged rates in general and the gold standard in particular are part of the problem or part of this solution will depend on the source of the boom – whether it is domestic credit as opposed to investment or

222

BARRY EICHENGREEN AND KRIS J. MITCHENER

asset-price inflation – and whether it is a global or country-specific phenomenon. The answer will also depend on how monetary policy is conducted, since even pegged rates provide some room for discretion under certain circumstances. Ultimately, then, whether credit booms were more or less pronounced under the gold standard is an empirical question. We can analyze it by constructing analogous indicators of credit booms using annual data for the period from 1880 up through 1997 and then calculating whether the volatility of our measure is greater in times and places when the gold standard was absent than when it was present. We have the necessary data for all or most of this period for nine countries: Australia, Canada, Denmark, France, Germany, Italy, Sweden, the United Kingdom and the United States.72 We detrended separately for the pre-1914, interwar, Bretton Woods and post-Bretton Woods periods. We then computed the standard deviation of the detrended composite, looking only at boom periods (observations for years when the composite was above trend). Finally, we compared years when countries were on the gold standard with years when they were not, and years when exchange rates were pegged to when they were floating (on the grounds that the same arguments regarding the external constraint that apply to the gold standard also apply, in principle, to other fixed-rate regimes). From the results (Table 4), it would appear that the amplitude of credit booms as measured by the standard deviation was greater in periods when exchange rates were pegged than when they were floating. The difference is significant at the 95% level.73 In contrast, when we compare when countries were on the gold standard with when they were not, we find no differences in volatility under the two regimes.74 These comparisons thus lend little support to the notion that credit booms were less of a problem under the gold standard. On the other hand, they are consistent with the view that pegged rates – which limit inflationary pressures but allow the demand for money to be endogenously determined, and which encourage policy makers to focus on the stability of prices and exchange rates but Table 4. Volatility of Credit Booms: The Gold Standard, Pegged Rates and Other Exchange Rate Regimes (Credit-Boom Observations Only). Comparison

Pegs v. others (equal weights) Gold standard v. others (equal weights) Pre-1914 gold standard v. non-gold (equal weights)

Pegs (Including Gold)

Flexible

Obs.

S.D.

Obs.

S.D.

216 130 88

7.468880 7.262342 5.373846

155 241 241

6.472419 6.949759 6.949759

Source: See text. The p value describes the exact level of significance.

p Value

0.0294 0.2786 0.0030

The Great Depression as a Credit Boom Gone Wrong

223

not the evolution of credit conditions – are conducive to credit booms, in individual countries or groups of countries at least, if not necessarily in the world as a whole. It could be that the interwar gold standard was special – that the credit boom of the 1920s was an anomaly and that similar phenomena were absent, or at least more muted, prior to 1914. This, after all, was the Mises and Hayek view. Hence, the final row of Table 4, instead of comparing gold-standard and non-gold-standard observations, compares pre-1914 gold-standard observations with non-gold-standard observations (that is, it eliminates the interwar gold standard years). As before, we restrict the analysis to credit-boom episodes – that is, to periods when the credit boom indicator is above trend. This change in periodization transforms the picture. The amplitude of credit-boom episodes appears to have been less before 1914 than in the non-gold-standard years (starting in the 1930s).75 Any evidence that credit booms were more pronounced under the gold standard than other monetary regimes, in other words, is attributable to the 1920s experience that is the subject of this paper.76 Our overall conclusion is that the gold standard was neither the cause nor the solution to the credit-boom problem; the effects depended importantly on how that gold standard was structured and managed. Similarly, this analysis does not support the notion that pegged exchange rates are either a fundamental cause or a solution to the credit-boom problem. Our own view is that an exchange rate rule, which focuses monetary policy makers’ attention on a particular asset price rather than on the broader constellation of asset and commodity market conditions, is not the optimal basis on which to formulate monetary policy. But the evidence of this section suggests that, more than the putative monetary regime, what matters is how monetary conditions are managed in practice.

7. CONCLUSION The 1990s was a decade of low and stable interest rates in many countries. Accommodating credit fueled increases in property prices and facilitated increasing consumer indebtedness, notably in the United States, while financing high investment rates. It encouraged rapid increases in securities prices. These developments heightened the vulnerability of financial systems and economies to a sudden reversal of sentiment, although the consequences to date have taken the form less of a bang than a fizzle. To be sure, credit market conditions do not provide the entire explanation for these developments. Investment and equity valuations were also stimulated by accelerating productivity growth, although the magnitude of this new-economy phenomenon remains a matter of dispute. But it is hard to dispute that credit market conditions at least played a supporting role.

224

BARRY EICHENGREEN AND KRIS J. MITCHENER

Among the consequences of these developments has been renewed interest in the work of Mises, Hayek, Robbins and Rothbard, who emphasized the role of credit dynamics in post-World War I cyclical developments. For a combination of domestic and international reasons, the Fed maintained a relatively accommodating stance for much of the 1920s. With inflation stabilization, other countries found themselves on the receiving end of capital inflows. Financial innovation magnified the impact of these accommodating credit conditions, and central banks did little to preempt their effects. The consequences, as in the 1990s, included property booms, increasing consumer debt, surging investment and rising securities prices, particularly those of high-tech firms. They included growing worries about the stability of financial institutions and markets. They culminated in the collapse of financial markets and institutions and the gravest macroeconomic crisis the modern world has ever seen. This characterization of the Great Depression as a credit boom gone wrong has much to recommend it as a cautionary tale for current-day policy makers. We wish not to be misunderstood: as emphasized above, we are not arguing that the credit-boom interpretation is a superior alternative to analyses of the Depression emphasizing the roles of the gold standard, the stock market boom, and monetary blunders. But a horserace is not the appropriate context in which to assess theories of the Great Depression. The Depression was a complex and multifaceted event. The perspective provided by the credit-boom view is a useful supplement to these more conventional interpretations. In particular, focusing on the credit boom of the 1920s directs attention to the role of the interwar gold standard in setting the stage for the slump of the 1930s. Our analysis suggests that equally-pronounced credit booms were not a facet of the classical gold standard. Notwithstanding the colorful accounts of Kindleberger et al. the amplitude of credit fluctuations appears to have been less under the pre-1914 gold standard than under the more flexible exchange rate regimes that followed. Evidently, however, the interwar gold standard was different. Our conjecture is that the strongly procyclical behavior of the foreign exchange component of global international reserves and the failure of domestic monetary authorities to quickly install stable policy rules to guide the more discretionary approach to monetary management that replaced the more rigid rules-based gold standard of the earlier era are important for explaining the fragilities that set the stage for the Great Depression. Previous work has emphasized the role of the interwar gold standard in the post-1929 collapse of foreign-exchange reserves and money supplies and in the international transmission of destabilizing impulses. But the credit boom view suggests that the structure and operation of the interwar gold standard also played a role in the expansion phase, when the endogenous response of the foreign exchange component of global reserves allowed credit to expand more rapidly than would

The Great Depression as a Credit Boom Gone Wrong

225

have been possible under traditional gold standard arrangements. This is an important extension of the conventional gold-standard-and-Great-Depression story. In addition, focusing on the role of credit conditions in the expansion of the 1920s and slump of the 1930s directs attention to two factors that warrant more attention than they have received in the recent literature on the Great Depression: the structure of domestic financial systems and the interplay of finance and innovation. Financial structure and regulation have featured in the comparative literature on the causes of banking crises in the 1930s (Grossman, 1994), but other channels through which they could have shaped and accentuated the boom of the 1920s and the subsequent reaction may have not received their due. The interplay of finance and innovation in stimulating the expansion and setting the stage for the crash has been the subject of even less attention, with recent authors tending to focus exclusively on one or the other of these two factors. It was of course precisely the experience of the 1920s and 1930s that provided the backdrop for Schumpeter’s great work, Business Cycles, where he characterized capitalism, and in particular its cyclical aspect, as “innovation financed by credit.” The experience of the 1990s reminds us that the development and effects of credit conditions may play out in quite different ways depending on the nature of the technological environment. It reminds us that the interaction of credit with innovation warrants additional attention. The implications for policy are less clear. One possible implication is that policy makers should act preemptively to prevent the development of unsustainable credit booms that might have seriously negative macroeconomic and financial consequences when they turn to bust. The strong version is that central banks should concern themselves not just with commodity price inflation but also with asset price inflation, especially in periods of technological dynamism when asset market inflation has a particular tendency to overshoot. They should tighten when they see credit expanding rapidly and asset-market conditions responding enthusiastically, and do so even if commodity-price inflation remains subdued. But most policy makers and analysts are reluctant to draw this conclusion. Central banks have no reliable way of determining when asset prices lose touch with fundamentals. This was as much a problem in the 1920s as the 1990s. It is only with benefit of hindsight that textbook writers refer confidently to a bubble, and even now, not all observers agree. In any case, monetary policy is a blunt instrument to deploy in response to increases in asset price valuations. The collateral damage to the real economy can be severe, as Hjalmar Schacht learned in 1927 and George Harrison learned in 1929. A more appropriate conclusion, in our view, is that although financial market conditions are important, they are first and foremost the responsibility of financial market regulators. In the interwar period, regulators should have concerned themselves with conflicts of interest between the underwriting and advising

226

BARRY EICHENGREEN AND KRIS J. MITCHENER

activities of the investment banks before as well as after the fact. They should have engaged in closer supervision of financial institutions if they saw signs that loan quality was deteriorating. They should have contemplated increases in capital and liquidity requirements to prevent the credit boom from developing in ways that heightened the vulnerability of the economy and the financial system to a subsequent downturn. This seems to us the right lesson for policy to draw also from the experience of the 1990s. The problem, of course, is that such lessons are always more evident after the fact.

NOTES 1. See for example Bernanke and Gertler (1999) and Tornell and Westerman (2002). 2. See Vila (2000), Borio, Fufine and Lowe (2001), and Borio and Lowe (2002). That this is the right policy conclusion is, of course, not universally agreed. On the controversy over the role of asset prices and credit conditions in the conduct of monetary policy, see Bullard and Schaling (2002), Bernanke and Gertler (1999), Cecchetti, Genberg, Lipsky and Wadhwani (2000), Filardo (2000) and Goodhart (2000). This same debate figures prominently in the literature on the Great Depression, as we describe momentarily. 3. We do not explicitly address the policy implications in this paper. One conceivable implication (which is implicit in Galbraith, 1972; Kindleberger, 1973) is that the Fed should have prevented the development of speculative excesses by maintaining a tighter policy stance toward the end of the 1920s, despite the absence of overt signs of inflation. Doing so, in this view, would have limited the build-up of vulnerabilities that became sources of financial stress when the economy eventually turned down. By limiting the extent of the credit boom in the late 1920s, it follows, a preemptive policy would have reduced the severity of the Great Depression in the early 1930s. There is of course an alternative view (e.g. Meltzer, 2003) that policy makers should have focused exclusively on inflation, with the implication that policy in the late 1920s was not too loose but too tight. To repeat, we do not tackle the policy controversy here. 4. Specifically, in the roles of the international financial system (Eichengreen, 1992) and the structure and regulation of domestic banking (Mitchener, 2003). 5. In this respect there are parallels between the Austrian model and Keynes’ Treatise on Money (1930), a fact appreciated by Keynes and emphasized by Laidler (1999). In addition, there are parallels between the Austrian view and the modern debate about whether central banks should simply concentrate on commodity price inflation or also be concerned about asset price inflation. We return to this below. 6. Mises and Hayek did not typically distinguish between asset and commodity price inflation, but when they did, they minimized the relevance of the distinction. Laidler (1999) argues that Hayek in particular saw the rate of interest (which affected the evolution of asset prices) as the key price (since it was what equilibrated or disequibrated saving and investment); by this interpretation, asset price inflation in fact mattered more than commodity price inflation. 7. Although Mises referred not to the build-up of indebtedness but to the inadequacy of saving, his point was essentially the same.

The Great Depression as a Credit Boom Gone Wrong

227

8. In Hayek’s (1932, p. 44) words, “any attempt to combat the crisis by credit expansion will . . . not only be merely the treatment of symptoms as causes, but may also prolong the depression by delaying the inevitable real adjustments.” 9. The Austrian views of the early Robbins were kept alive by, inter alia, Rothbard (1975). 10. As Robbins (1934, pp. 41–42) put it, “Sooner or later the initial errors are discovered. And then starts a reverse rush for liquidity. The Stock Exchange collapses. There is a shortage of new issues. Production in the industries producing capital-goods slows down. The boom is at an end.” 11. Robbins (1934, p. 62). 12. There is clear overlap between the Austrian view and the bubble interpretation; thus, Robbins points to the run-up in stock prices in the United States as a prominent consequence of the expansion of bank credit in the 1920s. 13. Eventually they also borrowed from corporations, as nonbank firms shifted funds into the money market in response to rising demand. 14. Estimates of asset returns are from Smiley and Keehn (1988). 15. Santoni and Dwyer (1990) and Meltzer (2003) argue that this evidence is not necessarily consistent with the existence of a bubble. Meltzer, similarly, rejects the bubble interpretation on the grounds that the rise in equity valuations in the late 1920s was not out of proportion to the rise of earnings. From the present point of view, the issue is not simply whether earnings rose as rapidly as equity prices in these years but also whether the magnitude of capital gains created expectations of further capital gains which were less obviously justifiable by fundamentals. See also White (1990a). 16. Others argued that the authorities should resist the temptation to stabilize commodity prices (which were now falling rapidly, doing considerable damage to the economy), much less asset prices, for fear that this would only encourage the development of another, even larger bubble that would be followed by a still more devastating crash. Thus, Robbins (1934) drew this conclusion, in advice he later came to regret. 17. Belgium experienced a more modest inflation, and it had a relatively modest credit expansion over this period, consistent with this interpretation. At the same time, the limited extent of the credit expansion in Denmark and Switzerland (where credit, although it did not decline, increased only modestly) suggests that the explanation for these trends is more complicated. 18. There was much concern in the 1920s about the possibility that the slow growth of global gold supplies was constraining the growth of money and thereby putting downward pressure on the money/GNP ratio. See League of Nations (1930). 19. We experimented with different filters and with filtering the data only through 1929 without substantively changing the results. 20. This trend may be indicative of the intensifying deflationary pressure exerted by the interwar gold standard, which constrained the growth of money and credit as economies recovered from World War I and expanded through the second half of the 1920s. In a number of countries, M2/GNP ratios then rise relative to this earlier trend in the 1930s as interest rates decline and the velocity of circulation falls. This tendency is documented by Bernanke (2000) and commented on further by Cole, Ohanian and Leung (2002). 21. And which presumably determines whether central banks should respond preemptively to the development of credit booms independent of their implications for inflation. 22. While the positive comovement of stock markets may strike some readers as puzzling in light of the steady flow of capital from Europe to the United States, this is to

228

BARRY EICHENGREEN AND KRIS J. MITCHENER

neglect flow savings by residents of other countries and the substitution of stocks for other investments as the decade progressed. The positive comovement of stock markets across industrial countries is of course the same phenomenon observed in the late 1990s. 23. Patat and Lutfalla (1990) observe that M2 continued increasing through the summer of 1930, unusually for the period, as a result of these capital inflows. This sequence of events and their connection with investment are analyzed by Eichengreen and Wyplosz (1988). 24. In addition the collapse of stock market valuations could have worsened the Depression by undermining bank balance sheets and leading to the bank failures that, observers widely agree, were a key engine of deflation in many countries (see, e.g. Bernanke & James, 1991). In fact, however, there is little correlation between q in 1928 and the incidence of banking crises thereafter. A probit regression of the Bordo-Eichengreen banking crisis dummy on the deviation of q from its 1920s trend, with and without a variety of controls, never yields a coefficient that differs from zero at standard confidence levels. On reflection this is not surprising. Consider, for example the contrast between the United States and Canada. Although both had exceptional stock market booms in the 1920s, one had a banking crisis while the other did not. Evidently, the absence of restrictions on branch banking in Canada and regulations limiting the ability of Canadian banks to lend against real estate dominated the impact of changing asset valuations on bank solvency and stability. Or contrast Britain and Argentina. Neither country experienced a pronounced credit boom or rapid stock market run-up in the 1920s, yet the latter had a serious banking crisis in the spring of 1931, while the former escaped the problem entirely. The reasons are not hard to see: Argentina’s terms of trade deteriorated in the Depression, while Britain’s improved, and Argentina had been on the receiving as opposed to the sending end of capital flows in the 1920s. The behavior of stock markets mattered for the subsequent evolution of output and employment, and for the banking systems whose stability was an important determinant of macroeconomic fluctuations; but they were not the only thing that mattered. 25. In particular, we know of little evidence that contemporaries expected the severe downturn that we now refer to as the Great Depression in advance of the event (a few prescient Austrian-school forecasters to the contrary notwithstanding). See Dominguez, Fair and Shapiro (1988), Hamilton (1992) and Cecchetti (1992). For those not convinced that timing provides identification, in Section 4 we also relate the development of credit conditions to deeper institutional and structural features of the economy (the monetary regime, the sectoral composition of activity, the structure of the financial sector) that are clearly predetermined with respect to the credit-market developments of the 1920s. 26. This points out the difficulty of distinguishing the credit and stock-market boom interpretations of the slump. While the stock market boom as a factor in the depression is a staple of history textbooks, it has not been much emphasized in the scholarly literature. In part, scholarly skepticism reflects problems with the thesis in the case of the United States, the country where the rise in stock prices was evidently most pronounced. The economic downturn in the U.S. preceded the stock market crash; while the business cycle peak was reached in August 1929, the Wall Street crash is conventionally dated as occurring in October. (The market reached its peak on September 3, 1929, but the big price drops associated in the popular mind with the Great Crash were Black Thursday, October 24, and Black Tuesday, October 29, well after industrial production peaked.) Moreover, the Great Depression in the United States was clearly compounded by the blunders of U.S. policy makers starting in 1930. Hoover’s tax increases and the failure of the Federal Reserve to stem the banking panics that ultimately forced a substantial fraction of all American

The Great Depression as a Credit Boom Gone Wrong

229

banks to close their doors were as important as any adverse consequences flowing from the run-up of the stock market. We will have more to say about this below. 27. To reconcile the predictive power of equity prices with the lack of predictive power of investment, it is only necessary to observe that while the link from equity prices to investment is loose (as shown in Table 2), the link from investment to subsequent output movements is even looser. 28. This is not exactly the procedure utilized by Borio and Lowe, who search for the best combination of weights that minimize the signal-to-noise ratio of subsequent banking crises correctly and incorrectly predicted. Below we experiment with some sensitivity analysis along these lines. 29. In Argentina and Australia, for example. 30. Robbins (1934, pp. 49–50) argues that the German credit boom persisted into 1928, as capital flows from the United States “overbore” the Reichsbank’s efforts to institute tighter conditions. Our composite indicators suggest that the boom ended earlier in Germany than the U.S., although one can quibble about the dating. 31. Schacht’s emphasis on the need to introduce an element of exchange risk into the market in order to discourage what we would now refer to as the carry trade suggests that the pegged exchange rates of the interwar gold standard were a factor in the development of the credit boom. It will remind readers of contemporary arguments (viz. Goldstein, 1998) that pegged rates can be an important source of investor moral hazard. We pursue these ideas in Section 5. 32. Other authors thus offer a more eclectic interpretation of the Fed’s motives than Robbins (1934), who focuses on the weakness of the British balance of payments and the Fed’s concern for the stability of the interwar gold standard. Note that the NBER placed the business cycle peak in October 1926, and industrial production hit its low in the final quarter of the subsequent year. The Fed’s interest rate cut was in the summer of 1927. 33. On the nation-wide character of the real estate boom, see Field (1992). The Florida land boom is a story in itself (we will have more to say about it below). Among other things it featured the involvement of no less than Charles Ponzi. Ponzi issued certificates of indebtedness promising a 200% dividend in two months’ time. He used the capital thereby raised to purchase land for subdivision, planning 23 lots to the acre. Ponzi’s advertising described the land in question as being “near Jacksonville,” where in fact it was 65 miles west of the city and covered with a thick growth of palmetto and other weeds. When he was unable to quickly sell the lots, Ponzi predictably found himself unable to meet his financial obligations and was subsequently indicted for violating Florida statutes regarding trusts and found guilty by jury. For details, see Vanderblue (1927a, b). 34. Of course, other factors also contributed to the bias toward monetary ease, including the fact that the economy experienced a slowdown in 1923–1924 and that the latter was an election year. On international motivations for 1924–1925 interest rate policies, see Wicker (1966), Chapter 7. 35. This was of course Temin’s (1976) objection to the older literature associated with authors like Robert A. Gordon emphasizing the rise and fall of fixed investment as a prime mover in the Depression. 36. Rothbard (1975), Table 8. 37. We elaborate the role of these factors in the sectoral studies of Section 5, below. 38. From 1920 to 1929, real private nonresidential construction spending in the United States rose by a cumulative 56%. Annual nonresidential real estate spending exceeded $5

230

BARRY EICHENGREEN AND KRIS J. MITCHENER

billion in each of these years (up from $3 billion in the immediately preceding period); construction activity was most intense in the central business districts of cities like New York, Boston and Detroit. The value of commercial contracts awarded peaked in 1927–1928, coincident with the peak in the composite credit boom indicator. Given the need for time to build, the process exhibited persistence: large commercial real estate projects like the Empire State Building were only finalized in 1929. (The Empire State Building actually broke ground only in March 1930; by 1931 it was being referred to as the “Empty State Building.”) See Hoyt (1933). 39. The classic reference is Toniolo (1980). 40. We will have more to say about some of these countries, Australia in particular, below. 41. See Johansen (1987). Denmark is not conventionally regarded as a country with chronic financial problems in the second half of the 1920s, although the analysis here suggests that it may have had more in common with Britain than commonly believed. Consistent with this interpretation, Denmark was quick to follow the U.K. off the gold standard in 1931 and then joined the sterling area. 42. The large flow of capital and gold to France in this period affected the rest of the world in the same manner, as observed in histories of the period (e.g. Johnson, 1998). 43. For example, a regression of the change in real GNP per capita between 1928 and 1932 on the absolute value of the trade balance relative to GNP in 1928 and the 1928 value of the composite indicator yields (with t-statistics in parentheses): y = −97.95 − 32.21 trade balance ratio + 54.83 credit boom (0.37)

(0.94)

(2.32)

F = 3.54, R-squared = 0.37. 44. Green and Sparks (1988) contrast the Australian and Canadian recoveries and attribute the timing of the turnaround to the identity of their principal trading partners: Australia’s main export market, the U.K., also began recovering at the end of 1931, whereas recovery in Canada’s principal export market, the U.S., was delayed until 1933. 45. Note also that the R-squared of the regression (of the fall in output between 1929 and either 1931 or 1932 on the one hand and the deviation from trend of the boom indicator in 1928) is higher when we use the deviation of share prices from trend than when we use the composite. One way of understanding this is that the impact of the stock market was felt partly insofar as it also affected the other components of the composite indicator. Although these linkages existed, they worked in opposite directions and were subject to variable lags. Table 2 above documented the positive association of equity valuations with investment. At the same time, however, the fluctuation of share prices affected the excess supply of money and credit in the other direction. A higher level of q which stimulated investment would have also raised the denominator of the credit/GNP ratio, other things equal. With the other two components of the composite indicator moving in opposite directions in response to the rise of share prices, but subject to complex and variable lags, it is not entirely surprising that these other two components added more noise than information content useful for forecasting output. 46. Borio and Lowe (2002) do something along these lines. When this is done separately for currency and banking crises, it yields slightly different composite indicators for the two cases, although the prevalence of twin crises in the 1930s dictates that the differences in the two variants are small. In practice, this means picking weights of 0.26, 0.40 and 0.34

The Great Depression as a Credit Boom Gone Wrong

231

on the M2/GDP, investment/GDP and equity price/CPI ratios for currency crises, and of 0.38, 0.32, and 0.30, respectively, for banking crises. Banking and currency crisis dates are taken from Bordo, Eichengreen, Klingebiel and Martinez-Peria (2001). Conveniently, this is the same source as used by Borio and Lowe for the recent period. 47. The significance levels declined, which is understandable given very limited degrees of freedom. The composite indicator was now significant at the 90% confidence level, while the two-way interaction of credit and the stock market was significant at the 95% level, and the two-way interaction of credit and investment just missed significance at the 90% level. 48. Indeed, Merritt (1991) criticizes the banks for the conservatism of their investment behavior in the 1920s. 49. And it is partly due to different macroeconomic policies after 1929 – Australia being early to abandon the gold standard, the U.S. being relatively late. 50. The speed with which those controls were removed thus played a role in shaping the construction boom. In countries where the removal of rent control was delayed, the incentive for the private sector to undertake new construction projects was correspondingly less. Countries such as Belgium, Denmark, Norway, and France that were slow to remove rent restrictions in the 1920s or only did so partially (Table 3) experienced delayed growth or only modest growth in residential housing. In contrast, Finland, Sweden, and the Netherlands abolished rent control altogether in the 1920s, the U.K. began to phase out its laws in 1923, and Canada and the United States never adopted comprehensive rent control at the national level. In these countries, prices were freer to respond to the increase in demand for housing. The construction industry in turn responded to the market signals, undertaking building activity that was fueled by ample credit from building societies, mortgage banks, and insurance companies. 51. Even after the initial postwar deflation, wage rates in the British building trades (circa 1923) remained 90% above 1914 levels for craftsmen and fully 115% for unskilled workers. Given the lag between price and wage adjustment in the 1920s, how and when countries stabilized their currencies appears to have mattered for the course of their subsequent housing booms. In particular, countries that deflated in the effort to restore prewar parities often saddled construction with higher labor costs that damped the response of the industry. These considerations go some way toward explaining the precocious timing of the U.S. construction boom. The country had no wartime depreciation to be reversed and no postwar depreciation to be halted; continual maintenance of the gold standard encouraged long-term financial commitments. It largely completed the necessary deflation in the initial postwar years, avoiding extended disjunctures between prices and labor costs. 52. Not surprisingly, analysts of the U.S. economy have placed considerable weight on the deterioration of household balance sheets as a factor depressing consumer spending in the subsequent slump (Mishkin, 1978). 53. This observation is not original. Crick (1929, p. 103) argues that installment credit did more to amplify the business cycle upswing in the U.S. because it started from a higher base and its use was more evenly spread over the population. In the U.K., in contrast, “the net result is a comparatively small expansion in the total volume of installment buying on the upward phase of the business cycle.” 54. The first instance of an installment credit plan in the United States of which we are aware was that introduced in 1807 by Cowperthwaite & Sons of New York, a furniture store. Scott (2003) argues that the phenomenon emerged in Britain in the second quarter of the century.

232

BARRY EICHENGREEN AND KRIS J. MITCHENER

55. The National Association of Finance Companies was then formed with the object of standardizing the installment business. The NAFC laid down rules for, inter alia, deposits and maturities, and in 1925 the American Rediscount Corporation was established to act as a kind of proprietary reserve bank for finance companies. The history of the ARC is yet to be written. 56. Bowden and Turner (1993, p. 252). 57. The first Canadian sales finance company, the Continental Guaranty Corporation Canada, was formed in 1916, coincident with the growth of motor vehicle purchases (Neufeld, 1972). A specialized company, the Fidelity Contract Corporation, had been established in 1904 to discount piano paper, and it had been joined by several competitors by 1916. 58. UMFC went public in 1928. In the U.K. the practice was known as hire purchase rather than installment purchase. Reflecting differences in legal convention, in the U.S. case ownership of the goods passed to the consumer, the seller or finance company merely retaining a lien. Under hire purchase, in contrast, the consumer leased the good (whose title thus resided with the financier) with an option to buy. 59. The Board of Trade estimated that by the late 1930s, hire purchase agreements were used in more than 70% of sales for cars and bicycles, working class furniture, and electrical household equipment (Hoovers, audio equipment), while trade estimates suggest that it accounted for at least this proportion of pianos and sewing machines. 60. Olney (1991, p. 27). 61. Olney (1991, Table 4.4) suggests that the share may have been only slightly lower than that for automobiles. 62. Bowden and Turner (1993) find that income distribution was very important for explaining the diffusion of motor vehicle ownership in the U.K. and that a more uneven income distribution than in the U.S. led to significantly slower diffusion. 63. Ford switched over to the Model A in 1927, as noted above, not without consequences for the course of the boom. 64. Although, just as in the case of the second half of the 1990s, there remains dispute over the precise magnitude of the acceleration and the sectors in which it was centered. 65. The quarterly market capitalization index is based on 10 industries, using SIC-level data from CRISP. The market capitalization figures for the 10 industries are summed and then indexed (1929:01 = 100). 66. This, of course, is a familiar argument, also highlighted by, inter alia, Schumpeter (1939). 67. Again, the argument has important precedents, such as Minsky (1986) and Kindleberger (1978). 68. Can the behavior of the high-tech sectors in the 1920s help us to distinguish between the credit-boom and bubble interpretations? That the run-up in the stock market was most pronounced among high-tech firms and particularly evident in the United States, the seed bed of the new industries, might seem like prima facie evidence for the bubble interpretation; as Perez argues, bubbles seem to be associated with the early emergence of network technologies of great promise but uncertain short-run profitability. However, accommodating credit conditions play an important role in the response of the securities markets, even in Perez’s own story. It is not technology but the interaction of technology with financial conditions that matters for her story. The technological impulse propelling the stock market may have been exogenous from the present point of view, but had credit market conditions been tighter, due to some combination of a more restrictive monetary

The Great Depression as a Credit Boom Gone Wrong

233

policy and a less dynamic financial structure, the response of securities prices would have been less. 69. To be sure, the inelasticity of the currency under the gold standard created other problems, such as pronounced fluctuations in money and credit over the cycle that could cause financial stringency and distress in the banking system, including, in the worst case, financial crises. It is revealing that the Federal Reserve System was established precisely in order to provide “an elastic currency.” 70. Mining and prospecting activity and the incentive to melt down jewelry for coinage in periods of deflation lent some elasticity to global gold supplies, as emphasized by contemporary observers, but the magnitude of this response was limited at business cycle frequencies; see Rockoff (1984). 71. Recall our discussion of the French credit boom of the late 1920s, which is couched in exactly these terms. 72. We are missing data for the early part of the pre-1914 period for Canada, Denmark, Italy and Sweden (due to gaps in the stock market series). 73. This is also true when we use signal/noise ratios constructed on the basis of banking crises or currency crises. 74. A variety of sensitivity analyses confirmed these results. Thus, we used different detrending schemes (for example, fitting log-linear rather than linear trends to the stock market series); the results were in all cases virtually identical. We weighted the components of the composite using signal/noise ratios, as explained above. We used different weights for the four subperiods (1880–1913, 1919–1938, 1945–1971, and 1973–1997); again, none of our results was affected. We computed the standard deviation of the entire series (rather than simply for those portions where the composite indicator was above trend), in which case we were able to detect no significant differences between the gold standard years and other periods. 75. Note that any bias in the volatility of estimates of pre-1914 national income would work against this conclusion (recent authors having argued that conventional estimates of pre-1914 GNP may be excessively volatile), which only reinforces our finding. 76. Recall also that we find that the credit boom of the 1920s was heavily concentrated in a handful of countries and that equity prices rather than the supply of domestic credit were the most important contributing factor. Our preceding discussion suggests that these are precisely the circumstances in which a pegged exchange rate would amplify such booms: as equity prices rose, stimulating investment and increasing the demand for credit, capital would flow in to arbitrage interest differentials, rendering credit more elastic and deactivating one mechanism (scarcity of funds and higher interest rates) that would work to limit the boom. The question, of course, is whether there was any difference in the factors initiating credit booms before and after 1914. The popular view of pre-1914 expansions and recessions is that they were mainly driven by investment and asset price booms and collapses, not by monetary policy. Thus, it is not clear that this provides an explanation for the apparent contrast between the pre-1914 period and the 1920s evident in the data.

ACKNOWLEDGMENTS We thank Pipat Luengnaruemitchai and Justin Jones for research assistance and Michael Bordo, Alex Field, Charles Goodhart, and Ian McLean for comments.

234

BARRY EICHENGREEN AND KRIS J. MITCHENER

We dedicate this paper to the memory of Charles Kindleberger, whose passing coincided with its completion.

REFERENCES Aukrust, O., & Bjerke, J. (1959). Real capital and Economic growth in Norway 1900–1956. Balboa, M., & Fracchia, A. (1959). Fixed reproducible capital in Argentina, 1935–1955. In: R. Goldsmith & C. Saunders (Eds), The Measurement of National Wealth, Series 8. London: Bowes & Bowes. Bernanke, B. (2000). Essays on the great depression. Princeton: Princeton University Press. Bernanke, B., & Gertler, M. (1999). Monetary policy and asset volatility. Federal Reserve Bank of Kansas City Economic Review, 84(4), 17–51. Bernanke, B., & James, H. (1991). The gold standard, deflation and financial crises in the Great Depression: An international comparison. In: R. G. Hubbard (Ed.), Financial Markets and Financial Crises (pp. 33–68). Chicago: University of Chicago Press. Bloomfield, A. (1959). Monetary policy under the gold standard. New York: Federal Reserve Bank of New York. Bordo, M., Eichengreen, B., Klingebiel, D., & Martinez-Peria, S. M. (2001). Is the crisis problem growing more severe? Economic Policy, 32, 51–82. Bordo, M., & Jonung, L. (1987). The long run behavior of the velocity of circulation: The international evidence. New York: Cambridge University Press. Borio, C., Furfine, C., & Lowe, P. (2001). Procyclicality of the financial system and financial stability: Issues and policy options. BIS Papers, 1, Basel: BIS. Borio, C., & Lowe, P. (2002). Asset prices, financial and monetary stability: Exploring the Nexus. BIS Working Paper, 114 (July). Bowden, S., & Turner, P. (1993). The demand for consumer durables in the United Kingdom in the interwar period. Journal of Economic History, 53, 244–258. Bullard, J. B., & Schaling, E. (2002). Why the Fed should ignore the stock market. Federal Reserve Bank of St. Louis Review, 84(March/April), 35–41. Cairncross, A. (1953). Home and foreign investment. Cambridge: Cambridge University Press. Cecchetti, S. (1992). Prices during the great depression: Was the deflation of 1930–1932 really unanticipated? American Economic Review, 82, 141–156. Cecchetti, S., Genberg, H., Lipsky, J., & Wadhwani, S. (2000). Asset prices and central bank policy. In: Geneva Reports on the World Economy (Vol. 2). London: CEPR. Clarke, S. V. O. (1967). Central bank cooperation 1923–1931. New York: Federal Reserve Bank of New York. Cole, H. L., Ohanian, L. E., & Leung, R. (2002). Deflation, real wages, and the International great depression: A productivity puzzle. Unpublished manuscript, Federal Reserve Bank of Minneapolis. Crick, W. F. (1929). The economics of installment trading and hire-purchase. London: Pitman and Sons. Dominguez, K., Fair, R., & Shapiro, M. (1988). Forecasting the great depression: Harvard vs. Yale. American Economic Review, 78, 595–612. Economist (2002). The unfinished recession: A survey of the world economy. 28 September (Suppl.). Eichengreen, B. (1992). Golden fetters: The gold standard and the Great Depression, 1919–1939. New York: Oxford University Press.

The Great Depression as a Credit Boom Gone Wrong

235

Eichengreen, B., & Wyplosz, C. (1988). The economic consequences of the Franc poincar´e. In: E. Helpman et al. (Eds), Economic Effects of Government Expenditure. Cambridge, MA: MIT Press. Field, A. J. (1992). Uncontrolled land development and the duration of the depression in the United States. Journal of Economic History, 52, 785–805. Field, A. J. (2003). The most technologically progressive decade of the century. American Economic Review (forthcoming). Filardo, A. (2000). Monetary policy and asset prices. Federal Reserve Bank of Kansas City Economic Review, 85(3), 11–37. Ford, A. (1962). The gold standard: Britain and Argentina. Oxford: Clarendon Press. Fratianni, M., & Spinelli, F. (1997). A monetary history of Italy. Cambridge: Cambridge University Press. Galbraith, J. K. (1972). The Great Crash. Boston: Houghton-Mifflin. George, H. (1879). Progress and poverty. Reprint. New York: Robert Schalkenbach Foundation. Goldsmith, R. (1985). Comparative national balance sheets: A study of twenty countries, 1688–1978. Chicago: University of Chicago Press. Goldstein, M. (1998). The Asian financial crisis. Washington, DC: Institute for International Economics. Goodhart, C. (2000). Asset prices and the conduct of monetary policy. Unpublished manuscript, LSE. Green, A. G., & Sparks, G. R. (1988). A macro interpretation of recovery: Australia and Canada. In: R. G. Gregory & N. G Butlin (Eds), Recovery from the Depression: Australia and the World Economy in the 1930s (pp. 89–112). Cambridge: Cambridge University Press. Grossman, R. (1994). The shoe that didn’t drop: Explaining banking stability during the great depression. Journal of Economic History, 54, 654–682. Hamilton, J. D. (1992). Was the deflation of the great depression anticipated? Evidence from the commodity futures market. American Economic Review, 82, 157–178. von Hayek, F. A. (1925). Die wahrungspolitik der vereingten staaten seit der uberwindung der krise um 1920. Zeitschrift fur Volkswirtschaft und Sozialpolitik, 5, 25–65, 254–317. von Hayek, F. A. (1932). Reflections on the ‘Pure Theory of Money.’ In: J. M. Keynes, Part II, Economica, 12, 22–44. Hoyt, J. (1933). One hundred years of land values in Chicago. Chicago: University of Chicago Press. Istituto Centrale di Statistica (1986). Sommario di Statistiche Storiche 1926–1985. Rome: ISTAT. Johansen, H. C. (1987). The Danish economy in the twentieth century. London: Croom Helm. Johnson, H. C. (1998). Gold, France and the great depression, 1919–1932. New Haven: Yale University Press. Kent, C., & D’Arcy, P. (2002). Cyclical prudence: Credit cycles in Australia. Unpublished manuscript, Reserve Bank of Australia. Keynes, J. M. (1930). A treatise on money. London: Macmillan. Kindleberger, C. P. (1973). The world in depression, 1929–1939. Berkeley: University of California Press. Kindleberger, C. P. (1978). Manias, panics and crashes. New York: Basic Books. Kuznets, S. (1946). National income since 1869. New York: National Bureau of Economic Research. Laidler, D. (1999). Fabricating the Keynesian revolution. Cambridge: Cambridge University Press. League of Nations. (1930). Interim report of the gold delegation. Geneva: League of Nations. League of Nations. (1936). World Economic Survey 1934–1935. Geneva: League of Nations. League of Nations. (1939). Urban and Rural Housing. Geneva: League of Nations. Lewis, W. A. (1949). Economic survey 1919–1939. London: Allen & Unwin. Maddock, R., & McLean, I. W. (Eds) (1987). The Australian economy in the long run. Cambridge: Cambridge University Press.

236

BARRY EICHENGREEN AND KRIS J. MITCHENER

McKinnon, R., & Pill, H. (1997). Credible economic liberalization and overborrowing. American Economic Review Papers and Proceedings, 87, 189–193. McNeill, W. (1986). American money and the Weimar Republic. New York: Columbia University Press. Meltzer, A. N. (2003). A history of the federal reserve, Vol. 1: 1913–1951. Chicago: University of Chicago Press. Merritt, D. T. (1991). Financial institutions under pressure: Are the right lessons being learnt from the 1890s? Economic Papers, 10, 1–10. Minsky, H. (1986). Stabilizing an unstable economy. New Haven: Yale University Press. von Mises, L. (1924). Theorie des geldes und der umlaufsmittel, (2nd ed.) Munich. Translated and reprinted as: The theory of money and credit. (1953). New Haven: Yale University Press. Mishkin, F. (1978). The household balance sheet and the Great Depression. Journal of Economic History, 38, 918–937. Mitchell, B. R. (1988). British historical statistics. Cambridge: Cambridge University Press. Mitchener, K. (2003). Bank supervision, regulation and financial instability during the great depression. Journal of Economic History, 63, 525–531. National Accounts of the Netherlands (2004), CBS Infoservice datafile. Neufeld, E. P. (1972). The financial system of Canada. Toronto: Macmillan. Noyes, A. D. (1938). The market place. Boston: Little, Brown. Nurkse, R. (1944). International currency experience. Geneva: League of Nations. Olney, M. (1990). Buy now, pay later: Advertizing, credit and consumer durables in the 1920s. Chapel Hill: University of North Carolina Press. Patat, J.-P., & Lutfalla, M. (1990). A monetary history of France in the twentieth century. New York: St. Martin’s Press. Perez, C. (2002). Technological revolutions and financial capital. Cheltenham: Edward Elgar. Rappaport, P., & White, E. (1993). “Was there a bubble in the 1929 stock market? Journal of Economic History, 53, 549–574. Robbins, L. (1934). The Great Depression. New York: Macmillan. Rockoff, H. (1984). Some evidence on the real price of gold, its costs of production, and commodity prices. In: M. D. Bordo & A. J. Schwartz (Eds), A Retrospective on the Classical Gold Standard, 1821–1931 (pp. 613–650). Cambridge: Cambridge University Press. Rothbard, M. N. (1975). America’s greatest depression. Kansas City: Sheed and Ward (3rd ed.). Santoni, G., & Dwyer, G. (1990). Bubbles or fundamentals: Evidence from the great bull markets. In: E. White (Ed.), Crashes and panics: The lessons from history (pp. 188–210). New York: Irwin. Schedvin, C. B. (1970). Australia and the great depression. Sydney: Cambridge University Press. Schumpeter, J. (1939). Business cycles. Philadelphia: Porcupine Press. Scott, P. (2003). The twilight world of interwar British hire purchase. Past and Present (forthcoming). Smiley, G., & Keehn, R. H. (1988). Margin purchases, brokers’ loans, and the bull market of the twenties. Business and Economic History, 17, 129–142. Temin, P. (1976). Did monetary forces cause the great depression? New York: Norton. Toniolo, G. (1980). L’economia dell’Italia fascista. Bari: Laterza. Tornell, A., & Westermann, F. (2002, December). The credit channel in middle income countries. NBER Working Paper No. 9355. Urquhart, M. C., & Buckley, K. A. H. (1965). Historical statistics of Canada. Toronto: Macmillan. Vanderblue, H. B. (1927a). The Florida land boom: Part I. Journal of Land and Public Utility Economics, III, 113–131. Vanderblue, H. B. (1927b). The Florida land boom: Part II. Journal of Land and Public Utility Economics, III, 252–269.

The Great Depression as a Credit Boom Gone Wrong

237

Vila, A. (2000). Asset price crises and banking crises: Some empirical evidence. www.bis. org/pubs/confer08.pdf. Voth, H.-J. (2002). With a bang, not a whimper: Pricking Germany’s ‘Stock Market Bubble’ in 1927 and the slide into depression. Unpublished manuscript, Pompeu Fabra University. White, E. (1990a). The stock market boom and crash of 1929 revisited. Journal of Economic Perspectives, 4, 67–84. White, E. (1990b). When the ticker ran late: The stock market boom and crash of 1929. In: E. White (Ed.), Crashes and Panics: The Lessons from History (pp. 143–187). New York: Irwin. Wicker, E. (1966). Federal reserve monetary policy 1917–1933. New York: Random House.

THE LENGTH AND THE DEPTH OF THE GREAT DEPRESSION: AN INTERNATIONAL COMPARISON Jakob B. Madsen ABSTRACT This paper examines the hypotheses that the length and the depth of the Great Depression were a result of sticky prices or sticky nominal wages using panel data for industrialized and semi-industrialized countries. The results show that price stickiness, particularly, and wage stickiness were key propagating factors during the first years of the Depression. It is found that prices adjusted slowly to wages, particularly in manufacturing. Manufacturing wages are also found to adjust relatively slowly to innovations in prices, but unemployment exerted strong downward pressure on wage growth.

1. INTRODUCTION The role of nominal rigidities during the Great Depression has a long history. Starting from Keynes (1936, 1939), Dunlop (1938) and Pigou (1933), a large body of literature has examined why the supply side was so slow to adjust to nominal demand shocks during the Great Depression. Following the long tradition of English economists, Keynes (1936) argued that wages are stickier than prices; however, based on the empirical findings of Dunlop (1938), Keynes (1939) conceded that prices may be stickier than wages due to monopoly pricing but, at the same time, he urged more empirical effort to uncover the relative importance Research in Economic History Research in Economic History, Volume 22, 239–288 © 2004 Published by Elsevier Ltd. ISSN: 0363-3268/doi:10.1016/S0363-3268(04)22005-5

239

240

JAKOB B. MADSEN

of price and wage stickiness in depressions in general. A more international approach was initiated by Eichengreen and Sachs (1985) using cross-country data for the industrialized and the semi-industrialized countries. Their evidence suggests that the length and the depth of the Depression across countries were strongly influenced by cross-country changes in nominal wages deflated by wholesale prices, and they argue that the price decline had adverse supply effects because sticky nominal wages prevented goods and labor markets from clearing. Using a more formal panel data approach, Bernanke and Carey (1996), henceforth B&C, show that the real wage channel was indeed a robust one in explaining the cross-section and time-series path of industrial production during the Depression. Furthermore, they find that nominal wages failed to respond to the increasing unemployment and only partially adjusted to the decreasing (wholesale) prices. Beenstock and Warburton (1986) argue that rising wages deflated by wholesale prices were the cause of the British unemployment over the period from 1929 to 1932.1 Bordo et al. (2001) show, from simulations of a general equilibrium model, that sticky nominal wages played an important role in propagating the monetary shocks during the 1929–1933 downturn in the U.S. Using the Layard-Nickell model of unemployment, Dimsdale, Nickell and Horsewood (1989) investigate sticky wages and sticky prices as possible explanations for the unemployment effects of the nominal demand shocks in the U.K. during the Depression. They find that nominal wage inertia and mismatch were mainly responsible for the increase in unemployment in the U.K. during the first years of the Depression and that lower import prices pushed up real wages in excess of their warranted level. The sticky wage hypothesis has not gone unchallenged, however. Bernanke and James (1991) and Eichengreen and Hatton (1988) suggest that there was not a clear-cut correlation between real wages and production when wages are deflated by a more appropriate deflator, namely product prices, for the U.K., the U.S., Germany, Japan and Sweden. Eichengreen and Hatton (1988) furthermore note that real product wages did not increase in Germany, despite Germany having one of the most severe depressions. Bernanke (2000) finds that “only when the [financial] PANIC variable is included does nominal wage growth have the correct (negative) sign . . . but it is not statistically significant” (p. 101). Regressing industrial production on prices and nominal wages, among other variables, Bernanke and James (1991) find that the estimated coefficients of nominal wages are insignificant. Dighe (1997) argues that the sticky wage hypothesis is inconsistent with the employment path from mid-1931 to mid-1933 in the U.S. and finds that the real wage increase from 1929 to 1931 was no more severe than in other contractions in the U.S. economy.

The Length and the Depth of the Great Depression: An International Comparison

241

Cole and Ohanian (1999) show that the detrended real wage in the U.S. between 1929 and 1933 decreased by almost 15% for the whole economy and rose only modestly for manufacturing. Based on this and other evidence, Cole and Ohanian (1999, 2001) argue that the sticky wage hypothesis is inconsistent with the evidence for the U.S. Based on a review of the collection of essays by Bernanke (2000), Margo (2000) concludes that wage stickiness remains an economic puzzle. Despite the critique, the sticky wage hypothesis has not been exposed to a systematic examination using more adequate cross-country sources of data. Previous international studies have focused solely on the manufacturing sector, have used severely biased price deflators, and have not investigated the sensitivity of the results to different sources of data, different sectoral coverage, different estimators, and cross-country variations in wage stickiness. More importantly, the alternative source of supply failure, the sticky price hypothesis, has not been rigorously examined, and rarely stressed as a possible source of supply failure. As shown below, sufficiently flexible prices could have cleared the goods and labor markets, except in the extreme case of completely sticky wages. Hence, some price stickiness must have prevailed under the sticky wage hypothesis. This paper examines sticky wages and sticky prices as possible sources of supply failure during the Depression, taking the approach that both sources of stickiness were potentially important and that one source of stickiness cannot be identified in isolation from the other. Acknowledging the less-than-perfect quality of the data, the sensitivity of the estimates to different sources of data, different sectoral coverage, different estimation periods, and different estimators that allow for errors-in-variables, is examined. Annual data are used throughout the manuscript due to data availability.

2. CASUAL EVIDENCE ON REAL WAGES AND OUTPUT DURING THE DEPRESSION Figure 1 presents the unweighted international average of real hourly wages for manufacturing and for the whole economy over the period from 1925 to 1937. The manufacturing wage data are from the same sources as those used by B&C, and are discussed, along with the other data, in the next section. Series 1 is average manufacturing wages deflated by wholesale prices and covers the 22 countries included in B&C’s sample. This series shows a 23% increase in real wages from 1929 to 1931, a slight increase from 1932 to 1933 and thereafter a slow decline. This time-profile suggests widespread nominal wage rigidity in the sense that wages failed to respond to the decreasing prices and increasing unemployment, at least in the first years of the Depression.

242

JAKOB B. MADSEN

Fig. 1. Real Wages. Notes: The following countries are included in the series. Series 1: Argentina, Austria, Australia, Belgium, Canada, Czechoslovakia, Denmark, Estonia, France, Germany, Hungary, Italy, Japan, Latvia, Netherlands, New Zealand, Norway, Poland, Sweden, Switzerland, the U.K., and the U.S. Series 2: Australia, Canada, Denmark, France, Germany, Italy, Japan, Spain, Sweden, the U.K., and the U.S. Series 3: Australia, Canada, Denmark, Finland, France, Germany, Italy, Japan, Spain, Sweden, the U.K., and the U.S. Series 4: Canada, Finland, France, Germany, Japan, Netherlands, the U.K., and the U.S.

Deflating manufacturing wages with the manufacturing value-added price deflator (Series 3) yields an 8% increase in real wages from 1929 to 1931 and an almost unaltered real wage thereafter. When using the manufacturing value-added price deflator, the modest increase in real wages in the first years of the Depression suggests that wage stickiness played a less important role in the supply failure than if wholesale prices are used as deflators. Using economy-wide data where economy-wide hourly labor costs are deflated by the economy-wide value-added price deflator for 8 countries for which data are available (Series 4) shows almost the same path as the manufacturing value-added deflated wages. Real wages increased by 7% from 1929 to 1931 and remained almost unaltered thereafter. Given that Series 3 is constructed for only 12 countries for which the manufacturing value-added price deflator or producer prices are available, Series 1 and 3 are not strictly comparable. However, using almost the same country set does not alter the conclusion. Series 2 shows manufacturing wages deflated by wholesale prices for the 10 countries that are jointly contained in Series 1 and 3.2 Series 1 and 2 move closely together, which suggests that the distinctive path between the

The Length and the Depth of the Great Depression: An International Comparison

243

wholesale deflated (Series 1) and value-added deflated (Series 3) series does not reflect a country selection bias. In conjunction with the doubts already raised in the literature as detailed in the Introduction, these results raise doubts about whether wage stickiness could have been solely responsible for the supply failure during the Depression. If the supply failure during the Depression was a result of increasing real wages, why did the real wage increase over the period from 1925 to 1929 not produce a Depression, assuming that the growth rates in the full employment marginal productivities of labor were the same over the periods 1925–1929 and 1929–1933? Wholesale-price-deflated wages (Series 1) increased by approximately 20% from 1925 to 1929, which corresponds to the increase in wholesale prices deflated wages over the period from 1929 to 1933. The value-added-deflated real wage series show’s an almost constant growth rate over the period from 1925 to 1933. If the supply failure was predominantly due to sticky wages, then real wages should have moved counter-cyclically and not pro-cyclically over the period from 1925 to 1929, assuming that the income fluctuations were mainly demand driven. Hence, real wages should have decreased, or at least have increased at a slower rate, over the period from 1925 to 1929. Turning to cross-country evidence, Fig. 2 shows little relationship between changes in manufacturing production and changes in manufacturing real product wages across countries over the downturn period from 1929 to 1932. Least squares regression suggests that there is not a statistically significant relationship between the change in real wages and the change in industrial production over the period 1929–1932, as shown in the first column of Table 1, regardless of whether wages are deflated by the value-added price deflator or wholesale prices. The estimates using wholesale prices include the 22 and 10 countries that are considered by B&C and Eichengreen and Sachs (1985), respectively. A weak,

Fig. 2. Changes in Real Wages and Industrial Production, 1929–1932.

244

JAKOB B. MADSEN

Table 1. Estimates of Cross-section Output Equations.

1 2 3

N = 12, PMAN N = 10, WPI N = 22, WPI

1929–1932

1932–1936

1929–1935

0.00 (0.01) 0.14 (0.64) 0.02 (0.17)

0.01 (0.04) −0.35 (0.67) −0.76 (1.76)

0.02 (0.17) −0.38 (1.99) −0.76 (1.76)

Notes: The change in the log of industrial production over the specified period regressed on the change in the log of real wages over the same period. WPI signifies that the wholesale price deflator has been used and PMAN signifies that the manufacturing value-added price deflator has been used. The numbers in parentheses are absolute t-statistics. Constant terms are included in the estimates but not shown. N = 22 are the 22 countries included in the estimates by B&C. N = 12 are the 12 countries for which manufacturing prices are available, and include the countries listed in the notes to Fig. 1. N = 10 are the 10 countries considered by Eichengreen and Sachs (1985).

statistically significant negative relationship between real wages can only be identified in the recovery phase from 1932 to 1936 if the B&C sample is used (second column). Considering the period from 1929 to 1935 in column 3, as done by Eichengreen and Sachs (1985), there is only a significant negative relationship between output and real wages if wholesale prices are used as the deflator and if the country sample of Eichengreen and Sachs (1985) is employed.3 The lack of a well-defined negative relationship between changes in industrial production and real wages is either: (1) because price stickiness played an important role and therefore that demand shocks did not result in movements along the labor demand schedule but a shift thereof; (2) because countries were exposed to supply shocks of different intensities; or (3) that the data are error-ridden. Hence, identification of sources of supply failure during the Depression requires a more careful analysis, which is undertaken in Sections 5, 6 and 7. Overall the casual evidence in this section suggests that neither sticky wages nor sticky prices could have been the sole explanation for the supply failure. First, if sticky wages were the sole explanation for the supply failure, then real product wages would have increased in all the countries that experienced a downturn during the Great Contraction from 1929 to 1932. However, real product wages decreased in Australia, Denmark and Finland over the period from 1929 to 1932/33. Second, given that industrial production fell on average by 20.6% in the 12 countries listed in Fig. 2, and real wages increased by less than 10% over the period from 1929 to 1932, an implausibly high real wage elasticity of output is required to render the sticky wage hypothesis as the sole explanation for the supply failure. Third, since real product wages increased almost as much from 1925 to 1929 as from 1929 to 1933, the 1925–1929 increase in real wages would have been consistent with a downturn, not an upturn. Fourth, since the average real

The Length and the Depth of the Great Depression: An International Comparison

245

wage increased from 1929 to 1932, the wage setting schedule must have shifted to the left in response to the adverse demand shocks. Nominal wage stickiness must consequently have played a role since supply shocks were predominantly positive, such as the reduction in real commodity prices.

3. DATA ISSUES AND THE MEASUREMENT OF WAGES AND PRICES One of the largest difficulties associated with the identification of the source of supply failure during the Depression is that wage and price data are measured with large errors and that the measurement errors in some of the data are likely to move systematically over time. The question is which data are the least likely to be measured with error and how to control for the errors-in-variables problem in the estimates. Why is the wholesale price index a misleading wage deflator? First, for most countries, non-manufactured raw materials weighed more than 50% in the wholesale price index (United Nations, 1954), and thus the collapse in commodity prices contributed substantially to the decline in wholesale prices in the first years of the Depression. For the U.S. over the period from 1929 to 1932, for instance, raw material prices fell by 47.8% (Warren & Pearson, 1937), whereas prices of industrial commodities decreased by only 23.2% (Liesner, 1989). Madsen (2001a) shows that the decline in agricultural prices dominated the decline in wholesale prices and argues that the agricultural decline was an important contributor to the Depression. Hence, the apparent influence on output of higher wages deflated by wholesale prices may be an illusion driven by the agricultural crisis. Another problem that is associated with the use of the wholesale price index is that it contains a large component of imported products (United Nations, 1954, p. 125). For New Zealand, for example, imported products had a weight of 58% in the wholesale price index. Furthermore, for a large proportion of countries, the wholesale price index included imports but excluded exports (United Nations, 1954, p. 154). This is contrary to the value-added principle, where imports are excluded but exports included. The heavy weight of import prices in the wholesale price index gives rise to serious problems in multi-country analyses of the downturn from 1929 to 1932/1933. In particular, one would expect to find that wholesale prices decreased the least in countries that depreciated their currencies and increased their tariffs on imports the most. This partly explains the finding of Eichengreen and Sachs (1985) that countries that were the first to go off the gold standard were also the first to recover from the Depression and experienced the least increase in real wages.4

246

JAKOB B. MADSEN

Deflating wages by wholesale prices is not only erroneous because prices of crude food and materials carried a large direct weight in the wholesale price index, but also because the output effects of changes in real commodity prices are not correctly accounted for when the wholesale price deflator is used. To see the latter more explicitly, consider the supply function for the firm under perfect competition:5


W Y = F K, Q , o P − P c ∂Q c /∂Y o o

c

 (1)

where Yo is gross output, K is the capital stock, Qc is the quantity of intermediate products, Po is the output price, and Pc is the price of intermediate products. Here gross output is a positive function of the capital stock and intermediate products and a negative function of real wages. Wholesale prices are given as a weighted sum of Po and Pc and prices of other products. The indirect bias from using wholesale prices as the deflator in real wages is that the decline in Pc lowered wholesale prices, thus increasing real wages, but should have resulted in a positive output response because [P o − P c ∂Qc /∂Yo ] increased. The only way to identify the supply schedule under the maintained hypothesis of perfect competition in the goods market is to deflate wages with either the value-added price deflator or with producer prices, while allowing real raw material prices to enter as separate arguments in the supply function. Since economy-wide value-added price deflators are available for some countries, the data requirements are met on an economy-wide scale. For manufacturing, however, the value-added price deflator is only available for a few countries, as detailed in the Data Appendix. For additional countries where manufacturing output prices are available, value-added price deflators are constructed as (0.8P o − 0.2P c ) where the prices were normalised to one on average over the period from 1926 to 1937.6 Although the value-added price deflator constitutes a major improvement over wholesale prices, it is still subject to measurement errors, which are dealt with in various ways as discussed below. The wage data are also measured with a large error that is probably much more serious than the measurement error in the value-added price deflator. There are two potential sources of errors in the wage data. First, hourly manufacturing earnings for both skilled and unskilled workers are only available for five out of the 24 countries considered here (U.S., Italy, Latvia, Poland and Spain). For the remaining 19 countries the wage data are of wider or different sectoral coverage than manufacturing, cover only a portion of the manufacturing employment (skilled workers, unskilled workers, capital city, men, etc.), or are recorded on a daily or weekly basis as detailed in the Data Appendix. These resulting of errors are likely to be substantial.

The Length and the Depth of the Great Depression: An International Comparison

247

Second, compositional changes may have biased wage changes. Cole and Ohanian (2001), Lebergott (1989) and Margo (1993) argue that the proportion of low wage earners decreased during recessions because the cyclical sensitivity of hours worked was higher for unskilled than skilled workers, thus biasing the fixed weight index of aggregate wages. They further argue that these compositional effects were particularly strong during the Depression, thus overstating the change in money wages over the period from 1929 to 1933. Moreover, Lebergott (1989) argues that the bankruptcy frequency was more pronounced among low pay firms. However, microeconomic evidence on wage rigidity is not consistent with these results (see Dighe, 1997; Hanes, 2000; Simon, 2001). Furthermore, comparing official aggregate wage indexes with fixed weight indexes derived from international occupational data for nine countries, B&C were unable to detect an aggregation bias in official aggregate wage indexes due to compositional effects. The importance of compositional effects in aggregate wage data is therefore not clear. What are the solutions to the measurement problems? The recommended solution to an errors-in-variables problem is to use instruments that are orthogonal to the measurement errors but highly correlated with the true variation in the variables and furthermore to use robust estimators if the instruments are weak. Due to the presence of both errors-in-variables and simultaneity problems in the estimates in this paper, however, consistent estimates can only be recovered if exogenous demand shifters, which are simultaneously uncorrelated with the measurement errors, are available. Unfortunately, this is not likely to be the case since the measurement errors in the data and particularly in wages, are likely to be correlated with the business cycle and therefore correlated with the exogenous demand shifters. It is therefore unlikely that instruments and robust estimators alone can solve the consistency problem. A complementary approach is called for. In addition to using instruments and alternative estimators, the sensitivity of the estimation results to different data sources, data coverage, instrument set, and restrictions imposed on the estimates, is examined. Furthermore, the sample is made as large as possible because the instrumental variable method does not yield unbiased estimates (Hahn & Hausman, 2002). The econometric methods that are used to deal with these issues are discussed in Section 5. Three different sources of wage data are used: Economy-wide hourly labor costs and two sets of hourly earnings, which in sectoral coverage are as close to manufacturing as possible using the same data sources as B&C (mainly ILO’s Yearbook of Labor Statistics). However, some countries have more than one potential useful wage series and it is not clear which is the best proxy for manufacturing hourly earnings. For Japan, for instance, two series for industrial daily earnings are reported in ILO’s Yearbook: One series is published by the Imperial Cabinet and

248

JAKOB B. MADSEN

another is published by the Bank of Japan. Without detailed knowledge of the data sources, sample, sampling method etc., it is not obvious which wage data are of the highest quality and the most representative for manufacturing hourly earnings. Hence, the data are subdivided into two sets. The first set is the same as the one used by B&C, judged from their Fig. 1, and the other set contains the same data as used by B&C except for six countries for which alternative data are available. Three different deflators are used; namely wholesale prices, the manufacturing value-added price deflator, and the economy-wide value-added price deflator. The reasons for estimating for manufacturing is because all international studies have focused on this sector and because the supply schedule is relatively well defined in this sector.7 Economy-wide data cover the whole economy and include indirect labor costs, such as contributions to social security and private pension funds, family allowances and private health insurance, which are not included in manufacturing wage data. Furthermore, economy-wide data are likely to be of a higher quality than manufacturing data given the importance that is attributed to economy-wide compensation to employees and value-added price deflators. Finally, only economy-wide data can be used to simulate the effects on economy-wide unemployment of various sources of supply failure using the model that is presented in the next section. These considerations suggest that manufacturing and economy-wide data are useful complements in identifying the sources of supply failure during the Depression.

4. A MODEL OF THE SUPPLY SIDE DURING THE DEPRESSION To analyse the sources and the consequences of the supply side failure during the Depression this section sets up a small supply model consisting of price and wage setting, which jointly determine the evolution of output at any given demand. The model draws on the model of Dimsdale, Nickell and Horsewood (1989), but differs in its treatment of import prices and allows for a Phillips curve effect in wage adjustment, that is, the deviation of unemployment from its natural rate is allowed to persistently affect wage growth. The latter feature is crucial for the persistence of shocks. Firms. The individual producer’s production function is given by: Q i = L ai , where Q is output and L is labor inputs and a is a constant, 0 < a ≤ 1.

The Length and the Depth of the Great Depression: An International Comparison

249

The demand for firm i’s products is given by the equation:
−␩ Pi = Y QD i P where ␩ is the price elasticity of demand, Y is aggregate value-added real income, Pi is the value-added price charged by the individual firm, and P is the general price level. The profit maximizing problem of the firm is given by:
−␩ Pi Pi W Pi W − Li (1) maxi /P = Q i − L i = Y P P P P P (Pi /P) where  is nominal profits. This optimization problem yields the solution as follows: Pi ␩ W L −1 (2) = a , P ␩−1 P Q which shows that the producer with market power sets price as a mark-up over marginal costs. Assuming asymmetry and that the price elasticity of demand is a function of demand, this equation can be written as:8
e L P = Y␪We a −1 Y where ␩/(␩ − 1) = Y ␪ and the superscript “e” stands for expectations. Decomposing expected wages and taking logarithms yields the price-setting equation as follows:9 p = ␪y + w − wu + (l − y)e − ln(a)

(3)

where lowercase letters signify logs of capital letters, and wu = w − we , is unexpected wages. This equation suggests that the mark-up of prices over marginal cost is a function of the output and wage surprises. Wage surprises influence mark-up by creating a wedge between the wage that firms expect to prevail when they set prices, and the actual wage outcome. If it is costly to change prices an unexpected wage reduction will increase mark-ups and reduce output. Wage setting. Consider the following standard wage setting model (Dimsdale et al., 1989; Madsen, 1998): w = ␣0 + p e + ␣1 (p cpi − p)e + ␣2 y + (y − l)e + z where pcpi is the log of consumer prices and z is a vector of wage push variables. Here, wages are indexed to the expected value of the value-added price deflator on a one-to-one basis following the natural rate hypothesis. Since the consumer

250

JAKOB B. MADSEN

price index is the relevant deflator for workers, the wedge between consumer prices and the value-added price deflator influences the wage outcome by the proportion ␣1 , which is determined by the relative bargaining position between firms and workers among other factors. Wage push factors, including the term (p cpi − p) are factors that push wages away from their full employment equilibrium. Excess demand for labor is measured by output to simplify the exposition in this section. This assumption is relaxed in the sections below. Decomposing expected prices yields: w = ␣0 + p − p u + ␣1 (p cpi − p)e + ␣2 y + (y − l)e + z

(4)

where p u = p − p e is unexpected prices and incorporates the wage-effects of price shocks due to contractual arrangements and co-ordination failures. If workers fail to coordinate a concerted reduction in nominal wages as a result of price deflation, then real wages will rise due to a widening of the gap between the expected and actual prices, (p e − p). Demand. Demand is a positive function of real money balances, M/P, and other demand shifters, X, y = (m − p) + x.

(5)

Equilibrium. Solving Eqs (3) and (4) yields the output equation as follows: y=

p u + wu − ␣1 (p cpi − p)e − z + ␣0 + ln(a) . ␪ + ␣2

(6)

In this model unexpected demand shocks have output effects because they change real ex post wages and mark-ups via the terms p u and wu . The output effects of demand shocks depend on four parameters: p u , wu , ␪, and ␣2 . Here, p u and wu , depend on the size of the demand shock and nominal rigidity. The parameters ␪ and ␣2 represent the adjustment of prices and wages to disequilibria in the labor and goods markets. If prices and wages are sensitive to demand shocks, then ␪ and ␣2 will be large and the output effects of demand shocks muted. In other words, the slope of the aggregate supply schedule is steep. In the model p u and wu need not only signify unanticipated outcomes but can also represent co-ordination failures among workers and firms. If a price deflation fails to cause a proportional reduction in nominal wages, then p u decreases and output declines. The same applies for firms if they fail to coordinate a price reduction when demand and wages fall. Under the assumptions of sticky wages but flexible prices and no price confusion, wu = 0 in the price equation, and demand shocks influence output via the p u term. Conversely, under the

The Length and the Depth of the Great Depression: An International Comparison

251

assumptions of sticky prices but flexible wages and no price confusion among workers, p u = 0 and demand influences output via the wu -term. When price and wage expectations are met, and hence p u = wu = 0, output settles at its potential, which corresponds to the natural rate of unemployment: y=

−␣1 (p cpi − p)e − z + ␣0 + ln(a) . ␪ + ␣2

Equation (6) is derived under the assumption that output, or more correctly the deviation of output from its potential, does not permanently affect the wage growth rate. This follows the framework of Layard and Nickell (1986) and is popularized by Blanchflower and Oswald (1994), who argue that the labor market is represented by a wage curve and not a Phillips curve, and, as a consequence, unemployment affects only the level of wages when unemployment changes. Once unemployment has stabilised after a shock, wages remain unaffected by the level of unemployment. This stands in contrast to the traditional Phillips curve framework where unemployment in excess of its equilibrium continues to put downward pressure on wage growth until the labor market disequilibrium is eliminated. The wage curve assumption is restrictive since several contemporaneous studies of the labor market find significant Phillips curve effects (see for instance the special issue on the Phillips curve in the Journal of Monetary Economics, 44(2), 1999). A Phillips curve effect can be incorporated into the model by writing Eqs (3) and (4) in first-difference form and adding to the first-differenced wage equation the terms ␣3 yt and ␣4 , where ␣4 is potential output corresponding to the natural rate of unemployment. This yields the output equation as follows: yt =

˙ u − ␣1 (p˙ cpi − p) ␣2 + ␪ ˙ e − z˙ + ␣4 p˙ u + w y t−1 + . ␣2 + ␣ 3 + ␪ ␣2 + ␣ 3 + ␪

(7)

This equation gives important information about the persistence of demand and supply shocks during the Depression. If wage growth is not allowed to depend on the level of output (unemployment), then ␣3 and ␣4 are zero, which implies that supply shocks have permanent output effects and demand shocks have persistent output effects. This aspect is analysed in Blanchard and Katz (1997). Suppose instead that there is no wage curve effect (␣2 = 0) and that prices are unaffected by demand shocks (␪ = 0), then Eq. (7) collapses to: yt =

˙ u − ␣1 (p˙ cpi − p) ˙ e − z˙ + ␣4 p˙ u + w , ␣3

which shows that supply shocks have only one-off effects on output and unemployment and demand shocks do not have persistent output effects because lagged

252

JAKOB B. MADSEN

output does not feed into current output. Intuitively, a shock that shifts the wage setting and the labor demand curves to the left increases the rate of unemployment. The higher unemployment puts downward pressure on wage growth and the wage setting schedule continuously shifts to the right until unemployment is eliminated. In the presence of a wage curve effect, output will also converge towards its initial equilibrium following a shock as long as the Phillips curve effect prevails, as seen from the coefficient of lagged output in Eq. (7). The adjustment in wages to the deviation in output and unemployment from their natural rates is therefore crucial to the persistence of demand shocks during the Depression, regardless of whether the supply failure was a result of labor market or goods market rigidities, and should therefore not a priori be ruled out in the empirical testing. Furthermore, since the adjustment of wages to price surprises, including downward nominal wage rigidity, cannot be adequately modeled given the difficulties that are associated with the measurement of price expectations and co-ordination failures, the unemployment term will automatically capture the part of the price surprise effect of a demand shock that is not captured by the price expectational error term. The change in the unemployment term, however, will not be able to capture the measurement errors in price surprises that last more than one year. Equation (7) is the key equation in the paper and will be used to determine the role of sticky wages and prices in the supply failure in the simulation exercises in Section 8. The model will be based on the estimates in the next sections. Before presenting the empirical estimates, the implications of wage and price rigidities on output and real wages are analyzed.

4.1. Demand Shocks Under Sticky Nominal Wages ¯ and prices are flexible. Suppose that nominal wages are constant, w = we = w, Then the solution to Eqs (3) and (5) gives the equilibrium output as follows: y=

¯ m + x + ln(a) w (y − l)e − + . 1+␪ 1+␪ 1+␪

In this scenario output is demand determined and is only influenced by the supply side due to productivity shocks that lower marginal costs. Since expected productivity probably did not change much during the Depression, this equation shows that output is solely demand driven and that wage push factors are unimportant for output since nominal wages are assumed to be fixed.10 This model corresponds to a flat aggregate supply schedule.

The Length and the Depth of the Great Depression: An International Comparison

253

Fig. 3. Sticky Wages, Flexible Prices, and Competitive Goods Market.

Figure 3 illustrates the behavior of real wages following an adverse demand shock, where (W/P) is real wages, L is labor, and LS and LD are labor supply ¯ and demand curves. The real wage is initially given by (W/P) 0 and the distance between A and B gives the number of unemployed. An adverse demand shock ¯ reduces prices and brings real wages up to (W/P) 1 and the number of unemployed increases to the distance between C and D. The labor demand curve remains unaffected by the demand shock because the goods market is assumed to be perfectly competitive and therefore price mark-ups over marginal costs are constant at one. In this case real wages are counter-cyclical and firms will always be on their notional labor demand curve given by Eq. (3).

4.2. Demand Shocks Under Sticky Prices If prices are sticky (p = p) ¯ but wages are flexible, then Eq. (5) reduces to: y = (m − p) ¯ + x,

(8)

for p ≥ mc, where mc is marginal cost. In this framework output is entirely demand determined up to the point where p = mc. This case corresponds to the case where the aggregate supply curve is horizontal and output is independent of the real wage path up to the point where p = mc, as illustrated in Fig. 4. The labor demand curve is vertical up to the point A in Fig. 4 and employment is demand determined. The labor demand curve becomes downward sloping when real wages exceed the point A because from that point, marginal costs exceed prices and producers will respond by lowering output until the marginal

254

JAKOB B. MADSEN

Fig. 4. Labour Market Under Price Stickiness.

conditions are satisfied. Below the point A, employment, and therefore output, is independent of real wages and lower nominal wages will not help clear markets. A negative demand shock shifts the labor demand schedule to the left and results in a reduction in real wages. If labor is on its supply schedule, real wages are pro-cyclical. If not, then the cyclicality of real wages is undetermined.

4.3. Some Degree of Both Wage and Price Stickiness In the intermediate case nominal wage and price stickiness coexist. Consider Fig. 5, where the wage-setting curve represents Eq. (4) and the labor demand curve represents the price setting model as given by Eq. (3). Unemployment is given by the horizontal distance between WS and the intersection between LD and LS , where LS is labor supply. Suppose that an adverse nominal demand shock shifts the wage-setting curve to the left from W S0 to W S1 , mainly because nominal wages are sticky. The labor demand curve shifts to the left by a magnitude that is determined by the cyclicality of the elasticity of demand, strategic interaction, the sensitivity of bankruptcies to demand, and other factors (Lindbeck & Snower, 1994). The direction of changes in real wages cannot be determined a priori. Therefore, real wages are either counter-cyclical or pro-cyclical; however, the cyclicality of real wages is muted compared to the extreme cases of only price rigidity or wage rigidity.

The Length and the Depth of the Great Depression: An International Comparison

255

Fig. 5. Labour Market Under Imperfections in Goods and Labour Markets.

5. THE NEXUS BETWEEN OUTPUT, PRICES AND WAGES This section tests the role of real wages for output and whether the identifying assumption of perfect competition in the goods market can be approximately maintained during the Depression. In the case of sticky wages but flexible prices it was shown in the previous section that output was a negative function of real wages and given by Eq. (3) for ␪ = 0. Following B&C, the output equation is estimated using annual data for country i at time t using the stochastic specification as follows: y it = ␤0 + ␤1 y i,t−1 + ␤2 wit + ␤3 p it + ␤4 Str it + ␤5 Panic it + TD ␰ + CD ␨ + ␧1,it ,

(9)

where Str is the log of the number of working days lost in strikes and lockouts, Panic is a binary dummy variable taking the value of one in periods of panics and zero otherwise, as classified by Bernanke and James (1991), TD is a N(T − 2) matrix of time-dummies, CD is a (N − 1)T matrix of fixed effect country dummies, ␰ and ␨ are vectors of fixed coefficients, where N is the number of countries and T is the estimation period, and ␧ is a zero-mean, independent, and identically distributed error term.11 Strikes, Panic dummies, and time-dummies are included in Eq. (9) following B&C. As discussed by Bernanke and James (1991), the banking

256

JAKOB B. MADSEN

panics hindered the functioning of intermediation and increased the effective cost of capital. The time-dummies capture omitted time-effects that are common across countries. They may help to identify the supply function, but may also reveal a source of supply failure that is different from wage rigidity, as discussed below.

5.1. Econometric Issues Equation (9) is estimated using non-linear generalised least squares to allow for first-order serial correlation in the residuals and the serial correlation coefficient is restricted to be the same across countries following B&C. The inverse of the estimated variances across countries are used as weights to allow for cross-country heterogeneity. Both the within (fixed effect) and the first difference estimators are used in the estimates of Eq. (9) for several reasons as given in the Appendix. Equation (9) is identified using exogenous demand shift factors under floating and fixed exchange rates as instruments. Two sets of instruments are used. Instrument set I is identical to the one used by B&C (str and Panic, lagged nominal wages and output plus M1 for off-gold countries and domestic discount rates, import prices for countries on gold, fixed effect dummies and time-dummies). Instrument set II contains the domestic stock of currency in circulation, which is used as instruments in off-gold standard periods, the world nominal short-term interest rate and export price competitiveness in on-gold standard periods, str, Panic, lagged output, fixed effect dummies and time-dummies. The world short-term interest rate is measured as a 3-month Treasury bill rate and is weighted by GDP in common currency in 1929, excluding the home country. Export price competitiveness is measured as a multilateral index where third country penetrations into the export markets are taken into account. Hence, the fact that U.S. producers compete with producers from other countries in, for example, the German market is taken into account in this competitiveness measure.12 All instruments enter contemporaneously and are lagged one period in all first stage regressions. There are two potentially serious problems associated with the use of instruments. First, demand shifters will not solve the errors-in-variables problem to the extent that the measurement errors are correlated with the exogenous demand shifters. Second, weak instruments will lead to inconsistent estimates and to a finite-sample bias that approaches that of OLS estimates (Bound et al., 1995; Hahn & Hausman, 2002). IV estimators are particularly biased in small samples (Donald & Newey, 2001). Various approaches are taken to deal with these issues. First, the estimates of the output equations are complemented with estimates using the limited information maximum likelihood (LIML) estimator to check the seriousness of the small

The Length and the Depth of the Great Depression: An International Comparison

257

sample bias following the recommendation of Donald and Newey (2001). Second, Sargan’s overidentifying restrictions test is undertaken to test whether the moment conditions are correct, that is, whether the instruments are valid (Verbeek, 2000). Third, following the recommendations of Bound, Jaeger and Baker (1995) first-stage F-statistics were undertaken for all equations. Bound, Jaeger and Baker (1995) show analytically that IV estimates are severely biased as the F-statistics in the first-stage regression approach 1 from above. However, since the F-statistics for overall significance of the excluded instruments in the first-stage regressions were significant at the 1% level, they are not shown in the estimates below.

5.2. Summary Statistics When examining the source of supply failure using regression techniques it is necessary to have identifying variations in the data. If not, then regression analysis may not reveal the source of supply failure because the coefficients may be biased towards zero. Suppose, for instance, that nominal wages are rigid and therefore show very little variation. From regression analysis one could conclude that wages are flexible because the variations in wages in the output and wage equations are too low for identification of the coefficients. The same argument applies for prices. The summary statistics in Table 2 suggest that the key variables, namely wages, prices, unemployment and production, all show sufficiently high variation for identifying purposes. All variables in the table are measured in percentage changes except the rate of unemployment, which is measured in first differences. Wages and the value-added price deflators have about the same variance, which is about half the variance of wholesale prices due to the heavy weight of commodity prices and prices of traded products in wholesale prices. Two sets of unemployment data are used, as discussed in detail in Section 6. One set is chiefly based on union statistics and is used by B&C (Ubc ). The other set is mainly based on economy-wide estimates for individual countries and is referred to as Ulow . Both sets show a large variation, especially Ubc , which varies substantially from year to year and shows quite extreme maximum and minimum values. Industrial production has the highest variance of the price and output series, and is more than three times as large as the variance in the manufacturing valueadded price deflator. This suggests that the manufacturing industry predominantly changed quantity and not prices in response to demand shocks and hence that price stickiness prevailed, although it does not reveal the extent of the price rigidity. By contrast, the variance in the economy-wide GDP is only about one and a half times the variance in the economy-wide value-added price deflator, which suggests that price-rigidity was much less widespread in the non-manufacturing sectors.

258

JAKOB B. MADSEN

Table 2. Summary Statistics in Percentage Changes. Variable

Mean

Var.

Min.

Max.

N = 22 Industrial production Wholesale prices Manufacturing wages Unemployment (Ubc )

1.31 −3.06 −0.75 1.22

1.27 0.62 0.21 40.0

−28.0 −23.0 −16.8 −15.7

30.6 15.4 16.0 21.5

N = 12 Industrial production Wholesale prices Manufacturing wages Man. value-added price deflator Unemployment (Ulow ) Unemployment (Ubc )

1.19 −2.41 −0.87 −2.02 0.33 0.50

1.10 0.63 0.34 0.33 7.59 21.9

−29.0 −21.2 −15.0 −20.8 −6.5 −15.7

20.3 15.4 18.6 14.4 9.3 13.8

N=8 Economy-wide GDP Hourly compensation Value-added price deflator Unemployment (Ulow )

1.02 −0.24 −2.07 0.50

0.40 0.43 0.25 7.26

−14.7 −24.7 −12.6 −6.5

13.0 11.8 10.8 7.6

Notes: Sample period: 1929–1936. All variables are measured as log first differences and multiplied by 100, except the rate of unemployment, which is measured in first differences. N = 22 is the country sample used by B&C, N = 12 is the sample of countries for which the manufacturing value-added price deflator is available (see Fig. 1), and N = 8 is the countries for which the economy-wide compensation to employees and the value-added price deflator are both available.

5.3. Estimation Results The results of estimating restricted and unrestricted versions of Eq. (9) are presented in Table 3. The estimated coefficients of panics, strikes, fixed effect dummies, constants and time-dummies are not shown in the table to preserve space. Rows 1 and 16 replicate the estimates of B&C using the same data, instruments, data period, and countries (henceforth referred to as the B&C baseline model).13 The estimates are remarkably similar to the estimates of B&C and suggest that real wages played a potentially important role during the Depression. The question is how robust the B&C results are to changes in data, specification, estimation period, and instrument set. Rows 3 and 18 show the B&C estimates with the only difference from the B&C baseline regressions being that wages for six countries show a stronger decline over the period from 1929 to 1933 than the series used by B&C. It is, however, not obvious which wage data are of the highest quality and are most representative for manufacturing hourly earnings. The absolute value of the estimated coefficients

The Length and the Depth of the Great Depression: An International Comparison

259

Table 3. Estimates of Output Equations. yt −1 Levels 1931–1936 N = 22, IV, WPI N = 22, WPI N = 22, IV, WPIa N = 22, IV, WPI, -TD N = 12, IV, WPI N = 12, IV, WPIb N = 12, IV, PMAN N = 12, IV, PMANb N = 22, LIML, WPI N = 12, LIML, PMANb N = 8, IVb 1927–1936 N = 22, IV, WPI N = 12, IV, WPIb N = 12, IV, PMANb N = 8, IVb First differences 1933–1936 N = 22, IV, WPI N = 22, WPI N = 22, IV, WPIa N = 22, IV, WPI, -TD N = 22, LIML, WPI N = 12, IV, WPI N = 12, IV, WPIb N = 12, IV, PMAN N = 12, IV, PMANb N = 12, LIML, PMANb N = 8, IVb 1929–1936 N = 22, IV, WPI N = 12, IV, PMANb N = 8, IVb

wt

pt

␹2 (6)

␹2 (1)

0.51(8.83) 0.55(8.56) 0.55(9.37) 0.63(11.3) 0.50(5.51) 0.61(5.93) 0.46(5.01) 0.65(6.51) 0.70(6.36) 0.67(4.79) 0.55(6.07)

−0.71(3.95) −0.30(2.49) −0.29(1.82) −1.68(7.48) −0.74(2.24) −1.24(2.89) −0.44(1.23) −0.59(1.38) −0.86(3.91) 0.02(0.00) −0.06(0.51)

0.65(5.21) 0.35(4.31) 0.52(4.00) 0.65(3.63) 0.67(2.71) 0.72(2.83) 0.25(1.22) 0.12(0.56) 0.67(3.35) 0.15(0.98) 0.84(7.61)

13.07 12.25 15.66 12.58 9.53 6.76 18.03 8.82 5.99 8.85 9.55

0.20 0.13 3.04 42.7 0.05 1.71 0.32 1.55 1.58 0.02 34.4

0.76(18.7) 0.82(13.5) 0.83(13.9) 0.67(7.66)

0.12(0.68) −0.29(3.36) −0.27(2.11) −0.07(0.93)

0.01(0.29) 0.05(0.44) 0.00(0.12) 0.59(4.10)

10.29 11.25 11.99 9.55

0.61 2.50 4.77 14.3

−0.14(0.82) −0.11(0.78) −0.12(0.73) 0.00(0.04) 0.31(1.72) 0.38(1.58) −0.02(0.13) 0.41(1.89) 0.01(0.05) 0.08(0.35) 0.51(3.80)

−0.74(3.95) −0.39(2.56) −0.11(0.27) −0.51(1.37) −0.58(1.76) −0.16(1.15) −0.99(2.60) 0.07(0.10) −0.96(2.47) −0.26(0.95) −1.20(3.50)

0.79(2.41) 0.12(1.39) 0.42(1.26) 0.78(2.35) 0.30(1.07) 0.28(1.75) 0.70(2.30) 0.01(0.04) −0.31(0.74) −0.10(0.25) 1.09(3.31)

10.02 11.33 9.91 9.77 5.62 12.69 15.82 12.89 12.28 12.32 9.55

0.03 2.76 1.07 1.03 2.50 0.02 0.17 0.02 5.45 1.43 0.20

0.86(3.81) 0.25(2.72) 0.17(1.09)

−0.09(0.21) 1.40(3.56) 0.24(0.86)

0.82(4.24) −0.28(0.81) 0.49(1.31)

11.04 15.80 10.01

6.97 9.88 6.62

Notes: The numbers in parentheses are absolute t-statistics. Panics, strikes, constants, fixed effect dummies (in level estimates) and time-dummies are included in all estimates but are not shown. N = 22 is the 22 countries included in the estimates by B&C. N = 12 is the 12 countries for which manufacturing prices are available, and include the countries listed the notes to Fig. 1. N = 8 is the 8 countries for which economy-wide valueadded price deflators and total labor costs are available, and include the countries listed the notes to Fig. 1. -TD signifies that time-dummies are excluded from the regression equations. The instruments used by B&C are used. IV = instrumental estimator. LIML = limited information maximisation likelihood estimator. PMAN = prices measured by value-added price deflator. WPI = prices are measured by wholesale prices. ␹2 (6) = Wald test for identifying restrictions, and is distributed as ␹2 (6) under the null hypothesis of instrument validity. ␹2 (1) = Wald test for the same coefficient of wages and prices and with opposite sign, and is distributed as ␹2 (1) under the null hypothesis of equality and opposite sign. a Alternative wage data for six countries as explained in the Data Appendix. b Instrument variable set II: Domestic stock of currency in circulation in the off-gold standard periods. The world nominal short-term interest rate and export price competitiveness are used as instruments in on-gold standard periods. One-period lags of dependent variables are included for all countries over the whole period.

260

JAKOB B. MADSEN

of wages is reduced substantially in the level estimates and is rendered statistically insignificant in the first-difference estimates, which suggests that the B&C results are sensitive to the choice of wage data. Excluding the time-dummies from the B&C base-line regressions renders the estimated coefficient of real wages substantially more negative in the level estimates and the hypothesis that ␤2 = −␤3 is rejected at any conventional significance level (row 4). In the first-difference estimates the estimated coefficients of wages is statistically insignificant (row 19). These results suggest that the time-dummies play an important role in the regressions and capture omitted variable effects. Since the time-dummies capture a common time-profile of the Depression among countries in the B&C baseline regressions, these results suggest that the B&C results are mainly driven by between country effects (cross-country variations) as opposed to within country effects. The estimated coefficients of the time-dummies were particularly significant in the level estimates in row 1: −0.136(11.98) in 1931, −0.183(5.64) in 1932, −0.095(3.15) in 1933, −0.045(3.35) in 1934, and −0.040(3.49) in 1935, where the numbers in parentheses are absolute t-statistics. The time-dummies therefore account for a 5% decline in production from 1931 to 1932, a 9% increase in production from 1932 to 1933, and a further 5% increase in production from 1933 to 1934. Extending the estimation period back four years to 1927 for the level estimates and 1929 for the difference estimates alters the estimation results substantially.14 The estimated coefficients of wages and prices are statistically and economically insignificant in the level estimates (row 12). In the first-difference estimates the estimated coefficient of wages is statistically insignificant but the estimated coefficient of prices remains significant and positive (row 27). This result is perhaps not surprising given that the real wage increase from 1927 to 1929, as seen from Fig. 1, coincided with a cyclical upturn. The result is consistent with Silver and Sumner’s (1995) finding of counter-cyclical real wages over the period from 1930 to 1939 for the U.S. using wholesale prices as deflators. The results from the longer estimation period raise two concerns. First, the B&C results are predominantly driven by the recovery phase although their focus is on the downturn, especially in their first-difference estimates, which only cover the recovery period from 1933 to 1936. Second, the coefficient estimates are highly sensitive to estimation period, which suggests that their estimates are either subject to an omitted variable or small sample bias. Under the alternative hypothesis of price rigidity, the omitted variable is mark-ups. Consistent estimates could be obtained from IV estimates if the instruments for wages are uncorrelated with mark-ups. However, it is difficult to find instruments that are uncorrelated with mark-ups and the coefficient of wages will consequently be biased under the alternative hypothesis of price rigidity.

The Length and the Depth of the Great Depression: An International Comparison

261

The LIML estimates in rows 9 and 20 are very close to the IV estimates in rows 1 and 16 and are consistent with the high significance of the F-tests for overall significance in the first stage regressions, and the tests for over-identifying restrictions in Table 3. The instrument validity tests are insignificant, except the estimates in row 7, which are marginally significant at the 1% level. These results indicate that the instruments have adequately dealt with the simultaneity and the errors-in-variables problem. Regressing the B&C baseline model for the 12 countries for which the manufacturing value-added price deflator is available does not alter the B&C results much (rows 5 and 21), except that the absolute value of the estimated coefficient of wages is reduced in the first difference estimates. Furthermore, using the alternative instrument gives very significant coefficient estimates of the coefficients of prices and wages (rows 6 and 22). These results suggest that the B&C results are not significantly affected by the country sample and the choice of instrument set. Using the value-added price deflator instead of wholesale prices alters the results substantially. For both manufacturing and the whole economy, the estimated coefficients of prices and wages are mainly insignificant regardless of instrument set, estimator, sectoral coverage, and estimation period (rows 7, 8, 11, 14, 23, 24, 25, 28 and 29). The sticky wage hypothesis is only supported by the economy-wide first-difference estimates in the recovery phase from 1933 to 1936 (row 26), for which the estimated coefficients of wages and prices have the right sign, and are significant, and the hypothesis that the estimated coefficients of wages and prices are the same and of opposite sign cannot be rejected at conventional significance levels. However, this result is not sufficient evidence for the sticky wage hypothesis given that the result changes significantly for changes in estimation method, estimation period, and sectoral coverage, and particularly when it is taken into account that these results only apply for the upturn phase. To investigate whether the estimates where prices are measured by the value-added price deflators, are biased due to errors-in-variables, the within and the first-difference estimates are combined to give consistent estimates using Eq. (A2) in the Appendix (corresponding to combining the estimates in rows 8 and 24 for manufacturing and rows 11 and 26 for the whole economy).15 In these models output is regressed on real wages and the deterministic variables since Eq. (A2) applies only for one stochastic regressor. The recovered coefficients of real wages are −0.29 and −0.22 for manufacturing and the whole economy, respectively, and are slightly lower in absolute terms if the longer estimation period is used. Using these estimated elasticities suggests that the approximately 10% real wage increase from 1929 to 1932 accounts for less than a 3% decline in output over the same period and therefore that real wages could not have been entirely responsible for the supply failure.

262

JAKOB B. MADSEN

These results are consistent with the results of three important other studies. For the U.K. Dimsdale, Nickell and Horsewood (1989) estimate labor demand in the interwar period and find that real product wages are insignificant but that cyclical demand is significant, where cyclical demand is used as a proxy for cyclical mark-ups. From labor demand estimates for 14 industrialized and semi-industrialized countries for the interwar period, Newell and Symons (1988) find that the estimated coefficients of real product wages are of either low significance or are insignificant. Clearly price stickiness must played a role if the labor demand schedule could not easily be identified. Similarly Bernanke and James (1991) regress the growth in industrial production on the growth in nominal wages and wholesale prices, among other variables, over the period from 1930 to 1936 using the ordinary least squares pooled estimator. Their estimated coefficients of nominal wages are statistically insignificant at any conventional significance level. This result is consistent with the OLS estimates in Table 3 where the estimated coefficient of nominal wages is insignificant in the first-difference OLS estimates over the period from 1929 to 1936 (row 27), which is similar to the estimation period used by Bernanke and James (1991). Overall the results in this section suggest that the results of B&C are sensitive to estimation period, instruments, measurement of wages and prices, model specification, and whether time-dummies are included in the estimates. The sensitivity of the results to estimation period is particularly serious because it does suggest that B&C’s baseline regressions are subject to omitted variables and measurement errors. In their estimates they omit the first crucial years of the Depression and miss out the decline almost completely in the first-difference estimates by starting in 1933. Combined with the fact that the time-dummies explain a large fraction of the time-profile of the Depression, this suggests that their results hinge very strongly on cross-country variations in the upturn. The results in this section, however, do not necessarily imply that real wages were unimportant during the Depression. The message of the results in this section is that the supply schedule has not been identified. The model in this section only identifies the supply schedule if prices are flexible, because the equations have been derived under the assumption of perfect competition in the goods market. The question here is therefore not whether the coefficients of real wages are significantly negative. If wages are rigid but prices flexible, then demand shocks will materialise in movements along the labor demand schedule, and the supply schedule is clearly identified and the coefficients of wages will be negative. If prices are sticky, however, nominal demand shocks will result in shifts in both the wage setting and the labor demand schedules, and the real wage-supply nexus becomes blurred, as is evident from Fig. 5. The supply schedule is consequently not identified. Furthermore, under imperfect competition, the causal link from

The Length and the Depth of the Great Depression: An International Comparison

263

prices to output is broken and firms determine output and prices simultaneously. The estimates in this section are therefore not an exercise in the importance of real wages on output but whether the hypothesis of full price flexibility can be maintained. Incorporation of mark-ups into the output equation should, in principle, enable identification of the source of supply failure. Unfortunately, the data requirements are too high for marginal costs to be estimated using the methods of Hall (1988) and Roeger (1995) for the interwar period. A simpler, and probably much more robust way of identifying sources of supply failure is to estimate wage and price equations separately and from these equations infer the sources of supply failure. This is done in the next three sections.

6. WAGE STICKINESS DURING THE DEPRESSION To test for the degree and the potential source of wage rigidity, the following equation, which nests the Phillips and the wage curves, is estimated over the period from 1927 to 1937: wit = ␵0 + ␵1 wi,t−1 + ␵2 p it + ␵3 U it + ␵4 U it + ␵5 prit + CD ␷ + ␧2,it ,

(10)

where U is the rate of unemployment measured in decimal points, and pr is the log of labor productivity, which is measured as real GDP divided by hours worked, but is only included in the economy-wide estimates since manufacturing data are not available for some countries.16 Log level estimates are not undertaken because no allowance for Phillips curve effects can be made in level estimates and, more seriously, because the residuals exhibited serious first-order serial correlation, which did not disappear when using the non-linear estimator to correct for firstorder serial correlation. The effects of the level and the change in unemployment on wage growth is again referred to as the Phillips curve effect and the wage curve effect, respectively. To gain efficiency and to minimise the small sample bias that is associated with the instrumental variable method, the estimation period starts approximately two years before the onset of the Depression and terminates one year after the estimates in the previous section. To gain further efficiency, the generalised instrumental variable method, where the covariance matrix is weighted by the correlation of the disturbance terms between countries, is used for the economy-wide estimates.17 This estimator is not used for manufacturing, where N = 12, because the generalised instrumental variable estimator is only feasible for T > N. Direct

264

JAKOB B. MADSEN

taxes, indirect taxes, and the ratio of import prices and the value-added price deflator, strikes and the panic dummies were initially included in the estimates for both manufacturing and for the whole economy, but were insignificant, which suggests that supply shocks were not sufficiently important during the Depression to be identified by the estimates. Prices are measured by the value-added price deflator and for comparison with the B&C estimates, also by wholesale prices. Modern versions of wage determination stress that both consumer prices and the value-added price deflator are relevant for wages because value-added price deflators are the important deflators for firms and consumer prices are the relevant deflators for workers (see Madsen, 1998). To allow for the possibility that both the value-added price deflator and consumer prices influenced wages, the log of the ratio between consumer prices and the value-added price deflator or wholesale prices was included as an additional regressor in Eq. (10). However, the estimated coefficient of the ratio was insignificant at the 5-percentage level in all estimates, suggesting that wages were not indexed to consumer prices but to the value-added price deflator. The stochastic regressors, except the lagged dependent variable, are instrumented using instrument set II (Str, Panic, the domestic stock of currency in circulation in off-gold standard periods, the world nominal short-term interest rate and export price competitiveness in on-gold standard periods, time-dummies, lagged dependent variable and two-period lags of the dependent variables as an instrument for the lagged dependent variable). An exception is in the estimates of the model where the sample of B&C is used (N = 22), for which instrument set I is used (str and Panic, lagged nominal wages and output plus M1 for countries off gold and domestic discount rates, import prices for countries on gold, time-dummies and two-period lags of the dependent variables as an instrument for the lagged dependent variable). The lagged dependent variable is instrumented using two-period lags of the dependent variable. Two sets of unemployment data, which are derived under two different sets of principles, are used in the estimates. One set is similar to the data used by B&C (Ubc ). The data are mainly from Galenson and Zellner (1957) and are based on trade union statistics. For countries for which unemployment statistics are not available, simulated unemployment rates are constructed using a method suggested by B&C. Namely, the rate of unemployment is regressed on manufacturing employment for the countries for which both series are available. The coefficient estimates are then used to simulate unemployment for the countries where manufacturing employment, but not unemployment, is available. The other set of unemployment data, Ulow , takes into account that the trade union records only cover a fraction of the labor force and often only include males (Grytten, 1995, p. 231). These unemployment estimates are substantially lower than the

The Length and the Depth of the Great Depression: An International Comparison

265

estimates reported by Galenson and Zellner. The source of unemployment data for each individual country is detailed in the Data Appendix. The country dummies in Eq. (10) do not have their usual interpretation of fixed effects dummies because the fixed effects have been removed by the first-difference transformation, but are included for an econometric and an economic reason. The econometric reason is that the country dummies remove the bias due to the possibility that the level of unemployment is measured with error. The economic reason is that it is the deviation of unemployment from its natural rate, and not the level of unemployment, that puts downward pressure on wage growth according to the natural rate hypothesis. Under the assumption that the “natural rate” of unemployment is constant but varies across countries, the country dummies will capture cross-country variations in the natural rate. No assumptions are made about the generation of price expectations in the estimates and the instrumental variable method certifies that the measurement error contained in actual prices as a measure of price expectations does not render the coefficient estimates inconsistent and biased. Lagged prices were initially included in the estimates to allow for slow adjustment of wages to innovations in prices but were statistically insignificant at the 5% level in all estimates. This suggests that instrumented contemporaneous prices best reflect expected prices. Furthermore, following Dimsdale, Nickell and Horsewood (1989), first differences of price changes were initially included in the estimates as proxies for inflation expectational errors under the assumption of extrapolative expectations. However, the estimated coefficient was significantly positive in all estimates. It cannot be excluded that this result reflects that accelerating inflation leads to workers intensifying their wage demands. However, the variable was deleted without affecting the estimates for the sake of simplification. The results of estimating restricted and unrestricted versions of Eq. (10) are presented in the upper half of Table 4. The instrument validation tests do not give evidence against the instruments. The estimated coefficients of the level of unemployment are consistently significant and negative, particularly in the economywide estimates, which suggests a strong Phillips curve effect. The estimated coefficients of the level of unemployment are generally higher when unemployment is measured by Ulow than by Ubc . The ratio of the coefficient estimates of Ubc and Ulow , however, approximately equals the ratio between the average values of Ulow and Ubc . Hence, the effects on wage growth of the level and change in unemployment are approximately the same regardless of whether Ulow or Ubc is used as the regressor. Turning to the estimated coefficients of the change in unemployment, the estimates suggest that the persistence of shocks on wages, and hence on unemployment, differs across sectors. The sign of the estimated coefficients of the

266

Table 4. Parameter Estimates of Philips Curves and Price Equations. Phillips Curve

N = 22, WPI, Ubc N = 12, PMAN, Ubc N = 12, PMAN, Ulow N = 12, PMAN, Ubc , LIML N = 12, PMAN, Ulow , LIML N = 12, WPI, Ubc N = 12, WPI, Ulow N = 8, Ubc , GIVE N = 8, Ulow , GIVE N = 8, Ubc , LIML N = 8, Ulow , LIML

N = 12, WPI N = 12, PMAN

wt −1

Ut

Ut

−0.10(1.46) 0.20(2.13) 0.19(2.04) 0.15(1.50) 0.13(1.42) 0.25(2.63) 0.23(2.57) 0.01(4.59) 0.02(7.01) 0.01(1.35) 0.01(0.91)

−0.20(3.27) −0.14(2.27) −0.27(2.79) −0.15(1.91) −0.28(2.23) −0.18(2.68) −0.33(3.26) −0.18(5.75) −0.43(12.6) −0.20(4.76) −0.40(4.25)

−0.39(2.34) −0.36(3.11) −0.61(3.69) −0.21(1.75) −0.31(1.55) −0.26(1.92) −0.47(2.40) 0.21(1.82) 0.20(1.35) 0.31(3.26) 0.43(2.36)

p t−1

wt

p ct

−0.03(0.62) 0.19(2.01)

0.14(1.24) 0.30(2.96) p t−1 0.07(0.71)

p t−1 N = 12, WPI, Ubc

−0.05

wt 0.07(0.63) p t−1

wt 0.54(4.55) p ct 0.52(8.35) wt

0.18(2.31) 0.36(3.90) 0.40(3.52) 0.39(5.15) 0.40(5.25) 0.13(2.26) 0.15(2.84) 0.84(7.81) 0.77(10.5) 0.93(10.4) 0.87(8.09) Price Equation p im t 0.31(7.26) y t 0.39(4.99) p im t 0.31(7.07) Ut

prt

0.87(18.1) 0.90(18.5) 0.88(8.80) 0.86(9.14)

␹2 (6) 7.1 14.14 14.03 5.34 6.02 14.14 16.07 5.22 4.02 10.24 11.12

y t

y ct

␹2 (4)

−0.03(0.60) 0.05(1.50)

−0.05(2.26) −0.02(1.20)

3.48 2.76

y ct −0.02(1.07) Ut 0.05(0.23) U t

prt −0.47(2.54) U t −0.07(0.56) prt

␹2 (5) 11.48 ␹2 (4) 7.07 ␹2 (4)

JAKOB B. MADSEN

N = 8, GIVE

0.48(6.09)

pt

0.17(1.79) 0.17(1.84) 0.12(1.93) 0.14(1.91) 0.10(1.43) 0.15(1.85)

0.24(2.74) 0.24(2.84) 0.53(6.13) 0.58(5.62) 0.65(8.45) 0.63(7.59)

−0.01(0.27) −0.01(0.18) 0.00(0.21) 0.03(0.22) 0.15(3.40) 0.28(2.89)

−0.45(2.49) −0.39(2.61) −0.49(6.88) −0.83(5.37) −0.44(6.38) −0.78(5.57)

−0.33(2.57) −0.33(1.87) −0.56(4.87) −0.53(4.41)

4.29 4.19 11.22 10.59 9.22 8.55

Notes: See notes to Table 3. GIVE = generalised instrumental variables estimator. ␹2 (i) = Wald test of overidentifying restrictions, and is distributed as ␹2 (i) with i degrees of freedom. The unemployment rate is measured in decimal points. Constants and fixed effect dummies are included in all estimates (fixed effect dummies are included in first-difference estimates of the Phillips curve but not the price equation) but are not shown. Instrument set I is used in the N = 22 estimates, and the instrument set II is used for the other estimates (see text). Ubc = unemployment data used by B&C. Ulow = unemployment rates based on estimations for individual countries. Estimation period 1927 to 1937.

The Length and the Depth of the Great Depression: An International Comparison

N = 12, PMAN, Ubc N = 12, PMAN, Ulow N = 8, Ubc , GIVE N = 8, Ulow , GIVE N = 8, Ubc , LIML N = 8, Ulow , LIML

267

268

JAKOB B. MADSEN

change in unemployment is consistently positive in the economy-wide estimates and therefore stands in sharp contrast to the predictions of wage equation based models, where only changes in unemployment are assumed to influence wage growth, and negatively. Coupled with the fact that the estimated coefficient of the level of unemployment is significantly negative, this result suggests that unemployment adjusts rapidly to shocks in the whole economy. A larger degree of persistence to shocks prevails in manufacturing where the estimated coefficients of the change in unemployment are consistently negative, although with varying degrees of statistical significance. The discrepancy in wage behavior between sectors is also evident from the parameter estimates of wage adjustment to innovations in prices. Wages adjust about 40% to price innovations within the first year in manufacturing using the value-added price deflator as the price variable (rows 2–5). Allowing for the effects of lagged wages, simulations of the models indicate that manufacturing wages decreased, on average, by approximately 6% over the period from 1929 to 1932 due to the price deflation. Taking account of the approximately 2.2–2.7% wage reducing effect of the level and the change in unemployment, depending on which measure of unemployment is used, over the same period, it can be concluded that wage stickiness in manufacturing could not have been solely responsible for the supply failure during the Depression.18 The economy-wide estimates show a rapid adjustment of wages to innovations in prices and productivity. About 85% of the adjustment takes place within the same year. Simulations of the models indicate that the 18% decline in valueadded prices and the 4% increase in productivity, on average, resulted in a 12% reduction in wages over the period from 1929 to 1932, using price and productivity elasticities of 0.85. Allowing for the approximately 3.5–4.5% wage reducing effect of the unemployment over the same period, depending of which unemployment measure is used, it can be concluded that wages were reasonably flexible on an economy-wide scale during the Depression. When wholesale prices are used as deflators, the significance of the estimated coefficients of prices is substantially reduced as compared to the estimates where the value-added price deflator is used (rows 1, 6 and 7).19 The estimated coefficients of wholesale prices are less than 0.2 in all the estimates and therefore point towards a large degree of nominal wage rigidity. The low estimated coefficient of wholesale prices is not surprising, however, given that commodity prices fell substantially during the first years of the Depression. Under perfect market clearing in the goods and labor markets, a reduction in commodity prices yields a positive output response and an increase in the nominal wage that is compatible with full employment. A reduction in the value-added price deflator, by contrast, necessitates a reduction in nominal wages to maintain full employment. Since

The Length and the Depth of the Great Depression: An International Comparison

269

wholesale prices are composed of both commodity prices and, implicitly, the value-added price deflator, the coefficient of prices in the Phillips curve will be biased downwards, particularly in periods of highly volatile commodity prices, and therefore do not give much information about nominal wage rigidity. How do the results compare with other studies? Surprisingly few estimates of the Phillips curve have been undertaken for the interwar period. Almost all studies of wage adjustment only allow for a wage curve effect and therefore a priori exclude the possibility that wages adjust to disequilibria in the labor market due to Phillips curve effects. Newell and Symons (1988) find the level of real wages to be fairly insensitive to the level of unemployment for 14 countries. Furthermore, they find a short-run price elasticity of wages of approximately 0.5. B&C estimate the short-run wholesale price elasticity of wages to be approximately 0.3 depending on model specification, which is close to the estimates in Table 4 using wholesale prices. Furthermore, they find the level of wages to be insensitive to the level of unemployment.20 Dimsdale, Nickell and Horsewood (1989) find the level of wages to be significantly negatively related to the level of unemployment for the U.K.

7. PRICE STICKINESS DURING THE DEPRESSION To investigate the degree of price stickiness, the following stochastic specifications of the price equation, which was derived in Section 4, are estimated: p it = ␸0 + ␸1 p i,t−1 + ␸2 wit + ␸3 p cit + ␸4 p im it + ␸5 prit + ␸6 y it + ␸7 y it + CD ␬ + ␧4,it cyc

(11)

and p it = ␭0 + ␭1 p i,t−1 + ␭2 wit + ␭3 p cit + ␭4 p im it + ␭5 prit + ␭6 U it + ␭7 U it + CD ␯ + ␧5,it ,

(12)

where ycyc is cyclical demand, pc is commodity prices, and pim is import prices. Cyclical demand is estimated as the residual from regressing income on a time trend and a squared time trend over the period from 1920 to 1939 for each individual country. The residual is then instrumented using the instruments as specified below. Equations (11) and (12) are estimated for manufacturing using the value-added price deflator and wholesale prices and for the whole economy using the valueadded price deflator. Some of the coefficients are restricted to zero as indicated in Table 4. All stochastic regressors, except the lagged dependent variables, are instrumented using instrument set II. A two-period lag of the dependent variable

270

JAKOB B. MADSEN

is used as an instrument for the lagged dependent variable. Log level estimates are not undertaken because the residuals exhibited serious serial correlation, which did not disappear using the non-linear estimator to correct for first-order serial correlation, and because they exclude the possibility of testing for output level effects. The non-linear generalised least squares estimator is used again. Commodity and import prices are only included as regressors in the estimates where wholesale prices are used as the dependent variable. Both output and unemployment are used as demand shifters. Although only output refers to the goods market, unemployment is also used to test the sensitivity of prices to demand shocks because desired effects may not be captured by income. The problems associated with the use of cyclical income is that it is sensitive to estimates of the income trend, whereas there is no need to remove a trend from the unemployment rate. The demand shift variables are measured in both levels and first differences to allow for the possibility that prices are sensitive to both the changes and the cyclical level of demand. A cyclical decline in demand may be perceived by firms to be temporary and they will consequently not alter prices. If demand remains depressed, it is possible that firms will start lowering their prices until the goods market has been cleared using the philosophy of the Phillips curve. The results of estimating restricted and unrestricted versions of Eqs (11) and (12) are shown in the lower part of Table 4. The instrument validity tests do not give evidence against the chosen instruments. The estimates in rows 13, 16 and 17 show that the manufacturing value-added price deflator is fairly insensitive to wage changes, with estimated short-run wage elasticities of approximately 0.25, but it is somewhat responsive to changes in unemployment. The estimated coefficients of cyclical demand, the change in output and the level of unemployment, however, were insignificant at any conventional significance level. Simulations of the models with unemployment as demand shift variables (rows 16 and 17) show that the 8–14 percentage point increase in the rate of unemployment from 1929 to 1932, depending on which measure of unemployment is used, resulted in a 3–6% reduction in manufacturing prices. The parameter estimates of the economy-wide price equations displayed in rows 14 and 18–21 show that that prices are quite sensitive to innovations in both demand and wages. Both the estimated coefficients of changes in output and changes in unemployment are statistically highly significant, which, unlike manufacturing, suggests that the estimates are robust to choice of the demand shift factor. Simulations of the models show that the 12.6% decline in real GNP, on average, contributed to a 5% decline in the value-added price deflator over the period from 1929 to 1932. An approximately 7% reduction in prices is predicted if the change

The Length and the Depth of the Great Depression: An International Comparison

271

in unemployment is used as the demand shifter, regardless of which measure of unemployment is used. This suggests that there is a robust positive relationship between changes in prices on an economy-wide scale and changes in demand. The estimated coefficient of wages is 0.54 when income is used as the demand shift variable (row 14) and about 0.6 when unemployment is used, regardless of which measure of unemployment is used (rows 18–21). These results suggest that the wage elasticities are somewhat insensitive to the choice of demand shift variable. Non-nested tests and Akaike’s information criterion are used to discriminate between the models. Both criteria favor the models using unemployment as demand shifters.21 Hence, the model with unemployment will be used as the benchmark model in the simulations in the next section. The estimated coefficients of productivity are statistically and economically significant in the economy-wide estimates, especially in the LIML estimates, and are in the range of 0.33 and 0.56. Hence, prices are less responsive to productivity advances than wages. This result is consistent with the coefficient estimates of the other variables in the models, which showed that wages are more responsive to price innovations than prices are to wage innovations and that unemployment, in excess of its natural rate, puts persistent downward pressure on wages, whereas changes in unemployment only have one-off effects on prices. The estimates for manufacturing where wholesale prices are used as the dependent variable (rows 12 and 15) show, perhaps, some surprising results. First, wholesale prices are not sensitive to y, U and U, and move counter to ycyc . In other words, movements in wholesale prices were more counter-cyclical than pro-cyclical. This result is remarkable because it shows that the decline in wholesale prices during the first years of the Depression was not a cyclical response, as is often stressed in macroeconomic textbooks. Second, wholesale prices are completely insensitive to wages. Third, wholesale prices are almost entirely driven by import and commodity prices, which reinforces the discussion in Section 2 that import and commodity prices had a heavy weight in wholesale prices and therefore that wholesale prices were misleading indicators of the value-added price deflator. Overall, these results show that the decline in wholesale prices at the onset of the Depression was a result of declining commodity prices, in particular, and declining import prices, and not due to the fall in aggregate demand. Comparing the results with the results in the literature it is remarkable how little attention has been given to estimating the degree of price flexibility. Dimsdale, Nickell and Horsewood (1989) find that prices are completely unresponsive to aggregate demand shocks, but do not test for nominal price stickiness because they restrict the coefficient of wages to one and only allow the deviation of prices from their long-run equilibrium to last for less than a year.

272

JAKOB B. MADSEN

8. MODEL SIMULATIONS The economy-wide estimates of the price equation and the Phillips curve, rows 8, 9, 18 and 19, are used in this section to simulate the effects of wage and price stickiness on unemployment. The restricted non-stochastic counterparts of Eqs (10) and (12) can be simplified as follows: wit = ␥0 + ␥1 p it − ␥2 U it + ␥3 prit + CD ␫

(13)

p it = ␺0 + ␺1 wit − ␺2 U it − ␺3 prit ,

(14)

where coefficients with low economic and statistical significance in the estimates in the previous section have been restricted to zero to simplify the exposition. All coefficients are expected to be positive. The price equation that contains unemployment as the demand shifter is used because the model selection criteria that were used in the previous section, favored this specification. Effects from the lagged dependent variables are suppressed because they were either negligible or insignificant. The equation system yields the following solution: U it =

␺2 (1 − ␥1 )p it (1 − ␺1 )wit (␥ − ␺3 )prit U i,t−1 − − + 3 ␥2 + ␺2 ␥2 + ␺2 ␥2 + ␺2 ␥2 + ␺2 +

␥0 + ␺0 + CD ␫ , ␥2 + ␺2

(15)

which corresponds to a restricted version of Eq. (7) with the exception that unemployment is used as the scale variable for output. Equation (15) shows how unemployment evolves over time following shocks to demand and productivity. Apart from productivity shocks, supply shocks are absent from the equation since they were insignificant in the empirical estimates in Section 6. Very similar results are obtained by Dimsdale, Nickell and Horsewood (1989), who find that the unemployment path was predominantly determined by real import prices and demand. The natural rate is ruled out by the estimates since unemployment is non-neutral to innovations in prices, wages, and productivity in the long run. Homogeneity in wages and prices was not imposed since the hypothesis that the long-run coefficient of wages and prices were rejected at conventional significance levels.22 Furthermore, the hypothesis of long-run productivity neutrality was strongly rejected. This need not mean that the natural rate did not exist, but that the labor and goods markets were sufficiently slow to adjust toward their long-run equilibrium to be identified by the estimates. Identification of the adjustment path was furthermore rendered difficult because goods and labor markets were constantly out of equilibrium during the Depression.

The Length and the Depth of the Great Depression: An International Comparison

273

The second and third right-hand terms in Eq. (15) represent the unemployment effect of demand shocks. A demand shock that drives prices and wages down leads to increasing unemployment to the extent that wages and prices fail to adjust to price and wage changes, respectively, and this generates a wedge between actual and full employment wages and prices. This in turn leads to a change in unemployment. The fourth right-hand side term is the effects on unemployment of productivity shocks. Since the estimates in the previous sections show that wages are more responsive to productivity shocks than prices, it follows that a positive productivity shock leads to higher unemployment because producers fail to lower their prices sufficiently to a positive productivity shock, but instead increase their mark-ups. The last right-hand-side term, divided by one minus the coefficient of lagged unemployment, defines the natural rate of unemployment, that is the level of unemployment at which price and wage expectations are borne out. Equation (15) is used to decompose the unemployment effects of the demand shocks into wage and price rigidity. One caveat of this decomposition is that unemployment effects of wage and price stickiness cannot strictly be decomposed to their source because of the interrelationship between the labor demand and the wage setting schedules. The unemployment effects of an adverse demand shock that shifts the wage setting schedule to the left due to nominal price rigidity, for instance, depends not only on the slope and the shift in the Phillips curve but also the slope of the labor demand curve. Hence, the more responsive is the goods market to demand shocks, the lower are the unemployment effects of a demand shock due to rigidities in the labor market. Similarly, the more reactive is the labor market to demand shocks the lower are the unemployment effects of the demand shocks, due to rigidities in the goods market. In the present case where price growth is affected by the change in demand, as opposed to the level of demand, the unemployment effects of nominal wage rigidity are muted on impact (␺2 in Eq. (15)) but zero in the long run as seen from the long-run solution to Eq. (15): U it = −

(1 − ␥1 )p it (1 − ␺1 )wit (␥ − ␺3 )prit ␥ + ␺0 + CD ␫ − + 3 + 0 . ␥2 ␥2 ␥2 ␥2

Hence, in the long run the unemployment effects of wage stickiness are not muted by price flexibility because the labor demand curve is vertical. Shocks are only muted by the wage growth response to unemployment in the Phillips curve. If the Phillips-curve effect did not exist, shocks would have permanent unemployment effects. Since the decomposition in Eq. (15) cannot account for this aspect, it underestimates the increase in unemployment that is due to price stickiness and overestimates the unemployment effects of wage stickiness.23

274

JAKOB B. MADSEN

Table 5. Decomposition of Unemployment Based on Simulations of Eq. (15). Wage Stickiness 1 Rows 8 and 18 (Ubc ) in Table 4 1930 1.38 1931 1.76 1932 2.86 Total Rows 9 and 19 1930 1931 1932 Total

6.00 (Ulow )

in Table 4 1.01 1.31 2.05 4.37

Simulated

U

2

3

4

5

−0.54 2.48 5.05

2.46 0.26 2.06

3.30 4.48 9.96

5.18 5.82 6.58

6.99

4.78

17.74

17.58

−0.19 1.17 2.30

1.28 0.14 1.13

2.11 2.62 5.47

2.22 2.56 3.70

3.28

2.55

10.20

8.48

Price Stickiness

Notes: The numbers are based on unweighted averages of the eight countries that are included in the economy-wide estimates. Column 1 is the price-induced change in unemployment due to wage rigidity; column 2 is the wage-induced change in unemployment due to price rigidity; column 3 is the productivity-induced change in unemployment because productivity changes mark-ups; and column 4 is the sum of columns 1–3, and column 5 is the actual change in unemployment. Unemployment is measured in percentages.

The simulations in Table 5 show the decomposition of the change in unemployment over the period from 1930 to 1932 using Eq. (15). The simulations in the upper half of the table are based on the estimates in rows 8 and 18 (Ubc ) and the simulations in the lower part of the table are based on the estimates in rows 9 and 19 (Ulow ) in Table 4. It is important to note that the changes in p, w and pr in equations do not refer to the source of stickiness in Eq. (15). Changes in wages influence unemployment because of price stickiness. Similarly, changes in productivity influence unemployment because prices fail to clear the market. Hence, the third and the fourth terms in Eq. (15) constitute the joint effects on unemployment of price stickiness. Finally, the price-induced change in unemployment, which is represented by the second term in Eq. (15), represents the unemployment effects of wage stickiness. The simulations explain the total increase in unemployment from 1929 to 1932 quite precisely. Based on Ubc , wage stickiness results in a 6-percentage point increase in the rate of unemployment, whereas price stickiness results in an 11.77-percentage point increase in the rate of unemployment over the period from 1929 to 1932. Based on Ulow , wage stickiness results in a 4.37-percentage point increase in the rate of unemployment, whereas price stickiness results in a 5.83-percentage point increase in the rate of unemployment over the period from

The Length and the Depth of the Great Depression: An International Comparison

275

1929 to 1932. These decompositions suggest that price stickiness was the main source of the supply failure, but that wage rigidity also played an important role. These results are largely consistent with the microeconomic evidence. Although there is an extensive micro-oriented literature on price rigidity in U.S. manufacturing during the Depression, the literature is old. The classic studies of Burns (1936), Means (1935), and Berle and Means (1991, first printed in 1933) showed that U.S. manufacturing was highly concentrated and had sticky prices during the Depression. In 1929 the 100 largest U.S. manufacturing corporations controlled 44% of net capital assets in the U.S. (Berle & Means, 1991, p. 353). There was, therefore, plenty of scope for price rigidity. Burns (1936) listed several products for which prices remained unchanged, or changed only a little from the beginning of 1929 to the beginning of 1933. In the important U.S. automobile industry, prices only decreased 10%, whereas production fell almost 70%, from 1929 to 1931 (Federal Trade Commission, 1932). Another example is the U.S. agricultural implements industry where employment fell 82% from May 1929 to July 1932, but output prices decreased less than 10% (Means, 1935). There is not much micro evidence on price rigidity during the Depression outside the U.S. There is, however, evidence that suggests a widespread and increasing degree of industrial concentration during the interwar period in most parts of the developed world. Supported by extreme nationalism among the public, governments were often the driving force behind the formation of cartels to strengthen the position of national industries in international markets and prevent rivals penetrating domestic markets (Wurm, 1989, p. 111; United Nations, 1947, pp. 8–9). The growing cartelization was often associated with the increasing tariffs and non-tariff trade barriers in the first part of the Depression, where the tariffs were imposed to strengthen the bargaining position of prospective cartel members in the cartel negotiations (United Nations, 1947, p. 21). Some governments even introduced legislation that made cartelization compulsory (United Nations, 1947, p. 10). Haussmann and Ahearn (1944) concluded from their statistical study that 42% of world trade over the period from 1929 to 1937 was cartelized or influenced by loosely knit associations or conferences. Recent evidence on wage rigidity in the U.S. during the Depression suggests that wages responded to market forces. Hanes (2000), Lebergott (1989) and Mitchell (1985a, b) find that nominal wage cuts were widespread among U.S. firms and Mitchell (1985a, b) argues that wages first became rigid after the Depression. Comparing the degree of wage stickiness during the Depression with earlier contractions in the U.S., Dighe (1997) concludes that wages were not particularly rigid during the Depression and that wage rigidity was not an important causal factor in the Depression. Simon (2001) finds that asking wages

276

JAKOB B. MADSEN

among female clerical workers seeking employment fell by 58% from 1929 to 1933, and that agricultural workers experienced almost the same decrease. Although there is no evidence of wage flexibility among workers who were not seeking employment, it nevertheless shows that there was a potential threat to clerical and agricultural workers who refused wage cuts.

9. SUMMARY AND CONCLUSION This paper has examined the extent to which the length and the depth of the Depression was a result of sticky prices and wages using a small supply side model which was estimated using annual data for industrialized and semi-industrialized countries for the interwar period. The model consisted of the labor market, which was represented by the Phillips curve, and the goods market, which was based on a price setting model. Based on panel data estimates, the model was simulated over the period from 1929 to 1932 and the unemployment effects of the demand shocks were decomposed into wage stickiness, price stickiness and productivity non-neutrality. The estimates of economy-wide wage setting equations showed that wages adjust relatively quickly to innovations in prices and that wage growth was very sensitive to the level, but not to changes in unemployment and therefore that shocks had neither persistent, nor permanent effects on unemployment. A shock that brings wages up to a higher level results in unemployment in excess of the equilibrium unemployment rate, which in turn, puts effective downward pressure on wages until unemployment is eliminated. More sluggish wage adjustment was identified for manufacturing. Only about half of the adjustment in wages to innovations in prices was accomplished within the same year, and both the level and changes in unemployment were significant determinants of wage growth. The latter implies that shocks do have persistent, although not permanent, output effects. Wages were initially slow to respond to the adverse demand shocks, but as unemployment continued to grow up to 1932/1933, it exerted increasing downward pressure on wages, and helped to clear the labor market. The estimates of the price equations showed that prices were slow to adjust to innovations in wages and that demand shocks only had one-off effects on prices, which suggests that there were no forces that automatically corrected disequilibrium in the goods markets and therefore that the labor market had to bear the adjustment costs. It appears that most of the economy-wide price stickiness stems from pronounced price stickiness in the manufacturing sector. The estimates indicated that manufacturing prices were slow to adjust to innovations in wages and their cyclicality was sensitive to measurements of demand.

The Length and the Depth of the Great Depression: An International Comparison

277

A remarkable finding of the paper was that wholesale prices were completely insensitive to innovations in wages and were either insensitive to demand shocks or moved counter-cyclically. These findings suggest that if the same data as used in other international studies were used to test the degree of price rigidity, then it would have been concluded that the Depression was propagated by sticky prices and that wage flexibility could not have alleviated the Depression. This result underscores the importance of using the correct price deflators in studies of the supply side. Robustness checks using various measurements of wages, prices and unemployment, different estimators, two different instrument sets, two different sectoral decompositions and estimators that were robust to errors-in-variables, indicated that the results were almost unaffected to these variations. An exception was the estimates using wholesale prices. The estimates confirmed the finding in the literature that wages are relatively insensitive to wholesale prices. It was shown, however, that this result is consistent with the fact that variations in wholesale prices were dominated by variations in commodity prices. The results that price stickiness played a primary role during the Depression and that wage rigidity only played a secondary role stand in contrast to the results of the international studies of B&C (1996) and Eichengreen and Sachs (1985), where price stickiness did not play any role in the propagation of the Depression. The finding that sticky prices played a key role in the propagation of the Depression needs to be backed up by more microeconomic evidence. Previous microeconomic studies point towards widespread price stickiness within U.S. manufacturing. These studies, however, are mostly 60–70 years old, so new studies that use the tools of modern corporate finance and microeconomic theory would be highly desirable. Furthermore, very few studies of price stickiness outside the U.S. have been undertaken. New studies of price stickiness outside the U.S. would be especially valuable.

NOTES 1. They also estimate a labor demand function conditional on wages deflated by producer prices, but do not use these estimates in their simulations. 2. Finland and Spain are not included in the B&C sample. 3. Eichengreen and Sachs (1985) obtain a more significant negative relationship between the change in output and the change in real wages than obtained here. The difference reflects different sources of wage data. 4. Regressing the log of wholesale prices, pwpi , on manufacturing hourly wages, wm , the log of the exchange rate relative to its 1929 gold parity, e, and the log of notes in circulation, h0, yields the following estimates using annual data over the period from 1927

278

JAKOB B. MADSEN

to 1937 for the 12 countries listed in Fig. 2: wpi

pˆ t

wpi

= −0.02 + 0.17p t−1 + 0.65wm t + 0.14e t + 0.30h0t , (2.25)

(1.60)

(4.95)

(4.13)

(4.07)

R 2 (B) = 0.39,

where R2 (B) is Buse’s raw-moment R2 and t-statistics are in parentheses. See Section 6 for the instruments used for wages and estimation method. The estimates show that wholesale prices are sensitive to exchange rates. 5. The supply function is derived from the following optimization problem of the perfect competitive firm where output is measured as gross output: max ␲ = P o Y o (K, L, Q c ) − WL − P c Q c . 6. The results below are not affected by alternative weightings (0.7 and 0.3, or 0.9 and 0.1). Similar results were also obtained if real commodity prices were used as separate arguments in the output supply function. 7. For agriculture, for instance, supply is over a wide range of output independent of real product wages since land is a fairly fixed factor of production. Land prices, rather than supply, are likely to be affected by real wages. 8. The cyclicality of prices need not depend on the cyclicality of mark-ups. Prices can also vary on a cyclical basis due to increasing or decreasing returns to scale and labor hoarding. 9. More complete pricing models that allow for number of firms and for strategic interaction among firms are derived in Hay and Morris (1991). 10. Actual labor productivity did not change much during the Depression. Whether this modest change was reflected in expectations cannot be tested. 11. Economy-wide non-residential capital stock was initially used as an additional regressor in the estimates of Eq. (9), however, its estimated coefficient was insignificant and the parameter estimates of the other variables were insensitive to its inclusion. It was, therefore, not included in the estimates because it would have increased the Data Appendix substantially. 12. See Madsen (2001b) for construction of the competitiveness index. 13. The only difference is that the import prices used by B&C in the instruments are derived from the wholesale prices of exporting countries, whereas this paper uses import unit values as instruments. 14. The data used by B&C commence in 1929. The same data are available before that time from their sources. Wages commence in 1927 from the ILO Yearbook and the same series are available in earlier years from the League of Nation, Monthly Bulletin of Statistics and League of Nation, Statistical Yearbook. 15. This exercise is not undertaken for the estimates where the wholesale price deflator is used because the measurement error in wholesale prices is likely to be highly serially correlated as argued in the previous section, and therefore renders the Griliches-Hausman transformation invalid. 16. Equation (10) was also estimated with asymmetric adjustment of wages to positive and negative price changes to allow for downward nominal rigidity but not upward rigidity as suggested by Keynes (1936, pp. 13–15). However, the null hypothesis of symmetrical wage adjustment could not be rejected at any conventional significance levels. 17. More formally the generalised instrumental variable method is used, where the covariance matrix is weighted by the correlation of the disturbance terms using the

The Length and the Depth of the Great Depression: An International Comparison

279

variance-covariance structure as follows: E{␧2it } = ␴2i , i = 1, 2, . . ., N, E{␧it , ␧jt } = ␴ij , i = j, where ␴2i = the variance of the disturbance terms for country i = 1, 2, . . ., N; ␴ij = the covariance of the disturbance terms across countries i and j; and ␧ is the disturbance term. The variance ␴2i is assumed to be constant over time but to vary across countries and the error terms are assumed to be mutually correlated across countries, ␴ij , as random shocks are likely to impact on all countries at the same time. ␴2i and ␴ij , are estimated using the feasible generalised least squares method described in Greene (2000, Chap. 15). 18. The estimates of the wage reducing effects of the level of unemployment are based on the deviation of unemployment from its “natural rate,” where the “natural rate” is the absolute value of the ratio of the constant term and the coefficient of unemployment. 19. Note that the estimates cannot be compared with the B&C estimates because their dependent variable is the level of wages, not the change in wages. 20. Unemployment is measured in first differences in their first-difference estimates. 21. Akaike’s information criterion gave the results as follows; −0.743 using Ulow , −0.716 using Ubc and −0.744 using output; thus favoring the models with unemployment as demand shift variables. Non-nested tests (J-tests) yielded the results as follows: Including the predicted value from the model with output as a regressor in the models with unemployment as regressor gave the t-values of 0.46 (the model with Ulow ) and 1.09 (the model with Ubc ), which suggests that the model with output as regressor does not encompass the models with unemployment as a regressors. Including the predicted value from the regressions with Ulow and Ubc as demand shift factors in the model with output as a regressor gave the t-values of 2.28 and 2.02, respectively. Hence, the models with unemployment as the regressors encompass the model with output as the regressor. 22. The tests of the null hypothesis that the coefficients of wages and prices divided by one minus the coefficient of the lagged dependent variable gave the following chi-squared test statistics: ␹2 (1) = 2.83 (row 8), ␹2 (1) = 7.26 (row 9), ␹2 (1) = 24.55 (row 18), and ␹2 (1) = 9.63 (row 19). 23. This is probably the reason why Dimsdale, Nickell and Horsewood (1989) do not decompose the unemployment effects of sticky wages and prices for the U.K.

ACKNOWLEDGMENTS Helpful comments and suggestions from participants at the 2001 Cliometrics conference and at seminars at University of New South Wales, University of Western Australia, and University of Waikato, and especially a referee, are gratefully acknowledged. Randall Fox, Paula Madsen, Signe Skarquist, Nick Tsitsianis and Wana Yang provided excellent research assistance. The estimates were programmed in Gauss and Shazam.

280

JAKOB B. MADSEN

REFERENCES Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte carlo evidence and an application to employment equations. Review of Economic Studies, 58, 277–297. Beenstock, M., & Warburton, P. (1986). Wages and unemployment in the interwar Britain. Explorations in Economic History, 23, 153–172. Berle, A. A., & Means, G. C. (1991). The modern corporation and private property. New Brunswick, NJ, 1932. First published in 1933. Bernanke, B. S. (2000). Essays on the great depression. Princeton: Princeton University Press. Bernanke, B. S., & Carey, K. (1996). Nominal wage stickiness and aggregate supply in the great depression. Quarterly Journal of Economics, 111(3), 853–883. Bernanke, B., & James, H. (1991). The gold standard, deflation, and financial crises in the great depression: An international comparison. In: R. G. Hubbard (Ed.), Financial Markets and Financial Crises (pp. 33–68). Chicago, IL: University of Chicago Press for NBER. Blanchard, O. J., & Katz, L. (1997). What we know and do not know about the natural rate of unemployment. Journal of Economic Perspectives, 11, 51–72. Blanchflower, D. G., & Oswald, A. J. (1994). The wage curve. Cambridge, MA: MIT Press. Bordo, M. D., Erceg, C. J., & Evans, C. L. (2001). Money, sticky wages, and the great depression. American Economic Review, 90, 1447–1463. Bound, J., Jaeger, D. A., & Baker, R. M. (1995). Problems with instrumental variables estimation when the correlation between instruments and the endogenous explanatory variable is weak. Journal of the American Statistical Association, 90, 443–450. Burns, A. R. (1936). The decline of competition. New York: McGraw-Hill. Cole, H. L., & Ohanian, L. E. (1999). The great depression in the U.S. from a neo-classical perspective. Federal Reserve Bank of Minneapolis Quarterly Review, 23, 2–24. Cole, H. L., & Ohanian, L. E. (2001). Re-examining the contributions of money and banking shocks to the U.S. great depression. NBER Macroeconomics Annual, 15, 183–227. Dighe, R. S. (1997). Wage rigidity in the great depression: Truth? consequences? Research in Economic History, 17, 85–134. Dimsdale, N. H., Nickell, S. J., & Horsewood, N. (1989). Real wages and unemployment in Britain during the 1930s. The Economic Journal, 99, 271–292. Donald, S. G., & Newey, W. K. (2001). Choosing the number of instruments. Econometrica, 69, 1161–1191. Dunlop, J. T. (1938). The movement of real and money wage rates. The Economic Journal, 48, 413–434. Eichengreen, B., & Hatton, T. J. (1988). Interwar unemployment in international perspective. In: B. Eichengreen & T. J. Hatton (Eds), Interwar Unemployment in International Perspective (pp. 1–59). Dordrecht, Germany: Martinus-Nijhoff. Eichengreen, B., & Sachs, J. (1985). Exchange rates and economic recovery in the 1930s. Journal of Economic History, 45(4), 925–946. Federal Trade Commission (1932). Senate committee on the judiciary hearings on amendment of the federal trade commission act. Washington, DC. Galenson, W., & Zellner, A. (1957). International comparisons of unemployment rates. In: The Measurement and Behavior of Unemployment (pp. 439–580). National Bureau of Economic Research. Princeton: Princeton University Press. Greene, W. H. (2000). Econometric analysis (4th ed). Upper Saddle River, NJ: Prentice-Hall. Griliches, Z., & Hausman, J. A. (1986). Errors in variables in panel data. Journal of Econometrics, 31, 93–118.

The Length and the Depth of the Great Depression: An International Comparison

281

Grytten, O. H. (1995). The scale of Norwegian interwar unemployment in international perspective. Scandinavian Economic History Review, 45, 226–250. Hahn, J., & Hausman, J. (2002). A new specification test for the validity of instrumental variables. Econometrica, 70, 163–189. Hall, R. E. (1988). The relation between price and marginal cost in U.S. industry. Journal of Political Economy, 96, 921–947. Hanes, C. (2000). Nominal wage rigidity and industry characteristics in the downturns of 1893, 1929, and 1981. American Economic Review, 90, 1432–1446. Haussmann, F., & Ahearn, D. (1944). International cartels and world trade, an explanatory estimate. Thought, Fordham University Quarterly. Hay, D. A., & Morris, D. J. (1991). Industrial organisation. Oxford: OUP. Hsiao, C. (1986). Analysis of panel data. Cambridge: Cambridge University Press. Keynes, J. M. (1936). The general theory of employment, interest and money. Reprinted in 1973. London: MacMillan. Keynes, J. M. (1939). Relative movements of real wages and output. The Economic Journal, 49, 30–51. Kiviet, J. F. (1995). On bias, inconsistency, and efficiency of various estimators in dynamic panel data models. Journal of Econometrics, 68, 53–78. Layard, R., & Nickell, S. (1986). Unemployment in Britain. Economica, 53, S121–S169. Lebergott, S. (1989). Wage rigidity in the depression: Concept or phrase? Mimeo, Department of Economics, Wesleyan University. Liesner, T. (1989). One hundred years of economic statistics. Oxford: Economist. Lindbeck, A., & Snower, D. J. (1994). How are product demand changes transmitted to the labour markets? Economic Journal, 104, 386–398. Madsen, J. B. (1998). General equilibrium macroeconomic models of unemployment: Can they explain the unemployment path in the OECD? The Economic Journal, 108, 850–867. Madsen, J. B. (2001a). Agricultural crises and the international transmission of the great depression. Journal of Economic History, 61, 327–365. Madsen, J. B. (2001b). Trade barriers and the collapse in international trade during the great depression. Southern Economic Journal, 67, 848–868. Margo, R. A. (1993). Employment and unemployment in the 1930s. Journal of Economic Perspectives, 7, 41–59. Margo, R. A. (2000). Review of Ben S. Bernanke essays on the great depression. Economic History Services, http://www.eh.net/bookreviews/library/0261.shtlm. Means, G. C. (1935). Industrial prices and their relative stability. U.S. Senate Document, 13, 74. Mitchell, D. J. B. (1985a). Wage flexibility in the U.S.: Lessons from the past. American Economic Review, Papers and Proceedings, 75, 36–40. Mitchell, D. J. B. (1985b). Wage flexibility then and now. Industrial Relations, 24, 266–279. Newell, A., & Symons, J. S. V. (1988). The macroeconomics of the interwar years: International comparisons. In: B. Eichengreen & T. J. Hatton (Eds), Interwar Unemployment in International Perspective (pp. 61–96). Dordrecht, Germany: Martinus-Nijhoff. Pigou, A. C. (1933). The theory of unemployment. London. Roeger, W. (1995). Can imperfect competition explain the difference between primal and dual productivity measures? Estimates for U.S. manufacturing. Journal of Political Economy, 103, 316–330. Silver, S., & Summer, S. (1995). Nominal and real cyclicality during the interwar period. Southern Economic Journal, 61, 588–601. Simon, C. J. (2001). The supply price of labor during the great depression. This Journal, 61, 877–903.

282

JAKOB B. MADSEN

United Nations (1947). International cartels. New York: Lake Success. United Nations (1954). Monthly bulletin of statistics, supplement. New York. Verbeek, M. (2000). A guide to modern econometrics. New York. Warren, G. F., & Pearson, F. A. (1937). World prices and the building industry. London. Wurm, C. A. (1989). International cartels, the state and politics: Great Britain between the wars. In: A. Teichova, M. Levy-Leboyer & H. Nussbaum (Eds), Historical Studies in International Corporate Business (pp. 111–122). Cambridge: Cambridge University Press.

APPENDIX: ECONOMETRIC ISSUES This appendix discusses the reasons why both the within (fixed effect) and the first difference estimators are used in the estimates of Eq. (9). The first-difference estimates may lead to less biased estimates of the lagged dependent variable than the within estimates. When T is low, the coefficient of the lagged dependent variable in the within estimates is negatively biased and converges towards approximately minus one for T = 2 and N → ∞ due to the correlation between the error terms and the fixed effects (see Hsiao, 1986). This bias does not disappear by taking first differences, even if T → ∞ (Kiviet, 1995). A second source of bias in estimates of the coefficient of the lagged dependent variable comes from the correlation between the lagged dependent variable and the error term. Using instruments for the lagged dependent variable or estimating in first differences to remove the fixed effects, will alleviate the problem of biased estimates of the estimated coefficient of the lagged dependent variable; however, Monte Carlo simulations by Kiviet (1995) show that various transformations of the data and alternative estimators, including the GMM estimator, do not the resolve the bias and consistency problems because of the correlation between the error terms and the fixed effects in small samples. To alleviate the consistency problem two period lags of the dependent variable are used as instruments for the lagged dependent variable, following B&C. This is not the most efficient method since all orthogonal conditions have not been exploited (Arellano & Bond, 1991). The first-difference models in this paper were also estimated using the generalised method of moments (GMM) estimator devised by Arellano and Bond (1991), and produced estimates of the lagged dependent variables that were higher than the estimates in this paper, but the estimates of the other coefficients were quite similar. It is, however, not clear which estimation method is the best. Hahn and Hausman (2002) show that the bias in the instrument variable estimator is positively related to the number of instruments. Since the GMM estimator uses a substantial number of instruments, it is not clear whether the efficiency gain from using the GMM estimator outweighs the instrumental variable bias. Second, the GMM estimator of Arellano and Bond is derived for N → ∞, which is not even approximately satisfied here.

The Length and the Depth of the Great Depression: An International Comparison

283

Another advantage of estimating in first differences is that it eliminates serial correlation in the residuals. Although serial correlation is allowed for in the level estimates, the serial-correlation coefficient needs to be restricted to be the same across countries because T is too low to relax this assumption in most of the estimates. If this restriction is violated, then the parameter estimates will become inconsistent and biased. If some regularity conditions are satisfied, consistent estimates can be recovered from a combination of first-difference and the within estimates as shown by Griliches and Hausman (1986). Suppose that the following model is estimated for a panel of i countries: y it = ␣i + ␤x it + u it , where u is a stochastic error-term and x is measured with an error: x ∗it = x it + vit where x∗ is the observed x and v is the measurement error, then we get the following probability limits for the within and the first-difference estimator, respectively:     2 2 ␴ 2␴ T − 1 v v P lim ␤ˆ d = ␤ 1 − 2 , P lim ␤ˆ w = ␤ 1 − (A1) N→∞ N→∞ T ␴2x˜ ␴dx where dx it = x ∗it − x ∗i,t−1 , x˜ it = x ∗it − x¯ it∗ , ␴2dx is the variance of the first differences of x, and ␴2x˜ is the variance of the within transformed x∗ ’s. These probability limits are derived under the assumption of no serial correlation in the measurement errors. From Eq. (A1) it follows that the within estimator is likely to be less sensitive to errors-in-variables than the first-difference estimator because the condition ␴2dx < 2␴2x˜ T/(T − 1) is likely to be satisfied (Griliches & Hausman, 1986). Equation (A1) can be solved to yield the following consistent estimate of ␤: ␤=

␾1 ␤ˆ w − ␾2 ␤ˆ d , ␾1 − ␾ 2

(A2)

where ␾1 = 2/␴2dx and ␾2 = (T − 1)/T␴2x˜ . This result is derived under the assumption that the measurement error is serially uncorrelated. If the measurement error is serial correlated, then the within estimator need not be less biased than the first difference estimator because the first difference transformation eliminates some of the measurement error. This can be seen from the following. Assuming that cov(x it , u it ) = cov(␣i , vit ) = cov(x it , vit ) = cov(␣it , u it ) = cov(u it , vit ) = 0, then the probability limit of the least squares

284

JAKOB B. MADSEN

estimator is (Griliches & Hausman, 1986): 
␴2v ˆ , P lim ␤w = ␤ 1 − 2 N→∞ ␴x + ␴2v 
␴2v ˆ P lim ␤d = ␤ 1 − , N→∞ (1 − ␳x )/(1 − ␳v )␴2x + ␴2v where ␳x is the first-order serial-correlation coefficient of (the level of) xit , and ␳v is the first-order serial-correlation coefficient of (the level of) vit . Comparing the two estimators, the first-difference estimator is more inconsistent than the level estimator if xit is more serially correlated than the measurement error, which is most likely to be the case for the variables used here except perhaps for wholesale prices. To examine the seriousness of the errors-in-variables problem Eq. (A2) is used to recover consistent estimates of the real wage elasticity of output in some of the estimates below. In these estimates the coefficients of the log of prices and wages are restricted to be the same and of opposite sign because the results become much more complex if more than one independent variable is measured with an error. Since, the restriction cannot be rejected at conventional significance levels in most circumstances, as shown below, this restriction will not impact on the results. Furthermore, the coefficients of Str and Panic are restricted to zero to simplify the analysis.

APPENDIX: DATA Economy-wide Compensation to employees. Canada. Series F1–13, F H Leacy (Ed.), 1983, Historical Statistics of Canada, Statistics Canada: Ottawa. Includes all sectors of the economy. USA. Table U.S.6, T Liesner (1989), One Hundred Years of Economic Statistics, Oxford: The Economist. Includes all sectors of the economy. Japan. Table A47, K Ohkawa, M Shinohara, and L Meissner, 1979, Patterns of Japanese Economic Development: A Quantitative Appraisal, London: Yale University Press. Includesthe non-agricultural sector. Finland. Table 12A, R Hjerppe, 1989, The Finnish Economy, 1860–1985, Helsinki: Bank of Finland, Government Printing Centre. Includes the non-agricultural sector. France. Table F.4, T Liesner, op cit. Includes the non-agricultural sector. Germany. Table 122, W G Hoffmann, F Grumbach, and H Hesse, 1965, Das Wachstum der Deutschen Wirtschaft seit der mitte des 19. Jahrhunderts, Berlin: Springer-Verlag. Includes all sectors of the economy. Netherlands. Table H1, C A Van Bochove and T A Huitker, 1987, Main National Accounting Series, 1900–1986, Occasional Papers No. NA-017, Central Bureau of Statistics, the Netherlands. Includes all sectors of the economy. U.K. Table 1, C H Feinstein, 1976, Statistical Tables of National Income, Expenditure

The Length and the Depth of the Great Depression: An International Comparison

285

and Output of the U.K. 1855–1965, Cambridge: Cambridge University Press. Includes all sectors in the economy. Economy-wide employment. Canada. Table C.7, T Liesner, op cit. Includes all sectors of in the economy. USA. Table U.S.11, T Liesner, op cit. Include all sectors in the economy. Japan. Table A53, K Ohkawa et al., op cit. Includes the non-agricultural sector. Finland. Table 11B, R Hjerppe, op cit. Include the non-agricultural sector. France. Table F.7, T Liesner, op cit. Includes the non-agricultural sector. Germany. Table 20, W G Hoffmann et al., op cit. Includes all sectors of the economy. Netherlands. Table XXIX, C Clark, 1957, The Conditions of Economic Progress, London: Macmillan. Includes all sectors of the economy. U.K. Table U.K.11, T Liesner, op cit. Includes all sectors of the economy. Average hours worked per employee. C Clark, op cit. Manufacturing value-added price deflator. Canada. Manufacturing nominal value added, Series F56–75, F H Leacy (Ed.), op cit, divided by manufacturing production, League of Nations, 1945, Industrialisation and Foreign Trade, Geneva. United States. BLS wholesale price index of manufacturing finished goods, Series E 84, Department of the Commerce, 1975, Historical Statistics of the United States: Colonial Times to 1970, Washington DC: Bureau of the Census. This index is based on articles for users, including raw foods and fuel. Japan. Net domestic product deflator in manufacturing and mining, Tables A11 and A12, K Ohkawa et al., op cit. The Bank of Japan’s wholesale price data have been the main source for the price deflator for manufactured goods. Australia. Table PC 61–70, W Vamplew (Ed.), 1987, Australians: Historical Statistics, Fairfax. Based on Butlin’s estimates, which are predominantly wholesale prices for manufacturing output. Denmark. Tables 2 and 3, S A Hansen, Økonomisk Vækst i Danmark, København: Akademisk Forlag. Based on consumer prices excluding housing, food, and taxes. Finland. Manufacturing nominal output divided by manufacturing production, Tables 4 and 6, R Hjerppe, op cit. Whether the deflators are value-added or output prices is not stated in the data source section of Hjerppe. France: Wholesale prices for manufactures, Series V16, J-C Toutan, 1987, Le Produit Interieur Brut De La France De 1789 A 1982, Economies et Societes. Germany. Wholesale price index of finished manufacturing products, Table A30, G Bry, 1960, Wages in Germany 1971–1945, Princeton: Princeton University Press. Italy. Manufacturing value-added price deflator, Table 5, G Fua, 1965, Notes on Italian Economic Growth 1861–1964, Milano: Mvlta Pavcis. Spain. Wholesale prices for manufacturing output, Tables 3 and 4, Instituto De Estudies Fiscales, 1976, Datos Basicos Para La Historia Financiera De Espana (1850–1975), Madrid: Ministerio de Hacienda. Sweden. ¨ Johansson, 1967, The Gross Manufacturing finished goods wholesale prices, O Domestic Product of Sweden and its Composition 1861–1955, Stockholm: Almqvist and Wiksell. U.K.: Manufacturing nominal value-added, Table 8, B R Mitchell, 1988, British Historical Statistics, Cambridge: Cambridge University

286

JAKOB B. MADSEN

Press, divided by manufacturing output, Table 51, Feinstein, C H, 1976, Statistical Tables of National Income, Expenditure and Output of the U.K. 1855–1965, Cambridge: Cambridge University Press. Hourly wage earnings in manufacturing. Mostly from ILO, Yearbook of Labor Statistics, and backdated from 1927 using League of Nation, Monthly Bulletin of Statistics and League of Nation, Statistical Yearbook. The coverage of the data, and if the sources differ from the sources above, for each individual country are the following, where I signifies agriculture, II mining, III, manufacturing, building, construction, power and water, IV, transport and communication, V, public administration, and IV, commerce and personal services, following the ILO classification. Hourly earnings across skills and genders are used unless mentioned. Argentina. III, IV, VI, Buenos Aires, monthly earnings, after 1929. Before 1929, J G Williamson, 1995, “The Evolution of Global Labor Markets since 1830: Background Evidence and Hypothesis,” Explorations in Economic History, 32, 141–196. The wage data are deflated by consumer prices and are therefore multiplied by consumer prices to convert them to nominal wages. Australia. II, III, IV, rates for males. Austria. II, earnings per shift. Belgium. II, III, IV. Before 1929, B R Mitchell, 1975, European Historical Statistics 1750–1975, London: Macmillan. Canada. Industry wages, B R Mitchell, 1983, International Historical Statistics: Americas and Australasia, London: Macmillan. Czech-Slovakia. III, hourly minimum rate in Prague. Denmark. III, IV. Estonia. III. Finland. Mitchell, 1975, op cit., money wages in industry. France. III, IV, hourly rates for skilled men in Paris. Germany. II, III, IV. 1925–1928, G Bry, 1960, op cit. Hungary. III. Italy. III. Before 1927, Mitchell, 1975, op cit., average daily wages. Japan. III, daily earnings, Imperial Cabinet. Latvia. III, skilled men. New Zealand. I, II, III, IV, men. Netherlands. II, III, men. Before 1928, Williamson, 1995, op cit., average daily earnings. Norway. II, III, daily earnings for men. Before 1928, Mitchell, 1975, op cit. hourly earnings for engineering. Poland. II, III. Spain. III, 1925–1933, Table 12.14, A Carreras, 1989, Estadisticas Historicas de Espa˜na, Madrid: Fundacion Banco Exteriror. 1934–1938, Williamson, 1995, op cit., workers in building trades in Madrid, unweighted average. Sweden. II, III, IV, VI. Switzerland. III, IV, men. Before 1929, Mitchell, 1975, op cit., weekly earnings for males involved in accidents. U.K. I, II, III, IV, V. Weekly rates. United States. III, C Hanes, 1996, “Changes in the Cyclical Behavior of Real Wage Rates, 1869–1990,” Journal of Economic History, 56, 837–861. Alternative wage set. The same wage data are used except for the follwing countries. Belgium. Average of several professions, Williamson, 1995, op cit. Canada. III, Unskilled men. Czech-Slovakia. I, III, VI, daily insured wages. France. Average for all sectors, Williamson, 1995, op cit. Japan. III, daily earnings, Bank of Japan. U.K. Compensation to employees in manufacturing, Table 8, Feinstein, op cit, divided by manufacturing employment, Table 59, Feinstein, op cit., and hours worked in

The Length and the Depth of the Great Depression: An International Comparison

287

manufacturing, ILO, Yearbook of Labor Statistics. Commodity price index. For all countries except Denmark: Warren and Pearson, 1937, World Prices and the Building Industry, London: John Wiley and Sons. Index for 40 basic commodities, where the same weighting was used for all countries. The weights reflect the importance of the commodity in the world. Consumer prices, wholesale prices, currency in circulation, and exchange rates. League of Nations, Monthly Bulletin of Statistics, Geneva. Indirect and direct tax rates. Direct and indirect taxes divided by nominal GNP. Direct and indirect taxes: B R Mitchell, 1982, International Historical Statistics: Asia and Africa, London: Macmillan, Mitchell, 1975, op cit. and Mitchell, 1983, op cit. Export price competitiveness. Multilateral index using export unit values as deflators. See J B Madsen, 2001, “Trade Barriers and the Collapse in International Trade during the Great Depression,” Southern Economic Journal, 67, 848–868, for construction and data sources. Unemployment. Two sets of unemployment data are used. The first set is based mainly on union unemployment statistics and is calculated by W Galenson and A Zellner (1957), Measurement and Behavior of Unemployment, NBER, Princeton: Princeton University Press, for 10 countries. ILO’s, Yearbook of Labor Statistics, Geneva, is used for the remaining countries. For Argentina, Austria, Estonia, Hungary, Italy, Latvia, New Zealand, Poland, and Spain, for which unemployment data are not available, unemployment is generated from manufacturing employment using the following method. The change in unemployment is regressed on the change in manufacturing employment over the period from 1930 to 1936 using pooled cross-section and time-series data for the countries where both series are available. The coefficient of the change in manufacturing employment is used to simulate unemployment series in first differences for the countries where unemployment data are not available. Finally, the level of unemployment is generated by accumulating the first-differencing unemployment data. The other set of unemployment data are based on estimates for the whole labor force and is only used for the 12 countries for which the manufacturing value-added price deflator is available. The following sources are used. Canada. A Maddison, 1964, Economic Growth in the West. Comparative Experience in Europe and North America. London: George Allen & Unwin LTD. USA. 1925–1930, C Romer, 1986, “Spurious Volatility in Historical Unemployment Data,” Journal of Political Economy, 94, 1–37. 1931–1939: M R Darby, 1976, “Three and a Half Million U.S. Employees have been Mislead: Or, and Explanation of Unemployment,” Journal of Political Economy, 84, 1–16. Japan. Galenson and Zellner op cit. Australia. M Keating, 1973, Australian Workforce and Employment 1910–1911 to 1960–1961, the Australian National University, Canberra. Denmark. P J Pedersen, 1977, “Langtidssammenhæng mellem Produktivitetsstigning og Beskæftigelsesgrad,” Nationaløkonomisk Tidsskrift,

288

JAKOB B. MADSEN

175–192. Finland. A Maddison, 1977, “Phases of Capitalist Development,” Banca Nazionale del Lavoro-Quarterly Review, 121, 103–137. France. Galenson and Zellner, op cit. Germany. Maddison, 1964, op cit. Italy. Maddison, 1964, op cit. Spain. Monthly Bulletin of Statistics, League of Nations, Geneva. Sweden. O H Grytten, 1992, Arbeidsledighetens Omfang i Mellomkrigstiden, Historisk Tidsskrift, 3. U.K. Feinstein, op cit. Industrial production. Mitchell, 1975, 1982, and 1983 op cit, I Svennilson, 1954, Growth and Stagnation in the European Economy, United Nations Economic Commission for Europe, Geneva, and T Liesner, op cit. Person-days lost in strikes and lockouts. Mitchell, 1975, 1982, and 1983 op cit. Gold standard periods. B Eichengreen, 1992, Golden Fetters: The Gold Standard and the Great Depression, 1919–1939, Oxford University Press: Oxford. Short-term interest rates. League of Nations, Money and Banking and Monthly Bulletin, Geneva. Economy-wide GNP deflator. Nominal GNP divided by real GNP. Canada. Table C.1, T Liesner, op cit. USA. Table U.S.1, T Liesner, op cit. Japan. Tables A1 and A2, K Ohkawa et al., op cit. Finland. Tables 1 and 2, R Hjerppe, op cit. France. Table F.1, T Liesner, op cit. Germany. Table G.1, T Liesner, op cit. Netherlands. Van Bochove and Huitker, op cit. U.K. Table U.K.1, T Liesner, op cit. Import unit values, import value and macro tariff rates. See Madsen, 2001, op cit. for sources.

THE DECLINE AND RISE OF INTERSTATE MIGRATION IN THE UNITED STATES: EVIDENCE FROM THE IPUMS, 1850–1990 Joshua L. Rosenbloom and William A. Sundstrom ABSTRACT We document long-run trends in interstate migration rates, using individuallevel data from the U.S. Census for the period 1850–1990. Two measures of migration are calculated. The first considers an individual to have moved if she is residing in a state different from her state of birth. The second considers a family to have moved if it is residing in a state different from the state of birth of one of its young children, allowing us to estimate the timing of moves more precisely. Overall migration propensities have followed a U-shaped trend since 1850, falling until around 1900 and then rising until around 1970. We examine variation in the propensity to make an interstate move by age, sex, race, nativity, region of origin, family structure, and education. Counterfactuals based on probit estimates of the propensity to migrate suggest that the rise in migration of families since 1900 could be explained by increased educational attainment, although education may be serving as a proxy for unmeasured covariates. The decline of interstate migration in the late nineteenth century remains to be explained.

Research in Economic History Research in Economic History, Volume 22, 289–325 © 2004 Published by Elsevier Ltd. ISSN: 0363-3268/doi:10.1016/S0363-3268(04)22006-7

289

290

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

1. INTRODUCTION The mobility of the American population has played an important role in the country’s economic development. The settlement of the frontier and urbanization are two of the great themes of American economic history. Efforts to study the history of internal migration in the United States have been hampered by a variety of data limitations, however. Since 1940 researchers have been able to make use of data on recent migration experience collected in the Census, the Current Population Survey, and panel data sets.1 For evidence that extends prior to that date, though, researchers have often been obliged to rely on indirect measures calculated using either census survival methods or aggregate data on the native population’s state of residence and state of birth.2 For the study of internal migration, such data have major limitations: the census survival method only measures net migration (rather than gross flows in and out of a location), and both measures are aggregate and thus are of limited use in examining the factors affecting individual migration decisions.3 In this paper we explore two ways of utilizing individual-level data from population censuses assembled in the Integrated Public Use Microdata Series, or IPUMS (Ruggles et al., 1997), to derive new measures of long-run trends in the migration of the native born within the United States. By using information on age in combination with state of birth and state of residence, we can follow interstate migration patterns for successive synthetic birth cohorts of individuals from 1850 to 1990. This allows us to describe changes in migration rates over time and how the propensity to leave one’s state of birth varied with individual characteristics. Unfortunately, this measure of migration has some serious deficiencies: it fails to indicate the timing of an individual’s move between birth and the census and fails to count moves subsequent to leaving the birth state. For families with children, however, we can obtain a much more precise measure of recent family moves by matching children with their parents and comparing a young child’s state of birth with the family’s state of residence at the time of the census.4 Both measures suggest that overall migration rates controlling for the agecomposition of the population followed a U-shaped trajectory between 1850 and 1990, falling from 1850 until roughly the turn of the century and rising thereafter, especially after 1940. Several points about this result warrant comment. The first is the fact that age-adjusted rates of migration at the end of the twentieth-century are no higher than they were in the mid-nineteenth century. In light of the dramatic declines in costs of communication and travel over this period the high levels of mobility in the last century are especially striking and serve to emphasize just how significant internal migration was in the nineteenth century.5 The second is that the U-shaped pattern of internal migration rates differs from the time

Decline and Rise of Interstate Migration

291

pattern of mobility found by earlier studies that have relied on aggregated data and consequently did not control for age. For example, based on comparisons of state of birth and state of residence data from published census returns Eldridge (1964, p. 109) finds no trend in the volume of gross population flows between 1870 and 1950.6 The third point is the negative correlation between rates of immigration and interstate migration of the native-born. That rates of internal migration reached a low point at the same time that immigration was reaching its peak suggests that immigration may have been at least a partial substitute for internal redistribution in response to shifting economic opportunities. To explore the relationship between the propensity to move and individual characteristics, such as gender, race, and region of origin, we estimate migration probits on census cross sections to identify the marginal impact of these characteristics. In the mid-nineteenth century, there was a large gender gap in migration propensities, presumably reflecting the low levels of female labor force participation at this time, and the predominant role of economic factors in directing population redistribution. From 1860 onward, we observe a decline in the gender gap, and, since 1940, female and male propensities to move between states have been insignificantly different. In the nineteenth-century there was also a pronounced racial gap in propensities to migrate. Only in the middle of the twentieth century, during the Great Migration of African-Americans, did overall migration propensities of blacks exceed those of whites: both before and since, African-Americans have been less likely to leave the state of their birth than whites. Among families with children, black families continued to have lower migration rates than whites even after World War II, suggesting that the Great Migration consisted disproportionately of single and/or childless individuals. Examining the role of region of birth, we find that individuals born in the Northeast region had the lowest migration propensities over much of the period, although regional differences had narrowed by 1990. In the postwar period, southerners and westerners have been the most likely to move between states. Perhaps our most interesting finding about the impact of personal characteristics on migration behavior concerns the potential role of rising educational attainment in explaining the upward trend in migration propensity over the past 100 years. Education is strongly correlated with mobility in each cross-section. Using our migration measure based on moves by families with children, we estimate the impact of education in cross section and use the coefficients from a single census year to calculate a counterfactual migration series for changing personal characteristics, taking advantage of Goldin’s (1999) recent estimates of high-school graduation rates for years before 1940. The counterfactual does a good job tracking the rise in migration propensity since the early 1900s, and changing educational attainment was the principal driving force. This finding cannot be considered an adequate

292

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

causal account, however, due to omitted variables that were likely to have been correlated with education, such as occupation and income. In the next section we review the literature on the history of internal migration in the United States, its causes and effects. The third section describes our measure of lifetime migration based on state of birth, discusses its limitations, and summarizes our findings on cohort and life-cycle migration patterns and differences by sex, race, and region of birth. The fourth section employs the family migration measure based on child’s state of birth, again examining covariates of migration, and presents the counterfactual exercise that shows the important effect of education. The fifth section offers conclusions and directions for further research.

2. INTERNAL MIGRATION IN THE UNITED STATES SINCE 1850 The geographical mobility of Americans is a well-known trait of the national character. According to figures cited by Greenwood (1997), as of 1970 the average American would make nearly twice as many residential moves during her or his lifetime as would the average resident of Britain or Japan. Migration – along with regional differences in rates of natural increase – has played a central role in the geographical redistribution of the U.S. population. Using aggregated data from published census volumes to examine the sources of population displacement, Eldridge (1964) showed the centrality of westward migration in the population increase of the West, the role of out-migration from the South in offsetting the South’s high rate of natural increase, and the net effect of European immigration to the Northeast in offsetting lower rates of natural increase there.7 Until the 1920s, net migration into the Northeast and North Central regions was dominated by European immigrants, while internal migration played a dominant role in the South and West. These net interregional migration flows obscure much of the underlying population movement. For example, after 1910, the net migration from the South was largely into the Northeast and North Central regions, which in turn contributed the bulk of the migrants to the West. Since 1950 there have been important changes in internal migration patterns. Most crucially, since about 1970 the flow of migrants out of the South has been reversed, with the South becoming the region with the largest net in-migration (Greenwood, 1997). Around the same time, the longstanding historical pattern of net migration from non-metropolitan to metropolitan areas also reversed. During the postwar period, differences in employment growth across states can be attributed primarily to migration, rather than differences in rates of natural increase (Blanchard & Katz, 1992).

Decline and Rise of Interstate Migration

293

Economic accounts of internal migration have often stressed locational differentials in wages or incomes as the driving force of migration patterns (see Greenwood, 1997, for an overview of the literature). In this sense migration can be seen as a process of equilibration of the national labor market. Using published census data from 1850, 1880, 1900, 1920, and 1960, Gallaway and Vedder (1971) examine the effect of income differentials on interstate migration flows, as measured by the number of individuals born in one state but residing in another. They find that the migration into a state increases with its per-capita income and decreases with its population density and distance from the state of origin. Over these years, the estimated elasticity of migration with respect to income increased, while that with respect to distance fell. The latter result is consistent with declining costs of moving over the period. Although Gallaway and Vedder find a significant role for geographical income differentials in determining migrant flows, other studies of this effect have obtained mixed results (Greenwood, 1997). Recent research suggests that other economic factors- in addition to local average wages or incomes-should be considered important determinants of migration patterns. Steckel (1983) and more recently Stewart (2003) stress the importance of specific human capital in farming as a factor directing westward migration flows historically. In particular, migrating farmers tended to move roughly along lines of latitude, presumably because of latitude-specific knowledge about crops and livestock. Theoretical models that assume heterogeneous worker skills imply that migration propensities should depend not only on the mean wage but also on the dispersion of wages (skilled workers want to migrate to places where skills are highly rewarded).8 This prediction is confirmed empirically by Borjas et al. (1992). Locational differences in employment growth or unemployment rates have been found in some studies to have a greater impact than wage differentials.9 Differences in government transfer programs or spending across location may also affect migration flows, as Fishback et al. (2001) find for New Deal spending. Given Gallaway and Vedder’s finding that internal migration redistributed labor from low- to high-wage states, it might be surmised that internal migration played a large role in the convergence of state wages and per-capita incomes over the past 120 years. Rosenbloom (1990, 1996) has argued that regional convergence in wage rates during the late nineteenth and early twentieth centuries, at least outside of the South, coincided with the emergence of cross-regional labor-market institutions and informational flows. However, the direct evidence linking labor-market and income convergence during the twentieth century is not strong. Barro and Sala-i-Martin (1991) find that migration explains only a small part of overall economic convergence across states. And as Kim (1998) notes, the process of convergence involved not only within-sector wage convergence but

294

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

also convergence in industry composition, which may have been due to causes other than the integration of labor markets. One enduring puzzle relating to convergence within the United States is the persistent difference in real wages (and per capita incomes) between the South and the rest of the country. The catch-up of the South, particularly after the Great Depression, was a significant source of economic convergence within the United States (see Barro & Sala-i-Martin, 1991; Mitchener & McLean, 1999; Wright, 1986). Yet why did labor migration fail to narrow this gap before the 1940s? Wright (1986) has argued that prior to the New Deal, the southern labor market remained isolated from the rest of the country, in large part because the demand for low-skilled labor in the industrializing North was satisfied by European immigrants, while flows of information and migrants between North and South were never established. Fishback (1998) has also noted that for a time southerners found high-wage opportunities by moving westward within the South. Better evidence on gross migration flows between the South and the rest of the country will help us better assess the degree and causes of southern isolation. The path dependence of migrant flows when migrant stocks affect the propensity of migration has been emphasized by Carrington et al. (1996) as a significant factor in the delay of African-American migration to the North, in spite of the lower wages and greater social and political oppression of blacks in the South.10 An alternative explanation of the delayed timing of the Great Migration is that prior to the 1920s black employment opportunities in the North were severely constrained by competition from unskilled European immigrants, given the racial preferences of northern employers (Collins, 1997). The special obstacles to black migration do not, of course, provide an explanation for the persistent North-South wage gap for white workers. Migration decisions are also influenced by the personal characteristics of potential migrants. The migration propensity of adults, at least for long-distance moves, tends to decline with age and to increase with education.11 Schwartz (1976) argues that these properties are implications of the equilibrium relationship between earnings, education, and experience; additionally, education may be associated with better sources of information, greater ability to process information about opportunities, or lower risk aversion. The age effect is also partly attributable to changes in family status and career that are correlated with age (Sandefur & Scott, 1981). Unemployed individuals are more likely to move than the employed, other things equal (DaVanzo, 1978). Married couples are less likely to move if both individuals are in the labor force (Greenwood, 1997). A distinct advantage of using the individual-level samples of the IPUMS to examine historical migration behavior is that it becomes possible to control for the impact of demographic characteristics and family labor allocation decisions on migration probabilities.

Decline and Rise of Interstate Migration

295

In this paper we focus on broad trends in the propensity to migrate and the demographic covariates of that propensity. We leave to future work an analysis of the responsiveness of migration to differences in economic opportunities across location.

3. MIGRATION PATTERNS REVEALED BY STATE OF BIRTH, 1850–1990 Using data from the IPUMS (Ruggles et al., 1997), our first measure of migration is based on whether an individual was living in the state in which he or she was born at the time of the census. We employ samples from thirteen census years spanning the period from 1850 to 1990.12 The IPUMS combines census microdata files produced by the U.S. Census Bureau since 1960 with new historical census files produced at the University of Minnesota and elsewhere.13 Each census in the IPUMS data included questions on the state of birth and state of residence of the native-born population along with questions about each individual’s age at the time of the census and other demographic characteristics, such as gender and race. Because we are concerned with internal migration between states, we restrict our attention to the native born. Our second measure of migration, using the birthplace of a child, allows us to consider interstate migration of families with foreign-born parents. Before exploring the historical trends in migration revealed by the data on state of birth, we consider their relationship to more direct measures of migration that are available for 1940 and later.

3.1. Reliability and Limitations of Measuring Migration Using State of Birth Using information on state of birth and state of residence to measure interstate migration will understate the size of gross migration flows for several reasons. First, we can only ascertain whether a person has ever moved between birth and the census; the number and exact timing of moves cannot be known. Second, some individuals who have moved during their lifetime will be missed by our measure, because they have moved away from their state of birth and later returned to it.14 Third, we do not capture within-state moves.15 The potential size of the first two biases can be examined using census data on five-year migration rates in 1940, 1960, 1970, 1980, and 1990. These censuses asked individuals where they were residing five years prior to the census date.16 For these years we can examine the correlation of ever moving (since birth) and five-year moving, as well as the frequency of return migration to the state of

296

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

Table 1. Comparison of Interstate Migration Rates Since Birth and in the Five Years Preceding Census (%). Age Group

Moved Since Birth

Moved Last Five Years

1940

10–19 20–29 30–39 40–49 50–59 60–69 70–79

12.6 23.3 30.4 33.1 34.5 36.1 38.2

4.5 8.7 7.5 5.1 3.9 3.3 2.6

1960

10–19 20–29 30–39 40–49 50–59 60–69 70–79

19.2 32.3 34.8 35.6 36.1 36.6 37.6

8.6 19.8 11.1 6.5 4.6 4.4 3.9

1970

10–19 20–29 30–39 40–49 50–59 60–69 70–79

19.1 32.9 34.9 35.5 35.4 35.8 35.8

8.5 19.9 12.2 6.6 4.2 4.3 3.7

1980

10–19 20–29 30–39 40–49 50–59 60–69 70–79

22.3 31.9 38.1 39.3 39.4 39.7 40.2

9.3 17.0 12.9 7.4 4.8 5.2 3.8

1990

10–19 20–29 30–39 40–49 50–59 60–69 70–79

21.8 32.7 36.7 40.8 41.0 41.2 41.6

9.3 16.8 12.1 8.0 5.5 5.2 3.7

Source: IPUMS samples of U.S. Census (Ruggles et al., 1997). Samples restricted to individuals with known state of residence at time of birth, 5 years prior to the census, and on the census date.

Decline and Rise of Interstate Migration

297

birth. Toward this end we have drawn samples of individuals satisfying each of the following selection criteria: (1) known state of birth (state or DC); (2) known state of residence (state or DC) five years prior to the census; (3) known state of residence (state or DC) at the time of the census; and (4) ages 10–79. Table 1 shows the percentages having moved since birth and in the last five years, broken down into 10-year age groups. The age pattern revealed by the table suggests that the probability of having moved in the past five years is greatest for young adults (20–29) and tends to tail off beyond that age, while the probability of ever having changed state since birth appears to be cumulative, as one would expect, but rises at a decreasing rate with age. By following birth cohorts across successive censuses, we can estimate the proportion of an age cohort who left their state of birth during the 10 years preceding the census. The implied age profiles are shown in Fig. 1.17 This can be compared with the age profile of the proportion reporting an interstate move in the five years prior to the census, in Fig. 2. Clearly, both measures show that the propensity to move declines dramatically with age. It is also clear that the measure based on state of birth misses a large number of moves. The 10-year propensity based on birth state is lower in nearly every case than the corresponding 5-year propensity based on the migration question. The proportional magnitude of the bias rises with age, presumably reflecting additional moves by individuals who have already left their birth state.

Fig. 1. Percentage of Age Cohort Having Left Birth State During Preceding 10 Years, by Age at the Time of the Census. Source: IPUMS samples (Ruggles et al., 1997).

298

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

Fig. 2. Census 5-Year Migration Rates, by Age at Census (Percent). Source: IPUMS samples (Ruggles et al., 1997).

Clearly, migration estimates based on comparing state of birth with state of residence are a better measure of gross migration the greater the proportion of migrants who are leaving their state of birth for the first time. Table 2 classifies individuals who moved within five years prior to the census into three categories: (1) those who returned to their state of birth after having lived elsewhere five years prior; (2) those who left their state of birth during those five years; and (3) those who were not living in their state of birth five years prior and moved to another (third) state by the year of the census.18 Note that the first of these categories consists of people who did migrate but who will not be counted as movers in our state-of-birth based measure, because as of the census they were back in their state of birth. For younger people (between 10 and 29), more than half of 5-year moves involved leaving the state of birth in 1940, and about half in 1970–1990. Not surprisingly, the percentage of 5-year moves that involved leaving the birth state falls with age. For individuals in their 40s, less than one-third of moves involved leaving the state of birth in most years. Clearly, age profiles of migration based on leaving the birth state are going to be excessively concave, because an increasing proportion of moves are not picked up by the measure. It is also noteworthy that the proportion of moves that involved returning to the state of birth from elsewhere shows no systematic age pattern.

Decline and Rise of Interstate Migration

299

Table 2. Distribution of 5-Year Movers (%). Age Group

Returned to Birth State

Left Birth State

Moved Between Two Non-Birth States

1940

10–19 20–29 30–39 40–49 50–59 60–69 70–79

18.1 13.7 17.3 18.9 18.0 18.8 17.1

62.9 62.4 46.3 37.5 35.1 33.3 35.8

18.9 23.9 36.4 43.7 46.9 47.9 47.0

1970

10–19 20–29 30–39 40–49 50–59 60–69 70–79

18.9 19.0 22.9 21.8 21.3 22.0 22.2

49.4 50.1 31.0 29.6 31.1 36.2 35.9

31.8 30.9 46.2 48.7 47.6 41.8 41.9

1980

10–19 20–29 30–39 40–49 50–59 60–69 70–79

17.6 17.8 19.9 21.4 18.5 17.3 17.7

48.8 47.2 31.7 27.8 32.5 39.1 35.9

33.6 34.9 48.5 50.7 49.0 43.7 46.4

1990

10–19 20–29 30–39 40–49 50–59 60–69 70–79

18.0 16.9 20.4 18.7 20.0 18.5 19.4

51.1 48.3 33.5 29.6 31.2 35.7 34.4

30.9 34.7 46.1 51.7 48.8 45.8 46.2

Source: See Table 1.

Individuals who have moved at least once in their lifetime are more likely to move again. This can be seen in Table 3, which gives the percent who moved during the five years preceding the census, conditional on whether they had moved between birth and 5 years prior. In each age category, those who had left their birth state by t-5 were much more likely to move to another state between t-5 and t. This is true even when we don’t count those who returned to their state of birth between t-5 and t (last column). Thus there appears to be persistent heterogeneity in migration propensities. Whether this is a trait of individuals or perhaps a characteristic of locations that tend to receive migrants (e.g. more volatile economic

300

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

Table 3. Percent Moving During 5 Years Prior to Census by Previous Migration Status. Age Group

Lived In Same State At Birth And 5 Years Before Census

Moved Between Birth and 5 Years Before Census Include Returns to Birth State

Exclude Returns to Birth State

1940

10–19 20–29 30–39 40–49 50–59 60–69 70–79

3.2 6.7 4.8 2.8 2.1 1.7 1.5

15.8 17.2 14.2 10.0 7.5 6.1 4.5

8.1 10.9 9.6 7.0 5.4 4.4 3.3

1970

10–19 20–29 30–39 40–49 50–59 60–69 70–79

5.0 13.6 5.7 3.0 2.0 2.4 2.1

26.0 37.1 24.9 13.3 8.3 7.9 6.7

16.3 23.0 16.6 9.2 5.7 5.2 4.4

1980

10–19 20–29 30–39 40–49 50–59 60–69 70–79

5.6 11.0 6.4 3.4 2.5 3.3 2.3

24.4 33.4 24.0 13.8 8.4 8.2 6.2

16.0 22.1 17.0 9.7 6.1 5.9 4.5

1990

10–19 20–29 30–39 40–49 50–59 60–69 70–79

5.8 11.2 6.3 4.0 2.9 3.1 2.2

24.3 31.6 23.0 14.2 9.4 8.3 6.0

15.3 21.2 15.9 10.4 6.6 5.9 4.2

Source: See Table 1.

opportunities, hence receiving but also sending away many migrants) remains to be determined. These findings suggest that migration rates based on state of birth must be treated with caution. They are likely to be more informative for younger people. In addition to their imprecision in terms of the timing of moves, their inability to capture moves subsequent to leaving the birth state implies that they systematically

Decline and Rise of Interstate Migration

301

underestimate the propensity to move, especially for older individuals. Although we report some general trends and demographic variation using this measure here, we think it is crucial to obtain more accurate estimates of migration, which we are able to do for families with children, as reported later in the paper.

3.2. Long-Run Trends in the Propensity to Leave the Birth State Figure 3 plots migration propensities for the full samples of the native-born as a function of age at each census. For most age groups, lifetime migration propensities exhibit a downward trend between 1860 and sometime around 1900. For 30–39 year olds, for example, the percent having left their birth state fell from 42% in 1860 to 31% in 1900. The trend is clearly upward after 1940, although measured migration rates appear to have been unusually low in 1940 at the end of the Great Depression.19 For ages 30–39, the propensity to migrate rose from about 30% in 1940 to around 37% in 1990. For individuals under 40, the pattern describes a long U shape, with the lowest point occurring around 1900, especially if we smooth over the unusually depressed migration rates of 1940. The trough occurs later for the older age groups, although as we have seen the measure based on birth state becomes a less

Fig. 3. Proportion Having Left State of Origin by Age Group, IPUMS Samples. Source: IPUMS samples (Ruggles et al., 1997).

302

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

reliable measure of migration rates as we examine older individuals. Interestingly, the trough in migration propensities among the younger native born coincides with the period of heaviest foreign migration into the country. The compression of the migration rates across the older age groups after 1920 suggests that migration has become increasingly concentrated among the young. By 1970, a person in her 70s was no more likely to be living outside the state of her birth than a person in her 30s. This flattening of the age profile of migration for older individuals is evident in cohort age profiles based on these migration rates.20 The phenomenon of the increased age concentration of migration is of some interest, although given the inability of this measure to capture a large percentage of moves among older individuals one must be wary of drawing strong inferences from the age profiles.

3.3. Covariates of Migration Propensity Using the IPUMS data we can estimate a simple descriptive model of migration propensity to see how demographic characteristics and location of origin affected the probability of leaving the birth state over time, and to determine whether changes in these characteristics might account for the observed historical trends. Here we present the results of estimating a probit model of the interstate move for 30–39 year olds.21 The regressors are age, gender, race, region of birth, and physical size of state. We include a control for state size because interstate moves are more likely to be observed from smaller states just because the border is likely to be closer. For this purpose we use the square root of the land area of the state of origin.22 Probit coefficients for each census year are provided in the appendix Table A1. Most of the coefficients are significantly different from zero in most years. To summarize trends in the effects of the variables, Figs 4-7 plot the time path of the estimated effect of each regressor on the move probability (dP/dX).23 For dummy variables, these effects represent the change in predicted probability of switching the variable from 0 to 1, evaluated at the means of the other variables. These marginal effects should be compared against an average move probability for this age group of between 0.3 and 0.4, depending on the year. Figure 4 shows the gender differential. Before 1940, women were less likely than men to report having left their birth state. After 1940 the coefficient on female is occasionally statistically significant, but is very small: there is today essentially no gender differential. Figure 5 shows the trend in the coefficient on black (the comparison group is all non-blacks, which consists predominantly of whites). The 1850 and 1860 censuses included only free persons, so the sample

Decline and Rise of Interstate Migration

303

Fig. 4. Effect of Gender on Probability of Leaving Birth State, 30–39 Year Olds, Differential for Female Relative to Male. Note: Effect based on probit coefficient, controlling for age, race, region of origin, and size of state of origin. Source: IPUMS samples (Ruggles et al., 1997).

Fig. 5. Effect of Race on Probability of Leaving Birth State, 30–39 Year Olds, Differential for Black Relative to White. Note: Effect based on probit coefficient, controlling for age, gender, region of origin, and size of state of origin. White includes all non-black races. Source: IPUMS samples (Ruggles et al., 1997).

304

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

Fig. 6. Effect of Region of Origin on Probability of Leaving Birth State, 30–39 Year Olds, Relative to Northeast (=0). Note: Effect based on probit coefficient, controlling for age, gender, race, and size of state of origin. Source: IPUMS samples (Ruggles et al., 1997).

Fig. 7. Effect on Probability of Leaving Birth State of Increasing the Size of the State of Origin by 100 Units (Square Root of Square km), 30–39 Year Olds. Note: Effect based on probit coefficient, controlling for age, gender, race, and region of origin. For comparison, a move from NY to CA would increase sqrt(area) by 286. Source: IPUMS samples (Ruggles et al., 1997).

Decline and Rise of Interstate Migration

305

of African-Americans shifts dramatically after the Civil War. Still, the picture is consistent with black migration rates being substantially lower until around 1900. By 1950, blacks were much more likely to have left their birth state, a reflection of the Great Migration out of the South. The racial differential has since then returned to within 5 percentage points of parity. Figure 6 shows the effect of region of birth, relative to the Northeast. The sample sizes of individuals born in the West for censuses before 1900 are very small, and the coefficients have been excluded here. The migration rates here include interstate moves within the region of birth, so these regional differentials do not necessarily represent the rate of out-migration from the region as a whole. Throughout the period, individuals born in the Northeast were the least likely to have left their state of birth by their 30s. Individuals born in the Midwest were the most peripatetic in the early 1900s, whereas southerners and westerners have generally been the most likely to move since 1940. By 1990, regional differences were quite small. Finally, Fig. 7 shows the effect of state size, as measured by the square root of area. As expected, the effect is usually negative, indicating that a person born in a large state is less likely to have reported leaving that state by her 30s. The effect is fairly large: the difference in move probability between an individual born in New York vs. one born in California would have been on the order of 6 percentage points in 1970, just because of the difference in area. Interestingly, this effect has tended to become stronger over time, contrary to what one might expect given declining real transportation costs.

3.4. Can Changing Demographic Characteristics Account for the Long-Run Trends in Mobility? The probit results reveal systematic variation in the probability of leaving the birth state by gender, race, region of origin, and state size. We can use a simple counterfactual to judge whether changes in these variables over time might account for the pronounced U shape in the migration propensity. Figure 8 shows one such counterfactual for 30–39 year olds. In the diagram, the actual migration rates are indicated by black diamonds. The white triangles show the results of a counterfactual. In each census year, the probability of an interstate move is predicted for each individual, using her or his actual X values and the probit coefficients from 1920.24 These predicted probabilities were then averaged over the whole sample to obtain the probabilities displayed in the figure. Clearly, changes in these observable characteristics cannot account for the actual historical experience.

306

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

Fig. 8. Actual and Counterfactual Rates of Leaving Birth State, 30–39 Year Olds. Source: IPUMS samples (Ruggles et al., 1997).

4. MIGRATION OF FAMILIES WITH CHILDREN The largest drawback of migration measures based on an adult individual’s place of birth is their inability to narrow down the timing of moves or measure repeat migration. For families with children, we can construct an alternative measure from census data that largely avoids these problems by comparing where the children were born with where the family was residing at the time of the census. For example, suppose a family living in Illinois reports that its five-year old child was born in Mississippi. Then we might conclude that the family moved sometime during the five or so years prior to the census date. In addition to allowing us to track migration over relatively short periods, the child-birthplace measure of migration has the advantage that we can use it to examine the internal migration of foreignborn as well as native-born adults.25 To construct the child-based measure from the IPUMS, we matched children ages 0–9 with their parents. Our sample consists only of families with both parents present and residing in an identified state or D.C. at the time of the census. For the years 1850–1870, the census data do not permit direct identification of spouses and own children; for those years, we used the IPUMS imputed family relationships, which are considered fairly reliable.

Decline and Rise of Interstate Migration

307

One child- referred to hereafter as the reference child- was selected at random from each such household, a procedure that allows us to avoid the problem of multiple observations for each family, which would give disproportionate weight to families with more children in the 0–9 age range.26 The unit of observation can thus be thought of as a two-parent family with at least one young child. Because we are interested in internal migration, an observation is included only if the reference child was born in the United States, in an identified state or D.C.27 We conclude that the family moved sometime between the birth of the reference child and the census if the child’s state of birth and the family’s current state of residence are different. Measuring the internal migration of adults using a child’s place of birth suffers from two obvious flaws. The first is that there is not a one-for one correspondence between the child being born in a different state and the family having resided in that state at the time of the birth. Some births may take place while a family is traveling, or perhaps because a mother may live temporarily with a relative during the period of the birth. The second problem is that families with children may not be representative of the migration behavior of the population as a whole. Recent evidence suggests, for example, that migration propensities fall after marriage and with the coming of children (e.g. Sandefur & Scott, 1981). The importance of the first objection appears to be relatively small. Two types of evidence support our claim that when the child’s state of birth and the family’s state of residence differ, it is likely that the family actually moved during the interim. First, as we show below, the probability that the states of birth and residence are different increases substantially with the age of the child, as it should if migration is at work, because older children have been at risk of moving for a longer period of time.28 Second, there is a high degree of correlation between the child-based migration measure and responses to the five-year migration questions available beginning in 1940. The censuses of 1940 and 1960 through 1990 asked individuals where they lived five years prior to the census. Children born during the year following that date would have been reported as four years old at the time of the census, while children born during the year preceding that date would have been reported as five. It seems reasonable, then, to compare the census five-year migration measure with a child-based migration measure for reference children ages four-five, an age range whose average birth date will be approximately five years before the census date. Table 4 compares our measure based on the reference child’s birth state with several alternatives based on the census five-year migration question: for parents in the family sample, and then for all men 20–49 years old. Although the kid-based measure tends to be a little greater than the census measure in the within-sample comparisons, it is generally quite close. Examining the last column, it is also

308

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

Table 4. Comparison of Kid-Based and Census 5-Year Move Probabilities (%). Year

Kid-Based Measure

Census 5-Year Migration Rate Family Sample

1940 1960 1970 1980 1990

7.0 13.3 14.3 15.2 14.7

All Men 20–49

Mothers

Fathers

Fathers 20–49

6.1 12.5 13.0 13.4 12.2

6.0 12.9 13.3 13.5 12.3

6.2 13.2 13.7 13.6 12.4

7.7 14.9 14.9 14.2 13.2

Notes: Kid-based measure is percentage of children aged 4 or 5 at the time of the census whose birth state differed from the family’s state of residence on the census date. Census migration rate is percentage of individuals reporting a state of residence five years before the census date different from their state of residence on the census date. Source: IPUMS samples (Ruggles et al., 1997).

interesting to note that the kid-based measure is rather close to the five-year measure for all men ages 20–49. Thus we can have some confidence that our measure is reasonably representative of the migration rate for the prime-age population as a whole, even though it is based only on a sample of two-parent families.29 Table 5 is a cross-tabulation of the two measures within the family sample. The measures are correlated, but not perfectly so. There are a number of moves Table 5. Cross-Tabulation of Kid-Based and Census 5-Year Migration Measures (%). Year

Kid

Father No Move

Move

1940

No move Move

92.0 2.0

1.0 5.0

1960

No move Move

84.2 2.9

2.5 10.4

1970

No move Move

83.1 3.6

2.6 10.7

1980

No move Move

82.5 4.1

2.3 11.1

1990

No move Move

83.8 3.9

1.6 10.7

Notes and source: See Table 4.

Decline and Rise of Interstate Migration

309

picked up by the kid-based measure that are not reflected in the father’s reported migration status, and somewhat fewer discrepancies in the opposite direction. The sources of these discrepancies are a topic for further inquiry, but it cannot be assumed a priori that the estimate based on the census’s five-year residence question is superior. For example, it seems plausible to us that parents’ memories regarding where they were living when a child was born may be more accurate than their memories about where they were living on a specific date five years ago. Given the substantial correlation between the measures, and the positive relationship between the child-based migration rate and the age of the reference child, we are confident that the child-based measure is at least a reasonable approximation of family migration propensities.30

4.1. Interstate Migration Propensities of Families with Children Figures 9 and 10 present the basic trends of interstate migration rates using the kid-based measure. Figure 9 presents the rates for each age of the reference child, while Fig. 10 is for four-five year olds, the measure closest to a five-year migration rate. The U shape of the series of migration propensities that appeared in the birthplace data for adults is also evident here. Migration rates tended to fall between

Fig. 9. Migration Rates by Age of Reference Child. Source: IPUMS samples (Ruggles et al., 1997).

310

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

Fig. 10. Five-Year Interstate Migration Rate Based on 4–5 Year Olds. Source: IPUMS samples (Ruggles et al., 1997).

1850 and 1880 and began to rise after 1900, particularly after 1940. In Fig. 9, the older the child observed, the greater the likelihood of a move (the curves shift up with age), consistent with the older child’s longer period at risk of migration. The strong dip in migration rates in 1940 is noteworthy. In Fig. 9, this dip is especially dramatic for nine-year-olds, and hardly evident at all for one-year-olds. A child of nine in 1940 was born in 1930 or 1931 and lived through the worst years of the Great Depression, when migration rates were apparently dramatically reduced. A child of age one in 1940 was born in 1938 or 1939, when the economy was in recovery and migration had presumably picked up again. The evidence of age compression in migration rates in 1940 illustrates the promise of the children’s birthplace measure as a means of pinpointing the timing of changes in migration propensity. It also suggests that 1940 was an unusual year for migration rates, and that the overall trend was probably upward after 1900.

4.2. Gross Interregional Migration Flows Using the family-based migration measure we can also track gross interregional flows over time. Figures 11–14 show gross in- and out-migration rates for the four U.S. regions over the period 1850–1990. Out-migration from the Northeast and

Decline and Rise of Interstate Migration

311

Fig. 11. Gross 5-Year Interregional Migration Rates of Families with Children, Northeast (As Percent of Families in Region 5 Years Prior to Census). Source: IPUMS samples (Ruggles et al., 1997).

Fig. 12. Gross 5-Year Interregional Migration Rates of Families with Children, Midwest (As Percent of Families in Region 5 Years Prior to Census). Source: IPUMS samples (Ruggles et al., 1997).

312

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

Fig. 13. Gross 5-Year Interregional Migration Rates of Families with Children, South (As Percent of Families in Region 5 Years Prior to Census). Source: IPUMS samples (Ruggles et al., 1997).

Fig. 14. Gross 5-Year Interregional Migration Rates of Families with Children, West (As Percent of Families in Region 5 Years Prior to Census). Note: Y-axis scale differs from Figs 11–13. Source: IPUMS samples (Ruggles et al., 1997).

Decline and Rise of Interstate Migration

313

the South exhibit the familiar U-shaped pattern, while in-migration was U-shaped for the Midwest and downward sloping for the West during the late nineteenth century. After 1900, out-migration rates rise in every region. An interpretation of these patterns would be that around the middle of the nineteenth century, migration rates were high as people left the Northeast and South and populated the Midwest and West. Overall migration propensities declined as the settlement process slowed in the late 1800s. After 1900, the rising propensity to move was associated with increased interregional flows in all directions.

4.3. Covariates of Five-Year Moves by Families with Children We estimated migration probits analogous to those presented in Section 3, using as a dependent variable the migration measure based on families with a reference child four or five years old. The independent variables included the age of the child (to control for the extra year at risk of moving for a five-year-old), the ages of each parent, the race and nativity of the father,31 the number of siblings in the household and the presence of a very young (under two) sibling, and two characteristics of the state of birth of the reference child: state size and region. Probit coefficients for each census year are provided in the appendix Table A2. The regressors used here are available for each year in our samples. Below we consider the effects of literacy and education, which are not consistently available across all years. In most years, the basic controls have the expected signs. The migration propensity tends to be greater if the reference child is a year older (five instead of four), which is consistent with the pattern in Fig. 9. Older parents tend to be somewhat less likely to move, although the coefficient on the age of the mother is not often statistically significant. Migration is significantly less likely from a larger state after 1940, although the effect is rather small in magnitude. Families with more children are less likely to move, although interestingly this effect is offset by a greater tendency for families with children under two to migrate. In most years, families with foreign-born fathers are less likely to move, but these coefficients are not consistently statistically significant. The probit results indicate that families with a black father were less likely to move in almost every census year. The difference is both statistically and economically significant. In 1950 and 1960, for example, near the height of the African-American migration, black families in the sample were actually about 5 percentage points less likely to move than whites, relative to an average migration rate on the order of 13%. This contrasts with the finding above (Fig. 5) that black adults were overall more likely than whites to have left their birth state at mid-century. The discrepancy is explained by the fact that the black migrants

314

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

during these years were disproportionately single, childless, and/or single parents, as compared with the white migrants. By 1990, the racial gap in migration propensities had closed completely, a change that may reflect convergence in migration behavior or possibly changes in sample composition.32

4.4. Increased Educational Attainment and the Rise of Migration Rates The IPUMS samples allow us to include an additional covariate for the father’s literacy for the years 1850 through 1920, and the father’s educational attainment for 1940 on.33 Education has a strong and statistically significant positive effect on migration propensity. Figure 15 shows the effects of literacy and educational dummy variables when they are added to the demographic variables considered above. The effect of literacy is generally rather small, but it is positive and significantly different from zero by 1900. From 1940 on, educational attainment has a large and significant positive effect. The strength of the education variable suggests that rising educational attainment over the course of the twentieth century might help explain rising rates of internal migration. Unfortunately, the census only began asking about educational attainment in 1940. Still, it is possible to use the estimated effect of education in

Fig. 15. Marginal Effect of Father’s Literacy or Educational Attainment on 5-Year Migration Propensity (Relative to Illiterate or 0–4 Years of Schooling). Source: IPUMS samples (Ruggles et al., 1997).

Decline and Rise of Interstate Migration

315

one of the census cross-sections to generate a counterfactual estimate of the impact of education using aggregate measures of educational attainment. Specifically, we make use of Goldin’s (1999, Table CG, A.11) series on high-school graduation rates to predict the propensity to migrate. We first estimate a model of migration, including as regressors the above demographic variables and a dummy variable for the father having 12 or more years of schooling. We then use the probit coefficients from one of the years and the mean values of the regressors and Goldin’s high-school graduation rate in each year to predict the time path of the migration rate. To the extent that this counterfactual tracks the increase in actual migration rates, we can conclude that rising education is a plausible explanation of the trend. The calculation is necessarily imprecise. First, census responses may overstate actual educational attainment, so the measure of schooling that we use in our estimates is subject to measurement error (Goldin, 1998).34 Second, using the means of the regressors to predict probabilities from the probit is not strictly correct, because of the non-linearity of the probit function. Third, we have had to estimate the stock of high-school graduates for synthetic cohorts in the prime-age population from Goldin’s “flow” series of high-school graduates among 17-year-olds. To do this, we have simply assumed that the percentage of high-school graduates in any given cohort remains constant, and calculated an unweighted average of the percentage of graduates among 27, 37, and 47-year olds each decade.35 Finally, the use of high-school graduation as the sole measure of educational attainment fails to capture the large impact of additional schooling on migration propensity at all schooling levels, which can be seen in Fig. 15. Undoubtedly more sophisticated estimates could be developed, but our purpose here is to see whether the rough magnitude of the education effect is large enough to have made a significant contribution to the rise of migration rates since 1900. We ran the counterfactual using probit coefficients from 1940, 1960, and 1990. All three tell a similar story. The results using the 1960 coefficients are presented in Fig. 16. The heavy series with diamond markers is the actual average 5-year migration rate from our samples. The series labeled “Mean HS (a)” is the counterfactual that uses the 1960 probit coefficients, sample means of the demographic variables each year, and the percent high school graduates based on the Goldin series for each year. The series labeled “Mean HS (b)” coincides with Mean HS (a) before 1940 and then replaces the Goldin high-school series with the reported high-school graduate rates from our samples after 1940. Table 6 reports the values of the two high-school variables used in the counterfactuals. Although the decade-to-decade changes differ between the actual and counterfactual series, it is clear that the counterfactuals account quite well for the

316

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

Fig. 16. Effect of Father’s High-School Education on Predicted 5-Year Move Probabilities, using 1960 Probit Coeffs. and Mean Characteristics. Source: IPUMS samples (Ruggles, 1997), Goldin (1999).

Table 6. High School Graduation Rates Used in Counterfactuals (%). Year

Mean HS (a)

Mean HS (b)

1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990

2.0 2.0 2.0 2.0 2.3 2.7 4.0 6.0 10.3 18.0 32.0 46.3 60.0 68.7 72.7

2.0 2.0 2.0 2.0 2.3 2.7 4.0 6.0 10.3 21.9 40.4 52.7 65.6 79.7 87.8

Source: Goldin (1999), Table CG. A.11, IPUMS samples (Ruggles et al., 1997).

Decline and Rise of Interstate Migration

317

basic upward trend in migration propensity over the past hundred years. And it is education doing the work here. The counterfactual series labeled “1920 HS” holds the high-school graduate rate fixed at its 1920 level, while allowing all the other regressors to vary over time. This counterfactual, which excludes the effect of changing educational attainment, exhibits a slight upward trend, but leaves most of the increase unaccounted for. The counterfactuals show that the association between education and migration is strong enough to account for essentially the entire increase in migration propensities of families with children over the twentieth century. Identifying the exact mechanisms whereby schooling might have encouraged or facilitated mobility are beyond the scope of this paper, but it is not hard to think of some. Educated workers may be better informed of distant opportunities elsewhere, they may possess general skills that are more readily transferred to new places or labor markets, and they may be less capital-constrained, which would matter if moving involves a large fixed cost.36 But correlation need not imply causation. Educational attainment could also be a proxy for omitted variables, such as occupation or income. Occupation is available in the census sample for every year, but we have not included occupational controls here because occupation is likely to be endogenous to interstate moves. An important area for future research is to investigate the underlying mechanisms whereby education was associated with the increasing geographical mobility of Americans.

5. CONCLUSIONS The IPUMS data show considerable promise for describing and analyzing internal migration over the past century and a half. In this paper we have presented two alternative measures of interstate migration derived from information on current residence and place of birth: one for individuals, based on the individual’s state of birth, and the other for families with young children, based on the child’s state of birth. Both measures suggest that for the country as a whole, migration propensities followed a broad U-shaped trend after 1850, falling until around the turn of the century and then rising gradually over much of the twentieth century. The migration measure based on moves by families with young children takes advantage of census information to pinpoint the timing of interstate moves with much greater accuracy than they can be by examining the state of birth of adults. Still, the measure has some disadvantages. Because we have restricted our sample to two-parent families, we do not capture the behavior of single parents or childless adults. The technique could easily be extended to the migration of single parents, but cannot be used for the childless. This appears to pose a particular

318

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

problem in examining the migration behavior of African-Americans during the Great Migration, for it would appear that these migrants were disproportionately childless. On the other hand, because so much of the Great Migration took place after 1940, the state-of-birth variables employed here can be augmented by direct migration questions from the census for those without children. Probit analysis of five-year migration rates derived from the family data suggests that rising educational attainment offers a potential explanation of the upward trend of migration propensity after 1900, although it could also be proxying for other factors. In future work with these data, we plan to explore the role of education in increasing mobility, geographic patterns of gross and net interstate migration, and the role of economic incentives in the migration decision. In particular, we are interested in the elasticity of migration with respect to locational differences in wages or incomes. This question is central to understanding the role of migration in the geographical integration of labor markets.

NOTES 1. See Shryock (1965, ch. 1). 2. The Census survival approach calculates net migration for a state or region as the difference between the actual change in population between successive censuses, and the predicted change, based on national survival rates for each age group within the population. The state-of-birth/state-of-residence approach looks at changes in the numbers living outside the state in which they were born between censuses. 3. Some specialized studies use linked census samples to follow individuals over a decade or two (see Ferrie, 1997, undated; Schaefer, 1989; Stewart, 2003). 4. The technique of using children’s birthplaces has been employed on a smaller scale in a number of other studies. See Schafer (1927), Atack and Bateman (1987, p. 79); Ferrie (1999, pp. 63–64). 5. This conclusion is not entirely novel. See Lee (1961, p. 79). 6. See also data on the percentage of the population residing in their state of birth reported in Historical Statistics of the United States (U.S. Census Bureau 1975, series C1). These data show no trend from 1870 to 1950, then decline slightly. Based on these data one would conclude that rates of migration were roughly stable until the mid-twentieth century and then increased slightly. Based on the underlying individual data, however, it would appear that the aging of the population in the twentieth century offset the decline in age-adjusted migration rates in the first part of the century, and then contributed to the apparent increase after 1950. 7. Eldridge used two kinds of evidence to trace population movements. The first was net migration based on census survival methods. In this approach age specific survival rates for the nation were applied to population figures for each region to predict the surviving population at the next census date. Differences between actual and predicted population were then used to calculate a net migration figure. The second approach used state of

Decline and Rise of Interstate Migration

319

birth and state of residence data to calculate gross flows of population between states. But because Eldridge did not have access to the underlying micro-data she was unable to control for differences in the age structure of the population at different dates. 8. Greater dispersion of opportunities may also attract job seekers who are willing to search long enough to find a high-wage offer (David, 1974). 9. See Treyz et al. (1993) and Blanchard and Katz (1992). The theoretical claim that unemployment rates should affect migration decisions was made by Todaro (1969), and has been explored using historical data by Hatton and Williamson (1992). 10. For reviews of the evidence on the importance of the stock of fellow migrants from the same source location in explaining migration flows (so-called chain migration), see also Greenwood (1975) and Rosenbloom (1994). The presence of fellow migrants may reduce migration costs by providing credit or housing and may increase expected benefits of moving by providing information on work opportunities and actual contacts or referrals with local employers. 11. See Long (1973) and Schwartz (1976). More educated individuals are also assumed to be more likely to engage in international migration. Recent work by Chiquiar and Hanson (2002), for example, finds that the typical Mexican immigrant to the United States is better-educated than the average Mexican who stays in Mexico. 12. Our analysis excludes date from the censuses of 1890, 1930, and 2000. Data from the 1890 census are not included in the IPUMS data set, because the original manuscript schedules of the 1890 census were destroyed by fire. At the time we conducted this analysis 1930 data were still subject to the 72-year census confidentiality rules, and the results of the 2000 census had not yet been released. 13. We use the full IPUMS general samples available for each year, with the following special cases: for 1850 and 1860, the samples of the free population; for 1900, the preliminary new sample drawn by researchers at the University of Minnesota; for 1970 the Form 2 State sample, for 1980, subsamples 0–19 of the 5% sample (thus generating a random 1% sample of the population); for 1990, the 1% unweighted state sample. Choice of samples in 1970 and 1980 was based on obtaining the most complete information about state of birth and state of residence. 14. These problems have been recognized for some time. An early exploration is Ross and Truxal (1931). 15. Ferrie (undated) found that three-fourths of all moves between 1850 and 1860 were between places within the same state. 16. The 1950 census asked where individuals resided one year before the census, but only the so-called sample line individuals, a subset of the full sample. We have not used this variable here. 17. For example, the figure of 6.5% for 30–39 year olds in 1960 is equal to the proportion of 30–39 year olds who were living outside their birth state in 1960 minus the percentage of 20–29 year olds who were living outside their birth state in 1950. This number can be negative as it is for some age groups in 1970, either because of sampling variation or potentially net return migration to the state of birth. 18. We cannot obtain these numbers for 1960 because the data do not report the state of residence in 1955, only whether that state was different from the state of residence in 1960. 19. Familiar tales of the “Okie” migration notwithstanding, interstate migration rates were low overall during the Depression.

320

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

20. Cohort age profiles are plotted in an appendix available from the authors. 21. Use of the 30–39 age group here is a compromise. For younger individuals, the rate of leaving the birth state is a more reliable indicator of recent migration, but their migration decision was likely made by someone else (parents). An individual who is 30–39 at the time of the census has been an adult for at least 10 years, and Figs 1 and 3 suggest that most individuals who will ever leave their birth state are likely to have done so by the time they are 39. 22. If all states had similar shapes, then the average distance to the nearest border would be proportional to the dimensions of the state, or the square root of the area. A more sophisticated measure would need to take account of different shapes and population concentrations within states. 23. We exclude a plot of the age variable, which only captures the effect of changing age between 30 and 39 and is thus of limited interest. 24. 1920 was chosen simply because it stands at the middle of our full span of years. 25. Steckel (1983) also used children’s birth states to examine the timing of moves. 26. In future research we will attempt to make use of the information on migration provided by multiple children. 27. The results presented here are unweighted; a check for 1940 suggests that the results are quite similar when we apply census household weights. 28. If births away from home were the principal cause of discrepancies between state of birth and current residence, the incidence of this discrepancy would not increase significantly with the age of the reference child. 29. It may be less representative for specific population sub-groups, such as AfricanAmericans (see below). 30. We can also examine the extent to which the birth state of children four – five years old and the state of residence of the father five years prior were the same. This percentage was 96.5 in 1940, 92.4 in 1970, 92.0 in 1980, and 93.1 in 1990. It cannot be derived for 1960 given the variables in the census sample. 31. Nativity and race of the mother tend to be highly collinear with these characteristics of the father, and added little explanatory power to the probit, so they were excluded in the estimates reported here. 32. Between 1970 and 1990, the percentage of black families with children headed by a single parent increased substantially. Consequently, our sample of two-parent black families represents a declining proportion of all black families, and cannot be assumed to be representative of the characteristics and migration behavior of all black families. 33. We have also run the probits including the mother’s literacy and education, with qualitatively similar results. 34. In addition, years of schooling fail to capture the quality of education, which varied by region, race, and census year. 35. For example, the percentage of 17-year-olds who graduated from high school was 6 in 1900, 9 in 1910, and 16 in 1920. Thus we calculate the percentage of 27, 37, and 47 year olds who were high-school graduates in 1930 as (16 + 9 + 6)/3 = 10.3%. These three ages are roughly consistent with the ages of most fathers in our family sample. This procedure simply ignores the diluting effect of immigrants who arrived after age 17 and were presumably less likely to be high-school educated than those who were children in the United States. 36. For further discussion see Chiswick (2000).

Decline and Rise of Interstate Migration

321

ACKNOWLEDGMENTS For helpful suggestions and comments on earlier versions, the authors thank Marigee Bacolod, Deborah Garvey, Kris Mitchener, two anonymous referees, and participants in the Economic History Association session of the ASSA meetings, New Orleans, January 2001, the meeting of the NBER Development of the American Economy group, Cambridge, MA, March 2001, and the All-U. C. Group in Economic History Conference, Irvine, CA, November 2002.

REFERENCES Atack, J., & Bateman, F. (1987). To their own soil: Agriculture in the antebellum north. Ames: Iowa State University Press. Barro, R. J., & Sala-i-Martin, X. (1991). Convergence across states and regions. Brookings Papers on Economic Activity, 1, 107–182. Blanchard, O. J., & Katz, L. F. (1992). Regional evolutions. Brookings Papers on Economic Activity, 1, 1–75. Borjas, G. J., Bronars, S. G., & Trejo, S. J. (1992). Self-selection and internal migration in the United States. Journal of Urban Economics, 32, 159–185. Carrington, W. J., Detragiache, E., & Vishwanath, T. (1996). Migration with endogenous moving costs. American Economic Review, 86, 909–930. Chiquiar, D., & Hanson, G. H. (2002). International migration, self-selection, and the distribution of wages: Evidence from Mexico and the United States. NBER Working Paper No. 9242. Chiswick, B. R. (2000). Are immigrants favorably self-selected? An economic analysis. In: C. D. Brettell & J. F. Hollifield (Eds), Migration Theory: Talking Across Disciplines. New York: Routledge. Collins, W. J. (1997). When the tide turned: Immigration and the delay of the great black migration. Journal of Economic History, 57, 607–632. DaVanzo, J. (1978). Does unemployment affect migration?–Evidence from micro data. Review of Economics and Statistics, 60, 504–514. David, P. A. (1974). Fortune, risk, and the microeconomics of migration. In: P. A. David & M. W. Reder (Eds), Nations and Households in Economic Growth: Essays in Honor of Moses Abramovitz (pp. 21–88). New York: Academic Press. Eldridge, H. T. (1964). Demographic analyses. In: H. T. Eldridge & D. S. Thomas (Eds), Population Redistribution and Economic Growth: United States, 1870–1950 (Vol. III, Part one). Demographic Analyses and Interrelations. Philadelphia: American Philosophical Society. Ferrie, J. P. (1997). Migration to the frontier in mid-nineteenth century America: A re-examination of Turner’s safety valve. Working Paper. Ferrie, J. P. (1999). Yankees now: Immigrants in the antebellum U.S., 1840–1860. Oxford: Oxford University Press. Ferrie, J. P. (undated). How ya gonna keep ’em down on the farm [when they’ve seen Schenectady]?: Rural-to-urban migration in 19th century America, 1850–1870. Working Paper. Fishback, P. V. (1998). Operation of ‘unfettered’ labor markets: Exit and voice in American labor markets at the turn of the century. Journal of Economic Literature, 36, 722–765.

322

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

Fishback, P. V., Horrace, W. C., & Kantor, S. (2001). Do federal programs affect internal migration? The impact of new deal expenditures on mobility during the Great Depression. NBER Working Paper No. 8283. Gallaway, L., & Vedder, R. K. (1971). Mobility of native Americans. Journal of Economic History, 31, 613–649. Goldin, C. (1998). America’s graduation from high school: The evolution and spread of secondary schooling in the twentieth century. Journal of Economic History, 58, 345–374. Goldin, C. (1999). A brief history of education in the United States. NBER Working Paper series on historical factors in long run growth, Historical Paper No. 119. Greenwood, M. J. (1975). Research on internal migration in the United States: A survey. Journal of Economic Literature, 13, 397–433. Greenwood, M. J. (1997). Internal migration in developed countries. In: M. R. Rosenzweig & O. Stark (Eds), Handbook of Population and Family Economics (Vol. 1B, pp. 647–720). Amsterdam: Elsevier. Hatton, T. J., & Williamson, J. G. (1992). What explains wage gaps between farm and city? Exploring the Todaro model with American evidence, 1890–1941. Economic Development and Cultural Change, 40, 267–294. Kim, S. (1998). Economic integration and convergence: U.S. regions, 1840–1987. Journal of Economic History, 58, 659–683. Lee, E. S. (1961). The Turner thesis reexamined. American Quarterly, 13, 77–83. Long, L. H. (1973). Migration differentials by education and occupation: Trends and variations. Demography, 10, 243–258. Mitchener, K. J., & McLean, I. W. (1999). U.S. regional growth and convergence, 1880–1980. Journal of Economic History, 59, 1016–1042. Rosenbloom, J. L. (1990). One market or many? Labor market integration in the late nineteenth-century United States. Journal of Economic History, 50, 85–107. Rosenbloom, J. L. (1994). searching for workers: U.S. labor markets after the Civil War. Social Science History, 18, 377–403. Rosenbloom, J. L. (1996). Was there a national labor market at the end of the nineteenth century? New evidence on earnings in manufacturing. Journal of Economic History, 56, 626–656. Ross, F. A., & Truxal, A. G. (1931). Primary and secondary aspects of interstate migration. American Journal of Sociology, 37, 435–444. Ruggles, S., & Sobek, M. et al. (1997). Integrated public use microdata series: Version 2.0. Minneapolis: Historical Census Projects, University of Minnesota. Sandefur, G. D., & Scott, W. J. (1981). A dynamic analysis of migration: An assessment of the effects of age, family and career variables. Demography, 18, 355–368. Schaefer, D. F. (1989). Locational choice in the antebellum South. Journal of Economic History, 49, 145–165. Schafer, J. (1927). Four Wisconsin counties. Madison: State Historical Society of Wisconsin. Schwartz, A. (1976). Migration, age, and education. Journal of Political Economy, 84, 701–720. Shryock, H. S., Jr. (1965). Population mobility within the United States. Chicago: Community and Family Study Center, University of Chicago. Steckel, R. H. (1983). The economic foundations of East-West migration during the 19th century. Explorations in Economic History, 20, 14–36. Stewart, J. I. (2003). Migration to the agricultural frontier and economic mobility, 1860–1880. Paper presented at 42nd Cliometrics Conference, North Carolina State University. Todaro, M. P. (1969). A model of labor migration and urban unemployment in less developed countries. American Economic Review, 59, 138–148.

Decline and Rise of Interstate Migration

323

Treyz, G. I., Rickman, D. S., Hunt, G. L., & Greenwood, M. J. (1993). The dynamics of U.S. internal migration. Review of Economics and Statistics, 75, 209–214. U.S. Bureau of the Census (1975). Historical statistics of the United States: Colonial times to 1970. Washington: Government Printing Office. Wright, G. (1986). Old South, new South: Revolutions in the Southern economy since the Civil War. New York: Basic Books.

APPENDIX Table A.1. Coefficients from Migration Probits for 30–39 year Olds with Known State of Birth. Variable Age SE Female SE Black SE Midwest SE South SE West SE Sqrt(land area)/100 SE

Age SE Female SE Black SE Midwest SE South SE West SE Sqrt(land area)/100 SE

1850

1860

1870

1880

1900

1910

1920

0.0075 0.0012 −0.0541 0.0072 −0.1492 0.0200 0.0754 0.0134 0.1949 0.0079 −0.3812 0.0296 −0.0123 0.0044

0.0019 0.0011 −0.0645 0.0063 −0.1325 0.0191 0.0787 0.0098 0.0992 0.0071 −0.3793 0.0184 0.0021 0.0039

0.0078 0.0009 −0.0476 0.0053 −0.1082 0.0077 0.0736 0.0077 0.0526 0.0066 −0.2411 0.0306 −0.0078 0.0034

0.0042 0.0008 −0.0440 0.0044 −0.0809 0.0067 0.0877 0.0062 0.0318 0.0058 −0.1499 0.0283 −0.0127 0.0027

0.0055 0.0008 −0.0372 0.0046 −0.0477 0.0077 0.1463 0.0064 0.0652 0.0068 0.0371 0.0195 −0.0248 0.0024

0.0033 0.0008 −0.0370 0.0046 −0.0135 0.0077 0.1617 0.0065 0.0537 0.0069 0.0403 0.0163 −0.0136 0.0021

0.0036 0.0005 −0.0238 0.0027 0.0120 0.0046 0.1380 0.0038 0.0678 0.0041 0.1007 0.0089 −0.0126 0.0011

1940

1950

1960

1970

1980

1990

0.0049 0.0004 −0.0026 0.0022 0.0655 0.0039 0.1140 0.0032 0.1015 0.0033 0.1893 0.0064 −0.0153 0.0009

0.0012 0.0004 −0.0061 0.0021 0.1096 0.0040 0.1025 0.0031 0.1130 0.0031 0.1775 0.0054 −0.0134 0.0008

0.0038 0.0004 −0.0028 0.0020 0.0694 0.0036 0.0889 0.0029 0.1469 0.0030 0.1149 0.0048 −0.0180 0.0008

0.0006 0.0004 −0.0128 0.0021 0.0305 0.0036 0.0638 0.0032 0.1203 0.0031 0.0987 0.0049 −0.0216 0.0008

0.0049 0.0003 0.0017 0.0018 0.0325 0.0032 0.0294 0.0027 0.0414 0.0027 0.0685 0.0039 −0.0267 0.0007

0.0032 0.0003 0.0040 0.0016 −0.0508 0.0026 0.0160 0.0024 0.0186 0.0025 0.0442 0.0033 −0.0248 0.0006

Notes: Dependent variable: Living in state different from birth state. Coefficients expressed as marginal probability (dF/dX) evaluated at sample means. Source: IPUMS samples (Ruggles et al., 1997).

324

JOSHUA L. ROSENBLOOM AND WILLIAM A. SUNDSTROM

Table A.2. Coefficients from Migration Probits for Families with 4–5 Year Old Reference Child. Variable Child age SE Mother’s age SE Father’s age SE Father black SE Father for-born SE Number of children SE Sibling under 2 SE Midwest SE South SE West SE Sqrt(land area)/100 SE

Child age SE Mother’s age SE Father’s age SE Father black SE Father for-born SE Number of children SE Sibling under 2 SE Midwest SE South

1850

1860

1870

1880

1900

1910

1920

0.0158 0.0102 0.0019 0.0011 −0.0008 0.0009 −0.0644 0.0243 −0.0049 0.0155 −0.0066 0.0031 0.0030 0.0115 0.0457 0.0157 0.0914 0.0157

−0.0121 0.0064

0.0221 0.0243 0.0004 0.0010 −0.0013 0.0008 −0.0748 0.0253 −0.0184 0.0099 −0.0061 0.0029 0.0059 0.0099 0.0167 0.0114 0.0536 0.0132 −0.0468 0.0343 −0.0064 0.0047

0.0087 0.0070 −0.0003 0.0007 −0.0005 0.0006 −0.0554 0.0088 −0.0158 0.0078 −0.0016 0.0022 0.0181 0.0085 0.0182 0.0095 0.0059 0.0103 −0.0381 0.0198 −0.0043 0.0037

0.0126 0.0054 −0.0006 0.0006 −0.0003 0.0005 −0.0524 0.0064 −0.0106 0.0062 −0.0079 0.0018 −0.0002 0.0063 0.0352 0.0079 0.0114 0.0086 0.0497 0.0249 −0.0041 0.0027

0.0046 0.0059 −0.0009 0.0006 −0.0001 0.0005 −0.0283 0.0085 −0.0011 0.0075 −0.0037 0.0020 −0.0174 0.0068 0.0367 0.0093 0.0214 0.0105 0.0382 0.0225 −0.0024 0.0026

0.0266 0.0071 −0.0011 0.0008 0.0000 0.0007 −0.0556 0.0097 0.0003 0.0093 −0.0097 0.0024 0.0119 0.0091 0.0347 0.0109 0.0129 0.0116 0.0354 0.0213 0.0008 0.0030

0.0091 0.0043 0.0000 0.0005 −0.0010 0.0004 −0.0152 0.0077 0.0025 0.0057 −0.0104 0.0015 −0.0042 0.0054 0.0331 0.0067 0.0384 0.0076 0.0630 0.0132 0.0006 0.0017

1940

1950

1960

1970

1980

1990

0.0047 0.0034 −0.0008 0.0004 0.0002 0.0003 −0.0207 0.0054 −0.0240 0.0051 −0.0094 0.0012 −0.0029 0.0049 0.0416 0.0063 0.0551

0.0210 0.0034 −0.0003 0.0004 −0.0029 0.0004 −0.0533 0.0051 −0.0162 0.0075 −0.0219 0.0012 0.0349 0.0049 0.0306 0.0055 0.0760

0.0201 0.0038 −0.0013 0.0005 −0.0033 0.0004 −0.0529 0.0057 0.0098 0.0106 −0.0102 0.0015 0.0151 0.0050 0.0307 0.0060 0.0910

0.0130 0.0041 0.0001 0.0005 −0.0038 0.0005 −0.0525 0.0061 −0.0071 0.0092 −0.0064 0.0016 0.0092 0.0056 0.0168 0.0064 0.0794

0.0133 0.0044 0.0011 0.0006 −0.0018 0.0005 −0.0391 0.0072 −0.0105 0.0081 −0.0078 0.0023 0.0206 0.0063 0.0091 0.0071 0.0440

0.0068 0.0042 0.0001 0.0006 −0.0020 0.0005 0.0017 0.0080 −0.0211 0.0065 −0.0043 0.0022 0.0158 0.0060 0.0075 0.0070 0.0601

Decline and Rise of Interstate Migration

325

Table A.2. (Continued )

SE West SE Sqrt(land area)/100 SE

1940

1950

1960

1970

1980

1990

0.0063 0.0851 0.0128 −0.0031 0.0013

0.0057 0.1358 0.0101 −0.0096 0.0013

0.0066 0.0852 0.0095 −0.0073 0.0013

0.0068 0.1259 0.0106 −0.0134 0.0015

0.0072 0.0950 0.0104 −0.0138 0.0016

0.0071 0.0569 0.0090 −0.0117 0.0014

Notes: Dependent variable: Reference child living in state different from birth state. Coefficients expressed as marginal (dF/dX) evaluated at sample. means. Source: IPUMS samples (Ruggles et al., 1997).