Migration and Inequality in Germany 1870-1913 0199276560, 9780199276561, 9781423770879

Grant's study is a rigorous analysis of migration in Germany within the demographic and socio-economic contexts of

405 37 5MB

English Pages 416 [415] Year 2005

Report DMCA / Copyright


Polecaj historie

Migration and Inequality in Germany 1870-1913 
 0199276560, 9780199276561, 9781423770879

Table of contents :
Contents......Page 8
1. Imperial Germany as an Example of Industrialization under Labour Surplus Conditions......Page 10
2. Sources of Inequality in Rural Germany......Page 31
3. The Pattern of Migration, 1870–1913......Page 65
4. Migration in Germany 1870–1913: A Statistical Analysis......Page 106
5. Demography and Migration......Page 133
6. Migration, Farm Size, and the Condition of the Agricultural Labourer......Page 190
7. Agricultural Productivity, Labour Surplus, and Migration......Page 224
8. Migration and Urban Labour Markets......Page 262
9. Industrialization, Migration, and Inequality......Page 302
10. Challenging the Kehrite View of Imperial Germany......Page 339
References and Sources......Page 370
A......Page 398
B......Page 399
C......Page 400
D......Page 401
E......Page 402
F......Page 403
I......Page 404
J......Page 406
L......Page 407
M......Page 408
P......Page 410
R......Page 411
S......Page 412
T......Page 413
W......Page 414
Z......Page 415

Citation preview

OXFORD HISTORICAL MONOGRAPHS editors j. maddicott j. harris j. robertson

r. j. w. evans b. ward-perkins r. service p. a. slack

This page intentionally left blank

Migration and Inequality in Germany 1870–1913




Great Clarendon Street, Oxford ox2 6dp Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With oYces in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York ß Oliver Grant 2005 The moral rights of the author have been asserted Database right Oxford University Press (maker) First published 2005. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover and you must impose this same condition on any acquirer British Library Cataloguing in Publication Data Data available Library of Congress Cataloging in Publication Data Data available Typeset by SPI Publisher Services, Pondicherry, India Printed in Great Britain on acid-free paper by Biddles Ltd. King’s Lynn, Norfolk. ISBN 0-19-927656-0 1 3 5 7 9 10 8 6 4 2

PREFACE This study has its origins in a D.Phil. thesis, ‘Internal Migration in Germany 1870–1913’, which was motivated by the view that migration is the central problem of an industrializing society. The initial stimulus came from a visit to China in the spring of 1995, a visit which reawoke my interest in the process of economic growth and its associated problems and opportunities. This combined with an older fascination with the course of German history to produce the thesis which has now evolved into this book. Comparison is one of the most powerful tools at the disposal of a historian, and one of the themes of this book is that historians should not be timid or reticent about this. They should be prepared to make broad comparisons across time and across continents. They should insist that theirs is a discipline which can contribute to the study of major issues of current concern: the causes of war and peace, economic growth and the creation of modern industrial societies, the foundations of democracy and the establishment of individual rights. If historians turn their backs on these issues, then others will Wll this gap. But, while historians can bring to these subjects the additional perspective which is gained by historical analysis, they should also be aware of the researches on contemporary societies which social scientists are making, and consider the ways that these can bring new perspectives to bear on historical issues. This means confronting the methods that social scientists bring to bear on these issues, in particular modelbuilding and the use of statistical analysis. Both will be used in the chapters that follow. The way that the thesis has evolved into this book has been inXuenced by the experience of designing and teaching a postgraduate course on comparative industrialization at Oxford University, entitled ‘Industrialization in Europe, North America and East Asia since 1700’. I would like to thank Professor Avner OVer for having encouraged me to do this. I also thank those students who have taken the course for many interesting discussions on the problems of industrializing societies. When the problems of Imperial Germany are set in this wider context, it soon becomes apparent that, to a large extent, this was a successful and progressive society, which dealt with the problems of rapid



industrialization unusually well. It was in many respects the Wrst ‘Modell Deutschland’. I owe thanks to many people who have helped directly and indirectly over the past three years. I have beneWted from exemplary supervision from Professors Charles Feinstein and Hartmut Pogge von Strandmann. I have much enjoyed our conversations. I have learnt much from their observations and suggestions. I had the additional good fortune to have had two excellent examiners in Professors Niall Ferguson and Albrecht Ritschl. This work has been considerably improved as a result of their comments. Individual chapters have been read by Anthony Atkinson, Avner OVer, Hans-Joachim Voth, Liam Brunt, James Foreman-Peck, Tom Nicholas, and Eric Grimmer-Solem. I thank all these for their time and attention. I owe much to the library of the London School of Economics. This has a large collection of German statistical materials on open access. I am particularly grateful that the German statistical collection has remained on open access during the rebuilding of the main library. I would like to express my thanks to the staV of the Nuneham Courtenay depository of the Bodleian Library, to Chris Newbould and John Slater. The depository is not well suited to the arrival of visitors with access permits, who sit in corridors and cause obstructions. Their tolerance was much appreciated. I owe thanks to my wife and children, the arrival of the latter having coincided with the beginning of this study. They have provided much diversion and, at times, a necessary sense of proportion. But my Wnal thanks are owed to my farm staV, to Paul Burns, Russell Donovan, and Alan Roberts, who have taken on additional responsibilities while I have been working on this study. It is beyond question that, without their eVorts, this would not have been possible. O.W.G. 18 June 2004

CONTENTS 1. Imperial Germany as an Example of Industrialization under Labour Surplus Conditions


2. Sources of Inequality in Rural Germany


3. The Pattern of Migration, 1870–1913


4. Migration in Germany 1870–1913: A Statistical Analysis


5. Demography and Migration


6. Migration, Farm Size, and the Condition of the Agricultural Labourer


7. Agricultural Productivity, Labour Surplus, and Migration


8. Migration and Urban Labour Markets


9. Industrialization, Migration, and Inequality


10. Challenging the Kehrite View of Imperial Germany


References and Sources




This page intentionally left blank


Imperial Germany as an Example of Industrialization under Labour Surplus Conditions 1. I n t r o d u c t i o n: ‘A h o use d iv ided’ —M i g r a t i o n a n d I nequa lity in I m p e r i a l G e r m a n y , 1870–1913 ‘A house divided cannot stand’ is Abraham Lincoln’s pre´cis of the issues facing the United States in the Civil War; it can also serve as a summary of the views of a school of German historians, the Kehrite school. The key concept of this school is the ‘primacy of internal politics’: the view that the internal weaknesses of pre-1914 Germany led to a series of expedients which made German foreign policy a disruptive force in pre-1914 European diplomacy, and which prevented the settlement of outstanding disputes by any means short of a general European conXict. The core of this policy was the building of a Xeet large enough to challenge British naval hegemony.1 What were the sources of the internal weakness of Imperial Germany? Historians have produced a list of suggested Xaws in the polity and social structure of Imperial Germany: a series of ‘skirted decisions’ which explain Germany’s failure to make political progress commensurate with its economic advances.2 In this analysis, the fundamental assumption is 1 Kehr’s own work, originally written in the 1930s, is available in English, Kehr (1975) and (1977). Other important works in the Kehrite tradition include Wehler (1969) and (1985), Berghahn (1971), Stegemann (1977), and the collection in Stu¨rmer (1976). The existence of a Kehrite school has been denied by Puhle (1978) and there is some disagreement over what the key Kehrite concept really is; for Eley (1986) it is either ‘the primacy of pre-industrial traditions’, p. 42, or the concept of Sammlungspolitik, p. 111. For the purposes of this study, the key Kehrite concept is the ‘primacy of internal politics’: meaning by this, that the Kaiserreich suVered from internal problems, which had an adverse eVect on foreign policy to an extent which is unusual compared to other states. This last point is rarely made explicitly, but it is important if the concept is to have any genuine force. For a simple statement of this view, see Ro¨hl (1969), 10. 2 The phrase ‘skirted decisions’ is taken from the title of an essay in Mommsen (1995). Mommsen’s work stresses the political crisis produced by the inability of the Bismarckian


Industrialization under Labour Surplus Conditions

that there were options which, if taken, would have alleviated the social strains and enabled Germany to become a normal, modern, democratic industrial state. And the implication of this is that such a state would have been a more stable and predictable element of the pre-1914 diplomatic order. These ‘skirted decisions’ were not just political and diplomatic: they were derived from a fundamental unwillingness of the ‘pre-industrial’ elites of Imperial Germany to modernize structures and institutions which had their roots in the traditions of an agrarian society.3 This produced a bureaucracy recruited from a narrow and unrepresentative social base, a political system which limited democratic control over the executive, and a social order in which large numbers felt themselves to be excluded or marginalized. In short, an outcome of ‘incomplete modernization’, derived in part from the Bismarckian origins of the Kaiserreich in a ‘revolution from above’. Historians outside the group normally identiWed as Kehrite, or Kehrinspired, have accepted parts of this argument, or put it forward in diVerent guises. The broader concept of a Sammlungspolitik, policies designed to rally ‘loyal’ forces in support of the Imperial government, has replaced Kehr’s narrower emphasis on the trade-oV of agricultural protection for battleship construction. Modern studies place greater stress on social and cultural divisions as the ultimate sources of the weakness of the political structure: the rise of extremist ideas, the isolation of important social strata, and the domination of political parties by economic interest groups. Instead of a conscious ‘Flucht nach Vorn’, an attempt to solve internal problems by foreign policy adventures, more recent studies have set German foreign policy in the context of a society where Social Darwinism had a ‘special resonance’, where ‘chauvinistic bile’ had a ‘broad appeal’, where paciWsm was ‘relatively weak’ and ‘its opposite was correspondingly powerful’. Under these circumstances, the ‘restless, querulous’ nature of German foreign policy, the ‘fatalistic brinkmanship’ of the civilian leadership in the constitution to adapt to changing economic and social conditions, and argues that ‘in July 1914 the ‘‘latent crisis’’ of the German political system spilled over onto the international stage, setting in train a world war’, Mommsen (1995), 204. This can be regarded as a variation on the basic Kehrite theme. 3 According to Wehler (1985), 236, Bismarck, and the ‘Wilhelmine polycracy’ that followed him, were determined to avoid the spread of democracy and moves towards equal rights. The militancy of the property-owning elites, and their determination to preserve their privileges at all costs, is stressed in, for example, Puhle (1976), 341.

Industrialization under Labour Surplus Conditions


crisis of July 1914, are seen, implicitly or explicitly, as expressions of deeper Xaws.4 This study puts forward an alternative hypothesis: that many of the internal stresses and strains of Imperial Germany were the inevitable consequences of industrialization, and that there was little that German political leaders could have done about this. Even if the political elites had been prepared to make concessions this would not have produced an immediate shift to a more harmonious society: the option of a ‘social market economy’ was not available in pre-1914 circumstances. Instead, the maturing of the industrial economy was a necessary precondition for the attainment of social stability and political maturity. The point that the transition from an agrarian to an industrial economy inevitably creates social and political strains is not a new one. Other societies had also experienced these tensions. But the process of industrialization in Germany was diVerent from that in the pioneer industrializing countries because Germany had the advantages, and the disadvantages, of ‘relative backwardness’, as originally pointed out by Alexander Gerschenkron.5 Germany could make use of advanced industrial technology developed elsewhere. This meant that the pace of German industrialization was faster. It meant that the gap between productivity in the mechanized industrial sectors and the traditional sectors (agriculture and handcraft) was much greater. Transfers of labour between these sectors was a major contribution to overall growth. But the same transfers brought about social and institutional disruption. Pre-1914 Germany was the Wrst economy to experience the telescoped process of industrial development typical of those countries which make their way to the front rank of industrial powers having started from a position of relative backwardness. Many historians have noted the rapidity of the process of social and economic transformation in Imperial Germany. But the full implications of this have not been examined.6 One important consequence is that the analysis of German economy in the period needs to make use of concepts and tools taken 4 The quoted phrases are all from Blackbourn (1997), 426, 433, 446, 452, and 459. This is not a work in the Kehrite tradition as it is normally understood, but it is indicative of the way that Kehrite ideas have penetrated the historical mainstream. Works which discuss foreign policy issues in isolation from the domestic context are rare. 5 Gerschenkron (1962); the main pioneer industrializing country was, of course, Great Britain, but France also experienced a prolonged period of relatively slow industrial development, as emphasized by Keyder and O’Brien (1978). 6 One study which has made use of insights from development economics in the analysis of inequality trends in nineteenth-century Germany is Dumke (1991).


Industrialization under Labour Surplus Conditions

from the Weld of development economics. The concept of development economics as a separate branch of economics rests on the idea that ‘relatively backward’ countries face diVerent problems and opportunities to those which arise in more mature economies, or in societies which have a more drawn out period of transition. Much of the analysis which will be presented in this study makes use of ideas whose origins lie in development economics: the ‘Kuznets Curve’ relating income inequality to industrialization; the ‘Lewis Model’ of development under labour surplus conditions. Many societies have found industrialization to be a disruptive process, and the path to maturity has often been a rocky one even for those societies which have successfully made the transition. The problems which Germany faced have reoccurred in other contexts. The historical debate over the pace of political progress relative to economic development in Imperial Germany has a modern counterpart in arguments over the role of authoritarian governments, or ‘Asian values’, in the success of the East Asian ‘newly industrializing countries’ (or NICs). Developmental dictatorships such as the Park regime in South Korea or the Kuomintang government in Taiwan have presided over the initial stages of rapid industrialization, and political liberalization has come later, when the most disruptive phase has been completed.7 It has been argued, by the World Bank amongst others, that an ‘insulated bureaucracy’ is an important component of a successful longterm growth strategy.8 If so, then the alleged backwardness of the German political system may have been advantageous in economic terms: the state was able to focus on long-term goals rather than being obliged, by the pressures of democratic politics, to resort to short-term expedients which had damaging long-term consequences.9 There are now a number of examples of societies which have had authoritarian governments during the early stages of industrialization, and which have successfully managed the transition to democracy once this phase is past. Besides the Korean and Taiwanese cases just cited, there are other examples: Chile post-Pinochet, Spain after Franco, the 7 The role of authoritarian states in economic development in East Asia has been examined in a number of works; see for example Amsden (1989), Amsden (2001), and Wade (1990). 8 World Bank (1994), 167. 9 Many Third World countries have a problem of ‘urban bias’ as a result of policies which were intended to improve urban conditions, but which then create privileged urban elites.

Industrialization under Labour Surplus Conditions


re-establishment of democracy in Brazil following military rule in the 1960s. Why were these societies able to make this transition when Germany could not? It cannot, surely, be argued that social and political tensions were any greater in Germany in the 1880s and 1890s than in Chile at the time of the coup against Allende, or Spain just before the outbreak of the civil war. There were, undeniably, particular factors in the case of each of these societies which helped to cause a reduction in the potential for social and political conXict as industrialization proceeded. This study does not propose a general theory of politics and industrial development. But one particular factor is that the process of transition was not derailed by any major external catastrophe. These countries did not face genuine strategic dilemmas which in turn impinged on domestic politics. They were not required to work out solutions to internal problems of transition at the same time as they dealt with foreign policy issues, which were themselves moving towards a crisis for independent reasons. This alternative view of pre-1914 Germany stresses the essential autonomy of the foreign policy problems which beset the German government. In this ‘anti-Kehrite’ formulation, ‘revolutions from above’ are commonplace; the processes of economic and political advance do not go hand in hand but follow diVerent trajectories, so that ‘incomplete modernization’ is the normal state for an industrializing society; Imperial Germany was not moving towards an internally generated catastrophe, but was a society with as good a chance of achieving full economic maturity, social modernization, and political democratization as any other. The decisive factor in the equation, which shifted German history onto a diVerent course, was the outbreak of war in 1914.10

2. Mi g r a t i o n a n d I n e q u a l i t y i n t h e D e v e l o p m en t P r o c e s s Migration has a key role in industrialization, and it is a principal cause of the social dislocation which accompanies the early stages of economic growth in modern industrial societies. This will inevitably be a period of high migration. It is the opportunity to shift labour out of low productivity sectors such as agriculture which makes it possible to 10 Arguments about the outbreak of war are summarized in Pogge von Strandmann (1988), see also the collection edited by Scho¨llgen (1990).

Industrialization under Labour Surplus Conditions


achieve high rates of growth in underdeveloped economies. Indeed, one of the deWning features which distinguishes ‘mature’ from ‘developing’ economies is that, typically, economic growth in the latter is accompanied by large-scale migration from the countryside into cities, while in the former it is not. One view which provides a simple analysis of the eVects of migration out of agriculture is the model of economic development Wrst set out by Arthur Lewis in 1955: ‘a model of development with unlimited supplies of labour’, generally known to development economists as the Lewis Model.11 In his article, Lewis analysed the eVects of industrial development in the context of a labour market where there is surplus labour, as commonly found in countries in the early stages of industrialization. The model predicts that, so long as capitalists can draw on these supplies of surplus labour, then the fruits of industrial progress will go largely to the holders of capital, in the form of rising proWts, and there will be little increase in the level of real wages. Given that the owners of capital are likely to be predominantly rich, and few in number, it can be seen that this prediction will lead to rising inequality in cities and other industrial areas, with a consequent increase in social divisions, and, also, to a feeling amongst industrial workers that they are not gaining a fair reward for their labours. For students of German history the attraction of the Lewis Model is that it provides a possible explanation both of the rapid growth of the German economy 1870–1914 and of the tensions that accompanied this. Moreover it also suggests a reason for one of these problems: the rise of the socialist vote. Lewis named the two phases he described ‘classical’ (when there is a labour surplus and wages are held down) and ‘neoclassical’ (when the labour surplus is exhausted and wages start to rise). But one might note the similarity between his model and that of Karl Marx and call the Wrst phase ‘Marxist’ or ‘quasi-Marxist’. According to the model, this is a period when the capitalist economy behaves in a way which seems to bear out some of Marx’s predictions: proWts rise as a percentage of the national income; wages lag behind productivity increases.12 11

Lewis (1954). For a modern discussion see Hayami (1997), 72–5 and 170–2. Lewis did not assume, as Marx did, that technology was always labour-saving. In a purely Marxist model capital accumulation leads to increased competition, a fall in the rate of proWt, and increasingly severe economic crises, so the ultimate level of the share of proWts in national income is not clearly determined. However, Kolakowski has pointed out that Marx’s views on whether capitalism led to the absolute or the relative impoverishment of the working class are also ‘by no means unequivocal’, Kolakowski (1978), i. 289. 12

Industrialization under Labour Surplus Conditions


If the economy behaved as Lewis predicted, then the experience of workers competing in urban labour markets with apparently unlimited supplies of migrants from rural areas would be a dispiriting one. The formation of trade unions could lead to an improvement in workers’ bargaining strength, but unions would have to be cautious about pressing claims for higher wages given the ability of employers to bring in large numbers of migrants with no previous experience of trade union activity. Trade unions would be more eVective where the workers they represented possessed skills which could not easily be taught to newly arrived migrants, as was the case for printers and other members of the so-called ‘labour aristocracy’. These would be easier to organize. In this context, the appeal of radical Marxist ideas would be understandable, together with a pessimistic view of the likely beneWts from non-revolutionary trade union activity. But the diVerence between Lewis’s model and Marx’s is that Lewis’s labour surplus phrase comes to an end eventually. When this happens, urban labour markets are no longer dominated by the Xow of surplus labour out of agriculture, and employers who need to attract labour to new industries are obliged to oVer superior wage levels to those obtained by fully employed workers elsewhere. Trade unions can bargain from a position of strength and union membership can oVer beneWts to unskilled as well as skilled workers. The economy enters a new phase in which wages rise in line with productivity and the proWt share stabilizes or may even decline. The operation of the Lewis Model is illustrated in Figure 1.1. The introduction of new technology causes a rise in productivity in the advanced or industrial sectors. But with unlimited labour supplies, the wage level is Wxed. Consequently, the gains from productivity increases go exclusively to the owners of capital, producing a rise in the proWts share. It is only when the labour surplus is exhausted—a moment sometimes called the ‘turning point’ in the development economics literature—that wages can rise above the Xoor set by the ready availability of surplus labour in the traditional or rural sectors.13 Once this has occurred the economy shifts to a new phase of development, represented by the neo-classical model of mainstream economics, where there are limitations on the supply of labour as well as capital. In this phase, both capital and labour share in the fruits of economic development. In mature capitalist economies real wages have tended to rise at least as fast as the growth of productivity, if not somewhat faster. This 13

The term ‘turning point’ was coined by Fei and Ranis (1964).


Industrialization under Labour Surplus Conditions

Productivity Profits Wages

Labour surplus model

Neo-classical model Turning point

Fig. 1.1. The pure Lewis Model: changing factor shares during industrialization

has meant that the share of wages in national output has been stable or tending to increase.14 The implication of the Lewis Model is that this historical moment, the ‘turning point’ as the economy moves out of the labour surplus phase, represents a new opportunity to achieve political and social stabilization. There are now reasons for workers to favour ‘normal’ trade union activity, aiming to secure steady improvements in wages and other working conditions, over more radical alternatives. ‘Revisionist’ views should gain ground. Provided that the political system oVers a reasonable chance of similar progress, leading to increased participation in political decisionmaking, and that the whole system is not destabilized by some major catastrophe, a society which has passed the ‘turning point’ of the Lewis Model has a much better prospect of achieving full political, social, and economic maturity—becoming a ‘normal’ modern, liberal, capitalist democracy—than one which is still in the labour surplus phase. The Lewis Model in this pure form has a clear and abrupt transition from the labour surplus to the neo-classical model. In practice, economic development does not divide quite so easily into phases. 14 The Wgure shows the eVect on factor shares if wages rise in line with productivity. This is not an automatic consequence of the transition to a ‘neo-classical’ economy, although it is broadly consistent with the experience of most OECD countries in the twentieth century.

Industrialization under Labour Surplus Conditions


Moreover, there is generally some rise in wages during the early stages of industrialization, even if the rise is not as fast as the increase in productivity. Initially, industry can rely on nearby sources of labour. But once this ‘local labour surplus’ is exhausted, wages have to be raised to attract migrants from regions which are more remote from the main industrial areas. This period of slow wage increases, which lag behind rising productivity, can last until all sources of surplus labour are used up. Then wages start to rise in line with productivity. Figure 1.2 illustrates this ‘modiWed Lewis Model’. In the early stages of industrialization the beneWts still go predominantly to the owners of capital, even though there is some increase in wages. The transition to the neo-classical model is a more gradual one. Instead of a sudden ‘turning point’ there is a slower transition to the neo-classical model. What evidence is there that wages lagged behind productivity? The actual record of German wage and productivity growth is set out in Figure 1.3. There are some caveats to note. The Wgures used for these calculations are taken from Walter HoVmann’s Das Wachstum der deutschen Wirtschaft. The wage series used is derived from a combination of diVerent sources, and this means that it must be treated with some caution.15

Productivity Profits Wages

Labour surplus model

Neo-classical model

Fig. 1.2. The modified Lewis Model: changing factor shares during industrialization 15 More reliable wage data becomes available from the 1870s and 1880s onwards, Bry (1960), Desai (1968), and Hohls (1995). However, the onset of German industrialization is


Industrialization under Labour Surplus Conditions





Real wages

100 90 80 1850











Fig. 1.3. Indices of productivity and real wages in the advanced sectors of the German economy (industry, mining, and transport), 1850–1913 Log scale; base is average for the years 1850–9 (¼100) Source: Calculated from Hoffmann (1965). The series are averages for the three sectors calculated from Hoffmann’s figures for employment, net value added, and wages. The real wage figures are produced using Hoffmann’s deflator for private consumption.

The second caveat is that wages may lag behind productivity for other reasons. If the terms of trade move against the industrial sector, then factor returns will not keep pace with productivity.16 In the period after 1895, the German export sector had to gain market share from Britain, and this led to downwards pressure on industrial prices at a time when world agricultural prices were recovering from the low point generally dated to the 1850s, and Figure 1.3 shows that the gap between the two series opened mainly in the 1860s and early 1870s. It would also be interesting to compare real wages and the return on capital in the industrial sector, but HoVmann’s proWt Wgures have come in for heavy criticism, Fremdling (1991). 16 The diVerence is that, with adverse movements of the terms of trade, factor returns in the industrial sector are depressed in general, so both proWts and wages are held down, but under the Lewis Model only wages are held back.

Industrialization under Labour Surplus Conditions


of the mid-1890s. This shift in the terms of trade meant that the impressive productivity performance of the German industrial sector in the years just before the First World War was not fully reXected in real wage gains. Figure 1.3 shows that, according to HoVmann’s Wgures, real wages did not keep pace with productivity in the modern sectors. Between 1850/9 and 1904/13 productivity rose by 202%; real wages rose by 119%. The rate of increase of productivity was 70% higher than the rate of increase of real wages. The Wgure shows that the gap opened substantially in the 1860s and early 1870s. It then closed in the 1880s before widening again in the 1890s and 1900s. The fact that so much of the widening of the gap occurred in the 1860s, which is a period for which there are few reliable wage statistics, means that these results need to be interpreted with caution. What can be said is that these are Wgures which are consistent with the modiWed version of the Lewis Model as shown in Figure 1.2. Wages did not keep up with productivity; most of the beneWts of industrialization went to the owners of capital.

3. T o w a r d s a M o r e G e n e r a l T h e o r y o f I n e q u a l i t y a n d I n d u s t r i a l i z a t i o n: T h e W e b e r –L e w i s –K u z n e t s H y p o t h e s i s The Lewis Model is not a complete model of inequality and industrialization: it lacks several important components. It does not explain how the labour surplus is created, and how this phase comes to an end. It provides a possible explanation of rising urban inequality, but to explain the general movement of inequality in an industrializing society it is necessary to widen the scope of the enquiry to include consideration of rural inequality and the relationship between the two sectors. Lewis’s view of the labour surplus is essentially a static one: lessdeveloped countries are overpopulated relative to economic opportunities; there is underemployment both in agriculture and in an informal urban sector.17 His views are, in some respects, very similar to those of Max Weber, writing about conditions in east-Elbian agriculture in the 17

In societies which do not have comprehensive unemployment insurance there is rarely much formal unemployment. Labour surplus conditions are more likely to Wnd expression in high levels of underemployment.


Industrialization under Labour Surplus Conditions

late nineteenth century. Lewis wrote: ‘in over populated countries the code of ethical behaviour so shapes itself that it becomes good form for each person to oVer as much employment as he can’.18 Weber held that ‘the old economic order asks: ‘‘how can I give, on this piece of land, work and sustenance to the greatest number’’ ’.19 Both these statements describe a system of values which gives rise to behaviour which departs from the proWt-maximizing norm. Lewis did not try to explain why this changed, but Weber did. The most important factor was the changing attitudes and values of the land-owners. These were obliged by their desire to maintain their position in society to match the consumption patterns of the rising urban bourgeoisie, and this in turn led to the adoption of entrepreneurial, proWt-maximizing behaviour. Under capitalism, entrepreneurial land-owners would employ the cheapest labour, use the best technology, and prefer market relationships to paternalistic systems of mutual obligation. Thus, while the ‘old economic order’ provided employment for surplus labour, this surplus was released when capitalist values came to predominate. In Weber’s view, therefore, it was the permeation of capitalist values throughout society which led to the creation and release of the rural labour surplus. This was an inevitable consequence of industrialization. While Weber’s views on the exact mechanism which brings about social change in rural societies during industrialization can be disputed, his basic point that there are institutional arrangements in traditional agriculture whose erosion during industrialization can create a period of heavy migration into urban labour markets is more in tune with modern analysis. This could arise for a number of reasons. Traditional marriage patterns may have held back demographic increase. There may be institutionalized underemployment in agriculture, supported by systems favouring revenue sharing or informal arrangements for mutual insurance in hard times. Institutions which were appropriate to an isolated agricultural sector with limited access to urban markets may hold back the use of new technology: improved transportation may bring modern inputs and a consequent increase in productivity, making it possible to 18

Lewis (1954), 142. The quotation is from a lecture Weber gave in the United States, Weber (1948), 367. Weber’s major work on east-Elbian agriculture formed part of a survey by the Verein fu¨r Sozialpolitik, Weber (1893). Other writings on this subject include an article which appeared in 1894, Weber (1894, trans. 1979) and various speeches and addresses, which can be found in Weber (1993). 19

Industrialization under Labour Surplus Conditions


shift large quantities of labour into the industrial sector while maintaining or increasing food production.20 A more general view of the relationship between industrialization and inequality in all sectors of society was provided by Simon Kuznets, writing, like Lewis, in the 1950s.21 He was inXuenced by the work of the Soviet economist Prokopovitch, who had used nineteenth-century Prussian tax data to show a rise in inequality during industrialization. As a result of Prokopovitch’s work, Kuznets was also aware of a gap between rural and urban inequality in nineteenth-century Prussia. He found a similar pattern in studies of inequality in the United States and India.22 This led him to a general statement of the relationship between inequality and economic development. His hypothesis was that the shifting balance between a relatively egalitarian, but poor, rural sector, and a relatively unequal, but richer, urban sector would produce a curve, an ‘inverted U’, as inequality Wrst rises with industrialization, and then declines (Figure 1.4). Migration has an important role in the mechanism which, according to Kuznets, produced his curve. It was migration between the two sectors which produced the shifting sectoral balance. Moreover, this eVect would be intensiWed if the initial eVect of industrialization was to raise the size of the rural–urban wage gap. Modern studies have shown that the movement of the rural–urban gap is an important factor in the overall income distribution: large gaps between rural and urban earnings are associated with high levels of overall inequality.23 Poorer rural workers and peasants account for a disproportionate number of the total numbers on low incomes even in advanced countries, so a worsening of the relative position of agriculture will tend to worsen the overall distribution of income. This factor in itself could explain part of the increase in inequality predicted by Kuznets. If modern industrial centres are to grow rapidly they will need to attract labour from rural areas. Initially this may come from nearby regions, where there may be surplus labour. But, when this surplus is exhausted, sources of labour must be found in further-oV 20

The view that many of the traditional institutions of agriculture in underdeveloped countries are eVective solutions to the problems of remoteness, uncertain yields, and limited access to markets is put forward in Dasgupta (1993). 21 Kuznets (1955). 22 Ibid. 7; the German Wgures came from Prokopovitch (1926). 23 Bourguignon and Morrison (1998), Bourgignon (1990); Basu (1997) discusses the theory behind this.


Industrialization under Labour Surplus Conditions


Gross Domestic Product per capita

Fig. 1.4. The Kuznets Curve: an inverted U-shaped relationship between inequality and economic growth during the early stages of industrialization

regions. This may require a substantial increase in the gap between rural and urban earnings to create the necessary incentives for would-be migrants. Although Lewis and Kuznets had diVerent mechanisms in mind, there are similarities between their views. The downward phase of the Kuznets Curve is brought about by two factors: the declining importance of the poorer rural sector in the overall distribution of income (which reduces the importance of the gap between rural and urban incomes as a contribution to general inequality) and a tendency for inequality within the urban sector to start high and fall over time. This last point can also be derived from the Lewis Model. Putting the views of these three authors together produces a general theory of the relationship between economic development and inequality: industrialization and the spread of modern values and relationships unleashes a process of social change in agriculture, which in turn leads to the transfer of surplus labour to the urban sector; labour surplus conditions produce an unequal distribution of income in the urban sector; the increased importance of the unequal urban sector leads to a general rise in inequality; the ending of the labour surplus period creates conditions which will eventually lead to a fall in inequality.

Industrialization under Labour Surplus Conditions


4. C r i t i c i s m o f t h e L e w i s Mo d e l a n d Al t e r n a t i v e V i e w s The Lewis Model has been widely used in development economics, and it is still one of the mainstays of this branch of economic analysis, but it has had its critics. 1. In the development economics literature the aspect of the Lewis Model which has attracted most criticism is the view that agriculture has only a passive role in economic development, and does not progress technologically in the same way as the urban or industrial sector.24 One of the most serious charges against the Lewis Model is that it has contributed to the tendency of modern developing countries to underinvest in agriculture. 2. Lewis makes a restrictive assumption about savings: ‘only capitalists save’. While this assumption may be justiWed for some developing countries where banks and other savings institutions are not well established, the experience of other industrializing countries shows that, given the right institutional framework, even people who are not particularly well oV will save, and that these savings can be an important component of total funds available for investment. A more realistic assumption might be that ‘only capitalists invest in industry’: meaning that, in the early stages of industrialization, only a minority have the specialized knowledge of the new technology to be able to take advantage of the opportunities it oVers. This, in turn, points to the importance of institutional development, in making it possible for the savings of those who lack technical knowledge to be channelled into industrial investments: the German banking system performed such a function. 3. Lewis has been accused of overemphasizing the importance of agriculture as a source of surplus labour. Surveys of modern developing economies show that the main area of underemployment is in fact the informal urban sector. Whether this also applies to nineteenth-century Germany is another matter. The phenomenon of high rates of urbanization, in countries which are still in the early stages of industrialization, is one which is found in post-1945 developing countries and not in nineteenth-century Europe. But an informal urban sector did exist: Tagelo¨hner, workers without permanent employment dependent on occasional day-labouring jobs, were a signiWcant part of the urban labour market. 24 Industrialization and urban growth produces change in agriculture in nearby regions, as was pointed out by Von Thu¨nen (1826).


Industrialization under Labour Surplus Conditions

4. In addition, there has been criticism of the concept of the labour surplus itself. When, in 1979, Lewis was awarded the Nobel prize for economics, it was shared with the Chicago agricultural economist Theodore Schultz. Both had made a range of important contributions to development economics, but on this subject they were in fundamental disagreement: Schultz denied the existence of a labour surplus in traditional agriculture, certainly in Lewis’s sense of a large body of labour with zero marginal productivity.25 The essence of Schultz’s critique was, Wrst, to deny that resource allocation was ineYcient in traditional agriculture, or that there was a preference for idleness or a bias against thrift. From this it followed that labour would be fully occupied to the point where marginal cost was equal to marginal product. Secondly, he made use of empirical studies to show that underemployment was not prevalent, and also demonstrated, using the eVects of the 1918–19 inXuenza epidemic in India, that an exogenous removal of agricultural labour did aVect food production.26 In Schultz’s view peasants were ‘poor but eYcient’.27 If investment was low it was because the return on capital was also low, or the discount rate was high. It was expensive for small subsistence farmers to search for innovations; this, rather than resistance to change, explained low technical progress. Peasant ineYciency was a myth imposed by Western agricultural experts who failed to understand the seasonal nature of much work in traditional agriculture. Although Lewis may well be the better known, or at least the more cited, author, Schultz’s critique has been highly inXuential. The view that peasants Wnd eYcient solutions to the problems they face is now a commonplace. Moreover, Schultz’s stress on the importance of the introduction of new factors—new productive knowledge, better education, human capital in the broadest sense—appears prescient in the light of subsequent developments in growth theory. Nineteenth-century German agriculture oVers an excellent example of the introduction of new, non-traditional factors exactly as described by Schultz: new scientiWc knowledge, heavy investments in education, and specialist training in agricultural sciences. But before setting up Lewis and Schultz as polar opposites it should be noted that in some respects they are not so far apart. Schultz thought 25

Possible deWnitions of the labour surplus are discussed in Reynolds (1975), 11. Major wars have also led to the sudden removal of large numbers from the labour market; Prest (1948) gives some interesting examples of cases where this led to a fall of production in less-developed economies, and some others where it did not. 27 Schultz (1964), 40. 26

Industrialization under Labour Surplus Conditions


that the introduction of new factors could produce a relatively ‘cheap’ rise in productivity: the transformation of traditional agriculture could be quite rapid. Now there is a clear distinction between Lewis’s view of a static rural labour surplus, which pre-dates industrialization, and Schultz’s more dynamic version, but the practical implications are similar: quite large amounts of labour can be released from agriculture while food production is maintained or increased.28 Moreover, it is worth considering what the source of these new inputs might be. Government investments in agricultural science and education in nineteenth-century Germany were not the whole story. Industrialization brought new agricultural inputs: basic slag from steelmaking; ammonium sulphate from gas production. Industrial progress stimulated scientiWc research as well as beneWting from it, and this created possibilities for agricultural advances in turn. Industrialization raised the demand for agricultural products, some of which, such as sugar beet, also required improved technical knowledge. Industrial growth led to improved communications, bringing to rural society better access to markets and to productive knowledge. Institutional developments, such as improved banking systems, also aVected agriculture. So, the transformation of traditional agriculture was, at least in part, a consequence of industrialization. Although Schultz and Weber came from diVerent traditions and were writing in diVerent centuries, there is a similarity in their views in that the creation of an agricultural labour surplus is a dynamic process, which comes about as a result of industrialization and the eVect that this has on agriculture. Weber’s primary stress was on the changes in social attitudes but he also considered the eVect of changing technology, in particular the introduction of sugar beet. This increased the seasonality of agricultural work, leading to the employment of seasonal migrant labour, which in turn brought about a weakening of social ties.

5. T h e L e w i s M o d e l a n d t h e H i s t o r i o g r a p h y of Imperial Germany Historical studies are mostly bounded in scope and in time, and this one is no exception: July 1914 makes a convenient date for the ending 28 Ibid. 24–5. This eVect is greatly strengthened if, as is often the case, industrialization is accompanied by increased population growth, an issue which will be examined in Chapter 5.


Industrialization under Labour Surplus Conditions

of a study of conditions inside Imperial Germany. The year 1913 is the last one for which a full set of data on peacetime conditions is available. But the temptation to see the events of 1914–18, not just the war but the revolutionary outbreaks which accompanied its ending, as the culmination of forces and developments already under way within the social structure of the Kaiserreich needs to be resisted. The sense that the narrative structure needs a natural full-stop, not an accidental one, pervades much that has been written about Imperial Germany. The massive disruption of German political and social development which followed is felt to need deep-seated explanations, not just the impact of chance events. Hence the appeal of Sonderweg arguments which trace the evolution of Xaws rooted in the character of the Bismarckian state and the social or cultural background to its political travails. The current consensus is not one to accept the Kehrite view of the absolute primacy of internal politics in foreign policy without reservation. Instead the view which prevails is one which stresses the interaction of external and internal politics: the way that interest group politics hemmed and hampered policy-making in both Welds; the constricting eVect of unresolved constitutional problems on government action; the eVect of a drift to radicalism and chauvinism on the tenor of debate and the ability of political actors to see issues as resolvable, given goodwill and a willingness to engage in rational analysis, rather than unbridgeable divisions, calling for recourse to apocalyptical solutions. The predominant view sees the intertwined problems of the Kaiserreich as a Gordian knot which, by implication, only a radical discontinuity could solve. Historians who take a positive and optimistic view, who stress the real prospects for peaceful evolution without violent confrontation, have had a poor reception from their colleagues.29 The value of economic models to historians lies in their ability to draw together diVerent pieces of evidence and present a coherent picture of social or economic evolution. The common features of the experience of many societies engaged in the process of industrialization has naturally excited the interest of theorists and led to the construction of theories which seek to explain these consistencies. While some historians may distrust these attempts to impose a common structure on historical events, the alternative approach, a history without theory, leads to the particularization of every historical event, and an undue


See, for example, the reaction to Rauh (1973) and (1977).

Industrialization under Labour Surplus Conditions


stress on the unique features of every society, ignoring many points of similarity. Migration is a particularly good example of this tendency to see events without considering the broader context. It is a subject which has rightly stimulated historical analysis. The movement of individuals or families from one social and economic context to a diVerent one is a dramatic historical event which brings about the disruption of traditional relations, and leads to the creation of new ones. But it also forms part of a larger picture, playing a role in the determination of wage levels and other factor returns in the economy as a whole. This in turn has an eVect on attitudes and motivations at the level of the individual. For a society in the upwards phase of the Kuznets Curve, a drift to radicalism is not inevitable, but the context is one which is, at the very least, conducive to alienation and a sense of social exclusion. The divergent paths of factor returns predicted by the Lewis Model, when translated into the actual experiences of workers exposed to the movements of urban labour markets under labour surplus conditions, meant that a generation of workers had poor living conditions, low life expectancy, and insecure jobs and saw only slow improvements in these circumstances, despite rapid economic growth. But these were not German peculiarities; they were the result of general forces unleashed by the introduction of modern technology. And as Germany began to climb out of the trap created by the labour surplus phase of economic development, conditions in German cities did improve, albeit slowly. This in turn created a more favourable background for social reconciliation and political progress. The Lewis Model suggests that the ending of the labour surplus phase may well represent an opportunity for a wider process of political and social modernization.

6. St a n d a r d s o f C o m p a r i s o n f o r t h e H i st or io g r a p h y o f t h e Kaiserreich A major concern of historians has been the extent to which Imperial Germany was an unusual state or society in the context of late nineteenth-century Europe. The view that there was something exceptional about Germany takes diVerent forms: Wolfgang Mommsen has placed the emphasis on Xaws in the political system; for Hans-Ulrich Wehler it is the Janusgesicht of incomplete modernization, which raises broader


Industrialization under Labour Surplus Conditions

questions of social and economic development. But one feature of these studies is that the standard of comparison is taken from the experience of the states of Western Europe: Wehler has written about choices between ‘Modernisierung und Tradition . . . okzidentalen Gemeinsamkeiten und deutschen Sonderkeiten’.30 The implication of the Weber–Lewis–Kuznets hypothesis is that this comparison may not be appropriate. In the pioneer industrial states the process of industrialization was more drawn out, the erosion of preindustrial, or pre-capitalist, values and institutions was slower.31 There may have been a labour surplus period, when labour supplies were abundant due to population growth or changes within agriculture, but this was less acute and spread over a longer period. The growth of cities was not so rapid. Patterns of inequality changed slowly, if at all. Political and social institutions could evolve in situations where there was less pressure for sudden change.32 In the British case, the existence of labour surplus conditions has been detected in the operation of the labour market in the period up to 1850.33 This was the result of a relatively drawn-out process: comparatively slow industrial growth combined with demographic pressure and the Wnal stages of the transition to large-scale agriculture which had started in the sixteenth century. But the German experience was a diVerent one. Although there had been pockets of proto-industrial activity in the eighteenth century, the onset of rapid economic growth was delayed until the ‘triple uniWcation’ of the mid-nineteenth century: economic uniWcation with the creation of the Zollverein in 1834; political and monetary uniWcation with the founding of the German Empire in 1871; and, crucially for a country with limited access to sea transport, the creation of a uniWed market as a result of the expansion of the railway network in the 1850s and 1860s. 30 Modernisation or Tradition . . . Convergence to Western Norms or German Exceptionalism, Wehler (1995), 1252; the use of the adjective okzidentalen obscures which Western state is being used as the standard of comparison. Since the strictures of Blackbourn and Eley (1984) appeared, German historians of the Sonderweg persuasion have been less speciWc on this point, although it is often Britain that is the implied model. 31 In the British case, the switch to large-scale capitalist agriculture came well before the onset of industrialization. 32 The main focus of comparison is obviously Britain, but Belgium and the Netherlands were also advanced industrial or commercial societies by 1830. Recent scholarship has also shown that the process of French industrial growth was a relatively drawn-out one. 33 As argued in Feinstein (1998), 651.

Industrialization under Labour Surplus Conditions


German economic development was, like that other of Continental countries, ‘jerked’ forwards by the application of British techology.34 In the German case, the jerk was made even more abrupt by the overcoming of barriers to internal market integration which had hitherto held back economic development. The result was that German industrialization from 1850 onwards was unusually fast. The process was telescoped in a way which was new in the experience of other European countries. This created problems of adjustment and adaptation which were unprecedented in their intensity. If so, then the standard of comparison for Imperial Germany should not be taken from Western Europe but, in the Wrst instance, from other late-industrializing European countries—Russia, Italy, and Spain, for example—and secondly, from other non-European countries which have gone through the process of industrialization. There is a large range of scholarship which has been devoted to the comparative study of developing economies, and the diVerent social, economic, and political problems associated with or resulting from the development process. It is the basic contention of this book that the tools developed within this Weld, the Weld of development studies, can be used to understand and elucidate the problems of Imperial Germany. This will, it is hoped, shed a new light on the diYculties, and the achievements, of the Kaiserreich. 34 Pollard (1981), 84. The transforming nature of the new British industrial technology is clear from the regional studies presented in chapter 3 of Pollard’s study.


Sources of Inequality in Rural Germany 1. I n t r o d u c t i o n The Kuznets Curve—the hypothesis that inequality will initially rise and then fall with industrialization—is predicated on the basis of a series of assumptions made by Kuznets in his original article. One of these is the assertion that, initially, inequality will be higher in the urban than the rural sector. However, consideration of the historical record, as well as the current position in developing countries, shows that there are a wide range of possible levels of inequality in the rural sector in underdeveloped or newly industrializing countries. Factors such as climate, crop specialization, the eYcacy of land reform, the legacy of colonialism or feudalism can produce very diVerent outcomes. The contemporary situation shows wide divergences. On the one hand there are the ‘plantation economies’ of central and southern America, which typically have large inequalities in land ownership and rural incomes; at the other extreme, the rural sectors of many newly industrializing East Asian countries are characterized by small-scale intensive rice production, with a consequent narrowing of disparities in income and wealth.1 Thus, a study of the evolution of inequality in an industrializing society needs to start with a consideration of the state of rural society: the allocation of property rights, the distribution of land holding and land ownership, the position of rural labourers. Even if the predictions of the Lewis Model concerning conditions in urban labour markets are correct, and this leads to high inequality in the urban sector, there are some societies in which the level of rural inequality is so high that industrialization produces little or no increase in the overall level of inequality.

1 Works which have stressed the importance of such geographical factors include Bray (1986) and Engerman and SokoloV (1997).

Sources of Inequality in Rural Germany


2. T h e A g r a r i a n R e f o r m Mo v e m e n t i n N i n e t e e n t h - Ce n t u r y G e r m a n y : G e n e r a l Co n s i d e r a t i o n s Historical analysis of the nineteenth-century agrarian reforms has tended to look for statistical indicators to judge the impact of the reforms: Wgures for peasant (or smallholder) land losses, information on the numbers of landless labourers and their incomes, the extent of peasant indebtedness, the overall structure of land holding before and after the reforms, the balance between rented land and land in owneroccupation. These are important issues, and it is right that they should be the main focus of research, but the use of statistical indicators is meaningless without ‘benchmarks’, standards by which to judge the diVerence between success and failure. Much of the modern historical analysis of the agrarian reforms has been inXuenced by the ‘incomplete’ or ‘absent’ revolution thesis of the Sonderweg school of German historians: the view that the deWning characteristic of German history has been a succession of ‘revolutions from above’, which necessarily lacked the capacity to achieve thoroughgoing social transformations of the type supposedly achieved in other countries which experienced ‘revolutions from below’. Implicitly, therefore, the standards of comparison for the agrarian reforms are the revolutionary land settlements brought about by the French or Russian revolutions (referring in the latter case to the pre-collectivization period). These are possible points of comparison, but not the only ones. There is now a large number of studies in the Weld of development economics on land reform and its problems. This has a number of lessons for historians. Perhaps the most important of these is the sheer diYculty of land reform: it is quite rare to Wnd examples of land reform programmes which have been even moderately successful. Agrarian reforms have too often been either ineVective or else led to results which were inimical to the development process.2 Land reform is a social and political process which places an unusual strain on the capacity of the political system to deliver an unbiased outcome: one which is not inXuenced either by the social composition of the bureaucracy or by the ability of special interest groups to aVect political decisions. It is a test of the ‘encompassing’ nature of the political process: the extent to which decision-makers put the long-term 2

Modern works that make this point include Low (1996) and Ghimure (2001).


Sources of Inequality in Rural Germany

interests of society, considered as a whole, ahead of considerations of short-term expediency (which militate against oVending powerful social groups) or the favouring of sectional interests.3 Moreover, it is a process which involves balancing a number of diVerent objectives. The history of land reform shows that there are many pitfalls. These include the following. (i) The eVect of expropriation on property rights Radical land reform without compensation can achieve a dramatic alteration of rural property rights. However, expropriation on such a scale will have adverse eVects on investment in other sectors: investors will wonder if similar violations of property rights are possible in other areas. This is one consideration which has led most governments to pay compensation to landlords and others adversely aVected by land reforms. The record of recent years shows that some compensation has nearly always been paid: the main exceptions being land reforms following the establishment of Communist governments which have been coupled with more general restrictions on private property rights. In the German case, compensation was paid by the peasants themselves, either in the form of monetary compensation or land given up to landlords, in order to secure the elimination of labour services and other onerous duties or restrictions. Arguably, the various German states could have provided more of these payments themselves, but the limited size of state budgets in the early nineteenth century probably precluded this option. (ii) The eVect on production The objectives of land reform are Wrst to reduce rural inequality, and secondly to promote increased agricultural productivity, the use of new technology, and rising food production. Otherwise, inadequate agricultural production may act as a brake on industrial development, producing problems such as the ‘scissors crisis’ of the Soviet economy in the NEP period.4 Land reform should not create a rural sector which is an obstacle to general economic development. It should not consist entirely of subsistence farmers unable to produce a marketable surplus. 3 The concepts of ‘encompassing’ and ‘sectional’ interest groups are taken from Olson (1982). 4 Nove (1969), 93–6.

Sources of Inequality in Rural Germany


The impressive productivity performance of the German agricultural sector in the second half of the nineteenth century showed that the reforms produced an eYcient market-orientated sector, which was well able to take advantage of the opportunities presented by expanding urban demand. (iii) The importance of rural credit A major theme which connects to this last point is the availability of rural credit. In cases where land reform has had an adverse eVect on production, the lack of credit has often been an important factor. Moreover, if peasants are obliged to pay compensation to former landlords, then some means of Wnance will be required for this. In cases where rural Wnance has been unobtainable or prohibitively expensive, there has often been a return to landlord control, as indebted peasants have been forced to sell land to local money-lenders or back to former landlords.5 This is probably the area where the Prussian reforms, in particular, were most deWcient. In the early nineteenth century, state-sponsored rural credit institutions, such as the Landschaften or rural mortgage banks, lent mainly to larger estates. Smaller farmers had limited access to credit, and this contributed to the extent of peasant land losses. By contrast, in areas such as the Kingdom of Saxony, where more eVort was made to give small and medium-sized farmers access to credit, the post-reform land distribution was more egalitarian. (iv) The importance of market access As the objective is to create an agricultural sector which is both egalitarian and also market-orientated, it follows that the costs of market access will be an important factor. For peasant farmers, with relatively small marketable surpluses, high transport costs and other obstacles can make the use of the market mechanism an unattractive proposition. The consequences are, Wrst, that peasant farmers revert to subsistence agriculture, which makes it diYcult to pay for modern inputs or maintain payments on debts, and secondly, that supplies of food and other materials from the rural sector are limited. 5

In the Philippines there have been 11 consecutive land reform programmes, but the lack of rural credit and other follow-up services has meant that land has tended to pass back into the hands of landlords and agricultural companies, Ghimure (2001), 13.


Sources of Inequality in Rural Germany

Inevitably, diVerent areas of Germany had very diVerent experiences. In the western regions the proximity of expanding urban markets made life easier for peasant farmers who could intensify livestock production or specialize in perishable high-value products for local markets. In the east transport costs were high, which meant that the main marketable crops would be cereals.6 This inevitably favoured the larger estates, which tended to be more orientated towards cereal production. But the elimination of internal tariVs and the creation of a single German market, together with the expansion of the railway network, did improve opportunities for farmers in some of the remoter areas.7 (v) The importance of contingency But even the best-designed land reform is still subject to a range of inXuences which are outside the control of the political authorities. Two of the most important are demographic pressure, and the movement of farm prices. Under circumstances where there is acute demographic pressure in rural areas, and the supply of fertile land is limited, it will be extremely diYcult to avoid a rise in the landless population, even with a highly egalitarian land distribution. The alternative is to permit the subdivision of holdings to an extent which would substantially reduce the marketable surplus, which would jeopardize other rural policy objectives. Another outside factor is the movement of farm prices. If land reform leaves the peasant sector with an increased level of indebtedness as a result both of compensation payments to landlords, and also investments consequent on the shift to a more market-orientated agricultural system, then the real burden of these debts will be raised if prices fall, and reduced if prices rise. Some notably successful land reform programmes, such as those of post-1945 Japan and eighteenth-century Denmark, have been considerably assisted by subsequent bursts of inXation which reduced the real value of peasant debts.

6 Which had a relatively high value-to-weight ratio; see Brinkmann (1935), 82–3 for the transport costs of various commodities; this uses German data collected by H. Settegast in the pre-1914 period. 7 East Prussia was an area which contained a large number of small or medium sized holdings, and which was able to expand livestock production substantially in the late nineteenth century, moving away from wool production into more intensive livestock systems.

Sources of Inequality in Rural Germany


Figure 2.1 illustrates this point by comparing the movement of agricultural prices in Prussia after 1816, and in Japan after 1946. The lower line gives the movement of rye prices in Prussia (1815 boundaries); superimposed on this is a second line giving the movement of rice prices in Japan in the years following the land reform introduced by the American occupation authorities in the years immediately after the Second World War. The Japanese reforms were mainly carried out in 1946–7. In the years which followed there was a burst of inXation which carried rice prices to levels two or three times those of the reform period. A more gradual increase in subsequent years took prices up to an index of 500 by the late 1960s. By contrast, the years which followed the Prussian reforms saw a fall in prices to levels which were around a third of the 1816–18 average by 600 400


100 80 60












Fig. 2.1. A comparison of agricultural price movements in Prussia (rye prices 1816–40) and Japan (rice prices 1946–69) Three-year moving averages as indices (base for Prussia is 1816–18, for Japan 1946–8). Japanese index is superimposed (1946 corresponds to 1816). Source: Japanese prices are from various issues of the Japanese statistical yearbook, 1950–70; the index is a chained index, using whatever was reported, so that the 1946–7 prices are wholesale prices (Tokyo), 1947–50 prices are black market prices (Tokyo), and the 1950–70 prices are producer prices (all Japan). The German index uses the average Prussian wholesale prices reported in ZdKPSB (1907), 84–92.


Sources of Inequality in Rural Germany

the mid-1820s. The reforms were carried out under circumstances which were considerably less favourable for peasant farmers than those in post-1945 Japan. The value to historians of the broader context provided by modern writings on land reform is that it emphasizes the point that agrarian policy needs to serve a range of objectives, and it must make a positive contribution to the general objective of social and economic development. Land reform cannot be judged according to a simple ‘black– white’ dichotomy, where a resulting egalitarian land distribution is ‘white’ and an inegalitarian one is ‘black’. These other objectives also need to be taken into account.

3. A C o m p a r i s o n o f A g r a r i a n R e f o r m s i n G e r m a n y and Denmark (i) The Danish experience The consideration of the Danish land reforms makes an interesting point of comparison for the later German reforms. Agrarian reform in Denmark was essentially an eighteenth-century phenomenon, which was largely completed by 1810. Most of the legislation was passed in the period 1768–81 (although the abolition of serfdom came earlier in 1701). There are a number of points of interest which can be compared and contrasted with the German experience.8 First, agrarian reform in Denmark was to a large extent a product of ‘enlightened’ state action. The state took the initiative and the reforms were designed to serve state interests. These included the improvement of state Wnances (which led the government to favour the peasantry who were not exempt from taxation over the nobility who were). There was also a conscious attempt to create a free independent peasantry; the abolition of serfdom in 1701, for example, was connected with a decision to move to a conscript army. Neither the French revolution nor industrialization had any part in the decision to introduce the reforms. Secondly, despite this the reforms did not operate against aristocratic interests in general. The nobility did give up their feudal rights but 8 The main sources used for this section are Baack (1977), Bjorn (1977), and Skrubbeltrang (1954); Tracey (1982) describes the evolution of Danish agriculture in the postreform period.

Sources of Inequality in Rural Germany


they were well compensated. Many estates chose to sell land to their tenants, and it was this which increased the proportion of freehold land held by peasants from the pre-reform level of under 10% to 40% in 1810. Between 1786 and 1807 226 out of 800 estates sold a majority of their land. However, the nobility did achieve an increase in rents on the land that remained in their possession, and the price of land rose by 93% between 1781–5 and 1796–1800. As land-owners, both nobility and peasantry gained from this increase. The main impetus to the reform came from the realization that improved agricultural techniques, coupled with the enclosure and consolidation of holdings, could bring about a large increase in output. Experiments by progressive landlords in the 1760s showed that these measures, which included the abolition of labour service and the granting of copyhold tenancies, could lead to a doubling of production (as occurred on the pioneering BernstorV estate between 1764 and 1783). The ‘increase in the cake’ was large enough to give everyone a share (grain production rose 40% 1770–1803). What the government did was to make sure that a substantial slice of the beneWts went to the peasantry. In other states, notably Britain, the gains from agricultural improvement went largely to landlords and large-scale tenant farmers. Thirdly, and this emphasizes the voluntary, co-operative nature of the reforms, some aspects were not the result of legislation. Thus the abolition of labour service or Hoveri was not decreed. Yet by 1807 this only applied to a quarter of Danish farmland, having previously been fairly universal. The reason was that it had become an ineYcient system which obliged the peasantry to keep large numbers of horses instead of cattle. So the reforms made it possible to raise cattle numbers, from 280,000 in 1774, to 560,000 in 1804 and 834,000 in 1837. At the same time the number of horses fell (from 400,000 in 1804 to 325,000 in 1837). Equally, the process of enclosure and consolidation was not regarded by the peasantry with hostility. Indeed, where they were freeholders or copyholders they often initiated such schemes themselves, and in areas where peasant freehold was common, enclosure was earlier.9 This points to the importance of the prior evolution of tenant rights. Where these had become hereditary and Wrmly entrenched, with legal support, the peasantry could bargain successfully with land-owners to secure enclosure and the commutation of feudal duties even if the state stood aside. 9

Bjorn (1977).


Sources of Inequality in Rural Germany

A fourth point is the importance of Wnance. Here the state did establish a Royal Credit Bank which made loans to peasant farmers for the purchase of freehold land or copyhold tenures. But the impact of this bank was not large compared to other sources of Wnance. It lent 442,000 Rigsdalers for this purpose in 1786–98 (compared to 1,058,000 for landlord improvements). In the overall Wnancing of peasant land purchase this was only a small share. In the 1790s it is thought that around 11⁄2 million Rigsdalers were lent by banks for this purpose and another 1 million by private investors. In 1797–1807 the Eukkase (or widow’s pension fund) lent 2,245,000 million for the purchase of freeholds or copyholds.10 As far as the Wnancing of the cost of enclosure and other improvements is concerned, this seems to have been mainly done by the landlords themselves. Tenants were required to repay the cost of enclosure at an interest rate of 4%, which would have been a very reasonable rate for agrarian credit in a pre-industrial society. However, landlords also beneWted from the increase in rents, which may explain why they were prepared to fund the costs of enclosure on favourable terms. A Wfth point is that although the Danish reforms had an egalitarian aspect, this did not mean that all beneWted equally. There were large numbers of cottagers who did not gain freehold or copyhold land, either through purchase or the enclosure of the common Welds. These bore the burden of such labour service as remained. It is probably wrong to blame the reforms for the creation of a ‘rural proletariat’ (which had probably always existed in some form or another), but the fact that some groups beneWted substantially while others did not was bound to increase social divisions (at least between these groups). This was probably inevitable. No eighteenth- or nineteenth-century government (and few since) had the power to push through a completely egalitarian land redistribution. The reforms had to take account of existing interests and work with these. What the state could (and did) do was to alter the terms on which the peasants bargained with the landlords. It forced the granting of longer leases, regulated the conditions under which landlords could evict, and obliged them to compensate tenants for improvements. The acquisition of peasant holdings by large estates was prevented (as was the splitting of holdings once consolidated).11 As one of the Danish ministers involved, Count von Eggers, wrote in an open letter to Stein in 1807: ‘it is not enough . . . just to dissolve personal bondage . . . One must secure them (the peasants) sustenance even as one gives them back their 10

Baack (1977), 37.


Tracey (1982).

Sources of Inequality in Rural Germany


freedom; one must ensure that they remain proprietors and do not become day-labourers.’12 Finally, the special conditions of Danish agriculture in the late eighteenth and early nineteenth century should be mentioned. In the Wrst place, even before the reforms Danish agriculture had not been dominated by ‘demesne’ farming (large estates typically using labour services provided in lieu of rent rather than wage labour). Large, directly farmed estates had occupied only 13% of the agricultural area before the reforms; this was reduced to 9% in 1835. Secondly, the reforms took place at a time of rising agricultural prosperity. Between the 1730s and the 1790s land prices rose sevenfold. Between 1771–80 and 1791–1800 rye prices rose 26%, oats prices rose 32%, and barley 45%. This made it easier to provide adequate landlord compensation without overburdening the peasantry. Thirdly, Denmark was well-situated for agricultural exports. By 1803, 12% of grain production was exported. With good access to foreign markets, in particular the expanding British market, there were favourable conditions for the establishment of a market-orientated peasantry. Danish geography was also well suited to an era when water transport was the only practical means of moving bulk cargoes such as grain. Finally, the debts that the peasants incurred as a result of the reforms were wiped out in a burst of inXation in 1807–10. This gave the reforms a redistributional aspect which was certainly unintended.13 (ii) Agrarian reform in central and north-west Germany Comparison with the Danish example shows that the German reforms, in particular those introduced in Prussia in 1807–11, diVered in several respects. One of the principal objectives of the centrepiece of the German reforms, the edict of 1811, was the abolition of labour services, an issue which was not addressed by legislation in Denmark. Peasant emancipation in the German case was, as Clapham put it, ‘twoedged’:14 the peasant was freed from restrictions and gained much in the way of rights, but at the same time the previous policy of the Prussian government—Bauernschutz, the deliberate protection of the peasantry against landlord demands—was also given up.15 12

Baack (1977), 39. A similar burst of inXation contributed to the success of the land reform imposed on Japan by the US occupation authorities in 1945. 14 Clapham (1921, repr. 1966), 43. 15 Knapp (1891), 35–6. 13


Sources of Inequality in Rural Germany

Labour services were not transformed into monetary obligations, but rather redeemed by the cession of land to the landlord by the peasant (a third of the total holding for tenants with hereditary rights, a half for the others). Thus, the reforms had a built-in tendency to expand the proportion of the total area farmed by large estates. Peasant land losses to landlords were, however, balanced by gains as a result of enclosure. The process still left many peasant holdings below the minimum needed to maintain a self-suYcient rural enterprise (Spannfa¨hige— capable of maintaining a team of horses or oxen—is the term used in the oYcial statistics).16 The enclosure movement in Germany started before the reforms and continued long after. The result was that enclosure and emancipation tended to be separate movements, with the consequence that the surge in production which followed enclosure could not be used to provide compensation for emancipation. However, despite these diVerences, the fact is that in a large slice of Germany, to the west of the Elbe and the north of the Main (and including the Kingdom of Saxony), the reforms worked very much as they had in Denmark: that is they produced a class of medium-sized peasant holdings, mostly owner-occupied although the majority did rent some land, free of labour service and other ‘feudal’ obligations, with the common Welds largely enclosed and some consolidation of holdings.17 There were a number of reasons for this. In many areas there was, as in Denmark, a deliberate policy bias in favour of the peasantry. The Kingdom of Saxony is perhaps the best example of this.18 In the Rhineland and other western regions, French inXuence in 1806–13 may have had some eVect, but was probably too brief to have made a major impact.19 The availability of credit was an important factor. In Saxony the Sa¨chsische Landesrentenbank lent 86 million Thalers to peasants for the purchase of labour service duties and other obligations in 1834–59.20 16 According to a Prussian survey of 1859, there were 344,701 Spannfa¨hige holdings and 604,501 Nichtspa¨nnfa¨hige (holdings below the minimum size needed to maintain a team of draught animals). The number of Spannfa¨hige had fallen from 351,607 in 1816. However, the same survey also showed that in many areas the Spannfa¨hige holdings had made net gains during the reform period; gains from the enclosure of common Welds outweighing losses through compensation for landlords, Meitzen (1868), 500. 17 In 1895 86% of agricultural land was owner-occupied, Rauchberg (1900), 580. 18 Conze (1969). 19 Dovring (1956), 144 emphasizes the impact of the French Revolution, but German historians have tended not to support this point. 20 Wehler (1995).

Sources of Inequality in Rural Germany


In general the western peasant farmer was more likely to get access to credit from Landeskreditkassen and Rentenbanken than his counterpart east of the Elbe, where the dominant credit institutions, the Landschaften, lent mainly to the large estates.21 The legal framework in the large parts of north-western and central Germany that came under Prussian administration was similar to that in the east, where the reforms led to the transfer of substantial amounts of land to the landlords, due to the fact that peasants were obliged to oVer land in return for the abolition of labour services. The fact that this did not happen suggests that labour services were either already rare, or naturally declining, or that the peasants were able to oVer monetary compensation instead.22 Local conditions could play a part. Thus in the Magdeburg region the prosperity brought about by the cultivation of sugar beet created a class of medium-sized peasant farmers who often had shares in the sugar reWneries. Labour services and payments in kind were ended by 1845, which was relatively early for central Germany, and the resulting debts seem to have been paid oV quite easily.23 In general the circumstances under which the reforms were introduced seem to have been at least as important as the actual legislation (and possibly more so). What these areas had in common with Denmark was a low level of demesne farming before the reforms, well-entrenched peasant rights which were often hereditary, and favourable economic opportunities thanks to the proximity of growing urban markets which also provided employment opportunities for surplus rural labour. It also should be remembered that the reforms of 1807–11 applied only to Prussia and to Spannfa¨hige peasants (who could provide a team of draught animals); only in the 1850s were the reforms extended to peasants who were Nichtspannfa¨hige and to important non-Prussian territories (the main reforms in Saxony date from this period). This coincided with the period of economic growth brought about by the railway boom. Almost certainly this was a favourable conjuncture. It meant that labourers who did not receive land as a result of the reforms could Wnd employment outside agriculture.24 21 Tcherkinsky (1921), 78. There were other sources of Wnance, but the Landschaften were the main sources of mortgage lending, ibid. 82. 22 This is the view of Conze (1969). 23 Plaul (1986). 24 Hempel (1957) describes how the Heuerlingen (the hired labourers of north-western Germany) became increasingly divided between agricultural and non-agricultural employment (having previously combined agricultural work and cottage industry).


Sources of Inequality in Rural Germany

(iii) The reforms in Prussia east of the Elbe This is the area which has received most attention from historians. There is a widely accepted view that something went wrong with the reforms in east-Elbian agriculture: ‘the beneWts of the agrarian reforms went largely to the feudal aristocracy’, as one historian has written.25 The aristocracy gained substantially from the surrender of peasant land as compensation for the loss of labour services, and also obtained 86% of the land made available by the enclosure of the common Welds. This view has been fairly well established since the writings of Theodor von der Goltz and Adolph Meitzen in the nineteenth century. In 1868 Meitzen analysed the movement of the land distribution in the original Prussian provinces over the 1816–59 period, and showed the impact on peasant land-holdings of the loss of land given up to landlords. However, his analysis also showed that the peasantry had had some compensation from gains made as a result of the overall increase in the land under cultivation. Writing slightly later, in 1875, von der Goltz examined the position of landless agricultural labourers, which appeared to have worsened as a result of the reforms (although this might also have been caused by demographic pressure).26 Subsequent writers, such as Max Weber and Gustav Knapp, endorsed the views put forward by Meitzen and von der Goltz, and the supposed failure of the land reforms in the Prussian east became a generally accepted part of ‘received opinion’ on the broader question of LandXucht: the depopulation of the rural east as a result of migration to cities in central and western Germany. There were Wve main reasons why the impact of the land reforms in the east diVered from the eVect in the rest of Germany. The Wrst was the remoteness of many eastern regions. In the period before the establishment of a railway network, marketing costs were extremely high for any farm which did not have easy access to water transport. While the east had its share of navigable rivers, such as the Oder and the Vistula, there were large areas between these rivers which had poor communications, and little access to the expanding urban markets of western Germany. In these circumstances, the establishment of a marketorientated peasant agriculture faced almost insuperable obstacles. 25 Koselleck (1967), 498. The aristocratic estates made substantial gains from enclosure, adding over 100 hectares on average according to one calculation, Berthold et al. (1990), i. 234. 26 Meitzen (1868), von der Goltz (1875).

Sources of Inequality in Rural Germany


Table 2.1. Communications in Prussian provinces east and west of the Elbe, 1867 Miles of improved Miles of Miles of canals and road (Chausee) railway line navigable rivers East and West Prussia Pomerania Posen Brandenburg Silesia Average east of the Elbe

53.4 63.0 67.2 62.8 82.9 64.6

9.0 10.8 10.7 18.1 24.3 14.4

14.2 13.4 13.0 28.4 9.0 15.7

Prussian Saxony Westphalia Rhineland Average rest of Prussia

105.9 159.6 178.9 148.0

26.4 30.1 37.0 31.3

20.7 14.0 24.2 20.1

Note: All figures are relative to the total land area (excluding water) in square miles (unit is pre-1870 Prussian mile). Source: Calculated from figures in Meitzen (1863), i. 65–6, iii. 291.

The poor state of communications east of the Elbe is demonstrated by the Wgures in Table 2.1, taken from a survey in 1867. The Wgures are expressed in relation to land area. An average square mile of territory east of the Elbe had only 44% of the total length of improved roads in a typical square mile in the provinces west of the Elbe. It had 46% of the railway mileage, and 78% of the total mileage of canals and navigable waterways, for an area of comparable size in the rest of Prussia. The chances of having reasonable access to improved roads or a railway were correspondingly low.27 Secondly, access to credit was limited. Only about a tenth of the rural population was able to obtain credit, and interest rates were in the range 12–30%,28 much higher than the rate of 4% charged to Danish peasants, and the similar rate which aristocratic borrowers paid on loans from the Landschaften.29 27 The importance of having good communications is well demonstrated in contemporary Argentina, where the lack of good roads in rural areas is an obstacle to the development of small farms. Only large estates can aVord the high cost of getting produce to markets. 28 Koselleck (1967), 507. 29 Tcherkinsky (1921), 43. Lu¨tge also stresses the role of credit, Lu¨tge (1963), 236.


Sources of Inequality in Rural Germany

Thirdly, the main areas of industrial expansion were not in eastern Germany but in the centre and the west. In the vicinity of these western cities, small-scale agricultural enterprises could prosper, supplying perishable goods, such as fruit and vegetables, and also livestock produce. The economics of livestock production were much better suited to peasant holdings than the extensive arable systems which characterized eastern Germany.30 Fourthly, the east was an area where the Prussian monarchy’s policy of Bauernschutz—the protection and entrenchment of peasant rights— had had less eVect. The position of the peasants before emancipation was less favourable, and their bargaining position was consequently weaker. This was an especially important factor in the parts of the east which were not part of Prussia (such as Mecklenburg-Schwerin and Mecklenburg-Strelitz) or where Prussian rule was more recently established (such as Pomerania).31 Finally, the Gutsherrschaft administrative system, common in the east, gave the aristocracy an important role in local government and the administration of justice. It is probable that this had some eVect on the agrarian reform process, either because the land-owners gained control over the operation of the reforms, or because the government was unwilling or unable to confront powerful local interests (a factor which would have applied a fortiori in areas where a German aristocracy ruled over a Polish peasantry). In assessing the reforms, historians have tended to divide between those who see this Wfth factor as the most important inXuence, and those who recognize that the other factors meant that, even if the government had been prepared to overrule aristocratic objections, the conditions in eastern Germany in the early nineteenth century were not favourable for small-scale, market-orientated, peasant farming.32 In which case, a more radical reallocation of rural property rights would have been damaging, holding back productivity gains, and producing a subsistence sector unable to produce a marketable surplus. For, considering the broader context of the contribution of the agrarian reforms to German economic development, it is clear that there are 30 Von Thu¨nen (1826) was the Wrst to analyse the eVect of urban proximity on agricultural systems; Levy (1911) explains the advantages that smallholdings specializing in livestock produce had over larger units. 31 For a more detailed discussion of the importance of the pre-reform legacy see, inter alia, Conze (1969), Lu¨tge (1963), Harnish (1986), and Perkins (1986). 32 Schissler (1986) stresses the way the aristocracy controlled the reform process east of the Elbe.

Sources of Inequality in Rural Germany


strong reasons to regard the eastern reforms as successful, despite the failure to match the egalitarian outcome in western Germany. Free labour, paid in cash or in kind, was, on some estimates, three times more productive than labour contributed under the pre-reform system.33 There was a large increase in investment: the new demesne farms required three times as much capital as the pre-reform holdings.34 And uncultivated land was reduced from 40.3% of the total agricultural area in 1816 to 19.0% in 1864. This, combined with increased yields (there was a 43% increase in the yields of the major crops 1805–63), meant that in the Wrst half of the nineteenth century the Prussian arable sector was sustaining growth rates of 1.0–1.3% per annum.35 In the context of an economy which, in this period, was largely pre-industrial (there were pockets of industrialization in Saxony and the Ruhr but no general upwards shift in the rate of growth before the 1850s), it appears that arable farming has a good claim to be regarded as a ‘leading sector’ in this period.36 The new, post-reform, ‘demesne-farming’ of east-Elbian agriculture was, in many respects, a highly Xexible contractual system. The landlord provided land to the peasant in return for labour services which were employed in the production of marketable produce on the demesne farm. In the pure form, the peasant received recompense only in the form of land. But in practice the system showed more variation. The Inste of eastern agriculture received some land and a share in the harvest. The Deputate or Kossa¨ten received a smallholding plus Wxed payment (in kind or later in money).37 This Xexibility allowed the system to adjust for the requirements of the production system. Thus peasants were often given responsibility for the care of animals and the upkeep of buildings, becoming in eVect small contractors, which 33

34 Tipton (1976), 24. Koselleck (1967), 507. Finck von Finckenstein (1960), 99–100 provides Wgures for yields and the cultivated area. The implied growth rate of production is easily calculated from these. The range of 1.0 to 1.3% per annum reXects the eVect of choosing diVerent base years. 36 HoVmann (1963), 100–4, produced much higher Wgures for the growth of agricultural production in Germany as a whole (1.9% per annum in 1800–1860/5) and derived from these an estimate of the increase in agricultural productivity of 1.3% per annum over this period. These Wgures were, however, based on estimates of livestock production which were much less well founded than those provided for the Prussian arable sector by Finck von Finckenstein. HoVmann also assumed that non-Prussian areas matched the Prussian increase in arable production, which is unlikely, given that the Prussian increase was made possible by the incorporation of previously uncultivated areas in the east, an under-utilized resource not available to other German states. 37 Finck von Finckenstein (1960), 197 gives payments to a Deputat around 1800. 35


Sources of Inequality in Rural Germany

reduced monitoring costs for the landlord. Granting harvest shares gave peasants an interest in the yield from the demesne Welds.38 The economics of the system shows that it was particularly well suited to the settlement of areas where land was relatively abundant, labour comparatively scarce, and markets either a long way away or inherently risky. The peasant household had no need to market any of its production beyond a small amount for local trade and barter. Marketing costs were borne entirely by the landlord. (iv) The extent of peasant land losses in Prussia The extent of peasant land losses has remained one of the most serious charges against the Prussian land reforms. Knapp claimed that the peasants ‘lost’ a million hectares as a result of the reforms. This Wgure is almost certainly an exaggeration. The situation was confused by the point that ‘peasant’ holdings, which in the literature of the period are generally deWned as holdings of at least 2 hectares, made both gains and losses as a result of the reforms (losing land given up as compensation for relief from labour services and gaining from the enclosure of the previously uncultivated areas). The most careful examination of the net gains and losses is probably that by Saalfeld, which is given in Table 2.2. So, in total, peasant losses were not so large when it is considered that labour services were removed from 16.95 million Morgen (the net loss of 5.5% of the aVected land is far less than the half to one third provided for in the legislation), but this has to be set alongside a substantial expansion of the total area under cultivation (the total arable area east of the Elbe rose by 61% 1800–61).39 Thus, of the land made available by the division of the commons only 14% went to the peasants; most of the rest went to the large estates. It is estimated that there was an overall increase in demesne land of 18%: 13.5% was due to the reforms and 4.5% due to purchases by the estates. Moreover, peasant losses were not evenly distributed. In Westphalia peasant land-holdings rose by 9.9% in area, and by 1.8% in Prussian Saxony, but falls of 13.2% and 12.8% were recorded for Pomerania and Silesia (respectively).40 38 It has been suggested that labour service was highly ineYcient compared to ‘free labour’ by a factor of 1 : 2 or 1 : 3 but this would have created a massive incentive to move to wage labour; the long survival of demesne farming suggests that any ineYciencies were not on this scale. 39 Dickler (1975), 273. 40 All Wgures are from Saalfeld (1963).

Sources of Inequality in Rural Germany


Table 2.2. Peasant gains and losses, Prussia 1816–59, in million Morgen


Gains: Net losses

to smallholdings to large estates total through enclosure

in Morgen (m.)

as % of all peasant land 1816

1.29 0.47 1.76 0.83 0.93

3.8 1.4 5.2 2.5 2.7

Note: ‘Peasant’ holdings are defined as between 2 and 100 hectares; a ‘Morgen’ is, according to the dictionary (Langenscheidt) ‘a variable unit of between 0.6 and 0.9 acres’ , but most commentators seem to assume that 4 Morgen ¼ 1 hectare. Source: Saalfeld (1963).

The losses have to be set against the removal of labour services. Of the spannfa¨higen Bauern 84% had had their service obligations removed (a total of 6 million days) by 1848 and of the 23.5 million Handdiensttage (days of labour service without animals) owed in 1816, 17 million had also been relieved. To get some perspective on this, it implies (assuming that the Wgure of 17 million Morgen for the aVected area is correct) that a holding of 20 hectare (or 80 Morgen) might have owed on average 80 days labour service without draught animals, and 28 with draught animals: a total of 108 days.41 It would have been a substantial gain to be free of this. The process by which the larger estates gained at the expense of peasant holdings was a complex one, in which free purchase played an important role. There was a high rate of turnover in the land market in the early part of the century, both for estates and peasant holdings. Thus 24% of Prussian peasant land changed hands in 1816–59, but most of these transfers (which totalled 8.2 million Morgen) were from one peasant to another.42 Transfers to smaller or larger holdings were only 21% of the total. Equally, the larger estates changed ownership 41 In practice it would have been more likely to be one or the other; so perhaps 50–60 days with animals and no Handsdienst, or 120–50 days’ Handsdienst on its own; calculations use Wgures from Koselleck (1967), 498; note, the estimates are much lower than the 5–6 days per week quoted by Harnisch (1986), but these levels could have been reached in some regions. 42 Saalfeld (1963).


Sources of Inequality in Rural Germany

quite frequently. Out of 888 estates in East Prussia in 1806 only 378 were in the same family in 1829.43 There were two main reasons for this. The Wrst was the cost of the reforms and associated investments which led to build-up of debt. The second was the fall in agricultural prices in the 1820s. The reforms led to a considerable expansion of the area under cultivation, but this was expensive. New houses had to be built for the labour needed to replace that previously provided by the tenants. The cost of enclosing (and where necessary draining) the common Welds had to be met. The peasants had to pay compensation where this had been agreed, meet their share of the enclosure costs, and replace livestock which were the landlord’s property. Rather more is known about the debts owed by the large estates which tended to be recorded (and periodically surveyed by the Prussian government) but the position of the peasantry was also serious.44 Whereas the estates could borrow from the Landschaften at 4% the peasantry had to pay considerably more, if they were oVered credit at all. Certainly the nobility had problems: a third of estates in Pomerania went into receivership during the period 1817–29.45 But the fact that they, and those investors who subsequently bought up the bankrupt estates, could borrow may have contributed to the gains made by the large estates during this period. If this is right then it was the economic success of the reforms which contributed to their ‘social’ failure. For it was the increase in the area under cultivation, and the consequent reduction in the uncultivated area from 40% of the Prussian land area in 1816 to 19% in 1864, that was the reform’s greatest achievement: ‘this increase of cultivated area and total production was only made possible by the new energies released by the reform process’.46 (v) The regional pattern of land losses and transfers in Prussia The impact of the reforms varied considerably in the diVerent Prussian regions. In a sense this was an intriguing ‘natural experiment’: a similar set of legal changes had diVerent eVects according to circumstances. 43

Koselleck (1967), 512. Meitzen (1885) is a rare attempt to survey debt on peasant holdings, which showed that debts could be as high relative to property values as on larger estates. 45 Koselleck (1967), 508. 46 Lu¨tge (1963), 228. 44

Sources of Inequality in Rural Germany


Government surveys showed that previously existing diVerences in agricultural structure, and in systems of agricultural tenure, aVected the way the reforms operated in the diVerent regions. Table 2.3 gives some Wgures for the most important reforms. The Wrst column gives the total number of holdings which were completely released from all services. The largest numbers of these were in Silesia and Prussian Saxony. The next two columns give Wgures for the two of the most important categories of labour service: Spanndienst (work on the demesne farm with a team of draught animals provided by the peasant household) and Handdienst (work on the demesne farm without the obligation to provide draught animals). The province where the amount of relief from Spanndienst was highest was Posen; the province with the highest amount of relief from Handdienst was Silesia. The Wnal column gives the area of the holdings whose land was ‘reformed’ (neu reguliert) through the enclosure of common Welds or the consolidation of strips. These areas were largest in the four eastern provinces: East and West Prussia, Posen, Pomerania, and Brandenburg. The other provinces had little land in this category. In general, the reforms are shown to have had little eVect in the western provinces of the Rhineland and Westphalia. In these areas feudal obligations were already in decline, as a result of developments in the pre-reform era (including the period of Napoleonic reform). Enclosure and the consolidation of holdings either took place in the Table 2.3. The impact of the agrarian reforms in Prussia, 1816–48

East and West Prussia Pomerania Posen Brandenburg Silesia Prussian Saxony Westphalia Rhineland

Number of holdings released from services

Given up days of Spanndienst (000s)

Given up days of Handdienst (000s)

Area of reformed holdings (000 Morgen)

8490 13,015 15,002 39,830 95,014 93,185 18,326 6789

375.1 818.0 1985.9 1145.1 195.1 33.4 0.0 0.0

568.2 1465.1 4344.4 2632.2 7547.5 252.8 59.5 0.0

1125.7 1208.3 1388.0 1231.3 205.3 0.2 0.0 0.0

Source: Figures from Meitzen (1868), iv. 302–3.


Sources of Inequality in Rural Germany

pre-reform era or else were carried out without direct involvement by the state or its agencies, which would exclude the reformed holdings from these statistics. Another province where there was little eVect from the reforms was Prussia (the province of Prussia was later divided into East and West Prussia), apart from the area recorded in the fourth column. Turning to the other side of the reform process, the provision of compensation to landlords for the loss of labour services and other feudal duties, this shows a similar variation between diVerent regions (Table 2.4). Large lump sum compensation payments were made in three regions: Brandenburg, Saxony, and Silesia. In Posen, compensation was more likely to take the form of an annual monetary payment. The heaviest payments in kind were made in Brandenburg. The Wnal column gives the amount of land given up as compensation, and this reXects both the extent of the reforms, and the amount of compensation paid in other ways. Thus, the low level of land losses in Saxony and Silesia can be connected to the relatively high level of capital sum payments in these provinces. In Posen, land losses were not so high, but there were substantial annual payments. The provinces

Table 2.4. The costs of agrarian reforms in Prussia, 1816–48: compensation payments by peasants to landlords Total compensation paid, in the form of:

East and West Prussia Pomerania Posen Brandenburg Silesia Prussian Saxony Westphalia Rhineland Source: As for Table 2.3.

Capital sums (000 Thaler)

Annual payments in money (0000 Thaler)

Annual payments in kind (Scheffel of rye)

Land given up (000 Morgen)

423.9 1403.4 104.7 4172.2 4192.5 5932.1 1965.6 350.3

153.2 140.2 492.6 266.4 338.9 134.0 68.7 5.9

15,300 36,900 10,400 108,500 51,900 23,700 2100 600

196.9 590.6 206.9 405.9 118.4 14.0 0.3 0.1

Sources of Inequality in Rural Germany


with the largest losses were Pomerania and Brandenburg. In the former, payments under the other categories were relatively low, and so it seems that giving up land was the preferred method of compensation in this province. In Brandenburg, the overall level of compensation appears to have been relatively high, with high payments in all categories. What this comparison shows is the variation in the reform process: a typical reform in Posen might involve relief from an obligation to provide Spanndienst, labour service with draught animals, and the compensation would be in the form of increased monetary payments; in Silesia, it was more likely to be relief from Handdienst, with compensation in the form of a lump sum payment. Concentrating on the extent of land losses and gains by peasant households obscures these other dimensions of the reform process. Although the reforms did involve some loss of land by peasant farmers, this has to be set against the gains that were made. The loss of grazing land may have been outweighed by the ability to devote more time to the holding (due to relief from labour services). The new holding was likely to be enclosed, which made it easier to introduce improved husbandry techniques. Cattle could be kept instead of horses or oxen to work the landlord’s Welds. For the landless labourers, however, the loss of common grazing was probably not outweighed by other gains. The main social cost of the reforms was the rise in the number of landless labourers, and the increase in the numbers of peasant holdings which were not large enough to be self-suYcient.47

Table 2.5. Consolidation of holdings in Prussian Saxony

Number of farms Total area Number of plots before consolidation Average size Number of plots after consolidation Average size



848 6101 Morgen 18920 0.32 Morgen 913 6.68 Morgen

1241 6803 Morgen 16100 0.42 Morgen 1584 4.29 Morgen

Source: Conrad (1891), 666. 47

Berthold (1978).


Sources of Inequality in Rural Germany

(vi) Consolidation and the reforms in southern Germany In southern Germany the reforms started late and in one important respect were unsuccessful: there was relatively little consolidation of holdings. Fragmentation of holdings had been a fairly general consequence of the open Welds system practised over much of Germany in the eighteenth century. It was greatly increased by the system of partible inheritance or Realteilung practised in the south (and other areas). In those areas of central Germany where the reforms were fully implemented there was considerable consolidation. Thus in two districts of Prussian Saxony, Altengotten and großengotten, the average numbers of separate plots were reduced as follows: In the article from which these Wgures are drawn Conrad gives seven examples of districts where consolidation had not taken place by the time the article was written (it was published in 1891). In these the number of separate plots per holding ranges from 21 to 59. Most are in the south, though some are in central Germany (the Koblenz area, for example). In general consolidation made slow progress in the Wrst half of the century, and eVective legislation was not passed until the 1850s in Prussia, Baden, and Hesse, and the 1860s in Bavaria and Wu¨rttemberg (further laws were passed in these states in 1886).48 Despite this there had already been some consolidation in the north-west. Even after laws had been passed there was much less consolidation in the south. In the south-west in particular, where peasant rights were strongly protected, there was opposition to the division of the common Welds. The system which evolved in the south was characterized by a relatively high population density, quite high reliance on non-agricultural earnings to supplement the income from the holding, few large estates but many smallholdings, and continuous division of holdings (not necessarily on inheritance—children could be given a share on reaching their majority). It was probably not very eYcient in economic terms. Germany came to import livestock products from Denmark and Holland which, arguably, could have been produced on southern farms. But it provided a reserve of labour which was available for industrial use when needed.49 And in downturns the rural economy could reabsorb surplus labour, at least for a time. Whether this system produced more or less discontent is hard to say. Contemporary observers thought it gave the peasants a higher social 48

Haushofer (1957), 58.


Catt (1986).

Sources of Inequality in Rural Germany


status.50 But, on the other hand, it was a vulnerable system, with a tendency to overpopulation, and not one which was conducive to agricultural progress. The dependence on non-agricultural earnings worked well when there were local sources of employment, but less well when craft earnings came under pressure due to competition from large-scale industry. In practice only relatively few of those with smallholdings found industrial employment: 27.5% of those with possession of holdings of less than 2 hectares in Baden in 1907, and 23.4% in Wu¨rttemberg, against 55.8% in Westphalia and 40.6% in Anhalt.51 (vii) Concluding comments The comparison with Denmark, and the consideration of the reform process in the diVerent parts of Germany, makes it plain that local circumstances and the timing of the reforms could have important eVects on the outcome. Past policies also had an eVect. The Wrm establishment of peasant rights before the reforms certainly aVected the subsequent distribution of land. Thus in Mecklenburg and the former Swedish Pomerania there had been little protection of peasant rights and these areas became dominated by large estates: 75% of agricultural land in the Stralsund region was in holdings of over 100 hectares in 1882. In Prussia east of the Elbe the proportion was 43%.52 So the Prussian policy of peasant protection did have some eVect. But in view of the cost of the reforms, the paucity of rural credit institutions, and the fact that labour was becoming less scarce, it was probably inevitable that the large estates were going to make gains, especially in the remote areas east of the Elbe.

4. G er man A g r i c u l t u r e i n C o m p a r i s o n with Other European Countries As already noted, the implicit assumption behind the ‘failed revolution’ hypothesis is that there was the possibility of a genuinely revolutionary outcome, which would have resulted in a more complete modernization of nineteenth-century German politics and society. For political 50

51 For example, Knapp (1887), 309. Dillwitz (1975), 110–11. Perkin (1986); Dillwitz (1973); Conze (1969) refers to the annihilation of the peasantry in Mecklenburg and Pomerania in the eighteenth century. 52


Sources of Inequality in Rural Germany

developments the implied point of comparison is usually Britain, whose path of gradual progress to a widened franchise and a fully democratic system of constitutional government contrasts with Germany’s halting and uncertain moves in the same direction. For agriculture, the most relevant comparison is the land settlement brought about by the French revolution. But it should not be assumed from this that all was well with rural France in the nineteenth century. While the revolutionary settlement had created a large number of small and medium-sized holdings, the level of owner-occupation was lower than in Germany. In 1882, only 60% of French agricultural land was in owner-occupation, 27% was rented, and 13% was in metayage, or share-cropping. By contrast, in Germany in 1895, over 86% of land was owner-occupied, just 12% was rented, and only 1% was in share-cropping (this adds in Deputatland, which was a form of share farming although applied to animal husbandry).53 A second problem was the failure of yields and productivity in France to rise as much as in Germany. According to Bairoch, around 1800 French wheat yields were 85% of the German level; by 1910 they were only 71% of the German Wgure. His estimates of agricultural productivity show a rise for Germany of 371% 1800–1910, for France the increase is 172%.54 There are various measures which can be used to compare the structure of German agriculture after the reforms with other countries. One is the division of the agricultural labour force between employed and self-employed workers. Labourers who were either landless, or who had insuYcient land to provide full-time employment on their own holding, would be obliged to take work on other farms or outside agriculture. These would then be classiWed as employed, in contrast to those whose own holding was large enough to support themselves and their family without working oV the holding. The proportion of the agricultural labour force classiWed as employed is thus an approximate measure of 53 German Wgures from Rauchberg (1900), 579; French from Auge´-Laribe´ (1950), 93. The German Wgure was just below that of South Korea in 1980, 92% owner-occupied, which had increased from 35% in 1950, Wgures from Asian Productivity Organisation (1994), 37. The South Korean example is generally considered the best-managed land reform in recent years. By contrast, despite various eVorts at land reform, Pakistan still has 36% of land in tenancy and 18% in share-cropping, ibid. 70. 54 Figures from Bairoch (1989); Bairoch’s measure of agricultural productivity is calories produced per male agricultural worker, which is not ideal but probably the only measure which can be constructed for this period given the deWciencies of the data sources.

Sources of Inequality in Rural Germany


the proportion that had access to holdings of a viable size. Table 2.6 provides a comparison drawn from national censuses carried out around 1900. Limitations in the data sources mean that the analysis has to be limited to the breakdown of the male labour force. This table shows that the German reforms produced an agrarian sector which predominantly consisted of self-employed owneroccupiers: the classic land reform objective. The percentage of the agricultural labour force which was classiWed as being employed was low by comparison with other European countries. The French level was notably higher. The high level of employment in England and Wales is not so surprising, given a historical process which had resulted in an unusually inegalitarian allocation of rural property rights. However, a number of other countries, not usually considered to be particularly unequal, had lower levels of self-employment than Germany. Quite evidently, the Danish land reforms did not produce a position in which all had equal access to land.55 Table 2.6. Percentages of the male agricultural labour force classified as being employed (including servants), excluding family members, various censuses circa 1900

Austria-Hungary Norway Finland Germany Belgium Italy France Sweden Switzerland Denmark Netherlands England and Wales

Date of census

% classified as employed

1900 1900 1910 1907 1901 1910 1901 1900 1900 1900 1899 1901

28.3 29.3 30.7 36.7 36.8 46.8 47.2 47.6 51.5 60.1 67.6 73.0

Source: Figures from Flora et al. (1983), adjusted to exclude family members and unclassified workers. 55 The extent of ‘landlessness’ is more diYcult to compare. Many employed workers in Germany, and elsewhere in Europe, had the use of a smallholding to grow potatoes and other vegetables. The Wgures given in Table 2.5 are not comparable with modern surveys of landlessness: see Jazairy et al. (1992), 406–7 for a summary of these.


Sources of Inequality in Rural Germany

Comparable Wgures on the distribution of agricultural land by size of holding are only available for a few countries. Table 2.7 provides some Wgures for England and Wales, Germany, and France. The Wgures have been adjusted to allow for diVerences in the classiWcations by size class. The British and German classiWcations use broadly comparable methodology: German holdings were classiWed by total agricultural area, the British by area under crops and grass. The French agricultural censuses classiWed holdings by total area, which tended to increase the importance of larger holdings, which would often include a disproportionate amount of woodland, moor, or other uncultivated areas. Comparison of the English and German Wgures shows that the distribution of land by size of holding was much more unequal in England: there was a smaller smallholding sector and the area in large holdings was proportionately greater. Typically, English agricultural land was concentrated in medium- or large-scale tenanted holdings of 100 acres or more. Comparison with France is more diYcult, due to the diVerent classiWcation methods used. An English statistician who examined this issue in 1887, P. G. Craigie, thought that, whereas three-quarters of the English agricultural area was in holdings of over 100 acres (roughly Table 2.7. A comparison of the percentage distribution of land by size of holding

Area in holdings of less than 20 hectares Area in holdings of between 20 and 40 hectares Area in holdings of over 40 hectares Area in holdings of between 40 and 100 hectares Area in holdings of over 100 hectares

England and Wales 1885

Germany 1882

France 1882
















Note: England and Wales: classified by area under crops and grass; Germany: classified by total agricultural area; France: classified by total area. Source: British figures from Ministry of Agriculture, Fisheries and Food (1968), 20; French and German figures from Tracey (1982). To produce comparable figures, the density functions were interpolated using a log-log interpolation method.

Sources of Inequality in Rural Germany


equivalent to 40 hectares), only a third of the equivalent French area was in this class.56 This is a considerably lower proportion than the 48.9% share given in Table 2.7. If this is right, then the distribution of agricultural area in France may have been rather more egalitarian than in Germany, contrary to the impression given by the French distribution by total land area, shown in Table 2.2. Either way, the diVerence was probably not that great. However, the objective of agrarian reform is not just to produce a relatively equal distribution of land-holding: it is also to produce a more egalitarian pattern of land ownership. Comparison of the distribution of land according to the size of the owned farm or estate is even more diYcult, as there are no oYcial Wgures available for this period. In England in the early 1870s there was a Local Government survey of land ownership, sometimes referred to as the ‘New Doomsday Book’.57 This survey showed that land ownership in England and Wales was highly concentrated: 74% of privately owned land was in estates of 300 acres or more. This is a distribution by size of total land ownership, not ownership of agricultural land. Making some allowance for this suggests that between 60% and 65% of agricultural land was in estates of 300 acres or more (the lower Wgure assumes that all nonagricultural land was in larger estates), and between 65% and 70% in estates of 100 hectares or more. This contrasts with the farm census Wgures which show that 28% of agricultural land was in holdings of 300 acres or more in 1875 (or 38% in holdings of over 100 hectares). Thus, the distribution of land by ownership was much more inegalitarian than the distribution by holding. The high level of owner-occupation in Germany meant that the gap between the distribution by ownership and by holding was nothing like as large as in Britain. The distribution by ownership may well have been more egalitarian than the distribution by holding size. This could arise if larger holdings rented from more than one landlord, or if they included some rented land as well as some which was owneroccupied. The best source of information on the structure of ownership in German agriculture is a survey carried out by the Reich statistical oYce in 1938. This produced a careful cross-tabulation of the distribution of 56 Craigie (1887), 110; Levy (1911) is also a useful source on the structure of English agriculture in comparison with other countries. 57 The original returns were summarized in Bateman (1879), but this study makes use of the adjusted results published in Brodrick (1881).


Sources of Inequality in Rural Germany

land by size of estate and by size of holding (something which is not available for British agriculture). The results are summarized in Table 2.8. This table shows a fairly close correspondence between the size distribution by holding size and the pattern of ownership. There are some regions where the ownership distribution was signiWcantly more equal than the holding distribution: Brunswick, Silesia, Westphalia, and Hanover are the most notable. There are two regions where the ownership distribution was signiWcantly less egalitarian: Mecklenburg and Anhalt. But in general the two series do not diVer by much.

Table 2.8. The distribution of land by size of holding and by size of estate, by province, 1938 % of land in individual ownership

% of agricultural land in holdings of over 100 hectares

% of all privately owned land in estates of over 100 hectares

East Prussia Brandenburg Pomerania Silesia Prussian Saxony Schleswig-Holstein Hanover Westphalia Hesse-Nassau Rhineland Bavaria Saxony Wu¨rttemberg Baden Thuringia Hesse Mecklenburg Oldenburg Brunswick Anhalt

72.0 71.1 71.8 71.7 72.1 83.8 77.1 81.1 48.2 59.9 72.1 72.4 63.9 61.5 62.3 62.7 82.8 84.4 56.2 55.5

32.2 35.7 46.5 38.1 25.9 16.8 19.7 19.1 6.4 10.5 5.7 17.4 1.1 10.9 9.8 9.9 35.0 11.8 13.2 26.1

32.0 31.1 40.0 26.7 25.7 14.4 10.1 5.7 6.7 4.7 4.5 13.8 3.9 5.9 11.3 5.1 48.2 19.3 2.9 35.2

All Germany




Source: SDR (1938).

Sources of Inequality in Rural Germany


The drawing of conclusions for the pre-1914 era from a survey undertaken in 1938 is obviously something which needs to be done with some care. The crisis which aVected German agriculture in the 1920s would have altered the relationship between the two series to some extent. One consequence could have been an increase in the ownership of land by institutions, both public sector and private (as a result of the repossession of mortgaged estates, for example). The Wrst column of Table 2.8 shows that, in some regions, the percentage of land in private ownership had fallen to under 60% (though this does not appear to have much eVect on the relationship between the two distributions). With these caveats in mind, a hedged statement is advisable: on the evidence currently available, there is no reason to anticipate that the ownership distribution in the German regions before 1914 would have been markedly less egalitarian than the holding distribution, with the exception of regions such as Mecklenburg where the 1938 survey showed this to be the case. In other words the holding distribution is a good guide to inequality in land ownership in pre-1914 Germany (which follows from the high level of owner-occupation in German agriculture). Looking at the map of German agriculture in 1895 (Figure 2.2), the only areas with a level of inequality in land-holding which came up to the English level (38.6% of land in holdings of over 100 hectares) were those east of the Elbe. Even in this area there were several regions where larger holdings were relatively rare: in the East Prussian region of Gumbinnen holdings with over 100 hectares accounted for 33.1% of the total agricultural area, in the Silesian Regierungsbezirke of Oppeln and Liegnitz they occupied 31.2 and 28.4% respectively. But when allowance is made for the substantial gap in Britain between inequality in land ownership and in land-holding, the diVerence between the two countries becomes a massive one. The English level of 65–70% of land in the ownership of estates with at least 100 hectares was way above the most probable level in most regions east of the Elbe. It was equalled only in parts of Pomerania (the Stralsund Regierungsbezirk and possibly Ko¨slin as well) and, most probably, in Mecklenburg. In Posen and Bromberg, the English level could have been reached if the ownership distribution had been substantially less equal than the holding distribution. The image of east-Elbian agriculture as one dominated by large estates, on the English pattern, is to a large degree a false one. The


Sources of Inequality in Rural Germany

25−30% 30−35% 35−40% 40−45% 65%

Fig. 2.2. Percentage of land in holdings of over 100 hectares, 1895

typical farm in Brandenburg, Silesia, East Prussia, and the Danzig region of West Prussia was more likely to be an owner-occupied holding of around 30–50 hectares. Even where larger estates predominated, they were very diVerent from the English model: an average Junker estate might consist of around 250 hectares farmed ‘in hand’; the equivalent English aristocratic estate in the 1890s would be almost entirely let out to tenants, and considerably larger. In Germany 14% of land was in some form of tenancy in 1895; the Wgure for England and Wales in 1891 was 85%.58 Tenanted percentages in German regions came nowhere near the English Wgures. Figure 2.3 gives the percentage of the total agricultural area which was rented, by region. The map shows that the highest percentages were in AlsaceLorraine and in Mecklenburg. The lowest was in Bavaria. In general, the rented percentage was not particularly high east of the Elbe, apart 58

Rauchberg (1900), 579; Ministry of Agriculture, Fisheries and Food (1968), 25.

Sources of Inequality in Rural Germany


from Mecklenburg. Levels in the other eastern provinces were below those in central and western regions such as Hanover, Prussian Saxony, and the Rhineland. In most of Germany, especially in the west and the south, large estates were a rarity. The typical farm was small, 10–20 hectares, and owner-occupied. There was little employed labour. The rural sector was therefore, by the standards of contemporary European countries, a relatively egalitarian one. The low proportion of landless labourers in the rural population as a whole, and the high level of owner-occupation, mean that the structure of nineteenth-century German agriculture compared well with the situation in many less-developed economies today.59

5. T h e Ku z n e t s C u r v e a nd G e r m a n A g r i c u l t u r e The Kuznets Curve is a hypothesis about the development of inequality in industrializing economies based on a series of assumptions about inequality in diVerent sectors. However, the analysis of particular countries shows that historical circumstances and geographical conditions matter a lot and can lead to large diVerences: Do these favour large estates or plantations? Has there been an attempt at land reform? Was this successful or unsuccessful? In the German case, the agrarian reforms produced a relatively egalitarian rural sector, which meant that the starting point for the Kuznets Curve was a low one.60 If inequality in the industrial sector turned out to be high, then a shift into this sector from the more egalitarian rural sector would tend to push up overall inequality. The structure of German agriculture also meant that the experience of rural–urban migration produced a much more dramatic change in social circumstances than in Britain. In Britain the tripartite division of industrial society—owner or shareholder at the top, salaried manager or foreman in the middle, wage-earning worker at the bottom—was not dissimilar to the tripartite division of rural society—large land-owner who rented

59 Prosterman and Riedinger (1987), 41 gives the current position in developing economies. 60 The Prussian rural distribution was an egalitarian one by modern standards. This issue is further discussed in Chapter 9, which includes estimates of income inequality derived from tax statistics.


Sources of Inequality in Rural Germany

15−20% >5%






Fig. 2.3. Rented land as a percentage of the total agricultural area, 1882

out land at the top, tenant farmer in the middle, landless rural labourer at the bottom. The position in rural Germany was a very diVerent one. Migrants from the rural areas in the south and the west came from areas where peasant proprietorship was the dominant system. If they were the sons and daughters of peasant farmers they would have had little experience of wage labour. If they came from the families of the relatively small number of landless labourers, the social context would still have been a dissimilar one (working alongside neighbours who did own some land) to that provided by an urban factory. Even in the east, the position of the agricultural labourer was not that of a wage-earning labourer paid by the hour. There were a variety of contractual arrangements, most of which gave a degree of autonomy to the labourer and his family, at least as far as the organization of the day’s work was concerned.61

61 The system was one which required the worker to supply speciWed hours of labour at harvest time and other peak periods, but which left the organization of duties such as cow-keeping up to the labourer. Outside peak periods, the worker and his family were

Sources of Inequality in Rural Germany


For these reasons, the structure of German agriculture had an eVect on the impact of industrialization and migration: the relative change was a greater one. Industrialization brought about a transformation of society which was more rapid than in early-industrializing economies such as Britain. It was also a more profound change than in Britain. expected to occupy themselves on a smallholding provided by the land-owner. See Weber (1893) and Chapter 6 below for further details.


The Pattern of Migration, 1870–1913 1. I n t r o d u c t i o n The discussion in the Wrst two chapters has made it clear that there are particular reasons to be concerned with the process of migration during the early stages of industrialization. First, migration is the conduit whereby a rural labour surplus holds down wages in urban labour markets during the labour surplus phase of the Lewis Model. Secondly, the transfer of population from a relatively egalitarian rural sector to a more unequal urban sector is an integral part of the Kuznets process, pushing up overall inequality. Both these factors make it important to consider the factors which aVected the pace of migration, and which eventually led to a change in the pattern of migration as the labour surplus phase came to an end.1 The important feature of the labour surplus phase is the movement of population from predominantly agricultural regions to the expanding urban centres, resulting in a high level of net losses in the former areas, and large gains in the latter. But this feature was not the only type of migration in industrializing societies. It was superimposed on a pattern which already existed in pre-industrial societies: of seasonal migration connected with the harvest demands and other types of work inXuenced by weather conditions. It also merged into a more modern pattern, characteristic of mature industrial societies, of migration between urban centres, responding to shifts in labour demand by industry. But neither of these other streams has the implications predicted by the Lewis Model. Other migration streams which have diVerent implications are those which cross international boundaries. Where there is a large outward 1 Internal migration is not the only source of downward pressure on labour markets. Kindleberger (1967), 20 argues that in late nineteenth-century America there were labour surplus conditions created by heavy immigration. In post-1945 West Germany there was a similar eVect as a result of the inXux of refugees, ibid. 34.

The Pattern of Migration, 1870–1913


Xow of emigrants this will transfer a portion of the rural labour surplus to foreign labour markets, thus acting as a ‘safety valve’, relieving pressure on over-supplied urban markets in the country of origin. But where there is large-scale immigration, then surplus conditions elsewhere can intensify the ‘labour surplus’ phase in the host country. EVectively, the labour market in the receiving country is integrated with that of the country of origin. Those parts of the migration process which have tended to attract most attention are moves which result in a permanent gain or loss of population. In the aggregate this is net migration: a surplus or deWcit is created when moves in are deducted from moves out. But in most societies net internal migration is relatively small compared to gross internal migration. There are many moves which are temporary or reversible. There are life-cycle moves: young men (less often women) who move to cities intending to return, perhaps hoping to have accumulated enough savings to buy a smallholding or some other business in their region of birth. There are seasonal moves: farmers who take winter jobs in cities but return for the harvest; other agricultural workers who move with the harvest, starting in the warmer southern regions, and moving north and east as the harvest season progresses. There are certain skilled occupations which have always been associated with high mobility: stonemasons and other skilled building workers moved with major construction projects in medieval Europe, such as bridges or cathedrals. Railway construction had a similar eVect in the nineteenth century, attracting large numbers of workers, skilled and unskilled.2 Most of the available statistical sources tend to measure net migration better than gross migration. Census returns generally record place of residence at the time of the census and place of birth: it is rare to Wnd any evidence on places of residence between these two points. This means that many temporary moves between these dates are unrecorded. Germany is relatively unusual in that, thanks to the system of MeldepXicht, compulsory registration after a change of residence, there is some evidence on gross migration.3 This makes it possible to examine the relationship between the labour surplus transfers and the general pattern of mobility. 2 Von der Goltz (1875) includes the results of a survey of rural migration in the early 1870s, which shows both the traditional pattern of seasonal migration, and the impact of railway construction. 3 The MeldepXicht system and its value as a migration source is described in Meyer (1938).


The Pattern of Migration, 1870–1913 2. G e r m a n M i g r a t i o n a n d U r b a n i z a t i o n in International Comparison

Most of this study is concerned with the eVect of internal migration in Germany. But it is worth considering whether the German experience diVered from that of other European countries, and if so, in what way? It is not easy to achieve comparability of migration statistics between countries. The drawing of the regional boundaries aVects migration records. A country with a large number of regional units will tend to have more recorded migration than one with a relatively small number of larger units. An accurate comparison needs to be based on a similar set of regional units for all countries. In the late nineteenth century, the American statistician, Adna Weber, attempted such a comparison. His results are given in Table 3.1. These show that, compared to all other European countries, the German states of Prussia and Saxony had the highest rates of internal migration. In many other countries there were relatively few internal moves. This is particularly striking for the Scandinavian countries, which had high rates of external migration, but much less internal migration.4 Table 3.1. Percentage of the population born outside the country where enumerated

Saxony Prussia England and Wales Denmark Austria France Switzerland Netherlands Norway Sweden Hungary

Date of census

% born elsewhere

1885 1890 1891 1890 1890 1891 1888 1889 1875 1880 1890

30.9 30.3 28.4 22.1 19.8 18.7 17.9 13.0 12.8 11.5 10.8

Source: Weber (1899), 249. 4 Norway and Sweden had the highest rates of emigration after Ireland, Hatton and Williamson (1998), 33.

The Pattern of Migration, 1870–1913


Another approach is to compare rates of urbanization. If cities are growing fast, so that the proportion of the population living in cities is rising, then this can come about either through higher rates of natural increase or through migration. In practice, in most European countries, the Wrst factor was not important: the rate of natural increase in cities was generally below that in rural areas. So, rising urbanization was a consequence of migration. Table 3.2 presents some Wgures for the proportion of the population living in cities in various European countries. These are countries for which results are available both for the period around 1870 and again for the period just before the First World War. The results for the Netherlands in 1909 are given in brackets as there are no comparable Wgures for the earlier period. The table is in two parts: Part A gives the percentage of the population living in cities of over 100,000 inhabitants; Part B gives the proportion living in towns and cities with over 20,000 inhabitants. The table shows that by 1910 Germany was the third most urbanized country in Europe after Britain and the Netherlands. There is a signiWcant gap between Germany and the fourth-placed country on both measures. In 1870, by contrast, Germany was behind France and Belgium on both measures, and around the same level as Italy, Denmark, and Ireland (the relative position depending on the measure used). The increase in urbanization between the two dates (measured as the annual gain in percentage points) is larger in Germany than in any other country. The German gain during the period 1871–1910 was 16.5 percentage points using the ‘over 100,000’ measure and 22.2 using the ‘over 20,000’ measure. The next-placed country was England and Wales with gains of 12.1 and 18.6 respectively (1871–1911). The gain for France was just 5.7 percentage points on the ‘over 100,000’ measure, and 9.3 percentage points for the ‘over 20,000’ measure. Urbanization in nineteenth-century Germany proceeded at a much faster pace than in other European countries. The pressure on urban services and amenities was correspondingly high, as was the pressure on urban labour markets in the Lewis Model framework. Where high urban growth resulted from high inward migration, this also meant that a higher portion of the population was exposed to the experience of social dislocation resulting from moves from a rural society still dominated by traditional attitudes and values to a very diVerent urban environment. Table 3.3 provides Wgures for a more detailed comparison of Germany with England and Wales over the whole of the nineteenth


The Pattern of Migration, 1870–1913

Table 3.2. Urbanization in various European countries, c. 1870 and c. 1910 A: Living in cities >100,000 Date of % census England and Wales Scotland Denmark France Belgium Ireland Italy Germany Austria Sweden Norway Switzerland Finland Netherlands not available

1871 1871 1870 1872 1866 1871 1861 1871 1869 1870 1875 1870 1870

25.8 22.7 10.2 9.1 8.1 7.8 5.4 4.8 3.7 3.3 0 0 0

Date of % census England and Wales Scotland (Netherlands) Germany Denmark Ireland France Belgium Italy Austria Norway Sweden Switzerland Finland

1911 1911 (1909) 1910 1911 1911 1910 1910 1911 1910 1910 1910 1910 1910

37.9 30.1 (29.7) 21.3 16.8 15.8 14.8 11.0 10.9 10.6 10.3 9.2 8.6 4.7

B: Living in towns and cities >20,000 Date of % census England and Wales Scotland Belgium France Germany Ireland Italy Denmark Norway Austria Switzerland Sweden Finland

1871 1871 1866 1872 1871 1871 1861 1870 1875 1869 1870 1870 1870

42.0 31.8 15.4 14.8 12.5 11.6 11.1 10.2 8.5 6.5 6.4 5.8 1.6

Date of census England and Wales Scotland Germany Belgium Italy Denmark France Ireland Switzerland Norway Austria Sweden Finland (Netherlands)

1911 1911 1910 1910 1911 1911 1910 1911 1910 1910 1910 1910 1910 (1909)


60.6 49.5 34.7 28.8 28.2 27.0 24.1 21.5 21.1 18.1 16.9 16.1 9.3 (40.4)

Source: Data from Flora et al. (1996), 249–80; the French and Belgian results used diVerent-size classes and were adjusted to make them comparable by extrapolating the density function.

The Pattern of Migration, 1870–1913


century. This shows that in 1801 England was much more urbanized than Germany in 1871. Interpolating between the census years, it appears that Germany reached the English 1801 Wgure of 23.0% in 1893. Germany then moved to 34.7% in 1910. Using the same approach on the results for England and Wales shows that this Wgure was attained around 1853. So, while it took Germany 17 years to get from 23.0% to 34.7%, it took England and Wales 52 years to make the same move. Urbanization was three times as fast in Germany than in England during the equivalent period.5 Table 3.3. Percentage of the population living in towns and cities >20,000 Germany Date of census

1871 1880 1890 1900 1910

England and Wales %

Date of census


12.5 16.1 21.0 28.8 34.7

1801 1811 1821 1831 1841 1851 1861 1871 1881 1891 1901 1911

23.0 24.4 25.9 28.5 30.7 33.6 38.2 42.0 47.9 53.7 58.2 60.6

Note: A comparison with France shows that the move from 4.8% of the population living in cities of over 100,000 to 14.8% took 59 years in France (1852–1911) but 26 years in Germany (1871–97). Source: Data from Flora et al. (1996), 262, 278.

Looking at results for individual cities, it is clear that one reason why urbanization was so fast in Germany was that all the main cities grew fast. The annual rates of growth of the Wve largest cities in 1871–1900 were: 5

There is no simple relationship between urbanization and economic growth, but this comparison provides another piece of evidence (if more were needed) indicating that the rate of growth in Britain during the industrial revolution was relatively slow.


The Pattern of Migration, 1870–1913

Berlin 3.70% Hamburg 2.85% Munich 3.52% Leipzig 3.66% Dresden 3.22%6 In Britain there were cities which matched these rates of growth: Liverpool grew at 3.30% per annum in 1801–31; Glasgow at 3.03% in 1801–41. But there were other cities which grew less fast: Birmingham at 2.4% in 1801–31, and London at 1.95% in 1801–41. The slow growth of London, which was already a large city in the eighteenth century, is particularly important in holding back the overall rate of increase.7 Comparison with cities outside Western Europe shows that there were American cities which matched or surpassed German rates of growth: New York grew at 3.9% per annum in 1825–71 and Chicago went from a population of 80,000 in 1855 to one of 420,000 in 1876, an annual growth rate of 8.2%. Massive immigration made these rates of growth possible. But the German rate of urbanization was unusually fast by the standards of nineteenth-century Europe: faster than any other European country had experienced. This meant that internal migration was also relatively high. The natural rate of population increase for Berlin was only 0.95% per annum in 1841–1900; it required high rates of inward migration to raise the annual growth rate to 3.7%.

3. M e a s u r i n g G ross a nd N e t Mi g r a t i o n As mentioned in the Introduction, there was an obligation on the German population to register a change of residence. This created a source of data on migration which is particularly valuable in that it gives information about total moves. The obligation did not apply to temporary changes and there may also have been some evasion. In 1902 the Berlin statistical oYce published adjusted Wgures which took account of under-recording, the extent of this being estimated from a comparison with the census results. No adjustment to the amount of 6 Figures from Schrott (1903), who provides a valuable adjustment to the population Wgures which takes account of boundary changes. 7 British Wgures taken from the 1879 edition of the Encyclopaedia Britannica, articles on various cities.

The Pattern of Migration, 1870–1913


inward migration was considered necessary, but the numbers of outward moves were adjusted upwards by 17%. However, it is possible that the Berlin police were unusually vigilant. Others may have been more lax. It is noticeable, for example, that recorded gross migration is signiWcantly lower for Bavarian cities than in other parts of Germany.8 Keeping these reservations in mind, these Wgures still show a society which had high levels of mobility. From 1898 onwards the available city statistics were collected and published in the Statistisches Jahrbuch Deutscher Sta¨dte. By 1907 this included Wgures for 48 cities, with a total population of 11.3 million. These cities recorded a total number of moves in of 2.06 million and moves out of 1.81 million. Total moves amounted to 346 per 1,000 head of population.9 There will, however, be some double counting in this (moves from one city to another will be counted twice). Moreover, some individuals will have left and come back in the same year, thus counting as two moves. The adjusted series produced by the Berlin statistical oYce runs from 1841 to 1902. For this study the series has been extended to 1912 using the same adjustment procedure, and the results of the 1905 and 1910 censuses. Figure 3.1 shows the results. There are four periods which can be distinguished. The Wrst runs from 1841 to the late 1850s. In this period overall mobility is relatively low. On average total moves are 141 per 1,000 head of population in 1841–59. Net gains are 1.2% per annum on average, but these Xuctuate considerably. In the second period, from 1860 to the early 1870s, overall mobility is much higher and rises to a peak of over 270 moves per 1,000 in 1871–3. Net gains are also high, averaging 3.4% in 1860–73, although these do still Xuctuate.10 It is worth taking note of the timing of this phase: it coincides with the onset of faster industrial growth brought about by the boom in railway construction. It is a period in which agricultural prices were tending to rise. The increase in mobility was not connected to any agricultural crisis. 8 Berlin Wgures from SJB (1903), 158; Wgures for all cities were published annually in SJDS. 9 SJDS (1908), 39; all the main cities are included with the exception of Munich. Total moves includes all moves to and from each city; moves within the city boundaries are not included. 10 This is partly due to the inXuence of the war years 1866 and 1870; in both these years net migration is low, presumably as a result of the conscription of many potential migrants.

The Pattern of Migration, 1870–1913

64 400



100 Net moves 0 Total moves

−100 1841









Fig. 3.1. Net and gross migration, Berlin 1841–1912: net moves and total moves per 1,000 head of population Source: The original series was in SJB (1903), 158; the extension uses annual data from SJDS, adjusted for the census results for 1905 and 1910.

In the third period, from 1874 to 1890, mobility is below the 1871–3 peak although there is an upward trend. Net gains are substantial, although, at 2.2% per annum, these are lower than in the previous period. From 1890 onwards net migration falls, while gross migration rises, eventually rising above the level of the 1871–3 peak. By 1910–12 Berlin was no longer gaining population through migration.11 So the Wnal period is one of high mobility, a Xexible pre-1914 labour market, but not one of high net migration. Whether this is typical of Germany as a whole is a question which will be considered later. What was the reason for this high mobility? There was a seasonal element, which leads one to conclude that some migrants did return home to help with farmwork in the summer. Figure 3.2 gives a monthly breakdown of migration in and out of Berlin in 1901–2.12 There was a 11 The chart shows a net loss; however, this is partly due to a shift in population to areas which were not included within the pre-1914 city boundaries: Charlottenburg, Neuko¨lln, Scho¨neberg, Spandau, Lichtenberg, and Wilmersdorf. Including these in a ‘Greater Berlin’ unit (and allowing for some under-recording of moves out) alters the overall result to one of balance: moves out equalled moves in. 12 Data from SJB (1903), 159; the moves out have been adjusted upwards by the statistical oYce.

The Pattern of Migration, 1870–1913






Net moves −1 Moves in −2

Moves out Jan.






Fig. 3.2. Net moves, moves out and moves in, as percentages of the total population, Berlin 1901–2 (averages of the two years) Source: Data from SJB (1903), 159.

substantial move out in March, at the start of the sowing season, and a smaller outwards movement in the summer months. In all Berlin lost 0.76% of the total population in the period February–July and gained 0.70% in the remaining months. However, this is probably a serious underestimate of the true movement; many of those who returned for just a few weeks in the summer may not have needed to register a change of address. In general, however, this is not a major contribution to overall mobility. As the chart shows, there were a large number of moves in both directions in most months. Monthly net migration is small relative to gross migration, as is annual net migration. Results from other cities also show a seasonal pattern. Figure 3.3 provides a similar analysis of moves to and from Cologne in 1912–13.13 Here the seasonal pattern is a rather diVerent one: there is net outwards migration in the June–August period, equivalent to 0.5% of the total population, and gains in the remaining months amounting to 2.9% of the total population. Again, monthly net migration is small relative to gross migration. 13

SJK (1914), 30; the moves out are not adjusted, so these may be underestimated.

The Pattern of Migration, 1870–1913





Net moves


Moves in −2

Moves out Jan.






Fig. 3.3. Net moves, moves out and moves in, as percentages of the total population, Cologne 1911–12 (averages of the two years)

It is also worth considering what aspects of migration are obscured or downgraded by the tendency to concentrate on net migration. Two studies are of interest. The Wrst is an analysis by the Berlin statistical oYce of the occupations of migrants in 1910. In this year there were 263,991 registered moves into Berlin and 206,552 moves out, an apparent gain of 56,839. Table 3.4a compares the breakdown by occupation of this Wgure with the distribution of total moves (in and out). In this year Berlin took in numbers of industrial workers, and these accounted for a third of net moves, but only a quarter of all moves. By contrast, the professional categories, oYcials, students, and workers in trade and commerce (a highly mobile group according to census returns) all occupy a larger share of total moves than net moves. The Cologne statistical oYce classiWed migrants in 1913 according to their last place of residence or intended destination. This again can be used to compare net and gross migration. The results (in Table 3.4b) show that net migration Wgures tend to be biased towards moves over longer distances, while the importance of moves within the locality, in this case the Regierungsbezirk Ko¨ln, is underestimated.

The Pattern of Migration, 1870–1913


Table 3.4a. A comparison of the distribution by occupation of net moves and total moves, Berlin 1910

Occupied in agriculture and Wshery Occupied in industry Occupied in trade and commerce Domestic servants General labourers Professional, state oVicials, and military personnel Pensioners and rentiers Students UnclassiWed

% share in net moves

% share in total moves

1.15 33.62 15.74 25.60 16.63 1.67 1.14 0.65 3.80

0.78 26.54 20.01 23.09 11.98 7.40 1.31 4.20 4.71

Note: Unoccupied family members are included, classiWed according to the occupation of the head of the household. Source: Calculated from SJB (1908–11), 215, which classiWed migrants into 43 occupational categories. These have been amalgamated for use in Table 3.4a.

Table 3.4b. A comparison of the distribution of net moves and total moves, by last place of residence or intended destination, Cologne 1913

Rest of Regierungsbezirk Ko¨ln Rest of Rhineland province Hesse-Nassau and Westphalia Rest of Germany

% share in net moves

% share in total moves

18.4 43.5 12.7 25.5

29.4 38.8 12.1 19.7

Source: Calculated from SJK (1913), 27.

To summarize, using Wgures for net migration concentrates attention on the movement of migrants over substantial distances into urban industrial jobs, but there are other short-distance moves which are also an important part of internal migration in the period: a student going to a large city to study; a doctor or oYcial moving to a diVerent post; a small trader trying his or her luck in a series of diVerent localities. Most of the rest of this study will be concerned with net migration, or moves revealed by the census comparison of place of residence with place of


The Pattern of Migration, 1870–1913

birth. But it should be borne in mind that these moves represent just the visible part of a much larger iceberg.

4. So u r c e s o f Mi g r a n t s Where did migrants come from? Did this change over time? The analysis of census data provides one way to answer these questions. By looking at cohorts, census returns sorted by the date of arrival in the city, it is possible to compare the origins of migrants in diVerent periods. Two points should be noted in advance. The Wrst, following from the discussion in the previous section, is that this will only capture permanent moves, or at least moves which had not been reversed by the time of the census. The second is that the analysis will be aVected if there are variations in death rates according to migrant origins. If migrants from the rural east, for example, live in poorer quality housing, and have lower life expectancy in consequence, then there will be a tendency for these to be under-represented in earlier cohorts. It is possible to make such an analysis by cohort of the 1910 Berlin census. This is shown in Table 3.5. In 1910 the Berlin census asked for the last place of residence prior to migration. This makes it possible to compare those migrants who moved directly from their place of birth to those who made an intermediate move. The Wrst point to note is the decline in the numbers of moves from the areas nearest to Berlin, especially the provinces of Brandenburg, Prussian Saxony, and Anhalt. These provided over half the migrants in the pre-1870 period, but only a third after 1900. There is a shift in favour of the eastern provinces in the period to 1890, but thereafter the east’s share declines. By 1906–10 migrants were drawn much more evenly. There were rises in the share contributed by the west and north-west, and by the other central provinces, and there was also a rise in the foreign contingent. There is a problem here: return migration could have the eVect of boosting the east’s share over time, as eastern migrants were less likely to return or make further moves. Thus, the eastern share will tend to rise for cohorts furthest from the census date. Fortunately, there are other ways to establish the trend of net migration and these will be used in a later section. Looking at the balance between migrants who moved from their place of birth, and those who had already made at least one move, it is

The Pattern of Migration, 1870–1913


Table 3.5. Analysis of the 1910 Berlin population by arrival cohort: percentage distribution of those born outside the city by last place of residence prior to migration (both sexes) Date of arrival in the city 1906– 1901– 1896– 1891– 1886– 1881– 1876– 1871– pre10 5 1900 5 90 5 80 5 1870 All moves: Regbez. Potsdam 16.5 16.0 16.4 Other Brandenburg/ 16.7 17.9 19.5 Pr. Saxony/Anhalt Other Centre 8.3 7.0 6.2 total centre 41.5 40.8 42.0 East 37.5 45.5 46.7 West and 10.6 6.9 5.8 north-west South and 3.1 2.1 1.7 south-west Foreign 7.3 4.8 3.7 Moved from place of birth: total 40.6 44.4 48.2 From towns and 29.2 34.4 39.4 rural areas From cities 8.7 8.1 7.3 Moved from somewhere other than place of total 59.4 55.6 51.8 From towns and 36.9 35.0 33.8 rural areas From cities 18.0 17.7 15.8

15.2 19.3

14.2 19.5

16.4 22.3

17.3 23.5

15.8 23.4

22.4 28.1

5.4 39.9 49.2 5.9

4.8 38.5 53.1 4.4

5.1 43.8 48.4 4.5

4.6 45.5 47.5 4.0

3.7 42.9 51.3 3.0

3.4 53.8 41.4 2.9













49.7 41.1

53.0 44.8

55.8 47.7

60.7 52.8

63.7 56.5

71.1 61.9

7.3 birth: 50.3 32.3






47.0 30.5

44.2 28.8

39.3 25.7

36.3 24.2

28.9 19.3







Source: These calculations were made using material from microWlm tapes kindly made available by Professor Dudley Baines, which are now in the library of the London School of Economics. These contain the full published record of the Berlin censuses 1861–1910. The results were summarized using the regional classiWcation set out in Appendix 2C. The data is drawn from table V.3 of the 1910 census, Berlin Statistisches Amt (1913–16), i. 32–9.

clear that there is a considerable change. Over 70% of the pre-1870 cohort were on their Wrst move. By 1906–10 this had dropped to below 30%. There is also a shift in the source of migrants, from smaller towns and rural areas to cities (the census data makes it possible to group migrants from 87 major towns and cities, the Berlin suburbs not


The Pattern of Migration, 1870–1913

incorporated in the city, and the industrial regions of Zabrze and Borbeck). But it should be noted that these urban areas only contributed 27% of migrants in 1906–10; the majority still came from rural areas and small towns. Comparison with the results of earlier censuses is made diYcult by the fact that these only recorded place of birth, not last place of residence. It is possible to take encouragement from the Wnding that most of those who arrived in the earlier years had come from their place of birth, but this is still an obvious source of discrepancy between the two data sources. Table 3.6 provides an analysis of the population resident in 1890, by place of birth and date of arrival. This shows a steady rise in the proportion coming from the eastern provinces. In the 1870s and 1880s about half of all migrants came from this area. At the same time there was decline in the proportion coming from the area around Berlin, which is particularly marked in the case of the nearby Kreise: Teltow, Niederbarmin, and Osthavelland. At this stage it would be premature to attempt to draw Wrm conclusions, but the pattern which is emerging suggests that the German Table 3.6. Cohort analysis of the Berlin population 1890: percentage distribution of those born outside the city by place of birth (both sexes) Date of arrival in the city 1881– 1871– 1861– 1851– 1841– 1831– 1821– pre90 80 70 60 50 40 30 1820 Nearby Kreise 6.3 Rest of Regbez. 9.4 Potsdam Other Brandenburg/ 20.7 Pr. Saxony/Anhalt Other Centre 3.5 TOTAL CENTRE 39.9 East 49.3 West and north-west 6.2 South and south-west 1.6 Foreign 3.1

6.4 11.2

7.4 13.5

10.2 16.1

12.3 17.8

16.4 20.7

20.0 21.5

22.6 20.9








2.6 43.3 50.3 3.9 0.8 1.7

2.3 49.8 45.0 3.5 0.5 1.2

2.5 59.6 34.9 4.1 0.5 0.9

3.2 65.3 28.6 4.3 0.8 1.0

3.0 70.4 24.2 3.8 0.7 0.9

4.2 73.8 20.4 3.3 0.9 1.3

5.7 74.0 15.2 5.1 2.4 4.9

Source: Calculated from data in Berliner Statistisches Amt (1893), table V.2; the regional classiWcation is set out in Appendix 2C.

The Pattern of Migration, 1870–1913


experience of migration during the period of industrialization can be divided into three periods: 1. Early (up to the 1860s): low mobility, most migrants drawn from the regions nearest to the cities. 2. Middle (from the 1860s to the 1890s): higher mobility, and a major shift of population out of the rural east. 3. Late (after the mid-1890s): a highly mobile, Xexible labour market, but one less dominated by the absorption of a labour surplus from the rural east. The cohort analysis made by the Berlin statistical oYce was also used to examine the extent to which migrants from diVerent regions made further moves after arrival. The basic requirement was an assessment of the probable death rate of a given cohort. After this had been deducted, a comparison of cohort numbers at diVerent census dates gave the percentages of those who remained and those who moved again. This was not an estimate of return migration as such, but return migration formed a part of the total of those moving. Ideally it would have been desirable to adjust for diVerences in likely death rates, but this information was not available, and the statistical oYce applied the same death rate to all groups (though diVerent rates were applied to each sex). Some error would have resulted from this. If migrants from the east, having lower wages and poorer living conditions than others, had higher death rates then there will be a tendency to overestimate the numbers moving: some of those assumed to have moved having, in fact, died.14 Table 3.7 provides a summary of this analysis, using the 1875, 1880, and 1890 censuses. The table shows that those who moved over short distances, coming from the adjacent areas of Brandenburg and Prussian Saxony, were less likely to make further moves. Of those who moved from more distant areas, those coming from the eastern provinces have much lower rates of further migration when compared to the rest of Germany. Two other points which emerge are the high rates of further migration for foreigners and the lower rates for women compared to men. 14 The view that the poor had higher death rates is well supported: see Spree (1988) for example. There is no evidence that eastern migrants had lower wages; nevertheless, given lower wage levels in the region of origin it can be assumed that reservation wage levels would be lower for eastern migrants facing similar costs of migration, and it is therefore probable that wages would have tended to be lower.


The Pattern of Migration, 1870–1913

Table 3.7. Estimates of the numbers making further moves after arrival, Berlin 1875–80 and 1880–90 % of each migrant cohort who move away from Berlin

Arrived from: Brandenburg/ Pr. Saxony East Rest of Germany Foreigners

Arrived pre-1875, moved in 1875–80

Arrived pre-1880, moved in 1880–90









27.6 43.6 51.0

16.8 23.4 35.6

25.4 46.4 53.3

19.0 32.1 44.7

Source: Calculated from data in Berlin Statistisches Amt (1893), 44.

This may appear a surprising result. The likelihood of a temporary move could be expected to be inversely related to distance: individuals might make short moves intending to return, to take advantage of unusual short-term employment opportunities, but over long distances the cost of the move would mean that most would be intended to be permanent. Instead the reverse seems to be the case. It might be that short-distance migrants are better informed, and therefore less likely to make mistakes, but examination of Figure 3.4, which gives the rate of further moves in 1880–90 by region, shows that there is no simple relationship with distance: the more distant provinces such as East Prussia, Alsace-Lorraine, and the Rhineland show very diVerent results. Figure 3.4 suggests that there are two factors which aVect the likelihood of further moves. The Wrst is what might be termed ‘cultural aYnity’: migrants from Alsace-Lorraine and from the Catholic areas of southern Germany were less likely to make permanent moves to Berlin. The same point applies to foreigners. The second is the income level of the region of origin. Migrants from the richer western and north-western provinces (Schleswig-Holstein, Hanover, Rhineland, and Westphalia) were more likely to make further moves. So this analysis shows that there was a fundamental diVerence between migration from the east and

The Pattern of Migration, 1870–1913













Fig. 3.4. Estimated numbers making further moves in 1880–90 by region: numbers moving again per 1,000 of those arrived in Berlin pre-1880 from each region (men only) Source: As for Table 3.7.

migration from other areas: the Wrst was a permanent shift of a population surplus in the sense that it represents the rectiWcation of an imbalance between the available labour force and perceived relative opportunities in the region of origin; the second was part of a general pattern of rising mobility. The east was the principal source of surplus labour for the Berlin labour market in the period. But there were other streams of longdistance migration, seasonal as well as permanent. Figures 3.5 and 3.6 illustrate this by making use of a survey of local authorities in the 1860s (at the Kreis level or equivalent). The administrators were asked about migration to or from their district. Many replied in general terms (‘much movement at harvest time’ or such like), but some identiWed particular destinations or sources, and sometimes gave reasons as well.


The Pattern of Migration, 1870–1913 Sweden - Harvest Peatcutting

Holland - Harvest

Peatcutting Sugar beet


Bohemia - Harvest

Italy - Harvest and brick-making

Fig. 3.5. Examples of long-distance migration connected with agricultural activities Source: Examples taken from survey replies given in von der Goltz (1875). There was substantial harvest-related migration within Germany over shorter distances; Bertold et al. (1990), 268 has a map of these movements.

The survey was not a comprehensive one, but it does provide a useful picture of movements in Germany at the time.15 Figure 3.5 gives some examples of long-distance migration connected with agricultural activities. It shows that, even in the 1860s, Germany was attracting foreign seasonal labour at harvest time, although in this period the main sources were Bohemia, Sweden, and Italy. There were also harvest-related movements inside Germany though these were mainly over relatively short distances. Mostly the movements were from 15 The authorities were not asked about the means of transportation used by these migrants. Some of these movements were suYciently short to have been possible on foot or using horse-drawn transport. Others were so long as to imply the use of the railway network.

The Pattern of Migration, 1870–1913


Areas reporting migration to railway construction

Fig. 3.6. Examples of long-distance migration connected with industrial or urban activities: areas reporting moves to Berlin or the Ruhr; areas reporting migration to take part in railway construction Source: As for Figure 3.5.

east to west, though there were some exceptions, such as the seasonal migration from the Eisenach region to eastern Silesia at harvest time. The sugar beet areas of central Germany were attracting seasonal migrants, although the authorities indicated that peat-cutting in these regions was also an important reason for migration to these areas. Figure 3.6 gives some examples of migration connected with industrial or urban activities. It shows, Wrst, those regions which speciWcally mentioned outward migration to Berlin or the Ruhr. The latter area was mainly drawing migrants from western or central Germany in this period (large-scale migration from the east came later). Berlin was attracting migrants from eastern Brandenburg, and parts of Silesia and Posen. These were, it appears, relatively concentrated streams, from a limited number of areas, compared to the more widespread eVect of railway construction. The Wgure shows the districts reporting major movements


The Pattern of Migration, 1870–1913

to railway construction. These were in predominantly rural areas in the south, the centre, and in the east. Comparison with the previous Wgure suggests that railway construction may well have Wtted in with the preindustrial pattern of seasonal migration at harvest time: young people from regions accustomed to provide harvest labour to other areas went to the railways instead. By contrast, moves to cities represented a more fundamental change in lifestyle and social surroundings. The transition from a pre-industrial to an industrial pattern of migration was not a sudden change for the supplying regions: there were several stages. Regions might start with a tradition of harvest-time outmigration, then move on to temporary migration connected with railway building or other major construction projects. It was then a possibility that a portion of those who made temporary moves for these reasons might decide to make permanent moves. This in turn would establish migrant cores, which could then expand as those making the initial moves were joined by friends or other family members. 5. D i f f e r e n c e s i n M i g r a t i o n b y S e x Many of the statistics on migration are available for both sexes. In this study these are amalgamated unless there is some evident diVerence which might be of interest, such as that shown in Table 3.6. Otherwise every table would be twice as long. There are indeed diVerences in the migration record between the sexes. In general men moved more. In 1901–11 in Berlin there were annually, on average, 301 moves per 1,000 for the male population, but only 222 moves per 1,000 for the female population (total moves to and from Berlin, Wgures adjusted by the statistical oYce). But this higher rate of gross migration did not mean that net migration was necessarily less for women. In fact in 1891–1912 the female population of Berlin gained proportionately more through migration: an average annual gain of 0.46% against one of 0.39% for the male population. Looking at the country as a whole, and dividing it into 17 fairly large regional groupings, the 1907 census found that 14.7% of men were living outside their region of birth, and 12.4% of women. This is a signiWcant diVerence, but one which shows that women were by no means immobile, even if they moved less often than men.16 16 Calculations from the analysis of the 1907 Occupational Census published in SJDS (1909).

The Pattern of Migration, 1870–1913


30 20 10 0 −10 −20 Men −30

Women 1891






Fig. 3.7. Net migration of men and women, Berlin 1891–1912, per 1,000 head of population of each sex (moves out adjusted by the statistical office) Source: Calculated from data in SJB (1913), 158.

Male migration was more volatile. Figure 3.7 compares annual net migration for each sex for Berlin during the period 1891–1912. The magnitude of the swings in male migration is clearly greater, and evidently related to economic conditions. For example, the so-called Elektroboom of 1898–1900, which was particularly marked in Berlin, produced a larger male inXow, and the collapse of this boom in 1901–2 led to a net outXow of men, but not to a net outXow of women. Looking at the historical record, there is evidence to show that female mobility rose more slowly than male. Figures 3.8 and 3.9 make use of the cohort analysis of the Berlin population in 1890 and 1910 given in Tables 3.5 and 3.6. In the Wrst of these, the proportion of migrants who came from the central provinces is compared for both sexes; in the second the analysis is repeated for the east. Two sets of lines are shown. As mentioned in the discussion in section 4 above, the census results are not comparable: the 1910 census classiWes by last place of residence prior to migration; the 1890 census classiWes by place of birth. So there are separate lines (A and B) for each census. The Wgures show that in the Wrst half of the nineteenth century there was a marked diVerence between the two sexes: female migrants were much more likely to be drawn from the central provinces, and less likely to come from the east. Female migration was predominantly

The Pattern of Migration, 1870–1913

78 80 Women (A)

70 Men (A)


50 Men (B) 40 Women (B)

30 1815










Fig. 3.8. Percentage of migrants drawn from the centre, by arrival cohort, Berlin 1815–1910 A 1890 census results classified by place of birth B 1910 census results classified by last place of residence Source: Figures given in Tables 3.5 and 3.6.

short-distance. As migration from the east rose, female migration lagged behind male migration, reaching the 30% mark about 20 years later, for example. But by the 1860s the gap had narrowed considerably. And from 1870 onwards there is no tendency for more female migrants to come from the central regions, and the eastern proportion is rather higher for female migration than for male. In general, the diVerences between male and female migration are not so large after 1870, at least when net migration is the main focus of interest. The main diVerences are in the pre-1870 period, and in the fact that men moved more often, and thus contributed a higher proportion of total moves. Young men made more frequent moves than young women, but women were nearly as likely to make permanent changes of residence.

The Pattern of Migration, 1870–1913


60 Women (B) 50

Men (B)


30 Men (A)

20 Women (A)

10 1815










Fig. 3.9. Percentage of migrants drawn from the east, by arrival cohort, Berlin 1815–1910 A 1890 census results classified by place of birth B 1910 census results classified by last place of residence Source: Figures given in Tables 3.5 and 3.6.

6. G e r m a n Em i g r a t i o n The Lewis Model deals with the operation of an economy seen in isolation, without considering the geographical context: How easy is migration across international frontiers? What are the conditions in neighbouring countries? Are wages in these countries lower or higher? In the model, the urban wage is determined by conditions in domestic agriculture: by the existence, or non-existence, of a labour surplus in this sector. But if there is migration from neighbouring countries with lower wages, then the whole wage structure will be inXuenced by conditions in these countries. And if there is substantial out-migration then


The Pattern of Migration, 1870–1913

the internal labour market will be aVected by developments in the recipient countries. German external migration underwent a major structural change around 1895. Comparison of census results with recorded births and deaths for the territories included in the post-1871 German empire show a total net loss of 2.34 million in 1841–71 (an annual average of 78,100) and 2.46 million in 1871–95 (an average of 102,500). But after 1895 this Xow came to an end. Between 1895 and 1910 the net loss was just 14,000, an annual average of less than a thousand. Calculations based on census returns may underestimate the extent of the change. By 1910 Germany was admitting large numbers of foreigners on temporary permits. Holders of these permits were required to leave the country by December, and then re-apply if they wished to continue to reside in Germany. As German censuses were carried out in December, those who complied with this regulation would have been omitted from the census total.17 A census taken in some other month would have shown a larger population, and the calculated rate of migration into Germany would have been greater. Germany may well have already been a net gainer through migration.18 If so then this was an unusual development. Of the other European countries, only Belgium was a net gainer before 1914. Britain did not start to gain population through migration until the 1970s. This change had the eVect of reinforcing the links between Germany and the other labour markets of central and eastern Europe, and decoupling Germany from the converging labour markets of the ‘Atlantic economy’. Discussion of German emigration in the nineteenth century is dominated by German emigration to the United States. Partly this is due to the nature of the statistical sources. There are records of German overseas migration, both from the ports of departure and from the recipient countries, but there is very little on movements across land frontiers. In 1912 the German statistical oYce made a survey of foreign censuses, to establish the numbers of Germans living outside the Reich. The Wgures are given in Table 3.8. The statistics are not fully comparable: some countries recorded all those born in Germany; others only included those who still retained German citizenship. Some countries which lack suitable census data are excluded: Brazil is the most important of these. 17 Those who chose not to comply with the regulation would have had a good reason to evade the census. 18 Return migration was also important: see Schniedewind (1994) for a study of this. Other studies of German migration include Bade (1980 and 1991) and Marschalck (1973).

The Pattern of Migration, 1870–1913


Table 3.8. Germans living in foreign countries 1910–11 United Statesy Switzerland Austria-Hungary France Belgium

2,501,000 220,000 126,000 90,000 57,000

Britain and Ireland y Netherlands Denmark Australiay




38,000 35,000 33,000

Luxembourg Chile Italy

22,000 11,000 11,000


Countries who recorded all those born in Germany, all others recorded only German citizens. Source: Kaiserlich Statistisches Amt (1916), 3*.

The pre-eminence of the United States as an emigrant destination is evident. But quite large numbers went to neighbouring states, although these in return provided migrants to Germany, reducing net migration. The most substantial Xow of net migration out of Germany to a European destination was to Switzerland (there were 68,000 Swiss living in Germany in 1910). The United States accounted for 91% of recorded overseas migration in 1871–1913. Figure 3.10 shows that there were years when this fell to 80% or below. Typically these were years when migration to the United States was low, but there is no evidence that migrants chose other destinations when conditions were unfavourable for migration to the United States (migration to the United States was not negatively correlated with Xows to other destinations). Towards the end of the period there is some evidence that German migrants were looking for alternative destinations. There was a rise in migration to Canada and Argentina. However, the Xows were not large compared to the movements to the United States in the period up to 1895. By contrast, after 1900 British emigration switched to the Empire, which took 65% of those leaving in 1911–13 (up from 28% in 1891–1900).19 Comparison of migrant destinations with the regional origins of migrants is possible from German Wgures from 1899 onwards. Figure 3.11 gives migration to the United States as a percentage of all recorded migration to non-European destinations, 1900–9. The main areas from which there was migration to alternative destinations were a group of central regions—Saxony, Prussian Saxony, and Berlin-Brandenburg. 19 British Wgures from Thomas (1973), 57; the German Wgures are calculations using Wgures for migrant destinations from SJDR.


The Pattern of Migration, 1870–1913



60 Asia 40

Africa Australia


Other North and South America Brazil

0 1871 1876 1881 1886 1891 1896 1901 1906 1911

United States

Fig. 3.10. Percentage distribution of recorded migration to non-European destinations (from German sources) 1871–1913 Note: The German figures initially covered only migration from German ports and Antwerp. Later this was expanded to include migration from French ports. Migration from British ports was not included, which affected the figures for migration to Australia. Source: Data from SJDR (1885), table 5, p. 23 and subsequent issues.

This group includes some important industrial centres. At the other end of the scale, in the agricultural regions of eastern and southern Germany there was little migration to destinations other than the United States. The result is that there is a positive correlation between the percentage occupied in agriculture in each region, and the percentage going to the United States (a coeYcient of 0.56), indicating that the Xow to the United States was more ‘agricultural’ than the Xow to other destinations.20 20 The negative correlation between the percentage going to Brazil (the next most important destination) and the percentage occupied in agriculture is even stronger (a coeYcient of 0.72). This Wnding that the ‘more advanced’ regions sent migrants to the ‘less advanced’ destination is intriguing although perhaps not of any great signiWcance. It has a parallel in the better-known division between northern and southern Italy (the south was more likely to send migrants to the United States), Hatton and Williamson (1998), ch. 6.

The Pattern of Migration, 1870–1913



Fig. 3.11. Migration to the United States as a percentage of all migration to non-European destinations 1900–9, by region Source: Calculated from data in SJDR (1901–10).

However, the basic conclusion has to be that the story of German emigration is, in aggregate, the story of German migration to the United States. There was no other Xow which was large enough to have a signiWcant eVect on labour market conditions inside Germany. The record of migration to the United States is presented in Figure 3.12. Only moves from Germany to the United States are given; there are no Wgures for return migration before 1909. The chart includes two series for German migration to the United States, one from German sources and the other from US immigration records. The two sources agree relatively well. Between 1890 and 1900 there is little diVerence. Before 1890 the discrepancy is probably due to gaps in the German records. After 1900, the diVerence is less easy to explain. The US authorities may have classiWed a number of Germanspeaking immigrants from Russia as Germans. This could explain why

The Pattern of Migration, 1870–1913

84 300

Emigration: US data US−German wage ratio 200

Emigration: German data


0 1820 1830 1840 1850 1860 1870 1880 1890 1900 1910

Fig. 3.12. Annual German emigration to the United States (in 1,000s) and the US-German wage ratio Source: Migration figures from US Department of Commerce (1975), and SJDR, annual issues; wage ratio calculated from Williamson (1995), 164–7. The source for the German wage data is a series for unskilled building wages in Rostock, Nuremburg, and Berlin, originally collected by Robert Kuczynski, also published in Kuczynski (1945) and Bry (1960).

the discrepancy rose after 1900. Some travellers, intending to stay for a limited period, may also be classiWed as immigrants in the US Wgures.21 The migration Wgures show three main peaks: in the early 1850s, in the late 1860s and early 1870s, and in the period 1881–93. After this German emigration to the United States remained at a low level. These changes can be linked to the relative movement of the German and US economies: emigration was high when conditions were favourable in the United States or unfavourable in Germany. But after 1895 this relationship becomes much weaker: the collapse of the Elektroboom in 1900–1 produced only a modest rise in emigration, for example.22 21 There are large discrepancies in statistics of German migration 1900–13 for all the major non-European destinations. These are discussed in Burgdo¨rfer (1926). One problem was German emigration through British ports, but there is no obvious reason why this should only have caused diYculty after 1900. 22 Grant (2002d ) contains a detailed analysis of the decline of German emigration to the United States.

The Pattern of Migration, 1870–1913


Figure 3.12 also gives the US–German wage ratio, calculated from Williamson’s series in purchasing power parities. There was a major disturbance to this series in the 1860s, as a result of the American Civil War. Once allowance is made for this, it is clear that there is no evident tendency for wages in the two countries to converge. Germany was an exception to the general tendency for wages in the Atlantic economies to move closer. One reason for this was that declining emigration to the United States, combined with increasing immigration from eastern and central Europe, had the eVect of strengthening links with other oversupplied labour markets in these regions, and weakening those with the converging Atlantic economies.23 German emigration to the United States declined for a number of reasons, due to changes within Germany and within the United States. One factor was increased competition within the United States from sources of ‘new immigration’: countries in southern and eastern Europe which supplied an increased proportion of US immigrants from the 1890s onwards.24 Another was a reduction in the availability of land in the United States which suited German farming skills as the frontier of settlement moved into more arid western regions. But a further possible reason is that Germany was moving out of the labour surplus phase when migration, internal and external, was dominated by the release of surplus agricultural labour. German emigration had a strong agricultural character. Figures 3.13 and 3.14 show the variation in the intensity of migration by region, for two periods: 1871–9 and 1890–9. The regional rates of emigration (per head of resident population) are expressed as percentages of the national average. The variation between regions is striking: in 1871–9 the average annual rate of emigration was 0.44 per 1,000 in East Prussia, and 4.53 in West Prussia, ten times as large. In the 1870s the highest rates of migration were in the rural east, in West Prussia, Mecklenburg, and Pomerania. There were also high rates in the north-west, and in the south-west, which had been an area of high emigration earlier in the 23 Important studies on convergence in the international economy of the late nineteenth century include O’Rourke et al. (1994 and 1996) and Taylor and Williamson (1997). Although Germany contributed large absolute numbers of migrants in the years before 1895, the numbers were not as high relative to the size of the German population as for Ireland and some Scandinavian countries. This meant that the convergence process was not a strong one even in the years when emigration was high. 24 In the literature on US immigration these are the ‘new immigration’ sources, to distinguish them from the traditional sources: Britain, Ireland, Germany, and the Scandinavian countries.


The Pattern of Migration, 1870–1913

>300 250−300 200−250 150−200 125−150 100−125 80−100 60−80 40−60 20−40 300 250−300 200−250 150−200 125−150 100−125 80−100 60−80 40−60 20−40 1










Fig. 3.15. Foreign-born workers as a percentage of all those occupied in mining and metals, quarrying, and construction, 1907 Source: 1907 occupational census, SDR (1908), n.f. 210.


SDR n.f. 210 (1908).

The Pattern of Migration, 1870–1913



Fig. 3.16. Foreign-born workers as a percentage of all those occupied in agriculture, June 1907 Source: 1907 occupational census, SDR (1908), n.f. 210.

workers is more or less as expected (Figure 3.16). The highest Wgure was the 11.3% of the Mecklenburg agricultural labour force which was foreign-born. Next came Prussian Saxony-Anhalt with 8.1%. Although the conventional view may be that the typical pre-1914 migrant was a Russian Pole working in eastern agriculture, this shows that a typical migrant was just as likely to be someone from Austria-Hungary working in a quarry or on a building site in south-western Germany.

8. M i g r a t i o n a n d L a b o u r Su r p l u s The previous section has stressed the growing links between Germany and the labour markets of eastern Europe, particularly Russia and Austria-Hungary. In combination with the decline in emigration to the United States, which weakened the connection between Germany and


The Pattern of Migration, 1870–1913

the converging Atlantic labour markets, this development reinforced the ‘Continental’ nature of the German labour market, insulating it from the forces driving the process of convergence in the Atlantic economies. It also meant that one source of elastic labour supplies—the labour surplus in German agriculture—was supplemented by another—migration from neighbours to the east. This in turn held back the increase in German wages. This was an inevitable consequence of the eastern movement of the process of industrialization in Europe. Both inside Germany and outside Germany industrialization led to increased mobility, traditional institutions which promoted immobility were eroded, and improved communications made migration attractive and aVordable. In the last quarter of the nineteenth century these same forces were unleashed in the countries to the east and the south of Germany. This in turn set oV a new wave of migration to the United States, displacing potential German emigrants, and also increased migration across Germany’s land frontiers to the south and the east. A theme which runs through the chapter is the identiWcation of periods within the 1870–1913 timespan covered by this study, and the characterization of these periods. Evidence presented points to the importance of the years around 1895 as a dividing point. After this external migration is greatly reduced, and the rate of migration out of the east falls. There is also a shift in the characteristics of the typical migrant, from being someone with no previous experience of urban life, to a person who had already moved at least once, and who was more likely to have come from another city. The importance of the 1895 caesura is demonstrated by Wgures which will be presented in Chapter 9. These show that the Kuznets Curve, the relationship between income inequality and economic development, peaks around these years. The years up to 1895 are characterized by rising inequality and high net migration: the labour surplus phase of the Lewis Model. After 1895 there was a highly mobile labour market, but static or falling inequality; Germany was moving out of the labour surplus phase.

A p p e n d i x 3A. A N o t e o n R e g i o n a l U n i t s The regions used in pre-1914 German statistics are a source of some diYculty in that they vary considerably in size, and the boundaries

The Pattern of Migration, 1870–1913


Table 3A.1. Regions and regional groupings Regional grouping

Regional unit used in this study

Includes the following regional units used in German statistical publications


East Prussia West Prussia Pomerania Posen Silesia Mecklenburg

East Prussia West Prussia Pomerania Posen Silesia Mecklenburg-Schwerin and -Strelitz


Berlin/Brandenburg Pr. Saxony/Anhalt Thuringia

Berlin and Brandenburg Provinz Saxony and Anhalt Saxony-Weimar, -Altenburg, -Meiningen, and -Coburg-Gotha, both Reub and both Schwarzburgs Saxony (Kingdom of)

Saxony West and north-west


Hesse-Nassau Rhineland Hesse

Schleswig-Holstein, Hamburg, and Lu¨beck Hanover, Oldenburg, Brunswick, and Bremen Westphalia, Waldeck, and both Lippes Hesse-Nassau Rhineland Hesse

Bavaria excl. Pfalz Pfalz Baden Wu¨rttemberg/ Hohenzollern Alsace-Lorraine

Bavaria excl. Pfalz Pfalz Baden Wu¨rttemberg and Hohenzollern Alsace-Lorraine

Hanover/Oldenburg/ Brunswick Westphalia

South and south-west


The Pattern of Migration, 1870–1913

reXect a historical process: Prussian expansion leading to eventual German uniWcation, not any Napoleonic striving for geographical logic. Therefore, all the statistics in this study are presented in tables or Wgures which use regional units designed to eliminate the smaller regions. On occasions, these are then grouped into four main regional categories: east, centre, west and north-west, and south and south-west. The basic system is shown in Table 3A.1. In addition, wherever possible, Oldenburg is split: the separate enclaves of Lu¨beck and Birkenfeld being included in Schleswig-Holstein and Rhineland respectively. For some purposes, chieXy agricultural, the term ‘east-Elbian’ is used: here Berlin/Brandenburg is added to the other eastern provinces. For the tables and maps presented in this study a certain amount of additional work was required to produce results which conformed to these boundaries. The Wgures could not be taken directly from oYcial publications.


Migration in Germany 1870–1913: A Statistical Analysis 1. I n t r o d u c t i o n: A N o t e o n Me t h o d o l o g y a n d So u r c e Ma t e r i a l s The study of migration has been one of the most active areas of German historical research in recent years. This has produced a number of excellent local studies, making use of regional archives. These have described the eVects of large-scale migration: the process of assimilation, the problems faced by rapidly expanding cities, the process whereby an industrial workforce was created from a stream of rural migrants.1 The response of the German state and the upper echelons of German society to these problems has been examined, and, at times, found wanting. Studies have repeatedly shown pre-1914 Germany to be a society of high inequality, and limited social mobility.2 The aim of this book is not to produce another local study. It is instead to try to grasp the process in its entirety: to put migration in a theoretical context which explains the origins of the problems described by the regional studies, and to answer the question ‘Why is migration in this period so important?’ The Lewis Model, the concept of a ‘labour surplus’ phase in economic development, provides such a context, and the application of this model is therefore a central theme. The remit of the study is therefore a wide one. The aim is to test certain hypotheses about the causes and consequences of migration. The level is the national one: not what was happening in certain cities or regions, but the eVect on Germany as a whole. And, wherever possible, hypotheses have been expressed in quantiWable terms. 1

Examples include Barfuss (1985), Bo¨lsker-Schlicht (1997), Del Fabbro (1996), Jackson (1997), Ko¨llmann (1960), Kosters-Kraft (1998), Lourens and Lucassen (1999), Oberpennig (1998), Reich (1998), Wenneman (1997), Crew (1979), Catt (1986), and the various studies in Hoerder and Nagler (1995), Ro¨bler (1995), and Bade (1987). Studies of migration with a wider remit include Matzerath (1985), Caumanns (1994), Ko¨llmann (1959), Tipton (1976), and Hochstadt (1999). 2 Kaelble (1986).

Migration 1870–1913: A Statistical Analysis


This is not because this is regarded as the only way to approach the subject, or as necessarily superior to other techniques of historical enquiry; it is because this is a type of study not currently available. The idea is to produce a work which is complementary to the regional and local studies: for example the collection published as People in Transit.3 Moreover, the main emphasis has been on the use of statistical sources available in nineteenth-century government publications. This period was one in which statistical enquiry was developing, both in scope, in the areas covered by published statistical series, and in the use that was made of the results. There was a belief that the proper use of statistics could lead to a better understanding of social problems and, at times, produce solutions. The classic case was the use of mortality tables to prove the connection between breast-feeding and low infant mortality.4 The new German empire had an Imperial Statistical OYce, which oversaw the collection of national statistics and the administration of censuses. There were also state statistical oYces, which collected a lot of the data and conducted enquiries of their own. The most important was the Prussian Statistical Bureau, headed by Ernst Engel in the 1860s and 1870s, but others also made interesting contributions: the Saxon Statistical OYce made a series of studies of income distribution in the 1900s which were unusually comprehensive for the time. In addition, all the major cities had statistical oYces, and typically published a statistical yearbook. The most important of these was the Berlin Statistical Bureau, whose director in the years before the First World War was Robert Kuczynski. The analysis of these materials used the techniques available at the time: where causation was suspected a series was ranked by one variable, and results given for a second variable. This could show a rough correlation: if the eye running down the page could pick up a tendency for the second variable to rise or fall. But regression analysis was not used, and any relationship involving more than one explanatory variable could not be studied. The result is a vast amount of statistical material presented in a form which is not easy to understand or digest. Taking one example: the 1907 Zeitschrift of the Prussian Statistical Bureau contained an article by Dr M. Broesike on internal migration in Prussia which included sixteen of these ‘visual correlation’ tables, some of which extended over two or more pages, and contained as many as 489 observations (an 3

Hoerder and Nagler (1995).


Berlin, Statistisches Amt (1893), 82.

Migration 1870–1913: A Statistical Analysis


example is shown as Figure 4.1).5 There were also a number of other tables containing basic statistical material. This was a barrier to comprehension greater, in most respects, than the obstacle presented by the terminology of modern econometrics. So the approach used in this study has been to seek, in the Wrst place, to achieve a thorough knowledge of the materials collected and made available by the various statistical oYces, and secondly to subject these to examination using modern econometric analysis. For the general reader the statistical results have been summarized in a series of tables which are included in the main text of this chapter,

Fig. 4.1. Part of a ‘visual correlation’ table from Broesike (1907) Note: This is part of a table covering 489 Prussian Kreise. Broesike has classified them by the dominant farm size; the first column gives the percentage of the total agricultural area in holdings of this size. He is looking for a relationship between this and rates of migration for 1895–1900 and 1900–5 (the two columns at the right). He also includes a classification (Kb or Gb) according to the next most important farm size class, and follows this with the percentage of the total agricultural area in holdings of this size. So, the reader is asked to look down the two right-hand columns, and consider if migration tends to rise or fall with the dominant farm size, keeping in mind the possible effect that might flow from the farm size sector ranked second. 5

Broesike (1907).


Migration 1870–1913: A Statistical Analysis

and in subsequent chapters which also make use of econometric analysis. These provide a ‘verbal exposition’ of the econometric results. A more detailed analysis is then presented in appendices, which may be of interest to those readers who are well versed in statistical matters. They can, however, be skipped without losing any essential part of the argument presented in each chapter.

2. T he C h a n g in g N a t u r e o f M i g r a t i o n , 1871–1913 The basic hypothesis considered in this study is that Germany went through a ‘labour surplus’ phase of economic development, and that this phase was coming to an end by the late 1890s. The shift to a more modern pattern of migration post-1900 created conditions which were favourable to political stabilization and the transition to a mature democratic society. The Lewis Model provides the theoretical basis for these claims. One of the best sources of Wgures for migration is the indirect calculation of net migration by region based on the comparison of census returns. For each census year from 1871 to 1910, the Imperial Statistical OYce compared the actual increase in population by region with the recorded numbers of births and deaths since the last census (generally Wve years before). Deducting the recorded natural increase from the actual increase produced a Wgure for net migration. This was done for eighty-one regions and eight periods, thus producing a panel data set of 648 observations. The reliability of these calculations will depend on the accurate recording of births and deaths. By and large, this was not unsatisfactory in Germany in the period. The main question raised by demographers concerns the recording of stillbirths in the Wrst half of the century, but this is not a problem as it does not aVect the overall rate of natural increase (the issue is whether these were recorded as a birth followed by a death or left unrecorded).6 One obvious problem is that the destination of migrants is not recorded by this procedure. Other data sources will be examined in the next section which do provide this information. It is, however, possible to make the same calculation for Germany as a whole, which provides a


See the discussion of this in Knodel (1974).

Migration 1870–1913: A Statistical Analysis


Wgure for net external migration. This gives an idea of the balance between external and internal migration. In 1900 the Imperial Statistical OYce extended the analysis back to 1841. In this pre-uniWcation period censuses were organized by the diVerent states, and did not always take place in the same year. In addition, boundary changes imply that a calculation of net external migration, for the territories included in the post-1871 state, will include a certain amount of internal migration within French and Danish territories (pre-1871 and pre-1864 respectively). For these reasons the pre-1871 Wgures should be treated with some care, although they are probably reliable for Prussian territories. The overall record 1841–1910 is given in Table 4.1. The table uses four regional groupings. Of these two are consistent sources of net outwards migration: east, and south and south-west. But the trajectories are very diVerent. The east was not a major source of net outwards migration prior to 1861. Indeed, parts of the east were gaining through migration: East Prussia and West Prussia both made gains in 1851–61. Thereafter, outwards migration rises until, in the decade 1880–90, the region is losing 1% of its total population annually

Table 4.1. Net migration 1841–1910, in average annual rates per 100,000 of the resident population

1841–51 1851–61 1861–71 1871–5 1875–80 1880–5 1885–90 1890–5 1895–1900 1900–5 1905–10



West and north-west

South and south-west

All Germany

55 65 342 738 431 959 1038 661 815 662 663

þ37 98 þ22 þ323 þ48 þ35 þ490 þ99 þ351 þ316 þ225

273 331 260 þ119 99 252 þ167 þ50 þ449 þ285 þ217

301 433 284 365 203 559 268 259 60 113 192

167 243 229 192 174 427 137 177 þ35 þ18 51

Source: Pre-1871 data from SDR n.f.150, pp. 190*–193*; post-1871 data from SDR a.f. ¨ bersicht 1B, pp. 50–2 (1871–5 Wgures) and subsequent census publications, 25(2) U combined in the regional units given in Appendix Table 3A.1.


Migration 1870–1913: A Statistical Analysis

through migration. This rate of out-migration then falls back in the decade after 1900. In the south and south-west, by contrast, outmigration was already substantial in the pre-1871 period. There is a continuous outXow, with Xuctuations, until 1895, although the rates are well below those in the east. After 1895, the rate of out-migration is much reduced. The experience of the two gaining regions is also diVerent. In the pre-1871 period the west and north-west was also a major source of out-migration. This was a period in which this region contributed heavily to German migration to the United States. But after 1871 this changed. There are periods of in-migration as well as out-migration. After 1885 the region consistently absorbs migrants, post-1895 this is on a larger scale. By contrast there is consistent migration into the central region; the only period in which outwards net migration is recorded is 1851–61. The rate of absorption Xuctuates, with peaks in 1871–5 and 1885–90, but it is more consistent after 1895. The story in summary is that, in the early part of the century, there was a substantial labour surplus in western Germany, north and south. This was chieXy exported, although some was transferred to the centre and east. In the rural east the amount of land under cultivation was still rising so any local surpluses could be mainly absorbed within the region. From the 1860s onwards industrial growth in the west and the centre meant that more migrants could be found employment within these regions, but at the same time the east emerged as a major source of out-migration. Before 1895 the rate of progress of the industrial regions was not suYcient to take in all the migrants out of the east and the south, resulting in a continuing high rate of emigration for Germany as a whole (Wnal column of Table 4.1). Statistical analysis of the migration Wgures can be used to extract additional information from the census results.7 Although the published census returns do not distinguish between agricultural and nonagricultural areas, the use of modern econometric techniques can produce estimated rates of net gain and loss for each period. These are given in Table 4.2. The table shows that there were four periods in which the net rate of out-migration from agricultural areas exceeded 2.0% per annum, and 7 Results of the regressions used to make these calculations are given in Appendix 4B, which also contains a discussion of the statistical issues involved, and further comments on the pattern of migration revealed by this analysis.

Migration 1870–1913: A Statistical Analysis


Table 4.2. Results of statistical analysis of net migration by region, 1871–1910: rates of net gain or loss through migration for agricultural and non-agricultural areas (annual rates of gain or loss as percentages of the total population)

1871–5 1875–80 1880–5 1885–90 1890–5 1895–1900 1900–5 1905–10

Estimated rate of net loss for an entirely agricultural region

Estimated rate of net gain for an entirely non-agricultural region

2.33 1.25 2.50 2.89 1.64 2.48 1.58 1.55

þ1.60 þ0.73 þ1.01 þ1.65 þ0.68 þ1.34 þ0.77 þ0.61

these also produced annual rates of gain for the non-agricultural areas of at least 1.0%.8 After 1900, the rate of migration drops to one which produces a signiWcantly lower rate of gain for the non-agricultural regions: the rate of gain for 1905–10 is the lowest of all. This does not mean that the overall level of migration had necessarily dropped; rather it had changed its character from one dominated by the transfer of surplus labour out of agriculture, to a pattern where this was one stream amongst others, although still important. One way of demonstrating this change is to consider what percentage of the pattern of net gains and losses can be explained purely by the extent to which the region specializes in agriculture. In a ‘labour surplus’ period a large proportion of migrant moves are from agricultural to non-agricultural areas, and most of the variation in net gains and losses by region is due to this factor. In other periods transfers between industrial regions due to the relative performance of diVerent industries will be an important inXuence and the role of rural–urban migration will be reduced. Table 4.3 gives Wgures produced by the statistical analysis of the census results, which show how German internal migration changed over the period covered by this study. 8 The rate of gain is lower relative to the size of the non-agricultural population for two reasons. First, the total non-agricultural population was larger. Secondly, some of those leaving agriculture emigrated and so did not transfer to non-agricultural regions inside Germany.


Migration 1870–1913: A Statistical Analysis Table 4.3. ‘Rural labour surplus’ migration as a percentage of all migration (the percentage of the variation in net gains and losses which can be explained by the proportion occupied in agriculture) 1871–5 1875–80 1880–5 1885–90 1890–5 1895–1900 1900–5 1905–10

44.2 43.5 57.1 63.9 39.8 48.9 33.1 26.9

Note: These are the R2 statistics from the regressions given in Appendix Table 4B1.

The table shows that, before 1900, labour surplus migration was the most important factor in internal German migration. Even in periods when the rates of net gain and loss were reduced, such as 1875–80, the importance of the ‘labour surplus’ component remained high. The peak for labour surplus migration came in the 1880s: in this period over 50% of the variation can be explained by this one factor: was the region agricultural or not? The 1890s are a transitional period, with a less pronounced agricultural character. By 1900–5 only a third of the pattern of net gains and losses can be explained by the proportion occupied in agriculture, and by 1905–10 this has dropped to just over a quarter. The labour surplus period was over. So, what happened in the 1890s that brought about this change? Was there a static lump of surplus labour in agriculture which had simply moved out by then? Or were there other changes which improved the prospects of the agricultural sector? Some Wgures on these issues are brought together in Figure 4.2. All are taken from HoVmann. The two indices of arable producer prices and imported cereal prices show the eVect that world market conditions had on German producer prices. World market prices fell from the 1870s onwards and this brought down German producer prices despite tariV protection. But from the mid-1890s onwards world market prices began to recover, and this in turn brought about a recovery of German arable producer prices. The consequences of this can be seen from the movement of the index of land prices. The index fell slowly from the late

Migration 1870–1913: A Statistical Analysis


200 Wage ratio

180 160 140 120 Import prices 100 80

Arable prices Land prices

60 40 1870









Fig. 4.2. Indicators of agricultural conditions, 1870–1913 (all indices have 1913 ¼100) ‘Wage ratio’ is the ratio of industrial to agricultural wages; ‘arable prices’ is an index of arable producer prices; ‘import prices’ is an index of the price of imported cereals. Source: Calculated from data in Hoffmann (1965), 492–5, 561–2, 569–70, and 610–11. Hoffman’s index of agricultural workers’ earnings is based on figures for cash payments, but figures given in Pohl (1908) also show gains for workers paid in kind or by harvest shares. Earnings of Insteleute with threshing shares in Oletzko (East Prussia) rose 73% between 1880 and 1903 for example.

1870s, when it was over 67, to a low point of 59.6 in 1895. From then onwards it was rising, the level reached in 1913 representing a rise of 68% in eighteen years. The uppermost line in Figure 4.2 gives industrial wages as a percentage of agricultural. In the early 1870s industrial wages were 40–50% above agricultural wages, and dipped below this in the late 1870s. Thereafter the ratio rose and remained in the range 170–180% from the late 1880s onwards. The improved agricultural conditions post1895 did not lead to a fall in this ratio, but agricultural wages did keep pace with industrial wages during this period.


Migration 1870–1913: A Statistical Analysis

The movement of this ratio can also be interpreted in the light of the migration patterns considered in the previous chapter. In the midcentury, the industrial regions were drawing migrants predominantly from nearby rural areas. Consequently, the premium required to persuade migrants to come from these areas needed to be only a moderate one. But by the late nineteenth century the premium had to be suYcient to persuade migrants to move over much longer distances. It is noticeable that the premium did not fall back after 1900, even though agriculture was then doing better. Industry still had need of longdistance migrants.

3. A n a l y s i n g N e t M i g r a t i o n b y R e g i o n, 1871–1910 Scrutiny of the pattern of net migration shown in Table 4.1 in the light of the evidence on changing agricultural conditions given in Figure 4.2 shows that there is a possible connection between migration and agricultural conditions: the improvement in land prices after 1895 appears to have had some eVect, reducing the rate of migration out of the east for example. This raises the possibility that a more precise modelling of the eVect of changing agricultural conditions on migration might produce an equation which can be estimated econometrically. The value of this does not lie only in giving precise values to relationships which can be guessed at from visual inspection—it also makes it possible to compare diVerent causes of migration. There are two issues which are of particular interest. The Wrst is the eVect of agricultural structure, the extent to which the prevalence of large estates had an eVect on migration. The second is the relationship between migration and demographic variables: how far was the regional pattern of migration caused by diVerences in rates of population increase? Both these issues will be examined in greater depth in subsequent chapters; the aim here is to show the part they played in the overall pattern of migration. The full set of econometric results is given in Table 4B.2 in the Appendix; Table 4.4 provides a summary of the results. The analysis shows that demographic factors had an important inXuence on the pattern of migration: regions with high birth rates tended to have high rates of out-migration; regions with high death rates took in migrants to replace these losses. The birth rate which matters most is the one which aVects current levels of entry onto the regional labour

Migration 1870–1913: A Statistical Analysis


Table 4.4. Summary of results of statistical analysis of net migration for 81 regions, Wve-year time periods, 1871–1910 I. Secure statistical results Factors which increase net outwards migration: High birth rates in the past Presence of large farms and estates Increases in the rural–urban wage gap Factors which decrease net outwards migration or lead to net gains from migration: High death rates in the current period High rates of non-agricultural employment II. Less secure statistical results Factors which increase net outwards migration: Low land prices Note: The classiWcation of results as ‘secure’ or ‘less secure’ is based on the analysis and discussion given in the Appendix. The assessment is based on the usual statistical tests of signiWcance, plus the ‘robustness’ of the results—the extent to which a coeYcient is aVected by changes in the form of the estimated equation or the inclusion of additional variables.

market. It is the lagged birth rate—the birth rate for 20 years before— which is used in the estimated equations. The estimates conWrm that the level of agricultural occupation was an important inXuence on migration: agricultural regions tended to shed population and non-agricultural regions made gains. But they also show that agricultural structure mattered: the presence of large farms and estates tended to increase out-migration. This question will be one of the main themes of Chapter 6, but the result appears to conWrm a point made by a number of contemporary authors, including Max Weber.9 The eVect of changing agricultural conditions on migration can be estimated as part of the overall procedure: eVectively this asks whether migration out of agricultural areas was higher in bad periods for the agricultural sector—when land prices were low or the rural–urban wage gap was high. The analysis shows that times of low land prices were indeed periods in which migration out of rural areas tended to be 9 It was one of the major themes of Weber’s study of the conditions of the agricultural workers east of the Elbe, Weber (1893).

Migration 1870–1913: A Statistical Analysis


higher, but this result is not a particularly secure one in statistical terms. A stronger result is obtained when the eVect of changes in the rural– urban wage gap is included in the estimated equations: upwards movements in this gap led to increased out-migration; falls held back the Xow out of rural areas. How important are these variables in the explanation of the overall pattern of migration? One way to approach this is to try to explain migration between the four main regional groupings. An important caveat should be noted: there is a lot of migration within these units which is not covered by this analysis. Table 4.5 compares the eVect of predicted migration using the coeYcients calculated for the two demographic variables with the actual Wgures for net migration previously given in Table 4.1.10 The results show that demographic factors contributed to out-migration from the east, though the eVect is not large in relation to the total Xow. But the

Table 4.5. Demography and migration: the estimated eVect of demographic variables on net migration by regional grouping (columns headed a) compared with actual net migration, annual rates per 100,000 (columns headed b) East

1871–5 1875–80 1880–5 1885–90 1890–5 1895–1900 1900–5 1905–10


West and north-west

South and south-west









24 27 29 34 41 46 48 58

738 431 959 1038 661 815 662 663

15 17 21 28 40 52 58 66

þ323 þ48 þ35 þ490 þ99 þ351 þ316 þ225

14 19 27 33 43 52 55 60

þ119 99 252 þ167 þ50 þ449 þ285 þ217

0 1 16 26 36 44 43 44

365 203 559 268 259 60 113 192

Note: The equation used in these calculations was NETMIGjt ¼ 0:45NETMIGjt1 þ 0:26 DEATHRTjt 0:23 BIRTHRTjt4 (the j subscripts denoting the use of regional groupings); these coeYcients are from column a of Appendix Table 4B.2. 10


The calculation is a simulation which allows for the eVect through the NETMIGkt1

Migration 1870–1913: A Statistical Analysis


same eVect should have led to a similar level of out-migration from the centre and the west, which in fact gained through migration. In the case of the south and south-west, there was a very small eVect from the demographic variables in the period of highest out-migration (1871–85) and a larger eVect in the period of lower out-migration after 1895. Some of the reasons for this can be seen from the comparison of the various regional groupings given in Table 4.6. The east did have a high birth rate, but it also had a high death rate (and more of the variation in the death rate fed into migration); the overall rate of natural population increase was not unusually high. The south had the lowest rates of natural population increase, yet initially a high rate of out-migration. The other factors show quite large variations which contribute rather more to the pattern of migration. Table 4.7 provides standardized estimates of the eVect of each factor on net migration over the whole period 1871–1910 (the standardization adjusts the calculated eVect so that the national average is zero). The impact of the demographic variables is small compared to the others. The most important eVect is the percentage of the population occupied in agriculture: the south and the east lost population through migration because they were more agricultural and because they had less industry. The south and the east diVered in the importance of larger farms and large estates, and this apparently caused a further rise in out-migration from the east. For the south, it caused a small reduction. This also had diVerent eVects on the two regions which gained through migration. The eVect of demographic factors on migration will be considered further in the next section and in the next chapter. In aggregate, it does Table 4.6. Average values by regional grouping, 1871–1910 East

Death rate, current period (annual %) Birth rate, 20 years previous (annual %) Percentage of land in holdings of over 100 hectares (1895) Percentage of the population occupied in agriculture (1895)


West and north-west

South and south-west

















Migration 1870–1913: A Statistical Analysis


Table 4.7. Estimates of the eVect of diVerent factors on the pattern of net migration by regional grouping (annual average net migration per 100,000), 1871–1910 East

Centre West and South and north-west south-west

Both demographic variables 3 2 Percentage of land in holdings of 135 31 over 100 hectares Percentage of the population occupied 290 þ331 in agriculture

3 þ70

þ9 þ96



Note: A similar equation to that given in the previous footnote is used for each variable; the coeYcients used are 0:32 for LAND > 100j and 0:168 for % AGOCCPjt .

not appear that out-migration from the eastern provinces to the rest of Germany was driven to any signiWcant extent by higher rates of population growth than those in other regions. But population growth was high relative to the limited employment opportunities available in eastern agriculture. This drove the demographic surplus to seek employment in cities both inside the east and in other regions. It was the combination of demographic increase with the relative decline of agriculture which made the east a major source of migrants in the ‘labour surplus’ period. But this uses a broad classiWcation. Results in subsequent chapters show a diVerent picture for individual Kreise and regions. High rates of population growth at this level were an important factor in migration.

4. R u r a l – U r b a n Mi g r a t i o n f r o m t h e 1890 Ce n s u s The analysis in the previous section used net migration Wgures calculated from census returns. This data gives no indication of the destination of the migrant Xows. But a comparison of place of residence with place of birth from census returns can provide information of this type.11 An individual census is a snapshot, describing a moment in time, which tells us quite a lot about migration but which will leave out many 11

This type of data has been used by Ko¨llmann (1959) amongst others.

Migration 1870–1913: A Statistical Analysis


moves. Those recorded as living away from their place of birth will include many who made other moves before reaching their current place of residence. Some of those living in the region of their birth may have left and then returned. In addition, there is no way of knowing which moves are temporary and which are intended to be permanent. Migration calculated from census records in this way is more than just net migration (since a calculation of the Xow from one region to another is possible in both directions) but it will also tend to understate the overall level of mobility. One advantage of this type of data is that it is possible to separate diVerent types of migration. The 1890 census is particularly useful. The published results provided an analysis of the population resident in major cities according to place of birth. This can be used to examine moves to the twenty-seven largest cities from areas outside these cities, thus excluding moves between major cities, and moves between rural areas. The analysis can consider whether the likelihood of a move was aVected by conditions in the region of origin or the recipient city, identifying factors which promoted or retarded migration between urban and rural areas. The full set of econometric results is given in Appendix 4B, Section (iii); Table 4.8 provides a summary of the Wndings. There are a number of general Wndings which provide an overview of the migration process in the period up to 1890. Migration was strongly aVected by the distance between the recipient cities and the regions of origin. If the recipient city lay within the region of origin this increased the likelihood of a move. This indicates a strong preference for short moves over long-distance ones. It also means that the proximity of major cities would have an eVect on overall rates of migration out of rural areas. In remote regions, the large distances needed to be travelled to reach the various major cities would have a depressing eVect on migration in general. High wages in the recipient cities, and low levels of unemployment or casual employment as revealed by the 1895 occupational census, tended to increase migration: moves were more likely to cities with stable or buoyant labour markets. There were certain industries which attracted migrants: clothing and construction oVered easy entry into the labour market for migrants who lacked previous experience of industrial work. By contrast, cities with high levels of employment in textiles did not attract migrants: the main period of rapid expansion of these cities had been in the 1860s and 1870s, and they had less need of migrants once this period had passed.


Migration 1870–1913: A Statistical Analysis

Table 4.8. Summary of results of statistical analysis of rural–urban moves from 1890 census I. Secure statistical results Characteristics of the regions of origin which increase outwards migration: Proximity to the recipient cities Presence of large farms and estates Poor economic prospects High rates of productivity growth in agriculture Characteristics of the recipient cities which increase inwards migration: High wages Low rates of natural population increase Low numbers of unemployed workers or workers without permanent jobs Presence of the construction and clothing industries Absence of the textile industry II. Less secure statistical results Characteristics of the regions of origin which increase migration: High rates of natural population increase High levels of agricultural employment

In the regions of origin, poor economic prospects, as indicated by a low rate of formation of Aktiengesellschaften (joint stock companies), had a positive eVect on outwards migration. This is interesting as it indicates a ‘forward-looking’ element to migration: migrants responded to the direction of economic change as well as current conditions. When these Wndings are combined with the result of those given in Table 4.4, they show how important economic factors could be in aVecting the course of migration: migrants were more likely to move if the urban–rural wage gap increased; they chose destinations with high wages and good chances of employment; they left regions which were lagging in terms of economic development. Demography was also important, although the strongest and most secure results are obtained for the recipient cities: high rates of natural population increase in these had a negative eVect on inwards migration, since such cities could Wll a greater proportion of available vacancies from their own natural increase. By contrast, while it appears that high rates of population increase in the regions of origin may have promoted out-migration from these regions, this Wnding is much less secure.

Migration 1870–1913: A Statistical Analysis


Agricultural conditions are shown to have had an important eVect on out-migration. The presence of large farms and estates increased the likelihood of outwards migration in the regions of origin. In addition, agricultural change, as measured by the rate of productivity increase between 1880/4 and 1893/7, is associated with high rates of out-migration. This could be an indication that labour was shed as a direct consequence of changes in agricultural technology—the use of mechanized threshing, for example—or it could show that there was a general process of agricultural transformation, involving the replacement of traditional social and economic relations with modern capitalist ones, which had the eVect of both raising productivity and increasing the likelihood of migration. Alternatively, improvements in transportation may have brought new inputs to remote areas, raising productivity, as well as making it easier for potential migrants to leave.

5. C o n c l u s i o n s In general the statistical results presented in this chapter and the appendices conWrm the analysis presented in the previous chapter. There was a period of intense rural–urban migration in the 1870s and 1880s, and this was aVected by conditions within agriculture relative to those in industry, and by changes within agriculture. Agricultural transformation brought about the release of surplus labour onto urban labour markets. After a period of transition in the 1890s, the pattern of migration changed to one which was no longer dominated by the movement of surplus rural labour. The ending of this ‘labour surplus’ phase created conditions which were favourable to the creation of a modern mature industrial society, and to a process of social and economic reconciliation which could have led to the emergence of democratic institutions and a reduction in social conXict.

Migration 1870–1913: A Statistical Analysis


Ap p e n d i x 4A. De s c r i p t i o n o f Re g r e s s i o n Va r i a b l e s a n d So u r c e s (i) Appendix Tables 4B.1 and 4B.2; Tables 4.2, 4.3, 4.4, 4.5, 4.6, and 4.7 in text NETMIGkt BIRTHRTkt DEATHRTkt %AGOCCPkt

LAND > 100k


Net migration, births, and deaths per 10,000 head of population from census results, from SDR a.f. 20 and 52, n.f. 32, 68 and 150–1, VSDR 1897 iv, 1906 iv, and 1911 iv. The population occupied in agriculture as a percentage of the total occupied population, from the occupational censuses, from SDR a.f. 4, n.f. 104–6 and 204–6: 1895 and 1907 results are also in Kaelble and Hohls (1995). The percentage of the total agricultural area in holdings of over 100 hectares, from 1895 Agricultural Census, SDR n.f. 212, pp. 489–500. HoVmann’s index of land prices, HoVmann (1965), 569–70. Change in the ratio of industrial and agricultural wages, using HoVmann’s Wgures, and comparing Wve-year averages, HoVmann (1965), 492–5.

(ii) Appendix Table 4B.3, Table 4.8 in text P(MIG)kj (dependent variable) NATPOPINCk



Numbers born in rural part of region k who are recorded as living in city j, divided by total numbers born in rural part of k, calculated from SDR n.f. 68, pp. 136–53. Natural rate of population increase in the region of origin, 1880–90, from 1885 and 1895 census results, per 1,000 head of population, SDR n.f. 52 and VSDR 1897 iv. The value of Aktiengesellschaften founded 1871–90 from VSDR 1907 iv, p. 373; the Wgures are then divided by the regional population 1895 (1895 census). The population occupied in agriculture as a percentage of the total occupied population, from the

Migration 1870–1913: A Statistical Analysis





1895 occupational census, Kaelble and Hohls (1995), 167–88. Percentage increase in agricultural value added per full-time labour unit, 1880/4 to 1893/7, from regional accounts prepared for this study (see Chapter 7). Natural rate of population increase for the recipient city, 1890–5, from SJDS (1892), table viii, p. 39, and subsequent issues. Average of Ortsu¨bliche Tagelo¨hne for men and women, 1892, from SJDS (1913), 827–8. Calculated from the results of the 1895 occupational census, given in SJDS (1898), 350–1; also in SDR n.f. 107–8.

Ap p e n d i x 4B. Ec o n o m e t r i c An a l y s i s (i) Analysis of net migration 1871–1910, using census data The post-1871 census results can be used to look at the rates of transfer between the eighty-one regions (the pre-1871 Wgures use diVerent regional units and so are not comparable). Analysing each time period separately can show if the course of migration changed—if there were some periods which were dominated by the transfer of surplus labour out of agriculture and others which show a diVerent pattern. Table 4B.1 presents the results of a series of regressions run for each time period in succession, with a single explanatory variable: the percentage of the total occupied population which is occupied in agriculture (%AGOCCP). What this table shows is the extent to which the pattern of net migration by region can be interpreted solely as a transfer of labour out of the more agricultural regions, either to the industrial regions or to foreign countries. The t-ratios are all signiWcant at the 95% conWdence level, which shows that the extent to which a region specializes in agriculture is always a factor in migration, but it is the R2 statistics in the Wnal column that are the most eloquent. These show the extent to which the incidence of this form of migration, the transfer of surplus labour out of agriculture, can explain all the observed variance in net


Migration 1870–1913: A Statistical Analysis

Table 4B.1. Regressions of net migration by period on % occupied in agriculture: dependent variable is net migration per 100,000 (annual rate) for 81 regions Period


coeYcient on %AGOCCP



1871–5 1875–80 1880–5 1885–90 1890–5 1895–1900 1900–5 1905–10

1595 734 1006 1653 677 1335 769 610

39.22 19.87 35.06 45.44 23.12 38.15 23.49 21.63

7.91 7.80 10.36 11.84 7.23 8.70 6.25 5.39

0.442 0.435 0.571 0.639 0.398 0.489 0.331 0.269

Note: Regression variables and sources given in Appendix 4A.

migration by region. At the peak in 1885–90, with an R2 of 0.639, this is evidently the dominant form of migration: 64% of the migration pattern can be explained by this factor alone. But by 1905–10 it only accounts for around a quarter of observed variance. By this time migration was more heavily inXuenced by other considerations.12 The slope coeYcients give an indication of the pace of the transfer. A high coeYcient indicates that a large proportion of the agricultural population migrated in that period. This peaks in 1885–90, although it is also high in 1871–5 and 1895–1900. By 1900–10 it is considerably lower. This table provides clear statistical support for the view that the pattern of migration shifted from a ‘labour surplus’ one in the 1870s and 1880s, to a more ‘normal’ one post-1900, where the movement of population out of agriculture (which still continues even in advanced industrial economies) is just one stream amongst other migrant movements, not a dominant inXuence on the labour market as a whole. (ii) Panel data analysis of migration 1871–1910 The separate analysis of each time period can produce interesting contrasts in the slope coeYcients, and in the Wt of the estimated equations, 12

The question of non-linearity in the functional form, which might aVect the R2 calculations, was considered. F-tests generally led to the rejection of additional X2 variables. No signiWcant improvement in the R2 s was found to be possible.

Migration 1870–1913: A Statistical Analysis


but combining the data in a single panel makes it possible to test hypotheses about the reasons for the changes in the slope coeYcients. The analysis in this section uses the same sources as the previous section but analyses the data as a single panel. There are some restrictions on the analysis which follow from the nature of the data. Only eight time periods are available for the 1871– 1910 period, and the pre-1871 data is not comparable either in terms of the regional units used or the lengths of the time periods. This means that any lag structure has to be a simple one: eVectively the only possibility is the inclusion of a NETMIGkt1 term (net migration for the kth region lagged one period) amongst the explanatory variables, implying, by way of the Koyck transformation, a lag structure of the form E(Xt ) ¼ aXt þ a2 Xt1 þ a3 Xt2 þ a4 Xt3 . . . : for the expected values of the explanatory variables. An additional problem is that the sum total of regional net migration is the external balance, which Xuctuates considerably, as shown in Table 4.2. Ideally, this would be modelled separately, and in a way which took into account the eVect of alterations in economic conditions in the main destination for German emigrants, the United States.13 But, given the limited number of time periods, it is not possible to estimate a complex model of this type for the regional data set. So, the data set was analysed as a Wxed eVects panel data set, including dummies for all time periods, which means that the external balance is treated as an exogenous factor. The estimated model only deals with the internal transfer of population between the diVerent regions. Results are given in Table 4B.2. The Wrst column gives the basic model: net migration as a function of two demographic variables (the current death rate and the birth rate lagged 20 years), the percentage occupied in agriculture and the amount of land in holdings of over 100 hectares. As shown in the previous section, the percentage occupied in agriculture has an important eVect on migration, but this table shows that there is a further eVect from the prevalence of larger farms and large estates: there is a signiWcant negative coeYcient on the LAND > 100k term, suggesting that this causes additional out-migration. This point merited further investigation, and it is one of the themes examined in the next two chapters. In brief, these results show that, 13

A model of this type, using annual data, was included in chapter 6 of the original D.Phil. thesis, Grant (2000). It is intended to publish a separate article based on this chapter.


Migration 1870–1913: A Statistical Analysis

once the issue is examined at a lower level of aggregation, using a Kreislevel data set (with data for around 400 Prussian Kreise), then it is discovered that what mattered was the geographical position of these estates, the fact that they were mainly located in remote eastern areas, not the fact that they were large as such. The demographic terms are both signiWcant. Their importance is somewhat obscured by the lag structure adopted; part of the eVect comes through the NETMIGkt1 term. The estimated coeYcients imply that, after Wve periods (or twenty-Wve years), 41% of an increase in the birth rate lagged four periods would be exported through increased net migration, and 47% of a decrease in the current death rate.14 The next two columns in Table 4B.2 introduce variables which are inter-acted with the %AGOCCPkt variable. These are variables which might aVect the rate of labour transfer between agriculture and industry, and thus explain the variation in the coeYcients on %AGOCCP shown in Table 4B.1. A negative coeYcient therefore shows that a rise in this variable increases the rate of transfer from agriculture to industry; a positive coeYcient shows that it holds it back. The second column uses the change in the ratio between agricultural and industrial wages to explain the shifts in the pace of migration. This model produces an improved Wt. Part of the increase in migration between the 1870s and the 1890s can therefore be attributed to the eVect of the rise in this ratio. Given the lag structure used, this would still be aVecting migration for some time after this, even though the ratio ceased to rise. The third column uses a diVerent variable: the level of land prices as an indicator of agricultural prosperity. The estimated coeYcient is positive, indicating that higher land prices did reduce the rate of migration, but it is not signiWcant at the 5% level (the hypothesis that there is in fact no eVect on migration from this variable cannot be rejected with 95% conWdence). So, migration is best explained by the equation used for column b, which uses the change in the agriculture–industry wage gap. It was the widening of this gap which was responsible for the high rate of rural urban migration in the 1880s. What is also notable about the results in Table 4B.2 is the stability of the coeYcients on the demographic variables, and on the structural 14 The reason for lagging the birth rate 20 years is that the Berlin census results show that most migrants arrived in their early twenties.

Migration 1870–1913: A Statistical Analysis


Table 4B.2. Fixed eVect regressions (OLS with time period dummies) of net migration, annual rates per 1,000 population, Wve-year time periods 1871–1910 a NETMIGkt1 DEATHRTkt BIRTHRTkt4 %AGOCCPkt LAND > 100k


þ0.454 (5.86) þ0.260 (3.43) 0.230 (4.26) 0.168 (6.60) 0.0320 (2.78)

%AGOCCPkt * DAGINDWGAPt %AGOCCPkt * LANDPRICEt N F-test of regression R-sqd adj R-sqd RSS

þ0.469 (5.98) þ0.231 (3.11) 0.214 (4.13) 0.163 (6.42) 0.0320 (2.84) 0.0119 (5.50)

c þ0.450 (5.73) þ0.247 (3.27) 0.223 (4.16) 0.170 (6.65) 0.0328 (2.85)

þ0.00380 (1.81) 648 89.10 [12,635] 0.627 0.620 112213990

648 89.02 [13,634] 0.646 0.639 106588139

648 83.18 [13,634] 0.630 0.623 111304403

Note: t-statistics (in brackets) calculated using White’s heterocedasticity-corrected standard errors. Appendix 4A gives a full description of variables and sources; alternative random eVects estimates of the equations are given in Appendix 2B of Grant (2000). These do not alter the basic results although some of the coeYcients did change.

variables LAND > 100k and %AGOCCPkt . The importance of these variables in the overall pattern of migration is conWrmed. The labour surplus was created by a combination of demographic factors and the relative decline of agriculture. (iii) Analysis of rural–urban moves from the 1890 census The 1890 census provides Wgures for moves to the twenty-seven largest cities from those living outside these cities, divided between twentynine provinces. This excludes city–city moves and rural–rural moves. This data has been used to create a variable Pkj , which is the proportion


Migration 1870–1913: A Statistical Analysis

of the total numbers born in province K, excluding those born in major cities, who move to city J. The model is a simple linear log-odds ratio model: log (Pkj =(1  Pkj ) ¼ f (Xk1 . . . :Xkn , Xj1 . . . :Xjm ), where f(.) is a linear function, and Xk1 . . . :Xkn , Xj1 . . . :Xjm are characteristics of the region of origin K and the recipient city J used as explanatory variables. The only nonlinear variable is the distance travelled which is entered in a log form. The functional form produces a logistic curve, thus preventing predicted values from falling below zero. The LOGDISTANCEkj term gives the log of the distance in kilometres between the recipient city and the nearest border of the region of origin. Where the recipient city was within the boundaries of the region a notional distance of 2 kilometres was used (this could not be zero as the term was to be in log form). To avoid distorting the results by this procedure a dummy variable DUM1CONTkj was introduced for moves to cities within the regional boundaries. Table 4B.3 gives the results of four diVerent regressions run using this data. In the Wrst three columns the only variables included for the

Table 4B.3. Regression results for linear log-odds migration model, 1890 census a Constant LOGDISTANCEkj DUM1CONTkj

5.982 1.331 (26.3) 0.655 (2.45)

Characteristics of the region of origin: NATPOPINCk þ0.01088 (1.26) %AGRICk þ0.01108 (2.64) LAND > 100k þ0.0154 (6.90) AGVALUEk DAGPRODk

b 5.620 1.334 (26.4) 0.676 (2.53) þ0.01118 (1.30) þ0.06684 (1.43) þ0.0141 (6.13) 0.00196 (2.13)



5.367 1.328 (27.2) 0.668 (2.55)

5.082 1.281 (30.4) 0.499 (2.21)

þ0.0156 (7.54) 0.00392 (4.63) 0.0140 (5.49)

þ0.0155 (8.73) 0.00381 (5.24) 0.0140 (6.37)

Migration 1870–1913: A Statistical Analysis Characteristics of the recipient city: NATPOPINCj 0.0120 (8.95) AVEWAGESj þ0.0195 (11.55) %CONSTRCTj

0.0120 (8.98) þ0.0195 (11.57)

0.0120 (9.13) þ0.0195 (11.79)

783 1.028 818.5 177.8 [8,774] 0.647 0.644

783 1.010 790.8 214.5 [7,775] 0.659 0.656

%CLOTHj %TEXTILESj %UNEMPj N S RSS F-Test R-sqd adj.R-sqd

783 1.031 823.3 201.6 [8,774] 0.645 0.642

121 0.0101 (5.13) þ0.0123 (7.896) þ0.162 (6.156) þ0.105 (5.283) 0.040 (9.74) 0.140 (7.29) 783 0.869 582.5 209.4 [11,771] 0.749 0.746

Note: RONATINC and RCNATINC are natural rates of population growth for the region of origin and the recipient city; %AGRIC is the % of the population occupied in agriculture; %LAND > 100HA and %LAND20/100 refer to the distribution of the usable agricultural area by size of holding; AVEWAGES is the average wages of daylabourers in 1892 for each city; AGPROD82 and DAGPROD82=95 are Wgures for the levels of agricultural productivity 1880/5 and the increase in 1880/5–1893/7. These make use of regional agricultural accounts prepared by the author. Further details on the sources are in Appendix 4A.

recipient city are the level of wages (AVEWAGESj) and the rate of natural population increase (NATPOPINCj ). A number of explanatory variables which were found to inXuence migration were introduced for the region of origin. The results in the Wrst column show that the prevalence of larger farms and estates (LAND > 100k ) has a signiWcant positive eVect on the likelihood of a move from the region of origin. This conWrms the Wnding reported in the previous section. The rate of natural population increase has a signiWcant negative eVect on the likelihood of migration into the recipient city (a high rate of population increase reduces migration into the city) but the coeYcient on the NATPOPINCk term (population growth in the region of origin) was not signiWcant. This is an intriguing result: variations in population growth between cities matter; those in rural areas are less


Migration 1870–1913: A Statistical Analysis

important. This theme will be examined in more detail in the next chapter. The percentage of the population occupied in agriculture in the region of origin (%AGRICk ) is a variable whose importance is considerably reduced when compared to the results given in the previous section. But this is not so surprising. The eVect of using rural–urban moves as the dependent variable is to eliminate much of the previous eVect of this variable. More urban regions had lower overall net migration for two main reasons: Wrst, there were proportionately less farmers and farm-workers, and secondly, there were more moves to cities within the region. Moves to cities within the region are now recorded; and by using moves relative to the rural population as the dependent variable the importance of the Wrst point is also reduced.15 The second column shows the eVect of including the term AGVALUEk . This gives the value per head of population of Aktiengesellschaften (joint stock companies) founded 1871–1910 by region. This was included in an attempt to examine the extent to which migration was aVected by the prospects of a region as well as its current position. The %AGOCCPkt term is a backward-looking measure: it reXects what has happened in the past, not what is likely to happen in the future. The formation of Aktiengesellschaften is a forward-looking decision by investors and should be related to the dynamism of the region, to the extent to which it has specialized in fast-growing industries for example. The introduction of the AGVALUEk variable reduces the signiWcance of the %AGRICk term (column b). For the last two columns this term and the NATPOPINCk term are both dropped. The coeYcient on AGVALUEk is raised as a result. Some of the eVect of the %AGRICk variable is captured by AGVALUEk (Aktiengesellschaften were more likely to be formed in areas which were already industrial). But the greater signiWcance of AGVALUEk is due to its ‘forward-looking’ or dynamic quality. What exactly were these dynamic factors which caused AGVALUEk to have this inXuence? There were three forces at work. The Wrst was an increase in the importance of large Wrms, which were more likely to be Aktiengesellschaften. The second was the impact of foreign trade on the German economy, which led to greater specialization in those areas 15 The fact that some eVect is still exerted by %AGRICk is explained by the fact that there are some regions which were industrial but which did not have large cities.

Migration 1870–1913: A Statistical Analysis


where Germany had a comparative advantage, and in turn led to faster growth in regions specializing in these Welds. The third was a tendency for regional specialization to rise, due to economies of scale internal and external to the Wrm. Increased vertical integration, leading to the combining of processing and primary production in a single region, was one aspect of this. In the third column the increase in agricultural productivity by region between the years 1880–4 and 1893–7 is introduced (DAG PRODk ).16 This variable is found to have a signiWcant positive eVect on the likelihood of migration: migration is higher from regions with high productivity increases. This result can be interpreted as supporting the view that there was a substantial agricultural labour surplus, and the release of this surplus both increased productivity and led to out-migration. But an alternative view is that these are both consequences of a change in institutions and social relations. The impact of industrialization led to a transformation of agrarian society which both raised productivity and led to increased out-migration. In the last column (d) the breakdown of the employed population in the recipient city by occupation is used to produce a series of additional explanatory variables. The results show that certain industries attracted migrants, leading to an increase in the likelihood of inward migration from rural areas. Two which had this eVect were construction and clothing (%CONSTRCTj and %CLOTHj ). Other industries had a negative eVect. One of these was textiles (%TEXTILESj ). The main period of rapid expansion for this industry had passed by 1890, and the main textile areas had less need of migrants. The results also show that there is a negative eVect from the percentage of workers who were without regular employment according to the census (the variable is called %UNEMPj although it is not the same as registered unemployment). 16

These Wgures are derived from regional accounts. The methods used are described in chapter 5 of Grant (2000), and also in Grant (2002a).


Demography and Migration 1. I n t r o d u c t i o n The last chapter presented some evidence on the inXuence of demographic variables on migration. The results were equivocal. Regressions showed an eVect from demographic variables on net migration for eighty-one regions 1871–1910, but, when migration between four major regional groupings was considered, the net eVect of demographic factors was shown to be small when compared with other explanatory variables. The analysis of the 1890 census showed that variations in the natural rate of population increase in the recipient cities had an important eVect on migration, but that the eVect from population growth in the regions of origin was less certain. This chapter will attempt to elucidate these issues, in the context of a general discussion of the relationship between migration and demography in Germany 1870–1913. In this period Germany was going through the ‘demographic transition’, usually deWned as a movement from a state in which both fertility and mortality are high to one in which both are considerably lower. This transition has been much studied by demographers: one conclusion which has emerged is that there is no single model of the demographic transition—much will depend on the circumstances of individual countries or regions. The German experience conWrms this: there were important diVerences between the regions and these have implications for migration. SpeciWcally, the demographic experiences of two regions which were major sources of potential migrants in the period, the south and the east, were very diVerent. This had a considerable eVect. It is one reason why there was heavy out-migration from certain heavily agricultural regions in the east, while migration from similar southern regions was lower. So, diVerences between the south and the east will be one theme of the chapter. A second theme will be a comparison of the demographic background to two periods of heavy out-migration: that which aVected southern and western Germany in the earlier part of the nineteenth century and the movement out of eastern Germany in the period after

Demography and Migration


1870. The intention here is to look for signs of demographic stress, direct evidence of severe population pressure. To what extent was migration caused by overpopulation? Are there diVerences between the two periods? Comparison between the south and the east leads to an examination of the role played by agricultural factors. The agricultural systems of the two regions were very distinct: diVerent sized-farms, diVerent types of labour contract, diVerent inheritance systems. Did these have demographic eVects? As regards the demographic experience of the recipient towns and cities, this has already been studied in some depth. The high fertility of the coal-mining regions in the west is a well-known fact.1 This has clear implications for migration. It is a fairly simple calculation to show that high fertility reduced the need for migration into the coal-mining areas. A higher proportion of the required increase in the total labour force was supplied through the natural increase of the resident population than was the case for Berlin, for example. This was an important factor in migration.2 A more general theme is the contribution of demographic movements to the ‘labour surplus’ phase of the Lewis Model. The period of the ‘demographic transition’, when mortality rates fall ahead of a reduction in fertility, is potentially one of high population growth, contributing to the overall labour surplus.

2. T h e G e r m a n D e m o g r a p h i c E x p e r i e n c e i n t h e N i n et e en t h Ce n t u r y An overview of German population movements in the nineteenth century is provided in Figure 5.1. This uses calculations made by the Prussian statistical oYce for all the territories included in Prussia after the incorporation of Hanover and Schleswig-Holstein in 1866. This series extends back to 1816. A second series is superimposed, which is for all the post-1871 German territories, and makes use of a series 1

Wrigley (1961); see also Haines (1977) and (1979). Valuable studies of demographic change in Germany in the period include Ballod (1914), Hohorst (1977), Ko¨llmann (1974), Knodel (1988), Marschalck (1987), Lee (1979a), and Pollard (1980). This chapter also draws on the theoretical analysis of Birdsell (1988) and Habbakuk (1958). The regional classiWcation follows the deWnitions set out in Appendix Table 3A.1. 2

Demography and Migration

126 500 Prussian births 400

German births 300 Prussian deaths

German deaths

200 Prussian natural increase 100

German natural increase

0 1816










Fig. 5.1. Births, deaths, and the natural rate of population increase, per 10,000 head of population, Prussia and Germany 1816–1913 (post-1871 territorial units) Note: Figures are three-year rolling averages. Source: The source used is SJPS (1915), 55; these estimates were endorsed by Ko¨llmann (1980), who also gives figures for the individual provinces.

produced by the Imperial Statistical OYce. This only goes back to 1862.3 There is a fairly close correspondence between the two series, which is as expected, given that Prussia accounted for around 61% of the total German population. The reliability of the earlier Wgures is open to question. The main area of concern is the Wgures for stillbirths prior to 1841. These had to be estimated as they were not then recorded. If these are underestimated then the birth rate and the death rate are aVected, but not the overall rate of population growth. The natural rate of population increase peaked in the 1816–22 period, possibly as a reaction to the ending of the disruptions caused by 3

Calculated from SJDR (1915), 22, SJPS (1910), 14.

Demography and Migration


the Napoleonic wars, but then fell back sharply in the agricultural crisis which followed, from 1.9% in 1822 to just 0.2% in 1831. This was brought about by a combination of a rise in the death rate and a fall in the birth rate. Thereafter there was an irregular recovery, which peaked in the mid-1870s, when the natural rate of increase reached 1.5% per annum. This period saw considerable Xuctuations in both fertility and mortality, which are apparent even when a three-year moving average is used. After the 1870s both series are much more stable, and tend to decline, with the decline in the death rate being more pronounced and starting much earlier. The birth rate does not show a substantial decline until after 1900, so that the rate of natural increase remained high, peaking at 1.56% per annum (reached in both 1898 and 1902).4 The decline of fertility after 1900 was too late to aVect the pre-1913 labour market. The combined eVects of high fertility in previous periods and declining mortality meant that the rate of increase of the potential labour force remained high right up to 1913. Comparison of the employment censuses of 1895 and 1907 shows the male labour force increasing at a rate of 1.5% per annum and an even faster rate for female employment.5 The pattern is a classic demographic transition curve. But the peak was reached at diVerent times in diVerent regions. Figure 5.2 illustrates this point, classifying each region by the timing of the peak rate of natural increase between census years (normally at Wve-year intervals). Most regions peaked in either 1895–1900 or 1900–5, as did the national average. But there was a group which peaked much earlier, in the 1870s. This included Berlin, Pomerania, and some other central regions. There is another group in which population growth was still rising by 1905–10, which had not, therefore, clearly peaked before the First World War. These are the regions shaded black. This is quite a scattered group of regions, with all areas represented apart from the centre. There is no clear pattern; it includes agricultural as well as industrial regions. Figure 5.2 helps to explain why there was relatively little eVect from demographic factors when the pattern of migration between the major regional groupings was considered in the last chapter. There were 4

German Wgures, SJDR (1912), 22. Data from Kaelble and Hohls (1989). There are some changes in the relative positions of the various regions as a result of diVerent rates of industrial growth between census dates, but the overall ranking remains remarkably constant. The classiWcation would be essentially the same if the 1882 or 1907 censuses were used in place of the 1895 census. 5


Demography and Migration









Fig. 5.2. The demographic transition in Germany 1871–1910: German regions classified by the period of highest natural population increase Source: Data for this figure, and the next two, taken from census publications, SDR (1876), a.f. 20, SDR (1881), a.f. 57, SDR (1886), n.f. 32, SDR (1891), n.f. 68, SDR (1901), n.f. 150–1; VSDR 1897 iv, 1906 iv, and 1911 iv.

diVerences within these groupings which had a tendency to cancel each other out. Population growth falling in Pomerania was balanced by rising population growth in other eastern regions, for example. Figures 5.3 and 5.4 demonstrate the eVect of these movements on the overall rate of population growth by region, comparing the 1870s with the 1900–10 period. The Wrst map has a ‘mottled’ appearance: most parts of the country contained regions both of high and of low growth. Again, this helps to explain why demography had a limited eVect on the major groupings. By 1900–10 there is a clearer pattern: the areas of high growth are predominantly in the west and in the east; the centre and the south contain the low-growth regions. In most areas the rate of increase was faster in the 1900–10 period. The main exceptions are in the centre, conWrming that this was the region which Wrst went into the downward phase of the demographic transition.

Demography and Migration



Fig. 5.3. The natural rate of population increase 1871–80 (annual averages) Source: As for Table 5.2.

From the viewpoint of a study of migration, the most important map is Figure 5.3. This is because diVerences in population growth in 1900–10, in so far as these reXect diVerences in birth rates, are not particularly relevant to migration Xows before 1914. A rise in the number of children will not increase migration immediately; in fact it could reduce migration if families with children are less likely to migrate. It is only when these children enter the labour market, and take decisions based on an assessment of their own prospects, that there is likely to be an increase in out-migration from a region of high population growth. Analysis of the Berlin Wgures from the 1885 census supports these points. The median age of migrants arrived in the previous year was 22.8; 52.2% of migrants were aged between 17 and 26.6 These Wgures also show that those aged between 20 and 21 had the highest probability of migration. 6

Calculated from Berlin, Statistisches Amt (1887), table Vb.

Demography and Migration




Fig. 5.4. The natural rate of population increase 1900–10 (annual averages) Source: As for Table 5.2.

So, it is the earlier, ‘mottled’ map of 1871–80 which has more signiWcance. The more pronounced regional pattern of 1900–10 was of little consequence for migration before 1914.

3. A C o m p a r i s o n o f D e m o g r a p h i c Mo v e m e n t s in th e Ea st a nd t he S o u t h The discussion in the previous section implicitly treated demographic factors as exogenous: as a pattern of behaviour which was unaVected by economic circumstances and prospects. But there are ways in which economic conditions can have eVects on population growth, and, once this is accepted, the main point about the interaction between demographic factors and migration becomes a rather diVerent one. It is the extent to which the natural rate of increase responds, or fails to respond, to diVerences in relative economic opportunities between regions that will aVect migration.

Demography and Migration


Contemporary observers were well aware of mechanisms whereby economic conditions could aVect both mortality and fertility. In 1879 von Fircks pointed to the coincidence of periods of rising mortality and high cereal prices.7 German demographers also recognized the importance of delayed marriage—the ‘Western European marriage pattern’— for the control of fertility. Indeed, they were particularly conscious of this, as Germany was very much on the frontier of this division: the area of delayed marriage in central and eastern Europe followed the line of German inXuence.8 In the German view, delayed marriage was a consequence of cultural progress, and this was the reason why Germany, and other parts of Western Europe, diVered in this respect from lands further east: ‘In primitive tribes or in countries with a low level of cultural development, such as Russia, Serbia or Bulgaria, the average age at marriage is very low; marriage takes place shortly after puberty for both sexes.’9 The same study argued that in Germany marriage was linked to economic conditions. The custom which delayed marriage until the couple were able to set up as an independent family unit meant that in favourable economic circumstances this would be attained earlier, while less good conditions would hold it back.10 Statistical analysis of marriage rates in Prussian provinces, 1816–75, supports the view that economic circumstances did inXuence marriage rates. High prices for foods consumed by the rural labourers and their families, principally rye and potatoes, were associated with lower marriage rates. However, high prices for goods which were produced in rural areas but mainly consumed by the better-oV or by city-dwellers, such as wheat and beef, were linked to higher rates. Marriage was put oV when times were bad and brought forward when times were good. The full analysis is given in Appendix 5E.11 With such a relationship, when there was out-migration caused by a rate of population growth which was excessive relative to available economic opportunities, this would set in motion demographic changes which would eventually reduce population growth in the regions of 7

Von Fircks (1879), 48. Including, for example, the future Baltic states, as can be seen from the map at the end of Coale and Watkins (1986). 9 Nadobnik (1908), 69. 10 Ibid. 80. 11 Hohorst (1977) also found a relationship between marriage rates and economic conditions for linen workers. 8


Demography and Migration

out-migration. A crisis of overpopulation would cause a deterioration of employment prospects. This would hold back marriage rates and act as a stabilizing mechanism. The question then becomes: why did this mechanism work in some regions and not in others? The importance of this point is illustrated by the results of a classiWcation of German regions given in Appendix 5A and summarized in Table 5.1. The data set is the one used in the previous chapter: net migration is calculated using the indirect method of census comparison, and the period of analysis is 1871–1910. The regions were divided into four equal groups according to the percentage of the population occupied in agriculture (1895 occupational census). These groups were then subdivided into two according to an estimate of predicted net migration resulting from population growth. This made use of the statistical analysis reported in Appendix 4B (Table 4B.2).12 As before, the birth rate is lagged four periods, or twenty years. The classiWcation shows that there were considerable diVerences between the demographic records of the diVerent regions. In the high population growth sub-group of the regions classiWed as highly industrial, the diVerence between the lagged birth rate and the current death rate was 1.66% per annum; for the low population growth regions it was 1.29%. The predicted eVect from these demographic diVerences amounts to a net diVerence of 137 persons per 10,000 resident population (108 plus 29). The actual diVerence was 227 (565 minus 238). So, these demographic factors accounted for 42% of the observed diVerence in net migration rates between the two sub-groups. At the other end, in the highly agricultural group, the demographic eVect accounts for 48% of the observed diVerence in net migration between the high and low population growth groups.13 Demographic factors played a large part in the determination of migration patterns. Looking at individual regions, there are some major diVerences between regions which are not so far apart geographically. In the highly industrial group there are two comparisons of this type which stand out: in the west, the Aachen region has one of the lowest rates of population growth, the Arnsberg region one of the highest; in Saxony 12 The equation used in these calculations was the same as that used for Table 4.5: NETMIGjt ¼ 0:44 NETMIGjt1 þ 0:26 DEATHRTjt  0:22 BIRTHRTjt4 . The regions were divided into eight sub-groups of ten; the extra region is in the ‘moderately agricultural–low population pressure’ sub-group. 13 For the ‘moderately industrial’ group, however, the predicted eVect is much greater than the observed diVerence.

Demography and Migration


Table 5.1. A classiWcation of 81 German regions according to the availability of non-agricultural employment and population growth, 1871–1910, unweighted averages (migration and demographic variables are per 10,000 head of population) Regions classiWed as Highly industrial

Moderately industrial

A. Regions with low population growth % occupied in 25.2 37.8 agriculture Average birth 3591 3588 rate (T-4) Average death 2305 2433 rate (T) þ29 þ82 Predicted demographic eVect Actual net þ238 206 migration B. Regions with high population growth % occupied in 16.7 35.6 agriculture Average birth 4079 3725 rate (T-4) Average death 2414 2217 rate (T) 108 72 Predicted demographic eVect Actual net þ565 246 migration

Moderately agricultural

Highly agricultural



All regions





























Source: ClassiWcation of regions uses 1895 occupational census, Kaelble and Hohls (1989).

the Zwickau region has a high rate of natural increase, while the Bautzen region is much lower. In the ‘highly agricultural’ group there is a clear geographical division. Six of the ten ‘high population growth’ regions are in the east, including the Wve highest; eight of the ten ‘low population growth’

Demography and Migration


regions are in the south, including the Wve lowest. The basic point is clear: in the south population growth was adjusted downwards in regions which had limited opportunities for non-agricultural employment; in the east this did not occur. Consequently there was more migration out of the eastern regions. The mechanism which achieved this was marriage restriction. Figures 5.5 and 5.6 make this point. The Wrst shows the overall record of population growth for the south and the east. It combines Wgures



South : births South : deaths

East : births East : deaths

South : natural increase

East : natural increase

Births 400

300 Deaths

200 Natural increase 100

0 1818







Fig. 5.5. Births, deaths, and the rate of natural increase, five-year average annual rates, 1816–1910 (per 1,000 head of population) Source: Calculated using data from Ko¨llmann (1980) for the pre-1871 figures; the post1871 figures come from the census publications. For this comparison MecklenburgStrelitz and Alsace-Lorraine were omitted from the regional groupings, thus ‘south’ is Bavaria, Wu¨rttemberg, and Baden.

Demography and Migration


0.6 East: Im index South: Im index 0.5 East: marriages South: marriages 0.4




0.0 1818







Fig. 5.6. Crude marriage rates and the Princeton Im index, 1816–1901, fiveyear average annual rates Note: Marriage rate is number of marriages per 100 head of population. Source: Data from Knodel (1974), Appendix table 2.1, and Ko¨llmann (1980).

given by Ko¨llmann for the pre-1871 period with Wgures produced by the Imperial Statistical OYce for the period after uniWcation. Ko¨llmann has estimated stillbirths for periods when these are not available, so there are reasons to treat the pre-1870 Wgures with some caution. However, the basic conclusions should not be seriously aVected. These are, Wrst, that the natural rate of increase was fairly consistently higher in the east, and secondly that this owed more to fertility than to mortality. Mortality Xuctuated up to 1880 (rather more in the east than in the


Demography and Migration

south), and thereafter fell in both regions. Fertility was generally higher in the east, apart from the 1871–85 period, when it was the same in both areas.14 Figure 5.6 compares the marriage Wgures for the two regions, and includes average values of the Im index used by the Princeton study of population growth.15 This is a weighted index of the proportion of women of diVerent ages who are married, the weights reXecting changes in the natural fertility of women in diVerent age groups. It should therefore represent an unbiased estimate of the potential eVect of marriage on birth rates. However, the limitations of the available census data meant that it could not be calculated before 1867. The crude marriage rate, which is available, is a much less valuable measure. As it includes remarriages, it is aVected by the death rate: higher death rates lead to more remarriages of those who have lost partners. The same problem applies to the average age of those getting married, if this does not distinguish remarriages (which is the case with the Prussian Wgures). Figure 5.6 shows that there is a relationship between the crude marriage rate and the Im index. The low marriage rate in southern Germany before the 1860s clearly aVected the Wrst Im index. Even though the southern marriage rate was rising in the 1860s this still left the south well below the east. The southern Im index was 79% of the eastern Wgure. A higher southern marriage rate in the years 1860–75 meant that the gap narrowed, and this is the period in which the southern birth rate rose to the eastern level. But in the late 1870s the southern marriage rate fell again, and both the Im indices and the birth rates moved apart. The low birth rate of southern Germany owed something to legal restrictions on marriage in Bavaria, but this was not the only factor. While these contributed to the extremely low Bavarian Im index of 0.380 in 1867, the indices for Wu¨rttemberg and Baden, where restrictions were lifted earlier, were also low (0.403 and 0.400 respectively). In addition, while the lifting of the Bavarian restrictions in 1868 certainly contributed to the high marriage rates in the following years, in the less favourable economic circumstances of the late 1870s there was a return to low marriage rates, which contributed to the fall in the southern Im 14

The reasons for high eastern fertility are examined in Section 3. All the Princeton indices for Germany are taken from the appendices in Knodel (1974). 15

Demography and Migration


index relative to the east. It seems therefore that the legal restrictions reinforced a tendency to low marriage rates which had other causes as well.16 Turning from the incidence of marriage to the level of fertility within marriage (the Princeton Ig index), there is no sizeable diVerence between the east and the south on this measure for the period 1867–1910. The southern average was slightly lower, 0.741 against 0.745, but this was balanced by a larger number of births outside marriage. Statistical analysis of the percentage unmarried did not produce a clear relationship between this variable and economic circumstances. The extent to which marriage was sensitive to economic conditions varied considerably between regions, especially once the non-Prussian regions were included in the analysis. One reason for this may be that some areas were restricting fertility within marriage. This was the conclusion reached by Knodel in his study of village fertility patterns. If so, then these areas would have had less reason to put oV marriage in times of economic hardship.17 Another statistical analysis which did not produce results was an attempt to Wnd causation running from migration to marriage patterns. If migration aVected the sex-ratio in regions of out-migration then it might also aVect the likelihood of marriage for women in these regions. The lack of any clear relationship between migration and the marriage prospects of women is demonstrated by Figure 5.7 which gives the proportion of women aged 49–54 who had never been married, by region, from the 1880 census. Figure 5.7 shows that there was no consistent relationship between migration and the likelihood of marriage. Marienwerder, a region of high out-migration, had one of the lowest proportions of never-married women. In general the regions of out-migration in the east had no tendency to have more never-married women than the recipient regions of the centre. Some of the reasons for this can be understood from the discussion in the previous chapter. Women did migrate, even if they 16 Works dealing with the eVect of the marriage restrictions in Southern Germany and Austria include Matz (1980), Ehmer (1991), and Mantl (1997). The eVect of these restrictions may have been to increase illegitimacy rather than reducing the overall birth rate. A study of 5,263 applications in Wu¨rttemberg during the restricted period showed that 6% were refused, but these couples may have applied successfully later, Matz (1980), 267. 17 Knodel (1988). This study of demographic behaviour in 14 villages shows considerable divergence between the diVerent villages, but the sample is too small, and the regional coverage is too restricted, to draw general conclusions about the reasons for these diVerences.


Demography and Migration


Fig. 5.7. Percentage of women aged 49–54 who had never been married, by region, 1880 census Source: Calculated from SDR (1881), a.f. 52, pp. 144–87.

made fewer moves than men.18 So women were quite capable of leaving regions where prospects were poor. Moreover, some of the men who made temporary moves out of their region of birth in their early twenties returned later to be married in that region. Thus out-migration did not inevitably lead to a decrease in the likelihood of marriage.19 But in the south and the south-west the position was a diVerent one. Here there are extremely high proportions of never-married women. The Bavarian Wgures are almost certainly aVected by the legal restrictions in force up to 1868, but high Wgures are also found in Baden, 18 Women also made temporary moves, Wnding employment as domestic servants, Lee and Marschalck (2000), 379, but men were more mobile in general, Hochstadt (1999), 93–4. 19 Analysis of Kreis-level results for 1895 did indicate that there was an eVect from migration on marriage rates inside Prussia by that date: see the note to Table 5D.2 in Appendix 5D.

Demography and Migration


Wu¨rttemberg, and Alsace-Lorraine.20 Figures in some western regions are also above average. These include many of the areas where there was high out-migration in the Wrst half of the nineteenth century. This leads to a more general point. Looking at the history of migration in Germany over the whole nineteenth century it is possible to identify two distinct movements of large-scale out-migration. The Wrst was centred on south-western Germany, began in the 1830s, and peaked in the 1850s. The second was centred on eastern Germany, began in the 1860s, and peaked in the 1890s. The Wrst was accompanied by clear signs of demographic distress. The death rate was high and the marriage rate fell. Partly in consequence, there was a rise in the number of illegitimate births. Ko¨llmann ¨ bervo¨lkerungskatastrophe’. 21 In a has referred to this as a period of ‘U process not untypical of pre-industrial Western Europe, an excess supply of labour relative to economic opportunities was eliminated, partly by out-migration and partly by reduced marriages. The excess supply of labour was most probably caused by falling mortality from the late eighteenth century onwards.22 The second period has none of these characteristics. The death rate was falling in all areas from the 1870s onwards, including the areas of out-migration. Marriage rates were high and tending to rise. Illegitimate births fell, and were below the national average in the east. This is not to argue that all was well in eastern Germany. There is evidence of poor conditions in an earlier period from a diVerent source, which is the height of recruits, generally taken as a good indicator of the nutritional standards of children and young adults.23 Figure 5.8 gives the proportion of recruits rejected for being less than 5 feet tall, in the 20 Marriage restrictions may also account for the relatively high Wgure for Mecklenburg, Lubinski (1995); for Bavaria see Lee (1977). The period after the ending of the marriage restrictions saw a gradual convergence of marriage rates in the south and south-west with those in other parts of the country, but this could also be explained by the improvement in economic circumstances in these regions after 1870. 21 Ko¨llmann (1976), 15. The reasons for this ‘overpopulation catastrophe’ have been debated: see inter alia Lee (1979 and 1980), Marschalck (1987), Ko¨llmann (1974). One important question is the impact of agrarian reforms introduced in the early nineteenth century. But it is quite possible to see it as a consequence of a period of relatively high population growth, set oV by declining mortality in the eighteenth century. 22 SchoWeld et al. (1991), Lee (1980), and Marschalck (1982) discuss this point. 23 See, inter alia, Komlos (1995) and Baten (1999). There are other possible reasons for the low stature of eastern recruits, such as the eVect of diseases, Leunig and Voth (1996), but in the absence of clear evidence of regional diVerences in disease epidemiology, it is reasonable to assume that the main cause of the variation in stature was nutritional.

Demography and Migration



Fig. 5.8. Numbers of possible recruits in the 1860s rejected for being less than 5 feet tall, per 1,000, by Regierungsbezirke Note: The unit of measurement is the old Prussian Fuâ. Source: Meitzen (1868), 322–3.

old Prussian provinces (post-1815 boundaries), taken from a survey in the 1860s. This would obviously reXect conditions in the 1840s when these recruits were born. The range is a large one: fewer than 2% of possible recruits were less than 5 feet tall in Berlin and Mu¨nster; in Marienberg 16.8% were, and in Posen the Wgure was 22.8%.24 This is a period when the east was attracting migrants from other parts of Germany. The map shows that conditions in the eastern Prussian provinces in these years were harsh and unappealing. It conWrms the view that only acute demographic pressure in other parts of Germany could have led migrants to move to these remote areas where living conditions were precarious and food 24 The Wgures are much higher for the Polish areas, which might reXect a disposition to reject Polish recruits, although putting down incorrect heights would have been an odd way to do this when there were other, less easily veriWable, reasons which could have been given (such as ‘bodily weakness’).

Demography and Migration


supplies often inadequate. What is striking is that, in the early part of the century, migrants moved from relatively rich provinces to the poorer ones in the east, but after 1860 they moved from the poorer east to the richer parts in the centre and the west.25 There are, therefore, reasons to see the two migration movements as representing diVerent types of labour surplus: the Wrst a pre-industrial labour surplus, driven by demographic conditions, the second, a labour surplus of the industrial period, not primarily caused by population growth, but rather by changes in the relative demand and supply of labour between sectors. Rising industrial demand for labour was supplied by the release of surplus labour from agriculture.

4. P o p u l a t i o n G r o w t h in t he R eg ion s of O r i g i n The argument has been made that the agrarian reforms of the early nineteenth century created the conditions for a large increase in population, and that this provided a stimulus to industrial growth. This is sometimes referred to as the ‘Ipsen–Ko¨llmann thesis’ after the two authors who are most closely associated with this view. The eVect of the land reforms has, however, been disputed. Some demographers, including Marschalck and Lee, have argued that the intensiWcation of work which, they argue, was brought about by the reforms contributed to a worsening of child care and a consequent rise in infant mortality.26 There is a problem with the chronology of the Ipsen–Ko¨llmann story which is apparent from Figures 5.1 and 5.5 in the previous section: the peak of German population growth in the years around 1900 is far removed from the period of the land reforms. Rising population growth after 1870 looks like a response to industrialization, not a cause. It can be argued that the surge in marriages and in births in 1816–22 owed something to emancipation (though the ending of the Napoleonic wars is also a possible explanation). This, in turn, meant that there was a ready supply of labour for aspirant industrialists, as well as providing

25 Ko¨llmann’s Wgures (calculated on a three-year basis using census returns) show that there were substantial gains in East and West Prussia, Silesia, and Posen through migration between 1816 and 1846. It is not possible to say exactly where these migrants came from, but there was out-migration from Brandenburg in 1816–31, Prussian Saxony in 1820–31, and Westfalia in 1831–7 and again in 1840–6. Ko¨llmann (1980). 26 Ko¨llmann (1976), Lee (1979a), Marschalck (1976); see also Harnisch (1977).


Demography and Migration

the labour required for the expansion of the area under cultivation in eastern Germany in the Wrst half of the century. The emphasis placed on emancipation has behind it a ‘top–down’ view of population growth: peasants and rural labourers will marry early, and have more children, unless laws or other restrictions imposed from above prevent them from doing so. But this underestimates the ability of rural society in Western Europe to regulate population growth in its own interests. Knodel’s 1988 study of fourteen German villages showed that some were restricting fertility within marriage as early as 1800, while the average age of women at Wrst marriage was raised from 25.5 in 1700–49 to 26.9 in 1825–49.27 This was not exactly a benign process. The burden fell heavily on women, particularly if reduced marriage opportunities led to an increase in illegitimate births. The high mortality of illegitimate children (84% higher than for those born within marriages in Prussia 1875–1900) is eloquent testimony to the social pressures placed on unmarried mothers.28 If rural society was able to exercise a crude but eVective control over population increase, then it is possible to interpret the increased birth rate which followed the agrarian reforms as a response to a perceived improvement in economic opportunities. But for some these hopes were disappointed. In the 1830s bad harvests and high prices brought about a worsening of the condition of the landless agricultural labourer. However, from 1860 onwards rising non-agricultural demand for labour brought about a recovery, both in real wages and in births.29 Agricultural institutions have consequences for the fertility decisions of households. One example is the contractual labour relationship typical on large estates in the east. The Inste or Deputate received a smallholding from the landlord (from which they derived much of their income) in return for an obligation to provide labour at harvest time. This obligation was not just to provide the labour of a single person, but of additional Scharwerker, who could be family members or Knechte or Ma¨gde (other labourers living in the Inste household). For a household living in this system it was valuable to have children who could 27

Knodel (1988), 125. Calculated from Wgures in SJPS (1915), 55. The Wgures show that on average 55.5% of illegitimate children died in their Wrst year, while 19.5% of those born within marriages died. This was not just a feature of rural society; the diVerence was just as large for those born in cities. 29 The role of economic factors in population growth is discussed in Birdsell (1988). 28

Demography and Migration


provide Scharwerker labour; it was also necessary that, if the family were to retain the smallholding when the parents were too old or inWrm to work during the harvest, there should be children willing and able to perform these duties.30 A second institution which may aVect fertility decisions is partible inheritance. There are reasons to expect this to aVect marriage decisions in particular. Studies show that landless labourers married later, so any institution which spread land ownership more widely should lead to earlier marriage and consequently higher fertility. But there are factors which might work against this. If the main concern of parents when making fertility decisions was the prospect of support from their children in old age, then there was a risk that, with partible inheritance, large family sizes could lead to a position where none of the children had holdings large enough to provide eVective support to their parents. Excessive subdivision of holdings would not be in anyone’s best interest.31 To examine the relationship between these institutional factors and fertility a data set was prepared which was based on the 1895 census results for the Prussian Kreise. Limiting the analysis to Prussian territories had the advantage that the legal framework would be the same for all areas. It also meant that the results of the 1892 and 1901 surveys of Ortsu¨bliche Tagelo¨hne could be used for rural wages. Finally, Wgures for Table 5.2. Summary of Wndings from statistical analysis of census results for 424 rural Prussian Kreise 1895: analysis of total fertility Factors which increase total fertility (numbers of children under 1 year relative to the total number of women of child-bearing age) I. Secure statistical results A large Polish-speaking proportion of the population Presence of large farms and estates High levels of rural wages Low levels of partible inheritance II. Less secure statistical results A high level of Catholicism A large Lithuanian-speaking proportion of the population 30 Weber (1895), 185; see also Harnisch (1984) for a diVerent perspective on tenurial relations in eastern Germany, but which also stresses the importance of the post-reform system as a source of labour for industrialization. 31 See Berkner and Mendels (1980) for a survey of the literature on this point.


Demography and Migration

Table 5.3. Summary of Wndings from statistical analysis of census results for 424 rural Prussian Kreise 1895: analysis of marriage rates Factors which increase the proportion of women aged 15–50 who are married I. Secure statistical results A low level of Catholicism A large Polish-speaking proportion of the population High levels of rural wages The proximity of cities (over 50,000 inhabitants) II. Less secure statistical results High levels of partible inheritance

partible inheritance were available from an 1896–1902 Prussian survey which recorded what was actually happening to property which changed hands as a result of inheritance. This is preferable to the use of a classiWcation of regions according to an assessment of the predominant system which does not record actual practice, since practice may diVer from custom. It is the actual splitting of holdings which is likely to have the strongest eVect on demographic behaviour.32 However, it does mean that the main areas of partible inheritance in the south and southwest are excluded from the analysis. The econometric analysis of this data set is reported in Appendix 5D. The results are summarized in Tables 5.2, 5.3, and 5.4. The Wrst of these shows the results of the analysis of total fertility: the numbers of children aged under 1 year relative to the total number of women aged 15–50. The main Wndings are that total fertility is highest in areas with a high Polish-speaking population, in areas where wages are relatively high and in areas where there were large farms and estates. In the rural east, the presence of large numbers of Poles, and of larger estates, would tend to raise fertility, but lower wages would have the opposite eVect. These results support the view that the institutions of eastElbian estates, in particular the Scharwerker system, did raise fertility, and thus lead to out-migration. Partible inheritance, on the other hand, does not appear to have led to an increase in fertility; on the contrary, fertility was lower in the 32 The survey was published in ZdKPSB (1905); studies of inheritance systems include Miaskowski (1882) and Sering (1899–1910). In the partial inheritance areas of southern Germany the problem of excessive subdivision aVected the attitudes of parents to later oVspring, and may have contributed to high infant mortality. Lee (1981), 33 and 96.

Demography and Migration


Table 5.4. Summary of Wndings from statistical analysis of census results for 424 rural Prussian Kreise 1895: analysis of fertility relative to the number of married women Factors which increase fertility measured by the number of children under 1 year relative to the number of married women of child-bearing age I. Secure statistical results A large Polish-speaking proportion of the population A high level of Catholicism Presence of large farms and estates High levels of rural wages Low levels of partible inheritance Large distances from cities (over 50,000 inhabitants) II. Less secure statistical results A large Lithuanian-speaking proportion of the population

Prussian areas which practised divided inheritance. If partible inheritance led to earlier marriage, then this did not feed through into higher birth rates. The analysis of marriage rates (summarized in Table 5.3) shows that partible inheritance did tend to be associated with a higher marriage rate, though the result is not a particularly secure one, and the size of the eVect is not large. The analysis also shows that Catholic areas tended to have low marriage rates, outside the Polish-speaking areas. This indicates that marriage restriction may not have been the only reason for low marriage rates in Catholic areas in the south and southwest: Catholics inside Prussia also had lower marriage rates. Marriage rates were higher in areas with good economic prospects: in Kreise with high wages and in areas close to major cities which would have beneWted from better access to urban markets and improved employment opportunities. However, the proximity of major cities did not lead to a rise in overall fertility. If the measure of fertility is altered to reXect the impact of variations in the marriage rate—dividing the number of children under 1 year by the number of women aged 15–50 who are married—then the statistical analysis shows that fertility on this measure was lower in areas close to major cities and higher in remoter areas (Table 5.4). This Wgure cannot be regarded purely as a measure of fertility within marriage since it will include children born outside marriage as well (these cannot be distinguished from the census results as published). However, it


Demography and Migration

will reXect changes in marital behaviour. The Wnding suggests that married couples in areas close to cities were restricting fertility and this was cancelling the eVect of higher marriage rates Similarly, in areas of partible inheritance, the tendency for marriage rates to be higher was cancelled by low rates of fertility relative to married women. The fear of excessive subdivision may have led married couples in these areas to restrict family size. Outside the Polish area, Catholic families followed the opposite strategy: high rates of fertility relative to married women were countered by low marriage rates. There were therefore three factors which had strong eVects on demographic behaviour, but which had less eVect on overall fertility, and hence on overall population growth, largely because of cancelling eVects from marriage rates on the one hand, and reproductive behaviour within marriage on the other. These factors were Catholicism, the proximity of large cities, and partible inheritance. This leaves three other factors which did have substantial eVects on overall fertility: high wages, the presence of large estates, and the Polish-speaking percentage. In these cases, the eVect on marriage rates and the eVect on fertility relative to married women both ran in the same direction. The impact of this on overall population growth is illustrated in Table 5.5, which gives predicted values for diVerent areas, distinguishing between the implied rate of fertility in areas dominated by larger estates and that for areas where smaller holdings predominated.33 Table 5.5. Predicted values for total fertility (numbers of children aged under 1 year per 1,000 women of child-bearing age)

Average values Polish areas Non-Polish areas Daily wages of 1.51 Marks Daily wages of 1.08 Marks

On large farms and estates (over 100 hectares)

On other holdings (under 100 hectares)

66.0 79.6 64.3 68.4 63.3

57.9 71.5 56.2 60.3 55.2

Note: The predictions use the coeYcients given in column a of Table 5D.1. 33 These are illustrative Wgures, which assume that the behaviour of the nonagricultural population in Kreise dominated by larger estates followed the example set by

Demography and Migration


Outside the Polish areas, fertility in the rural east was held back by low wages: the predicted level for a larger estate paying the average wage east of the Elbe (1.08 Marks) is only slightly higher than that for a smaller holding in the western area where wages reached the average for Kreise west of the Elbe (1.51 Marks). However, in the Polish areas population growth was much higher, despite low wages. The previous section raised the question why population growth was high in the east even in regions which had few opportunities for nonagricultural employment, creating a labour surplus which had to migrate to Wnd work. A contrast was drawn between the east and similar regions in the south, where population growth was restricted. The analysis presented in this section provides answers to this question. It shows that there was a tendency for total fertility to fall in regions where wages were low, but the operation of this adjustment mechanism in the east was limited by other structural factors: 1. A higher proportion of women were married in the east; the east does not appear to have practised marriage restriction as a means of population control to the same extent as the German-speaking Catholic areas of southern and western Germany, or the Catholic areas inside Prussia. 2. The Catholic population of the east was predominantly Polishspeaking; this minority had high fertility. 3. There was a tendency for fertility to be higher in rural areas where the farm structure was dominated by large farms and estates; this was much more common in the east. 4. Partible inheritance was rare in the east; this was a factor which tended to reduce overall fertility.34 But the demographic experience of the east can be seen in a wider context. Both partible inheritance and the presence of large farms and estates are related to the overall distribution of property and income. In practice, inequality in the distribution of income in rural areas is closely the estate labourers. This may, however, have been the case. If the presence of large estates led to a change in demographic behaviour only on these estates, then an interaction term, combining the percentage of land in larger holdings and the percentage of the total population occupied in agriculture should have a signiWcant and positive eVect on overall fertility. A variable of this type was tried, but it had no signiWcant eVect. 34 The evidence suggests that partible inheritance areas were quicker to move to some form of fertility restriction within marriage in the late nineteenth century. In the earlier part of the century, this pattern of behaviour was not widespread in rural areas, so that the demographic problems of the south and south-west led instead to a policy of fertility control by marriage limitation (which was encouraged by legislation).


Demography and Migration

linked to inequality in the distribution of land. And, over time, partible inheritance should tend to reduce inequality in property distribution. Modern demographic studies of developing countries have shown a link between high inequality and population growth. The Wndings reported in this section point in the same direction.35

5. M i g r a t i o n a n d P o p u l a t i o n G r o w t h i n R u r a l Ar e a s The eVect of population growth on migration can be studied using a data set for the Prussian Kreise prepared by the Prussian Statistical Bureau in the interwar years and used in an article by Peter Quante, published in 1933.36 This made estimates of the rate of migration out of the population occupied in agriculture by applying to the agricultural population rates of fertility and mortality recorded for the population residing in rural areas. These rates of natural increase were then used to derive estimates of net migration, using the indirect method (subtracting the estimated rate of natural increase from the actual increase as recorded by the occupational censuses).37 There are some problems with this approach. The population occupied in agriculture was a large component of the total population resident in rural areas but there were others, occupied outside agriculture, who also lived in rural areas. If rates of mortality or fertility for this rural, non-agricultural, population were to diVer markedly from those for the agricultural population, there would be a bias in the estimates.38 The second problem is that, although the data set is described as one which gives migration out of agriculture (landwirtschaftliche Abwanderung) it is not in fact one which necessarily records migration. Someone who switched from agricultural employment to a non-agricultural occupation without leaving the Kreis would be recorded as a move. Given the nature of agricultural employment in the late nineteenth century, 35 The empirical evidence, and the theoretical reasons for expecting such a link, are surveyed in Birdsell (1988). 36 Quante (1955); the data was also used in Quante (1958). 37 These estimates are derived from the occupational censuses, which were conducted in June. These Wgures would therefore include some movements which were purely seasonal (such as migration to sugar beet areas to do hoeing and other cultivations) although the main period of seasonal migration came later in the year, at harvest time. 38 In 1907 55.2% of the German population was occupied in agriculture, while 41.5% lived in rural areas (taking an average of the 1905 and 1910 censuses). So around 15% of the rural population had an occupation outside agriculture.

Demography and Migration


which typically involved living on the farm where the worker was employed, a switch to non-agricultural work would usually involve a move of some description, but this could be no more than a shift to a nearby town or village. The data set is best described as one which gives the net agricultural ‘quitting rate’: the rate at which persons occupied in agriculture switched to other occupations. The results of this analysis are summarized in Table 5.6; the full econometric results are given in Table 5D.3 in Appendix 5D. The most striking result from this analysis is the close relationship between natural population increase and migration. The estimated relationship indicates that at least 100% of any population increase in rural areas was exported to other sectors. There was no capacity to absorb additional labout within agriculture. Any population growth fed directly into migration (at least within the Kreis). The quit rate was heavily inXuenced by the position of the Kreis; it was much higher in areas close to the border with Russian Poland and fell progressively for Kreise further and further to the west. This would have led to a high rate of migration out of the east but for the tendency for the quit rate to be higher on smaller holdings than on larger estates. Both the presence of large estates (over 100 hectares), and, to a lesser extent, the presence of medium-sized holdings (20–100 hectares) are associated with lower quit rates than those on smaller holdings of below 20 hectares. Table 5.6. Summary of results of statistical analysis of moves out of agriculture for 390 rural Prussian Kreise, 1882–95 I. Secure statistical results Factors which increase net moves out of agriculture: A high rate of natural population increase Being closer to the border with Russian Poland Factors which decrease net moves out of agriculture or lead to net gains: Presence of large farms and estates (over 100 hectares) Presence of medium-sized farms (20–100 hectares) A large Polish-speaking proportion of the population A large Lithuanian-speaking proportion of the population II. Less secure statistical results Factors which increase net moves out of agriculture: High levels of non-agricultural employment in the Kreis


Demography and Migration

This Wnding contrasts strongly with the view expressed by many contemporary writers that large estates led to a higher rate of outmigration. As the previous section has shown, one explanation of this is that large estates were associated with higher birth rates and it was this, not any tendency to shed labour as such, that produced an association between large estates and out-migration. This issue will be considered in more detail in the next chapter. Quit rates were also lower in the Polish and Lithuanian areas. In these areas population pressure was high, particularly so in Polishspeaking areas, and this was the main factor driving out-migration. Table 5.7 gives the predicted quit rates for the 1882–95 period for Kreise at diVerent distances from the border with Russian Poland, contrasting the predicted rates for smallholdings and larger estates. In areas close to the border with Russian Poland the predicted rates are considerably higher than in the centre of Germany (approximately 500 km from the border). The predicted quit rates are also much lower for the Polish areas, but the calculations are based on the assumption that the rates of natural population increase were similar, and they were in fact much higher in these areas. If the Polish-speaking population had had a similar rate of population increase to the Germans living in the rural east, then Polish migrants would have been a relatively unimportant factor in the overall pattern of German migration. Table 5.7. Predicted quit rates (per 1,000 of the total agricultural population) for rural areas with similar rates of natural population increase, based on analysis of Wgures for Prussian Kreise 1882–95 On large farms and estates (over 100 hectares)

On small farms (under 20 hectares)

Average values Distances from the border with Russian Poland: 50 km 500 km



225.8 148.0

372.7 294.9

Polish areas Non-Polish areas

69.2 182.6

214.9 328.3

Note: The calculations use the coeYcients from column c of Table 5D.3, and average values for all other variables.

Demography and Migration


On average the quit rate for Kreise east of the Elbe was 21% higher than for Kreise west of the Elbe (284.5 per 1,000 versus 235.5 per 1,000). About half of this can be attributed to higher rates of population increase feeding directly into migration. The remainder was due to other factors. Chief amongst these was the proximity of the border with Russian Poland. Cheap seasonal migrant labour from Russian Poland could be used to displace permanently employed workers. The availability of this labour source lowered wage rates in the areas close to the border, providing an additional reason for farm labourers in these regions to look for employment elsewhere in Germany.

6. Po p u l a t i o n G r o w t h in t h e R ecip ient C i t i e s The regressions in Chapter 4 showed that the rate of natural population increase of recipient cities was an important inXuence on migration rates into those cities. Table 5.8 compares lists of the 20 cities with the highest rates of natural population increase and those with the lowest rates (selected from a list of 96 for which Wgures were available). The signiWcance of these diVerences is evident when compared to the average rate of growth of German cities in this period, which was 2.6% per annum (total increase of population from all sources). The cities in the Wrst list could have supplied much of this increase from the natural increase of the resident population; those in the second list would have needed to take in migrants. A few points emerge from this comparison. The cities in the Wrst list are predominantly industrial cities, with a heavy concentration in the Ruhr. Those in the second list are well scattered around the country, and tend to be longer-established cities, often the capitals of their regions, or cities which for other reasons might be expected to have a strong middle-class presence.39 There are two views of the demographic relationship between the cities and their surrounding regions. The Wrst stresses the leadership role of the cities, and of the middle class within these cities. Middleclass attitudes and behaviour would Wrst aVect the working class living within the cities, and then transfer to the surrounding rural areas. The 39

Laux (1989) shows that natural population growth rates were lower in cities which were predominantly commercial or administrative, and these cities tended to have higher rates of in-migration in consequence.


Demography and Migration

Table 5.8. German cities with highest and lowest rates of natural population increase 1908–12 20 cities with highest rates

20 cities with lowest rates

Annual rate of increase (%) Hamborn Borbeck Herne Gelsenkirchen Recklinghausen Oberhausen Bochum Ko¨nigshu¨tte Essen Gleiwitz Dortmund Du¨sseldorf Rixdorf Mu¨lheim a.d. Ruhr Mannheim Plauen Ludwigshafen Duisburg Saarbru¨cken Osnabru¨ck

3.16 3.09 2.93 2.93 2.63 2.61 2.51 2.31 2.21 2.18 2.12 2.10 2.04 1.99 1.92 1.88 1.86 1.86 1.75 1.74

Annual rate of increase (%) Potsdam Hildesheim Regensburg Ulm Go¨rlitz Mu¨lhausen Frankfurt a. Oder Metz Braunschweig Liegnitz Dessau Rostock Koblenz Bromberg Wiesbaden Berlin Solingen Augsburg Wu¨rzburg Brandenburg

0.28 0.34 0.47 0.51 0.51 0.62 0.66 0.70 0.72 0.73 0.75 0.76 0.76 0.77 0.79 0.82 0.83 0.83 0.84 0.84

Source: Calculated from SJDS (1911), table xii, pp. 56–7, and subsequent issues.

second view is the one expressed by Ko¨llmann: migrants from rural areas bring their behaviour patterns with them to the cities. The results reported in Tables 5.2, 5.3, and 5.4 showed that there was a relationship between demographic behaviour and the proximity of major cities. In areas nearer to cities the marriage rate was higher, but fertility relative to the number of married women was lower compared to more remote regions. This could be interpreted as supporting the view that changes in demographic behaviour were initiated in the cities and then moved outwards to contiguous areas. However, the analysis also indicated that fertility in rural areas responded to factors which were speciWc to those areas: the presence of large estates and the

Demography and Migration


institutions associated with these in eastern agriculture, and the prevalent inheritance system. Demographic behaviour outside cities was not just a lagged response to the changes taking place in urban areas. If migrants from rural areas did transfer demographic behaviour from their regions of origin then it should be possible to Wnd this eVect by statistical analysis. The birth rate in rural areas was higher than in cities, and the diVerence was increasing. The crude birth rate in Prussian cities was 97.6% of the rural level in 1876–80 but only 82.5% in 1906–10.40 Despite this it was not possible to Wnd a statistical relationship between migration and urban fertility. Two data sets were tried. The Wrst made use of the 1895 census results for the major German cities (showing the proportion of the population born in rural areas); the second used Wgures for movements into German cities in the period 1905–10.41 One reason for this may be that the demographic behaviour of migrant families diVered in ways which tended to cancel each other out. Migrants tended to marry later, which should have reduced overall fertility.42 But they may have been more reluctant to adopt ‘modern’ patterns of reproductive behaviour, which would have pushed fertility up. Studies based on household Wgures show relatively small diVerences in the size of completed families.43 Demographic behaviour was inXuenced by conditions within the cities. Table 5D.4 in Appendix 5D gives the results of an econometric analysis of the birth and death rates in 79 German cities in 1910–12. These are summarized in Table 5.9. The main inXuences came from the occupational structure of the cities. In cities which had high levels of industrial employment—which also meant relatively low levels of employment in the professions and other middle-class activities, and in domestic service—the 40

SJPS (1915), 55–7. Using the 1905–10 data set, no signiWcant coeYcients were found either for net migration or total moves per head of population. The dependent variable was the crude birth rate. For 1895, a relationship was found between the average marital fertility of the regions from which migrants were drawn (using the Princeton Ig index) and marital fertility in the recipient cities. The problem here is that most migrants were drawn from the provinces within which the cities were situated, and so this result could also be interpreted as demonstrating the inXuence that demographic behaviour in cities had on the surrounding countryside. An important drawback with this approach is that it does not consider the actual behaviour of migrant families, only the aggregate Wgures for cities with high levels of migration. A family-level study may well show diVerent results. 42 Lee and Marschalck (2000), 383–4. 43 Lee and Marschalck (2000), 384 report that completed family size was 4.3 children for in-migrants in the city of Bremen 1820–75, but 4.6 for native-born households. 41


Demography and Migration

Table 5.9. Summary of results of statistical analysis of birth rates and death rates in 79 German cities, 1910–12 Secure statistical results I. Factors which are associated with high birth rates: A high level of industrial employment relative to other occupations A high level of employment in mining and metal production relative to other industrial occupations A low Socialist vote in the 1912 election II. Factors which are associated with low death rates: High wages A high level of industrial employment relative to other occupations A low level of employment in mining and metal production relative to other industrial occupations (less secure result) III. Factors which are associated with high rates of natural increase: A high level of industrial employment relative to other occupations A high level of employment in mining and metal production relative to other industrial occupations High wages A low Socialist vote in the 1912 election Factors which have no consistent eVect on the rate of natural increase: Catholicism A high level of migration

birth rate tended to be higher. It was higher still for the cities dominated by ‘heavy industry’, mining and metal production. There was an association between low birth rates and a high vote for the Social Democratic party in the 1912 election. These were both indicators of a change in working-class attitudes. Arguably, both were signs of a greater self-conWdence: an attempt on the part of workingclass households to control or at least inXuence their own destiny. They were no longer prepared to accept that their lives should continue to be dominated by forces outside their control. This turned out to be a more important and consistent eVect than Catholicism, which had little eVect on demographic behaviour once industrial structure had been included in the statistical analysis. Death rates were also inXuenced by industrial structure and by wage levels: low death rates were found in cities with relatively high levels of industrial occupation compared to other forms of employment, and in

Demography and Migration


cities with high wages. On the other hand, the heavy industry sector was associated with higher death rates when compared to other industries (although the result is not a particularly secure one in statistical terms). This was most probably due to the nature of the work, and the illnesses and accidents which resulted from this. The rates of natural increase which resulted from these birth rates and death rates show eVects from the same factors: population growth was highest in industrial cities, with high wages, and relatively low socialist votes. The positive eVect of high birth rates in the ‘heavy industry’ cities outweighed the tendency for these to have higher death rates, so the overall eVect from this factor on population growth was a positive one. Again the analysis shows no consistent eVect from Catholicism, and migration, as measured by moves in and out relative to population, was not found to have any eVect on demographic behaviour. Table 5.10 illustrates these points for ten selected cities, those with the highest and lowest rates of population increase. The list of cities with low rates of increase includes several with small Catholic percentages, but alongside them is Mu¨hlhausen, which was 75.8% Catholic.

Table 5.10. Predicted rates of population increase (per 100,000 resident population) for selected cities, 1910–12 Catholic %

Wages 1912

Industry %

A. 5 cities with lowest rates of increase Potsdam 8.7 3.25 32.8 Go¨rlitz 14.0 2.50 49.7 Mu¨lhausen 75.8 2.40 63.3 Wiesbaden 31.8 3.20 37.4 Brunswick 6.5 3.20 50.2 a v e r ag e 27.3 2.91 46.7 B. 5 cities with highest rates of increase Beuthen 55.1 3.40 66.4 Duisburg 52.8 3.25 65.8 Bochum 50.4 3.30 68.2 Gelsenkirchen 49.7 3.40 74.3 Recklinghausen 75.2 3.30 73.5 a v e r ag e 56.6 3.33 69.6

Mining/ metal %

0.15 0.13 0.01 0.13 0.13 0.11 42.8 33.0 45.8 56.5 64.2 48.5

SDP vote 1912

Rate of increase

47.4 48.8 56.1 35.8 55.6 48.7

23.3 38.7 40.3 50.7 53.7 41.3

35.7 30.9 36.8 36.8 27.3 33.5

189.7 192.3 229.0 252.0 262.5 225.1


Demography and Migration

Other Catholic cities with low rates of population increase were Wu¨rzburg, Augsburg, and Koblenz. It is this inconsistency that makes the relationship between Catholicism and population growth such a weak one. On the other hand the low growth cities are distinguished by generally low levels of industrial employment, relatively low wages and high SDP votes. But above all, they are not cities which have large mining or metal production sectors. By contrast the cities with high rates of population growth are all ones where the heavy industry sector was important. They tended to have lower socialist votes, but wages were high. The classiWcation of regions given in Appendix 5A shows that, over the whole 1871–1910 period, comparison of the death rate and the lagged birth rate identiWes the Ruhr regions, Arnsberg and Du¨sseldorf, as two of the regions which were both highly industrial and which had high population growth. But a comparison with other industrial regions shows that the Ruhr was not so unusual when values for the whole 1871–1910 period are considered. Population pressure was as high in parts of Saxony, Leipzig, and Zwickau, for example. What made the natural rate of increase in the Ruhr so distinctive is that it remained high while population growth in other urban areas was tending to fall. The more isolated working-class culture of the Ruhr was a factor in this delayed demographic transition. 7. P o p u l a t i o n G r o w t h a n d t he L e w i s M o d e l So far the analysis in this chapter has concentrated on relative diVerences in population growth between regions. For migration this is the most important factor. Is the pattern of demographic change in line with the distribution of economic opportunities? If it is not, then there will be an adjustment role for migration. But, in the context of the Lewis Model, the aggregate level of population increase also matters. Lewis argued that demographic response would be a component of the ‘unlimited supplies of labour’ assumed by his model. He placed most emphasis on the decline in the death rate: Because the eVect of development on the death rate is so swift and certain, while its eVect on the birth rate is unsure and retarded, we can say for certain that the immediate eVect of economic development is to cause the population to grow; after some decades it begins to grow (we hope) less rapidly. Hence in any

Demography and Migration


society where the death rate is around 40 per thousand, the eVect of economic development will be to generate an increase in the supply of labour.44

The reasons Lewis gave for the ‘swift and certain’ decline in the death rate were, Wrst, that improved communications would reduce the incidence of local famines, and secondly, that there would be improvements in public health and medicine. The relationship between development and the death rate does not appear quite so certain now as it did to Lewis: rather more is known about high urban mortality in industrializing cities. Life expectancy in German cities remained low into the twentieth century. Table 5.11 gives life expectancy at birth for the Prussian population, as calculated by the Prussian Statistical OYce using mortality tables for 1901–5. This Table 5.11. Life expectancy at birth, Prussia 1901–5 Men

East Prussia West Prussia Berlin Brandenburg Pomerania Posen Silesia Pr. Saxony Schleswig-Holstein Hanover Westphalia Hesse-Nassau Rhineland


Towns and cities

Rural areas

Towns and cities

Rural areas

37.6 37.5 41.8 41.7 37.9 38.7 36.5 41.1 44.9 44.7 41.7 45.2 42.9

42.0 43.4 43.2 46.5 45.5 40.0 44.9 52.9 50.3 46.6 49.6 45.8

44.8 42.9 47.3 47.0 43.6 43.9 41.7 45.2 49.7 49.4 46.6 49.5 47.6

45.2 45.9 47.5 48.6 48.5 43.5 47.7 54.8 51.4 48.6 50.9 47.7

Note: The reasons for high mortality in the east are discussed by Spree (1988), 58. See also Hennock (2000) for a discussion of attitudes to urban sanitation. These Wgures may understate the problems of eastern cities. Appendix 5B gives life-expectancy Wgures for individual cities. The results show that male life expectancy in the major eastern cities was 29–35 years. There had been an improvement in urban life expectancy since the 1870s, but this was still below rural life expectancy. Source: Zdp/SB (1908), i. 26–7. 44

Lewis (1954), 144.


Demography and Migration

table shows that in all regions, and for both sexes, life expectancy in urban areas was below that in the countryside, and the gap could be a substantial one: 7 years for men in Posen, 81⁄2 years in Pomerania. National averages tend to obscure this diVerence, as the urban population was concentrated in the west and centre, regions where life expectancy was high, while the east had proportionately more of the rural population. The ‘centre of gravity’ of the rural population was further to the east compared to that of the urban population.45 One evident conclusion is that, in the absence of balancing variations in fertility, cities would have to draw in migrants from rural areas to make up for their higher mortality rates. Lower rural mortality should have made it easier to Wnd migrants from the countryside nearby. Migration over longer distances had a mixed eVect on mortality. On the one hand, it transferred population from rural areas to cities, which would tend to increase mortality. On the other, it moved population from the east to the west, which should have led to a reduction. But, as most moves were over shorter distances, and did result in net gains by cities, it can be seen that the overall eVect would have tended to raise mortality. If urbanization tended to raise mortality, then the decline in the German death rate from 1870 onwards becomes even more impressive. However, by European standards German life expectancy in the period just before the First World War was not high. At 49.0 years in 1910–11 (both sexes) it was below France (50.4), England and Wales (53.5), and the Netherlands (56.1). In Sweden, Norway, and Denmark life expectancy was over 57 years. Comparative studies show that the decline of mortality in Germany was relatively late; in other Western European countries it had started earlier.46 45 Life-expectancy tables for the early 1900s show a substantial diVerence between urban and rural areas for the male population at all ages, while there was little diVerence for women past the Wrst year of life (urban infant mortality was high for both sexes). For men the gap was 3.8 years at birth and 3.6 years at age 10, Vo¨gele (1998), 50–1. This calculation, for the population as a whole, underestimates the gap between urban areas and contiguous rural areas. 46 SchoWeld et al. (1991), 47. There had been considerable progress in reducing mortality in German cities since the 1870s: see Vo¨gele (1998) and Vo¨gele and Woelk (2000) for studies of this. The fact that German life expectancy was still not high by European standards is an indicator of the severity of the problems faced in the 1870s and 1880s. The high rates of migration in this period had contributed to a major crisis of mortality, which was only gradually overcome. By 1905, however, migrant life expectancy in the more advanced cites was below that of the native-born population, Lee and Marschalck (2000), 381.

Demography and Migration


On the other hand, Lewis’s ‘uncertain and retarded’ decline in fertility set in fairly quickly. The Princeton study showed that fertility within marriage started to fall from the late 1870s onwards. This was initially oVset by a rise in the proportion married. If there had not been this decline in marital fertility, then population growth in the 1880s and 1890s would have been much higher. The Princeton study showed that the decline in marital fertility started earlier in urban areas, but village studies, and the evidence of the regressions presented in this chapter, make the point that this was not a completely new phenomenon. Some rural areas had already reduced fertility within marriage. If rural areas tended to place more emphasis on marriage restrictions, it may well be that the reason for this was that the degree of social control required to maintain a high average age at Wrst marriage was easier to achieve in village communities. It seems therefore that the decline in marital fertility which began in the 1870s had its roots in attitudes and objectives which were also found in rural society, and which had already found expression in the Western European pattern of delayed marriage. EVectively, population growth was linked to economic objectives: the household unit sought to regulate population increase so as to achieve a desired standard of living for itself and its children. Late marriage was particularly linked to the institution of the nuclear family, and to the custom that children should set up a separate household on marriage. Parental approval of marriage would depend on the couple having suYcient means to do this, and to achieve a respectable standard of living in the new household.47 This implies that there was a gradual upwards adjustment of the objective. Even in pre-industrial Europe there was some increase in living standards, although this was intermittent and uncertain. If the household made no adjustment of the desired standard of living, then in times of prosperity there would have been a surge of population growth suYcient to wipe out any gains achieved. By and large this did not happen. If this argument is correct then the German rural population which moved into industrial cities carried with it attitudes which predisposed it to control fertility within marriage. The incentive for a married couple to move out of a family home consisting of a one- or two-room 47

This meant that eVective population control could be exercised simply by refusing permission to set up a new household, as happened in Mecklenburg for example, Lubinski (1995).


Demography and Migration

apartment in a Berlin Mietskaserne (or tenement block) was at least as great as the incentive to move in a rural village. The strategy may have changed. Increased employment opportunities for married women without children would have made it more attractive to marry early but put oV having children. But the target was the same.48 Moreover, the ability to adjust objectives would have also aVected behaviour. Cities oVered new opportunities: white-collar jobs, apprenticeships, accessible secondary schools. Even if migrants themselves were not able to take advantage of these possibilities, their children might, if there were not too many of them. In cities with a strong middle-class presence these opportunities may well have been greater, or at least more apparent.49 The demographic aspect of the Lewis Model was therefore curtailed. Germany had to deal with quite a high rate of population increase in the 1880s and 1890s, but this would have been higher still if cities in central Germany had not started to reduce fertility from the 1870s onwards. This in turn made these cities take in migrants from other areas where the shift in demographic behaviour was less advanced. The intention in this chapter has been to concentrate on those aspects of demographic behaviour which aVect migration. This meant that particular attention was given to mechanisms which might adjust population growth to available economic opportunities within a region. The main conclusion is that this did not happen in the rural east, and that this was an important reason why the region was a major source of migrants in the 1871–1913 period. The evidence of the regressions at the Kreis level shows that demographic factors had major eVects on migration. But this does not mean that the east was in a crisis of overpopulation in 1870–1913 similar to that of the early nineteenth century. There was less evidence of demographic distress. It was not just population pressure which led migrants to leave the east.

48 Neumann (1978) provides evidence of fertility control in urban working-class households. 49 The idea that marriage in Western Europe was linked to economic circumstances is fairly widely accepted. This may have been a long-established pattern of behaviour, as argued by Habbakuk (1958). But this has the implication that European households were accustomed to adjust demographic behaviour in general to economic circumstances, and may therefore have been unusually predisposed to reduce fertility in response to improved opportunities to invest in human capital, leading to the downward phase of the demographic transition. The extent of the responsiveness of demographic variables to economic conditions may be a European peculiarity.

Demography and Migration


The next two chapters will look more closely at conditions in the rural east. The position of the rural labourer will be considered. Evidence on wage movements, and on changing social relationship will be presented. The role of changing technology will also be examined. But it is important to remember that the east was not the only agricultural region. Changes were taking place in agriculture in other regions. It was the combination of the fact that the east was predominantly agricultural, with a rate of population increase which was high relative to opportunities in local agriculture or industry, that made migrants leave the east.

A p p e n d i x 5A. A C l a s s i f i c a t i o n o f G e r m a n R e g i o n s A. Highly industrial regions 1. Regions with high population growth: Arnsberg (43.0/23.3) Starkenburg (37.4/22.4) Zwickau (48.6/28.9) Brunswick (35.6/20.7) Dresden (40.3/24.7) Du¨sseldorf (41.0/22.8) Leipzig (42.3/25.1) Berlin (38.7/23.7) Reub -a.l. (42.4/25.5) Potsdam (38.1/23.9)

2. Regions with low population growth: Wiesbaden (33.7/20.5) Aachen (36.1/23.2) Bremen (35.5/22.9) Ko¨ln (38.7/24.9) Saxony-Coburg-Gotha (35.4/22.2)Hamburg (35.1/22.8) Lu¨beck (32.0/20.3) Rheinhessen (35.3/22.4) Bautzen (36.6/24.6) Reub -n.l. (40.3/26.2)

B. Moderately industrial regions 1. Regions with high population growth: 2. Regions with low population growth: Magdeburg (38.8/24.7) Sch.-Lippe (33.1/18.7) Neckarkreis (41.6/24.5) Hildesheim (34.0/22.3) Breslau (40.6/28.4) Saxony-Meiningen (35.9/21.4) Mannheim (38.0/24.8) Erfurt (39.2/22.9) Schleswig-Holstein (33.6/20.0)Schwarzburg-Rudolfs. (35.1/21.9) Upper Alsace (34.6/25.1) Anhalt (38.3/21.8) Saxony-Weimar (35.0/22.0) Mu¨nster (32.0/23.3) Merseburg (40.7/24.6) Saxony-Altenburg (40.9/27.9) Lorraine (29.4/22.0) Schwarzburg-Sond. (35.7/21.1)Hanover (35.8/21.7) Karlsruhe (38.0/24.5) C. Moderately agricultural regions 1. Regions with high population growth: Trier (37.5/22.5) Danzig (44.2/27.0) Frankfurt a. Oder (38.3/24.1) Stettin (40.1/24.3) Oppeln (43.7/27.3) Schwarzwaldkr. (41.7/26.7)

2. Regions with low population growth: Upper Alsace (34.7/24.0) Kassel (36.0/22.4) Oldenburg (33.0/20.7) Mittelfranken (38.4/27.6) Liegnitz (37.4/28.2) Koblenz (35.3/22.7)

Pfalz (38.1/22.8) Lippe (36.0/21.1)

Minden (35.8/21.8) Mecklenburg-Schwerin (32.9/20.4)

Oberpfalz (40.6/30.9) Donaukreis (41.5/27.9) Oberbayern (39.6/30.3) Stralsund (36.1/24.0) Mecklenburg-Strelitz (32.1/21.7)

D. Highly agricultural regions 1. Regions with high population growth: Ko¨nigsberg (40.7/22.0) Aurich (32.8/18.9) Gumbinnen (42.4/27.1) Marienwerder (46.7/26.7) Stade (34.7/21.0) Potsdam (44.2/25.1) Bromberg (45.8/27.1) Jagstkeis (40.3/25.8) Osnabru¨ck (34.3/20.8) Koslin (42.6/26.5)

2. Regions with low population growth: Konstanz (34.7/24.2) Waldeck (35.2/21.5) Unterfranken (34.2/24.1) Sigmaringen (37.8/26.1) Oberfranken (34.6/23.5) Freiburg (31.5/22.5) Schwaben (39.6/29.8) Oberhessen (31.2/20.7) Niederbayern (39.5/31.0) Lu¨neburg (31.0/20.8)

Note: Regions classiWed according to percentage occupied in agriculture 1871–1910. Figures in brackets are average birth rate per 1,000 lagged 20 years followed by the average death rate.

Demography and Migration


A ppend ix 5B. L ife Ex p e c t a n c y a t B i r t h i n Ma j o r C i t i e s , in Y e a r s , C a l c u l a t e d f r o m 1893–4 Mo r t a l i t y Ta b l e s Men East Prussia West Prussia Brandenburg Pomerania Posen Silesia Prussian Saxony

36.2 38.6 37.9 40.5 39.5 34.9 39.6

Ko¨nigsberg Danzig Berlin Stettin

Breslau Magdeburg Halle Schleswig-Holstein 43.9 Altona Hanover 43.2 Hanover Westphalia 40.9 Hesse-Nassau 42.6 Frankfurt am Main Rhineland 39.6 Ko¨ln Du¨sseldorf Elberfeld Barmen Crefeld Aachen All Prussia 39.0 Women East Prussia West Prussia Brandenburg Pomerania Posen Silesia Prussian Saxony

40.1 42.2 41.9 43.5 42.9 38.5 42.8

Ko¨nigsberg Danzig Berlin Stettin

Breslau Magdeburg Halle Schleswig-Holstein 46.4 Altona Hanover 44.8 Hanover Westphalia 42.9 Hesse-Nassau 44.7 Frankfurt am Main Rhineland 42.5 Ko¨ln Du¨sseldorf Elberfeld Barmen Crefeld Aachen All Prussia 42.1 Source: Kucyzinski (1897).

29.5 Rest of Province 36.6 33.6 Rest of Province 38.9 35.8 29.2 Rest of Province 41.0 30.0 36.2 36.0 37.0 41.0

Rest of Province 35.2 Rest of Province 39.8 Rest of Province 44.4 Rest of Province 43.3

38.5 Rest of Province 42.8 33.6 Rest of Province 40.1 37.5 39.7 42.0 36.6 35.0

35.8 Rest of Province 40.3 39.5 Rest of Province 42.4 41.1 35.2 Rest of Province 43.8 35.2 39.6 41.9 42.1 43.5

Rest of Province 38.6 Rest of Province 42.9 Rest of Province 46.7 Rest of Province 44.8

44.4 Rest of Province 44.7 38.0 Rest of Province 42.6 43.0 46.6 45.8 41.4 41.1

Demography and Migration


A p p e n d i x 5C. D e s c r i p t i o n o f R e g r e s s i o n Va r iabl es a nd So u r c e s (i) Tables 5.2 and 5.5 TOTFERTILITY






Numbers of children aged under 1 year per 1,000 women aged 15–50, 1895 census, PS (1897–8) vol. 148, pp. 60–176. As above, children per 1,000 married women aged 15–50. As above, percentage of women aged 15–50 who were married. The percentage of the total agricultural area in holdings of over 100 hectares, from 1895 Agricultural Census, SDR n.f. 212. The percentage of the population who were Catholic, 1890 census, PS, vol. 121, pp. 152– 98. As above, the percentage of the population who spoke Polish. As above, the percentage of the population who spoke Lithuanian. Average wages (Ortsu¨bliche Tagelo¨hne) for rural areas; the Wgures are averages of male and female daily rates, 1892 and 1901, ZdKPSB (1904), 320–8. The percentage of land transferred by inheritance in 1896–1902 which was divided, ZdKPSB (1905), 262–7. Distance in kilometres from the Kreis midpoint to the nearest city with at least 50,000 inhabitants in 1900 (measured from contemporary maps).

(ii) Tables 5.4 and 5.6 Net quit rate (dependent variable)

The estimated number of net moves out of agriculture, by Kreis, 1882–95 and 1895– 1907, per 1,000 of the average population occupied in agriculture, from Quante (1933), 38–58.


Demography and Migration



As above, estimated numbers of total births less total deaths, per 1,000 of the average population occupied in agriculture. As for RURALWAGES, using only 1892 Wgures. Shortest distance from the Kreis mid-point to the border with Russian Poland. Distance to the nearest city with at least 20,000 inhabitants in 1900.

(iii) Appendix Tables 5D.4 and 5D.5 Birth rates, death rates, rates of natural increase (dependent variables) CATHOLIC% WAGES1912





All Wgures are rates per 10,000 head of population, taken from SJDS (1911), 67–8; (1912), 70–1; and (1913), 76–7. The percentage of the population who were Catholic, 1910 census, SJDS (1911), 678–9. Average wages (Ortsu¨bliche Tagelo¨hne) for male workers, aged 16 and over, from SJDS (1913), 826. The share of the total occupied population (both sexes) occupied in industry, 1907 Occupational Census, SDR (1909), n.f. volumes 207 and 209. As above, the share of the industrial workforce occupied in the mining and metal production sector. Social Democrat share of the vote from Die Reichstagswahlen von 1912, SDR, vol. 250 (1913). Data from SJDS 1911–13: total moves for each city in 1910–12 divided by total population.

(iv) Appendix Table 5E.1 MARRIAGEt (dependent variable)

The number of marriages in each province per 1,000 head of population in each period (annual data), from Ko¨llmann (1980).

Demography and Migration



Dummy for war years (1866, 1870, and 1871). Average rye prices for each province, from RYEPRt ZdKPSB (1907), 84–92. As above, for potatoes, wheat, beef, pork, POTATOPRt, WHEATPRt, BEEFPRt, and butter. PORKPRt and BUTTERPRt Wages paid to male workers in Prussian state WAGESt forests, daily rate in Marks, from Eggert (1883), decade averages.

A p p e n d i x 5D. E c o n o m e t r i c R e s u l t s (i) The analysis of fertility in the rural Prussian Kreise The aim of this analysis was to examine factors which aVected fertility decisions in rural areas, so as to consider the eVect of agricultural conditions on migration when these factors operated indirectly— through their eVect on the demographic balance in the rural area. Migration would tend to occur when the supply of labour exceeded the ability of the local economy to supply employment. Information on fertility was derived from the 1895 census, by comparing the numbers of children to the numbers of women of child-bearing age. Two measures of fertility were used as dependent variables: TOTFERTILITY is the number of children aged under 1 year per 1,000 women aged 15–50; MARRFERTILITY is the number of children aged under 1 year per 1,000 married women aged 15–50. This second variable has the disadvantage that it includes all children, not just those born within marriage (these are not distinguished by the published census results). A third dependent variable is the percentage of women aged 15–50 who were married.50 The explanatory variables include two which measure the ethnic composition of the population, speciWcally the presence of Polish- or 50 In societies with high infant mortality these fertility variables will be biased due to deaths in the Wrst year. On average around 8% of children would have died before they were recorded by the census. However, being rather brutal about this, from a migration viewpoint, what matters is not the number of births but the number of children who survived. So this bias is not a serious problem.


Demography and Migration

Table 5D.1. Regression results for the analysis of fertility in rural Prussian Kreise, 1895, estimation by OLS, t-ratios in brackets Dependent variable is: a b c TOTFERTILITY MARRFERTILITY %MARRIED Constant LAND>100ha %CATHOLIC %POLISH %LITHUANIAN RURALWAGES N S R2 adj R2 F-test of regression RSS

89.5 þ0.214 (4.90) þ0.0406 (2.01) þ0.324 (9.44) þ0.216 (1.32) þ21.0 (6.92)

183.3 þ0.342 (4.56) þ0.494 (14.25) þ0.234 (3.96) þ0.508 (1.81) þ24.2 (4.64)

477 13.90 0.319 0.311 44.13 91148

477 23.89 0.442 0.436 74.84 269470

23.0 0.0007 (0.13) 0.0399 (15.5) þ0.0350 (8.00) 0.0244 (1.17) þ2.10 (5.42) 477 1.77 0.385 0.378 58.99 1485

Note: DeWnitions and sources are given in Appendix 5C. The analysis is a crude one by the standards of modern demographic studies: there is no attempt to measure fertility by cohort or to make allowance for diVerences in the age structure of the female population in the diVerent Kreise. The main reason for this was that the data needed to do this were not available from the sources used. The analysis serves the needs of this study, which is to examine the eVect of demographic factors on migration.

Lithuanian-speaking minorities. The inclusion of these variables is important if the eVect of Catholicism is to be accurately estimated. The coeYcients on %LITHUANIAN are not signiWcant, which is not so surprising as there were only six Kreise where there was a substantial minority (or majority—Lithuanian-speakers were 58% of the population of Kreis Heydekrug in East Prussia). The variable was left in the equations to avoid creating a bias in the other estimates. The behaviour of the Polish-speaking and German-speaking Catholics is shown to have been very diVerent. Overall fertility was 3% higher, compared to the Prussian average, for German Catholics, but

Demography and Migration


27% higher for Polish Catholics. The reasons for this are of interest. Although fertility within marriage was high for German Catholics, they were much less likely to be married (16% less likely than the rest of the Prussian population).Thus the practice of late marriage found in southern Germany was also found in the Catholic areas of Prussia (something which can be seen from Figure 5.7 as well). The presence of large farms and estates is shown to raise fertility, though it does not aVect the likelihood of marriage. The coeYcient implies that the rural population in areas where farms were over 100 hectares had fertility rates 16% higher than the rest of the population. But the eVect of this on the east was cancelled by the Wnding that low wages depressed fertility. In Table 5.D2 the eVect of adding additional variables is considered. The addition of %PARTIBLE (the percentage of land transferred by inheritance in 1896–1902 which was divided) and CITYDISTANCE(a) (the distance to major cities) did not have a large eVect on the other estimated coeYcients, although the overall eVect of Catholicism was reduced.51 The eVect of partible inheritance was to reduce overall fertility, which may surprise some. The likelihood of marriage was increased, however, which leads to the conclusion that, barring major changes in the percentage of children born outside marriage, there must have been a major diVerence in fertility within marriage in the regions where partible inheritance was common. The implied reduction in marital fertility is about 15%. What this suggests is that, in these regions, families were making eVorts to reduce fertility in order to avoid further subdivision of the family property.52 Remoteness, as measured by the distance from major cities, had the eVect of raising fertility within marriage, and reducing the percentage married. Rural areas closest to cities had a more ‘modern’ demographic pattern: earlier marriage, and lower fertility within marriage. The overall eVect on total fertility was not signiWcant. The same data set was also used for the analysis of mortality rates. The results were of no particular interest. Neither LAND>100ha nor 51

One reason could be that there were 55 Kreise observations which had to be dropped because %PARTIBLE was not available, and these were predominantly in the Catholic areas of the Rhineland. 52 According to a contemporary observer, Wydgozinski in Sering (1899–1910), i. 116, there was indeed a tendency in the partible inheritance areas to restrict fertility in order to avoid the excessive splitting of holdings: a ‘two-child’ system was practised (Zweikindersystem).


Demography and Migration

Table 5D.2. Regression results for the analysis of fertility in rural Prussian Kreise, 1895, estimation by OLS, t-ratios in brackets a b c TOTFERTILITY MARRFERTILITY %MARRIED Constant LAND>100ha

88.4 þ0.195 (4.34) %CATHOLIC þ0.0393 (1.61) %POLISH þ0.336 (9.17) %LITHUANIAN þ0.216 (1.31) RURALWAGES þ23.0 (6.58) %PARTIBLE 0.149 (2.94) CITYDISTANCE(b) 0.00013 (0.11) N S R2 adj R2 F-test RSS

424 13.9 0.345 0.334 31.4 80443

166.7 þ0.346 (4.52) þ0.0435 (10.45) þ0.267 (4.33) þ0.411 (1.47) þ33.0 (5.58) 0.330 (3.81) þ0.00601 (3.14) 424 23.7 0.436 0.426 45.9 233943

24.8 0.0040 (0.72) 0.0350 (11.50) þ0.0337 (7.43) 0.0147 (0.72) þ1.33 (3.05) þ0.0112 (1.77) 0.000658 (4.74) 424 1.72 0.333 0.321 29.6 1238

Note: This data set was also used to examine the eVect of out-migration on marriage rates. The inclusion of a term giving rates of out-migration between 1895 and 1905 improved the Wt, and the variable had a t-statistic of 11.00, indicating a highly signiWcant result. As the migration Wgures refer to a period after the census used for the dependent variable, the result cannot be regarded as an entirely satisfactory one, and there must be a question about the direction of causation. For this reason, the result has not been reported, although it certainly suggests that marriage rates were depressed in regions of out-migration.

%PARTIBLE had any signiWcant eVect on mortality. The only clear result was a tendency for mortality to rise where wages were low. (ii) Analysis of migration by Kreis, 1882–1895 The connection between migration and population growth can be examined by making use of an analysis conducted by the Prussian Statistical

Demography and Migration


Bureau and used in an article in 1933.53 This made indirect estimates of the rate of migration out of agriculture by applying rates of fertility and mortality recorded for the rural population. These estimated rates of natural increase for the population occupied in agriculture were then used to derive estimates of net migration, using the indirect method (comparing census returns with the predicted natural increase). The dependent variable is best described as the net agricultural ‘quitting rate’: the rate at which persons occupied in agriculture switched to other occupations, since it is not clear whether or not the people who left agriculture actually migrated. It may be surmised that most did make some move: although some may have found non-agricultural employment in their village, the majority probably had to leave for a local town at the very least. But many may not have left the Kreis, particularly if this was in an area where non-agricultural employment was rising rapidly. The data set makes it possible to examine the eVect of variations in population growth on the net quitting rate, along with the inXuence of other variables, such as agricultural structure. Bivariate correlation analysis shows a positive relation between the presence of large farms and estates (where LAND>100HA is high) and the net quitting rate: outward migration was apparently higher on the larger estates. However, in multiple regression analysis, this eVect is reversed once the rate of natural population increase is introduced into the estimated equations. In other words, the apparent relationship between large estates and outmigration is entirely due to demographic factors. The Wrst column of Table 5.D3 gives the results of Ordinary Least Squares estimation of the basic equation. The estimated coeYcient on LAND>100HA is negative, as is the coeYcient on LAND20–100HA, though smaller. Large estates were the ones with the lowest quit rates; quit rates were highest on holdings of less than 20 hectares. The quit rate was reduced by the presence of Poles or Lithuanians. The fact that Poles, in particular, have Wgured quite heavily in accounts of migration, has much to do with high birth rates: these non-German groups had no great inclination to migrate, and were rather less likely to leave agricultural employment than Germans in similar circumstances. It was population pressure which drove up migration from the Polishspeaking areas.54 53

Quante (1955); the data were also used in Quante (1958). If NATINCREASE is omitted from the estimated equation, the coeYcient on %POLISH becomes positive and signiWcant. 54


Demography and Migration

Table 5D.3. Regression results for moves out of agriculture by Kreis, 1882–95 Estimation method




IV c

IV d

184.4 1.245 (6.42) 0.585 (3.76) 0.507 (3.19) 0.161 (10.30) 0.826 (5.33) 1.687 (2.99) þ1.051 (21.39)

182.2 1.230 (6.30) 0.565 (3.58) 0.540 (3.27) 0.159 (10.00) 0.843 (5.38) 1.698 (3.00) þ1.043 (20.70) þ0.110 (0.76)

155.9 1.469 (4.10) 0.643 (3.65) 0.361 (1.42) 0.173 (7.95) 1.109 (2.70) 1.749 (3.01) þ1.218 (5.32)

190.4 1.550 (3.84) 0.624 (3.10) 0.431 (2.15) 0.154 (5.21) 1.183 (2.25) 1.711 (2.94) þ1.271 (3.66)


0.17 (0.02) 0.39 (0.84)


390 46.45 0.672 0.666 111.84 824158

390 46.47 0.673 0.666 97.82 822914

390 47.15 0.474 0.464 49.16 849148

339 46.41 0.483 0.469 34.12 708756

Note: Dependent variable is the estimated net quitting rate per 1,000 for the total agricultural population.

The negative coeYcient on %AGOCCP1882, the percentage occupied in agriculture, shows that the availability of opportunities to Wnd non-agricultural employment within the Kreis increased the quit rate. This, however, would not lead to increased net out-migration. In the

Demography and Migration


more industrial areas the quit rate was high, but many of those leaving found non-agricultural work within the locality. But the most important result to emerge from this analysis is the importance of the rate of natural population increase. The eVect is a large one. The estimated coeYcient in column a is just over 1.0, indicating that, in Kreise with above-average rates of population increase, over 100% of this surplus left agriculture. There was no capacity to absorb surplus labour within agriculture. In regions where non-agricultural employment was increasing, surplus labour could Wnd work without migrating. In other regions where there were limited non-agricultural opportunities the surplus had to migrate. Demographic pressure fed directly into migration in these regions. There are problems of possible endogeneity bias in this equation. High migration rates aVect the population balance and this will have an inXuence on the rates of natural increase. In areas with high inward migration, the population will tend to be younger, and this will reduce the crude death rate, pushing up the rate of natural increase. One way to deal with this is to use instrumental variables, if a suitable instrument can be found. The distance to large cities turned out to be a possible instrument. Its inclusion had little eVect on the estimated equation (column b) but it was well correlated with the rate of natural increase.55 Using this as an instrument produced some changes in the estimated values (comparing column a with column c), and reduced the overall Wt of the equation, but pushed up the coeYcient on NATINCREASE to over 1. The new coeYcient suggests that 120% of the natural increase in the population left agriculture. This implies that the overpopulated Kreise were releasing a surplus built up in previous periods of high population growth as well as the surplus created by high current rates of population increase. The Wnal column introduces other variables. There is no eVect from partible inheritance on the quit rate. High rural wage levels have a small negative eVect on the quit rate, but the coeYcient is not signiWcant. (iii) Analysis of demographic behaviour in urban areas Econometric analysis of demographic variables for German cities can be used to show the eVect of industrial structure and other factors on rates 55

A correlation coeYcient of 0.274 (p-value of 0.000).


Demography and Migration

of natural increase. Cities with low rates of natural increase would need higher rates of inward migration to provide the increased labour force needed for industrial expansion. The basic model uses three variables: the Catholic percentage of the total population, the level of wages, and the percentage of the occupied population who are occupied in industry. In column a of Table 5D.4 the dependent variable is the birth rate for the years 1910–12. The results show that the birth rate was highest in Catholic industrial cities. The Wnding that the industrial percentage has an eVect on the birth rate is consistent with the view that the working class in these cities was relatively isolated and therefore less likely to emulate middle-class reproductive behaviour, although other explanations cannot be ruled out. In column b the eVect of industrial structure is examined by including MINING/METALS% (the percentage of the industrial workforce occupied in mining and metal production) amongst the explanatory variables.56 The inclusion of this variable has an eVect on the coeYcients estimated for the other variables. The eVect of Catholicism is greatly reduced, as is the eVect of industrial employment in general. The mining and metal production sector was characterized by largescale enterprises and a strong ‘isolated’ working-class culture. The eVect of Catholicism is further reduced, eVectively eliminated, by the inclusion of SDPVOTE1912, the Social Democrat vote in the 1912 election as a percentage of all votes. This should not be interpreted as a direct causation; rather, both changed demographic behaviour, and a willingness to vote for the SPD, should be taken as indicators of changes in working-class mentality. The more conWdent, and independent, working class which was emerging in many German cities was prepared to express this in votes for the more radical programme of political and social change espoused by the SPD. Similar changes in social attitudes led to an alteration in reproductive behaviour, and a reduction in family size, which led to a fall in the birth rate. The Wnal column of Table 5D.4 gives the results of a similar analysis for the death rate. Here, the importance of wage levels is much greater, having had an insigniWcant eVect on birth rates. Mortality was considerably lower in better-oV cities. Mortality was also lower in more indus-

56 Other variables of this type were also considered (for other industrial sectors and for the service sectors). This was the one which the analysis showed to have had a consistent and signiWcant eVect on demographic behaviour.

Demography and Migration


Table 5D.4. Regression results for the analysis of birth and death rates in 79 major cities, 1910–12 Dependent variable is:


Birth rate a

Birth rate b

Birth rate c

Death rate d

180.3 þ0.641 (3.67) 6.8 (0.57) þ1.609 (4.14)

232.0 þ0.293 (2.03) 8.6 (0.93) þ0.738 (2.16) þ2.220 (6.88)

241.2 þ0.038 (0.22) þ0.8 (0.08) þ0.932 (2.76) þ2.057 (6.48) 0.916 (2.57)

259.4 þ0.162 (1.88) 25.1 (4.51) 0.680 (3.33) þ0.352 (1.82)

77 33.11 0.584 0.561 25.29 78929

77 31.89 0.620 0.593 23.13 72210

77 19.79 0.361 0.325 10.16 28194

MINING/METALS% SDPVOTE1912 N S R2 adjusted R2 F-test of regression RSS

79 42.78 0.297 0.269 10.56 137246

Note: Dependent variables are annual rates per 10,000 inhabitants.

trial cities, as long as the industry involved was not mining or metal production: these two variables have opposite signs. Combining the birth rates and the death rates produces the overall rate of natural increase, which is the most signiWcant variable for the explanation of migration patterns. This is analysed in the regressions reported in Table 5D.5. These show that the eVect of Catholicism is not secure or consistent. High wage levels have a positive eVect on rates of population increase, and this coeYcient is signiWcant at the 5% level once MINING/METALS% is included in the explanatory variables (columns b to d). The industrial structure variables (INDUSTRY% and MINING/METALS%) have a consistent and signiWcant eVect on population growth. The eVect of heavy industry is somewhat reduced by the tendency of this sector to produce high death rates as well as high birth rates, but the overall eVect is a positive one.


Demography and Migration

Table 5D.5. Regression results for the analysis of the population growth in 79 major cities, 1910–12 a Constant CATHOLIC% WAGES1912 INDUSTRY%



69.2 þ0.432 (3.01) þ16.9 (1.74) þ2.186 (6.84)

27.4 þ0.131 (1.13) þ16.4 (2.21) þ1.418 (5.18) þ1.868 (7.22)

18.6 0.112 (0.83) þ25.5 (3.34) þ1.603 (6.04) þ1.713 (6.87) 0.873 (3.12)

11.5 0.176 (1.05) þ28.9 (3.42) þ1.620 (5.34) þ1.726 (5.67) 0.971 (2.97) 0.3568 (0.9)

79 35.19 0.460 0.439 21.34 92860

77 26.54 0.697 0.680 41.36 50721

77 25.07 0.733 0.714 39.03 44620

69 26.15 0.700 0.671 24.09 42410



Note: Dependent variable is the annual rate of natural increase per 10,000 inhabitants.

The socialist vote continues to be associated with a reduced rate of population increase (column c). The Wnal column includes MOBILITY (the number of total moves in and out of each city relative to the total resident population) amongst the explanatory variables. This has no signiWcant eVect on demographic behaviour. This conWrms the Wnding, reported in the chapter, that it is not possible to discover any relationship between migration and demographic behaviour in urban areas, using the sort of aggregated analysis reported in this chapter. A microlevel study, using individual household data, might produce a diVerent result. Statistical analysis of results for an earlier period, the years 1899– 1901, broadly conWrmed the Wndings of this section. The more limited data availability meant that the analysis had to be conWned to 27 major cities, with a resulting loss of degrees of freedom. But the econometric results also showed a strong eVect on the rate of natural increase from

Demography and Migration


the percentage occupied in industry, and from heavy industry, and also conWrmed that population growth was highest where wages were high. Catholicism was found to be relatively unimportant.

Ap p e n d i x 5E. Ma r r i a g e a n d E c o n o m i c C i r c u m s t a n c e s The argument that late marriage was a distinctive Western European institution, which linked marriage rates to economic conditions, and Table 5E.1. Panel data analysis of marriage rates in Prussian provinces 1816–75


Random eVects model

Fixed eVects model





3.83 þ0.515 (15.48) 1.176 (8.09) 0.0149 (5.90) 0.0012 (2.30) þ0.0120 (6.55) þ0.0266 (3.52) 0.0104 (1.69) þ0.0014 (0.52) 0.0001 (0.02) 0.0072 (1.67)

3.98 þ0.526 (16.09) 1.151 (7.93) 0.0155 (6.19) 0.0011 (2.01) þ0.0123 (6.72) þ0.0301 (4.11) 0.0141 (2.44) þ0.0011 (0.39) 0.0050 (1.37)

 þ0.503 (9.44) 1.209 (7.07) 0.0140 (5.16) 0.0014 (2.76) þ0.0116 (5.91) þ0.0261 (3.14) 0.0098 (1.45) þ0.0014 (0.48) þ0.0027 (0.47) 0.0101 (1.79)

 þ0.52143 (10.50) 1.161 (6.78) 0.0151 (5.93) 0.0011 (2.29) þ0.0121 (6.30) þ0.0306 (3.73) 0.0146 (2.05) þ0.0011 (0.38) 0.0048 (1.42)

711 0.668 0.256

711 0.667 0.246

711 0.687 0.226

711 0.685 0.228

Note: t-statistics in brackets, calculated using White’s heterocedasticity-corrected standard errors.


Demography and Migration

thus formed a ‘positive check’ on population growth, to use Malthus’s terminology, is associated with the work of Hajnal, although it is certainly of much older origin.57 The idea can be found in the work of German demographers in the late nineteenth century, and indeed delayed marriage is Malthus’s moral check on population growth. There has, however, been criticism of the view that non-Western societies inevitably tended to push population growth to the point where wages were reduced to the subsistence level. Some form of population control seems to have operated in Tokugawa Japan and in Imperial China, although the mechanism did not involve delayed marriage.58 If marriage was related to economic circumstances in Western Europe, then it should be possible to Wnd some connection between economic variables and marriage rates. In the period before the transformation to an industrial society these variables will be primarily related to agricultural conditions. Good harvests should produce gains in real wages for agricultural workers, as a result both of lower living costs and of higher demand for rural labour. High urban demand for agricultural produce should also raise rural living standards, and lead to higher marriage rates. To test this hypothesis, a data set was prepared which included annual marriage rates for the original Prussian provinces (1816 boundaries), and average prices for each province from the oYcial price surveys. The data set ran from 1816 to 1875. This made it possible to include wage data from a survey of day-labourer rates in the Prussian state forests for each region, which should be a good guide to rural wage rates. The wage Wgures refer to decades. It would have been desirable to have had annual wage data, but a source of this type is not available for the period. The analysis runs for a period when Prussia was still a predominantly rural society. The data set was analysed as a panel data set; the results are given in Table 5E.1. The Wrst two columns give the results when a random eVects model is used; the second two when the analysis uses a Wxed eVects model (with dummies for each province). The two sets of results are very close, indicating that the random eVects results can be regarded as secure.


Hajnal (1965). Infanticide was one means employed: Lee and Wang (1999), Hanley and Yamamura (1977). 58

Demography and Migration


The eVect from wage levels is weak and not secure in statistical terms. The introduction of a time trend (columns a and c) eliminates the eVect of this variable. Butter prices were also found to have little eVect on marriage rates. The other price variables were associated with signiWcant changes in the marriage rate. There are three price series which have negative coeYcients: rye, potatoes, and pork (although the result for pork is not a robust one). These were the basic items of consumption for the poorer sections of the agricultural population, and the less well-oV urban dwellers as well. High prices for these commodities hit the living standards of the agricultural workers and led to a postponement of marriage plans. The eVect of rye prices is a particularly strong one. There are two price series which are positively linked to marriage rates: those for beef and wheat. These are items which did not form a major part of the diet of the agricultural labourers and their families, but which were mainly consumed by the better-oV in the countryside, or sent to urban markets. High prices for these commodities were an indicator of high urban demand, and a general rise in prosperity. As the agricultural sector exported these products to the cities, while consuming relatively little itself, high prices raised rural incomes. So, the positive link with marriage rates shows that rising rural prosperity was a factor which brought forward marriages. It can also be argued that high prices for these ‘luxury goods’ were indicators of rising urban wages, which may have led to increased marriage rates in urban areas. In general these results support the view that marriage in Prussia in the pre-industrial phase and the early stages of industrialization was inXuenced by economic circumstances, particularly in the rural sector. When conditions were favourable, leading to gains by the rural sector, then the marriage rate rose. More couples were able to aVord the cost of setting up a household on their own. But when times were hard, then this was put oV. This analysis leaves open the question whether regions which had poor economic conditions had permanently low marriage rates. The fact that no satisfactory result is obtained from the wage data argues for the view that there was no permanent eVect; marriages were brought forward or put oV according to circumstances, but the rate was not kept down over the long term in regions where wages were low. The price data might obscure this eVect if poorer regions had relatively high prices for rye and potatoes but relatively low prices for beef and wheat, though there is no evidence that this was in fact the case. Against this,


Demography and Migration

the econometric results for the Kreise given in Section 5D(i) do show a positive eVect from rural wages on marriage rates in 1895, so it may be that the analysis reported in this section is too aggregated to Wnd this eVect. It might be found at the Kreis level but not at the provincial level.


Migration, Farm Size, and the Condition of the Agricultural Labourer 1. I n t r o d u c t i o n One theme which ran very strongly through contemporary writings on migration, particularly migration out of eastern agriculture, was that there were institutional problems east of the Elbe which caused the high rates of migration. These were particularly related to the prevalence of large estates, and the contractual labour system practised on these estates. The property-owning peasant farmers typical of other regions were, it was thought, less likely to leave. The landless eastElbian agricultural workers were drawn to the cities, leaving empty dwellings and uncultivated Welds behind them. Or, even more seriously, they were replaced by Polish workers, which threatened the population balance in the disputed eastern provinces which Prussia had gained as a result of the partitions of Poland in the eighteenth century. Many of these studies expressed the view that there was something wrong with agriculture east of the Elbe, and this feeling has tended to be reproduced in modern historical studies: the east was backward, bound by tradition, hostile to modernity, and so on.1 This chapter and the next one question some of these stereotypes. The theme of this chapter is the institutional structure of east-Elbian agriculture, and the part this played in migration out of the rural east. The next chapter considers the technological transformation of eastern agriculture in the late nineteenth century. 2. A M o d e r n P e r s p e c t i v e o n t h e I n s t i t u t i o n s of Eastern Agriculture Adam Smith and the classical economists of the eighteenth and early nineteenth centuries were hostile to share-cropping and other contractual systems which deviated from the employer–wage labourer 1 A work published in 1898 described large eastern estates as the principal obstacles to the establishment of a ‘just society’ and a ‘modern exchange economy’, Oppenheimer (1898), 9. See Frauendorfer (1957), i. 407–8 for a summary of Oppenheimer’s work.


Migration, Farm Size, Condition of the Labourer

or landlord–tenant norms. The reason for this was the eVect that such institutions had on the incentives faced by the share-cropper or tenant farmer: a share-cropper would only receive a proportion (typically 50% or less) of the return to additional eVort, or to investments in new productive knowledge. By contrast, a tenant farmer who paid a Wxed money rent would receive the entire reward, which should lead to the adoption of more progressive techniques, greater eVort on the part of the farmer, and higher output. The classical economists also favoured the use of cash payments: the certainty provided by wages paid in a unit of known and stable value would be preferable to the uncertainty of payments in kind, or payment in the form of access to land or other beneWts. It would take extreme conditions, where the value of a currency was uncertain or changing rapidly (such as a period of hyper-inXation), to persuade rational and well-informed economic agents to give up the beneWts of money and return to barter and other forms of non-monetary exchange. From this perspective, the traditional institutions of eastern agriculture were not conducive to the rational pursuit of economic gain. An example of payments and beneWts provided to a Deputat household on an eastern estate in the 1800s included: free accommodation and heating; 1 ⁄2 Morgen fruit or vegetable garden; 1 Morgen potato land (fertilized and cultivated); housing and fodder for one cow, plus a calf every other year; the possibility of having hens or geese, or 1–2 sheep; permission to keep a pig; 68 ScheVel of rye and other cereals for each adult male worker and 30 ScheVel for each female worker.2 As the nineteenth century progressed, there was a gradual shift to cash payments, but, even in the years just before the First World War, payments in kind remained an important source of income for eastern agricultural workers. Table 6.1 gives income sources for households in Wve eastern regions (East Prussia, Pomerania, Brandenburg, Silesia, and Mecklenburg-Schwerin) compared to the average for all regions.3 This shows that eastern workers received 60% of total household income in 2

Finck von Finckenstein (1960), 198. Averaging all the eastern regions obscures the fact that there were considerable diVerences between the eastern provinces—cash payments were much more common in Silesia, for example. See Wygodzinski (1917), 26 for a more detailed analysis of payment systems. 3

Migration, Farm Size, Condition of the Labourer


Table 6.1. Sources of income of workers in eastern agriculture, 1910–14 (in Marks), compared with the German average

Value of house plus Wrewood Value of household production using land supplied by landlord Value of harvest shares Other payments in kind Other beneWts Cash payments Total household income Cash payments as % of total

East of the Elbe

All Germany

127 138

103 94

210 29 18 338 860 40.5

124 34 13 501 870 57.6

Source: Averages calculated from budgets given in Mittler (1929), 210–15.

the form of harvest shares and other payments in kind, considerably more than the national average. EVectively, the eastern worker and his family had two roles: the household practised smallholder production on the land supplied by the landlord; but the household was also a small contractor, who cultivated and harvested the estate Welds, and looked after estate livestock in return for a share in production, as well as the land allocated to the household (and some cash payments). The critical view of share-cropping which can be found in the work of the classical economists was widely accepted by economists and other social scientists until the 1950s and 1960s, and formed a part of a generally dismissive attitude to the institutions and values of traditional rural society. But from the 1970s onwards a series of studies have appeared which have taken a more positive stance. Although the main focus has been on the advantages and disadvantages of share-cropping, the analysis has broader implications for the study of agricultural institutions. The authors responsible for this new approach have been working mainly in the Welds of development economics and principalagent theory; but their insights can be applied to the historical study of east-Elbian agriculture. The essence of this approach lies in the concepts of ‘monitoring costs’ and ‘missing markets’, particularly for credit and insurance. The monitoring of a worker’s eVort and diligence is costly for employers, and this is particularly the case where the workplace is a diVuse one


Migration, Farm Size, Condition of the Labourer

(like an eastern estate) rather than concentrated (as in a modern factory). But the obvious answer to this would be to make the workers into independent tenant farmers paying a Wxed rent, who would then have every incentive to monitor their own activities as they would gain the full rewards from additional eVort or increased diligence. What is the reason why this option is not taken? One possibility lies in the economies of scale available to large holdings, which may have access to advanced machinery and management skills which are not available to smaller farmers. In this case the ‘missing market’ is the one which might allow other farmers to use such factors. Another possible reason is that credit markets are closed to smaller farmers, or that credit is prohibitively expensive. A third reason may be that smaller farmers wish to have some insurance against crop failures and other disasters, and, in the absence of formal insurance markets, they may be obliged to enter into contractual arrangements with landlords which provide some element of support in bad years. Finally, the costs of getting goods to market may be so high that only larger estates can run the risk of producing marketable crops. Recent theory predicts that, in circumstances where there are missing markets and high monitoring costs, then crop-sharing contracts will be eYcient solutions to these problems. They combine an incentive element to promote self-monitoring and higher eVort, and also provide solutions to the problems which hinder smaller farmers’ access to product markets, credit, insurance, and modern technology.4 The situation in east-Elbian agriculture in the early part of the nineteenth century was one which combined many features of the various ‘missing markets’ hypotheses: credit was not easy to obtain, crop yields and prices were highly variable but insurance against these risks was not available, new technologies such as sugar beet cultivation were being introduced which favoured larger estates, urban markets were distant and diYcult to predict, and transport was costly.5 4 Articles which set out the modern approach to agricultural contracts include Stiglitz (1974) and Newbery (1977). Bardhan (1989) provides an overview of the various approaches. Two articles which were particularly useful in the preparation of this section were Newbery (1989) and LaVont and Matoussi (1995). Newbery deals with the eVect that marketing costs have on the optimal choice of contract, and argues that a mixture of a rental contract and a wage labour contract might provide an eYcient solution. The institutional arrangements found in east-Elbian agriculture provided such a combination. LaVont and Matoussi analyse the eVect that missing credit markets have on the optimal choice of contract. 5 The analysis of the importance of marketing costs follows Dasgupta (1993), 232: ‘The Wxed rental system is particularly ineYcient . . . where information about market conditions

Migration, Farm Size, Condition of the Labourer


The contractual system of eastern agriculture was well suited to these circumstances. It provided a guaranteed amount of work for the farmworker and his family at harvest time, which protected them against Xuctuations in labour demand. Although payments by harvest share meant that they were exposed to these Xuctuations, the crops grown on the household smallholding provided an alternative source of food. Access to buildings and fodder for livestock reduced the capital cost of keeping a cow and other animals. The main disadvantage from the viewpoint of the farmworker was that it only provided full employment at harvest time (when the system often required the labourer to provide additional Scharwerker). It could be described as a form of institutionalized underemployment: in oV-peak periods the family occupied itself as best as possible on the household smallholding or in other tasks. This was equally one of the principal advantages from the landlord’s point of view: there was no requirement to provide tasks to keep the estate workers busy during the winter. The second advantage for the land-owner was that it economized on cash outgoings. Although eastern estate owners had advantages over small farmers in that they had better access to credit, and that they were also better placed to face the costs and the risks of bringing crops to market, they were also credit-constrained even if at a rather higher level, and marketing costs were still high. There could be years when market prices would not cover the costs of bringing in the crops from remote estates. The normal practice in such years was to feed grain not required for human consumption to livestock. This would not produce any income until the animals were sold. EVectively, by making payments in kind, part of the risk associated with price Xuctuations was transferred to the estate workers, the value of whose ‘wages’ would Xuctuate with price changes. But as most of their income was spent on food anyway (or would have been if they had been paid money wages), this was not such a serious disadvantage. Industrialization promotes institutional change in rural society partly because it provides solutions to many of these ‘missing market’ dilemmas. The spread of banks and other credit institutions eventually makes it possible for small farmers to borrow, and also to save (as a form of self-insurance). Improved employment opportunities can also reduce the need for insurance: if work on construction sites, for example, can be is of importance in generating farm proWts, and where the tenant does not possess this information to any great extent.’


Migration, Farm Size, Condition of the Labourer

found in times when agricultural incomes are low then it would be less necessary to accept disadvantageous labour contracts which provide an element of insurance. As urban centres grow, and transport links are improved, market opportunities improve, integrating markets which were previously separated, and thus reducing price variability. Institutional arrangements such as the contractual relationships of eastern agriculture form a part of a wider system of values and arrangements, behavioural norms and beliefs, which characterize pre-industrial rural societies. Revenue-sharing norms can support high levels of underemployment, causing families to share resources with other members of an extended kinship network, and with other villagers. In traditional societies revenue-sharing and other forms of reciprocal behaviour can be substitutes for missing markets in credit and insurance. Industrialization will tend to alter these attitudes, mainly because wider employment opportunities will reduce the need for insurance mechanisms. The move to increased mobility will damage support mechanisms which rely on a close-knit network of mutual obligations between diVerent kinship groups. Partha Dasgupta has written about this: Costs of mobility involve those associated with transportation, but they involve many other things besides. Markets being thin, the outside world presents considerable uncertainties to the individual. Life in the village may be harsh, but there is an element of security provided by the family and the community. This is sustained by norms of behaviour by which people abide. You can’t simply pack up and leave for better prospects in a neighbouring village. For one thing, you may not be welcome; for another, your departure would be seen as an unsocial act, and this could inXict costs on your kin through a reduction of the support system your village provides for them.6

The implications of this are, Wrst, that the cost of migration includes not only the cost to the individual of giving up the support available in the home village, but also the consequences for those family members that remain of this ‘unsocial’ act. But, secondly, the system of reciprocity itself will be at risk once there is signiWcant out-migration. If families with members who migrate are perceived as having withdrawn from a system of mutual obligation, then the system will cease to be one which involves the whole community.7 While it might survive a few 6

Dasgupta (1993), 234. Dasgupta does not draw this conclusion, but it is a logical consequence of the system he describes. The system can, however, survive considerable seasonal or short-term migration. 7

Migration, Farm Size, Condition of the Labourer


defections, there will be a critical level below which it will no longer function. Once this happens, the cost of migration will fall dramatically, since migrants will no longer have to set the loss of this insurance beneWt against the gains from higher urban wages. In this analysis, as in Max Weber’s, one consequence of industrialization is the erosion of traditional relationships in rural society. Initially, this may be a slow process, but the system could collapse quite quickly once migration is established. These traditional relationships confer a beneWt on those living in rural areas which is additional to the income received from farming (or other activities). The consequence of the removal of this beneWt will be the release of surplus labour onto urban labour markets. Dasgupta also provides an analysis of the eVect of eYciency wages on rural unemployment. The model is a simple one: workers are either permanently employed or employed only at peak periods or permanently unemployed. Those without work are obliged to feed themselves using common-property resources. Labour eVort is related to nutritional status, therefore land-owners pay eYciency wage rates which exceed the market-clearing rate, so there is involuntary unemployment. If there is a reduction in the rural population through out-migration, then the Wrst consequence is an improvement in the average amount of nutrition available from the common Welds and other common resources. This in turn provides an improved nutritional carryover for those who go on to work during the peak period. In consequence, landowners are able to cut peak period eYciency wages, resulting in increased employment and a general rise in production. As Dasgupta puts it: ‘agricultural output would be expected to increase were population to decline in size. This is surplus labour with a vengeance.’8 More generally, the existence of eYciency wage rates either in industry or agriculture will tend to create unemployment if wages are thereby raised above market-clearing levels. In modern economies most of this will be open and recorded, due to the availability of beneWts. In the nineteenth century, when the availability of beneWts was restricted or non-existent, underemployment was hidden; people scraped by on the occasional casual job, joined the informal urban sector, went into domestic service, or returned to family farms. The size of this pool of underemployed labour is not easy to estimate, and much will depend on


Dasgupta (1993), 509 and 518–23.


Migration, Farm Size, Condition of the Labourer

the deWnition of underemployment. Nevertheless, the reabsorption of discouraged or temporarily unemployed urban workers was one way in which agriculture helped to stabilize urban labour markets. In this case, the source of the surplus may lie as much in the operation of the urban labour market as in the state of agriculture.

3. Co n t e m p o r a r y w o r k s o n M i g r a t i o n out of Eastern Agriculture There are a number of distinctive features of Germany’s position which aVected contemporary attitudes to migration.9 These included: (a) Germany’s geographical position: Germany was not an island, but linked to the labour markets of central and eastern Europe. Migration could Xow across Germany’s eastern borders relatively easily. (b) Political considerations: the size of the rural vote was an important factor in the political structure of imperial Germany, and those who sought to maintain it had positions of inXuence or authority within the system. (c) The strength of German socialism: from an early stage Germany had a well-organized social democratic party, which polled over 50% of the vote in three German cities in the Reichstag elections of 1877, and nearly 40% in Berlin in that year. (d) The British example: Germans had the opportunity to study conditions in British industrial cities. Many found this an alarming prospect for Germany: something to be avoided, or improved upon, if possible. (e) Ethnic composition: parts of the German east contained substantial numbers of Poles and Lithuanians, and the departure of Germanspeaking migrants from such areas had implications for the government’s policy of strengthening the German identity of these regions. Attitudes to migration were certainly inXuenced by the social problems associated with urbanization. Adolf Weber, writing in 1908, was able to use oYcial statistics to demonstrate that crime was higher in cities, as were divorce, suicide, adultery, the incidence of venereal disease, and the number of children on welfare, while church attendance was lower. He added a number of other charges: city art was ostenta9

Haberle (1938) surveys German studies before 1938.

Migration, Farm Size, Condition of the Labourer


tious, there was less patriotism, and city life hindered the moral and spiritual progress of the nation. Despite this, Weber did not condemn all aspects of city life: he thought that productivity was higher and educational opportunities were better.10 Most serious commentators recognized that urbanization had its beneWts, and that it was an inevitable consequence of economic growth. So, although the debate over the question ‘Agrar- oder Industriestaat’ attracted attention in the 1890s, in practice no serious eVort was made to hold back urbanization or the growth of industry. The most that it was hoped to achieve was that Germany should remain both ‘Agrarund Industriestaat’, to use the title of a work by one participant in the debate, the historical economist Adolf Wagner.11 It was also considered to be important to preserve the Mittelstand, and peasant proprietors constituted a large part of this class. This led to a concern with the interests of small and medium-sized farmers: as Gustav Schmoller put it, ‘the issue of the preservation of the position of the peasantry involves our entire future as a society’.12 This piece of rhetoric requires some elucidation. Schmoller feared that industrialization would lead to the concentration of property, and the polarization of society between a propertyless working class and a rich capitalist minority, just as Marx had predicted. Membership of the Mittelstand was deWned not just by income level but also by the ownership of at least a small amount of property. Gustav Knapp, speaking after Schmoller in the same debate, praised labour relations in agriculture in western Germany: the Heuerlingen of this region had some animals of their own, long-term contracts with their employers (themselves not aristocratic land-owners but larger peasant farmers), and worked ‘shoulder to shoulder with the employer’.13 The consequence was that social democracy had made no progress in this area.14 Migration transferred agricultural workers into a very diVerent society. In large-scale industry there was little personal contact between employers and employees; employment was insecure and contracts were 10 Weber (1908). Adolf Weber (1876–1963), economist, posts included chairs at the universities of Cologne, Breslau, Frankfurt, and Munich. 11 Wagner (1902); the debate is described in Lebovics (1967) and Barkin (1970). Adolf Wagner (1835–1917), economist, professor at Berlin university, leading member of the historical school, and one of the founders of the Verein fu¨r Sozialpolitik. He represented the Christian Social Party in the Prussian State Assembly 1882–5. 12 Verein fu¨r Sozialpolitik (1893), 2. 13 Ibid. 12. 14 Ibid. 29.


Migration, Farm Size, Condition of the Labourer

short-term; workers had little property of their own. This was fertile ground for socialism. The concern of the historical economists with the polarization of society and the decline of the Mittelstand was intensiWed by a series of studies of Prussian tax returns which appeared to show that inequality was widening. The Wrst of these, by Engel, appeared in 1868. Some of the subsequent studies were produced by members of the Verein fu¨r Sozialpolitik; Schmoller and Wagner were amongst those who worked on these estimates.15 These studies led to a debate about the reasons why inequality was widening. Adolf Wagner thought that the cause was technological progress favourable to large enterprises. But the essence of the historical school’s analysis of this, and other, economic problems was to stress the importance of the historical context: society did not move along predetermined lines. Decisions were made which aVected the outcome. They held the view that the state could bring about the reconciliation of capital and labour, both by direct action and by encouraging favourable institutional developments.16 Some made the same connection between rural conditions, migration, and urban labour markets which lies behind the Lewis Model. Max Sering, in a report presented to the Verein fu¨r Sozialpolitik in 1893, blamed excessive migration for depopulating rural areas in the east, as well as holding down the wages of industrial workers: while the cities and industrial districts are accumulating an ever larger reserve army, which rarely Wnds full employment, even in times of maximum economic activity, and which is forcing down the living standards of other workers, the wide Welds of the eastern arable districts are short of labour . . . 17

Interesting points in this passage include the use of the term ‘reserve army’, familiar from Marxist terminology. 15 Engel (1868 and 1875); Schmoller (1895) and Wagner (1904); there had been a tradition of concern about the distribution of income in German political economy, going back at least as far as von Thu¨nen (1826), whose views are given in Pogge von Strandmann (1999). Ernst Engel (1821–96), statistician, director of the Prussian Statistical OYce 1860–82, represented the National Liberal Party in the Prussian State Assembly, 1867–70, Neue Deutsche Biographie, iv. 500–1. 16 Wagner (1871) and Schmoller (1900); also HelVerich (1915); the views of the historical economists on income distribution are discussed in Dumke (1991) and Grimmer (1999); other writers considered ways to reconcile the interests of capital and labour, amongst them Walter Rathenau, Pogge von Strandmann (1988a). Karl HelVerich was a director of the Deutsche Bank and secretary of the Treasury during the First World War. 17 Sering (1893), 7.

Migration, Farm Size, Condition of the Labourer


For a group of writers on agricultural matters, the extent of migration out of agriculture was far greater that could be justiWed by the diVerence between urban and rural wages. This group, which included Knapp, Max Weber, and Sering, blamed developments within agriculture.18 Sering noted that in 1885–90 the eastern provinces had lost 640,000 people through migration, 75% of the natural increase in the population. This was an indication of a profound social malaise: ‘it suggests that all social classes had a feeling of profound dissatisfaction, that there was a sickness of the social organism itself’.19 The view that migration was a symptom of some major problem within rural society was an inXuential one, and led to schemes of ‘internal colonization’: the settlement of landless rural workers, or disaVected urban workers, in colonies of land-owning peasant farmers.20 Migration from rural areas into German cities was a matter of concern to a range of opinion in Germany. A number of authors took the view that it was caused by a deterioration in agricultural conditions, in particular by changes in the position of landless agricultural workers. One strand of this analysis concerned the eVect of the agrarian reforms of the early nineteenth century. In 1873 Theodor von der Goltz pointed out that migration was highest from areas of Großgrundbesitz—the large east-Elbian estates—and unknown in areas dominated by smaller peasant holdings. In 1887 Gustav Knapp argued that the reforms had led to a loss of traditional rights and increased the number of landless labourers:21 In Knapp’s view it was the poverty and lack of freedom of the eastern agricultural worker which was responsible for out-migration. In the western provinces, where there is often serious poverty amongst small farmers as a result of the splitting of holdings, there are certainly poor agricultural workers, but nobody who feels to the same extent the lack of freedom in every aspect of their lives. . . . The poorest forestry worker in the Black Forest, the most down-trodden hay-maker in Kanton Uri, have a higher station in society than the Insten in our wide eastern provinces.

18 Knapp (1887), Weber (1893 and 1894): also von der Goltz (1875). G. F. Knapp (1842–1926), economist and statistician, professor of statistics at Leipzig and Strasburg, NDB xii. 152. T. A. von der Goltz (1836–1905), farmer and agricultural historian, held chairs of agriculture at Ko¨nigsberg and Jena, NDB vi. 635–6. See Frauendorfer (1957), i. 392–4, 394–6 and 436–7 for the background to this debate. 19 Sering (1893), 7. 20 These schemes also served nationalist aims, Grimmer (1999), 293–4. 21 von der Goltz (1875), Knapp (1887 and 1891).


Migration, Farm Size, Condition of the Labourer

Knapp blamed the operation of the Stein-Hardenburg laws for the original loss of land and status by the agricultural workers east of the Elbe. But he also thought that their position was deteriorating as the land-owners intensiWed their demands for labour to work the estate Welds, thus reducing the time the labourers had for their own smallholdings: The aristocratic landlords want to have all the labourers’ time for their own enterprises . . . Where this development has been completed . . . the worker on an estate will be turned into a state of utter dependency . . . he is no longer a true Inst: he has no house, he has no arable acres of his own, he no longer has a cow.

These early studies had an eVect on opinion. The view that large estates were predominantly to blame for migration out of agriculture became widely accepted. A study published in 1930 commented: ‘It is a commonly held view that the large landholdings considerably increase migration out of agriculture.’22 The debate produced three major surveys: one of the conditions of the peasantry in 1883, another of the conditions of agricultural labourers in 1890–1, and a survey of Prussian inheritance systems in 1899– 1910. These are valuable surveys of conditions in German agriculture in the period.23 There are, however, some diYculties in using the information they provide. They are, in eVect, collections of articles, which reXect the concerns and interests of individual authors. Data is not generally provided on a comparable basis.24 The 1890–1 survey was in three parts. The third of these, covering Germany east of the Elbe, was written by Max Weber. It was his second major publication. His contribution is useful for a number of reasons. It was more systematic than the others. It includes useful data on wage rates and payments in kind, and attempts to link these Wgures to those produced by earlier surveys. And Weber attempted to provide an explanation of the process of social change revealed by the survey.

22 Golding (1930), 214. See Oppenheimer (1898) for a particularly forceful presentation of this view. 23 These were published in the Schriften des Verein fu¨r Sozialpolitik (1883), vols. 22–4 and (1892–3), vols. 53–5, and in Sering (1899–1910). 24 A comment which reXects the frustrations of the modern statistician, faced with what could have been immensely valuable sources of data. There are some very good articles in these surveys: one example is the article by Paasche on Mecklenburg-Schwerin, Paasche (1883).

Migration, Farm Size, Condition of the Labourer


4. M a x w e b e r a n d t h e C o n d i t i o n o f t h e Ag r i c u l t u r a l L a b o u r e r i n t h e E a s t Of all Weber’s major works this is probably the least known. It remains untranslated. Yet it is a thorough survey of conditions in eastern agriculture, a valuable source of material, and one of the earliest works to consider the question of a ‘labour surplus’ in traditional agriculture: how is it created? under what conditions is it released? The questionnaire results were used as a basis. Weber reported some of the more reactionary views of the respondents without comment, as for example the opinion that it was not desirable to educate the rural population.25 But he also used the survey to express his own views on the social relationships found in eastern agriculture. He developed this analysis in an article in the 1894 Archiv fu¨r Soziale Gesetzgebung und Statistik, which also responded to some of the criticism made of the original study at the 1892 Colloquium of the Verein fu¨r Sozialpolitik. In 1908 he returned to the theme in a lecture in the United States.26 Weber followed Knapp in that he thought that the larger estates were primarily responsible for out-migration. He saw the creation of the labour surplus as a result of an alteration in social relationships on these estates. Paternalist attitudes were replaced by a commercial, capitalist approach. This in turn led to the introduction of more intensive arable production, associated with sugar beet cultivation, and increased use of mechanization. Contractual German workers were replaced by seasonal Polish workers. Weber’s attitude was inXuenced by political concerns, a desire to maintain and, if possible, increase the German population in the disputed lands of the Prussian east. He was a member of the nationalist organization, the Ostmarkverein (also known as Hakatisten from the initials of the organization’s founders), which argued for the compulsory purchase of Polish estates and the settlement of German farmers on lands previously occupied by Poles.27 Weber supported the organization’s aims. A newspaper report of a speech he gave in March 1893 went: ‘above all else, Herr Weber argued for the pursuit of a policy 25

Weber (1893), 197–8. Weber (1894 and 1948). See Honigsheim (1949) for an assessment of Weber’s work in this area. 27 See Hagen (1980) and also Galos et al. (1966) for accounts of the conXict between Germans and Poles in the east. Schmoller and Adolf Weber were also members of the Ostmarkverein, Hagen (1980), 269. 26


Migration, Farm Size, Condition of the Labourer

which served German interests, he was against the equal treatment of Poles and Germans’.28 His criticism of the Junker estates should be seen in this context. In the same month, in another speech, he said: ‘the presence of large estates is the factor which is, currently, most responsible for the increase in the Polish population in the East’.29 However, in the 1890s the concern over foreign agricultural workers was premature. In 1907 there were 195,000 foreign-born agricultural workers in Prussia, constituting 1.8% of the agricultural labour force: 1.0% in East Prussia, 1.6% in West Prussia, 1.3% in Posen, and 2.7% in Pomerania. The largest concentration was in Prussian Saxony, where they amounted to 4.8% of the population occupied in agriculture.30 This suggests that, up to this point, the main cause of migration from outside Germany had been the high seasonal demand for labour in the sugar beet Welds of central Germany. It has been estimated that the number of foreign workers rose to over a million by 1913 (including 356,000 employed in industry). This is diYcult to check given the lack of an occupational census after 1907. The Preubische Feldarbeiterzentrale issued a total of 412,000 permits in 1907–14, but, as Russian-born Poles, in particular, were obliged to return home every year, it is possible that these were only used for a few years on average.31 In general, permanently settled foreigners do not seem to have replaced Germans to any considerable extent before 1907. There was some displacement by temporary seasonal migrants, although this could be interpreted as the replacement of German labourers who left voluntarily to seek better wages in the cities. This question is related to the claim that the position of the agricultural labourers in the east was deteriorating, socially or economically. In this respect there is a diVerence between Knapp on the one hand, who saw in the persistence of payments in kind and contractual labour services evidence of the dependent status of the eastern workers, and explained migration as an attempt to escape this, and Weber on the other, who took a rather more favourable view of the traditional labour relationships. Weber’s view was that the Inste and Deputate of eastern agriculture had had the status of Kleinwirte or small contractors, bound individually to the landlords in a system which was certainly one of subordination, but which also emphasized common interests (through harvest shares 28 31

29 Weber (1993), 825. Ibid. 179 Figures are from Wehler (1995), 546.


Ibid. 489.

Migration, Farm Size, Condition of the Labourer


and access to communal grazing). The keeping of livestock by the labourers was a crucial sign not only of material well-being, but also of status and of ‘cultural-level’. The point here was that, in Weber’s view, a reduced level of cow-keeping (such as Weber found in Silesia) was generally associated with a higher level of female participation, in work both oV and on the holding, and this was a backward step culturally: ‘the ability to keep, preferably, a milking cow, or at least some cattle, is the most important precondition for a well-adjusted family life’.32 Weber held that this system of patriarchal labour relations had come under threat from two directions. The Wrst was the shift to a ‘capitalist’ agriculture: more intensive use of the land and greater livestock production. This had led to the reduction of the smallholdings given to workers, and the division of the common grazing Welds: ‘the loss of access to grazing on the village commons at the time of enclosure was the most grievous blow that could have been struck at the property owning labourer’.33 He described the reduction of the Masurian labourers to the status of Milchdeputate (looking after the landlord’s cows instead of their own) as ‘something which embittered social relations and which destroyed the sense of having interests in common’.34 The second problem was the unwillingness of the children of agricultural labourers to remain part of the system. The Instleute households were required to provide two or three workers at harvest time, so the participation of children was vital, especially as it became more diYcult to Wnd substitutes (‘cripples, old men, women of loose morals, have been oVered as such’).35 Weber did not believe that there was a connection between the migration of the children of agricultural labourers to the cities and the level of wages. Indeed, he thought that migrants experienced a decline in wages. In his view it was a somewhat perverse desire for an illusion of freedom which led the children of eastern agricultural labourers to prefer proletarianization and lower wages in cities to the paternal system and higher ‘cultural level’ on oVer in eastern agriculture: ‘according to the survey returns, the motive was a desire for independence even if it could only be achieved at the cost of a reduction in living standards’.36 It should be borne in mind that Weber was making these remarks while reporting on the results of a survey of the condition of agricultural labourers, conducted by means of a questionnaire sent out to 32 35

33 Weber (1893), 184. Ibid. 36 Ibid. 185. Ibid. 193.




Migration, Farm Size, Condition of the Labourer

landlords. Weber had no other evidence to support the statement that migrants out of agriculture did not improve their economic position apart from the opinions expressed by these respondents. The eVect of out-migration was damaging to the institutional structure of eastern agriculture as a result of its impact on the Scharwerker system. If the Inste were unable to provide additional workers, then they were not fulWlling their side of the bargain which led to the landowner oVering a smallholding in return: The farmer’s own children move away, at the latest, after completing military service, though many leave earlier, so that they can devote themselves to life in the cities as soon as possible, without considering whether this will bring about the collapse of the family business. As a result, the Inste are no longer able to fulWl a contractual obligation to provide three workers to the estate, as used to be commonly required in East Prussia, many are unable to even provide two, if their wives don’t work with them.37

The allegation that, as a result of the shift to capitalist values, labourers had less opportunity to keep animals themselves is not supported by the evidence, which shows continued cow-keeping on smallholdings, although it was probably true in certain districts.38 However, the fact that cow-keeping was lower in industrial districts like Silesia is not necessarily an indicator that the position of the labourers was worse in these areas. It could just as well show that better opportunities to supplement agricultural earnings through industrial work made the keeping of cows less attractive. Whether or not the increase in the keeping of livestock on smallholdings was suYcient to compensate for the undoubted reduction in grazing rights on uncultivated lands, as these were enclosed and brought into arable production, is a more open question. The argument that changed values had brought about a deterioration in the living standards of the agricultural workers is also not supported by the evidence. On the contrary, there is good evidence (some provided by Weber himself) to show that agricultural wages were rising. Weber supplied a mass of evidence on the conditions of employment of 37 Weber (1893), 185. Another factor which might have aVected the Scharwerker system was the enforcement of laws against child labour during school hours. Weber did not mention this as a factor, but, according to other authors involved in the survey, this had a signiWcant eVect on the use of child labour in some regions, Schriften des Vereins fu¨r Sozialpolitik, 52, p. 332. 38 Between 1881 and 1895, the value of livestock kept by holdings of less than 2 hectares rose by 16.2% and it rose by 17.4% on holdings of between 2 and 5 hectares; the rise on holdings of over 100 hectares was only 9.5%, Rauchberg (1900), 590.

Migration, Farm Size, Condition of the Labourer


diVerent classes of agricultural labourers, and comparisons with the position in 1849 and 1873 (the latter date uses von der Goltz’s Wgures). Where a high proportion of the payments are in kind or through harvest shares or the allocation of smallholdings the comparisons are hard to make. There seems to have been an increase in cash payments to groups traditionally paid in kind. But the value of these other beneWts may have altered, and there is some evidence that the labour contributions demanded by landlords had risen. However, comparisons are possible for those workers paid in cash ohne Kost (without provision of board and lodging). There are a number of localities where Wgures are available for 1849 and 1892 and these have been brought together to form Table 6.2. What this table shows is a very considerable improvement in the wages paid to agricultural workers in the east over the period. The last line gives indices which show wages rising by a factor of 2¼–21⁄2 1849–92. This is important for a number of reasons. First, the suggestion that eastern agricultural workers experienced declining wage rates is encountered in some publications.39 This does not appear to be the case. Secondly, it puts the problems of eastern grain producers in a new perspective. Between 1870/4 and 1890/4 arable producer prices fell by 10.3%, while this table shows wages rising by 34–35%. Producers were being squeezed from both directions.

5. Mi g r a t i o n , F a r m Si z e a n d A g r i c u l t u r a l T e c h n i qu e s The regressions in the previous chapter showed that the apparent relationship between farm size and migration was reversed once allowance was made for demographic factors. Large farms contributed to outmigration indirectly, through the positive relationship between farm size and the natural rate of population increase. There was no tendency for the quit rate to be higher on larger holdings once the eVect of high population increase was removed. Indeed, the quit rate was shown to be lower on larger holdings. The data set used in this analysis made use of a 1933 study of net migration in the Prussian Kreise from 1882 to 1895. In this, an indirect estimate of the quit rate for the agricultural population was produced


Dickler (1975), for example.


Migration, Farm Size, Condition of the Labourer

Table 6.2. Daily earnings in Marks of ‘free’ agricultural day-labourers (ohne Kost) in full-time employment where possible (earnings of those who are only temporarily employed are in bold type) Summer wages 1849 East Prussia: Gumbinnen Ragnit 0.5/0.6 Gumbinnen 0.5/0.6 Insterburg 0.6/0.7 Sensburg 0.5/0.6 East Prussia: Ko¨nigsberg Mohrungen 0.5/0.6 Ro¨ssel; Allenstein; 0.5/0.6 Ortelsburg West Prussia: Marienwerder Stuhm 0.8/1.0 Rosenberg 0.8 West Prussia: Danzig Stargard 1.0/1.5 Putzig 0.75 Pommern: Ko¨slin Schlawe; 0.8 Rummelsberg; Stolp Pommern: Stettin Anklam 1.0 Greisenberg 0.75 Pommern: Stralsund Greiswald 1.0/1.2 Silesia: Oppeln Neustadt 0.5 Silesia: Liegnitz Liegnitz 0.5 Brandenburg: Frankfurt Lebus 0.6 Brandenburg: Potsdam Barnim 1.0 Unweighted 0.74 averagesa as an index 100 (1849¼100) a

Winter wages






0.85 0.85 1.55 1.1

1.0/1.8 1.2/1.4 1.5/2.0 1.0/1.6

3/0.5 0.3/0.4 0.4 0.4

0.62 0.62 0.66 0.57

0.8/1.2 0.8/1.0 1.0/1.2 0.8/1.2

1.23 1.4/2.0 1.25/1.3 1.0/1.5

0.4/0.5 0.73 0.9/1.2 – 0.7/0.81 0.7/1.0

2.03 –

1.25/2.5 – 1.25/1.75 0.4

1.41 –

– 1.0/1.2

1.62 –

2.75/4.0 1.0/1.75

– 0.5

0.92 –

1.0/1.5 0.75/1.0




1.68 1.53

1.5/2.0 1.5/2.0

0.62 0.6

1.0 1.31

1.25/1.75 1.2


















1.56 1.37

2.1 1.83

0.75 0.52

1.06 0.87

1.2 1.17






Gaps in the data have been Wlled by extrapolation (such as using the average relationship between summer and winter wages for the year in question); ranges were entered as the mid-point values.

Migration, Farm Size, Condition of the Labourer


using a combination of results from the occupational censuses of these years, and records of births and deaths by Kreis. An alternative approach is to examine the movement of the agricultural population directly using the results of the occupational censuses. This eVectively excludes the inXuence of demographic factors by looking only at the trend of agricultural employment. Implicitly, it assumes that any demographic surplus will be exported, either to nonagricultural sectors within the region, or to other regions. The analysis presented in the previous chapter showed this to be a relatively wellfounded assumption. The data set prepared for this analysis covered a rather longer period, 1882 to 1907, and included 797 Kreis-level units, covering the whole of Germany not just Prussia. The results reinforced the analysis in the previous chapter. Although there is a simple negative relationship between the movement of the agricultural population and the presence of large holdings, this is a spurious result. More sophisticated analysis showed that the reason why the permanent agricultural population declined in areas where there were larger farms and estates (or rose less rapidly) is that these estates were situated in areas where the agricultural population was declining for diVerent reasons: these were remote areas, with poor access to urban markets, and they were situated in regions close to the border with Russian Poland, which meant that seasonal Polish workers could be used.40 The results of the statistical analysis are summarized in Table 6.3. The most important inXuences on the movement of the agricultural labour force were the level of non-agricultural employment within the Kreis in 1882—there tended to be a decline in the more agricultural regions—and geographical location. In purely agricultural areas the predicted level of agricultural employment was static, meaning that any demographic surplus would be exported, while in the areas which had high levels of non-agricultural employment, agricultural employment rose by 35–40% (37% for a region which was 80% non-agricultural in 1882). In these regions, quite substantial rates of population increase could be accommodated within the agricultural sector. Remoter regions tended to maintain a higher level of agricultural employment. The predicted level of employment in regions 100 kilometres from major cities (over 100,000 inhabitants) is 7% higher in 1907 compared to a region with a similar starting point in 1882, but 40 The full details of this analysis were originally given in chapter 4 of Grant (2000) and have been published separately in Grant (2002c).


Migration, Farm Size, Condition of the Labourer

Table 6.3. Summary of results of statistical analysis of the change in the agricultural population 1882–1907 in 793 German Kreise Secure statistical results Factors which are associated with a less rapid rise in the population occupied in agriculture or with a decrease in the agricultural population: A high rate of agricultural employment in 1882 A situation close to the border with Russian Poland Factors which are associated with a less rapid decrease in the population occupied in agriculture or with an increase in the agricultural population: The presence of large farms and estates Being a long distance from large cities (over 100,000 inhabitants) Having a high level of sugar beet production in 1907 Having high use of mechanized threshing in 1907

which was 20 kilometres from a major city. This suggests that regions closer to major cities were quicker to shed surplus labour. There was also a major east–west division: the agricultural population declined fastest, or rose less rapidly, in the rural east. Table 6.4 illustrates this for two hypothetical regions: Region A with 60% of land in holdings of over 100 hectares, Region B with only 10%. Both have 65% of the population initially occupied in agriculture, and 75 kilometres between their mid-point and the nearest large city. This table shows the importance of location. In areas close to the border there is little predicted increase in the agricultural population. The average German rate of natural population increase of 1.3% (1871–1913) implies an increase of 38% over 25 years. The rates in Table 6.4. Predicted change in the agricultural population over 25 years for two hypothetical regions Distance from the border with Russian Poland in km 50




Predicted change in the agricultural population over 25 years A 60% in þ3.5 large farms B 10% in þ3.3 large farms







Migration, Farm Size, Condition of the Labourer


some eastern provinces were substantially higher (1.6% per annum in West Prussia, 1.7% in Posen). In consequence, the anticipated rate of out-migration would be high from all the agricultural regions close to the border. But at a distance of 500 kilometres (approximately the German average) the position changes. Here the larger estates increased employment opportunities more than small or medium-sized farms did, and so Region A is predicted to increase the agricultural population by more than Region B. The implied rate of out-migration from Region A is substantially lower than from Region B. The same analysis can also be used to examine the eVect of more advanced agricultural techniques, such as the introduction of sugar beet cultivation and also the use of labour-saving machinery such as steamthreshing. The results given in Table 6.3 show that the intensiWcation of arable production associated with the cultivation of sugar beet led to an increase in employment opportunities: when controlling for other factors the eVect of sugar beet cultivation is a positive one. The more intensive agricultural crops such as sugar beet made increased demands on the farm labour force, for weeding and harvesting, and also were associated with increased livestock numbers (who could be fed on the pulp left after the sugar had been extracted), which also raised labour requirements. Similarly, the use of steam-threshing was not associated with a decline in employment opportunities, when this factor was considered separately from other factors inXuencing agricultural employment. Rather, the use of steam-threshing is associated with rising agricultural employment. The larger estates did often use steam-threshing, but, as already noted, these holdings were tending to increase employment for other reasons. If steam-threshing did replace labour previously used to thresh by hand (and this was the main advantage of the technology) then it appears that these workers were redirected to other tasks. The positive connection between the use of something which was quite clearly a labour-saving technology, and the change in agricultural employment, suggests that this technology tended to be used mainly in areas which were already short of labour.

6. I n f l u e n c e s o n R u r a l W a g e Le v el s It was a commonly held view that large holdings caused out-migration partly because they paid low wages. A study published in 1907


Migration, Farm Size, Condition of the Labourer

commented: ‘the accusation has often been made against large farms and estates that by their nature, as well as by paying poor wages, they have created migration out of agriculture, or at least caused it to increase’.41 If so, it should be possible to Wnd this eVect by examining variations in rural wage levels. These can be studied making use of the surveys conducted to provide information for the Reichsversicherungsamt. The accident insurance legislation required that the Ortsu¨blicher Tagelohn should be recorded for each locality. The Prussian surveys are particularly useful in that wage rates in rural areas are distinguished from those in towns. The data was used to construct estimates of average rural wages for each Kreis, combining the 1892 and 1901 surveys. The full statistical results are given in Appendix 6B, Table 6B.1; Table 6.5 provides a summary of the main Wndings. The main inXuences on wage levels came from geographical location and from the costs of living in the locality: wages were higher in areas where food prices were high. The impact of the border with Russian Poland was also an important factor. Wage rates were substantially lower in Russian Poland: male day-labourer rates were 0.52 Marks in Russian Poland in 1890, compared to 1.13 Marks in a German border region such as Gumbinnen (in 1892), and 1.66 Marks in a more prosTable 6.5. Summary of results of statistical analysis of rural wages levels in 392 Prussian Kreise 1892/1901 I. Secure statistical results Factors which are associated with a higher level of rural wages: High food prices A situation further from the border with Russian Poland Factors which are associated with lower wage levels: A high rate of agricultural employment in 1882 A high level of household employment in 1882 Having a high level of sugar beet production in 1897 II. Less secure statistical results Factors which are associated with a higher level of rural wages: The presence of large farms and estates The proximity of large towns and cities (over 50,000 inhabitants) 41

Broesike (1907), 22.

Migration, Farm Size, Condition of the Labourer


perous central region such as Magdeburg.42 The equivalent Wgures for female day-labourers were 0.37, 0.71, and 1.01 Marks respectively. In the areas closer to the border, the availability of cheap migrant labour had a depressing eVect on German wage rates. Table 6.6 gives the predicted levels of wages for rural areas at diVerent distances from the border with Russian Poland.43 The strong eVect of the border on rural wage levels is shown. The importance of nonagricultural employment is also demonstrated. Wages were low in the east mainly because of the proximity of the border and the relative shortage of non-agricultural employment. Once allowance is made for these factors, it can be shown that the presence of large farms and estates did not lead to lower wages. On the contrary there is a slight tendency for wages to be higher in areas where there were signiWcant numbers of holdings of 100 hectares or more. There is no justiWcation for the view that larger estates held down wages.44 Wages were low in the rural east for other reasons. Other factors which were associated with low wages include a high level of household employment in 1882. Cottage industry tended to be found in areas of high population pressure, and could therefore be an Table 6.6. Predicted average daily wages in Pfennigs, in rural areas, 1892–1901 Distance from the border with Russian Poland (km) 50 Predicted level for a region with 20% 121.6 of the population occupied in agriculture Predicted level for a region with 80% 98.2 of the population occupied in agriculture










42 Figures for Russian Poland are from Knoke (1911), 32, who obtained them from the Warsaw Statistical OYce and converted these into Marks; German data from Neuhaus (1904), 320–28. 43 The calculations use the coeYcients given in column a of Appendix Table 6B.1 (other variables take average values). 44 The statistical result showing that large holdings tended to push up wages is not a strong one. However, the alternative hypothesis, that large estates paid signiWcantly lower wages, can be rejected with a high degree of conWdence.


Migration, Farm Size, Condition of the Labourer

indicator of surplus labour conditions within the Kreis.45 The Wnding that wage levels within agriculture were lower in regions with high levels of household employment supports this view. The statistical analysis also found an association between low wages and sugar beet cultivation: wages were lower in Kreise with high areas under sugar beet relative to other arable crops in 1897. This is a surprising result when considered alongside the Wnding reported in the previous section, that sugar beet cultivation led to an increase in agricultural employment. It appears to support Weber’s contention that there was a change in the behaviour of land-owners in marketorientated agriculture, such as the areas practising intensive arable farming associated with sugar beet cultivation. Sugar beet cultivation raised the demand for labour but led to a fall in the price of labour, in apparent contradiction of the basic laws of supply and demand. The reason seems to lie in the relationship revealed by the importance of the proximity of the border with Russian Poland: the availability of migrant labour from Russian Poland was a major factor in wage determination in German agriculture. Much of the eVect of migrant labour is captured by the distance variable, as the most important factor was the proximity of the frontier. However, in the sugar beet areas the use of foreign migrant labour was established early, and this led to the development of networks of agents who supplied labour on a regular basis. It was therefore easier for farmers in these regions to obtain migrants for sugar beet cultivation and other tasks. This had a depressing eVect on wages.

7. D ir ect E v i d e n c e o n U n d e r e m p l o y m e n t i n R u r a l A r e a s This section looks for evidence that underemployment in German agriculture was a factor in the relative decline of the rural population. Was there a shift away from areas where there was underemployment? If so, then this would support the concept of a pre-industrial rural labour surplus, which is then released during industrialization. The analysis reported in Section 5 made use of a data set covering all the German rural Kreise. This section restricts the analysis to the Prus45

See for example the description of conditions in the Taunus uplands, in Ba¨uerlicher Zustand in Deutschland (1883), i. 156–7, where the main cottage industry had been nailmaking.

Migration, Farm Size, Condition of the Labourer


sian Kreise, which makes it possible to use other data sources, in particular the more detailed publication of the occupational census results found in Prussian statistical sources. Table 6.7 provides a summary of the Wndings; the full econometric results are given in Appendix 6B, Table 6B.2. The main results repeat those shown earlier when the full data set was used: the agricultural population declined faster or rose least in areas close to the border with Russian Poland, and in areas where there was a high level of agricultural employment in 1882. There was a tendency for the decline to be held back in remote areas (measured by the distance from major cities). A number of authors thought that the excessive division of land in areas of partible inheritance had led to surplus population and underemployment.46 If so, then this might have produced a negative relationship between partible inheritance and the movement of the labour force, as migration reduced the labour surplus in these regions. However, the analysis shows that the relationship is a positive one not a negative one: partible inheritance is associated with a slightly larger

Table 6.7. Summary of results of statistical analysis of the change in the agricultural population 1882–1907 in 397 Prussian Kreise I. Secure statistical results Factors which are associated with a less rapid rise in the population occupied in agriculture or with a decrease in the agricultural population: A high rate of agricultural employment in 1882 A situation close to the border with Russian Poland A high level of household employment in 1882 A high level of self-employment in 1882 Factors which are associated with a less rapid decrease in the population occupied in agriculture or with an increase in the agricultural population: Being a long distance from large cities (over 100,000 inhabitants) II. Less secure statistical results Factors which are associated with a less rapid decrease in the population occupied in agriculture or with an increase in the agricultural population: A high level of partible inheritance


Sering (1899–1910), Conrad (1891).


Migration, Farm Size, Condition of the Labourer

increase in the agricultural labour force in the Kreise where this was practised, although the result is not a secure one in statistical terms. The previous chapter has provided a possible reason for this. In areas of partible inheritance demographic behaviour had adjusted so as to avoid excessive subdivision. There was, therefore, no reason for there to be any more underemployment, or surplus labour, in these regions than there was in other rural areas.47 The analysis also considered some factors which might be indicative of a high level of rural underemployment. These are the percentage of the non-agricultural labour force which is self-employed and the percentage occupied in cottage industry. The justiWcation for including these variables is that underemployment within agriculture in a particular Kreis may be related to underemployment outside. If there is surplus agricultural labour, some of this will take non-agricultural work, even if there is not full employment in this sector either. Expected returns in both sectors will tend to equalize. And total annual returns will be a multiple of expected daily earnings and the anticipated number of fully employed days. So, unless there is a compensating diVerence in expected earnings, underemployment in one sector will spill over into the other.48 The results show that high levels of household employment and selfemployment are associated with slower increases in agricultural employment. In Kreise with this ‘labour surplus’ pattern of employment the rate of natural population increase was considerably greater than the predicted increase in agricultural employment, leading to out-migration. Table 6.8 illustrates this point, by providing predicted rates of increase in agricultural employment for diVerent Kreise, ranked by the estimated labour surplus (which is a weighted combination of the percentage in self-employment and the percentage in household employment). The estimates show that, in a Kreis in the top decile for ‘estimated labour surplus’, with high levels of household employment and selfemployment, the predicted increase is well below the average rate of natural increase (38% over the whole period). The implied rate of outmigration from rural areas for such a Kreis is 25% of the population in 1882. By contrast, a Kreis in the lowest decile would have had a 47 A number of regressions, using diVerent data sets and diVerent periods, were run looking for a relationship between migration and partible inheritance. All these produced results which were insigniWcant. 48 This implies that there is some Xoor to wages in both sectors, so eYciency wages are paid.

Migration, Farm Size, Condition of the Labourer


Table 6.8. Analysis of the change in the agricultural population, 1882–1907: predicted rates of change for 397 Prussian Kreise ranked by estimated surplus labour Ranking of Kreis by estimated labour surplus

Household employment 1882 (%)

Self-employment 1882 (%)

Predicted change in agricultural population 1882–1907 (%)

Top decile average Top quartile average Average all Kreise Lower quartile average Lower decile average

18.74 9.10 3.68 1.60 0.87

39.36 38.05 28.89 19.51 15.59

þ12.95 þ16.84 þ23.34 þ28.85 þ31.09

Note: The ranking is obtained by using the coeVicients given in column c of Table 6B.2 for %HOUSEIND and %SELFEMPLOYED, and applying them to the Kreise data to produce an estimate of the labour surplus. Note also that the increase in the agricultural labour force shown between 1882 and 1895 was at least in part due to the underrecording of female labour in the earlier census.

predicted increase in agricultural employment nearly equivalent to the rate of population increase, leaving a migration Xow equivalent to just 7% of the population in 1882. This measure provides strong evidence for the view that labour surplus conditions in rural areas contributed heavily to migration in the period.

8. T he Le w i s M od el and Ea st - E l b i a n Ag r i c u l t u r e The focus in this chapter has been on the eVect that large farms and estates had on migration out of rural areas. A central theme has been to examine the view of Max Weber that there was a change in the values and attitudes of land-owners which led to deterioration in the condition of the agricultural worker and the release of surplus labour. The evidence does not support the view that there was a general fall in real wages after 1870, nor was there any overall tendency for larger farms to shed labour. On the other hand, it does appear that the shift to more intensive arable production systems, characterized by the introduction of sugar beet, did lead to a change in labour relations, involving the increased use of migrant labour, and this had a depressing eVect on


Migration, Farm Size, Condition of the Labourer

wages in the areas most aVected. This eVect was strongest in the east, due to the proximity of the border with Russian Poland. One constant theme running through the results presented in this chapter is the inXuence of Russian Poland, both on wage rates and on the numbers occupied in agriculture. This inXuence seems to have increased. In Weber’s view this was due to a change in the behaviour of land-owners. There are, however, other factors which might account for this. In the Wrst place, Russia was starting to industrialize. From the 1880s onwards growth was rapid.49 This would have set oV a similar process of increased mobility and migration to the one this study has described for Germany. This could have made it easier for German farmers, and their agents, to obtain migrant Polish workers. Another consequence of this rise in mobility was a dramatic increase in the emigration of Russian Poles to the United States. Secondly, there may have been a change in the attitude of the workers themselves. The bargain represented by the traditional contractual relationship of eastern agriculture was one-sided. Although Weber interpreted the contractual relationships of eastern agriculture as reXecting a pre-capitalist desire on the part of the land-owners to employ as many as possible, the evidence he provides is not consistent with this. The essence of the Inste system was that it did not oVer yearlong full-time employment (as some landlord who sought only to maximize employment might have done). Rather it ensured that the peak seasonal labour requirements of the land-owner were met, even if this meant considerable underemployment for the rest of the year. The provision of smallholdings and grazing rights meant that this ‘surplus labour’ could be absorbed in the Inste household economy during the slack periods. The system he described was one motivated by self-interest not paternalism. Industrialization made it less necessary for agricultural workers to accept this. There is evidence that some eastern workers, and in particular their children, gave up their contractual status voluntarily, in response to a general rise in the demand for labour and increased wages outside eastern agriculture.50 So land-owners were obliged to look elsewhere for labour to meet peak seasonal requirements. Russian Poland was the nearest alternative. 49

Gregory (1997). There is a report on this in the Jahrbuch fu¨r die Amtlich Statistik des Preußischen Staats II (1867), 284. 50

Migration, Farm Size, Condition of the Labourer


Taking the long view, the movement of German farmers into the eastern provinces, and into other regions of eastern Europe, was a product of high population pressure in other parts of Germany, and a lack of suitable land for settlement nearby. In the eighteenth century the Prussian state was able to attract settlers from Wu¨rttemberg and Switzerland.51 But around 1860, this eastwards shift of the German population, which had been going on for centuries, came to an end. Instead there is a movement of the German population to the centre and west, and this is followed by an inXow of migrants from lands further east. The reason for this was the rise of industry in the west and the centre. Weber, and other nationalists, saw this as a problem. But for those who left eastern agriculture it was an opportunity. Conditions in industrializing cities were harsh, but for many they represented an improvement on what they had left behind. It was no longer necessary to accept institutional underemployment. The broad conclusion of this chapter is favourable to the Lewis Model and to the concept of the rural labour surplus. The evidence points to the existence of a reserve of underemployed rural labour available for use in industry. But it also makes the point that geographical factors are important. Germany was not an island. It was connected to conditions in the labour markets of eastern and central Europe. Surplus labour in these markets was available for use in Germany. The operation of the Lewis Model in one country can be aVected by conditions in neighbouring economies. 51

Keyser (1959), Hauser (1965).

Migration, Farm Size, Condition of the Labourer


A p p e n d i x 6A. D e s c r i p t i o n o f R e g r e s s i o n V a r i a b l e s a n d So u r c e s %DAGPOP

Percentage change in the agricultural population (includes dependants of occupied population) 1882–1907, from occupational censuses, SDR n.f. 2 (for 1882) and SDR n.f. 209 (for 1907), adjusted to 1882 Kreis boundaries where these had changed. FOODPRICES A weighted index of food prices for 1899–1901, for 165 Prussian towns, weights are 1900 shares in private consumption from HoVmann (1965), applied to the nearby Kreise, prices from Neuhaus (1904). RURALWAGES Average wages in Pfennigs (Ortsu¨bliche Tagelo¨hne) for rural areas; the Wgures are averages of male and female daily rates, 1892 and 1901, ZdKPSB (1904), 320–8. %LAND>100HA The percentage of the total agricultural area in holdings of over 100 hectares, from 1895 Agricultural Census, SDR n.f. 112, pp. 489–500. %AGOCCP1882 The population occupied in agriculture as a percentage of the total occupied population, from the 1882 occupational census, SDR n.f. 2. RUSSDISTANCE Shortest distance in kilometres from the Kreis mid-point to the Russian border (measured from contemporary maps). CITYDISTANCE(a) Distance in kilometres from the Kreis midpoint to the nearest city with at least 200,000 inhabitants in 1900. CITYDISTANCE(b) As above, for cities with at least 100,000 inhabitants. %SUGAR1897 Sugar beet area as % of the total arable area, by Kreise, for 1897, from PS vol. 154 (1898), 8–163. %PARTIBLE The percentage of land transferred by inheritance in 1896–1902 which was divided, ZdKPSB (1905), 262–7. %HOUSEIND The percentage of the non-agricultural occupied population occupied in Hausindustrie, 1882 occupational census, from PS vol. 83 (1884), part 2, pp. 3–466. %SELFEMPLOYED As above, for those recorded as self-employed.

Migration, Farm Size, Condition of the Labourer


A p p e n d i x 6B. R e s u l t s o f E c o n o m e t r i c A n a l y s i s (i) Statistical analysis of wage levels in rural Prussian Kreise 1892/1901 The dependent variable is taken from the Wgures provided by the Reichsversicherungsamt. These were used to calculate average rural wages for each Kreis combining the 1892 and 1901 surveys.52 In the surveys diVerent wage rates were given for male and female workers, and for those aged below 16. The Wgures used are weighted averages of these, the weights being the share of these categories in the total agricultural Table 6B.1. InXuences on rural wages in the Prussian Kreise






69.0 þ0.581 (4.40) þ0.0615 (15.08) 0.390 (8.38) 0.308 (3.94)

84.6 þ0.486 (3.72) þ0.0649 (16.05) 0.458 (9.45) 0.288 (3.77) 4.310 (4.39)

77.8 þ0.516 (3.86) þ0.0681 (14.35) 0.459 (9.31) 0.271 (3.48) 4.142 (4.13) þ0.060 (1.15) þ0.011 (0.66)

87.8 þ0.491 (3.78) þ0.0626 (14.96) 0.442 (9.04) 0.282 (3.70) 4.953 (4.82)


0.0480 (2.03)

CITYDISTANCE(b) N S R2 adj. R2 F-test of regression RSS

392 13.76 0.738 0.735 272.9 73440

392 13.42 0.752 0.748 233.6 69564

392 13.43 0.753 0.748 167.0 69253

392 13.37 0.754 0.750 196.9 68830

Note: Dependent variable is average daily wages in Pfennigs, 1892 and 1901 surveys combined (RURALWAGES), estimated using OLS, t-ratios in brackets. 52

Figures are from Neuhaus (1904).


Migration, Farm Size, Condition of the Labourer

workforce for the whole of Germany (from the 1907 occupational census). They are thus standardized estimates, removing the eVect of variations in the age or sex distribution of the labour force by Kreis. These were then used as the dependent variable in the regressions shown in Table 6B.1. The Wrst column gives the basic results. One important inXuence on wages was the cost of living in the locality. Ideally, it would be possible to construct an index of real wages, making allowance for housing costs, and the prices of all consumer goods, but the data sources are not available for such a calculation. However, there was a survey of food prices in 165 Prussian market towns, published in the Zeitschrift des Ko¨niglich Preußischen Statistischen Bureaus. The data is not given at the level of the individual Kreise, and only Wve items are covered (rye, wheat, potatoes, beef, and pork). This survey was used to construct a weighted index of food prices (FOODPRICE) for nearby Kreise. The Wrst line of Table 6B.1 shows that this does indeed have an important eVect on wages. Wages are higher in areas with high food prices. Other inXuences include the level of employment in non-agricultural sectors in the Kreis: the coeYcient on %AGOCCP1882, the percentage occupied in agriculture according to the 1882 occupational census, is strongly negative. Wages were lower in regions where there was little alternative employment. However, the coeYcient on employment in ‘cottage industry’ (%HOUSEIND—non-agricultural household employment) is a negative one. As the presence of other forms of non-agricultural employment has a positive eVect, this implies that the presence of cottage industry may well be an indicator of underemployment. Cottage industry was found in areas which had a long-term problem of overpopulation and underemployment. However, the most important eVect on the structure of wages in German rural areas was the distance in kilometres from the frontier with Russian Poland (RUSSDISTANCE). This is an important eVect: wages rise by 6–7 Pfennigs per 100 km. A regression equation with just RUSSDISTANCE as the sole explanatory variable has an R2 of 0.673, indicating that this factor alone explains over two-thirds of the observed variance in money wages. Adding sugar beet (column b) improves the overall Wt. The coeYcient is negative, indicating that wages in sugar beet regions were lower than would be expected. This coeYcient is not greatly aVected by the inclusion of other variables in columns c and d.

Migration, Farm Size, Condition of the Labourer


The third column introduces %LAND>100HA, the percentage of land in large holdings. If large estates did pay low wages then this variable should have a negative coeYcient. In fact the estimated coeYcient is positive although not signiWcant at the 5% level. There is no evidence to support the view that larger farms and estates paid low wages; the evidence suggests that they may have raised wages. The CITYDISTANCE variables introduced in columns c and d are used to test the robustness of the estimated coeYcients on the other variables. The strongest eVect on wages comes from CITYDISTANCE(b) (the distance to cities with over 100,000 inhabitants). The coeYcient is negative—wages are lower further away from the major cities—which is as expected. The other results are not seriously aVected.

Table 6B.2. Underemployment and the change in the agricultural labour force a Constant %AGOCCP1882 RUSSDISTANCE CITYDISTANCE(a) %PARTIBLE

21.6 0.500 (7.63) þ0.0468 (10.52) þ0.123 (5.14) þ0.107 (1.30)

%HOUSEIND %SELFEMPLOYED N S R2 adj. R2 F-test of regression RSS

347 17.47 0.409 0.402 59.1 104361

b 30.3 0.458 (5.90) þ0.0485 (11.09) þ0.123 (5.24) þ0.099 (1.23) 0.330 (3.25) 0.359 (2.58) 347 17.05 0.442 0.432 44.7 98505

c 35.2 0.555 (6.86) þ0.0477 (11.86) þ0.181 (7.39)

0.334 (3.00) 0.512 (3.50) 397 19.47 0.432 0.424 59.4 148182

Note: Dependent variable is change in total agricultural population 1882–1907 as a percentage of the 1882 Wgure (%DAGPOP), estimated using OLS, t-ratios in brackets.


Migration, Farm Size, Condition of the Labourer

(ii) Underemployment and migration The analysis in this section looks at the connection between underemployment and migration, making use of data sources which are only available for the Prussian Kreise. The dependent variable in Table 6B.2 is the change in the total agricultural population in the Prussian Kreise between the two occupational censuses of 1882 and 1907. The Wrst three variables are those found to have the largest eVect on the movement of the agricultural population. The decline was steepest in areas which were closest to the border with Russian Poland: by contrast, farms in central and western Germany tended to raise employment. The decline was also steeper in areas close to major cities, as is shown by the positive coeYcient on CITYDISTANCE(a). There is a strong negative eVect from the percentage occupied in agriculture in 1882 (%AGOCCP1882). This can be interpreted as showing that there was underemployment in many predominantly agricultural areas at that time, so that the decline in the agricultural labour force was higher in Kreise which had little non-agricultural employment. The data set was used to study the eVect of partible inheritance, using %PARTIBLE (the percentage of land transferred by inheritance 1896–1902 which was partitioned). Columns a and b show that the coeYcient on this variable is positive and not signiWcant. There was no tendency for partible inheritance to produce out-migration; there was a slight eVect in the opposite direction. Column b adds two other variables. These are employment in cottage industry and self-employment (%HOUSEIND and %SELFEMPLOYED). Both are expressed as percentages of total non-agricultural employment, so they are possible measures of underemployment outside agriculture. The results in columns b and c show that coeYcients on both these variables are negative and signiWcant. So, summarizing these results, there is good evidence that there was underemployment in rural areas in 1882, and that, in 1882–1907, this was reduced through migration, as the rural labour surplus was transferred to urban labour markets.


Agricultural Productivity, Labour Surplus, and Migration 1. I n t r o d u c t i o n In the previous chapters a number of diVerent explanations of the agricultural ‘labour surplus’ have been considered. Some have been rejected, some have found at least partial support: the agrarian reforms did create a class of landless labourers, there was considerable demographic pressure in parts of rural Germany, there was underemployment in the agrarian sector and its ‘cottage industry’ oVshoot, there were institutional and social changes in agriculture which contributed to out-migration. In Chapter 1 there was a discussion of Theodore Schultz’s critique of the concept of the labour surplus. The essence of Schultz’s view is that, although productivity may be low in traditional agriculture, this is not due to ineYciency or underemployment, but rather to a lack of resources, in particular to a lack of ‘non-traditional resources’: education, science, new productive knowledge. Schultz also stressed the importance of access to produce markets and to capital. Industrialization brings to agriculture access to improved inputs. The creation of a railway network and other transport improvements make it possible to transport bulky inputs like fertilizers and animal feedstuVs over long distances. This also makes it easier to get agricultural produce to market, and, by integrating markets, helps to reduce price volatility. Education is another productive factor which is likely to be aVected by industrialization. Industrializing societies, particularly ones like Germany which develop a specialization in advanced technology, Wnd it desirable to invest in educational improvements so as to raise the skill level of the workforce. If the system is a universal one, then the rural sector will also have access to a higher level of general education. In specialist Welds such as scientiWc education, advances in basic science often have implications for agriculture. Nineteenth-century Germany was a classic case of this: advances in the understanding of organic


Agricultural Productivity, Labour Surplus, Migration

chemistry fed into an improved understanding of plant nutrient needs and how to supply these. Industry also supplies improved inputs directly. Steam power could be applied to agricultural tasks, such as ploughing and threshing. Industrial by-products such as ‘basic slag’ turned out to be valuable fertilizers. Improvements in mechanical engineering made it possible to design and build machines such as reaper-binders or seed drills, which were suYciently robust to stand up to being pulled over Welds while in operation. These ‘non-traditional inputs’ can, therefore, lead to productivity increases and improved yields as farmers react to the stimulation provided by rising urban demand and better transport links, by investing in new technology. This does not lead directly to out-migration, but it does have indirect eVects which might produce this. In the Wrst place, it will lead to social change within agriculture: as this becomes more market-orientated, traditional social attitudes and institutions will tend to be eroded. Secondly, it creates the possibility that the nutritional needs of the country could be supplied by a much smaller agricultural sector (measured by the relative size of the agricultural population). Whether or not there is a relative decline in the agricultural sector as a result depends on the evolution of the foreign trade position. It is quite possible for countries to industrialize, to develop modern industrial sectors, and at the same time remain major exporters of primary products, including agricultural goods. Examples include Brazil and Australia in the twentieth century and the United States in the nineteenth. In these examples, the industrial sector concentrated on supplying the home market. Under these circumstances, the relative importance of the primary-producing sectors can remain high despite industrialization. But consideration of these examples shows quite clearly that they were all ‘resource rich’ countries, with comparatively small populations relative to the availability of land and other resources. Late nineteenthcentury Germany was not ‘resource rich’ in these terms, especially as the expansion of European settlement and falling transport costs were bringing new areas into the world trading system which had far more favourable land–labour ratios. Germany’s comparative advantage in world trade lay in her scientiWc base, in the education system, and in the skills and knowledge possessed by urban workers. Germany was ‘skill rich’ not ‘resource rich’. The logic of Germany’s position was, therefore, that the German economy should export skill-intensive industrial goods and import food

Agricultural Productivity, Labour Surplus, Migration


and raw materials. This meant that demand for home-produced agricultural goods was limited to a declining share of the home market. Under these circumstances, improvements in agricultural productivity meant that the available demand could be supplied by a smaller labour force. As Germany became more integrated in the world trading system, resources, including labour, would shift to the exporting industrial sector, and leave the sector exposed to rising imports, which was agriculture. The context for the discussion of these issues is the more general question of the relationship between agriculture and industry in the circumstances of the late nineteenth century, when cheap US grain exports reduced prices on world agricultural markets. European countries reacted in diVerent ways, inXuenced by internal political and social pressures, and by economic conditions. Many tried to mitigate the eVect on their agricultural sectors by increasing tariVs on agricultural goods. Germany was not exceptional in this. An important point about Germany is that it was already industrializing, and that it had a comparative advantage in industrial goods. This meant that it was able to pay for increased agricultural imports by exporting industrial goods. This was not an option available to all European countries. Perhaps only Britain and Belgium were in a similar position.1 A second point about the German position is that the arrival of cheap US grain exports in the 1870s came at a time when Germany was still predominantly rural, and when German cities were already under pressure from rapid growth. By contrast, the agricultural population in Britain had become relatively small by that time, and British cities had been growing steadily for a century or more. The consequences to British society of a free-trade policy were absorbed much more easily. A simple calculation illustrates this point. Suppose that a country with the rural–urban balance of Britain in 1871 (42% of the population in towns with at least 20,000 inhabitants) maintained a free-trade policy in face of rising agricultural imports, and that this caused 10% of the rural population to move to the cities over the next ten years. The additional migrants would amount to 14% of the initial urban population: a burden for the urban areas but not an insurmountable one. The same calculation for a country with a similar balance to Germany in 1 For a general discussion of these issues, see O’Rourke (1997). There is an important distinction to be made between countries which had the capacity to raise industrial exports and other less industrial countries, whose best option was probably to reorientate agricultural exports to the livestock sector, as Denmark did.


Agricultural Productivity, Labour Surplus, Migration

1871 (12.5% in towns with at least 20,000 inhabitants) would produce a movement equivalent to 70% of the initial urban population. This would have raised the growth rate of German cities from 3.0–3.5% per annum to over 8%. Such a rate would have created intolerable strains. If there is a relationship between the social costs of migration and the rate of urban growth, there may be good reasons for a government to seek to slow down the decline of the agricultural sector. Agricultural protection is probably undesirable on economic grounds, but it may be a reasonable response for a society which fears the consequences of a collapse of the agricultural sector.2

2. A g r i c u l t u r e i n t h e G e r m a n E c o n o m y The relative importance of the German agricultural sector fell in 1870– 1913, despite protection. Table 7.1 presents some calculations, from Wgures provided by HoVmann, which show how the changing structure of the German economy aVected the position of agriculture. HoVmann made a reasonable adjustment to the agricultural labour force Wgures to allow for under-recording of family labour in the earlier censuses.3 The resulting estimates show the agricultural labour force rising at a slow rate, 25% in 42 years (1871 to 1913), while the nonagricultural labour force more than doubled (a rise of 125%). The estimated rise in agricultural productivity 1870/2 to 1911/13 is 1.0% per annum, which is relatively high compared to other European countries (such as Britain), but still below the rise in the non-agricultural sector (1.4% per annum). Consequently, the fall in the agricultural share of total net value added is rather greater than the fall in the share of the total labour force.4 It is possible to decompose the fall in the agricultural share of net domestic product into two parts: that which is due to changes in the structure of domestic demand and that which is due to changes in foreign trade. As the table shows, there was a small net deWcit on 2 There are other reasons to support agriculture, including the strategic one. A country dependent on food imports is in a vulnerable position, as was demonstrated by the German experience in the First World War, a point discussed in OVer (1989). 3 On the under-recording of female labour in the 1882 census, see the discussion in Tipton (1976) and also comments in the introduction to the 1895 survey, SDR n.f. 112. 4 The increase in labour productivity in British agriculture averaged 0.5% per annum 1873–1913, Matthews et al. (1982), 228–9.

Agricultural Productivity, Labour Surplus, Migration


Table 7.1. Agriculture in the German economy 1870–1913 (all calculations use 1913 prices) Agricultural share in: Exports as % of NDP NDP

1870–2 38.7 1911–13 23.0

Imports as % of NDP

Labour force

NonAgricultural NonAgricultural agricultural agricultural

49.2 34.9

10.1 18.0

2.7 1.8

13.0 16.4

3.4 6.3

Source: Data from Hoffmann (1965); Hoffmann builds up his accounts from estimates of net value added by sector, the final figure is therefore net domestic product (NDP).

agricultural trade in 1870–2, equivalent to 0.7% of NDP (1913 prices). By 1911–13 this had risen to 4.5% of NDP. The change in the external balance contributed 20% of the total fall in the agricultural share of net value added.5 But this shows that there would have been a substantial fall in the agricultural share even without the contribution of the external sector. The shift in the structure of domestic demand had rather more eVect than the change in the foreign trade position. Comparison of the net import Wgures with the Wgures for total agricultural production (not just value added) shows that agricultural selfsuYciency in all sectors, deWned as total domestic production as a percentage of total consumption, fell from 98.7% in 1870–2 to 88.1% in 1911–13. This exaggerates Germany’s ability to feed itself without using imports, since it takes no account of imported fertilizers and other non-agricultural inputs. Moreover, the importance of fertilizers, in particular, was far greater than that implied by the Wgure of their cost relative to total production.6 Figure 7.1 concentrates attention on trade in cereals. In the period before 1870 Germany exported cereals in most years. Germany supplied 23.5% of British cereal imports in 1856–60, but this share was declining, in 1871–5 it was only 8.2% (the share of the United States 5 This Wgure is based on an estimate of the size of the agricultural share in NDP assuming that the deWcit on agricultural trade had remained at the 1870–2 Wgure. It takes account of the point that imports were themselves Wnal products; their importance tends, therefore, to be exaggerated by a comparison with domestic value added. 6 A calculation of self-suYciency including the eVect of imported fertilizers is not easily done. Some fertilizers also had non-agricultural uses (nitrates were used in the manufacture of explosives). Germany exported potash and imported phosphate and nitrogen. Simply adding up tons imported and exported will give a misleading impression.


Agricultural Productivity, Labour Surplus, Migration





−10 1850







Fig. 7.1. Net imports of cereals as a percentage of total domestic consumption (including use as animal feed), Germany (post-1871 boundaries) 1850–1913 Note: It is difficult to find any clear evidence in this diagram of effects from changes in tariff policy, Bismarck’s duties of 1879–80 produced, at the most, a brief interruption in the upwards trend in imports. The diagram also prompts the question: was it conceivable, in the context of pre-1914 power politics, for any country to have been as dependent on overseas trade for vital supplies, and yet not build a navy to protect this trade? See Offer (1989) for a discussion of the consequences. The issue is also considered in Chapter 10 of this study. Source: Calculated from Hoffmann (1966), 292–3, 530–5, and 537–43.

rose from 18.8% to 40.9% over the same period).7 But before concluding that it was US competition that was the cause of this loss of share, it should be noted that German producer prices actually rose in 1856/60 to 1871/5: by 13% for wheat, and 15% for rye. The eVects of increased US competition in third markets was outweighed by buoyant domestic demand. A careful examination of the timing of the fall in cereal producer prices in the 1870s (given in Figure 7.2) shows that this came after the shift to a net import position. The 1870s were a period of relatively high producer prices, certainly better than the early 1860s. There is no evident downward trend in cereal prices until after the peak in 1873–4. 7

Lambi (1963), also Hobson (1997).

Agricultural Productivity, Labour Surplus, Migration






100 Wheat Rye


Barley 0 1850

Oats 1855








Fig. 7.2. Cereal producer prices 1850–95 (in Marks per ton) Note: The ranking of the crops by production in 1870–2 is, first, rye (40% of the total by weight), then oats (28%), wheat and spelt (17%), and finally barley (15%). Source: From Hoffmann (1966), 552–4.

It was the low prices of the 1878–9 period that changed the attitude of the German agricultural lobby, which had previously supported free trade. At the same time domestic demand for food was buoyant, increasing the viability of a strategy which relied more on protected domestic markets, and less on exports. Table 7.2 gives some Wgures for per capita consumption. Between 1850–4 and 1870–4 there were substantial rises in most categories, apart from pulses and other cereals. The increases in the consumption of sugar and pork are particularly notable. As the German population rose by 16% over this period, the overall increase in demand was high. Rising demand for livestock produce increased the amount of cereals required as livestock feed: principally oats, but also some barley and rye; potatoes were also fed to livestock. This reduced the area that could be allocated to exports. One consequence of this period of relative agricultural prosperity was a rise in average rents, from 14.10 Marks


Agricultural Productivity, Labour Surplus, Migration

Table 7.2. Per capita food consumption: five-year averages Wheat Potatoes Pulses and Sugar Beef & Pork Milk Eggs and rye other cereals veal Annual consumption per capita in kg: 1850–4 1870–4 1890–4 1909–13

90 114 127 128

120 169 242 200

26 18 12 11

2 5 9 18

In numbers: 10 12 15 17

7 13 18 25

272 47 296 54 313 72 381 102

Note: There are reasons to question the 1850–4 figures. The implied increase in per capita calorie consumption of these foods between the first two periods is high (20.8% in 1850/4 to 1870/4 against an increase of 7.8% in 1890/4 to 1909/13). Some of this may be explained by the increased use of cereals and potatoes for domestic alcohol production. But, in the earlier years, Hoffmann may also have underestimated personal consumption of pulses and other cereals (as opposed to their use as livestock feed). High potato consumption in 1890–4 may also be spurious, if a high proportion was fed to livestock in these years. But in general, the implied calorie consumption figures for the period after 1870 are not unreasonable. Source: Calculated from Hoffmann (1966), 622–33.

per hectare in 1850 to 38.16 in 1880.8 This is a clear indication that problems in third markets had not been the principal cause of the reduction in exports. After 1880 the connection between rising imports and falling producer prices becomes clearer. There is a rise in net imports from 7% of total consumption in the period around 1880 to 15% in the mid-1890s, and there is an evident downward trend in producer prices. TariVs did have some eVect in holding back the fall in prices, as Table 7.3 shows. The fall in producer prices was 53 Marks per ton (1871/80 to 1891/ 1900), but the Hamburg price fell by 93 Marks per ton over the same period. Comparison of this price with the prices in Nebraska and New York shows the eVect of falling transport costs: a total reduction of 55%, 70% of this coming from a reduction of costs within the United States. After 1895 net imports of cereals remained at around 15% of total consumption, but net imports of all agricultural produce continued to rise. In 1911–13 net imports were 15.6% of total food consumption; net imports of foodstuVs other than cereals and sugar having risen from 8

Henning (1977), 223.

Agricultural Productivity, Labour Surplus, Migration


Table 7.3. Wheat prices, tariffs and transport costs (a) Average wheat prices, by decade (Marks per Ton)

Nebraska New York Hamburg (f.o.b.) Average German producer prices




106 195 227 224

99 155 164 187

81 123 135 171

(b) German duties on imported wheat (minimum duty payable) 1879–85 1885–7 1887–91 1891–1902 1902–6 1906–10 Tariff in Marks/ 10 Ton




55–75 55 (sliding scale)

Source: Prices in Nebraska, New York, and Hamburg taken from Henning (1977), 219, German producer prices calculated from Hoffmann (1966), 552–4, tariffs from Tracey (1982). These are the minimum rates, so the lower rates for 1891–1902 reflect the trade treaties with Austria-Hungary, Romania, and Russia. American grain was still paying the 1887–91 rate.

2.7% of total consumption in 1870–2 to 12.0% in 1894–6 and 15.5% in 1911–13. The rise after 1894–6 was not due to a increase in meat imports, which fell slightly in the period to 1911–13 (including trade in live animals), but to rises in other categories such as fruit (up 168%) and eggs and dairy products (up 180%) over the same period. By 1911–13 Germany was importing nearly 11% of total consumption of eggs and dairy produce.9 The impressive performance of German agriculture in raising production has, therefore, to be set against a background of buoyant demand. Domestic production was not suYcient to supply all the food required by the expanding urban population. But it was important that productivity rose by as much as it did. A fairly simple calculation shows that, if agricultural productivity had risen by 0.8% per annum instead of 1.0%, domestic food production in 1911–13 would have been 26% 9 Figures from HoVmann. The question of the relative rates of protection given to diVerent agricultural sectors has been a matter of controversy since Gerschenkron’s 1943 study, Gerschenkron (1943), Hardach (1967), and Webb (1982).


Agricultural Productivity, Labour Surplus, Migration

lower. Under these circumstances, the achievement of a similar level of self-suYciency to that actually attained in 1911–13 would have required additional capital and labour. Using a joint marginal output coeYcient of 0.85 for capital and labour (and assuming that the capital input rose in proportion with labour), it can be calculated that the agricultural labour force would have had to be 40% higher, so that agriculture would have continued to account for 49% of the total labour force. This would have had a major eVect on German industrial growth. It also implies a much lower rate of rural–urban migration.10 A similar calculation shows that the cost of achieving full agricultural self-suYciency would also have been high. To achieve the necessary increase in production 16% more labour would have been needed in agriculture. This would have meant that 40% of the labour force would have still been occupied in agriculture in 1911–13 instead of 35%. Again, this would have held back industrial growth.11 There was a cost to protection. Food prices were raised, though by less than the price diVerences given in Table 7.3. Calculations using price ratios from the 1905 survey by the British Board of Trade, shown in Table 7.4, show that food prices were 14.5% above the British level (weighted average of the Wgures in the table). According to the household budget surveys, these items accounted for 40% of total household expenditure. So there was certainly an eVect on real wages from protection. But this was oVset by lower prices for milk and potatoes, presumably reXecting lower German production costs in these sectors. Germany faced some diYcult choices about the position of agriculture within the whole economy, and it is not surprising that this was an issue which was much debated. The outcome which was chosen was one that involved a slow decline in self-suYciency, a rate of industrial growth which was fast by the standards of the time, large-scale migration to the cities, and a level of agricultural protection which imposed some costs on urban consumers.12 It was in many ways a sensible 10 Van Zanden (1991) suggests coeYcients of 0.35 for land, 0.5 for labour, and 0.15 for capital (livestock). This gives a joint coeYcient on mobile factors (capital and labour) of 0.65. This is probably too low. Additional investment could have raised the value of land (by drainage for example) even though there was little suitable land left uncultivated. A joint coeYcient of 0.85 is more reasonable. 11 Full self-suYciency is deWned as an increase of production suYcient to reduce net imports to zero. Germany would still have imported some products and exported others. There would also have been imports of fertilizers. 12 There was, however, a political price. The crisis in agriculture, even if it was a crisis of adjustment, contributed to the radicalization of right-wing politics, Rettalack (1988), 101.

Agricultural Productivity, Labour Surplus, Migration


Table 7.4. German food prices as percentages of British prices, 1905 (percentage of total German household expenditure spent on each category shown in brackets) Wheat and rye flour Wheat and rye bread Sugar Potatoes Milk

140 131 119 88 75

(2.3) (10.0) (1.4) (2.6) (5.8)

Beef Mutton Pork and lard Butter

122 137 123 105

(4.4) (0.3) (8.4) (4.6)

Source: Price ratios from Board of Trade (1908); household expenditure proportions taken from Desai (1968), 122, who used the results of a household budget survey published in the 1909 Reichsarbeitsblatt. Additional information on the consumption of different meats was taken from Hoffmann (1966), 632. Price ratios for pork were also applied to lard; those for wheat flour were used for rye as well. The price ratio for bread was derived using the baker’s premium found by Peterson (1995), 271: 24% of the British price for wheat flour. This was used as an estimate of the absolute cost of producing bread from flour in both countries. The Board of Trade survey provides a better estimate of the effect of German tariffs than calculations based on the movement of prices inside Germany. World prices were falling in the late nineteenth century; German tariffs held back this decline.

solution, which avoided the social costs of a complete agricultural collapse, but still allowed urban living standards to increase. 3. ‘F e w b e t t e r f a r m e r s i n E ur op e’ : Te c h n o l o g i c a l Pr o g r e s s i n G e r m a n A gr i c ul t u r e 1870–1913 In 1921 J. H. Clapham published The Economic Development of France and Germany 1815–1914. Writing about German agriculture in comparison with France, and, by implication, in comparison with Britain as well, Clapham was expressing opinions widely held by contemporary observers of pre-1914 German agriculture. Germany was perceived as the leader of ‘scientiWc agriculture’: there were few better farmers in Europe than the best of the Junkers in the later nineteenth century . . . after the decade 1840–50 Germany had little to learn from any country in agriculture and forestry. In agricultural chemistry she was undoubtedly the leader.13

In his writings Clapham placed great emphasis on the role of the larger farms and estates of eastern Germany: ‘these were the Welds on 13

Clapham (1921), 206.


Agricultural Productivity, Labour Surplus, Migration

which the Wght for scientiWc agriculture was won’. But he also pointed out that these diVered from the typical English estate of the time, which was rented out to tenant farmers: the German Empire was . . . a land of free land-owning peasants and powerful cultivating squires . . . In a broad survey of modern German agriculture tenant farming might almost be omitted . . . East of the Elbe where big estates were common they were nearly all in hand. The typical eastern Junker lived on and directed the management of an estate of perhaps 2000 acres of cultivated land . . . 14

ScientiWc agriculture involved, in particular, the application of the discoveries of organic chemistry to agricultural practice. But it was also recognized that knowledge of these new practices had to be diVused. This involved a major expansion of agricultural education. By 1908 Germany had 26 Agricultural Mittelschulen, 45 Ackerbauschulen, 43 Agricultural Fachschulen, and, for those without access to a full-time secondary school (or unable to aVord the time), there were Winterschulen (204 of these in Prussia alone).15 Agricultural faculties had been established at many universities. There was a network of research stations and experimental farms run by the provincial authorities and the local Landwirtschaftsgesellchaften.16 The new agricultural science was based on an improved knowledge of the nutrients required by plants, and the extent to which these would be removed when the crop was harvested. Soil testing was used to establish the extent to which nutrients could be supplied from resources already present in the soil. DeWciencies could then be made up using a combination of animal manure, artiWcial fertilizers, and ‘green manure’: nitrogenous plants which were ploughed in to provide food for subsequent crops.17 The story of one estate, Kittnau in Kreis Graudenz in West Prussia, illustrates the extent of the changes: 14 The entrepreneurial abilities of the Junkers have been recognized by some authors who are critical of their political inXuence, for example Rosenberg (1978), 94. On the other hand, Puhle (1986), 86, claims that the Junkers refused to invest or modernize their estates, and had a ‘general disinterest in the nature of agricultural production’. According to Schissler (1986), 36, ‘historians agree that the Junkers represent a great political, economic and social burden for Prusso-German history’. 15 Figures from Mu¨ller (1985). 16 The German agricultural education system was surveyed in 1881–3 by the Royal Commission on technical education. This part of the report was published in the Journal of the Royal Agricultural Society of England and Wales (1885), 126–64 and 518–46. 17 Valuable studies of late nineteenth-century agricultural technology and its eVect on productivity include Evershed (1884), Dyer (1896), Rauchberg (1900), Sering (1932), Schremmer (1988), Haushofer (1957), Frauendorfer (1956), Rolfes (1976), Henning (1977), Perkins (1986), Reif (1994), and Fussell (1962).

Agricultural Productivity, Labour Surplus, Migration


Previously there had been little cultivation of root crops. In 1884 a sugar beet factory was constructed at the railway station at Melno, 5 km from the estate, and the manager, Oberamtmann Mu¨ller, immediately switched to intensive sugar beet cultivation. Clover was replaced with a rotation using lucerne, and a third of the farmed land was put into sugar beet. The arable land was drained, a Fowler steam plough was bought and sugar beet was cultivated according to the practices developed in Prussian Saxony. In order to move the sugar beet promptly to the factory, around 5 km of roads were built at a total cost of 40,000 Marks. At the same time cereal yields also increased as a result of the improved husbandry system.18

What evidence is there of a major change in the performance of German agriculture? The Wrst place to look is movement of yields relative to other countries. Figure 7.3 gives average wheat yields in four countries 1848–1913. In the 1850s German yields were at a similar level log scale 4


Germany France Denmark Britain Germany, east Germany, rest


1 0.9 1848







Fig. 7.3. Wheat yields 1848–1913, 1,000 kg per hectare, rolling five-year averages Source: Calculated from data in Mitchell (1981), from national sources, but then adjusted using the 1905–9 Institut International d’Agriculture comparison of cereal yields: Institut International d’Agriculture (1915), 28. The British figures were converted to metric tons using estimates of bushel weights from the 1879 Encyclopaedia Britannica. 18 Dade (1913), i.19. This is a collection of farm histories, providing a valuable description of German agriculture in the period.

Agricultural Productivity, Labour Surplus, Migration


to France, some way below those in the leading countries: Britain and Denmark (yields were also high in Belgium and the Netherlands). From the late 1850s onwards, German yields increased faster than in other countries, so that they caught up with the leading group, and overtook Britain. The German performance was even more impressive when the fact that Britain and Denmark were reducing their wheat acreage, while it was expanding in Germany, is taken into account. The Wgure also shows average yields for east-Elbian Germany, and for the rest of Germany, for the years for which this is available. The convergence of yields in eastern Germany with those in the rest of the country is evident. Part of the improvement in German yields was due to a more even performance within Germany itself. Wheat provides an interesting comparison because it was a widely grown crop. But rye was the more important crop in Germany. Figure 7.4 repeats the analysis for rye. Britain has to be omitted as the rye crop was insigniWcant. This again shows that German yields caught up with those in the leading country, Denmark in this case. German yields showed no upwards trend until 1860. After that they started to rise, but slowly. There log scale 2 Germany France Denmark Germany, east Germany, rest

1 0.9 0.8 1848







Fig. 7.4. Rye yields 1848–1913, 1,000 kg per hectare, rolling five-year averages Source: As Figure 7.3.

Agricultural Productivity, Labour Surplus, Migration


was a marked acceleration after 1890. Again, this was accompanied by a strong process of convergence within Germany, although a gap between the east and the rest remained at the end of the period. Another basic indicator of agricultural conditions in the nineteenth century was livestock intensity. Even though artiWcial fertilizers were coming into use, animal manure was still the main source of plant nutrients—nitrogenous fertilizers in particular were not much used. So, high livestock intensity would not only be associated with a high level of production of livestock produce, but it would also have a beneWcial eVect on arable yields. International comparison of livestock numbers (Table 7.5) shows that overall livestock intensity rose by 39% in Germany between 1873 and 1913, which brought the German Wgure up to the British level. German livestock intensity was some way above France, but well below the Danish Wgure. Danish agriculture had reorientated away from cereal exports and was now concentrating on the export of livestock produce. The composition of the German livestock Wgures changed. Sheep numbers fell sharply, and there was a large increase in pig numbers. Table 7.5. Livestock numbers (in thousands) and livestock intensity

Germany 1873 1913 % change France 1882 1913 % change Denmark 1871 1914 % change Britain 1872 1912 % change






3352 4558 þ36.0

15777 20994 þ33.1

7124 25659 þ260.2

24999 5521 77.9

45.0 62.5 þ38.9

2838 3222 þ13.5

12997 14788 þ13.8

7147 7036 1.6

23809 16131 32.2

37.3 39.9 þ6.9

325 487 þ49.8

1239 2463 þ98.8

442 2467 þ458.1

1842 515 72.0

54.6 105.1 þ92.4

1258 1441 þ14.5

5625 7026 þ24.9

2772 2656 4.2

27922 25058 10.3

57.8 64.9 þ12.3

Source: Data from Mitchell (1981), conversion to Livestock Units (LU) uses weights from Wagner (1896).


Agricultural Productivity, Labour Surplus, Migration

The livestock structure in 1913 resembled the Danish pattern, rather than that of France or Britain, where sheep numbers remained high. The more intensive livestock systems of Germany and Denmark had an eVect on the value of manure as fertilizer. The manure of in-housed cattle and pigs would be stored and spread on the Welds in the spring, which reduced nutrient loss through leaching; sheep by contrast would be left out on the Welds all year, which meant that part of the nutrients would have leached away by the time the crops were Wnally planted.19 Going from these comparisons of livestock numbers and yields to comparisons of productivity involves much greater problems of comparability. It is not a job for the faint-hearted. One of the more ambitious exercises is Bairoch’s series of estimates of calorie production per male agricultural worker reaching back to 1800. There are two problems here. Female labour is ignored, so genuine variations in the male– female ratio by country will aVect the results, and the aggregation of production by caloriWc content can be criticized on the grounds that calorie content is a poor guide to economic value. The calorie content of a potato is greater than that of a bunch of grapes but the price of the latter is much higher. These objections apply most strongly to Mediterranean agriculture, whose products—fruit, olives, wines—often have a high value relative to their calorie content. Productivity in these countries will be underestimated. Conversely, it will be overestimated in countries which specialize in root crops. The calorie output per hectare of potatoes is much higher than the equivalent yield of wheat, yet the value of output is not so diVerent. However, for all these problems Bairoch’s estimates are the only ones available for a substantial group of countries in the early nineteenth century, and some relatively crude aggregation is probably the only option for this period. The results are given in Figure 7.5. According to Bairoch, agricultural productivity in the United Kingdom was substantially greater than in any other European country around 1800. It was a third higher than the Dutch level, the nearest at that time. In the 1860s Britain was overtaken by Denmark, and around 1890 by Germany. Germany rose from Wfth position, level with France, to be second only to Denmark in 1910. Of the other countries, relative losers included Austria-Hungary, France, and (perhaps more surprisingly) the Netherlands; the other relative gainer was Sweden. 19 Nutrient loss through leaching is now a major agricultural concern, on environmental grounds, but it was recognized in the nineteenth century: see Voelcker (1883) on this.

Agricultural Productivity, Labour Surplus, Migration


50 40 Denmark







10 Netherlands 9 8 7 6



5 4 Sweden 1800 1810







1880 1890



Fig. 7.5. Bairoch’s estimates of agricultural productivity 1800–1910 (calories produced per male occupied in agriculture) Source: From Bairoch (1989).

Some of the German improvement is due to the ‘root-crops’ bias. Germany had a proportionately larger acreage of potatoes and sugar beet, and this would produce an overestimate of the value of German agricultural output.20 The value of French ‘Mediterranean’ produce would be similarly underestimated. The use of male labour as the denominator also favours Germany. In the early twentieth century around half the German agricultural labour force was female; only 10% of the British was (as recorded). Thus, dividing by the male labour force raises estimates of German productivity in 1910 by 73% relative to Britain. It also raises it against other countries: by 70% against the Netherlands, by 26% against Denmark, and 16% against France, for example.21 20 In 1913 19.7% of the German arable area was down to root crops: three times the French proportion, and eight times the British, Perkins (1981), 80; this had risen from 14.8% in 1883 and 2.3% in 1800, Schremmer (1988), 50. 21 Calculated from estimates in O’Brien and Prados de la Escosura (1992).


Agricultural Productivity, Labour Surplus, Migration

So, if Bairoch’s estimates produce a distorted picture how can they be improved? The Wrst step is to express output in some unit of value, to deal with the problems created by the calorie measure. The output of diVerent commodities in the various countries should be valued at uniform prices, and combined as an index of production measured in some common unit of account. Generally the unit of account is some monetary measure of given purchasing power (a purchasing power parity or PPP). However, one author (Van Zanden) has chosen to express all prices in terms of wheat equivalents (thus a cow might be worth the equivalent of four tons of wheat, a ton of sugar beet half a ton of wheat, etc.). Table 7.6 gives a range of estimates from diVerent sources, indicating the unit of account, and the productivity measure, which are used by each study. The table concentrates on the comparison between Britain and Germany. This table shows once again that the use of male labour as the denominator aVects all these results. The use of value added as the numerator is also important. This is because German agriculture was more ‘self-suYcient’, it relied less on inputs from other sectors. So ‘value added’ is a higher share of total output. The fact that British agriculture made much greater use of employed labour, which was predominantly male, is one explanation of the diVerent rates of female participation between the two countries. However, Table 7.6. German agricultural productivity compared to the United Kingdom, 1910–13

O’Brien/Prados de la Escosura (PPP) Bairoch (calories) Van Zanden (wheat equivalent) Fremdling (PPP) Broadberry (PPP)

Gross output per head

Value added per head

Both sexes

Men only

Both sexes

Men only






127 133


51 67

Source: This is a reworking of a similar table in O’Brien and Prados de la Escosura (1992): their estimates are combined in a range (German weights and British weights); Broadberry’s estimate is added, from Broadberry (1997); and Bairoch’s estimate is revised from 106 (which came from an earlier study) to 127, from Bairoch (1989). Quite why Bairoch made this revision is not clear.

Agricultural Productivity, Labour Surplus, Migration


the diVerence between Germany and the Netherlands was almost as great, and this is more diYcult to explain. Institutional diVerences were less important. So it may be that the German occupational censuses were rather more diligent in the recording of the labour contributed by wives and other female family members than surveys in other countries. If this is right, then comparisons which use both sexes may understate productivity in German agriculture. Looking at the diVerent estimates, those of Fremdling use a small number of prices to establish the PPP: just the prices for wheat and beef for the whole agricultural sector. These are probably unfavourable to the German side (potatoes, pork, and rye might have produced a diVerent Wgure). The estimates using purchasing power parity calculations and value added per head are the ones which should be preferred on a priori grounds. Discounting the Fremdling estimates, this leaves those by Broadberry and O’Brien/Prados de la Escosura. There should probably be some allowance for the superior coverage of female labour in Germany. This suggests that the most likely level of productivity in German agriculture was around 75–80% of the British level in 1910–13. As far as other countries are concerned, estimates by O’Brien and Prados de la Escosura (Table 7.7) show Germany to be ahead of France (on most measures) but behind Denmark. The comparison with the Netherlands is heavily aVected by the inclusion or exclusion of female labour. Allowing for some under-recording in the Netherlands would put productivity in Germany ahead of the Dutch level in 1910. Between 1890 and 1910 Germany seems to have made gains relative to Britain, France, and the Netherlands, but not relative to Denmark (where the increase was exceptionally high). Van Zanden has constructed an Table 7.7. Production and productivity in selected countries, as compared to the British level, 1890 and 1910 (UK ¼ 100)

Netherlands Denmark France Germany

Output per worker

Output per male worker

Output per hectare







82 44 52 63

89 82 72 89

89 82 72 89

92 147 82 118

192 140 128 148

237 202 136 205


Agricultural Productivity, Labour Surplus, Migration

index of total productivity (combining land, labour, and livestock). This shows that total productivity rose by 1.53% per annum in Germany in 1870–1910, which is just ahead of Denmark (1.31%) and substantially more than in the Netherlands (0.82%), France (0.46%), and Britain (0.19%).22 Broadberry’s estimates also show German productivity rising from 55.7% of the British level in 1871 to 67.3% in 1911.23 So a tentative conclusion is that productivity growth in German agriculture in 1870–1913 was higher than in most other countries with the possible exception of Denmark. This was partly due to the use of techniques already established in other countries. One study has oVered the following view of the increase in German production per hectare 1870–1913: it was 50% due to increased fertilizer use (including both artiWcial fertilizers and livestock manure), 15% due to improved breeding, and 35% due to improved husbandry and reduced losses, including harvest losses.24 Breeding was mainly new technology, although some crops and improved breeds came from foreign countries. Husbandry improvements, on the other hand, often involved the use of machinery, and livestock systems, which had been largely developed outside Germany: drills, reapers, the stall-feeding of cattle. The major German contribution was the development of a husbandry system for lighter soils: the Lupitz system.25 This section will conclude with an attempt to compare crop nutrient supply from artiWcial fertilizers to that available from livestock manure. The use of artiWcial fertilizers was clearly new technology, and an area where Germany was an acknowledged leader. The Wgures presented in Table 7.8 make use of Wgures produced by Schremmer for the nutrients supplied by artiWcial fertilizers. He also provided estimates of the total tonnage of animal manure, but not the nutrient content. The table gives the estimates of nutrient value, using Wgures for the analysis of animal manure given by the German scientist, Augustus Voelcker, who became 22

Van Zanden (1991). Broadberry (1997), 18; Broadberry makes use of the 1937 benchmark comparisons made by Rostas (1948); these incorporate a crude, but possibly justiWable, adjustment of the German agricultural labour force (removing three out of four wives); Broadberry also uses HoVmann’s estimates of the agricultural labour force 1870–1913, thus incorporating (as does everyone else who uses HoVmann in this respect) HoVmann’s upwards adjustment of the German agricultural labour force in the 1870s and 1880s to allow for a supposed under-recording of female labour in the census of 1882. This has the eVect of reducing estimated productivity growth in German agriculture. 24 Estimates given in Mu¨ller (1895). 25 Brouwer (1957) and Dyer (1896). 23

Agricultural Productivity, Labour Surplus, Migration


Table 7.8. Plant nutrient supply, 1800–1914, average annual applications divided by the total agricultural area 1800




A. Estimates of nutrients available from animal manure Nitrogen (kg/ha) 5 14 19 Phosphate (kg/ha) 4 12 16 Potash (kg/ha) 4 11 15 B. Nutrients available from artificial fertilizers Nitrogen (kg/ha) 0.7 Phosphate (kg/ha) 1.6 Potash (kg/ha) 0.8


33 28 26

2.2 10.3 3.1

6.4 18.9 16.7

Source: Schremmer (1988) and Voelcker (1883).

consulting chemist to the Royal Agricultural Society of England and Wales. These Wgures should be regarded as approximations. Schremmer’s Wgures are themselves estimates. The nutrient value of manure depends on factors such as storage practice, time of application, losses through leaching, and the way that animals are fed. Margins of error are likely to be high. Keeping these reservations in mind, the estimates show that, by the years just before the First World War, artiWcial fertilizers were supplying around 40% of the potash and phosphate applied, but only 16% of the nitrogen (an additional source of nitrogen came from the use of clover and other nitrogen-Wxing legumes in the rotation). But they did represent a large share of the increase since 1878–80: over 50% for potash and phosphate, and 23% for nitrogen. Total nutrient applications rose twice as fast after 1878/80 as in the period 1800–1878/80, and this acceleration was entirely due to the use of artiWcial fertilizers. So, the contribution of new technology was, in this area, a major factor.

4. R e g i o n a l Pa t t e r n s o f P r o d u c t i v i t y G r o w t h A problem which has restricted discussion of agricultural developments in Germany has been the lack of estimates of regional productivity levels. HoVmann’s accounts provide national Wgures, but no attempt was made to provide a regional breakdown. Thus, it was not possible to


Agricultural Productivity, Labour Surplus, Migration

consider issues such as whether the problems of eastern agriculture were due to poor productivity rather than low prices. The best way to create Wgures for regional productivity levels is to construct agricultural accounts for each region: dividing estimates of total net value added by the labour force (in full-time labour units to allow for the eVect of parttime employment) produces Wgures for labour productivity. It was considered that the provision of regional accounts would be of value, not only to this study, but also to others working on German agriculture, so three sets were prepared, for 1880–4, 1893–7, and 1905–9. The details of the calculations have been published separately. The results are given in Table 7.9, which shows estimates of labour productivity in German agriculture for 21 regional units for three Wveyear periods: 1880–4, 1893–7, and 1905–9. The table also gives the Table 7.9. Productivity in German agriculture: estimates from regional agricultural accounts in 1913 prices, 95% confidence intervals in brackets Net Value Added per FLU (Full-time Labour Unit)

Annual % rates of growth

1880–4 (Marks)

1905–9 (Marks)


934 (877–992) 1078 (1022–1134) 1222 (1163–1280) 1433 (1363–1503) 1179 (1124–1234) 960 (911–1009) 1388 (1330–1445) 1709 (1600–1819) 1136 (1075–1197) 834 (787–881) 769 (727–811) 757 (713–800) 667 (623–712) 699 (664–735) 1395 (1316–1474) 661 (619–704) 635 (599–671) 1016 (965–1067) 1716 (1632–1801) 1088 (1034–1142) 676 (639–713) 982

2.71 (2.87–2.54) 2.07 (2.18–1.95) 2.08 (2.19–1.97) 2.11 (2.22–1.99) 2.51 (2.63–2.38) 2.26 (2.38–2.13) 0.98 (1.03–0.93) 1.62 (1.74–1.50) 1.53 (1.62–1.43) 1.48 (1.58–1.37) 1.55 (1.65–1.45) 1.41 (1.50–1.31) 1.08 (1.16–1.00) 1.53 (1.61–1.44) 2.32 (2.46–2.17) 0.44 (0.47–0.41) 0.83 (0.88–0.77) 1.60 (1.68–1.5) 1.23 (1.3–1.16) 1.50 (1.59–1.42) 0.51 (0.54–0.47) 1.53

1893–7 (Marks)

East Prussia 480 (454–506) 745 (703–786) West Prussia 648 (615–680) 923 (879–967) Berlin/Brandenburg 731 (697–764) 1001 (957–1045) Pomerania 852 (811–893) 1175 (1121–1229) Posen 636 (607–664) 891 (850–932) Silesia 550 (521–579) 765 (728–802) Pr. Saxony/Anhalt 1089 (1046–1131) 1356 (1304–1407) Schleswig-Holstein 1145 (1072–1218) 1323 (1243–1403) Hannover/Old./Bruns. 779 (737–821) 1075 (1023–1126) Westphalia 579 (544–614) 865 (822–908) Hesse-Nassau 524 (497–552) 798 (760–837) Rhineland 535 (504–565) 761 (723–799) Bavaria excl. Pfalz 510 (484–536) 691 (654–729) Pfalz 480 (458–501) 791 (756–826) Saxony 789 (744–834) 1052 (999–1105) Wu¨rttemberg/Hoh. 594 (565–622) 641 (608–674) Baden 517 (492–543) 631 (601–661) Hesse 685 (654–716) 949 (908–990) Mecklenburg 1267 (1211–1323) 1442 (1380–1505) Thuringia 750 (717–783) 1029 (984–1075) Alsace-Lorraine 596 (571–622) 609 (580–638) Germany 672 855

Agricultural Productivity, Labour Surplus, Migration


annual rate of increase 1880/4 to 1905/9. As the calculations involve a degree of uncertainty, conWdence intervals have been estimated both for the levels of productivity and for the rates of growth. The regional units are those deWned in Appendix 3A.26 The Wrst point which emerges clearly from Table 7.9 is the high rates of productivity growth in eastern Germany. Productivity growth rates of over 2.0% per annum are impressive by any standards. While this does not prove that there had been underutilized resources in these provinces, this is certainly consistent with this explanation. In Chapter 4 regressions were presented which showed that agricultural productivity growth had a positive eVect on migration. The table shows that the reason for this is the high rate of increase in the east, in regions which also had high out-migration. There is a contrast with southern and south-western Germany, which was also agricultural, but where productivity growth was slower. Out-migration was also less. This is the eVect which is picked up by the regression results. If there had been a labour surplus in eastern agriculture then it could have caused both high out-migration, as some of those who were underemployed left, and high productivity growth, as others became more fully employed. A relationship between high productivity growth and high out-migration would be one consequence of this. Looking at the maps of regional productivity (Figure 7.6), it is evident that, in the Wrst period, there was a small group of central regions with high productivity: Schleswig-Holstein, Mecklenburg, and Prussian Saxony. By 1905–9 these regions were still amongst the leaders, but the gap with regions to the west and east had closed. The south and southwest, on the other hand, were lagging. Table 7.10 provides a summary of the changes in the relative position of the east-Elbian provinces. Their share in the total labour force was relatively static, as was the share in livestock production. The factor which increased the share of total value added, and thus produced faster than average productivity growth, was the gain in the share of total production of major roots and cereals (sugar beet, potatoes, wheat, barley, spelt, oats, and rye). It was the gains made by the arable sector which drove the process of convergence. Table 7.11 provides a decomposition of the increase in the production of major roots and cereals, comparing the east-Elbian provinces with the rest of Germany. The major part of the shift in the balance of production 26

Details of the methods used to construct these accounts are given in Grant (2002a).


Agricultural Productivity, Labour Surplus, Migration

Productivity in 1880-4 Net valve added per full-time labour unit in 1913 Marks









900−1,000 Productivity in 1905-9 Net valve added per full-time labour unit in 1913 Marks













Fig. 7.6. Agricultural productivity by region

Agricultural Productivity, Labour Surplus, Migration


Table 7.10 East-Elbian share in total production and value added

Production of: Major roots and cereals Other crops Livestock produce Total net value added Labour force in FLU




33.5 31.6 33.0 32.8 32.7

37.7 27.7 32.2 33.7 32.7

43.3 27.8 33.3 37.6 32.2

Source: Calculated from accounts given in Grant (2002a).

Table 7.11. Decomposition of change in production of major roots and cereals 1880/4–1905/9

Total % increase in production Increase in production per hectare Increase in areas Decomposition of increase in production per hectare: Composition effect Weighted increase in yields

East-Elbian Germany

Rest of Germany

118.5 95.5 11.8

48.0 39.7 5.9

2.9 89.8

0.9 38.1

Note: Calculations use 1905–9 prices.

in favour of the east came about as a result of a faster increase in yields. There was a contribution from an increase in areas, and another from a composition eVect (the substitution of higher yielding, or higher value, crops for others which were less productive or less valuable). But the main factor was the strong tendency for yields in eastern Germany to converge on the higher levels of central and western Germany. What was the cause of this convergence? Clapham suggests that sugar beet cultivation played a crucial role in making producers aware of the importance of artiWcial fertilizers. One source which conWrms this is a survey of fertilizer production and use in Prussia published in the 1887 Landwirtschaftliche Jahrbu¨cher.27 This brought together reports from sixteen regional research stations. It showed that, typically, fertilizer use was two or three times as high for sugar beet as for other crops. It also 27

Thiel (1887).


Agricultural Productivity, Labour Surplus, Migration

showed that most fertilizer was supplied from Wrms within the region. This suggests that sugar beet may have stimulated the use of fertilizers in other ways besides the ‘demonstration eVect’ mentioned by Clapham. The introduction of sugar beet into a region would raise demand, and this might then lead to the construction of additional fertilizer factories. Having a fertilizer factory nearby would then reduce transport costs and so lead to increased use on other crops. Sugar beet cultivation expanded faster in the east. Sugar beet rose from 0.6% of the eastern arable area in 1880–4 to 2.2% in 1905–9, which was a larger increase than in the rest of Germany (1.7% to 2.5%). So the sugar beet eVect is one which promoted convergence. But the pattern of convergence was much wider than this. Sugar beet cultivation was limited to areas with suitable soil and climate (Figure 7.7). Initially, the centre of cultivation was Prussian Saxony, especially the Magdeburg region. From there it spread west and east. But certain regions

>10 8−10 6−8 4−6 3−4 2−3 1.5−2 1−1.5 0.5−1 10 10−15 15−20 45−50










Fig. 8.10. Day-labourer rates in urban areas in German regions, 1912 (weighted averages of male and female rates) Note: Regional averages were calculated by weighting the rates for each city by its share of the total urban population in each region (where the number of cities in the region is greater than one). Regional units which did not have a major city were amalgamated with nearby regions (Koslin with Stettin for example). Source: Data for 94 cities from SJDS (1913), 826.

Migration and Urban Labour Markets


Ortsu¨bliche Tagelo¨hne (day-labourer rates) for urban areas throughout Germany in 1912. The Wgures are weighted averages of male and female rates, using the share of each sex in the total occupied population for the whole of Germany (the male share was 66%). Figure 8.10 conWrms the point that urban wages in the east were below levels in the rest of the country, and it shows that this applied just as much to the industrial areas, such as Silesia, as to the other, more agricultural, eastern regions. It also shows that wages in many southern cities were high. Day-labourer rates in Freiburg, Mannheim, Stuttgart, and Munich were as high as anywhere in Germany. As in the earlier analysis of rural wage rates, the inXuence of the Russian–Polish border is strong. There is a clear east–west gradient, although high rates were also found in the centre, in the Berlin area, and in Saxony. Relating this to the discussion in earlier chapters, it provides a reminder of the point that German labour market conditions were inXuenced by Germany’s position in central Europe, and the proximity of Russian Poland in particular. (ii) Migration and skill endowments There is one possible consequence of migration which is less favourable for the region of origin: if migration depletes the skills available in a region signiWcantly faster than the overall rate of out-migration then the region will become an unattractive location for advanced industries with high requirements for skilled labour. A vicious circle could develop: a lagging region loses skilled workers and then lags even more as a result. The German occupational census of 1907 is well suited to the analysis of this question, as it recorded the place of birth of respondents, as well as classifying them by the skill requirements of their occupations. Consequently, it is possible to compare the numbers born in a given region, of a particular skill level, with the numbers resident at the time of the census. Regions which experience out-migration will have a ratio of numbers born to numbers resident of less than 1. The question is: is this ratio lower for the higher skill categories? Table 8.9 provides an analysis of this type for the main regional groupings. The results show that the east had a net loss of around a quarter of the men born in the unskilled category, but a lower Wgure of just 10% of women. However, female participation rates fell with age, so this may just reXect the fact that many unskilled female migrants withdrew from the labour market after migration. Net losses for the east in the skilled


Migration and Urban Labour Markets

Table 8.9. Ratios of numbers born to numbers resident, by skill category and region, 1907 occupational census East

Centre West and South and north-west south-west

Entrepreneurs and Men 0.803 1.077 self-employed: industry Women 0.845 1.148

1.062 1.044

0.995 0.993

Entrepreneurs and self-employed: services

Men 0.713 1.134 Women 0.706 1.200

1.075 1.059

0.996 0.980

Managerial and technical: industry

Men 0.587 1.129 Women 0.622 1.153

1.070 1.021

0.950 0.960

Skilled workers: all sectors

Men 0.740 1.069 Women 0.744 1.185

1.090 1.024

1.015 0.964

Unskilled workers: all sectors

Men 0.746 1.273 Women 0.903 1.202

1.152 1.035

0.973 0.996

Source: Calculated from 1907 occupational census, SDR n.f. 210.

category are about the same for men and women, which means that they are no higher than the unskilled for men, but rather higher for women. In general the table shows that the east did not lose disproportionate numbers of skilled or entrepreneurial male migrants, with the exception of the ‘managerial and technical’ category, where there were high losses both of men and women. So the question revolves around two issues: how important were the high losses in this category? and can the high losses of skilled and entrepreneurial women relative to unskilled be disregarded? Looking at the absolute numbers, it is clear that the most important ratio in the more skilled categories is the ratio for skilled male workers: there were more of these than any other category. The fact that the east did not lose disproportionate numbers of these leads to the conclusion that the eastern economy did not experience a major loss of economic potential through migration. On the other hand the Wgures do suggest that for white-collar workers of both sexes, and for entrepreneurs and the self-employed in the service sector, the east was a relatively unattractive place to work. This would have contributed to migration. It

Migration and Urban Labour Markets


would also have had some detrimental consequences for the growth of non-agricultural employment. Turning to the other regions: the centre and west made net gains in all categories. It is noticeable that, in both regions, the highest net gains were in the ‘male, unskilled category’; the educational and training system within these regions was supplying most of the skills needed from those born within the region. The south had net losses in all categories except skilled male workers. So the migration balance was slightly favourable to the ratio of skilled to unskilled in the south. The eVect was not so large: migration raised the ratio of skilled to unskilled male workers from 0.0835 to 0.0871. In general, therefore, the migration pattern of the more skilled was not massively diVerent from that of the less skilled, and skill transfers were not a major contribution to the economic success or failure of the diVerent regions. (iii) Migration and the position of women The analysis in Sections (i) and (ii) has shown up some diVerences between the position of men and women. According to the Wgures in Table 8.9, the gap between male and female wages was larger in the east: female pay was 78.8% of the national average in eastern cities in 1912, while the Wgure for men was 85.5%. The analysis of skill transfers has indicated that opportunities for white-collar employment, an important area for women, were better outside the east. So, it would appear that migration out of the east should have led to an improvement in women’s prospects. There are two important problems with this view. The Wrst is the eVect of migration on women’s place in the household economy. Analysis of the budgets of rural households in 1910–14 shows that the household economy contributed as much as 44% of total household income in some rural areas.27 This cannot be exactly equated with the female contribution, but it does give an idea of the potential size of the contribution which could be made by a married woman who did not have formal employment. There is good evidence that migrant households preferred, if possible, to replicate this position in industrial areas, by obtaining smallholdings. This was one factor which contributed to the


Mittler (1929), 210–15.


Migration and Urban Labour Markets

large rise in small agricultural holdings in industrial regions.28 There was an increase of 72,975 in the numbers of holdings of less than 2 hectares in Westphalia in 1882–1907.29 However, this was only possible in industrial areas where agricultural land could be found in close proximity. This was often the case in mining areas but it was rare in the larger cities. The second point is that the gap between male and female earnings was higher in urban areas than in the countryside, and it was higher still in the major cities. Table 8.10 gives Wgures calculated from the 1892 and 1902 surveys of Ortsu¨bliche Tagelo¨hne. The second part also shows that the gap between rates for men and women in cities was tending to rise up to 1902, although there was some reduction in 1902–12. In summary, migration had, on the whole, negative consequences for the economic position of women. Although some may have beneWted Table 8.10. The relative position of women: ratios of male to female wages (Ortsu¨bliche Tagelo¨hne) (a) Average ratios for diVerent localities in Prussia (1892 and 1902 surveys combined) Cities 1.646

Male to female ratio

Towns 1.532

Rural 1.519

Source: Data from Neuhaus (1904). Note: Neuhaus (1904) gives, for all the Prussian Kreise, Wgures for rural and urban rates. These have been divided between ‘cities’ (urban areas with a population of at least 50,000 in 1912) and ‘towns’ (other urban areas). The averages given in the table are unweighted averages following this classiWcation.

(b) Average ratios for 86 German cities

Unweighted Weighted by population





1.591 1.585

1.618 1.639

1.679 1.696

1.657 1.665

Source: Calculated using SJDS (1913), 826. 28 The smallholding could support the family in times of unemployment or sickness. ‘A small piece of land is eVectively a savings bank for the worker’, Wygodzinski in Sering (1899–1910), i. 140. 29 Dillwitz (1973), 98; see Tenfelde (1977), 114–18 for the importance of smallholdings to the Ruhr miners.

Migration and Urban Labour Markets


Table 8.11. Comparison of life expectancy in major Prussian cities to expectancy in the rest of the province, calculated from mortality tables for 1893–4 At birth

At 1 year

At 20 years


Major cities Rest of province DiVerence

34.7 40.0 5.3

47.7 51.8 4.1

56.9 60.8 3.9


Major cities Rest of province DiVerence

39.9 42.9 3.0

52.2 53.5 1.3

62.4 62.8 0.4

Source: Calculated from tables in Kucyzinski (1897); Wgures for individual cities given in Appendix 5B.

from the improved opportunities to achieve white-collar employment, or to become self-employed in the service sector, these were a minority. Wages for women in cities were lower relative to men than in the countryside, and there was a loss of economic status as a result of reduced opportunities within the household economy.30 However, life expectancy Wgures show a rather diVerent picture (Table 8.11). Here, it seems that men were the main losers. At a typical migrant age of around 20, men experienced a loss of nearly 4 years in life expectancy by moving to a city. This was 3.5 years more than the loss experienced by women. Possible reasons for this include the more dangerous nature of the work done by men in urban areas, and the fact that more women were employed as domestic servants, which may have involved better housing or residence in areas with better sanitation. On this point, therefore, it does not appear that women were the main losers from migration.

6. Co n c l u s i o n s The main theme of this study is the movement of labour out of agriculture into urban labour markets, and the conditions and the 30 Kocka (1990), 465 gives Wgures for the gap between male and female earnings: Ritter and Tenfelde (1992), 215 also show that there was a gap in status at the workplace as well. On the other hand, mortality Wgures, as given in Appendix 5B, show that male life expectancy fell by more when rural and urban rates are compared.


Migration and Urban Labour Markets

circumstances under which a ‘labour surplus’ is Wrst created and then released. The preceding chapters have, therefore, concentrated on the rural sector—on technological, demographic, and institutional developments within this part of German society. But the urban context is also important. If the urban sector is under pressure—from capital shortage or from adverse movements in the international terms of trade—then the ability of the urban economy to absorb the rural surplus, without putting downwards pressure on wage rates, is impaired. From this point of view, there were two main phases in the development of the industrial sector in this period. The Wrst lasted from the 1870s to the 1890s, and was characterized by favourable movements in the terms of trade for the export sector. However, it was also a period when the German capital markets were suVering from the after-eVects of the Gru¨nderzeit boom of the 1870s. This restricted access to capital at a crucial time in German economic development. The second phase began in the 1890s and lasted until the First World War. In this period the capital markets still experienced periodic crises, but these were relatively quickly overcome. Investment levels were high, and only temporarily aVected by cyclical downswings. On the other hand, the export sectors were adversely aVected by the movement of the terms of trade. The recovery of world food and raw material prices from the trough of the 1890s had a part in this, but German industry was also having to cut prices to gain a share on world markets: the price of German manufactures was falling relative to those of other countries. In the long run, Germany overcame this problem by moving up the technological ladder, by developing new products rather than competing in established markets. This process was already under way before 1914. The Wgures for ‘high technology’ exports make this clear. This created the prospect for rapid export growth without adverse movements in the terms of trade, as would be the case for the Federal German Republic after 1945. If world trade had not been disrupted by the First World War and its aftermath, then the Kaiserreich might well have moved into this third phase, which would have made it much easier to achieve real wage improvements.

Migration and Urban Labour Markets


Ap p e n d i x 8A. De s c r i p t i o n o f Re g r es s i o n Va r i a b l e s a n d So u r c e s (i) Appendix Table 8B.1 Dependent variables

Independent variables

For each industry: % change in numbers occupied, calculated from occupational census data given in Kaelble and Hohls (1989). As above, % of the total occupied population occupied in each industry.

(ii) Appendix Table 8B.2 As Table 8B.1, industry variables as indicated. (iii) Appendix Table 8B.3 Dependent variable




Logistic transformation, Yjk ¼ log (r= 1  rjk ), where rjk is the average annual numbers of moves from region j to region k, 1909–13 divided by the occupied population in region k in 1907. Movement data from Reichsarbeitsblatt (annual issues) 1910–14; occupied population from Kaelble and Hohls (1989). The distance in kilometres between each region measured from maps, being the smallest distance between the frontiers. The distance between contiguous regions was entered as 10 kilometres, this being the value which was found to reduce the coeVicient on a contiguity dummy to zero. The total population of the region of origin 1907, data from Kaelble and Hohls (1989). As above, the percentage of the total occupied population occupied in industry, 1907, in the recipient region. As above, for services. As above, for chemicals. As above, for Maschinenbau.

Migration and Urban Labour Markets



As above, for Metallverarbeitung. As above, for construction. As above, for Versorgungsleistung (gas, water, and electricity). As above, for agriculture in the region of origin.

(iv) Appendix Table 8B.4 Dependent variable (MIGt )




Data from SJDS (1907), 69, and subsequent issues, combining annual results for 25 cities: total net moves in was divided by total population. The selected cities were those with complete sets of data for the period. The annual average rate of unemployment amongst trade union members, data from Bry (1960), 325–6. Data from SJDS 1904–13, combining results for 23 cities: total numbers of registered applicants at labour exchanges divided by the number of vacancies, annual Wgures. Industrial production, data from Hofmann (1965), 451–2.

A p p e n d i x 8 B. E c o n o m e t r i c R e s u l t s (i) Analysis of employment change by industry, 1895–1907 The changing structure of German industry had a direct eVect on patterns of migration. As industry became more concentrated and more export-orientated, migrants were drawn towards these areas. The tendency to concentration can be demonstrated using the occupational censuses. Table 8B.1 gives the results of a series of equations estimated using the form Yki ¼ a þ bXki , where Yki is the percentage increase of employment in industry k in region i in 1895–1907 and Xki is the percentage of the total occupied population in region i occupied in industry k in 1895. Any tendency towards concentration should produce a positive estimate for b.

Migration and Urban Labour Markets


Table 8B.1 Regressions of change in employment 1895–1907 on the percentage of the total occupied population employed in each industry in 1895, estimated using ordinary least squares and a two-stage weighted least squares procedure (81 regions) Equations estimated using OLS

constant All industry All services

18.04 2.40

Mining and 93.33 quarrying Metal 274.57 production Metal-working 16.02 Machinery 64.16 Chemicals 82.69 Textiles 2.15 Clothing 9.65 Food and drink 20.46 Construction 52.00 Utilities 170.32 Other industry 10.26


t-ratio R2 0.406 (2.74) 0.087 0.797 (4.26) 0.187

2.74 70.4 4.12 28.1 28.2 0.71 4.11 1.29 1.79 164.3 1.47

Equations estimated using 2S WLS adj. R2 0.075 0.177



0.574 (3.91) 0.895 (4.00)

(0.42) 0.002




(1.65) 0.035




(3.50) (3.18) (1.14) (1.17) (3.68) (1.20) (1.22) (1.41) (4.00)

0.124 4.12 0.102 40.2 0.004 1.24 0.000 0.18 0.135 5.44 0.005 1.16 0.006 0.47 0.012 87.3 0.158 1.34

(3.50) (4.18) (0.10) (0.54) (3.33) (1.01) (0.35) (1.73) (4.04)

0.134 0.113 0.016 0.017 0.146 0.018 0.018 0.025 0.169

There was a problem with heteroscedastic errors in some of the estimated equations. So alternative estimates were produced using a two-stage weighted least squares procedure. There are few major diVerences between the two sets of results, but the 2S WLS results are somewhat clearer, due to the elimination of some nearly signiWcant, but in fact rather spurious, results.31 31 The 2S WLS procedure follows the procedure set out in Greene (2000), 514–15. A weighting variable was calculated, which was wi ¼ 1=(Xi )z for each industry, the value z being obtained by an interactive maximum likelihood procedure. This was then used in WLS estimation of the equation. The reason why this problem arose was that there were some industries where employment was low and highly variable. In metal production, for example, employment was often in single Wgures: 1 person was employed in Bromberg in 1895, but 9 in 1907, an increase of 800%. Using OLS produced absolute errors strongly related to the explanatory variable.


Migration and Urban Labour Markets

There is good evidence of a move towards concentration both in industry and services. Analysis of the service sectors (not reported) showed that this result was heavily aVected by the domestic service sector.32 Within the industrial sector there are four categories which show clear trends towards concentration: metal-working, machinery, clothing, and ‘other industries’. The sector with the best evidence of a move towards dispersion is utilities, although this is not a result signiWcant at the 5% level. The occupational censuses can also be used to look for evidence of vertical integration. This analysis is reported in Table 8B.2. Correlation analysis revealed a number of industries which were linked geographically. The highest value in 1895 was 0.83 for the correlation of employment in mining and metal production. But these industries were already so well integrated that there was little scope for further movement in this direction. Column a of Table 8B.2 shows that there was no signiWcant eVect on the change in employment in metal production from the percentage employed in mining in 1895. Column e shows that there was little eVect from textile employment on the change in the numbers occupied in the clothing industry: the coeVicient on textile employment is positive but not signiWcant. Similarly, the results in column b indicate that there was some eVect from employment in metal production on the increase in employment in metal-working, but the coeVicient was not signiWcant. The machinery industry was the one which showed the strongest inXuence from other industries. As column c shows, the increase in employment in this industry was inXuenced by the presence of metalworking in 1895. In column d other variables are added. These results indicate that the strongest eVect on the rate of growth of machinery employment came not so much from the percentage employed in machinery in 1895, but rather from the presence of metal-working and from employment in utilities. This last eVect may well reXect the increased importance of electrical power, as the electrical industry was included in this category. The machinery industry was moving to the areas most advanced in the use of electricity.

32 Which suggests a ‘middle-class agglomeration’ eVect. However, this result is not unimportant: domestic service was an important source of employment opportunities for female migrants, and the availability of such opportunities was therefore a factor in migration decisions by women.

Migration and Urban Labour Markets


Table 8B.2. Regressions of change in employment 1895–1907 on the percentage of the total occupied population employed in related industries in 1895, estimated using OLS (81 regions) Metal Metal-working Machinery production a Constant MINING95 METALPROD95

281.12 17.55 (0.63) 30.52 (0.40)









þ2.63 (1.10) þ3.47 (2.65)


þ14.96 (4.36) þ18.27 (2.20)



þ9.38 (1.45) þ9.96 (2.71) þ8.76 (1.01) þ295.72 (3.96) þ4.02 (3.57) þ0.20 (0.78)

CLOTHING95 TEXTILES95 N S R2 adj R2 F-test of regression RSS


78 398.80 0.040 0.014 1.56 11926363

81 20.09 0.148 0.126 6.76 31491.7

81 56.48 0.287 0.269 15.73 248817

81 51.45 0.424 0.393 13.97 201211

81 14.09 0.153 0.131 7.04 15480.2

Note: t-ratios in brackets. As before there were heteroscedasticity problems with column a; however, no corrective mechanism was found which produced signiWcant results for this equation, so the results are not reported.

(ii) Industrial development and migration in 1909–13 The eVects of industrial developments on migration can be studied using data provided by the various Landesversicherungsanstalten—the oVices which administered accident and sickness insurance. When an insured worker left one region for another, a Quittungskarte was exchanged between the regions. Total numbers of Quittungskartenaustauch were published annually in the Reichsarbeitsblatt from 1910 onwards.


Migration and Urban Labour Markets

This provides a more complete picture of annual movements than any other published source. Even though data are only available for Wve years, 1909–13, it still gives a valuable insight into the operation of the German labour market in the years just before the First World War. One reservation is that the data covers only workers who were already insured. Younger migrants, who had not been insured, would not be included. In Table 8B.3 the linear log-odds model, used in Chapter 5, is applied to the Quittungskartenaustauch data. Here, the focus is on the characteristics of the recipient region (k), and so the dependent variable is a logistic transformation of the number of recorded moves into region k from region j, divided by the total occupied population in that region (from the 1907 occupational census). The Wrst column gives the basic model. The probability of a move into a given region is inXuenced by distance, by the size of the population of the region of origin, and by the amount of non-agricultural employment in the recipient region. The much higher coeVicient on service employment conWrms that this sector was an important source of opportunities for migrants. The percentage occupied in agriculture in the region of origin was also entered, but this variable was not found to be signiWcant and so was dropped.33 Columns b, c, and d show that, entered separately, the export-orientated industries have a positive eVect on the probability of inward migration. Due to collinearity, when entered together only chemicals and metal-working have signiWcant coeVicients. The Wnal column shows the results of a step-by-step procedure, in which all eleven industries and seven service sectors were entered. The procedure identiWed four industries as having a signiWcant eVect on migration (no service sectors survived the procedure). These are chemicals, metal-working, and construction, with positive coeVicients, and utilities, with a negative coeVicient. The attraction of work in construction to migrants is not unexpected: it oVered easy entry into a relatively unskilled urban occupation. Moreover, the transitory nature of the work meant that in any given period a number of workers would be on the move, as projects in one area came to an end and others started up elsewhere. The negative coeVicient on utilities suggests that few jobs in this sector were open to migrants. 33 This might show that agricultural regions were no longer major sources of migrants, but as many migrants out of agricultural regions would not have had insurance cards, this conclusion may not be justiWed.

Migration and Urban Labour Markets


Table 8B.3. Regression results for linear log-odds migration model: 1909–13 a Constant LOGDISTANCEkj POPREGORIGINj INDUSTRY%k SERVICES%k REGORIGIN%AGj

5.685 0.454 (38.4) þ0.378 (16.29) þ0.0170 (5.62) þ0.0637 (15.94) 0.0003 (0.14)

b 6.104 0.455 (38.9) þ0.380 (18.09) þ0.0138 (4.49) þ0.0633 (16.03)

c 5.830 0.451 (38.1) þ0.379 (17.97) þ0.0162 (5.37) þ0.0630 (15.83)

d 5.822 0.453 (38.4) þ0.379 (17.94) þ0.0164 (5.44) þ0.0571 (11.91)

þ0.0762 (3.11)


7.723 0.451 (38.8) þ0.380 (18.32) þ0.0267 (5.21) þ0.0799 (6.35)

þ0.0699 (2.74) þ0.0356 (2.46)

MACHINERY%k þ0.0553 (4.56)


þ0.0510 (4.00) þ0.0495 (4.38) 0.7230 (2.34)

CONSTRUCTION%k UTILITIES%k N S R2 adj R2 F-test of regression RSS


930 0.9393 0.713 0.712 460.1 815.2

930 0.9289 0.720 0.718 474.6 797.2

930 0.9344 0.716 0.715 466.9 806.7

930 0.9362 0.715 0.714 464.3 809.9

930 0.9169 0.728 0.725 307.9 774.2

Note: t-ratios in brackets.


Migration and trade cycles 1904–12

Annual data for movements into and away from major cities can be obtained from the published series of town migration Wgures produced as a result of the system of MeldepXicht, compulsory registration after a change of residence. For 1904–12 a series can be produced covering 25 cities, which includes all the major cities with the exception of Munich.


Migration and Urban Labour Markets

Table. 8B.4. Regression analysis of migration into major cities 1904–12 Dependent variable is: MIGt a Constant UNEMPLOYMENTt

4.335 1.320 (3.88)


DMIGt b 0.187

DMIGt c 0.190


1.046 (5.78)


2.616 (7.44)



0.223 (2.31) 9 0.568 0.683 0.638 15.08 2.259 0.55

8 0.341 0.848 0.822 33.4 0.698 3.08

8 0.274 0.902 0.886 55.29 0.449 2.91

8 0.636 0.471 0.383 5.34 2.424 1.01

Note: t-ratios in brackets.

Although there are only Wgures for nine years, it was still possible to produce satisfactory econometric results (Table 8B.4) once Wrst diVerences have been used (to deal with the high autocorrelation of residuals shown in column a). Various diVerent dependent variables related to economic conditions were tried. Column b uses the change in unemployment; column d uses the change in industrial production. The coeVicients on both are signiWcant. The best results are those given in column c, possibly because the applicants/vacancy ratio was obtained from the cities themselves, and thus recorded conditions in these cities rather than conditions in the economy as a whole. The results in column d are the weakest, and the Durbin Watson statistic is within the inconclusive range (dL ¼ :773, dU ¼ 1:332) for a sample of this size.


Industrialization, Migration, and Inequality 1. I n t r o d u c t i o n The analysis contained in the preceding chapters has provided ample evidence of the existence of labour surplus conditions in Germany in the late nineteenth century. The sources of this surplus were various: population pressure as a result of the demographic transition; the release of underemployed labour in agriculture and cottage industry; the increased use of seasonal migrant labour in agriculture; technological change which raised agricultural productivity; altered social relations inside agriculture as a result of changed economic circumstances. The combined impact of these changes was to create a period when labour supplies were highly elastic. High levels of rural–urban migration provided the motor for rapid economic growth. But this also meant that Germany faced social and economic problems which were unprecedented in the experience of European industrialization. This amounts to a qualiWed endorsement of Max Weber’s views about the origins of the agricultural labour surplus. The main qualiWcation is that the release of surplus labour was brought about by other factors besides Weber’s mechanism involving the decline of the patriarchialische Arbeitsverfassung due to the spread of capitalist values. But the basic point, that there is surplus labour within agriculture and that this surplus is released as a result of changes associated with industrialization, is endorsed. Comparison with other countries shows that the German experience was unique. No other European country had such a rapid transition to an urban industrial society. The rate of growth of the major cities was unusually high; the numbers migrating internally was also above the levels in other countries. Only the United States went through a similar experience, but this was in a very diVerent context, where most migrants were immigrants.


Industrialization, Migration, Inequality

Intertwined with these problems were other sources of diYculty. As a late-industrializing country, with a massive need to import food and raw materials, Germany had to expand exports rapidly, gaining share at the expense of other countries. This meant that wages had to be kept down. Germany’s position in central Europe made it a natural destination for migrants from other countries in central and eastern Europe. This connected Germany to other ‘labour surplus’ countries and prolonged the labour surplus period in German industrialization. The requirements of industrial growth also placed strains on the German capital markets, resulting in the Wnancial crisis of the 1870s. A ‘capital shortage’ phase coincided with the period of labour surplus. The implications of the labour surplus period are set out in the propositions of the Lewis Model: growth will be high as underemployed labour is transferred to productive occupations; wages will tend to lag behind productivity in the advanced or industrial sector; the proWts share will rise; the beneWts of industrialization will go disproportionately to the owners of capital. Inequality in the urban sector will be high and remain so until the labour surplus phase is over, when the ‘turning point’ is reached. As noted in Chapter 1, the Lewis Model is one possible explanation of the Kuznets Curve: the prediction that inequality will rise during the initial stages of industrialization, and then fall as the economy reaches maturity. Bringing together the views of Weber, Lewis, and Kuznets produces a coherent model of economic development: the impact of industrialization and the spread of capitalist values brings about the release of surplus labour from agriculture; this holds back wage increases in industry and raises the proWt share; which in turn leads to a rise in inequality in the economy as a whole. This chapter examines the evidence of a rise in inequality, making particular use of Prussian tax statistics. This is an unusually valuable source, which makes it possible to study the evolution of income inequality in an industrializing society over a long period, something which is rarely possible even with modern statistical materials.1

1 An important point is that there were no conXicts long enough to aVect the distribution of income in the 1815–1914 period, something which makes it hard to draw general conclusions from the movement of inequality in the major industrial countries in the twentieth century.

Industrialization, Migration, Inequality


2. T he K u z n e t s C u r v e a n d G e r m a n H i s t o r y The idea that, during industrialization, the relationship between growth and inequality follows the path of an inverted ‘U’, with inequality Wrst rising and then falling, has long been associated with Simon Kuznets, whose 1955 article gave it wide currency. However, the idea that the distributional consequences of industrialization might not be favourable was certainly not a new one. Apart from its role in the socialist critique of capitalism, it was also a concern of the German historical school, and discussion of income distribution was an important component in the German debate over Manchestertum: the pros and cons of laissez-faire and free trade economics. Evidence that the income distribution was worsening fuelled fears that the lower and middle classes were not beneWting from industrialization to the same extent as the rich. Kuznets was inXuenced by the same evidence, drawn from Prussian tax records. The Soviet economist Prokopovitch had used these in a study of income distribution, and, at the time of the 1955 article, this was the only empirical support that Kuznets had for the initial rise in inequality.2 As a result of Prokopovitch’s work, Kuznets was aware of the diVerence between rural and urban inequality in nineteenth-century Prussia. He found a similar pattern in studies of inequality in the United States and India.3 This led him to a more general statement of the relationship between inequality and economic development. Migration had a central role in Kuznets’s own explanation of his ‘inverted U’ relationship. Kuznets’s model had three parts: 1. There are two sectors A (rural) and B (urban); incomes are higher in B; there is migration from A to B; 2. Income inequality is higher in B; 3. Income inequality in B will tend to fall.4 Kuznets showed, using mathematical examples, that under certain assumptions the Wrst part, on its own, could produce a ‘Kuznets Curve’. This eVect was intensiWed by the second part. But Kuznets also thought that income inequality in the three countries he was considering (Britain, Germany, and the United States) had in fact stabilized while the Wrst two components were tending to cause it to rise, and he 2 Prokopovitch’s work was published by Keynes in the Economic Journal, Prokopovitch (1926). 3 Kuznets (1955), 7; the German Wgures came from Prokopovitch (1926). 4 Kuznets (1955), 7–8 for parts 1 and 2, p. 17 for part 3.


Industrialization, Migration, Inequality

attributed this to the third component (for which he had no direct evidence). The Lewis Model provides an explanation for parts 2 and 3 of Kuznets’s model: it explains high inequality in sector B, and it also predicts that this will fall once the labour surplus phase is over. Later Kuznets made a cross-section study of inequality in developing countries which appeared to support his hypothesis. However, this result has been disputed, and more recent studies of contemporary evidence from developing countries have tended not to show an inverted U-shaped curve, so that one participant at a conference was led to declare: ‘the Kuznets curve is a Wction’.5 If the Kuznets Curve is a contemporary Wction, then this leaves some questions. Does the historical evidence for the curve stand up? And, if it does, how could it be that the relationship held for some industrializing countries in the nineteenth century and not for their twentieth-century successors? Is the Kuznets Curve ‘historical fact, but contemporary Wction’? Is it just a ‘peculiarity of German history’? Since Kuznets wrote his article some additional evidence of high inequality during the early stages of industrialization has emerged. Lindert and Williamson concluded, with reference to the United States in the mid-nineteenth century, that ‘there apparently was a Kuznetsian fall from grace as America embarked on the path of modern economic growth’.6 Charles Feinstein has reinforced the pessimistic view of working-class living standards during the British industrial revolution: ‘the majority of the working class . . . had to endure almost a century of hard toil before they really began to share in any of the beneWts that they had helped to create’. In the British context this was a continuation of a pattern of high inequality already established in agriculture.7 In the next section of this chapter, new estimates of inequality, derived from Prussian tax statistics, are presented which show that Kuznets’s ‘inverted U’ hypothesis is relatively well supported for the 1820–1914 period. The peak of the curve was reached around 1900. Thereafter inequality tended to decline. So, the Kuznets Curve is not a Wction as far as nineteenth-century Germany is concerned.

5 Dani Rodrik, the discussion was published in Bruno et al. (1996), the remark is on p.159. For examples of the studies which led to this remark, see Anand and Kanbur (1993), Ravaillion and Chen (1997), and Dollar and Kraay (2000). 6 Lindert and Williamson (1980), 284. 7 Feinstein (1998).

Industrialization, Migration, Inequality


If the Kuznets Curve has some basis in historical fact, then this suggests that it is one possible outcome of the process of industrialization, but not a relationship set in stone. Given favourable circumstances, some countries can avoid the initial rise in inequality; others, facing less favourable conditions, may not experience the subsequent fall postulated by Kuznets. Which leaves the question, what are these favourable or unfavourable factors? In Germany’s case, what were the reasons for the existence of a Kuznets relationship in the 1820–1914 period?

3. T h e G e r m a n D e b a t e Germany, as one of the second wave of industrializing powers, could observe the eVects of industrialization on British society and draw conclusions. There were horriWed reactions from German visitors ranging from Friedrich Engels to Richard Wagner, who saw London as ‘the fulWlment of Alberich’s dream—Nibelheim, world dominion, activity, work, everywhere the pressure of steam and fog’.8 The German debate began once it was clear that Germany was industrializing, and facing similar problems. The eVect of industrialization on income distribution was one aspect of this debate. Gustav Schmoller, writing in 1895, summarized the position as follows: Optimistic free-traders and defenders of the current social order have certainly asserted that economic freedom and the current organisation of society will, on their own, guarantee an increasingly equal distribution of income, while the historical school, the more pessimistic critics of our time, and above all the socialists have, in the main, come to the opposite conclusion.9

As this quote indicates, concern over the movement of the income distribution was not limited to the left. A wide range of opinion feared that industrialization would lead to widening inequality unless government action was taken to prevent this. Considerable impetus was given to the pessimistic view by an article by Ernst Engel, the director of the Prussian Statistical OYce, published in the Zeitschrift des Ko¨niglich Preußischen Statistischen Bureaus in 1875. This contained an analysis of the Prussian tax statistics which appeared to show that the beneWts of 8 9

Following a boat trip down the Thames to Greenwich in 1877, C. Wagner (1978–80). Schmoller (1895), 1067–8.


Industrialization, Migration, Inequality

industrialization were going, in the main, to the higher-income groups. Engel speciWcally referred to the way that the data seemed to support Schmoller’s concerns: ‘There is no doubt that the concentration of wealth has increased.’10 The historical school placed much more emphasis on the role of institutions, particularly the state, in the determination of income inequality. Schmoller argued that there was a tendency for income inequality to rise but that state action could counteract or even reverse this. This was a small, but still signiWcant, breach with free market economics. Although the historical school was not a large group, they were very active, founding journals and producing surveys (including many used as sources for this study), so they had a considerable eVect on opinion.11 The view that the income distribution was worsening was challenged by Soetbeer, who argued that the lower classes were in fact doing better than Engel had thought. Engel was using income tax Wgures at a time when the Prussian tax system was a hybrid, with the lower-income groups paying a Klassensteuer, a tax which was intended to reXect ability to pay, but which may have been less representative at times of rising money wages. Soetbeer made a more systematic analysis of the income distribution, which was less aVected by changes in the tax system. Adolf Wagner, writing in 1904 but using tax data from 1822 onwards, took the view that both the upper- and the lower-income groups were doing well, but at the expense of the middle classes.12 In 1913 HelVerich made an analysis of developments since 1896 which came to the conclusion that, ‘Contrary to the much asserted view that society is developing in a plutocratic direction, the lower-income classes are gaining their full share of the increase in the national income’.13 HelVerich had made a full analysis of the national income and its distribution. His work was one of a series which made use of the data produced by the revised Prussian tax system, which had been in operation since 1892. This was showing a declining trend in inequality, at least amongst taxpayers. It had also stimulated a discussion of the 10

Engel (1875), 142. For discussion of Schmoller’s views, and the role of the historical school, Dumke (1991) and Grimmer (1999). 12 Soetbeer (1882) and Wagner (1904). Soetbeer was a statistician who published on a number of issues, including monetary reform. 13 HelVerich (1915), 128. Karl HelVerich was a director of the Deutsche Bank and secretary of the Treasury during the First World War. 11

Industrialization, Migration, Inequality


statistical problems involved in the construction and analysis of income distributions.14 Despite this progress, there are serious problems with the work of these authors. One is the lack of an agreed standard for the measurement of inequality. A second is a tendency to underestimate gains by those below the income tax threshold. Both Soetbeer and HelVerich put in unchanged Wgures for average incomes in this group. Yet, if an income distribution is shifted upwards, so that the numbers falling below some Wxed threshold falls, the average income of this group will tend to rise unless there is some balancing shift within the exempt group.15 A third problem is that comparisons are often between single years, and over short time periods. A much longer view, less dependent on estimates for single years, is needed.

4. T h e Pr u s s i a n T a x D a t a : Pr o b l e m s a n d O p p o r t u n i t i e s There are three main problems with the use of tax data to examine income inequality. First, part of the population is normally exempt from the tax, due to income levels below the tax threshold, or some other reason for exemption. So, coverage is inevitably incomplete. Secondly, as the income assessment is linked to the tax payment, there is an obvious incentive to minimize declared income and conceal sources of income wherever possible. Some downwards bias must be anticipated. Thirdly, the tax system tends to change over time, which has eVects on coverage, on incentives to evade, on deWnitions of income, on procedures for checking returns, and so on. These are serious problems, but not insuperable. There are two ways of dealing with the Wrst problem. One is to divide the income received by diVerent groups of taxpayers by total personal income, obtained from national income accounts. This is an approach used in modern studies by Atkinson and Piketty.16 It has been applied to the Prussian income tax data by Geisenberger and Mu¨ller.17 There are, however, a 14 See Bresciani (1907) for a discussion of the statistical issues; this article anticipates many of the techniques used in modern studies, including this one. Other studies of income (and property) inequality include Kuhnert (1906 and 1916) and Biederman (1918). Constantine Bresciani (or Bresciani-Turroni) was an Italian economist who studied in Berlin in 1902–7. 15 A point which can be established by simulation using a log-normal distribution. 16 Piketty and Saey (2001), Atkinson (2002). 17 Geisenberger and Mu¨ller (1972).


Industrialization, Migration, Inequality

number of problems with this approach. The main one is that it can only be used for periods for which national income accounts are available, so the Geisenberger and Mu¨ller study starts in 1874. The second is that the income side of the national income accounts is generally the least reliable. Data on physical production is normally available from an earlier stage than information on wages, rents, and proWts. HoVmann’s estimates of income from capital and land in the nineteenth century were derived using methods which have come in for criticism.18 Moreover, for a series which is derived from Prussian tax statistics, the divisor should be based on income accounts for Prussia, and these are problematic, as there is no information on inter-regional trade or transfers of property income (post-1871 Prussia contained 61–62% of the total German population).19 An alternative approach (which was the one used to prepare the estimates given in this study) is to Wnd other sources of information on the incomes of those excluded from the tax system. In nineteenthcentury Prussia, the most important low-income group was that containing the landless agricultural labourers. Information on these, and on incomes in low-paid non-agricultural occupations, can be used to construct an overall distribution. This in turn can be used to derive a GinicoeYcient, or other measure of general inequality, whereas the Wrst approach only yields a measure of the share of top taxpayers in total income. The second problem, the problem of evasion, is inherent in the use of income tax data.20 It is a serious source of possible error, particularly in cross-section comparison with other countries (where enforcement procedures may be diVerent). It is a problem in a time series for a single country if there are changes in the proportion of income which is not declared, as a result of improved methods of oversight, for example. It is clear from contemporary comment on the tax system that there were 18

Problems with HoVmann’s Wgures are discussed in Fremdling (1988 and 1995). The best income distribution Wgures are those derived from the Saxon tax statistics, Jeck (1970); this source has also been used by Hentschel (1979) and Kiesewetter (1988). However, Saxony was a small and not very typical part of Germany, which industrialized early and had a relatively equal land distribution. Given the importance of Prussia, with 62% of the population in 1907, it is clear that any study which attempts to draw conclusions for Germany as a whole must make use of the Prussian data. More recently German studies have tended to look at other aspects of inequality, such as social mobility—see Kaeble and Federspiel (1990) and Kocka (1983)—and there has been little work on income distribution. Kaelble (1986) surveys work up to that date. 20 For a more detailed examination of the pros and cons of tax data as a source of information on income distribution, see Atkinson (2002). 19

Industrialization, Migration, Inequality


changes of this type.21 The reform of the Prussian tax system in 1891–2 led to a more rigorous examination of tax returns. It is important to make some allowance for these changes. The tax system was not constant in Prussia in the period, but evolved in stages from a Klassensteuer, under which households paid a Wxed charge, reviewable every three years, based on an assessment of ability to pay (which took account of property owed, and indebtedness as well as expected income), to a modern income tax system. Comparability between the diVerent systems is not to be expected. These last two problems were addressed by producing estimates of income inequality for periods when the tax system was not subject to major changes (1822–46, 1853–73, 1874–91, and 1892–1914), and then looking for contemporary estimates or other data sources to chain the series together. This proved to be possible for 1873–4 and 1891–2, but not for 1846–53. During this period the Klassensteuer went through a series of changes: urban areas which previously paid a diVerent tax were brought into the system and the top bands were changed to a system more explicitly linked to income. It was not possible to Wnd evidence from other sources which could cover this period. In addition to selecting an approach which should minimize potential sources of error, an attempt has been made to quantify the eVect of those errors which remain. By constructing diVerent variants for each period, based on alternative sets of assumptions, it was possible to examine the sensitivity of the Wgures to the assumptions used in the central estimates, and Wnd extreme values produced by a range of possible assumptions. These variants were then used to calculate variance estimates which were, in turn, employed to construct conWdence intervals for the central inequality measures.22 Appendix A contains the resulting estimates and conWdence intervals.23 The approach is one which is appropriate for a data source which has serious deWciencies by modern standards. It generates a range of probabilities as well as a single set of point estimates. Taking as an example the weakest period, 1822–46, the tax system in these years was only 21

Meisel (1911) gives data on the number of returns rejected or challenged under the post-1892 system. On average, a challenge to an assessment led to an increase of 25–30% in the amount of taxable income. See also Wagner (1880), 612, on the pre-1892 system. 22 This follows the approach set out in Feinstein and Thomas (2001). 23 A full discussion of the issues involved in these calculations, together with details of the calculations and references to the sources used, has been published separately in Grant (2002b).

Industrialization, Migration, Inequality


indirectly related to income and coverage was not universal. Nevertheless, the data does provide certain pieces of evidence. The numbers in Stufe 1, the highest band, rose by 60% between 1822 and 1846, while the total population in the areas subject to the tax rose by only 39%. The numbers exempt from the tax on the grounds of having insuYcient property or income rose steadily, from 36.6% of the population in 1822 (including dependants) to 41.5% in 1846.24 These two factors strongly suggest a hollowing out of the distribution, with numbers rising at each extreme. Moreover, there is evidence from other sources which suggests that real wages for those at the bottom of the distribution were falling in this period. Table 9.1 gives an index derived from Wgures for earnings in the Prussian state forests, deXated by local food prices. Taking these factors into account, the balance of evidence strongly suggests a worsening of the income distribution in the period. But by how much? This is less easy to say. The variance calculations produce a 95% conWdence range for the increase in the Gini-coeYcient of 12.2% to 16.2%, assuming that the central estimates are subject to uncorrelated Table 9.1. Rural wages in Prussia 1820/9–1875/9 (post-1815 boundaries) 1820–9 1830–9 1840–9 1850–9 1860–9 1870–4 1875–9

104.6 101.1 88.7 81.9 91.4 91.3 100.0

Note: Index based on earnings of male day-labourers in the Prussian state forests (1875–9 ¼ 100), deXated by local food price indices. Source: Earnings in Prussian state forests come from Eggert (1883); this gives the results produced by a circular sent to all Oberfo¨rster in November 1879 asking for the movement of wages since 1800. The survey was not repeated. Prices were taken from ZdKPSB (1907), 84–92.

24 The Wgures of those exempt have been adjusted to allow for the eVects of an extension of exemptions to those aged 14–16 and over 60 in 1831.

Industrialization, Migration, Inequality


errors. But examination of the highest and lowest variants (considering the possibility that there are correlated errors produced by incorrect assumptions), and applying variance calculations to these estimates, produces a much wider range: an increase of between 4.3% and 21.4%. This demonstrates that the statement that inequality worsened in the period can be made with a high degree of certainty, even though the exact scale of the increase is less easy to determine. Thus, for all these caveats and notes of caution, the Prussian tax data does present a major opportunity, provided that the data is handled with caution. For 91 years, 1822–1913, the Prussian population paid some form of income-related tax. Coverage was, at times, remarkably high (74% in 1891). The period covers a society going through a process of transformation from an essentially rural, pre-industrial economy in the 1820s to a position as one of the world’s most advanced industrial economies in 1913. Moreover, and in contrast to the twentieth century, there were fewer unexpected shocks to the income distribution: world wars with concomitant rises in taxation and prices, major world depressions on the scale of the 1930s, unanticipated periods of high peacetime inXation as in the 1970s. It is a unique opportunity to study the evolution of inequality in an industrializing society.

5. T h e Ku z n e t s C u r v e i n N i n e t e e n t h - Ce n t u r y Pr u s s i a (i) Estimates for Prussia Figure 9.1 gives the base estimates chained together to produce an index. Because of the gap between 1846 and 1853 there are two base years, 1846 and 1853. The estimated Gini-coeYcients are converted to indices with these years set to 100.25 The shape is approximately that of the Wrst part of Kuznets’s ‘inverted U’. There was a downwards movement in inequality after a peak in the early 1900s, though the extent of the fall was not large (a fall of just 5% in the average Gini-coeYcient estimated for 1910–14 compared to 1903–7). The main downwards phase of the Kuznets Curve for 25 The Gini-coeYcient is a measure of the extent to which the income distribution diVers from perfect equality, all divergences being given equal weight. It is, in eVect, an average of the absolute values of all deviations from the average income level, expressed on a scale of 0 to 1, where 0 is perfect equality (all incomes are the same) and 1 is perfect inequality (one person has all the income).

Industrialization, Migration, Inequality

304 130





80 1822










Fig. 9.1. Estimates of inequality from Prussian tax data 1822–1914: indices of estimated Gini-coefficients (1846/1853¼100)

Germany, as for other industrialized countries, came in the twentieth century. Modern estimates give a Gini-coeYcient for Germany in the 1990s of 0.25, which is approximately the same level as the chained estimate (using 1892–1914 as the base estimates and making no allowance for changes in 1846–53) for Prussia in the 1820s.26 So the Prussian or German Gini-coeYcient appears to have risen 35–40% in the nineteenth century, and fallen by a similar amount in the twentieth. This is in line with Kuznets’s hypothesis. In the Wrst period, 1822–46, industrialization was just beginning, and Prussia was still a predominantly rural society. In these years, the main inXuences on the level of inequality came from developments within the rural sector. These include the eVects of the Prussian land reforms, the consequences of a relatively high level of population growth, leading to a period of quite severe demographic pressure, and the eVect of rural expansion in the east, bringing into cultivation land in remote areas which were not well suited to market-orientated peasant agriculture. In addition, agricultural prices fell in the 1820s and 1840s, which would have disadvantaged the sector as a whole, without necessarily leading to a worsening of inequality within agriculture. 26

Atkinson et al. (1995), 40.

Industrialization, Migration, Inequality


The second period saw the onset of a period of more rapid growth in Germany. The start of more widespread industrialization is generally dated to the 1850s. The tax statistics analysed by Engel seemed to show that the fruits of industrialization were not being equally distributed. Engel’s analysis was probably too pessimistic. The tax system in this period was a hybrid. The richer taxpayers were subject to an income tax, which was responsive to increases in earned and unearned incomes; the less well-oV paid a revised version of the Klassensteuer. This was a period in which, according to available evidence, wages were rising in industry and agriculture. This may have led to a lag in the adjustment of Klassensteuer rates to improved earnings. More importantly, it means that any estimate of the overall income distribution should allow for some improvement in the incomes of those below the tax threshold. The base estimates for this period show an increase in the GinicoeYcient of between 4.3% and 6.0%, but the systematic error variants show a much wider range, from 20.2% to 3.5%. This uncertainty is largely caused by the question of the extent to which the incomes of the exempt population moved up with other incomes. However, this period is also identiWed as one when inequality most probably moved upwards. The fruits of the early stages of German industrialization were not evenly distributed. In the next period, 1873–82, there was a change in the terms of trade which was unfavourable to agriculture. The widening gap between agricultural and non-agricultural earnings had the eVect of pushing up inequality. As the exempt population contained large numbers of landless agricultural labourers, the assumptions made with regard to this group have a signiWcant eVect on the level of overall inequality. The base estimates show an increase in the Gini-coeYcient of between 11.6% and 13.3%. The systematic error variants show a range from 13.6% to 5.7%. The lower Wgure comes from a variant which is based on a more optimistic view of the movement of real earnings in agriculture in the period. So the Wrst three periods studied, covering 1822 to 1892, less the omitted six years in 1846–52, all show strong evidence of rising inequality, even when allowance is made for possible sources of error or bias. The upwards phase of the Kuznets Curve is relatively established for nineteenth-century Prussia. There were a number of diVerent factors at work, but the general trend was towards greater inequality. The Wnal period, 1892–1914, can be divided into two: a Wrst part in which inequality may have continued to rise, and a second one in which


Industrialization, Migration, Inequality

inequality began to fall, albeit slowly and uncertainly. The break came around 1906. Given the nature of the data, there are considerable diYculties with annual movements in the inequality measures. The main problem is the estimation of the movement of the income levels of the exempt. There are a number of possible wage series, but these show diVerent patterns of cyclical movements.27 Thus, while the base estimates for the period 1892–1906 show an increase in the Gini-coeYcient of 5.5–6.0%, the range of possible variants runs from þ8.3% to 1.3%. The period is best regarded as one of relative stabilization, with Xuctuations around a core level which may have still been moving upwards. The exact timing of the peak is uncertain, but it probably lay in the early 1900s. In this period, the recovery of the German capital markets, combined with the ending of the main period of labour surplus migration out of agriculture, produced an improvement in conditions in urban labour markets. The increase in real wages still lagged behind productivity, but this owed more to the adverse movement of the terms of trade. As German industry forced its way onto world markets, it had to keep prices low, and this restricted wage rises. From 1906 onwards there was a slight decline. The base estimates for 1906–14 show a fall of 3.1–3.4%, with a range for the other variants of between 5.4% and 2.7%. There was a cyclical rise in inequality in 1913–14 which reduced the overall decline to 1914.28 But, making some allowance for these Xuctuations, the evidence for this period indicates that the downward phase of the Kuznets Curve had begun. Although the timing of some of these movements can be uncertain, the overall trends are clearer. Throughout most of the nineteenth century inequality was rising. In the 1890s this upwards movement came to an end, and, after a period of Xuctuation, with no strong trend upwards or downwards, inequality began to fall, although the fall before 1914 was a slight one. (ii) Comparison with other countries To get some idea of the scale of inequality in Prussia in the period, it is necessary to make comparisons with other countries. This is the only 27 The main wage series are those given in HoVmann (1965), Desai (1968), Bry (1960), and Hohls (1995). 28 The tax system worked in arrears, so the assessments for 1914 were not aVected by the outbreak of war in that year (as would have been the case with a PAYE system).

Industrialization, Migration, Inequality


way to create a standard of measurement, to decide whether a given Gini-coeYcient is ‘high’ or ‘low’. The peak Gini-coeYcient for Prussia in 1906 is estimated to have been 0.34 (Table 9A.5 in Appendix 9A). The World Bank data set measuring inequality in modern economies shows that only seven out of the 70 developing economies had lower Gini-coeYcients than this. So, even at the 1906 peak, Prussian incomes were more equally distributed than in most developing countries today.29 On this basis the overall level of inequality in Prussia in the period appears to have been relatively low for a country going through the disruptions of industrialization. However, the increase over levels in the earlier part of the century may well have been a factor creating resentment and pessimism over the future distribution of the rewards of industrial progress. Another point of comparison is the experience of other European countries in the late nineteenth century. Figure 9.2 compares the GinicoeYcients estimated for Prussia with estimates for Britain given by Charles Feinstein. The Prussian estimates are chained backwards from the estimates for 1892–1913 (which are the most reliable). ConWdence intervals are also shown (from Appendix 9A). It is not possible to apply the same procedure to the British Wgures, but, given the nature of the data, it can be assumed that the conWdence intervals would be larger than the Prussian ones, even those given for the 1822–46 period. It appears that the movement of inequality in Britain and Prussia followed diVerent paths. Inequality in Britain was consistently high, and not much aVected by industrialization. Estimates produced by Peter Lindert show a rise in inequality in the late eighteenth century, but Britain was already a highly unequal society before industrialization.30 The estimated Gini-coeYcient for 1913 (0.48) would put Britain in the top half of the World Bank data set, with a higher level of inequality than 45 of the 70 developing economies. This is not surprising given the much less equal distribution of land ownership in Britain. Industrialization brought about a transfer of population out of an agricultural sector which was itself highly unequal. The fact that inequality did not rise as a result of industrialization was due to the fact that inequality was just as high inside agriculture as outside. 29 Deininger and Squire (1996), table 1; the calculation excludes Eastern Europe and the former Soviet economies. There are massive problems of comparability in an exercise of this type. For a discussion of some of these see Atkinson and Brandolini (1999). 30 Lindert (1994).

Industrialization, Migration, Inequality


Gini-coefficients 0.6 Prussia

Lower 95% C I

Upper 95% C I






0.1 1802








Fig. 9.2. Estimates of inequality from Prussian tax data, 1822–1914 (including calculations of upper and lower 95% confidence intervals), compared with figures for Britain Note: The 1846–53 gap is covered by assuming that the Gini-coefficient for 1846 was the same as that for 1853. An important point to note is that the probability of a movement from the upper limit of the 95% range to the bottom is 1/1600, which means that the confidence intervals for rates of change are tighter than might be assumed from this graph: see the calculations presented in Appendix 9A. Source: British figures from Feinstein (1988), 723.

The Prussian experience was very diVerent. Inequality in the early part of the century was relatively low, reXecting the more equal distribution of land ownership. But industrialization was accompanied by an increase in overall inequality. There are a number of possible reasons for this. There was an increase in the number of landless labourers in the early part of the century, at a time when real wages in agriculture were falling. This could have been responsible for part of the increase in inequality in the 1822–46 period. There was a rise in the urban–rural wage gap in the 1870s, which would have increased overall inequality in this period. Both these factors were, however, relatively short-term eVects. The longer trend needs a diVerent explanation.

Industrialization, Migration, Inequality


The most likely cause was a shift of population from a relatively equal rural sector to a less equal urban or industrial sector. There are two sources of information on inequality in the urban sector. One is the tax data for the period after 1896. A breakdown into two sectors is available for a number of years. These have been used to produce estimates of inequality which are given in Table 9.2. Some caveats should be noted. These are based on a much cruder breakdown than that used for the estimates given in Appendix 9A, just six tax-paying classes plus the exempt. Consequently margins of error are likely to be rather larger than those given in the appendix. In addition, the same Wgures for household size by income class are used for both sectors, which is a possible source of bias.31 The results show that urban Gini-coeYcients were 28% higher than rural on average. So, a shift of population from one sector to another would have an eVect on overall inequality.32 The table also shows that inequality in both sectors fell substantially in 1900–13. This, however, would have been partially balanced by the eVect of the continuing shift of population into the less equal urban sector. A second source of evidence comes from budget surveys in urban areas. Results from these are given in Table 9.3. Again there are caveats to note. These are based on household frequency distributions with no information about household size. This has the eVect of raising the

Table 9.2. Gini-coeYcients estimated for rural and urban sectors from Prussian tax data, 1896–1913

1896 1900 1905 1910 1913



0.429 0.438 0.396 0.362 0.355

0.351 0.349 0.321 0.261 0.265

Source: Income tax Wgures from Biedermann (1918); income of the exempt calculated using HoVmann’s Wgures for wages in industry and agriculture. 31 Household size was deWnitely larger in agriculture, but this would only be a source of bias if the diVerence was not the same for all income classes. 32 Overall inequality is not, however, just a weighted average of the two sectors. The size of the gap between the sectors is also important.

Industrialization, Migration, Inequality


Table 9.3. Gini-coeYcients calculated from household budget surveys Hamburg Hamburg Hamburg Hamburg Hamburg

1867–8 1873–4 1881–2 1890–1 1900–1

0.651 0.467 0.720 0.534 0.496

Leipzig 1875 Leipzig 1885 Leipzig 1900 Dresden 1880 Breslau 1880 Breslau 1900

0.643 0.613 0.605 0.559 0.542 0.595

Source: Calculated from survey results given in Kuczynski (1908).

estimated Gini-coeYcient by around 0.08 (many low-income households had few dependants). The surveys were undertaken at diVerent times, using methods of recording and selection which could have varied considerably. This may well explain some of the surprising jumps in inequality shown for Hamburg, for example. Allowing for these reservations, the results still show a pattern of inequality substantially higher than the estimates given for the rural sector. They indicate that rural–urban migration would have been a powerful force driving up overall inequality. And, more broadly, they support the view that Germany went through the ‘labour surplus’ phase of the Lewis Model in this period, and that this was driving up inequality in urban areas. These were the circumstances under which the urban voters started to turn to the Social Democratic Party.

6. T he K u z n e t s C u r v e: C o n t e m p o r a r y F i c t i o n? H i s t o r i c a l F a c t ? G e r m a n P ec uliarity ? If the Kuznets Curve was a product of historical circumstances, which may be speciWc to Germany, this leaves a question: what was it that produced the relationship in this period and not in others? Two sector models, connected by migration, have a certain value in the elucidation of points such as the importance of migration costs and the signiWcance of expectations. But they also obscure other points, such as the role of technological change within sectors. They focus attention on changes in the relative importance of the diVerent sectors, not on changes within sectors. Looking at the process of development in the more advanced economies over the past 200 years, this can be summarized as an alteration

Industrialization, Migration, Inequality


of the production function for the economy as a whole. The pre-industrial production function was one in which, by deWnition, the level of productive knowledge was low, and the stock of tangible capital was also low. Physical eVort played an important part in the production process, and the other major factor was the quantity of suitable agricultural land. It might also be noted that the productive knowledge which was available was rarely covered by patent rights, and so there were few ‘owners’ of such knowledge able to charge for its use: medieval harnesses for draught animals were major technological improvements but nobody held a patent.33 The modern industrial or post-industrial economy is one in which agricultural land has little importance in the overall production function. Physical eVort is also relatively unimportant. The stock of tangible capital is much higher. There has also been a massive increase in the level of productive knowledge, and a substantial part of national income now goes to those who possess this knowledge, as skills, patent rights, experience gained through learning-by-doing, or unique capabilities possessed by Wrms and other institutions.34 The modern debate about income distribution mainly revolves around the distribution of skills and the relative rewards received by skilled and unskilled labour. Ways to improve the overall income distribution include the provision of training for those at the bottom of the income distribution, and the achievement of high and uniform standards in education. The modern German vocational training system, which had its roots in the pre-1914 period, is generally seen as a particularly good example of these measures.35 German vocational training was already admired and studied by other countries in this period. So, if the wider diVusion of skills is indeed a major factor tending to bring about a more equal distribution of income, then this factor was already operating in pre-1914 Germany. There are some reasons to expect that skill acquisition and diVusion may lag during the earlier stages of industrialization and migration has a part in this. Vocational training typically takes place immediately after the end of formal education, which was around the age of 14 in Germany in the period. Relatively few migrants had arrived at their destination by this age. Only 13.9% of those arriving in Berlin in 33

Mokyr (1990), 36–8. The role of human capital in the movement of the income distribution is discussed in Atkinson (1996a). 35 Nickell (1996). 34


Industrialization, Migration, Inequality

1881–5 were aged less than 14 at the time of arrival.36 Migrants arriving later would Wnd it more diYcult to obtain formal vocational training. They would be too old to be accepted as apprentices. Employers might be more likely to oVer training to the children of resident workers than to ‘footloose’ migrants. Moreover, from the point of view both of employer and of employee, the prospective rate of return on investment in vocational training would fall with age, as the anticipated period of future employment was reduced. Migrants could receive training prior to migration, but this would be less valuable. Most vocational training involves a practical element, training ‘on the job’. This is only possible in the vicinity of the industry concerned: shipyard apprenticeships will only be available in towns where there are shipyards. Even more general courses are less likely to be available in sparsely populated rural areas. The establishment of a course on electrical engineering is only possible when there are a suYcient number of prospective students in the catchment area to make the course a viable proposition. Such agglomeration eVects will tend to raise skill levels in cities and the areas around them; rural migrants will not have the same opportunities. Kuznets made a similar point, though with less emphasis on skill acquisition: The very fact that after a while, an increasing proportion of the urban population was ‘native’, i.e. born in cities rather than in the rural areas, and hence more able to take advantage of the opportunities of city life in preparation for the economic struggle, meant a better chance for organisation and adaptation, a better basis for securing greater income shares than was possible for the newly ‘immigrant’ population coming from the countryside or abroad.37

The consequence of this is that there will be a delay after the onset of industrialization before skill diVusion starts to aVect the distribution of income. As for the distribution of rewards to holders of other forms of productive knowledge, such as patent rights, in the modern economy this knowledge is more likely to be held by corporations than individuals, and so this is bound up with the broader question of the distribution of capital ownership. However, it can be said that the initial establishment of property rights over certain forms of productive knowledge, through patent laws and laws on copyright, would tend to increase 36

Calculated from Berlin, Statistisches Amt (1887), table V. Kuznets (1955), 17; skill acquisition is also an important component in Williamson’s model, Williamson (1985). 37

Industrialization, Migration, Inequality


inequality unless there are oVsetting factors (the increase in the importance of this source of income might be at the expense of another form of income which was even less evenly distributed, such as agricultural rents). In modern societies the ultimate ownership of capital is relatively widely spread, with institutional investors such as pension funds, with large numbers of beneWciaries, holding a high percentage of total equity. Institutional improvements which widen the range of savings options available to the less wealthy, and which, by bringing down the cost of intermediation, reduce the gap between the rate received by small savers and the rate of return on industrial investments, will tend to decrease inequality. The arguments presented in the last chapter suggested that only a small group of relatively wealthy investors were able to share in the high proWts available from industrial investments in the early stages of industrialization. As the Wnancial system developed, a greater range of investors came to share in these beneWts. If the mature industrial economy is one in which a wider diVusion of skills and improved savings opportunities are tending to reduce inequality, it does not necessarily follow that, in the absence of these factors, a country in the early stages of industrialization will experience a rise in inequality. The rising importance of capital may give a laboursaving bias to the development of technology in the early stages of industrialization, as argued by JeVrey Williamson, but the eVect of this factor could be negated by a decline in the share of rents, if land ownership is highly concentrated, as agriculture becomes less important.38 Much will depend on the starting position: the distribution of income in the pre-industrial economy. This in turn will be heavily inXuenced by inequalities in land ownership. Kuznets’s view was that ‘inequality in the percentage shares within the distribution for the rural population is somewhat lower than in that for the urban population’.39 His evidence came from a comparison of income distributions in India, from US data for the period before the Second World War, and from Prokopovitch’s study of Prussian tax data. But these studies may demonstrate the importance of institutional factors in patterns of land ownership. The United States, for example, had a relatively equal rural income distribution partly as a result of

38 39

Williamson (1985), 86–90; Williamson’s model is discussed in Feinstein (1988). Kuznets (1955), 7–8.


Industrialization, Migration, Inequality

successive Homestead Acts which had favoured the establishment of medium-sized family farms. Other factors which aVect land ownership include climate, infrastructure, and geographical isolation. There are certain tropical crops which are associated with ‘plantation’ production systems, due to economies of scale favouring larger units.40 Where public sector provision of infrastructure is poor, so that farmers have to provide this themselves, the cost of this will be a problem for all except the largest estates. This eVect is intensiWed by geographical isolation. The current expansion of large-scale soya-bean production in the Brazilian Mato Grosso is a good example of the eVect produced by a lack of infrastructure when combined with geographical remoteness. The example of US agriculture shows that soya beans can be successfully produced on small and medium-sized units, given favourable circumstances.41 The state has an important role in this, both because it sets the legislative framework and because it is responsible for infrastructure investments. The United States in the nineteenth century was suYciently democratic to be able to carry through a land settlement policy broadly favourable to the interests of small and medium-sized units; the modern Brazilian state does not appear to have either the desire or the capacity to do this.42 Even in conditions favourable to large-scale production the state can alter the playing Weld so as to prevent the establishment of larger units. British colonial policy in Fiji was hostile to the cultivation of sugar cane on plantations, with the result than many Indian immigrants brought in as indentured labourers were eventually able to set up as small farmers (on leased land). The much-disputed current EU banana import policy is an attempt to protect small-scale producers on Caribbean islands from competition from plantations in central America. Societies make choices which aVect land ownership, and these are often choices which reXect the balance of political power. Two eighteenth-century constitutional monarchies, Britain and Denmark, made diVerent choices, which resulted in very diVerent patterns of land ownership. The range of crops grown was similar (the Danish shift to 40

Engerman and SokoloV (1997) make this point. Article on the Mato Grosso in the Financial Times, Survey of Brazil (3 Oct. 1994), 16. In the Brazilian case the problem may well be that the cost of dealing with the social problems created by one of the world’s most unequal income distributions leaves little money for infrastructure investments in remote regions. In which case, this is an example of the way that high inequality can breed further inequality. 41 42

Industrialization, Migration, Inequality


livestock production came later); neither country was geographically remote. In the British case, the political system produced an outcome in which the interests of land-owners were placed above all other considerations.43 The Danish reforms were more beneWcial to the peasantry. It has also been argued that a more equal distribution of land, and a more equal starting point, will promote skill acquisition and thus prevent a rise in inequality in the urban or industrial sector.44 Putting a child into secondary education or some form of vocational training is a considerable investment for families in poorer countries: there is the cost of feeding and clothing, there are often fees to be paid, but most important of all there is the loss of potential earnings, a valuable addition to the family budget. With high initial inequality, and borrowing constraints on poorer families, rather fewer will be able to aVord this investment. Statistical results which support these views are inXuenced by the fact that, since 1945 there have been a number of ‘high growth–high equality’ cases, countries with relatively equal land distributions which have achieved high growth rates: Japan, South Korea, Taiwan, and the Irish Republic are some examples.45 There have been others with high levels of inequality in land ownership, which have had diYculty sustaining high growth rates even when these have been achieved for short periods: Brazil is a good example. Consideration of the four ‘high growth–high equality’ cases reveals that in all these countries there was a reallocation of property rights in agriculture at a time when some exceptional event made it possible to achieve this without it being seen as a precursor of further measures. If there was an expectation of repeated expropriations, this would aVect investment in agriculture and in other sectors where there might be political pressure for a reallocation of property rights. These events include defeat, or near-defeat, in war, or in a civil war, and the end of colonial rule. Prussia is a good historical example. Defeat by Napoleon at Jena made it possible to carry out a land reform without creating the expectation that similar measures detrimental to the free exercise of property rights would be introduced at some future date. These ‘Jena-like’ windows of opportunity appear rarely, if at all. In their absence, property rights tend to evolve slowly. Sometimes, 43

The importance of starting points was also emphasized by Williamson (1991). Perrson and Tabellini (1994), Galor and Tsiddon (1994), and Clarke (1995); see also Bernabou (1996) for a survey of the arguments. 45 There is an interesting study of the Taiwan case in Fei et al. (1979). 44


Industrialization, Migration, Inequality

technological changes, such as those which (arguably) made the enclosure of common Welds advantageous, can create conditions where suYcient gains can be made to compensate some owners for the loss of their property rights, while reallocating these rights to others. In general, the declining relevance of feudal rights and obligations made it necessary for these to be recast in a more modern form, but this did not determine the eventual outcome. These factors produce a more complex mechanism than the one outlined by Kuznets, and one which is heavily inXuenced by historical circumstances. There are now many possible paths, not just one. Alternative paths include some which are radically diVerent. A country with a extremely high level of inequality in land ownership might experience little or no increase in inequality during the early stages of industrialization simply because the starting position was so high.46 If political problems, or low skill acquisition, then holds back economic growth, this country may never break through to economic maturity and the downward phase of the Kuznets Curve. It could become stuck in a high inequality–low growth trap: an initial pattern of high inequality being perpetuated through low skill acquisition. A second country with a highly equal land distribution may avoid the Kuznets Curve for completely diVerent reasons. The transition to the downward phase could be so fast that the initial rise in inequality would be insigniWcant. If high skill diVusion is established right at the onset of industrialization then this factor could outweigh other forces tending to increase inequality. In which case there would be no Kuznets Curve: inequality would start low and remain low. Kuznets explained high inequality in his sector B, the urban or industrial sector, using an argument which contains elements of the Lewis Model: It seems most plausible to assume that in earlier periods of industrialisation . . . [sector B’s] income distribution was more unequal than that of the agricultural population. That would be particularly so during the periods when industrialisation and urbanisation were proceeding apace and the urban population was being swelled, and fairly rapidly, by immigrants—either from the country’s agricultural areas or from abroad.47

46 Lindert’s Wnding that inequality in Britain started high and remained high may indicate that Britain was in this category, Lindert (1994). 47 Kuznets (1955), 16.

Industrialization, Migration, Inequality


He also saw the concentration of savings as an important factor tending to raise inequality in sector B.48 Lewis himself endorsed the use of his model as an explanation of high inequality: ‘the elasticity of supply of unskilled labour is so high (whether in the dual or the classical models) that it dampens real wages . . . The fruits of growth are concentrated in few hands.’49 This study has looked at the labour surplus assumption of the Lewis Model in some depth. The conclusions are broadly favourable to the view that there is a labour surplus phase in industrialization, though much depends on circumstances: in the German case the inXuence of geographical location, in particular the proximity of Russian Poland, has been shown to be considerable. However, there are two points which have emerged which have important implications for the Kuznets Curve. The Wrst is demographic. In Germany the onset of the fall in birth rates characteristic of the ‘demographic transition’ came early: there was a downward trend in the Princeton Im index (the index of births within marriage allowing for the age distribution of married women) from the late 1870s onwards. This was led by the urban areas. In 1880–1 marital fertility in cities of over 20,000 inhabitants was 12.9% below the Wgure for rural areas; by 1905–6 it was 31.4% below, having fallen by 25.6% compared to the fall of just 5.5% in rural areas. But this fall was itself a continuation of a pattern of behaviour already present in rural society in Western Europe: the restriction of population growth principally through delayed marriage. The main diVerence was that the population in the urban areas began to marry earlier and to restrict fertility within marriage. This, however, was not unusual, as Protestant rural areas followed a similar pattern of behaviour; in Catholic areas, by contrast, people married later. The situation in many contemporary developing economies is very diVerent. As Todaro put it: ‘LDC (less-developed country) birth-rates today are substantially higher than they were in pre-industrial Europe. This is largely because of early and almost universal marriage in contemporary Third World countries . . . since the initial level of birth-rates in Western Europe was generally low as a result of either late marriage or 48

Ibid. 6–8. In Ranis and Schultz (1988), 14; Williamson commented (at the same conference): ‘the upswing of the Kuznets curve is not an unambiguous fact of history’, ibid. 23. Taylor (1979) and Taylor and Anda (1988) explore the distributional implications of the Lewis Model. 49

Industrialization, Migration, Inequality


celibacy, overall rates of population growth seldom exceeded the 1% level even at their peak.’50 The result has been that many development economists take a relatively pessimistic view of the likely pace of population increase: Hayami writes (in 1997): ‘the probability of low-income countries escaping quickly from the strong pressures of exogenously given population growth appears to be relatively slim over the next couple of decades’.51 If the Kuznets Curve is indeed ‘historical fact but a contemporary Wction’ then this very diVerent pattern of demographic behaviour is probably part of the explanation. The labour surplus phase in the industrialization of Western Europe was curtailed, partly by the rapid decline of the urban birth rate, and partly by the fact that birth rates were already controlled in rural areas. Hence, the more rapid transition to the downward phase of the Kuznets Curve. The point is also of interest for the light it sheds on household motivation and behaviour. Take, for example, the Wnding reported in Chapter 5, that birth rates were lower in areas of partible inheritance. This suggests a concern for the well-being of subsequent generations and a willingness to adapt behaviour to secure an improvement. Too many children threatened the viability of the family holding. So numbers were restricted. This may not be entirely altruistic; parents may have considered that their own interests, in having children able to provide support in their old age, were better served by having relatively few children who were better-oV. But it also suggests that, if a move to a city brought improved educational and training opportunities, then similar considerations would lead to a decision to restrict family size so that available resources were suYcient for children to make use of these opportunities. The second point concerns the capital side of the Lewis Model. DiVerential saving rates, especially in the form used by Lewis, ‘only capitalists save’, create a powerful mechanism which may well drive up inequality in an economy which does not draw on international sources of capital. However, once the possibility of large-scale international capital Xows is introduced into the equation, then the domestic savings rate becomes less relevant. With perfect capital mobility, and zero transactions costs (for turning international savings into domestic investment), the whole Lewis Model disappears, as does any concern


Todaro (1997), 196 and 200.


Hayami (1997), 60.

Industrialization, Migration, Inequality


with the rate of population growth and the agricultural labour surplus.52 There was certainly a degree of capital mobility in Western Europe in the mid-nineteenth century, but the largest Xows were those associated with the international bond market: there was little direct investment in industry. In the last quarter this began to change, and there was much more direct investment in those countries which began to industrialize in this period: Italy, Russia, and Austria-Hungary. The electrical industry, which also emerged in this period, was dominated by a small group of German and American Wrms which had subsidiaries in diVerent countries. But this was a relatively new development. The position facing contemporary developing countries is not the same. There is a large pool of international capital available for investment in all sectors of the economy. Provided it is able and willing to oVer conditions favourable to the interests of foreign investors, a successful developing economy should be able to attract suYcient investment to avoid Wnding itself with surplus labour, even with high population growth and large-scale migration out of agriculture.53 Given these factors, it is not surprising that modern studies do not Wnd a Kuznets relationship between economic growth and inequality. The more successful developing economies have managed to avoid the rise in inequality that Kuznets predicted. The less successful have failed to make the breakthrough to economic maturity and the downward phase of the curve. The Kuznets Curve is one path among many possibilities.

7. S o u r c e s o f I n e q u a l i t y i n G e r m a n y 1870–1913 (i) General comments From the discussion in the previous section it appears that there are ways whereby a society going through the early stages of industrialization can avoid a rise in inequality. How well does nineteenth-century 52

The consequences of diVerential savings were examined by Kaldor (1956) and Pasinetti (1962). 53 This is not, however, something that is either easy to achieve, or costless to the developing country. Dependence on external Wnance can lead to instability as a result of changing sentiments in international Wnancial markets, which is one reason why successful developing economies such as Singapore or South Korea have taken steps to encourage domestic savings.


Industrialization, Migration, Inequality

Germany measure up to these standards? Was the rise shown in Figure 9.1 inevitable? Germany started with a relatively equal distribution of income, due to a more equal distribution of land ownership, which was generally more egalitarian outside Prussia, with some exceptions such as Mecklenburg. Historians have criticized the land reforms of the early nineteenth century on the grounds that they left a relatively unequal distribution of land ownership east of the Elbe and also that there was a rise in the numbers of landless labourers. But a broader comparison shows that the reforms, coupled with the previous policy of Bauernschutz which protected and entrenched peasant rights, were at least partially successful.54 Unlike the most successful of the modern developing countries, Germany was not able to avoid a rise in inequality during industrialization. In comparison with these countries, Germany suVered from two main disadvantages: there was little foreign direct investment (by modern standards), and Wnancial institutions were relatively underdeveloped. In particular, the system of universal banks closely tied to industry, which Germany pioneered and which has reappeared in other countries such as Japan or South Korea, was then relatively unknown. In consequence, there was a period when industrial investment was held back, more by institutional problems than by a general shortage of capital, which ran from the 1870s to the mid-1890s. The shift of the terms of trade against agriculture in the 1870s would not, under reasonable assumptions about factor immobility, have led to a permanent rise in inequality. However, it had a powerful short-term eVect, and it was particularly unfortunate that, just as the movement of the international terms of trade required an increased rate of migration from rural areas into cities, the rate of industrial investment should be held back by the institutional problems mentioned in the last paragraph. It also coincided with a period when surplus labour was being released from agriculture as a result of institutional changes, such as the decline of the contractual labour system studied by Weber. German conditions were also inXuenced by geography, in particular the proximity of Russian Poland, which had a considerable inXuence on the structure of rural wages. By restricting wage growth this factor 54 Historians who have analysed the agrarian reforms include Conze (1969), Dickler (1975), Harnisch (1986), Koselleck (1967), Lu¨tge (1963), Plaul (1986), Perkins (1986), and Saalfeld (1963).

Industrialization, Migration, Inequality


contributed to overall inequality even in areas where the numbers of migrants were small. Competition from migrants from eastern and central Europe also aVected German migration to the United States, but it was fortunate that this eVect was mainly felt after 1895, by which time the ‘labour surplus’ period was over. During the period up to 1895, the safety valve of emigration to the United States was open. It reduced the pressure on urban labour markets inside Germany, and thus helped to prevent an even steeper rise in inequality. Given these factors, it is probable that high levels of inequality were inevitable in the urban, or industrial, sector in 1870–95. Nothing much could have been done to avoid this. Higher levels of agricultural protection could have held back the rate of decline of rural employment, but at the cost of a signiWcant reduction in industrial growth, and there would also have been a detrimental eVect on urban wages. Examples from modern developing countries show that, under similar circumstances, the introduction of measures to raise urban wages creates ‘urban bias’, and a division of the urban labour market between a relatively well-paid formal sector, and a low-paid, or underemployed, informal sector. Some measures might record this as an increase in inequality. The same point applies to the activities of trade unions. In this period, trade unions were better organized in the more skilled sectors, so their inXuence, in so far as they did aVect wage levels, would have tended to widen divisions within the urban labour market. The deliberate restriction of internal migration through government action, a practice followed in a number of developing countries (notably China), would have had the eVect of widening the overall income distribution by holding a number of potential migrants in low-paid rural employment. It is a policy which is only likely to appeal to governments which give little weight to the interests of those living in rural areas. The factors that tend to bring the initial rise in inequality to an end have been identiWed as higher investment and increased skill diVusion. In both these areas the German record can be described as exemplary: it was one which other countries have sought to emulate. Both the vocational training system, and the Wnancial system as it developed in the 1890s, were well suited to the requirements of the German economy in this period, and these were the forces which brought about the decline in inequality, in combination with the recovery of German agriculture after 1895, which reduced the urban–rural gap and also decreased the pressure on urban labour markets.


Industrialization, Migration, Inequality

(ii) Analysis by simulation The Wrst year for which separate distributions for both the rural and urban sectors are available is 1896. These can be used to simulate the eVect on overall inequality of diVerent scenarios. Between the 1820s and the 1890s there was a total increase in inequality of around 40%; this is the change that has to be explained. Table 9.4 presents simulations which show that, using Kuznets’s assumptions, migration will push up inequality (scenarios C and E) but that this eVect can be reversed if more realistic assumptions are made, so that migrants come from the lower part of the rural distribution and move to the lower part of the urban distribution (scenarios D and F). So migration, in itself, is not a very convincing explanation of the Kuznets Curve. Table 9.4. Simulations using 1896 rural and urban income distributions I. Using 1896 sector weights: 40% urban, 60% rural Scenario EVect on overall inequality A All rural incomes fall 10% B Incomes of poorest 50% in rural sector fall 10% C 10% of rural population moves to urban sector—no eVect on sector distributions (Kuznets process) D 10% of the poorest half of the rural population move to bottom half of the urban sector

Gini-coeYcient rises 2.6% Gini-coeYcient rises 3.8% Gini-coeYcient rises 1.8%

Gini-coeYcient falls 2.1%

II. Using circa 1850 sector weights: 10% urban, 90% rural E 10% of rural population moves to urban sector—no eVect on sector distributions (Kuznets process) F 10% of the poorest half of the rural population move to bottom half of the urban sector G Overall eVect of move from 10% urban/90% rural to 40% urban/60% rural (Kuznets process)

Gini-coeYcient rises 7.1%

Gini-coeYcient falls 3.8%

Gini-coeYcient rises 14.5%

Industrialization, Migration, Inequality


However, the Kuznets process as a whole, the shift from a predominantly rural society to one in which the urban sector is substantially increased, will increase overall inequality, as scenario G shows. But the eVect only accounts for around a third of the increase in inequality which actually occurred between the 1820s and the 1890s. Other factors which might have contributed are simulated in scenarios A and B: a shift in the terms of trade which has an adverse eVect on agriculture raises inequality, but the eVect is greater if it is strongest on those at the bottom of the rural distribution (scenario B). In general, the strength of the rise in inequality shown by the analysis of the Prussian income tax data is such that it was almost certainly brought about by a combination of adverse factors. These were: 1. a shift of population from a more equal rural sector to a less equal urban sector (Kuznets process); 2. increased inequality in the rural sector (due to a rise in the numbers of landless labourers); 3. increased inequality in the urban sector (due to the operation of the Lewis Model); 4. a shift of the terms of trade against agriculture. (iii) Econometric analysis The Prussian tax statistics can also be used to examine some of the causes of inequality in the rural Kreise. In 1916 an analysis of tax returns in 1914 at the Kreis level was published in the Zeitschrift of the Prussian Statistical Bureau. The assessments were based on income in the previous year, so this is an analysis of the Prussian income distribution in the Wnal year before the war.55 The data gives the numbers of taxpayers, and their dependants in four categories: those with annual incomes below 900 Marks, those with between 900 and 3,000 Marks, those with between 3,000 and 9,500 Marks, and those with over 9,500 Marks. This made it possible to produce estimates of inequality at the Kreis level, and these could then be subjected to statistical analysis. The results of the overall analysis are summarized in Table 9.5. The analysis showed that diVerent-sized cities were associated with diVerent levels of inequality in nearby regions. Larger cities tended to 55 There is a full report on the sources used and the results of the econometric analysis in Grant (2002b).


Industrialization, Migration, Inequality

Table 9.5. Summary of results of statistical analysis of inequality in 583 Prussian Kreise, 1913/14 I. Secure statistical results Factors which tend to increase inequality: Proximity of the border with Russian Poland Remoteness (as measured by the distance from towns and cities with at least 20,000 inhabitants) Factors which tend to decrease inequality: High rates of agricultural employment II. Less secure statistical results Factors which tend to increase inequality: Proximity of large cities (at least 100,000 inhabitants) Source: Summary of the results given in table 13 of Grant (2002b).

have high levels of inequality, and this appears to spill over into nearby regions. But the eVect of smaller towns was more beneWcial: inequality was lower in regions which were close to these. Kreise which were remote from these smaller towns tended to have high inequality. Inequality was lower in the more agricultural Kreise, conWrming the sectoral diVerences shown in Table 9.2. There was also a pronounced east–west division: inequality tended to fall as the distance from Russian Poland increased. However, in the rural areas the land distribution was a major inXuence on overall inequality, so it may have been the unequal land distribution which was causing inequality to be high in the rural east. This can be demonstrated by considering the rural Kreise separately; an analysis of this type is summarized in Table 9.6. This conWrmed some of the points made in Table 9.5: inequality was high in areas close to big cities, and in those which were remote from all urban areas. But the inXuence of Russian Poland was considerably weakened, and may well be insigniWcant in statistical terms. While the earlier table might be interpreted as showing that inequality was ‘imported’—the availability of cheap labour from Russian Poland was pushing up German inequality—it now appears that inequality in the east was to a large extent ‘home-grown’. It was the unequal land distribution in the east which caused inequality to be high in these areas. Turning to the major cities, Table 9.7 reports the results of the analysis of inequality in these urban areas. There was no tendency for inequality to be higher in cities further to the east, which conWrms that the

Industrialization, Migration, Inequality


Table 9.6. Summary of results of statistical analysis of inequality in 478 rural Prussian Kreise, 1913/14 I. Secure statistical results Factors which tend to increase inequality: Remoteness (as measured by the distance from towns and cities with at least 20,000 inhabitants) Presence of large farms and estates (over 100 hectares) Factors which tend to decrease inequality: High rates of agricultural employment Proximity of large cities (at least 100,000 inhabitants) II. Less secure statistical results Factors which tend to increase inequality: Proximity of the border with Russian Poland Presence of medium-sized farms (20 to 100 hectares)

Table 9.7. Summary of results of statistical analysis of inequality in 44 Prussian cities, 1913/14 I. Secure statistical results Factors which increase inequality: High levels of migration (moves both in and out) II. Less secure statistical results Factors which tend to increase inequality: Presence of textile, mining and printing industries Presence of utilities Factors which tend to decrease inequality: Presence of leather industry

east–west division was mainly due to diVerences in the land distribution. However, inequality was higher in cities with high levels of migration, measured by the total number of moves, in and out, relative to the resident population. In these cities, with highly Xexible labour markets and elastic labour supplies, there was a larger gap between rich and poor. This provides support for the view that the migration of the rural labour surplus into urban areas was a factor driving up inequality. However, the main ‘labour surplus’ period was over by 1914. If an eVect can be


Industrialization, Migration, Inequality

found for this date, it suggests that the impact would have been much greater in the 1870–95 period, when the bulk of the transfer took place. The industrial structure of the cities was found to have an eVect on inequality, although the statistical analysis was not entirely satisfactory. Industries which were associated with high inequality included textiles and mining, which may reXect the larger scale of operation of these industries, but also printing, which was not organized on a large scale. The leather industry, which was relatively unmechanized and organized in small workshops, was associated with low levels of inequality.

8. Co u l d t h e R ise o f I n e q u a l i t y h a v e b e e n Av o id e d ? The results produced so far are in line with the conclusion reached by some modern cross-section studies: there is a conditional Kuznets Curve, which can emerge when the historical circumstances are right but which, under diVerent conditions, might be obscured by other factors. The question then is, what were the forces that produced the Wrst stage of the Kuznets Curve in nineteenth-century Prussia? Could the increase of inequality have been prevented? The Wrst point is that, even at the peak, inequality was not particularly high in Prussia by modern standards. Although income inequality in the urban sector was quite high, the overall Wgure was held down by a relatively egalitarian rural sector. What created the Kuznets Curve in Prussia was the shift from a rural society with a low Gini-coeYcient to an urban or industrial sector which had a level of inequality which was high by comparison with the rural sector though not so unusual when compared to other countries going through the process of industrialization. The second point is that, from the 1870s onwards, there was a shift in the terms of trade against the agricultural sector as a result of developments on world markets. This had the eVect of widening the gap between rural and urban incomes, and thus worsening the overall income distribution. Although this was not the only cause of the rise in inequality in the period, it contributed to the rise in the GinicoeYcient. This was due to the changing terms of trade on world markets. Only a massive rise in agricultural protection could have kept this from aVecting German agriculture. A third point is that, in the early part of the century, inequality was rising in the rural sector, due to a range of factors: demographic pressure, expansion into remote eastern areas, the eVects of the agrarian

Industrialization, Migration, Inequality


reforms, and changing social relations. This was only indirectly connected to industrialization. These factors provided a considerable reinforcement to the mechanism speciWed by Kuznets himself. It would be quite possible for a society which industrialized under diVerent circumstances to escape the Kuznets Curve. The Prussian example points to some factors which might help to achieve this: such as a relatively low level of demographic pressure, a prosperous agricultural sector in which incomes kept pace with those in urban areas, agrarian policies which promoted an equal distribution of income in this sector, and better credit provision for small farmers. These are the ‘good’ ways to escape a Kuznets Curve. But the comparison with Britain also reveals that there is a ‘bad’ way to escape it, and that is to start oV with such a high level of inequality in the rural sector that the transition to an industrial society does not push up overall inequality even though incomes in the urban sector are also unequally distributed. In short, the Kuznets Curve, while not just a peculiarity of German history, owes its origins in Prussia to some features of nineteenthcentury Prussia which may or may not be replicated in modern developing countries. It is the product of circumstances rather than an automatic consequence of industrialization. But there was not much that government action in nineteenth-century circumstances could have done about this. A period of rising inequality was probably something that just had to be endured.

Ap p e n d i x 9A. E s t i m a t e s o f i n c o m e i n e q u a l i t y f r o m Pr u s s i a n t a x s t a t i s t i c s Table 9A.1. Estimates of income inequality in Prussia 1822–46 1822






Gini-coeYcient 0.260 0.266 0.274 0.278 0.286 0.297 as indices (1846¼100) Gini-coeYcient 87.6 89.4 92.3 93.5 96.4 100.0 Upper 95% CI of 96.5 98.5 101.7 103.1 106.2 110.2 Gini-coeYcient Lower 95% CI of 78.7 80.3 82.9 84.0 86.6 89.8 Gini-coeYcient

Table 9A.2. Estimates of income inequality in Prussia 1853–73

0.345 100.0 111.7 88.3


0.348 100.9 112.7 89.0


1864 0.354




0.355 102.8 114.9 90.7

102.6 114.6 90.5

102.3 114.3 90.3

102.1 114.1 90.1



103.4 115.5 91.2

1873 0.363 105.2 117.5 92.8

Table 9A.3. Estimates of income inequality in Prussia 1873–5

U10/L40 Gini-coeYcient as indices with 1853¼100 Gini-coeYcient




1.44 0.315 105.2

1.51 0.325 108.5

1.47 0.318 106.2

Table 9A.4. Estimates of income inequality in Prussia 1875–91 1875










0.379 0.382 0.374 0.370 0.374 0.373 0.357 0.340 Gini-coeYcient 0.371 0.341 as indices (1853¼100) 120.5 119.5 117.9 118.0 116.7 116.8 117.7 112.5 107.1 Gini-coeYcient 107.6 126.1 123.6 125.1 123.5 123.3 122.2 122.3 Upper 95% CI of Gini-coeYcient 112.1 112.7 117.8 114.8 112.5 113.8 112.4 112.2 111.2 111.3 Lower 95% CI of Gini-coeYcient 102.1 102.5 107.2

Industrialization, Migration, Inequality

Gini-coeYcient as indices (1853¼100) Gini-coeYcient Upper 95% CI of Gini-coeYcient Lower 95% CI of Gini-coeYcient




Table 9A.5. Estimates of income inequality in Prussia 1892–1914 1892























0.318 0.329 0.315 0.319 0.334 0.316 0.333 0.338 0.335 0.331 0.340 0.332 Gini-coeYcient as indices (1853¼100) 122.0 117.0 117.9 117.3 123.5 118.2 124.0 126.0 123.1 125.5 124.3 Gini-coeYcient 122.6 126.8 121.6 122.6 122.0 128.4 130.4 122.9 128.9 131.1 128.0 129.3 Upper 95% CI 127.5 117.1 112.3 113.5 113.3 112.6 118.6 120.5 119.1 121.0 118.2 119.4 Lower 95% CI 117.7

Table 9A.6. Analysis of the estimated changes in each period

Source: ***Take in JJ From fo. 386***





þ14.2 þ16.2 þ12.2 þ21.4 þ4.3

þ5.2 þ6.0 þ4.3 þ20.2 þ3.5

þ12.5 þ13.3 þ11.6 þ13.6 þ5.7

þ5.7 þ6.0 þ5.5 þ8.3 1.3

3.2 3.1 3.4 5.4 2.7


Change in Gini-coeYcient (%) Upper 95% CI of change Lower 95% CI of change Upper 95% CI of highest variant Lower 95% CI of lowest variant


Industrialization, Migration, Inequality

0.334 0.333 0.327 0.331 0.324 0.315 0.318 0.323 0.324 Gini-coeYcient 0.324 0.321 as indices (1853¼100) 124.1 122.9 123.6 121.4 120.0 117.0 118.1 119.8 120.1 Gini-coeYcient 120.1 119.2 128.6 129.0 127.8 124.8 126.2 122.8 124.6 121.6 124.8 123.9 124.8 Upper 95% CI 118.7 119.1 118.0 116.5 115.3 112.4 113.4 115.1 115.3 Lower 95% CI 115.3 114.5


Challenging the Kehrite View of Imperial Germany 1. I n t r o d u c t i o n The view that the Kaiserreich suVered from internal Xaws is so deeply entrenched in German historiography that one doubts if it will ever be dispelled, however much evidence is produced against it.1 The list of Xaws has changed over time: concepts such as the lack of a middle-class identity, ‘organized capitalism’ as a distinctive economic system, ‘social imperialism’, the importance of the ‘failed revolution’ of 1848 have been given up or downgraded. The term Sonderweg is used less. But new lists are produced, which stress diVerent aspects of the social order, internal structure, political system, and cultural outlook of Imperial Germany: the importance of organized economic pressure groups, the impotence of parliamentary control, the appeal of Social Darwinism, the drift to extremist nationalism, and so on. This study has not attempted to confront these arguments directly; instead it has put forward an alternative view, taking many of its cues from the comparative study of developing countries in general, and from development economics in particular. The argument in its essence is that the Kaiserreich suVered from no more than the normal problems of a rapidly industrializing society. This is not to argue that there were no special features and characteristics of German society in the period. There were indeed unusual and unique attributes, as there are in all societies. But as Richard Evans has 1 Even historians specializing in foreign policy have accepted much of the ‘Xawed nation’ thesis: see Hildebrand (1995), 5–6 and 9. Hildebrand also accepts the interlocking of domestic and foreign policy, Hildebrand (1989), 77. The view that a more realistic foreign policy, involving an alliance with either Britain or Russia, was blocked for internal reasons is set out in Geiss (1976), 116, and Kehr is cited in support of this. Calleo (1978) provides a more sympathetic view of the ‘German problem’, and argues that German uniWcation should not be seen as ‘a malevolent accident which befell an otherwise harmonious European state system’, ibid. 3, but then goes on to accept the now discredited argument about the connection between ‘organized capitalism’ and German colonial expansion, put forward in Wehler (1969).

Challenging the Kehrite View


pointed out, commenting on the evolution of the Sonderweg thesis in the works of Hans-Ulrich Wehler, the list of Xaws which Wehler has put forward in recent work as an explanation of the particular course of German history contains nothing that cannot be found in some degree in other European states in the late nineteenth and early twentieth centuries.2 And, it might be asked, why is the comparison limited to Europe in this period? Industrialization is no longer a purely European phenomenon. It is an experience through which a number of non-European societies have passed or are passing. And in virtually every case it has been a disruptive and dislocating process. Evans also commented that the Sonderweg argument has become one which is increasingly concentrated on the failure to make political progress commensurate with the pace of social and economic transformation. This is a point which is widely accepted even by historians who are not associated with the Sonderweg view. It is one of the commonplaces of Wilhelmine historiography.3 But a broader comparison of societies, inside and outside Europe, going through the initial stages of industrialization reveals that few if any have made political progress during this period. Even in Britain, the earliest phase of industrial growth, usually dated 1780 to 1830, was not a period which saw much in the way of progressive democratic politics. In many other countries, there have been declines into civil war, authoritarianism, or military government. When this broader context is considered, the insistence that Imperial Germany should have made political progress commensurate with its economic advance can appear rather bizarre.4

2. So c i a l , E c o n o m i c , a n d P ol itical Trajectories in Imperial Germany In the history of the Kaiserreich the results of the 1912 election have a particular signiWcance: the Social Democratic Party gained 110 seats, 2

In a review of Wehler’s Deutsche Gesellschaftsgeschichte, reprinted in Evans (1997). Theodor Schieder was one example of a historian of a diVerent persuasion who made this point, Schieder (1960), 85. 4 The view that industrialization and progress towards democracy are linked is a component of ‘modernization’ theory, and this is often regarded as a typically European phenomenon; see Eley (1996) for comments on this from a German perspective. 3


Challenging the Kehrite View

reversing the losses made in the ‘Hottentot’ election of 1907, and exceeding the previous high-water mark of 81 seats achieved in 1903. It was now the largest party in the Reichstag; if it could gain the support of the left liberals and the parties of the national minorities (Poles, Danes, and Alsatians) it would form a ‘disaVected’ block of 185 out of a Reichstag of 397 deputies. This had a number of important consequences. On the right it produced a sense of marginalization, which was intensiWed when the government made concessions to the SPD, including the introduction of direct taxation at the Reich level, in order to secure the joint passage of the 1913 Finance Bill and the Army Bill of that year. Kehrite historians such as Dirk Stegmann have argued that the radicalization of the right, and the consequent abandonment of Sammlungspolitik, were important factors in the intensiWed foreign policy crisis which culminated in the events of July 1914.5 More broadly, the results of the election called into question the quasi-democratic system of constitutional checks and balances which maintained the Kaiserreich’s uneasy compromise between diVerent social and regional groups. A Reichstag which possessed legislative powers but which did not control the executive had been a successful solution to many of the problems faced in 1871: how to provide representation for others in an empire dominated by Prussia; how to secure local autonomy from encroachments by a central executive; how to reconcile previously dominant groups (the landed aristocracy, the urban Burgertum) to the experiment of universal male suVrage.6 But the continuation of this compromise required that all participants accepted the implied limitations, and this was particularly diYcult for the Social Democrats, because they were largely excluded from participation in the political process at the level of local and regional government. Any compromise which would have a reasonable prospect of securing Social Democratic support over the long term, rather than as a shortterm tactical expedient (as in 1913), would have had to have included reform at the local and regional level as well as the strengthening of 5 Stegmann (1970), 257–93; according to this, the election led to a renewed imperialist agitation, ibid. 277. The ‘shameless, chronic persecution’ of Bethmann-Hollweg by these groups had, it is argued, an eVect on his policy in July 1914, Chickering (1984), 289. 6 The Bismarckian constitution has attracted much criticism from historians: ‘monstrous’ and ‘irregular’ are two adjectives that have been used, Sauer (1968), 413. But this underestimates the problems it had to resolve. For an alternative view, see Rauh (1973), 47 V.

Challenging the Kehrite View


Reichstag authority.7 There are many historians who would argue that such a compromise was not acceptable to powerful interest groups, and that a military coup, or Staatsstreich, would have occurred had not the outbreak of war intervened to make this unnecessary.8 The war was welcomed by such groups as a means of halting the progress towards democracy. Thus, in the Kehrite scheme, the election has a special signiWcance: it is the point where sharpening internal divisions directly impinged on Germany’s strategic position, making the need for a solution to the problems of Germany’s diplomatic isolation into a pressing imperative. It also made it clear that any internal solution would have to involve major concessions to a party with a revolutionary tradition and regarded by its conservative opponents as disloyal.9 Against this, there is the more favourable picture presented by the economic trends discussed in the previous chapters. Combining these trends it is possible to construct, as a mental picture, a curve of the intensity of the socio-economic problems facing Imperial Germany, meaning by this both purely economic problems, and social problems having economic roots. The trajectory of this curve shows a rising trend to 1895–1900 and then declines. But this was not the only set of problems which beset the Kaiserreich. There were diVerent sets involving socio-cultural issues (dissatisfaction with or alienation from social and cultural structures and trends) and socio-political questions, deWning these to include purely political problems and social divisions which manifest themselves within the operation of the political system. Voting behaviour is one symptom of the latter, though the movement of opinion within parties is another aspect which is also important although less easy to measure. If the shares of the diVerent parties are plotted (Figure 10.1) this shows the steady gains made by the Social Democrats from the late 1880s onwards and the way that this squeezed 7 For a discussion of the tactical nature of the SPD’s support of the government in 1913, see Groh (1973), 434–6. The background to the vote is analysed in Stargardt (1994), 108–24. 8 The danger of a Staatsstreich was probably not that great, Pogge von Strandmann and Geist (1965). 9 The extent to which the SDP was ever a revolutionary party is debatable. Marxist ideas certainly inXuenced the leadership, but had less appeal at lower levels, Lidtke (1985). Its ‘disloyalty’ would seem to have been at least partly disproved by its vote for the 1913 Army Bill. Despite this, Epkenhans (1991) has argued that there was a fear that a political compromise would lead to a ‘storm of disarmament’, ibid. 413, and that this, together with the internal political situation, led to the Flucht nach Vorn of July 1914.


Challenging the Kehrite View



Minorities and others

Conservatives and other right 60 National liberals 40


Free conservatives

Lift liberals Centre

0 1871 1874 1877 1878 1881 1884 1887 1890 1893 1898 1903 1907 1912


Fig. 10.1. Percentage shares of different parties in Reichstag elections, 1871–1912

other parties. However, it would be wrong to conclude from this that there was a radicalization of internal politics in the Wilhelmine period. The internal politics of the Social Democratic Party have also to be considered. These were moving in a more centrist direction, and the party was making a determined eVort to broaden its support and move outside its natural constituency of the urban working class. This was rewarded by a gain in electoral support. The area of German politics which was shifting in a more radical direction was the conservative right. But this was losing ground. The best result for the right-wing parties came in the 1893 election, when the combined right-wing vote was 16.9%, but this had fallen to 9.6% by 1912. There was a fall in seats gained from 88 to 46. The shift to more radical policies was not beneWcial in electoral terms. In short, the pattern of electoral politics in Imperial Germany was not one of increasing polarization. Parties which adopted more moderate policies gained votes; parties which shifted towards the extremes lost votes. This was exactly as would be expected in a normal, stable,

Challenging the Kehrite View


democratic system. But the gains made by the SPD meant that the party had to be admitted into the Wilhelmine political establishment: the option of continuing to ignore it no longer existed. 3. T h e D o g t h a t D i d n ’t B a r k : Ec o n o m i c I s s u e s a n d t h e 1 9 1 2 E l e c t i o n The SDP’s ability to break out from its core constituency depended on its success in appealing to artisans, tradesmen, workers in small workshops in the urban areas, and in attracting votes in smaller towns which were not major industrial centres.10 For these voters, the importance of the economic issues stressed in earlier party policy, such as public ownership of major enterprises, was at best peripheral. It is worth considering quite how small the Social Democrats’ core constituency was. In 1910 the urban population, those living in major towns and cities (91 cities with at least 50,000 inhabitants), was 17 million out of a total German population of 64 million. A survey of employment in these cities showed a total of 1.6 million men (and 0.5 million women), aged 16 and above, employed in enterprises with at least 10 employees. Bearing in mind that some of these would have been too young to vote, and also that 29% of the urban population were Catholics, this suggests that the Social Democratic core, urban Protestant working-class men, was between 1.2 and 1.3 million voters. But the SDP vote was well above this level. It totalled 4.25 million in 1912. It was 1.78 million in 1893 (when the size of the core was considerably smaller). Figure 10.2 contains a comparison of the SDP share of the vote with the percentage of the German population living in urban areas. This illustrates the way that structural change in German society was favouring the party’s electoral advance. It also suggests that the party broke out of its core constituency relatively early, in the 1890s. In the early period the party’s share of the vote was roughly in line with the share of the German population living in cities of at least 100,000: the major industrial centres where large-scale enterprise was common. But from 1890 onwards the party’s share moves with the percentage of the population living in towns and cities with at least 20,000 inhabitants, many of which were not industrial centres at all. 10 Sperber (1997), 64, estimates that the party had 45% of the Protestant middle-class vote, and 13% of the Catholic middle-class vote in 1912.

Challenging the Kehrite View

336 40

SPD vote % in cities >20,000 30

% in cities >100,000



0 1871









Fig. 10.2. The share of the Social Democratic Party in Reichstag elections compared with the proportions of the German population living in urban areas, 1871–1912

The graph shows the deviations from trend caused by the anti-socialist laws of the 1880s, and it also shows the exceptional nature of the 1907 election: the 1912 election represented a recovery of the relationship previously established. It is also worth noting that the proportion of the German population living in towns with at least 20,000 inhabitants was rising by 6 percentage points each decade. If the SDP could hold onto the electoral share implied by this trend, its vote would rise proportionately—by over a million votes every ten years. The ability of the party to address broader social and political issues made it attractive to a swath of opinion outside the larger cities. It was also helped by the ineVectiveness of other opposition parties: voters who wanted signiWcant change in the political structure of Imperial Germany had no other party to turn to. The statistical analysis of election results can elucidate some of these points, though it is necessary to stress that statistical tests can only be run where there are questions involving measurable issues. It is possible to look at questions such as ‘Did high unemployment in particular constituencies cause a rise in the socialist vote?’; others, such as ‘Did the desire for constitutional change lead to an increase in the SDP

Challenging the Kehrite View


vote?’, are not so easily answered. Moreover, statistical analysis at the aggregate level does not provide direct evidence on voter motivation; this would require surveys of individual voter behaviour, attitudes, and social background. It shows which factors are associated with higher or lower probabilities of a vote for a given party. It tests the inXuence of the general social and economic environment on voting behaviour.11 The data set prepared for this analysis was split in two. The Wrst included the major cities and the second looked at results in Prussia outside these urban areas. This enabled statistical sources giving information on economic conditions in the urban areas to be used in the Wrst data set, and sources on agricultural structure to be included in the second. Table 10.1 gives a summary of the statistical analysis of voting behaviour in 70 large towns and cities (the full results are given in Table 10B.1 in Appendix 10B). The analysis Wrst demonstrated the importance of city size: the SDP share was higher in the larger cities. The second important factor was religious aYliation: Catholics were much less likely to vote for the SDP (the proportion of non-Catholics voting for the SDP was 2–21⁄2 times that for Catholics).12 Table 10.2 shows the implications of these results for cities of diVerent sizes.13 The importance of city size is clearly demonstrated. The growth of cities tended to favour the SPD. Migration patterns which caused cities to expand were therefore a factor in the party’s advance. Catholicism remained an obstacle even in the larger cities. However, when Prussian cities are separated from cities outside Prussia an important diVerence is revealed: the eVect of Catholicism was considerably stronger inside Prussia. In Prussian cities the Catholic population was much less likely to vote SPD. In a city of 500,000, non-Catholics would be three times as likely to vote SPD as Catholics. But outside Prussia 11 To give an example of this: only a survey of individual voters could show whether unemployed workers were more likely to vote for the SPD. A statistical association between high unemployment in a given constituency and a high SPD vote might show this, but it might also show that employed workers in a constituency with high unemployment were more likely to vote SPD. The two eVects cannot be distinguished. All that can be shown is a general result that unemployment inXuences voting behaviour. 12 The SPD was gaining ground with the Catholic working class, though its share was only 22% against 54% for Protestant workers, according to estimates by Sperber (1997), 37–8. Contrary to the view put forward by Schma¨deke (1995), 183–91, Catholicism was not an insuperable obstacle to the SPD. Schma¨deke’s conclusion, that the SPD’s rise was coming to an end, does not appear to be justiWed by the evidence. 13 The coeYcients used are those given in column a of Appendix Table 10B.1.


Challenging the Kehrite View

Table 10.1. Summary of results of statistical analysis of the 1912 SPD vote in 70 major German cities I. Secure statistical results Factors which were associated with a higher SPD vote: Size of the city (larger cities had higher SPD shares) A large SDP vote in 1878 Factors which were associated with a lower SPD vote: Catholicism Catholicism within Prussia (additional eVect) Factors which had little or no inXuence on voting behaviour: Levels of inequality (measured by the U10/L40 ratio) Local labour market conditions (the ratio of applicants to vacancies) II. Less secure statistical results Factors which were associated with a higher SPD vote: High wages High mobility Large-scale industry (large factory size)

Table 10.2. Predicted SPD share of the vote for cities of different sizes and different religious affiliations Number of inhabitants in the city

City population is: 80% Catholic 20% Catholic





20.5 37.8

24.6 43.4

35.7 56.7

41.1 62.2

this was less important. In a similar-sized non-Prussian city the ratio between non-Catholics and Catholics would only be 1.3–1.14 It might be thought that this was a reXection of the importance of the Polish-speaking Catholic population in Prussian cities. But inspection 14 This conWrms a point made by historians who have looked at regional election results. The SPD was a strong contender for the Catholic working-class vote in nonPrussian areas. Blackbourn (1980), 193, comments on the strength of the party in working-class Catholic areas in Baden and Wu¨rttemberg. The SDP had 28.3% of the vote in Baden in 1912 (Baden was 59.3% Catholic) but only 9.8% in Regierungsbezirk Koblenz (65.5% Catholic) and 4.6% in Regierungsbezirk Trier (79.3% Catholic). Figures from Ritter (1980).

Challenging the Kehrite View


of the data reveals that this was not the case. The socialist vote was also unusually low in Rhineland cities like Aachen and Trier which had few Poles. Rather it seems to reXect the strength of Rhineland Catholicism, the social cohesion of a minority in a predominantly Protestant society.15 The after-eVects of the Kulturkampf almost certainly played a part as well. There was a strong element of persistence in the socialist vote: the cities which had a relatively large vote in 1877 tended also to have high votes in 1912. Table 10.3 gives the Wgures for the two elections for the ten cities with the highest votes in 1912. The SPD share was already above 40% in six out of the ten cities. Much of the growth in the party’s overall share came from the spread of socialist ideas from these core cities, initially to other large cities, and then to smaller towns and other areas.16 Table 10.3. Cities with the highest SPD vote in 1912; comparison with 1877 percentage share of votes cast (first-round votes)

Berlin Chemnitz Altona Hamburg Nuremberg Zwickau Charlottenburg Mu¨lhausen Brunswick Solingen

SPD vote 1912

SPD vote 1877

75.0 64.3 63.4 61.2 60.7 60.6 58.7 56.1 55.6 55.6

39.2 54.9 51.0 40.0 48.3 61.4 16.2  46.1 35.8

15 Another possible reason, suggested by Rohe (1982), 236, is that in Prussian areas such as the Ruhr there was a confessional aspect to employment relations: typically, Catholic workers were employed by Protestant entrepreneurs. This meant that, in these areas, Catholicism developed a ‘social’ dimension which was less evident in non-Prussian Catholic regions, where workers and employers were both likely to be Catholics. See also Evans (1997) and Sperber (1983) for the consequences of the Kulturkampf. 16 Hamburg and Saxony were strongly represented at the original meeting of the Allgemeiner Deutscher Arbeiterverein in 1863, Berger (2000), 49. See also the maps of SDP support in Schma¨deke (1995) which show the strength of the party in areas around Saxony, particularly Thuringia and Franconia. There was a geographical element to the diVusion of SPD support.


Challenging the Kehrite View

The analysis also showed that there was only a limited eVect on the socialist vote from economic conditions within the cities. There was some evidence that the socialist vote was higher in the more prosperous cities (where wages were higher), and in cities with large-scale industry, measured by factory size, though the connections were not strong ones. This in turn reXected the success of the party in widening its appeal beyond the conWnes of its core constituency: workers in large-scale industry.17 The presence of this core constituency is just discernible in the results, but the eVect is only a slight one. Another weak Wnding (in statistical terms) is the existence of a relationship between high mobility, the total number of moves in and out of each city in 1910–12 relative to population, and the size of the SPD vote. The socialist vote rose in cities with more Xuid labour markets, but the result is not a strong or a secure one. Two factors which were shown to have virtually no eVect on voting behaviour were labour market conditions and inequality. The Wrst was measured by the ratio of applicants to vacancies at town labour exchanges. Where economic conditions were adverse, the ratio of applicants to vacancies would be high. This was not a factor which had any discernible impact on the SPD vote. Inequality was measured by the ratio between the estimated share of the top 10% of households in total income and the estimated share of the bottom 40%. In the more unequal towns, the top 10% had a share which was Wve or six times as large as that of the bottom 40%; in the more egalitarian towns the shares were about equal. But these diVerences had no signiWcant eVect on votes cast in the 1912 election.18 The analysis of the vote in rural or small-town Prussian constituencies is summarized in Table 10.4 (full results are in Table 10B.2). The SPD share in these constituencies averaged 20.6%, which was rather less than in the towns analysed earlier (42.2% on average). However, the SPD advance in these small town and rural constituencies was an important component of the party’s overall gains. The average share had risen from 16.3% in 1907. 17 The SPD no longer represented a ‘homogeneous group of proletarians’, Fairbairn (1997), 216. 18 The methods used to produce these estimates of the income distributions in the Prussian Kreise are described in Grant (2002b). Eley (2002), 56–7, comments on the lack of a relationship between economic factors and the SPD vote in a comparison of Remscheid and Hamborn: Hamborn workers were the ‘very epitome of the brutalised and exploited factory proletariat’ but the SPD was weaker there than in Remscheid.

Challenging the Kehrite View


Table 10.4. Summary of results of statistical analysis of the 1912 SPD vote in 204 constituencies outside the major cities (Prussian constituencies only) Secure statistical results I. Factors which were associated with a higher SPD vote: A high level of non-agricultural employment Large numbers of Gutsbezirke relative to other Gemeinden A high Lithuanian-speaking population II. Factors which were associated with a lower SPD vote: Catholicism A high Polish-speaking population Presence of large farms and estates (over 100 hectares) Remoteness from major cities (over 100,000 inhabitants) and from towns (over 20,000 inhabitants) III. Factors which had little or no influence on voting behaviour: Levels of inequality (measured by the U10/L40 ratio) Wage rates

The analysis shows that there were two factors which explained much of the variation in the SPD vote: the percentage of the population occupied outside agriculture and the Catholic percentage. The agricultural population was less likely to vote for the SPD, and the relative decline of agriculture was a factor working in the SPD’s favour in these rural constituencies.19 Catholicism was an important factor working against the SPD especially in these Prussian areas. In the Polish-speaking regions the SPD vote was held down even further, although it was higher in constituencies where Lithuanians were present. The SDP vote was highest in areas near to the larger cities (with over 100,000 inhabitants) but it was also high in areas near to towns with a population of at least 20,000. But the party’s share was reduced in the 19 This result is also found in the analysis by Fairbairn (1997), 228. Fairbairn also shows that the negative eVect from agricultural employment became stronger between 1890 and 1912, ibid. 230. As pointed out by Steinbach, electoral research in German history is ‘relatively immature’ and the type of multi-variable statistical analysis of electoral results presented in this chapter is rare for pre-1914 Germany, Steinbach (1992), 121. Fairbairn’s analysis is an exception. Suval (1985), 122, produces correlations which are interesting but does not use multivariate analysis. Nor does Schma¨deke (1995). Winkler (1993), 251 does not give t-statistics, so there is no way of assessing the quality of the results shown.


Challenging the Kehrite View

more remote constituencies, and this was an eVect which was independent of the occupational structure of these constituencies. Towns and large cities were the centres of socialist activity, and this had an eVect on nearby areas. Economic variables had little eVect apart from the occupational structure. Neither wage levels nor inequality (as measured by income inequality) had any eVect on voting behaviour. This repeats the Wnding reported earlier from the analysis of votes in urban constituencies: economic issues do not appear to have been important factors in the 1912 election. Inequality in land ownership, however, as measured by the percentage of the agricultural area which was in large farms and estates (over 100 hectares), was a factor working against the SPD. Traditions of deference may have kept the socialist vote down, or fears of retaliation if the secrecy of the ballot was not thought to be absolutely guaranteed.20 However, a factor which operated in the opposite direction was the presence of Gutsbezirke. The SPD vote was higher in constituencies where a large percentage of the local Gemeinden were classiWed as Gutsbezirke: aristocratic estates which had in the past had an important administrative function. The other categories were Landgemeinden and Stadtgemeinden. Around 25% of the administrative units in these constituencies were Gutsbezirke, though this exaggerates their importance as they were usually much smaller than the Landgemeinden. In the past the aristocratic owners of the Gu¨ter had had considerable control over the local police and administration, and judicial functions of their own. These were largely abolished by the Kreisordnung of 1872.21 What remained was an important position in local society, and some Wscal privileges, notably an exemption from the supplement charged on the Prussian income tax assessment to Wnance the expenditure of the Landgemeinden and Stadtgemeinden. Against this, the owners of the Gu¨ter had to provide some services which would otherwise have been paid out of local government funds.22 20 Anderson (2000), 157–9, argues that the secrecy of the ballot was not assured in some rural areas. 21 The judicial role of the Rittergu¨ter was removed in 1849–51; the police functions were abolished in 1872, von Below (1891), 675 V. However, the Rittergu¨ter remained separate administrative districts until 1927, Puhle (1986), 85, and some historians have argued that much of their authority and inXuence survived the 1872 Kreisordnung, Carsten (1989), 120. 22 The supplement could be a substantial one, an increase of 50% or more in 663 Gemeinden according to Meisel (1911), 298. The aristocratic owners of the Gu¨ter also beneWted from favourable tax treatment from the Landra¨te (district administrators), according to Witt (1985). This would have been a further cause of resentment.

Challenging the Kehrite View


Not all large farms and estates were Gu¨ter: there were many estates that had been sold by their original owners or that had never the functions of a Gut. And not all Gu¨ter were large agricultural holdings: some estates had been broken up or rented out while still retaining the title and privileges of a Gut. In places the system was a relic with little practical relevance. What these results indicate is that this system still caused resentment. In general, economic factors (income inequality, wage levels, unemployment, and enterprise size) did not have much eVect on the SPD vote. The party was appealing to a broader constituency, concerned with other issues. But political inequality, the sense of being excluded from decision-making, whether at the local or the national level, did matter. And this is the main signiWcance of the result with the Gutsbezirke variable.23 It is not the case that the Gutsbezirke were a particularly onerous part of the political and administrative system of Prussia in 1912. There were other more important concerns for electors willing to vote for radical change. Their presence happens to be something which can be measured, and which varied considerably between diVerent constituencies. Other issues, such as the three-class electoral system, aVected all constituencies, so there is no variation in the importance of the variable which can be picked up by statistical analysis. These results indicate that there was a desire for change in the political and administrative system, and that this desire for change was an important factor in the SDP vote. Voters who wanted progress towards democracy turned to the SDP as the party most likely to achieve this.24 The SPD responded to this new role. Its policy statements and leaXets emphasized constitutional issues, and it signalled an increased willingness to play a constructive role in the political process.25 Moreover, and more dramatically, there was a shift in the social make-up of the 23 The question of political ‘fairness’, the sense of being excluded by the suVrage system, was one of the central issues of Wilhelmine politics, Fairbairn (1992). 24 Another way to protest against the three-class electoral system was to abstain in local elections. The turnout in Landtag elections could fall as low as 8% for electors in the third class, Anderson (2000), 196. In some rural areas the left-liberals also beneWted from dissatisfaction with the electoral system, Thompson (2000), 301. The electoral system also kept many German city administrations in liberal hands, despite SPD majorities, and it may be surmised that resentment of this was a factor in the strength of the SPD vote in urban areas, Pogge von Strandmann (1992). 25 Bertram (1964), 169; this trend towards an emphasis on political objectives was in part a reaction to the SDP’s setback in the 1907 election: Crothers (1941), 213 and 229; Schorske (1955), 225–6.


Challenging the Kehrite View

SDP’s Reichstag candidates. In 1907 the party had Welded 132 workingclass candidates (Arbeiter, Handwerker, and Angestellte); by 1912 this had fallen to 57. By contrast, the numbers of private sector managers and oYcials (Privatbeamte) had risen from 50 to 175.26 Table 10.5 gives the social breakdown of the SPD candidates. Private sector managers and businessmen constituted a majority of the candidates oVered in 1912. The other striking feature of the table is the under-representation ¨ Ventliche Beamte, of anyone in public sector employment: oYcials (O Justizbeamte), army oYcers, and teachers. These were just 1.5% of Table 10.5. Social origins of the SDP candidates, 1907 and 1912 elections, compared to those of other parties (as percentages)

Entrepreneurs, businessmen: Managers, officials:

Religious: Professional:

Agriculture Commerce and industry Public sector Legal Private sector Protestants Catholics Army officers Lawyers Doctors Writers Teachers Workers

Employees: Others: Total number of candidates

SPD candidates

Candidates of other parties





0.8 16.6 3.0 0.0 12.6 0.3 0.0 0.0 2.8 0.5 23.1 0.3 33.2 7.0 398

1.5 10.0 1.5 0.0 43.9 0.0 0.0 0.0 3.5 0.5 23.3 0.0 14.3 1.5 399

21.4 11.1 10.9 12.3 1.6 1.4 4.7 1.2 7.1 2.3 9.5 6.3 5.3 4.8 1285

19.5 10.4 9.1 8.0 4.9 3.3 3.6 1.8 10.0 2.7 9.5 9.0 5.6 2.6 1153

Source: Calculated from figures in Bertram (1964), 158.

26 The size of the shift raises the question whether it represented a genuine change, or whether candidates previously identiWed as Angestellte or even Arbeiter were now classifying themselves as Privatbeamte. If so, it still represents a shift in the party’s approach: candidates were stressing their middle-class aYliations instead of seeking to appeal to a narrow working-class constituency.

Challenging the Kehrite View


SPD candidates in 1912, but 28.6% of those put forward by other parties.27 The composition of the SPD’s candidates reXected the shift in its policy orientations: it represented the politically excluded middle classes as well as the working class. The statistical analysis presented in this section conWrms that economic issues, which would have been major concerns of working-class electors, played a relatively unimportant role in the explanation of the size of the SDP vote. The party was appealing to a broader constituency: those who wanted political change as well as those who wanted radical alterations in the structure of society.

4. E c o n o m i c s a n d t h e T i r p i t z Pl a n : T he Fo r c e s D r i v i n g G e r m a n y t o w a r d s W a r The weakness of the political structures of the Bismarckian Reich had left voters who wanted political change with little alternative but to vote for a party which, in its original policy statements, had stood for a farreaching programme of revolutionary social and economic change. The rapprochement between the aims of the SDP leadership and those of its newer and wider base was a gradual one, as the party shifted its emphasis from economic issues to the pursuit of political reform, and began to indicate its willingness to work within the system. However, the growing strength of the party and the desire for change which it represented presented a serious problem for the Bismarckian system. Could the system evolve in a democratic direction without a violent upheaval? It is an open question whether this could have been accomplished if the only issues facing pre-1914 Germany had been internal ones. Could concessions have been made which would have strengthened parliamentary control without conceding full majority rule? How would the strengthened Reichstag authority have been regarded by the nonPrussian states? How far were the conservative forces opposed to further democratization prepared to go to obstruct such a settlement? Would the radicals within the SPD have accepted a compromise? But this is all speculation. For the issues of democratic control and accountability which were raised by the trajectory of electoral results 27 It is likely that public sector employees would have been reluctant candidates, fearing discrimination against anyone who put themselves forward as candidates for an ‘unpatriotic’ party.


Challenging the Kehrite View

meant that the Kaiserreich confronted the problem of constitutional change at just the same time as the trajectory of Germany’s diplomatic isolation and weakening strategic position caused a series of crises in external politics, one of which ended in the outbreak of the First World War. If there was ever a chance of achieving a normal transition to a democratic settlement, this disappeared in July 1914. To consider the background to the deteriorating external position it is worth returning to one of the arguments put forward in Chapter 8, when explaining why economic developments had created the conditions for political stabilization post-1900: the favourable prospects for German trade. Germany was able to use its comparative advantage in the production of industrial goods, particularly in the ‘high technology’ sectors, to gain market share in world trade, and it could then use these revenues to import an increasing volume of food and raw materials.28 This was a development which had profound strategic implications, and it meant that relations with Britain, previously a peripheral issue in German diplomacy, were now of central importance. In the 1860s Bismarck was able to largely ignore British positions on issues such as the annexation of Schleswig-Holstein and the progress towards German uniWcation: little more than lip service was paid to British concerns, British oVers of mediation were treated with barely disguised condescension, and substantive British proposals were mostly ignored. But this reXected the realities of the position: Britain was in no position to intervene eVectively in a land war in continental Europe; Prussia was a largely self-suYcient economy, whose ability to wage war would not be aVected by a British blockade and which had few nonEuropean interests vulnerable to British sea power.29 To have maintained a position of self-suYciency and strategic independence would have required policies which supported a higher level of agricultural production than that achieved under the Kaiserreich. TariVs would have had to be high enough to enable the grain sector to match production costs in the American Mid-West. Resources would have had to have been maintained in agricultural production, not switched to industry, requiring a substantial reduction in the rate of rural–urban migration. To achieve this, agricultural prices and incomes 28 The shift in the German position with regard to cereals was given in Table 7.1. There were substantial increases in the importation of other foodstuVs and raw materials. 29 Bismarck’s remark that Britain was a quantite´ ne´gligeable in European politics was quite justiWed from a Prussian point of view as long as Prussia remained solely a Continental power, Kennedy (1980), 191.

Challenging the Kehrite View


would have had to be increased relative to urban levels. German farmers would have beneWted at the expense of the urban consumer. By choosing a moderate level of protection, which did not prevent a substantial rise in imports, Imperial Germany adopted a policy which provided some relief to agriculture but which did not prevent an increase in living standards for the urban working class. Improved incomes for workers contributed to the post-1896 fall in the urban Gini-coeYcient. This in turn created economic conditions favourable to the normalization of internal politics. It raised the likelihood that the Social Democratic Party could be persuaded to turn away from revolutionary politics and become an accepted part of the political system. Germany faced a genuine socio-strategic dilemma. Economic selfsuYciency could be maintained only by the pursuit of policies which were clearly hostile to the interests of the urban working class and which would have perpetuated the alienation of that class. By contrast, rising urban living standards could be secured only by recourse to imported foodstuVs and raw materials. To a considerable extent these came from British possessions, or from areas dominated by British capital, and the overwhelming proportion crossed seas controlled by the Royal Navy.30 Even in the event of a war with France and Russia, from which Britain stood apart, the state of the German navy around 1900 was a serious strategic deWciency. In 1903 Germany was outnumbered by France alone. Germany had 9–12 ships of the line; France had 21. It would only be possible to have a local superiority over the French Atlantic Xeet if the French Mediterranean Xeet were prevented from moving north and if the Russian Baltic Fleet did not join with the French.31 To see Germany’s Flottenpolitik solely in terms of its eVect on internal politics, as historians in the Kehrite school tend to do, to see it as a part of a Sammlungspolitik, a cause for the rallying of forces loyal to the Empire, is to ignore or obscure this genuine dilemma.32 The economic facts are clear: Germany in the late 1890s was not in the same 30

That there were particular problems associated with Germany’s geographical position is a point made in Stu¨rmer (1990), but this study does not deal with the interaction of geography and the evolution of foreign trade. 31 Bu¨chel memorandum, 27 Mar. 1903, Lambi (1984), 203. 32 There is a way in which some Kehrite historians have linked the Flottenpolitik to trade issues and that is to argue that German industry needed colonial expansion to gain new markets, which in turn required a powerful Xeet, Wehler (1969), 112–26.


Challenging the Kehrite View

position as Prussia in the 1860s. Ignoring Britain was no longer an option. Bismarckian condescension was not an appropriate policy. The trade routes which brought imported food and colonial products to Germany crossed those which brought the same goods to Britain. Only one power could control these seas. Germany would have to confront or conciliate. In the Kehrite literature, the existence of a genuine strategic dilemma is either downplayed or ignored altogether. Economic factors are not considered as a part of the geo-political equation. Yet the German government was certainly aware of the importance of foreign trade to the German economy. Hohenlohe wrote in a letter in 1897: ‘Germany’s trade has grown to such an extent that the government is obliged to protect it. This can only be done by a Xeet which is more than just a coastal protection force, but one capable of ensuring the safe passage of our imports.’33 And the Kaiser wrote ‘richtig’ in the margin of a government report against the following passage: ‘without her overseas trade Germany could not exist, and this trade would not prosper for long without the protection of a Xeet capable of commanding respect’.34 Tirpitz himself was well aware of the economic implications of the plan. He saw an intimate connection between the expansion of the Xeet and the growth of German trade. In his public statements he tended to play down the trade implications in order to avoid antagonizing the agrarian lobby, who would have preferred Germany to remain selfsuYcient in food, rather than exporting industrial products to pay for food imports. But there was an awareness of the strategic consequences of the changing structure of German imports.35 Critics of the Tirpitz naval programme need to explain how Germany could have resolved this dilemma. What was the alternative to building a navy? A British alliance was one possibility, and this was considered at various times by Germany’s leaders. Yet this had the clear implication that Germany could be drawn into conXict with other European powers (most probably Russia) as a result of disputes arising from Britain’s imperial interests and possessions. The British Empire would 33

Letter from Hohenlohe to So¨ldendorV, 7 Nov. 1897; quoted in Berghahn (1971),

133. 34 The document was a report from the Prussian representative in Bavaria, Berghahn (1971a), 134. In 1899 Tirpitz predicted that, without a Xeet, Germany would sink back to the status of a poor agricultural country, Herwig (1987), 35. 35 Deist (1976), 110–12. Trade issues were, however, used to gain the support of inXuential lobby groups and industrial organizations, Pogge von Strandmann (1972).

Challenging the Kehrite View


be defended ‘on the banks of the Elbe’, at immense cost to Germany.36 To have eschewed naval expansion in the absence of a formal British alliance would have meant that Germany would have little choice but to accept British mediation in the event of a prolonged European war. Britain would be the arbitrator of Europe, with unpredictable consequences. In addition, German trade would have been unprotected from attack by the French or Russian navies should such a war involve these powers.37 Eventually, after 1945, West Germany was able to reap the rewards, in terms of rising living standards and political stabilization, of a tradeorientated economic policy. But this was done in the context of an American alliance which did not just provide security to West German trade routes but also oVered an eVective guarantee of land frontiers as well. A pre-1914 British alliance oVered only the Wrst part of this: trade security at the expense of a more dangerous position on land. To recognize this dilemma is not to deny that internal weaknesses had an adverse eVect on German foreign policy. A more far-sighted foreign policy, more Xexible and less inXuenced by domestic political considerations, might have kept potential opponents isolated, might have prevented the formation of a power bloc hostile to Germany, might have pursued a policy of naval expansion less obviously driven by hostility to British interests, and might, in short, have avoided the drift towards war on terms which were unfavourable to Germany. The contrast with British policy is instructive. Once German policy had shifted to one of evident confrontation, the British response was indeed far-sighted and Xexible. Outstanding disputes with France, Russia, and the United States were settled, even at the cost of giving up longheld positions (conceding a Russian sphere of inXuence in Persia, recognizing the Monroe doctrine). An empire built on an implicit sense of racial superiority concluded an alliance with Japan. German policy, by contrast, was hesitant and needlessly antagonistic. Internal weaknesses had a part in this. But Germany was not the only country pre-1914 whose foreign policy was adversely aVected by internal weakness. 36

As the elder Pitt had claimed to have won Canada in 1759, by distracting France in Germany. 37 The Kehrite historians’ concentration on the Tirpitz plan has obscured the fact that much of German naval planning was concerned with France and Russia—for example, the 1899 plan, Lambi (1984), 193. Any navy large enough to match the combined strength of France and Russia, and therefore give Germany the option of planning for a prolonged war against these two countries, was also large enough to be a threat to Britain.


Challenging the Kehrite View

Russia and Austria-Hungary are two obvious examples. It was the quality of the British policy response to Germany’s challenge which was truly exceptional.38 Kehrite historians have ignored the economic dimension of the Tirpitz plan, and concentrated on the internal political aspects. The result has been to underestimate the genuine dilemma faced by German policy-makers, and to present the plan as a cynical attempt by an outmoded elite to retain power and obstruct political reform. Even if such motives did play a part, the success of the naval programme in attracting widespread support owed a lot to the economic imperatives which lay behind it.

5. G r o wt h i n M a t u r e a n d i n I n d u s t r i a l i z i n g S o c i et i e s : T h e Ec o n o m ic Pr e c o n d i t i o n s f o r P o l i t i c a l Pr o g r e s s Economic growth in mature industrial societies diVers from that in industrializing societies in that the main engine of growth is improvements in productivity within each sector not the transfer of labour and other resources from sectors where productivity is low into more modern sectors. In mature economies the proportion of the labour force in agriculture and other lagging sectors (if agriculture is still lagging) is typically a low one, below 10%. Further transfers from this sector can make only a modest contribution to overall growth. In the early stages of industrialization, structural shifts are an important component of overall growth. Indeed, productivity growth within each sector is often quite modest, so that the contribution of structural change is a vital one. Thus a high rate of migration out of agriculture into the urban or industrial sector is intrinsic to the growth process, as well as a major force of social transformation. The Lewis Model is valuable in that it provides an explanation for some features of the operation of urban labour markets in industrializing countries, and may help to explain the high level of urban inequality found in Germany and other countries. But this is only part of the broader process of social change brought about by rural–urban 38

The contempt with which German foreign policy is treated by German historians can come as a surprise: was German policy really that of a ‘deluded upstart’, Hildebrand (1995), 10.

Challenging the Kehrite View


migration: a rural society bound by traditional ties of obligation, deference, and reciprocity is cast aside and a new urban structure, built on very diVerent principles, has to be constructed in its place. Inevitably, this takes time. Initially, the sense of dislocation produced by migration is intensiWed by the contrast between the stability of rural institutions and values compared to those of an urban society which is in a state of Xux. As the growth process continues, the contribution made by structural change will tend to fall, unless exceptional factors intrude (such as the check to structural change experienced by Germany and many other European countries in 1913–50, which left room for a rapid catch-up in 1950–73). As the rural labour surplus is used up, the opportunities for rapid growth as a result of the redeployment of labour which was previously underemployed will be reduced. Migration in and out of urban labour markets will become a stabilizing rather than a destabilizing process. The analysis of the previous chapters has concentrated on the social and economic consequences of the ‘labour surplus’ phase of industrialization. It has looked at the eVects of migration out of agriculture into urban labour markets, it has considered the movement of inequality in the rural and urban sectors, and it has examined the eVects of change and transformation in agriculture. Standing back from these trends, it is possible to discern the broad movements: a pattern of sharpening tensions and divisions up to the 1890s followed by a period of stabilization and improvement. The elements of this better position post-1895 are the following: 1. An end to the ‘labour surplus’ pattern of migration dominated by the transfer of surplus population out of agriculture. The sharp decline in emigration to the United States after 1895 is a further indication of the changing nature of migrant behaviour. 2. The stabilization and slight decline in the overall Prussian GinicoeYcient after 1900, combined with the fact that inequality within the rural and urban sectors was declining from 1896 onwards. 3. The improved position of German agriculture after 1895, as indicated by rising land prices, the reduction in forced sales, and increases in rural wage levels. This in turn reduced the pressure on urban labour markets. 4. The recovery of the German capital markets from the crisis of the 1870s was complete by 1900; the robustness of the new system


Challenging the Kehrite View

was demonstrated by the fact that investment levels remained high despite short-term Wnancial crises in the post-1900 period. The urban sector was better able to Wnance the creation of jobs for rural migrants. 5. The strong growth of German exports, particularly in ‘high tech’ sectors, promised well for the prospects of the German economy in the twentieth century. So long as world trade remained open, and volumes increased, Germany could expect a steady improvement in its share of world exports, making it possible to Wnance an increased bill for imported raw materials and foodstuVs, and to deliver continued improvements in living standards for German citizens. From the mid-1890s onwards the German economy developed in directions which created the preconditions for social and political stabilization.39 These favourable preconditions were not enough, in themselves, to guarantee the successful transformation of the Kaiserreich into a modern democracy. Even if there had been no outbreak of war in 1914 there was still much that could have gone wrong. But circumstances were more favourable in the early twentieth century than they had been in the 1870s and 1880s. In the end, though, the weakness of the Kehrite argument lies in the normality of the pre-1914 German experience. Germany went through a process of rapid industrialization, and this brought with it economic and social problems common to many countries going through this phase. The tools needed to understand this are available in the comparative study of industrializing societies. There was no Sonderweg before 1914. The outbreak of war in 1914 had other roots, other causes.

6. J ul y 1914: A D e r a i l m e n t o r t h e E n d o f a S a c k g a s s e ? The idea of this study was conceived during a visit to China in 1995. There was an evident labour surplus in the countryside; the migration of this surplus to already crowded cities was clearly going to have serious social, economic, and political consequences. This observation 39 This view contrasts with the one put forward by some historians that industrialization inevitably led to new social conXicts, Puhle (1976), 340. For GeoV Eley, the ‘central contradiction of Wilhelmine politics’ was failure of the state to adapt to the new task of social reconciliation caused by the basic ‘antagonism of capital and labour’, Eley (1980), 349. The eVect of economic and social change has also been blamed for the readiness of the middle classes to embrace radical nationalist ideas, Chickering (1984), 302.

Challenging the Kehrite View


prompted a general reXection. If there is a labour surplus period in economic development—a ‘hump’ which many countries have to surmount—does this aVect the stability of countries going through this process? Are they potentially dangerous factors in world politics? Is this an explanation of the political problems of pre-1914 Germany? Did these in turn aVect German foreign policy? Will a similar process make China a disruptive force in world politics in the twenty-Wrst century? The conclusions are in many ways more optimistic than the fears which were prompted by this visit. The autonomy of the foreign policy problems and the internal problems of the German empire is the main conclusion to emerge. While this is not a conclusion which necessarily applies to every state in every era, it is at least more comforting than the implication of the Kehrite view: that a state which lacks the ability to tackle its internal problems will turn to an aggressive foreign policy in an attempt to defuse domestic tensions. The picture given in this study is one of a society which, by 1914, was having considerable success in dealing with the problems of rapid industrialization and the ‘labour surplus’ period. These were problems which were unprecedented in their scale and intensity. Germany 1870– 1913 was the Wrst ‘developing economy’, in the sense that it was the Wrst to go through a process of telescoped industrialization and fast economic growth characteristic of countries ‘catching up’ with a technological leader. The speed of German industrialization and urbanization was rapid, perhaps three times as fast as Britain at a similar stage. If there is a connection between the economic and political problems of Imperial Germany, it probably lies in an earlier period, before 1895. Evidence presented in this study shows that the time of maximum rural–urban migration was also one of intense urban problems: high inequality and low life expectancy. It was during this period that Marxist ideas became inXuential in working-class districts. This development, and the violence of the reaction by the Bismarckian state which it provoked, had a profound eVect on the political system of Imperial Germany. It is in this period that the ‘Weber–Lewis–Kuznets’ hypothesis has most value for those seeking to explain the course of German history. But by 1914 the chances of democratic transformation were much improved. There remained massive obstacles. For many historians, these were so great as to be insuperable.40 Consequently, it can be 40

It was a situation of ‘intolerable complexity’, according to Eley (1980), 350.

Challenging the Kehrite View


argued that the rising demand for change inevitably created tensions which, even if these were not directly linked to German policy during July 1914, produced a background which aVected the decision-makers’ ability to resolve the crisis. The atmosphere of pessimistic fatalism, the general weariness with the political process and the compromises it required to keep operating, a willingness to take risks and consider radical solutions—these were dangerous sentiments in a context which required clear thought and an ability to appreciate the consequences of a mistaken policy.41 The view that the Kaiserreich was in a political cul-de-sac, a Sackgasse, where the options open to the executive were narrowing, is one which is held by many historians outside the Kehrite school.42 The arguments of Manfred Rauh, that a process of ‘quiet Parliamentarization’ was already under way, have been widely derided.43 However, if the option of peaceful progress was ruled out, then the solution would have had to have been a violent one of some sort. But, the implications of the arguments about economic preconditions put forward in this study are that the path to democracy was getting easier, not more diYcult. In which case, even if Rauh overstated the extent of the progress made by 1914, the prospects for further advances were improving. The events of 1914 represented a derailment, not the hitting of the buVers at the end of a Sackgasse.

A p p e n d i x 10A. D e s c r i p t i o n o f R e g r e s s i o n V a r i a b l e s a nd S o u r c e s (i) Appendix Table 10B.1 SDP1912 (dependent variable)

Social Democrat share of the vote in urban constituencies, from Die Reichstagswahlen von 1912, SDR vol. 250 (1913).

41 The view that the policy which led to the outbreak of war was based on an inability to appreciate the consequences of German policy, leading to the mistaken view that the conXict could be limited, was put forward in Jarausch (1969). But the consensus now is that Germany pushed for war, Pogge von Strandmann (1988). 42 The term Sackgasse is taken from the title of a paper by Volker Berghahn, Berghahn (1971). 43 The phrase is found in Rauh (1973), 347. The point is made more forcefully in Rauh (1977).






Log of total population in 1910, from SJDS (1911), 67–8. The percentage of the population who were Catholic, 1910 census, SJDS (1911), 678–9. As above, Prussian cities only. As above, non-Prussian cities only. Average moves in and out, divided by total population. Data from SJDS (1911), 67–8, and subsequent issues. Average wages (Ortsu¨bliche Tagelo¨hne) for male workers, aged 16 and over, from SJDS (1913), 826. Total number of employed workers divided by the number of enterprises, 1912, from SJDS (1916), 386–7. Social Democrat share of the vote in urban constituencies, from SDR a.f. vol. 37 (1879). Estimated levels of inequality, as ratios of the income shares of the top 10% to the bottom 40% (U10/L40 ratios). Data from Kuhnert (1916); estimation procedure is given in Appendix B of Grant (2002b).

(ii) Appendix Table 10B.2 SDP1912 (dependent variable)

Social Democrat share of the vote in nonurban constituencies, from Die Reichstagswahlen von 1912, SDR vol. 250 (1913). %AGOCCUP07 The percentage of the population occupied in agriculture, from the 1907 occupational census, SDR vol. 209 (1909). %CATHOLIC The percentage of the population who were Catholic, 1890 census, PS vol. 121, pp. 152–98. %SPEAKPOLISH As above, the percentage of the population who spoke Polish. %SPEAKLITHUAN As above, for Lithuanian speakers. CITYDISTANCE(a) Distance in kilometres from the Kreis midpoint to the nearest city with at least 200,000 inhabitants in 1900 (measured from contemporary maps).

Challenging the Kehrite View




As above, for cities with at least 50,000 inhabitants. The percentage of the total agricultural area in holdings of over 100 hectares, from 1895 Agricultural Census, SDR n.f. 212. As above, for holdings of between 20 and 100 hectares. The numbers of Gutsbezirke as percentages of all administrative units, from PS vol. 81 (1884), part 1, pp. 20–65. As Table 10.1 (averages of Kreis results for each constituency). Average wages (Ortsu¨bliche Tagelo¨hne) for rural areas; the Wgures are averages of male and female daily rates, 1892 and 1901, ZdKPSB (1904), 520–8.

Ap p e n d i x 10B. Ec o n o m e t r i c An a l y s i s The dependent variable in these regressions was the share of the Social Democrats in the 1912 election. The analysis showed that the SDP vote was highest in the larger cities (the log of the resident population has a strong eVect on voting behaviour), and also that the presence of a large Catholic population held it down (column a of Table 10B.1). In the second column a distinction is drawn between non-Prussian cities and Prussian cities. The inXuence of Catholicism is shown to have been signiWcantly greater inside Prussia. In the other columns a variety of economic variables are introduced. These turned out to have, at the most, a slight eVect on voting behaviour. Two labour market indicators were used. MOBILITY is a measure of migration (total moves per head of population for 1910–12); this was weakly linked with an increase in the SPD vote, but the result was neither robust nor statistically signiWcant. A second indicator, the ratio of applicants to jobs reported by the labour exchanges, was completely insigniWcant (a t-statistic of 0.16 when added to the equation estimated in column c).44 Short-term movements in job market conditions did not 44 The variable gave the ratio of applicants to vacancies for 1910–12, derived from town labour exchange statistics (SJDS 1911–13).

Table 10B.1. Regression results for linear log-odds voting model, 1912 election: dependent variable is logistic transformation of the SDP vote in major cities; estimated by OLS








1.51 þ0.332 (3.79) 0.0143 (5.72)

1.55 þ0.341 (4.28)

1.94 þ0.350 (4.10)

2.30 þ0.314 (3.48)

1.22 þ0.167 (2.26)

0.0183 (7.34) 0.0068 (2.29)

0.0183 (6.98) 0.0026 (0.67) þ0.0098 (1.22)

0.0188 (7.72) 0.0079 (2.65)

0.0086 (2.87) 0.0039 (1.46)

2.91 þ0.271 (2.35) 1.985 (6.90)

þ0.263 (1.51) þ0.0022 (1.55)

þ0.228 (1.75) þ0.0016 (1.27) þ0.232 (4.61)

70 0.567 0.435 0.418 25.82 21.563

70 0.516 0.540 0.519 25.84 17.560

65 0.512 0.539 0.519 19.70 15.699

70 0.503 0.576 0.543 17.40 16.185

55 0.332 0.706 0.669 19.19 5.302

þ0.617 (2.75) þ0.0010 (0.57) 0.063 (0.72) 46 0.520 0.637 0.592 14.03 10.822


Challenging the Kehrite View

appear to matter at all. Wage levels were positively linked with the socialist vote (column c), as was factory size (the average number of employees per enterprise with at least 10 workers), but both results are below the 10% signiWcance level. In the fourth column the socialist share of the vote in 1878 was included (SPD1878). As might be expected, this has an eVect on the other coeYcients. It shows the persistence of the socialist vote and the importance of the early establishment of a separate working-class culture and organization. Finally, measures of income inequality derived from Prussian tax statistics were included: the ratio of the share of the top 10% of the distribution to that of the bottom 40% (INEQUALITY). This meant that the analysis could only be run for the 46 Prussian cities. The results showed no connection between inequality and the socialist vote. Inequality does not appear to have been a factor in the electoral performance of the SPD in urban areas. Table 10B.2 gives the results of a similar analysis for the remaining Prussian constituencies. This is a data set covering 204 rural or small town constituencies. Two variables have a major eVect on voting behaviour: the percentage occupied in agriculture and the Catholic proportion. Both coeYcients are strongly negative, and the two variables together explain much of the variance in the dependent variable (column a). Again, the slow relative decline of the agricultural population was a factor working in favour of the SDP. Adding the percentage of the population who spoke Polish (column b) conWrms the point made earlier, that there remains a strong negative eVect from Catholicism even when allowance is made for this factor. However, Lithuanianspeakers showed a preference in favour of the SPD, so the presence of this group tends to raise the SPD vote (these were found in Memel and a few other constituencies in the Gumbinnen region of East Prussia). Cities also had an eVect on the surrounding countryside. The two city distance variables, measuring the distance from major cities, both have coeYcients that are negative and signiWcant: the SPD vote tended to fall as the distance from major cities rose. The eVect of the larger cities (CITYDISTANCE(a) over 200,000 inhabitants) is larger than that for smaller cities (CITYDISTANCE(c) over 50,000 inhabitants), as would be expected given the strength of the SPD vote in the major cities. The eVect of large farms and estates was examined using two variables. The Wrst was the percentage of agricultural land in holdings of

Table 10B.2. Analysis of the SDP vote in Prussian constituencies (excluding urban constituencies), 1912. Dependent variable is logistic transformation of vote a



1.417 0.0473 (14.9) 0.0240 (15.4)

1.262 0.0371 0.0181 0.0127 0.0110 0.0122 þ0.0281 0.0199 þ0.0296

N S R2 adj R2 F-test RSS

204 0.810 0.680 0.677 214.00 131.75

204 0.661 0.793 0.785 93.44 85.30

(12.1) (11.2) (4.37) (2.37) (4.25) (2.37) (3.80) (5.67)



1.588 0.0383 (11.9) 0.0182 (11.2) 0.0128 (4.42) 0.0110 (2.37) 0.0127 (4.41) þ0.0294 (2.48) 0.0182 (3.37) þ0.0295 (5.67) 0.225 (1.28)

0.816 0.0360 0.0179 0.0126 0.0102 0.0122 þ0.0298 0.0206 þ0.0321

204 0.660 0.795 0.785 83.52 84.59


(10.8) (10.9) (4.30) (2.15) (4.27) (2.47) (3.88) (5.31)

þ0.150 (0.83) 204 0.662 0.794 0.784 83.00 85.00

1.307 0.0370 0.0183 0.0128 0.0107 0.0121 þ0.0281 0.0200 þ0.0294

(12.0) (10.8) (4.37) (2.27) (4.24) (2.36) (3.80) (5.61)

0.00124 (0.36) 204 0.663 0.793 0.784 82.70 85.25


Challenging the Kehrite View

over 100 hectares (LAND>100HA). This variable had a negative eVect on the SDP vote: workers on large farms and estates tended to be less likely to vote for the SPD, perhaps for reasons of deference or loyalty to the estate-owner, or possibly out of fear that such a vote would be discovered and lead to retribution. However, the second variable, %GUTSBEZIRKE, has a positive sign. This had a persistent eVect on voting behaviour, and this result is robust to the inclusion or exclusion of other variables. The estimated coeYcients for this variable are unexpectedly large given the relative unimportance of the Gu¨ter in this period. They are rather higher than the coeYcients on LAND>100HA, indicating that resentment of the Gu¨ter was apparently a rather stronger eVect on voting behaviour than deference on larger holdings. The remaining columns introduce some other variables: the results in column d show that the level of wages did not aVect voting behaviour, in contrast to the results for the larger cites. Those in column e show that the presence of medium-sized holdings did not aVect the results either (these holdings typically employed some labour). But perhaps the most interesting result from this table is in column c, which shows that inequality at the Kreis level, measured by income inequality, had no signiWcant eVect on the SPD vote. This conWrms the result given in Table 10.B1, column f, that the level of inequality was not a factor in the SDP vote in the cities. Political inequality, as represented by the %GUTSBEZIRKE variable, had much more eVect.


a . a b b r e vi a t i o n s u s e d f o r m a i n s t a t i s t i c a l s o u r c e s 1. Publications of the Imperial Statistical OYce ErgVSDR MSDR SDR SJDR VSDR

Erga¨nzungsheft zur Vierteljahrsheft zur Statistik des Deutschen Reichs Monatsheft zur Statistik des Deutschen Reichs Statistik des Deutschen Reichs Statistisches Jahrbuch fu¨r das Deutsche Reich Vierteljahrsheft zur Statistik des Deutschen Reichs

2. Publications of the Prussian Statistical OYce PS SJPS ZdKPSB

Preußische Statistik Statistisches Jahrbuch fu¨r das Preußischen Staat Zeitschrift des Ko¨niglich Preußischen Statistischen Bureaus

3. Other publications SJB SJDS SJK

Statistisches Jahrbuch der Stadt Berlin Statistisches Jahrbuch deutscher Sta¨dte Statistisches Jahrbuch der Stadt Ko¨ln

b. references

A b e l , W. (1962), Geschichte der deutschen Landwirtschaft vom fru¨hen Mittelalter bis zum 19. Jahrhundert (Stuttgart: E. Ulmer). A b e l s h a u s e r, W. (1983), Wirtschaftsgeschichte der Bundesrepublik Deutschland 1945–1980 (Frankfurt am Main: Suhrkamp). A l b r e c h t , K. (1927), ‘Die Preußischen Gutsbezirke’, Zeitschrift des Preußischen Statistischen Landesamts, 3: 344–477. A m s d e n, A. (1989), Asia’s Next Giant: South Korea and Late Industrialization (Oxford: Oxford University Press). —— (2001), The Rise of ‘the Rest’: Challenges to the West from Late-Industrializing Economies (Oxford: Oxford University Press).


References and Sources

A n a n d , S. and Ka n b u r, S. M. (1993), ‘The Kuznets process and the inequality–development relationship’, Journal of Development Economics, 40/ 1: 25–52. A n de r s on, M. L. (2000), Practicing Democracy: Elections and Political Culture in Imperial Germany (Princeton: Princeton University Press). Asian Productivity Organisation (1994), Agrarian Structure and Reform Measures (Tokyo: APO). A t k i n s o n, A.B. (1996), ‘Income distribution in Europe and the United States’, Oxford Review of Economic Policy, 12: 15–27. —— (1996a), ‘Seeking to explain the distribution of income’, in J Hills (ed.), The New Inequalities: The Changing Distribution of Income and Wealth in the United Kingdom (Cambridge: Cambridge University Press). —— and Br a n d o l i n i , A. (1999), ‘Promises and Pitfalls in the Use of ‘‘Secondary’’ Data-Sets: Income Inequality in OECD Countries’, manuscript, NuYeld College, Oxford and Banca D’Italia, Research Department. —— Ra i n w a t e r , L. and Sm e e d i n g, T. M. (1995), Income Distribution in OECD Countries, OECD Social Policy Studies no. 18. A ug E´ -La r i b E´, M. (1950), La Politique agricole de la France de 1800 a` 1940 (Paris: Presse Universitaire). B a a c k, L. J. (1977), Agrarian Reform in Eighteenth Century Denmark, University of Nebraska Studies 56. B a d e, K. J. (1980), ‘German emigration to the United States and continental emigration to Germany in the late nineteenth and early twentieth centuries’, Central European History, 12/4: 348–77. —— (1984), ‘Vom Auswanderungsland zum ‘‘Arbeitseinfuhrland’’: kontinentale Zuwanderung und Ausla¨nderbescha¨ftigung in Deutschland im spa¨ten 19. und fru¨hen 20. Jahrhundert’, in K. J. Bade (ed.), Auswanderer, Wanderarbeiter, Gastarbeiter (OstWldern: Scripta Mercaturae Verlag). —— ed. (1987), Population, Labour and Migration in 19th and 20th Century Germany (Leamington Spa: Berg). —— (1991), ‘German transatlantic emigration in the nineteenth and twentieth centuries’, in P. C. Emmer and M. Mo¨rner (eds.), European Expansion and Migration (New York: Berg). B aines , D. (1994), ‘Population, migration and regional development, 1870–1939’, in R. Floud and D. McCloskey (eds.), The Economic History of Britain since 1700, vol. 2 (Cambridge: Cambridge University Press). B a i r o c h, P. (1989), ‘Les trois re´volutions agricoles’, Annales, 44/2: 317–54. B a l l o d , C. (1914), ‘Die Bevo¨lkerungsbewegung der letzten Jahrzehnte in Preußen und in einigen anderen wichtigen Staaten Europas’, Zeitschrift des Ko¨niglich Preußischen Statistischen Bureaus, 3: 237–90. B a r d h a n, P. (1989), ‘Alternative approaches to the theory of institutions in economic development’, in P. Bardhan (ed.), The Economic Theory of Agrarian Institutions (Oxford: Oxford University Press).

References and Sources


B a r f u s s, K. M. (1985), Gastarbeiter in Nordwestdeutschland 1884–1918 (Bremen: Staatsarchiv). B arkin , K. D. (1970), The Controversy over German Industrialisation 1890–1902 (Chicago: University of Chicago). B asu, K. (1997), Analytical Development Economics: The Less Developed Economy Revisited (Cambridge, Mass.: MIT Press). B a t e m a n , J. (1879), The Great Landowners of Great Britain and Ireland (London: Harrison). Baten , J. (1999), Erna¨hrung und wirtschaftliche Entwicklung in Bayern (1730–1880), Beitra¨ge zur Wirtschafts- und Sozialgeschichte 82 (Stuttgart: F. Steiner). Ba¨uerlicher Zustand in Deutschland (1883), Schriften des Verein fu¨r Sozialpolitik, vols. 21–4. B e l o w, G. von (1891), ‘Rittergut’, in Handwo¨rterbuch der Staatswissenschaft, suppl. vol. I: 672–5. B e n a b o u, R. (1996), ‘Inequality and growth’, NBER Macroeconomics Annual, xi: 11–74. B erg er , S. (2000), Social Democracy and the Working Class in Nineteenth and Twentieth Century Germany (Harlow: Longman). B erg hahn , V. (1971), ‘Das Kaiserreich in der Sackgasse’, Neue Politische Literatur, 16: 494–506. —— (1971a), Der Tirpitzplan: Genesis und Verfall einer innenpolitischen Krisenstrategie unter Wilhelm II (Du¨sseldorf: Droste Verlag). —— (1994), Imperial Germany,1871–1914: Economy, Society, Culture and Politics (Providence and Oxford: Berghahn). B erk ner , L. K. and Me n d e l s, F. (1980), ‘Inheritance systems, family structure and demographic patterns in Western Europe, 1700–1900’, in C. Tilly (ed.), Historical Studies of Changing Fertility (Princeton: Princeton University Press). Berlin, Statistisches Amt (1887), Die Berliner Volks-Za¨hlung vom 2. Dezember 1885 (Berlin). —— (1893), Die Bevo¨lkerungs- und Wohnungs-Aufnahme vom 2. Dezember 1890 in der Stadt Berlin (Berlin). —— (1913–16), Die Grundstu¨cks-Aufnahme vom 15. Oktober 1910, sowie die Wohnungs- und Bevo¨lkerungs-Aufnahme vom 1. Dezember 1910 in der Stadt Berlin und 44 Nachbar-Gemeinden (Berlin). B e r t h o l d , R. (1978), ‘Die Vera¨nderungen im Bodeneigentum und in der Zahl der Bauernstellen, der Kleinstellen und der Rittergu¨ter in den preußischen Provinzen Sachsen, Brandenburg und Pommern wa¨hrend der Durchfu¨hrung der Agrarreformen des 19. Jahrhundert’, Jahrbuch fu¨r Wirtschaftsgeschichte, Sonderband, 10–116.


References and Sources

Be r t h o l d , R. et al., eds. (1990), Geschichte der Produktivkra¨fte in Deutschland von 1800 bis 1945. Band 1: Produktivkra¨fte in Deutschland 1800 bis 1870 (Berlin: Akademie Verlag). B e r t ra m , J. (1964), Die Wahlen zum Deutschen Reichstag vom Jahre 1912. Parteien und Verba¨nde in der Innenpolitik des Wilhelminischen Reichs (Du¨sseldorf: Droste). B i e d e r m a n n, E. (1918), ‘Die Einkommens- und Vermo¨gensverha¨ltnisse der preußischen Bevo¨lkerung im Zeitraum 1895–1914, Zeitschrift des Ko¨niglich Preußischen Statistischen Bureaus, 60–86. B i r d s e l l, N. (1988), ‘Economic approaches to population growth’, in H. Chenery and T. Srinivasan (eds.), Handbook of Development Economics, vol. I (Amsterdam: North-Holland). B j o r n, C. (1977), ‘The peasantry and agrarian reform in Denmark’, Scandinavian Journal of Economic History, 25/1: 117–37. B l a c k b o u rn , D. (1980), Class, Religion and Local Politics in Wilhelmine Germany: The Centre Party in Wu¨rttemberg before 1914 (New Haven: Yale University Press). —— (1997), The Long Nineteenth Century: The Fontana History of Germany 1780–1918 (London: HarperCollins). —— and El e y, G. (1984), The Peculiarities of German History (Oxford). Board of Trade (1908), ‘Report of an Enquiry by the Board of Trade into Working Class Rents, Housing and Retail Prices Together with Rates of Wages in Certain Occupations in the Principal Industrial Towns of the German Empire’, Parliamentary Papers, Accounts and Papers (47), cd 4032. B o¨ hme , H. (1966), Deutschlands Weg zur Groâmacht: Studien zum Verha¨ltnis von Wirtschaft und Staat wa¨hrend der Reichgru¨ndungszeit 1848–1881 (Berlin and Cologne: Oxford University Press). B o¨ l s k e r -Sc h l i c h t , F. (1997), ‘Deutsche Saisonarbeiter in den Niederlanden bis zum 19. Jahrhundert’, in H.-H. Nolte (ed.), Deutsche Migrationen (Mu¨nster: LIT Verlag). B o r c h a r dt , K. (1982), ‘Was there a capital shortage in the Wrst half of the nineteenth century in Germany?’, in K. Borchardt, Perspectives on Modern German Economic History and Society (Cambridge: Cambridge University Press). B o r j a s, G. J. (1994), ‘The economics of immigration’, Journal of Economic Literature, 23 (Dec.), 1667–1717. B ou rgign on, F. (1990), ‘Growth and inequality in the dual model of development: the role of demand factors’, Review of Economic Studies, 57: 215–28. —— and Mo r r i s o n, C. (1998), ‘Inequality and development’, Journal of Development Economics, 57: 233–57. B ow ley , A. L. (1946), Elements of Statistics, 6th edn. (New York: Staples). B oy er , G. R. (1990), An Economic History of the English Poor Law, 1750–1850 (Cambridge: Cambridge University Press).

References and Sources


B r a y, F. (1986), The Rice Economies: Technology and Development in Asian Societies (Oxford: Blackwells). ¨ ber die Methoden der Einkommenverteilungstatistik’, B resciani . C. (1907), ‘U Jahrbu¨cher fu¨r Nationalo¨konomie und Statistik, 33: 577–600. B rezis , E., Kr u g m a n, P. and Tsiddon , D. (1993), ‘Leapfrogging in international competition: a theory of cycles in national technological leadership’, American Economic Review, 83/5: 1211–19. B r i n k ma n n, T. (1935), Theodore Brinkmann’s Economics of the Farm Business (Berkeley: Social Science Research Council). B r o a d b e rr y , S. N. (1997), ‘Anglo-German Productivity DiVerences: A Sectoral Analysis’, unpubl. —— (1997a), The Productivity Race: British Manufacturing in International Perspective, 1850–1990 (Cambridge: Cambridge University Press). B r o d r i c k, G. (1881), English Land and Landlords (London: Cassell, Petter, Galpin). B r o e s i k e, M. (1907), ‘Die Binnenwanderung im Preußischen Staat’, Zeitschrift des Ko¨niglich Preußischen Statistischen Bureaus, 1: 1–62. B r o u w e r, W. (1957), ‘A. Schultz-Lupitz zum 125. Geburtstag’, Zeitschrift fu¨r Agrargeschichte und Agrarsoziologie, 5/1: 69–75. B runo , M., Ra v a i ll o n, M. and Sq u i r e, L. (1996), Equity and Growth in Developing Countries: Old and New Perspectives on the Policy Issues, Policy Research Working Paper 1563 (Washington: World Bank). B r y, G. (1960), Wages in Germany 1871–1945 (Princeton: Princeton University Press). B u c h s t e i n e r, I. (1994), ‘Besitzkontinuita¨t, Besitzwechsel und Besitzverlust in der Gutswirtschaft Pommerns 1879–1910’, in H. Reif (ed.), Ostelbische Agrargesellschaft im Kaiserreich und in der Weimarer Republik (Berlin: Akademie Verlag). B u r g d o¨ r f e r, F. (1931), ‘Migration across the frontiers of Germany’, in I Ferenczi and W. F. Willcox, International Migrations, vol. 2 (New York: NBER). C alleo , D. (1978), The German Problem Reconsidered: Germany and the World Order, 1870 to the Present (Cambridge: Cambridge University Press). C a r s t e n, F. (1989), A History of the Prussian Junkers (Aldershot: Scolar Press). C a t t , C. (1986), ‘Farmer and factory worker: rural society in Imperial Germany, the example of Maudach’, in R. J. Evans and W. R. Lee (eds.), The German Peasantry (London: Croom Helm). C a u m a n n s, U. (1994), Technischer Fortschritt und Sozialer Wandel in deutschen Ostprovinzen (Bonn: Kulturstiftung der deutschen Vertriebenen). C h i c k e r i n g , R. (1984), ‘We men who feel most German’: a cultural study of the Pan-German League (London: George Allen & Unwin).


References and Sources

C l a p h a m , J. H. (1921, repr. 1966), The Economic Development of France and Germany 1815–1914 (Cambridge: Cambridge University Press). C l a r k e , G. (1995), ‘More evidence on income distribution and growth’, Journal of Development Economics, 47: 403–27. C o a l e, A. J. and Wa t k i n s, S. C., eds. (1986), The Decline of Fertility in Europe (Princeton: Princeton University Press). C o n r a d , J. (1891), ‘Bodenzersplitterung’, in Handwo¨rterbuch der Staatwissenschaft, vol. 3 (Jena: G. Fischer). C o n z e, W. (1969), ‘The eVects of nineteenth century liberal agrarian reform on social structure in central Europe’, in F. Crouzet (ed.), Essays in European Economic History (London: E. Arnold). C r a i g i e , P. (1887), ‘The size and distribution of agricultural holdings in England and abroad’, Journal of the Royal Statistical Society, 100: 86–142. C r e w, D. (1979), Town in the Ruhr: A Social History of Bochum 1860–1914 (New York: Columbia University Press). C r o t h e r s, G. D. (1941), The German Elections of 1907 (New York: AMS). D a de , H. (1913), Die deutsche Landwirtsaft unter Kaiser Wilhelm II (Halle: C. Marhold). D a s g u p t a, P. (1993), An Inquiry into Well-Being and Destitution (Oxford: Oxford University Press). D ei nin ger , K. and Sq u i r e , L. (1996), ‘A new data set measuring income inequality’, World Bank Economic Review, 10/3: 565–91. —— and—— (1997), ‘Economic growth and income inequality: re-examining the links’, Finance and Development (Mar.), 38–41. —— and —— (1998), ‘New ways of looking at old issues: inequality and growth’, Journal of Development Economics, 57: 259–87. D ei st , W. (1976), Flottenpolitik und Flottenpropaganda: das Nachrichtenbureau des Reichsmarineamtes 1897–1914 (Stuttgart: Deutsche Verlagsanstalt). D el Fa b b r o, R. (1996), Transalpini: Italienische Arbeitwanderung nach Su¨ddeutschland im Kaiserreich 1870–1918, Studien zur historischen Migrationsforschung 2 (Osnabru¨ck: Rausch). D e s a i, A. V. (1968), Real Wages in Germany, 1871–1914 (Oxford: Oxford University Press). D i c k l e r, R. A. (1975), ‘Organisation and change in productivity in Eastern Prussia’, in W. N. Parker and E. L. Jones (eds.), European Peasants and their Markets (Princeton: Princeton University Press). D illwitz , S. (1973), ‘Die Struktur der Bauernwirtschaft von 1871 bis 1914’, Jahrbuch fu¨r Geschichte, 9: 47–127. D odd , D. (1993), Historical Statistics of the United States: Two Centuries of the Census, 1790–1990 (Westport: Greenwood). D o l l a r , D. and Kr a a y, A. (2000), ‘Growth is Good for the Poor’, manuscript, www.worldbank.org/research

References and Sources


Do v r i n g, F. (1956), Land and Labour in Europe, 1900–1950 (The Hague: Martinus NijhoV). Du mk e, R. (1991), ‘Income inequality and industrialisation: the Kuznets curve re-examined’, in Y. Brenner, H. Kaelble, and M. Thomas (eds.), Income Distribution in Historical Perspective (Cambridge: Cambridge University Press). Dy er , B. (1896), ‘Green manuring’, Journal of the Royal Agricultural Society of England and Wales, 773–9. E dwardes , J. and Og i l v i e, S. (1995), Universal Banks and German Industrialisation: A Reappraisal, CEPR Discussion Paper 1171. E g g e r t , U. (1883), ‘Die Bewegung der Holzpreise und Tagelohn-Sa¨tze in den preußischen Staatsforsten von 1800 bis 1879’, Zeitschrift des Ko¨niglich Preußischen Statistischen Bureaus, 23/1: 1–44. ¨ konomisches Wandel, KriE h m e r, J. (1991), Heiratsverhalten, Sozialstruktur, O tische Studien zur Geschichtswissenschaft 92 (Go¨ttingen: Vandenhoeck & Rupprecht). E l e y, G. (1980), Reshaping the German Right (New Haven: Yale University Press). —— (1986), From UniWcation to Nazism: Re-Interpreting the German Past (Boston and London: Allen & Unwin). —— , ‘German history and the contradictions of modernity: the bourgeoisie, the state and the mastery of reform’, in G. Eley (ed.), Society, Culture, and the State in Germany, 1870–1930 (Ann Arbor: University of Michigan Press), 43–103. —— (2002), Forging Democracy: The History of the Left in Europe 1850–2000 (Oxford: Oxford University Press). E ng el , E. (1868), ‘Die Ergebnisse der Classensteuer, der classiWcierten Einkommensteuer und der Mahl- und Sachlachtsteuer im preußischen Staate’, Zeitschrift des Ko¨niglich Preußischen Statistischen Bureaus, 1: 25–84. —— (1875), ‘Die Ergebnisse der Classensteuer und der classiWcierten Einkommensteuer im preußischen Staate’, Zeitschrift des Ko¨niglich Preußischen Statistischen Bureaus, i. 108–48. E ng erman, S. L. and So k o l o f f, K. L. (1997), ‘Factor endowments, institutions, and diVerential paths of growth among new world economies: a view from economic historians of the United States’, in S. Haber (ed.), How Latin America Fell Behind: Essays on the Economic Histories of Brazil and Mexico, 1800–1914 (Stanford: Stanford University Press). E p k e n h a n s , M. (1991), Die wilhelminischen Flottenru¨stung 1908–1914 (Munich: Oldenbourg). E r i c k s o n , C. (1957), American Industry and the European Immigrant (Cambridge, Mass.: Harvard University Press). E v a n s , E. L. (1997), The German Centre Party 1870–1933: A Study in Political Catholicism (Carbondale: Southern Illinois University Press).


References and Sources

E v a n s, R. (1997), Rereading German History: From UniWcation to ReuniWcation (London: Routledge). E v e r s h e d , H. (1884), ‘Improvements of the plants of the Weld’, Journal of the Royal Agricultural Society of England and Wales, 77–113. F a i r b a i r n , B. (1992), ‘Interpreting Wilhelmine elections: national issues, fairness issues and electoral mobilisation’, in E. L. Jones and J. Rettalack (eds.), Elections, Mass Politics and Social Change in Modern Germany: New Perspectives (Cambridge: Cambridge University Press). —— (1997), Democracy in the Undemocratic State: The German Reichstag Elections of 1898 and 1903 (Toronto: University of Toronto Press). F e i, J. E., Ra n i s , G. and Kuo, S. Y. (1979), Growth with Equity: The Taiwan Case (World Bank and Oxford University Press). F e i n s t e i n , C. H. (1972), National Income, Expenditure and Output of the United Kingdom 1855–1965 (Cambridge: Cambridge University Press). —— (1988), ‘The rise and fall of the Williamson Curve’, Journal of Economic History, 47/3: 699–729. —— (1998), ‘Pessimism perpetuated: real wages and the standard of living in Britain during and after the Industrial Revolution’, Journal of Economic History, 58/3: 625–58. —— and Th o m a s , M. (2000), ‘A Plea for Errors’, manuscript, All Souls College, Oxford. F e l de n k i r c h e n, W. (1982), ‘Zur Kapitalbescha¨ftigung und Kapitalverwendung bei Aktiengesellschaften des deutschen Maschinenbaus im 19. und beginnenden 20. Jahrhundert’, Vierteljahresschrift fu¨r Wirtschaftsgeschicht, 49: 38–74. F erenczi , I. and Wi l lc o x , W. F. (1929–31), International Migrations, 2 vols. (New York: NBER). F e r g u s o n, N. (1994), ‘Public Wnance and national security: the domestic origins of the First World War revisited’, Past and Present, 142: 141–68. —— (1998), The Pity of War (London: Allen Lane). F i n c k v o n Finck en stein, H. W. Gr a f (1960), Die Entwicklung der Landwirtschaft in Preußen und Deutschland 1800–1930 (Wu¨rzburg: Holzner). F ircks , A. v o n (1879), ‘Ruckblick auf die Bewegung der Bevo¨lkerung im Preußischen Staat 1816–74’, Preuâische Statistik, 48/1. F l o r a , P., Kr a u s, F. and Pfennig , W. (1983), State, Economy and Society in Western Europe 1815–1973: A Data Handbook, 2 vols. (Frankfurt am Main: Campus). F oh lin , C. (2002), ‘Corporate capital structure and the inXuence of universal banks in pre-World War I Germany’, Jahrbuch fu¨r Wirtschaftsgeschichte, 113–34. F r a n k e n s t e i n , K. (1893), Die Arbeiterfrage in der deutschen Landwirtschaft (Berlin: R. Oppenheim).

References and Sources


F rauendorfer , S. (1956), ‘Naturwissenschaften und Technik als dynamisches Moment’, Zeitschrift fu¨r Agrargeschichte und Agrarsoziologie, 4/2: 113–27. —— (1957), Ideengeschichte der Agrarwirtschaft und Agrarpolitik im deutschen Sprachgebiet, Band 1 (Munich: Bayerischer Landwirtschaftsverlag). F r e m d l i n g, R. (1991), ‘Productivity comparisons between Great Britain and Germany, 1855–1913’, Scandinavian Journal of Economic History, 39/1: 28–42. F ussell , G. E. (1962), ‘The early days of chemical fertilisers’, Nature, 195: 750–4. Ga l o r , O. and Ts i d d o n, D. (1994), Human Capital Distribution, Technological Progress and Economic Growth, CEPR Discussion Paper no. 971 (London: CEPR). Ga l o s, A., Gen tzen , F. and Ja ko´k c z y c k, W. (1966), Die Hakatisten: Der Deutsche Ostmarkverein 1894–1934 (Berlin: VEB). Ge i s e n b e r g e r , S. and MU¨ller , J. H. (1972), Die Einkommensstruktur in verschiedenen deutschen La¨ndern 1874–1914, Schriften zu Regional- und Verkehrsproblemen 10 (Berlin). Ge i s s, I. (1990), ‘The German version of imperialism, 1898–1914: Weltpolitik’, in G. Scho¨llgen (ed.), Escape into War? The Foreign Policy of Imperial Germany (Oxford: Berg). Ge r s ch e n k r o n, A. (1943), Bread and Democracy in Germany (Berkeley: University of California Press). —— (1962), Economic Backwardness in Historical Perspective (Cambridge, Mass: Harvard University Press). Gh i m u r e, K., ed. (2001), Land Reform and Peasant Livelihoods: The Social Dynamics of Rural Poverty and Agrarian Reform in Developing Countries, UN Institute for Social Development (New York: UN). Go l d i n, C. (1994), ‘The political economy of immigration restriction in the United States, 1890–1921’, in C. Goldin and G. Libecap (eds.), The Regulated Economy: A Historical Approach (Chicago: University of Chicago Press). Golding , F. (1930), ‘Die Wanderbewegung in Ostpreuâen seit der Jahrhundertwende mit besonderer Beru¨cksichtigung der Abwanderung vom Lande’, Zeitschrift des Preußischen Statistischen Landesamts, 69/2: 200–34. Go l t z, T. Fr h r . v o n d e r (1875), Die Lage der la¨ndlichen Arbeiter im deutschen Reich (Berlin: P. Parey). —— (1893), Die la¨ndliche Arbeiterklasse und der preuâische Staat (Jena: G. Fischer). Go¨ m m e l , R. (1979), Realeinkommen in Deutschland: ein internationaler Vergleich (Nu¨rnberg: Im Selbstverlag). Gr a n t , O. (2000), ‘Internal Migration in Germany, 1870–1913’, Oxford D.Phil. thesis.


References and Sources

Gr a n t , D. (2002a), Productivity in German Agriculture: Estimates of Agricultural Productivity from Regional Accounts for 21 German Regions: 1880/4, 1893/7 and 1905/9, Oxford University Discussion Paper in Economic and Social History, available at http://www.nuV.ox.ac.uk/ Economics/History/ —— (2002b), Does Industrialisation Push Up Inequality? New Evidence on the Kuznets Curve from Nineteenth Century Prussian Tax Statistics, Oxford University Discussion Paper in Economic and Social History, available at http://www.nuV.ox.ac.uk/Economics/History/ —— (2002c), ‘Max Weber and Die Lage der Landarbeiter im ostelbischen Deutschland: a statistical examination’, forthcoming in the Jahrbuch fu¨r Wirtschaftsgeschichte, 2: 61–84. —— (2002d ), ‘Explaining the Decline of German Emigration to the United States after 1895’, manuscript, NuYeld College, Oxford. —— (2002e), ‘ ‘‘Few Better Farmers in Europe?’’ Productivity, Change and Modernisation in East-Elbian Agriculture 1870–1913’, manuscript, NuYeld College, Oxford. G reene , W. H. (2000), Econometric Analysis, 4th edn. (New Jersey: PrenticeHall). G r e g o r y, P. R. (1997), ‘Searching for consistency in historical data: alternative estimates of Russia’s industrial production, 1887–1913’, Journal of Economic History, 57/1: 196–202. G rimmer -So l e m, E. (1999), ‘The Science of Progress: The Rise of Historical Economics and Social Reform in Germany, 1864–1894’, Oxford D.Phil. thesis. G roh, D. (1973), Negative Integration und revolutiona¨rer Attentismus: Die deutsche Sozialdemokratie am Vorabend des Ersten Weltkrieges (Frankfurt am Main: Propyla¨en). G r u m b a ch , F. and Ko¨n ig, H. (1957), ‘Bescha¨ftigung und Lo¨hne der deutschen Industriewirtschaft 1888–1954’, Weltwirtschaftsarchiv, 79: 125–54. G u t t s m a n, W. (1981), The German Social Democratic Party 1875–1933 (London: Allen & Unwin). H a b ak k u k , J. (1958), ‘Population growth and economic development’, in J. Habakkuk et al., Lectures on Economic Development (Istanbul: Sermet Matbaash). H a b e r l e , R. (1938), ‘German approaches to internal migration’, in D. Thomas (ed.), Research Memorandum on Migration DiVerentials (New York: Social Science Research Council). H agen , W. (1980), Germans, Poles and Jews: The Nationality ConXict in the Prussian East, 1772–1914 (Chicago: University of Chicago Press). H aines , M. (1977), Economic–Demographic Interrelations in Developing Agricultural Regions: A Case Study of Prussian Upper Silesia, 1840–1914 (New York: Arno).

References and Sources


—— (1979), Fertility and Occupation: Population Patterns in Industrialisation (New York: Academic). Ha j n a l , J. (1965), ‘European marriage patterns in perspective’, in D. Glass and D. Eversley (eds.), Population in History (Chicago: Aldine). Ha n l e y , S. B. and Ya ma m u r a , K. (1977), Economic and Demographic Change in Pre-Industrial Japan, 1600–1868 (Princeton: Princeton University Press). Ha r d a c h, K. W. (1967), Die Bedeutung wirtschaftlicher Faktoren bei der Wiedereinfu¨hrung der Eisen- und Getreidezo¨lle in Deutschland 1879 (Berlin: Duncker & Humblot). Ha r n i s ch , H. (1977), ‘Die Bedeutung der kapitalistischen Agrarreform fu¨r die Herausbildung des Inneren Markte und die industrielle Revolution in der o¨stlichen Provinzen Preußens in der ersten Ha¨lfte des 19. Jahrhunderts’, Jahrbuch fu¨r Wirtschaftgeschichte, 63–82. —— (1984), Kapitalistische Agrarreform und Industrielle Revolution (Weimar: H. Bo¨hlhaus). —— (1986), ‘The background to the agricultural reforms in feudal Prussia east of the Elbe 1760–1807’, in R. J. Evans and W. R. Lee (eds.), The German Peasantry (London: Croom Helm). Ha t t o n, T. J. (1993), A Model of UK Emigration 1870–1913, CEPR Discussion Paper no. 771. —— (1997), ‘The immigrant assimilation puzzle in late nineteenth century America’, Journal of Economic History 57: 34–62. —— and Wi l l i a m s o n, J. G., eds. (1994), Migration and the International Labour Market, 1850–1939 (London: Routledge). —— and —— (1996), ‘What drove the mass migrations from Europe in the nineteenth century?’ Population and Development Review, 20: 533–59. —— and —— (1998), The Age of Mass Migration: Causes and Economic Impact (Oxford: Oxford University Press). Ha t z i u s, J. (1994), Regional Migration, Unemployment and Vacancies: Evidence from West German Microdata, Oxford, Institute of Economics and Statistics, Applied Economics Discussion Paper no. 164. Ha u s e r , A. (1965), ‘Schweizer Bauern als Kolonisten in Preußen und Litauen’, Zeitschrift fu¨r Agrargeschichte und Agrarsoziologie, 13: 212–19. Ha u s h o f e r , H. (1957), Die deutsche Landwirtschaft im technische Zeitalter, vol. 5 of G. Franz (ed.), Deutsche Agrargeschichte (Stuttgart: Ulmer). Ha y am i , Y. (1997) Development Economics: From the Poverty to the Wealth of Nations (Oxford: Oxford University Press). He l f e r r i c h, K. (1915), Deutschlands Volkswohlstand (Berlin: G. Stilbe). He l l i n g, G. (1966), ‘Zur Entwicklung der Produktivita¨t in der deutschen Landwirtschaft im 19. Jahrhundert’, Jahrbuch fu¨r Wirtschaftsgeschichte, 129–41. He m p e l , L. (1957), ‘Heuerlingswesen und Croftersystem’, Zeitschrift fu¨r Agrargeschichte und Agrarsoziologie, 5/2: 169–79.


References and Sources

H en ni ng, F. W. (1977), ‘Produktionskosten und Preisbildung fu¨r Getreide in den letzten Jahrzehnten des 19. Jahrhunderts’, Zeitschrift fu¨r Agrargeschichte und Agrarsoziologie, 25: 214–36. H en nock , E. P. (2000), ‘The urban sanitary movement in England and Germany, 1838–1914: a comparison’, Continuity and Change, 15/2: 269–96. H entschel , V. (1978), Wirtschaft und Wirtschaftspolitik im wilhelminischen Deutschland (Stuttgart: Klett-Cotta). —— (1979), ‘Erwerbs- und Einkommensverha¨ltnisse in Sachsen, Baden und Wu¨rttemberg vor dem Ersten Weltkrieg’, Vierteljahrschrift fu¨r Sozial- und Wirtschaftsgeschichte, 66/1: 26–73. H e r b e r t , U. (1990), A History of Foreign Labour in Germany 1880–1980 (Ann Arbor: University of Michigan Press). H e r w i g, H. (1987), ‘Luxury Fleet’: The Imperial German Navy 1888–1918 (London: AshWeld). H es s, K. (1994), ‘Zur wirtschaftlichen Lage der Groâagrarier im ostelbischen Preuâen 1867/71 bis 1914’, in H. Reif (ed.), Ostelbische Agrargesellschaft im Kaiserreich und in der Weimarer Republik (Berlin: Akademie Verlag). H iggins , M. and Wi l l i a m s o n, J. G. (1999), ‘Explaining Inequality the World Round: Cohort Size, Kuznets Curves and Openness’, manuscript, Federal Reserve Bank of New York and Harvard University. H iggs , R. (1971), ‘Race, skill and earnings: American immigrants in 1909’, Journal of Economic History, 31: 420–9. H i l d e b r a n d, K. (1989), Deutsche Auâenpolitik, 1871–1918 (Munich: Oldenbourg). —— (1995), Reich–Nation-State–Great Power: ReXections on German Foreign Policy 1871–1945 (London: German Historical Institute). H obson , J. M. (1997) The Wealth of States: A Comparative Sociology of International Economic and Political Change (Cambridge: Cambridge University Press). H o c h s t a d t , S. (1996), ‘The socio-economic determinants of increasing mobility in nineteenth century Germany’, in D. Hoerder and L. P. Moch (eds.), European Migrants (Boston: Northeastern Press). —— (1999), Mobility and Modernity: Migration in Germany 1820–1989 (Ann Arbor: University of Michigan Press). H oerder , D. and Na g l e r , J., eds. (1995), People in Transit (Cambridge: Cambridge University Press). H o f f ma n n, W. (1963), ‘The take-oV in Germany’, in W. Rostow (ed.), The Economics of Take-OV into Sustained Growth (London: Macmillan). —— (1965), Das Wachstum der Deutschen Wirtschaft seit der Mittel des 19. Jahrhunderts (Berlin: Springer Verlag). H oh ls , R. (1995), ‘Germany’, in P. Scholliers and V. Zamagni (eds.), Labour’s Reward (Aldershot: E. Elgar).

References and Sources


Ho h o r st, G. (1977), Wirtschaftswachstum und Bevo¨lkerungsentwicklung in Preußen 1816 bis 1914 (Mu¨nster: University of Mu¨nster). Ho n i g s h ei m, P. (1949), ‘Max Weber as historian of agriculture and rural life’, Agricultural History, 23/3: 179–212. Ho u r w i c h , I. (1922), Immigration and Labour: The Economic Aspects of European Immigration to the United States (New York: Huebsch). Hu nt , J. C. (1974), ‘Peasants, grain tariVs and meat quotas’, Central European History, 7: 311–31. Hu t ch in so n, E. P. (1956), Immigrants and their Children 1850–1950 (New York: Wiley). Institut International d’Agriculture (1915), Annuaire International de Statistique Agricole 1913/14. J ac k ma n , R. and Sa v o u r i , S. (1992), ‘Regional migration in Britain: an analysis of gross Xows using NHS central register data’, Economic Journal, 102: 1433–50. J ac k so n, J. J. (1997), Migration and Urbanisation in the Ruhr Valley 1821–1914 (New Jersey: Humanities Press). J ar a u s c h, K. (1969), ‘The illusion of limited war: Chancellor Bethmann Hollweg’s calculated risk’, Central European History, 48–76. J az a i r y , I., Al a m g i r , M. and Pa n u c c i o, T. (1992), The State of World Rural Poverty: An Enquiry into its Causes and Consequences, International Fund for Agricultural Development. J eck, A. (1970), Wachstum und Verteilung des Volkseinkommens (Tu¨bingen: Mohr). J ei dels, O. (1905), Das Verha¨ltnis der deutschen Groâbanken zur Industrie, Staats- und Sozialwissenschaftliche Forschungen, 24. J en ks, J. W. and La u c k, W. J. (1926), The Immigration Problem: A Study of American Immigration Conditions and Needs (New York: Huebsch). Ka e l b l e, H. (1976), ‘Social stratiWcation in Germany in the 19th and 20th centuries: a survey of research since 1945’, Journal of Social History, 10: 144–65. —— (1986), Industrialisation and Social Inequality in Nineteenth Century Europe (Leamington Spa: Berg). —— and Fe d e r s p i e l , R. (1990), Soziale Mobilita¨t Berlin 1825–1957, Historische Statistik von Deutschland, vol. X. —— and Hoh ls , R. (1989), Die Regionale Erwerbsstruktur im Deutschen Reich und in der Bundesrepublik 1895–1970, Quellen und Forschungen zur historischen Statistik von Deutschland. Ka e r g e r, K. (1890), ‘Die Sachsenga¨ngerei’, Landwirtschaftliche Jahrbu¨cher, 19: 237–522. Kaiserlich Statistisches Amt (1916), Die Deutschen im Ausland und die Ausla¨nder im Deutschen Reich, Erga¨nzungsheft zur Vierteljahrsheft zur Statistik des Deutschen Reichs 4.


References and Sources

Ka l d o r, N. (1956), ‘Alternative theories of distribution’, Review of Economic Studies, 23/2: 94–100. Ka m p h o e f n e r , W. D. (1986), ‘At the crossroads of economic development: background factors aVecting emigration from nineteenth-century Germany’, in I. A. Glazier and L. De Rosa (eds.), Migration across Time and Nations (New York: Holmes and Meier). —— (1987), The Westfalians: From Germany to Missouri (Princeton: Princeton University Press). —— (1999), ‘Deutschen-Amerikaner: Musterknaben der Einwanderung?’, IMIS-Beitra¨ge, 12. Ke h r , E. (1975), Battleship Building and Party Politics in Germany (Chicago: University of Chicago Press). —— (1977), Economic Interest, Militarism and Foreign Policy (Berkeley: University of California Press). Ke n n e dy , P. (1980), The Rise of the Anglo-German Antagonism, 1860–1914 (London: Allen & Unwin). Ke y d e r , C. and O’Br i e n, P. (1978), Economic growth in Britain and France, 1780–1914: Two Paths to the Twentieth Century (London: Allen & Unwin). Ke y s er, E. (1959), ‘Die Ansiedlung von Schwaben in Westpreuâen 1776–1805’, Zeitschrift fu¨r Agrargeschichte und Agrarsoziologie, 7/2: 169–76. Ki e s e w e t t e r , H. (1988), Industrialisierung und Landwirtschaft: Sachsens Stellung im regionalen Industrialisierungsprozess Deutschlands im 19. Jahrhundert (Cologne: Bo¨hlau). Ki n d l e b er g e r, C. (1967), Europe’s Postwar Growth: The Role of Labour Supply (Cambridge, Mass.: Harvard University Press). Kl e m m , V. (1994), ‘Die Agrarwissenschaften und die Modernisierung der Gutsbriebe in Ost-und Mitteldeutschland’, in H. Reif (ed.), Ostelbische Agrargesellschaft im Kaiserreich und in der Weimarer Republik (Berlin: Akademie Verlag). Kn a p p , G. F. (1887), Die Bauernbefreiung und der Ursprung der Landarbeiter in den a¨lteren Theilen Preussens (Leipzig: Duncker & Humblot). —— (1891), Die Landarbeiter bei der Stein-Hardenbergischen Gesetzgebung (Strassburg: Heitz and Mundel). Kn od e l, J. (1974), The Decline of Fertility in Germany 1871–1939 (Princeton: Princeton University Press). —— (1988), Demographic Behaviour in the Past (Cambridge: Cambridge University Press). Kn o k e , A. (1911), Ausla¨ndische Wanderarbeiter in Deutschland, Diss., Leipzig. Ko c k a , J., ed. (1975), ‘Soziale Schichtung und Mobilita¨t in Deutschland im 19. und 20. Jahrhundert’, Geschichte und Gesellschaft 1/1 (Go¨ttingen: Vandenhoeck und Ruprecht). —— (1990), Arbeitsverha¨ltnisse und Arbeiterexistenzen: Grundlagen der Klassenbildung im 19. Jahrhundert (Bonn: Dietz).

References and Sources


Ko l ak o w s k i, L. (1978), Main Currents of Marxism, 3 vols. (Oxford: Oxford University Press). Ko¨ l l m a n n , W. (1959), ‘Industrialisierung, Binnenwanderung und ‘‘Soziale Frage’’ ’, Vierteljahrschrift fu¨r Sozial- und Wirtschaftsgeschichte. —— (1960), Sozialgeschichte der Stadt Barmen im 19. Jahrhundert (Tu¨bingen: J. C. B. Mohr). —— (1964), ‘The population of Germany in the age of industrialisation’, in H. Moller (ed.), Population Movements in Modern European History (London: Collier-Macmillan). —— (1974), ‘Demographische ‘‘Konsequenzen’’ der Industrialisierung in Preuâen’, in W. Ko¨llmann (ed.), Bevo¨lkerung in der Industriellen Revolution (Go¨ttingen: Vandenhoeck und Ruprecht). —— (1976), ‘Bevo¨lkerungsgeschichte 1800–1870’, in H. Aubin and W. Zorn (eds.), Handbuch der deutschen Wirtschaft- und Sozialgeschichte, vol. 2 (Stuttgart: Klett-Cotta). —— (1980), Quellen zur Bevo¨lkerungsgeschichte Deutschlands, vol. 1 of Quellen zur Bevo¨lkerungs-, Sozial- und Wirtschaftsstatistik Deutschlands (Boppard: Boldt). —— and Ma rs c h a l c k, P. (1973), ‘German emigration’, Perspectives in American History, 7: 499–554. Ko ml o s, J. (1995), The Biological Standard of Living in Europe and America 1700–1900: Studies in Anthropometric History (Aldershot: Variorum). ¨ ber das Verha¨ltnisse zwischen Rohertrag und ReinKo¨ n i t z, C. von (1895), ‘U ertrag in landwirtschaftlichen Betrieben verschiedener Wirtschaftssysteme’, Landwirtschaftliche Jahrbu¨cher 24: 309–444. Ko s e l l e c k, R. (1967), Preuâen zwischen Reform und Revolution (Stuttgart: Klett-Cotta). Ko s t e r s-Kr a f t, M. (1998), Wanderarbeit und Groâbaustelle: Niederla¨nder beim Bau des Dortmund-Ems-Kanals 1892–1900, Studien zur historischen Migrationsforschung 8 (Osnabru¨ck: Rausch). Ku c z y n s k i, J. (1945), A Short History of Labour Conditions in Germany, 1800 to the present day (London: F. Muller). Ku c z y n s k i, R. (1897), Der Zug nach der Stadt (Stuttgart: J. G. Cotta). —— (1908), Einkommen und Miete (Berlin: Berlin Statistische Amt). —— (1913), Arbeitslohn und Arbeitszeit in Europa und Amerika (Berlin: J. Springer). Ku h n e r t , F. (1911), ‘Die Ergebnisse der preuâischen Einkommens- und Erga¨nzungssteuer: Veranlagung nach Sta¨dtegruppen fu¨r das Jahr 1908’, Zeitschrift des Ko¨niglich Preuâischen Statistischen Bureaus, 1–54. —— (1916), ‘Einkommensgliederung der preuâischen Bevo¨lkerung im Zeitraum 1902–1911’, Zeitschrift des Ko¨niglich Preuâischen Statistischen Bureaus, 269–309.


References and Sources

Ku zn et s , S. (1955), ‘Economic growth and income inequality’, American Economic Review, 45: 1–28. L af f o n t, J.-J. and Mato uss i, M. (1995), ‘Moral hazard, Wnancial constraints and share-cropping in El Oulja’, Review of Economic Studies, 62/3: 381–99. L am b i , I. N. (1984), The Navy and German Power Politics 1862–1914 (Boston: Allen & Unwin). —— (1963), ‘Free trade and protection in Germany 1868–1879’, Vierteljahrschrift fu¨r Sozial- und Wirtschaftsgeschichte, Part 44. L au x , H.-D. (1989), ‘The components of population growth in Prussian cities, 1875–1905, and their inXuence on urban population structure’, in R. Lawton and W. R. Lee (eds.), Urban Population Development in Western Europe from the Late Eighteenth to the Early Twentieth Century (Liverpool: Liverpool University Press). L e b o v i c s, H. (1967), ‘ ‘‘Agrarians’’ versus ‘‘industrializers’’: Social Conservative resistance to industrialism and capitalism in late nineteenth century Germany’, International Review of Social History, 12: 31–65. L ee , J. Z. and Wang, F. (1999), One Quarter of Humanity: Malthusian Mythology and Chinese Realities, 1700–2000 (Cambridge, Mass.: Harvard University Press). L ee , W. and Mar s c h a l c k, P. (2000), ‘Demographic change and industrialisation in Germany, 1815–1914: Bremen in comparative perspective’, History of the Family, 5/4: 373–90. L ee , W. R. (1977), Population Growth, Economic Development and Social Change in Bavaria 1750–1850 (New York: Arno). —— ed. (1979), European Demography and Economic Growth (London: Croom Helm). —— (1979a), ‘Regionale DiVerenzierung im Bevo¨lkerungswachstum Deutschlands im fru¨hen 19. Jahrhundert’, in R. Fremdling and R. H. Tilly (eds.), Industrialisierung und Raum (Stuttgart: E. Klett). —— (1980), ‘The mechanism of mortality change in Germany 1750–1850’, Medizinhistorisches Journal, 15. —— (1981), ‘Family and ‘‘modernisation’’: the peasant family and social change in nineteenth century bavaria’, in R. Evans and W. R. Lee (eds.), The German Family: Essays on the Social History of the Family in Nineteenth and Twentieth Century Germany (London: Croom Helm). L ee s, A. (1979), ‘Critics of urban society in Germany, 1854–1914’, Journal of the History of Ideas, 40: 61–83. L epsius , R. (1966), ‘Parteinsystem und Sozialstruktur. Zum Problem der Demokratisierung der deutschen Gesellschaft’, in W. Abel (ed.), Wirtschaft, Geschichte und Wirtschaftsgeschichte: Festschrift zum 65. Geburtstag von Friedrich Lu¨tge (Stuttgart: G. Fischer).

References and Sources


Le u n i g, T. and Voth , H.-J. (1996), ‘Did smallpox reduce height? Stature and the standard of living in London, 1770–1873’, Economic History Review, 49/3: 541–60. Le v y, H. (1911), Large and Small Holdings: A Study in English Agricultural Economics (Cambridge: Cambridge University Press). Lewis , W. A. (1954), ‘Development with unlimited supplies of labour’, Manchester School of Economics and Social Studies, 20: 139–92. Li dt k e, V. (1985), The Alternative Culture: Socialist Labour in Imperial Germany (New York: Oxford University Press). Li n d e n l a u b , J. and Ko¨hn e-Li n d e n l a u b , R. (1991), ‘Business Wnancing by Krupp 1811–1848’, German Yearbook on Business History, 9–86. Li n d e r t, P. (1994) ‘Unequal living standards’, in R. Floud and D. McCloskey (eds.), The Economic History of Britain: Volume One: 1700–1860 (Cambridge: Cambridge University Press). —— and Wi l l i a ms o n , J. G. (1980) American Inequality (New York: Academic Press). Li p t o n, M. (1980), ‘Migration from rural areas of poor countries: their impact on rural productivity and income distribution’, World Development, 8: 1–24. Lo u r e n s, P. and Lu c a s s e n, J. (1999), Arbeitswanderung und beruXiche Spezialisierung: die lippischen Ziegler im 18. und 19. Jahrhunderts, Studien zur historischen Migrationsforschung 6 (Osnabru¨ck: Rausch). Lo w, D. (1996), The Egalitarian Moment: Asia and Africa 1950–1980 (Cambridge: Cambridge University Press). Lu b i n s k i, A. (1995), Entlassen aus dem Untertanenverband: Die Amerikaauswanderung aus Mecklenburg-Strelitz (Osnabru¨ck: Rausch). L U¨ t g e , F. (1963), Geschichte der deutschen Agrarfassung von fru¨hen Mittelalter bis zum 19. Jahrhundert (Stuttgart: E. Ulmer). Mc Go u l d r i c k, P. F. and Ta n n e n, M. B. (1977), ‘Did American manufacturers discriminate against immigrants before 1914?’, Journal of Economic History, 37: 723–46. M a d d i s o n, A. (1991), Dynamic Forces in Capitalist Development (Oxford: Oxford University Press). —— (1994), ‘Explaining the economic performance of nations, 1820–1989’, in W. Baumol et al. (eds.), Convergence of Productivity (Oxford: Oxford University Press). —— (1995), Monitoring the World Economy, 1820–1992 (Paris: OECD). Ma n t l , E. (1997), Heirat als Privileg: obrigkeitliche Heiratsbeschra¨nkungen in Tirol und Vorarlberg 1820 bis 1920 (Vienna: Verlag fu¨r Geschichte und Politik). ¨ berseewanderung im 19. Jahrhundert (StuttMa r s c h a l c k, P. (1973), Deutsche U gart: Klett).


References and Sources

M a r s c h a l c k, P. (1987), ‘The age of demographic transition: mortality and fertility’, in K. Bade (ed.), Population, Labour and Migration in 19th and 20th Century Germany (Leamington Spa: Berg). M a s s e y, D. S. (1988), ‘Economic development and international migration in comparative perspective’, Population and Development Review, 14: 383–413. M a t t h e w s , R., Fe i n s t e i n , C. and Od l i n g-Sm e e, J. (1982), British Economic Growth 1856–1973 (Oxford: Oxford University Press). Ma t z , K.-J. (1980), Pauperismus und Bevo¨lkerung: Die gesetzlichen Ehebeschra¨nkerungen in der su¨ddeutschen Staaten wa¨hrend des 19. Jahrhunderts (Stuttgart: Klett-Cotta). M a t z e r a th , H. (1985), Urbanisierung in Preuâen 1815–1914 (Stuttgart: W. Kohlhammer). Me i s e l , F. (1911), ‘Moral und Technik bei der Veranlagung der preußischen Einkommensteuer’, Jahrbuch fu¨r Gesetzgebung, Verwaltung und Volkswirtschaft im Deutschen Reich, 35/1: 285–374. Me i t z e n, A. (1868), Der Boden und der landwirtschaftlichen Verha¨ltnisse des Preußischen Staates (Berlin: Finanzministerium). —— (1885), ‘Ermittelungen u¨ber die durchschnittliche Ho¨he der Grundbuchschulden der ba¨uerlichen Besitzungen in 52 Amtsgerichtsbezirken der Preußischen Staat nach dem Stand des Jahres 1883’, Landwirtschaftliche Jahrbu¨cher, 14/2: 1–32. Me y e r , F. (1938), ‘German internal migration statistics: methods, sources and data’, in D. Thomas (ed.), Research Memorandum on Migration DiVerentials (New York: Social Science Research Council). Mi a s k o w s k i , A. von (1882), ‘Das Erbrecht und die Grundeigentumverteilung im deutschen Reich’, Schriften des Vereins fu¨r Sozialpolitik, 20. M i l w a r d, A. and Sa u l , S. B. (1977), The Development of the Economies of Continental Europe, 1850–1914 (London: Allen & Unwin). M i n i s t r y o f A g r i c u l t u r e , Fi s h e r i e s a n d F o o d ( 1 9 6 8 ) , A C e n t u r y o f A g r i c u l t u r a l S t a t i s t i c s : G r e a t B r i t a i n 1 8 6 6– 1 9 6 6 (L on don: H M SO ) . M i t c h e l l , B. R. (1981), European Historical Statistics 1750–1975, 2nd edn. (London: Macmillan). Mi t t l e r , E. S. (1929), Untersuchungen u¨ber Landarbeiterverha¨ltnisse, Berichte u¨ber Landwirtschaft (Berlin: P. Parey). Mo k y r , J. (1990), The Lever of Riches: Technological Creativity and Economic Progress (Oxford: Oxford University Press). M o m m s e n, W. (1995), Imperial Germany: Politics, Culture and Society in an Authoritarian State (London: Arnold). N a do b n i k , M. (1908), ‘Die Abnahme des durchschnittlichen Heiratsalters in Deutschland’, Zeitschrift des Ko¨niglichen Preussischen Landesamts, 48: 69–84. N eu bu rger , H. and St o k e s , H. H. (1974), ‘German banks and German growth, 1880–1913’, Journal of Economic History, 34: 710–31.

References and Sources


Ne u h a u s, G. (1904), ‘Die ortsu¨blichen Tagelo¨hne gewo¨hnlicher Tagearbeiter in Preussen 1892 und 1901’, Zeitschrift des Ko¨niglich Preuâischen Statistischen Bureaus, 310–46. Ne u m a n n, R. P. (1978), ‘Working class birth control in Wilhelmine Germany’, Comparative Studies in Society and History, 20: 408–28. Ne w b e r y, D. (1977), ‘Risk sharing, share-cropping and uncertain labour markets’, Review of Economic Studies, 23: 187–207. —— (1989), ‘Agricultural institutions for insurance and stabilization’, in P. Bardhan (ed.), The Economic Theory of Agrarian Institutions (Oxford: Oxford University Press). Ni c h t w e i s s, J. (1959), Die ausla¨ndischen Saisonarbeiter in der Landwirtschaft der o¨stlichen und mittleren Gebiete des Deutschen Reichs: Ein Beitrag zur Geschichte des preußisch-deutsch Politik von 1890 bis 1914 (Berlin: Rutten und Loening). Ni c k e l l, S. (1996), ‘The low-skill low-pay problem: lessons from Germany for Britain and the US’, Policy Studies, 17: 17–21. No v e , A. (1969), An Economic History of the U.S.S.R. (London: Allen Lane). Nu g en t , W. (1992), Crossings: The Great Transatlantic Migrations, 1870–1914 (Bloomington: Indiana University Press). Nu r k se , R. (1953), Problems of Capital Formation in Underdeveloped Countries (Oxford: Basil Blackwell). O’B r i e n, P. and Pr a d o s d e l a Es c o s u r a , L. (1992), ‘Agricultural productivity and European industrialisation’, Economic History Review, 45: 514–36. O’R ourk e, K. (1997) ‘The European grain invasions, 1870–1913’, Journal of Economic History, 57/4: 775–801. —— Taylor , A. M. and Wi l l i a m s o n, J. G. (1996), ‘Factor price convergence in the late nineteenth century’, International Economic Review, 37/3: 499–530. —— Wi l l i a m s o n , J. G. and Ha t t o n, T. (1994), ‘Mass migration, commodity market integration and real wage convergence’, in T. J. Hatton and J. G. Williamson (eds.), Migration and the International Labour Market, 1850–1939 (London: Routledge). Ob e r p e n n i g, H. (1998), Migration und Fernhandel im ‘Todden-System’: Wanderha¨ndler aus dem no¨rdlichen Mu¨nsterland im mittleren und no¨rdlichen Europa des 18. und 19. Jahrhunderts (Osnabru¨ck: Rausch). Of f er , A. (1989), The First World War: An Agrarian Interpretation (Oxford: Oxford University Press). Ol s o n , M. (1982), The Rise and Decline of Nations: Economic Growth, StagXation and Social Rigidities (New Haven: Yale University Press). Op p e n h e i me r , F. (1898), Groâgrundeigentum und soziale Frage (Berlin: VITA).


References and Sources

Or s a g h, T. (1969), ‘Lo¨hne in Deutschland 1871–1913: neuere Literatur’, Zeitschrift fu¨r die gesamte Staatswissenschaft, 125: 476–83. P aasche , H. (1883), ‘Die rechtliche und wirtschaftliche Lage des Bauernstandes in Mecklenburg-Schwerin’, in Ba¨uerliche Zustand in Deutschland, vol. 3, Schriften des Verein fu¨r Sozialpolitik, 24. P a s i n e t ti , L. (1962), ‘Rate of proWt and income distribution in relation to the rate of economic growth’, Review of Economic Studies, 29: 267–79. P erk ins , W. A. (1981), ‘The agricultural revolution in Germany’, Journal of Economic History, 10/1: 71–118. —— (1984), ‘The German agricultural worker, 1815–1914’, Journal of Peasant Studies, 11/3: 3–27. —— (1986), ‘Dualism in German agrarian historiography’, Comparative Studies in Society and History, 28/2: 287–306. P errson , T. and Ta b e ll i n i, G. (1994), ‘Is inequality harmful for growth?’, American Economic Review, 84/3: 600–21. P eterso n, C. (1995), Bread and the British Economy: c.1770–1870 (Aldershot: Scholar). P h e l p s Brown , E. H. and Br o w n e , M. (1968), A Century of Pay (London: Macmillan). P l a u l , R. (1986), ‘The rural proletariat: the everyday life of rural labourers in the Magdeburg region, 1830–80’, in R. J. Evans and W. R. Lee (eds.), The German Peasantry (London: Croom Helm). P ogg e von St r an d m a n n , H. (1972), ‘Nationale Verba¨nde zwischen Weltpolitik und Kontinentalpoltik’, in H. Schottelius and W. Deist (eds.), Marine und Marinepolitik im kaiserlichen Deutschland 1871–1914 (Du¨sseldorf: Droste). —— (1988), ‘Germany and the coming of war’, in R. Evans and H. Pogge von Strandmann (eds.), The Coming of the First World War (Oxford: Oxford University Press). —— ed. (1988a), Walter Rathenau: Industrialist, Banker, Intellectual and Politician: Notes and Diaries 1907–1922 (Oxford: Clarendon). —— (1992), ‘The Liberal power monopoly in the cities of Imperial Germany’, in E. L. Jones and J. Rettalack (eds.), Elections, Mass Politics and Social Change in Modern Germany: New Perspectives (Cambridge: Cambridge University Press). —— (1999), ‘Revolutionsfurcht, Sozialreformen und wirtschaftliches Gleichgewicht: Anmerkungen zur Thu¨nenrezeption’, in O. N. Haberl and T. Korenke (eds.), Politische Deutungskulturen: Festschrift fu¨r Karl Rohe (Baden: Nomos). —— and Gei ss , I. (1965), Die Erforderlichkeit des Unmo¨glichen: Deutschland am Vorabend des Ersten Weltkrieges (Frankfurt am Main: Europa¨ische Verlagsanstalt). P oh l , E. (1908), Die Lohn und Wirtschaftsverha¨ltnisse der Landarbeiter in Masuren in den letzten Jahrzehnten, Diss., Ko¨nigsberg.

References and Sources


—— (1984), ‘Forms and phases of industrial Wnance up to the Second World War’, German Yearbook on Business History, 75–94. Po hl , M. (1976), Einfu¨hrung in die deutsche Bankengeschichte (Frankfurt am Main: Knapp). —— (1986), Entstehung und Entwicklung des Universalbankensystems (Frankfurt am Main: Knapp). Po l l a r d, S., ed. (1980), Region und Industrialisierung (Go¨ttingen: Vandenhoeck und Ruprecht). —— (1981), Peaceful Conquest: The Industrialization of Europe 1760–1970 (Oxford: Oxford University Press). Pr a d o s d e l a Es c o s u r a , L. (2000), ‘International comparisons of real product: an alternative data set’, Explorations in Economic History, 37/1: 1–41. Pr e s t, A. R. (1948), War Economies of Primary Producing Countries (Cambridge: Cambridge University Press). Pr o k o po v i t c h, S. (1926), ‘The distribution of national income’, Economic Journal, 36: 69–82. Pr o s t e r m a n, R. and Ri e d i n g e r, J. (1987), Land Reform and Democratic Development (Baltimore: Johns Hopkins University Press). Pu h l e , H.-J. (1972), Von der Agrarkrise zum Pra¨faschismus (Wiesbaden: F. Steiner). —— (1976), ‘Parlament, Parteien und Interessenverba¨nde 1890–1914’, in M. Stu¨rmer (ed.), Das Kaiserliche Deutschland (Du¨sseldorf: Droste Verlag). —— (1986) ‘Peasants and Lords in the Kaiserreich’, in R. Moeller (ed.), Peasants and Land in Modern Germany: Recent Studies in Agricultural History (London: Allen & Unwin). Qu a n t e, P. (1933), ‘Die Flucht aus der Landwirtschaft’, Zeitschrift des Ko¨niglich Preuâischen Statistischen Bureaus, 3: 277–380. —— (1958), Die Abwanderung aus der Landwirtschaft, Kieler Studien 48. —— (1959), ‘Die Bevo¨lkerungsentwicklung der preuâischen Ostprovinzen im 19. und 20. Jahrhundert’, Zeitschrift fu¨r Ostforschung, 8/4: 481–99. R a de t z k i, W. (1930), Die inla¨ndischen landwirtschaftlichen Wanderarbeiter Deutschlands, Arbeiten aus dem Institut fu¨r Wirtschaftslehre der Universita¨t Breslau. R anis , G. and Sc h u l t z , T. P. (1988), The State of Development Economics: Progress and Perspectives (Oxford: Basil Blackwell). R a u c h b e r g, H. (1900), ‘Die Landwirtschaft im deutschen Reich’, Archiv fu¨r soziale Gesetzgebung und Statistik, 15: 554–97. R a u h, M. (1973), Fo¨deralismus und Parlamentarismus im Wilhelminischen Reich (Du¨sseldorf: Droste). —— (1977), Die Parlamentarisierung des Deutschen Reiches (Du¨sseldorf: Droste Verlag).


References and Sources

R avallion , M. and Ch en , S. (1997), ‘What can new survey data tell us about recent changes in distribution and poverty?’, World Bank Economic Review, 11: 357–82. R e h b e i n , F. (1911), Das Leben eines Landarbeiters (Jena: E. Diederichs). R e i c h, U. (1998), Aus Cottbus und Arnswalde in die Neue Welt: Amerikaauswanderung aus Ostelbien im 19. Jahrhundert (Osnabru¨ck: Rausch). R e i f , H., ed. (1994), Ostelbische Agrargesellschaft im Kaiserreich und in der Weimarer Republik (Berlin: Akademie Verlag). R e t al l a c k, J. (1988), Notables of the Right (Boston: Unwin Hyman). R e y n o l ds , L. G., ed. (1975) Agriculture in Development Theory (New Haven: Yale University Press). R ich ardson , H. W. (1965), ‘Over-commitment in Britain before 1930’, Oxford Economic Papers, 17: 237–62. R iehl , W. H. (1854), Die Naturgeschichte des Volkes als Grundlage einer deutschen Sozial-Politik, vol. 1: Land und Leute (Stuttgart: J. G. Cotta). R i e s s e r , J. (1911), The German Great Banks (Washington: National Monetary Commission). R i t t e r , G. (1980), Wahlgeschichtliches Arbeitsbuch: Materialien zur Statistik des Kaiserreichs 1871–1918 (Munich: C. A. Beck). —— and Ten felde, K. (1992), Arbeiter im Deutschen Kaiserreich 1871 bis 1914 (Bonn: Dietz). R oh e, K. (1982), ‘Die ‘‘verspa¨tete’’ Region’ in P. Steinbach (ed.), Probleme der politischen Partizipation im Modernisierungsprozess (Stuttgart: Klett-Cotta). R o¨ hl , J. (1969), Germany without Bismarck: the Crisis of Government in the Second Reich, 1890–1900 (Tu¨bingen: Rainer Wunderlich Verlag). R o l f e s, M. (1976), ‘Landwirtschaft 1850–1914’, in H. Aubin and W. Zorn (eds.), Handbuch der deutschen Wirtschaft- und Sozialgeschichte, vol. 2 (Stuttgart: Klett-Cotta). R o m e r, C. (1986), ‘Spurious volatility in historical unemployment data’, Journal of Political Economy, 94: 1–37. —— (1989), ‘The pre-war business cycle reconsidered: new estimates of gross national product, 1869–1908’, Journal of Political Economy, 97: 1–37. R o s e n b e r g, H. (1985), ‘The pseudo-democratisation of the Junker class’, in G. Iggers (ed.), The Social History of Politics (Leamington Spa: Berg). R o¨ ßl e r , H., ed. (1995), Es zieht eben einer den Anderen nach: Wanderungen und ihre Wirkungen auf ausgewa¨hlte Gebiete Schleswig-Holsteins und Ostelbiens (Scripta Mercaturae Verlag). R o s t a s, L. (1948), Comparative Productivity in British and American Industry, NIESR Occasional Paper no. 13. S a a l f e l d , D. (1963), ‘Zur Frage des ba¨uerlichen Landverlustes im Zusammenhang mit den preuâischen Agrarreformen’, Zeitschrift fu¨r Agrargeschichte und Agrarsoziologie, 11: 163–71.

References and Sources


Sauer , W. (1968), ‘Das Problem des Nationalstaats’, in H. Bo¨hme (ed.), Probleme der Reichsgru¨ndungszeit 1848–1879 (Cologne: Kiepenheuer und Witsch). Sc h a m a , S. (1995), Landscape and Memory (London: HarperCollins). Sc h i e d e r , T. (1961), Das Deutsche Kaiserreich von 1871 als Nationalstaat (Cologne: Westdeutscher Verlag). Sc h i s s l e r, H. (1986), ‘The Junkers: notes on the social and historical signiWcance of the agrarian elite in Prussia’, in R. Moeller (ed.), Peasants and Land in Modern Germany: Recent Studies in Agricultural History (London: Allen & Unwin). Sc h m A¨ d e k e , J. (1995), Wa¨hlerbewegung im Wilhelminischen Deutschland. Vol. 1: Die Reichstagswahlen von 1890 bis 1912: eine historisch-statistische Untersuchung. Vol. 2: Die Reichstagswahlen von 1890 bis 1912: Wahlergebnisse und Strukturen im Kartenbild (Berlin: Akademie Verlag). Sc h m o l l e r , G. (1895), ‘Die Einkommensverteilung in alter und neuer Zeit’, Jahrbuch fu¨r Gesetzgebung, Verwaltung und Volkswirtschaft im deutschen Reich, 19: 1067–94. ¨ ber die Einkommensverteilung’, Jahrbuch fu¨r Gesetzgebung und —— (1895a), ‘U Statistik. —— (1900), Grundriss der allgemeinen Volkswirtschaftlehre (Munich: Duncker und Humblot). Sc h n e p p e, F. (1960), ‘Die Ertragssteigerung auf den leichten Bo¨den der Lu¨neburger Heide’, Zeitschrift fu¨r Agrargeschichte und Agrarsoziologie, 8/1: 14–22. Sc h n i e d e w i n d, K. (1994), Begrenzter Aufenthalt im Land der unbegrenzten Mo¨glichkeiten: Bremer Ru¨ckwanderer aus Amerika 1850–1914 (Stuttgart: Franz Steiner). Sc h o f i e l d , R. S., Re h er , D. and Bideau , A., eds. (1991), The Decline of Mortality in Europe (Oxford: Oxford University Press). Sc h o¨ llgen , G., ed. (1990), Escape into War? The Foreign Policy of Imperial Germany (Oxford: Berg). Sc h o r s k e , C. (1955), German Social Democracy 1905–1917: The Development of the Great Schism (Cambridge, Mass.: Harvard University Press). Sc h o t t , S. (1903), ‘Das Wachstum der deutschen Grossta¨dte seit 1871’, Statistisches Jahrbuch deutscher Sta¨dte, 11: 129–49. Sc h r e m m e r , E. (1988), ‘Faktoren, die den Fortschritt in der deutschen Landwirtschaft im 19. Jahrhundert bestimmen’, Zeitschrift fu¨r Agrargeschichte und Agrarsoziologie, 36/1: 37–77. Sc h u l t z , T. W. (1964), Transforming Traditional Agriculture (New Haven: Yale University Press). Se r i n g, M. (1893), ‘Die Innere Kolonisation im o¨stlichen Deutschland’, Schriften des Verein fu¨r Sozialpolitik, 56.


References and Sources

Sering, M. ed. (1899–1910), Die Vererbung des la¨ndlichen Grundbesitzes im Ko¨nigreich Preuâen, 4 vols., Landwirtschaftliche Jahrbu¨cher, vol. 28, supp. vol. 1; vol. 29, supp. vol. 3; vol. 37, supp. vol. 5; vol. 34, supp. vol. 2. —— ed. (1932), Die Deutsche Landwirtschaft, Berichte u¨ber Landwirtschaft Sonderheft 50 (Berlin: P. Parey). S i e m e n s, W. vo n (1893, trans. 1966), Inventor and Entrepreneur: Recollections of Werner von Siemens (London: Lund Humphries). S i e v e r s , K., ed. (1981), Die deutsche und skandinavische Amerikaauswanderung im 19. und 20. Jahrhunderts (Neumu¨nster: Karl Wachholtz). S k r u b b e lt r a n g, F. (1954), Agricultural Development and Rural Reform in Denmark, FAO Agricultural Studies no. 22. S o e t b e e r, A. (1882), Umfang und Verteilung des Volkseinkommens im Preuâischen Staate, 1872–1878 (Leipzig: Duncker & Humblot). S p e r b e r , J. (1983), ‘The shaping of political Catholicism in the Ruhr Basin, 1848–1880’, Central European History, 16/4: 347–67. —— (1997), The Kaiser’s Voters: Electors and Elections in Imperial Germany (Cambridge: Cambridge University Press). S pree , R. (1988), Health and Social Class in Imperial Germany (Oxford: Berg). S q u i r e, L. (1981), Employment Policy in Developing Countries. (Washington: World Bank). S t a r g a r dt , N. (1994), The German Idea of Militarism: Radical and Socialist Critics 1866–1914 (Cambridge: Cambridge University Press). S t a r k, O. (1991), The Migration of Labour (Oxford: Basic Blackwell). Statistisches Reichsamt (1932), Das deutsche Volkseinkommen vor und nach dem Krieg, Einzelschriften zur Statistik des Deutschen Reichs 24 (Berlin: R. Hobbing). S t e g ma n n, D. (1970), Die Erben Bismarcks: Parteien und Verba¨nde in der Spa¨tphase des wilhelminischen Deutschlands—Sammlungspolitik, 1897–1918 (Cologne: Kiepenheuer & Witsch). S t e i n b a c h, P. (1992), ‘Reichstag elections in the Kaiserreich: the prospects for electoral research in the interdisciplinary context’, in E. L. Jones and J. Rettalack (eds.), Elections, Mass Politics and Social Change in Modern Germany: New Perspectives (Cambridge: Cambridge University Press). S t e l l e r , P. (1903), ‘Die Maschinenindustrie Deutschlands’, Schriften des Vereins fu¨r Sozialpolitik, 106/3. S tern , F. (1961), The Politics of Cultural Despair (Berkeley: University of California Press). S tiglitz , J. (1974), ‘Incentives and risk sharing in share-cropping’, Review of Economic Studies, 41: 219–55. S t U¨ b e l , H. (1935), Staat und Banken im Preuâischen Anleihewesen von 1871–1913 (Berlin: C. Heymann). S t U¨ r m e r , M., ed. (1976), Das Kaiserliche Deutschland (Du¨sseldorf: Droste Verlag).

References and Sources


—— (1990), ‘A nation state against history and geography: the German dilemma’, in G. Scho¨llgen (ed.), Escape into War? The Foreign Policy of Imperial Germany (Oxford: Berg). Su v a l, S. (1985), Electoral Politics in Wilhelmine Germany (Chapel Hill: University of North Carolina Press). Ta y l o r , L. and Anda , P. (1988), ‘Long run income distribution and growth’, in H. Chenery and T. Srinivasan (eds.), Handbook of Development Economics, vol. 1 (Amsterdam: North-Holland). Ta y l o r , A. M. and Wi l l i am s o n , J. G. (1997), ‘Convergence in the age of mass migration’, European Review of Economic History, 1: 27–63. Ta y l o r , L. (1979), Macro Models for Developing Countries (New York: McGraw-Hill). Tc h e r k i n s k y , M. (1922), The Landschaften and their Mortgage Credit Operations in Germany 1770–1920 (Rome: International Institute of Agriculture). Te m p l e, J. (1999), ‘Income Distribution in the Harris–Todaro Model’, manuscript, NuYeld College, Oxford. Te n f e l de , K. (1977), Sozialgeschichte der Bergarbeiterschaft an der Ruhr im 19. Jahrhundert (Bonn: Verlag Neue Gesellschaft). Th ie l , H. (1887), ‘Der Verbrauch von Kunstdu¨nger im Preussischen Staate’, Landwirtschaftliche Jahrbu¨cher, 16: 447–79. Th o m a s, B. (1973), Migration and Economic Growth: A Study of Great Britain and the Atlantic Economy, rev. edn. (Cambridge: Cambridge University Press). Th o m p s o n, A. P. (2000), Left Liberals, the State and Popular Culture in Wilhelmine Germany (Oxford: Oxford University Press). Th U¨ n en , J. H. von (1826), Der Isolierte Staat in Beziehung auf Landwirtschaft und Nationalo¨konomie (Berlin: Wiegandt, Hempel und Parey). Ti ll y , R. (1986), ‘German banking 1850–1914: development assistance for the strong’, Journal of European Economic History, 15: 113–52. Ti pt o n, F. R. (1976), Regional Variations in the Economic Development of Germany in the Nineteenth Century (Middletown, Conn.: Wesleyan University Press). To da r o , M. P. (1969), ‘A model of labour migration and urban unemployment in less developed countries’, American Economic Review, 59: 138–48. —— (1997), Economic Development (New York: Longman). Tr a c e y, M. (1982), Agriculture in Western Europe: Challenge and Response, 1880–1980 (London: Granada). Twose , A. (2001), Statistics and the German State, 1900–1945: The Making of Modern Economic Knowledge (Cambridge: Cambridge University Press). US Department of Commerce (1975), Historical Statistics of the United States: Colonial Times to 1970. US Immigration Commission (1909–10), Immigrants in Industries, 61st Congress, 2nd Session, Senate Documents, no. 633, vols. 68–85.


References and Sources

US Immigration Commission (1910–11), Statistical Review of Immigration 1820–1910, 61st Congress, 3rd Session, Senate Documents, no. 756, vol. 20. V a n Za n d e n, T. (1991), ‘The Wrst green revolution’, Economic History Review, 44: 643–64. Verein fu¨r Sozialpolitik (1893), ‘Verhandlungen des 20. und 21. Ma¨rz 1893’, Schriften des Vereins fu¨r Sozialpolitik. V o e l c k e r, A. (1883), ‘Manure’, Encyclopaedia Britannica, 15: 505–12. V o¨ g e l e , J. (1998), Urban Mortality Change in England and Germany, 1870–1913 (Liverpool: Liverpool University Press). —— and Wo e l k, W., eds. (2000), Stadt, Krankheit und Tod: Geschichte der sta¨dtischen Gesundheitsverha¨ltnisse wa¨hrend der Epidemiologischen Transition (Berlin: Duncker & Humblot). Wa d e , R. (1990), Governing the Market: Economic Theory and the Role of Government in East Asian Industrialisation (Princeton: Princeton University Press). Wa g n e r , A. (1871), Rede u¨ber die Soziale Frage. Gehalten am 12 Oktober 1871 (Berlin: Wiegandt und Grieben). —— (1902), Agrar- und Industriestaat (Jena: G. Fischer). —— (1904), ‘Zur Methodik der Statistik des Volkeinkommens und Volksvermo¨gens’, Zeitschrift des Ko¨niglich Preuâischen Statistischen Bureaus, 2: 41–122. —— (1904a), ‘Weitere statistische Untersuchungen u¨ber die Verteilung des Volkseinkommen in Preuâen’, Zeitschrift des Ko¨niglich Preuâischen Statistischen Bureaus, 4: 229–67. Wa g n e r , C., trans. G. Skelton (1978–80), Cosima Wagner’s Diaries (London: Collins). Wa g n e r , P. (1896), Die Steigerung der Rohertra¨ge der Landwirtschaft im Laufe des 19. Jahrhunderts, Diss., Jena. Wa l k e r, M. (1964), Germany and the Emigration, 1816–1885 (Cambridge, Mass.: Harvard University Press). W e b b , S. (1982), ‘Agricultural protection in Wilhelmian Germany: forging an empire with pork and rye’, Journal of Economic History, 42: 309–26. We b e r , A. (1908, trans. 1976), ‘From ‘‘Die Grossstadt und ihre soziale Probleme’’ ’, in A. Lees and L. Lees (eds.), The Urbanisation of European Society in the Nineteenth Century (Lexington: Heath). We b e r , A. F. (1899), The Growth of Cities in the Nineteenth Century: A Study in Statistics (New York: Macmillan). We b e r , M. (1893), ‘Die Lage der Landarbeiter im ostelbischen Deutschland’, Schriften des Verein fu¨r Sozialpolitik, 55. —— (1894, trans. 1979), ‘Developmental tendencies in the situation of East Elbian rural workers’, Archiv fu¨r soziale Gesetzgebung und Statistik, 7: 1–41, translated in Economy and Society (Berkeley: University of California Press).

References and Sources


—— (1948), ‘Capitalism and rural society in Germany’, in H. H. Gerth and C. Wright Mills (eds.), From Max Weber: Essays in Sociology (London: Kegan Paul). —— (1993), Landarbeiterfrage, Nationalstaat und Volkswirtschaftspolitik, ed. W. Mommsen, Vol. 4 of M. Weber, Gesamtausgabe Abteilung I: Schriften und Reden, 1892–1899. We g g e , S. (1997), Migration Decisions in Mid-Nineteenth Century Germany, Diss., Northwestern University, Evanston, Illinois. —— (1998), ‘Chain migration and information networks: evidence from nineteenth century Hesse-Cassel’, Journal of Economic History, 58/4: 957–86. W e h l e r , H.-U. (1969), Bismarck und der Imperialismus (Cologne: Kiepenheuer & Witsch). —— (1985), The German Empire 1871–1918 (Leamington Spa: Berg). —— (1995), Deutsche Gesellschaftsgeschichte. Vol. 3: 1849–1913 (Munich: C. Beck). We l l ho¨ n e r , V. (1989), Groâbanken und Groâindustrie im Kaiserreich (Go¨ttingen:Vandenhoeck und Ruprecht). We n c k s t e r n , H. von (1911), Existenz-Bedingungen seâhafter Landarbeiter, Diss., Rostock. We n n em a n , A. (1997), Arbeit im Norden: die Italiener im Rheinland und Westfalen des spa¨ten 19. und fru¨hen Jahrhunderts, Schriften des IMIS 2 (Osnabru¨ck: Rausch). Wi l l i a ms o n , J. (1988), ‘Migration and urbanisation’, in H. Chenery and T. Srinivasan (eds.), Handbook of Development Economics, vol. 1 (Amsterdam: North-Holland). Wi l l i a ms o n , J. G. (1985), Did British Capitalism Breed Inequality? (Boston: Allen & Unwin). —— (1991), Inequality, Poverty and History (Cambridge, Mass.: Basil Blackwell). —— (1995), ‘The evolution of global labour markets since 1830: background evidence and hypotheses’, Explorations in Economic History, 32: 141–95. —— (1997), Growth, Distribution and Demography: Some Lessons from History, NBER Working Paper no. 6244. Wi n k l e r , J. (1993), Sozialstruktur, politische Traditionen und Liberalismus: eine empirische La¨ngsschnittstudie zur Wahlentwicklung in Deutschland, 1871–1933 (Opladen: Westdeutscher Verlag). Witt , P.-C. (1970), Die Finanzpolitik des Deutschen Reiches von 1903 bis 1913 (Hamburg: Matthiesen). —— (1973), ‘Der preußische Landrat als Steuerbeamter 1891–1918: Bemerkungen zur politischen und sozialen Funktion des deutschen Beamtentums’, in I. Geiss and B. Wendt (eds.), Deutschland in der Weltpolitik des 19. und 20. Jahrhunderts (Du¨sseldorf: Bertelsmann Universita¨tsverlag).


References and Sources

Witt , P.-C. (1985), ‘The Prussian Landrat as tax oYcial, 1891–1918’, in G. Iggers (ed.), The Social History of Politics (Leamington Spa: Berg). World Bank (1994), The East Asian Miracle: Economic Growth and Public Policy, World Bank Policy Research Report. Wrigley , E. A. (1961), Industrial Growth and Population Change: A Regional Study of the CoalWeld Areas of North-West Europe in the Later Nineteenth Century (Cambridge: Cambridge University Press). W y g od z ins k i , W. (1917), Die Landarbeiterfrage in Deutschland (Tu¨bingen: J. C. B. Mohr).

INDEX absent revolution thesis 23 agrarian reform 45 and aristocracy 34, 36 and banking system 32–3 in central and north-west Germany 31–3 and child care 141 and compensation 24, 41–2 and consolidation of holdings 44 and credit 25, 32–3, 35 and demographic pressures 26 and Denmark 26, 28–31 and difficulty of 23 east of the Elbe 34–8 aristocracy’s role 36 distance from industrial areas 36 limited credit 35 peasants’ weak position 36 poor communications 34–5 success of 36–8 supposed failure of 34 and effects of 191 egalitarianism 53 and farm prices 26–8 and infant mortality 141 and labour service 31–2, 39, 41 and land ownership distribution 320 and landlords’ land gains 32, 33 and market access 25–6 and objectives of 24, 28 and population growth 141 and productivity 24–5, 29 in Prussia 31 cultivated land area 40 peasant land losses 38–40 regional variation in impact of 40–3 and regional variations in 43

and social costs of 43 and Sonderweg school 23 in southern Germany 44–5 and standards of comparison 23 and statistical analysis of 23 as test of political process 23–4 agriculture: and contemporary views of 181 and cultivated land area 40 and debt levels 245–6 and demesne-farming 37–8 and economic development 15 and the economy: decline of importance in 218–25 domestic demand 219, 220, 221, 223 international trade 216–23 protection 217–18, 221, 222, 224 and education 226 and emigration 81–2, 85–8 and enclosure 32, 34, 41–2 and farm size 48–9, 51–2 and immigration 92, 93 and impact of capitalist values 12 and industrialization 17, 185–6 consequences of 187 impact of 215–18 and institutional problems with 181 and Kuznets Curve 53–4 and labour contract system 183, 185, 208 and labour force 218 employed 46–7 productivity increase 246–7 self-employed 47



agriculture: (Continued) and labour surplus 193, 209 and land ownership distribution 49–53 tenancies 52–3 and Lewis Model 207–9 and migration 5, 6, 12–13, 102–4, 106, 109 concerns over 190–2 consequences of 187 costs of 186–7 large estates 192, 193, 194 market conditions 104–5, 107–8, 113 population density 249–50, 251–2 seasonal labour 74–5 as symptom of social problems 191 and missing market hypothesis 184 and mutual obligations 186–7 and owner-occupation 46, 47 and payments-in-kind 182–3 and population changes 199–201 and prices 104–5, 220–1, 222 fall in 197 and productivity 16–17, 24–5, 29, 215–16 artificial fertilizers 234–5, 239–40, 242, 244 breeding improvements 245 compared with France 46 impact of increase 216 improvements in 218, 223–4 international comparisons 230–4 labour force 246–7 labour-saving technology 247–8 large estates 241–2 livestock intensity 229–30, 242, 245 migration 113, 123, 237, 246 regional patterns of 235–45 soil quality 242–4

sugar beet cultivation 239–41 yield improvement 227–9, 239, 244 and quit rate from 148–51, 171–3, 197–9 farm size 149–50, 171, 197 geographic variations 150–1 population growth 173 and structure of: fertility decisions 142–3, 169 migration 55, 107, 113 and technological improvements 201, 216, 225–6, 234–5, 242–5 labour-saving 247–8 and transport costs 25–6 and underemployment 12, 16, 185–8, 204–7, 214, 247–9, 273 partible inheritance 205–6 and wages: influences on 201–4, 207–8, 211–13 rising level of 196–7 and Weber (Max) on 193–7 see also agrarian reform Argentina, and emigration to 81 aristocracy, and agrarian reform 34, 36 Army Bill (1913) 332 artificial fertilizers, and productivity 234–5, 239–40, 242, 244 ‘Asian values’, and industrialization 4 Atkinson, A B 299 Australia 216 Austria-Hungary, and migration to Germany 91 authoritarianism, and industrialization 4–5 Baden, and agrarian reform 44 Bairoch, P 46, 230, 232

Index banking system: and agrarian reform 32–3 Denmark 30 and consolidation of 257 and expansion of 185 and investment 15, 254 financing of 257–9 and rural credit 25 and underdevelopment of 320 Bauernschutz 31, 36 Bavaria, and agrarian reform 44 Belgium, and immigration 80 Berlin, migration to 63–5 and further moves 71–3 and migrant occupations 66–7 and sources of migrants 68–70, 75 Berlin Statistical Bureau 98 Berliner Handelgesellschaft 257 birth rate 127 and cities 153–4 and decline in 317 and economic conditions 142 and less-developed countries 317 and migration 106–7 and rural-urban differences 153 Bismarck, Otto von 346 Blackbourn, D 3 n4 Brandenburg, and agrarian reform 41, 42, 43 Brazil 216 and industrialization 5 and land ownership distribution 315 and soya-bean production 314 Broadberry, S N 233, 234 Broesike, Dr M 98 Browne, M 263 bureaucracy, and unrepresentativeness of 2


Canada, and emigration to 81 capital: and mobility of, and Lewis Model 318–19 and ownership distribution 312, 313 Catholicism: and demographic behaviour 154, 168–9 and population growth 146, 155–6, 174 and Social Democratic Party support 337–9 child care, and agrarian reform 141 Chile, and industrialization 4 China, and labour surplus 352, 353 cities: and birth rates 153–4 and contemporary views of 188–9 and death rates 154–5, 157–8, 174–5 and employment 268 and fertility rates 145–6, 152, 169 and impact of agricultural decline 217–18 and inequality 323–4, 325–6 and life expectancy 157–8, 283 and migration to 63–6, 111–12 congestion effect 275 further moves 71–3 migrant occupations 66–7 sources of 68–70 and opportunities in 160 and population growth 151–6, 173–7 and rate of growth of 59, 61–2, 151, 293 and skill levels 312 see also urbanization Clapham, J H 31, 225–6, 239



Cologne: and migrant occupations 66 and migration to 65–6 communications: and agrarian reform 34–5 and improvements in 157 and migration 113 compensation, and agrarian reform 24, 41–2 Conrad, J 44 consolidation, and agrarian reform 44 constitution, and election of 1912 332 corporations, and investment 254 Craigie, P G 48–9 credit: and agrarian reform 25, 32–3, 35 and Danish agrarian reform 30 and expansion of 185 Darmsta¨dter Bank 257 Dasgupta, Partha 184 n5, 186, 187 death rates 127 and cities 154–5, 157–8, 174–5 and economic conditions 131 and economic development 156–7 and migration 158 and migration analysis 71, 106 and wages 154–5, 170 debt, and agriculture 40 n44, 245–6 demography: and birth rates 127 cities 153–4 decline in 317 economic conditions 142 less-developed countries 317 rural-urban differences 153 and comparison of east and south 147 fertility rates 137 marriage and migration 137–9 marriage rates 134–7 migration movements 139–41

non-agricultural employment 132–4 and death rates 127, 131 cities 154–5, 157–8, 174–5 economic development 156–7 wages 154–5, 170 and demographic transition 124 regional variations 127–8 and economic factors: influence of 130–1 non-agricultural employment 132–4, 172–3 and fertility rates 127, 131, 137 agricultural institutions 142–3 decline in 159 factors affecting 144, 146, 147, 152–3 large estates 146, 169 partible inheritance 143, 144–5, 169, 318 proximity to cities 145–6, 152, 169 in rural Prussian Kreise 142–5, 167–70 and infant mortality 141 and life expectancy 283 international comparisons 158 regional variations 157–8 and marriage rates 131 economic conditions 131, 137, 177–80 migration 137–9 proximity to cities 152 regional variations 134–7 and migration 112, 121–2, 124–5, 160 birth rate 106–7 death rates 106 population growth 129–30 regional variations 108–10 and population growth 126–7 Catholicism 146, 155–6, 168–9, 174

Index in cities 151–6, 173–7 delayed marriage 177–8 industrial structure 156, 174, 175 industrialization 141 inequality 147–8 Lewis Model 156–60 regional variations 128–30, 132–4 rural migration 148–51, 170–3 rural regulation of 142, 159, 317 wages 175 and population movements 125–7 Denmark, and agrarian reform 26, 28–31 and agricultural productivity 29 and context of 31 and financing of 30 and nobility 28–9 and non-legislative aspects 29 and peasants 30–1 and state initiative 28 and unequal benefits 30 Deutsche Bank 257 development economics: and relative backwardness 3–4 see also Lewis Model Disconto-Gesellschaft 257 Dresdner bank 257 economic development: and agriculture 15 as alteration of production function 310–11 and death rates 156–7 and financial sector 254 and inequality 14, 295 and investment 253–4 and land ownership distribution 315 and ‘leapfrogging’ 264–5 and migration 6, 112, 122–3, 350 and model of 294


economy: and agriculture: decline in importance of 218–25 domestic demand 219, 220, 221, 223 industrialization 17, 185–6, 187, 215–18 international trade 218–21, 222–3 prices 220–1, 222 protection 217–18, 221, 222, 224 and consequences of migration: skill endowments 279–81 wage convergence 274–9 and economic stabilization 351–2 and migration and economic cycles 271–4, 291–2 and structural changes in 255–66 Anglo-German rivalry 263–4 export growth 255–6, 260–1, 265–6 financial sector 257 growth potential 265 investment 256–7 investment finance 257–9 new industries 264–5 raw material costs 262–3 recovery from recession 259–60 terms of trade 261–2, 264 wages 263–4 education: and agriculture 226 and income distribution 311–12 and industrialization 215–16 Eggers, Count von 30–1 election of 1912: and foreign policy 333 and political consequences of 332–3 and Social Democratic Party support: agricultural population 341 aristocratic estates 342–3



election of 1912: (Continued) candidate composition 343–5 city size 337, 339 desire for change 343 econometric analysis 356–60 economic conditions 340, 342, 343 feelings of political inequality 343 increase in 331–2 inequality 340, 342 population mobility 340 proximity to towns/cities 341–2 religious affiliation 337–9, 341 rural/small town areas 340–3 and statistical analysis, limitations of 336–7 electoral politics: and radical right 334 and urbanization 335–6 and voting trends 333–4 see also election of 1912 Elektroboom 256, 259 Eley, G 1 n1 elites, and unwillingness to modernize 2 emancipation, and population growth 142 emigration: and agricultural character of 85–8 and decline in 89 and destination countries 81 and Germans living abroad 80–1 and labour surplus 56–7, 89 and regional variations in 85–6 and structural change in 80 to United States 81, 83–4, 321 agricultural employment 88–9 decline in 85 influence of agriculture 81–2 previous occupations 86–8 wage ratio 85 see also immigration; migration

employment: and agriculture 46–7, 218, 246–7 and export-oriented industries 266–7 and household 206, 212, 214, 249 and migrant participation rates 273 and non-agricultural 132–4, 172–3, 199 industrial concentration 268–9, 286–8 regional pattern of 267–8, 271 vertical integration 268, 269–71, 288 and self-employment 47, 206, 214 and sugar beet cultivation 201 see also underemployment enclosure 32, 41–2 and aristocracy 34 and Danish agrarian reform 29, 30 Engel, Ernst 98, 190, 297–8, 305 Engels, Friedrich 297 Evans, Richard 330–1 exports: and finished goods 261 and growth in 255–6 and terms of trade 10, 254 expropriation, and agrarian reform 24 famines 157 farm size: and agriculture 48–9, 51–2 and productivity 241–2 and quit rate from agriculture 149–50, 171, 197 and wages 201–2, 203, 213 see also large estates Feinstein, Charles 296, 307 fertility rates 127 and agricultural institutions 142–3 and decline in 159 and delayed marriage 131, 159, 177–8 and economic conditions 131

Index and factors affecting 144, 146, 147, 152–3 and large estates 146, 169 and partible inheritance 143, 144–5, 169, 318 and Polish-speaking areas 145, 146, 147, 167–8 and proximity to cities 145–6, 152, 169 and regional variations in 137 in rural Prussian Kreise 142–5, 167–70 and wages 146, 147 see also farm size Finance Bill (1913) 332 financial sector: and development of 254, 257 and financing of investment 257–9 and underdevelopment of 320 Finckenstein, Finck von 37 n35 Fircks, A von 131 foreign direct investment 320 foreign policy: and election of 1912 333 and industrialization 5 and internal politics 1 and naval expansion: economic motives for 347–8 lack of alternatives 348–9 and social context of 2–3 France: and agriculture: farm size 48–9 owner-occupation 46 productivity 46 and urbanization 59 Fremdling, R 233 Geisenberger, S 299, 300 gender, and migration 76–9, 137–8 see also women Gerschenkron, Alexander 3 Gilbert and Lawes 244


Goltz, Theodor von der 57 n2, 191 and Prussian land reform 34 Gutsherrschaft administrative system, and agrarian reform east of the Elbe 36 Hajnal, J 178 Handdienst, and agrarian reform 41 Hayami, Y 318 Helfferich, K 298, 299 Hesse, and agrarian reform 44 historical school, and income distribution 297–8 historiography: and economic models 18–19 and German exceptionalism 19–20 of Imperial Germany 1–2, 18, 331–2 and Tirpitz naval plan 350 see also Kehrite school Hoffmann, Walter 9, 37 n36, 104, 218, 235, 257, 300 household behaviour, and fertility decisions 318 household employment: and underemployment 206, 214, 249 and wage levels 212 illegitimacy 139 and death rates 142 immigration 90–3 and agriculture 92, 93 from Austria-Hungary 91 and countries of origin 91 and increase in 90 and industrial workers 92, 93 and labour surplus 57 and net gains from 80 and regional variations in 92–3 from Russia 90, 91



Imperial Germany: and Anglo-German relations 263–4, 346 and democratic transition 345–6 economic stabilization 351–2 potential for 353–4 and electoral politics of 331–5 growth of Social Democratic Party 335–45 and German exceptionalism 19–20 and industrialization 3, 331–2 standards of comparison 20–1, 332 success in coping with 353 and internal weakness of 1–3, 330, 349 and naval expansion: economic motives for 347–8 lack of alternatives 348–9 naval weakness 347 and problems faced by 333 and socio-strategic dilemma of 346–7 Imperial Statistical Office 98 income distribution, see inequality incomplete revolution thesis 23 industrial structure: and birth rates 155 and death rates 154–5 and inequality 326 and migration 266–71 export-oriented industries 266–7 industrial concentration 268–9, 286–8 patterns of non-agricultural employment 267–8, 286–90 vertical integration 268, 269–71, 288 and population growth 156, 174, 175 industrialization: and adoption of capitalist values 12 and agriculture 17

eastern 185–7 impact on 215–18 structure of 55 and authoritarianism 4–5 and contemporary concerns with 189–90 and education 215–16 and foreign policy 5 and German debate on 295 and inequality 6, 13–14, 297–8, 305, 307–8, 316, 320 and internal weakness of Imperial Germany 3, 331–2 and Kuznets Curve 13–14, 53, 295 and Lewis Model 6–11 and migration 5–6, 7, 56, 94 periods of 71 sources of 75–6 and nature of German 3, 20–1 in pioneer countries 20 and population growth 141 and productivity growth 350 and relative backwardness 3–4 and rural society 185–6 consequences for 187 and speed of 59, 61, 62, 151, 293, 353 and standards of comparison 20–1, 332 and trade unions 7 and urbanization 15 inequality: and avoidance of 326–7 and cities 323–6 and contemporary concerns with 190 and debate over 297–9 and economic development 14, 295 and electoral politics 340, 342, 343 and historical school 297–8 and industrial structure 326 and industrialization 6, 13–14, 297–8, 305, 307–8, 316, 320

Index and international comparisons 306–8 and Kuznets Curve 13–14, 53, 295 and labour surplus 310 and land ownership distribution 49–53, 313–16, 324 and Lewis Model 6–11, 317 and migration 53, 56, 309–10, 316, 322, 325 and population growth 147–8 and productive knowledge 311–13 skills 311, 316, 321 vocational training 311–12, 321 and proximity to Russian Poland 324 in Prussia, nineteenth century 303–6 and rural society: divergences in 22 rural-urban gap 13–14, 54–5, 108 and savings’ concentration 317 and skill endowments 311–12, 315–16, 321 and sources of 319–21 analysis by simulation 322–3 econometric analysis 323–6 and tax data: opportunity provided by 303 problems with use of 299–302 and terms of trade 320, 323, 326 and urbanization 316, 323, 326 and worsening of 302–3 infant mortality: and agrarian reform 141 and illegitimate children 142 inheritance, see partible inheritance insurance 289–90 internal politics: and foreign policy 1 and industrialization 3, 331 and primacy of 1


international trade: and agriculture 216–17, 218–21, 222–3 and Anglo-German rivalry 263–4, 346 and export growth 255–6, 260–1, 265–6 employment 266–7 and German naval policy 347–8 and impact of 254–5 and new industries 264–5 and raw material costs 262–3 and terms of trade 10, 254, 261–2, 264 investment: and banking system 254 and capital mobility 319 and corporations 254 and economic development 253–4 and financial sector 254 and financing of 257–9 and improved opportunities for 313 and inequality 321 and institutional constraints on 320 and pattern of 256–7 and savings 253–4 Ipsen-Ko¨llmann thesis 141 Ireland, and land ownership distribution 315 Italy, and migration to Germany 90 Japan: and agrarian reform 26 and agricultural prices 27 and land ownership distribution 315 joint-stock companies, and investment 254 Junkers, and opinions of 226 n14



Kehrite school: and election of 1912 333 and internal weakness of Imperial Germany 1–2 and key concept of 1 and radicalization of the right 332 and Tirpitz naval plan 350 and weakness of argument 353 Kindleberger, C 56 n1 Knapp, Gustav 189 and migration from rural east 191–2 and peasant land losses 38 and Prussian land reform 34 Knodel, J 137, 142 knowledge, productive 311–13 and skills 311, 316, 321 and vocational training 311–12, 321 Ko¨llmann, W 135, 139, 152 and Ipsen-Ko¨llmann thesis 141 Koselleck, R 34 Kuczynski, Robert 98 Kuznets, Simon 13, 295, 312, 316–17 Kuznets Curve 4 and agriculture 53–4 and assumptions of 22 and conditionality of 326, 327 as contemporary fiction 296 and disputing of 296 and escaping from 327 as German peculiarity 296, 297 as historical fact 296, 297 and inequality and industrialization 13–14, 53, 295 and Lewis Model 294, 296 and migration 295, 322 in Prussia, nineteenth century 303–6 international comparisons 306–8

labour market: and ‘Continental’ nature of 80, 94 and impact of capitalist values 12 and industrial development 6–8 and integration of 276 and non-agricultural employment, and migration 132–4, 172–3, 199 and population increase 127 see also employment labour service: and abolition in Denmark 29 and agrarian reform 41 abolition of 31–2, 39 and fertility decisions 142–3 and inefficiency of 38 n38 labour surplus: and criticism of concept of 16, 215 and eastern agriculture 193, 209 and emigration 56–7, 89 and immigration 57 and impact of capitalist values 12 and implications of 294 and inequality 310 and Kuznets Curve 13–14 and Lewis Model 6–9, 294 and migration 56, 93–4, 97, 100, 102, 103–4, 141 and political stability 353 and sources of 293 and underemployment 204–7, 249–50 land ownership: and agrarian reform 320 and Danish agrarian reform 29 and distribution of 49–53 and income distribution 313–16, 324 and marriage rates 143 and political power 314–15 and re-allocation of 315–16 and skill acquisition 315

Index and the state 314 and turnover of 39–40 land prices 104–5, 106 and migration 107–8 land reform, see agrarian reform land tax assessment 244 Landesrentenbank 32, 33 Landflucht 34 land-owners, and adoption of capitalist values 12 Landschaften 25, 33 large estates: and economies of scale 184 and fertility rates 146, 169 and migration 192, 193, 194 and productivity 241–2 and quit rate from agriculture 149–50, 171 and wages 201–2, 203, 213 Laux, H-D 151 n39 ‘leapfrogging’, and economic development 264–5 Lee, W 141 less-developed countries, and birth rates 317 Levy, H 36 n30 Lewis, Arthur 156–7, 317 and employment 12 and the Lewis Model 6 Lewis Model 4 and capital mobility 318–19 and criticisms of 15–16 and eastern agriculture 207–9 and explanatory value of 350 and inequality 6–11, 317 and Kuznets Curve 294, 296 and labour supply 6–8 and labour surplus 6–9, 294 and migration 79–80, 97, 100 and modification to 9 and population growth 156–60 and propositions of 294 and shortcomings of 11


and ‘turning point’ 7, 8, 294 and wages 9–11 Liebig, Justus von 244, 245 life expectancy: and international comparisons 158 and migration 283 and regional variations in 157–8 Lindert, Peter 296, 307 living standards: in eastern Germany 139–41 and regulation of population growth 159 and wage levels 212–13 local government, and Social Democratic Party 332 Lupitz system 244, 245 Magdeburg, and peasant farmers 33 Malthus, Thomas 178 market access, and agrarian reform 25–6 marriage: and decline in fertility rates 159 and delayed 131, 159, 177–8 and economic conditions 131, 137, 177–80 and land ownership 143 and migration 137–9 and partible inheritance 145 and proximity to cities 152 and regional variations in 134–7 and wages 178–80 Marschalck, P 141 Marx, Karl 6 Mecklenburg, and agrarian reform 45 medicine 157 Meitzen, Adolph: and peasant debt 40 n44 and Prussian land reform 34 Meldepflicht system 57 middle class, and cities 151, 160



migration: and agriculture 5, 6, 12–13, 102–4, 106, 109 concerns over 190–2 large estates 192, 193, 194 market conditions 104–5, 107–8, 113 productivity 113, 123, 237, 246 quit rate from 148–51, 171–3, 197–9 structure of 55, 107, 113 and demographic factors 112, 121–2, 124–5, 160 birth rate 106–7 death rates 106, 158 population growth 129–30 regional variations 108–10 and economic cycles 271–4, 291–2 and economic development 350 and economic factors 112, 122–3 non-agricultural employment 132–4, 172–3, 199 and factors affecting attitude towards 188 and gross migration 57 and historical research 97 and impact of 19 and industrial structure 266–71 export-oriented industries 266–7 industrial concentration 268–9, 286–8 patterns of non-agricultural employment 267–8, 286–90 vertical integration 268, 269–71, 288 and industrialization 5–6, 7, 56, 94 and inequality 53, 56, 309–10, 316, 322, 325 rural-urban gap 13–14, 54–5, 108 and international 56–7 and international comparisons 58 and intra-urban 56

and Kuznets Curve 295, 322 and labour surplus 56, 93–4, 97, 100, 102, 103–4, 141 and Lewis Model 79–80, 97, 100 and marriage patterns 137–9 and measurement of 62–8 and net migration 57 and periods of 71, 94, 139 and population density 249–50, 251–2 and population mobility 63–6 migrant occupations 66–7 and regional variations in 101–2 and rural population growth 148–51, 170–3 and rural-urban 56, 102–4, 110–13 concerns over 190–2 consequences of 187 costs of 186–7 as symptom of social problems 191 and seasonal 56, 64–5, 74–5 by sex 76–9, 137–8 and skill endowments 279–81 vocational training 311–12 and social change 350–1 and sources of migrants 68–76 agriculture-related 74–5 further moves 71–3 industrial-related 75–6 and statistical analysis of: changing nature of (1871–1913) 100–6 net migration by region (1871–1910) 106–10, 115–16 panel data analysis (1871–1910) 116–19 regression variables and sources 114–15 rural-urban (1890 census) 110–13, 119–23 and statistical sources for 98 and underemployment 214, 249

Index and unemployment 271–4 and urbanization 59–62, 111–12 and wages 56, 94, 111 wage convergence 274–9 wage dispersion 276 and women 76–9, 138–9 life expectancy 283 wages 281–3 see also emigration; immigration; Lewis Model missing market: and industrialization 185–6 and share-cropping 184 Mittelstand, and preservation of 189 mobility 63 and cities 63–6 and migrant occupations 66–7 and sex of migrants 76–9 and sources of migrants 68–76 modernization, and incompleteness of 2 Mommsen, Wolfgang: and German exceptionalism 19 and ‘skirted decisions’ 1 n2 monitoring costs, and sharecropping 183–4 Mu¨ller, J H 299, 300 Newbery, D 184 n4 newly industrializing countries (NICs) 4 nobility, and Danish agrarian reform 28–9 nutrition, and indicators of 139–40 O’Brien, P 233 occupations, and migration 66–7 Ostmarkverein 193 Pakistan, and land reform 46 n53 partible inheritance: and fertility decisions 143, 144–5, 169, 318


and marriage rates 145 and underemployment 205–6, 214 patents, and productive knowledge 311, 312 peasants: and agrarian reform: bias in favour of 32, 320 credit availability 32–3 Denmark 28, 30–1 east of the Elbe 36 labour service 31–2, 39 land losses 38–40 and demesne-farming 37–8 and efficiency of 16 and non-agricultural earnings 44–5 and preservation of 189 Phelps-Brown, E H 263 Philippines, and agrarian reform 25 n5 Poland, see Russian Poland Polish-speakers: and fertility rates 145, 146, 147, 167–8 see also Russian Poland Pomerania: and agrarian reform 41, 43, 45 and large estate failures 40 and peasant landholdings 38 Posen, and agrarian reform 41, 42 Prados de la Escosura, L 233 prices: and agrarian reform 26–8 and agriculture 104–5, 197, 220–1, 222 and death rates 131 and marriage rates 178–9 and wages 212 production, and economic development 310–11



productivity: and agrarian reform 24–5, 29 and agriculture 16–17, 24–5, 29, 215–16 artificial fertilizers 234–5, 239–40, 242, 244 breeding improvements 245 compared with France 46 impact of increase 216 improvements in 218, 223–4 international comparisons 230–4 labour force 246–7 labour-saving technology 247–8 large estates 241–2 livestock intensity 229–30, 242, 245 migration 113, 123, 237, 246 regional patterns of 235–45 soil quality 242–4 sugar beet cultivation 239–41 yield improvement 227–9, 239, 244 and industrialization 350 and wages 7, 9–11 see also knowledge, productive profits, and industrial development 6 Prokopovitch, S 13, 295 property rights: and agrarian reform 24 and evolution of 315–16 and exceptional re-allocation of 315 and productive knowledge 312–13 protection, and agriculture 217–18, 221, 222, 224 Prussia: and agrarian reform 31, 42 consolidation 44 peasant land losses 38–40 regional variation in impact of 40–3 and agrarian reform east of the Elbe 34–8

aristocracy’s role 36 distance from industrial areas 36 limited credit 35 peasants’ weak position 36 poor communications 34–5 success of 36–8 supposed failure of 34 and agricultural prices 27–8 and rural credit 25 see also taxation data, Prussian Prussian Statistical Bureau 98 public health 157 Quante, Peter 148 railway construction, and migration 75–6 Rauh, Manfred 354 recruits, and height of 139–40 regional government, and Social Democratic Party 332 regions 94–6 Reichstag, and constitutional position of 332–3 relative backwardness: and development economics 3–4 and industrialization 3 religious affiliation, and voting behaviour 337–9, 341 Rentenbanken 33 ‘revolutions from above’ 23 Rhineland, and agrarian reform 41 Royal Credit Bank (Denmark) 30 Ruhr, and sources of migrants 75 rural society: and efficiency wages and unemployment 187 and fertility rates 142–5, 167–70 and impact of capitalist values 12 and industrialization 185–6 consequences of 187

Index and inequality 22 rural-urban gap 13–14, 54–5, 108 and migration: concerns over 190–2 consequences of 187 costs of 186–7 large estates 192, 193, 194 population density 249–50, 251–2 as symptom of social problems 191 and mutual obligations 186–7 and population changes 199–201 and population growth: migration 148–51 regulation of 142, 159, 317 and wages, influences on 201–4, 207–8, 211–13 see also agrarian reform Russia: and industrialization 208 and migration to Germany 90, 91 Russian Poland 201 and inequality and proximity to 324 and influence on wage levels 202–3, 204, 208, 213, 279, 320–1 as source of labour 194, 199, 208 Saalfeld, D 38 Sammlungspolitik 2 savings: and economic development 15 and improved opportunities for 313 and inequality 317 and investment 253–4 Saxony, Kingdom of: and agrarian reform 32, 41, 42 and peasant landholdings 38 and rural credit 25


Saxony Statistical Office 98 Schaaf’hausen Bankverein 257 Schissler, H 36 n31 Schmoller, Gustav 189, 190, 297 Schremmer, E 234, 235 Schrott, S 62 n6 Schultz, Theodore 16–17, 215, 246 Schultz-Lupitz, A 244, 245 science, and agriculture 225–6, 244–5 seasonal migration 56, 64–5 self-employment: and agriculture 47 and underemployment 206, 214 serfdom, and abolition in Denmark 28 Sering, Max 190, 191 share-cropping: and criticisms of 181–2 as efficient solution 184 and missing market 184 and monitoring costs 183–4 and positive views of 183 Silesia: and agrarian reform 41, 42 and peasant landholdings 38 skill endowments: and income distribution 311, 316, 321 and land ownership distribution 315 and migration 279–81 and vocational training 311–12 ‘skirted decisions’, and internal weakness 1–2 smallholdings, and migrants 281–2 Smith, Adam 181 social change, and rural-urban migration 350–1 Social Darwinism 2, 330 Social Democratic party 310 and broadening appeal of 335



Social Democratic party (Continued) and electoral support for: agricultural population 341 aristocratic estates 342–3 candidate composition 343–5 city size 337, 339 desire for change 343 econometric analysis 356–60 economic conditions 340, 342, 343 feelings of political inequality 343 gains in 1912 election 331–2 growth of 333–6 inequality 340, 342 population growth 154, 156, 174, 175 population mobility 340 proximity to towns/cities 341–2 religious affiliation 337–9, 341 rural/small town areas 340–3 and internal politics of 334 and local/regional government 332 Soetbeer, A 298, 299 Sonderweg school 23, 330, 331 South Korea: and industrialization 4 and land ownership distribution 315 and land reform 46 n53 soya-bean production 314 Spain, and industrialization 4 Spanndienst, and agrarian reform 41 Spannfa¨hige, and agrarian reform 32, 33 state, the, and land ownership distribution 314 statistics, and contemporary use of 98–9 steam-threshing 247–8 Stegmann, Dirk 332 Stein-Hardenburg laws 192 stillbirths 126, 135

sugar beet cultivation: and artificial fertilizers 239–40 and employment 201 and productivity 239–41 and wages 204, 212 Switzerland, and emigration to 81 Tagelo¨hner 15 Taiwan: and industrialization 4 and land ownership distribution 315 tariffs, see protection taxation data, Prussian: and income distribution 303–6 econometric analysis 323–6 measurement of 297–9 worsening of 302–3 and opportunity provided by 303 and problems with use of 299–302 technology: and agriculture 216, 225–6, 234–5, 242–5 impact on employment 201 labour-saving 247–8 tenancies, and agricultural land 52–3 tenant rights, and Danish agrarian reform 29, 30–1 terms of trade: and inequality 320, 323, 326 and movement in 10–11, 254 Thu¨nen, J H von 36 n30 Tirpitz, Alfred von 348 Tirpitz naval plan 348, 349 and Kehrite school 350 Todaro, M P 317 trade unions: and industrialization 7 and wage levels 321 transport costs, and agriculture 25–6 ‘turning point’, and Lewis Model 7, 8, 294

Index underemployment 11 n17 and agriculture 12, 16, 185–8, 204–7, 214, 247–9, 273 partible inheritance 205–6 and definition problems 250 and informal urban sector 15 and labour surplus 204–7, 249–50 and migration 214, 249 unemployment: and efficiency wage rates 187 and migration 271–4 United Kingdom: and Anglo-German rivalry 263–4, 346 response to German foreign policy 348, 349 and farm size 48 and immigration 80 and land ownership distribution 49, 51 and political development of 46, 331 and urbanization 59–62 United States: and agriculture 216 exports 217, 219–20 and city growth rates 62 and emigration to 81, 83–4, 321 agricultural employment 88–9 decline in 85 influence of agriculture 81–2 previous occupations 86–8 wage ratio 85 and income distribution 313–14 urbanization: and contemporary views of 188–9 and death rates 157–8 and electoral politics 335–6 and impact of agricultural decline 217–18 and industrialization 15 and inequality 316, 323, 326

405 and international comparison of rates of 59–62 and life expectancy 157–8, 283 and migration 59–62, 111–12 and population growth 151–6, 173–7 and Social Democratic Party vote 335–6 and social dislocation 59, 351 and speed of 59, 61, 62, 151, 293, 353

Van Zanden, T 233–4 vocational training: and income distribution 311, 321 and migration 311–12 Voelcker, Augustus 234–5 voting behaviour, see electoral politics wages: and advantages of cash payments 182 and agricultural: agricultural/industrial ratio 105–6 influences on 201–4, 207–8, 211–13 rising level of 196–7 and Anglo-German comparison 263–4 and death rates 154–5, 170 and decline in rural 302 and efficiency wage rates 187 and fertility rates 146, 147 and German/American ratio 85 and labour supply 6, 7 and marriage rates 178–80 and migration 56, 94, 111 impact of 190 wage convergence 274–9 wage dispersion 276 women 281–3 and payments-in-kind 182–3

406 wages: (Continued) and population growth 175 and productivity 7, 9–11 and proximity of Russian Poland 202–3, 204, 208, 212, 279, 320–1 and regional variations 275–6, 278–9 and rural-urban gap 108 and trade unions 321 Wagner, Adolf 189, 190, 298 Wagner, Richard 297 Weber, Adna 58 Weber, Adolf 188–9 Weber, Max 11–12, 17, 107, 187, 191, 192, 207 and eastern agriculture 193–7 capitalist organization 195 children of labourers 195 impact of migration 196 labour relationships 194–5

Index large estates 193, 194 political concerns 193–4 and labour surplus 293 and labour-saving technology 247 and Prussian land reform 34 Wehler, Hans-Ulrich 2 n3, 331 and German exceptionalism 19–20 Westphalia: and agrarian reform 41 and peasant landholdings 38 Williamson, J G 264, 296, 313 women: and illegitimate children 142 and migration 76–9, 138–9 life expectancy 283 wages 281–3 working class 154, 156, 174 World Bank 4, 307 Wu¨rttemberg, and agrarian reform 44 Zollverein, and creation of 20