Political Origins of Firm Strategies a Dissertation

POLITICAL ORIGINS OF FIRM STRATEGIES

A DISSERTATION SUBMITTED TO THE DEPARTMENT OF POLITICAL SCIENCE AND TO THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

Alexander G. Kuo September 2010

This work is licensed under a Creative Commons Attribution- Noncommercial 3.0 United States License. http://creativecommons.org/licenses/by-nc/3.0/us/

This dissertation is online at: http://purl.stanford.edu/jj651ff0390

ii I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Jonathan Rodden, Co-Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Isabela Mares, Co-Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Stephen Haber

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

David Laitin

Approved for the Stanford University Committee on Graduate Studies. Patricia J. Gumport, Vice Provost Graduate Education

This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file in University Archives.

iii Abstract

What explains the emergence of different kinds of capitalist systems across advanced industrialized democracies? In particular, why do highly coordinated employer associations form in some states and not others, and why do such organizations change over time? Because different kinds of contemporary employer coordination or varieties of capitalism are linked to long-term outcomes of unemployment, inequality, social policy generosity, and product innovation, it is important to understand their historical origins and development. Given the recent attention to historical economic institutions and their long-term policy consequences, understanding their origins and divergence is of obvious historical, normative, and theoretical interest.

In this dissertation I propose an alternative account of the development of highly coordinated employer associations and their absence in advanced industrialized states. I provide a new framework for conceptualizing these outcomes as strategic decisions of firms. I go on to argue that major collaborative employer associations emerged largely out of major threats to firms posed by workers, during key historical moments in the early twentieth century and the inter-war period. When the threat posed by workers was extremely high, firms agreed to form institutionalized arrangements with unions, such as encompassing collective bargaining institutions.

When firms felt that the threat could be contained, they pursued strategies of collective repression of workers, in the form of repressive employer associations. However, the collective strategies that these firms pursued could only succeed if there were not incentives to defect from such arrangements. Lower industrial heterogeneity in

iv economies that facilitated collective action increased the likelihood of collective firm strategies succeeding. This proposed theory is more consistent with the historical evidence of the actual development and changes in employer organizations than existing literature.

To test the theory I provide evidence that confirms the preferences and strategies of firms, and demonstrate how they change in response to differing levels of the variables described above. I use a combination of empirical approaches to achieve these goals. First, to substantiate the predictions about the strategies pursued by firms,

I gathered primary data on the main employer organizations in the United States and

Germany from the late nineteenth to mid twentieth century. The new evidence, constituting publications, meeting notes, and internal correspondence supports the main predictions of the theory about the determinants of firms‘ strategies and formation of early employer associations. Second, I leverage the extensive sub- national variation in the strategies pursued by firms in both countries to confirm observable implications of the theory. I statistically test the key mechanisms of the theory using new sub-national data on membership in different employer organizations in the US and Germany between 1880 and 1930. Third, to test the portability of the theory, I present evidence of changes in firm strategies between 1880 and 1950 for other countries, including Australia, Britain, Denmark, France, Japan, and Sweden.

I conclude the dissertation with discussion of broader implications for comparative politics, in particular the importance of understanding collective action among firms and instruments of repression.

Acknowledgments

I take great pleasure in thanking all the individuals who have helped make this dissertation possible. I would first like to thank the members of my dissertation committee. My greatest emotional and intellectual debts go to my advisor, Isabela

Mares. Since the day I walked into her office with a set of barely formed ideas about the history of varieties of capitalism, she has been a wonderful mentor through my years as a graduate student. Her critical and helpful intellectual insights, amazing patience, generosity with her time, sense of humor, and contagious enthusiasm have encouraged my interest in comparative politics, and vastly improved this manuscript. I have met with her to talk about the project in a variety of geographical locales, and each time, she devoted maximum energy and effort to improve what I had written and to constructively criticize what I was thinking. I feel lucky to have had her as a mentor and I look forward to learning more from her throughout the rest of my professional career. I could not have asked for a better or more caring advisor.

I owe a great deal of thanks to Jonathan Rodden. I feel lucky that he joined the department just as I was coalescing some of the ideas and drafting more of the manuscript. Jonathan always provided a host of useful suggestions and energetic encouragement on the project, and had particularly insightful and creative ideas for gathering new data and for framing the ideas. His generosity with his time, patience, and obvious interest in the implications of my research are much appreciated.

David Laitin taught my first few courses in comparative politics in graduate school, and I have learned a tremendous amount from him and his approach to doing social science; this manuscript is my first real (and hopefully not last) attempt to do

vi justice to his teaching. At critical points throughout the writing, he gave important suggestions about improving the argument, testing, and pushing to make the project more relevant to a broader audience of comparative politics scholars. I look forward to continue learning from him the professor who some students affectionately and admiringly refer to as ―Gandalf.‖

Steve Haber, the final member of the dissertation committee, encouraged my early interest in comparative political economy and the study of business history and the role of business organizations in politics. He also pressed me to think about how the arguments in the manuscript could generalize to developing states, and to apply insights from other literatures to my own thinking and writing about advanced industrialized states. I will continue to learn more from him and from his work.

Fieldwork for the dissertation was partially funded by the Littlefield Graduate

Fellowship and the Graduate Research Opportunity Dissertation Grant. Gerald

Feldman kindly met with me before I left for Germany and gave great feedback on the project, as well as how to get acquainted with German historical research. I spent a stimulating year as a visiting doctoral student at the Max Planck Institute for the Study of Societies (Max Planck Institut für Gesellschaftsforschung) in Cologne, Germany, for which I have Jens Beckert and Wolfgang Streeck to thank. I had a wonderful time doing research there, due in large part to the community of doctoral students and scholars, including Michael Blauberger, Helen Callaghan, Steffen Ganghof, Philipp

Klages, Martin Höpner, Olga Maletz, Philip Manow, Cornelia Woll, and Armin

Schäfer. These friends and colleagues made me feel at home while doing research.

Many archivists and librarians in Germany assisted my research and the gathering of

vii many historical materials there. I would like to thank in particular Christian Hillen at the Rhineland-Westphalian Economic Archive; Hans-Hermann Pogarell at the Bayer

Archive; and Norbert Uersfeld at the Institute for the German Economy in Cologne.

A host of librarians and archivists in the United States were also extremely helpful in tracking down sources and data. I would like to thank the staff at the

Indiana Historical Society, the US National Archive, and Hagley Museum and library, in particular Marge McNinch. At Stanford, Tony Angiletta and Ron Nakao patiently helped with numerous requests and questions. Patrick Galloway shared data on

Prussia demographics and kindly answered questions about data sources. Gerald

Friedman kindly shared his data on unionization in the United States, used in chapter

7. Edward Anderson shared data on cross-national wages, used in Chapter 8.

Throughout graduate school and in writing this dissertation, the staff at Encina

Hall has been there at every step to help. Chandelle Arambula, Judy Low, Jeanette

Lee-Oderman, Eliana Vasquez, and Jackie Sargent were always willing and eager to navigate any administrative problem.

This dissertation has benefited greatly from comments on either written or oral presentations of the manuscript from a number of friends and colleagues. I am fortunate to have received feedback from so many individuals. Peter Hall wrote an especially thoughtful set of comments on an early incarnation of the proposal for the dissertation. I received helpful comments on written portions of the manuscript from

Matt Carnes, Luke Condra, Roy Elis, Victor Shih, Jed Stiglitz, Kathleen Thelen, Julia

Tobias, Jeremy Wallace, and Anne Wren. Many friends and colleagues gave helpful advice on presentations of the material, including Andrea Abel, Claire Adida, Mike

viii

Albertus, Yuen Ang, Sarah Anzia, Kendra Bischoff, Thomas Brambor, Jowei Chen,

Dara Cohen, Josh Cohen, Tiberiu Dragu, Jesse Driscoll, Desha Girod, Judy Goldstein,

Laurel Harbridge, Jim Fearon, Danielle Harlan, Stephen Jessee, Karen Long Jusko,

Kimuli Kasara, Candy Ku, Charlotte Lee, Aila Matanock, Sonia Mittal, Maggie

Peters, Connor Raso, Natan Sachs, Emily Shafer, Luke Stein, Rachel Stein, Alex

Tahk, Mike Tomz, and Jessica Weeks.

There are many friends to thank. Luke Condra provided much encouragement and warm support during the writing of this dissertation. I would like to thank in particular Neil Malhotra, who has been a wonderful friend and colleague throughout my graduate school days. I will have many fond memories of our friendship forged at

Stanford. Along the way he has given me generous advice and insightful criticism of my research ideas. I owe a special debt to Yotam Margalit, who remains an ideal colleague. Yotam has been an especially kind and compassionate friend through the years. I look forward to the times ahead with them. It would be difficult to list all the reasons to thank them, but hopefully, they know most of them.

My parents, Ming and Mei, and my sister, Michelle, have been the most loving family that one can possibly hope for. Their emotional support and unwavering confidence in my efforts made this work possible. They encouraged me to pursue my passion, and for that, I am eternally grateful.

Table of Contents Chapter 1: Introduction ...... 1

Motivating Observations ...... 1 Section 1 Existing perspectives on industrial institutions and firm strategies ...... 9 Section 2 Outline of dissertation ...... 17 Chapter 2: A New Conception of Firm Strategies ...... 25

Section 1 A New Framework ...... 25 Section 2 Coding the Framework ...... 36 Section 3 Distinguishing Firm Strategies, Labor Strategies, and Outcomes ...... 41 Chapter 3: A Theory of Firm Strategies ...... 46

Section 1 Main Explanation and Summary of Predictions ...... 47 Section 2 Mechanisms of the Threat from the Left and Endogeneity ...... 50 Section 3 Industrial Heterogeneity, Variance in Cost of Repression ...... 60 Section 4 Alternative Explanations and Concerns ...... 68 Chapter 4: Evidence from Imperial Germany ...... 83

Section 1 The Industrial and Repressive Context ...... 88 Section 2 Transition to Collective Strategies ...... 107 Section 3 The Creation of Peak Level Employer Associations ...... 136 Section 4 Overcoming Differences and Further Collective Strategies ...... 162 Chapter 5: Testing Mechanisms and Implications from Imperial Germany ..... 177

Section 1 Explaining Strategies of Collective Action: Sector Level Evidence ..... 178 Section 2 Association Level Evidence ...... 183 Section 3 Explaining Membership in Collective Repressive Associations ...... 188 Section 4 Conclusions ...... 196 Chapter 6: Evidence from Weimar Germany ...... 205

Section 1 From Collective Repression to Collective Collaboration ...... 207 Section 2 The Collapse of Collective Collaborative Strategies ...... 229 Section 3 Repressive Strategies in Weimar Germany ...... 247 Section 4 Implications and Alternative Views ...... 271

Chapter 7: Evidence from the United States Inter-War Period ...... 278

Section 1 Overview of Changes in Firm Strategies ...... 281 Section 2 The United States and the ―Open Shop‖ Movement ...... 284 Section 3 Testing the Theory with Micro Data ...... 293 Section 4 Results from Micro-Tests ...... 300 Section 5 Qualitative Evidence ...... 306 Section 6 Conclusions ...... 317 Chapter 8: Cross-National Evidence ...... 323

Section 1 Summary of Cross-national Evidence and Approach ...... 323 Section 2 Country evidence ...... 327 Section 3 Conclusions ...... 345 Chapter 9: Conclusions ...... 353

Bibliography ...... 361

List of Tables

Table 2-1: Classifying Strategies of Firms ...... 45

Table 2-2: Historical Examples of Firm Strategies ...... 45

Table 3-1: Main Theoretical Predictions of Firm Strategies ...... 81

Table 3-2: Current Hypotheses about Firm Strategies ...... 82

Table 4-1: Characteristics of Employer Associations, 1911-1913 ...... 175

Table 5-1: Explaining Employer Coordination at the Sector Level ...... 198

Table 5-2: Major National Level Employer Associations ...... 199

Table 5-3: Explaining Strike Insurance Provisions ...... 200

Table 5-4: Explaining institutional connection ...... 200

Table 5-5: Membership of the HDA by Industry and SPD Threat ...... 201

Table 5-6: The Left Threat and HDA ...... 201

Table 5-7: Membership by vote share ...... 202

Table 5-8: Explaining Collective Repressive Strategies ...... 202

Table 7-1: County-Level Repressive Employer Coordination ...... 320

Table 7-2: Substantive Effects of Union Threat ...... 322

Table 8-1: Dates of Firm Strategies and Electoral Systems ...... 347

xii

List of Figures

Figure 5-1: 1907 Distribution of SPD Vote Share ...... 203

Figure 5-2: Threat from Left and Repressive Activity ...... 204

Figure 6-1: Change in Firm Strategies in Germany ...... 277

Figure 7-1: Change in US Firm Strategies ...... 319

Figure 8-1: Change in Firm Strategies in Sweden ...... 348

Figure 8-2: Change in Firm Strategies in Britain ...... 349

Figure 8-3: Change in Firm Strategies in Denmark ...... 350

Figure 8-4: Change in Firm Strategies in France ...... 351

Figure 8-5: Change in Firm Strategies in Australia ...... 352

xiii

Chapter 1: Introduction

Motivating Observations

In 1907, J.W. Van Cleave, the president of the US National Association of

Manufacturers (NAM), addressed the member firms at a convention of a fellow employer organization, the National Metal Trades Association (NMTA). The purpose of the address was to convince firms of the fruits of a formal association of firms against organized labor. He tried to persuade the audience of company owners and representatives that they should organize collectively to counter the losses that unions were imposing on firm owners. Specifically, he argued that a federation of all employer organizations could be effective in ―head[ing] off all foolish or vicious propositions before they were introduced on our…legislature,‖ and that a ―great council‖ of employers would still be able to preserve the independence of its members

(Van Cleave, 1907).

Van Cleave‘s address was one of many efforts made by American regional employer organizations to form strident ―open shop‖ organizations before and after

World War I. These organizations aimed to prevent unions from organizing across firms to dictate wages, working conditions, and restrictions on apprenticeships. The open-shop efforts were quite successful in reducing inter-firm union influence before the Great Depression (Ingham 1974; Wakstein 1964). Since the end of World War II, however, no movement like the open shop has taken place in the United States. In the

United States today, negotiations between firms and workers largely occur within the firm; disputes are handled in courts of labor law. But ―organization men‖ like Van

Cleave have no bearing on the wages and working conditions of the American worker.

In 1918, amidst concern about economic chaos that could follow the end of the

First World War, several prominent German business leaders met to form a collaborative agreement with unions, which would re-organize industrial relations in

Germany and establish the first time national-level collective bargaining institutions.

The leaders managed to present this agreement to the government as a fait accompli.

Hans von Raumer, the business manager of the German electro-technical industry, headed the negotiations. Raumer was persuaded by his colleagues to quickly start an

―organic collaboration‖ with unions before the ―flood of events [the expected post-war chaos] overtook us all‘‖ (Feldman 1970). The resulting negotiations led to the 1918

Stinnes-Legien Accord, the first formal definition of the rights of unions at the national level, and the first German government-sanctioned ―cross-class‖ agreement between business and labor. This agreement would provide the ideological template for a similar framework of laws that were implemented in the immediate post-WWII period. The willingness of business leaders to negotiate the Stinnes-Legien Accord contrasts sharply with their attitudes towards workers in the early twentieth century.

In the same year as the Stinnes-Legien agreement was being quickly negotiated, representatives of the most important British employer organization, the

Engineering Employers Federation (EEF), contacted the head of the War Ministry with an urgent request: they sought from the Ministry additional public enforcement of a law that was supposed to prohibit firms from poaching workers from each other with lures of higher wages and benefits. The head of the EEF, Alan Smith, succinctly stated to Ministry officials the motivation for such action: ―we want to subject ourselves to discipline as much as anybody else, because we have to protect ourselves

2 from ourselves‖ (Zeitlin 1991, 64). These efforts ultimately collapsed as the Ministry disallowed local employer associations to restrict worker mobility.

In 1938, the Swedish Employers‘ Confederation (Svenska

Arbetsgivareföreningen, or SAF), negotiated a sweeping agreement with its main union counterpart, the Swedish Confederation of Labor (Landsorganisationen I

Sverige, LO) at Saltsböjaden. This famous ―Basic Agreement‖ stipulated a number of provisions that would benefit workers and gave unions an institutional voice in wage setting. The key aspects of the agreement included centralized mediation of wage disputes between firms and workers. The famous ―cross-class‖ agreement contrasts sharply with the strategies of harsh lockouts that the SAF and related associations, particularly the Engineering Employers‘ Association (Verkastadsföreningen, VF), had engaged in thirty years prior, which had the effect of almost bleeding union treasuries dry.

These four historical examples of firm decisions to collectively act illustrate much cross-national variation in the strategies pursued by firms with respect to workers. What compelled the Van Cleave in the United States and his colleagues to take hostile action, while Raumer and his colleagues sought a national cross-class agreement? Why were the German industrialists represented by Raumer concerned with presenting a fait accompli to the government—essentially trying to minimize the state‘s initial involvement—while Smith was in the opposite position of begging the government for enforcement? In all these instances, prominent industrialists made decisions about whether and how to expand the reach of their nascent collective institutions, and what actions such organizations should take vis-à-vis organized labor

3 and the state. Annals of business history of both developed and developing economies are replete with similar examples of industrialist calls to action and changes in fundamental strategies about how to deal with workers.

The historical or temporal variation in such firm strategies is also striking.

Why is someone like Van Cleave relatively unimportant for American industrial relations today, while the representatives and member associations of the

Confederation of German Employers‘ Associations (Bundesvereinigung der

Deutschen Arbeitgeberverbände, BDA) remain important industrial and political actors? Why did the SAF apparently radically change its use of lockouts to sign one of the most labor-friendly accords? Why did people like Smith of the British EEF engage in different kinds of collective strategies after the inter-war period, while today the EEF retains only a fraction of its functions from its early twentieth century origins?

These incidents show that firms have altered strategies greatly in response to changing labor and economic conditions—both in terms of acting collectively and in terms of treatment of workers. They also raise broader questions in comparative political economy. What explains the emergence of different kinds of capitalist systems across advanced industrialized democracies? In particular, why do highly coordinated employer associations form in some states and not others, and why do such organizations change over time? In some countries, many firms in a country are members of large, encompassing organizations that obligate them to comply with a range of rules regarding the treatment of workers. In other countries, firms act in a more laissez-faire economic environment, and even if employer organizations exist,

4 they have little impact on the major labor-market actions of firms and serve only as lobbying associations. Because different kinds of contemporary employer coordination are linked to outcomes of unemployment, inequality, worker training regimes, social policy generosity, and product innovation, it is important to understand their historical origins and development.1

Beyond the normative policy consequences of different forms of employer coordination, understanding long-term historical change in firm strategies and treatment of workers is relevant for contemporary institutional dilemmas facing both advanced industrialized and middle-income states. A central concern of modern economies that have a reputation for stable collective bargaining institutions is whether such institutions can survive. Scholars have noted a pattern of increasing defection of member firms and organizations from national-level employer confederations in previously ―stable‖ coordinated market economies such as Germany and Sweden.2 That is, member firms and sector-level employer associations are voluntarily deciding that they are better off on their own than being subjected to organizational rules about how they should negotiate with their own workers. There is now a lively debate about how whether employer confederations will endure, and with them, whether the welfare state they help support can survive (Hall 2007; Thelen

2000; Swenson and Pontusson 2000). In the developing world today, open questions include whether nascent business and employer organizations will repress or deal with

1 See the discussion in Hall and Soskice (2001). 2 For discussion of the exit of German firms from national associations, see the discussion in (Hassel 2002, 1999) and (Silvia and Schroeder 2007).

5 workers; how influential they will be on the state; and whether they will engage in corruption or rent-seeking.

Recent scholarship in the field of comparative political economy has begun to address these questions by trying to determine historical ―root causes‖ and tracing the development of different forms of capitalist systems within current advanced industrialized states. Within this field, more attention is being paid to the choice of industrial institutions and identifying divergence in forms of economic coordination, and in particular, the relationship between political institutions and corporatist economic institutions.

Yet the most recent accounts on the subject of the development of industrial relations institutions in the late nineteenth and early twentieth century remain curiously disengaged from the kinds of incidents and decisions discussed above in the four example countries. I discuss these extant theories more extensively in the following chapters. In particular, they posit an account of long-standing coordination among firms and between firms and workers that is at odds with the historical record that highlights the real redistributive threat posed by workers to firms during this time period, and a set of other incentives firms had to act collectively. In this dissertation I address the question of the development of different forms of employer coordination by positing a more historically grounded range of strategies that firms pursued vis-à- vis workers.

I propose a new way of conceptualizing the outcomes and a theory that can explain this variation. This conception captures the logic of the decisions of organizational entrepreneurs that opened this introduction: whether to continue

6 repressive strategies against workers or to collaborate with them; and whether to act individually or collectively. I argue that the major outcomes to explain are not just

―degrees of coordination,‖ but when countries experienced qualitative movement from repression of workers to allowance of workers to ―have a seat at the bargaining table‖; and whether countries experienced the development of organizations that obligate member firms to comply with institutional rules.

I then offer a theory for explaining this range of outcomes that is based on more historically grounded view of the design and change in industrial institutions over time, but hypothesize independent variables and mechanisms that can be generalized to other countries and eras. I argue that major collaborative employer associations emerged largely out of major redistributive threats to firms posed by workers, during key historical moments in the early twentieth century and the inter- war period. When the threat posed by workers was extremely high, firms agreed to form institutionalized arrangements with unions, such as encompassing collective bargaining institutions. When firms felt that the threat could be contained, they pursued strategies of collective repression of workers, in the form of repressive employer associations that sought to crush unions. However, the collective strategies that these firms pursued could only succeed if structural conditions prevented firms from defecting from such arrangements. In particular, lower industrial heterogeneity in economies that facilitated collective action increased the likelihood of collective firm strategies succeeding. Lower industrial heterogeneity meant that there was a lower variance in the costs of repression, which increased the probability of agreements among firms and sector-level associations. If these structural conditions

7 did not allow for collective action, then individual firm-level strategies with respect to workers were pursued.

This explanation is more consistent with the historical evidence of the actual development and changes in employer organizations. The theory incorporates an explanation for why some firms engaged in repressive coordination or strategies against workers. This outcome of repression (either by firms or employer organizations) has been theoretically understudied in the political economy literature on the development of varieties of capitalism. While the existing accounts focusing on electoral systems and guild traditions account for some cross-national and temporal variation in employer coordination, the proposed theory here explains a wider array of strategies of firms. Given the recent attention to endogenous electoral and economic institutions and their long-term policy consequences, understanding their origins and long-term divergence is of obvious normative and theoretical interest.

To test this theory of firm strategies, I use a research design that involves in- depth comparison of substantively important cases as well as exploitation of sub- national variation within both cases. Additionally, I use historical evidence that tests the theory over time. I present evidence that confirms the different preferences and strategies of firms, and shows how their strategies changed in response to differing levels of the variables described above. I use a combination of empirical approaches to achieve these goals.

First, to substantiate the predictions about the strategies pursued by firms, I gathered extensive original archival data on the main employer organizations in the

United States and Germany from their advent through the inter-war period. The new

8 evidence, constituting publications, meeting notes, and internal correspondence supports the main predictions of the theory about the determinants of firms‘ strategies.

I present much descriptive evidence that links the hypothesized variables to the predicted outcomes.

Second, I leverage the extensive sub-national variation in the strategies pursued by firms in both countries to confirm observable implications of the theory. I statistically test the key mechanisms of the theory using new sub-national data on membership in different employer organizations in the US and Germany between

1880 and 1930. This dataset includes original data on firm strategies as well as new data on covariates at the local level.

Third, to test the portability of the theory, I present short case descriptions of changes in firm strategies between 1880 and 1950 for other important advanced industrialized states. This research design is an advance over previous studies that rely on limited cross-national regressions.

In the rest of the introduction, I briefly discuss the development of the literature on employer coordination, discuss some of the prominent flaws in the conceptualization, and then preview the rest of the dissertation.

Section 1: Existing perspectives on industrial institutions and firm strategies

In this section I present a brief overview of how firm strategies with respect to workers, particularly collective organizations, have been discussed. Much of the voluminous literature in comparative politics, industrial organization, economic sociology, and business history examine the redistributive and political consequences of industrial institutions and the impact that employer associations have on wage-

9 setting or related outcomes. In particular, recent literature in comparative political economy theorizes why the presence or absence of inter-firm coordination links to other economic institutions and normatively important social outcomes. The ascendant ―varieties of capitalism‖ (VOC) research agenda started by (Hall and

Soskice 2001, 2001) takes employer associations a pre-existing part of a set of coordinated economic institutions. In their framework, countries can be coded as

―coordinated market‖ (CME) or ―liberal market‖ economies (LME), with the former group having encompassing employer associations that cover a large plurality (or majority) of major firms in an economy. LMEs are economies where employer associations play a minimal role in relations with workers. This agenda links the existence and functions of modern employer organizations and their counterpart unions to long-term economic outcomes such as unemployment, inequality, social policy innovation, and responses to globalization. This research agenda also argues that industrial and financial institutions can be clustered in to CMEs and LMEs, and that such institutions constitute equilibria.

This literature is an advance with respect to the previous generation of scholarship that distinguished between corporatist and pluralist economic systems

(Schmitter and Streeck 1991; Katzenstein 1985; Schmitter 1974). This influential categorization of different economic systems implies that in corporatist systems, firms pursue a collective strategy, as opposed to an individual strategy, to maximize profits.

By contrast, in ―pluralist systems,‖ firms either choose not to pursue collective strategies, or are unable to. One of the key contributions of the VOC body of research is that it takes the firm as the key actor, and introduces their preferences and strategies

10 vis-à-vis other economic actors such as banks and unions. Previous literature on industrial institutions which focused only on union preferences and demands ignored the varying preferences of firms in designing or joining industrial institutions.

Recent arguments consistent with the assumptions of the VOC approach takes firm preferences and strategies as key factors in explaining variation in social policy and labor market institutions (Mares 2003; Swenson 2002). These studies convincingly demonstrate that variation in firm preferences and strategies are crucial to understand why social insurance provisions and labor market institutions across countries differ. Mares argues that differences in skill profiles of firms affect preferences for any social protection for workers; in addition, such profiles explain support for targeted as opposed to publicly funded national social policy (2003).

Swenson argues that structural conditions can explain why employers favor solidarist versus segmentalist labor market strategies. These studies have convincingly critiqued the previous arguments that welfare states and labor legal arrangements are a consequence of labor ―power‖ alone (Korpi 2006).

Both the VOC literature and these recent accounts which share some of its assumptions assume the existence of employer organizations at the sector or national level. The point of departure for these arguments is that firms have somehow overcome the collective action problem to lobby for or act to achieve social policy or labor-market objectives. However, this recent literature does not consider the choice of firms and their resulting institutions as an outcome variable. The question of whether firms would prefer or be able to form such institutions, and what actions such organizations would undergo, was an older but still unresolved question in the

11 academic fields noted above. A critical component of the discussion of industrial institutions is whether firms would be able to collectively act to form such organizations and institutions that would be sustainable beyond just being pure lobbying associations. One perspective was that formal organization of firms would be unnecessary. Friedrich Engels implicitly argued that we should not observe anything like formal associations: ―Capitalists…are always organized. Their small number, the fact that they constitute a particular class, [and] their social and commercial relations, make formal organizations superfluous‖ (cited in (Friedman

1991)). This argument is based on an assumption of the basic commonality of interests of capitalists, obviating the need for any kind of formal coordination. By contrast, Olson argued that in the absence of selective incentives or agents willing to pay the costs of public good provision, there should be a similar outcome of relative absence of collective organizations of firms (1968). The research agenda that Olson spawned was about the failure of collective action: even if individual firms all desired a collective organization, a set of structural conditions would prevent its formation.

In response to these predictions of relatively little collective action of firms,

Offe and Wiesenthal, in a re-opening of this debate, argued that firms should find it relatively easier than workers to organize, due to ontological ease in communicating interests (firms do not have to have profit motives ―redefined‖ for them; firms possess higher organizational capacity than workers; and they have lower heterogeneity of inputs (Offe and Wiesenthal 1980). One of the main implications of this agenda of

―two logics of collective class action‖ was that economies should experience a limited but powerful degree of capitalist action, in contrast to the predictions of Engels and

Olson. Streeck and Schmitter countered that the same theoretical framework proposed by Offe and Wiesenthal implied that we should observe firms acting on a interests in both product and labor markets, and so economies should be composed of many different kinds of business associations with different functions (Streeck 1992;

Schmitter and Streeck 1991).

This lively debate about the collective action of firms is in some ways more precise than the previous literature on the consequences of corporatist versus pluralist economic systems and varieties of capitalism, with regards to measurement of the outcome variable, as the debate in economic sociology became preoccupied with the number and degree of employer associations (Traxler 2000). But these literatures still do not adequately distinguish the different types of collective strategies of firms to be explained. Both the more historical and recent literatures conflate scopes and purposes of organizations of firms. Scholars discuss many different kinds of inter- firm coordination—employer associations (dealing with labor market obligations of member firms); trade associations (dealing with information sharing and lobbying); and cartels (highly coordinated product market control of either quotas or prices).

Descriptive literatures in different fields focus on one narrow type of collective organization, while the theoretical literature in political science has lagged in failing to account for different kinds of collective action of firms. Much of the vast empirical literature on trade associations, employer associations, and cartels takes the organization (or institution of inter-firm coordination) as the unit of analysis, while the comparative political economy literature naturally takes the country-time period as the unit of analysis.

The descriptive literature in the fields of industrial sociology and business history do give useful descriptions of different kinds of business organizations (at the industry or peak level. However, there remains confusion about coding and clarifying the dependent variable and unit of analysis, and accounts do not describe testable general propositions about their formation (Gladstone and Windmuller 1984; Sisson

1991; Tolliday and Zeitlin 1991). In addition to the conflation of different kinds of firm strategies, the literature has failed to provide a clear set of scope conditions under which we should predict collective action of firms to be successful at the national level and when such activities would change over time.

The problems of a focus on explaining corporatist versus pluralist economic systems, and the conception of employer coordination assumed in this classification, become clearer when we consider recent explanations for employer collective action.

Scholars in comparative politics have begun to address the question of the formation of specific kinds of employer associations, though, as discussed below, their characterization of such associations remains incomplete. This literature dwells on the link between economic coordination and electoral systems. They focus on two variables: the electoral system and pre-existing guild institutions.

Recent important and influential literature has begun systematically exploring the origins and mechanisms of historical forms of employer associations and coordination. This literature links systems of employer coordination to political institutions (and vice versa), focusing on the importance of the electoral system and pre-existing guild traditions. Cusack, Iversen, and Soskice (henceforth CIS) argue that different forms of pre-existing economic coordination affect the choice of electoral

14 system. Martin and Swank (henceforth MS) argue that the relationship is in the opposite direction; PR electoral systems are more likely to cause the development of highly coordinated employer organizations. Both accounts argue that employer organizations historically secured or protected investments in specific skills for workers (Cusack, Iversen, and Soskice 2007; Martin and Swank 2008; Kreuzer 2010;

Cusack, Iversen, and Soskice 2010).

While CIS are not concerned with the development of this forms of economic coordination per se, it is important to outline their mechanisms linking economic to political institutions, as those same mechanisms are used to explain both historical and modern employer coordination by other scholars. They argue that higher degrees of economic coordination among firms, and between firms and labor unions, lead to adoption of proportional representation (PR) electoral systems; lower degrees of such coordination lead to the maintenance of majoritarian single-member district (SMD) electoral systems. Martin and Swank (MS) propose a similar causal logic but advocate the opposite direction: in their account, proportional electoral systems promote and solidify corporatist employer associations while majoritarian systems inhibit them.3 In both accounts, the kind of economic coordination that exists is assumed to be the result of longstanding or historical guild traditions as described by

Thelen (2004).4 These accounts both argue there is a correlation between the proportionality of electoral systems and the development of coordination among firms.

Such coordination institutions help sustain joint investment in specific skills, and

3 The mechanisms linking electoral systems to economic coordination differ somewhat between these two accounts, which I address in the first section. 4 See as well as Hansen (1998) for explication on the origins and functions of different guild systems in terms of provision of skills.

15 incentives for firms to establish cooperative institutions with workers. I discuss these accounts in much greater detail in chapter 3 of the dissertation.

Both the recent and older accounts of employer coordination suffer from several deficiencies, particularly regarding their conceptualization of the outcome variable. First, existing frameworks of firm strategies fail to explain why some countries have firms that pursue labor-market goals individually, or why inter-firm coordination occurs only in select sectors. The preoccupation has been with variance within employer associations, and not in explaining the predominance of large firms not participating in such associations. The literature generally ignores the possible outcome that firms can individually pursue outcomes.

Second, studies of employer coordination have assumed only on a collaborative coordination that focuses on the presence of centralized or coordinated wage-setting and the industry-level provision of goods like vocational training for skilled workers. The collective bargaining outcome is the most frequently studied one by comparative scholars, as evidenced by datasets capturing differences within peak associations. Most of these datasets focus on Western Europe and the English- speaking OECD states (Golden, Lange, and Wallerstein 2006; Kenworthy 2001;

Traxler 2000), although there has been recent attention to coding the degree of collective employer coordination before WWII in Western Europe (Cusack, Iversen, and Soskice 2007; Martin and Swank 2008). Both recent pre-WWII and post-war datasets try to capture ―how collective‖ or ―degree of coordination‖ of firms and unions in their datasets. Kenworthy (2001) and Ebbinghaus and Visser (2000) measure at what level (firm, industry, or national-level) wages are set, as a proxy for

16 centralization of wage bargaining. Golden, Lange, and Wallerstein (henceforth GLW) measure a number of indicators for employer associations across twenty OECD states for the post-war period. These binary indicators of associations include: whether the employer confederation has appointment power of its board; whether it has veto over wage agreements and veto over lockouts; whether its own conflict funds; and whether a peak (national-level) confederation exists.

The indicators collected in these kinds of datasets take the existence of collective collaborative strategies as a given. As I discuss more extensively in Chapter

2, they do not consider repressive strategies in the choice set of firms. Moreover, they conflate self-fulfilling strategies of cooperation (the existence of collective bargaining, for example) with other indicators that could also denote the existence of repressive strategies (such the existence of lockout funds and control over affiliate members).

This problem is not as theoretically severe for the post-war data coding. But given the increased attention to understanding causes and consequences of early employer coordination (Cusack, Iversen, and Soskice 2007; Thelen 2004; Martin and Swank

2008), the assumption of a uni-dimensional ―collective cooperative‖ dimension to firm strategies going back to the late nineteenth century is problematic. I return to discuss these issues extensively in the next two chapters of the dissertation.

Section 2: Outline of dissertation

Chapter 2 presents a new framework of firm strategies that incorporates the choice of firms to repress workers. The first part of the chapter describes the framework. In brief, I argue that two dimensions characterize variation in the strategies of firms. First, employers can act individually or collectively; they can either

17 address the concerns of workers individually, or join a formal employer association.

Second, firms can battle repressively against workers, or they can collaborate with workers, by recognizing the latter as a formal negotiating partner with attendant responsibilities in the wage-setting process and establishment of basic working conditions. These strategies characterize different forms of inter-firm coordination, such as employer associations that seek to crush or promote the role of organized labor, or choices by firms to pursue these goals alone. This mapping of firm strategies contrasts with existing literature which only views them as either collectively collaborative or not collectively collaborative at all (Hall and Soskice 2001). The second part of the chapter discusses the coding procedure for this new outcome, and relates the conceptualization to existing approaches.

Chapter 3 describes the theory and predictions. The variation in firm strategies depends on two variables: the redistributive threat of workers posed to firms, and the degree of industrial heterogeneity. The former affects the incentives to pursue a collaborative or repressive approach; the latter shapes incentives to act collectively or individually. The two main insights to the theory involve the response of firms to the threat posed by the politicization of labor, and the enforceability of any institutions they agree to design to address the threat. I then discuss specific mechanisms that link the independent variables to variation in firm strategies. The threat from workers affects the level of the costs of repression that firms must engage in. Industrial heterogeneity affects the variance in the costs of repression, which captures the difficulty in maintaining self-enforcing collective strategies. I discuss some potential extensions to the theory, including joint operation of both mechanisms, and the

18 incentives for firms to share the costs of repression or enforcement with the state.

This parsimonious explanation accounts for a wide range of temporal and cross- national variation in strategies pursued by employers.

The next section of the chapter contrasts the theory with existing theories that emphasize other variables, including the role of electoral systems, legacies of guilds, initial structural conditions, and degree of unionization. I demonstrate how the new theory contrasts with these accounts because it clarifies a strategic logic from the key actors—firms and nascent organizations—regarding their response to changes in political and economic conditions, and the theory explains a much larger range of firm behavior. The theory gives predictions for when firms would engage in repressive practices that both inhibit democratization and reduce the welfare of workers. The chapter concludes by discussing and addressing potential objections to the theory.

The next five chapters describe various tests of the theory. I perform extensive testing of the theory with detailed evidence from Germany. The theory accounts for change in the political strategies of German firms, from a strong collective and repressive stance against organized labor to a willingness to recognize unions.

Chapters 4 and 5 test the theory with evidence from Imperial Germany. In Chapter 4 I document the transition from individual to collectively repressive strategies against workers. I use original archival evidence from correspondence and reports from important firms and employer organizations to confirm the predictions of the theory: the initially low political threat posed by the workers‘ parties and low heterogeneity facilitated a collective repressive set of strategies against workers. However, the

19 moderate increase in the political threat posed by the Social Democratic Party (SPD) before World War I prompted a shift to a more collaborative strategy among firms.

In Chapter 5 I exploit sub-national variation in the adoption of firm strategies in Imperial Germany to test mechanisms and observable implications of the theory. I do so by using new data on such strategies at the sector, associational, and voting district level. I demonstrate that higher degrees of industrial heterogeneity affect the enforceability of agreements. I constructed an original dataset of repressive firm strategies, by using data from national yearbooks, and membership data by economic region and city of repressive employer associations, most prominently the members of the Administration of German Employer Associations (Hauptstelle deutscher

Arbeitgeberverbaende, HA). Using another original dataset on Prussian voting districts, I find that support for the Social Democratic Party leads to more repressive firm strategies at the sub-national level. But, such collective repressive strategies were only possible in areas with lower industrial heterogeneity. This micro-evidence supports the main predictions of the theory. This evidence contrasts with standard accounts of firm strategies, which argue that firms in Germany were generally collaborative and had static incentives to protect specific assets with the assistance of unions (Cusack, Iversen, and Soskice 2007; Thelen 2004; Martin and Swank 2008).

In Chapter 6 I further test the theory with evidence from war-time and Weimar

Germany. I demonstrate that exogenous changes in the costs of repression of workers explain the change in firms‘ collective repressive strategies to much more collaborative strategies after 1918, and back again to repressive strategies after 1925.

The primary evidence on firm strategies indicates that higher enforcement costs

20 undermined the returns to strategic collective action because such agreements among firms. Firm strategies in the aftermath of World War I were strongly conditioned by the sudden rise in the political threat posed by organized labor. This accounts for the change to collaborative action through the important Central Working Community

Agreement (Zentralarbeit-gemeinschaft; ZAG). Internal evidence, including personal correspondence and organizational reports by the firms and organizations that were pursuing this strategy indicates that the changes in the reduced threat posed by the

Social Democratic Party facilitated a shift in firm strategies to repressive approaches, as evidenced by the defection from the collaborative ZAG. Changes in heterogeneity during the latter part of this decade then provided firms with higher incentives to adopt individual repressive strategies during the rest of the Weimar period. I then document the different ways in which repressive strategies were pursued in the latter half of the

Weimar era.

In Chapter 7 I test the theory with evidence from a vastly differently country time period, the United States in the inter-war Period. In this chapter I show how the theory can explain why American employers have chosen to act individually, or established repressive institutions that repressed workers. I begin by demonstrating that the United States shifted from largely individual repressive strategies in the late nineteenth century to a more collective repressive strategy right after WWI and during the inter-war war period. I also show that prior to the Great Depression, firms made strategic choices in their response to changing conditions to begin to revert back to individual repressive strategies. The variation in strategies was caused largely by

21 exogenous changes in industrial heterogeneity and the perceived political threat of organized labor.

I first test the importance of the threat and industrial heterogeneity variables using an original dataset on membership in collective repressive strategies during the inter-war period. I gathered original sub-national data on firm strategies from membership data for the ―Open Shop‖ movement, and membership from the most important manufacturing associations during this period, including the National Metal

Trades Association (NMTA), National Association of Manufacturers (NAM), and

National Erectors‘ Association (NEA), as well as reports from the Federal Trade

Commission (FTC). The statistical tests confirm the hypothesis that a lower threat from workers increased the pursuit of collective repressive strategies (the adoption of anti-labor employer organizations) at the county level.

As a second empirical test of the theory, I use archival evidence from key employer organizations to show how firm strategies at the national level responded to changes in the enforceability of agreements, and the redistributive threat posed by the workers. The records of these organizations indicate that firms chose to take advantage of lower enforcement costs to form influential anti-labor institutions. But towards the end of this period, these firms made a strategic choice not to develop (and partially dismantle) a more coordinated and sustained institution at the national level. In fact, the success of this repressive strategy removed incentives for its continuation. I trace this outcome during this period to specific firm concerns about the enforceability of inter-firm institutions. This chapter provides a more nuanced explanation of temporal

22 changes in American ―exceptionalism‖ in industrial relations, and challenges the conventional view of the US as an always ―liberal‖ regime.

In Chapter 8, the final empirical chapter of the dissertation, I test the portability of the theory with cross-national evidence from Sweden, Denmark, Britain, France,

Australia, and Japan. I compiled data on firm strategies and employer associations for other cases between 1870 and 1950. The case discussion includes the adoption and changes of firm strategies, as well as data on the two main independent variables: political threat posed by workers and heterogeneity. I demonstrate that these variables are correlated with the changes in predicted firm strategies, through the posited mechanisms. The descriptive data here indicate that the theory can explain patterns of firm strategies during much of the twentieth century. Where possible I demonstrate the validity of the proposed theory over competing explanations, in particular those that focus on longstanding ―cooperative‖ worker training systems and electoral systems.

The final section of the dissertation concludes by summarizing the findings, discussing possible extensions to the research agenda, and implications for comparative politics. This manuscript about the causes of different firm strategies and how they choose various institutions is embedded in a larger research agenda on instruments that businesses can use to affect democratization and redistribution. The strategy to form and maintain institutions, in response to structural conditions and political threats, is a historically important example. These strategies and institutions also plausibly inhibited democratization and income inequality as well. These research avenues indicate that the historical and contemporary political activities of firms are

23 relevant to the study of inequality, redistribution, and comparative political economy more generally.

Chapter 2: A Conception of Firm Strategies

In this chapter I provide a new conception of variation in the strategies of firms described in the previous chapter, and then an explanation for such variation. Section

1 presents the logic and framework of the dependent variable. Section 2 discusses issues regarding coding the variable. Section 3 concludes by relating the outcome of firm strategies to other outcomes of interest.

I offer a new conceptualization of firm strategies that allows for firms to act collectively versus individually as well as repressive against workers or collaboratively with them. In this section I propose a new framework that captures a greater range in the outcome variable. As Chapter 1 discussed, existing cross-national indices rely only on a ―cooperative‖ conception of employer coordination. Due to this limited conceptualization, extant theories can account for some cross-national variation in the co-emergence of political institutions and industrial institutions. But in their conception of in the dependent variable, they neglect important theoretical and historical variation in the possible functions of employer organizations. In this chapter

I argue that these indices can be thought of as a subset of the possible universe of strategies that firms can pursue in response to labor market conditions to maximize profits.

Section 1: A New Framework

Conditional on a set of firms that decides not to deal with concerns about workers or products individually, how do we think about their collective action?

Independent of collective action, what are the most important strategies they can take with respect to workers? Regarding collective action, the most intuitive way to code

25 collective strategies of firms is to code the existence of a formal organization.5 At the most basic level the organization is comprised of member firms. Organizations that operate at the national level (peak-level associations) can have sector or regional organizations as members. The organization of firms can also have institutional rules that constrain and encourage firm behavior to achieve collective goals in product or labor markets. Some of the rules are meant to facilitate pursuit of collective goals.

Other rules are designed to enforce such activity, such as by punishing member firms who do not comply. These organizations can have institutions that deal with representation of firms and administration of rules, such as voting boards, committees, and executive decision-making bodies. These collective strategies of firms can have features similar to those of interest groups (as posited by Olson), but also institutional features that constrain and guide member behavior (as posited in the modern VOC literature). But, the features of an organization of firms described above are fairly general and can apply to organizations that in practice are quite different.

Cursory examination of industries in the early twentieth century shows much variation in the kinds of firms strategies adopted within each industry. For example, in the United States, machine construction, publishing/printing, marble cutting, clothing manufacturing industries adopted early on collective strategies where unions were recognized and played an integral role in the wage-setting process (Willoughby

1905). But later on, machine manufacturing industries, as discussed in next empirical chapter, switched to repressive strategies and were involved in bitter industrial conflicts with workers. In Germany, skilled manufacturing industries in chemicals

5 This definition ignores informal ―gentlemen‘s agreements.‖

26 and high-end manufacturing were relatively organized and collaborative early on.

Other early organized industries included timber, printing, and shipyard workers

(Geary 1989). But heavy industry and raw materials organizations from early on, by contrast, fought against unionization.

The current conception of the dependent variable of employer coordination wrongly presumes that higher levels of coordination necessarily imply higher levels of cooperation with labor. In particular, existing theorizing of ―high‖ degrees of employer coordination obscures the fact that such firms could also be highly coordinated for purposes of repressing workers (both skilled and unskilled). Firms can choose to repress workers as a strategy, as well as cooperate with them. Moreover, repression can occur by individual firms, but also be done collectively, or with assistance from the state. In particular, collective repression played a critical in inhibiting organized labor, but it is under theorized as a possible outcome in existing literature. As I argue below, there was much discussion among firms within states about which strategy—repression of or collaboration with workers—firms should pursue. The resulting national-level firm strategies can be thought of as choices of firms to pursue repression or not, and as bargains between different industrial sectors.

The outcome of ―positive/cooperative employer coordination‖ with unions that is currently discussed in the literature can thus be viewed as a subset of the possible strategies of firms to either respond to demands of workers, or to preemptively deal with workers. The formation and tasks of an employer organization can be conceived as one possible strategic decision of firms in response to demands from workers, but

27 firms can choose from a set of other strategies. We can think of this as a specific kind of employer organization that negotiates with unions.

The outcome of strategic decision of firms can be characterized by two dimensions: First, employers can act individually or collectively. This means that firms can either address the concerns of workers independently, or they can join a formal employer association. Second, firms can choose to repress workers, or they can collaborate with workers. Collaborating with workers is defined as recognizing unions as a formal negotiating partner with attendant responsibilities in the wage- setting process and establishment of basic working conditions. This set of strategies characterizes different forms of inter-firm coordination, such as employer associations that seek to crush or promote the role of organized labor, or choices by firms to pursue these goals alone. The existing literature presumes that employer coordination is mainly collective collaborative strategies, ignoring the fact that employer organizations can also repress workers, or not be able to form such organizations, or have no incentives to form them. These choices of firm strategies are characterized as outcomes displayed in Table 2-1. In chapter 4 I discuss specifically a theory to explain these outcomes.

Collective versus Individual Collaborative Strategies

Firms can collectively decide to collaborate with workers, displayed in the lower right quadrant of Table 2-1. I term this a ―collective collaborative‖ strategy, which is characterized by an employer association that recognizes and negotiates with a counterpart union. This strategy is a formal term for the oft-studied ―corporatist‖ or

―class compromise‖ institutions, or modern coordinated market economies. The

28 features of such employer associations have been well-documented in the literature on corporatist institutions and varieties of capitalism. Under this strategy, firms agree to the rules of an employer association and recognize unions as a legitimate counterpart in the administering and enforcing of rules that constrain both firm and worker behavior. This kind of strategy is what is often assumed with current studies of the origins of employer coordination.

The employer association can negotiate a variety of issues with unions.

Member firms (or member associations, if the employer association is a peak level organization) must comply with the negotiated outcomes and institutional rules enforced by the association. Issues of negotiation can include wages, working conditions, insurance provision, training, and other worker concerns. Employer associations that oversee peak or sector-level collective bargaining frameworks, such as the modern BDA in Germany and the Swedish Employers‘ Federation (Svenska arbetsgivareföreningen, SAF), are examples of collective collaborative strategies.

Examples from the literature on employer coordination in continental Europe also focus on the major ―class compromises‖ such as the 1938 Swedish Saltsbajoden

Agreement between the and the SAF and the Swedish Trade Union Confederation

(Landsorganisationen, LO), and the 1899 ―September Agreement‖ in Denmark

(Galenson 1952; Swenson 2002; Martin and Swank 2008).

Under collaborative employer associations, firms and unions jointly administer and enforce rules that subject member firms and workers. These rules can include regularly scheduled collective-bargaining agreements, and measures to prevent and settle industrial disputes like strikes and lockouts. Additional tasks of collaborative

29 employer organizations can vary, such as the provision of worker training, or management of relations with relevant business lobbying associations. Overall, collaborative employer associations and unions constitute the building blocks of the

―class compromise‖ and industrial relations institutions in coordinated-market economies.

The enforcement of collectively negotiated wage agreements is an important part of the collective collaborative strategy (unless otherwise noted, I use the terms

―collective collaborative strategy‖ and ―collaborative employer association‖ interchangeably). The association can enforce agreements regarding minimum wages, maximum wages, or both. Modern post-war collective agreements often stipulate both, but initial employer organizations at the sector level often focused on either a wage floor or ceiling. As Swenson describes, firms might agree to set a minimum wage to remove wages as a source of competition (2002). Some firms would agree to this because other firms have a comparative advantage in producing goods at low wages (this may come from being better at resisting workers‘ demands). If all firms can abide by a strategy to pay below a certain wage, and that strategy is enforceable by the organization, they would all be better off. Other kinds of policies that are analogous to this issue include minimum working conditions.

A collective collaborative strategy might also entail an employer association in which firms agree to set a maximum wage. In a model of wage bargaining,

Wallerstein and Moene (2003) argue that conditions of labor scarcity of labor, particularly regarding workers with high skills, facilitate coordination to constrain wages. Unions that prefer reducing equilibrium unemployment would consent to such

30 inter-firm coordination over wage ceilings (Wallerstein 1990). The ZAG in Germany and SAF in Sweden in the 1930s and, discussed in chapters 6 and 8 respectively, are examples of such strategies of wage compression. A key objective of an employer association‘s wage ceiling is to prevent poaching of workers from firms; this assumption guides recent research about the historical development of collaborative employer associations. The consequences of the strategic dilemmas posed by imposing wage floors and ceilings are fully explicated in the next chapter.

As displayed in the lower left quadrant of Table 2-1, firms might decide to pursue a strategy of individual collaboration. They might decide to recognize a union within a firm, but not join any organization that would obligate it to do and comply by the organization‘s rules. This means they would not have to pay the cost of joining any organization in terms of dues, obligations involved with compliance, and wage premia above or below the designated collectively set wage. Still, the individual firm would negotiate terms with a recognized counterpart union. The firm would have no obligations to any associational rules.

Collective versus Individual Repressive Strategies

Firms can also choose to be highly coordinated to repress workers. I define employer associations that seek to undermine unionism and prevent workers from organizing as collective repressive strategies. This outcome is displayed in the upper right quadrant of Table 2-1. These repressive organizations contrast with the collaborative employer associations that recognize unions as bargaining counterparts.

Both collaborative and repressive organizations are collective strategies that firms can pursue in response to demands of workers and unions. While both kinds of employer

31 organizations constrain member firms with rules regarding the treatment of workers, they fundamentally differ in their view of the rights and responsibilities of organized labor. 6 Such employer associations were prominent during the late nineteenth and early twentieth century, and are an important part of industrial relations in many developing economies.

What are the objectives of repressive employer organizations? Repressive strategies are designed to reduce organizational capacities of workers. Firms in repressive employer organizations coordinate to prevent or minimize union presence and influence within individual firms and plants. They exist to prevent firms from committing to agreements with unions that would impinge upon firms‘ flexibility in setting wages. Other goals of such organizations can include the reduction of wages and working conditions, and autonomy in management of the labor market.

Repressive employer associations provide services to help firms to prevent or undermine union activity and preserve the employer‘s ―right to mange.‖7

What do repressive employer organizations do, and how do they function? As with collaborative employer organizations, firms can pay membership dues and agree to comply with the rules of the organization, in exchange for services that the organization provides. Rules of the organization include constraints on member firms from recognizing unions (though organizations can vary in the enforcement capacity of such rules). These repressive associations can use similar institutional apparatus as collaborative associations, but for different objectives. As with collaborative

6 Through the dissertation I use the terms ―collective repression‖ or ―repressive coordination‖ synonymously. 7 The term comes from the title of Harris‘ book, which discusses the history of attempts of firms to preserve autonomy before World War II (Harris 1982)

32 institutions, the most common arrangement is for member firms to pay membership dues into a central organizational fund. The organization has rules that constrain member firms from recognizing or making unapproved agreements with or concessions to unions. These repressive associations use similar ―institutional apparatus‖ as collaborative associations, but for different ends. The organization has rules that constrain member firms from recognizing or making unapproved agreements with or concessions to unions. In return, member firms have access to funds, strikebreakers, or replacement labor in the event of industrial conflict. The American

―open shop‖ movement during the early twentieth century and inter-war period, in particular the National Metal Trades Association (NMTA) in the United States, and the Central Office of German Employers‘ Associations (Hauptstelle deutscher

Arbeitgeberverbaend, HA) in Imperial Germany, are examples of a collective repressive strategy.

The repressive collective strategy entails a number of instruments to achieve their objectives. The repressive strategy can involve a number of different instruments with different costs, but the prominent one has been the use of lockouts and strikebreakers, with the effect of bleeding union funds and preventing further unionization. A common one as discussed above is the use collective funds to support firms when the organization decides to lock out workers in response to, or in anticipation of, labor strikes. The use of this instrument had an important goal of draining union funds and preventing further unionization. The employer association could target unions and union treasuries, with the aim of compelling workers to withdraw from unions or preventing them from joining. The lockouts can also be

33 targeted at firms or workers not involved in the initial industrial conflict to prevent further unionization. Other instruments include the use of strikebreakers and importation of substitute labor during labor industrial disputes; use of blacklists to alert firms to employees‘ previous affiliations with unions or political organizations; and lobbying the state for use of coercive personnel to enforce lockouts or repress workers. More extreme instruments include collective agreement to infiltrate unions and explicitly exclude union workers from employment. They can also include violent tactics, such as physical repression of strikers and forced support of strikebreakers and other forms of intimidation or coercion.

How would the collective repressive strategy of the lockout work? A common example is that repressive employer associations would form to respond to or prevent

―whipsaw‖ strikes from unions. These are strikes where unions can collectively isolate and target an individual firm and compel it to abide by union conditions. Such conditions might be higher wages, improved working conditions, and union authority over managerial decisions involving hiring and firing. Workers in the targeted firm or locality can strike, and its workers can survive without wages based on support from the collective union. If the firm is not part of an association and has difficulty finding replacement labor (or if unions are capable of preventing such replacement workers from appearing), than it would have to accede to demands made from the union in order to survive. If unions target firms one by one (thus the ―whipsaw‖ analogy), then they can impose on each individual firm union conditions. A collective repressive strategy could prevent unions from imposing inter-firm union agreements by responding to the union strategy with industry, region, or nation-wide lockouts. This

34 strategy of the ―punitive‖ lockout would drain union treasuries and prevent workers from making demands or organizing.8 The repressive strategic lockout can include targeting unions and union treasuries, with the aim of compelling workers to withdraw from unions or preventing them from joining. The lockouts can also be targeted at firms or workers not involved in the initial industrial conflict, or so-called ―innocent‖ third parties.

More extreme repressive strategies include infiltration of unions and explicit exclusion of union workers from employment. They can also include violent tactics, such as physical repression of strikers and forced support of strikebreakers and other forms of intimidation or coercion. Existing comparative politics literature ignores these kinds of strategies, despite their historical importance.

As with collaborative strategies, firms might decide to pursue such repressive strategies individually. They can choose to repress workers without joining an employer organization to do so. This might be because they do not want to be obligated to follow the organization‘s rules in case of industrial conflict. This outcome is depicted in the upper left quadrant of Table 2-1. Under this strategy, firms can use many of the same instruments as an employer association would, including lockouts, strikebreaking, importing substitute labor, lobbying for coercive measures, and intimidation. However, they do not pay any membership dues to an organization.

The critical distinction regarding lockouts is that firms can still lock out workers but play no role in the lockout decisions of other firms. Similarly, they do not receive any external assistance from other firms. Firms would choose an individual-level strategy

8 See Swenson (2002)and Fitch (1924) for discussion of this function of employer organizations.

35 if the costs are lower; the costs of a collective strategy include the membership dues as well as the obligation to participate in a collective lockout in the event of a labor dispute that may not involve that particular firm. The specifics of the theory of strategy choice are discussed below.

Section 2: Coding the Framework

In this section I address issues of coding the above firm strategies, and the appropriate units of analysis. The more complex framework proposed Table 2-1 contrasts with existing views which only classify firm strategies as either collectively collaborative or not collaborative. The proposed framework incorporates the choice of firms to organize to work against unions, and highlights the fact that firms or sectors with differing abilities to repress workers may not be able to collaborate with each other. Both the collective collaborative and collective repressive strategies include employer organizations that seek to constrain firms‘ freedoms to deal with workers.

But, they differ fundamentally in their view of how firms should treat their workers.

The theoretical framework also considers the possibility that firms may act collectively or individually. Firms may prefer, not to, or not be able to, form organizations as a strategy. This broader coding of business coordination is important for several reasons. The oft-cited examples in contemporary Western Europe in fact constitute only a portion (albeit an incredibly important one) of the possible strategy and coordination space. The early twentieth century as well as the post-war era has had employer institutions that varied widely in the sectors they encompassed and their stances on recognizing and policing agreements with unions. As Table 2-1 also displays, to differentiate countries simply along the dimension of ―high‖ and ―low‖

36 business coordination (as is common practice in current studies) comes at the expense of obscuring qualitatively different kinds of employer coordination.9 Further, this framework can be extended to current middle-income or developing economies.

How should these outcomes of different firm strategies be conceptualized temporally? A more practical and historically accurate approach is to consider what the most important firm strategy was for a country-time period. This is because with a few exceptions, as the empirical chapters demonstrate, there was relatively little annual fluctuation in the levels of firm coordination. Ultimately we should be interested in the most substantive qualitative changes in industrial relations institutions, which occur over longer periods of time. The theoretical task then is to account for these major changes in firm strategies.

The typology allows one to map temporal changes in business coordination— one can better measure what it means when business in a polity becomes ―more coordinated‖ over time, because firms can move to one part of the coordination space to another. As Table 2-2 illustrates (and is discussed in the empirical sections), the

United States transitioned from a collective repressive equilibrium to a liberal-market one, in which employer organizations play a minimal role in industrial relations.

Japan transitioned from a repressive equilibrium to a more collaborative one where industry and national-level employer associations cooperate with enterprise unions.

Germany and Sweden also transitioned from a repressive regime to a much more

9 The scale measures three indicators of degree of employer organization: scope of employers‘ organization (share of employers organized in one or coordinated multiple peak national association); centralization of power of employer organizations and unions; and integration of national peak associations into national policymaking forums. Each score sums on a 1-3 index; the sum of these indicators is the degree of employer organization (so the index runs 1-9). They do this for 1900, 1914, 1925, and 1938 for 16 states.

37 collaborative one; as is discussed in Chapter 6, the second half of the Weimar republic saw repressive employer associations try to undo collaborative institutions.

Repressive institutions continue to flourish, particularly in developing societies, and accounting for their emergence is necessary for understanding how the beneficial collaborative equilibrium in some states emerged.

Of course, within each country time period variation exists at the sub-national level in terms of the strategies firms pursued. Firms in different sectors may pursue different strategies (some sectors may have employer associations; others can have firms acting individually). Firms within a sector may pursue different strategies; for example, some firms in a sector can form an employer association while others choose not to join it. And across countries, some sector-level organizations can form inter- sector or peak-level employer associations that could be repressive or collaborative.

We can think of one of the four main outcomes as the strategy pursued by the modal large firm in a country. We can code what strategy that firm adopts in a particular year, or time period. We can then place that modal strategy and time-period into one of the four stylized categories in Table 2-2. This rule is clear and advantageous because it captures firm strategies that are affecting a plurality of a country‘s workers. We should want to code firm strategies in a way that captures the most important firm activity in labor markets. Table 2-2 presents examples of how specific country time-periods fit in this new framework; the examples in the table are discussed in the empirical chapters. This coding scheme allows us to map more specific changes in firm strategies over time as well. In the subsequent empirical chapters that test the validity of the new coding scheme and theoretical explanation, I

38 use detailed historical documentation of the actual strategies to justify the characterization of each country time-period.

Substantively, the outcome of interest is the formation of employer associations in major industrial sectors, and whether they collaborate with or repress workers. A country can be characterized by a firm strategy that is collective repressive even if it does not have an inter-sector or peak-level employer organization that represses workers. For example, if a country has a series of sector-level employer associations that each pursue a repressive strategy, then by definition the modal strategy is collective repressive (but it exists only at the sector or regional level). This might be because the state shares the costs of repression (see the theoretical and empirical discussion in the rest of the text). This logic applies analogously to collaborative employer associations. As the empirical section demonstrates, much of the substantive outcomes of interest involve major sector-level employer associations deciding to switch strategies or solidify them by coordinate with other sector associations. The most interesting variation cross-nationally and over time, particularly before WWII, are the major industrial institutions that characterize whether firms have engaged in a strategy of ―class war‖ and whether they can pursue such a strategy collectively.

It is important to note that just because an economy has some sectors that do recognize workers, it does not necessarily follow that the modal strategy is collective collaborative; if the economy has major sectors that still repress workers, we would code it as a repressive strategy. Conversely, many collective strategies in states take the form of agreements among industrial sectors (and by extension, among firms in

39 those sectors). To explain this inter-sector outcome, we need to account for the fact that some firms can collectively act to form sector-level associations, while others firms cannot form such associations. I argue that the ultimate national-level outcomes are the result of failures or success of sectors to bargain with each other. In some cases, outcomes are the result because one set of firms is unable to collectively act and form a faction, giving other firms that have collectively acted an advantage in attaining their preferences. I return to this coding point when discussing the theoretical predictions.

The coding rule I present does suffer from some deficiencies shared by other existing coding schemes. The main problem is that the blunt coding rules ignores to some extent heterogeneity within an economy, in that it reduces each country-time period to one ―modal firm strategy,‖ which, as discussed below, I argue constitutes on form of institutional equilibrium. As discussed above, multiple sectors in an economy could be pursuing different strategies. Even within one industry, firm strategies could differ. For example, large firms in a raw-materials industry may find it easier to repress workers, whereas firms in highly skilled manufacturing industries in the same economy could engage in collaborative strategies. Coding the country time period as one out of four possible strategies in Table 2-1potentially neglects other strategies enacted at the sector level. Moreover, coding the modal firm strategy will involve some qualitative judgment if data are not easily available. However, the modal firm strategy, classified as one of the four, is still informative about the general state of industrial relations in a state. In the next chapter, I elaborate on how each of the four categories of firm strategies can arise as a consequence of different strategies pursued

40 between industrial sectors. The theoretical mechanisms posited can partially deal with the problem of coding a singular ―national-level‖ firm strategy in heterogeneous economies.

Section 3: Distinguishing Firm Strategies, Labor Strategies, and Industrial Outcomes

The above characterization of the dependent variable focuses on the actions or strategies of firms, and characterizes such collective strategies an industrial institution which can be repressive or collaborative. I do so to allow for parsimonious and easy coding of country time-periods as described above. Of course, though, the existence of such strategies cannot be studied independently of the strategies of workers, the state, and the actual implied outcomes of such strategies. I deal with these concerns more fully when I describe the theory for variation in firm strategies, but in this sub- section I address how labor strategies and actual industrial outcomes fit into the proposed coding framework.

Labor strategies and equilibrium outcomes

We can characterize the outcomes as equilibrium labor market and equilibrium institutional outcomes. The guiding assumption of the above is that firms are choosing strategies in response to workers‘ demands; the nature of those demands is fully explicated below in the next chapter. Each outcome of firm strategy choice entails a counterpart action on the part of workers. Workers, anticipating the firm strategies pursue set of strategies that are the ―counterpart‖ to those described in Table

2-1. Consider first the outcomes of firm strategies to pursue objectives individually.

If the modal strategy of firms is at the individual level, then by construction workers are not organized in encompassing unions as well; there may be sector or firm-level

41 unions, but negotiations between firms and workers are done without an institutional framework of an association. Now consider the group of collective strategies. Under a collective repressive strategy, there may be a counterpart union that is organized or is trying to organize that the employer association is acting against. Under a collective collaborative strategy, by extension there is a counterpart organized union that obligates workers by the similar guidelines as firms. The goal of the theory then is to predict a set of different equilibrium outcomes that characterize industrial relations, wherein firms and unions are engaged in ―best reply‖ strategies. In chapter 3 I describe the underlying variables that explain the resulting matrix of such strategies in a polity.

Relation to industrial outcomes

The outcomes in Table 2-1 describe the different of industrial institutions we are likely to observe. However, these strategies and institutions are distinct from other measurable outcomes such as degree of industrial conflict. The choice of firm strategies displayed in Table 2-1 does not necessarily imply that those strategies are successful from the firms‘ perspective. As the theoretical section below clarifies, firms choose based on relative net benefits of each strategy. Thus, choice of a particular strategy does not necessarily mean we observe the ideal outcome of that strategy. As a result, the actual observed industrial outcomes—such as degree of wage equality or industrial conflict—can differ from the observed choice of firm strategy.

As an example of this divergence between strategy choice and industrial outcome, the choice of firms to engage in collective repression does not necessarily mean they will be successful in crushing of the unions. The only observable outcome

42 is the existence of an association that pursues such a strategy. Unions can still be organized or try to organize against a repressive employer association: the result is a form of familiar class conflict. Alternatively, repressive collective strategies can be successful, and we can observe reduced union density and the success of such organizations. Another example of divergence between strategy choice and observed outcomes is that industrial conflict in the form of actual strikes and lockouts, or violent activity, can occur in an economy if a firm chooses repressive individual or repressive collective strategy.10

Repressive strategies and elite coordination

The above distinctions are also important because the distinction between repressive and collaborative strategies of firms also informs a growing literature in comparative politics that examines conditions under which democratization occurs and when democracy is sustainable (Boix 2003; Acemoglu and Robinson 2006). One of the claims of the literature is that structural conditions—predominantly initial inequality in wealth—creates incentives for elites to either repress or redistribute via extension of democratic (franchise) rights to the poor. The incentives to redistribute exist because elites want to prevent a revolution which would be costlier to them then democracy. In Boix‘s account, higher initial inequality coupled with lower asset mobility gives elites a larger incentive to engage in repressive strategies.

Missing from these accounts of inequality and democratization is a discussion of the role that firms and other important elites played in retaining status quo institutions. Importantly, the mechanisms they posit regarding establishment of

10 An analogous example of this divergence is in the study of international relations. Theories can predict decisions to go to war or not, without necessarily predicting the victor, or number of casualties.

43 democratic institutions relates to strategy choice of firms regarding treatment of workers. Firms and their respective business organizations played an important role in repressing workers directly or in lobbying the government to do so. The availability of repressive strategies to firms—either individually or collectively—is an important aspect of variation that current scholars of both contemporary industrial relations and historical causes and consequences of ―varieties of capitalism‖ have ignored.

One of the implications of the models about incentives for elites to democratize is that firm owners might face similar incentives to grant labor rights to workers. These models are attentive to the collective action problem faced by the poor

(or by workers). They assume first that industrialists are inherently pro-labor (with very different preferences from large landholders, for example). Second, they assume that the industrialists and other elites are capable of collectively acting. Both assumptions are problematic. As the preceding discussion shows, collective action among elites—specifically important firms—varies greatly as well. In this recent influential framework for understanding democratization and inequality, problems of collective action of elites and firms are ignored. By contrast, scholars of varieties of capitalism and the historical origins of economic institutions have ignored the existence of repressive strategies. The conceptualization of firm strategies presented here addresses both kinds of outcomes and addresses these omissions. In the following chapter, I turn to the task of explicating a theory that can help explain this richer set of outcomes and changes in industrial outcomes in countries over time.

Table 2-1: Classifying Strategies of Firms at the National Level

Individual Collective

Repressive Firm represses Organization that workers represses workers

Collaborative Firm recognizes Collective bargaining union agreements

Table 2-2: Historical Examples of Firm Strategies

Individual Collective

Repressive United States, United States, 1880-1914; 1973 – 1918-1929 Germany, 1895-1914

Collaborative United States, Germany, 1918-1924, 1950-1973 Germany, 1948 – United States, 1935- 1950

Chapter 3: A Theory of Firm Strategies

In this chapter I outline a new theory of firm strategies that can explain cross- national and temporal variation in the choice of firm strategies displayed in Table 2-2.

As the new coding of the outcome variable in Table 2-2 implies, the existing accounts of variation in industrial regimes underemphasize two historical and strategic dilemmas which form the basis of my alternative theory. First, they underemphasize the redistributive threats posed by unions and workers‘ parties to firms in the early twentieth century and inter-war period, and how such threats relate to repressive actions of firms. Second, the existing accounts do not theorize about the dilemmas of collective action that faced nascent employer organizations at the sector and inter- sector level. Specifically, there was heterogeneity of interests among firms and sectors that affected the viability of long-term collective action among firms. As subsequent chapters demonstrate, the evidence on the development of the largest employer organizations, and institutions of economic coordination between firms and workers, is more consistent with this proposed theory.

In this chapter I present a theory that addresses these flaws and that can account for the variation described above. The first section describes two key variables that affected decisions of firms to form different kinds of employer organizations and gives predictions based on levels of those variables. Sections two and three discuss the most important mechanisms linking the changes in the variables to changes firm strategies. These sections also address concerns about endogeneity and conditions under which the theory holds. Section four contrasts the theory with existing approaches to employer coordination.

Section 1: Main Explanation and Summary of Predictions

In examining the decisions of firms in the late nineteenth and early twentieth century across most democracies in that period, the most important considerations were about the responses to the threats posed by socialist or labor-supported parties and growing unions that had made varying demands regarding managerial control in the workplace, working conditions, wage schedules, and in some cases, changes to the social order. Firms and business organizations faced varying uncertainty and magnitudes of threats from workers, as well as the degree of economic and political mobilization of workers. Firms confronted the questions: should they battle against or cooperate with workers? Should individual firms band to do so. If they agreed to collectively act, or to bargain with unions, what would make either agreement self- enforcing? This set of dilemmas much more cogently captures the formation and dynamics of initial employer organizations and industrial institutions.

Two key variables account for the variation in firm strategies cross-nationally and temporally. The first is the redistributive threat to firms posed by workers. This is characterized by the amount of expropriation that workers‘ organizations (unions and political parties) sought. The threat to firms posed by workers affects the incentives of firms to pursue collaborative or repressive strategies. This threat captures the costs of repression that firms would need to undertake in response to demands from workers, to reduce that threat. The second variable is industrial heterogeneity which captures the mechanism of ease of collective action among firms.

This structural variable affects incentives of firms to act collectively or individually.

Table 3-1displays the predictions of this theory based on low and high levels of the two main independent variables. I first preview the predictions of the theory, and then describe the logic of each prediction. It is important to note that of course these outcomes are probabilistic—we are interested in outcomes that are most likely, given changes to the independent variables. Consider first the upper right cell. If industrial heterogeneity is high and the threat posed by workers‘ parties is low, firms are more likely to pursue an individualized repressive strategy. Given the same low level of labor threat, lower heterogeneity allows for easier enforcement of agreements among firms in different industrial sectors. This will make collective repressive strategies more likely (the outcome displayed in upper left cell).

If industrial heterogeneity is low, and if the threat to firms posed by political threat of labor is low, firms have less of an incentive to collaborate with workers. At the lowest levels of these variables, industry-specific structural variables (such as industrial concentration within a sector) will have the greatest role in determining whether firms engage in individual or collective strategies against workers.

But, if the threat from workers increases to moderate levels (and if heterogeneity remains low), firms will have a greater incentive to resisting worker demands by forming collective and repressive organizations. When the threat from workers becomes high, though, firms have incentives to pursue a strategy of inter- class collaboration (the bottom left cell). Under conditions of higher industrial heterogeneity, collective action is more difficult; as a result, individual firm strategies of collaboration with workers would be more likely (the lower right cell). This theory can explain much of the cross-national and historical variation in the industrial

48 institutions and strategies of firms. It explains why in some country time-periods individual firms or sectors act without a collective organization, while in others industrial relations are characterized by an inter-sector, cross-class agreement. As I illustrate in the empirical chapters, the transitions from these initial equilibria better capture the important events and patterns in industrial relations throughout the late nineteenth and early twentieth century.

At historical junctures where workers mobilized, firms faced higher costs of expropriation and higher costs of repression; as a result, they had incentives to switch strategies and engage in collaborative agreements with workers. In some instances such agreements pre-empted intervention from the state or further radicalism from workers. This explains the timing of key industrial agreements and why they were cooperative with workers. In economies with lower industrial heterogeneity, prevalent in many small European states, collective action was easier. But even in these states, the initial strategies pursued by firms against demands from workers were repressive.

The threats posed by workers‘ parties and the fears of social revolution, both when labor parties began contesting elections and their increased influence after WWI, induced adoption of more collaborative strategies. By contrast, in country-time periods where such threats were much lower, firms preferred to adopt individual or collective strategies of repression. The main insights to the theory involve the response of firms to the potential distributional threat posed by the politicization and mobilization of workers, and the enforceability of any collective response of firms to the threat.

This parsimonious explanation accounts for a wide range of temporal and

49 cross-national variation in strategies pursued by employers. It explains why American employers have established antagonistic institutions that repressed workers. The theory also accounts for change in the political strategies of German firms, from a strong collective and repressive stance against organized labor to a willingness to recognize unions. Additionally, the theory predicts when we would be more likely to observe collective strategies at the national level as a consequence of bargains between different factions of capital, and when we should observe strategies that emerge because one faction can imposes its preferences.

There are three contributions of this theory over existing explanations. First, it specifies conditions under which firms repress workers and undermine labor organizations. Second, it specifies when firms decide and are able to act collectively to achieve labor-market goals. Firms might prefer to act collectively to either repress or cooperate with workers, but be unable to do so. Third, the theory incorporates the role of the state more directly by integrating the costs of repression—it demonstrates how firm strategies relate to the actions of the state, and can be designed to influence state activity.11

Section 2: Mechanisms of the Threat from the Left and Endogeneity

I now turn to a fuller explication of each of the independent variables and describe the most important mechanisms and observable implications of the theory.

While there are many reasons why the threat to firms posed by workers and the degree of industrial heterogeneity would affect firm strategies at the sector and country level,

I focus on the most important mechanisms that link the two variables to decisions of

11 In particular, the state and firms can share the costs of repression.

50 firms. I focus on the costs of repression, which affects choice of repressive strategies, and difficulties in enforcing agreements, which affects collective strategies.

The threat from the left and the level of cost of repression

The threat posed by workers affects the choice of collaborative or repressive strategies through the mechanism of the level of the cost of repression. I argue that firms must incur some cost for repressing these workers. This can be thought of as lost capital stock, lobbying the state for intervention, or reduced profit stream from expending resources to repress workers.12 Substantively important examples of those costs include costs of administering and forgone profit from lockouts, use of strikebreakers, and search costs for substitute (and possibly lower quality) labor during industrial conflict. If the threat posed by workers remains low, firms prefer to repress workers. If the threat posed by them increases, and costs of repression increase, at some threshold firms will prefer to pursue collaborative strategies with workers.

These collaborative strategies meant formal agreements with unions. Rather than bear higher costs of repression as well as the probability of higher conflict with workers, firms in some countries chose a moderate strategy of accommodation. This strategic logic best explains the ―class compromises‖ that emerged before WWII. This mechanism also implies that some employer associations collaborated with moderate unions and their political parties against radical workers, but also against radical firms that still pushed for repression; this element of collaborative strategies is discussed below.

12 Boix (2006) models repression as a parameter ρ ; firms‘ profits are after it has paid out wages and beared this cost. Acemoglu and Robinson (1986; Acemoglu and Robinson 2006) model the costs of elite repression as lost capital stock K or loss output from conflict.

What was the threat to firms posed by unions and workers‘ parties? The threat posed by workers can be decomposed into a union wage threat posed by workers and taxation threat posed by a workers‘ party. Direct strategies of redistribution included imposing costs on firms through strikes, and making demands regarding wages, working conditions, and policies. As discussed in the previous section, unions could target individual firms and impose costs.13 An important aspect of this ―threat from the left‖ was strikes that could threaten the industrial order. The discussions of the initial unions spanning geographical or industrial boundaries, as well as those of the early left-wing political tacticians, are replete with contemplation of mass versus local strikes, and the most effective means of securing working conditions and wages for workers. During the late nineteenth and early twentieth century as well as during the inter-war period, the use and threat of strikes were a way to extract concessions. This forced firms to confront the two dimensions of action: collective versus individual action and repression versus collaboration. Even though some social democratic or labor party leaders thought that mass strikes would not necessarily lead to revolution, and of course much disagreement existed within leaders of unions and left-wing parties as to which tactics would best achieve goals of redistribution, firms reacted to the threat posed by such strikes.14

13 Kennan (1989) provides an overview on the costs of strikes to firms to unions. In a model by Wallerstein (1991), this parameter of repression costs is taken as exogenous. He models profits during a strike as S = S(µ, x); the key parameter is x, which is a vector representing all factors besides unionization that affects profits during a strike, such as ―level offered by an employers' association, the likely reaction of the government, legislation covering industrial conflict, etc‖ (Wallerstein 1989), 123). This section more explicitly theorizes about the sources of variation in x and links it to individual firm strategies. 14 Rosa Luxembourg in fact had observed industrial unrest in Belgium and concluded that a mass strike (or threat of such a strike) would not work to bring about socialist revolution in Germany. She reached this conclusion based on dispersion of industrial centers, and the lack of immediate threat that such a

The debates among the German SPD leaders in the early twentieth century are indicative of the threat posed by unions and labor parties to firms, as well as the strategic decisions that unions faced. Regarding the crippling strikes in Belgium in the early 1900s, Rosa Luxemburg noted the proximity of Belgian industrial centers to one another as a factor that would increase the chances of success for a mass strike.

Luxemburg thought that that strikes would be most effective in geographically concentrated states, but that Germany had too much distance among its industrial centers for a mass strike to be successful.15 Rudolf Hilferding countered that the threat of a mass strike could be used to fight for universal suffrage. Even the more conservative party leaders like Eduard Bernstein ―advocated consideration of the mass strike to defend or acquire universal, equal suffrage.‖16 Similar kinds of debate about the viability of mass strikes and actions of unions to bring about redistribution or revolution existed of course in other states.

In response to varying levels of demands from workers, if the firm wants to individually repress workers, its profits Π are what is left over after it pays wages of workers and bears the costs of repression.17 Let this cost of repression for each firm be ρ. The higher the cost the firm bears to repress workers, the lower wages it will have to pay. But, of course such costs of repression detract from the firm‘s total profit. If the firm wants to individually repress workers, its profits Π are what is left strike would pose to political authorities in the center. But, the mass strike was still debated in 1912 and again after World War I (Schorske 1955). 15 See Schorske 1955, 34, excerpted from Luxemburg, Werke, 3, 356-357. 16 These discussions are from the SPD publication, the Neue Zeit 22 (1), 137-141. Of course unions and workers‘ parties are strategic actors as well, and can have different preferences about the amount of redistribution they seek. The implications of this are discussed in the next section. 17 This notation is meant to be consistent with Boix (2003). Acemoglu and Robinson (2006) model the cost of repression as some loss of capital stock K. The cost can be normalized as a cost per worker, for purposes of simplicity I ignore such notation. Formally, let the firm‘s profit be Π = r – w(l) - ρ

53 over after it pays wages of workers w, and bears the costs of repression. Costs of repression can include a range of activities, including funds for internal strikebreaking or monitoring union activity. If a firm expends a higher cost to repress workers, it will get to pay lower wage to workers. But, of course such costs of repression detract from the firm‘s total profit. For example, large firms like US Steel in the late eighteenth century and Krupp in Germany in the early twentieth (among many other large manufacturing firms) simply dismissed workers if they attempted any inter-plant unionization. If the firm individually collaborates with workers, they pay a higher wage to workers, but do not pay costs of repression. If the costs of repression fall, then firms have a greater incentive to switch from collaborative to repressive strategies. The problem for individual firms in this scenario is straightforward: once ρ

< w, there is an incentive for firms to individually repress.18

Clarifying the redistributive threat and workers’ strategies

Before I turn to the mechanisms that link heterogeneity to collective strategies,

I address concerns about endogeneity and clarifying elements of the independent variables. In particular, how do we know that the proposed causality and sequencing of actions runs between the redistributive threat posed by workers and left-wing parties to firm strategies, and not the other way around? This theory takes as exogenous the redistributive threat and the structural conditions affecting enforcement among firms. But repressive strategies were designed to inhibit worker organization

18 Larger firms should have a comparative advantage with lower costs of repression (per worker); it should be easier for them to inhibit union organization as they may have higher initial funds and capacity for internal strikebreaking or monitoring union activity; this lower cost of repression. Small firms by contrast would have to expend a lot, and therefore could find it less costly to accede to some union demands; or, they may simply be unable to bear any costs.

54 and respond to and prevent the emergence of such threats, and so firms may have engaged in such strategies to pre-empt such threats, as discussed above. A fuller version of the theory would have to account for a dynamic process in which firms and workers engage in multi-period responses to each other‘s actions. As Chapter 2 describes, though, within a certain time period, the modal firm strategy can be characterized as one of those from Table 2-1.

To address these concerns I describe the set of strategies that allows for the outcomes to be considered equilibrium firm and worker strategies in Table 2-1. I focus on the strategy set of workers needed to defend the assumptions to obtain the predictions in Table 3-1. In particular, I focus on the strategic problem of workers‘ organizations in deciding how much of a redistributive threat to pose. I argue that workers can choose a moderate degree of redistributive threat that would still produce a repressive (collective or individual) response from firms, but ex ante the expected value of that outcome could still be preferable to choosing a high redistributive threat to try and impose on firms. Recall from section 2 in the description of the outcome variable, that outcomes such as repression do not necessarily imply complete success of such repression (such as complete victory in class war or complete submission of workers); it is only that firms prefer such strategies to collaboration.

Strategies of workers

In order for the theoretical predictions in section 1 to be coherent, some discussion of workers‘ incentives in making demands for redistribution is warranted. I theorize that workers can choose different levels of redistributive threats to pose. In order for a pronounced amount of redistribution or policy goal such as implementation

55 of a socialist program to be more than a minimal threat, workers must have some cost to bear in demanding redistribution, and workers can choose this level. This cost can also signal the resolve of workers to attain their redistributive demands.19 An important example of this cost that workers bear would be the choice to participate in local, sector, or national-level strikes that can be supported by other unions and nascent workers‘ parties. This explains why workers would not be able to pose a credible threat to firms at all times, as workers must pay the costs for an increased threat.

The discussions within the German SPD about the utility and costs of a mass strike noted above, which divided leadership among unions and within the party between 1905 and 1912, are illustrative about the costs workers bear in posing a redistributive threat to firms. In some cases, the willingness to bear such costs might be because of extreme political orientations from workers, whose demands could include total expropriation of firms and worker cooperatives in the place of private ownership. These unions could have syndicalist, communist, or variants of socialist demands. Such radical unions existed in some mining sectors in France, and dockworkers and some manufacturing sectors in Germany before WWI and in the immediate post-WWI period. But in other cases it could be due to highly skilled workers seeking more control over local labor supply, training, and wages—higher infringements into managerial autonomy then firms desire. This is characterized for example by the Swedish construction industry in the late 1800s and early 1900s and in the craft-based US manufacturing industries. As discussed more in the empirical

19 As the industrial relations literature in economics discusses, workers can signal resolve by bearing costs. See the discussion of models by Kennan (1986) and Ashenfelter and Johnson (1969).

56 evidence from the US Open Shop period, craft-based unions, in particular the

International Molders Union (IMU), and affiliate unions of the American Federation of Labor (AFL), triggered a series of industrial conflicts with member firms of the

NMTA and NEA due to conflict over compliance with rules about wages and control over apprenticeships (Harris 2000; Fine 1995). These were higher skilled workers participating in an intense class war (by American standards) with firms. This last example is the classical kind of class conflict described by Korpi and standard descriptive histories of local employer associations (Leckebusch 1966; Korpi 1978).

In order for the theoretical predictions to hold, the outcomes for workers under each set of firm strategies displayed in Table 2-1 have to be preferable to some other set of strategies that workers could choose which might induce a different response from firms. In other words, as described above, the theory needs to predict outcomes that are indeed workers‘ and firms‘ best responses (Przeworski and Wallerstein 1982).

For example, if firms choose either individual or collective repression as a strategy, in response to worker demands, the outcome for workers must be preferable for them than the outcome had they not made such demands. Analogously, if firms choose to collaborate with workers, the outcome implies that workers agreed to this set of institutions and preferred it to some other outcome (such as imposing a higher degree of taxation on firms, without such institutions, or outright ownership). As discussed below, each of the outcomes represents an equilibrium set of strategies pursued by firms and workers.

I describe the theory symmetrically from the workers‘ perspective. The key to this aspect of the theory is that firm strategies have a limited impact on the level of

57 redistributive threat that workers ―choose‖ to pose. Workers can make high demands

(where they bear higher costs for redistribution) or low to moderate demands (where they bear lower costs of redistribution). High threat demands are those that involve bearing some costs, such as a major strike. That is, making demands without imposing a cost is not very credible (demanding higher redistribution in a party platform, for example). In effect, then, workers are choosing how much costs to bear and impose on firms, to induce a set of firm strategies.

Workers do not know with certainty ex ante what level of threat of redistribution—what costs they would bear—would induce firms to engage in collaborative strategies. If workers make high demands for redistribution and bear the high costs of doing so, that means the cost of repression for firms is too high and collaboration with workers is preferred. The result is collaboration, though this might occur after some protracted conflict. Again, only sufficiently high demands from workers can induce such collaboration, though workers do not know that ex ante.

But because workers do not know what level of threat will induce collaborative strategies from firms, if workers make insufficiently large demands for redistribution and bear those costs of those threats, firms may respond with repression as long as it is less costly on their end (as long as ρ

58 firms‘ perspective, such moderate workers‘ demands mean firms have less to lose if they do not repress, but repressing is still less costly for firms than acceding to such demands.

Finally, if workers make small demands (presenting a lower threat for redistribution and incurring themselves limited costs), the cost to repression for firms is lower. Under this scenario, workers suffer from repression, but the expected loss if they had demanded more and suffered more costs of repression is greater; thus, workers could choose a lower threat in terms of redistribution.

This logic helps explain why the assumption that firm strategies respond to the threat from workers is defensible, and addresses the question of why workers would not pursue a strategy such as mass strikes more often. This is because such a strategy—posing a credible threat of redistribution—is costly for workers and entails a gamble that bearing such costs will be sufficient to induce firms to engage in collaboration. The strategy is worth it if workers can force collaboration. If they cannot force collaboration, then the more extreme the demand is, the worse consequences they suffer from firm repression. For workers, suffering these costs of repression would be preferable to bearing the costs in demanding more redistribution and failing. Because they cannot be sure that bearing the costs of demands for redistribution would result in collaboration, for some workers, moderating demands is preferable. This assumption is consistent with the historical pattern of a general monotonic increase from very low degrees of worker mobilization and demands for improved working conditions, to moderate levels and calls for more state protections of working conditions and higher wages. It is at this point where firms have

59 historically engaged in repression. Generally, workers and labor parties increased their efforts at organization and electoral contestation, even in the face of repressive strategies from firms and the state.

Empirically, this assumption is partially confirmed by the noted ―lag‖ between unionization and the employer response to such threats. The exogenous shifts in the threat posed by workers‘ parties and worker mobilization, as well as changes in structural conditions, are more consequential than the dynamic effect that firm strategies have. The sudden intervention of the state before and after both World Wars suggests that parameters changed suddenly, inducing radical changes in firm strategies. The return of working soldiers in Germany after WWI, for example, posed a much higher cost of repression for firms. Additionally, state imposed wage agreements during the war suddenly became less enforced, giving firms who had pre- existing incentives to defect the opportunity to do so. Naturally, future research designs, beyond the scope of this dissertation, should orient themselves toward a more rigorous identification strategy to isolate causal effects of the independent variables posited here.

Section 3: Industrial Heterogeneity, Variance in Cost of Repression, and Costs of

Enforceability.

I now turn to the mechanism linking structural conditions of heterogeneity to collective responses. The variable of industrial heterogeneity affects the outcome of whether strategies are individual or collective through the mechanism of enforceability of agreements among firms. In particular, higher heterogeneity means there is higher variance in the costs of repression and costs of enforcement ρ among firms and

60 sectors, making it less likely that sectors will collectively act.20 Consider an economy with a distribution of costs to repression across different industries. If there is a lower/higher variance in the costs of repression ρ within the economy (due to lower/higher economic heterogeneity across sectors), both firms and sector-level associations will be more/less likely to share preferences to collectively act, and the costs of sanctioning or deterring potential defectors from any agreement would be lower/higher. Similarly, whether such organizations occur at the sector level depends on heterogeneity within the sector. Lower heterogeneity increases the likelihood of collective action and the formation of employer organizations. This mechanism of enforceability affects whether employer associations—collaborative or repressive— remain viable. Heterogeneity then predicts the difficulty of enforcing agreements among firms and the difficulty of inter-sector cooperation.21

To make the theory tractable I assume that ρ characterizes the individual cost of repression pays, the share of the total collective repressive strategy P, as well as the individual cost to the firm of enforcing a collective collaborative strategy. Consider the cost side of this collective repressive strategy. Firms have to contribute some

20 For tractability of the theory, I let ρ row refer to both the costs of repression, and for an individual firm, the individual cost of a firm to participate in enforcing in collective collaborative strategy. 21 Other structural factors within an industry affect whether collective action in the form of an employer association for firms within a sector is possible. Much of the literature in the field of industrial organization argues that industrial concentration within a sector should affect the probability of cooperation within a sector. The proliferation of sector-level associations is evidence that more homogeneous firms within a sector (increasing substitutability of products and similarity in firm sizes) are most likely to form associations. This is consistent with the exhaustive literature of cartel formation. Empirically, most initial employer associations (as well as trade associations) are formed at the sector level (often within a sub-national, such as regional or city-level unit). Organizations of firms that have product homogeneity (in fact, the same product) have been historically more likely to form. (The classical literature on firm theory also predicts that firms that produce goods that are perfect complements have incentives to merge or vertically integrate). An alternative argument is that in industries with high levels of HHI, costs of collective action should be lower. The largest firm is more willing to bear costs of collective action to provide the public good for all firms, and maximize profits. See McCaffree (1962) for discussion of this point.

61 portion of the total cost P required to set up a central association that can repress workers. Part of this cost is simply annual dues (based on number of workers or production). These dues allow the association to use a wider range of instruments of repression than the firm can individually. In fact, some of the instruments—such as targeting third parties and providing strikebreakers—may not be possible for individual firms to pursue. In return, firms obtain a benefit of getting to pay lower wages to workers. If each firm contributes enough portion of the total cost ρ, the net effect of lower wages means a collective strategy is preferable. This cost ρ also includes the costs required to set up the employer association‘s monitoring and enforcement apparatus. For some firms, the individual cost of ρ is analogous to an insurance premium, in the event of industrial conflict.

However, another component of ρ is that membership in the association obligates a firm to participate in a lockout or industrial conflict even if the firm is not experiencing higher wage demands from workers. The firm must bear foregone profits from work stoppages due to industrial conflict. For firms that are not facing problems from workers demanding higher wages, this of course can be viewed as unfairly costly. This is the power of the collective strategy: by punishing workers who have not made demands, the association creates internal pressure within unions and pressure from other workers for unions to end the strike. Funds from the employer association can support firms who engage in the lockout. Without an employer association, an individual firm could have to accede to union demands. So the cost includes this premium and expected cost of lost profits from having to lock out workers who may not be involved in a strike. The issue is how much this varies

62 across sectors. This in fact is the standard view in the older descriptive literature on local or sector-level employer associations, as aggressive responders to union efforts

(Bonnett 1956; Gladstone 1984).

Note that even if the variance in ρ across sectors is lower, this only means there is a higher likelihood that collective action is possible. Even with a moderate number of firms, and lower variance in the costs of repression, the classic free rider problem still holds. This problem exists even if all firms want to pay workers‘ minimal wages, and even in the face of a threat from unions targeting firms individually. How is free- riding under this scenario possible? Even with a small number of firms, enforcement and monitoring can be imperfect. At first glance, this would seem implausible—in an employer association, the central office keeps records of members, what purposes funds are used for, who has contributed, and various attributes of the firm. In times of industrial conflict a firm can appeal to the central office for assistance (or the office itself is proactive), and on the basis of its previous contributions, the office can decide whether the firm merits support from the association. But a logical form of free-riding involves firms covertly not participating in the association mandated lockout.22 If enough firms choose not to participate in the association‘s mandated strategies, the lockout will not have the desired effect of reducing union mobilization and wages.

The structure of the collective action problem for each firm can be represented as follows. Let Πu be the expected total long-run cost to an individual firm of acceding to demands from an industrial union. Let Πnu be the profits from not having

22 Of course, unions face a symmetrical problem. As is told frequently from the industrial sociology literature, workers may not want to participate in a strike called if they do not have a particular demand for higher wages or other demands of the unions. Workers might also have an incentive to free ride off the strike if firms have symmetrical preferences in keeping them producing.

63 to accede to demands. For these kinds of firms, Πu < Πnu . We can measures the net benefit of a collective strategy by the difference in profits it generates, represented by the quantity (Πnu - Πu). Again, let P < Πu be the total expected cost to form a collective association to deal with the labor threat. The value of P can be thought of as the sum of individual firm contributions to an employer association, or the employer association‘s central budget. If there are n firms in a sector for whom the cost of repression is less than the lower profits from acceding to union demands, they would all be better off to strike a deal in which they jointly contribute a total of P to the organization, with each one contributing no more than the reservation wage bill that they would have to accede to, in the absence of an organization. The problem is that P is divided among the n firms, and firms may disagree over how much of the total cost of the association each should contribute. The higher variance in ρ across sectors still lowers the probability of agreement upon a collective strategy, whether it the strategy is repressive or collaborative. The point is that when ρ differs greatly across sectors, collective enforcement of a strategy becomes more difficult.23 The conclusion from this simple description is that even if firms would be collectively better off with a collective strategy, they might not be able to act due to differing costs of repression, as well as free rider problems. Unions might have their way, as happened in the early textile industries in the United States (Carpenter and Thomas

23 For example, if the cost to each individual firm of the threat from organized labor is 100 (Πu = 100), and if the sector has 10 firms, the total cost of the threat is 1000. As long as the cost of pooling resources to form an employer organization is less than 1000 (P = 1000), the firms would be better off forming such an organization. But no firm would like to contribute more than 10 each to the organization. If each firm contributed an equal amount, they would contribute Πu /n to form the organization. We can think of n as the number of effective firms or the degree of industrial concentration. If n is high, collective action is difficult. As n decreases collective action among firms should be more likely. At lowest of levels of n, collective action should be most probable.

1972). But, if large firms can provide selective incentives, or are willing to bear a disproportionate cost of the collective strategy, firms may have an incentive to pay the cost for such a strategy. And as described later in the chapter, if the state is willing to assume some cost of repression, collective action might also be possible.

Generalizing the collective action to other cases

The stylized example and logic was described for a particular industry where all firms preferred to pay workers minimal wages. But the notable and useful insight of this example is that it applies to almost all instances of firms attempting to achieve jointly higher profits, by regulating the labor markets. Regardless if the collective strategy among firms is repressive or collaborative, firms face incentives to defect from the organization. As discussed in chapter 2, collaborative action among firms that recognize and work with unions can have the following functions:

Establishing minimum wages. Suppose firms want to take wages out of competition and wish to organize to monitor each other to prevent ―chiselers.‖ Unions can play a role in helping to enforce such agreements. Consistent with the above logic, individual firms have an incentive to under-cut the agreement and pay workers less; this is the same logic for why producers have incentives to cheat on price or quota cartels. Note that this kind of collective strategy means that firms agree to help prevent ―bad‖ firms from reverting to individual level repressive strategies.

Establishing maximum wages. Firms may wish to set a maximum wage to prevent poaching of workers. Again unions can play a role in helping to enforce such agreements. Particularly if unions value lower unemployment in equilibrium, they help enforce wage restraint. Again, though, individual firms have incentives to defect

65 from wage ceilings and pay efficiency wages to lure workers. The example from the

British EEF at the start of chapter 1 was an example of firms‘ admitting that their organization was incapable of preventing defection from this kind of collaborative strategy after WWI.

For both collective repressive and collective collaborative strategies, then, firms have incentives to defect. This incentive has two implications. First it means under certain conditions firms will not be able to form collective strategies. This could be because they recognize such incentives of defection and decide it is not worth it, as is the case for standard commitment problems. As with any commitment problem, even if firms recognize that they jointly they would be better off if they could comply with the organization‘s rules, the temptation to defect means they could decide not to join. Entrepreneurial firms, perhaps those who suffer more from union penetration, for example, can try to form an organization by recruiting members, but fail in establishing a formal organization. Allen‘s address to the American NMTA at the start of Chapter 1 was an example of successful persuasion to get firms to recognize the benefits of collective action, at least at the sector level.

Second, firms could successfully form a collective organization, but shifts in economic or political conditions, coupled with the baseline incentive to defect, mean that the organization unravels. Even if firms are able to overcome initial problems

(whether through provision of selective incentives, expending extra costs on monitoring and prevention from defection, or having the state assume partial costs of enforcement and repression), changes in initial conditions (for example, higher costs in repression or exogenous economic shocks) might give firms an incentive to defect.

Joint Operation of Both Mechanisms: Sanctioning Defectors

The implication of the above theory is that the challenge facing nascent employer associations involves sanctioning member firms or sector-level associations which may be more ―militant,‖ to ensure the viability of the collective strategy. A plurality of employers could prefer to engage in collaborative strategies when its costs of repression become too high relative to collaborating with workers. This can be done through agreements with moderate unions which consent to agreement with employer associations, because they prefer gains from redistribution through agreements with firms as opposed to more extreme tactics (such as mass strikes or revolution). Moderate employer associations and unions have a mutual interest in sanctioning or deterring extremist workers (who prefer to expropriate more) and extremist firms (who are against any kind of collaborative strategy, and have lower costs of repression). Moderate employer associations, or those that have less to lose from collaboration due to lower costs of repression, have an incentive to similarly deter radical actors. Moderate unions have a similar incentive to prevent radical unions from sabotaging cross-class agreements.24

The empirical chapters document this kind of joint logic in the change of strategies in response to failures to sanction militants. The NMTA‘s switch to

24 A potentially complementary logic on the union side for explaining unionization is offered by Wallerstein (1990) and Horn and Wilensky (1988). In these accounts, a key parameter is the elasticity of substitution of the products different unions produce (whether those goods are complements or substitutes). In these models, the prediction is that workers who are substitutes have an incentive to organize in a single bargaining unit to maximize ability to strike effectively; amalgamation decreases the strike threat of workers who are complements. The optimal strategy for unions would be for unions to unite workers who are substitutes in production, and remain separate from unions that represent workers who are complements in production. In these models, it is a ―union maximizer‖ that exogenously sets the relative decision-making weights of unions that produce substitute and complementary goods. In my account, it is possible that firms or sectors with very different costs of repression can produce complementary or substitute goods.

67 repressing unions and punishing firms that did not sign union agreements in the twentieth century captures this logic. The organization had initially signed agreements with the IMU. But because the IMU was unable to punish unions from defecting, the

NMTA projected (correctly) increased wage demands. The cost of repression had stayed the same but the expected wage bill (in the form of IMU demands) would increase; thus the NMTA switched from a collaborative to a repressive strategy.

An observable implication of this is employers can sometimes pre-emptively negotiate with unions to prevent further state intervention. They might do so to prevent state management of institutions. Or, employers could agree to a pre-emptive collaborative agreement with workers to prevent further state expropriation (as in the case of Swedish SAF action to prevent further anti-lockout legislation, which is discussed below). In both cases, pre-emptive action by firms to switch to collaboration occurs because of lower costs of repression and preventing of additional intervention by the state. Once costs of repression escalate, for unions, the costs of sanctioning radicals become too high for unions. Pre-emptive collective collaborative can be preferable for firms because it prevents them from incurring costs of repression of workers, and state can share in costs of enforcing this kind of bargain.25

Section 4: Alternative Explanations and Concerns

25 The mechanisms advanced by CIS and MS could apply more directly for a subset of industries within a country. One interpretation is that these are the industries that have higher costs of repression. But, this approach addresses less the existence of repressive strategies and organizations, and, how bargains between collaborative and repressive organizations were key to producing the most important industrial institutions. Also, in some cases the variables of skill specificity may not explain the initial formation of collaborative employer organizations in some sectors. For example, in Germany, it was the case that some industries which were highly skilled were also disproportionately represented by radical workers and areas that did not support the SPD after WWI. Industries that emerged with higher wages and that grew quickly during World War I, for example, had many skilled workers who did not have previous ties to craft unions before World War I, and thus had much more extreme redistributionist goals. See Geary 1978 for discussion of this point.

In this section I discuss why the theory is an improvement over existing explanations, and address potential flaws in the argument. The main advantage of the theory over existing explanations, as discussed in chapter 2, is that it explains a much wider range of outcomes. Most theories in comparative political economy that deal with industrial institutions as a dependent variable are concerned with explaining variation within a given category (such as collective collaboration), or are only interested in explaining the emergence of collective bargaining period, without attention to the fuller range of outcomes. My theory accounts for why firms are able to collectively act to form employer organizations, and sometimes why we observe the failure or collapse of such collective action. Specifically, it can account for why we observe firms collectively acting only in certain industrial sectors, and not across sectors. Such outcomes of elite coordination are not addressed in the models by Boix

(2003) and Acemoglu and Robinson (2006).

Second, the theory more explicit incorporates a strategic logic to firms‘ decisions and a mutual strategic logic for both workers‘ organizations and firms. It explains why firms sometimes engage in repressive strategies against workers. These outcomes are not theorized in recent accounts of historical corporatist institutions

(Cusack, Iversen, and Soskice 2007; Thelen 2004; Martin and Swank 2008). Previous arguments make assumptions that firm reactions to labor mobilization are going to be uniformly positive (for those who focus on the problem of labor scarcity) or negative

(for those who focus on the traditional ―taxation by workers‖ argument). The theory does not arbitrate between the older ―traditionalist‖ view of industrial relations as a product of pure class conflict (Korpi 1983) and the newer perspective which

69 recognizes incentives of firms to work with unions in the provision of social policy or labor market outcomes (Mares 2003; Swenson 2002). Rather, it provides an explanation for why we would observe both within a state (at the regional or sector level), and why we would observe bargains that explain the modal firm strategy.

The theory articulates a key mechanism that incorporates the structural conditions of industrial heterogeneity to explain variation in enforceability of collective agreements. The literature that focuses on explaining just the emergence of collaborative strategies emphasizes other variables, including the role of electoral systems, legacies of guilds, initial structural conditions, and degree of unionization— but these accounts are less precise on how the mechanism of each variable operates.

Of course, some of these alternative explanations predict the same outcomes, and so historical evidence discussed in the empirical chapters will help arbitrate between competing explanations. For now I focus on how my theory provides mechanisms that other theories do not, and how broad historical evidence does not comport with the predictions of competing theories. I also discuss some internal flaws to the logic of each theory, which mine attempts to correct for. Table 3-2 compares the theory to existing accounts.

Electoral systems, co-specific assets, and firm strategies. The most recent arguments about the causes and consequences of industrial institutions link historical employer coordination to electoral systems, summarized in chapter 1. Cusack,

Iversen, and Soskice (2007; Cusack, Iversen, and Soskice 2010) argue that under certain pre-existing economic systems, specifically long-standing guilds, firms and conservative parties have incentives to advocate and adopt PR electoral systems. They

70 argue that a PR electoral system allows for protection of investments in specific skills or assets that firms and unions have jointly maintained. These investments are characterized by pre-existing cross-class coalitions which sometimes arose out of patterns of regional economic coordination. Building upon Thelen‘s important characterization of pre-industrial forms of economic coordination, CIS argue that in some economies, historical guild traditions created incentives for firms and unions to co-invest in skills, which facilitated cross-class alliances. Such alliances between workers and firms increased the probability of adoption of PR, because they argue that

PR systems better protect the interests of these cross-class alliances. In other economies, due to the absence of guild traditions, craft-based unions formed that inhibited industrial unionization. This created little incentive for co-investment between firms and workers in skill development. The consequence of this was more general conflict between workers and firms; because there were no specific assets to protect, there was an incentive for firms to maintain existing majoritarian SMD electoral systems.26 CIS argue that firms and right-wing political parties positively supported and adopted PR to ―lock in‖ benefits of pre-existing economic coordination.

The authors test the theory with cross-national estimations using data on indices of employer coordination and proportionality of electoral systems.27

26 They argue that as a consequence, PR systems redistribute more and SMD majoritarian systems redistribute less. This theory of PR adoption is a direct response to Boix (1999), who argues that ―fear of the left‖ led right-wing parties to adopt PR to maintain some degree of representation. 27The dataset uses a 5-point scale of economic coordination that consists of binary measures of the following: strong guild traditions; widespread rural cooperatives; high employer coordination; industry- based or nationally centralized unions (as opposed to craft or fragmented unions); large skill-based export sector (which they argue is an indicator of the necessity for compromises over wages and training) (CIS 2007, 385).

This account suffers from several flaws. First, they take as pre-existing the existence of some kind of collaborative employer coordination. But the theory proposed above does not take such coordination as exogenous, and hypothesizes why we might observe initial variation at the national level. Second, as the preceding discussion demonstrates, though, there has been much intra-country, cross-national, and temporal variation in the kinds of firm strategies countries experience. In particular, their theorizing of ―high‖ degrees of employer coordination obscures the fact that such firms were highly coordinated for purposes of repressing workers (both skilled and unskilled), not for protecting investments in specific assets.

The most important mechanism that this competing theory posits is that workers and firms have incentives to jointly collaborate for asset-specific reasons; employers have incentives engage in coordinating systems to protect such assets.

These arguments rely on an insurance model. In chapters 4, 6, and 7 I present evidence challenging these claims. The evidence demonstrates that the choice of collaboration was because it was viewed as less costly for employers than repression, as opposed to the mutual benefits from asset specific logic posited by existing scholarship. The latter logic can exist, but its empirical application is limited to a few sectors, and is relatively less important than the actual forms of coordination that existed in other prominent sectors. An important different empirical implication that is born out is that associations switched to collaboration once the costs of repression increase.

Another theoretical objection is that existing authors specify a production regime where incentives to jointly invest in specific skills entail what I code as

72 collective collaborative reason. Yet embedded within this logic is one of ―joint hold up costs‖—that is, once workers incur some cost to demand redistribution, it is quite costly for firms to repress. However, this mechanism of joint hold up costs does not imply some mutual gain in asset specificity: rather, it can be that firms prefer not to pay the costs of repression. This is an important distinction because it implies that under a different labor-market equilibrium, firms would prefer to repress. The observable implication is that firms should initially express a preference to repress, and only deviate when it becomes too costly. The evidence from Germany and the

United States also confirms this prediction.

Martin and Swank (MS) argue the causal logic runs in the opposite direction: the presence of a PR electoral system increased the probability of highly coordinated employer organizations. They focus in particular on the multi-party systems that develop under PR versus two-party systems that emerge under majoritarian SMD rules. Under multi-party systems, they argue that parties can ―credibly commit‖ to party supporters on benefits that they will receive while the parties participate in government; this creates incentives for parties to form alliances with other parties that represent other interests, such as workers or farmers. MS go on to argue that this cooperative political system ―transfer[s] this spirit of cooperation among employers‘ associations.‖28 Under a PR system, ―…mangers calculate that their changes of capturing a party likely to win an electoral majority are minimal; therefore, they have an interest in the creation of policymaking channels that work outside the government

28 Specifically Martin and Swank state that ―…dedicated business parties have incentives to form alliances with parties representing workers, farmers, and other interests in order to join ruling coalitions, and one might expect these parties to transfer this spirit of cooperation to employers‘ associations‖ (Martin and Swank 2008, 184).

73 by delegating considerable power to social actors. Highly organized employers‘ associations may emerge to coordinate labor and industrial relations systems and to help with state development strategies; yet, these groups should be less involved in electoral politics (the party‘s prerogative).‖29

They also argue that under two-party systems, political parties are more likely to cut across class lines. As a result, employers are dispersed across parties, and therefore act mainly as interest groups to fill a ―representation gap‖ of firms. ―When employers are dispersed across parties, a common corporate voice may be more difficult to establish. …When employers in two-party systems do capture one of the parties, they still have no reasons to compromise with the other party.‖30 The authors also test their theory with cross-national evidence, using a similar scale of coordination in conception to that of CIS.31 In the MS account, strong preindustrial guilds are treated as part of a collection of features of industrial relations which explain employer organizations.32 The alternative theories of interest in this account are the degree of industrialization, federalism, and party systems. The authors also describe some of the activities of the National Association of Manufacturers (NAM) in

29 Martin and Swank 2008, 184. 30Martin and Swank 2008, 184. 31 The scale measures three indicators of degree of employer organization: scope of employers‘ organization (share of employers organized in one or coordinated multiple peak national association); centralization of power of employer organizations and unions; and integration of national peak associations into national policymaking forums. Each score sums on a 1-3 index; the sum of these indicators is the degree of employer organization (so the index runs 1-9). They do this for 1900, 1914, 1925, and 1938 for 16 states. 32 Martin and Swank 2008, 183.

74 the United States and the Danish Employers‘ Federation in Denmark as support for their theory.33

This account also suffers from the latter two flaws of the CIS argument: they conflate different kinds of employer coordination, and assume that because parties have an interest in economic cooperation, that that by extension means industrial actors will cooperate. By contrast, my proposed theory argues that it is the degree of threat posed by workers that gives firms differing incentives to collaborate with workers, depending on underlying structural conditions. It accounts for the ability of firms to collaborate, as well as their varying preferences. Finally, as an empirical matter (shown in the German and cross-national evidence chapters), many different kinds of employer coordination pre-date the adoption of PR in most states. Crucial intra and cross-national variation in employer coordination systems pre-dates WWI.

Structural explanations. Another hypothesis is that structural conditions in the labor market alone can account for variation in labor-market institutions. Swenson presents the most historically detailed version of this theory (2002). The argument is that initial differences in market structural conditions explains why the United States has a more labor unfriendly set of institutions since the post-war era, while Sweden adopted solidarist policies. The implication is that changes in such conditions explain changes in firm strategies over time. Swenson‘s theory was one of the first to articulate why firms would prefer to organize in collaboration with centralized unions.

33 The political institutions are hypothesized to have an impact beyond standard variables such as degree of industrialization and size of state. There are obviously other explanations that both sets of authors consider. But I focus on these explanations related to electoral systems and pre-existing guilds because they are the most prominent in their work and because such works also draw on other influential discussions of the importance of skill development among firms and workers (Thelen 2004; Herrigel 2000).

It presents a theory of unions almost welcoming lockouts so that more militant unions can be effectively disciplined, with the outcome of a broad cross-class set of industrial institutions; this explanation is partially modeled by Moene and Wallerstein (2003).

The most important condition for Swenson is the initial scarcity of skilled labor.34 He defines solidarist policies that emerged in Sweden in the 1930s as a cross- class coalition where key sectors had firms and unions agree to wage restraint to lower unemployment (this account is elaborated on in the cross-national chapter). A major strength of Swenson‘s account is that it carefully traces historical change in firm strategies through the twentieth century—he documents the change in firm and employer organizations strategies from repressive to collaborative, or vice versa.

Swenson‘s account, as detailed as it is, suffers from some theoretical lack of clarity, which the above account tries to provide. First, Swenson does not really theorize the role of the state and the costs of repression, and how they affect firm strategies. (The previous accounts by CIS 2007 and MS 2008 do not either, but

Swenson provides some examples in that validate the importance of employer repression as a strategy).

Second, his focus on structural explanations undoubtedly accounts for some of the variation, but he is unclear as to what the particular mechanisms linking the conditions to the outcomes are. For example, it is unclear, for example, why the LO to

―rein in‖ militant construction unions, and the parallel ability of the SAF to do so with the building employers). I propose a simple theory of heterogeneity that affects the probability of collective action, via the mechanism of variance in the costs of

34 There are of course other conditions, but this is the most important one. In the cross-national chapter, I address more directly how my theory better

76 repression. I argue that structural conditions (which, like Swenson, I take to be exogenous) affect the variation in the costs of repression within an economy, which in turn affects the enforceability and ease of collective action—this mechanism is clearer and more testable. More broadly, my theory does not discount older perspectives that the threat posed by a radicalized labor force does matter.35 I argue that the key mechanism of enforceability is what makes structural conditions important.

Left power and unionization. Korpi and Shalev (1979) argue that corporatist institutions were simply attributable to the power of the left, measured by either union density or socialist party vote share. Therefore, we should only observe collective collaborative strategies when some pre-existing support for the ―left‖ exists. The main mechanism was that collective institutions were supported by unions, and the more power they had, the more likely they could be implemented.

The main problem with the original ―left power‖ hypothesis is that it ignored heterogeneity of firm preferences—it ignores the theoretical possibility of firms preferring to cooperate with workers to help take wages out of competition or prevent wage drift. It mistakenly attributed outcomes to the ability of workers to impose their preferences without considering that some firms might have wanted such policies.

Second, this theory does not account for the wider variation in firm strategies shown in

Table 2-1—in particular, it neither theorizes individual firm strategies, nor does it consider firm repression as an outcome. By contrast, my theory accounts for this variation across sectors; the theory acknowledges that the ―left threat‖ argument holds only for some sectors. I propose a specific mechanism—the preference of firms to

35 Wallerstein (1991) argues that size of labor force should be correlated with union density due to lower costs in recruitment and expected benefits of each additional recruit. See

77 pre-empt state involvement and not incur the higher costs of repression—to explain why we observe collaborative strategies.

Cartelization. An alternative argument is that initial degrees of cartelization and high degrees of financial sector concentration explain variation in firm strategies.

The cartelization argument has not been thoroughly articulated in the literature, but it is intuitive: if firms have higher coordination institutions in the product market

(represented by cartels), they should find it easier to coordinate in labor markets. An example of this intuition is that if unions are allowed to help police cartels and institutions designed to remove wages out of competition (essentially wage and price controls) then collective collaborative institutions should be more likely. If such behavior is not allowed due to early anti-trust restrictions (as was the case in the

United States after the passage of the Sherman anti-trust legislation), then firms may not be able to commit to such collaborative institutions because they do not have enough enforcement capacity to monitor them. Therefore, either variables that explain cartelization, or cartelization itself, should account for firm strategies in labor markets.

This theory is a powerful alternative, but my explanation partially subsumes it.

First, degree of cartelization alone cannot explain repressive versus collaborative strategies. In the United States, for example, the anti-cartel legal environment indirectly facilitated repressive strategies in several respects. First, it caused a wave in mergers which resulted in many large individual firms that practiced labor repression

(Bittlingmayer 1985). Second, legal precedents often meant anti-trust law could be used against unions (which were in some cases deemed as cartels of labor). Firms‘ use of injunctions against workers to prevent strikes is an example of firms and states

78 sharing in the costs of repression (made relatively lower for firms by the willingness of the state to bear them, and the lack of a widespread industrial union movement).

This meant also, as stated above, that unions were barred in many industries from helping to police wage agreements among firms. By contrast, in Germany and

Sweden, cartels were encouraged, and cartels in heavy industry became the vanguard of repressing organized labor, particularly in Germany. Thus ex ante, cartels alone cannot predict repressive versus collaborative strategies. Cartels themselves though can be a component of collective strategies, or exist for specific sectors in a country where the modal strategy is still repressive.

The other response to the cartelization argument is that structural variable of industrial heterogeneity should help account for cross-national variation in cartelization, as well as intra-country variation. In particular, the theory explains inter-sector as opposed to single sector collective action. Industrial organization research is replete with examples and models of why cartels fail due to incentives to defect and under-price; a similar logic explains why employer organizations sometimes unravel or fail to form at the inter-sector level. Thus, structural differences in the costs of repression across sectors should explain cross-national differences in the longevity of cartels as well as employer organizations.

Having set up the theoretical expectations and outlined the mechanisms, the rest of the dissertation tests the theory with exhaustive evidence from Imperial

Germany, war-time and Weimar Germany, the inter-war period in the United States, and cross-national evidence. I deviate from the main approaches used in the literature of cross-national regressions of a limited number of observations on a small number of

79 predictors. These regressions are often sensitive to specifications, although they generally establish that degree of unionization, left-wing vote share, and higher degrees of employer coordination tend to be correlated. As noted, however, such measures of coordination conflate repressive strategies, or miscode strategies.

Chapter 4 provides detailed evidence of the advent of individual repressive strategies in Imperial Germany and the transition to collective strategies. Chapter 5 statistically tests mechanisms of the theory at the micro-level, using new sub-national data on membership in different employer organizations and firm strategies in this time period Germany. Chapter 6 documents how changes in the proposed variables affected firm strategies in the Weimar era. The evidence, constituting publications, meeting notes, and internal correspondence, supports the main predictions of the theory about the determinants of firms‘ strategies. Chapter 7 uses a mix of archival evidence from the United States and new data on employer repression to test the theory. The idea is that such evidence should show firms responding to redistributive threats from workers, and take into consideration costs of repression and enforceability of agreements when deciding whether to collectively act. The final chapter shows that changes in firm strategies do follow changes in heterogeneity and costs of enforcement and the threat posed by workers for other states. The concluding chapter takes up some of the objections raised in this chapter.

Table 3-1: Main Theoretical Predictions of Firm Strategies

Industrial heterogeneity

Low High

Redistributive threat posed Low Collective or Individual Individual Repressive by workers Repressive Strategy Strategy

Medium Collective Individual Repressive Repressive Strategy Strategy

High Collective Collaborative Individual Strategy Collaborative Strategy

Table 3-2: Current hypotheses about firm strategies

Main Predictor Mechanism Outcome

Proportional representation Allows for Coordinated political system institutionalized industrial coordination between relations versus (Martin and Swank, 2008) firms and labor not

Early guild systems Initial coordination; Coordinated path dependence industrial (Thelen, 2004) relations & skill-specific training institutions

Unionization alone Employers ―must Initial respond‖ employer (Historical labor literature) associations

Level of industrialization / Scarcity of labor creates ―Solidarist‖ vs. Structural scarcity incentives for certain ―non-solidarist‖ institutions coordination (Swenson, 2002)

Threat of political Incentives to forestall Different firm representation of workers x expropriation; strategies industrial heterogeneity at t0, enforcement of inter- (individual t1,… firm and inter-sector versus agreements collective; repressive versus collaborative)

Chapter 4: Evidence from Imperial Germany

―Scratch one of these gentlemen, and you get the police.‖ – Max Weber

―The spirit of solidarity among employers needed to be aroused and continually strengthened, until individual interest came to accept the principle of voluntary submission to the common interest‖ – Henry Axel Bueck, head of the

Central Office of German Employers‘ Federation36

The sentiments expressed by Max Weber and Henry Axel Bueck in the early twentieth century encapsulate the arguments described in the previous chapter about the adoption by firms of repressive or collaborative strategies vis-à-vis workers, and when such strategies were executed individually or collectively as part of employer associations. Weber‘s statement about ―gentlemen‖ referred to the owners and managers of the largest firms in Germany, who he argued were aligned with the state authorities in preventing workers from mobilizing. Bueck‘s statement refers to a plea to organize firms to combat what he viewed to be the growing threat of social democratic trade unionism. The two men had radically differing political views and prognostications about the future of the German capitalism, but their impressions of how firms were responding and should react to political and industrial events on the ground indicate the importance of the variables described in the previous chapter.

They indicate that the threat from workers and the problem of collective action were the most important challenges facing enterprises at the turn of the twentieth century.

In this chapter I use evidence from Imperial Germany to show that firms‘ responses to organized labor were conditioned on the threat posed by workers and a

36 Weber‘s quote is from Anderson (2000), 226. Bueck‘s statement is from Krüger (1926), 331.

83 set of structural conditions, namely the industrial heterogeneity that allowed for effective collective action. The evidence documents the transition from the use of individual to initial collective level strategies of repression, and the transition from initial regional collective associations to peak-level associations. The evidence on the evolution of German employer associations overwhelmingly indicates that the initial response of the most important firms in the largest industries of Germany was repressive, because the firms could incur the costs of such actions. Collective strategies in the forms of influential sector, regional, and national-level employer associations could be adopted because the variance of costs of enforcing agreements among different firms and sectors were low. This was particularly true in industries which had a previous history of heavy cartelization and geographic concentration.

The evidence also confirms the importance of heterogeneity in undermining the expansive and ambitious goals to form a peak association that would obligate member firms to abide by a central office‘s regulations. Tensions existed among different industries in terms of costs of maintaining employer associations. Even though firms were able to initially overcome the collective action problem, there were inter-industry differences about who should bear the costs of peak associations indicates‘ duties.

I use different pieces of qualitative and quantitative historical evidence from specific sectors and within major organizations to test the theory. In addition to showing how historical trends, events, and descriptive statistics are consistent with the theoretical predictions, I also use primary evidence from the main industrial and political actors to illustrate the mechanisms of the theory. This chapter focuses on the development of the largest employer associations and relates their incentives of

84 formation to the redistributive threat they faced from mobilizing workers. Chapter 5 uses specific quantitative data on the development, activities, and membership of important employer associations to test the theory.

The evidence from this time period also poses a difficult test for the theory, because the SPD never won a majority of electoral districts and was known as a perennial opposition party; in addition, unionization rates were comparatively lower than other industrialized economies. Thus the magnitude of its threat to major firms can be viewed as relatively small compared to that of Weimar Germany or in other

Western European states. That is, the empirical tests leverage smaller changes in the redistributive threat posed by workers to show that they matter for repressive versus collaborative strategies. The quantitative and qualitative evidence indicates that contrary to the existing theories about the origins of capitalist institutions, firms in

Germany before WWI pursued repressive strategies, continuing after the lifting of the ban of Socialist activity in 1890. Understanding why these anti-labor and repressive strategies were so predominant before World War I can help explain their subsequent historically important to shift to collaborative strategies at the start of Weimar

Germany which I address in Chapter 5, and again in the post-WWII era.

The evidence presented that supports the theory helps rectify a curious disjuncture in the study of the development of industrial-relations institutions by social scientists, as described in Chapter 2. The literature that focuses on the incentives of firms as actors, and their incentives to form associations, has recently emphasized the supposed ―cooperative‖ nature of such institutions, by focusing on examples from

Denmark, Sweden, and inter-war Germany. The recent political science literature has

85 focused on the existence and operation of firms in handicraft or craft-based skills sectors to confirm theories about electoral systems or employer coordination. But in examining these cases, this literature ignores some of the important examples of employer repression. By contrast, the earlier influential literature on power resources of the left, as noted in the previous chapter, ignored the strategic choices of firms and the costs of repression that the incumbent conservative elite had to make. Nor did this literature engage in the problem of collective action that faced these firms. Although this slightly older literature did not draw much on evidenced from Imperial Germany, the attitudes of influential heavy industry firms would support this older theory of pure class conflict.

The discussion here demonstrates that much of the existing attention in political science on the joint-incentives of firms and workers to form collaborative associations incompletely characterizes the development of capitalist systems: the repressive logic of employer coordination was more important in this time period.

The existing theoretical accounts that use German evidence as a test of the ―asset specificity‖ or positive employer coordination focus on the development of handicraft industries in the late 1800s and the Weimar period. But the development of associations in the early twentieth century in heavy industry and skilled manufacturing, described below, is at odds with these accounts. Further, I demonstrate that the development of handicraft employer associations is more consistent with the composed theory.

Thus the existing literature has focused on different pieces of the puzzle. From a theoretical perspective, it is puzzling why firms in different product markets in

86 different regions should be willing to subjugate some of their managerial autonomy to a central office in Berlin or elsewhere. In the theoretical account from chapter 2 and the evidence from Imperial Germany, I argue that firms in different sectors that are the subject of different traditions of social science inquiry did express such reluctance to cooperate, but it was initially overcome in the face of a prominent redistributive threat that made repressive collective action a preferable response. Such collective institutions experienced their own internal tensions; this chapter documents them as support of the proposed theory.

The chapter proceeds as follows. Section 1 describes the political and economic background that puts the strategies of political and individual firm repression in historical context. I provide evidence of the initial use of individual level strategies to repress workers. Section 2 provides evidence on the transition to collective strategies at the regional and sector level and their importance. I demonstrate the transition to such strategies in light of changes to the economy and the increase in use of repression in response to worker mobilization; I also show how structural variables that facilitated collective strategies. Sections 3 and 4 show how the logic of formation of important peak-level associations in the early 1900s—the further adoption of collective strategies—confirms the theory. Section 3 of shows how the emergence of two separate peak associations is consistent with the theory.

Section 4 shows how the eventual merger of these two associations, despite its limited in time is existence, is similarly consistent with the theory. Both sections: a) link the threat posed by workers to collective repression as a strategy; b) demonstrate how industrial heterogeneity affected the fragility of such institutions.

Section 1: The Industrial and Repressive Context

1.1 The political repressive context

In this overview I describe the political climate in which firm repression operated. During the two decades before the outbreak of World War I, the SPD and general trade unions did make significant inroads in terms of popular support and labor force that was unionized. There was a tenfold increase in unionization between the years 1890 and 1913, and increasing contestation of the SPD in national constituencies during the early twentieth century. Still, despite these trends and the lapsing of the anti-Socialist laws in 1890, the adoption of the secret ballot in 1903, and the SPD‘s increasing political contestation and representation in the national legislature (Reichstag), the overall legal environment in Imperial Germany was quite hostile for workers (Hall 1974). Besides the restricted three-class franchise system in the dominant and large Land Prussia, workers faced a legal environment that allowed the state to disrupt and punish attempts at organizing into unions and making demands upon firms. In 1887 the German Supreme Court (Reichsgericht) upheld the legality of measures to prevent the association of workers. According to an 1889 ruling, freedom of combination among workers was not a basic social right, and therefore firms were allowed to prohibit employees from joining unions or using union membership as grounds for dismissal (Saul 1985). The Prussian Ministry of the Interior enforced these measures. In 1904 a court in Kiel ruled that dismissal of an employee due to his membership in a union was permissible; in 1905 a regional Dresden court recognized the ―self-evident right‖ of the employer over employees (Saul 1974).

Additionally, during Bismarck‘s ―Kartell Reichstag,‖ which was governed by a coalition between the National Liberal and Conservatives parties between 1887 and

1890, the official election commission often ruled against workers in contestations about electoral fraud. The Reichstag refused to discard votes that had been alleged to be extorted by employer pressures. The state apparatus helped bear some of the costs of repression, as this gave firms freedom to continue practices of voter intimidation.

This practice changed only when the SPD won a plurality of seats in 1912; the

Commission then considered cases of fraud regarding the employer‘s extortion or intimidation of employee voting intentions (Anderson 2000).37 The result of this legal environment and its general enforcement by the state was that unions faced the risk of and actual police harassment; they could not depend upon state support during industrial disputes. As an anonymous drafter of the German Civil Code articulated in government decisions against liberalizing association laws, the goal was to ―wrench the working population from the utopia of social democracy and win them over for the existing state and social order‖ (Kögler 1974).38

Overall, the state apparatus was geared toward the containment of the socialist threat and maintenance of the existing political order, and helped blunt the mass mobilization of workers. As Saul writes, ―The bias against the working class in the judicial system and the war of the police against all socialist organizations including the Free Trade Unions were part and parcel of the social reality of the time‖ (1985,

349). Repression was a means of ―defending civil order‖ from the ―onslaught of

37 It is difficult to estimate, of course, how much electoral fraud there was in this period. For an alternative view of the Election Commission, see Ziblatt (2009). 38 The quotation is from an anonymous proposal from the second Civil Code committee from 1891.

89 social democracy‖ (Saul 1985, 347).39 In the next sections, I show how firms uniquely contributed to this environment, and shared the costs of repression by adopting individual and then collective repression against workers. Firms and the state shared the costs of repression. However, when such costs became too high for state authorities to bear, individual firms took on a greater proportion of such costs.

1.2 Industrial repression—individual level strategies and context

The costs of repression in the late nineteenth century were also shared by individual firms. The repressive climate meant that large portions of the workforce in key industries in Germany operated at the whims of German firms, and had a smaller role in determining equilibrium wages and working conditions. In this section I discuss the initial importance of individual firms in helping make these dismal outcomes for workers possible. For most of Germany‘s important industries during this time period, repression was the default strategy for large firms and their employer associations. In this section I discuss some of the techniques, many of which were also pursued, despite a change in labor law, during the Weimar era. In the next section

I link the strategies of this environment to the threat posed by workers and show the transition to collective strategies in the form of repressive employer associations. The evidence presented here is in contrast to the current hypotheses which argue that

―cooperative‖ relations emerged out of joint investments in skills.

As of 1914, in most major industries including agriculture, mining, heavy industry, the large chemical industries, the electrical and engineering industries, commerce, transport, insurance and banking, sector-level collective bargaining was

39 See further documentation in Knips (1996).

90 absent. From 1905, in some large metalworking, electrical and engineering factories, firm-specific or company-level unions existed, but these were distinct from and in some cases helped undermine social democratic trade unions. In industries with largely skilled workforces, employment bureaus were still run by firms that dictated outcomes and punished union affiliation. In much of the mining industries, for some of the largest companies (such as the Prussian-Hessian State Railway Company, which was then the world‘s largest firm), and certainly on large agricultural estates, unionism and enforced formal protections for workers were non-existent.

Other industries that had lower industrial concentration, more intensive labor competition, and were composed of small firms—such as printers, construction workers, carpenters, painters, and tailors—did have some collective wage agreements.

However, these industries, which are the substantive focus of existing political science research on the development of different employer coordination regimes, composed a smaller portion of output. Further, as discussed in section 3 below, the advent of these more collaborative strategies in specific sectors, in these less important industries, is consistent with the theoretical predictions discussed in chapter 2. They emerged because the costs of repression in such industries were higher, and the variance of such costs was lower.

As described in chapter 3, the individual firms that initially pursued repressive strategies faced two sources of distributive threats from workers. The first was the rising demands from workers about workplace conditions and wages. The second was increasing attempts from the state to intervene on such matters. Some firms were disappointed that franchise restrictions and other repressive means were off the table;

91 this meant that firms would have to incur a higher share of the costs of repression.

Incidents of conflict and attempted intervention from the state were important in giving firms an incentive to collectively act—in some cases, to take existing product market coordinating institutions and extend them into the labor sphere. Until the early

1900s, large firms felt that formal employer associations were unnecessary to handle labor disputes. For example, in the Ruhr area, the mining industry was not heavily unionized, and thus large firms, concerns, and cartels felt that formal employer associations for much of the late 1800s were not necessary.

The major industries of the German economy toward the end of the nineteenth century were in the raw production and processing of coal, steel, and iron and related industries; railways and transportation; and to a lesser extent, chemicals and construction. The explosion of production in heavy industry and manufacturing was characterized by high degrees of concentration, cartelization, and a proliferation of large firms.40 Between 1890 and 1910, 30 of the largest 50 firms in Germany in terms of production and number of employees were involved in heavy industry, coal, or the manufacturing of related goods (Feldenkirchen 1988; Fiedler 1999); many of the other large manufacturing firms outside of these top 30 were connected through cartel arrangements and trade associations. The economic influence of the ―chimney barons‖ thus stretched to multiple industries. For example, 20 percent of all coal production in the Ruhr by 1900 was controlled by steel firms. Heavy industry accounted for over 60 percent of all ―giant‖ employers in Germany (those that employed more than 1,000 workers). As early as 1882, one-third of firms in heavy

40 See also Groh (1978) and Kinghorn and Nye (1996).

92 industry fell into that category.41 Six syndicates were responsible for 90 percent of all hard-coal production and 58 percent of the mining labor force worked in pits employing over 1,000 men in 1907. There was a similar dominance of large firms in the chemical and electric-technical industries (Geary 1991, 149-150).

The influence of these large firms extended beyond Germany‘s economic production: these firms and their associated cartels and organizations extended formal and informal repressive control into the lives of workers. Many of these large firms in manufacturing and raw materials repressed workers who either participated in organized labor movements or supported the SPD, or were suspected of either activity.

As the economist and social reformer Lujo Brentano wrote in 1905: ―The power the giant firms attain over those they employ extends even beyond the work situation; it extends to their entire social, religious, and political being. Even the shopkeepers who do business with the workers can come under the dominion of the managers of these firms. Regions are emerging within the Reich where it is not the will of the legislator, but that of the owner, that is law‖ (Anderson 2000, 199).42 These firms had lower costs of repression, and could pursue a variety of strategies to prevent worker participation in social democratic trade unions by repressing or dividing workers; they could also afford to ―sit out disputes‖ and refuse to negotiate with them (Geary 1991).

Of course, there was much variation across and within regions in Germany in this

41 By 1914, the Krupp concern employed more than 70,000 workers; Deutscher Kaiser and Harpener Bergbau-AG employed 20,000 each; the Bochumer Verein, Hibernia, Hoesch, and Rhenisch Steelworks each had more than 10,000 workers. 42 Max Weber thought class differences were ingrained in employer sense of hierarchy. At a speech for the Verein für Sozialpolitik, he debated Alexander Tille, who was a spokesman for Saar industrialists; Tille justified the behavior of these large firms on the grounds that ―they must be also be able to document outwardly the subordination of others‖ (Anderson 2000, 227). As Weber described these firms, ―Scratch one of these gentlemen [i.e., the bosses], and you get the police‖ (Anderson 2000, 227). Tille corresponded with Bueck about the ideological propaganda of the SPD and its effectiveness.

93 period, but the dominant areas of production were home to firms in these basic raw materials, manufacturing, and transport industries.

In the reminder of this section I first discuss strategies of individual firm strategies vis-à-vis the state, and then briefly turn to repressive strategies that intervened in the labor market. With respect to the state, the costs of repression can be understood as costs that individual firms expended to lobby the state. These can be for strategies for which the costs are shared between individual firms and the state. With respect to direct intervention in the labor market, these costs can be viewed as those that incurred from dismissing workers or maintaining other forms of repression described below. These two categories of strategies are obviously not mutually exclusive, nor is there always a clean division between the labor-market strategy and the state strategy. In section two, I show how changes in industrial heterogeneity and a rapid increase in the threat posed by the SPD and growth in unionization increased the incentives for firms to take these individual-level strategies and execute them collectively.

1.2.1 Direct repression from firms – punishment for affiliation with left-wing institutions

One set of strategies involved direct punishment and prevention of workers who supported the SPD. Anderson (2000) and Saul (1974) document a number of such examples. This strategy of punishment for political affiliation had several potentially overlapping elements. First, firms could fire or fine workers who were affiliated with the SPD. They could gather this information from ―informal‖ hearing of alleged support from informant workers. For example, participation in SPD related

94 cultural or social clubs could be grounds for dismissal. Second, firms could credibly threaten workers with the above penalties, but with the goal of deterring them from turning out to vote. Third, firms could intimidate workers into voting for the firm owners‘ preferred parties, such as the National Liberal or conservative parties. A final strategy in this set was that firms might have simply prevented workers from leaving the factory or workplace on the day of the election.

Where did worker intimidation occur? Anecdotal evidence indicates that these strategies of intimidation of workers who supported the SPD could be found in almost all major industries. Intimidation that utilized this ―social control‖ of worker space was used in firms in paper and glass manufacturing; coal and iron; craft, bench, and assembly line industries. At the Baden Aniline and Soda Company of Ludwigshafen, where the managers also comprised local leadership of the Liberal Party, the foremen informed factory workers that anyone who voted for the Center or SPD parties against the Liberal incumbent would be immediately dismissed. Firm owners could extract oral or written promises, and threaten dismissal or foreclosure in the case of firm- provides housing. The anti-SPD pressure was particularly strong in forestry, railways, and heavy industry (including coal mines, processing of iron and steel, and production of goods made of iron and steel). These industries had large workforces and strong ties to the state. Even with the privatization of mining and forestry industry, managers of these enterprises had often begun careers working for the bureaucracy. (As discussed above, average firm size in these industries was much larger).

Major companies that explicitly threatened workers if they supported the SPD included the largest heavy industry firms, such as Phoenix Steelworks,

Guttehoffnungshütte, the Krupp concern, firms in the Bochumer Verein, Hoesch,

Rhenisch Steelworks, and Burbacher Steel works in Saarbrücken. In a representative example, after 1890 Krupp made his workers sign a pledge not to vote for the SPD, and extracted similar promises from workers to not support the Christian Social

Verein, which was affiliated with the Center party (Möllers 1955). The Breslau factory owner Hermann Seidel supported Conservative candidates for national elections and coerced his workers into turning out to vote for them (Anderson 2000).43

In a revealing statement, the porcelain manufacturer Egmont Tielsch told Reichstag investigators: ―The men named had been pointed out to me by a competent authority as belonging to a Social democratic organization. I have posters in my factory making it known that I will not keep Social Democrats in my employ, but will dismiss them immediately…I expressed the wish that they then give me proof of their assurances by abstaining them from the election on 17 February 1887. This they voluntarily promised to do‖ (Anderson 2000, 200).

Another example of firm intimidation and revealed preference if workers voted for the SPD came from Preuss and Company, a factory in Leipzig. The owner, in a speech to his workers, stated: ―Each may vote freely…but consider the following consequences: If the government party wins, then industry will flourish again, peace will be maintained. If the opposition wins, then we will surely have war, and immediately after such an election result we will dismiss half of the workers and have

43 Anderson discusses the case of Egmont Tielsch, a porcelain manufacturer. Before 1887 election, the head painter in Tielsch‘s factory called painters to resign from Social Democratic glee clubs (2000, 199). Tielsch made clear that workers should vote for National Liberal industrialist candidate Egmont Websky (200), and intimidated workers to do so, while getting them to see that to vote for Websky was their own ―enlightened self-interest‖ (201). Tielsch said explicitly, ―Either each individual one of ye gives me the promise that he will abstain from this election, or I‘ll sack the lot of ye immediately.Later though canst vote for whomever thou willt, only not Socialist.‖

96 to introduce the eight-hour day [i.e., short hours]. All the other factories will find themselves in the same situation. Therefore, workers, be well warned‖ (Anderson

2000, 203; Müller 1963). As discussed in section 1.1, the political authorities did not view such efforts as illegal. Employer authority over workers was asserted through supervisors: firms screened workers for union loyalty, SPD affiliation, political

Catholicism, and affiliation with left-liberal movements. Some firms only hired

Protestant workers because they were viewed to be less likely to affiliate with activist unions (Spencer 1984, 84). The use of these screening processes as a separate technique of repression is discussed below on labor-market intervention as a repressive strategy.

The example of German railways illustrates the use of firm attempts to intimidate voters into voting for preferred business interests. Politicians from the

Center Party and Progressive Party had opposed nationalization of German railways because it would concentrate the voting blocs that these companies could generate in the hands of the government. By 1888 the Prussian state had employed nearly

250,000 personnel; this company was the largest in Germany. The state-owned companies performed extensive surveillance of the political views of workers and forbade formation of trade unions in public sectors; as late as 1907 the German rail companies outlawed festivals and public celebrations for Catholic or labor organizations. Railway companies required employees to present on demand work certificates (Arbeiter Nachweise), similar to what employer associations did, as discussed in section two. This procedure facilitated screening out workers with previous union histories. By 1890 the rapidly growing industry had 350,000

97 individuals, and 700,000 by 1910. Such intimidation of workers who affiliated with unions or the SPD was prominent in Hanover, Silesia, Saxony, and Bavaria (Müller

1963).

Some firm owners questioned the cost of such repression. Krupp, before turning to repression, mused, ―It costs thousands…and is it worth that much?‖ He suggested that wholesale firing of SPD-affiliated with workers would naturally drastically reduce the available labor supply. But Krupp‘s view changed. He put his workers under surveillance, and sent stenographers to cover SPD rallies, arguing that informants were necessary to uncover the extent of formal support of the SPD among his workers.44 Krupp went from not caring, to adopting a repressive stance. ―If one wanted to chuck the Social Democrats out…then all the people would have to be sent away‖ (Anderson 2000, 221). His comment demonstrates awareness of the labor market equilibrium conditions: the supply of workers available allowed him to practice such repressive strategies.

Incidents of industrial conflict, discussed more extensively below, led directly to repressive responses from firms that punished workers affiliated with unions involved in the conflict. This was partially because such conflict could make SPD electoral campaigns salient. For example, the large May1889 Ruhr miners‘ strike involved 87,000 men and spread to Saar, Upper Silesia, and Saxony. SPD activists used the strike as a rallying point for the 1890 national election; the SPD forced the

44 ―Without secret police…we will not get shut out of the thing…‖; he posted political proclamations in his factory and had them guarded (Anderson 2000, 222). Of course, there were scope conditions on such repression. In 1877, Germany suffered a severe economic downturn, resulting in the mining industry letting to 10 percent of the workforce. But Krupp‘s operations were expanding. The problem is that with high worker demand, he found it more difficult to punish or fire workers for their SPD allegiance. In fact in Krupp personally traveled to Berlin to plead the case of an SPD-affiliated worker to allow him to keep his job (Anderson 2000, 222-23).

National Liberal incumbent, the landowner Ferdinand von Reden, into a runoff. Chief

Mining Director von Detten, who maintained the state‘s mining operations in the

Hameln district, argued that repression was a necessary response to the strike. The miners‘ rebellion had reached Hameln‖ (Franz 1951). Von Detten reacted directly to the 1890 electoral results: ―Be warned at this last minute, ye miners, thy work, thy wage, thy pensions, thy support funds are all at stake if the Social Democrats move in through thy support‖ (Anderson 2000, 205). Workers in mills, brickworks, and glassworks, were threatened with dismissal if an SPD candidate won.

1.2.2 Connections between individual firms and the state

The previous section discussed prima facie evidence of state and individual firm repression of workers and the general hostile legislative and judicial environment, despite occasional victories. I now present more specific ways in which these strategies dovetailed. The empirical implications in terms of testing the theory are threefold. First, important representatives often also served in the state, either in a bureaucratic capacity, or were in the legislature. Thus there was definitional alignment of incentives—as representatives of firms and as part of the state, they desired the same outcome. Second, through such an overlap in responsibilities, they shared the costs of repression; such costs for the individual firm were lower if they either in fact were or had informal affiliations with public officials who had direct business representation. I provide evidence of this alignment in terms of the connection between state repression and employer associations; firm support of repressive political parties; and businessmen who also acted as agents of the state.

There were informal and formal affiliations between industrialists who led key repressive employer associations, and politicians and bureaucrats who assisted employer associations in their repressive activities. Sometimes prominent industrialists or managers of employer associations also served as politicians.

Members of these associations allied mostly with the conservative wing of the

National Liberal Party, and supported Bismarck in the 1870s; this was not just because of the anti-socialist laws in place but also due to his advocacy of naval expansion.

Prominent industrialists supported the Prussian Landtag National Liberal faction, and voiced opposition to the SPD and Progressive parties. The Conservative party was also considered useful to these firms and employer associations, but it was more electorally weak in the Ruhr area and dominated by agrarian interests.45 Carl Lueg of the GHH and Krupp had linkages to the Free Conservative (Frei Conservativ, FC) party (Spencer 1984, 54).

The Ruhr business elite lobbied and established connections with the Prussian

Landtag House of Lords. In fact, some industrialists were appointed to House of

Lords; this was a way for ministerial bureaucracy to guarantee goodwill of entrepreneurial elite. Krupp was appointed in 1901; Lueg in 1903, and Franz Haniel in 1905. Krupp was the ―personal choice of the emperor‖ (Spencer 1984, 49). The

House of Deputies in Prussian Landtag also had a majority hostile to organized labor, aided by the three-class voting structure.46 Coal, iron, and steel industrialists opposed

45 As Spencer writes, ―…close identification with the conservative party at the other end of the political spectrum was impeded by the dominant role within it of eastern agrarian interests…‖ (1984, 54). 46 Voters in Prussia were divided into three classes according to amount in taxes they paid– each class chose the same number of electors, who makes final choice of district‘s deputy—this heavily favored

100 any reform of the franchise of this body (Spencer 1984, 58). Of the 38 members of the business elites in the Ruhr area, three were representatives in the Reichstag: August

Servaes of Phoenix (1880-85) and Eduard Kleine (1887-89) were representatives of the National Liberal party, and Krupp (1893-98) was a member of the Free

Conservative party. Overall, the business elite that led repressive employer associations during this time period generally resented any attempts of the state to extend regulation of working conditions. They opposed universal suffrage and adoption of the secret ballot in 1903.

There were important connections between these major firms and the state, in particular local and national-level political elites. For example, Krupp in Essen directly dictated the entire slate of voters in the first class in municipal elections in

Essen in 1874, 1896, and 1894. Krupp personally named one-third of Essen‘s aldermen until 1918. The Krupp concern in Essen also had a company union that explicitly supported NL candidates; such industrial authoritarianism happened at the

BASF in Ludwigshafen and the Howald concern in Kiel. BASF chemical workers were expected to support the NL candidates. As another prominent example, managers at the Bochum employer association (Bochumer Verein, BV) constituted

Bochum‘s entire first class; aldermen seats became the discretion of the BV‘s directors. Even the second and third classes were dominated by Bochum management. These same men were on the campaign committees for National Liberal or Free Conservative Parties in national elections (Tenfelde and Volkmann 1981). 16 out of the 38 region‘s leading industrialists served on city councils (Anderson 2000, the interest of wealthy industrialists. Until 1908 not a single Social Democrat was elected in the Prussian Landtag (Spencer 1984, 58).

101

228).47 Louis Baare, the BV managing director from 1855 to1895, had connections to other key political organizations: he was a deputy to the provincial diet and Prussian

Landtag; head of Bochum‘s Election Association; and was involved in financing of

National Liberal electoral campaigns. In fact, Baare helped script his district‘s Liberal candidate‘s campaign policy views on social and economic questions: he threatened to remove the BV‘s funding for the campaign and divert resources to the Free

Conservative party if his (Baare‘s) demands were not met.

1.2.3 Individual lobbying for anti-worker legislation

Firms also adopted individual-level repression by seeking state legislation that would make it more difficult for workers to organize. As discussed in section one, the legal environment in the early twentieth century was hostile to such organizations.

Repressive firms also managed to successfully defeat legislation that would have increased the cost of such strategies. In 1885, the Center party sponsored legislation to protect voter‘s freedom from threats to his employment. The bill was drafted by

Viktor Rinteln, a judge (Geheimer Oberjustizrat) in Berlin. Rinteln wanted to amend civil code number 109; his proposed legislation would punish any employer who fired or threatened with penalties a worker because of the exercise or non-exercise of political opinions. Conservative and NL party representatives objected to the bill.

Anderson (2000) quotes a summary of conservative opposition to the legislation:

―Under what principle could the Reichstag deny a man‘s right to remove from his

47 When Karl Ferdinand Stumm in the Saar did not get his way in terms of fighting SPD, he withdrew as a candidate, throwing pro-government region into disarray. Stumm allowed the Center party to win the first round of election, forcing the right-wing party candidate into a runoff; after Stumm‘s transition to serve in the Prussia‘ House of Lords, the Saar area reverted to normal Free Conservative/National Liberal electoral pattern (Anderson 2000, 229).

102 payroll a person whose party characterized employers as bloodsuckers, parasites, thieves, and called for their expropriation?‖ (Anderson 2000, 233; Möllers 1955).

Representatives of the Center party and SPD contended that this bill would protect the principle of free elections. The Rinteln bill eventually failed, and the Reichstag would not attempt to legislate directly against employer influence until World War I.

Another important debate over state-sanctioned repression was the

―penitentiary bill‖ (Zuchthausvorlage) which officially protected the actions of strikebreakers, and allowed for firms to engage in anti-striker violence. This law reduced the effectiveness of the strikes (Spencer 1984, 69).48 But prominent industrialists including Bueck, Emil Kirdorf, and Hanns Jencke made a point of emphasizing the threat of working classes in their support of the bill (Spencer 1984,

69, note 32). This legislation that would ultimately have benefited repressive employers ultimately failed in the Reichstag, as the majority of National Liberals,

SPD, Center, and Progressive political party deputies voted against it. This legislative failure caused Jencke to want the Ruhr elite to sever ties with National Liberal Party

(Spencer 1984, 70, note 34).49

This episode demonstrates that there were limits to state repression because of electoral considerations and costs of repression. The situation was not a completely one-sided class war were workers lost every battle: the state was willing to share in some costs of repression, but not a sufficient amount. Industrialists were concerned

48 Some firms outside of the Ruhr heavy industry circles did not favor proposal, even firms that operated within the ZDI. The chief opposition to the anti-worker bill came from German textile manufacturers; their spokesman Karl Krafft claimed that the law was unnecessary. 49 This indicates that the National Liberals only real support of anti-worker initiatives came from the right-wing faction of that party.

103 about this: in the 1908 meetings of the board of directors of a major business lobbying association (discussed below), industry representatives fretted that the NL party had drifted too leftwards for firms. After the failure of this legislation in 1899, these repressive employer associations demonstrated less desire to be dependent on government initiatives to share in the costs of repression. They turned to strategies that member firms could pursue. Consistent with the theory presented chapter 3, firms turned to bearing their own costs of repression (Niederschrift der Sitzung des

Direktoriums des Zentralverbandes deutscher Industrieller 1908; Spencer 1984)

1.2.4 Labor market strategies: blacklists and undermining workers

Section 1.2.1 documented directly how individual firms shared costs of repression with the state to repress workers. This strategy was consistent with, and facilitated by, the hostile attitude of the state apparatus towards the SPD. A direct labor market strategy that dovetailed with this strategy was the use of the blacklist— not employing workers who had previous union or SPD affiliation. Note that a blacklist could be managed individually or collectively; the latter approach is discussed in section 2. Their application was not restricted to heavy industry; textile firms in Augsburg and chemical firms in Ludwigshafen used them as well.

Large firms in Imperial Germany also could engage in strategies to divide workers, with the intent of preventing workers from joining an inter-firm union, or union that would receive support from the SPD. This ―divide and conquer‖ strategy involved giving some workers incentives not to join trade unions (Geary 1991, 143).

Some firms offered bonuses to workers who informed on their colleagues. Other instruments of worker differentiation included different incentives for blue and white-

104 collar workers; physical separation of workers from different regions; exploitation of ethnic tensions (between German and Polish workers for example); high differentiation based on skills; and importantly, creation of company unions that agreed to join broader inter-firm unions.

These firms also enjoyed an advantage in competing for workers—they had lower labor costs per unit of production, and could offer higher pay and more benefits to induce workers not to support unions. The costs of this strategy were low because

Krupp and other giant firms had less to worry about regarding labor supply, and had more applicants (as opposed to SMEs). These company-specific ―yellow unions,‖ did sometimes engender genuine loyalty among its members. They were useful for detecting abuses of lower supervisory personnel, and grew after the massive strike wave of 1903-06. By 1914 they had about 279,000 members and were strong in the

Ruhr and in mining, iron, steel, electrical, and chemical industries—industries that comprised the vast portion of German output. For example, in BASF in

Ludwigshafen in 1913, company unions were supported by over half the labor force.

Over 80 percent of the Siemens workforce in Berlin in 1913 belonged to a company union. These unions were important in administering company welfare schemes.

These company payments served as efficient wages according to the classical economics literature. But an alternate purpose of these institutions was that they helped large firms divide workers and prevent unions from gaining strength; they reduced incentives of workers to go on strike or join alternative, more threatening unions. The greatest extension of company welfare schemes and yellow unions came after 1904 and were simultaneously implemented with repressive measures.

105

How replicable were the company-welfare schemes of BASF, Siemens, and

Krupp? They were not copied by the smaller concerns because such schemes were too costly. Smaller firms, by contrast, were forced to conclude deals with strikers and trade unions. Of the 1.4 million workers covered by collective wage agreements in

1913, two-thirds were in firms employing fewer than 20 workers. These were generally in printing and construction (and related building trades); there were fewer in metal work and textiles and none in mining (Geary 1991, 145). The printing industry was dominated by small enterprises with higher skilled workers, and high competition. These variables made collective collaborative strategies more viable

(Geary 1991, 146). The key difficulty though is distinguishing between the decisions of firms to ascent to collective collaborative strategies because of the high costs of repression, versus desiring such strategies because of joint incentives with workers to invest in skills. In section 3.2, I discuss how the participation of these employer associations that had made ―union concessions‖ supports the first explanation more.

This evidence demonstrates the use of repression by individual firms in the late nineteenth and early twentieth century. In Section 2, I show how homogeneity within sectors, characterized by higher levels of industrial concentration, as well as lower heterogeneity across sectors—characterized by lower variance in the costs of repression—facilitated the rise of collective organizations that eventually became employer associations. Dissatisfaction with the unwillingness of the state to further roll back workers‘ rights led to a substitution of collective for individual strategies.

These employer associations often performed similar strategies as described above, and were seen as a more effective way to achieve labor market ends.

106

Section 2: Transition to Collective Strategies

In this section I first provide data on the spread of employer associations by sector; I then turn to explaining the emergence of specific trade associations and their transition to dealing with labor market issues and repressive strategies. I link the emergence of such strategies to the threat posed by workers and the structural conditions that enabled collective action. Throughout, I demonstrate the strategies of these employer associations were similar to those at the individual level, only the threat posed by workers allowed for such strategies to be more efficiently pursued collectively.

2.1 Prevalence of collective strategies

In this section I document the advent and proliferation of collective strategies among firms, in product and labor markets. This provides evidence of the proliferation of collective strategies in different industries. The period industrialization of Germany from 1871 was synonymous with an increase in trade and then employer associations, many of which initially operated as cartels. The pre- existing degree of coordination in product markets in the form of cartels facilitated the transition and creation of organizations to coordinate responses to labor concerns.

Many initial trade associations gathered information from member firms to use for lobbying purposes and managed quotas and prices in specific regions. These associations formed at the local, regional (Länder), and national levels, and generally operated with a specific territorial scope and/or over a specific industry.

In March 1903, Bueck presciently argued that in geographical or industrial areas that did not have strong SPD support, the formation of pre-emptive employer

107 associations could actually provoke workers into joining trade unions and the SPD

(Saul 1974, 99). But firms could not ignore the fact that unions were much more politically oriented; SPD gains in the Reichstag threatened employers. The question was which instruments now could be used to counter this threat. As Geary describes, before WWI, industrialists ―became increasingly shrill in their demands for state intervention against pickets and unions‖ (Geary 1991, 151). The strategies discussed above, including locking out workers, using blacklists, punishing workers for supporting the SPD, and lobbying the state, could also be pursued collectively. As the theory presented in chapter 2 discussed, firms are more likely to pursue collective strategies as long as heterogeneity within the sector is low. This is because the cost of joining the association and costs of enforcing the collective agreements are lower than paying the cost individually.

The first recorded employer associations to form were the Association of

German Master Printers (Deutsche Buchdruckerverein) in 1868, in response to strikes from the more centralized printers‘ union.50 Employer associations generally first formed in these small-scale handicraft industries in the 1870s (Spencer 1984, 319).51

The earliest incarnations of coordinated market activity occurred in industries that were less economically and politically influential. I discuss below, though, that even for these industries, their development is consistent with the theoretical predictions.

In 1889, 14 employer associations formed; the following year there were 29 documented. Before the early twentieth century, much of the coordinated activity among firms was occurring at the regional and local levels within important industrials

50 See Swenson (1989) and Kessler (1907). 51 These types of associations are emphasized in the account by Thelen (2004).

108 sector. I discuss in section 3.3 below, though, that even for these industries, their development is consistent with the theoretical predictions.

However, eventually more important, associations formed later in the industries of raw materials and manufactured goods. The development of employer associations in industries that would run the German economy (heavy industry, manufacturing of products based on such raw materials, chemicals) did not occur until later, until the early 1900s. In these industries in the late 1890s and early 1900s, they were dominated by large firms or concerns with mostly unorganized and unskilled workforces. This was true as well for industries in metallurgy and chemicals. As discussed, up until the early1900s, these large firms were capable of individually paying the costs of repression in preventing skilled and unskilled workers from organizing. Or, as in the Saar and Silesian areas, in concentrated industries like mining, specific regional-industrial associations were sufficient to contain labor disputes.

The main purposes of these employer associations that developed later in major industries was protecting the supposed ―right to work‖ of workers and to assist workers who were willing to work. As I discuss in Chapter 7, early associational leaders in the United States used similar rhetoric against a much weaker labor movement). This purpose applied to associations organized as well as the local and industry levels (Saul 1974, 101). Before 1904, employer associations existed mainly in areas of volatile union agitation, as well as in the Hansestädten in large German cities in the north and south. In southern Germany, including the Saar area, Alsass-

Lorraine, and in Silesia, eastern Prussian provinces, where union density was

109 relatively lower, there were fewer institutionalized employer associations (Saul 1974,

101). For example, the Association of German Metal Industry (Gesamtverband der

Deutschen Metallindustriellen, GDM) and the Employer Association of Hamburg-

Altona (Arbeitgeberverband der Hamburger-Altona, AHA) were important industrial and regional employer associations that responded to increasing union efforts. These two associations also had ties with each other; the Hamburg Association of Iron

Industry (Hamburger Verband der Eisenindsutrie, HVE) was institutionally tied to the

GDM, and was the most important member association of the AHA.

I now turn to systematic data on the employer associations and their functions, which only are available for a number of industries after 1909. What was the initial scope and coverage of employer associations? Systematic data on the number of employer associations and their functions only became available in 1909. Related data on the diffusion and variation in employer associations, and their relevance to firms‘ interests, are discussed in the annual memoranda of a prominent trade association, the

Association for the Protection of Common Economic Interests in Rheinland and

Westphalia (Verein zur Wahrung der gemeinsamen wirtschaftlichen Interessen in

Rheinland und Westfalen). I discuss this organization, dubbed by Bismarck the

Langnam Verein (―Long-named Association‖) more extensively in this section below.

For ease of abbreviation, I refer to it by the acronym LNV hereafter. These memoranda summarized and discussed the newly gathered government statistics regarding the growth in employer associations in the twentieth century.

According to its 1909 memoranda, Germany had a total of 2,591 employer associations. Of these associations, 127 were classified as independent associations

110 that were responsible for organizing membership of either member associations or member firms. These associations had ―affiliate‖ associations that were ―sub members.‖ Of these 127, there were 46 employer associations at the national level; these associations had an additional 22 national-level, 212 regional-level, and 1,442 local-level associations attached.52 There were 38 independently operating regional employer associations (Landes or Bezirk level); these associations had 16 sub-regional and 80 local level associations connected to them. Finally, there were 43 independent associations at the local level.53

The total number of firms covered by these associations was 159,304, many of which were the largest firms in Germany; these member firms employed almost 3.7 million workers (Beumer 1909). The authors of the LNV memoranda also optimistically pointed out that this total was much greater than the contemporaneous union density, which was about 2.4 million at the time. In the memoranda for 1910, the LNV also reported figures on the growth of employer associations, classified by industrial sector. The number of firms that were member of associations fell to about

115,000, but the number of workers covered by such member firms increased to 3.85 million, again outpacing union growth. The major industrial sectors of minerals and mining manufacturing metal products, and construction had the highest number of employees who worked for firms that belonged to employer associations. In mining

52 Discrepancies between the LNV reported data and the 1909 data as reported by the German government are discussed. 53 This data comes from the Mitteilungen des Vereins zur Wahrung der gemeinsamen wirtschaftlichen Interssen in Rheinland und Westfalen, 1909 and 1910. In these discussions the original author is not noted, though sometimes interjections are made by noted members, such as Wilhelm Beumer. More information about specific associations is available from the German statistical yearbook, discussed below. The LNV Mitteilungen discussions are of aggregate counts of associations, not discussions of specific associational features.

111 and related industries, there were nine associations with about 240 member firms, responsible for over 450,000 workers. In the industry of metal finishing and machine building, there were 168 associations covering over 12,000 firms and 750,000 workers. In textiles, there were 88 associations covering 600 firms and 470,000 workers. The last ―major‖ associational industry, construction, had 1,226 associations covering 51,000 firms and over 520,000 workers (Mitteilungen 1910, 231).

The data from the 1909 German statistical yearbook (Reich 1909) further validate the claim that firms participated in collectivist strategies in the early twentieth century, and that such participation was not limited to a few industries.54 The government collected data and calculated a total of 46 national-level associations with

27 national-level associations as attached members, across all major industries. The most important national-level associations, the Central Office of German Employer

Associations, and the Union of German Employer Associations, are discussed in section 2. Other national-level associations with large numbers of member firms and workers included the German Employer Associations for Construction (Deutscher

Arbeitgeber. F. d. Baugewerbe), with 21,000 firms and 250,000 workers; and the

Association of German Metal Industry (Gesamtverband Deutscher Metallindustrielle,

GDM), with 2,800 member firms and over 500,000 workers. In terms of services rendered, 89 percent of associations collected data on member firms, and 77 percent collected data on the employees of the member firms (data might include wages or worker histories).55 In chapter 4, I use this data on national-level associations to

54 See the Statistisches Jahrbuch für das Deutsche Reich (1909). 55 Examples of these early important employer associations include the Association of German Bottle Manufacturers (Vereinigung deutscher Flaschenfabriken), the Union of German Boot and Last

112 systematically test which aspects of employer associations are more likely to predict the use of specific collective strategies, namely the adoption of strike insurance.

More systematic data on employer associations by industrial sector exist for the period 1911-1913. Table 4-1 displays characteristics of employer associations.

For simplicity of presentation the table displays period averages for a number of characteristics by major industry category. The table shows the distribution of the number of employer associations by industry for the period 1911-1913. I use the raw data from the statistical yearbooks to calculate a number of indicators of each industry, regarding the coverage of employer associations. Columns 2-5 display characteristics of employer associations by industry, without reference to the number of firms and workers in the industry as a whole. Columns 2, 3, and 4 display for each industry the number of employer associations, member firms, and employed workers, respectively.

Column 5 displays the number of workers per association. The construction industry had the highest number of associations in this period, with nearly 1,400. This industry also had the highest number of firms that were members of employer associations, with about 49,000 member firms employing over 500,000 workers. But the major finished metals and manufacturing industries had associations with member firms employing the most number of workers; over 860,000 workers were covered by such associations. The mining industry had the highest ratio of workers per association, nearing 50,000 workers per association.56

Manufacturers (Verband der deutschen Schuh- und Schäftefabrikanten), the General Union of German Metal Manufacturers (Gesamtverband deutscher Metallindustrieller), and the Employers‘ Union of Hamburg-Altona (Arbeitgberverband Hamburg-Altona). 56 The count of associations includes associations at the national, regional (Landes or Bezirk), and local levels.

113

The two rightmost columns of Table 4-1, Columns 6-7, display characteristics of employer associations with respect to attributes of the entire industry. The data on characteristics of different industries come from the 1909 German Census (Statistik des Deutschen Reichs).57 In chapter 4 I conduct more specific econometric tests using such data. The columns display one basic measure of the degree of associational coverage with respect to the number of firms and workers in the industry. Column 6 displays the ratio of the number of firms in the sector to the number of member firms in the association. Column 7 displays the ratio of the number of workers employed in the sector to the number of workers employed by member firms in the employer associations of that sector. Lower ratios suggest that the given industry has associations that encompass more of the industry‘s firms. For example, approximately

25 percent of all firms in the agricultural, polygraphics, and construction industries were members of employer associations. If we consider the ratio of number of workers in the sector to the number of workers employed by firms that are members of employer associations (column 7), the ―most represented‖ industries are the three aforementioned industries, as well as the most important industries of manufacturing and mining and heavy industry, and textiles. Approximately one-third of workers in manufacturing were in firms covered by employer associations; this ratio is estimated to be one-half for the textile and raw materials/minerals industries. This sector produced the largest amount of German output, and had many workers covered under employer associations at the local, regional, and national levels.

57 See Statistik des Deutschen Reichs (1909)

114

These measures of ―organizational scope‖ likely underestimate the substantive impact of these associations, as in many of the most important industrial sectors, the most important members were the largest firms, discussed in detail below. This evidence demonstrates the proliferation of collective strategies in Germany at this time period, as well as their concentration in the most important industries. In chapter 5, I use the data presented in Table 4-1 as well as further data on industry characteristics to estimate the effect of industrial heterogeneity on adoption of collective strategies by firms. I use similar data on industrial characteristics to predict which characteristics of the national-level employer associations made them most likely to adopt the collective strategy of strike insurance.

2.2: From individual to collective strategies: the formation of regional repressive employer associations

Table 4-1 demonstrates the prevalence of collective strategies of firms in

Germany, although employer associations varied considerably in the degree of collective action and obligations of member firms, as well as the actual services provided to member firms. I now provide evidence on how the emergence of these associations was due to increases in threats posed by unions. The most important initial collective activities of firms were in cartels and trade associations in the industries of coal, iron and steel manufacturing, and the finished goods used from those inputs. These trade associations turned to repressive strategies against workers in the early twentieth century.58 These trade associations, even before they shifted

58 They also formed to lobby for investments in transportation infrastructure, as well as government recognition of official cartels. The most important cartels included the Rheinisch-Westfäliches Kohlensyndikat, which formed in 1893; it was a very rigid cartel, represented 87 percent of all coal

115 resources to forming employer associations to deal with unionization and workers‘ demands, were involved in attempts to counter the increasing political contestation and electoral success of the SPD.59

This section provides evidence that the threat posed by workers induced the adoption of firms‘ collective repressive strategies. As I show below, a commonality of these collective strategies was that they emerged in response to threats posed by workers—either electoral threats posed by the SPD or unionization threats across different industries. Once the redistributive threat by workers increased, firms had an incentive to form associations to respond to the threat. Collective strategies in product markets—the organization of cartels and a lobbying infrastructure (plus associational personnel also serving in political capacities)—facilitated the creation of organizations, with the same member firms responding to increasing contestation of electoral districts by the SPD and increased unionization with repression (Spencer

1984; Mitteilungen der ―LNV‖ 1909, 1910). These regional and sector-level associations came to overtake unions in numerical strength and defeated them in industrial disputes (Geary 1991, 144).

Incidents of industrial conflict provided crucial movements for collective mobilization of firms within regions and industrial sectors as well as among regions and sectors. These incidents included demands for strikes; actual strikes; and

produced in Dortmund. It was more difficult to form cartels in the steel industry due to higher product diversity. The Stahlwerksverband formed in 1894, representing 84 percent of steel production in Ruhr. 59 Additionally, some cartels had provisions which compensated members for lost output during industrial conflict. For example, coal syndicates had provisions where member firms would take over supply functions and compensate firms for losses during industrial conflict. But these cartels did not have formal procedures for joint lockouts or collaborate with other sectors or associations. It was members of these cartels that eventually saw the need to use their existing collective institutional infrastructure to form employer associations (Spencer 1984)

116 sympathy walkouts by workers. To be sure, there was much variation in the approaches and reasons for conflict across different industrial sectors and firms.

Similarly, when such conflict occurred in proximity of national elections, the redistributive threat posed by workers‘ organizations was made even more salient. As discussed in chapter 3, the ability of firms to pursue collective lockouts was an important repressive strategy in response to demands from workers and initial industrial conflict. Lockouts were often used in anticipation of strikes or in response to strikes. During major incidents of industrial conflict in this time period, they allowed for firms to further mobilize. In observing the problems caused by strikes,

Heinrich Kamp of the Phoenix industrial concern argued that employer organizations were needed before, rather than after industrial conflicts.60 Employer associations resorted to lockouts and occasionally police backing of such lockouts to prevent pickets. Such techniques increased during recessions. Lockouts increased during the

1901-02 recession; 600,000 male-worker days were lost; this increased to 1.3 million in 1903, and 1.7 million in 1904, to 4.4 million in 1905 (Spencer 1984, 106, note 37).

Where did lockouts happen? Lockouts were most prevalent in handicrafts and finished product manufacturers–metal, construction, machine building, electrical industries (Spencer 1984 106, note 39) These industries made extensive use of lockouts, and had common employment of high percentage of skilled workers and already highly developed organizations of both labor and management. Note, however, as discussed in chapter 2, collective lockouts imposed costs on firms as well, and thus were not the first preferred strategy. Where employers were not already

60 Spencer 1984, 104, note 32

117 highly organized, the lockout was ineffective. Firms in highly concentrated mass industries could use other low-cost methods to repress workers. As Spencer writes, the ―…the employer had to be certain that other employers in the same or related industries would not hire his workers, and if need be, would even instigate sympathy lockouts…where workers were not highly organized, as in the primary iron and steel industry, a lockout was not necessary. Less drastic methods sufficed‖ (Spencer 1984,

106). Large firms with unskilled workers had alternative repressive strategies besides mass lockouts—they could just dismiss workers summarily, and undermine unionism.

By contrast, coal, iron, and steel firms were reluctant to dismiss entire workers who were already hard to recruit in coal. One fear, more extensively discussed below, was that large firms did not want to lock out workers and drive them into unions (Spencer

1984, 106, note 40).

2.2.1 Regional coordination in coal mining

I now turn to specific cases of industrial conflict and the threat posed by workers as a mobilizing incentive for the shift to collective strategies to deal with workers. In 1858, an influential mining association, the Bergbau Verein (BV) formed in the Ruhr area formed to secure cheaper and more efficient transportation, and coordinated with other mining associations. The BV became the main economic association representing Ruhr mining, particularly regarding relations with state

(Spencer, 1984). What kinds of industrial disputes faced the BV and similar associations? The regulation of the workday was an obvious central issue. In the mining industry, the main concern was penalties of the Wagenullen—employers‘ refusing payment for underperforming or insufficient loads of coal; similarly disputes

118 existed regarding whether workers would be paid in the rolling mills for producing low quality sheet metal. Other issues of dispute included non-uniform policies; forced overtime and Sunday work; and arbitrary re-calculation of wage rates. A critical problem for firms was the continuous production with rolling mills, and irregular pauses regarding problems with processing molten metal. Disputes existed over the separation of workers into 12-hour versus eight-hour shifts (a transition to three shifts would have caused increases in labor costs and worker shortages) (Spencer 1984, 115, note 2).

In May 1889 the first major mass strike in mining occurred since 1873. 87,000 out of 104,000 Ruhr miners struck. The main issues were regulation of miners‘ demands for eight-hour shifts as opposed to existing 12-hour shifts. It was a serious enough disturbance that the Prussian Ministry of the Interior and provincial governors discussed the possibility of extending the civil forces (i.e. state police or municipal police, instead of using military assistance (Johansen 2004). The behavior of the BV and its employer association demonstrate the importance of repression as a response, both in terms of collective action against firms as well as associational lobbying of the state. The BV was the employer association that was most hurt by the strike and it initially rejected any industry-wide negotiation with worker representatives; opening negotiations would be an automatic concession (Spencer 1984, 64). It also opposed state intervention to preserve their autonomy in negotiations with unions; this displeased state authorities, as Chancellor William II wanted to avoid a protracted labor dispute. Friedrich Hammacher, chairman of the BV as well as deputy for the

National Liberal Party in the Reichstag, eventually entered into negotiations with the

119 miners‘ representatives and urged firms to engage in negotiations with miners

(Spencer 1984, 65, note 8). This government interference irked the BV members (the mine owners), who thought Hammacher had betrayed them; he was replaced as head of the BV in 1890 by the hardliner Hanns Jencke. Jencke pursued a strategy after

1890 of constantly supporting firms in violating contracts with workers. For example,

Jencke successful convinced his fellow industrialist Krupp to purge all workers who were Social Democrats from his firm (Spencer 1984, 67). This industrial incident demonstrates the importance of a threat from workers in terms of repressive responses, both political and direct labor market intervention.

The recession years of 1908-09 allowed the industrial elite in heavy industry the opportunity to increase repressive strategies against workers, who were in a more vulnerable position (Spencer 1984, 114). In 1908, the BV founded another important trade association that was more oriented towards the large industrial concerns, the

Zechenverband (ZB). The ZB was designed to manage conflict against workers and to respond to unionization concerns, and inhibited unionization in the coal mines

(Spencer 1984, 104, note 35). The ZB disseminated anti-worker propaganda and coordinated efforts to prevent firms from acceding to trade union demands among pre- existing local organizations. It was intended to advertise to workers and the general public that industrialists were meant to be unified, to publicize, and to deter workers‘ efforts at organization. The ZB ―…included almost all important leaders of Ruhr mining‖ (Spencer 1984, 105).

The BV and ZB use these strategies to regulate employment on a permanent, industry-wide basis, using the information on worker lists as a weapon. Industrial

120 concentration and the design of formal associations facilitated such action (Spencer

1984, 108). Members of the mining employer associations were obligated to not to employ workers for at least six months if they had failed to give proper notice since leaving their last job, or had previous ties to unions; employers who were discovered to have not complied were fined 1,000 marks for each violation. Lists of names of workers who were not to be employed circulated among firms (Spencer 1984, 107, fn

46). Blacklist or ban on workers with previous SPD ties was enforced in the Ruhr area except in a few small mines outside the Zechenverband.61 As a result of this collective repression, by 1910, in the coal mining industry, only 82 individuals were covered by collective wage agreements. This is the canonical highly cartelized and concentrated sector with effective repressive employer organizations that employed both yellow unions and repression. The workforce was generally unskilled, though higher wages were needed to attract workers to the more dangerous and isolated mining areas.

There were similar circumstances in the chemical and electrical industries, textiles, and sections of engineering (Geary 1991, 146).

2.2.2 Regional coordination in heavy industry and beyond

61 There were extensive Reichstag debates on this black list issue (Spencer 1984, 108, fn 48). Some of the more left-wing NL politicians and business lobbyists like Gustsav Stresemann spoke against bans and blacklists (Spencer 1984, fn 49). These worker blacklists in the coal industry were replaced by a less extreme form of labor exchanges, discussed above (Arbeitsnachweise) (Spencer 1984, 108). How did these ―labor exchanges‖ work? They let workers know when and why they were being punished and provided possibilities for appeal (Spencer 1984, 109). They were a little less repressive, but still preserved firm autonomy in dealing with unions. However, the Ruhr iron and steel producing and consuming firms declined to follow the lead of coal industrialists in creation of labor exchanges. Major firms rejected such a labor-controlling proposal in 1914. The Phoenix steelworks argued that the interests of iron and steel were different from those of coal industry; the coal industry was more cohesive, had a greater homogeneity of work force, and therefore regional coordination was easier. But the key reason for the RHI rejection of the policy was that the creation of labor exchanges would destroy good relations between Ruhr metallurgical concerns and their workers, and would have given the unions a further basis for agitation (Spencer 1984, 109, fn 56). This is evidence consistent with the theory: permitting labor exchanges would have led to higher costs of repression. Thus the original strategy of repression using lower cost strategies was preferred.

121

In 1874, the largest industrial businessmen in iron and steel formed the influential Association of German Iron and Steel Industrialists (Verein deutscher

Eisen- und Stahlindustrieller, VdESI). The VdESI‘s initial goal was also to lobby for tariffs on competing goods. The VdESI was composed of eight sub-groups, including six regional groups and two groups representing industrial branches that also included finishing and manufacturing firms. The VdESI also formed as a reaction to the depression of 1873, and supported the alliance between heavy industry and large agriculture estates.62 The six regional groups of the VdESI were the Northwest, East,

Central German, North German, South German regions, each with their own respective branch offices. The two groups of industrial branches were the North

German Railroad Car Builders Association and the Association of German

Shipbuilders. Among these eight groups, the Northwest Group (NWG) had the most influence on the VdESI‘s policy positions (Poensgen 1929). This association also had firms that produced finished manufacturing goods as well, and more industrially complex firms such as the electro-technical giant Siemens–Schukert concern.63 The

Southern and Central German groups were composed almost entirely of machine construction works (Kaelble 1967; Feldman and Nocken 1975).64

62 The VdESI members coordinated on hours and over-time regulations. Within the VdESI, there were distributional concerns between large and small firms. In May 1913, the problem of excessive overtime was discussed; they tried to reduce voluntarily excessive hours. One problem was that smaller, less modern, and less diversified firms did not want to implement changes (Spencer 1984, 137, fn 45). The substitution of shifts was harder for such firms because of higher labor costs and difficulty in finding workers; there the ―…problem of obtaining enough additional workers was a bigger obstacle‖ (Spencer 1984, 138). 63 Feldman 1975, 417 64 See Winschuh (1932). As an example of coordinated lobbying for repressive measures, members of the Northwest Group had specific meetings to discuss government regulation in workers‘ affairs. On February 6 and September 23, 1908, representatives met in Düsseldorf to discuss problems of government intervention, and agreed not to concede to calls for more collaborative strategies with workers (Spencer 1984, 116). Each employer was to send a letter to the Regierungspräsident of the

122

These associations expressed opposition to government efforts to intervene in labor markets and internal operations of firms. Their lobbying efforts were a reaction to perceived encroaching government authority. For example, the above associations protested the proposed government introduction of worker safety inspectors

(Sicherheitsmänner) in 1909. Employer associations and firms that were members of the Bochumer Verein and the Northwest Group regularly condemned working-class presses. These associations also made financial contributions to Reichstag parties

(Spencer 1984, 118, note 18). During the 1907 election campaigns, the Northwest

Group and the Bergbau Verein, contributed 15,000 marks. Krupp director Alfred

Hugenberg tried to centralize political tactics and sought to coordinate political financing; he wanted all contributions from the BV to go through his office, and wanted to make sure that financial aid resulted in benefits for donors (Spencer 1984,

119; (Weisbrod 1978). (Hugenberg‘s extremism and further attempts at centralizing repressive strategies are discussed in Chapter 6). Managing Director Paul Reusch of the GHH wanted the Northwest Group to be involved in more anti-socialist interest groups (Spencer 1984, 55). In fact, Reusch sponsored the anti-socialist interest group the Deutsche Vereinigung, which was formed in 1908.65

The logic of collective repression as a response to the threat posed by workers extended is evidenced in the behavior and formation of the Northwest Group. Two

district in which his firm was located, asking that he be given the privilege of limiting workers‘ lunch breaks. These protest letters were sent regardless of whether the sender viewed longer breaks as costly (Spencer 1984, 116). 65 Resuch threatened to resign from the Northwest Group of the Association of German Iron and Steel Industrialists and its employer association, the Arbeitnordwest, if those organizations denied his request for a contribution to the Deutsche Vereinigung. Although these industrialists despised hated social democrats and some even universal suffrage, some thought that political activities did divert them from more useful work (Spencer 1984, 56).

123 strikes by skilled ironworkers in Dortmund in 1905 increased interest in the association.66 However, even within the relatively homogeneous membership of the organization, there were different costs of repression among member firms, and such cleavages undermined its effectiveness. There was a split between iron and steel producers and iron and steel consumers. Primary producers thought the finished product firms were too conciliatory towards workers; symmetrically, the machine- building firms thought iron and steel producers were too rigid, and that their repressive tactics, ―invited attacks from labor‖ (Spencer 1984, 104, note 34). The producers were able to dictate policy because they were the dominant supplier for metal industries in the region. This cleavage between these kinds of producers would affect firm strategies toward workers during Weimar Germany, discussed in Chapter 6.

The repressive efforts of the NWG are also discussed in the memoranda of the aforementioned Association for the Protection of Common Economic Interests in

Rheinland and Westphalia (the ―LNV‖, formed in 1871). The LNV formed to manage cartels (quotas and prices); lobby for transportation infrastructure spending; and lobby for higher tariffs.67 The LNV was founded by the major iron, textile, and coal concerns in Rheinland and Westfalen. These associations in heavy industry generally formed initially. The founding members of the LNV were all influential industrialists in heavy industry and had investments in other firms, and some would become

―association men,‖ that is men who in addition to their posts as firm owners, would manage funds and institutional structures for collective organizations. Prominent

66 Spencer 1984, 104, note 33 67 See Winschuh (1932); 1886 and 1907 reports of the Verein zur Wahrung der gemeinsamen wirtschaftlichen Interessen in Rheinland und Westfalen (1886).

124 members included William Thomas Mulvany, who had founded the Prussian Mining

Stock Company (Preussische Bergwerks- und Hütten-Aktiengesellschaft, PBHAG) and was one of the founding members of the LNV; Ernst Poensgen of

Stahlwerksverband-AG; Paul Reusch of the Gutehoffnungshütte (GHH); and Fritz

Springorum of Gelsenkirchener Bergwerks-AG. These men would represent the

―Herr-im-Haus‖ attitude that of large firms that guarded their autonomy against the incursion of workers. As Chapter 6 discusses, some of these men would become important in the inter-war German period for trying to revive the repressive tactics that they employed in the early twentieth century.

The memoranda provide useful clarification as to what firms thought at the times the best benefits of collective strategies against workers were. The LNV discussed both the NWG‘s and its own valuing of the importance of a counter- response to union mobilization and electoral contestation of the SPD. The reports praised the anti-labor and union restricting activities employer associations, in particular their role in defending against strikes and their role in managing employment exchanges. The LNV credited employer associations with helping to exhaust union treasures during industrial conflicts, and saving firms from long, drawn out lockouts. The reports also discuss supporting the coal industry during industrial conflict, even though some coal industry firms were not formally affiliated with the

LNV. The memoranda also detail their opposition to collective agreements; the organization‘s support for peak associations‘ positions in recent industrial disputes and support of firms in the printer trades (though the specific conflict and region were not mentioned in the discussion). The report authors also approvingly cite other

125 interventions of peak employer associations‘ roles in helping firms win those battles, and the importance of employer associations to restrain demands of workers and combat their radical tendencies (Mitteilungen 1909, 217).

These reports also discussed the treasures of trade unions and compared membership growth in the unions versus the count of member firms in employer associations. Beumer mentioned the ―willingness of workers to be ‗victims‘‖ during industrial conflict, and that firms should have a similar degree of fortitude

(Mitteilungen 1909, 217). (The LNV report estimated total trade union membership at

2.5 million, with a treasury of approximately 41 million marks, down from the previous year‘s estimate of 57 million marks). The report also mentioned the specific threat posed by social democratic unions and their role in promoting other labor organizations. Beumer in the report also exhorted employer associations and member firms to recognize the importance of the political intervention in assisting firms.

In the 1910 report, the LNV noted the smaller union growth , crediting employer associations for the higher relative growth in employer organizational membership, from 3.9 million workers, as compared to 3.7 million from the year before (though membership declined to slightly 115,000).68 The report also noted the strength of the employer associations in the construction, metal, and textile industries, as well as iron and mining. They warned not to let down after their initial successes, and to prevent member firms from signing agreements with unions. The strikes and organizations in the construction and shipping industries demonstrated the need for employer associations to ―remain vigilant‖ (Mitteilungen, 1910).

68 They noted that union growth was smaller: overall trade unions, there was an increase in 46,000 as opposed the 200,000 more workers that associations were responsible for.

126

LNV reports from this time period also summarized the discussion from a meeting from the Northwest Group of the VdESI of that year (1909, 218). The

Northwest Group also referred to the ―secret terror‖ of trade unionists. (Mitteilungen,

1910, 232). The Northwest Group report praised the reduced number of strikes as a result of the successful unification of firms. Like the LNV, the Northwest Group also compared the growth trends in trade unions to that of employer federation. The

Northwest Group report concluded that in light of unions‘ increased militancy, cooperative strategies were not possible. The group also praised recent efforts of the association to gather data, and how the circulation such data on worker histories and wages were useful in efforts to end industrial conflict, such as the strike in Aachen in

1906.69 The LNV then went on to discuss recent efforts of the national government to collect data on employer associations. The Northwest Group praised the higher number of workers who were employed by member firms of employer associations, as opposed to unionized workers—they compared the former count of 3,648,579 workers to the unionized number of 2,421,185. The Northwest Group also noted that there would be a merger between the two peak employer associations at some point, and praised the recently signed cartel contract between the two associations. The

Northwest Group report discussed the ―pressure‖ and ―counter-pressure‖ logic of employer associations versus unions, and in particular emphasized that employer organizations needed to form to respond to the ―arbitrariness‖ of the disputes initiated by the powerful social democratic unions. They argued that the importance of the

69 The Northwest Group report also noted positive trends in the organization‘s five-year history and most recent trends.

127 financial defense provided by employer associations could not be captured by data alone.

The Northwest Group discussions also argued against collective agreements and how they bound workers. The NWG report discussed collective bargaining statistics: 2,090 collective bargaining agreements/associations. The metal working industry had 128 such agreements, covering 230,195 workers. These agreements generally covered the handicraft industry; an agreement covered on average about 11 firms and 10 workers. The group positively interpreted this data: the iron industry though had on average only 14.4 percent coverage in firms with more than 50 workers, while the participation of workers of that firms of that size was about 60 percent of working population. Thus the Northwest Group was satisfied with the efforts of the industry thus far in resisting collective agreements.70

Influential trade associations that would develop repressive strategies also formed in industries that did business with the above associations—in particular industries that purchased input materials from the firms and cartels represented by the above economic associations. These associations also had large firms, but with more skilled and complex products. The Association of German Machine Builders (Verein deutscher Maschinenbauanstalten, VDMA) was founded in 1892 by 27 firms that were economically linked with the iron industry. Its goal was to coordinate with the mines and steel mills that they supplied with machinery (Feldman and Nocken 1975,

420). By 1914, the VDMA had 246 member firms covering 189,000 workers. The

VDMA had a cartelized agreement with the VdESI, under which member firms

70 The LNV also noted concerns about anti-picketing legislation.

128 received export rebates were granted by the VdESI to member firms. Disputes existed regarding the amount of export rebates, as the steel cartels that constituted the VdESI and ZDI dictated that only member firms could benefit from such rebates. As within the Northwest Group, the cartelized agreements between these two factions of capital would play an important role in helping repressive strategies during the Weimar era; by 1923, 90 percent of machine construction industry firms belonged to the VDMA.

Of course, the opposition to increased unionization and collective bargaining was prevalent in other sectors. But, in other industrial and more heterogeneous sectors, the relative quiescent labor movement meant that firms did not need to bear the costs of forming an interventionist, repressive employer association: individual firm repression was sufficient, and thus employer associations were less substantively important for industrial conflict. The interest group association representing the chemical industry, the Association for Protection of the Interests of Chemical Industry in Germany (Verein zur Wahrung der Interessen Chemischen Industrie Deutschland,

VWICID) for example, warned against unionization (Saul 1974, 76). At this time in the early twentieth century, the chemical industry was relatively strike free and employed unskilled workers (Saul 1974, 108). The chemical industry became more organized in a formal association in 1913. In February 1904, representatives of the

Chemieverein, argued that ―…a need for an employer organization does not exist in the chemical industry, since the relationship between employers and working takers was satisfying generally so far‖ and ―…the union of the employers would presumably drive workers into the arms of social democracy,‖ referring to workers who were not yet unionized (Saul 1974, 108). These concerns were similar to concerns expressed by

129

Bueck: forming pre-emptive employer associations in areas where the workers posed little redistributive threat would only backfire. This further indicates that collective repression was more likely a response to the threat posed by workers.

2.3. Transition to national collective repression

Many of the cartels and concerns that were influential in forming the LNV and

VdESI then formed the national level industrial association, the Central Association of

German Industry (Zentralverband der Deutschen Industrie, ZDI) in 1876, which claimed to speak for all of German industry, but its leadership was in reality composed of heavy industry firms. The ZDI had institutional ties with the VdESI, LNV, and

NWG. Like the other sector-level associations, the ZDI‘s initial purpose was to lobby for tariffs. Henry Axel Bueck was leader of the LNV and NWG in 1873, and between

1893 and 1912 was the business leader (Geschäftsführer) of the ZDI and the VdESI.71

There was much firm and personnel overlap within this constellation of associations.

The personnel that lead the Northwest Group of the VdESI were also leaders within the LNV as well as the ZDI. These association managers also had overlapping roles in politics, discussed more in section two of this chapter. Wilhelm Beumer was the business manager of the LNV and Northwest Group within the VdESI; he also served as a member of the executive committee of the ZDI and as a Deputy in the Prussian

Lower House and the Reichstag. Bueck was business manager of the VdESI and the

ZDI and a member of the Prussian House of Deputies.72

71 Spencer 1984. In a letter from Bueck to Alexander Tille, Bueck lamented the influence of social democratic ―theories‖ and the public hostility towards firms (Bueck 1901). 72 The NL party was divided in terms of business support. Only a minority—conservative parties and the right wing of National Liberals, supported heavy industry consistently.

130

As with the industrial and regional-level evidence, the most direct evidence supporting the link between the threat posed from workers and the formation of national-level repressive employer associations comes from the numerous initial statements of the leaders of these associations when they took on responsibilities regarding industrial relations, or founded affiliated associations to handle relationships with workers. The founders of the ZDI clearly expressed their preferences against cooperative strategies towards workers. The ZDI in early circulars opposed any negotiations with trade unions. Bueck remarked in 1890 that union interference in production would be firmly rejected employers, and that the ZDI would never negotiate with unions. Similarly, a 1905 ZDI declaration said that collective agreements with workers would not be pursued, out of concern of appeasing their radical notions.73 Bueck remarked: ―The employers‘ organizations have for their object a common defense against the unjustified efforts and claims of the workers and their organizations, who seek to impose uniform labor conditions or to interfere in the administration of undertakings, and the organization of resistance to projected strikes‖

(Krüger 1926, 330).74 Their goal was ―…solely resistance to excessive and unjustified claims and measures of organized labor‖ (Krüger 1926, 331). The employer organization‘s function was to support individual firms, and to help them to prevent threatened strikes, or to resist strikes forced upon them. Bueck succinctly captured the importance of collective repression: ―The spirit of solidarity among employers needed

73 See Kessler (1907), and the ZDI resolution May 5th, 1905. 74 Krüger does not provide the original documentation for this quote, but similar sentiments are expressed in the aforementioned letter to Alexander Tille in 1901 as well as in Bueck‘s autiobiography. See Bueck (1997).

131 to be aroused and continually strengthened, until individual interest came to accept the principle of voluntary submission to the common interest‖ (Krüger, 1926, 331).

To that end the ZDI founded the German Employer Newspaper (Deutsche

Arbeitgeber Zeitung, DAGZ), which argued that employer associations need to defend against unions and social democracy and urged disorganized firms to join organizations. Articles in the DAGZ referred to the importance of workers‘ class consciousness as a threat, and the danger of firms being lost to union influence.

Members of the ZDI argued that higher coordination among employer associations was a necessary response to the strengthening social democratic movement, as seen in

SPD‘s improved electoral performance in 1903. By this time, the SPD increased the number of districts they contested in to over 90 percent of all districts. The ZDI supported the formation of interest group associations that wanted to reduce in the influence of the SPD, including the National Association against Social Democracy

(Reichsverband gegen die Sozialdemokratie, RGS); Bueck thought at the time that these anti-worker and anti-democratic interest groups could work with German political parties. The 1903 Reichstag electoral results symbolized ―flammable writing on the wall.‖

Members of the ZDI supported restricting the franchise and became disillusioned with a majority of parties in the Reichstag that did not. By the early

1900s, only the Free Conservative party and National Party supported such measures.

Regarding political interventions, employer associations wanted to support only reliable candidates would ―energetically represent‖ business interests. After 1903,

Bueck hoped to form an ―industrial-conservative‖ party that, given the difficulties in

132 enacting franchise restrictions, would at least reduce the political clout of the Center party and SPD. They turned to other measures utilizing instruments between firms and in the workplace to restrict demands from workers. To respond, institutional organizers had to mobilize existing local-level associations (Saul 1974, 102). Articles in the DAGZ specifically called for a joint effort of bottom up (using pre-existing local level employer associations) and top down (powerful sector-level associations would have to organize peak federations). This movement was against not just unions, but against the threat of social democratic ideology.75

ZDI staff also expressed concern about the effective limits to the cooperation with the state in terms of repression. Industry representatives expressed some suspicions of the pro-labor leanings of the Center party, and concern that the National

Liberal party had drifted leftwards. There was explicit concern that elections might be called early, and thus they pushed for a larger effort on election spending against SPD candidates. However, these trade associations also decided against forming a central political association that would be directly linked to a supportive political party, particularly the Free Conservative Party. The argument against such a move was that they were afraid of popular backlash of obvious association of employer associations with just one political party. Industry representatives at the CDI meeting advocated instead local level funding against SPD candidates (Niederschrift der Sitzung des

Direktoriums des Zentralverbandes deutscher Industrieller 1908).

75 Geary writes, ―…the wave of organization that embraced German industrialists between 1905 and 1913 and which saw the creation of the yellow unions came in the wake not only of the great Crimmitschau strike but in the wake of impressive Social Democratic gains in elections to the Reichstag‖ (Geary 1991, 151).

133

The ZDI also battled legislation that would have forced collaborative strategies with workers. In February 1908 the Reichstag accepted a resolution from the Center party that would have allowed for collective bargaining in the maritime and shipping industries.76 Acceptance of the resolution and similar legislation could have legitimized or supported similar demands for such legislation or outcomes in other industries. The Center party in fact hoped that this resolution would help the Christian trade unions in the Saar and Ruhr areas. The Center, Social Democratic, and National

Liberal parties supported this proposed legislation.

The industry that would be most directly affected by this legislation, the shipbuilders, opposed this resolution; the Association of German shipyards lobbied against it and warned that union approval of collective agreements would be grounds for an industry-wide lockout. The resolution, they argued, would threaten the industry and the existence of the federation. The ZDI argued against legislation; as noted, the

ZDI viewed collective agreements as a symbol of workers‘ rights that could not be tolerated. Other employer associations argued that if this resolution were adopted by the naval authority, shipyards and the metallurgical plants in the Ruhr and Saar areas would also have been forced to negotiate collective agreements with workers as well, in conjunction with smaller firms in these industries. The VDMA, which was composed of firms with higher skilled workers, also opposed introduction of mandatory collective bargaining (Saul 1974, 62, note 67). In fact the ZDI observed the higher influence of workers within the VDMA after some member firms negotiated agreements, and used that as evidence against further provisions (Saul

76 This was accepted on February1, 1908 by a vote of 213 to 67. See Saul 1974, 63.

134

1974, 62). The opposition of the VDMA and similar industries that had higher skilled workers demonstrates that even in industries composed largely of higher skilled workers, there was opposition among firms of collaborative strategies. Both associations argued that proposed resolution would also have undermined firms‘ abilities to fire workers for union associations. Thus other employer associations had incentives to oppose the adoption of collective bargaining in related industries. This is indicative of a pattern of behavior where employer associations composed of larger firms made efforts to prevent the extension of collaborative strategies in other industries.

Opposition to these proposals regarding collective bargaining existed outside the economically powerful Ruhr area and heavy industry. In Bavaria, for example, the

Bavarian Industry Association (Bayerischen Industriellen-Verband, BIV) expressed opposition. The BIV counsel Alfred Kuhlo, speaking for the association‘s member firms, referred to collective agreements as a source of ―eternal discord,‖ and that firms that signed collective agreements increased worker agitation and interest in unions; firms bound by collective agreements were under the ―constant control‖ of trade unions (Saul 1974, 63).77 In 1905, both the BIV and ZDI had lobbied against the proposal of the Bavarian Prime Minister to legalize and enforce collective bargaining.

Additionally, the Association of Bavarian Metal Industries (Verband Bayerischer

Metallindustrieller, VBM) publicly argued that these proposals would only lead to the

77 These appear in Saul 1974, notes 70-71. The original source is a 1913 letter from Kuhlo.

135

―conquest‖ of trade unions and successfully demanded a withdrawal of the order (Saul

1974, 63).78

In June 1908, representatives of the Prussian and national interior ministries expressed opposition to legislative interference in the independence of firms. They also noted that such an attempt to regulate working conditions in the maritime industry could easily spread to others (as the industry associations had argued). Executive authorities in the Prussian and national bureaucracies ended up pulling their support for such measures, due partially to industry pressure. Still, despite their success, they complained of the state‘s inability to suppress increasing demands from workers.

Section 3: The Creation of Peak Level Employer Associations

Section 2 documented the advent of regional trade associations that shifted to collective repression institutions in response to the threat posed by workers. These trade associations were composed of individual firms that throughout the late nineteenth century had initially acted individually. In this section I demonstrate with more specific evidence that collective repressive strategies emerged as a result of the increased threat posed by workers, and the lower costs of adopting a collective response. The major industrial and regional associations described above, particularly in coal, iron, and steel, had formalized relations among many of the largest firms in terms of cartelization and lobbying mechanisms. But these did not coordinate formally with respect to industrial relations until labor strife and the electoral threat posed by the SPD and the redistributive threat posed by social democratic labor unions

78 See Steller (1906). At this time, there were some stipulations of collective bargaining in Hessen and Würrtemberg that only affected the printing and construction industries, and were less of a concern to manufacturing industries elsewhere.

136 increased in the early twentieth century. This latter threat was punctuated by the

Crimmitschau strike discussed below.79 As discussed in section one above, in many of the smaller handicraft traditional sectors, associations between firms had formed earlier, but the industrial associations in the sectors that were most important to the

Germany economy started to address industrial relations issues and demands from workers in the early twentieth century. Despite the formation of powerful trade associations in the more important industries, some industrialists, though they were accustomed to negotiating cartel quotas with each other, expressed reluctance in paying costs of joining formal employer associations. These costs included not just annual dues and the commitment to resources in the event of a lockout that other firms might experience, but also of course the infringement upon autonomy. But as the threat from industrial conflict increased in the early twentieth century, firms chose to respond with collective repression.80

3.1 The Crimmitschau conflict

The Crimmitschau strike of August 1903 was an incident of industrial conflict that spurred the development of national-level associations. The reaction of major firms and regional and industrial employer associations demonstrates the importance of the threat posed by workers in inducing national-level repressive strategies as a response to labor activity and demands. The failure of local and regional-level employer associations to address threats from Crimmitschau, as well as the backing of

79 Spencer (1984) writes, ―…delayed formation of employer organizations in heavy industry reflected both the later development of labor unions in those industries as compared to other types of production and the great resources of the large corporations, which long sustained the illusion that they could regulate relations with their employees on their own. 80 Spencer writes, ―…Events of 1904-05 ended that illusion by dramatizing the growing number and importance of outside influences on labor-management relations‖ (Spencer 1984, 104).

137 the local union by the general trade union and the SPD, led to a desire to form more peak organizations that could pool resources for common lockout funds and networking of employment agencies. Regional-level associations took advantage of the salient threat posed by union mobilization and SPD support of such actions to organize more institutionalized collective strategies.

Textile workers in Crimmitschau had unsuccessfully demanded a ten-hour workday since 1899. After a set of failed negotiations, and in reaction to an anticipated lockout, 7,800 workers across five factories struck on August 22, 1903.

(Interestingly, textile manufacturers in the Rhenisch-Westphalian region had previously argued that a ten-hour workday was likely the maximum amount the female workers in the textile industry could bear, and thus were not immediately sympathetic to the refusal of the Saxon manufacturers to negotiate). Soon after the strike, the Union of Textile Manufacturers of Crimmitschau exhausted its funds to support the striking workers, and its leadership turned towards the free trade unions for financial assistance. The executive committee of the free trade unions initially criticized Karl Hübsch of the Saxon textile union for drawing on central treasures and for diverting attention from other industrial conflicts (Saul 1974, note 66).

Still, the general trade unions assisted the local association, and the SPD press advertised for donations to contribute to the cause. The initial contribution was quite successful, bringing in over one million marks and creating a wave of worker sympathy. The textile manufacturers and existing employer associations in other industries viewed the Crimmitschau conflict, due to the intervention of the SPD and general trade unions, as ideologically elevated from an industrial conflict to a struggle

138 for political power and the prospect of social democracy. It became, according to a

ZDI circular, a question of whether firms would run their own businesses or be subordinate to the influence of the SPD (VdESI 1903).

The financial support by the general trade unions and the SPD for the local textile union outpaced that of their counterpart employers. Local employers in

Crimmitschau responded by using strikebreakers.81 Crimmitschau textile manufacturers asked the Textile Association of Saxony for assistance for funds to support member firms during the strike, but such funds were insufficient in supporting the struck firms. The employer association of the Saxon textile industry

(Arbeitgeberverband der Sächsischen Textile Industrie, ASTI) exhausted its immediate funds by December 1903. The efforts of this local association were financially inadequate (Krüger 1926, 320). In response to the crisis, the association called a meeting of the German textile manufacturers from the region in Kottbus to organize broader support for the firms in Crimmitschau (DAGZ 1903, 2).

Representatives of firms and employer associations from 25 cities decided unanimously upon a levy of two percent of their firms‘ weekly wages to assist

Crimmitschau firms for the duration of the strike; the collection and management of funds went along with an effort to organize a broader association of textile firms.

Hermann Vogel, the chairman of the Saxon textiles employers‘ association, warned that a social democratic victory at Crimmitschau would constitute a major defeat for

German industry (DAGZ 1903, 2).

81 Spencer 1984, 103.

139

The ZDI saw this opportunity to expand its influence, by leveraging the national attention and importance of this local conflict. It saw also the opportunity for the organization to expand intervention in social policy, and to expand its philosophy of ―protection‖ of those who were ―willing to work,‖ against the ―terrorism‖ of trade unions (Saul 1974, 422). The ZDI set a condition that its support for the firms would have to lead to an organization that would punish union agitators and to prevent their reprisals after the ending of the conflict (Saul 1974, note 84; Niederschrift der ZDI,

1904). Importantly, the ZDI at this point was rallying against social democracy as much as union penetration of plants. Representatives of the VdESI supported these

ZDI efforts and argued that trade unions needed to be smashed to pop the ―charm‖ of social democratic teachings. They viewed the defeat of these textiles workers as partial payback for the Social Democratic electoral victories of 1903 (Saul 1974, note

90). Vogel expressed a concern about a chain reaction from the Crimmitschau strike.

The external support of unions aroused fears from regional employer associations that the conflict would become radicalized and spread across other industries.

One day after the Kottbus meeting called by textile manufacturers, on

December 16, 1903, the ZDI board of directors convened, and following Bueck‘s request, called on member associations for financial assistance to the firms of

Crimmitschau and began discussions of a formal employer association that would be the largest in Germany. The ZDI appealed for support of the struck manufacturers and firms hurt by sympathy strikes, by means of voluntary contributions to (Krüger 1926,

320). The ZDI director, Henry Axel Bueck, toured Germany to obtain such support; the efforts of the ZDI ultimately helped end the strike in the manufacturers‘ favor.

140

The ZDI before the Crimmitschau defeat had not called on member firms or associations to financially contribute to assistance to firms to end labor disputes; a resolution passed by the board of directors on December 17 officially approved of such assistance. Bueck engaged in personal negotiations with member firms to convince them of the worth of contributing, and that it was in individual firm‘s interests to not only make sure Crimmitschau ended in the employers‘ favor, but also that further attempts would be deterred; free enterprise versus control of social democratic unions was at stake, given how thoroughly outspent the local employer association had been relative to the trade union contribution (Mallmann 1990). On

January 19, 1904 the strike ended in Crimmitschau, with the textile employers victorious.

Formally, the executive committee of the Northwest Group of the VdESI, proposed on January 4, a levy of 50 pfennigs per worker per month from member associations of the ZDI. The stated goal was to preserve firm autonomy from encroachment of unions (Saul 1974, Winschuh 1932). The members in the Saar region paid the same amount, while member firms in Cologne paid a levy of 0.5 percent of total wages for a Crimmitschau assistance fund. Other initial contributors to this fund included the Rhenish-Westfalian mining industry, the Bavarian industrial federation, large chemical companies, and a banking organization. The original goal was to supply 50,000 Marks per week, and the ZDI succeeded in raising 170,000

Marks (VMB 1904). The associations that formed in response to the strike were

―children of emergency‖ (Saul 1974, 100).

3.2 Repressive collective action: evolution of a peak association

141

3.2.1 Redistributive threat and formation

The board of the ZDI met on January 13, 1904, to establish the General

Administration of German Employers‘ Associations (Hauptstelle Deutscher

Arbeitgeberverbände, HA). In this sub-section, I discuss how the formation of the HA confirms the prediction of the theory about the importance of the redistributive threat of workers; what repressive strategies the HA pursued; and how the variable of industrial heterogeneity produced divisions within the organization that reduced the effectiveness of these strategies. The HA was the first peak level employer association, and its object was to ―combat unreasonable demands by workers‖ (Krüger

1926, 321).82 The founders of the HA discussed the necessity of uniting existing employer associations and apparatus of the ZDI to respond to the Crimmitschau strike; an organization was needed to counter the ―terrorism‖ of the social democratic movement, as evidenced by the strike. There was also discussion of maintaining the

HA‘s independence from the ZDI, despite having overlapping membership, and how to balance the autonomy of regional members with the central lockout fund.

It should be noted that before the HA was officially founded, representatives from the ZDI, Paul Heckmann, Hermann Vogel, Hermann Blohm, and Johannes

Menck, who had been on the organizing committees and were prominent in other regional employer associations, tried to develop a truly pan-industrial peak organization that would have obligated member firms to contribute more than the final version of the HA charter would end up doing. They represented the GDM, AGSTI

(Arbeitgeberverband der Sachsischen Textilindustrie, and the Employer Association

82 See also Spencer 1984, 103.

142 of Hamburg-Altona (AHA). These associations had been attached to the ZDI, but the

ZDI leadership outside of Bueck was initially skeptical of another association that would challenge its economic authority (DAGZ 1904; Rundeschreiben des ZDI 1904).

Leaders of the employer associations of Hamburg-Altona and the metal industry association asked the Secretary General of the Bavarian Industry Association, Alfred

Kuhlo, to draft a program for a peak German employer federation. But Kuhlo‘s efforts, as well as those of Heckman failed because other regional employer associations discussed above did not want to give up their autonomy or change their organizational structure (Saul 1974, 105, 109; DAGZ 1904). Kuhlo explained the establishment of a general German employer federation for one ―urgent necessity‖

(Saul 1974, note 119; Jahrbuch des BIV 1904) but a pan-industry association that obligated members to comply was not accomplished. Menck had been a member of the National Liberal party and legislator. He too had been concerned with the threat posed by the SPD and the fact that the strength of firms would be broken by legislation.83

Despite this setback, ZDI representatives still moved forward to create a national-level association that would increase resource sharing among existing employer associations and possibly found new associations in industries. Bueck and the ZDI intended for this trade association to create a specific employer organization to nationalize and further institutionalize the anti-union policies that already existed in various member regional and industrial organizations. To this end, the first unofficial

83 Menck left this wing of the NL party and became a member of the Free Conservative party (Saul 1974, 52). This party advocated the further restriction of the franchise. In contrast, the NL party advocated preserving the franchise.

143 meetings of the HA in January 1904 were attended by approximately eighty individuals, representing approximately 60 different major employer organizations.

The largest manufacturing areas and industries were either directly represented or had institutional affiliation, including the large Association of the German Metal Industry

(Gesamtverband Deutscher Metallindustrieller, GDM), the German Brown Coal

Industry Association (Deutscher Brauhkohlen Industrieverein, DBI), the Employer

Association of Hamburg-Altona (Arbeitgeberverband Hamburg-Altona), and the aforementioned LNV and VDMA. There were also representatives from organizations in Saxony (Verband der Arbeitgeber der Saechsischen Textilindustrie) and Bavaria

(Bayerischer Industriellenverband). The main representatives were those representing the previously formed regional employer associations in heavy industry and manufacturing, as well as employer associations in the textile industry (Mitglieder

Verzeichnis Hauptstelle Deutscher Arbeitgeberverband, 1904).84 Representatives from the BDI and VWICI (the chemical firms‘ association) also took part in the

January 12, 1904 meeting.

Thirty representatives of employer associations, predominantly from the ZDI, met on February 15, 1904 to decide the structure and goals of the HA. A commission of six individuals, who had been selected in January, drafted the statutes. The

84 See letters on opening meetings of the HA, 1904. Despite opposition from BDI members, they wanted to initially present a unified front (109). Representatives of the BDI in principle accepted the philosophy of the HA, because they wanted to present a unified front (Saul 1974, 109). The BDI did not put much effort into stopping the creation of the HA, although it condemn its ―egoistic propaganda‖ that came at the expense of the efforts and autonomy of existing federations. The GDM and largest textile associations were not formally part of the HA. The GDM and Employer Association of the German Textile Industry (Arbeitgeberverband der deutschen Textilindustrie, ADT) had 489,300 and 310,000 member workers, respectively. The largest regional employer association was the Employer Association of Unterelbe (Arbeitgeberverband Unterelbe) with 11,500 member firms employing 153,000 workers. Eventually the GDM joined the HA in 1904, despite its skepticism that the HA was too attached to the leadership of the ZDI (Saul 1974, note 148).

144 committee representatives were: Vopelius and Leidig, representing the ZDI; Beumer who represented the Rhenisch-Westphalian iron industry organizations; Menck who represented the GDM and AHA; Emil Stark who represented the Saxon textile industry; and Wilhelm Wendlandt who represented the other national lobbying association that represented higher end manufacturing goods, the Association of

German Industries (Bund der Deutschen Industrie, BDI). All except for Wendlant were involved with the governing bodies of the ZDI (Saul 1974). The sub-committee debated about whether the resulting employer association would be bound to recognize the ZDI as a veto player in its decisions; Wendlandt and Menck objected to a provision that would allow for ―statutory influence‖ of the ZDI executive committee

(Jahrbuch des BDI 1903/04). But the ZDI leadership on March 18, 1904 refused to consider this provision, and in fact ruled that the employer association would be independent and that the ZDI would only intervene if required. Effectively, then, employer associational autonomy for the new organization was written in its founding statutes.

The HA was officially created on April 12, 1904. Even though it had formal independence from the ZDI, ten of the twenty-two committee members of the HDA were selected by the delegate conference of the CVDI. These clauses could be changed only with agreement of the delegate conference of the ZDI. But, the two organizations maintained separate organizing lists and administrative bodies. The membership in the HA was thus not attached to the membership in the ZDI.

3.2.2 Purpose and activities: repression

145

The records of the first meetings of the HA, as well as its regional constituent organizations, emphasize the importance of the perceived increase in strikes and rising success of the SPD. The threat posed by the unions and SPD support in a massive

1899 mining strike and the 1903 Crimmitschau strike prompted the ZDI leadership to establish organizations that would allow employers to assist each other during strikes and to engage in pre-emptive lockouts when necessary (Bueck 1997; Leckebusch

1966).85 In a meeting of the board of directors of the ZDI, Bueck lamented the unwillingness of some employer associations to join the HA and referred to the

Crimmitschau strike as a reminder of the importance of collective strategies

(Niederschrift der ZDI, 1908). The by-laws of the HA indicate that the purpose of the organization was to unite member associations and firms against threats posed by workers, and indicated a common purpose with a lockout fund against workers and strikes (HA 1904; Leckebusch 1966).

The founding charter of the HA from May 1904 demonstrates the basic importance of a full response to union mobilization as the rasion d’etre of the administrative structures of the new peak federation. Its stated reason for existence was to protect firms against workers‘ strikes and boycotts and to assist firms in such events. Much of the charter deals with the authority of the executive committee regarding actions to repel union threats. In particular, the founding documents discuss the importance of connecting existing employment agencies that would share information on workers; the criteria by which the executive committee would accept

85 The HA‘s main goal was to have adequate funds to respond to strikes and to sometimes preemptively engage in industry-wide lockouts. These organizations formed in direct response to perceived increasing threats posed by workers.

146 members; and the authority of the committee to decide whether assistance to member associations that either were suffering strikes or wanted to initiate lockouts was warranted. Member firms and associations were obligated to comply with executive committee rulings. For example, this obligation is articulated in a provision that required member associations to comply with committee rulings about not accepting workers who were locked out and seeking new employment, for example.86 The main instruments of the newly formed repressive peak federation were to centralize lockout funds among pre-existing member associations and to create institutional linkages among such associations (HA 1904).87 The HA confined itself to the organization and grouping of industrial associations, regarding the ―unification of the great trade associations as the keystone of their power‖ (Saul 1974, 322).

The HA had at its founding had two national-level federations, 67 district federations, and 129 local federations. Membership increased to 6,664 companies with 1,092,789 workers in 1909. These federations had as members in total 6,144 member firms, covering 988,142 workers. The average member firm had on average

156 workers.88 Maintaining such membership was an important priority of the organizational leadership; there was concern that membership had reached a ―steady state‖ and HA leaders were concerned with keeping pace with unionization efforts— which the HA had tried to slow down. Vogel expressed concerns about stagnating

86 The charter also described voting rules for the member association. Each member association automatically was credited one vote; member associations received two votes if there were at least 10,000 member workers employed by member firms of the association. Beyond the 10,000 worker benchmark, each additional 1,000 workers translated into an additional vote. 87 The initial contribution rate was 50 Pfennigs per worker; the ZDI ended up paying 170,000 Marks to start the association. 88 There was an increase in 1,700 members and 150,000 workers due to the entry of the German employer federation for the construction industry.

147 membership not keeping up with the growing labor force (Saul 1974). As discussed below, the merger of the HA with a counterpart peak association did address concerns about membership.

The peak organization also accelerated the development of the employer associations and provided support for existing regional and industry associations. For example, after Bueck‘s efforts, the German textile manufacturers formed their own industrial association in January 1904, called the German Textile Employers‘

Association (Arbeitgeberverband der deutschen Textilindustrie) (Krüger 1926, 321).

The peak organization was also supportive of heavy industry in the west, paper, mining, and glass. About one-eighth of its funds went towards this purpose of a general lockout fund (HA 1904). Another major purpose of the organization was the diffusion of blacklists and information on workers, as well as use of labor exchanges to prevent hiring workers who were striking or who had been locked out; the organization also helped with assistance of legal fees for firms (Saul 1974, 446). The organization was also particularly active in the areas surrounding Aachen, the Saar area, lignite mining of the province Saxony, and textile employers' associations in

South Germany, Elsass Lorraine, and Silesia. By contrast, the other prominent peak level association, discussed below, was more active in central and northern Germany and in the metallurgy industry of South Germany. In eastern Prussia, employer associations overall were less active, due possibly to lower levels of industrialization.

The HA, as the ZDI had argued before, contended that employer associations needed to protect the right of firms to hire and fire, and were opposed to legislative efforts of the NL party and SPD to increase collective bargaining rights. These

148 formerly individual level strategies at the firm level were aggregated and pooled at the national level. The HA supported firms during the 1905 strike of Ruhr miners, and argued that the successful intervention of employer associations and the government demonstrated the importance of such organizations to respond to the threat posed by unions and the SPD (Bueck and Leidig 1905).89

The HA was a success in establishing a true, expansive employer association and coordinating efforts of existing regional employer associations. As discussed above, the existing regional employer associations in the late 1800s had concerns about obligations of membership and encroachment of managerial authority that a peak association would have. But the incidents of industrial conflict helped moderate these concerns; such associations were willing to cede some autonomy. The majority of heavy industry firms and their related manufacturing consuming firms of Rhineland

Westphalia had overcome in 1904 their old reluctance of employer associations that intervened in the regulation of the employer-employee relationships. Much orientation toward the collective solution had occurred. These new collective strategies helped complement, or share the costs of, individual strategies.

3.2.3 How industrial heterogeneity undermined collective action

89 Members of the HA resented pressure from the government on business leaders to have a conciliatory stance in 1905 strike. For example, representatives of the Center party encouraged Chancellor Bülow to intervene on behalf of the miners; he was concerned that mining strikes would undermine government efforts regarding military preparation (Spencer 1984, 100). Most government officials expressed disapproval of requests for military action from employers. The strike lasted from January 14 to February 9, and involved 73 percent of all Ruhr miners. On March 8, the Prussian government placed before the Landtag a proposed mining law revision—miners gained some demands, creation of worker committees, abolition of Wagennullen, and other improvements to their working conditions. Ruhr employers of course feel betrayed, though not greatly inconvenienced—they thought they should have ―won‖ against these demands and that the state should have stayed out (Spencer 1984, 101). From the HA‘ view, the main lesson of 1905 was that total intransigence against workers demands and total employer domination over workers was impossible; some modification of strategy needed (Spencer 1984, 102).

149

Despite the initial unity of the most important large firms and regional associations as a response to the threat posed by social democratic unions and the party, inter-industry differences in the costs of repression did undermine unity. The

HA was more effective at facilitating information sharing and centralizing propaganda and advertising efforts; it also was better at forming new associations, and assisting specific industries in need of assistance than in creating durable inter-industrial institutions. These divisions had consequences even though member firms and associations shared the same general goals. Even though they had lower costs of repression, these differences meant those with lower costs still had incentives not to subsidize repression that would disproportionately favor other sectors. Inter- association negotiations within the peak association confirm one of the predictions of the theory: when the costs of joining the association are higher than the costs of repressing individually, firms will not join a collective association.

This prediction is evidenced in the problem of maintaining membership within the HA and in the unwillingness of some of the member associations to subsidize the general lockout fund proposed in the organization‘s initial charter. For example the

Essen mining association, and nascent chemicals industry associations resigned from the HA. In 1908 the Ruhr mining industry founded its own organization, the

Zechenverband (which did not belong to the HA formally, and was not obligated to contribute to lockout funds, but did support its efforts). As discussed above, the

Zechenverband came into existence after the mining worker strike of 1905. In a

February 1904 letter to Bueck a representative of a chemical association noted that a need for an employer organization did not exist in the chemical industry, since large

150 firms were sufficiently able to restrict unionization. The representative stated, ―The union of the employers would presumably drive also those elements of the workers who do not currently belong to social democracy into its arms‖ (Saul 1974, 126). As discussed above, the large chemical firms had sufficient measures to deter unionization, such as firm level unions and ―work associations‖ which were designed to keep workers from joining social democratic unions. Bueck mentioned his regret in being unable to recruit Duisberg‘s organization to join the HA (Bueck 1908).

The main cleavage was that heavy industry associations were reluctant to subsidize textile industries, as the textile industry was one of the largest beneficiaries of the HA. Its firms employed around 285,000 followed workers. The Saar industry and iron industries in Rhineland and Westphalia did not contribute to mass lockout funds. They used firm-level unions and associations to prevent workers from joining social democratic trade unions. The HA supported such efforts, though it is unclear whether they contributed to such efforts financially. Instead, its main efforts were actually most evident in assisting the textile industry. The industries in Rheinland-

Westphalia and Saar area actually demanded that they only have voluntary contributions, and wanted to only compensate their own member firms. They did not look kindly upon the appeals to ―charity‖ from the comparatively weaker textile industry, and some viewed it as a ―disgrace‖ (HA 1907). This demonstrates heterogeneity in firm interests and the difficulty in sustaining a collective strategy, even though both sets of industries—heavy industry and textiles—were against social democratic unions. Heavy industry associations were better able to withstand the costs of work stoppages as well and thus had less inclination to heavily subsidize

151 weaker associations. Still, these regional associations in heavy industry did contribute to assist the textile industry, but chafed at long-term obligations to the central lockout fund. Vogel and Stark, as powerful representatives within the HA and the textile industry, managed to establish a strike fund (HA 1907). Although the regional heavy industry associations initially rejected the demands of contributions of 250 per thousand wages in Marks, they did eventually make some partial subsidies, but simply ignored obligations for a full central fund (HA 1907).

3.3: ―Mixed strategies‖? The formation and activities of the VA

The above discussion of the advent, operation, and cleavages of the HA demonstrates that the formation of this peak level association is consistent with the theoretical predictions of chapter 3. The HA had a membership of firms and employer associations which shared similar goals countering union mobilization and the SPD‘s political program, but the organization still faced inter-associational and inter-sector distributional issues. Even though member firms in this association had overall lower costs of repression, there were still distributional conflicts about subsidizing collective strategies. In this section I show how this logic also applies to the emergence of the other peak association, the Association of German Employer Organizations (Verein deutscher Arbeitgeberverbände, VA), which had a much higher variance in the costs of repression among its members firms. Some of its members at the time of the formation of the association had already concluded collective bargaining agreements with unions because their costs of repression were much higher. This meant higher variation in the kinds of strategies that member associations of the VA adopted, and an overall more mixed approach regarding demands from workers.

152

The origins of the counterpart employer association come from the longstanding antipathy of handicraft and finished-product industries towards the alliance of the major semi-manufactured goods and raw material industries, as they were punished most by cartel arrangements and the passing of such arrangements on to their input costs (Nussbaum 1966; Ullmann 1976). Partially out of concern with the institutionalized representation of large manufacturing and raw-materials firms, the main lobbying association that represented medium-sized finished-product manufacturing interests the Federation of Manufacturers (Bund der Deutschen

Industrie, BDI, founded in 1895) established its own coordinated employer association. The BDI had served as a competing lobbying association and

―counterpart‖ to the ZDI. The BDI‘s newly founded employer association had one goal of ―harmonizing‖ pre-existing collective agreements that had existed at the regional and sector level. The resulting association, the Association of German

Employer Organizations (Verein deutscher Arbeitgeberverbände, VA) formed in

1904. The organizational leadership was greatly influenced by the aforementioned

United Association of German Metal Industries (Gesamtverband deutscher metallindustrieller, GDM), the employer organization of finished metal products

(Spencer 1984, 103). GDM representatives were installed in the management offices of the VA until the VA and HA eventually merged in 1913, an event discussed below.90

The resulting peak association had seven national-level federations, 96 district federations and 1,092 local associations. The membership in these associations

90 The GDM in 1904 had 2,549 member firms that employed over 315,000 workers (Saul 1974, 186).

153 consisted of 32,500 member firms that employed 1.45 million workers in 1904. By

1909, the VA covered approximately 50,000 member enterprises employing 1,800,000 workers. In contrast to the HA, most member firms of the VA were small or medium- sized enterprises; the average member firm of the VA employed 32 workers

(compared to the average HA firm that employed 161 workers).91 The consequences of this different composition in membership are discussed below. Other important VA member associations were the AHA, the Bergisch regional employer association, and timber employer associations. After the VA‘s founding, individual smaller federations of the finished industry, bakers, stepped fitter, and construction industries, local employer associations of the chemical industry, as well as a number of mixed regional associations joined the VA (Saul 1974). Many influential industry associations, such as the GDM and AHA, were members of the VA but were not composed of craft- based industries.92

3.3.1 Repression within the VA

What differentiated the VA from its counterpart peak association was its higher degree of industrial heterogeneity among its members, as characterized by a higher variance in the cost of repression. This ―original sin‖ led to a wider diversity of strategies, in particular a set of mixed strategies of repression and collaboration within the organization. The pursuit of repressive strategies is emphasized in the charter of the VA, which allowed and advocated the use of repressive strategies, in particular the use of lockouts (Krüger 1926, 322). Even though some member firms of the VA had

91 Data are from the Statistiches Jahrbucher das Deutsche Reich, 1909. 92 The GDM and VA, both of which had member firms that signed agreements with workers, both objected to industry or association-wide bargains with workers Saul 1974, 59). The GDM was the critical organization with in the VA (it had 2,549 members with 316,210 workers) (112).

154 signed agreements with workers, the VA‘s stated policy was that it wanted to maintain the ―fundamental right of employers to choose their own workmen‖ (VA 1904). Thus the VA was not in principle completely collaborative; its stated policy was to support managerial rights, and not one of promoting co-determination or the protection of workers. Despite the existing of bargaining agreements within the VA, the founding charter of the VA demonstrates some important similarities to the HA in terms of the members‘ tolerance for repressive strategies.

The July 16, 1904 charter of the VA described one of the main purposes of the

VA was, like the HA, was to assist firms in cases of industrial conflict, and to prevent firms from suffering from unwarranted strikes and boycotts from workers. The VA, despite having member associations with collective bargaining agreements, had a stated purpose of assisting member associations and firms to combat industrial conflict, and in fact had a clause to assume the responsibilities of firms in this matter if warranted. This involved compensation of member firms and associations involved in work stoppages.

The HA and the VA also both supported repressive strategies that both established and connected existing employment offices, which would improve information-sharing services regarding the hiring of workers who had been locked out or had previously participated in strikes. The VA similarly called for connections between sector federations to share information and resources against strikes and boycotts. Like the HA, its declared goal was also protection of firms from unauthorized work stoppages, and to help firms combat them (VA 1904)). As a VA official described, small and medium sized enterprises would have to become much

155 tougher and give up previously ―sympathetic positions‖ if the strike compensation mechanism of the VA was to be effective. The VA also supported using lockouts in industries that were composed of craft-based firms, and acknowledged that such actions would be more difficult (DAGZ 1905). VA representatives thought that strike insurance and institutional support of lockouts would help contain strike epidemics, and praised strike insurance and lockout assistance as an ―instrument of the social peace‖ (Saul 1974, 447). Once the threat from unions increased, even employer associations firms formerly skeptical about strike insurance shifted their position to support it.

Some VA members did initially have concerns about the utility of strike insurance, as it could aggravate labor disputes, and potentially stir further union organization. As early as August 1903, employer associations that would join the VA though strike insurance would undermine relationships with workers. However, the

Crimmitschau conflict convinced them to change their minds, and in 1905, they recommended strike insurance as a suitable means for ―the containment of the strike epidemic‖ (Saul 1974, 447). Johannes Menck, the aforementioned head of the GDM, advocated the use of lockouts, though he argued they should be strategically used. He was concerned that the use of general lockouts could provoke government intervention or that their overuse would result in anti-lockout legislation that would undermine employer efforts. Heckman, who in late 1905 became chairman of the VA and GDM, opposed negotiations with workers and collective agreements, unless they were absolutely necessary. His view was also adopted by the large member firms in metal processing and manufacturing of the GDM. Heckman wanted to extend this principle

156 to prevent and weaken the accommodating, negotiating stances of the smaller and medium-sized firms in the handicraft associations (Saul 1974, 447). Though the VA charter specified a more de-centralized structure in terms of member associations‘ obligations (it did not establish a centralized lockout fund like the HA, despite the proclamation of similar goals in the charter), the executive committee still had to approve of assistance to member associations that wanted to hire workers who had been locked out. The executive committee had the authority to approve of assistance to member firms. In fact, the committee argued that assistance was only warranted if the strike was ruled by the committee to be unauthorized. The VA proclaimed in cases that it had the right to override decisions of member associations, and member associations were obligated to comply with VA regulations.93

The use of repressive strategies of the VA is also evidenced in its opposition to the discussed 1908 collective bargaining extensions in the shipping industry.

Consistent with the position of the ZDI, HA, and Bavarian metal and peak associations, the VA was also opposed these proposals; on October 29, 1908 the VA issued a statement against the resolution regarding collective bargaining. It stated that collective agreements are generally ―impracticable in the exporting industries‖ (Saul

1974, 64). The VA position was also not unusual among associations with smaller firms or higher skilled workers: other associations with member firms employing

93 In terms of membership dues, the VA structured contributions by number of employees. Each member association paid in contributions two marks per hundred employed workers. The differences between the HA and VA illustrate the different strategies vis-à-vis workers. Overall, the collective repressive organization was more coordinated and was better able to adopt a coherent strategy vis-à-vis its regional members. The HA‘s main instrument against unionization was a central treasury to fund pre-emptive lockouts and to support firms during strikes (Krueger, 322). The VA did not have such a treasury. The VA thought that employers should be indemnified against all strikes, and not merely those involving a fundamental principle; ―…a balance should be maintained between contributions and benefits, and insurance against strikes should be organized on a grand scale‖ (322)

157 higher skilled workers also opposed collective bargaining extensions. In March 1909 the VDMA also argued that collective bargaining expansion would increase the influence of trade unions and impinge upon firm flexibility to integrate wages and technological changes as they saw fit.

3.3.2 Collaboration within the VA

Despite the repressive strategies embedded in the charter of the VA, some member associations and firms had already concluded agreements with unions.

Because of this membership of industries that had in the late nineteenth century negotiated agreements with nascent craft-based unions, and the organization was from its inception forced to adopt a more flexible strategy from the start, as opposed to the

HA. As a result, complete repression was not a viable strategy. Some firms that were members of the VA‘s constituent employer associations had already concluded collective wage agreements with unions before the formation of this association, in areas where the costs of repression were too high (Geary 1989).94 Such members included firms in metallurgy, building, timber, which had trade union presence.

However, as the discussion in the above section demonstrates, VA leadership advocated use of repressive strategies to try and undo this collaboration within the organization.

These associations that were more collaborative in nature that eventually came to be represented in the VA had responded in the late 1890s to the pressures that earlier guilds placed on firms. Employer associations formed earlier in these

94 The VA membership consisted of smaller firms and craft-based, and relied mainly on inter-trade associations (322). (In terms of division of responsibilities between the HA and ZDI, the HA limited itself to defense against strikes; the ZDI handled social policy lobbying. The VA devoted some initial attention to social policy questions.

158 industries where guilds were powerful, but they were (like their successor organizations would become) formed in response to the redistributive threat posed by the more skilled workers in these industries. As discussed in section 2, initial employer associations were in the craft and small-finishing industries, which depended more on craft0based workers. An 1897 law made membership in these guilds compulsory (Thelen 2004). Since this legislation made certain kinds of labor intervention illegal, independent employer associations were formed in response to try and structure the labor market (Saul 1974, 99). This is further evidence that even in the industries dominated by smaller firms relying on craft skills, the cost of repression determined the kind of response of firms. For example, 44 percent of workers of bricklayers had won collective agreements by 1911. These firms were vulnerable to strike action due to seasonal factors and completion-date fines. As one building employer said in 1901, not to conclude agreements with workers was ―impractical‖ and ―obsolete‖ (Geary 1991, 146). But the reasons had to do with the high cost of the alternative strategy of repression.

These industries that were populated with smaller firms and had overall much lower degrees of industrial concentration were compelled to deal with unions.

Consistent with the theory in chapter 2, structural variables affecting collective action—in particular high numbers of small firms—left firms vulnerable to coordinated union demands. Large firms had lower costs of repression and so were less likely to have to sign agreements with workers. Their relative financial weakness benefited unions; therefore collective agreements emerged in giving union greatly increased negotiating power. Such agreements would not emerge in the more

159 economically important industries populated by large firms until 1918; this event is discussed in Chapter 6. The occurrence of collective agreements among industries populated by smaller firms and/or higher skilled labor force does not mean that this was the first preference of those firms. It more likely meant that such agreements were the least costly strategy for firms and that the firms‘ first preference was repression, which may have been too costly. (Geary 1991, 147)

The VA‘s tolerance of membership of these associations divided the associational leaders. Bueck viewed the issue of membership of firms with handicraft technologies of production to be the main reason why an initial merger between the

HA and VA was unacceptable, because those firms generally already recognized unions. He expressed the view that handicraft firms were a point of weakness in the battle against trade unions, as they were obviously vulnerable to union intervention and had a long history of dealing with unions. By contrast, Menck of the VA thought it was self-evidently in the interest of large-scale industry to strengthen these firms

―since their defeat indirectly also meant a defeat of industry‖ (Saul 1974, 158). Thus

Menck thought a short-term sacrifice of institutional ―recognition‖ of these temporarily collaborative associations was worth the benefit of trying to deploy the repressive apparatus of the VA against them. Because local employer associations were perceived to be important defenses against gains in social democratic unions,

Menck wanted to include employer associations composed of craftsmen and smaller manufacturing goods. VA officials argued that if large-scale employer associations from more important industries did not support them, then unions would likely be able to impose more conditions on other industries. This would be a threat to large-scale

160 industry, since skilled workers would begin demanding equivalent wages to other skilled workers in the smaller firms. An effective peak organization would have to exercise control over industries populated by smaller firms. The ideal was to make sure that such associations were ―doubly‖ protected by local and peak associations.

Thus even though there were sector differences in strategies in Germany due to differences in the cost of repression, even within the peak association that tolerated collaborative associations, the logic posited in chapter 3 operated twofold. Within the

VA, the craft based industries and other set of industries populated by smaller firms adopted collaborative strategies largely due to the impractical and higher costs of repression. Second, the VA charter and its institutional apparatus were designed to help counter the ―pre-existing‖ collaboration that existed in some of its member associations.

At the end of 1904, despite these differences, the HA and VA representatives opened discussions about merging, which would have meant exchange of member lists, and a common institutional and financial structure to counter strikes and boycotts

(Saul 1974, 160). This supports the idea that the handicraft firms that populated the

VA were ―co-opted‖ into a peak association. The co-existence of the HA and VA demonstrates the dominance of collective strategies. The modal strategy was a repressive one, though, as the HA was responsible ultimately for larger firms and more employees, and the VA promoted repression despite having some collective bargaining arrangements. This evidence contrasts with recent studies study of employer coordination of Herrigel (1998), Thelen (2004), and building on them,

Iversen, Cusack, and Soskice (2007, 2010), that focus on industries that were more

161 likely to be members of the VA. But a fuller picture shows that firms adopted a wider array of strategies. Moreover, the more politically important strategies were collective repressive organizations.

Section 4: Overcoming Differences and Further Collective Strategies

4.1 The cartel contract

At the time of the founding of these two peak associations, despite differences over recognition of collective agreements, there was already discussion of a merging of institutional apparatus into one association. The 1904 VA charter discussed the administration of cartel contracts of some member associations with counterparts in the HA. The charter explicitly discussed the potentially cooperative relations between the VA and HA, and pursuit of ―common purposes‖ of both peak organizations regarding industrial relations. These two employer associations co-existed from 1904 to1913. This division between the HA and VA reflects the split between the two major trade associations, the ZDI and BDI (Krüger 1926, 321), but their respective employer associations at the time of their founding laid down plans to further

―collectivize‖ strategies by creating one peak association.

On March 14, 1909, the HA signed a cartel agreement with the VA, but did not officially form a joint peak association until 1913. The agreement was signed by Max

Roetger for the HA and Paul Heckmann from the VA, and laid the foundation for a merger, to be established at an unspecified date, between the two peak associations.

The agreement formalized pre-existing repressive strategies of both associations. The basic tenets of the cartel agreement involved the creation of a central office, merging of membership lists, and centralized management of the employment exchange offices

162 and sharing of information regarding worker histories. The main reasons given for this cooperation were increased efficiency of managing collective strategies through one main office (Kartellvertrag zwischen ―Hauptstelle‖ und ―Verein‖, 1909).95 The resulting proto-peak organization greatly increased the total number of firms and workers covered by associations to 38,644 employers and 2,438,142 workers.

What would become the cartel contract was discussed at a meeting of the board of the directors of the ZDI in Heidelberg in September 1908. The position of the VA representatives at these negotiations (which Bueck and Rötger wanted to do more privately) demonstrates that despite common repressive preferences, there were concerns about disproportionate decision-making power being given to the HA. In particular, VA representatives expressed concern that smaller craft firms that constituted the VA membership would be consistently outvoted. The HA and VA were divided over issues of contribution rates; protection of craft industries, and access to strike defense funds (Tänzler 1929). The main issue was whether industries populated by smaller firms and craft firms would be protected. Menck, as discussed, thought they had to be protected, because defeats there could spread to other industries

(Saul 1974, 110). The other issue in these discussions was the levels of contributions that member associations of the VA; Menck wanted to ensure that any resulting peak federation would have some exceptions to prevent some member associations from paying full fees (Niederschrift der ZDI, 1908). This divergence in interests illustrates the importance of industrial heterogeneity in inhibiting collective action at the peak level.

95 The original documentation is found in Leckebusch, 1966.

163

Still, less than one year after this meeting, a cartel contract emerged that resolved some of these disagreements. The cartel agreement demonstrates that even despite differences in industrial composition of these two associations, the threat from organized labor was large enough that a peak association was deemed necessary as a response. Despite the disagreements over relative contributions, the industry representatives had argued in 1908 that unification between the two organizations was necessary to counter the expanding efforts of unions. Bueck observed that social democratic supported trade unions had accumulated 40 million marks, and that existing employer associations would have to pool resources to keep up with such funds.

The cartel agreement allowed for enforcing lockouts and specified the goal of the newly formed organization in punishing workers who initiated industrial conflict.

The central committee, as the committees for the existing peak associations had done, would rule on whether member associations would be allowed to hire locked out workers, though initial procedures for this were deferred to the member associations of the constituent peak associations. (In the text of the agreement, though, the difficult issues of working out shares of contributions to the centralized lockout fund and strike insurance from member associations were tabled for future discussion).

The cartel arrangement that laid the groundwork for the eventual merger between these two peak federations was viewed with suspicion and concern by unions, as evidenced by discussion of the agreement in the union publication Soziale Praxis.

An article published immediately after the cartel agreement in 1909 observed that the cartel agreement between the HA and VA would eliminate any rivalries between the

164 two central employer associations, and therefore it was important for unions to try and counter with their own associations. The author of the piece advocated union organization as a response to the cartel agreement (Sociale Praxis, 1909, 661).

4.2 Formation of the VDA

In 1913 these two peak associations merged to form the Federation of German

Employers Associations (Vereinigung der deutschen Arbeitgeberverbände, VDA).

This organization was the institutional predecessor of the contemporary German employer federation, the BDA. This organization, and its predecessor cartel contract, formed largely as a result from the increase in threat posed by unions as well as the lower costs of enforcement. There is little primary documentation on the actual mechanics and negotiations of the merger, but from existing documents, the 1913 agreement that formed the VDA merged out of ―friendly relations‖ and signified ―...a complete amalgamation of their internal and external structure‖ (Krüger 1926, 323).

This agreement meant that the HA accepted the ―principle of affiliation of handicrafts and attention to social questions, and on the part of the VA of the payment of heavier rates of subscription.‖ The HA was willing to trade less influence over lobbying on social policy issues, in exchange for the creation of a larger lockout fund. The ultimate success of the VDA is difficult because of the onset of World War I.

However, in chapter 6, I describe how the reconstitution of the VDA after the conflict supports the theory‘s prediction about the shift to collaborative strategies.

In the years preceding the organization of the VDA, Fritz Tänzler, who became its executive director, discussed its reasons for formation. Attempts at creating an organization like the VDA were made after the Crimmitschau strike, but failed due to

165 obscurely mentioned ―personal‖ and ―tactical‖ reasons. But the foundations for the

VDA had been laid in discussions between the VA and HA in 1904. Tänzler argued that both peak federations had common interests in forming a peak organization, and that there had been a longstanding effort to organize member associations to form a peak level association that could effectively execute lockouts. He argued that there was a mutual interest in repelling disputes, which were acted upon in the 1909 cartel agreement (Tänzler, cited in Sociale Praxis, 1913). Tänzler in fact argued that the two peak associations, the HA and VA, differed only in details. He emphasized that from the founding of the two peak associations after the Crimmitschau strike, discussions of joining the two ―organically‖ had taken place, which was finally able to happen in

1912. The draft produced by the cartel committee was submitted to the two federations for approval, which was received on April 4, 1912.

The newly formed VDA‘s constituent employer organizations oversaw 10,000 collective agreements covering close to 1.4 million workers; these agreements were legacies from the VA‘s institutional structure. Key leaders from within the VDA and its constituent organizations would go on to play an important role in the influential

1918 Stinnes-Legien agreement, the first formal codification of a nation-wide agreement between much of organized capital and labor, discussed in Chapter 5

(Erdmann 1966). The association further institutionalized repressive strategies, but also adopted mix strategies of cooperation in sectors where the costs of repression were too high. Still, with the merger, the HA essentially gained more resources to help execute its own repressive strategies.

4.2.1 Explaining institutional merger

166

A strictly repressive association in the early twentieth century ended up helping to administer a giant institution that essentially oversaw a series of pre- existing collective agreements—a qualitative change from its previous position. What explains this shift in position of the HA, even before the chaos of World War I? The large increase in the electoral success of the SPD in January 1912 helps account for this shift to a collaborative strategy. The SPD increased the number of seats in the

Reichstag from 43 to 110 (out of a possible 397) between 1907 and 1912. While there had been discussions of a merger between these two associations since 1905, it was finally completed several months after the election. The initial reports of the VDA and final reports of the HA discuss the increasing electoral threat of the SPD between

1907 and 1912 constituted one reason for this merger: both the VA and HA thought it would be more useful to have a national bulwark institution against workers, but also an institution to manage agreements with them. A clear point of compromise was that the HA believed that the VA should contribute a higher share of lockout funds. Thus the HA successfully bargained to retain its instruments to enforce anti-labor policies, but now they served as basic policing procedures of previously existing collaborative strategies (VDA 1913).

The success and challenge of the labor party was one of the major causes, but not the only reason, for the shift in the HA‘s strategies. The meetings of the minutes demonstrate pre-occupation with increased efficiency of managing industrial relations through a joint organization, as well as increased access to the VA‘s own treasury.

But the basic shift in strategy—one where the HA tried to stamp out organized labor to a strategy of recognizing their existence—happened after a massive SPD electoral

167 victory. This incarnation of the VDA would soon face a much bigger threat from radicalized labor after WWI. A cursory glance at the mechanics of the post-WWI major collaborative strategies shows that much of the institutional apparatus for effective national employer coordination was laid down already before WWI; this apparatus survived emergency decrees during WWI and became folded into the administration of collective bargaining in the early 1920s.96

The formation of the VDA was the culmination of the ―…unity of German employers that they had sought for almost a decade‖ (Spencer 1984, 121). In its brief time of operation as a peace-time peak-level employer association, the VDA oversaw a mix of strategies of individual employer associations. The VDA still campaigned for strategies to restrict unionism, in particular legislation that would prohibit picketing from striking workers (Spencer 1984, 121, note 37). Additionally, for some areas, the VDA tried to restrict worker mobility and prevent workers from finding new employment should they be unsuccessful during a strike; this was a continuation of the

HA‘s policies (Krüger 1926, 326). The VDA also sought to prevent overlapping membership if firms when possible.97 Where collective agreements were not allowed, the VDA continued to execute the strategies of the HA. These included lockouts and provision of supplementary funds during work stoppages (either due to lockouts or strikes). They included also using employment offices which kept information on worker histories (particularly regarding political activity, so member firms could screen workers); and the legal defense of firms in the event of industrial conflict. As

96 See VDA report of 1913. 97 ―…an inter-trade association will refuse to accept an individual employer as a member if there exists in the neighbourhood or region a trade association for the industry concerned, but will invite him to become a member of that association‖ (Krüger, 326)

168 the HA had done before, the merger, the VDA also campaigned for legal prohibition of picketing.98

As with the 1909 cartel contract between the VA and the HA, union representatives expressed concern at the centralization of employer coordination.

They were particularly concerned because the HA leadership which pushed for and coordinated the unification had a previous history of course in repressing workers. As an article in Soziale Praxis expressed, the impact of this merger on working conditions was not to be underestimated (Soziale Praxis, 1913, 866). The specific concern expressed was with the new leadership direction of the VDA. Though the VA was acknowledged to have a more conciliatory stance towards workers, unions expressed concern that the influence of the HA meant the VDA would continue the former‘s repressive policies, despite the appointment of leadership who had previously been with the GDM. There was concern that the VDA would not simply represent the

VA‘s policies.

The VDA also created a common lockout fund pool, levied at 20 pfennigs per

1,000 marks in wages for each member association. Tänzler pointed out that the central fund would enable handicraft industries to deal with larger strikes for longer periods of time. The large industrial concerns, he mentioned, had interests in making sure even small handicraft firms were not subjected to strikes and union imposition.99

Another advantage of the unified organization was that it would expand employment

98 It is difficult to assess what direction the VDA would have taken in terms of repression, because its activities were interrupted by the transformation of the domestic economy into a war-time one and the domestic instability wrought by World War I and its end. The consequences of these events for firm strategies are discussed more fully in Chapter 6. 99 This contrasts with existing accounts which argues that handicraft industries had different interests from larger firms.

169 exchanges and offices, which had been before more of the purview of the HA. The

VDA, as Tänzler described, would substantially increase the resources to battle unions

(Tänzler, 1913, quoted in Soziale Praxis). He also emphasized the importance of moral suasion and propaganda efforts to win public opinion battles against workers.

The VDA would be an expansion of repressive strategies against workers.

The VDA also helped to institutionalize conflicts between different industries, in particular small and medium sized enterprises (SMEs) and larger enterprises, and

―smoothed out‖ disagreements over treatment of workers (Krüger 1926, 324). Note there were some areas, as discussed, where due to higher sector level costs of repression, collaboration was allowed. Under Tänzler‘s direction, the VDA ―…called upon its affiliated organisations to accede to the legitimate demands of labour‖

(Krüger 1926, 325). The VDA adopted a more conciliatory position towards workers and did not assist individual firms that locked out workers, if such behavior was viewed to be illegitimate.100

4.2.2 Managing heterogeneity

A key issue was whether the newly formed peak associations would support the industries populated by smaller firms and craft-based production. The Hamburg- based leadership wanted to incorporate those associations, because otherwise they would be ―too weak without organizational support. Echoing Menck‘s arguments about the VA, they argued that if the more economically powerful industries did not support the handicraft and the small firms, trade union victories were more likely. This

100 ―In cases where individual employers showed a tendency to reject claims considered legitimate by their colleagues and to impose less favourable conditions upon their workers, it [the VDA] refused to support them‖ (Krüger 1926, 325)

170 would, they argued, constitute an indirect defeat of the large-scale industry, since skilled workers would be able to use their collective agreements to force larger firms to comply with these agreements. The main concern was that skilled workers in the craft industries would be able to impose their improved wage and working conditions on other firms outside of the immediate coverage of these bargaining agreements (Saul

1974, 102). A peak organization that would provide strike reinsurance for craft and smaller firm industries would address this problem, by undermining collective agreements in the smaller industries. This contention too is supported by the aforementioned memoranda from the LNV, who argued that it was beneficial for the

HA to intervene on behalf of the printers‘ employer associations.

This compromise in strategy of the peak association was partially dictated by the structural reality of heterogeneity in sectors and regions in terms of demands from workers and wage conditions. The peak association‘s executive committee controlled a central fund, but the ―center of gravity‖ of the VDA was the local or regional association in terms of initial decision-making. This allowed for some de- centralization of wage management, which could not practically be done from Berlin.

The explicit policy of the VDA was to allow for inter-regional variation in wages, due to variation in local-level transport costs, prices, and working conditions. The mix of repressive and collaborative strategies helped to enforce these wage contracts. The

VDA, as with its predecessor organizations, pursued a collective strategy such that workers could not take advantage of inter-firm wage differentials by migrating.101

101 The internal organization and governance of these associations were adapted to the law on associations in the existing Civil Code. Membership to the VDA was determined by an executive committee (Krüger 1926, 327). The VDA made it difficult to exit the organization. If a firm wanted to

171

The VDA had difficulty legally enforcing its obligations. But there were administration rules to fine members for non-compliance and procedures for terminating membership. The VDA obligated member firms to deposit funds to the central committee to be used as a deposit in case the firm failed in its obligations.

(The VDA and its constituent regional associations tried to amend the German laws so that they would be allowed to sue firms for non-compliance, particularly for exiting the organization without committee authorization). The VDA in fact did expel members for non-compliance and effectively blacklisted firms, preventing them from receiving industrial conflict assistance from other firms.

There was also recognition of heterogeneity of interests within the VDA.

Union representatives discussed how the formation of the VDA would heighten workers‘ awareness of the need to unionize, and that unions did not have the luxury of divisions that prevented coordinated action. The official merger of the HA and VA induced a renewed call for trade union unification, as evidenced in the discussion in the Soziale Praxis article. However, there was recognition of the problem of overcoming some divisions among workers. A specific danger that was mentioned was the long-standing efforts of unification among employer associations in the metal manufacturing industries, which were successful despite differences among

resign from the VDA, it was only allowed to do so at the end of an administrative year; it was required to give two to three months notice (some regional associations insisted on six months notice). Firms were generally forbidden to resign during a strike. Firms paid membership dues in return for services, which included funds during a lockout or strike. Subscriptions were determined on the basis of wage sheets or workers employed in addition to flat entrance fees for all members. The voting power of individual firms in the VDA was determined by the number of employed workers and wages paid. The administration of the VDA was composed of the committee (Vorstand), council (Ausschuss), and the general meeting of members (Mitgliederversammlung). The general meeting occurred annually and elected the executive committee (327).

172 employers. Of particular concern was the citation by unions of discussion for combinations of employer confederations ―without consideration‖ for political and economic differences, as these federations had a joint goal in fighting social democratic unions.

The creation of peak level associations, and the proliferation employer coordination in the early twentieth century, was described as a ―new section in the history of class warfare‖ (Saul 1974, 447). But still, as the first section that opened this chapter described, such efforts had mixed results. Major lockouts, particularly those after the formation of the peak associations which took place in 1904 and 1905 did not reduce union membership. Trade unions increased their membership after

1904. The effect of bleeding union treasuries was partially offset by increasing desires for workers to join unions, as workers still were able to join because unions could provide for protection during unemployment. Still, employer associations could claim more influence over workers than trade unions. In 1910, about 35 percent of workers were employed by firms who had obligations to employer associations, as opposed to 22 percent who belonged to trade unions (and that estimate includes non social democratic trade unions.

After the failures of major industry lockouts to wipe out social democratic trade unions, employer associations turned increasingly to lobbying for legislation.

There were concerns too from employer associations that mass lockouts would become too costly; this was coupled with the fact that such attempts were not always successful. In chapter 5 I test some of the above implications more rigorously with data from this time period. In chapter 6 I show that the concern about whether the

173

VDA could continue its strategies was partially obviated by the advent of world war and economic chaos; a fundamental shift to collaboration was to come when repression was simply not a viable option.

174

Table 4-1: Characteristics of Employer Associations, 1911-1913 Industrial Number of Number Number Sector associations of of member member firms workers Agriculture 53 7,735 73,648 Mining 10 257 495,702 Minerals 133 4,075 218,161 Finished Metals & 343 14,970 862,888 Manufacturing Chemicals 4 91 23,227 Textiles 98 2,725 495,685 Paper 42 1,125 56,949 Leather Goods 69 2,778 19,900 Timber 252 7,803 80,934 Foodstuffs 154 12,918 189,589 Clothing 237 11,143 187,467 Cleaning 57 1,371 27,774 Construction 1,399 49,342 528,060 Polygraphics 130 5,223 74,431 Tradeables & 156 6,389 117,203 Transport Services 11 5,389 9,400 Arts 103 529 25,000 Mixed 144 14,598 800,933 Note: Cells report period averages of 1911-1913. Source: Leckbusch 1964 and Statisisches Jahrbücher das Deutsche Reich, 1911, 1912, 1913.

175

Table 4-1: Characteristics of Employer Associations, 1911-1913 Industrial Workers per Ratio of total Ratio of Sector association firms in sector total to member workers in firms in sector to association member workers in association Agriculture 1,413 4 2 Mining 49,570 19 2 Minerals 1,644 13 4 Finished Metals & 2,525 18 3 Manufacturing Chemicals 5,807 159 11 Textiles 5,076 52 2 Paper 1,361 19 4 Leather Goods 291 38 13 Timber 330 40 11 Foodstuffs 1,237 31 7 Clothing 787 66 9 Cleaning 495 101 12 Construction 379 4 3 Polygraphics 573 4 3 Tradeables & 752 204 25 Transport Services 771 767 913 Arts 243 71 5 Mixed 5,791 NA NA Note: Cells report period averages of 1911-1913. Source: Leckbusch 1964 and Statisisches Jahrbücher das Deutsche Reich, 1911, 1912, 1913.

176

Chapter 5: Testing Mechanisms and Observable

Implications from Imperial Germany

In Chapter 4 I demonstrated that during Imperial Germany, the rise of the

Social Democratic Party and trade unions posed a threat to firms, and they responded to this threat with individual repressive strategies and then transitioned to collective action. This latter activity took the form of important peak-level employer confederations that supported regional and industry-level associations. However, consistent with the hypotheses regarding collective action, differences among sectors in terms of costs of repression made management of such collective strategies difficult. The patterns of development of these associations, evidence regarding motivations for formation, and observable implications are all consistent with the theory.

In this chapter I use more fine-grained data from the same time period to test mechanisms and observable implications of the theory. I exploit sub-national variation in Germany, using different units of analysis, to test different mechanisms and predictions of the theory. The first empirical test uses data at the industrial sector level to test the impact of industrial heterogeneity on the degree of collective strategies. The second quantitative test assesses if industrial characteristics lead to differences in degree of collective action. This test focuses on whether institutional connections between associations matter for substantive collective action, in the form of strike insurance adoption. The third and final test uses new data on local-level adoption of repressive firm strategies. I use data at the Prussian voting district

177

(wahlkreis) level and new data on firm membership in key repressive employer institutions to show that the threat of politicized labor affects the adoption of repressive firm strategies against labor. Specifically, firms adopt collective repressive strategies against workers when there is a moderate change in the electoral threat posed by the SPD, but such collective action is only possible when low industrial heterogeneity allows for collective action.

Section 1: Explaining Strategies of Collective Action: Evidence at the Sector Level

I describe and present a simple test of the main hypotheses from Chapter 3 that links industrial heterogeneity with degree of collective strategies of firms. I use the sector-level data on employer associations from the years 1911-1913 discussed in

Chapter 4 and presented in Table 4-1 to test whether heterogeneous or fractionalized industries affect the degree of collective action. The yearbook data are of aggregate characteristics of the associations, coded by sector. The data described for each sector include: the total number of employer associations, total number of member firms in these associations, and total number of workers employed by such firms.

Unfortunately, data on individual number of firms and employees for each association are unavailable, and so only inferences from the sector level are available. In the next section I use a separate dataset that of a sample of national-level associations to test other observable implications of the theory. The relevant unit of analysis is the industrial sector (n = 17). Across all sectors for 1911-1913, the mean number of associations by sector is 113, with a standard deviation of 256. For example, the metal manufacturing industry in 1911 had a total of 294 associations; these associations covered 13,752 member firms, employing a total of nearly 800,000 workers. By

178 contrast, the textile sector had 95 associations with 2,616 member firms, employing nearly 500,000 workers.

I consider different outcome variables that capture the concept of degree of collective action at the sector level. To measure the extent of collective action by sector, I measure the fraction of workers in the labor force in each sector who are employed by firms in that are members of associations.102 This captures the degree of coverage of associations. The mean fraction of employer associational coverage of the labor force is 0.24, with a standard deviation from 0.19. The smallest fraction is near 0 (in the industry low skilled guest services), and the largest fraction is 0.64 in the agriculture industry. The fraction for the metal manufacturing industry is 0.39; for the large mining sector it is 0.55. I use data from the 1909 German Census (Statistik des

Deutschen Reichs) to obtain decomposition of the labor force by sector. This constitutes the denominator of the dependent variable; the numerator is number of workers covered by associations.

To construct the main independent variable of interest, the degree of industrial heterogeneity, I use the same data source to construct indices of fractionalization for each industry at the national level. The Census gives detailed decompositions of the number of firms and workers by sub-categories of industries for each of the main industries. For example, the main industrial category of chemical industries includes

17 sub-industries, for which data on the number of firms and workers in each of those

102 An alternative measures if the fraction of firms covered by association, but such a measure does not account for differences in firm size and amount of product or labor markets covered by associational activity. We can consider outcome variables that do not normalize by conditions of the sector such as: number of associations; number of workers; number of firms covered by the association; ratio of the number of workers per association; and ratio of number of firms per association), but these are less appropriate for reasons discussed in the text.

179

17 industries exist.103 For each sub-industry, I first calculate its share of total employment in the main industry. I use the share of employment in each sub-sector, as that gives a more accurate approximation of the market share of each industrial sub- sector.104 Using this data on employment share by sub-sector, I then calculate the

Herfindahl index of employment shares for each industrial sector (the inverse of the sum of squares for each sub-industry).

The average degree of fractionalization by employment across all sectors is

6.8, with variation 6.0. The least fractionalized industry besides agriculture was the low-skilled guest services industry with a measure of 1.8. The most fractionalized industry is one labeled ―trade and transportation‖ with a measure of 21.2. The industrial category of finished metal manufacturing has a high fractionalization of

19.0. By comparison, the mining industry has a fractionalization measure of 3.0.

I estimate a simple linear regression model where the outcome variable is the fraction of labor force in a sector that is covered by employer associations (the fraction of labor force that is employed by firms that are members of employer associations).

The main covariates are the number of associations in the sector, number of workers covered by associations, number of member firms that are member of the associations, and industrial fractionalization by employment (as calculated above). All sector-level covariates except for industrial fractionalization are logged. Consistent with the hypothesis from Chapter 3, I expect that higher degrees of fractionalization should

103 As discussed above, the data from German statistical yearbooks (Statistik Jahrbücher) contain data on 22 industrial sectors. To match the industrial categories from these two sources, I combined a few industrial categories from the Census categories. The metal manufacturing industry is a combination of metal manufacturing employment and machine instruments employed. 104 I also calculate each sub-industry‘s share of the total number of firms. The share of number of firms for each sub-industry is also instructive, but detailed data on actual market share of the firms in each sector do not exist.

180 lead to lower degrees of collective action. The results are displayed in Model 1 of

Table 5-1. Unless otherwise noted, all analyses are at the sector-level and for the year

1911, the year for which complete data for all major sectors are first reported. The use of 1911 data also allows for appropriate lags from the Census data from 1909. The main results do not change when period averages of 1911-1913 data are used; nor do the results change when a fractional logistic model to estimate proportions is used.

The result is consistent with the theory. Regarding the control variables, sectors that have more associations have a smaller fraction of the labor force covered under them. Sectors that have more absolute numbers of workers covered by associations have a higher proportion of workers covered. In terms of directly testing the theory, as hypothesized, higher degrees of fractionalization by employment reduce the associational coverage of the labor force; the coefficient is negative and statistically significant at p<.09. Substantively, moving from the 10th to 90th percentile of fractionalization in an industrial sector reduces labor force coverage from.28 to .08.

The negative effect of this change in fractionalization on employer coordination is equivalent to moving from the minimum number of associations in an industry to the median number of associations in an industry. While there is a downward trend, the result treat should be treated with caution because of the small sample size, as even with large changes in fractionalization, the 95 percent confidence intervals still overlap. The effect of fractionalization remains controlling for the number of workers in the industry.105

105 The coefficient remains negative but the standard errors increase when the number of firms in the sector is controlled for. But, given that we do not have data on fractionalization by firm‘s profits or market share, ex ante it is not clear why it should relate to the degree of employer coordination.

181

As an alternative measure of degree of employer coordination, I also consider an alternative measure of the dependent variable of the degree of collective action: the ratio of workers covered by employer associations to the number of associations in the sector. This ratio captures a rough density of associations, conditional on some labor force participation in associations. That is, it codes a hypothetical sector that has one association that covers 1,000 workers denser than a sector with two associations covering the same number of workers, and so on. The mean number of workers per association in the period 1911-13 was 4,400, with a standard deviation of 11,250. In the finished metals and machine industry, the average ratio was 2,500. The highest ratio was in the mining industry with nearly 50,000 workers per association. One of the lowest ratios was in the construction industry with about 400 workers per association.

Model 2 presents the results of estimating a different linear regression model where the outcome variable is this alternative measure of coordination. Again, industrial fractionalization, which is a proxy for heterogeneity within a sector, is negatively correlated with associational coverage. Moving from the minimum to the median degree of industrial fractionalization reduces the number of workers per association from about 6,000 to 3,500. As above, though, the 95 percent confidence intervals overlap; this is suggestive evidence that industrial fractionalization has an effect on degree of collective action at the sector level. This simple test confirms one of the predictions from the theory. Heterogeneity of firms makes collective action

182 more difficult, as evidenced in the lower presence of employer associations in this time period.106

It is careful to state the limits of what sector-level data can reveal about the degree of collective action within each sector. The employer association data at the sector level do not reveal the distribution of actual number of workers and firms per association. Consider for example an industry that has a high number of workers per association. It could be because one association has a very high number of workers, and the rest of the associations are limited in their membership and do not have many workers. Or, it could mean that there is an even distribution of workers per association. Similarly, using just the number of employer associations in sector would not tell us the degree of coverage of such associations. Both of the measures displayed in Table 5-1 are fairly blunt ways to measure collective action, but the negative correlation of industrial fractionalization confirms the importance of the structural side of the theory posed in Chapter 3.

Section 2: Explaining Strategies of Collective Action: Evidence from Associations

In this section I test another observable implication of the theory, as well as a substantive claim that arose from the qualitative evidence presented in the previous chapter. I examine whether institutional ties employer associations, which I argue is a proxy for a minimal degree of collective action, leads to substantive collective action.

That is, does being officially attached to an association actually bring any benefits? I proxy higher degree of collective action with an indicator for if the association

106A problem with the data is that we cannot precisely trace employer associations to specific locations. Thus there is an ecological inference problem is that specific industrial characteristics may vary by region or locality. Section 3 of this chapter uses more fine-grained local-level variation in data to test the hypothesis.

183 provides strike insurance. As discussed in the previous chapter, associations had to pay costs to provide for strike insurance; the adoption of such insurance implies sufficient concern with the threat from workers to pay for such insurance. In this section I test for the correlation between institutional ties that are ―on paper‖ and the actual enactment of the collective strategy.

To this end, I use another dataset on characteristics of a sample of major national-level employer associations to test the theory for different associations. The data also come from the German statistical yearbook of 1909. I use data on 73 major associations at the national level. The data include for each association the number of member firms, workers employed by those firms, the sector in which the association operates, and whether the association provides or plans on providing strike insurance for member firms. For example, the peak associations the Hauptstelle and Verein discussed extensively in the previous chapter are listed. In addition, there are other associations at individual sectors, based out of different cities.107 I also use the sector- level data discussed in section 1 to test whether sector-level factors affect characteristics of the associations.

Table 5-2 gives descriptive statistics of the characteristics these national-level associations, as classified by industrial sector. The table displays the number of employer associations by sector; the total sum of firms that are members of all the associations, and the sum of workers employed by such firms. Unsurprisingly, the metal manufacturing industry has the most number of major national-level employer associations (14), as well as the highest number of workers covered by employer

107 For example, the Verband der elektrotechnischen Installationsfirmen in Deutschland, Frankfurt a. Main is an example association in the dataset.

184 associations.108 Additionally, over 90 percent of these associations gathered data on member firms and workers.

As a proxy for degree of collective action, I consider which associations have provided or are planning to provide strike insurance. As discussed in the previous two chapters, the provision of strike insurance (in addition to managing of lists of workers) was a crucial provision as well as issue of contention in the formation of employer associations. The presence or discussion of such provisions is a proxy for a more coordinated body. This variable is binary (1 indicating the presence of strike insurance or such plans, 0 otherwise). Column 1 displays the number of associations in each sector that have provisions for strike insurance or are planning to adopt them.

Of the 73 national-level associations, 30 percent had such strike insurance plans.

Much of the activity is driven by the metal manufacturing industry, as 6 out of 14 associations fall in this category. Five out of twelve such associations exist in the raw materials (earth and stone) industry. Thus much of the associational activity and strike planning is driven by, but not exclusive to, these major industries. Of the associations in these two industries, 44 percent provided strike insurance, as opposed to 22 percent of associations in all other industries. Of the associations that are connected to another national-level employer association, 48 percent have strike insurance plans; whereas only 20 percent do for other non-connected associations. These initial differences are more systematically tested below.

108 Note that these sums only apply to associations that claimed national representation; they ignore coverage of regional and local associations.

185

As with the sector-level analysis in section 1, I consider the variables of the

(logged) number of member firms and workers, only now I use such data for the specific association as opposed to industrial sector. To test for the relationship between institutional connection and strike insurance provision, I consider the variable which captures whether the association in question has any institutional attachment to any other national-level association. This variable is a binary variable (1 indicating attachment, 0 if not). This variable tells us if some institutional connection exists between associations. I hypothesize that this variable as well should be positively correlated with adoption of strike insurance.

To systematically test this relationship, I estimate a logistic regression model where the dependent variable is adoption or consideration of strike insurance, controlling for the above association and sector-level characteristics. Regarding the main independent variable of interest, the connection to another national-level association, a simple difference of means of tests shows the initial importance of this variable. Associations with such an institutional connection constitute a higher proportion of associations that provide or consider strike insurance (.48 versus .20, p <

.01). The results are displayed in Table 5-3. The only consistent statistically significant predictor institutional attachment to another national-level association, although the binary indicator for mining and manufacturing approaches conventional levels (there results hold under other specifications). As Models 1-3 shows, the coefficient on the binary variable indicating institutional connection to another national association remains substantively large and statistically significant (p<.04), as does the coefficient for whether the association is in the metal manufacturing industry.

186

The predicted probability of having or planning strike insurance goes from 0.20 (0.05,

0.36) to 0.53 (0.31, 0.75) when an association has an institutional connection. This is a slightly larger effect than an association being in the manufacturing industry, where the change in probability is from 0.24 (.08, 0.41) to .49 (.24, .74). This indicates that types of collective action activities—connections to institutions and provision of strike insurance—are correlated. More importantly, it validates the claims discussed in

Chapter 4 about the importance of ―institutional connection‖ to peak associations: such ties imply a higher likelihood of adopting strike insurance. This provides some quantitative confirmation for the cases discussed in Chapter 4.

Turning to the rest of the covariates, neither the number of firms and workers in the association, nor the structural characteristics of the sector, predict an association‘s adoption of strike insurance. Neither the number of member firms or workers predicts the adoption of strike provisions. Similarly, the hypothesized structural conditions (number of member firms and workers in the industry, and the degree of industrial fractionalization) do not predict strike insurance adoption.

Regarding the latter set of variables, this lack of effect might be because of the mismatch between the units of analysis (sector versus associational variables).

Additionally, as discussed above, the degree of fractionalization might be correlated with the development within the industry. Metal manufacturing is coded as a highly fractionalized industry partially because of the high number of sub-industries that emerged due to the development of a high number of sub-occupations within the industry (as opposed to say textiles, even though the latter industry had many more small firms). Unfortunately, the rapid development in some industries meant that

187 more sub-sectors were measurable and classifiable, potentially overstating the actual degree of heterogeneity in an industry.

The importance of the binary variable of institutional connection to a national- level association raises the question: what explains such institutional connections? I estimate another logistic regression model where the national connection variable is the outcome with the same set of structural covariates. Table 5-4 displays the results.

As models 1-3 shows, the only structural variable that is statistically significant is that of number of firms: associations with lower number of firms are more likely to be connected to national employer associations. Moving from the 10th to the 90th percentile of firm membership reduces the probability of being connected to a national institution from .64 (.40, .88) to .20 (-.04, .44). This is consistent with the notion that lower numbers of firms make collective action and institutional connection easier; further, such associations might be more vulnerable to threats from unions and thus would benefit more from such a national connection.109 The conclusion is that structural conditions do matter, but in an indirect way: the associations that were most likely to have adopted strike insurance were those in the metal manufacturing industry and that possessed some institutional connection to another national-level association.

In turn, associations with smaller number of firms were most likely to have such national-level connections.

Section 3. Explaining Membership in Collective Repressive Associations

In this section I conduct another specific test of the theory by leveraging sub- national variation in the adoption of collective repressive strategies at the local level.

109 Ideally we would like a pre-1909 measure of ―labor threat‖ for each association.

188

To test the hypothesis that the redistribution threat from workers should affect firm strategies, I constructed an original dataset on adoption of collective repressive strategies at the voting district level in Prussia in 1910. I use membership in the

Hauptstelle Deutscher Arbeitgeberverbände (HA), the peak-level anti-labor employer association discussed in the previous chapter, as a proxy for a repressive strategy of firms (HA 1910). I also use data that allows me to directly test the variable of industrial heterogeneity on collective action. I find that low to medium threats from workers have a positive effect on the adoption of repressive firm strategies. Where the

SPD had more electoral support, firms had a greater incentive to respond by joining repressive employer organizations. This finding is consistent with the theory: firms have an incentive to adopt a collective repressive strategy if there is a threat, but only when that level is low to moderate. In short, when the workers posed a redistributive threat, firms have an incentive to repress and respond collectively, but such collective action needs to be facilitated by structural conditions to make it self-enforcing.

The relatively weak support of the SPD in Prussia during this time period provides a difficult test for the hypothesis because the overall level of threat posed by this party had not reached a level where the SPD could plausibly take a majority of the legislature or implement its preferred social policies. The 1907 election was seen as a relative loss for the SPD and a re-affirmation of some of Chancellor Bulow‘s domestic policies (Sperber 1997; Suval 1985). Thus testing for responses of firm strategies, as measured with 1910 data, to the 1907 election represents a hard test for the theory.

This empirical test leverages the dip in support of the SPD between 1903 and 1907 to provide the most stringent test of the theory. When comparing the behavior of firms

189 across voting districts, support for the SPD in 1907 can be coded as national-level moderate threat posed by labor.

3.1 Descriptive evidence from a major repressive association, German level evidence

Before I provide support for the above statements, I first examine descriptive evidence of repressive firm strategies across all of Germany. I gathered data on firm strategies at the district level using reconstructed lists of the 1910 membership list of the peak anti-labor employer organization, the Hauptstelle Deutscher

Arbeitgeberverbände (HA). As discussed in chapter 4, this organization had a stated anti-labor objective. The membership list was composed of two broad categories:

―mixed‖ organizations (which had organizations that covered multiple industries) as well as ―pure industrial‖ organizations (which had organizations that covered one industry but possibly multiple geographic areas). This data can be exploited because almost all member organizations include a geographic place of reference in their name, indicating their geographical area of service. An example of a member organization‘s name was the ―Employer Organization of the Textile Industry, Aachen‖

(Arbeitgeberverband der deutschen Textilindustrie, Aachen). None of the listed members from this year are individual firms. The membership of the HA in 1910 was composed of 154 employer organizations in the metal manufacturing, raw materials, glass, paper, and textile industries. Some member organizations were also formally listed as cartels.

Table 5-5 describes the composition of HA membership by general category and industry. The majority of the members of the organization were textile associations. While the membership counts do not give a sense of relative power or

190 size of each member (as textile associations were generally populated by small and medium-sized enterprises), the frequency of textile associations is not surprising, as the original spur for efforts at repressive collective strategies, in the form of coordinating anti-labor institutions, was a response to strikes in the Saxon textile industry. Also, recall from chapter 4 that there were a series of distributive conflicts between metal manufacturing and textile associations in terms of subsidizing central lockout funds. About 75 percent of the members were ―pure‖ industrial associations, indicating that much of the membership was denoted by local associations operating in specific geographic areas.

To determine the effect of the workers‘ threat on the adoption of repressive firm strategies, I test at the micro-level which voting districts in Germany were likely to have members of the HA. To do this, I first trace each geographically-based employer organization to an electoral district (wahlkreis). For example, several textile employer organizations were based out of the electoral district Chemnitz, where the

SPD won a majority in the 1907 election on the first ballot. For comparability purposes, I truncate the HA sample to just members that had their base of operations in one electoral district. This constitutes 68 percent of the member associations of the

HA. I can then use the data to determine if firms were more likely to adopt collectivist repressive strategies (proxied by joining the HA) if the SPD won on either the first or second ballot. The 105 members that signify repressive firm strategies that are members of the HA came from 64 electoral districts out of 397 districts in Germany.

Did firms adopt collectivist repressive strategies against workers, in areas where the SPD was successful in the previous year? Yes. Table 5-6 demonstrates

191 initial support for the ―left threat‖ hypothesis. 27 percent—the highest percentage—of the members joined the HA where the SPD won in the previous election—this is the highest percentage among all political parties. But firm participation in this anti-labor organization is not totally explained by SPD threats. Firms did adopt repressive strategies where parties of both left and right-wing ideological support won. At the other end of the ideological extreme, about 10 percent and four percent of the members appeared in districts where the German conservative and national parties won, respectively. After the SPD, however, firms adopted collectivist repressive strategies most frequently in electoral districts where the Center and National Liberal

Parties won (23 percent and 15 percent, respectively). It is important to note that part of the ―left‖ threat effect is not driven by a few politically important electoral districts; with a few exceptions, each HA member is traced to one unique electoral district. As

Table 5-7 shows, if we consider SPD victories by industry, the textile industry stands out as having an anti-labor strategy in areas where the SPD won the most seats. These results provide initial evidence for the ―left threat‖ hypothesis. While the SPD won in the plurality of districts that experienced anti-labor firm strategies, such organizations formed in other areas where other political parties did win.

3.2 Specific test of the theory in Prussian voting districts – the data

But what happens when we consider other covariates besides political ones, and also consider SPD vote share instead of just victory? To test the theory of firm strategies using statistical estimation, I also construct a new dataset that tests the importance of the left threat, economic heterogeneity, and basic demographic characteristics at the level of national voting district in Prussia. I consider the subset of

192 just the members of the HA in Prussia that can be traced to one electoral district, truncating the data in a different way from the previous section. This dataset allows for better testing of competing independent variables to explain which voting districts in Prussia were most likely to have important repressive firm strategies.

I test the determinants of repressive firm strategies in Prussia, using control data at the Kreis (city or village) and voting district levels. I constructed a dataset of basic demographic controls at the national voting-district level by using Kreis-level demographic data in Prussia (Galloway 2007) and appropriately ―nesting‖ this Kreis- level data within voting districts. In 1907 Prussia had 231 out of 395 Reichstag voting districts; the SPD minimally contested in all of them by this time. To do this analysis

I restrict the data to Prussia; there are 58 out of 64 HA members based out of Prussia.

(Again, this accounts for only the members based out of one electoral district; this indicates we are not dismissing too many ―single-district‖ organizations). This new dataset includes at the voting district level for the year preceding 1910 for which data are available: SPD vote share (1907); logged population (1905); percent urbanization

(1905); various measures of agricultural holdings (1905); and a measure of industrial heterogeneity. The heterogeneity measure of employment fractionalization is similar to that used in sections 1 and 2, but measured at the local as opposed to the national- sector level. The difference is that for each Prussian kreis, I calculate the percentage of employment by the six major industrial categories for which data are available. I then aggregate these percentages up to the voting district level. I then take the inverse of the sum of squares of these percentages. The resulting measure is the inverse

Herfindahl Index (HHI), giving us an effective number of sectors. This heterogeneity

193 measure ranges from 1.7 to 5, with a median of 3 and mean of 3.1. Higher values indicate higher heterogeneity, and this greater difficulty in collective action. Industrial heterogeneity and the vote share of the SPD in 1907 are correlated at .30.

3.3 Descriptive statistics for repressive firm strategies and membership in Prussia

What explains why firms pursued these repressive strategies in some Prussian voting districts but not others? To what extent do the left threat and heterogeneity affect firm membership in this organization? I find that higher support for the SPD is correlated with adoption of anti-labor firm strategies, supporting the argument that firms respond to moderate threats from the left. Before turning to regression results, I present basic descriptive statistics of both SPD vote share and distribution of HA membership.

displays the distribution of SPD vote share across Prussian voting districts in

1907. The modal district had less than 20 percent support of the SPD. Only 13 districts (6 percent of the districts) had a district where the SPD won on the first ballot.

The average vote share for the SPD in Prussia was 20.1 percent, a 2.5 percent average reduction from the previous election year of 1903.

Which Prussian districts had associations that were members of the HA? Out of 231 Prussian voting districts, 58 (or 25.1 percent) of the districts contained at least one associational member of the HA. Members of the HA in Prussia were involved in a wide range of industries, including raw materials, heavy industry and manufacturing, and textiles. Table 5-7 shows the distribution of HA members across the Prussian voting districts, broken down by quartiles of SPD vote share in the 1907 election. As

194 the electoral success of the SPD increases, a higher percentage of districts experience repressive firm strategies.

3.4 Estimation results

To test the relationship between SPD support, industrial heterogeneity, and HA membership, I estimate a probit regression model where the dependent variable is whether the Prussian voting district in 1910 has a member of the HA. Table 5-8 presents the results and lists the models with each of the classes of variables, representing different hypotheses. Model 1 confirms the main hypothesis that low to moderate threat from the left facilitates collective repressive firm strategies. Further the results of Model 1 show that there is a conditional marginal effect: the left threat only has an effect on collective strategies when that strategy can be facilitated by lower heterogeneity. Model 2 shows that this effect holds even when controlling for urbanization. Model 3 shows that this effect is robust to inclusion of other district- level controls. Thus firms respond collectively and in a repressive manner to the electoral threat posed by workers, but a collective response is only possible when structural conditions permit collective action.

Figure 5-2 presents further empirical support of the argument. This figure displays the relationship between the support of the SPD and predicted probability of adoption of HA membership at the district level, interacted with degree of heterogeneity. For low heterogeneity districts, an increase support for the SPD increases the probability of a collective repressive firm strategy; this increase is about

22 percent (the confidence interval ranges from 0.1 to 1). For high heterogeneity districts, the increase in support of the SPD does not affect adoption of collective

195 repressive strategies, as predicted by the theory. The change in percentage is -18 percent, and the confidence interval overlaps 0. However, there are few observations where the SPD threshold is about 40 percent; thus the effect of high support of the

SPD should be interpreted with caution, as the confidence intervals become significantly larger. The fact that even modest increases in SPD vote share have such an impact is partial evidence for the theory that repressive firm strategies to respond to moderate increases in the left threat, but only in less industrially heterogeneous areas.

The statistical test provides some confidence about the hypotheses about the determinants of repressive firm strategies. It demonstrates that repression is a direct response to the threat posed by workers, but collective repression is only possible when facilitated by more promising structural conditions that are for self-enforcing bargains among firms or constituent associations.

Section 4. Conclusions

In this chapter I use a number of different datasets on the participation in and activities of employer associations to test the importance of structural conditions that affect collective action as well as the redistributive threat from workers to explain repressive strategies. The first two empirical sections isolated the importance of industrial heterogeneity and institutional connections in explaining collective strategies. Section 1 demonstrated the importance of industrial heterogeneity in explaining the prevalence of collective action by sector. Section 2 demonstrated that institutional connections to other associations facilitate substantive collective action in the form of strike insurance. Section 3 specifically tested the redistributive threat aspect of the theory and showed that collective repression occurs most in areas where

196 heterogeneity is low and the threat from workers is salient. The overall evidence indicates that repression is substantively important and its emergence is consistent with the expectations of the theory discussed in Chapter 3. In the next chapter, I provide further confirmation of the theory by demonstrating the change in firm strategies to a temporary collaborative regime in the aftermath of WWI and the initial years of the Weimar republic.

197

Table 5-1: Explaining Employer Coordination at the Sector Level Model 1 Model 2 Outcome: Outcome: Fraction of labor force Member workers per covered employer association Total number of employer associations (logged) -0.10* -4,441* (0.047) (2522)

Absolute number of workers covered by 0.11 6,463*** associations (logged) (0.025) (1349)

Absolute number of firms 0.051 -2,710 covered by associations (0.041) (2191) (logged)

Sector fractionalization by -0.011* -543* employment (0.0058) (310)

n 17 17 Standard errors in parentheses. *** significant at p<.01, ** significant at p<.05, * significant at p>.10.

198

Table 5-2: Characteristics of Major National-Level Employer Associations Number of Number of Number of Number of associations associations member workers with/planning firms in employed by to provide associations member strike firms insurance Agriculture, 2 1 21 23,500 Nurseries, Fisheries Mines, Iron/Steel, 1 1 70 30,589 Salt refineries Earth and stone 12 5 2,864 169,390 Finished/metal & 14 6 6,996 635,874 Manufacturing Chemical 2 0 563 2,142 Textile 2 0 18 850 Paper 7 1 4,962 86,425 Leather 1 0 250 8,000 Wood, Cut Materials 4 1 5,217 79,558 Consumption/Food 5 1 45,106 43,930 Clothing Trade 4 1 3,494 55,580 Cleaning Industry . . . . Construction 2 0 20,994 251,600 Polygraphics 4 2 601 22,199 Trade, 4 1 78 43,500 Transportation Tourist 2 0 5,602 12,000 Housing/Services Arts 5 0 23,958 78,000 Mixed Organizations 2 2 38,644 2,438,142

199

Table 5-3: Explaining Strike Insurance Provisions in Employer Associations

Model 1 Model 2 Model 3 Institutional 1.58*** 1.37** 1.49** connection (.59) (.67) (.70) Metal manufacturing 1.27** .96 1.11 industry (.59) (.65) (.71)

Absolute number of member firms .075 0.074 (logged) (.22) (.22) Absolute number of member workers -.0046 0.015 (logged) (.25) (.25) Sector fractionalization by -0.033 employment (.057) n 51 51 51 Standard errors in parentheses. *** significant at p<.01, ** significant at p<.05, * significant at p>.10.

Table 5-4: Explaining Institutional Connections in Employer Associations

Model 1 Model 2 Model 3 Metal manufacturing .53 .064 -.084 industry (.33) (.70) (.76)

Absolute number of member firms -.34** -.40* -.39* (logged) (.18) (.21) (.22) Absolute number of member workers .11 .098 (logged) (.64) (.24) Absolute number of workers in sector .52 .088 (logged) (.40) (.49) Sector fractionalization by .11 employment (.071) n 51 51 51 Standard errors in parentheses. *** significant at p<.01, ** significant at p<.05, * significant at p>.10.

200

Table 5-5: Membership of the Hauptstelle Deutscher Arbeitgeberverbände by Industry and SPD Threat

Industry % of HA % of Membership Organizations that come from an SPD district Raw Materials 0.95 0% Leather 0.95 0% Paper 1.9 0% Services 1.9 0% Mines, Iron/Steel, Salt refineries 2.86 0% Polygraphics 4.76 20% Construction 7.62 0% Finished/metal & Manufacturing 9.52 30% Mixed Organizations 17.14 17% Textile 52.38 42%

Table 5-6: The Left Threat: Repressive Firm Strategies Occur in Districts with the SPD Winning Number of Districts the % of Total Districts

Political Party Party Won the Party Won

SPD 28 26.7

Center 24 22.9

Free People‘s Party 18 17.1

National Liberal 16 15.2

German National Party 10 9.5

Conservative 4 3.8

Polish Party 3 2.9

German Reform Party 1 1.0

% Total refers to the total number of members in the Hauptstelle Deutscher Arbeitgeberverbände

201

Table 5-7: Percent of Prussian Districts with Membership in HA Organization Increases with SPD Vote Share SPD vote share SPD vote share SPD vote share SPD vote share is 0-5% is 5-17% is 17-33% is >33% All districts 12.68 19.72 34.00 46.15 High 7.50 33.33 43.48 52.38 heterogeneity Districts Low 19.35 7.89 25.93 38.89 heterogeneity districts

Table 5-8: Explaining Collective Repressive Firm Strategies in Prussia

Model 1 Model 2 Model 3

1907 SPD 0.04** 0.11* 0.10* vote share (0.01) (0.052) (0.053)

Heterogeneity -0.18 0.33 0.29 (0.26) (0.42) (0.43) 1907 SPD vote share* -0.027* -0.026** -0.026** Heterogeneity (0.015) (0.008) (0.0016) Urbanization 0.026*** 0.03*** (0.009) (0.001) Agricultural 0.062 production (0.17) Agricultural -4.73 inequality (7.43) Pseudo R- squared 0.05 0.13 0.13 n 226 226 226 Standard errors, clustered by voting district, in parentheses. * significant at p < 0.1; ** significant at p < 0.05, *** significant at p < 0.01

202

Figure 5-1

1907 Distribution of SPD Vote Share

39.39

30 20 16.02 13.42 13.42

% of total districts 10.82 10 5.195

.8658 .8658 0

10 20 30 40 50 60 70 80 1907 Wahlkreis, SPD voteshare in 1907

203

Figure 5-2

Left Threat Predicts HA Activity

.8 .6

Homogeneous

Heterogeneous

Predicted Probability of HA Membership 0

0 20 40 60 80 1907 SPD Vote Share

Note: Estimation is of conditional marginal effect of SPD vote share. Controls: logged urbanization; agricultural inequality; logged population

204

Chapter 6: Evidence from Weimar Germany

―…[the] flood of events overtook us all‖ – Hans von Raumer110

The reflections of Hans von Raumer about the rapid adoption of a new labor law regime in Germany at the end of World War I illustrate some of the most important predictions and confirmations of the theory. The flood of events—the uncertainty that would accompany the return of soldiers after the war and the nature of the state-run transition economy—drove firms to strategically act to adopt a less costly regime for them. In this chapter I expand considerably on this point and test the theory with evidence from the inter-war period in Germany. I demonstrate that large changes in the two key variables—the redistributive threat to firms posed by workers and changes in industrial heterogeneity—are linked to corresponding changes in firm strategies. I also show that the main mechanisms for both variables—the level of the cost of repression and the variation in the cost of repression—operated to affect the changes in firm strategies. In particular, changes in industrial heterogeneity affected ability of employer organizations to enforce collective strategies.

As discussed in the previous chapter, Wilhelmine Germany was characterized by a set of repressive industrial institutions that were largely designed to inhibit unionization and were especially predominant in Prussia. This is most prominently evidenced by the relative strength of employer associations that use the various kinds of repression described in Chapter 2 as a response to unionization and electoral strength of the SPD. But after WWI and during the first part of the Weimar era, there was a qualitative shift to cross-class collective employer institutions that recognized

110 Raumer (1954)

205 unions. Most theories that discuss the historical causes or consequences of industrial institutions and social policy often reference these collaborative institutions during inter-war Germany. Yet, current theories and evidence either downplay the actual transition to such cooperative institutions, or are insufficient in explaining them. The advent of widespread, state-sanctioned collective bargaining in one of the world‘s largest economies, and the fact that it is a modern template for ―coordinated market economies‖ or the possible ―European model,‖ means social scientists often label

Germany as a ―critical case‖ for competing theories. The discussion in this chapter also goes one step further than current accounts and shows why collaborative strategies during Weimar Germany actually changed within the time period, and why repressive strategies were adopted and returned before the republic‘s ultimate collapse.

The chapter proceeds as follows: section one presents evidence of the shift to collective collaborative strategies during and at the end of WWI. The section links the variables and mechanisms of the theory to this shift. Section two presents evidence that changes in these variables led to a collapse in the adoption of collective collaborative strategies. The section describes the reduction of the threat from workers in the immediate aftermath of WWI, and the difficulties posed by exogenous economic conditions in enforcement of collaborative strategies among employers.

The joint affect of these variables was to make a return to repressive strategies preferable. Section three presents evidence of the use of such repressive strategies in the late 1920s, and tests a key observable implication of the theory described in

Chapter 3: the failure of actors with preferences for collaborative strategies to succeed due to problems of enforcement and incentives of other firms. This section presents

206 several pieces of evidence of employer association repression. I show the how employer associations challenged state laws and unionization; how employer associations using collective strategies to inhibit and constraint the actions of collaborative associations; how repressive employer associations lobbied the state as part of their costs of repression; and finally, how heterogeneity within the repressive faction affected its political lobbying.

The fourth section briefly discusses how the evidence from inter-war Germany confirms the theory from chapter 3, and challenges existing accounts and logic of employer coordination. In particular, changes in firm strategies cannot be attributed to changes in institutions that protected co-investments in specific assets, or the electoral system. Evidence that focuses on the variable of electoral systems cannot account for the change in strategies that took place within the same PR system. Evidence that focuses on the mechanism of co-investment in assets and the cooperative nature of PR cannot explain the distributional conflict that dominated much of the period. For each section, I use several pieces of evidence regarding historical events and data, including primary evidence, to show the causal strength of the variables. Where possible, I try to show that the changes in the posited variables are indeed exogenous and that firm strategies can be thought of as an important response to such changes.

Section 1: From Collective Repression to Collective Collaboration: the Legacy of

War-time Policies and Stinnes-Legien

In this section I show that firms that had previously supported collective repression as a strategy were willing to switch to a strategy of collaboration once the costs of maintaining that strategy became too high. When the threat from workers

207 rapidly increased following the end and chaos of WWI, firms preemptively signed with unions the Stinnes-Legien agreement in 1918, named after industrialist Hugo

Stinnes and union leader Carl Legien. The agreement has been heralded as one of the most important cross-class compromises in labor history. I describe the agreement more below; the broad contours of the arrangement were that firms and employer associations recognized unions, and unions in turn agreed not to overthrow the system of private property. As I argue below, this agreement largely came about due to firms‘ concerns about higher costs of repression: they would not be able to continue their pre-war strategies after the war. But, as discussed in sections two and three, the collective collaborative strategy ultimate collapsed once those costs fell and when the collective nature of the agreement was no longer self-enforcing. In the rest of this section, I discuss how the exogenous increase from the threat from workers during the war and at the end of the war caused the first major shift in firm strategies.

1.1 War-time institutions

Germany‘s participation in World War I and the unprecedented massive intervention of the state in all aspects of economic activity—in product and labor markets, as well as adoption of rations and prices controls—radically changed the legal position of workers and unions in the German economy and polity. The drafting of approximately eight million soldiers for the war meant the country faced a large labor shortage. The German government leadership, and the Economic and Prussian

War Ministries calculated by 1915 that the war-time effort could not afford labor disruptions in the forms of strikes or steady demands for higher wages (Feldman

1965).

208

What explains the initial recognition of unions, which involved a switch to collaborative strategies with firms? In autumn 1916, the German government under the leadership of Hindenburg and Ludendorff to engage in a ―total war‖ strategy which involved a large increase in mobilization efforts and shift of bureaucratic initiatives to further ration the economy. The exogenous conditions and demands of the war forced the German government to declare by fiat that firms and the state had to recognize unions. This was encapsulated in the Auxiliary Service Law (ASL) of December

1916, the first official act that recognized unions (Feldman 1981). The change in the relative threat posed by unions induced the state to recognize them, lest they be paralyzed by union intransigence during the war; the German government was reliant on unions to prevent strikes. This law was a monumental change in the status of organized labor in German society, and officially recognized German unions as legitimate bargaining counterparts. The Auxiliary Service law was a large departure from the previous doctrine which effectively allowed firms to pursue repression as a strategy.

Under the law, firms were compelled to recognize unions, and they became part of economic decision-making procedures within firms and were to be consulted by the relevant war authority boards. Firms had to negotiate with trade union secretaries on war-time bodies and establish or recognize war-time worker committees in the plants. After the law passed, the average worker wage increased and membership in both Socialist and Christian trade unions increased. Unions were exempted from conscription, and were given a representation in the procedures of

209 collective bargaining under the terms of the ASL of 1916. They were also allowed to recruit in previously forbidden areas, namely among state employees.

War-time mandated participation of unions in the economy created a joint set of concerns for many firms: higher wages for some firms that were already being squeezed by a set of price controls and rations, and concerns about state interference in decision-making. Of course, there was much firm opposition to these laws, and this opposition would return in the debate among employer associations over post-WWI industrial institutions as well as strategies after the war; the opposing arguments that

Silesian industrialists like Ewald Hilger or some of the Ruhr magnates would be used later during the Weimar period. Some firms and employer associations, despite the exigencies of war, opposed these new rules, and objected to their lack of participation in the war-time executive decision. Silesian industrialists like Ewald Hilger raged against concessions. But other prominent industrialists, who were also powerful within manufacturing employer associations, like Albert Vögler, Wilhelm Wilhelm

Beukenberg, and Hugo Stinnes were more flexible and accepted quickly the provisions of the law. These individuals would come to play a critical just under two years later, in the actual negotiation of post-war institutions.

On the union side, the 1916 Auxiliary Service Law strengthened the reformist tendencies of the Socialist Free Trade Unions. Union leadership discussed after the law their goal of building upon the law and forming a set of collaborative ―working community‖ institutions called the Arbeitgemeinschaft, a goal that was eventually realized at the end of the war. As discussed in section 1b and at the start of section 2, there was tension within both the main trade union and among newly formed left-wing

210 political parties about whether the ASL constituted sufficient gains for workers, or whether more extreme redistributive policies should be pursued.

It is important to note the context of the Auxiliary Service Law: the executive war ministries had already made other large significant interventions in the economy by this point. The War Office in fall of 1916 had substantially increased state intervention in the economy. This intervention included almost all policy areas relevant for firms: wage and price controls; freight rates, social legislation; military contracts. There were imposed cartels and wage and price controls. Employer associations and unions were forced by government mandates to navigate a series of price and export controls, and maintain workers‘ food supplies when government allocation schemes broke down; they also had to communicate supply needs to the local and national offices of the temporarily decreed war-time boards. While many firms also opposed the Auxiliary Service Law, both firms and unions wanted to lift export and price controls during and after WWI. By 1916, there was fairly uniform business opposition to continued state intervention after the war. The expansion of both bureaucratic influence, as well as union and employer associational capacity to inflict costs on each other and the state—meant that there was going to be conflict over how to reform war-time labor institutions once the war was over.

The demands of the transition of a peace-time to full war-time economy meant that unions and employer associations were more directly integrated into the decision- making and execution of the decisions of powerful ministries. During the conflict, and especially after 1916, pre-existing cartels and employer associations became more influential, as their staffs grew and came into regular contact with ministry planning

211 officials.111 Cartels were used to allocate raw materials; membership in such cartels was imposed by the government. War-time Germany embodied the highest degree of corporatist integration of interest groups into economic decision-making.

How did the variable of industrial heterogeneity and the mechanism of enforceability of collective institutions matter for the formation of collective strategies? Regarding differences among industries and the structural problems of collective action, the state essentially forced industrial cooperation in some areas.

Collective strategies had existed before World War I, as described in the previous chapter, so it was less difficult for the war ministries to use existing employer associations to ration and implement price controls and solidify cartels. But business opposition to intervention existed well before actual government policies took place.

Some firms joined cartels to prevent blunt the impact of state intervention—in fact they anticipated more intervention, and so explicitly expressed a desire to collectively act to prevent further state intervention (Feldman 1970).112 Business organizations that formed before WWI preferred minimal state intervention. For example, there were protests against the German Economic Ministry (RWA) by Hamburg business groups and the Association of Southwest German Chambers of Commerce in spring

1918.113 The ASL of 1916 and its attendant empower of organized labor, happened because the costs that workers could inflict had dramatically risen in the preceding two years.

111 See also discussion in 1916 meetings of the VdESI (1917; VdESI 1916). 112 One of the iron cartels representatives noted, ―…we should voluntarily join together in our own most basic interest, and not wait pressure from above‖ (Feldman 1970, 317, note 14) 113 Feldman 1970, 321. Interestingly, the protests of these groups provides some evidence against the notion that Southern German businesses were more pro union. This is discussed at the conclusion.

212

1.2 Transition to collective collaboration: overview of Stinnes-Legien

In this sub-section I show that the concerns about potential post-war chaos after WWI and the potential threat posed by radicalized workers and soldiers returning from conflict was the key variable in inducing firms to continue and lock in a collaborative strategy vis-à-vis unions. This collaborative strategy was embodied in the important Stinnes-Legien Accord of 1918. Unlike previous war-time mandates, though, this strategy was initiated by firms and intended to be more permanent, not a temporary war-time act. The evidence supports the theory that threats from the left compelled firms to switch to collaborative strategies. Large firms and unions were both concerned about managing demobilization of soldiers and the potential chaos after the end of WWI. They were also worried that the administrative authorities, especially the RWA, would either be unable to manage the demobilization, or intervene with its own set of procedures. For the firms which had been members of repressive employer associations, the costs of repression increased.

The November 15, 1918 adoption of Stinnes-Legien agreement was a formal shift from previously repressive to collaborative strategies of firms. The agreement was the first major cross-class agreement in Germany that guaranteed minimal working conditions and legitimized unions as a bargaining partner in the workplace. It is often considered a template for other major cooperative arrangements or social pacts. The main provisions of the agreement were a guarantee of a maximum eight- hour day for workers, and a basic collective bargaining framework between unions and firms. The agreement‘s procedures were ultimately executed by an institution that was managed by peak employer organizations and unions, the Central Working

213

Community (Zentralarbeitgemeinschaft, ZAG) (Kaun 1938). The main outlines of the agreement were that firms officially recognized trade unions as negotiating partners and had to consent to bargain with them. Mediation agencies and industrial committees to arbitrate disputes were created based on parity in representation of firms and workers. Existing employer associations joined a peak association called the Reichsverband der Deutschen Industrie (RDI).

In exchange for this recognition and provision, trade unions acceded to recognition and acceptance of private property, abandoning discussion of overhaul of the capitalist system. This was an important concession the industrialists extracted: the essential creation of a ―bourgeois republic.‖114 The main issues of contention in the negotiations leading up to November 1918 were compromises on yellow unions

(these were firm-sponsored worker organizations that had existed during the early twentieth century, which were designed to undermine representation in inter-firm unions); the maximum eight-hour day; and whether there would be parity representation on central committees that were administering collective bargaining agreements. Ultimately unions consented that any yellow unions that still existed after six months would be officially recognized by trade unions. The agreement stipulated joint management of industrial relations between unions and employer associations.

The overarching issue towards the end of WWI was whether the re- organization of the economy that gave unions more economic influence was only a short-term, crisis-contingent set of institutions, or rather, a permanent re-organization of society. This was the issue of fundamental disagreement between business leaders,

114 Feldman 1970, 314, in a phrased coined by Arthur Rosenberg.

214 union officials, and the ministry leaders. There was also of course disagreement over the temporal nature of these war-time institutions within the German bureaucracy, disagreements which increased as Germany‘s defeat became clear. Lieutenant

Colonel Joseph Koeth, the head of the powerful Raw Materials Section of the Prussian

War Ministry, viewed the reforms as something more permanent. Business leaders such Walther Rathenau advocated initially that the institutions should be short-term.

As it became clear that Germany would lose the war, employer associations and firms faced the decision of whether to advocate reversion to previous repressive strategies, or maintain in some form the collaborative institutions that had developed.

Unions faced the decision about how much of a change to the war-time institutions they wanted. And both unions and employer associations had ―extremist‖ factions— some firms thought that preferred total repression a complete reversion to pre-war strategies, and some unions thought that post-war uncertainty and the return of soldiers was the best opportunity to launch a revolution and build upon the ASL. Moderate firms and unions therefore had some mutual interests in preventing the extremists from winning.

Why would industrialists have wanted a pre-emptive agreement with workers to present to the German authorities? What were the advantages of this pre-emptive agreement? Steel industrialists thought the organization would have more lobbying leverage and would be able to negotiate legislation before it was passed. The conditions for a collaborative and a pre-emptive collective strategy had been set by war-time intervention of the state. An important priority for firms was the removal of

215 state regulations on prices, quotas, and wages.115 On the trade union side, leaders wanted to contain radical demands from workers for the overthrow of the capitalist system and focus on locking in the benefits engendered by the Auxiliary Service Law, and other war-time regulations that gave organized labor more influence.

Of course, strategies regarding labor markets were not the only major concerns of firms after WWI. A core business preference towards the end of WWI was autonomy from the state (Selbstverwaltung). As discussed in the previous chapter, in the early twentieth century there had been little intervention of the state with private setting of wages; the ―Herr im Haus‖ attitude meant freedom from demands of workers as well as those of the state. And as discussed above, firms wanted a removal of war-time price and export controls and a reduction in the power of various boards that been set up to administer them. Industrialists feared that government interference and surveillance would continue after the war, particularly after the passage of the

ASL. They viewed cooperating with unions ―as the best defense‖ against state intervention and post-war chaos. The desire was attaining autonomy from workers‘ demands and from the state.

1.3 The build-up to the switch to collaboration

The monumental shift to collective collaborative strategies can be traced to the crucial year leading up to November 1918. The shift was because of the initiatives of a handful of industrialists who had close connections to employer associations and/or were prominent industrialists with contacts in the war-time and economic ministries.

115 In February 1918 the CDI created a fund to fight postwar ―socialization‖ schemes. See the September 1917 discussion in the Executive Committee of the Association of German Iron and Steel Industrialists (VdESI). The Mitteilungen des Kriegsausschusses der deutschen Industrie, from October 1918 discussed post-war transition planning, in particular reducing the role of the state (Industrie 1918).

216

Hans von Raumer, the business manager of the Association of the German Electro-

Technical Industry was one of the major initiators of discussion over post-WWI industrial relationships with unions. Raumer felt assistance of organized labor in setting up post-war economic institutions was necessary: without it, there would be a higher possibility of chaos, social tension, and bureaucratic interference. He thought there should be an ―organic collaboration with the unions‖ before the ―flood of events overtook us all‖ (Raumer 1954).116 Raumer and his colleagues wanted a set of institutions that would manage the demobilization process and re-integrate soldiers into society, while containing radical elements. He spoke with the leaders of industries of manufactured finished products (Carl Friedrich von Siemens, Walther

Rathenau, and Felix Deutsch, who were all critical of the attitudes of heavy industry).117 In their view, pre-emptive cooperation with the major unions would also help the neutralize socialist threat.

Prominent industrialists and representatives of employer associations feared both continued government intervention and the inability of the government to prevent chaos from workers after the war. They were also concerned that a left-wing

Reichstag would make the wartime gains and power of trade unions permanent.118

Raumer also pushed for further encourage centralization and branch organization across all industrial sectors to create a suitable set of partners to negotiate with organized labor.119 It was not just skilled industries that would benefit from this action, as previously existing heavy industry associations would see expansion of

116 Raumer, Deutsche Rundschau, 80, 428 117 See discussion in Raumer (1954), 428. 118 See Raumer (1954) and the discussion in the VdESI meetings of 1917-18. 119 Feldman 1981, 167

217 influence. Industrialists thought that during the close of World War I that membership for trade unions could serve as a way to integrate returning soldiers and prevent further radicalization and manage the demobilization of soldiers. The rhetoric used was to ―save‖ Germany from the threats of bureaucratic and parliamentary intervention and revolution. Raumer thought that without the role of trade unions, radicalized masses might have undermined the war effort. The negotiations that started in 1917 were a deliberate effort on the part of both sets of industrial actors was needed to pre-empt state intervention and possible civil war or chaos.

Where did the source for agreement come from? The actual intellectual origins of regular post-WWI institutional contact between firms and workers came from

August Müller and Hermann Schumacher. Müller was a conservative member of the

SPD who worked for the War Food Office in 1916 and the Imperial Economic Office

(Reichswirtschaftsmat, RWA) in fall 1917 and had contact with trade unionists.

Schumacher was an academic with contacts in business circles. In May 1917

Schumacher arranged a meeting between Müller and Stinnes in Berlin; two larger meetings followed between industrial and union leaders on August 9 and October 2,

1917 (Feldman 1973; Kaun 1938). The war-time imposition of temporary union recognition as well as expansion of employer associations meant that high-level functionaries from both sides had been in regular contact managing rations and quota compliance (or non-compliance).

The primary evidence from the actual negotiations among firms, and between firms and unions, demonstrates the importance of the ―shadow of the future‖ in terms of the threat posed by post-war chaos and a potentially interventionist left-wing

218 government for decisions of firms. In particular, reports from the meetings of the

Mitteilungen des Deutschen Kriegsauschusses and Association of German Employer

Associations (VDA) from early 1918 demonstrate the concern with returning soldiers and whether the ASL was going to be extended. At the meetings, outlines for the agreement for what would become Stinnes-Legien were debated (largely proposed by

Raumer, Reichert, and Stinnes). Stinnes had become one of the important liaisons between the firms and the war ministry. He at first thought cooperation with unions was a secondary issue to securing primary ―war aims‖; he was concerned about what territory would be annexed in Belgium.

The initial concessions from employers to unions in this process came as early as September 1917, when Beukenberg said it would be permissible for industries to have trade union secretaries involved in discussions with the RWA regarding future social legislation. Vögler in April 1918 stated, ―Employers and workers have very similarly directed interests. For this reason they ought to get together and represent them in common. If they can only find themselves in agreement on the great economic questions, then perhaps they will have found the bridges which may lead to a mitigation of social conflicts‖ (Varain 1973).120 Beukenberg indicated that industry was prepared to have trade union secretaries participate in the discussions between the

RWA and the interest groups concerning future social legislation. Note that this is a logic of cooperation not based on co-investment of assets. Rather, it is a logic based on mutual interests in preventing state interference in industrial affairs and management of demobilization and integrating workers without state control. The

120 This quote appears Varain 1973, 115. Beukenberg‘s concession is recorded in the VdESI meeting of September 24, 1917.

219 purpose of this proposed cooperation was to contain potentially radical or recalcitrant factions on both sides.

After this initial concession, in the summer of 1918, Raumer expressed fears that if unions were not involved in management of demobilization, there would be either chaos or bureaucratic interference. Raumer voiced the position that industry should undertake an ―organic collaboration with the unions‖ before the ―flood of events overtook us all.‖121 On October 2, 1918, Raumer met with the following representatives of the trade unions: Carl Legien, Gustav Bauer, and Alexander

Schlicke. By then, Raumer reported more business propensity to form an agreement with unions, given the worsening conflict situation. Ludendorff demanded an immediate armistice and the government led by Max von Baden formed on October 3,

1918: this government would have Gustav Bauer as State Secretary of the newly formed Labor Office, and would include Social Democrats as part of the new government. Why did the unions trust Raumer? Raumer‘s overtures to unions were made, in a letter from Raumer to Siemens shortly after the meeting, ―at a time when the military situation was favorable (my emphasis). This fact made the gentlemen

[union delegates] confident that our negotiation was not created by the changed circumstances but sprang rather from a freely made and therefore honest decision‖

(Feldman 1970, 325-326).122

121 Raumer, Deutsche Rundschau, 80, 428. 122 The original source is in a letter from Raumer to C.F. von Siemens, Oct. 8, 1918, from the Siemen Archiv, Nachlass Henrich, 11/Lg736. As Feldman discusses, Raumer‘s elder brother was a friend of Stinnes, and Raumer had known Stinnes since 1899. They served as DVP deputes in the Reichstag; Raumer became Treasury Minister in the spring of 1920.

220

Raumer distinguished his preference which was labeled the ―Berlin‖ stance, for collaboration, from that of the coal and Ruhr industries, which he anticipated would be more difficult: he thought more moderate terms could be negotiated in Berlin, but they would have to deal with the objections of the coal industry. Raumer and Stinnes had to win over the ―Ruhr contingency,‖ the faction that had been more repressive before the war. Of course, Stinnes had investments and affiliations with some of these repressive industries and firms. At this point both Raumer and Stinnes thought agreement with the unions to manage post-war issues was necessary. Forging such an agreement became a critical item on Stinnes‘ agenda in October 1918.

What did the Ruhr industrialists say regarding this? They were not the initiators of these negotiations, even though Stinnes had financial interests in heavy industry. But, the repressive faction of capital—represented by leading iron industrialists who had been members of the Ruhr employer associations—met after the formation of the new government and Raumer‘s successful meetings with trade union representatives on October 9, 1918, in Düsseldorf. There they ―officially‖ consented to have Stinnes negotiate with the unions over post-war industrial institutions. Why did they do this? What was the counterfactual of not doing so? Jakob Reichert, who played an integral role in these negotiations, recounts the logic of the employers‘ move in a 1919 address:

―The assembled [heavy industrialist representatives] were unanimous in feeling that the government of Prince Max von Baden and Herr von Payer was untenable under the existing circumstances and that it would soon be brought down. One calculated the life of this government to be not more than 4 or 5 weeks…In any case, the iron industrialists could expect no help from a weak government. If one looked elsewhere and asked if the middle class (Bürgertum) might become a strong help and support for German economic

221

policy in the future, then one was compelled to say, in view of the many regrettable events and frequent disappointments experienced over the decades, that there was no counting upon the middle class, as it is constituted in Germany, in economic matters. Only organized labor seemed to have a commanding influence. Form this fact the conclusion was drawn that, in the midst of the failing power of the state and the government, industry could only find strong allies in the labor camp, and they are the unions.‖123

Discussions after the Dusseldorf meeting took place among Ruhr representatives,

Berlin representatives (industrialists), and unions, on October 22, 1918 (between

Raumer and unions), and between Stinnes, Vögler, and union representatives in

Düsseldorf on October 26, 1918.124 This is evidence that eventually previously repressive firms and employer associations were grudgingly supportive of a collaborative approach because it was the least costly option.

What were these common interests regarding demobilization and state interference for unions and firms? Both objected to the RWA plan for demobilization, which would have involved War Office Bureaus that were scatted throughout the country. They objected to these agencies because their view was such agencies were both an infringement and that they were not capable of managing the return of the soldiers; they thought that only branch organizations of industries and unions could work. As Ernst von Borsig, a Berlin machine builder articulated, ―90 percent of the statistical surveys gathered by the war Office Bureaus are not worth the paper they are written on…‖; he ―…would rather have the unions do the job [i.e., of locating work for returning soldiers and unemployed munition workers alone‖ (Feldman 1970,

123 This address comes from Jacob Reichert, Enstehung, Bedeutung, und Ziel der ―Arbeitgemeinschaft‖ (1919), 6. 124 Stinnes wanted to avoid the sense to his colleagues that industrialists were ―chasing‖ after unions.

222

328).125 In fact, Joseph Koeth, the previously discussed head of the Raw Materials

Section of the Prussian War Ministry, thought it was obvious that the employers and trade unions should organize the demobilization.

Rathenau explicitly feared by mid-October 1918 that disorganized demobilization, which would be partially imposed by the allies, could lead to ―civil war and Bolshevism.‖126 Director Henrich of Siemens addressed Rathenau‘s concerns explicitly with a proposal presented to the German Society of Engineers on

October 20, 1918. His idea was to have a new demobilization office but have that office staffed by representatives of unions and industrialists, as well as some government representatives. Henrich thought that such co-management was necessary for an effective demobilization.127 The issue for Henrich was getting approval of

Socialist State Secretaries Scheidemann and Bauer, and the left-wing Centrist leader

Erzberger. Stinnes supported this plan and in fact thought it should be marketed such that the unions would take a more active role. The October meetings afterwards discussed implementation of this plan. In Berlin, Stinnes and his colleagues asked the major employer associations to recognize unions; allow for their organization; enter into collective bargaining agreements; assist with labor exchanges; and have mediation organs based on parity. Employers were asked to renounce financial

125 The original documentation is from Bayerisches Hauptstaatarchiv München, Abt. IV, MKr, Nr. 14- 412. 126 This quote is from Walther Rathenau who was concerned with speedy demobilization which the allies would impose on Germany. He wrote War Minister Scheüch a letter on October 15 that urged formation of a special Demobilization Agency to coordinate the employment of returning masses in collaboration with industry. See Feldman 1970, 329; Feldman 1973. 127 This statement comes from the Henrich Nachlass, Siemens Archiv, 11/Lg. 736. Other people at the meeting were Directors Rieppel, Borsig and Deutsch.

223 support of ―yellow‖ unions.128 So by mid-October 1918, the core issues were set on the table. In Düsseldorf, meanwhile, union leadership was pressured by rank and file members to secure an agreement that there would be a maximum eight-hour day.

Another, almost equally important objective was that unions were to pressure employers to bring ―yellow‖ union representatives to negotiations, despite the refusal of the ―fighting‖ unions to recognize them as legitimate worker organizations.129

These issues were not settled in October. Stinnes expressed the view that ―trade unions were the one group in society that had the power to contain the revolutionary situation and that no confidence could be placed in the weak German middle class

(Bürgertum).‖

Events towards the end of the war and the dismissal of Ludendorff on October

26, and the fact that an armistice was going to be called on allied terms, increased the urgency of firms to negotiate a deal. The major initiators of these negotiations were aware that enemy territory they had occupied was going to be evacuated quickly and that institutions were going to be needed to foment an orderly demobilization. On

October 30, 1918, Stinnes, Raumer, Legien, and Bauer approached Ludendorff and convinced him to keep Hindenberg as head of the army throughout the demobilization process. They presented to Ludendorff the draft agreement: ―The employer and worker organizations are resolved to work together in unity concerning all questions

128 Raumer, Deutsche Rundschau, 80, 428 and Kaun, 48-49. Feldman notes that the ―yellow union‖ issue wasn‘t regulated until November 5, and the acceptance of collective bargaining agreement sin all industries was not made until November 12. But the Ruhr industrialists were less conciliatory. Feldman 1970. 129 Reusch was insistent that all unions participate. He corresponded with Beukenberg. Feldman also documents the labor shortage issue from Stinnes: ―I expect a gigantic labor shortage about 12 months after the demobilization which will then last a few years. These two or three years must be used by us in Germany. After that the whole world will start saving. By then we must sit in the saddle abroad once again.‖ From correspondence in HA / GHH, Nr. 3001242/7.

224 pertaining to the demobilization and transition from a wartime to a peacetime economy.‖130 This document asked for a Demobilization Office with executive powers to be composed of an equal number of worker and employer representatives.

This program became an official policy on November 9, 1918.

On November 3, 1918, Raumer presented this proposal to Vice Chancellor von

Payer, communicating to him that the employers and their respective organizations and unions were the only effective organizations capable of running the demobilization. Stinnes, Rathenau, Legien, and Christian trade union leader

Stegerwald discussed with Chancellor Max von Baden and threatened the government that they would not collaborate with each other or with the government, if it did not submit to the idea of the demobilization office (Kaun 1938, 50). The von Payer

Government consented and appointed Koeth as head of the newly established

Demobilization Office.131 Koeth had been the employer associations‘ preferred choice to run such an office. Raumer contacted other prominent employer associations and convinced them to be part of a grand Arbeitgemeinschaft, culminating in a central organization. There would be an ―almost complete organization of German industry on the basis of agreement‖ (Feldman 1970, 334).

The actions of some the local workers‘ councils established in 1918 validated the concerns of firms regarding costs of repression. For example, provisional workers‘ and soldiers‘ councils in Berlin and Cologne called for a socialist republic in

130 Letter from Raumer to Siemens, Oct. 30, 1918, Siemen Archiv, Nachlass Henrich, 11/Lg 736). Feldman 1970, 333. 131 It is not clear what made the threat from the union and business leaders credible: certainly they had privately expressed to their respective colleagues that the whole point of the procedure was to avoid post-war chaos and/or government intervention to stem such chaos.

225 early November 1918. On November 7, the executive of the Hamburg councils declared: ―This is the beginning of the German Revolution. The World Revolution.

Long live Socialism. Long live the Workers‘ Republic.‖ The soldiers‘ council of

Brunswick thought power should remain in the hands of these councils, and should have more power than any national assembly (Mommsen 1978; Sollmann 1918).

Militant declarations were made by similar works‘ councils of Dresden, Leipzig,

Chemnitz, Cologne, Cuxhaven, Bremen, Berlin, as well as the Berlin Factories of the

German Motorworks, Fritz Wener AG, and Siemens-Schukert (Geary 1978; Ryder

1959). In early November 1918, workers and soldiers‘ councils of Brunswick, Berlin,

Chemnitz, Munich, and Stuttgart were under direction of left-wing radicals.

These high threats near major industrial centers indicate that indicate that the solution to co-opt workers was preferred to state intervention and the chaos of return soldiers without union institutions to integrate them. The threat posed by a left-wing government induced firms to switch strategies. Firms in fact preferred a political alliance with workers to reduce the risk of expropriation. The actual German revolution increased the urgency of solidifying the agreement, which was drawn up on

November 12 and ratified on November 15. The violence accompanying the revolution made the clear the importance of committing to a collaborative strategy, as the costs of repression now became more salient to firms. Raumer and Stinnes expressed fears at this time of ―anarchy, Bolshevism, rule of the Sparatcists or chaos.‖132

132 Kaun (1938), 52.

226

The formal agreement included the following provisions: the introduction of eight-hour day in all industries; recognition of unions in all industries was to take place through collective agreements on wages and working conditions; existing employer associations would administer such agreements; the agreements would be binding on all industrial groups not previously involved in the negotiations.133 There were also specific compromises regarding the yellow unions. Unions agreed to recognize such unions were still intact after six months as representatives of workers.

This was worth compromising to achieve their goal of Article 10 of the agreement, which guaranteed parity in representation.134 The agreement also introduced collective wage agreements on an industry-specific basis.

The Stinnes-Legien agreement created the Central Working Community

(ZAG) and stipulated the creation of a new peak association of employers. The Reich

Association of German Industry (Reichsverband der Industrie, RDI) was created after

WWI in 1919 in place of older business groups, but the institutional structure of the sector-level organizations was retained. The RDI had 78,000 member firms in 61 affiliated associations. Its structure was based industry-level organizations which monitored collective bargaining arrangements. The newly formed RDI also represented machine-building, electro-technical, and chemical industries in negotiation with trade unions, as well as the older manufacturing and heavy industry associations.

As discussed below, though, employer associations which had previously repressive

133 There were ―secret agreements‖ that made eight-hour day introduction contingent based on the agreement of other industrial nations. 134 The text of Article 10 reads: ―To carry out this agreement as well as to regulate further measures needed for the demobilization, the maintenance of economic life and the securing of the possibilities of existence for the workers, especially the war disabled, the participating employer and worker organizations will establish a central committee based on parity with a substructure organized by branches‖ (Preller and Tennstedt 1978), 53-55, also Kaun 1938.

227 strategies resented this representation of new industries in the RDI and would undermine the RDI‘s operation. Gustav Streseman in fact expressed concern at the preservation of powerful sector-level organizations and officials in the RDI, and was concerned with the predominance of branch organizations in new organizations.135

The ZAG successfully achieved another goal of firms: it was in charge of overseeing collective bargaining and managing the industrial demobilization. It played the role that pre-existing bureaucracies like the RWA would have taken on. Its constituent interest groups demanded executive authority and new demobilization office under control of Koeth. The policies of the newly demanded demobilization office were determined in collaboration with the ZAG, and implemented on the basis of self-administration by the Arbeitgemeinschaften.136

The protections to workers enshrined in 1918 in the Stinnes-Legien Agreement were also legitimized later in the Weimar Constitution of August 1919, which recognized unions as a legitimate economic organization with functions in the state.

The Weimar Constitution also authorized the government to act as an arbiter in settling disputes. Article 154 obligated the government to protect workers‘ rights; article 159 guarantee freedom of association; article 123 affirmed their right to hold

135 Streseman wrote a letter to the League of German Industrialists: ―I am very concerned by the strong predominance of the branch organizations called forth by the tasks of the Arbeitsgemeinschaften, and, as a consequence of this development, the apparent intention of staffing the future presidium of the Reichsverband with new personalities, who will no longer by the spokesmen for the specific views on economic policy of a central organization but rather shall collaborate in the leadership of the Reichsverband as representatives of their particular branch of industry. Things being as they are, the coal industry, the steel and iron industry, the electrical industry and the chemical industry want to be represented by their leaders in the presidium. To this will probably be added representatives of large scale industry in Berlin. That means strong influence by the persons like Hugenberg, Borsig, Stinnes, Deutsch, Duisberg in the new presidium…‖ Stresemann, March 31, 1919, Roll 3051, Serial 6816H, Frame H/122937. 136 Feldman 1981. The original documentation is in Internationale Korrespondenz, 85.

228 public meetings and distribute leaflets. Article 165 provided for works councils and district and national economic councils, and granted the worker ―an equal voice with the employer in the regulation of salaries, working conditions, and economic development.‖137

This condensation of an incredibly complex set of events from 1916 to 1918 provides observable implications and evidence that support the theory. When the threat from workers to firms exogenously increased in terms of the demobilization and uncertainty about radicalism, firms had an incentive to adopt collaborative strategies, as repressing workers and returning to pre-WWI institutions would have been too costly. Some industrialists that led the charge were also those that were involved in repression before the war; other repressive associations were convinced to join. The reason for the cooperative strategy was less about returns to co-specific assets, but about prevention of state intervention and higher threats from workers. The theory in chapter 2 demonstrates that pre-emptive collaboration is an observable implication that happened in the German case. Further, the collective strategy allowed for both sides to effectively sanction extremists in both camps. Carl Duisberg, who would be head of the newly formed RDI, mentioned of Germany‘s transition to parliamentary democracy and cooperation with the unions: ―I work hand in hand with the unions and seek in this way to save what can be saved. You see, I am an opportunist and adjust to things as they are…‖138

Section 2: The Collapse of Collective Collaborative Strategies

137 Braunthal 1978, 143. 138 This statement is from a letter from Duisberg to Dr. E.A. Merck, October 17, 1918, from personal records of Duisberg, Bayer Archiv.

229

In this section I demonstrate that changes in the threat from workers and industrial heterogeneity meant that firms ended collaborative strategies and switched to repressive strategies during the Weimar Republic. The first section documents extensively the period from the signing of Stinnes-Legien and early Weimar as one of reduction in the threat from the left, as compared to the end of World War I. The second section shows that this reduction in the threat meant lower costs of repression, as well as increased industrial heterogeneity induced by the passing of hyperinflation, which induced higher variance in the costs of repression. This caused firms to end the set of collective collaborative strategies enshrined in the ZAG. As a result, the ZAG unraveled. Exogenous industrial conditions made the ZAG impossible to enforce.

2.1 Reduction in the redistributive threat from workers

Not long after the ZAG was formally implemented, large firms that signed on to Stinnes-Legien out of fear of further expropriation from workers and the state immediately began to try and find ways out around the agreement. I argue this was a response to the lower threat posed from workers. The passing of the threat of social revolution and the moderation of the SPD meant that firms actually had lower costs of repression. At first, it might seem strange to argue that in the early years of Weimar

Germany, the threat from workers to the modal firm had fallen. Metalworkers had played major roles in strikes in 1916, 1917, and 1918. In the first Weimar elections of

January 1919, the SPD received 37.9 of the vote share; the USPD received 7.6. One interpretation is that this constituted a high threat, because in areas with high degrees of manufacturing output, support for the radical left-wing USPD party was high. In particular, the USPD cut into SPD vote share in areas concentrated with

230 manufacturing workers: in Berlin, it was 36.4 (SPD) versus 27.6 (USPD); in

Düsseldorf, 34.6 versus 22.5. Overall, in the 1920 elections, the USPD did well in heavily industrial areas, indicating some more radicalist support than previously thought (Geary 1978). The USPD moved even more left, affiliating with the Third

International in 1920 (Prager 1970).

In addition to the USPD‘s electoral contestation, there was the matter of its influence on unions. By October 1919 the USPD had taken significant membership control within the metalworkers‘ union, the largest in Germany (Geary 1978; Opel

1957). In January 1919, there had been a push by the USPD and affiliated unions for socialization and state ownership in the Ruhr area; a Spartacist rising in Berlin; socialist republic proclamations in Cuxhaven on January 11, 1919, with similar proclamations in Bremen, Brunswick, Munich, and Mannheim. Massive worker demonstrations against the Ebert-Scheidemann government took place in Brunswick,

Stuttgart, Halle, as well as insurrectionary strikes against the Kapp Putsch in 1920 and communist uprisings Hamburg, Saxony, and Thuringia in 1923. Firms observed these events after having just committed in 1918 to granting unions vastly more powers than before, in conjunction with an SPD run government with an SPD Labor Minister.139

Despite these events, though, I argue that the absolute level of threat from workers did fall in this time period and shortly after. Compared to the costs of repression that concerned employers in the final months of World War I, what

139 The demonstrations of November 9, 1918 in Berlin were mainly by workers from northern armaments factories. Many of the war-time strikes were organized by the Revolutionary Shop Stewards Organization (RSSO), as were the Sparatacist risings of January 1919. The RSSO was based upon the turners‘ branch of the Berlin Metalworkers‘ Union Metalworkers. Through the 1920s, 145 out of 285 Communist Party (KPD) functionaries worked in the metallurgical industries. The core of KPD support came from metalworkers.

231 transpired was a largely successful attempt of the SPD and General German Trade

Unions (Allgemeiner Deutscher Gewerkschaftsbund, ADGB) to signal and execute a strategy of punishing radicals and tempering demands for extreme redistribution. Of course, after the war, high level and party functionaries had bitterly debated the future of socialism. For some, the Stinnes-Legien agreement entailed the abandonment of socialist revolution as a goal of workers. I now turn to three pieces of evidence that illustrate why and how unions and the SPD were able to moderate their demands.

2.1.1 Moderation, ADGB leadership, and institutional structure

In 1918, the ADGB had 5.5 million members; this grew to 7.9 million in 1920.

85 percent of all ADGB members belonged to one of the 15 industrial unions; this was immense compared with the number of 20 percent in 1891. At the 1919 Congress of

ADGB, 645 delegates represented the 5.5 million members; 20 percent of whom were from the more extreme Metal Workers Union. The ADGB had proportional voting in their annual meetings. Despite those foreboding numbers for firms, combined with the incidents of industrial unrest above, the ADGB was able to contain radical threats because the actual executive decision-making bodies granted heavy influence to leaders of more moderate industrial unions, and because there was ―push back‖ against radical positions from the moderate constituent unions, which were less radical than the metalworkers‘ union. The Federal Executive Committee of the ADGB

(Bundesvorstand) had 15 members was the primary decision-making body; seven were paid functionaries elected by Congress delegates. The other members, selected at the Federal Executive Board, or the Bundesausschuss, were president of the eight constituent unions (Braunthal 1978, 91). This Board was another important body: it

232 was composed of the president of each affiliated union, and had additional representatives for every 300,000-500,000 members.140

This executive was steered by conservative union leaders who could ally with conservative members of the executive board. Leipart took over for Carl Legien as the national ADGB Commissioner in 1920, and stayed in that position until 1933. At the time of Leipart‘s promotion, the executive members of the ADGB were affiliated with or loyal to the long influential ―right wing‖ of the SPD, which included important party entrepreneurs like Herman Müller, Peter Grassmann, and Lothar Erdmann. This shift in conservatism started at the 1919 ADGB Congress, when members affiliated with the SPD and USPD were rivals. Local unions attempted to put up USPD- affiliated representatives for nomination to the ADGB executive boards, but were outvoted.141 Leipart was never challenged as president of the ADGB: he had the support of majority of Congress delegates and Executive and Board members throughout his tenure. Union bureaucracy—the institutional restraint of radical representation—meant some dissatisfaction among rank and file: during crises, workers went on wildcat strikes or looked for extremist political solutions, and union leaders in response tried to sanction such behavior.

The right-wing or ―reformist‖ wing was headed by conservative unionists

Leipart and Grassmann: they set the ADGB‘s more conservative policy goals and

140 The ADGB executive had some agenda-setting power, but could not overrule the leaders of other constituent unions. For example, the executive body could not have ordered the mine workers in Ruhr to strike; workers in that union had to vote for such action. The ADGB institutional structure went to regional and city/town levels. In 1930, the ADGB had 1,300 local units. Within ADGB, the affiliated unions maintained rights over strikes and their own budgets, and acted as counterweight to centralized system (Braunthal 1978). 141 The ―radical‖ wing did have some seats on the Board: they included Robert Dissman (President of the Metal Workers Union), who had been one of the USDP founders.

233 endorsed political cooperation with the SPD. Although there was Communist support in the Metal and Textile Workers Unions, and in the works councils in Ruhr and

Upper Silesia (in areas where employer associations been largely collective repressive), the Communist faction never seriously threatened the policy positions and direction of the national-level ADGB executive board.142

The ADGB leadership also involved itself in framing economic clauses of SPD program to be more conservative and steered it away from using Marxist ideology in the SPD Congress statements. For example, at the 1925 Heidelberg Convention, the

ADGB wanted to influence clauses on party ideology. Leipart wrote to the SPD

Chairman Müller (the SPD Chairman) of his objection to the draft paragraph on the struggle of working class. Leipart opposed the strong declarations: ―The claim of the

SPD for the ideological leadership of the unions can be as little recognized as the leadership claim of the Communist International, which has been vainly attempting for years to force its goals and its tactics on the unions‖ (Braunthal 1978, 124). As a result of Leipart‘s objections, the SPD reworded the clause and put the economic aspects of the struggle on part with political aspects reference to the primacy of political activity was omitted. The SPD leadership ended up conceding to ADGB objections on acceding to reformist and not revolutionary principles in a statement.143

2.1.2 Evidence of moderation: sanctioning radicals

142 Communist-affiliated workers were about 10 percent of membership in 1925, and in the Metal Workers Union this was about 16 percent; this was a decline relative to the 1922 Congress. 143 Note that these policy positions were not due to concerns about preservation of skill-training institutions. They were due to preserving the redistributional influence of the ADGB. In fact, there were battles in the early 1920s about the integration of craft unions into the structure of the ADGB.

234

Sanctioning radicals: Communists. In 1919, the communist faction of the

ADGB argued that industrial unions were superfluous and urged workers‘ councils take over; they demanded revolutionizing of existing unions through communist cells and ousting SPD officials. In 1920 they accepted the ―Red Trade Union International‖ and thought that communists had the duty to wage a struggle within the socialist unions. KPD leaders in November 1923 organized a conference in Weimar which was designed to counter the annual ADGB Congress. In response, the ADGB executive board simply expelled local councils that had communist affiliations, depriving them of access to funds. For example, the construction union branch of the ADGB revoked the membership of 4,000 unionists. This is an example of the ADGB not tolerating competing radical ideology, and making good on its threats to sanction its own radicals. The KPD tried to set up alternative unions (the Revolutionary Trade Union

Opposition, the RGO), but it was never that successful (its peak membership in 1932 was 312,000), and this alternative union controlled less than two percent of ADGB local councils. This KPD backed alternative also failed in its calls for a general strike in 1932 and 1933.144 The KPD and Communist unions did not really move overarching ADGB policy leftward during the Weimar era.

Containing radicals: USPD supporters. There was also the serious matter of the radical faction of workers who supported the USPD. At the 1919 ADGB

Congress, the USPD claimed to have support of one-third of the member workers, but this was the maximum level, and would decline in the 1920s. At this congress, the

USDP succeeding in forcing conservative union officers to abandon their previous

144 Note that the ADGB had an incentive to exaggerate the communist threat to its own members in order to ―hide their own failures‖ (Braunthal 1978, 103-104).

235 policy of supporting the SPD only, and to accept a resolution on political neutrality.

In1922, the ADGB pushed for supporting institutional changes within the SPD that would have strengthened the party‘s more conservative wing, but this effort created a backlash from the USPD members: they formed in response a left wing in the reorganized party. Who led this radical faction within the ADGB? The leaders included Robert Dissman and Siegfried Aufhäuser , who were supported by the Metal

Trade Unions and Textile Trade Unions, as well as minorities of other free unions— these unions had high membership in Saxony and Thuringia.145 Still, despite this pushback from the USPD affiliated unions, they were not able to sway the ADGB leadership positions, and besides the political resolution, extracted little. As an example, at the 1919 convention of the Construction Workers Union, KPD and USPD delegates at least succeeded in getting their calls for disbandment of the traditional structure of the industrial unions on the agenda, but nothing came of it.

Containing radicals: not supporting wildcat strikes. The ADGB and SPD also demonstrated the moderation in the threat of redistribution—and thus lower cost of repression o firms—by containing and preventing potentially massive strikes. On

March 4, 1919, the USPD and Berlin Trade Union Commission, called a general strike to enforce demands for their program, which included ―…the socialization of the entire economy‖ (Braunthal 1978, 164). The socialist-led government cabinet

―assured a strike delegation that the socialization laws would shortly be submitted to the National Assembly…‖ (Braunthal 1978, 164). But in a turnaround, the cabinet after the strike announcement ordered the Reichswehr to quell the strike. The General

145 Note these are the areas that are claimed to be ―high skill‖ areas, but in fact there was a wide distribution of types of workers; radical union leaders came from these areas.

236

Commission of the ADGB denounced the Berlin Trade Union Commission for having endorsed a political strike damaging to the economy. An ADGB representative argued that ―…socialization would best be achieved through the action of the responsible and elected national leaders‖ (Braunthal 1978, 164). And as will be noted below, the

ADGB did not pursue this course of action either.

The miners‘ union also opposed wildcat strikes and radical workers‘ goals of state ownership of mines. A February 1919 report of the union proposed the nationalization of all coal deposits, and government control over the market, prices, and the opening of new mines. The report caused a mixed reaction among union and socialist leaders, no immediate commitment from the government. Left-dominated workers‘ and soldiers‘ councils in the mining sector initiated a series of strikes and armed clashes between troops and workers. 400,000 miners supported a strike; its stated intent was to induce state ownership of the mines. However, the Mine Workers

Union did not endorse this strike, nor did they approve of funds to these local councils.

As a result, the strike collapsed swiftly, and was followed by armed clashes between the workers and government troops under the order of SPD War Minister Noske.146 In not endorsing these wildcat strikes, the general union and mining unions signaled their lack of support for extreme left-wing measures.

2.1.3 Evidence of moderation: failures of the ADGB ―socialization‖ commissions

In November 1918, the socialist-led government considered a set of commission proposals from the ADGB regarding the objective of socialization, or which industries might be overtaken by the state. But importantly, the commission‘s

146 Radical miners again endorsed a new unsanctioned strike in early April, but the Mine Workers Union again refused to endorse; as a result, the strike ended quickly.

237 proposals were never supported or implemented by the ADGB Executives. The

ADGB General Commission for example supported socialization initially, but without that much enthusiasm: even the Metal Workers journal warned that ―…the metal industry should not be socialized in times of grave political upheavals.‖147 This is evidence that the threat from workers was no longer extreme: neither the socialist party, nor the main industrialized union which had been legitimized during and after the war, were advocating full state ownership of industry. As Otto Huë declared of the mining union, which had some of the most radical workers and affiliations with the

Communist party, ―But today the question arises as to whether the time is ripe for the socialization of our extraordinarily complicated industry…This period is not the present. The socialization of such powerfully developed undertakings as the mines and iron and steel industries cannot be carried out according to the dictates of the minority but only at the right time and for the benefit of the community.‖148 The

ADGB Commission in December 1918 and January 1919 further moderated or quashed proposals from internal committees that discussed socialization policy goals.

In February 1919, the Commission report concluded that if any industries were to be nationalized, they should only be the most basic ones, and that industries related to foreign trade and the food supply would not be socialized (ADGB Commission report

1919).

147 Braunthal 1978, 163 148 Braunthal 1978, 163

238

Both the ADGB and SPD leadership in the government rejected proposals made by socialization commissions, and ended them.149 Rudolf Wissell, an ADGB leader and SPD Minister of Economic Affairs, suggested to the commission that

Germany could adopt a planned economy. But Leipart and other ADGB leaders disapproved of the plan, for they feared that would undermine the ability of unions to strike against the state. The SPD cabinet rejected the plan in 1919. The ADGB approved of the SPD cabinet rejection of the plan because they viewed it to be too far reaching: the SPD selected Robert Schmidt, who had been a former ADGB General

Commission Executive member and an opponent of a planned economy, to succeed

Wissell as Minister of Economic Affairs; Schmidt was more conservative, and had a much more limited plan granting socialization. He proposed the government would only take ownership of the electrical plants, and would compensate firms should it do so.

Still, the ADGB had to appease its left-wing supporters. After the fallout from the Kapp Putsch in 1920, the ADGB re-established a broader-based socialization

Commission. The result was a second commission with 23 members, including the previous Economics Minister Wissell of the ADGB, but also representatives from the more moderate Christian and Hirsch-Duncker federations.150 This second committee again debated socialization proposals, and again, there was much internal division. A coal industry report of July 1920 had a conservative majority support a ―gradual public

149 This is evident in the policy preferences of the leadership, as well as those of the ADGB. Overall, the government did not implement the ADGB Socialization Commission‘s recommendations, which resulted in its resignation. The majority of the Socialization Commission‘s 11 members were socialist economists. 150 Braunthal 1978, 165

239 take-over,‖ while a minority of other union representatives supported an immediate takeover. The Commission was deadlocked and ended up endorsing neither position.

The more conservative cabinet of July 1920 led by Konstantin Fehrenbach (Center) did not act on the Commission‘s recommendations. This is an example of the inability of the left-wing to agree on socializing policy. Both Socialization Commissions failed: both ADGB and SPD leaders opposed both times proposals for immediate socialization out of concern for the electoral power that might give to its radical supporters.

The actions of the ADGB indicate a gap between their rhetoric and actual policy initiatives. As discussed in Chapter 3, this indicates that the ADGB found it costly to be a substantive redistributive threat during this uncertain time. More conservative labor representatives constituted the majority of the voting bloc in the

ADGB. They professed faith in socialist ideology: in 1919 the ADGB policy declarations were that socialism was the highest form. But they opposed in practice the socialization of industry. They swayed party leaders to be more moderate. In fact, as late as 1931, Tarnow, the ADGB leader, rejected left-wing ideas to let the capitalist system crumble, and he the urged SPD and unions to protect the capitalist system.151

The ADGB was able to contain the radical left within its own ranks and the SPD, but did so while paying lip service to faith in socialism. Despite the passage of laws and emergency provisions in the Weimar Constitution (article 156) that authorized government takeover of enterprises, the Socialist-led cabinets in 1918-1920 never did.

151 Braunthal 1978, 173

240

2.2 The response to moderation in the threat: firm strategies return to repression & collapse of the ZAG

The previous sub-section described the overall reduction in the threat from workers. In this section I present evidence of the theory by confirming the observable implications of the theory: once the threat from the left moderated, firms had an incentive to revert to repression. The result was subversion of and defection from the

ZAG. Further, exogenous changes in heterogeneity made enforcing the ZAG—the maintenance of collective strategies—more difficult. In particular, cleavages between heavy industry and finished-product manufacturing became more divisive. The lower costs of repression meant that firms had a higher incentive to roll back the ZAG and to end the compulsory arbitration decrees. Also, higher heterogeneity and variance in the costs of repression, as consequence of the ending of hyperinflation, meant there was more difficulty in enforcing collective strategies.

The political instability and uncertainty, partially described above, and economic crisis in the early years of Weimar Germany, certainly affected beliefs about the stability of economic institutions. So on the one hand, perhaps the relatively short- lived nature of this collective collaborative strategy is not surprising. On the other hand, the ZAG did survive these initial crises and played a role in preventing further radicalism. As a test of the theory, it is important because its demise, and the subsequent strategies of firms, relate to changes in the key variables.

How did the ZAG operate? The ZAG functioned as both a way to structure union agreements with employer associations as well as a way to help both subsidize exports. The organization managed a network of control boards to make sure that

241 goods were sold at sufficiently high prices and for foreign currency. It managed a centralized system of export controls, subsidizing exports. It was composed of employer associations and led by prominent directors of industrial concerns, including

Hugo Stinnes, Otto Henrich, Carl Friedrich von Siemens, Ernst von Borsig, Walther

Rathenau, and Anton von Rieppel. Hans von Raumer and Jakob Reichert were highly placed functionaries in major trade associations. These associations managed with counterpart unions‘ collective bargaining agreements.

2.2.1 Changes in the threat from the left and incentives for repression

Why did the ZAG not succeed in the long-term? First, on the repression- collaboration dimension of firm strategies: not long after the adoption of Stinnes-

Legien and the ZAG, major employer organizations expressed their opposition to the union-friendly aspects of the accord. Opposition came after the ZAG constitution was drafted in December, and even as RDI preparation for its administration began. Major sector associations disliked the wage concessions to workers, and thought that unions had been given too much power. Employer associations that had been previously repressive before the war and grudgingly consented to collaborative strategies during and after the war, found reasons to undermine the ZAG once the threat from the left had passed. Reports from the membership meetings of the VDA and LNV express immediate discontent with the enforcement of the eight-hour day and a desire to rollback collective bargaining arrangements.152 The economic stabilization helped reveal fundamental distributional disagreements between workers and firms. What

152 See Association of German Employer Organizations (VDA) in December 18, 1918 in January 2, 1919 and the inaugural meeting of the RDA from Feb. 3-4, 1919 in the Mitteilungen des Deutschen Industrierates und des Kriegsausschusses der Deutschen Industrie, Feb. 8, 1919. VDA and LNV reports from 1919-1922 also discuss this issue.

242 happened during the period of stabilization? Employers ended up rebelling against the eight-hour day and reducing wages. ―Stabilization formulas‖ came, much to the displeasure of unions, in the form of longer hours and lower wages. Employer demands for ―moderation‖ from unions coincided with their increasing autonomy.

Part of the new firm opposition came to the operation the ZAG was its objection to government decrees that were issued to help enforce it. These decrees, including laws on compulsory arbitration (discussed more fully in section 3), meant that firms had greater incentives to revert to repressive strategies. Firms and unions disagreed over the implementation of the ZAG. The ADGB unions thought that the rules that oversaw bargaining and negotiations would take place should be securely grounded in voluntary agreements. Contrary to what contemporary unions might expect, in negotiating the ZAG, unions were initially opposed to government decrees about compulsory arbitration and concerned that government decrees were reversible, and thought that voluntary agreements signed by employers would be more enforceable. They put little faith in the German legislature at that point to enforce such agreements. Stinnes efficiently summarized the logic of reversibility in 1918:

―…the trade unions say very correctly that which is now introduced by the government under compulsion and which does not have any legal legitimization, does not have the same significance as a free agreement between the unions and the participating industry. The latter will remain, while the former is something that can be taken away again.‖153 Many of the social arrangements of early Weimar were based on these broad emergency powers. Both unions and firms did not like the emergency

153 The original source is Feldman, Internationale Korrespondenz, 88.

243 government decrees, but they both recognized that voluntary arrangements were insufficient. Stinnes‘ cynicism was well-founded: in fact, as discussed below, the voluntary agreements were violated by firms.

2.2.2 Changes in heterogeneity: consequences for enforceability and incentives for defection

Regarding the effect of changes in heterogeneity and the mechanism of variation in costs of enforcement, cleavages within industry mattered undermined its operation. With the moderation of the threat from workers, the common threat helped expose different interests across industries for abiding by a cross-class agreement.

These structural conditions, which actually worsened after the passing of the hyperinflation in 1923, helped undermine the ZAG. In fact, it was during the ―golden years‖ of Weimar that the ZAG collapsed. This is at odds with the normal views of crises undermining cooperation. It was the increase in wealth and recovery, and stark increase wage dispersion (see Figure 6-1) that made the structural condition of industrial heterogeneity—and the mechanism of enforcement among firms—more important. Moreover, as shown above, in this period meant costs of repression went down.

Many important firms and employer associations defected from the ZAG because they did not want to comply with the regulations regarding eight-hour days and wanted to deviate from the wage minimums; they sought to reduce wages

(Feldman 1981, 171). The ending of the occupation of the Ruhr meant that heavy industry associations wished to capitalize and not be constrained by regulations about

244 hours. Employer associations that had pursued repressive strategies were among the first to undermine the ZAG.

Evidence for this comes from cleavages within the new peak association, the

RDI. Representatives of the RDI at the annual meeting in Jena criticized the ZAG.

The harshest critiques of the ZAG came from representatives of business associations representing smaller and medium-sized enterprises (SMEs), as they were less able than larger firms of affording wage concessions to workers. SME associations complained that large firms were better able to pass costs of wage concessions on to consumers, and more resources at their disposal to engage in negotiations with the control boards.154 Another piece of the evidence of tensions was the dispute over leadership of the RDI. Some ZAG supporters proposed the appointment of the right- wing Social Democrat, August Müller, to be the business manager of the RDI.155

Müller, as noted above, was one of the original initiators of the cross-class alliance during World War I. This motion was defeated. This is a microcosm of the larger issue: pro-ZAG businesses now saw the problem as not one of winning the loyalty of labor, but persuading other firms why they should remain in the ZAG.

There was also the problem of defection on the union side. The more radical metal workers‘ union officially left the ZAG in October 1919. These union leaders supported principles of collective bargaining, but argued that the ZAG obligated too much from member unions. They disagreed with the idea that ―…the ZAG was some

154 Given that there were many SMEs in highly skilled industries, this is evidence against the claim that skilled-industry firms were more likely to favor cooperative employer associations with unions. 155 See Feldman 1981, 175, note 23.

245 kind of logical and necessary extension of the principle of collective bargaining…‖156

These labor leaders ended up reversing their position on the utility of voluntary arrangements, as they expressed in 1919 the view that legislation would be more useful than voluntary agreement with firms that could not be trusted.157 By May 1919, union leaders were of view that ZAG was ―lacking in interest to them because they will be able to push through most of their desires legislatively later on.‖158

Another example of heterogeneity of interests and problems of compliance was the divisions within the Ruhr mining industry. In 1920, Ruhr mine union leaders refused to associate with the ZAG proclamation of industry-labor leaders in Berlin that protested against the six-hour shift. They feared being accused of betrayal of working-class interests, but not because they disagreed with the proclamation. Rather, they were concerned with the response of mine owners should they complied. Ruhr mine owners similarly in 1923 tried to extend working day beyond eight-hour shifts.

This move on the part of employers helped instigate the ultimate collapse of the ZAG

(Feldman 1980).

Bücher of the RDI tried to persuade unions to stay in the ZAG. In fact, Bücher thought industry should be risk-adverse and attempt to lock in benefits secured by the

ZAG as opposed to constantly trying to undermine the organization. As he articulated,

―It would be wrong to disband it simply because the employers might win the upper hand in the near future, and he is of the opinion that the employers in such a case would, in the interest of the general good, be more moderate than the workers were

156 See the debate in the 14. Ordentlichen Generalversammlung des Deutschen Metallarbeiterverbandes, abgehalten in Stuttgart vom 13. Bis 23. Oktober 1919, 244. 157 See the Höltzendorff report, May 22, 1919, from the HAPAG Archiv. 158 See the Höltzendorff report, May 22, 1919, from the HAPAG Archiv.

246 after the Revolution. They would also realize that a reaction might otherwise come later…‖159 This sentiment nicely illustrates that there were firms for whom collaboration was still a viable strategy, because of concerns that costs of repression would rise again if they were to revert back to repression.

The massive inflation in the early years of Weimar helped employer associations and unions pass on the costs of higher wage contracts to consumers. The

ZAG collapsed as the economy stabilized and the hyperinflation passed. It was actually profitable for firms and unions during periods of high inflation, as control boards were connected such that goods were sold at high prices for foreign currency, allowing Germany to maintain an export advantage. As the German economy recovered, firms had less of an incentive to maintain this system.

The collapse of the ZAG illustrates that once conditions improved, increasing the difficulty of enforcing both collaborative and collective contracts, such cross-class strategies were not tenable. There is an interesting tension illustrated in the previous sections in terms of the strategic problem faced by unions: with the passing of the revolution, the costs of repression fell—and they demonstrated considerable moderation in their policy positions and ability to sanction defectors. But with that also came an incentive for firms to reduce their level of cooperation. Once ―good times‖ returned, distributional conflicts between firms and workers and among firms helped undermine this peak association. The next sections discuss, though how collective repression was maintained in the remainder of the Weimar era.

Section 3: Repressive Strategies in Weimar Germany

159 The original source is from the meeting of the ZAG directors, November 27, 1923, from the Zentrales Staatsarchiv der DDR, Abt. 1, Potsdam, ZAG, Nr. 7, Bd. 31/1. Br. 78

247

In this section I present evidence of the theory by describing the reversion to repressive strategies after the collapse of the ZAG. Employer associations and individual firms undermined the efforts of existing collaborative associations and engaged in successful lockouts, waging conflict against unions and rolling back the agreements they had grudgingly signed when the threat was the highest. As discussed in chapter 2, this repressive strategy had a number of elements: use of lockouts; blacklists; dividing workers by only supporting firm-specific unions; support of political parties that favored their policies. After the collapse of the ZAG, employer associations pursued all these strategies. I identify the following categories of repressive strategies pursued and discuss each of them: directly attacking unions and pro-union legislation; undermining collaborative employer associations; ―locking in‖ other collaborative employer associations to tolerate or support repressive strategies via collective agreements and cartels; and contributing to political parties to repress firms.

Associations representing Ruhr heavy industry (RHI) were particularly influential and engaged in many repressive tactics, despite the Weimar legal protections of workers. But other associations that had manufacturing goods and finished-goods also engaged in repressive strategies. These tactics included direct challenges to unions in the form of demanding longer hour shifts and unilateral wage reductions. Additionally, these employer associations lobbied for anti-labor political parties. The RHI and other associations scuttled later attempts at collaborative strategies; once enforcement conditions fell, they switched strategies to repressive.

After the collapse of the ZAG, firms continued their repressive strategies and tried to

248 further undermine collective bargaining agreements. They pursued an explicit economic strategy against unions and electoral strategy against the SPD.

3.1 Return to repression: direct assault on hours, working rights, and arbitration

The six most influential Ruhr concerns were: Vereinigte Stahlwerke,

Guttehoffnungshuette (GHH), Krupp, Hoesch, Klöckner, and Mannesmann. Their leaders complained frequently about rise in ―political‖ costs of wages and social insurance (even though real wages throughout the 1920s remained nearly constant). In addition to being influential leaders in the Weimar era repressive employer associations, they were also part of the ―Ruhrlade‖ that was a conglomeration of representatives of major employer associations that tried to undermine the RDI. In

October 1928, one month before the large Ruhr lockout that I discuss below, one average worker week‘s wage (with taxes and social insurance rates deducted), was worth only two percent more than before the war (Weisbrod 1979, 250). The ratio of labor to total costs remained at about one-third. (Worker productivity increased, but purchasing power lagged far behind progress in productivity). Employers for these concerns had effectively restrained wages for their workers, but, the industrial concerns still viewed unionism as a threat. The priorities of heavy industry were the abrogation of the eight-hour day (the working regulation of the Arbeitszeitverordnung in December 1923). The unifying ideology of these concerns was ―restraining social democratic‖ influence.

These firms or concerns mentioned above faced the issue of whether to continue the collective strategies by consolidating existing employer organizations and further centralizing the administration. Should the LNV, the NWG, the employers‘

249 association of the NWG (which included machine builders as well as iron and steel producers), the Zechenverband (employers‘ association of the Ruhr coal industry) all form one formal association? Once again, heterogeneity and cleavages between sectors mattered. Iron consuming and manufacturing industries resisted these centralization efforts because they rightly feared that heavy industry wanted to have sole leadership of the Ruhr. But it is not clear how much contestation of forming a massive LNV-type organization there was. As discussed below, they were able to from cartel arrangements with other raw industry consuming associations.

Who initiated the LNV‘s renewed attack on collaborative strategies with labor?

Paul Reusch of the GHH, Fritz Thyssen, and Poensgen from the Vereinigte Stahlwerke were the leaders of the main cartels and employer organizations who wanted to enlarge the LNV membership and make it more influential from Düsseldorf; the NWG stayed separate, but assisted in this effort.160 The LNV and related associations were also opposed to collective bargaining and the compulsory arbitration system; and favored a return to direct employer-worker relations and ―yellow unions.‖

3.1.1 Early repressive successes in rights and hours and wages

The LNV and NRG wanted to dismantle the eight-hour day and the system of collective bargaining. Both associations favored twelve-hour as opposed to eight-hour shifts. They had an early success with the Arbeitszeitverordnung of December 1923, which revoked the eight-hour day achievement. The government replaced it with

―voluntary‖ agreements under Labor Minister Heinrich Brauns, which effectively

160 Weisbrod 1978; 1979

250 meant firms could adopt twelve-hour shifts.161 Member firms of these associations took advantage of this law and similar decrees. Firms for example reduced the number of blast-furnace workers favored by the decree of January 1925 and prevented the Working Hours Emergency Act of April 1927 from declaring a return to the eight- hour day. Once they had been successful in this initial repeal, they fought hard to keep it in place.

3.1.2 Fighting the conciliation system

To help enforce the ZAG collective bargaining procedures, the government adopted conciliation boards in the early 1920s. These boards made binding decisions regarding industrial or contract disputes. Unions thought they could benefit from arbitrators, whereas employers uniformly wanted nonbinding awards, essentially rendering the decisions of such boards unenforceable. By 1920, labor leaders favored granting joint conciliation boards the authority to make binding awards—compulsory arbitration was a way to avoid costly strikes. Unions calculated that otherwise they would be the weaker partner, and that net wage decisions from the conciliation boards would benefit unions. The Center Party and SPD ministers of labor defended the system of compulsory arbitration, but the repressive employer associations like the

LNV wanted to dismantle it. SPD ministers viewed it as essential (―compulsory arbitration merely trimmed the peaks of the ups and downs‖ (Wesibrod 1978, 258).

Compulsory arbitration in their view was a cornerstone of collective labor law and was a natural product of securing equal rights for employers and workers and collective

161 In exchange, Brauns negotiated for unions some maintenance of collective bargaining institutions. Unions objected to this deal.

251 bargaining. The SPD Minister of Labor Wissell, thought it was necessary, as well as consistent with article 165 of the Weimar constitution.

Unions did not initially favor such policies. Chancellor Bauer in 1920 initiated discussion of a permanent compulsory arbitration system. The Labor Ministry drafted a proposal but pressure from the firms discussed above led the Ministry to add a clause that would limit awards for certain strikes. Unions rejected this counter-proposal, and the Labor Ministry formed a committee with all interested parties. In 1921, the Labor

Ministry prepared a bill based on these proposals; it reached the Reichstag in March

1922, but no legislative action was taken. In October 1923, the Stresemann government bypassed the legislative process and officially decreed a conciliation and arbitration system—the decree created boards to be headed by government-appointed arbitration officials. Under the decree, the Labor Ministry was assured of a decisive influence in labor disputes—the arbitration board‘s jurisdiction included everything from industry-wide agreements to individual shops. The policy was that ―If the parties could not reach an accord, then the arbitrator would act as a voluntary mediator…if he

[the arbitrator] failed he could declare the board‘s award binding upon an entire trade or industry, even if one party did not accept the decision‖ (Braunthal 1978, 151).

Firms and unions thus had to make calculations of what the arbitrator could decide if they did not settle on an agreement. By 1929, 52 percent of all collective agreements were concluded on the basis of awards from the arbitration process.

The circumstances and effects of such legislation are important because they provide evidence that employer associations had reverted to repression. They resorted to attacking this apparatus that tried to keep collaboration in place. This is most aptly

252 displayed in the lockouts in response to such decisions. In April 1927, the main employers‘ associations of heavy industry, which also encompassed small and middle- sized firms in machine building, started an offensive strategy. A labor court declared in July 1927 a mandatory reduction in hours in steel and rolling mills was to be implemented in January 1928. This would have forced firms to give up the twelve- hour shift they had implemented. This declaration was going to affect the negotiation of new contracts; the negotiations were to be concluded in December 1927. In response to the decree, firms accumulated a fund that equaled a pay rise of about two percent. The major firms in iron and steel, including those outside the Ruhr district, agreed to shut down their plants if the decree took effect. They also forced smaller firms within the tariff district to follow suit, because a central committee of the association was to decide the length of the lockout and the distribution of the fund.

When the firms officially announced the lockout (legally four weeks before the deadline in December 1927), this essentially forced the district arbitrator and Labor

Minister Brauns to limit the rise in wages to a minimal margin and allow for a new court decree. This new decree left large sections of the working force on the twelve- hour shift right through 1928. The actual lockout was called off by the associations once Labor Minister Brauns brokered an agreement and agreed to limit the wage increase.

The same repressive logic returned in the October 1928 Ruhr iron dispute, a key example of successful collective repression. This was a case of unions demanding higher wages, not liking employer counteroffer, and in response, employers locked out

250,000 metal workers. Metal workers unions of three labor confederations demanded

253 a 15 pfennig per hour wage increase, which was rejected by the all the metal-working employer associations; the union‘s revised request of 12 pfennig per hour was also rejected). The arbitrator awarded 6 pfennigs per hour to several categories of workers.

The unions requested that it be made binding (reluctantly accepting it). Minister of

Labor Wissell agreed, but the employer association rejected its legality, arguing that the industry would not be able to tolerate such a wage hike. On November 1, 1928 the association locked out more than 250,000 metal workers. The employer association also launched appeals through labor courts: it lost the Land level appeal, but in

January 1929, the Reich Labor Court sustained objections, and accepted the employer association‘s claim. This repressive measure preemptively threw out workers in anticipation of the arbitration decision. It was a direct challenge to the SPD leader

Hermann Mueller. This pre-emptive locking out of workers was the most repressive strategy: employer associations could have refused to pay the wage increase; they could have closed the plant with the legal four weeks‘ warning, but they chose the most damaging outcome for workers. This was a major defeat for unions, as well as a weakening of the principle of collective bargaining.162

The 1928 lockout was expensive: it cost half of Krupps‘ 1927-28 profit, so the

LNV and other firms sought to devise another solution in response to the threat from the left, as the cost of this repressive strategy was quite high. It was part of a general effort of a lobbying campaign against the arbitration boards. The SPD was proposing increased national control of heavy industry; in response the LNV and related

162 The legislature‘s response to the lockout was to partially support the workers: the right-wing party the DVP had to consent to the government support as a way to prevent other pro-worker measures from being instigated, including further intervention in heavy industry (Weisbrod 1979, 259).

254 associations sent a memo to SPD Interior Minister Carl Serving that indicated their readiness to the cabinet accept a unique exclusive arbitration at the hands of Severing.

The intent was to show that there existed some policy options that they could be amenable to, in trying to get unions to concede (Vogt 1970). Severing eventually brokered an agreement that reduced the wage increase in the industry to about one percent over 18 months. Employer associations still, however, called for the entire system of compulsory arbitration and collective bargaining overturned. (In fact,

Reusch of the GHH internally offered to resign after Severing‘s decision, given that he thought the brokered agreement was a failure).

3.2 Repression: undermining attempts at collaborative strategies and other associations

After the collapse of the ZAG‘s official peak and sector-level managed collective bargaining arrangements, employer associations continued their assault on the basic tenets of the Stinnes-Legien accord. In this section I demonstrate these associations that were engaged in repression and also undermined collaborative employer associations. In particular, they tried to undercut the positions of the RDI.

The LNV changed from being an important regional predecessor, to the regional rival of the Weimar-era peak association, the RDI. The LNV during the Weimar era, especially after the collapse of the ZAG, became the main rivals to the RDI, which was represented by Duisberg. The LNV viewed the RDI as speaking for, even after the collapse of the ZAG, pro-union positions. The inter-war period in Germany was characterized by a rivalry between the two organizations (Zunkel 1973). The standard accounts of industrial institutions in political science ignore these organizations, but

255 they were successful in engaging in repression. One of their main objectives was to provide coalition governments with alternative sources of information and impose costs on collaborative employer associations for voicing support for pro-union legislation. These efforts attained some success despite the presence of SPD led coalition governments in 1928.

Employer associations, in particular, the LNV and the NWG, engaged in a strategy of undermining the RDI to set up a credible opposition body with which to persuade firms and transmit information to the government, as well as support preferred candidates and parties. Its goal though was not to make itself an object of popular targeting, and so engaged in ―behind the scenes‖ politicking to counter positions of the RDI. The LNV disagreed with the industrial relations positions of some the constituent associations of the RDI, including the more moderate institutions such as the chemical and electro-technical industries. The goal of the LNV was not necessarily to destroy the RDI, but rather to undermine its authority as the ―sole‖ voice of business, and to lobby the government with its own policy suggestions. Reusch in fact advocated establishment of a more authoritarian economic ministry, Wirtschafts- und Spardiktator, who would have an extra-parliamentary authority to give emergency economic decrees by article 48 of the Weimar constitution.

How did the LNV do this? The LNV would meet to directly respond to RDI meetings, and the content of these were communicated to the relevant economic departments in the government. These meetings were designed to undermine SPD legislative proposals, and were during the Chancellorships of the Brüning, von Papen, and Schleicher governments. An important example of this counter-strategy and the

256 conflict between these factions of capital occurred in 1926. Paul Silverberg, a prominent Cologne industrialist, addressed the RDI members at a meeting in Dresden in September 1926 on behalf of the brown coal industry associations. He argued that the major industrial firms were going to remain loyal to the Weimar Republic, and offered his firm‘s cooperation with the SPD Reichstag Fraktion.163 Silverberg and

Klöckner proposed the renewal of a collaborative institution like the ZAG. This overture in 1926, after the official ending of the ZAG, represented an opportunity for a re-institutionalization of relations between unions and employer associations. Further, this attempt at reviving institutional coordination would have occurred under a coalition government that included the SPD, increasing its chances for success.

But in response to Silverberg‘s speech, the LNV held a meeting one month later. At the meeting, Thyssen spoke for the LNV and vociferously refuted Silverberg.

The LNV representatives rejected this extension of compromise and said no member firms would cooperate with Silverberg‘s proposals (Thyssen 1926).164 Thyssen reiterated the LNV‘s commitment to fighting social democracy and socialist unions.

Reusch and Vögler at this meeting supported Thyssen‘s speech and led the LNV‘s efforts to scuttle Silverberg‘s proposal (Stegmann 1974). As an ―alternative,‖ they advocated strengthening the Werkgemeinscahft (―yellow company unions‖ which had been partially banned in the Stinnes-Legien agreement). Duisberg, in response to intransigence of the LNV, wanted to avoid potential controversy—he did not want the

RDI to be embroiled in a battle with the LNV and its related associations—and so he renounced responsibility on the part of the RDI for Silverberg‘s proposal. This is one

163 This overture was made in 1926. See Silverberg (1926). 164 Thyssen gave this speech at a meeting of the LNV on October 1, 1926.

257 key example of the LNV trying to undermine the RDI‘s proclamation of support for collective collaborative strategies (Stegmann 1974).165

The same personnel also played the role of spoiling overtures of the RDI for a revival of the ZAG again 1930. Repressive employer associations including the LNV responded to Chancellor‘s Bruning‘s speech to the RDI in November 1930 with a counter-protest meeting in June 1931. These are two moments where employer associations with collaborative preferences attempted but failed to form peak-level management of collective bargaining, but could not because the economy‘s variance in the costs of repression were too high. Existing repressive employer associations preferred to keep their own strategies.

3.3 Repression: co-opting the collaborative employer associations

The evidence of collective repressive strategies presented in sections 3.1 and

3.2 only partially addresses the question of how repressive employer associations were able to directly challenge collaborative employer associations. In this section I present evidence of the relative weight of repressive versus collaborative organizations in terms of affecting actual industrial outcomes. I provide evidence that repressive organizations were able, through forming collective institutions that spanned the major manufacturing goods industries, able to constrain the more pro-worker/collaborative

165 Another example of how the LNV tried to undermine the RDI is in the negotiations over the Young Plan in 1929. Albert Vögler, general director of the Vereinigte Stahlwerke, renounced his post as German delegate to the negotiations to be held in Paris over the Young Plan. The RDI remained politically silent on these issues and did not want to divide its membership regarding choice of a successor. As a result, heavy industry voiced support for Vögler and implicitly against the RDI chief of staff Ludwig Kastl, who had taken Vögler‘s place at these negotiations. This position was taken at a July 1929 meeting of the LNV, which adopted a resolution that showed the ―fullest appreciation of Vögler‘s decision and cleared heavy industry from all responsibility for the fulfillment of the Young Plan.‖ The goal was to undermine consensus within the RDI, and to show that on major initiatives, the LNV would take a different stance from the RDI.

258 strategies of other organizations, further undermining the RDI and other moderate employer associations. They used a set of institutions that took advantage of the external economic environment and promoted cartelization to do so.

In the post-ZAG era, repressive employer associations used their economic position and influence over production and processing of raw materials to take advantage of consumer and finished goods industries in new cartel arrangements. The effect of these agreements was to restrain the degree to which consumer industry associations and their member firms could effectively abide by any collective bargaining agreements, and after the ZAG‘s collapse, these agreements undercut their engagement in collaborative strategies. This dimension of collective action among firms in product markets, encapsulated in inter-industry cartels, helped constrain collaborative strategies of other associations.

How did this economic bargain between repressive and collaborative employer associations work? The cooperative deal between these associations was the following: the associations of finished-goods industries would purchase the cartelized inputs from heavy industry at higher cartelized prices. But these industries‘ firms also would then receive in return export subsidies paid for by the firms from repressive associations; the subsidies guaranteed the final products would sell at world market prices. Finishing industry firms and their associations (particularly machine-building firms) had been essentially cut off from cheaper French raw material, and heavy industry firms benefited in the mid 1920s were participants in international cartels.

The net price result of these policies was higher prices for the finished-industry goods for domestic consumers, as the costs of subsidies were passed on to domestic

259 consumers. But another important result was that finished-goods industries were committed to heavy industry‘s anti-union politics in the form of supporting lockouts.

This kind of collective strategy that ―locked in‖ collaborative associations was characterized by the AVI agreement. Through these agreements, finished industries; they were prevented from favoring collaborative or more worker friendly legislation regarding compulsory arbitration. The AVI agreement demonstrates how collective action and incentives to repress or prevent collaboration are connected. As part of these cartels, iron consuming industries had less market influence and were put in a position where they had to pass on prices to consumers and have no flexibility in dealing with unions.

The AVI agreement consisted of three agreements between December 1924 and June 1926. The name comes from the Working Community of the Iron Finishing

Industry (Arbeitsgemeinschaft der Eisen Verarbeitenden Industrie, AVI), which was an industry association that represented branches of machine building, electro industry, iron and steel wares, and steel construction. The basic feature of the AVI agreement was that the iron industry assumed the legal obligation to provide raw steel and other inputs for finished export products at world market prices. The agreement made possible the re-establishment and extension of the iron industry‘s control over the domestic market. The agreement would allow for inter-industry cartel and manipulation of the domestic market. The effective consolidation of the domestic manufacturing allowed for heavy industry to adopt a more repressive strategy against workers between 1927 and 1930.

260

What explains this inter-industry agreement? The historical background was the attempt of the largest German iron and steel concerns and finished goods firms to collectively act regarding the passing of a post-inflationary period and exogenous changes in the supply of raw materials, and foreign competition. Starting in

September 1923, imports from producers in the Lorraine and Saar areas were entering

Germany under tariff-free provisions of the Versailles Treaty and depressing domestic iron prices. At this point German raw materials concerns and producers had to decide how to react to new post-inflationary conditions (Nocken 1973). The percentage of syndicated output in this industry went from pre-war level of 43 percent to 90 percent in 1926; lots of syndicates formed in this period. The new huge steel trust of the

United Steel Works (Vereinigte Stahlwerke, VS) in 1926 had 43 percent of the quota of the crude steel cartel (Rohstahlgemeinschaft, RG). The threat to German firms came from Lorraine steel firms and Belgian and Luxemburg producers. The

Versailles Treaty‘s tariff-free provisions were supposed to expire on January 10, 1925.

During the Imperial Germany era, these large firms would have just successfully lobbied for tariffs on competing imported raw materials. But during this time period

Weimar, the finishing industry associations opposed such tariffs.

Who represented these finishing industries? The most important association was the Association of German Machine Building Firms (Verein Deutscher

Maschinenbau-Anstalten,VDMA) which ended up administering the AVI. The

VDMA was also an important member association organized within the RDI. The

AVI agreement meant that the VDMA would support the other employer associations‘ call for a low tariff on competing iron materials, if the iron industry agreed to

261 reimburse the manufacturing firms for the difference between the world and domestic price for the amount of iron used in an exported product. Under the agreement, an exporting manufacturer could use the world iron price as the basis for calculating and negotiating input costs, and therefore would not be at a relative disadvantage to competing foreign firms. The finishing-product industries under the AVI agreement would at least not oppose the iron tariff: as a result, the iron tariff was lower than what the iron producers wanted, but preferable to no tariff at all (Nocken 1973, 697). These different business associations essentially worked out agreements over tariffs and cartel pricing policies without government interference.

The important point is that this inter-industry agreement meant that repressive strategies were less costly, and, collaborative employer associations were locked into supporting them. The AVI agreement allowed for renewed attention from business associations in iron and steel on compulsory arbitration. As discussed in sections 3a and 3b, they opposed government social policy and unions‘ demands regarding wages and hours, as well as the government-enforced arbitration decisions. The finishing industry associations had a less extreme view of repressive strategies: in their view the collective bargaining and compulsory arbitration systems had prevented long drawn- out strikes (note this logic is based on one of avoiding costly disputes, not joint skill provision). Importantly, the system had also protected financially weaker firms that otherwise might have had to capitulate to union demands.

But iron producers were able to use the AVI alliance to secure the support of finishing industry against the arbitration system. How did it do this? Under the agreement, they were able to automatically raise the cartel price for iron and steel in

262 response to any arbitration decision which raised costs through higher wages, shorter hours, or both. Signatories of the AVI were obligated to accept price increases as valid, and thus these industries bound to the agreement were economically compelled to accompany repressive employer associations in criticizing government arbitration decisions. This united front also mattered for other pricing policies of the government: for example, in January 1928, the AVI and iron industry jointly confronted government industry against cartel supervision.

Another example of the effect of the inter-industry agreement on repression was that during the massive Ruhr lockout in 1928 described above, the finishing industry associations supported the iron industry in their preemptive repression of workers. The finishing industry also used its contacts in the left-liberal media to support the repressive strategies of employer associations. Machine building associations put pro-lockout articles in the Berliner Tageblatt, Vossische Zeitung,

Frankfurter Zeitung, which had been more pro-worker in the past. The AVI agreement allowed repressive employer associations to cartelize the industry and attack pro- worker social policies more directly.

Employer associations that initiated the AVI could also use to keep collective strategies afloat. For example, when the coal industry association representatives of the Zechen-Verband threatened to resign from the RDI in 1929 in a dispute over the

Young Plan and thus cause a potential split of the RDI, the iron industry association threatened to use the AVI to undermine the Zechen-Verband. This forced the coal industry representatives to remain in the RDI (Nocken 1973, 703). This is evidence of

263 how industries tried to sanction and deter potential defectors to maintain repressive strategies and prevent a return to collaborative strategies.

The AVI agreement had negative economic consequences: the result was the costs were passed on to consumers in the form of domestic prices and dampened domestic demand. Essentially higher domestic prices were being used to subsidize lower export prices for the entire iron industry. The more moderate labor-relations views of finishing industry became weakened as the industry was actually politically weaker on labor regulations. The consequence was to make the AVI more dependent on repressive employer associations. Large small and medium sized finishing firms ended up playing a smaller role in labor politics. It is fair to say that the exigencies of the collective strategy meant that unions lost a true ―cross-class‖ alliance. The politics of this are discussed below.166

3.4 .Repression: supporting political parties

In this sub-section I present evidence of another form of repressive strategy.

This involves firms bearing some costs to support political parties that challenged SPD and union policy positions. In fact, most political lobbying activity of firms was directed toward such political parties. This evidence is consistent with the theory. It shows that firms invested some amount in the state to get preferred anti-labor policies.

This sub-section concludes with evidence that further confirms the mechanism of heterogeneity: problems of enforcement within a repressive capitalist

166 During the depression years of 1930-1933, these artificially high iron prices were detrimental to Brüning‘s deflationary policies. Because of the AVI-agreement, the government was unable to convince the finishing-industries to artificially lower their iron prices—they could not simply de- legitimize the AVI-agreement and cartels (their attempts to lower prices by decree were opposed by other employer associations. These associations essentially broke ties with the Brüning government (discussed below) with the declaration on December 9, 1931 that called for a 10 percent wage reduction along with a 10 percent reduction of all prices bound by agreements such as cartels.

264 faction made pursuing labor market goals—through the strategy of state lobbying— difficult. These problems occurred even though different employer associations and industrialists shared joint interests in continuing the undoing of Weimar-era labor reforms.

3.4.1 Mechanics of political repression

Employer associations‘ political strategies were influencing these parties to support their positions against collective bargaining and the arbitration courts. To use that influence—a strategy of repression—they used funds. The major industrialists formed a group that represented a break from the previous system of collecting and disbursing political money of big business. The two parties that most represented the views of repressive capital and employer associations were the German People‘s Party

(Deutsche Volkespartei, DVP) and the German National People‘s Party (DNVP). The right-wing faction of the DVP was particularly sympathetic to the views of heavy industry. In the 1890s, as discussed in chapter 3, campaign funds had been managed through a centralized agency established under the auspices of CDI—but the formation of the RDI made it difficult to continue channeling such funds to campaigns because the RDI represented a wider variety of industries (including finished manufacturing and heavy industry).

In 1924, coal and iron industries under the LNV and related associations wanted to restore the centralization of campaign funds to support favored candidates.

In 1927, this system ran into opposition: Alfred Hugenberg a manufacturing industrialist, wanted to defer less to his fellow industrialists, particularly those in the influential associations. He was focused on winning the party leadership of the

265

DNVP, and wanted to attain control the funds and subsidies extended to DNVP by these firms. The existing DNVP leader Count Kuno von Westarp asked Reusch and

Springorum for assistance: they supplied Westarp with the funds from a new tax on iron firms (Turner 1970).

To institutionalize the above political support for the DNVP, the employer association leaders from the LNV and Northwest Group decided on March 5, 1928 to manage political funds through a source called the Eisenseite. Springorum became the treasurer. A parallel treasury organized by the coal industry (the Kohlenseite) was run by Ernst Brandi, who was head of the Mining Association (Verein für die

Bergbaulichen Interessen, BV). The actual amount of electoral funds managed by the

Ruhrlade is unknown. Turner estimates that the Eisenseite levy produced a yearly yield of approximately 620,000 marks for each year 1928-1930. The range of total funds (Konto Wirtschaftshilfe) during 1928-30 around 1,200,000 to 1,500,000 marks; thus the contributions of the iron industry based employer associations amounted to 40 to 50 percent of the total. In 1932, though, the BV desired independence and assumed management of the Kohlenseite‘s funds independently.

What did these firms and associations do with these funds? They subsidized other anti-union organizations and publications; contributed to political parties and politicians; and supported headquarters of the parties favored by the groups. Some of the prominent industrialists and association leaders were officially affiliated with political parties. Krupp von Bohlen, Silverberg, and Vögler were all supporters of the

DVP; Springorum and Thyssen were affiliated with the DNVP; and Klöckner was a member of Center Party. Reusch had been with the DVP until 1923, until he resigned

266 in protest over Chancellor Stresemann‘s decision to terminate passive resistance in the occupied Ruhr. These industrialists, as discussed above, viewed politics as one of socialist interests versus non-socialist interests. For the 1928 Reichstag elections, they allocated their political funds among those parties they regarded as bourgeois

(bürgerlich) and sufficiently anti-worker: the DVP, DNVP, and Center party (Turner

1970; Heidenheimer and Langdon 1968).

I present evidence of the political strategy used by firms to get the government to support its repressive strategies. The DNVP received 200,000 marks from the

Ruhrlade (supported by funds from the mentioned associations) for the 1928 election campaign.167 After the 1928 election, the Ruhrlade gave monthly subsidies of 5,000 marks to the DNVP and DVP, as well as 3,000 marks going to the Bavarian People‘s

Party (Bayerische Volkspartei, BVP) (Turner 1970, 207). For the 1928 election, the

DNVP received 487,000 marks in total for campaign contributions, so the Rurhlade‘s

200,000 was a substantial portion—about 40 percent—of the party‘s total campaign contributions. Springorum explicitly brought up the contributions in his discussions with DNVP party leader Count Westarp—in exchange for these funds, he wanted industrial and associational decision-making influence regarding the ranking of nominees on the party‘s lists of candidates for the next elections. Springorum however was unsuccessful: he had extracted from the DNVP leadership a commitment to place nominees favored by the repressive employer associations higher on the electoral list than the more extremist Hugenberg‘s preferred choices. However,

Westarp failed, and Hugenberg‘s preferred nominees ranked higher on the final list.

167 Turner speculates that the DVP and Center Party received similar amount.

267

3.4.2 Evidence of sanctioning defectors: divisions with the repressive faction

The problem of dissension within the faction of repressive employer associations illustrates the mechanism of heterogeneity operating within a repressive faction. Hugenberg became chairman of the DNVP in October 1928. This was in fact a defeat for the industrialists representing most repressive employer associations: they preferred the relatively moderate leadership of Westarp, who had led the DNVP into a coalition government in 1927-28. Of course, still, these industrialists and associations objected to the coalition government‘s acceptance of laws regulating the length of industrial workday, as well as the establishment of an unemployment insurance program based in part on employer contributions. But, they still wanted the DNVP to work within the confines of the parliamentary system. They were against any result that would minimize the influence of the DNVP because they were concerned that under Hugenberg, the DNVP would become too obstructionist. Reusch himself encouraged industrial opposition to Hugenberg‘s bid for DNVP chairmanship

(Stegmann 1976).

In 1930, the Reichstag dissolved, and in July of that year there were new elections scheduled for September. The repressive employer associations and industrialists set as their goal to reduce the fragmentation of non-socialist parties with the goal of creating a unified party that could challenge the SPD within the parliamentary system. Their goal was a ―bürgerliche Einheitspartei,‖ which partially came out of concern about Hugenberg‘s potentially radical leadership of the DNVP.

At a July 28, 1930 meeting of the most influential industrialists in the repressive faction, the members resolved to support all parties from the Hugenberg-led DNVP on

268 the right to the former Democratic Party (Staatspartei) on the left, in an effort to confront the SPD while neutralizing Hugenberg‘s influence. They resolved to aid potentially secessionist DNVP groups only if they combined with each other and with the older middle parties to present ‖joint lists‖ of electoral candidates to challenge

Hugenberg. At this point, Reichart, who was head of the national headquarters of the

VdESI in Berlin, was heavy industry‘s leading spokesman in the DNVP Reichstag

Fraktion. He had opposed Hugenberg‘s candidacy and resigned in 1930. The goal of this new anti-socialist effort was to keep open the possibility of cooperation with the moderate groups and keep alive the idea of bourgeois cooperation as a viable political strategy—the choice between submitting to the SPD and extremist choice of undoing

Weimar democracy. The idea was to force compromises from Hugenberg and get the

DNVP to cooperate.

This attempt to moderate the DNVP and simultaneous create a grand anti- socialist coalition failed. The industrialists Blank, Springorum, and Baron

Wilmowsky met with DNVP spokesmen, and tried to induce it to change its position with the promise of regular campaign subsidies. Blank persuaded the leaders of the

DVP, the Wirtschaftspartei, and the new Konservative Volkspartei to draft a joint campaign declaration: the Wirtschaftspartei disavowed the declaration upon learning about it. Still, these industrialists and their associations contributed funds to parties that were reliably bourgeois, but this time without extracting any promises from the parties.168 These industrialists did believe at first in the Brüning government: they expected abandonment of compulsory arbitration and binding state arbitration of wage

168 Turner speculates money was given to them in 1928, also to the conservative Volkspartei.

269 disputes, and the ending of the ―inviolability of wage contracts.‖ Repressive employer associations wanted the Brüning government to disavow the SPD and battle against the unions, and to do away not only with compulsory state arbitration, but with the whole system of collective bargaining as well. Brüning did meet with them, but he also depended on the SPD for electoral support, who wanted to maintain existing laws.

By summer 1931, these associations and firms realized Brüning would not waver, and by autumn of 1931, they openly opposed the government coalition.

For the spring 1932 elections, the repressive faction was still divided. Reusch and Krupp von Bohlen were against Hugenberg: he was their obstacle to cooperation among non-socialist parties. They proposed that the associations and firms should withhold financial contributions from the DNVP until Hugenberg was replaced by a more moderate figure who could credibly unite parties against the SPD. But

Springorum and Vögler resisted this proposal; they thought such intervention would only cause more internal party chaos. As a result, the Kohlenseite raised and disbursed its levy separately; Brandi, the head of the mining association, had long wanted more independence from the iron and steel-centered industrialists who were influential. These associations ended up not coordinating electoral funds for the 1932 electoral campaign: they let individual firms decide whether to contribute to the industrial fund, sponsored by Duisberg.169 This, along with other policy disagreements, helped divide the repressive faction of capital. It should be noted, of

169 Another non-political cause for internal disagreement among the Ruhrlade occurred in July 1932: in mid-June the Reich had purchased a majority of the stock of the Gelsenkirchener Bergwerks-AG, and Gelsenkirchen stock carried controlling interest in the Stahlverein. Reusch was upset with the Stahlverein and attacked Poensgen, Thyssen, and Vögler for ―acquiescing in the establishment of a massive state beachhead in heavy industry‖ (Turner 1970).

270 course, that one reason why these associations might have failed in their negotiations was that they could not deliver sizeable blocks of votes to the DNVP or other related parties. This evidence illustrates that divisions within the repressive faction—which already had been quite successful with the collapse of the ZAG and anti-labor consequences of the AVI agreement—prevented it from attaining all of its objections.

Even this faction had more extremist anti-democratic ones which it could not adequately sanction and control.

Section 4: Implications and Alternative Views

Before turning to interpretation of the evidence from this time period, I summarize the main trend in changes in firm strategies in Germany. Figure 6-1 displays the change in firm strategies, going back to Imperial Germany and continuing through the inter-war period. The template for Figure is used as well for discussion of other countries in Chapters 7 and 8. The solid and dashed vertical lines denote the years in which changes in major firm strategies occurred. The x-axis displays decades.

The y-axes display values of the main independent variables. To proxy the threat from workers and costs of repression I use data on union density and percentage of votes for all left-wing parties. To proxy for industrial heterogeneity and mechanism of variance in the costs of repression, I use data on wage dispersion between skilled and unskilled workers and where data exist.170

As Chapter 4 and the beginning of this chapter discussed, the first major shift in firm strategies occurred when the two major peak employer associations, (the

170 The wage dispersion data are from Edward Anderson and Concha Betrán. The support for left-wing parties comes from (Bartolini 2000), and the data from union density comes from (Bain and Price 1980).

271

Verein Deutscher Arbeitgeberverband and Hauptstelle Deutscher

Arbeitgeberverband) merged to form a national organization in 1913. This new organization was the Association of German Employer Associations (Verinigung der

Deutschen Arbeitgeberverband, VDA). When the electoral support for the SPD substantially increased, as in 1912, there was an incentive to switch to more collaborative strategies. In 1913 these two peak associations—the HA and the VA— unified to form the Federation of German Employers Associations (VDA). By 1913, the VDA‘s constituent employer organizations saw over 10,000 collective agreements covering close to 1.4 million workers. Key leaders from within the VDA and its previous constituent organizations would go on to play an important role in the influential 1918 Stinnes-Legien agreement, the first formal codification of a nation- wide agreement between much of organized capital and labor. What is notable is that the most virulent anti-labor association in the early twentieth century ended up helping to administer a giant institution that essentially oversaw collective agreements—an important moderation of its previous position.

When the threat from workers and the costs of repression rapidly increased towards the end of WWI, firms and unions preemptively formed a cross-class collaboration, encapsulated in the famous Stinnes-Legien agreement of 1918. This agreement was a formal shift from previously repressive to a collective collaborative strategy of firms. The agreement authorized the first major institutions in Germany that legitimized unions as a bargaining partner in the workplace and guaranteed minimal working conditions. This shift is displayed in the next line on the figure.

272

But the landmark collective collaborative strategy was not sustainable; this change is indicated in the last vertical line in Figure 6-1. The ZAG collapsed in 1925 due to exogenous changes in the enforceability of the agreement. After the passing of the economic crisis and hyperinflation, as well as the lower threat posed by the SPD and ADGB in terms of their redistribution demands, firms which had prior to World

War I engaged in collective repressive strategies had an incentive to renege and return to such strategies. Many employer associations and member firms directly renounced the eight-hour day and re-activated repressive strategies. In particular, employer associations in manufacturing lobbied heavily against the compulsory arbitration system that the government had decreed to help resolve industrial disputes when intra-

ZAG procedures were insufficient. In 1930, when moderate leaders of the RDI wanted to re-introduce collective collaborative strategies, the same associations lobbied against it threatened to veto such an idea. Besides the defection of employer associations, more extremist unions, like the Berlin Metalworkers‘ Union, left the

ZAG as well. Once economic growth increased again in the mid 1920s, the variance in the costs of repression across industries increased, and the incentives for collective strategies to be maintained disappeared. While the ZAG collapsed in 1925, it served as a viable template for future legislation that sought to guarantee union and employer association bargaining procedures.

The preceding discussion of war-time and inter-war Germany demonstrates the importance of the redistribute threat posed by workers and the heterogeneity of the economy in conditioning firm strategies. Firms only forged a coalition with workers out of necessity due to the uncertainty and chaos engendered by the end of World War

273

I. Once the threat from revolution passed, though, the costs of repression fell (as evidenced by the ability of the ADGB and SPD to contain militant threats) firms had an incentive to switch strategies. Collaboration with workers was fragile: with the passing of hyperinflation; an end to occupation of the Ruhr; and a ―moderation‖ in the threat posed by workers, collective collaboration was not sustainable. This chapter has documented the different ways in which repression was used and how even collaborative employer associations were locked into participating. It demonstrates the theoretical importance of costs of repression and the variance in such costs for explaining the broad contours of industrial institutions.

This confirmation of the theory contrasts with existing accounts of the coevolution of electoral systems and economic coordination. In particular, it proposes a more convincing explanation of the general strategies of firms than arguments that focus mainly on electoral systems and the incentives of parties and politicians. As discussed in chapter 3, Martin and Swank‘s account argues that PR transfers a ―spirit of cooperation‖ between employer associations and unions and that dedicated business parties participate in coalition governments to attain cooperative ends with workers.

MS propose a number of explanations for this outcome. They argue that firms have specific parties represent businesses which then work with parties representing workers or farmers.

The precise mechanism in their theory that explains the form of employer cooperation is not supported with the evidence presented above. Further, the theory from chapter 3 and the subsequent evidence cuts against their claim. First, the fact that firm strategies changed so dramatically at the end of World War I and during the

274

Weimar republic indicates that the electoral system cannot account for such important temporal changes in firm strategies. Second, the evidence presented on the actual adoption of wide-spanning employer coordination actually cooperation pre-dates the adoption of PR: it was the feared chaos and attendant rise in the costs of repression that prompted important firms to switch strategies and seek unions‘ assistance in managing the demobilization.

Third, as discussed in chapter 2, existing accounts which focus on coding such coordination as collaborative misses the importance of collective repression, which was important during the Weimar era. In fact, repressive factions of employers helped to undo the ZAG. Fourth, the repressive positions of the political parties DVP and

DNVP, and the fact that they participated in coalition governments, indicate that employer associations still invested resources in government to have the state share the costs of repression when possible. This evidence means that distributive conflict, not just a spirit of cooperation, can characterize industrial relations within a proportional electoral system.

This chapter also presents evidence inconsistent with the mechanism posited by CIS regarding the co-evolution of PR and systems of economic coordination. This theoretical perspective argues that certain long-standing systems of economic coordination, due to historical guilds, explains why some countries adopted PR. This account argues as well that an ―inclusionary‖ logic explains the link between these two institutions; certain economies have economic institutions that jointly invest in workers‘ skills. In particular, they cite Thelen‘s 2004 discussion of the ZAG as being supportive of their mechanism.

275

This manuscript does not address the formation of PR. But the above evidence on the ZAG indicates that its initial formation and subsequent collapse had less to do with joint incentives of workers and firms to invest in specific assets. Rather, the

ZAG was formed out of fear that without some institutionalized cooperation, either workers or returning soldiers from WWI would become further radicalized and threaten production and the viability of the economic order. These reasons ultimately mattered more for why the ZAG was established. The preceding chapter establishes the importance in Weimar Germany of the threat from workers and changes in enforceability of agreements for explaining variation of firm strategies displayed in

Table 2-1. In the next chapter I discuss how the absence of these existential threats to the United States meant that repression was sustainable for a more continued amount of time.

276

Figure 6-1: Change in Firm Strategies in Germany

277

Chapter 7: Evidence from the US Inter-War Period

―I am to set before you the advantages of organization. The benefits of organization are so numerous and so diversified, and in some instances so obscure that to discuss all of them would require considerable time…[particularly] the existence of the National Metal Trades Association and kindred organizations of employers, originated in the militant attitude assumed by aggressive labor leaders towards the individual employer.‖ – Andrew J. Allen, Secretary of the Indianapolis Branch of the National Metal Trades Association, addressed to the Ninth Annual Convention of the National Metal Trades Association, at Boston, Massachusetts, March 21-22, 1907 (Reprinted in The Open Shop, 1907, 269, 272).

Andrew J. Allen‘s address to the member firms at the annual convention of one of the most prominent American employer organizations had many goals. Its main purpose was to exhort the NMTA‘s member firms to commit more resources to the organization. It was also a plea for firms to invest more resources and funds to enforce the principles of an organizational movement that would unite firms against the perceived increased aggressiveness of American unions. Allen‘s address was representative of that of other prominent leaders of national and regional employer organizations. It illustrates that even when some company managers and owners thought that the benefits from joining an organization were obvious, organizational leaders still felt the need to persuade member firms to commit resources. Most strikingly, Allen cites the ―militant attitude‖ of workers in justifying the organization‘s purpose.

Allen‘s dilemma illustrates the difficulties of collective action of firms to form associations. Further, it demonstrates the importance of repressive associations against unions: the existence and influence of these associations in the United States during the inter-war period provides further confirmation of the theoretical framework

278 presented in chapters 2 and 3. In this chapter I use evidence from a considerably different country time period to test the hypotheses laid out in the theoretical chapter: variation in the adoption firm strategies can be attributed to the redistributive threat posed by organized labor as well as problems of enforceability of agreements.

The United States poses a useful contrast against the German experience, because, as discussed below, there was historically a lower redistributive threat from workers and much higher degrees of industrial heterogeneity and structural obstacles to collective action among firms. As a result, we should observe a different set of outcomes: difficulties forming employer coordination at a meaningful national level, and a reluctance to switch from repression to collaboration. I demonstrate that during periods of exogenous shifts in heterogeneity—war time intervention of the state— collective repression did emerge, which has been previously undocumented in the comparative political economy literature. In the United States inter-war period, I demonstrate that the initial adoption of collective repression and the failure to make it sustainable the national level are a result of the hypothesized variables. While there has been much research on how the United States functions as a canonical liberal market economy, the role of repressive employer organizations in its labor history has been understudied.

Specifically, to confirm the existence and importance of repressive employer coordination and test the proposed theory, I use new quantitative and qualitative evidence from the important anti-labor Open Shop movement in the United States during the early 1920s. This movement represented the height of manufacturing firms‘ collective repressive strategies against workers: key organizations that

279 spearheaded the movement espoused an ideology and set of policies that restricted union presence at the plant level and undermined the nationalization of the labor movement. Despite the importance of the aggressive anti-labor movement in the

United States, the mechanics and politics of it remain poorly understood I use a new dataset on participation in the Open-Shop movement at the US county level during the inter-war period. I find firms‘ adoption of repressive employer organizations against workers arose in response to higher levels of unionization. Collective repressive employer organizations, controlling for other structural conditions, emerged at the local level in response to threats from unions.

The second empirical part of the chapter demonstrates that these initial collective repressive strategies were not sustainable at the national level during the early inter-war period. I use primary and archival evidence from records of major employer organizations, including meeting synopses, publications, and internal records, to show that these nascent organizations did not engage in increased national- level coordination among member firms. Nascent employer organizations made a strategic choice to limit the collective strategy, and ultimately favored reversion to individual-level repressive firm strategies. Incumbent employer institutions thought that expanding the collective apparatus could possibly backfire and undermine existing achievements. The increase in heterogeneity during this time period— proxied by the exogenous removal of state intervention after WWI—made the continuing of such collective strategies untenable and unenforceable. Further, these incumbent institutions thought expanding the collective apparatus could possibly backfire and undermine existing achievements. The self-imposed institutional

280 constraint—the choice to revert to more individual firm strategies—also contrasts sharply with the experience of some continental European states. As the previous chapters demonstrate, Germany experienced much more sustained collective collaborative strategies from firms after World War I due to the sharply increased redistributive threat from workers.

This particular research design and test for the US experience is an improvement over existing explanations of US industrial relations. By using county- level analyses, I am better able to arbitrate between competing explanations of firm strategies. Naturally, such analyses cannot help us explain the firm strategy at the national level, but they allow one to isolate the importance of individual mechanisms which might work at the national level. The use of primary evidence of course allows one to assess whether the logic of the strategies described in chapter existed.

Unfortunately, specific data on industrial heterogeneity such as that used in Chapter 5 for Germany are not available for the United States. However, I use qualitative evidence regarding the expansion of employer associations to test the mechanism of enforceability of agreements.

Section 1 of the chapter provides an overview of the US case and the main trends in the outcome variables. Section 2 discusses the particular relevance of the US time period considered. Section 3 operationalizes the theory in terms of specific variables, and then describes the original data used. Section 4 presents the results for testing one prediction for the theory. Section 5 tests another prediction of the theory with qualitative evidence. Section 6 concludes.

Section 1. Overview of Changes in Firm Strategies

281

Figure 7-1 displays the main changes in employer strategies in the United

States since the late nineteenth and early twentieth century. The main strategies that firms pursued for most of the nineteenth and twentieth century were characterized by individual and then collective repression of workers to prevent unionization. As the figure shows, the adoption of the Open-Shop movement occurred after a sharp government-induced compression of wages during World War I. An exogenous imposition of wage dispersion reduction facilitated a collective action movement that was repressive towards workers. A slight increase in industrial heterogeneity, which is proxied by the increase in wage dispersion towards the end of the inter-war period, is linked to a shift in firm strategies.

The figure shows the inter-war period switch to a modal strategy of individual repressive strategies later on in the mid 1920s. These strategies predominated up until the implementation of the Depression-era National Recovery Act (NRA) in 1933 and the National Labor Relations (the Wagner Act) in 1935, which legalized collective bargaining and made illegal many repressive tactics of firms. During the period of repression, the costs were shared among employer organizations and the state, including the local, state, and federal government. The graph shows that changes in the theorized variables correlate with the major shifts in firm strategies.

Overall, firms were able to engage in the lower cost strategy of repression due to the absence of a coordinated labor movement that would have posed a large redistributive or revolutionary threat. A multitude of reasons have been documented for the absence of such a threat from workers. These reasons document the structural difficulty in posing a large threat from workers: a much larger workforce and

282 geographical scope served as inhibitors of labor collective action; divisions between craft and industrial unions; and the general absence of a workers‘ party (Montgomery

1989; Perlman and Taft 1955).171 But less documented is the importance of collective repression from firms as a strategy. Firms engaged in individual-level repressive strategies when costs of enforcing a collective agreement among firms were high.

After WWI, they were able to adopt more collective repressive strategies, though they were not of the same scope as those in other states.

From the late nineteenth century until World War I, firms engaged largely in individual-level repressive strategies. Major steel companies and manufacturing firms could engage in repressive strategies, preventing and ending strikes. For example, between 1881 and 1900, firms hired over 500,000 replacement workers of the 4.8 million who went on strike; strikebreakers were used in 40 percent of strikes in this period and substantively increased the chance of success for firms (Rosenbloom

1998). Unions increasingly struck during this period to attain bargaining rights, and both large and small firms responded with repressive strategies, and there were very few sector-level agreements between firms and workers. The judicial apparatus and contested legal status of unions after 1890 also greatly helped firms engage in other strategies of repression and inhibit unionization (Bok 1971). Following the passage of the Sherman anti-trust act in 1890, there was a long legal battle in state and federal courts between firms and unions on whether unions themselves were illegal agreements, and whether strikes were legal. Firms could file injunctions against unions in anticipation of a strike, resulting in costly legal battles and delays for unions

171 These reasons demonstrate the possibility that structural factors that affect employer coordination can also affect worker‘ collective action.

283

(Fitch 1924; Letwin 1965).

The United States then in the early inter-war period was characterized by a strategy of collective repression. Collective repression emerged as a strategy to dismantle war-time labor institutions which had also compressed wages. These strategies were pursued by multiple influential sector-level employer organizations.

These strategies constituted part of the ―Open Shop‖ (OS) movement, which is a key example of firms coordinating to prevent workers from organizing. Important employer associations included the National Metal Trades Association (NMTA), the

National Erectors‘ Association (NEA), the National Association of Manufacturers

(NAM) – mainly a lobbying group, and the National Founders‘ Association (NFA). 172

Under the movement, firms joined local-level employer associations which had the goal of undermining craft unionism and preventing union infringement on firms‘

―right‖ to set wages, working conditions, and control of training.

Section 2. The United States and the ―Open Shop‖ Movement

172 Martin discusses the turn of the National Association of Manufacturers (NAM) towards a more neo- liberal ideology but does not explicitly address the rise of other anti-labor movements or repression as a strategy before this transition. Her excellent account focuses on the link between the Republican Party and the NAM, and discusses why this particular organization did not become a cooperative encompassing employer institution like its European counterparts. One of the key explanations is that the NAM was perceived to be too linked to the Republican Party, preventing it from being a true corporatist institution. In this paper I focus on a different outcome: the rise of repressive employer coordination before the NAM became involved in the decision ―nationalize‖ such movements. See (Martin 2006) Several important studies in the field of labor history and industrial sociology examine specific organizations, but without specifying generalizable theory. A wealth of literature exists regarding the formation of US industrial-relations institutions For development of the NMTA and its role in the Open Shop see a series of articles by Haydu (1988; Haydu 1999) Harris provides a detailed account of the Philadelphia branch of the National Metal Trades Association (2000). Fine examines the history of the important National Erectors‘ Association (1995). Derber provides a description of important industrial-sector employer organizations (1984) In this chapter I quantify and test some of the implied theories in these historical accounts regarding development of employer associations.

284

The experience of the United States after World War I is a useful country and time period in which to examine variation in the development of repressive employer coordination. Few systematic studies in the fields of comparative political economy and American political development examine the development of repressive employer organizations. In particular, the development of the Open Shop (OS) movement in the

United States after World War I presents a unique opportunity to explore the emergence of repressive employer coordination. It is theoretically important because it represents a clear instance of firms coordinating to prevent workers from organizing.

Existing theoretical literature on the emergence of industrial institutions neglects the formation and actions of major employer organizations. In particular, they ignore the importance of repression. However, the variation in the emergence of repressive employer coordination during the inter-war period—across different industrial sectors as well as different geographical regions—indicates important intra-state as well as temporal variation in the actions of employer organizations. Both kinds of variation can be leveraged to differentiate between possible explanations of firm strategies.

Under the Open-Shop movement, firms joined local-level employer associations which had the goal of undermining craft unionism and preventing union infringement on firms‘ ―right‖ to set wages, working conditions, and control of training. It is empirically useful because there was much intra-state variation in the degree of employer coordination, as discussed below. These occasionally brutal activities had great implications for the worsening of working conditions in the United

States before World War II, but we do not have any systematic understanding of what explains their variation.

285

The influential onset of the movement can be traced back to the collapse of attempts to form encompassing collective strategies after WWI. Firms and prominent employer associations discussed after the conflict how to dismantle the federal government-imposed wage agreements that came into existence during the war

(Wakstein 1964).173 Their preferences regarding which strategy and industrial institutions they wanted were articulated in a number of publications, but also prominently at President Woodrow Wilson‘s National Industrial Conference in

September 1919.174 The purpose of this conference was to agree upon post-WWI industrial institutions and to resolve labor disputes that had emerged since the end of the war. At this conference, prominent employer organizations, including the NAM,

NMTA, and NICB, spoke against union preferences to maintain government-imposed industrial institutions; unions wanted expansive collective bargaining with counterpart associations. Employers viewed such ideas as a threat to managerial independence.175

The conference representatives of employer organizations included Magnus

Alexander, the head of the National Industrial Conference Board (NICB), James A.

Emery of the National Association of Manufacturers (NAM), J.W. O‘Leary of the

National Metal Trades Association (NMTA), and representatives from individual large

173 See the useful accounts provided by Hurvitz (1977) and Wakstein (1969). The original source for the positions of different representatives comes from the Proceedings of the First Industrial Conference, October 6 to 23 1919 (1920). 174 They also discussed their position on repression and support of the Open Shop in a variety of publications, such as the Iron Trades Review and Metal Trades Bulletin. 175 The conference representatives of employer organizations included Magnus Alexander, the head of the National Industrial Conference Board (NICB), James A. Emery of the National Association of Manufacturers (NAM), J.W. O‘Leary of the National Metal Trades Association (NMTA), and representatives from individual large corporations. Representatives from firms included Bethlehem Steel, DuPont, General Motors, International Harvester, General Electric, and Standard Oil of New Jersey. The labor delegation had leaders of the American Federation of Labor (AFL), including Samuel Gompers, as well as leaders of major railroad unions.

286 corporations.176 The labor delegation had leaders of the American Federation of Labor

(AFL), including Samuel Gompers, as well as leaders of major railroad unions.

Union leaders at the conference wanted to continue the framework that had been established by the National War Labor Board (NWLB), which had given workers the right to bargain collectively. As Hurvitz notes, the business delegates‘ preferences were ―diametrically opposed‖ to those of unions (Hurvitz 1977). Employer representatives at this summit advocated a return to pre-WWI institutions and a dismantling of collective bargaining. To accede to union demands, as one industrial representative put it, would be ―nothing short of a national calamity, for it would set up a dominating power side by side with the National Government which would have an influence and a force…[that] no private organization should possess.‖177 Magnus

Alexander, the head of the NICB, advocated at this conference the principle of the

Open Shop. As Hurvitz describes, the First Industrial Conference ended without any resolution; it was at this conference where the anti-union and repressive preferences of key employer associations were articulated. They argued that unionism imposed unfair restrictions on firms‘ managerial authority.178 The two sides did not come to any agreement that would have kept war-time wage agreements and labor regulations

176 Representatives from firms included Bethlehem Steel, DuPont, General Motors, International Harvester, General Electric, and Standard Oil of New Jersey. 177 See Proceedings of the First Industrial Conference, October 6 to 23 1919, 196. 178 There were other third-party representatives appointed by President Wilson, but the main issues of contention were discussed by the organizations representing unions and employers. At this conference, employer representatives also advocated another strategy to undermine unions, by using ―shop councils‖ which would be forums for workers to air grievances, but they would not be able to call strikes or make decisions without managerial consent. At this conference, collective employer associations advocated also that firms could solve labor problems locally: they opposed the AFL‘s vision of large scale labor unions bargaining on an equal footing with employer associations. Firms also expressed anti-union preferences through their role in the National Civic Federation, which had minimal impact on actual employer association policy (Green 1956)

287 in place.

The Open Shop movement that emerged after WWI and in the wake of the collapsed national conference was a direct challenge to organized labor and the efforts of unions to maintain and extend such wage agreements. Wilson‘s attempts to solidify the government-imposed national cooperative industrial institutions failed. From the employers‘ perspective, the perception was that firms, after the period of war-induced government intervention private intervention, needed to ―win back‖ managerial authority over workers. William H. Barr, the President of the National Founders‘

Association (NFA), stated one succinct goal of the organization after the war: ―War- time wages must be liquidated‖ (Dunn 1927), 22).

The Open Shop movement was characterized by the diffusion of employer associations across counties that supported ―open shop‖ or anti-unionization principles for member firms. These associations generally formed in the metal-manufacturing and finished-goods industries. The phrase ―Open Shop‖ refers to an objective of the owners of a firm to have the choice and ―right‖ to hire and fire workers regardless of their union affiliation. The ―open shop‖ also refers to the firm owner‘s right to manage worker training without interference from apprenticeship demands from workers, and pay wages and benefits as they saw fit. The Open Shop was a battle of employer associations and individual firms against craft-based unionism over the

―right to manage‖ and a strategy to win back and solidify the industrial relationships that had put workers at a bargaining disadvantage before WWI. The movement had an ideological component that used rhetoric of firms ―rights‖ to be unconstrained by union-imposed restrictions. These publications also provided information on the status

288 of legislation and union demands in other states and localities. Many of the associations were formed to prevent recognition and growth of unions, both within and across firms. The movement, despite objections to such labels by some firms, was intended to oppose collective bargaining rights for workers, and to discredit and reduce the spread of unionism.

The Open Shop movement promoted itself as a necessary institutional response to the rise in ―closed shop‖ ideologies advocated by unions such as the International

Association of Machinists (IAM) and the longstanding threat posed by member unions of the American Federation of Labor (AFL). Firms involved in the Open Shop movement and staff members for regional and local offices produced an impressive array of publications and propaganda to promote its policy goals as part of a strategy to reduce the political and economic influence of workers. For example, the National

Metal Trades Association published The Open Shop, with articles from prominent industrialists on threats posed to firms by unionism and pro-labor legislation. The

NAM published in 1921 a guide to the Open Shop for distribution in public schools and libraries (Open shop encyclopedia for debaters 1921). Other publications like the

Iron Trades Review and Metal Trades Bulletin discussed the progress of legislation at the state and federal level that was perceived to be harmful to firms. They also included useful information on industrial trends and economic data.

This diffusion in local associational activity coincided with increased activity from national-level employer associations that used aggressive and sometimes violent means to defeat unionism, particularly in the steel-making and steel erection industries. The explicit goals of these local and state organizations were what one

289 historian called ―undoubtedly the most militant and widespread offensive ever undertaken by the American employing class—a class which has probably shown greater vigor and fighting spirit than in any other country‖ (Dunn 1927, 22). This period in industrial relations after World War I was characterized by a mix of vibrant sub-national employers‘ activity from manufacturing firms, along with a nationally promoted anti-union ideology. They existed in over 250 US metropolitan areas, many of which were medium-sized or the largest industrial areas.179 This burst of collective repression helped reduce unionization in the inter-war period.180

These local and regional collective repressive associations, like their German counterparts, sought to achieve anti-union goals through several instruments. They imposed rules prohibiting employers from hiring union or formerly union-affiliated workers; coordinated ―blacklists‖ that prevented workers with any union history from being hired; imported strikebreakers; and tried to consign member firms to participate in lockouts to enforce their aggressive anti-union stance. As described in Chapter 2, an important instrument of such associations was the multi-employer sympathy lockout in response to strikes. Employers united in a city would agree to lock out union-affiliated workers in the event of a strike. The motivation was to force unions to use as much of their strike-support funds as possible. This strategy countered the union strategy of engaging in whipsaw strikes, which consisted of targeting and striking against individual firms and ―picking them off‖ one at a time. Sustained

179 The 1936 La Follette Senate Committee report documents a variety of tactics used by companies to undermine unionism, including the use of labor spies, intimidation, violence, and coordinated lockouts. 180 Elsewhere I document why these initial employer organizations chose not to form a peak inter-sector body that would have management over local employer associations‘ efforts. These organizations made a strategic decision that local-level repression was sufficient.

290 employer coordination at the local level, with the weapon of the lockout, was designed to counter such union activities. Beyond these instruments that directly targeted union workers and union treasuries, some state-level employer associations lobbied state governments for legislation against unions, such as outlawing pickets and allowing the use of court injunctions against strikes.181

The Open Shop movement in the inter-war period began as firms across cities in the United States, beginning with waterfront employers in Seattle, and manufacturers in San Antonio, Toledo, and Indianapolis, independently established employer associations that espoused Open Shop ideals about the right for employers to manage without labor interference. These movements were often established by specific industries, though they sometimes adopted city titles for their movements.

For example, the organization the ―Associated Industries of Seattle‖ was the Seattle- based open shop employer association. These early local employer coordination movements started largely independently of each other. By 1922, over 200 US cities had employer organizations (Wakstein 1961).182

These newly formed local employer associations were called ―Open Shop

Associations‖ (OSAs). These OSAs functioned much like the kind of repressive employer coordination regimes described in Chapter 2. They generally functioned as follows: an OSA‘s member firms paid dues and agreed to comply with institutional rules such as participating in a lockout, but in exchange they received information and

181 Swenson discusses the logic and use of these lockouts. See Swenson (2002). Harris describes their use in Philadelphia. See Harris 1991, 111-132. The Senate committee hearings chaired by Senator La Follette provide much detail about the use of lockouts, labor spies, and blacklists. See U.S. Congress, Senate, Subcommittee of the Committee on Education and Labor. Hearings Pursuant to S. Res. 266, Violations of Free Speech and Rights of Labor. 74th–76th Cong., (1936–1940). 182 See discussions in various trade publications, including: (Public Sponsors Open Shop Associations 1920; Declare for Open Shop in 78 Cities 1920).

291 industrial-relations services described above, such as assistance in times of strikes, and supporting funds to compensate for lost productivity during lockouts. Some OSAs also published local trade journals and circulated information on labor conditions and wages of firms. OSAs distributed information about ongoing and potential labor organization, and were to provide strike assistance when a member firm asked for it.

The existence of an OSA in a city or county is a straightforward indicator of a collective repressive coordination, as it constitutes a successful attempt on the part of firms in a manufacturing industry to institutionalize their operations against organized labor.183

The preceding discussion has focused on local diffusion of OSAs and ideological elements to this anti-union effort, but national employer associations also existed. The National Metal Trades Association (NMTA) and National Erectors‘

Association (NEA) were key actors in managing labor markets in select industrial areas, and were the national unions‘ largest adversaries. As discussed above, the

NMTA had representatives at Wilson‘s industrial conference, and Allen‘s address at the start of this chapter was intended to persuade NMTA members. These associations shared similar Open Shop goals. The major manufacturing lobby, the

National Association of Manufacturers (NAM) also became more oriented towards espousing anti-labor policies, which I discuss below. The Open Shop movement— both the locally coordinated and national institutions—helped reduce unionization rates during the 1920s (Wakstein 1961).

Studying variation in OSAs allows us to test hypotheses about what explains

183 We do not have local (city or county) level data on the ―success‖ of each OSA regarding its attainment of its objectives.

292 variation in repressive employer coordination. To measure instances of repressive employer coordination, I use a new dataset on local-level adoption of OSAs. Local level OSAs serve as the best available proxy for repressive employer coordination.

Despite the influence of the Open Shop movement on US industrial relations, and its brand of virulent anti-unionism, we know very little about its diffusion, and how these independent local repressive movements related to existing national institutions.

Historians have provided important descriptions of specific associations‘ activities during this time period, but there has been little use of systematic data to explain why

OSAs emerged in some counties and not others.184

Two empirical questions arise regarding variation in repressive employer coordination and the test of the theory. First, what factors explain variation in local adoption of OSAs? Understanding such variation allows for testing of the theory at a more micro level. Second, why did local level OSAs not expand the scope or further institutionalize to form a true ―national federation‖ of employers that would have been a more effective bulwark against organized labor? In short, why was there no

―American Federation of Employers‖ to counter the AFL? This would have been a fortuitous period for US employers to solidify and engage in repressive coordination at the national level. As I discuss below, issues of enforceability help explain why national-level strategies were not adopted.

Section 3: Testing the Theory with Micro Data

Hypotheses

Because of the scarcity of empirical studies of the development of the Open

184 Harris in particular provides a detailed account of the role of the Open Shop in employer coordination of labor markets in Philadelphia. See Harris (2000).

293

Shop movement in the United States, and the absence of studies regarding variation in repressive employer coordination more generally, I outline in this section a basic set of hypotheses that follow from the discussion in Chapter 3. I consider variables that should affect the probability of the adoption of an OSA at the local level. To measure these instances of repressive employer coordination, I use a new dataset on local-level adoption of OSAs. The existence of a local OSA at the city or county level is the best available proxy for firms‘ repressive coordination against workers, as it is a discrete unit of a collective repressive employer organization. For the following hypotheses, the question is why a specific county would adopt an OSA.

Economic variables. I first consider demographic and economic structural variables that should affect the adoption of repressive employer organizations. First, I expect that areas with higher manufacturing output should be more likely to have repressive employer organizations. Higher output means there is enough economic activity such that firms should either have more resources to form organizations, or have an interest in maintaining high output through an employer organization. This hypothesis is consistent with previous descriptions of national-level employer organizations.185 Second, following the same logic about industrial output, I expect areas with higher population to be more likely to have repressive employer organizations. Third, I hypothesize that the number of firms in an area should be negatively related with the adoption of repressive employer organizations. This is because an increased number of firms could reduce the probability of successful

185 See McCafree (1962) for discussion of this point.

294 collective action among firms (Olson 1968).186

Non-economic variables. I now consider hypotheses that link non-economic variables to the adoption of repressive employer coordination. I hypothesize that repressive employer organizations should be more likely to form the higher the threat to firms posed by worker unionization. Firms should be more likely to form repressive employer organizations as a counter to workers‘ demands. This is hypothesis is consistent with the descriptive evidence about employer organizations‘ preferences (as evidenced by their publications, discussion at annual meetings, and their positions taken at Wilson‘s Industrial Conference). Many other employer organizations outside of the United States also initially formed to respond to demands from unions.187 However, this hypothesis has not been rigorously tested.

I consider now hypotheses regarding political variables. Firms should be responsive to political conditions in making decisions about forming repressive organizations. Specifically, I hypothesize that repressive employer organizations are more likely to form to respond to left-wing electoral support. This is because firms would be concerned that local support for left-wing politicians would make pro-union policies more likely. As discussed in chapter 3, this could be because of the concern that the local population will support unions through favoring legislation (such as laws that prohibit the use of injunctions to prevent strikes, or that exempt unions from anti- trust regulation), or through coercive intervention in labor disputes that would favor workers. Under this hypothesis, employers coordinate to form repressive

186 An alternative hypothesis is that because the number of firms in a locality might cause or be a consequence of higher economic production, more firms in a locality could be positively correlated with the development of employer associations. 187 See examples from Gladstone and Windmuller (1984).

295 organizations in response to the activities of the state, as opposed to those of unions.188

In the first empirical section, I test these competing hypotheses with a new dataset on US county-level variation in repressive employer organizations from the inter-war period. In the second empirical section, I address the question of why these local movements did not organize with existing sector-level organizations to form a more organized national confederation of employers after the formation of local-level

Open Shop associations. The evidence, gathered from reports and internal discussions from regional and national employer institutions, indicates that the US Open Shop movement did not nationalize because existing employer organizations chose not do so, due to concerns about the costs of enforcing such an agreement. The qualitative evidence indicates that the interests of incumbent institutions matter for explaining the emergence of repressive employer coordination at the national level.

Data Sources, Variables, and Estimation

In this section I describe the sources of data and coding procedures used to code variables to test the above hypotheses.

Measuring the dependent variable: city/county level repressive employer coordination. To measure the dependent variable of the existence of repressive employer coordination, I use results from a 1920 questionnaire distributed to firms and associations administered in 1920 by the Associated Employers of Indiana (AEI). The

188 An alternative hypothesis is that lower support of the left could mean that that the population would be willing to support actions by employer organizations (such as lockouts or importation of substitute workers during a labor conflict). Therefore, left support could be negatively correlated with more employer organizations. Political support for employer organizations could take the form of pro-firm or anti-union legislation, or executive intervention in industrial disputes. The empirical test below arbitrates between these competing hypotheses. One difficulty is that the Democratic Party was not nearly as affiliated with workers as Socialist parties were in European states, indicating that Democratic support is not necessarily a threat to firms (Sanders 1999); (Montgomery 1989).

296

AEI supplemented the questionnaire with attempts to find out which cities in the

United States had active Open Shop Associations (OSAs). As noted above, the OSA is a useful proxy to denote the existence of repressive coordination. The adoption of an OSA in a city or county indicates that firms in a given locality felt strong enough that a formal association was needed.189 I use the AEI survey data to generate a list of

US counties that contain an OSA because more precise economic data from the US

Census are available at the county level.190 The dependent variable of interest for all county level analyses is binary: I code whether a particular county had an OSA. I code the county with a ―1‖ if an OSA in the inter-war period was present; ―0‖ indicates no OSA was present.

Independent variables.

Economic control variables. I use county level economic data from the 1920

US Census. I include lagged control variables at the county level for logged population, amount of manufacturing output, and number of firms (ICPSR 2004). 191

Non-economic variables: Threat posed by unions. To code the threat posed to firms by unions, I use data that quantifies the amount of union presence at the county level. This dataset indicates whether a US county had a union local for 23 industries or

189 The AEI‘s survey data are discussed in several editions of the Iron Trades Review. The existence of an OSA in a county does not indicate that the city government or the electorate chose to have an OSA; it only indicates the collective decision of firms within a given locality to form an OSA. 190 This results in a reduced number of observations because some counties have more than one city which had an active OSA). Unfortunately the sampling methodology used by the AEI to form their list of active OSAs is not available. 191 Haines, Michael R., and the Inter-university Consortium for Political and Social Research. Historical, Demography, Economic, and Bibliographic Citation: Social Data: the United States, 1790- 2000 [Computer file]. ICPSR02896-v2. Hamilton, NY: Colgate University/Ann Arbor: MI: Inter- university Consortium for Political and Social Research [producers], 2004. Ann Arbor, MI: Inter- university Consortium for Political and Social Research [distributor], 2005-04-29. Other economic variables are discussed in the results section.

297 craft professions for three different time periods between 1890 and 1915. The three time periods for which data on the existence of a union local are coded are 1890-1895,

1895-1910, and 1910-1914. Data on the presence of union locals by county are available for 595 cities in 17 states.192 For each industry at each of the three different time intervals, there is a binary indicator for local union presence.

To construct the measure of union threat to firms at the county level, I first code a binary variable for each county that indicates whether county j was home to a union local in industry x for any period t (t represents one of three time periods of

1890-1895, 1895-1910, and 1910-1914). A ―0‖ indicates that for the given county- time period-industry in question, no union local was present; a ―1‖ indicates that a union local in that county present. I then sum for each county j each of these binary indicators. This sum indicates the total number of binary indicators for a union local in a country, throughout the entire time period, 1890-1914. For example, if a county had no union local until the period 1910-1914, but then had union presence in 10 different industries in that period, its summed union threat would be 10.193 The variable

Unionthreat denotes this sum. Higher values of Unionthreat indicate that throughout the time period 1895-1914, the county had a higher number of different unions

192 The industries or crafts where union locals were coded are: Boiler Makers, Boot and Shoe makers, Bricklayers, Carpenters and Joiners, Cigar makers, Coopers, Flint glass workers, Gold beaters, Hod Carriers, Machinists (IAM) , Electrical Workers (IBEW) , Ladies Garment Workers (ILGWU) , Molders (IMU) , Typographers (ITU) , Leather workers, Metal polishers, Painters, Postal workers, Pattern makers, Sheetmetal workers, Stove mounters, Amalgamated Wood Workers. The states for which county level data on union local presence were available are: Alabama, Arkansas, Florida, Georgia, Illinois, Kentucky, Louisiana, Michigan, Minnesota, Mississippi, Missouri, North Carolina, South Carolina, Tennessee, Texas, Virginia, Wisconsin. 193 Similarly, if a county had one union local in the same industry across all three time periods, its value on the Unionthreat variable would be three. Unfortunately, such data at the annual level do not exist. Nor do data on union density by county level exist for this period. But the data at the time-period level capture much of the threat posed by unions to firms by documenting the number of locals in different industries.

298 represented in the county. The mean of this union threat variable is 4.8 and the standard deviation is 5.9.

I also code a variable that measures the growth in union presence at the county level. To do so, I follow the same procedure as above—summing the binary indicators for a union local for each county—for each of the three time periods. The difference between the total number of unions across all time periods (Unionthreat) and the sum for just the first two time periods captures the number of new union locals that emerged. This variable is called Uniongrowth. This variable captures changes in union presence in the most recent period before employer organizations emerged after

WWI. The mean of this union growth variable is 2.3 and the standard deviation is 3.0.

35 percent of counties experienced no growth in union presence between 1910 and

1914. 20% of counties had one additional union local in this five-year period. This was also the modal amount of union growth, conditional on a county experiencing any growth in union presence. Both variables of Unionthreat and Uniongrowth are the best available measures of the threat posed by unions to firms during this time period.

Political variables. To measure the threat from a left-wing electorate, I use county-level Democratic presidential vote share from the 1920 election. This is a proxy for political support for the left and the potential threat from the left to firms. I also use data for the Socialist presidential vote share from the same election. I also use the difference in Democratic presidential vote shares between 1916 and 1920.

Estimation procedure. To test the competing hypotheses about the development of repressive employer coordination, I estimate a probit regression model

299 with standard errors clustered by county. 194 The unit of analysis is the US county.

The dependent variable for each specification is binary: whether the county was noted to have an OSA or not. I use the existence of an OSA as an indicator of repressive employer coordination. The model takes the form:

probit (Yij) = i + 1Xi + i (1) where Yij indicates the binary presence of an OSA for county i, and X is the vector of relevant covariates for each county. In all specifications the subscript j is binary, taking a value of 1 if the county has an OSA, and 0 if not.

Section 4: Results from Micro-Tests

Before presenting the results of different estimations, I discuss some basic patterns in the data that are supportive of the hypothesis that the threat posed by unions is a predictor of repressive employer coordination (which I proxy by adoption of an OSA). In 1921 there were 272 cities (217 counties) in the United States that had at least one OSA. The cities that had such associations were not just limited to the largest metropolitan areas; in fact, such employer associations were prominent in

194 One methodological concern could be that given the limited data available regarding county-level union presence, there could be selection bias when estimating the effect of union presence on OSA adoption. That is, it could be the case that for the counties for which data on union presence are available, are also counties in which OSAs are adopted, but it may not be the case that the union presence has an independent causal effect on OSA adoption. To address the possible selection issues, I also estimate a Heckman probit model with sample selection. This is the equivalent of a Heckman selection model, except the model consists of a probit model in the selection equation, and a probit model in the outcome equation. In this model, the selection equation has the binary dependent variable indicating if data were available on union presence or not for a specific county. The main dependent variable was the same as in equation (1). The intuition is that some of the economic independent variables might explain why union data are more available in certain counties. In the Heckman probit model, the statistical significance of the parameter estimate of ρ is the correlation of errors between the two questions. If ρ is zero, then the unobserved variables that affect whether union data area available for a county are independent of the unobserved variables which might affect adoption of an OSA. In all specifications of the selection model, ρ was statistically insignificant indistinguishable from zero, and the coefficient on the variable Unionthreat was the same as those estimated in the probit equation (1). Interested readers should consult Dubin and Rivers and Greene 1997 for details on derivation of the model (Dubin and Rivers 1990); (Greene 1997). See Berinsky for an application of the model (1999).

300 many medium-sized cities throughout the Midwest and South. 2,889 counties did not have an OSA, so the dependent variable of interest for all US counties is seven percent of the sample. Only one US county had an OSA that also had no union local presence in any industry (that, is where the value on the union threat variable is zero). This is strong baseline evidence that the minimal existence of a union in a county is necessary for an OSA to exist. In other words, the set of counties that experienced an OSA constitutes almost an entire subset of cities with at least some union presence in the previous time period 1885-1914.

For the entire set of counties, population and manufacturing output are correlated at r2 = .56. Manufacturing output and the number of firms are correlated at r2 = .57. The 1920 Democratic presidential vote share is negatively correlated with population, manufacturing output, and number of firms (r2 = -.047, -.13, -.31, respectively). Regarding union presence or union threat, interestingly both union presence and union growth are negatively correlated with Democratic presidential support (r2 = -.20, -.21, respectively). Union presence is also positively correlated with population, manufacturing output, and number of firms (r2 = .69, .47, .28, respectively).

How are counties that have an OSA different from those without? I conduct difference of means tests between the means of the economic variables of the OSA and non-OSA counties. As expected, OSA counties are much richer and more populous counties than non-OSA counties. OSA counties have four times the amount of manufacturing output per capita as non-OSA counties and are about nine times as populous (p<.001 for both tests). OSA counties also have 30% more firms than non-

301

OSA counties (p<.001). Regarding differences in Democratic presidential vote share,

OSA counties have a 10% less electoral support for the Democratic candidate than do non-OSA counties (36% versus 46%, p<.001). OSA counties also have a mean of approximately 14 different union locals between 1885 and 1914, whereas non-OSA counties only have a mean of three union locals for this time period (p<.001). Thus the initial evidence indicates that union presence is an important factor that distinguishes OSA from non-OSA counties.

Table 7-1presents results of different probit estimations of equation 1. Overall, the results confirm the hypothesis that union presence affects the adoption of repressive employer associations, even controlling for important economic variables.

Models 1-3 consider just the effect of economic covariates on adoption of an OSA.

Models 4-6 consider union characteristics. (Recall that due to more limited data on union characteristics, there is a smaller sample of counties. For models 4-6, 15 percent of the counties have an OSA). Models 1 and 2 show that population and manufacturing output are positively correlated with a higher probability of OSA adoption. The coefficients for both variables are positive and statistically significant at conventional levels. However, as Model 3 shows, the number of firms is not correlated with the probability of OSA adoption in a county. The non-effect could have several interpretations. It is reasonable that higher number of firms in a county would make it more difficult for a collective employer association to form. But it is possible that such a relationship could be non-linear: a small number of firms may decide they are capable of repressing workers without the need for a formal open-shop association, and thus not form an OSA. Conversely, if there are a large number of

302 large firms, they may be better able to overcome collective action problems. The number of firms does not capture the relative sizes of firms, which may better proxy for difficulty in collective action as a predictor for formation of an employer organization.195

Models 4-6 control for the variables of the threat posed by unions, changes in this threat, and the Democratic support of the electorate. The coefficient on

Unionthreat for all three estimations is statistically significant and substantively meaningful. Importantly, these results hold controlling for general industrial productivity and number of firms. It is evident that overall union presence affects the adoption of an OSA. Note that when the effect of union presence and union growth are accounted for, the standard errors for the coefficients for manufacturing output and number of firms increase dramatically. This is one of the few systematic tests using sub-national variation to link the threat posed by organized labor to the response of repressive employer coordination.196

Table 7-2 displays the substantive effect of the union threat variable. The table shows the changes in the probability of a county adopting an OSA when the value of

Unionthreat variable changes (with 95% confidence intervals in parentheses, and holding all other variables at their means). Overall, there is a steady increasing positive relationship between the degree of union threat and the probability of the adoption of repressive employer coordination. A county with no union presence has

195 Specifications that include a square term for logged number of firms per capita also yielded no statistically significant coefficient for either the original term or square term. 196 All models in Table 7-1 were also estimated with the economic variables normalized by number of workers in the county (as opposed to per capita); the coefficient on Unionthreat does not substantively change and remains statistically significant. The inclusion of other economic controls—percent of owned homes, amount of manufacturing value added—did not affect the results.

303 almost no chance of having an OSA. Moving to just one union during the time period gives just a probability of .01. But a county having an increased union threat level at eight union locals has a higher probability of OSA adoption of .17. A county with a union threat level of 16 has a probability adoption of OSA adoption of .80. It should be noted that despite the relative statistical importance of union presence over economic variables, only relatively high values of unions (13) change the probability of OSA adoption above .50. Still, there is a definite positive impact of union threat on the probability of repressive employer coordination.

The results from models 5-6 show that the coefficient on the variable union growth is negative. Counties that face more union growth in the five years preceding

WWI actually have a lower probability of having an OSA, though the trend is less pronounced. The substantive impact of the coefficient is not as large as that of the variable union threat, nor is the effect monotonic. Following the same procedure used to generate the results in Table 7-2, the predicted probability of an OSA for a county with no union growth is .12. An increase in the value of the union growth variable to two (the mean of the sample) lowers the probability of OSA adoption to .06. An increase of four union locals between 1910 and 1914 generates a probability of OSA adoption that is statistically indistinguishable from zero. Thus the negative effect of union growth on OSA adoption is limited to changes involving less than four union locals, though this constitutes about 75% of the distribution.

This result suggests that a very high amount of union growth right before WWI makes it more difficult for employers to establish OSAs. Thus, repressive employer coordination might only be possible under conditions where unions pose a threat to

304 firms, but faster union growth can prevent firms from organizing in such a way. From this data alone it is difficult to infer whether this is because firms try to coordinate repressively and fail to do so, or whether they sign agreements with workers under collaborative arrangements (such as consenting to collective bargaining), or both. The negative effect of union growth on OSA adoption indicates that the arrival of specific union locals, indicating perhaps the creation of ―union towns,‖ can have disproportionate effects that undermine coordinated employer repression.197 But, the overall results indicate that union presence going back to the early twentieth century is correlated with firms‘ response through collective repressive strategies.

Model 6 controls for the threat posed by a left-wing electorate, as indicated by

Democratic presidential vote share in 1920. The coefficient for the variable is statistically significant, but the substantive effect is quite small. Following the same calculation of predicted probabilities as shown in Table 7-2 (holding all other covariates at their means), a presidential vote share of 10% yields a probability of

OSA adoption of .11 (.037, .17). A higher presidential vote share of 50% leads to a predicted probability of OSA adoption of .05 (.015, .091). A much higher presidential vote share of 70% leads to a predicted probability of .03 (.0017, .070). While there is a negative relationship, the effect is not large and the confidence intervals overlap for much of the range of data.198 Overall, the results indicate that it is not the electoral concern at the county level that affects firms‘ decisions to create an OSA, but rather

197 This finding is robust to alternative specifications of ―union growth‖ for the time period until 1914. One interpretation of this negative effect is that there is variation in the degree of militancy and nature of union demands among the 23 industries and crafts. 198 Specifications controlling for difference in presidential vote share between 1916 and 1920 yield similar results. Specifications controlling for Socialist presidential vote share yield a statistically significant but substantively small positive coefficient.

305 specific concerns about union presence. One reason for this set of results is that the

Democratic Party, while perceived to be pro-labor, was far less extreme in its demands than counterpart Socialist parties in other states. Firms were more responsive to direct threats posed by union organization than Democratic politicians.199

Overall, the results demonstrate both the existence of repressive employer coordination and confirm the hypothesis that union presence predicts repressive employer strategies. The above results help isolate individual factors that explain why repressive employer coordination was adopted in some counties and not in others.

Once the effect of union threat is accounted for, number of firms and manufacturing output are not correlated with repressive coordination; only population is still correlated with probability of OSA adoption. These results confirm the theory and previous historical evidence that repressive firm strategies were primarily designed as defensive organizations against union penetration of plants. The next section considers whether these hypotheses can help us understand decisions to form collective strategies at the national level.

Section 5.Qualitative Evidence: Why did the Open Shop movement fail to generate

―An American Federation of Employers?‖

In this section I turn to explaining why repressive coordination at the local county level did not necessarily translate into repressive coordination at the national level. In contrast with the German experience, influential peak employer confederations were limited in ability to compel member associations, and in fact were short lived. Why was a strategy of repressive coordination not sustained beyond this

199 See Martin (2007) for discussion of the relationship between the Republican Party and the NAM.

306 initial post-WWI Open Shop movement? Given the virulence of some of the anti- union rhetoric used by both employer organizations, and the potential collective benefits that a more coordinated national organization could have attained, it is surprising that a federation, or more nationally coordinated set of movements, did not emerge. The Open Shop seems like it should have naturally segued into an American

Federation of Employers, or some similar organization. But it did not. Why?

In this section I use qualitative evidence on the attempted formation of a collective repressive strategy at the national level to show that the mechanisms posited in the theory—that heterogeneity and resulting high variance in costs of repression across different localities—large explain the choice not to form a more influential peak association. I provide evidence that key industrial actors—initial employer organizations—chose not to do so, but also against the preferences of other employer organizations, for several strategic reasons. It was not due to a failure of collective imagination. Existing employer institutions did not support plans for nationalization of the Open Shop because of: 1) concern with radicalization of craft labor movements; and 2) protection of their incumbent institutional authority. I argue that in particular, the second reason is consistent with the mechanism of difficulties of enforcing agreements. Specific records from employer organizations provide evidence of preferences and strategies of the major actors in this time period.

After World War I there was much discussion among existing employer organizations regarding the question of national-level collective action; these associations were those that had participated in the aforementioned industrial conference. Many national level organizations viewed their task as defending against

307 what they perceived to be encroaching union presence in plants, both in finished metal goods as well as raw metals processing (Fine 1995). Such organizations included the

NMTA, NEA, and the NAM. The NAM, the most publicly visible organization at the time, had demonstrated previous interest in labor issues but did not until after World

War I have an official committee that addressed them and drafted policy responses.

The NAM had little to do with earlier local-level pre-WWI anti-labor organizations.

But they directly confronted the choice about whether to nationalize this movement after WWI, and discussed the possibility of forming inter-regional bodies to deal with union in internal correspondence and meetings (Wakstein 1961).

As early as 1907, some individual executives in the NAM had advocated a form of national federation of employers that would counter organized labor directly.

J.W. Van Cleave, the president of the NAM, addressed member firms of the NMTA, and thought that the modal manufacturing firm should have a preference for repressive coordination at the national level, as well as across sectors. In particular, he voiced support for a national federation:

―To a large degree we have organization among employers already, but we do not extend the principle far enough. We must supplement organization with federation. The National Metal Trades Association, the founders‘ Association, the National Association of Manufacturers, the National Typothetae and the other organizations of great interests should be brought into constant and intelligent cooperation. All these and other great national organizations of employers should come together in a spirit of mutual fraternity and establish a central body in which each would be represented. This central body, or council, could deal with the interests common to all organizations. These interests are not only large in number, they are large in magnitude. And in each particular they are rapidly expanding...One of the evils which every organization of employers is frequently called upon to combat is silly or pernicious legislation…‖ (Van Cleave 1907)

Van Cleave singled out his dissatisfaction with employers‘ existing individual state

308 level and somewhat ad hoc approach to organization, regarding labor legislation and unionization:

―I have said that each organization of employers is often called upon to combat these menaces [demands raised by unions] to its activity and prosperity. But each organization acts separately. Each fights its own battles, in its own time, and in its own way. Each makes its own experiments, and learns its own lessons from its own losses. And the losses are frequent and humiliating. By an intelligent concentration of effort these reverses to us would be avoided, and we would be spared the losses and humiliations which beset us under our present lack of cooperation. We could avert these losses in one or other of two ways. By coordination of effort in diffusing education among the people we would have a chance to head off all foolish or vicious propositions before they were introduced in our national or state legislature, or by a prompt massing of forces, we could defeat them after they were introduced…Let us have a federation of all the organizations of employers, in which each organization, while preserving its independence as absolutely as at present, will come into cooperation with all the other organizations in a great council, in which each will have a choice and in which all will meet on terms of fraternity and equality‖ (Van Cleave 1907)

From Van Cleave‘s address, and other early NAM documentation from this period, it is unclear whether he intended for a federation that would do more than coordinate information flow and lobby. At the time of this address, the NAM had at the time no explicit labor-oriented institutional apparatus that would have supported such a massive integration of local-level employer organizations; it is unclear whether the proposal there could have even been implemented. But the expressed initial preference for such an organization raises the question of why it did not emerge in the wake of local-level Open Shop movements after WWI.

In the immediate post-WWI period, several influential leaders of regional employer organizations met to discuss a national federation; one leader was Andrew

Allen of the Associated Employers of Indiana (AEI), whose address was noted at the outset of this paper. Allen was one of the most vocal proponents of further developing

309 a national federation that would coordinate local-level Open Shop movements. In

1919, the AEI undertook a research endeavor to assess local employer-organization interest in such a federation, part of which involved collecting the survey data analyzed in the first empirical section. Allen desired that the AEI be the leader behind such a movement. The AEI had successfully challenged the efforts of the machinists‘ union in Indianapolis, and Allen wanted to take this victory nationally and establish a comprehensive collective institution of employers.

Allen‘s own correspondence with the NAM and with the executive committee of the AEI reveal a central tension among local employers‘ preferences. On the one hand, some organizations did not support the idea of placing organizational decisions in the hands of a national federation. A summary of Allen‘s committee report stated,

―In some quarters at first it was felt that a local association could do more effective work by ‗going it alone,‘ but when it was explained that a national federation does not presume to dominate the affairs of the local, but to offer such assistance as the local could not possibly obtain otherwise, the local invariably has favored ‗going in.‘‘200

This concern reveals that local associations that already existed had concerns about giving up autonomy of management of labor relations to national-level employer organizations.

Internal correspondence between Allen and officials of other state and local- level employer organizations reveals some interest from local employer organizations in a national-level organization that could provide assistance, in the form of information about union activity or strike support. The records of the November 11,

200 This comes from articles in the Iron Trades Review , (Public Sponsors Open Shop Associations 1920)

310

1920 meeting of the AEI executive committee include a memo from Allen to other

AEI executives. The letter describes Allen‘s account of his attempt to get employers to sign on to his proposed ―American Plan‖ and his previous meetings with representatives of the NAM. Allen argued that the AEI had been ―besieged on all sides by local employers‘ associations in all sections of the country‖ to establish a central national organization to assist businesses that would ―[bring] about the exchange of information, counsel, advice and mutual assistance and guidance for the national good in association endeavor with respect to right industrial relationships.‖

(Allen 1920). The minutes of this meeting also note Allen‘s belief that the execution of the ―plan of mutual cooperation in behalf of the open shop on a national scale‖ would start in December of that year. The AEI itself would presumably have had a large administrative role in such a national-level organization, although this is not detailed in the letter.

The precise outline of the AEI plan that Allen had come to advocate, stemming from internal AEI suggestions, as well as consultations with ―hundreds of

Associations throughout the country,‖ is unknown. But Allen details a number of aspects of the proposed organization. The new organization would have centralized existing state-level ―American Plan Associations‖ (i.e., state-level employer organizations) into a national organization based in Washington. This national organization would have required annual meetings from representatives of all state- level employer organizations; had national-level employment of ―field agents‖ to promote the Open-Shop organization (partially funded by state-level organizations); clear hierarchy of state and city-level employer organizations and coordination among

311 them; establishment of an information exchange service; inauguration of a department to deal with standardization of employment contracts and civic and industrial coordination; and funding of the plan through taxation of existing employer associations to support these measures (Allen 1920). Such a proposal would have been institutionally quite different from the status quo, characterized by the patchwork nature of the Open Shop movement and the tremendous local-level variation in repressive employer organizations.

It is difficult to assess the counterfactual of how many of the above measures would have been implemented, had Allen‘s movement to nationalize employer coordination succeeded. The proposals discussed by the AEI were not completely based off of other national organizations like the NMTA, which had more explicit rules regarding support for lockouts and support for member firms during strikes. The

AEI meetings on this subject do not reference the existing NMTA or NEA institutional structures and rules. So it is unknown how much of an overhaul to the status quo that the AEI proposal would have been, in comparison to the then locally-driven Open

Shop movement. However, given the lack of centralized efforts that actually directed the Open Shop movement, and absence of national-level institutional apparatus, it seems the AEI proposal would have constituted a fairly radical departure and overhaul from locally-driven status quo.

The Death of the Federation Movement

The AEI plan foundered because the necessary support required from existing national organizations, particularly the NAM, did not exist. The actual death of the

AEI proposal occurred when the NAM refused to consider it in the NAM Open Shop

312

Committee (OSC), on November 16, 1920. Allen himself attended this meeting to help influence the institutional goals of the OSC, and broached his proposal to the

OSC committee members (Allen 1920). But key NAM officials including James A.

Emery, the NAM legal counsel, were present. The NAM OSC did not consider Allen‘s proposal formally after Emery expressed his opposition (Emery 1920). Emery held particularly important sway in the organization. In his correspondence with AEI member firms and other regional employer organizations, Allen expressed disappointment at the NAM and NEA‘s shelving of the nationalization plan.

Why did the NAM and other affiliated national employer organizations stop

Allen‘s AEI plan to nationalize the OS movement? Two reasons were given. The first reason was that nationalization of an employer movement would radicalize labor, and just as critically, inflame public opinion against employers. Walter Drew, the president of the pro-Open Shop NEA, wrote Allen on October 29, 1920: ―The appearance of a national organized movement for the purpose of crushing labor unions, as it would be interpreted, would be unfortunate…If the employers make the move that you suggest, the public will say, ‗ it is only the same old fight between capital and labor,…[Our] worst enemy at this time is the employer who attempts to take advantage of the reaction against the unions in order to return to old methods of exploitation and unfairness‖ (Allen 1920, 2). While Drew was one of the most vocal proponents of the Open Shop philosophy, he also took pride that the movement‘s goals had been attained at the local level without arousing public anger against employers (Fine 1995). The NAM also shared Drew‘s concern, regarding federation, that the ―appearance of a new national association would give organized labor the

313 opportunity it desired to start counter propaganda on the ground that capital was making preparations to curtail the power of trade unions‖ (Allen 1920, 1). Drew‘s concerns may have been cold comfort to counties that either could not form an OSA, or could not get NEA assistance. They also illustrate an inherent tension in the differing strategies of firms when dealing with workers—they had to pitch their battle as partly ideological, no doubt upsetting some firms that found it advantageous to collaborate with unions, but they also had to restrain from nationalizing the policies that logically followed from the rhetoric. Existing employer organizations viewed local-level strategies to be sufficient in dealing with union demands.

A second explanation for the NAM killing of the AEI plan was that AEI plan would have posed a serious threat to the NAM‘s desire to be the institutional leader of employer actions. The NAM had already in May 1920 established the OSC to concern policy proposals regarding union threats, and AEI leadership on the matter would have detracted from NAM‘s hegemonic position on the matter, which was needed to attract and maintain membership, the financial lifeline of the organization (Steigerwalt 1952);

(Griffin, Wallace, and Rubin 1986). The NAM did not want to cede any control to the

AEI. As Emery himself stated: ―…what is needed is not more organizations, but more co-operation; that new machinery is not necessary for greater efficiency and that conservatism is better than an unnecessary display of aggression‖ (Emery 1920).201

Emery expressed to NAM leadership the importance of turning the NAM into the primary institutional advocate for labor relations issues of member employers, as well

201 These come from the Minutes of NAM OSC Meeting, November 16, 1920. The original source of this documentation is the OSC Meeting minutes from the National Association of Manufacturers Accession 1411. Series VII Box 127. Hagley Library Museum and Archive. Wilmington, DE.

314 as for potential member firms.202 In fact Allen, the initial entrepreneur of nationalization, expressed a suspicion in his letter, after his plan was shelved, that the

NAM voiced disapproval because ―… prominent eastern national associations interested in the furtherance of the open shop, [voiced] disapproval of the proposed plan to coordinate local associations‘ activities because (a), they either misunderstood the motives and plans or (b) were fearful that the Associated Employers [of

Indianapolis] would be successful in doing the very work these national associations should have undertaken and accomplished long ago‖ (Allen 1920. In a meeting of the

NAM‘s OSC in November 1920, Walter Drew of the NEA suggested that Allen from the AEI ―…should be given to understand that the great national associations would not go along with him in the plan proposed.‖203

What exactly did the NAM‘s Open-Shop Committee do, if its committee chairs and parent institution did not want to take the federation approach and really create a national institution with repressive coordination? Their efforts in centralization of information and lobbying were still considerable, but they did not constitute the collective strategy that was common in other employer associations in European states. The committee was originally proposed by Counsel James Emery to President

Stephen Mason on February 12, 1920. In Emery‘s letter, his principal concern was unions‘ attempts to ―destroy‖ independence of the judiciary, and that an OSC should promote the ideology of anti-unionism. The OSC‘s initial responsibilities included

202 This sentiment came from a letter from James Emery to Stephen Mason, February 12, 1920. National Association of Manufacturers Records. Accession 1411. Series VII Box 127. Hagley Library Museum and Archive. Wilmington, DE. 203 NAM OSC meeting notes, November 16, 1920. National Association of Manufacturers Records. Accession 1411. Series VII Box 127. Hagley Library Museum and Archive. Wilmington, DE.

315 compilation of information, presenting it to member firms, contacting and persuading the media, and becoming the principal medium of ―propaganda.‖204 The NAM succeeded in this information clearinghouse role.

But Emery was particularly concerned with limiting NAM intervention in regional employer movements around the country. The OSC codified what had already been a local movement. The newly created OSC did not obligate local employer associations at all to the dictates of the NAM board. As reported in the ITR,

―…local associations are not affiliated with one another in an organized movement, nor are they under specific obligations to one another, although they are promised the active support of a national association.‖205 The OSC committee resolved only to have informal cooperation between the NAM and local bodies.206 There were explicit provisions in OSC meetings for the NAM not to interfere in local employer associations or open-shop movements, and to leave local organizations to engage in any militant or anti-labor activities.207 In fact an NAM administrator, J. Constantine argued that regarding Open Shop policy, the NAM in ―cannot actively enter the field of specific local labor troubles.‖208 The most succinct expression of this sentiment came from Noel Sargent, head of the OSC during this period:

―The Committee…opposed recommendation by Mr. Nelson of Tacoma that the National Association of Manufacturers should attempt to coordinate the work of local bureaus. Such coordination can best be handled by trade bodies, such as the National

204 Letter from James Emery to Stephen Mason, cited in Wakstein, ―The Origins of the Open-Shop Movement, 1919-1920,‖ 473. 205 (Wages Working to Normal Basis 1921), 7-10, 8. 206 NAM OSC Meeting Notes, October 10, 1922. National Association of Manufacturers Records. Accession 1411. Series VII Box 127. Hagley Library Museum and Archive. Wilmington, DE. 207 NAM OSC Meeting Notes, December 14, 1923. National Association of Manufacturers Records. Accession 1411. Series VII Box 127. Hagley Library Museum and Archive. Wilmington, DE. 208 NAM OSC Meeting Notes, February 6, 1925. National Association of Manufacturers Records. Accession 1411. Series VII Box 127. Hagley Library Museum and Archive. Wilmington, DE.

316

Metal Trades Association. Any such coordination on a national scale by a national organization such as the N.A.M. would moreover tend to substantiate union claims of central direction of the open shop movement. The Committee expressed opposition to any such central direction or control, believing that the Open Shop Department should remain as an agency of research and advice, with the open shop movement continuing its spontaneous character.‖209

This qualitative internal evidence supports that conclusion that non-federation of repressive coordination was a choice on the part of employers, particularly those in already incumbent positions, and that institutions like the NAM wanted to keep the status quo.210 In these particular discussions from the OSC there is little mention of partisanship of the government or discussion of whether certain kinds of governments would better support their decisions on this matter, although there had been institutional connections between the NAM and the Republican Party. It may be the case that they felt this de-centralized form of institutionalization would be invulnerable to future partisan intervention.

This narrative of the non-development of a national-level peak federation in the inter-war period demonstrates some of the mechanisms posited in Chapter 3. In particular, it shows that incumbent employer institutions found it too costly to try and form further collective associations. This evidence can be interpreted with a logic that is consistent with the idea that different costs of repression across different localities made a peak-level confederation that set a uniform policy too difficult to adopt.

Further evidence that details this particular mechanism is warranted, though.

209 Noel Sargent, OSC Meeting, February 6, 1925, pg. 4. Cited in Wakstein, ―The Origins of the Open- Shop Movement, 1919-1920.‖ 210 This evidence and claim here are not inconsistent with Swenson‘s observation that a national multi- employer lockout in the United States would have been difficult due to the size of the state. The evidence suggests that even if such a lockout were possible, existing nationally based employer organizations thought local-level coordination was sufficient. See Swenson (2002).

317

Section 6: Conclusions

This chapter presented a micro-level test of the theory explicated in chapter 2: the redistributive threat from workers and industrial heterogeneity can account for the emergence of specific repressive kinds of firm strategies. The micro-level response of adoption of open-shop associations in response unionization confirms the first prediction of the theory. This supports the idea that low to moderate threats from workers—political and economic—give firms incentives to be repressive in response.

Part of the nature of this relatively lower threat is due to the non-revolutionary and relatively moderate nature of the Democratic Party, relative to its European counterparts. It is of particular importance that these variables predict a unique repressive and anti-labor strain of firm activity. However, as predicted, by the theory, low to moderate threats from workers meant firms could still repress in response. Part of the nature of this relatively lower threat is due to the non-revolutionary and relatively moderate nature of the Democratic Party, relative to its European counterparts.

The fact that a nascent, anti-labor coordinated movement only became possible after the war-time imposition of wage compression and encompassing agreements, and became difficult to enforce across localities, is evidence of the importance of variation in repression costs in explaining collective action. I suggest that from a broader perspective, this perverse strain of coordination played a large role in making the

United States the liberal-market economy that it is today. The advent of the Great

Depression and political re-alignment it induced brought a burst of pro-labor legislation in the 1930s, resulting in a reorientation away from collective repression as

318 a strategy. The National Recovery Act (NRA) allowed for legalized cartelization and allowed for collective bargaining (Hawley 1995). In the next chapter, I move beyond the United States and test the theory with evidence from a much wider sample of cases.

319

Figure 7-1:

320

Table 7-1: Determinants of County-Level Repressive Employer Coordination Model 1 Model 2 Model 3 Logged 1.26*** 1.08*** 1.09*** Population (.082) (.086) (.088) Logged per .30*** .27*** capita output (.069) (.076)

Logged per .11 capita # firms (.104)

Union threat

Union growth

Democratic Pres. vote share -14.66*** -14.44*** -13.67*** Constant (.876) (.771) (1.007)

Observations 3048 2857 2857

Adj. R- .47 .51 .51 squared Robust standard errors in parentheses, clustered by county. *** p<0.001, ** p<0.01, * p<0.05

321

Table 7-1: Determinants of County-Level Repressive Employer Coordination Model 4 Model 5 Model 6 Logged 1.15*** 1.09*** 1.20*** Population (.261) (.256) (.282) Logged per -.03 .01 .01 capita output (.141) (.153) (.148)

Logged per .55* .60* .32 capita # firms (.270) (.280) (.334)

.13*** .21*** .22*** Union threat (.030) (.047) (.049)

-.17* -.20* Union growth (.084) (.085) Democratic -.01* Pres. vote (.005) share -10.60*** -9.99*** -12.26*** Constant (2.944) (2.912) (3.703)

Observations 459 456 456

Adj. R- 0.57 .57 .58 squared Robust standard errors in parentheses, clustered by county. *** p<0.001, ** p<0.01, * p<0.05

322

Table 7-2: Changes in Probability of OSA Given Levels of Union Threat Union Predicted threat Probability level of OSA adoption

1 .01 (.005, .018) .02 3 (.003, .043) 0.05 5 (.012, .093) .12 7 (.057, .19) .17 8 (.084, .26) .24 9 (.11, .36) .31 10 (.15, .47) .39 11 (.18, .60) .48 12 (.23, .72) .56 13 (.29, .84) .73 15 (.44, 1) .80 16 (.52, 1) .90 18 (.69, 1) .96 20 (.81, 1) Predicted probabilities based on holding all covariates, except for Unionthreat variable, at means. 95% confidence intervals in parentheses. n = 456.

323

Chapter 8: Cross-National Evidence

The previous chapters demonstrate the existence of the repressive collective strategies, test the mechanisms of the theory by exploiting sub-national variation at the local level, and use extensive qualitative evidence regarding the decisions of firms and employer associations. In particular, concerns about responding to organized labor and labor parties mattered for decisions of pre-existing employer organizations to engage in repression.

In this chapter I demonstrate the portability of the theory and the importance of the typology of different kinds of firm strategies. I show that the broad historical experiences of the development and change in functions of employer organizations across countries are consistent with changes in the posited variables of the threat from workers and exogenous changes in industrial heterogeneity can be linked to the hypothesized variables. I show that the mechanisms of the costs and variance in costs of repression affected the decisions of firms to form repressive employer associations.

I present comparative historical case evidence from Germany, the United States,

Sweden, Denmark, Britain, France, Australia, and Japan. The evidence demonstrates the validity and importance of the new conception of firm strategies in Chapter 2, and shows how the proposed theory can better explain the cross-national and temporal variation. Section 1 summarizes the approach and general expectations from cross- national trends. Section 2 discusses each of the cases. Section 3 concludes.

Section 1: Summary of Cross-national Evidence and Approach

The case evidence confirms the main predictions of the theory. The stylized facts from the case evidence are consistent with the theory: first, the modal initial

324 employer organizations were repressive. They arose as an immediate response to workers‘ attempts to unionize and sought to resist worker demands. The cases of collective collaboration with workers only occurred in a specific set of industries which were not always large or politically influential. Firms generally initially adopted collective repressive strategies, and only switched to collaborative strategies when the threat from workers increased dramatically. Early employer organizations, whether locally or nationally, formed to push back against workers. Firms initially organize when they can as employer organizations at the sector level: they organize to work against unions that are generally demanding higher wages, control over working training and wage schedules, and other infringements against firm ―managerial authority.‖ When possible, firms engaged in collective repressive strategies against initial unions to maintain managerial authority and prevent union dictation of wages and working conditions. But, if the threat from left-wing parties rose and costs of repression of workers also increased, such firms had incentives to switch to collaborative strategies.

Second, the evidence also shows that national level collective strategies are difficult to maintain because they involve bargaining between sectors with divergent interests, in particular sectors with higher variance in the costs of repression. When industrial heterogeneity increases, the variance in the costs of repression within an economy increase, and sector members and firms have incentives to defect. This is evidenced in the devolution from inter-sector to single-sector level strategies, or as firms defect from sectors. An observable implication of this is the lack of success in some countries of national employer organizations in engaging in long-term successful

325 lockouts, or in failures of employer organizations to retain membership. Where national multi-industry lockouts and organizational use of strikebreakers as a collective repressive strategy were not possible, such as in the United States, other repressive strategies were used, including state intervention.

Third, we observe that changes in strategies occur because firms do not want to the state to bear the entire costs of repression or enforcement. This accounts for the pre-emptive nature of some cross-class agreements, most prominently in Germany,

Sweden, and Denmark. Finally, many of the major changes in firm strategies occurred within country time periods where the electoral system did not change. This casts doubt on the size of the causal impact of electoral systems on firm strategies.

Throughout discussion of the cross-national evidence, we observe some common initial phenomena that should generalize to a wider set of states. Employer organizations, if they form, form at the sector level against unions that are generally demanding higher wages, control over working training and wage schedules, and other infringements against firm ―managerial authority.‖ In some cases, like the United

States, these initial employer organizations were able to inflict great damage against workers‘ movements, by allying with major firms, other employer organizations, or with assistance from the state. American individual and collective repressive strategies were successful, and pitched battles today over worker legislation are a consequence of this initially successful strategy that put unions on the long-term defensive. In other cases, like Germany, the rapid rise in support for organized labor, the concerns about repressing workers towards the end of WWI, and the initial electoral threat posed by the Social Democratic parties after WWI created incentives

326 for firms to switch to collaborative strategies. In other cases, like the Scandinavian states, firms pursued initially collectively repressive strategies, but found it less costly to engage in collaboration, with one of the goals being wage restraint. The important outcome of other countries, including Britain and France, is the failure of collective strategies—repressive or collaborative—to stay collective.

The broader conception of firm strategies described in chapter 2 means it is difficult to test cross-national correlation relationships between the hypothesized variables and changes in firm strategies, in a multivariate regression framework.

Ideally, one would estimate multinomial probit models for a complete dataset for the period of 1880 to the present, with each kind of firm strategy as a possible outcome.

One of the difficulties with the approach, as discussed in Chapter 2, is that there is little annual fluctuation in the independent and dependent variables; rather, there are shifts in the qualitative kinds of firm strategies adopted across time periods. With the universe of observations for the set of OECD states, there are too few observations and too little change year to year to estimate such a model without large standard errors for each coefficient. Moreover, as discussed below, a theory should be able to explain the substantively large shifts in industrial regimes that occurred. Current cross-national regressions that link employer coordination to electoral systems (or vice versa) depend on a small number of observations, and as discussed extensively in Chapter 2, have too narrow a theoretical conception of employer coordination. Moreover, such estimations can be sensitive to specifications (Martin and Swank 2008; Cusack,

Iversen, and Soskice 2007).

327

As an alternative approach I show how the major changes in firm strategies in a series of countries can be traced to changes in the variables hypothesized in the theory chapter: threat from the left-wing party and exogenous changes in industrial heterogeneity. The case study approach of course makes ruling out alternative explanations more difficult, but I show that the changes in firm strategies within countries are consistent with my theory. Where possible in the interpretation of the historical evidence I demonstrate how existing explanations are insufficient in explaining changes in firm strategies. In particular I show changes in firm strategies across or within states are uncorrelated with changes in the electoral system (Martin and Swank 2008), and I also show how structural conditions as posited by Swenson, while important, are insufficient in accounting for the main outcomes (Swenson 2002;

Swenson 1991).211 Swenson discounts the role that left-wing parties play in changing the incentives of firms to switch strategies on both the collective-individual and repressive-collaborative dimensions. The state played a crucial role in affecting costs of repressing workers, and also incentives to act collectively. As Table 8-1 shows, from this cursory view of several cases, the existence of collective repressive strategies and their conversion to collaborative strategies has little to do with the electoral system.

Section 2: Country evidence

Sweden

211 The account described in Swenson (1991, 2002) is persuasive. Swenson hypothesizes that structural conditions allowed for cross-class alliances in Sweden, in particular within the export-oriented manufacturing sector. This account challenges previous theories that posit that labor power explains the lack of industrial conflict and that firms as a unified class grudgingly (or were unwillingly compelled) to accept centralization of industrial relations. For discussion of the Danish and Swedish cases, I rely on wage dispersion data and Swenson‘s own evidence to re-interpret what transpired regarding changes in firm strategies.

328

The proposed theory can explain the timing and qualitative changes in firm strategies in Sweden. In Sweden, firms organized at the sector and eventually national level to counter unions, engaging in collective repressive strategies. Once costs of repression increased and the state threatened to adopt a more interventionist role regarding costs of enforcement, pre-existing employer organizations had an incentive to switch to collaborative strategies. Throughout this process, low industrial heterogeneity allowed for high collective action within Swedish manufacturing sectors. As Swenson shows, wage restraint was as major goal of the early employer organizations. The challenge that faced the early sector-level Swedish employer associations was to force moderate counterpart unions to effectively constrain and sanction more extremist unions. This account re-interprets the evidence presented in

Swenson.212

Figure 8-1 shows that the initial firm strategies in Sweden were repressive in the early twentieth century. The first vertical line indicates the formation of initially collective repressive employer organizations. Once costs of repression rose, however, firms shifted to more collaborative strategies in the 1920s. This culminated in the

1938 Basic Agreement in Saltsjöbaden. The second (dashed) line indicates the transition to collaborative strategies. Recent research on Sweden focuses on the presumed collaborative nature of the famous Agreement between the LO and SAF

(which had begun negotiations for this agreement starting in 1936). Key provisions of the agreement called for centralized mediation of disputes, abolishment of targeting

212 Swenson 2002; 1991

329

―third parties,‖ and preventing industrial conflict. This agreement is often held up as the template of cross-class alliances.

But, initial collective firm strategies were considerably more complicated, and had a repressive nature early on. Early important organizations were in the building, timber, and manufacturing industries. The manufacturing organizations formed to wrest and secure managerial authority from unions. The first solid line indicates this adoption of collective repressive strategies. In 1902, the Swedish Engineering

Employers Association (VF) locked out the Swedish Metal workers‘ Union, to prevent the unions from asserting control over the labor supply and wages. The lockout punished nearly 14,000 workers. The result was an agreement between both actors in

1905, which resulted in some guarantees for workers, but was a major victory for the

VF because it established managerial control. According to the agreement, unions allowed firms to hire workers with skill levels and pay them as they saw fit; as in the

United States, unions conceded to a firms‘ practicing an ―open shop.‖ (Swenson 2002,

79). (In fact, as Swenson points out, unions here conceded on issues of craft control that their American counterparts did not).

The same repressive strategy was threatened on a much broader scale just shortly after this victory for the VF. In 1905, the national-level employer organization, the SAF, threatened to engage in its first major multi-industry lockout of the LO with similar goals, which would have threatened potentially all manufacturing activity in Sweden. This threat (it is not clear how credible it was) did induce the LO to agree to the December ―Compromise‖ of 1906. This agreement again not only recognized managerial authority, but it also explicitly permitted the use of such

330 strategies (the multi-industry lockouts) to continue. In fact, the LO agreed to let SAF firms use sympathy lockouts (targeting workers who had not participated in a strike).

And again in 1908, the sawmill industry would achieve a similar outcome with its workers. One of the goals of these strategies were, as discussed in the description of repressive strategies in the above section, was to force LO to spend much of its money supporting workers, weakening its capacity early on. But another key goal was of course securing managerial freedom.

The problem was that such repressive strategies could not be indefinitely maintained for several reasons. The first was that employers were concerned about increasing radicalism and local strikes that its lockouts were inducing. Firms realized after the initial successes of the actual lockouts and threats of lockouts that syndicalist or more recalcitrant unions could not be adequately ―restrained.‖ Swenson quotes the

SAF leader von Sydow, ―it is hard to declare a lockout when there is nobody to declare a lockout against‖ (Swenson 2002, 84). Thus eliminating these kinds of unions would have involved a higher cost of repression.

Second, labor scarcity in key manufacturing sectors was inducing up-bidding of wages. The SAF realized that the LO could assist in dealing with both problems.

The latter reason is especially highlighted as the most convincing for the cross-class coalition in Sweden: the Swedish manufacturers credibly threatened the LO with lockouts. This induced the LO to centralize wage institutions with the explicit goal of keeping militant building and construction unions in line and restraining their wages.

The result was wage restraint nationally and a reduction in inter-sector wage differentials. Manufacturing unions in the LO (who had long been intolerant of

331 militancy in the protected building trades), helped achieve that goal because of a preference for wage restraint and lower unemployment. In fact, Swenson interprets the LO as almost ―winking‖ when it communicates the SAF‘s threats of lockouts to building unions (knowing that such threats would induce cooperation). This pattern indicates that repressive strategies can be used to achieve the firm goal of wage restraint—but in the Swedish (and Danish) cases, the goal was to get a union authority to be able to effectively sanction its own potential defectors. This is evidence of the broader theoretical mechanism posited above about the importance of a degree of heterogeneity low enough such that moderate employer associations and their union counterparts are capable of punishing and deterring would-be defectors from collective strategies.

A third reason why a switch from collective repressive strategies was preferred was that the firms had an incentive to prevent the state from taking on costs of enforcement, and intervening in industrial affairs. Because pre-existing collective employer organizations were already successful, there was no need for the state to take over costs, either of repressive or collective strategies. An important aspect of the

1938 Saltsjöbaden pact is that the agreement between the SAF and LO was done preemptively to reduce state intervention into industrial affairs. That is, the SAF was willing to undertake some of the costs of enforcement of its collective strategies if it meant that the state would not be able to interfere. The SAF initiated such discussions with the LO to in anticipation of (and to prevent) the Riksdag from considering further anti-lockout legislation, which had been the SAF‘s main weapon for 30 years. The

LO, as discussed above, thought such weaponry was useful in helping to discipline

332 militant unions too.213 The key piece of historical evidence needed to prove this argument is some documentation from firms linking the 1932 electoral victory of the

Social Democrats to the belief that anti-lockout legislation would be pursued, to the

SAF negotiations regarding the pre-emptive nature of this collaborative agreement.

However, throughout the 1930s, the Swedish Social Democratic Party occasionally intervened, generally helping to enforce firm strategies of lowering wages.

As an important caveat to the previous discussion, the case of Sweden illustrates that an employer organization that engages in a repressive lockout that can both increase managerial authority and reducing local level union penetration and wages, but also increase union centralization and union power to sanction militant labor branches from conducting strikes. Note that a repressive strategy can have both goals. The strategy to achieve these results can be repressive, including lockouts on member and non-member ―innocent‖ workers. The repressive lockout, as Swenson details, can have as one of its goals centralization to promote wage restraint. But, note that this strategy is only chosen once it becomes too costly to repress all workers.

Given that the employer association cannot ―on its own‖ successfully defeat unionism at the local level, the next best strategy is to have labor organizations adequately sanction and control local affiliates.

Britain

The development of employer organizations in Britain since the late nineteenth century confirms the observable implications of the theory. Britain is characterized as a modern liberal market economy, with a general lack of coordination among firms.

213 Swenson, 2002, 114

333

This was not for want of trying, however—the story of British employer organizations is one of general failure to maintain collective action and to successfully execute collective strategies. The main story of the development of employer organizations in response to union growth in Britain is that the low left threat gave incentives for employer organizations to pursue repressive strategies. Collective strategies, due to extreme differences across sectors, meant that such collective action was difficult to maintain—employer organizations faced continual defection from member firms and sector organizations.

Early employer organizations formed at the regional and sector level in response to increased unionization and demands from craft-based manufacturing workers. In the 1870s there was a series of sector-level employer organizations that emerged in response to major strikes. They were designed to counter such demands.

Thus as predicted by the theory, due to a low threat from the left and higher wage dispersion initial firm strategies were individually repressive. Early British employers formed industrial associations after 1850 to counter unionism. They were explicitly anti-labor, and were formed to dictate labor conditions to organized labor; they were not formed to ―negotiate‖ with labor but rather to remove their influence from plants

(Yarmie 1980). The most important early regional anti-labor association was the Iron

Trades Employers‘ Association (ITEA), which formed in 1872. The ITEA resisted union demands by using strikebreakers and replacement labor, using ―black lists‖ and supplying data to firms on the union history of previous workers. But the ITEA was not a national level organization. There were several initial attempts at inter-sector collaboration, but they foundered because of conflicts between sectors over

334 sanctioning of member firms (in particular, conflicts between shipbuilding and manufacturing sectors). For example, in 1889, the Federation of Shipbuilding and

Engineering Employers tried to engage in initial lockouts, but collapsed because of differences in what each sector (shipbuilders and manufactured goods) wanted.

The clearest evidence from Britain that supports the theoretical predictions comes from the development of the most important and influential employer from the early twentieth century, the Engineering Employers‘ Federation (EEF). Figure 8-2 displays this strategy with the solid vertical line in 1896. The EEF formed in 1896 in response to a labor dispute in Clyde-Belfast, and was initially designed to battle with its main counterpart union, the Amalgamated Society of Engineers (ASE), over wages and control over the shop floor. The EEF engaged in two national lockouts in 1897-98 and 1922 against the ASE to force recognition of EEF member rights in regulating wages and apprenticeships (Zeitlin 1991, 1983). The ASE launched strikes and pressed for wage advances, and hour reductions, including a campaign for a maximum eight hour day—hallmarks of the aggressive craft labor movement. The EEF formed as coordinated ―resistance‖ to union demands, and the story of the EEF is one of central apparatus having limited success in getting local associations to comply with its rules, and preventing them from engaging unions independently. The particular issue was battle over control over worker assignment and compensation, or simply

―who would own the machines‖ (Zeitlin 1991, 55).

At the end of the first EEF directed lockout in January 1898, which involved around 45,000 workers, the ASE membership agreed to the basic right of managerial control over wages, and firms were free to do a number of things antithetical to

335 unions, such as the freedom to hire non-unionist workers and pay them equally as union workers (just as the VF in Sweden had won in 1905). But the agreement also established a framework for basic collective bargaining; EEF members recognized this framework as legitimate. The upshot of this agreement was that the ASE could not allow any strike from a local union to proceed until it had received national approval.

Like the SAF, the EEF threatened to ―discipline‖ the ASE by threatening national lockouts in response to failures of the ASE to restrain its members. EEF members regarded the lockouts in 1898, and again in 1922, as successful. But, the lockouts did not get rid of worker militancy—it was not just that the ASE was incapable of sanctioning its own members; it was the ASE felt that after the passage of years and especially after WWI, it could gain the upper hand in re-opening conflict about control over worker autonomy.

But unlike the SAF, the EEF‘s collective capabilities in forcing discipline were far more limited (Wigham 1973). The problem was that after 1922 the EEF had difficult in maintaining its collective strategy. In fact, even though membership in the

EEF increased after the 1898 lockout, it had difficulty getting member associations and firms to comply with organizational rules. There was concern about moves toward centralization. EEF head Alan Smith suggested that it was unlikely that the

EEF‘s membership would support a national lock-out over district-level wage disputes. An EEF subcommittee reported that the obstacle to a successful national lockout was ―want of common interest on the part of firms…textile machinery might be very busy when marine depressed. Similarly with other branches of trade a serious difficulty would accordingly arise in the way of getting the Federation to act as a body

336 on a wage question‖ (Zeitlin 1991, 62). The EEF neither was unable to successfully negotiate an enforceable agreement with the ASE after 1922, nor was is it able to sanction its own members, due to heterogeneity of interests between its own member sectors. In fact, EEF members expressed concern about financing strikes that they did not want to participate in; membership fell from 810 to 744 firms in the years before

WWI (Zeitlin 1991). Ultimately, the EEF could not consolidate its position after

WWI, indicated by the change in firm strategies in the dashed line of Figure 8-2.

Even after the success of the 1922 lockout, the EEF had no desire to merge with other employer organizations and give the executive powers of the newly formed

National Confederation of Employer Organization (NCEO) more authority over firms

(Zeitlin 1983). They also rejected other legislative measures that would have impinged upon managerial authority. At this point in time, the Labor government had not interfered much in industrial affairs, and the unions and labor party—in contrast to those in Sweden in the run-up to WWI—did not pose as high a revolutionary or extreme distributive threat. This suggest that the EEF had little incentive to accommodate and opt for more centralized collaborative strategies with the main unions (Wigham 1973).

The EEF is an example of a collective repressive strategy in response to a low threat from workers—a redistributive threat that engages firms to respond, but not enough such that a collaborative strategy becomes preferable to repression. We observe through the EEF a collective repressive strategy that had occasional success, but was unable to sustain long-term collective action. The ultimate evolution with the

337 advent of labor was a change from initially collective repressive strategies to individual repressive, and finally to individual collaborative strategies.

Denmark

Figure 8-3 shows the same key variables and change in firm strategies for

Denmark. The development of the Danish Employers‘ Confederation

(Arbejdsgiverforeningen, or DAF) and its relationship with the Danish Labor

Confederation (DSF) is consistent with the predictions of the theory. The Danish case is also characterized by the existence of initial repressive firm strategies and the incentives for firms to switch and stay collaborative. The first vertical (solid) line indicates the foundation of the first peak level organization of employers, the DAF.

The second vertical (dashed) line indicates the September Agreement which recognized workers‘ rights to strike, but also managerial autonomy. The third vertical

(dashed) line indicates the 1933 Kanslergade agreement, which substantively expanded labor rights. The Danish evidence supports the idea that firm strategies were initially repressive, but organizations switched strategies that resulted in recognition of managerial rights and centralization of union authority. As in Sweden, firms engaged in repressive strategies—lockouts—with one repressive goal: to lower wages of militant workers in protected sectors (generally in construction or building).

But to achieve that, they had to give central unions the authority to sanction their own.

As in Sweden, this was because it was too costly to engage in total repression. The main difference is that this transition occurred much faster in Denmark.

Employers in the building and metal trades formed the Danish Employers‘

Confederation in 1896. This initial organization had repressive tactics and goals: they

338 were formed in response to organized labor demands, and engaged in lockouts with the goal of reducing union penetration in plants. Its predecessor sector-level organizations (in building and manufacturing) had successfully won concessions on managerial authority and resisted local strikes at the local level, but still faced militant unions. As an ultimatum measure, in 1899 the DAF issued a multi-industry lockout that initially targeted carpentry workers who had gone on strike, and then workers in other industries. Importantly, these other potential targets were ―innocent‖ industries that had not yet experienced higher wage demands from workers, such as engineering industries. The result of this conflict was the 1899 September Agreement with the

DSF (preceding Saltsböjaden by nearly 40 years) (Galenson 1952).

This agreement is hailed as a cross-class agreement, but as with others, unions conceded the concept of ―managerial authority.‖ The agreement attained the DAF‘s main goals with the lockouts: unions agreed to sanction their own defectors

(particularly workers in unskilled unions). After the September compromise of 1899, there were concerted efforts by the DAF to force centralization of DSF behavior to make strikes more difficult to authorize. Ultimately, these repressive strategies became collaborative—unlike Sweden or the United States, this repressive period was far shorter. But, the DAF did resort to lockouts still throughout the 1920s and 1930s, pushing for eventual higher centralization of the DSF in the 1930s (Swenson 1991;

Swenson 1991; Schiller). As with the Saltsböjaden and Stinnes-Legien Agreements, the 1899 agreement was done with minimal state monitoring or interference, and done so preemptively.

France

339

In France, as in Britain, we observe initial collective repressive firm strategies that were not sustainable. These initial employer organizations ultimately reverted to individual collaborative firm strategies as enforceability of collective strategies decreased, and the threat from the left-wing parties increased. Figure 8-4 displays this trend. The secular decline in wage dispersion in the first half of the century would seem to predict increasing collective action capability among French employers, but in fact they had difficulty engaging in any kind of collective strategy despite multiple attempts. The vertical solid line indicates the advent of collective repressive strategies. The dashed solid line indicates that such strategies ultimately became individual collaborative after World War I.

After a generation of repression of workers, in 1884 the Waldeck-Rousseau

Act legalized trade unions, and there was an increase in city-level subsidized trade unions. Employer resistance to improving working conditions helped lead to a major mining strike in 1890. In 1892, the French government passed laws allowing for state intervention in labor disputes. The early twentieth century saw increasing unionization and support for the Socialist party. These trends posed a direct threat and spurred initial spikes in employer organizational activity at the local level.

The period 1889-1914 saw a parallel development of sector-level employer associations that had initially repressive strategies. Sector-level employer associations had formed early in the 1800s. The Sainte Chapelle group represented builders; the cotton industry had an organization in 1835. Prominent associations included manufacturing and mining associations (Confederation des groups commerciaux et industriels and Union des industries metallurgiques et minieres, both formed in 1901).

340

These associations had their largest membership concentrations in large cities and areas with high socialist political activity and unionization. The resistance to trade unionism and government intervention provided the most compelling reasons to unite.

Their initial goals were to defeat unions in strikes; as with other employer organizations, their instruments included using strikebreakers and organizing lockouts.

But, these organizations were not as well coordinated as their German and

Scandinavian counterparts: by 1910, over 80 percent of such associations had almost no central funds to support lockouts and member activities. Only 12 percent provided strike assistance. They were generally too weak to actually affect industrial dispute outcomes, and there was nothing approaching a multi-industry lockout in the cards for these firms. Local level employer associations in fact had little impact on the resolution—positive or negative—on industrial disputes in the early twentieth century

(Friedman 1991).

Local level employer associations were directly set up against socialist organizations and aimed to reduce SFIO vote share. But such associations, while capable of acting at the local level, never successfully engaged in national-level lockouts to compel and undermine French labor unions. As Friedman concludes, when French firms did decide to organize in response to radical unions, they did so

―too late,‖ and such efforts proved insufficient. This partially accounts for the contemporary fractured and ―intermediate‖ nature of French industrial-relations institutions: fairly uncoordinated employer organizations from before WWI. French employer organizations were even weaker than their British counterparts, who had similar difficulties in defeating unionism. The French case demonstrates that even in

341 the face of rising costs of repression and a credible threat from left-wing parties, employers may not be able to collectively act.

Australia

Australian firms never faced a high threat of expropriation from the labor party, or high costs of repression due to a historically weak labor movement that was dependent on the state for support. This absence facilitated initial collective repressive strategies. However, high heterogeneity and differences among sectors made sustaining peak-level organizations difficult. As Figure 8-5 shows, there was much volatility in wage dispersion in the early twentieth century. This helped undermine collective repressive strategies. The first solid vertical line indicates the initial collective repressive strategies, but they did not exist long. Even though there was increasing support from the state for organized labor throughout this period, the threat from support for labor was not large enough to give sector-level organizations incentives to switch to collaborative strategies. Thus the main trend we observe is initially collective repressive strategies transitioning to individual level repressive strategies. Australian employer organizations never formed a sustained peak organization that could wield national lockouts. Similarly, there was not a strong enough expropriation threat from the left-wing party to induce a full shift to collaborative strategies.

The initial modal strategy in Australia was a collective repressive strategy in response to the threat of trade unionism. Such organizations started initially as trade associations formed to lobby for tariffs, and later became employer organizations in response to local demands for the eight-hour day and local unionism. Dufty

342 documents a pattern, consistent with that described above, of employer organizations forming in the late nineteenth century in response to local threats from workers and strikes, successfully putting them down and reducing union growth, and then dissipating once the initial union threat had been neutralized. The largely decentralized nature of Australian government helps explain the initial autonomy of individual regional associations (Dufty 1984). As an example of this regional-sector level organization, the Victoria Boot and Shoe Manufacturers formed in 1883 to counter union wage claims in that industry. The Iron Trades Employers Association

(ITEA) formed in New South Wales for the same reason. Most other associations, including those facing craft unions (the Master Bakers‘ Association, Printing and

Allied Trade Employers‘ Association) had similar anti-union goals.

Australian national-level employer associations had similar goals to these local level associations—they countered organized labor and also lobbied against initial compulsory arbitration legislation (at the state and federal level). Collective repressive strategies arose to keep unions down and to prevent state interference in individual firm affairs. The Federal Council of the Chamber of Manufactures (which later became the Associated Chambers of Manufactures of Australia, the ACMA), was founded in 1903 because employers wanted a centralized body to lobby against enforcement of the recently passed arbitration legislation, which was perceived to grant workers more organizing rights and permission of state intervention. Another peak organization, the Central Council of Employers of Australia (ACEF), had similar anti-union and state intervention intentions. Both ―pro-labor‖ legislation and the concurrent repressive employer organizations developed before the formation of the

343

Australian Labor Party in 1910 (Plowman 1989). After the party began contesting elections, the employer organizations also expended much of their resources challenging arbitration court decisions, and tried to limit the federal jurisdiction of the courts. The Australian government played a pivotal role in granting workers‘ rights, and established compulsory arbitration tribunals: employers organized to counter the activities of the state as well as those of unions. Throughout this battle, Australian employer associations never tried to engage in any systematic collaborative strategies with workers.

However, these initial collective repressive organizations could not build on these efforts to form more well-coordinated peak institutions. During the 1920s, there were several failed attempts at cooperation between sector-level employer organizations (similar to such failed attempts in the US after the initial success of the

Open Shop Movement in the 1920s). The Central Council of Employers (ACEF) proposals to merge with other organizations were rejected, on the simple grounds that there was not enough commonality of interests to continue merging organizational apparatus. In 1937, the Australian Council of Employers‘ Federation tried to form a permanent committee representing all national employers‘ bodies, but another major organization—the Associated Chamber of Manufacturers—rejected that proposal on similar grounds. This cursory summary indicates that Australian firm strategies had fragile collective repressive strategies, which before World War II, became more individualistic in nature. The second vertical (dashed line) indicates the transition of these collective repressive to individual repressive strategies during the inter-war period (Plowman 1989, 1985).

344

Japan

In Japan, the absence of a strong labor party and labor unions through the twentieth century helps explain the general repressive strategies of firms until after

WWII. With the Japan case we can examine what strategies firms pursued in an industrializing state in the absence of a strong redistributive threat from a workers‘ party. (Unfortunately, wage dispersion data for this time period are not available).

The strategies through the early twentieth century were generally repressive at the individual firm and sector level, as labor unions were attacked by both businesses and the state. The general consensus is that incipient industrial unions were continually weakened throughout the first half of the twentieth (Tolliday and Zeitlin 1991; Levine

1984).

Japan also is characterized in the early twentieth century by sector-level employer organizations that tried to manage labor markets and control wages without labor unions. Hundreds of local and sector-level employer associations existed in the late nineteenth century. Early business associations were intended to appropriately

―segment‖ the labor market and prevent worker poaching among firms. For example, the Japan Federation of Cotton Spinning Trades Companies formed in 1888; this

Federation and other similar organizations established agreements among member firms to avoid excessive competition in worker recruitment. They divided the labor market into company-assigned territories (Crawcour 1978).

These early employer associations also generally opposed early government intervention to expand workers‘ rights. In 1884 these chambers opposed a government proposal to legislate minimal working conditions; in 1891 they also

345 opposed legislation from the Ministry of Agriculture to establish health and safety standards for workers. While the 1911 Factory Act supposedly guaranteed workers the right to organize in the workplace, it was not rigorously enforced—firms use an array of techniques, including lockouts and strikebreaking, to undermine unions. In

1917, major sector-level organizations and the zaibatsu managers formed the first truly national organization called the Japan Economic Federation (Nihon Keizai

Renmei), principally for the purpose of preventing recognition of labor unions (Levine

1984). (The zaibatsu managers argued, in addition, that unionism was contrary to

―Japanese values‖ and that a labor movement would lead to radicalism and disorder).

The Federation made a point of vowing to defeat union ―rights‖ (Levine 1984, 321).

Up until after WWII, then, Japan is characterized by consistent repressive strategies individually, and then collectively. While we do not have enough evidence to test the role of initial heterogeneity, the formation of key sector-level organizations indicates that individual repression played a critical role in stunting the growth of the Japanese labor movement.

Section 3: Conclusions

In this chapter I show test the theory with descriptive evidence from other advanced industrialized states. The evidence shows the validity of the new explanation for employer coordination and firm strategies that focuses on the decisions of firms in response to demands and threats from worker institutions, and the enforceability of such responses. Existing theories that focus on the mechanism of joint investment in skills and electoral systems are a critical step forward, and can account for the emergence of a specific kind of coordination within some industries.

346

However, they are less able to explain other important sub-national, cross-national, and temporal variation over a wider set of firm strategies. The theory here accounts for why firms are able to collectively act to form employer organizations, and sometimes why we observe the failure or collapse of such collective action. In particular, it explains why firms sometimes engage in repressive strategies against workers. The proposed theory does not arbitrate between the older ―traditionalist‖ view of industrial relations as a product of pure class conflict (such as that posed by

Korpi and Shalev) and the newer perspective which recognizes incentives of firms to work with unions in the provision of social policy or labor market outcomes. Rather, it provides an explanation for why we would observe both within a state (at the regional or sector level), and why we would observe bargains at the national level that explain the modal firm strategy. The case evidence presented indicates that the formation and evolution of initial employer organizations is a consequence of changes in the costs of repression as well as in the enforceability of institutions.

347

Table 8-1: Dates of firm strategies and electoral systems

Country Year of Year of Date of PR Collective Collective adoption Strategy collaborative Founding strategy switch (major EO) Denmark 1895 1899 1918 Germany 1905 1918 1918 Britain 1896 None None United States 1899 1933 None Sweden 1904 1938 1909 Japan 1916 None None Australia 1904 None None France 1895 None 1945

348

Figure 8-1: Changes in Firm Strategies in Sweden

349

Figure 8-2: Changes in Firm Strategies in Britain

350

Figure 8-3: Changes in Firm Strategies in Denmark

351

Figure 8-4: Changes in Firm Strategies in France

352

Figure 8-5: Change in Firm Strategies in Australia

353

Chapter 9: Conclusions

In 1941, the American labor scholar Whiting Williams and the head of the San

Francisco Employers‘ Council Almon Roth had an exchange in the magazine Rotarian that debated the responses of employer associations and firms to increasing enforcement of the Wagner Act and labor militancy on the west coast. The specific debate was if US employers should attempt a national confederation similar to those observed in Britain and Sweden. Both writers expressed sympathy for such a project, though Roth ultimately favored the idea while Williams opposed it. Williams argued that an attempt at federation would backfire, simply because obligating different local associations to confirm to national organizational strictures would sow discontent; the resulting defections would do more damage to dealing with unions. Roth countered that absent such organization, unions would succeed in defeating local employer federations. He wrote, ―The question…is…whether we shall curtail the power of highly organized existing pressure groups by setting up organizations to create that balance of power which is so essential…‖ (7). He argued that the threat from workers necessarily implied, ―What else can the employer do but organize to protect his business against local, state, and national organizations of employees…?‖ (6).

Williams had a different view of the degree of threat that firms faced: he concluded,

―…every well wisher of the American employer and employee alike will continue to hope that no similar crisis may come to enough factor gates throughout the land to force similar federation on a national scale‖ (5). In his essay, Williams contrasted the relative weakness of the US movement relative to its higher militancy in Britain and

Sweden, and argued that such militancy explained and justified the coordinated

354 responses of those firms, but that the absence of it in the United States meant that local level employer associations did not have the same urgency as their European counterparts (Roth 1941; Williams 1941).

The arguments in this exchange nicely illustrate the central dynamics spelled out in this dissertation, about the development of different regimes and degrees of collective action of firms. Williams and Roth clashed over the main issues of how to enforce agreements in localities and sectors that might be different (and differences within each), and whether the threats unions posed to firms merited such a drastic national response. In this dissertation I have proposed a new range of strategies that encapsulates the kind of responses these authors discusses, and outlined a parsimonious theory that can account for the immense cross-national and historical variation in firm strategies. I argue that the outcomes previous scholars have studied regarding the formation of different degrees of employer coordination should be studied as firm choices to choose strategies that can be collective or individual, or repressive or collaborative. I show that the redistributive threat posed by workers affected firms‘ decisions to repress, as the greater the threat meant more costly repression. Beyond a certain point, collaboration was pursued as a less costly option.

Similarly, I argue that structural conditions that inhibited collaborative action—best characterized by initial degrees of industrial heterogeneity—affected collaborative action because of higher variances in the cost of repression. This explains why such collective action was difficult in the United States and why exogenous changes in those structural conditions, from war-time interventions to the state to technological change—affect collective action.

355

I have used in-depth tracing of substantively and theoretically import cases—

Germany and the United States—to confirm the theory. In addition, I leverage sub- national variation in both states to test observable implications and mechanisms of the theory. Finally, in a set of other states, I demonstrate that the proposed theory can make better sense of the industrial outcomes we observe.

In posing this new theory and providing some exhaustive testing, this dissertation hopes to resolve some of the previous debates about the origins of modern divergence in capitalist systems. The most recent scholarship about the development of different coordination regimes takes as its starting point an assumption of cooperative coordination between firms and workers as the outcome to be explained.

This research, operating on innovative theorizing about joint incentives of firms and workers to invest in co-specific assets, focuses on a theory of skill formation and links that to institutions. This same pattern of scholarship also assigns an important role to electoral systems in promoting such co-investment. This literature links directly too to the modern conception of ―coordinated market economies‖ as synonymous with a set of institutions which protect those investments. An older generation of literature on the descriptive features and development of corporatism had a similar set of conclusions, though the logic was not based on skill-formation. Rather, it assumed that the ease of collective action of unions and associations in small states necessarily implied positive cross-class collective action.

The recent scholarship in some ways challenges the previous generation of literature which focused on the ability of the ―left‖ broadly conceived to impose its will. While these accounts have been convincingly challenged by authors showing

356 that firms in some cases shared preferences for social protection or collective bargaining, this literature also gave short shrift to the strategic incentives of firms and the issue of the costs and variance in costs of repression across an economy. In particular, it ignored the collective action problem among firms and took the formation of employer organizations for granted.

In the dissertation the proposed theory helps reconcile these literatures and shows how each one is insufficient to explain the range of firm strategies. As discussed in the empirical chapters on Germany, and in the cross-national chapter, the logic of joint investment in skills applies only to a subset of industries and regions.

The proposed theory can better explain the important outcomes of repression and provide an alternative logic of why we would observe collaborative strategies in some areas. With respect to the older literature on the threat posed by workers, the dissertation accounts for the outcome of the failure of collective action on the part of firms and demonstrates the symmetric strategic from the firm‘s perspective.

The variables and mechanisms discussed and tested here were particularly relevant for a historical time period in which there was great uncertainty about the redistribution consequences of the mobilization of workers. Further, given the upheavals wrought by WWI, increased state intervention in economies, and economic crises in the inter-war period, a plausible argument can be made that the kind of strategies this theory can explain are limited to a particular era. Yet efforts should still be made to test this theory on a broader sample of cases, in particular on middle- income developing states. First, the kind of repressive logic extensively documented is extant in employer associations across Latin America, East Asia, and Eastern

357

Europe. Second, given increased public sector intervention in these economies and competition from trading partners, it is clear that there will be exogenous changes in the ability for associations to maintain membership and continue effective collective action. Such problems face modern contemporary economies as well, with trade competition and de-industrialization.

I now consider some theoretical omissions and suggest possible extensions.

As previewed in chapter 3, future work should address concerns about endogeneity and demonstrate that the threat posed by workers and conditions that affect the enforceability of agreements affect firm strategies do have a causal impact on firms‘ choices, and that the actions of firms themselves have less of an impact on these independent variables. No doubt, as the case evidence displays, there is a dynamic process at work: successful repression can reduce the redistribute threat posed by workers; certain types of collaboration can of course reduce wage dispersion, thus increasing the enforceability of collective strategies. But the assumption of the theory has been partially proven by the evidence in the case studies: in each case, firms reacted to varying degrees of demands from workers (in some cases exogenous crises and shifts in the threat due to war-time conditions). With respect to collective versus individual strategies, the historical literature documents a frequent time lag between changes in worker efforts to mobilize and firm responses. Often employer associations do form in response to worker demands, and the responses are conditioned on the kinds of demands. There are important exceptions to this rule—for example, the Swedish SAF‘s encouragement of centralization of union authority

358 across industrial sectors—but the theory is partially confirmed by the noted ―lag‖ between unionization and employer response.

Still, a fuller version of the theory would incorporate the dynamic process— initial firm strategies that subsequently influence labor conditions and heterogeneity, which makes other kinds of firm strategies possible. In Chapter 3 I outlined a set of conditions under which workers will still incur costs to pose a redistributive threat, but a strategic setting that takes both actors equally seriously is warranted.

A second theoretical extension would be to seriously consider the interactive effect between industrial heterogeneity and the threat posed by workers. As discussed in Chapters 6 and 7 about the development of union incentives in Germany and the

United States, there may be asymmetries in the capacity of employer associations and unions to sanction defectors, or to organize in the first place. Other factors, such as ethnic heterogeneity and fractionalization, may have an additional inhibiting impact on the organization of workers. In some sense this suggestion is a return to the ―two logics of collective action‖ that Offe and Wiesenthal expounded upon, wherein they argued collective action among workers could be more difficult.

A third potential fruitful area of research is clarifying conditions under which firms and the state will share the costs of repression or enforcement of collective strategies. Throughout the cases, we observe a common impulse of firms, when forming organizations preemptive agreements with workers, to initially minimize state intervention. And yet under forms of collective repression and collaboration, the state ends up playing a role in management of both strategies. Under what circumstances was the state willing to bear more of the costs? If the state was willing, did this

359 necessarily imply that collective repressive organizations were less necessary? A similar question applies to collaborative arrangements. If one of the logics of the collaborative arrangement is that the state can assist in enforcement, why did employer organizations sometimes resist such intervention? There is an interesting disjuncture here between current models and the actual strategies that employer organizations pursued that should stimulate more historical gathering of evidence on firm preferences regarding state intervention. I have omitted discussion of electoral politics, wherein a set of voters may prefer further state intervention in the face of uncertainty.

A fourth extension is empirically demonstrating that the instances of collective collaboration we observe—the assumed form of collective action that previous scholars study—are the result of a switch from repression, and not co-investment in assets from firms and workers. While this is evident at the national level in many states, in individual sectors or regions this might not be the case. It would be particularly useful to apply the theory to the development of worker training institutions and demonstrate that their advent was because other costs of repression were too high.

Fifth, the logic of discussion between different factions of capital—between collaborative and repressive sectors—remains underspecified. It may be the case that certain structural conditions not identified here induce asymmetries in the capacity of collective action. Some firms may be able to collectively act, due to conditions in the sector, but others fail to. This creates an asymmetry where one ―faction‖ is able to impose its preferences. Similarly, if sectors with different costs of repression are able

360 to bargain on another dimension, higher differences in those costs need not impede collective action.

Another possible extension is more consideration of how broader kinds and purposes of coordination among firms effects industrial regimes. I suggest two. First, coordination on product markets—legal cartels—may facilitate collective action in the labor market. This was demonstrated at least in the German case; in the US inter-war period, strict judicial enforcement of anti-cartels did not inhibit mergers or repression by other means. Testing if these findings travel to other states would be a useful extension. Second, collective action in firms in developing states can have rent- seeking purposes. In fact, as discussed in Chapter 4, the initial employer associations in Germany developed out of protectionist preferences. The nature of rent-seeking and corruption as a result of employer coordination remains unexplored.

The broader agenda that this dissertation raises, beyond providing a new explanation for the emergence of different capitalist regimes, is about how certain economic actors use repressive strategies and institutions to inhibit democratization.

Existing accounts link some initial conditions, such as inequality, to democratization and redistribution through electoral channels. For example, a common mechanism is that these conditions affect demands of the median voter for spending. These accounts ignore, however, important non-electoral channels—the actions of firms—that affect the welfare of citizens. This dissertation hopefully will open up an agenda of more analytical and empirically rigorous examination of the relationship between firm strategies, democratization, and inequality.

361

Bibliography

Acemoglu, Daron, and James A Robinson. 2006. Economic Origins of Dictatorship and Democracy: Cambridge University Press. Allen, Andrew. 1920. Correspondence with AEI members. Indianapolis. Anderson, Margaret L. 2000. Practicing democracy: elections and political culture in Imperial Germany: Princeton Univ Pr. Ashenfelter, O., and G. E. Johnson. 1969. Bargaining theory, trade unions, and industrial strike activity. The American Economic Review 59 (1):35-49. Bain, G. S., and R. Price. 1980. Profiles of Union Growth: a comparative statistical portrait of eight countries: Blackwell. Bartolini, Stefano. 2000. The Political Mobilization of the European Left, 1860-1980: The Class Cleavage. Cambridge: Cambridge University Press. Berinsky, Adam J. 1999. The two faces of public opinion. American Journal of Political Science 43 (4):1209-1230. Beumer, Wilhelm. 1909. Mitteilungen des Vereins zur Wahrung der gemeinsamen wirtschaftlichen Interessen in Rheinland und Westfalen. Düsseldorf. Bittlingmayer, G. 1985. Did Antitrust Policy Cause the Greater Merger Wave. Journal of Law and Economics 28:77. Boix, Carles. 2003. Democracy and Redistribution. New York: Cambridge University Press. Bok, Derek C. 1971. Reflections on the Distinctive Character of American Labor Laws. Harvard Law Review 84 (6):1394-1463. Bonnett, Clarence E. 1956. History of Employers' Associations in the United States. First ed. New York: Vantage Press. Braunthal, Gerard. 1978. Socialist Labor and Politics in Weimar Germany. Hamden, CT: Archon Books. Bueck, Henry Axel. 1901. ———. 1908. ———. 1997. Mein Lebenslauf. Stuttgart: Franz Steiner Verlag. Bueck, Henry Axel, and Leidig. 1905. Der Ausstand der Bergareiter im Ruhrkohlenrevier: Hauptstelle Deutscher Arbeitgeberverbände. Carpenter, J. T., and J. Thomas. 1972. Competition and collective bargaining in the needle trades, 1910-1967: New York State School of Industrial and Labor Relations, Cornell University Ithaca. Crawcour, Sydney. 1978. The Japanese Employment System. Journal of Japanese Studies. Cusack, Thomas R, Torben Iversen, and David Soskice. 2007. Economic interests and the origins of electoral systems. American Political Science Review 101 (03):373-391. ———. 2010. Coevolution of Capitalism and Political Representation: The Choice of Electoral Systems. American Political Science Review 104 (02):393-403. Cusack, Thomas R, Torben Iversen, and David Soskice. 2007. Economic Interests and the Origins of Electoral Systems. American Political Science Review 101. Declare for Open Shop in 78 Cities. 1920. Iron Trades Review, August 12, 440.

362

Derber, Milton. 1984. Employers Associations in the United States. In Employers Associations and Industrial Relations, edited by J. P. Windmuller and A. Gladstone. Oxford: Clarendon Press. Dubin, Jeffrey A, and Douglas Rivers. 1990. Selection Bias in Linear Regression, Logit and Probit Models. Sociological Methods and Research 18 (2/3):360-90. Dufty, Norman F. 1984. Employers associations in Australia, edited by A. Gladstone and J. P. Windmuller. Dunn, Robert W. 1927. The Americanization of Labor: The Employers' Offensives Against the Trade Unions. New York: International Publishers, Inc. Ebbinghaus, Bernhard, and Jelle Visser. 2000. Trade unions in Western Europe since 1945, The societies of Europe. New York: Grove's Dictionaries. Emery, James. 1920. Minutes of Meeting, November 16, 1920: National Association of Manufacturers. Erdmann, Gerhard. 1966. Die Deutschen Arbeitgeberverbände im Sozialgeschichtlichen Wandel Der Zeit. Neuwied und Berlin: Hermann Luchterhand Verlag GmbH. Feldenkirchen, Wilfried. 1988. Concentration in German Industry 1870-1939. In The concentration process in the entrepreneurial economy since the late 19th century, edited by H. Pohl and W. Treue. Stuttgart: Franz Steiner. Feldman, Gerald. 1965. Army, Industry and Labor in Germany. ———. 1970. German Business Between War and Revolution: The Origins of the Stinnes-Legien Agreement. In Entstehung und Wandel der modernen Gesellschaft: Festschrift für Hans Rosenberg zum 65. Geburtstag, edited by G. A. Ritter. Berlin: Walter de Gruyter & Co. ———. 1973. The Origins of the Stinnes-Legien Agreement: A Documentation. Internationale Wissenschaftliche Korrespondenz zur Geschichte der deutschen Arbeiterbewegung 20:45-110. ———. 1981. German Interest Group Alliances in War and Inflation, 1914-1923. In Organizing interests in Western Europe: Pluralism, corporatism, and the transformation of politics, edited by S. Berger. Cambridge: Cambridge University Press. Feldman, Gerald. 1980. Arbeitskonflikte im Ruhrbergbau 1919-1922. Zur Politik vom Zechenverband und gewerkschaften in der übersiehtenfrage. Viertel jahreshefte für Zeitgeschichte 28 (2):1-56. Feldman, Gerald, and Ulrich Nocken. 1975. Trade Associations and Economic Power: Interest Group Development in the German Iron and Steel and Machine building Industries, 1900-1913. Business History. Fiedler, Martin. 1999. Die 100 größten Unternehmen in Deutschland - nach der Zahl ihrer Beschäftigten - 1907, 1938, 1973, und 1995. Zeitschrift für Unternehmengeschichte 44 (1):32-66. Fine, Sidney. 1995. Without Blare of Trumpets: Walter Drew, the National Erectors' Association, and the Open Shop Movement, 1903-57. Ann Arbor: University of Michigan Press. Fitch, John A. 1924. The Causes of Industrial Unrest. New York and London: Harper & Brothers.

363

Friedman, Gerald. 1991. The Decline of Paternalism and the Making of the Employer Class: France, 1870-1914. In Masters to Managers: Historical and Comparative Perspectives on American Employers, edited by S. M. Jacoby. New York: Columbia University Press. Galenson, W. 1952. The Danish system of labor relations: a study in industrial peace. Cambridge, Mass.: Harvard University Press, 1952. Galloway, Patrick R. 2007. Galloway Prussia Database 1861 to 1914: www.patrickgalloway.com. Geary, Dick. 1978. Radicalism and the Worker: Metalworkers and Revolution 1914— 1923. Society and Politics in Wilhelmine Germany. ———. 1989. Socialism and the German Labour Movement. In Labour and Socialist Movements in Europe Before 1914, edited by D. Geary. New York: St. Martin's Press. ———. 1991. The Industrial Bourgeoisie and Labour Relations in Germany 1871- 1933. D. Blackbourn u. R. Evans (Hg.), The German Bourgeoisie. Essays on the Social History of the German Middle Class from the Late Eighteenth to the Early TWentieth Century, London. Gladstone, Alan. 1984. Employers associations in comparative perspective: functions and activities. In Employers Associations and Industrial Relations, edited by J. P. Windmuller and A. Gladstone. London: Routledge. Gladstone, Alan, and John Windmuller, eds. 1984. Employers Associations and Industrial Relations. London: Routledge. Green, Marguerite. 1956. The National Civic Federation and the American Labor Movement. Washington, DC: The Catholic University of America Press. Greene, William H. 1997. Econometric Analysis: New Jersey: Prentice Hall. Griffin, Larry J, Michael E Wallace, and Beth A Rubin. 1986. Capitalist Resistance to the Organization of Labor Before the New Deal: Why? How? Success? American Sociological Review 51 (2):147-167. Groh, Dieter. 1978. Intensification of Work and Industrial Conflict in Germany, 1896- 1914. Politics and Society 8 (349-397). HA. 1904. Satzungen, edited by H. D. Arbeitgeberverbände. ———. 1907. ———. 1910. Mitglieder Verzeichnis Hauptstelle deutsche Arbeitgeberverband. Hall, Alex. 1974. By Other Means: The Legal Struggle against the SPD in Wilhelmine Germany, 1890-1900. Historical Journal 17 (2):365-386. Hall, Peter A., and David G. E. Soskice. 2001. An introduction to varieties of capitalism. Varieties of capitalism (S. 1-70):1-70. ———. 2001. Varieties of capitalism, the institutional foundations of comparative advantage. Oxford [u.a.]: Oxford Univ. Press. Hansen, Hal. 1998. Caps and Gowns, History, University of Wisconsin, Madison. Harris, Howell J. 1982. The right to manage: Industrial relations policies of American business in the 1940s: University of Wisconsin Press. Harris, Howell John. 2000. Bloodless Victories: The Rise and Fall of the Open Shop in the Philadelphia Metal Trades, 1890-1940. Cambridge: Cambridge University Press.

364

Hassel, Anke. 1999. The Erosion of the German System of Industrial Relations. British Journal of Industrial Relations 37 (3):483-505. ———. 2002. The Erosion Continues. Reply. British Journal of Industrial Relations 40 (2):309-317. Hawley, Ellis. 1995. The New Deal and the Problem of Monopoly. New York: Fordham University Press. Haydu, Jeffrey. 1988. Employers, Unions, and American Exceptionalism: Pre-World War I Open Shops in the Machine Trades in Comparative Perspective. International Review of Social History XXXIII (1):25-41. ———. 1999. Two Logics of Class Formation? Collective Identities Among Proprietary Employers, 1880-1900. Politics & Society 27 (4):505-25. Heidenheimer, AJ, and F Langdon. 1968. Business Associations and the Financing of Political Parties: a comparative study of the evolution of practices in Germany, Norway and Japan: Martinus Nijhoff. Herrigel, G. 2000. Industrial constructions: The sources of German industrial power: Cambridge Univ Pr. Horn, H., and A. Wolinsky. 1988. Worker substitutability and patterns of unionisation. The Economic Journal 98 (391):484-497. Hurvitz, Haggai. 1977. Ideology and industrial conflict: President Wilson's first industrial conference of October 1919. Labor History 18 (4):509-524. Industrie, Mitteilungen des Kriegsausschusses der deutschen. 1918. Ingham, Geoffrey K. 1974. Strikes and Industrial Conflict: Britain and Scandinavia, Studies in Sociology. London and Basingstoke: Macmillan. Johansen, Anja. 2004. Military involvement in the policing of French and German industrial areas, 1889-1914. European History Quarterly 34. Kaelble, Hartmut. 1967. Industrielle Interessenpolitik in der wilhelminischen Gesellschaft: Walter de Gruyter. Katzenstein, Peter. 1985. Small States in World Markets: Industrial Policy in Europe. Ithaca: Cornell University Press. Kaun, Heinrich. 1938. Die Geschichte der Zentralarbeitsgemeinschaft: der industriellen und gewerblichen Arbeitgeber und Arbeitnehmer Deutschlands, Wirtschafts Hochschule, Berlin. Kennan, J. 1986. The economics of strikes. Handbook of labor economics 2:1091- 1137. Kenworthy, L. 2001. Wage-setting measures - A survey and assessment. World Politics 54 (1):57-+. Kessler, Gerhard. 1907. Die deutschen arbeitgebervebände: Im Auftrage des Vereins für Socialpolitik: Duncker & Humblot. Kinghorn, Janice Rye, and John Vincent Nye. 1996. The Scale of Production in Western Economic Development: A Comparison of Official Industry Statistics in the United States, Britain, France, and Germany, 1905-1913. Journal of Economic History 56 (1):90-112. Knips, Achim. 1996. Deutsche Arbeitgeberverbände der Eisen-und Metallindustrie, 1888-1914: Franz Steiner Verlag.

365

Kögler, P. 1974. Arbeiterbewegung und Vereinsrecht: ein Beitrag zur Entstehungsgeschichte des BGB: Duncker & Humblot, Berlin. Korpi, W. 1978. The working class in welfare capitalism: Routledge & Kegan Paul Boston. ———. 1983. The democratic class struggle: Routledge Kegan & Paul. ———. 2006. Power resources and employer-centered approaches in explanations of welfare states and varieties of capitalism: Protagonists, consenters, and antagonists. World Politics 58 (2):167. Korpi, W., and M. Shalev. 1979. Strikes, industrial relations and class conflict in capitalist societies. British Journal of Sociology:164-187. Leckebusch, Roswitha. 1966. Entstehung und Wandlungen der Zielsetzungen, der Struktur und der Wirkungen von Arbeitgeberverbänden. Berlin: Duncker & Humblot. Letwin, William. 1965. Law and Econoic Policy in America: Yale Law Journal. Levine, Solomon B. 1984. Employers associations in Japan, edited by A. Gladstone and J. P. Windmuller. Mallmann, Luitwin. 1990. Perspektiven aus Tradition: 100 Jahre Gesamtmetall. Gesamtverband der Metallindustriellen Arbeitgeberverbände., ed. Köln: Deutscher Instituts-Verlag. Mares, I. 2003. The politics of social risk: business and welfare state development: Cambridge University Press. ———. 2003. The sources of business interest in social insurance - Sectoral versus national differences. World Politics 55 (2):229-+. Martin, Cathie Jo. 2007. Sectional parties, divided business. Studies in American Political Development. Martin, Cathie Jo, and Duane Swank. 2008. The political origins of coordinated capitalism: business organizations, party systems and state structure in the age of innocence. American Political Science Review 102 (2):1-18. Martin, CJ. 2006. sectional Parties, Divided business. Studies in American Political Development 20 (02):160-184. McCaffree, Kenneth M. 1962. A Theory of the Origin and Development of Employer Associations. Paper read at Proceedings of the 15th Annual Meeting of the Industrial Relations Research Association, December 27-28, at Pittsburgh, PA. Mommsen, WJ. 1978. Die deutsche Revolution 1918-1920. Politische Revolution und soziale Protestbewegung. Geschichte und Gesellschaft:362-391. Montgomery, David. 1989. The fall of the house of labor: the workplace, the state, and American labor activism, 1865-1925: Cambridge Univerity Press. Müller, Klaus. 1963. Politische Strömungen in den rechtsrheinischen Kreisen des Regierungsbezirks Köln (Sieg, Mülheim, Wipperfürth, Gummersbach und Waldbröl) von 1879 bis 1900: Rheinische Friedrich-Wilhelms-Universität. Niederschrift der Sitzung des Direktoriums des Zentralverbandes deutscher Industrieller. 1908. edited by Z. d. Industrieller. Heidelberg. Nocken, Ulrich. 1973. Inter-Industrial Conflicts and Alliances as Exemplified by the AVI-Agreement. Paper read at Internationales Symposium Industrielles System und Politische Entwicklung in der Weimarer Republik, at Bochum.

366

Nussbaum, H. 1966. Unternehmer gegen Monopole: Akad.-Verl. Offe, Claus, and Helmut Wiesenthal. 1980. Two Logics of Collective Action: Theoretical Notes on Social Class and Organizational Form. Political Power and Social Theory 1:67-115. Olson, Mancur. 1968. The Logic of Collective Action. Opel, F. 1957. Der deutsche Metallarbeiterverband. Hannover/Frankfurt/M. Open shop encyclopedia for debaters. 1921. New York: National association of manufacturers. Perlman, Selig, and Philip Taft. 1955. History of Labor in the United States, 1896- 1932. New York: Macmillan Company. Plowman, David H. 1985. Industrial legislation and the rise of employer associations 1890-1906. Journal of Industrial Relations 27 (3):283-309. ———. 1989. Countervailing Power, Organisational Parallelism and Australian Employer Associations. Australian Journal of Management 14 (1):97-113. Poensgen, Ernst. 1929. 25 Jahre Arbeitnordwest 1904-1929. Berlin: Verlag Von Reimar Hobbing. Prager, E. 1970. Geschichte der USPD: Entstehung und Entwicklung der unabhängigen sozialdemokratischen Partei Deutschlands: D. Auvermann. Preller, L, and F Tennstedt. 1978. Sozialpolitik in der Weimarer Republik: Athenaum. Proceedings of the First Industrial Conference, October 6 to 23 1919. 1920. Przeworski, Adam, and Michael Wallerstein. 1982. The Structure of Class Conflict in Democratic Capitalist Societies. American Political Science Review 76 (2):215-238. Public Sponsors Open Shop Associations. 1920. Iron Trades Review, November 11, 1339-1348. Raumer, Hans von. 1954. Unternehmer und Gewerkschaften in der Weimarer Zeit. Deutsche Rundschau, 428-431. Reich, Statistisches Jahrbuch für das Deutsche. 1909. In Berlin: Puttkammer & Mühlbrecht: Amt, KS. Reichert, Jacob. 1919. Enstehung, Bedeutung, und Ziel der ―Arbeitgemeinschaft‖. Reichs, Statistik des Deutschen. 1909. Verlag des Königlich Preussischen Statistischen Bureaus. Rosenbloom, Joshua L. 1998. Strikebreaking and the Labor Markets in the United States, 1881-1894. Journal of Economic History 58 (1):183-205. Roth, Almon. 1941. Should U.S. Employers Organize? Yes. Rotarian, 6-9. Ryder, AJ. 1959. The German revolution of 1918: Routledge and Kegan Paul. Sanders, Elizabeth. 1999. Roots of reform: farmers, workers, and the American state, 1877-1917: University of Chicago Press. Saul, Klaus. 1974. Staat, Industrie, Arbeiterbewegung im Kaiserreich: Zur Innen- und Aussenpolitik des Wilhelminischen Deutschland 1903-1914: Bertelsmann Universitätsverlag. ———. 1985. Repression or Integration? The State, Trade Unions and Industrial Disputes in Imperial Germany. The Development of Trade Unionism in Great Britain and Germany, 1880-1914. Schiller, B. Storstrejken 1909: förhistoria och orsaker.

367

Schmitter, Philippe. 1974. Still the Century of Corporatism? The Review of Politics 36 (1):85-131. Schmitter, Philippe C, and Wolfgang Streeck. 1991. The Organization of Business Interests: Studying the Associative Action of Business in Advanced Industrial Societies. In MPIfG Discussion Paper. Cologne. Schorske, Carl. 1955. German Social Democracy 1905-1917: The Development of the Great Schism: Harvard University Press. Silverberg, Paul. 1926. Das deutsche industrielle Unternehmertum in der Nachkriegszeit, edited by V. d. R. Mitgliederversammlung des RDI am 3. Und 4. September 1926 in Dresden. Silvia, Stephen J, and Wolfgang Schroeder. 2007. Why are German Employers Associations Declining? Arguments and Evidence. Comparative Political Studies 40: 1433-1459. Sollmann, W. 1918. Die Revolution in Köln. Ein Bericht über Tatsachen. Köln. Spencer, Elaine Glovka. 1984. Management and Labor in Imperial Germany: Ruhr Industrialists as Employers, 1896-1914. New Brunswick: Rutgers University Press. Sperber, Jonathan. 1997. The Kaiser's Voters. Cambridge: Cambridge University Press. Stegmann, D. 1974. Unternehmerpolitik zwischen Reform und Restauration. Sozialgeschichte Heute. Festschrift Hans Rosenberg, hrsg. von Hans-Ulrich Wehler, Göttingen:594-610. Stegmann, Dirk. 1976. Hugenberg Contra Stresemann: Die Politik der Inudstrieverbände am Ende des Kaiserreichs. Vierteljahrshefte für Zeitgeschichte 24 (4):329-378. Steigerwalt, Albert K. 1952. The National Association of Manufacturers: organization and policies, 1895-1914. Ann Arbor: University of Michigan Press. Steller, K.G. 1906. Die Bedeutung der Tarifverträge und die Lage der Industrie in Bayern r.d. Rhn Nürnberg. Streeck, Wolfgang. 1992. Interest Heterogeneity and Organizing Capacity: Two Class Logics of Collective Action? In Social Institutions and Economic Performance: Industrial Relations in Advanced Capitalist Economies, edited by W. Streeck. London: SAGE Publications. Suval, Stanley. 1985. Electoral Politics in Wilhelmine Germany. Chapel Hill: University of North Carolina Press. Swenson, P. 1991. Bringing capital back in, or social democracy reconsidered: employer power, cross-class alliances, and centralization of industrial relations in Denmark and Sweden. World Politics: A Quarterly Journal of International Relations:513-544. Swenson, Peter. 1989. Fair Shares, Unions, Pay, and Politics in Sweden and West Germany. Ithaca: Cornell Univ. Press. ———. 1991. Bringing capital back in, or social democracy reconsidered: employer power, cross-class alliances, and centralization of industrial relations in Denmark and Sweden. World Politics: A Quarterly Journal of International Relations 43 (4):513-544.

368

———. 2002. Capitalists against markets, the making of labor markets and welfare states in the United States and Sweden. Oxford: Oxford Univ. Press. Tänzler, Fritz. 1929. Die Deutschen Arbeitgeberverbände 1904-1929: Ein Beitrag zur Geschichte der deutschen Arbeitgeberbewegung: Otto Elsner Verlagsgesellschaft. Tenfelde, Klaus, and Heinrich Volkmann. 1981. Streik: zur Geschichte des Arbeitskampfes in Deutschland während der Industrialisierung: Beck. Thelen, Kathleen. 2004. How Institutions Evolve: Cambridge University Press. Thyssen. 1926. Mitteilungen des Vereins zur Wahrung der gemeinsamen wirtschaftlichen Interessen in Rheinland und Westfalen. Düsseldorf. Traxler, Franz. 2000. Employers and employer organisations in Europe: membership strength, density and representativeness. Industrial Relations Journal. 31 (4):308–316. Turner, H.A. 1970. Ruhrlade, Secret Cabinet of Heavy Industry in the Weimar Republic. Central European History:195-228. Ullmann, Hans Peter. 1976. Der Bund der Industriellen: Vandenhoeck und Ruprecht. VA. 1904. Satzung der Verein der Deutschen Arbeitgeberverband. Van Cleave, J.W. 1903. Trade Unions and the National Guard. The Bulletin of the National Metal Trades Association, 101-102. ———. 1907. "The Labor Question and Employers' Organizations." The Open Shop, 343-349. Varain, Heinz Josef. 1973. Interessenverbände in Deutschland. Köln: Kiepenheuer und Witsch. VDA. 1913. Jahrliche Bericht. VdESI. 1903. ———. 1916. Jahrliche Berichte. ———. 1917. Jahrliche Berichte. Vogt, M. 1970. Akten der Reichskanzlei. Weimarer Republic. Das Kabinett Müller II (2 vols., Boppard am Rhein, 1970), I 643. Wages Working to Normal Basis. 1921. Iron Trades Review, January 6, 7-10. Wakstein, Allen M. 1961. The Open-Shop Movement, 1919-1933, Department of History, University of Illinois, Urbana, Urbana. ———. 1964. The Origins of the Open-Shop Movement, 1919-1920. Journal of American History 51 (3):460-475. ———. 1969. The national association of manufacturers and labor relations in the 1920s. Labor History 10 (2):163-176. Wallerstein, Michael. 1989. Union Organization in Advanced Industrial Democracies. American Political Science Review 83 (2):481-501. ———. 1990. Centralized Bargaining and Wage Restraint. American Journal of Political Science 34 (4):982-1004. ———. 1991. Industrial concentration, country size, and trade union membership. American Political Science Review 85 (3):941-953. Wallerstein, Michael, and Karl Ove Moene. 2003. Does the Logic of Collective Action Explain the Logic of Corporatism? Journal of Theoretical Politics 15 (3):271-297.

369

Weisbrod, B. 1978. Schwerindustrie in der Weimarer Republik: P. Hammer. Westfalen, Mitteilungen des Vereins zur Wahrung der gemeinsamen wirtschaftlichen Interessen in Rheinland und. 1886. Mitteilungen des Vereins zur Wahrung der gemeinsamen wirtschaftlichen Interessen in Rheinland und Westfalen. Düsseldorf. Wigham, Eric. 1973. The Power to Manage: A History of the Engineering Employers' Federation. London: Macmillan Press. Williams, Whiting. 1941. Should U.S. Employers Organize? No. Rotarian, 1-5. Willoughby, William Franklin. 1905. Employers' Associations for Dealing with Labor in the United States. The Quarterly Journal of Economics 20 (1):110-150. Winschuh, Josef. 1932. Der Verein mit dem langen Namen. Geschichte eines Wirtschaftsverbandes, Berlin 182. ———. 1932. Der Verein mit dem langen Namen. Geschichte eines Wirtschaftsverbandes, Berlin. Yarmie, Andrew H. 1980. Employers' Organizationos in Mid-Victorian England. Review of Social History 25:209-235. Zeitlin, Jonathan. 1983. The Labour Strategies of British Engineering Employers, 1890-1922. In Managerial Strategies and Industrial Relations: An Historical and Comparative Study, edited by H. F. Gospel and C. R. Littler. Hampshire: Heinemann Educational Books. ———. 1991. The internal politics of employer organization: The Engineering Employers' Federation 1896-1939. In The Power to Manage? Employers and industrial relations in comparative-historical perspective, edited by S. Tolliday and J. Zeitlin. London and New York: Routledge. Ziblatt, Daniel. 2009. Shaping Democratic Practice and the Causes of Electoral Fraud: The Case of Nineteenth-Century Germany. American Political Science Review 103 (01):1-21. Zunkel, F. 1973. Die Gewichtung der Industriegruppen bei der Etablierung des Reichsverbandes der Deutschen Industrie. Paper read at Internationales Symposium Industrielles System und Politische Entwicklung in der Weimarer Republik, at Bochum.

370