Organization matters: Trade Union Behavior during Peace and War Ralf R. Meisenzahl Abstract This paper employs a comparative institutional analysis to explain the origins of different trade union organizational structures in Germany and Britain as of 1913 and their macroeconomic implications. We first model coordinated industry-wide unions and centralized employer organizations in Germany, on one hand, and fragmented unions and decentralized employer organizations in Britain, on the other hand, as two equilibria of a union-employer game. Next, I use WWI as a “natural experiment” to study the impact of the union structures on union strategies and bargaining outcomes during the war in a union-government game and discuss their long-run ramifications.

We were the worst organised nation in this world for this war... It is a war of munitions. We are fighting against the best organised community in the world, the best organised either for war or peace... David Lloyd George, Minister of Munitions, to trade unionists in Manchester 3.6.1915 In March 1892 delegates of the German Free trade unions voted for a new organizational structure: centralized, industry-wide unions with the Generalkommission der Freien Gewerkschaften Deutschlands (General commission of German’s Free Unions, henceforth Generalkommission), the elected leadership of all Free unions, as a powerful coordinating body for member unions. This centralized structure was unique for a large, industrialized country at Department of Economics, Northwestern University, 302 Arthur Andersen Hall, 2001 Sheridan Road, Evanston, IL 60208-2600, [email protected] I am indebted to Joel Mokyr for guidance and comments on this project. I also would like to thank Joe Ferrie, Sebastian Karcher, Chiaki Moriguchi, Kathleen Thelen, and the participants of the Third Max Planck Summer Conference on Economy and Society and seminars at Northwestern University and University of Wisconsin, Milwaukee for helpful comments. A Summer Research Travel Grant from the Center for International and Comparative Studies at Northwestern University is gratefully acknowledged.

1

the time. Over 1100 British trade unions, for instance, remained fragmented and the main overarching body, the Trades Union Congress (TUC), weak. So far the literature on trade unions has discussed the impact of different organizational modes on the postwar economy.1 However, the origins of the observed organizational modes and the consequences of each mode for the war economy was neglected.2 What caused this major institutional change in pre World War I Germany? And can disparate trade union structures account for the differences in union strategies during the first truly exogenous macroeconomic shock of World War I? This paper studies the origins and consequences of disparate union structures, using a comparative institutional analysis. This methodology laid out in Greif (2006) views institutions as multiple equilibria.3 We explain why multiple equilibria in trade union structure arise and provide detailed evidence on the sources of equilibrium selection in Germany and Great Britain.4 Divergent paths in union strategies as consequence of equilibrium selection lead to differences in labor allocation, strikes frequencies, and wage development during peacetime. The selected equilibrium is also the departure point for the different varieties of capitalism observed by Hall and Soskice (2001) in both countries. Moreover, we extend our analysis of the consequences of different union structures by examining the path dependent responses of trade unions to and during World War I. The Great War offers crucial insights as trade union organizations faced a new bargaining partner, the government. This exogenous change in the bargaining partner allows us to identify in the impact of previously chosen trade union structure on union strategies. As we will see, union strategies during World War I set the stage for industrial relations today. As such Germany and Great Britain, large industrialized participants in the Great War with different trade union structures, serve as important case studies. To better understand the mechanisms of equilibrium selection, we model the trade unions and employers choice over their organizational form as a coordination game. Olson (1982), studying macroeconomic outcomes, argues 1

For instance, Hoel (1990) and Mares (2006) discuss wage bargaining and employment, Lingens (2004) growth, Iversen and Soskice (2000) monetary policy and Hall and Soskice (2001) the structure of the economy. 2 Eisenberg (1986) studies differences in the early trade union organization. 3 See also North (1990) on multiple equilibria of institutions. 4 Greif (1993) and Moriguchi (2003) study equilibrium selection and consequences, using the same methodology.

2

that collective action problems on the trade union side might be substantial. Hence, we embed a collective action problem among trade unions in a bargaining game to determine the payoffs of the coordination game. In our peacetime model two equilibrium modes of organization exist. In one equilibrium both employers and trade unions choose a weak national organization with strong local entities and firm level bargaining (Great Britain) and in the second both establish strong national industry wide organization and national level bargaining (Germany). We argue that the expectations of German employers about the future structure of German trade unions triggered an “arms race” in which both sides overcame their collective action problems and centralized their respective organizations. Our peacetime model predicts that fractionalized trade unions will bargain for exclusive rights for their members and thereby distort the allocation of labor within and between firms. When bargaining with the employers, unions with strong national organizations will instead focus on the wage increases, internalizing externalities arising from distortions of labor allocation. We provide pre World War I evidence on average wage development, strike behavior, and labor allocation within a firm that is consistent with the peacetime model’s predictions. Olson (1982) claims that centralized trade union internalize macroeconomic conditions. We use World War I to test this claim. Firstly, World War I is an exogenous shock in which an output maximizing government replaces profit maximizing employers as the bargaining partner. This change was not foreseen by trade unions when they chose their organizational structure during peacetime. Hence, this “natural experiment” allows us to identify the impact of trade union organization on union bargaining strategies. Since the length of war was unknown, we model wartime bargaining as a repeated game with stochastic end. From the perspective of the wartime bargaining partners the next round of negotiations is reached only with a certain probability. If union leaders can credibly promise support for the government’s effort to organize the war economy, then they are able to push for legal changes in industrial relations such as worker participation in the firms’ decision making process (co-determination), a central goal of the trade union movement. Instead of supporting the government, unions can strike and bargain for wage raises. The wartime model implies that strong national organizations can increase their bargaining power by providing significant support for the government in its effort to organize the war economy. Fractionalized unions embedded in weak national organizations, however, are more inclined 3

to strike to prevent a decline in the standard of living for their members and thereby further decreasing the leverage and credibility of their national leaders. In fact, German trade unions suppressed strikes, and thus faced a decline in real wages. But they successfully demanded that the government enacts legislation forcing firms to allow worker participation in the firms’ decision making process in 1916; a triumph that has affected industrial relations in Germany until today. The road to this triumph proved to be costly. German trade union membership declined by 60% from the beginning of the Great War to mid 1916 as a result of the draft and dissatisfaction of union members with the trade union leadership’s war policies. In the wake of the successful negotiations of worker participation it recovered to prewar levels in 1918. Frequent strikes, disrupting wartime production, kept up British wages with the cost of living. British trade union membership increased by more than 50% during the Great War, but British trade unions did not achieve a change in industrial relations. This confirms Olson (1982). Centralized trade union do in fact internalize macroeconomic conditions, while fractionalized trade union organizations don’t. A possible alternative explanation for wartime bargaining outcomes is representation of trade unions in their respective parliaments and their close ties to the respective Labor parties. We examine this political channel and argue that trade union structure rather than representation in the parliament accounts for the observed outcomes of wartime bargaining. Lastly, we examine the long run ramifications of the outcome of wartime bargaining.

Trade unions in Germany and Great Britain: A brief history Trade union organizations in Germany and Great Britain developed differently in the last 25 years before World War I.5 After the expiration of the Sozialistengesetz (Anti Socialist Law) of 1878, which outlawed the Sozialistischen Arbeiterpartei (Socialist Workers Party) and trade unions, on September 30, 1890 German trade unions could re-organize. For the duration of the Sozialistengesetz workers could organize in Fachvereinen, clubs restricted to admit members of one profession only.6 The conference of the Free trade 5

For a comparison of the early trade union development in both countries, see Eisenberg (1986). 6 See Schr¨ oder (1965) for a survey.

4

TABLE 1 TRADE UNION MEMBERSHIP IN 1913 Germany Total membership 2,973,395 Total Number of Trade Unions 99 Membership in main federal organization 2,525,042 Trade unions in main federal organization 49

Great Britain 3,993,769 1,135 2,232,466 n.a.

Source: Statistisches Jahrbuch f¨ ur das Deutsche Reich 1914, p. 433-434, Labour Gazette August 1914, p. 283, and Lloyd, Trade Unionism, p. 66.

unions, by far the largest trade union federation in Germany, in 1890 mandated the Generalkommission to propose a new organizational form. In 1891 the proposal suggested umbrella organizations (Zentralverb¨ande) with local branches for related trades, basically industry wide trade unions with the Generalkommission as coordinating body.7 The two small competing federations, Christian and Hirsch-Duncker, chose a similar structure. Both federations accounted for 25 trade unions each. Since the Generalkommision as executive committee of the Free trade unions represented roughly 84% of all trade unionists, we focus on it and on the Free trade unions. As coordinating body the Generalkommission, which received parts of the union membership fees, had significant influence by moderating between the Zentralverb¨ande and setting the date and the agenda of the trade union general meetings. It was credited with the increase in trade union membership and became more influential within the trade union organization. In fact, the Generalkommission reserved to itself the right to call for a general strike. Unlike their German counterparts, British trade unions were never dissolved by law. British labor remained organized by profession. At the eve of the war some four million union members were organized in 1,135 separate trade unions which were overlapping within industries, professions, and firms.8 The TUC, main federal organization, represented about 50% of all 7 Gimbel (1989), pp. 69 ff. During the first trade union general meeting in Halberstadt (Halberst¨ adter Kongress) several organizational forms, including local and industry wide unions, were considered. However, both extremes fell short in votes. Overlapping of trade union activities was minimized. Exceptions were the printing, construction and textile industries, where some trades had separate organizations originating in the medieval guild/apprenticeship system. See Umbreit (1928), pp. 31-32. 8 Lloyd (1921), p. 35.

5

TABLE 2 TRADE UNION MEMBERSHIP BY INDUSTRY IN 1913 Germany* Great Britain Industry Number Members Number Members of Unions of Unions Building 8 472,186 65 248,568 Mining and Quarrying 1 104,113 91 915,634 Metal, Engineering and 2 560,644 183 538,275 Shipbuilding Textiles and Clothing 6 268,053 314 518,408 Transportation 1 229,785 61 699,952 Printing 6 135,300 35 84,414 Other 25 803,617 700 988,518 Notes: * Free Unions only. Source: Statistisches Jahrbuch f¨ ur das Deutsche Reich 1914, p. 433, Labour Gazette August 1914, p. 283.

unionized workers, in 1913. The powerful Amalgamated Society of Engineers (A.S.E), however, did not join. Likewise, the miners had their own federation, accounting for about 700,000 workers, leaving more than one million trade union members unfederated.9 Since the TUC as a loose federation was not able to enforce decisions upon individual trade unions, its resolutions were consistently ignored. This structure was criticized as highly inefficient by the members of the Fabian Society.10 The Triple Alliance, consisting of the National Union of Mineworkers, the National Union of Railwaymen and the National Transport Workers’ Federation, founded in 1914, rather than the TUC, was seen as the only institution within the system of trade unions which was able to call for a general strike in Great Britain. However, these three federations were not competing among each other for members. The Alliance, formed to strengthen its members against other trade unions in 9

Lloyd (1921), p. 70. Beatrice and Sidney Webb described the TUC as “an unorganised public meeting, unable to formulate any consistent or practical policy” (cited in Lloyd (1921), p. 201). George Lloyd saw a ‘disastrous’ amount of independence and confusion. He attributes some weaknesses of the federal organizations to selfishness and jealousies of trade union leaders (Lloyd (1921), p. 69). 10

6

their industries, ended due to a lack of solidarity after only seven years.11 On the eve of the Great War German trade unions were centralized with a powerful coordinating body, the Generalkommission, and minimal overlap. The Generalkommission represented the vast majority of German trade unionists. In contrast, British trade unions remained fragmented. They competed among each other and were reluctant to join or cede power to the TUC.

Bargaining in Peace and War Peacetime Bargaining A model of peacetime bargaining should accommodate the two observed modes of trade union organization and take a stand on costs and benefits of each mode. Bargaining over wages and exclusive assignments of tasks to certain trades, so-called demarcations, introduces a collective action problem on the trade unions side. While securing work for its members, fragmented trade unions impose a negative externality each other by restricting the work of each trade. The Report on collective agreements between employers and workpeople in the United Kingdom (1910) states that there existed “...schedules of the work which it is recognised belong to a certain branch of trade.” 12 Centralize trade unions, internalizing this externality, focus on wage development. Centralized trade unions also increase their bargaining power as strikes of all employees are more effective. For the same reason, strikes should be on average shorter than strikes of fractionalized trade unions. Similarly, centralized employer organizations reduce the bargaining power of trade unions by being able to break strikes at the cost of more flexible bargaining outcomes. The following peacetime model, based on historical evidence, integrates all aspects mentioned above. It allows us to study the implication of different modes of employer and trade union organization in more general setting. To address the implications of differences in the structure of trade union organization, we model the peacetime bargaining process between trade unions and firms within a coordination game. 11

Perkin (1989), p. 175. Report on collective agreements between employers and workpeople in the United Kingdom (1910), p. XXIX. 12

7

GRAPHIC 1 COORDINATION GAME

Employers

Trade unions

weak

weak x,z

strong 0,y

strong

w,0

me ,mu

with mu ,me > 0, z > y and x > w.

To justify our payoffs of the coordination game, we assume that on the trade union side the wage increase due to higher bargaining power does not offset the loss in job security provided by demarcations. Similarly, firms prefer the ability to cut firm specific deals rather than lower wages due to coordinated strike breaking effort. Hence, two equilibria, one in which both employers and trade unions have weak national organizations and one in which both support strong national organizations, exist.13 In fact, these are the two modes of organization observed in industrialized countries. We adopt the static asymmetric information bargaining set up from Hayes (1984).14 We modify that approach by allowing for bargaining about so-called demarcations, i.e. exclusive assignments of tasks to certain trades, providing job security. Trade unions impose externalities on each other, by imposing demarcations. Therefore trade unions, representing one trade only, face a collective action problem with respect to imposing demarcations. To illustrate how demarcations affected the firm’s labor allocation, we specify a nested CES production function with heterogenous labor. Suppose a firm can hire labor from different skill groups. Each skill group i has a comparative advantage in executing a respective task i. An end product Y requires N different tasks. Task i can be carried out by specialized labor i with productivity ai or others j with productivity bij and ai > bij , ∀i, j. Task i (Qi ) is executed 13

Without specifying the functional form of the utility function, we cannot rank the equilibria. Note that from a social planner’s perspective z < y or x < w or both are possible and hence the (weak, weak)-equilibrium may not exist. 14 Tracy (1986) and McConnell (1989) provide empirical evidence for asymmetric information in wages bargaining.

8

by 

Qi =

ai L

σ σ−1

i

+

X

σ σ−1

bij (1 − Ri )Lj

 σ−1 σ

,



j6=i

where Ri ∈ [0, 1] measures the extent to which a task is exclusive or not. The elasticity of substitution between the skill groups is σ. The end product is the aggregate of tasks with θ as elasticity of substitution. Each task has a weight ηi . Y (L1 , ..., LN ) = A

"N X

θ θ−1

ηi Qi

# θ−1 θ

.

i=1

If tasks are sufficiently complementary and different types of labor are imperfect substitutes, demarcations increase the labor demand to produce the same output. In fact, depending on ai and bij , Ri shifts demand to group ∂Li i. In the following analysis we will consider the case ∂R > 0. Note that i 2 ∂ Y ≤ 0. The trade union of worker type i demands a wage wi and de∂Li ∂Rj marcations Ri , i.e. only workers of groups i are allowed to execute task i. Trade unions maximize membership. The objective function of union i, Ui , depends positively on the wage wi , employment Li , and worker participation, the right of workers to participate in the firm’s decision making process (co-determination), valued with Ω. As argued above Li is increasing in own demarcations Ri and decreasing in other trade unions’ demarcations R−i . N trade unions share the influence within a firm (for an industry wide union, N = 1). While worker participation could be industry specific, it was never observed to be skill group specific. Hence, we assume that worker participation is a public good, i.e. either all or no trade unions are included in the decision making within a firm. While co-determination was a long run goal of the trade union movement, it was not vital to trade unions. Hence, we Ω . assume that Ui (wi , Li , Ω) = Ui (wi , Li ) + N Since demarcations distort the allocation of labor within the firm, the firm’s profit Π(wi , w−i , Ri , R−i ) − c(Ω) is decreasing in all variables. Moreover, we assume that worker participation incurs possibly high fixed cost c(Ω) in case of implementation. H¨ ubler and Jirjahn (2003) argue that worker participation on a firm level leads to profit reducing rent seeking.15 Addison et al. 15

In their model, worker participation can be beneficial if an establishment is covered by collective bargaining. The latter was introduced in Germany after the Great War.

9

(2000) find that worker participation reduced firm performance in unionized firms in Great Britain and small firms in Germany in the 1990s.16 There are two states k = h, l of the profit levels due to changes in demand, where state h occurs with probability ph . The firm knows the current state, the trade union does not. The length of the bargaining is T . The trade union offers a set of wages wik , demarcations Ri , and strike length sk , and the firm choose the optimal contract given its profit levels. Note that the cost of Ω further reduces profits, making an implementation of workers participation less likely. Since workers participation enters additively in the objective function of the trade union, we solve the trade unions problem for a separating equilibrium neglecting Ω. The trade unions problem, given other trade unions’ demarcations R−i , is: max

wih ,wil ,Ril ,Rih ,sli ,sh i,

ph (T − sh )Ui (wih , Lhi ) + (1 − ph )(T − sl )Ui (wil , Lli )

(1)

subject to incentive compatibility constraints for firms’ to truthfully report the state of profits: h

l

(T − sh )Πh (wih , wh−i , Rih , R−i ) ≥ (T − sl )Πh (wil , wl−i , Ril , R−i ) l

h

(T − sl )Πl (wil , wl−i , Ril , R−i ) ≥ (T − sh )Πl (wih , wh−i , Rih , R−i ),

(2) (3)

participation constraints: h

(T − sh )Πh (wih , wh−i , Rih , R−i ) + sh Πh−i ≥ Π (T −

l sl )Πl (wil , wl−i , Ril , R−i )

+ sl Πl−i ≥ Π,

(4) (5)

and feasibility constraints: (T − sh ) ≥ 0 (T − sl ) ≥ 0.

(6) (7)

Π−i can be interpreted as the profit level during strikes and Π as the firm’s outside option. By the standard arguments, the participation constraint of the low type, equation (5), and the incentive constraint of the high type, 16

Addison et al. (2000) note that in Germany this result does not hold for large firms. However, employers were very resistant to the introduction of worker participation either expecting high cost or unwilling to share power.

10

equation (2), always bind. Combining the first order conditions with respect to wih and Rih , we obtain ∂Ui ∂wih h ∂Ui ∂Li h ∂Li ∂Rih

=

∂Πh ∂wih ∂Πh ∂Rih

.

(8)

In the equilibrium, each trade union chooses demarcations Ri optimally given ∗ all other trade unions demarcations R−i . Hence, Ri (R−i ) = Ri∗ , ∀i. To be clear, demarcations in Britain reserved the right to execute a certain task for a particular skill group. For instance, it was recognized that the shipwrights were to “case all pipes leading through tunnel”, but the joiners were to “case in all pipes in cabins, deck houses, and any other room in connections with joiners’ work.”17 Therefore, demarcations reduce the ability of employers to allocate workers to certain tasks and, as Lorenz (1991) shows, severely affect efficiency. In the case of an industry-wide union, i.e. a trade union representing workers of all skills, the payoff of the union is the weighted sum of the individual skill groups with weights φi . For simplicity, assume two symmetric skill groups i = 1, 2 only. We restrict the strike length to be the same for both groups. The maximization problem becomes max l l h

2 X

∀i{wih ,wi ,Ri ,Ri },sl ,sh i=1

h

i

φi ph (T − sh )U (wih , Lhi ) + (1 − ph )(T − sl )U (wil , Lhi ) (9)

subject to constraints similar to equations (2) - (7). The optimality condition for the industry-wide unions is ∂Ui φi ∂w h i

h ∂Ui ∂Li φi ∂L h ∂Rh i i

∂U−i + φ−i ∂L h

−i

∂Lh −i ∂Rih

=

∂Πh ∂wih ∂Πh ∂Rih

.

(10)

−i Since ∂L < 0, an industry-wide union internalizes the externality of one ∂Rih group’s demarcation on the other groups and will demand less demarcation per group (if any). In other words, an industry-wide union does not care about the skill identity of its members but overall membership. Fractionalized trade unions, maximizing over the employment of their skill group only, have no incentive to coordinate.

17

Report on collective agreements between employers and workpeople in the United Kingdom (1910), pp. 111-112.

11

The length of strikes depends implicitly on the profit level during strikes. Since the punishment of industry-wide unions in more severe, i.e. Π−i = 0, strikes of industry-wide unions are shorter. Trade unions have an incentive to coordinate if employers’ organizations are able to break small and medium scale strikes, i.e. increasing the value of Π−i . However, firms are facing cost of organizing, e.g. they lose the ability to cut deals, satisfying their firm specific needs. Introducing a collective action problem on the trade unions’ side implies the (weak,weak)-equilibrium as “natural” state of the coordination. In this state we should observe a high density of demarcations, a slow wage growth, and long strikes. A second implication of the peacetime model is that, as one side overcomes the collective action problem, the other side unifies, too. In this case, the peacetime model predicts a low density of demarcations, high wage growth, and short strikes. These testable implications allow us to evaluate the validity of our assumptions. In particular, the model sheds light on the mechanisms and the potential impact of the collective action problem on the trade union side. Wartime Bargaining During wartime an output-maximizing government replaces profit maximizing industrialists. However, the length of the relationship is uncertain as it depends on the length of war. The bargaining between the government and the trade unions during wartime is therefore modeled as repeated game with stochastic end. This allows us to derive conditions under which a trade union organization would abandon the strike weapon and support the government’s effort to organize the war economy in hope of later rewards. Governments, in need of maximum production to win the war, will not accept any restrictions on the allocation of labor. Hence, Ri = 0, ∀i. The Ω Ω = Ui (wi , Li ) + N . trade union’s utility can be now written as Ui (w) + N After the first war enthusiasm died away and it became clear that the war would last longer than expected, trade unions could either 1) support the government and hope for a later reward, 2) strike for wage increases, or 3) oppose the war. For simplicity we restrict union strategies to these observed outcomes. Again, trade union structure plays a crucial role in the decision making process. While individual trade unions assume initiative in decision making in an economy with a weak national leadership, a strong national leadership, due to control over organization and strike funds, can set the trade unions’ agenda during wartime. 12

The government can either 1) actively ask trade unions for extended support in organizing the war economy, 2) passively react to demands of (striking) trade unions, or 3) dissolve trade unions. A government will seek only for extended support, if it is confident that the trade unions are a) loyal and b) able to carry out the particular policy. The government uses earlier support of the trade unions as an indicator about trade union loyalty and ability.

Government

GRAPHIC 2 WARTIME GAME

ask

support P Ω δti − N ,

k:τk =1

react dissolve

Ω N

Trade union strike P ∗ − ζ −w ,U (w∗ ) + k:τk =1 ∗

−w∗ ,U (w ) −γ,−X

δ, −ζ −γ,−X

Ω N

oppose P −Z,Z + k:τk =1

Ω N

−Z,Z −γ,−X

The wartime model consists of 2 periods. The first period (P = 1) represents the initial transition to a war economy. With probability p the war lasts longer and P = 2 is reached. In P = 2 trade unions and the government renegotiate. Wages increases, valued with Ui (w∗ ), cost the government w∗ . The trade unions can support the war effort by abandoning the strike weapon without prior negotiations with the government. This can be either out of hope for future concessions or out of fear of being dissolved. In this case of support the government receives a payoff of δ. The government can also ask for extended help. The value of extended help increases if the government is confident that the trade unions can carry out the demanded policy. If the trade unions supported the government before, the indicator variable ti increases. The government receives a payoff ti δ with ti = 2 if trade unions were supportive in an earlier period and ti = 1 otherwise. Support costs trade unions ζ as real wages of their members and therefore membership declines. The government rewards the whole labor movement for extended support of the war effort. It offers to realize the trade unions’ long run goal: workers participation. Therefore, the reward for trade union support is the same Ω which was, due to high cost on firms’ side, not implemented Ω for each supporting trade union. in the peacetime bargaining. It offers N The indicator τi is set to 1 if trade union i supports the government. We 13

assume that for a government, the trade union’s support outweighs its cost Ω > δ. We consider N -symmetric unions. In case of strong national 2δ > N leadership N = 1 and N large otherwise. A strong leadership fixes the strategy for all trade unions and receives N times the individual payoff. Due to symmetry the government plays the same strategy against all trade unions. Trade unions opposing the war receive and cost the government Z. Z can be seen as the value of public support or as bribe necessary to appease trade unions. Governments can dissolve trade unions at cost γ and the game ends. Trade unions receive −X in this case. Governments, engaged in war have a high time preference. They need production now to win the war. Workers, facing steep price increases and a fast deterioration of the standard of living, demand that the trade union leaders deliver instant wages raises, so the time preference of trade unions increases in wartime, too. For simplicity, governments and trade unions discount future payoffs at rate β. We can solve for a pure strategy equilibrium in which a strong trade union leadership supports the government during all phases of the war by backwards (U (w∗ )+ζ) induction. Having a strong leadership implies N = 1. If Ω ≥ 1+βp βp holds then one equilibrium outcome of this game is [(support, react),(support, ask)]. A second Nash equilibrium is [(strike, react),(strike, react)]. Now we consider the case of weak trade union leadership. A profitable deviation by one trade union from [(support, negotiate),(support, ask)] in the last stage Ω ≤ U (w∗) + ζ. Under this condition [(strike, react),(strike, react)] requires N is the outcome in a weak trade union leadership setting. Supporting the government in the first stage signals the government that its demands will be met in the second stage. Thus, the trade unions and the government can coordinate the outcome of the second stage.18 Note that [(oppose, dissolve), (oppose,dissolve)] constitutes another Nash equilibrium. The peacetime model serves the understanding of the trade-off trade unions face in their choice of organizational structure. Introducing demarcations in the peacetime bargaining results in a collective action problem on the trade unions side. Centralization would yield higher wages for trade unionist but also less job security. The testable implications of the peacetime bargaining are: 1) Fractionalized trade unions and employers remain fractionalized without outside pressure, 2) economies with industry-wide unions have less demarcations and higher wages, and 3) economies with fractionalized trade unions lose more workdays due to strikes. In the next section we 18

Follows from Crawford and Sobel (1982).

14

identify the circumstances leading to centralization in Germany and provide evidence that supports the peacetime model. The wartime bargaining allows us to identify the impact of trade union structure chosen in peacetime. Since the war imposes an exogenous change in the bargaining partner, we can use the wartime model to test whether centralized trade unions are in fact more likely to internalize macroeconomic conditions as Olson (1982) claims. The testable implications of wartime bargaining are that centralized trade unions (N = 1) are more likely to support the government throughout the war. They experience a reduction in membership as long as they do not deliver results to their member. However, centralized trade unions can achieve long run goals of the trade union movement by internalizing macroeconomic conditions. In particular, by abandoning strikes and restrained wage demands they are able to signal their credibility as bargaining partner. Secondly, fractionalized trade unions, focusing on their respective members, strike for higher wages.

Evidence I: Coordination in Germany The peacetime model shows that multiple equilibria in union structure can arise due to collective action problems on the unions’ and on the employers’ side. We now focus on equilibrium selection, in particular why German unions and employers overcame their respective collective action problem. Mallmann (1990) notes that the foundation of German employers organizations, in particular the powerful Gesamtmetall, the steel and electronics employers’ umbrella organization, can be directly linked to the expiration of the Sozialistengesetz in 1890. Between January 1889 and April 1890, while the Sozialistengesetz was still effective, 1,131 strike are reported. As a result of calls for a general strike on May, 1 1890 of the Internationale and organizing attempts during the Sozialistengesetz, employers expected the formation of industry-wide unions. At the same time, arguing that it was in the public interest to have less strikes, the advisers of Kaiser Wilhelm II. suggested compulsory arbitration. Averse to any interference by the state, the employers proposed forming organizations able to prevent and break strikes, increasing Π−i . The first circular letter of Gesamtmetall from April 15, 1890 therefore emphasized that the main goal of the organization would be to prevent and fight off strikes. It claimed that experience proves 15

that only joint effort will be successful in the future. While between 1882 and 1888 only 11 mostly regional employer organizations were established, 43 employer umbrella organizations with strike prevention as principal goal were founded in 1889\90. Third parties mocked the new organizations as Antistreikvereine (Anti Strike Clubs).19 In fact, the beliefs about the organizational form of labor triggered the emergence of industry-wide national employers organizations.20 Employer umbrella organizations are listed in the German national statistics for the first time in 1909. Gesamtmetall was then by far the largest umbrella organization measured by the number of employees of its members.21 German employers found federations in 1905. The members of the two main federations, the Verein deutscher Arbeitgeberverb¨ande and the Hauptstelle deutscher Arbeitgeberverb¨ande, accounted for 22 out of 41 employer umbrella organizations and for roughly 80% of all workers in firms which were members in employer organizations. Both federations had their own strike insurance systems. Some umbrella organizations paid strike support to their members. The newly formed employer organizations drew the attention of trade unionists when Hamburg’s employer organizations unified and locked out several thousand workers on May 1, 1890. Hamburg’s employers had threatened to lock out workers if they were to follow the call to strike of the Internationale. Ignoring this threat, roughly one third of all workers (organized and unorganized) in Hamburg went on strike on May 1, 1890. At that time 30,462 workers were organized in 84 Fachvereinen in Hamburg.22 The employers aimed to force their employees to leave their respective Fachvereinen. Despite nationwide solidarity and donations, trade unions suffered a severe blow. In reaction five officials of the metal workers published an open letter to all trade unionists on August 17, 1890, asking how trade unions can fight off these new employer organizations. Employer organizations were the second of four agenda points at the first German trade union conference in November 1890. The third was the structure of the trade unions organization. Employer organizations were used to push for industry-wide trade unions. The final resolution of the conference explicitly states the need for centralized industry-wide unions because of the new employer organiza19

Mallmann (1990), pp. 17-25. In coordination games, signals about future actions are informative. Crawford and Sobel (1982). 21 Later statistics lack this level of detail. 22 Kutz-Bauer (1988), p. 412 and Albrecht (1982), p. 514-515. 20

16

tions.23 Heilborn (1907) sees the concentration of employers as reason for trade union centralization.24 Paul Umbreit, a trade unionist and a close ally of the chairman of the Generalkommission Carl Legien, dismisses the idea that the Social Democratic Party’s call for unity affected the trade unions’ decision to centralize. He concludes that “the struggle [with the employer organizations] led to the unity of trade unions.” 25 By 1913, 99 Zentralverb¨ande with 2,973,395 members existed. At the same time we observe 111 employer umbrella organizations had 145,207 members who employed 4,631,361 workers.26 Lacking a similar threat, British employers founded 1,435 mostly local associations and did not attempt to centralize.27 German trade unions, facing strong organizations with the ability to break small and medium scale strikes, reorganized in national organizations with strong leadership, following the employers’ strategy. This is consistent with the first prediction of the peacetime model. Eisenberg (1986) in her pre-Sozialistengesetz study identifies socialism, the existence of a nation state, and the stage of industrialization as possible candidates for differences in trade union organizations in Germany and Great Britain.28 She argues that German unions were founded by the political elite in the 1860s. But she also acknowledges that the German union organizers, who regarded the British trade union structure as outdated, saw the British skill-based structure as the only feasible way to sell the idea of trade unions in Germany. In her view, the Sozialistengesetz eventually led to the take off of German unions as response to and as defense against future repression by the state implying, that the repressive policies also account for the centralized union organization.29 However, capitalists and state were not as unified 23

Umbreit (1915), pp. 7-9 and pp. 156 ff. and Ritter and Tenfelde (1975), p. 95. Heilborn (1907), p. 5. 25 Umbreit (1915), p. 7. 26 Developments in neighboring Austria highlight the importance of employer organizations. Employers did not strongly centralize when trade unions were allowed. Despite late industrialization and the development of the Social Democratic Party in a repressive state, trade unions remained organized by trade with a low degree of centralization. In fact, the centralization of employers, which was considered to be weak, followed the weak centralization of trade unions. See Grandner (1992) for details on Austrian trade unions and employer organization. 27 Waddington (1995), p. 113. 28 See Marks (1989) for details on trade unions in politics. 29 Eisenberg (1986), pp. 259-263. 24

17

as this view suggests. The emperor, unable to stop production stoppages, pressured the employers to prevent strikes. Moreover, an extension of the Sozialistengesetz was seen as impossible. It is therefore unlikely that the fear of repression by the state in the future motivated the centralization. Thelen (2004) points to differences in the control of skill formation. While the German employers’ control of skill formation and late industrialization may have reduced the threat potential of German skilled workers at the point of formation of trade unions, unskilled workers were still not easily substitutable for skilled worker and the latter were already organized in Fachvereinen in 1890. Moreover, employers gained control over skill formation with the Handwerkergesetz (Handicrafts Protection Law) of 1897, which is too late to explain the timing of centralization.30 Hence, as argued above, the emergence of employer organizations, which threatened the success of fragmented trade unions, played a critical role in the decision to centralize. According to the peacetime model we should also observe a favorable wage development and a lower density of demarcations in Germany relative to Great Britain. In fact, between 1899 and 1913 the share of productivity increases going to the workers, measured as the real wage increase relative to the increase in output per worker, was 79.5% in Germany and 73.8% in Great Britain.31 We find no evidence of demarcations in Germany. The Report on collective agreements between employers and workpeople in the United Kingdom (1910) lists over 1600 formal agreements. Some deal with demarcations only. The report states that “in a large number of cases the question is between two different classes of workpeople each class claiming that a particular piece of work belong to it.”32 Most agreements on demarcations, however, were informal. Lloyd (1921) notes that, for instance, the Engineering and Shipbuilding Trades Federation, with 28 affiliated trade unions, was mostly concerned with mediation in demarcation questions arising between trade unions.33 The wage and demarcation data are also consistent with the hypothesis that collective bargaining maximizes the sum of shareholders’ and union members’ wealth.34 30

Thelen (2004), p. 47. Real wages from Kl¨ onne (1981), p. 99 and Feinstein (1990), p. 344. Output from Broadberry (1997), p. 43. 32 Report on collective agreements between employers and workpeople in the United Kingdom (1910), p. xxvii 33 Lloyd (1921), p. 64. 34 See Abowd (1989) for a discussion and empirical evidence. 31

18

TABLE 3 STRIKES 1900 - 1913 Year

Strikes

1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913

648 642 442 387 355 358 486 601 399 436 531 903 853 1,497

Great Britain Number of Workdays Workers lost 189,000 3,153,000 180,000 4,142,000 257,000 3,479,000 117,000 2,339,000 87,000 1,484,000 94,000 2,470,000 218,000 3,029,000 147,000 2,162,000 296,000 10,834,000 301,000 2,774,000 515,000 9,895,000 962,000 10,320,000 1,463,000 40,915,000 689,000 11,631,000

Strikes 806 692 805 1,200 1,513 2,010 3,059 2,469 1,795 1,839 2,224 2,707 2,469 2,173

Germany Number of Workers 101,081 40,062 48,922 75,830 104,555 363,917 222,686 176,292 66,307 96,750 142,113 211,89 397,490 177,594

Workdays lost 1,051,159 1,070,278 847,121 1,109,461 1,512,886 5,565,546 3,991,606 2,147,695 1,139,636 1,587,257 2,193,452 4,359,062 3,447,729 4,139,720

Average 610 393,929 7,759,071 1,840 158,964 2,440,186 Source: Dearle, Labour Cost, p. 122 and Hohorst, Kocka, and Ritter, Sozialgeschichtliches Arbeitsbuch, p. 132.

Comparing the two countries with respect to strikes, Table 3 shows that on average more strikes occurred in Germany. However, Table 4 shows that on average fewer workers participated per strike in Germany, and that strikes were, on average, shorter in Germany. A possible explanation for this pattern lies in the sectoral composition of strikes. While in Great Britain 76.6% of the strikes occurred in the concentrated sectors (mining, textiles and transport), strike activity in Germany was largest in sectors in which small and middle sized firms without an overarching employer organization predominated. Construction, accounting for about 15% of the total labor force and 8 Zentralverb¨ande, experienced 25% of all strikes with 25.8% of all striking workers participating. Of all firms affected by strikes, about a third were construction companies. Notably the German construction industry founded the largest non-federated umbrella organization. It had neither strike in19

TABLE 4 STRIKE STATISTICS 1900 - 1913 Country Average Num- Average Dura- Workdays lost per ber of Workers tion (days) year and industrial worker* Germany 86 16.5 0.9 Great Britain 645 19 1.5 Notes: * 1899-1914. Average duration is calculated by dividing total workdays lost by the total number of striking workers reported in Table 3. Source: see Table 3 and Boll, “International Strike Waves”, p. 122.

surance nor assistance payments. Likewise the German textile and clothing industry, with 12 employer umbrella organizations and 6 Zentralverb¨ande in 1913, experienced small scale strikes.35 This is consistent with the model, since it predicts more strikes in sectors with more trade unions. Thus, we see more frequent small scale strikes in the fragmented sectors, explaining the low number of workers per strike in Germany. The cross country comparison and the variation within Germany, therefore, confirm the prediction of our model with respect to differences in strike behavior.

Evidence II: Wartime Behavior The outbreak of the war forced the governments to organize the war economy.36 Trade unions as representatives of labor were regarded as critical in the effort to transform the peacetime economy. Demarcations, distorting the allocation of labor and thereby reducing output, or interruptions of production due to strikes endangered war success. If centralized unions in fact internalize macroeconomic conditions, they should support the government’s war effort for later rewards. Fractionalized trade unions, however, would continue to strike for the well-being of their members only. The German Generalkommission immediately met with delegates of the Zentralverb¨ande on August 2, 1914. Within 24 hours of the declaration of war, the conference presented its fundamental position for wartime: 1) 35 36

Boll (1985) and Statistisches Jahrbuch f¨ ur das Deutsche Reich, multiple years. Ritschl (2005) discusses the impact of WWI on the German economy.

20

TABLE 5 STRIKES DURING WAR TIME Great Britain Germany Year Strikes Number of Workdays Strikes Number of Workdays Workers lost Workers lost 1914 793* 414,000* 8,923,000* 1,115 61,000 1,715,000 Aug.-Dec. 206 34,000 1,188,000 26 2,101 n.a. 1915 707 453,000 3,040,000 137 14,000 42,000 1916 578 281,000 2,581,000 240 129,000 245,000 1917 803 885,000 5,809,000 561 667,000 1,862,000 1918 1,194** 928,000** 4,845,000** 531 391,000 1,452,000 Notes: * from Jan-July 1914, ** from Jan-Oct 1918. Source: Dearle, Labour Cost, p. 122. and Petzina, Abelshauser, and Faust, Sozialgeschichtliches Arbeitsbuch III, p. 114.

strikes were to be abandoned immediately, 2) no strike money would be paid to any worker for the duration of the war, and 3) trade unions would focus on the support of war families and the unemployed. This, as Table 5 shows, resulted in only 26 strikes in Germany from August to December 1914, a sharp reduction compared to the whole year. The number of strikes not aiming at wage increases or reductions of hours worked increased from 26% in 1915 to 63% in 1917. The reasons for strikes include protests in the wake of the trial against the most prominent anti-war politician Karl Liebknecht in the summer of 1916 and food shortages in 1917. Democracy within the trade union organization was abolished and internal opposition to the war censored.37 With the war economy picking up pace, the government, maximizing war production, subsequently took over the position of bargaining partner of trade unions. It negotiated agreements and legislation with the trade unions , restricting the rights of workers, to enhance production. The German department of labor within the Ministry of War relied heavily on the support of trade unions. The Generalkommission, having access to every region through local branches, helped to allocate unemployed workers into 37

In fact, as the war continued more and more members of the lower and middle management of trade unions were drafted, whereas the leadership, thanks to its political and military contacts, was able to avoid the draft or limit its duration. This tilted the power balances even more towards the well connected top level (Bieber (1981), p. 114).

21

TABLE 6 SECTORAL DISTRIBUTION OF WARTIME STRIKES IN GERMANY 1915 1916 1917 Industry

Strikes

Number of Workers 1,506 5,327 2,317

Strikes

Building 29 44 Mining 22 77 Metal and 30 57 Engineering Textiles and 8 413 8 Clothing Transport 7 173 7 other 41 1,909 47 Source: Statistisches Jahrbuch f¨ ur das Deutsche

Number of Workers 1,890 44,166 65,081 618

Strikes

37 209 286

Number of Workers 5,460 162,625 170,938

30

4,641

427 8 12,233 95 Reich, multiple years.

1,682 305,312

agriculture and construction and also partly financed reeducation of workers. Although trade unions were able to conserve some of their achievements, reallocated unemployed workers experienced wage cuts relative to their previous profession. Lastly, the Generalkommission gave advice on food supply politics.38 The promised rewards for the trade unions’ help, however, were rather vague. Nevertheless, the Generalkommission, suppressing opposition within the trade union organization, actively supported the war effort, even publishing war propaganda in their trade union newsletters.39 When Great Britain entered the war on August 4, 1914 over 100 strikes were in progress. Three weeks later the TUC pledged to end all strikes and avoid further disputes as the contribution to the war effort of the labor movement. At the end of August all but 20 strikes had been settled.40 Despite the national trade union leaders’ promises, rising prices from January 1915 onwards led to an increase in strike activities back to levels of the early 1900’s. The British national leadership was, due a lack of control over its 38

Pohl and Werther (1989), pp. 123 ff. See also Feldman (1966), Chap. I. Grandner (1992) notes that the Austrian trade unions’ involvement was strikingly similar. However, lacking the same degree of centralization they were less efficient in suppressing strikes. Fearing a loss of control, the leadership eventually challenged the government openly when it became clear that the war could not be won. 40 Aris (1998), p. 99. 39

22

TABLE 7 SECTORAL DISTRIBUTION OF WARTIME STRIKES IN GREAT BRITAIN 1915 1916 1917 Industry

Strikes

Number of Workers 15,925 298,138 46,710

Strikes

Building 68 76 Mining 85 74 Metal and 191 113 Engineering Textiles and 109 38,632 119 Clothing Transport 86 26,600 60 other 169 26,556 139 Source: The Labour Gazette, multiple years.

Number of Workers 7,679 310,660 75,628

Strikes

53 128 181

Number of Workers 6,897 247,145 426,699

77,032

107

75,929

34,578 26,510

40 179

26,740 50,317

local members, unable to end strikes effectively or to support the efforts to reorganize the economy. While the leadership of the TUC favored the entry into the war, trade union publications and some trade unions published anti war propaganda.41 David Lloyd George, then Chancellor of the Exchequer, led the organization of the British war economy. He realized that the rigid unwritten customs and privileges of skilled workers, were limiting output.42 On March 17, 1915 his negotiations with the TUC and the leaders of the 36 trade unions, involved in war production controlled by the government aimed to enhance production. In exchange for the abandonment of all independent union rights, including the possibility to employ un- and semiskilled workers in jobs reserved for skilled workers (dilution), it offered some safeguards and the restoration of pre-war practices after the war. Stressing the fact that the soldiers at the front were suffering from the delays in production, Lloyd George was able to convince most trade union leaders to allow dilution with the promise of restoration of the pre-war practices. However, nationalism itself did not unify the trade unions. In fact, the miners’ leaders, unwilling 41

Parliamentary Debates, House of Commons, 1915, Vol. LXXII, pp. 167-168 and The Lloyd George Papers, D/20/2/9. 42 Official History of the Ministry of Munitions, Vol. 2, p. 4048.

23

to accept compulsory arbitration, left the conference after the first day. The A.S.E renegotiated and succeeded in demanding more specific protection of their skilled workers in this so-called Treasury Agreement. Trade union behavior in the early stage of the Great War shows cooperation of the national leadership with the government in Germany. Control over the individual members was crucial for the success of the Generalkommission’s support of the government. In Great Britain, the TUC lacked the power to set the agenda for the trade union movement. Despite the TUC’s pledge of no strikes during wartime, individual trade unions chose confrontation to offset price increases. The British government, noting the weakness of the TUC, negotiated with the trade unions involved in war production controlled by the government directly.

Evidence III: Wartime outcomes The Munitions of War Act in Great Britain and the Auxiliary Service Act in Germany were the main wartime legislation organizing labor. Both laws are similar in their aim and their means to enhance wartime production by regulating allocation of workers. But the differences between the two in the rights of workers shed light on the impact of trade union strategies. The negotiations and implementations of these laws exhibit differences due to trade union structure. In particular, while the duties of workers are alike in both countries, the rights and their distribution are quite different. In Germany, the trade unions, having supported the war effort unconditionally so far, built up a reputation as moderate, credible, and loyal to the Reich. General von Gr¨oner, in charge of mobilization, announced: “We can’t do it without the unions.” 43 The trade union leadership demanded worker representation, appeal courts for wage settlements, arbitration councils (all three with equal representation), exceptions from the draft for welfare insurance workers, unions’ and employers’ officials, and wage safeguards for their support of industrial conscription.44 Industrial conscription, the obligation for all men at ages 18 to 60, who did not join the armed forces, to work in the war economy, became law on December 12, 1916. In the law almost 43

cited in Umbreit (1928), p. 146. Feldman (1966), p. 218. For a detailed description of the power struggle over this law Shakespearian in its magnitude, see Feldman (1966), Chap. IV. 44

24

all trade union demands were met. German unions achieved their long run goal: worker participation. In addition, members vital to the trade union organization were exempted from the draft. The Generalkommission then issued a statement in favor of the auxiliary service and offered an explanation why industrial conscription was necessary and supported by trade union leadership.45 The Generalkommission was not able to suppress strikes fully as Table 5 and Table 6 show. But in Germany, strikes were used to protest particular policies rather than demanding wage increases. In the wake of the trial against Karl Liebknecht in the summer of 1916 the first massive strikes broke out. In the two biggest strike waves, in April 1917 and January 1918, were due to food shortages. In both cases trade union leaders helped the government to divide the strikers. Trade union leaders, pointing to a promised suffrage reform, reduction in hours worked without wage cuts and minimum wages after the war, managed to appease moderates. The government dissolved the meetings of the radicals with brutal force.46 However, the long support of the government imposed a cost for the German trade union movement. As a result of the abandonment of the strike weapon to increase wages, the minimum wage was stagnant, while prices for essential food tripled during the war. In the best case the average nominal wage for a day laborer went up by only 150%.47 Dissatisfaction with the strong war support by German trade union leadership and the draft cost the German free unions three fifth of its members until the end of 1916.48 The Generalkommission’s achievements in the negotiations of the Auxiliary Service Act led to an increase of membership to prewar levels at the end of the war. The British war economy was regulated earlier than the German. The first Munitions of War Act, an expansion of the treasury agreement, came into effect on July 2, 1915.49 The Act provoked a series of strikes, peaking in the industrial unrest of 1917, despite the safeguards including wage rates and a renewal of the promise of restoration of pre-war practices.50 The Munitions of War Act was amended in 1916 and 1917 regulating the 45

The statement is printed in Umbreit (1928), pp. 245 ff. Pohl and Werther (1989), pp. 164 ff. 47 In the counties Lippe-Detmold and Mecklenburg-Schwerin. Statistisches Jahrbuch f¨ ur das Deutsche Reich (1919), pp. 104-105. 48 Umbreit (1928), p. 161. 49 Fyfe (1917) states the Act and its amendments. 50 Cole (1923), Chap. V. 46

25

dilution of skilled labor in the private sector as well. The TUC played no substantial role in the negotiation with the government over this law. The weakness of the TUC allowed the individual trade unions to negotiate separately. The A.S.E. and others, for instance, negotiated an exemption from the draft for their members. Shortages of skilled workers already threatened war production, and thus this concession was almost costless to the government. British national trade union leaders, disappointing their members by conceding demarcations, lost more and more influence to the radical so-called ‘shop steward’ or ‘rank and file’ movement. The latter called more frequently for strikes to improve the standard of living. Trade union leaders, failing to centralize control of strike funds during peacetime, were unable to suppress strikes. The number of workdays lost increased above the average of the first decade of the 20th century. This strategy kept the real wage at least stagnant over the period of war. Succeeding in increasing wages and by offering protection from the draft, trade unions increased their membership by more than 50% during the war.51 The British government had to negotiate with individual trade unions rather than with the TUC to achieve its goals in organizing the war economy. It bought out moderates with minimal concessions and dissolved the revolutionary wing as, for instance, in the case of the Clyde Workers’ Committee.52 The Generalkommission had promised to end strikes and to support the government’s war effort, and it had delivered in the first two years of war. The German government, now seeing the trade unions as moderates and useful, actively sought the support of trade unions in implementing industrial conscription. The Generalkommission named its price for support: the inclusion of co-determination in the Auxiliary Service Act. The German government, desperate to increase production and to prevent strikes against the law, agreed. The promises of British trade union leaders, on the other hand, had proven worthless. Strike activity had increased throughout the war. The British government did not actively seek the help of trade unions in drafting the Munitions of War Acts. It rather passively responded to the demands of trade unions by increasing the wage or in the case of crucial trade unions by granting other concessions like exceptions from the draft for A.S.E. members.

51 52

Mitchell (1988), p. 137. Perkin (1989), p. 208.

26

Political Ties Not only trade unions themselves but also their political wings, the respective labour parties, could influence legislation. To separate the effect of trade union structure, we examine the behavior of the German Sozialdemokratische Partei Deutschlands (SPD) and of the British Labour Party before and during the war. In addition, we analyze the links between trade unions and labor parties in both countries. After the expiration of the Sozialistengesetz the trade unions regarded themselves as the recruiting school of the SPD.53 In 1912, the SPD emerged with biggest fraction (34.8% of the votes). Lacking a coalition partner, this was, however, not enough to have a leverage to obtain concessions from the Emperor or to dominate politics in the Reichstag. Only 46 out of 110 MPs of the SPD fraction in the Reichtag were trade union officials including six members of the Generalkommission and 30 of the extended national leadership.54 Trade union membership was, however, not mandatory for members for the SPD.55 When the martial law was declared on July 31, 1914 the SPD faction in the Reichtag met to discuss whether they would support to government’s request to be allowed to issue war bonds, which were essential to the ability to fight the war. In spite of previous concerns and because of fear of repression by the empire, the faction waved the patriotic flag and voted unanimously for the first war credits indicating the support of the politics of the Reich during World War I. To explain their voting behavior to war sceptics, they later argued to have acted to defend the Reich.56 The Social Democrats also supported the Enabling Act on August 4, 1914, which empowered the Bundesrat to enact emergency economic legislation (kriegswirtschaftlich notwendige Verordnungen), including all interventions necessary to run the war economy, for the time of war.57 The strong war support led to a split of the SPD into pro-war (MSPD) and anti-war (USPD) factions in December 1915. The trade union leadership remained in the pro-war camp, another signal of credibility of the national trade union leadership. The representation in parliament simplified 53

Mommsen (1985), pp. 372 ff. Miller (1975), pp. 49-50 and Varian (1956), p. 45. 55 The attempt to introduce it failed in 1905. Umbreit (1915), p. 76. 56 Kruse (1993), pp. 26 ff. 57 The abandonment of convertibility of money into gold was part of the bill. Feldman (1966), p. 29. 54

27

the negotiations between the unions and the government over the details of the Auxiliary Service Act in 1916. The unions successfully insisted on co-determination. However, Feldman (1966) notes that the behavior and the pressure of trade unions rather than the party accounted for the worker participation clauses in the Auxiliary Service Act.58 In Great Britain the Parliamentary Labour Party, formed in 1906, understood itself as the political wing of the trade unions. It did not admit non-union members until 1918.59 After the elections of 1910, the Liberal government had to rely on the 42 Labour Party MPs (out of 670 total), including nine out of 13 members of the TUC’s leading committee, to govern. However, the question who was to deal with the government, the TUC or the Labour Party, was not solved. The TUC’s Parliamentary Committee, which was responsible for lobbying at Westminster, was not dissolved but in fact, demanded to be included in future dealings with the government.60 The Fabian Beatrice Webb noted that with few exceptions the Labour Party MPs were a “lot of ordinary workmen who neither know nor care about anything but the interests of their respective trade unions and a comfortable life for themselves.” 61 This suggests that the collective action problem of British trade unions carried over into their representation in parliament. In fact, trade unions of sectors under special legislation, like the miners or the railway workers, defended individual interests.62 The Labour Party’s war position was neither fully supported by all Labour members nor by all TUC members. This led to two splits, one at the outbreak of the war and one in August 1917.63 After the Liberal split in December 1916, David Lloyd George needed the support of the Labour Party to govern. He offered a seat in the war cabinet, and promised not to introduce industrial conscription and to put coal mines under state control. Conscription had already weakened the Liberal Party. Industrial conscription was infeasible without the trade unions. Since even more unskilled worker would have threatened the status of skilled workers further, British trade unions fiercely opposed industrial conscription. Therefore, preventing industrial conscription was hardly progress for the trade union movement, but rather another 58

Feldman (1966), Chap. IV. Perkin (1989), p. 102. 60 Martin (1980), pp. 116-127 and pp. 151-157. 61 Webb (1984), p. 196. 62 Lovell (1991), p. 43. 63 see McKibbin (1974), pp. 88-110 for details. 59

28

attempt to preserve the status quo. Being recognized as moderates, German trade unions succeeded in demanding worker participation in the negotiation over the Auxiliary Service Act. The British Labour Party ensured the majority of the Liberal Party’s minority government and therefore had a bigger lever to push favorable legislations compared to its German counterpart. The Labour Party succeeded only when joining Lloyd George’s government but neither the trade unions nor the Labour Party was able to crucially influence in the Munition of War Acts. Thus, trade union structure, not political representation, led to the observed wartime bargaining outcome.

Impact of the War Legislation Representatives of German industrialists started talks with the trade union leadership about peace time arrangements in April 1917 because the Generalkommission had built up a reputation as being moderate rather than revolutionary and therefore, was seen as credible bargaining partner.64 With their wartime behavior, the trade union representatives set the stage for the Stinnes-Legien Agreement of November 15, 1918, including general wage negotiations, trade unions access to all companies and a reduction in working hours.65 The Weimar Constitution of 1919 acknowledged the role of the workers as equal to the employers (Artikel 165). The plant representation of workers became law one year later. Laws regarding worker committees in firms, weakly included in the constitution, used the Auxiliary Service Act as a blueprint. The idea of equal representation can be found throughout the law.66 Today, industrial relations in Germany are still shaped by equal representation in arbitration councils and plant representation. In Great Britain, after the expiration of the Munitions of War Acts, trade customs came into effect again. The coal mines were handed back to private control on March 31, 1921. Having learned from the past, trade union leaders were more inclined to delegate power to the TUC when negotiating with the government. When the miners called for a general strike in 1926, the TUC expressed its support and with its newly won power helped to organize 64

In December 1917 the talks were expanded see Feldman (1976). Clearly, the negotiation process was speeded up by the Novemberrevolution. Varian (1956), p. 124. 66 Harrer and Roßmann (1989), pp. 182 ff. 65

29

the general strike. As an irony of history, the greater power of the TUC ultimately led to its decline when the general strike was defeated in 1926. Lacking a moderate reputation and being defeated in the first direct confrontation, British national trade union leadership never fully recovered, and the fractionalized system with its implication for labor allocation persisted.67 The lack of coordination left demarcations in place. Lorenz (1991) studies the decline of the British shipbuilding industry in this context. Pratten (1976) estimates the distortions of labor allocation within firms. He finds a significantly reduced competitiveness of British firms even after World War II. Conlusions Collective action problems on the trade union side proved to be decisive for the history of labor unions. These problems define trade union structure and thereby trade union behavior. During peacetime, collective action problems led to a misallocation of labor and lower wages; in wartime they reduce workers’ grievances but they also prevent the achievement of long run goals. Without external pressure neither trade unions nor employers have no incentive to centralize. In Germany, the institutional change was triggered by pressure from the emperor and employers’ expectation about trade unions’ organization after the upcoming expiration of the Sozialistengesetz. They formed umbrella organizations to prevent and fight strikes which forced German trade unions to unify in order to withstand the employers’ strike breaking attempts. Due to their centralized structure, German trade unions were able to play an active part in the organization of the war economy by suppressing strikes and criticism, supporting the reallocation workers, providing training, and helping distribute food rations. The government accepted the demands of the helpful, moderate trade union leadership in the Auxiliary Service Act as reward for their continuing support. Neither British trade unions nor British employers were subject to comparable pressures and thus kept their fractionalized structure. Lacking the infrastructure and the willingness, British trade union leaders never played the active part in the war economy that their German counterparts did. Only trade unions crucial for war production, such as the A.S.E., were able to negotiate favorable terms for themselves, e.g. exemption from the draft. 67

Dorfman (1983) p. 5 ff.

30

We find that identifying the source or absence of pressure to overcome the collective action problem is crucial to explain the divergent paths of trade union structure. World War I as a natural experiment confirms Olson’s claim that centralized trade unions internalize macroeconomic conditions. The different structure of trade unions, leading to different strategies during the war and their implications, favored the achievement of long run goals in Germany (worker participation) whereas the factionalized British trade unions successfully fought for short run goals (wage increases and restoration of prewar practices). The outcome of the coordination game, not representation in the parliament per se, accounts for the differences in industrial relations and laid the foundation for Germany’s variety of capitalism: the coordinated market economy. Official Publications Parliamentary Debates (1915), House of Commons, Vol. LXXII. Official History of the Ministry of Munitions (1920-1924), 12 Vols. Report on collective agreements between employers and workpeople in the United Kingdom, Board of Trade, submitted by G. Askwith (1910), Cd. 5366. Statistisches Jahrbuch f¨ ur das Deutsche Reich, multiple years. The Board of Trade Labour Gazette, multiple years. Primary Sources The Lloyd George Papers, Archive of the House of Parliament.

References Abowd, J. M. (1989). The effect of wage bargaining on the stock market value of the firm. American Economic Review 79 (4), 774–800. 31

Addison, J., S. Siebert, J. Wagner, and X. Wei (2000). Worker participation and firm performance: Evidence from germany and britain. British Journal of Industrial Relations 38, 7–48. Albrecht, W. (1982). Fachverein - Berufsgewekschaft - Zentralverband: Organizationsprobleme der deutschen Gewerkschaften 1870-1890, Volume 11 of Politik und Gesellschaftsgeschichte. Bonn: Verlag Neue Gesellschaft. Aris, R. (1998). Trade Unions and the Management of Industrial Conflict. London: MacMillan Press LTD. Bieber, H.-J. (1981). Gewerkschaften in Krieg und Revolution: Arbeiterbewegung, Industrie, Staat und Milit¨ar in Deutschland 1914-1920, Volume 1. Hamburg: Hans Christians Verlag. Boll, F. (1985). International strike waves: A critical assessment. In W. Mommsen and H.-G. Husung (Eds.), The Development of Trade Unionism in Great Britain and Germany 1880-1914. London: Allen & Unwin Publishers. Broadberry, S. (1997). The productivity race: British manufacturing in international perspective, 1850-1990. Cambridge and New York: Cambridge University Press. Cole, G. (1923). Trade Unionism and Munitions. Economic and Social History of the World War (British Series), Carnegie Endowment for International Peace. Oxford: Clarendon Press. Crawford, V. and J. Sobel (1982). Strategic information transmission. Econometrica 50 (2), 1431–51. Dearle, N. (1940). The labour cost of the world war to Great Britain, 19141922; a statistical analysis. New Haven: Yale University Press. Dorfman, G. (1983). British Trade Unionism against the Trade Union Congress. Stanford: Hoover Institution Press. Eisenberg, C. (1986). Deutsche und englische Gewerkschaften: Entstehung und Entwicklung bis 1878 im Vergleich, Volume 72 of Kritische Studien zur Geschichtswissenschaft. G¨ottingen: Vandenhoeck and Ruprecht.

32

Feinstein, C. (1990). What really happened to real wages?: Trends in wages, prices, and productivity in the united kingdom, 1880-1913. Economic History Review 43, 329–355. Feldman, G. (1966). Army, Industry and Labor in Germany 1914-1918. Princeton: Princeton University Press. Feldman, G. (1976). German Business between War and Revolution: The origins of the Stinnes-Legien Agreement. Berkeley: University of California, Institute for International Studies. Fyfe, T. (1917). Employers & Workmen under the Muntions of War Acts (2 ed.). William Hodge & Company. Gimbel, H.-D. (1989). Sozialistengesetz und große depression: Die deutsche gewerkschaftsbewegung von der wirtschaftskrise 1873 bis zum k¨olner parteitag der deutschen sozialdemokratie 1893. In F. Deppe, G. F¨ ulbert, and J. Harrer (Eds.), Geschichte der deutschen Gewerkschaftsbewegung (4 ed.). K¨oln: Pahl-Rugenstein Verlag. Grandner, M. (1992). Kooperative Gewerkschaftspolitik in der ¨ Kriegswirtschaft: Die freien Gewerkschaften Osterreichs im ersten Weltkrieg, Volume 82 of Ver¨offentlichungen der Kommission f¨ ur neuere ¨ Geschichte Osterreichs. Wien: B¨ohlau Verlag. Greif, A. (1993). Contract enforceability and economic institutions in early trade: The maghribi traders’ coalition. American Economic Review 83 (3), 525–548. Greif, A. (2006). Institutions and the Path to the Modern Economy: Lesson from Medieval Trade. Cambridge: Cambridge University Press. Hall, P. and D. Soskice (2001). Varieties of capitalism : the institutional foundations of comparative advantage. Oxford and New York: Oxford University Press. Harrer, J. and W. Roßmann (1989). Gewerkschaften in der weimarer republik. In F. Deppe, G. F¨ ulbert, and J. Harrer (Eds.), Geschichte der deutschen Gewerkschaftsbewegung (4 ed.). K¨oln: Pahl-Rugenstein Verlag.

33

Hayes, B. (1984). Unions and strikes with asymmetric information. Journal of Labor Economics 2, 57–83. Heilborn, O. (1907). Die “freien” Gewerkschaften seit 1890. Jena: Gustav Fischer Verlag. Hoel, M. (1990). Local versus central wage bargaining with endogenous investments. Scandinavian Journal of Economics 92, 453–469. Hohorst, G., J. Kocka, and G. Ritter (1975). Sozialgeschichtliches Arbeitsbuch: Materialien zur Statistik des Kaiserreichs 1870-1914. M¨ unchen: C.H. Beck. H¨ ubler, O. and U. Jirjahn (2003). Works councils and collective bargaining in germany: The impact on productivity and wages. Scottish Journal of Political Economy 50, 471–491. Iversen, T. and D. Soskice (2000). The nonneutrality of monetary policy with large price or wage setters. Quarterly Journal of Economics 115 (1), 265–284. Kl¨onne, A. (1981). Die deutsche Arbeiterbewegung: Geschichte, Ziele, Wirkungen (2 ed.). D¨ usseldorf: Diederichs Verlag. Kruse, W. (1993). Krieg und nationale Integration: Eine Neuinterpretation des sozialdemokratischen Burgfriedensschlusses 1914/15. Essen: Klartext Verlag. Kutz-Bauer, H. (1988). Arbeiterschaft, Arbeiterbewegung und b¨ urgerlicher Staat in der Zeit der Grossen Depression: Eine regional- und sozialgeschichtliche Studie zur Geschichte der Arbeiterbewegung im Grossraum Hamburg 1873 bis 1890, Volume 20 of Politik und Gesellschaftsgeschichte. Bonn: J.H.W. Dietz Nachf. Lingens, J. (2004). Union Wage Bargaining and Economic Growth. Berlin and Heidelberg: Springer. Lloyd, G. (1921). Trade Unionism (2 ed.). London: A&C Black LTD. Lorenz, E. (1991). Economic Decline in Britain: The Shipbuilding Industry, 1890-1970. Oxford: Clarendon Press. 34

Lovell, J. (1991). Trade unions and the development of independent labour politics. In B. Pimlott and C. Cook (Eds.), Trade Unions in British Politics: the first 250 years (2 ed.). London and New York: Longman. Mallmann, L. (1990). 100 Jahre Gesamtmetall - Perspektiven aus Tradition, Volume 1. K¨oln: Deutscher Institutsverlag. Mares, I. (2006). Taxation, Wage Bargaining, and Unemployment. New York: Cambridge University Press. Marks, G. (1989). Unions in Politics: Britain, Germany, and the United States in the Nineteenth and Early Twentieth Centuries. Princeton: Princeton University Press. Martin, R. (1980). TUC: The Growth of a Pressure Group 1868-1976. Oxford: Clarendon Press. McConnell, S. (1989). Strikes, wages, and private information. American Economic Review 79 (4), 801–815. McKibbin, R. (1974). The Evolution of the Labour Party 1910-1924. Oxford and New York: Oxford University Press. Miller, S. (1975). Burgfrieden und Klassenkampf, Volume 53 of Beitr¨age zur Geschichte des Parlamentarismus and der politischen Parteien. D¨ usseldorf: Droste Verlag. Mitchell, B. (1988). British historical statistics. Cambridge and New York: Cambridge University Press. Mommsen, H. (1985). The free trade unions and social democracy in imperial germany. In W. Mommsen and H.-G. Husung (Eds.), The Development of Trade Unionism in Great Britain and Germany 1880-1914. London: Allen & Unwin Publishers. Moriguchi, C. (2003). Implicit contracts, the great depression, and institutional change: A comparative analysis of u.s. and japanese employment relations, 1920-1940. Journal of Economic History 63 (3), 625–665. North, D. (1990). Institutions, Institutional Change and Economic Performance. Cambridge: Cambridge University Press. 35

Olson, M. (1982). The rise and decline of nations : economic growth, stagflation, and social rigidities. New Haven: Yale University Press. Perkin, H. (1989). The rise of professional society: England since 1880. New York: Routledge. Petzina, D., W. Abelshauser, and A. Faust (1978). Sozialgeschichtliches Arbeitsbuch III. Materialien zur Statistik des Deutschen Reiches 1914-1945. M¨ unchen: C.H. Beck. Pohl, K. and F. Werther (1989). Die freien gewerkschaften im ersten weltkrieg. In F. Deppe, G. F¨ ulbert, and J. Harrer (Eds.), Geschichte der deutschen Gewerkschaftsbewegung (4 ed.). K¨oln: Pahl-Rugenstein Verlag. Pratten, C. (1976). Labor productivity differentials within international companies. University of Cambridge Department of Applied Economics Occasion Paper 50. Ritschl, A. (2005). The pity of peace: Germany’s economy at war, 1914-1918 and beyond. In S. Broadberry and M. Harrison (Eds.), The Economics of World War I. Cambridge: Cambridge University Press. Ritter, G. and K. Tenfelde (1975). Der durchbruch der freie gewerkschaften deutschlands zur massenbewegung im letzten viertel des 19. jahrhunderts. In H. Vetter (Ed.), Vom Sozialistengesetz zur Mitbestimmung: Zum 100. Geburtstag von Hans B¨ockler. K¨oln: Bund Verlag. Schr¨oder, W. (1965). Trib¨ une Verlag.

Klassenk¨ampfe und Gewerkschaftseinheit.

Berlin:

Thelen, K. (2004). How Institutions Evolve: The Political Economy of Skills in Germany, Britain, the United States and Japan. Cambridge: Cambridge University Press. Tracy, J. (1986). An investigation into the determinants of u.s. strike activity. American Economic Review 76, 423–436. Umbreit, P. (1915). 25 Jahre Deutscher Gewerkschaftsbegewegung 1890-1915. Berlin: Verlag der Generalkommission der Gewerkschaften Deutschlands.

36

Umbreit, P. (1928). Die deutschen gewerkschaften im kriege. In P. Umbreit and C. Lorenz (Eds.), Der Krieg und die Arbeitsverh¨altnisse, Wirtschaftsund Sozialgeschichte des Weltkrieges, Carnegie Endowment for International Peace. Stuttgart: Deutsche Verlagsanstalt. Varian, H. J. (1956). Freie Gewerkschaften, Sozialdemokratie und Staat: Die Politik der Generalkommission unter der F¨ uhrung Carl Legiens (18901920). D¨ usseldorf: Dorste Verlag. Waddington, J. (1995). The Politics of Bargaining: the Merger Process and British Trade Union Structural development 1892-1987. London: Mansell Publishing Limited. Webb, B. (1984). The Diary of Beatrice Webb, Volume 3. Cambridge, Mass: Belknap Press.

37