Chapter One
Introduction

The rapid growth of the Japanese economy in the 1960s and the competitiveness of Japanese firms in international markets in the 1970s and 1980s have attracted considerable attention from researchers in economics, political science, and sociology. These researchers have attempted to explain Japan's "economic success," identifying such factors as the role of the state in economic development, inter-firm relations, and the characteristics of Japanese people. Industrial relations have been considered as another important reason for the competitiveness of the Japanese economy and/or firms. But analyses of organized labor have focused on its contribution to good economic performance and the competitiveness of firms. Thus, previous studies have paid attention mainly to unions in export-oriented industries, and have associated cooperative enterprise unions in these industries with the labor movement in postwar Japan as a whole. Often obscured is the other side of the labor movement, i.e., the militant unions of public-sector enterprises, such as the Japan National Railways (JNR).

In the dissertation, I explore how cooperative enterprise unionism spread in the labor movement in postwar Japan, and show that its spread was not self-evident but a contested process. Cooperative enterprise unionism was not the dominant feature of the labor movement from the beginning of the postwar period, but developed out of conflicts with militant and political unionism. Conflicts between cooperative and militant policy orientations existed both between public and private sector union movements and within the union movement in each sector. To see the conflicts between the two types of unionism in the history of postwar Japanese labor movement (from the early 1950s through the mid 1980s), I compare the unions in the steel industry and the Japan National Railways (JNR), as representatives of the union movements in the private and public sectors, respectively, examining their internal political dynamics. Specifically, I examine the histories of Tekkororen (Japan Federation of Steel Workers' Unions, the industrial-level federation of enterprise unions in the steel industry) and its member unions in the major steel firms, and Kokuro (the Japan National Railway Workers' Union), the largest union in JNR. A comparative historical analysis of these unions will demonstrate how polarization in the policy orientations of the unions in the two industries- between the cooperative unionism in the steel industry and the militant unionism in the national railways- grew in the course of the 1960s and 1970s, then went through the process of "de-polarization" in the mid 1980s when the militant unionism represented by the railway workers' union was considerably weakened as a result of the division and privatization of JNR. The analysis will also show that the cooperative unionism of Tekkororen and the militant unionism of Kokuro developed out of factional conflicts and other forms of internal tension within the unions.

I analyze the alternative policy orientations of the unions in terms of the distinctions between political and economic unionism and between cooperative and militant unionism. The former distinction is concerned with goals of union activities: whether a union aims at improving the economic conditions of union members within an existing political system or aims at reform or overthrow of the system itself. The latter distinction is concerned with the methods through which the union tries to attain its policy goals: whether the union pursues its goal through cooperation and consultation with management or through adversarial collective bargaining, strikes, and other forms of industrial actions. Tekkororen and its member unions, after the mid 1960s, sought to achieve economic goals, such as wage increases and employment security, through cooperation with management at the enterprise level in increasing the competitiveness of firms and did not adopt strike-based strategies as a means of pressing their economic goals (thus, their policy orientation was different from that of unions seeking economic goals through strikes). Kokuro, on the other hand, sought to achieve political goals, such as the repeal of legal restrictions on union activities of public sector workers, utilizing strikes and other forms of mobilization of union members (thus, its policy orientation was different from that of unions under neo-corporatist arrangements which have sought political goals through consultation with management and the state). Although the railway workers' union was also interested in economic issues, economic demands such as increased wages were made as part of broader political demands. In sum, the history of the postwar labor movement in Japan can be characterized by conflicts between economic and cooperative unionism (as represented by Tekkororen) and political and militant unionism (as represented by Kokuro), and by the spread of the former at the expense of the latter.

To explain the increasingly contrasting behaviors of the unions, I will focus on their internal politics. By the internal politics of unions, I mean conflicts and compromises among factions and between officials and members over union policies. The concept of "the internal politics of unions" is different from the concept of "union democracy" commonly used in sociological studies of trade unions. While the latter concept is concerned with the relations between union leaders and members and with the characteristics of "a trade union government" (whether oligarchical or democratic), the former pays attention to the internal political processes of unions including not only relations between leaders and members, but also factional conflicts and conflicts between top leaders and lower-rank officials. Thus, even if a union is characterized as being oligarchical, the union may still have some sort of internal political process, such as contested union elections or debate over policies at union congress. In addition, the concept of union democracy treats a union as a political system relatively independent from its external contexts, while the concept of internal politics of unions is attentive to relations between the internal politics of unions and the external contexts of unions, inter alia, changing relations between unions and management.¹

In applying the concept of the internal politics to the unions of steel and railway workers, I argue that the cooperative economic unionism of the former and the militant political unionism of the latter derived from the different ways that the internal politics of these unions developed: while the internal politics of the steel unions became "under-politicized," in the railway union it became "over-politicized." By "under-politicization," I mean the absence of factional conflict in a union under the oligarchical rule of a dominant faction. Characteristics of under-politicized unions include uncontested union elections, lack of policy debate at union congresses, and a widespread indifference of rank and file members to union activities. By "over-politicization," on the other hand, I mean factional conflicts in the union that are so intense as to undermine the integrity of the union organization. Characteristics of over-politicized unions include union factions established along political and ideological lines, the weak grip of union leadership over the factionalized union organization, and possible splits of a union along factional lines.

Why did the internal politics of these unions develop in two such extreme ways? One important influence on the internal politics of the unions I examine is the interaction among unions, management, and the state. I argue that how the internal politics developed in the unions had much to do with the patterns of interactions among unions, management, and the state. The under-politicization of the steel workers' unions derived from the intervention of management into the internal politics of these unions in such a way that a pro-management faction was able to maintain its oligarchical rule of the unions. The over-politicization of Kokuro, on the other hand, derived from the intervention of the state into union-management relations of JNR, which in turn made union-management relations at JNR an issue of national politics and strengthened the influence of political parties on the internal politics of the union. In other words, labor politics within unions are embedded in a broader context of labor politics at the levels of industrial relations and national politics. We cannot fully explain the policy orientations of unions unless we examine the linkage between the internal politics of unions and the broader context of labor politics.

The next section puts the research agenda laid out above into theoretical perspective. It first examines two contrasting theories for defining workers' interests, then explores theories of the relationship between the internal politics of unions and the politics of unions, management, and state interaction, and finally presents a theoretical proposition and two empirical claims based on a critique of previous theories of the internal politics of unions.

The broad theoretical framework of this study is that the policy orientations of a union do not directly reflect the position of the union in structures of product and labor markets, but are mediated by politics over the definition of union members' interests. Such politics is played out not only within unions, but also in the larger arena of industrial relations (i.e., relations among the union, management, and the state). I argue that, while market conditions constituted important contexts for the contrasting policy orientations of Tekkororen and Kokuro, the difference between them was reinforced by politics played in the two arenas.

Previous studies of the militant policy orientation of Kokuro or of such a policy orientation in comparison with the moderate policy orientation of private sector unions suggest that union behavior is closely related to the extent to which its industry is exposed to market pressures. Thus, Kokuro's militant policy orientation has been explained by the fact that the nationalized railway was not exposed to market forces, while the moderate policy orientation of private sector unions was due to the fact that private sector industries, especially export-oriented ones, were exposed to strong market pressures. In this explanation, one of important reasons for the militant policy orientation of Kokuro was that the union did not have to be overly concerned about the negative impact of union-management conflicts on the competitiveness of JNR, because JNR, as a public corporation, was protected from market forces (e.g., Hanami, 1989). On the other hand, private sector unions, such as those in the major steel firms, could not afford to take militant strategies because union-management conflicts would negatively affect the market competitiveness of firms (e.g., Kume, 1993). Thus, the latter unions responded to market pressures in a rational way and adopted moderate strategies and developed cooperative relations with management. Underlying this explanation is a theoretical assumption that the interests of workers are largely shaped by their position in product or labor market structures, and that union behaviors reflect such economically-defined interests of workers.²

The theoretical assumption that interests are defined by economic structures has been criticized by scholars who argue that "politics" mediates between economic structures and definition of interests. The critique of the economically-based definition of interests is made forcefully by S. Berger. She points out that there is a wide-spread assumption among scholars taking the structural-functionalist view that interests are "the unmediated reflection of fundamental properties of socioeconomic structures," and that these scholars do not pay much attention to problems of representation, i.e., how interests are defined, and how relations within a group could be structured so that there would be less divergence between interests between the leaders of a group and its members (Berger, 1981: 6). Rather than seeing interests as the unmediated reflection of socioeconomic structures, Berger argues that interests are defined not only by socioeconomic structures, but also by political factors, such as politics within interest groups and state policies that structure politics among interest groups (ibid.: 10). Thus, the critique made by Berger suggests that unions should be seen not as unitary actors based on the homogenous interests of members defined by economic structures, but as potentially divided groups whose interests are defined through internal politics and the politics among unions, management, and the state.

Some recent studies of unions in Western countries have paid more attention to the internal political processes within unions, and have pointed out that the definition of union members' interests is a contested process, taking the form of factional conflicts and/or tension between union leaders and rank and file members (e.g., Golden, 1988; Stepan-Norris and Zeitlin, 1989, 1995; Yates, 1992). These studies reach different conclusions about the relationship between the internal political process of unions and union behavior. Some argue that the internal structures of unions, especially whether they are democratic or not, affect union behavior, while others argue that union behavior is largely influenced by the worldview held by union officials. Nevertheless, these authors share a common theoretical perspective that posits a plurality of interests within unions, and assume that policy choices made by unions largely reflect their internal political dynamics. One study strongly informed by this theoretical perspective is by Stepan-Norris and Zeitlin on the impact of intraclass struggles on the policy orientation of the leadership of the new CIO unions (Stepan-Norris and Zeitlin, 1989). They argue that the policy orientation was not directly decided by socioeconomic structures, but was a result of internal struggles in the labor movement between Communists and rival factions over the definition of workers' interests, suggesting a "relative autonomy of politics" in which "political struggles have relatively independent effects on class and intraclass relations" (ibid.: 503).

Although these studies stress the importance of political processes in explaining union behavior, they tend to focus on the internal politics of unions as explanations of union behavior, rather than on the politics among unions, management, and the state. Nor do they offer theories of how the two levels of labor politics (i.e., intra-class politics and inter-class politics) might be related. Previous studies have not paid much attention to the two levels of labor politics, because the unions analyzed were industry-based, whose internal political processes were assumed to be relatively immune from the intervention of management and the state. In the case of Japan, however, a theoretical framework which takes into consideration interactions between unions, management, and the state, in addition to the internal politics of unions, is required for analyzing union behavior, because many unions (including the unions examined in this study³) are enterprise-based and are subject to strong influence by management and/or the state. We need to consider how management and the state are involved in the politics of different union factions over the definition of members' interests. More specifically, we can ask such questions as: how policies taken by management in the steel firms contributed to the depoliticization of the internal politics of the steel workers' unions and to the identification of union members' interests with those of management, and how policies taken by the state and management of JNR contributed to the intensification of internal politics of Kokuro and to the definition of union members' interests that was confrontational to management and, especially, the state. The following section considers the theoretical issues concerning how the internal politics of unions is influenced by unions' interactions with management and the state.

Previous studies on the internal politics of unions in Western countries, particularly those in the U.S. and Britain, have tended to treat a union as an organization whose internal political process was relatively independent from the external contexts. In their study of union democracy, S.M. Lipset, et. al. analyze unions as independent political systems that are usually under an oligarchical rule.⁴ In their analysis, the external context, such as relations between unions and management, becomes relevant to the internal politics of the unions only during "organizational crises," i.e., the periods when relations between unions and management are unstable. Under unstable union-management relations, the oligarchical rule of union elites is disrupted by the intensification of factional struggles over union strategies. However, when stability is restored to union-management relations, the unions come under oligarchical rule again under the dominance of the faction that won the struggle and took over the control of the union leadership (Lipset, et. al., 1956: 242-3). Studies of union democracy and union politics that followed Lipset, et. al. continued to treat a union as a relatively independent political system, focusing analysis on such factors as: formal organizational structures of unions (Edelstein and Warner, 1976), status of opposition factions (Martin, 1968), and a "polyarchical" (rather than oligarchical or democratic) process of decision-making in unions (James, 1981). While these studies treat a union as an increasingly complex political system, pointing out the internal political dynamics which tend to be overlooked by the Lipset, et. al. model of oligarchical rule, they do not analyze the internal politics of unions in the context of changing relations among unions, management, and the state.

While students of the sociology of trade unions continued to treat a union as relatively independent political system, some political scientists turned to the question of how the changing institutional arrangements of industrial relations have influenced the internal politics of unions. They are concerned with the internal politics of unions because stable neo-corporatist arrangements depend on the ability of union leaders to control the decision-making process of unions and to prevent "political exchanges" from being undermined by dissenting union members. Thus, political economists investigate the conditions under which union leaders succeed or fail in gaining the consent of union members for "deals" made between union leaders, management, and the state in tripartite concertation (e.g., Regini, 1984; Wolfe, 1985). C. Sabel theorizes an inherent instability of the internal politics of unions under neo-corporatist and other types of industrial relations arrangements. He identifies two forms of politics within unions. One is the struggle over "the rules governing the relations between groups (i.e., union-management relations)," and the other is the struggle over "the rules controlling relations between leaders and rank and file in any one group (i.e., unions)" (Sabel, 1981: 210-11). When union leaders have established stable relations with management and the state, they try to depoliticize the internal politics of unions by "choking off" internal debate over both kinds of rules (ibid.). The issue is the extent to which union leaders are able to depoliticize internal politics. Sabel argues that internal politics is a "continuous feature" of unions, because union leaders cannot suppress internal debate over the rules governing union-management relations for a long period of time. Union leaders, in pursuit of their self interests, try to establish institutional arrangements that support stable relations between unions and management and stable control over rank and file members. However, there does not exist any "self-regenerating set of institutional relations" that ensures the oligarchical rule of unions. This is because the shifting balance of power between unions and management makes any union strategy vis-a-vis management "indeterminate" in the long run, thus reintroducing political debate in the form of factional struggle among union leaders over "the correct course of action" vis-a-vis management. And such political debate over the rules governing union-management relations necessarily engenders debate over the governing relations between leaders and rank and file (ibid.: 212-3, 225-6).

Some studies informed by class analysis analyze the internal politics of unions in the context of class relations in a capitalist economy. For example, C. Offe and H. Wiesenthal examine how the definition of workers' interests is influenced by class relations between capital and labor. They argue that interests are defined by political processes as well as economic structures (as Berger did), but that the process of defining interests is much more political for labor than for capital. This is because, while the interests of capital are readily defined in terms of rational adaptation to market conditions, the ambivalent position of labor in a capitalist economy prevents labor's interests from being defined based on one principle. As they put it, "[t]he problem is that workers can neither fully submit to the logic of the market..., nor can they escape from the market.... Caught in this trap, workers and workers' organizations are constantly involved in the immensely complicated process of finding out what their interests are....(Offe and Wiesenthal, 1980: 104)." They criticize the assumption of interest group theory that interest groups operate on an equal basis, arguing that unions are more disadvantaged in representing member interests than employer groups. This is because the strength of unions depends on the "willingness to act" of union members, which in turn depends on the ability of unions to construct politically a collective identity for workers as members of the working class to counter the pervasive individualist ideology of the capitalist market economy. In other words, the interests of union members are defined by the two contradictory principles as individual participants of the capitalist market economy and as members of the working class (what Offe and Wiesenthal called "monological" and "dialogical" patterns of collective action). Their analysis suggests, as that of Sabel did, that the internal politics of unions is continuously contested, because unions can never completely reconcile the two contradictory logics of defining workers' interests.

These studies in political economy suggest a general theoretical framework for analyzing the two levels of labor politics. In this framework, the internal politics of unions is inherently unstable, due to the ambivalent position of unions in a capitalist economy. The ambivalent position is due to the unions' dual function as a voluntary (one might say class-based) association based on mobilization of membership, and as an actor in an industrial relations system (and/or a neo-corporatist system). If a union is to become a reliable actor in the system, it must impose bureaucratic control over internal politics. However, this poses a dilemma to union leaders. If they take too accommodative a stance in relation to management and the state and tighten the bureaucratic control of union organization, their legitimacy will be questioned by union members and challenged by opposition factions. On the other hand, if they elect strategies based on mobilization in response to member pressures and adopt more confrontational attitudes toward management and the state, the legitimacy of the union leaders will then be questioned by their bargaining partners. Thus, if the internal politics of unions is relatively depoliticized in one phase, that phase is likely to be followed by one of increasing politicization. And the policy orientations of unions correspondingly experience alternate phases of relative moderation and militancy.

Although the theoretical framework outlined above may provide a general framework for analyzing the union movements in the steel industry and in the national railways, there are some limitations in applying the theoretical framework to these cases. For example, while the theoretical framework suggests that the internal political processes of unions are inherently unstable, the internal politics of the enterprise unions of the major steel firms in Japan, once depoliticized around the mid 1960s, has continued to be largely depoliticized since then under the dominance of a faction advocating cooperative economic unionism, even though the context of industrial relations in the steel industry has changed a great deal in the period. Until the mid 1960s, however, the internal politics of the steel workers' unions had been active between leftist and rightist factions. Why did such a change in the union movement in the steel industry take place? And how was the depoliticized state of the internal politics maintained? The theoretical framework does not provide sufficient answers to these questions. One reason for the difficulty in applying the framework to the case of the steel workers' unions is that the framework assumes unions and management interacting with each other as two independent actors (even if the power balance between them is not equal), leaving no room for the possibility that personnel policies of management may fundamentally transform both the nature of union-management relations and the internal politics of unions. As the subsequent historical analysis shows, the personnel policies adopted in the steel firms played a crucial role in depoliticizing the internal politics of the unions and in maintaining this state over time.⁵

The theoretical framework appears to be more applicable to the case of Kokuro than to the steel workers' unions, since the internal politics of the former has been politicized throughout the period covered by the study. As the subsequent historical analysis shows, the leadership of Kokuro faced a strategic dilemma much more pronounced than what the leaders of the steel workers' unions faced. However, the theoretical framework does not explain why Kokuro was deeply factionalized, even though the factions shared the same radical socialist rhetoric. While the theoretical framework suggests that the internal politics of unions concerns the choices of union strategies and is fought between two camps (one advocating strategies based on bureaucratic control, and the other advocating mobilization of union members), in the case of Kokuro union factions were in fact divided along party lines. Thus, the internal politics of the union reflected not only disagreements over union strategies but also over the policies of different parties (the Japan Socialist Party and the Japan Communist Party) or party factions with which union factions were affiliated. The theoretical framework, however, does not take into consideration the influence of political parties on the internal politics of unions. This is because the neo-corporatist studies around which the framework has been developed assumed a politically unified labor movement interacting with management and the state, not a divisive competition between leftist parties (socialists and communists) within a union movement (see Golden, 1986). As the subsequent historical analysis will show, the state implemented policies hostile toward Kokuro in the 1950s that provoked militant reactions of the union. The influence of the political parties on the union's internal politics increased as the conflict between the union and the state became a part of the overall context of national politics, i.e., the conflict between the socialist and communist parties, on one hand, and a strong ruling conservative party (the Liberal Democratic Party) on the other.⁶

As indicated above, the general theoretical framework developed from the previous studies in the area of political economy is inadequate in explaining the internal politics of the steel workers' unions and Kokuro. Based on this critique, I put forth the theoretical proposition that the internal politics of unions is not necessarily autonomous from the influence of other actors, such as management, political parties, and the state. While the previous studies in this area assume that labor unions interact with management and the state as relatively independent actors, and that the internal political dynamics of unions are not directly influenced by other actors, I argue that the degree to which a union's internal politics is autonomous depends on how industrial relations are institutionalized. In postwar Japan, industrial relations were institutionalized in such a way that unions interacted with management at the enterprise level in the case of the private sector but confronted the state directly in the case of the public sector. Under this institutional context, the unions were not able to develop their own internal political space independent from the influence of outside actors. By the political space of unions, I mean a range within which unions make their policy choices through debate among union members. When unions do not have their own political space, the terms of debate over union policies are to a great degree conditioned by the outside actors with which unions interact.

Based on the theoretical proposition discussed above, I make two empirical claims concerning the internal politics of the unions of steel and railway workers. First, I argue that state policies played an important role in the development of internal politics in the steel workers' unions and Kokuro. This argument is raised as a critique of previous studies of neo-corporatism that are mainly concerned with the extent to which unions and management move beyond their own economic sphere and extend their activities into public decision-making, and thereby treat the state as an actor playing a relatively passive role, mostly as a facilitator for social contracts between unions and management (see Zeitlin, 1985: 16). However, the case of state-labor relations in Japan shows that the state developed a wider range of policies having direct or indirect influence on the development of union-management relations and the internal politics of unions. The influence of state policies on the internal politics of Kokuro is obvious: the repressive policies of the state politicized labor-management relations in JNR and promoted factionalization of the union along party lines. Later, the policy of privatization and break-up of JNR considerably weakened Kokuro. In the case of the steel industry, the state was not directly involved in labor-management relations in the steel firms, but did influence the development of labor-management relations by adopting industrial policies that promoted intense competition among the major steel firms. Under this politically-constructed competitive environment, management of the steel firms introduced, especially from the mid 1950s through the mid 1960s, new production systems and new systems of personnel management, which in turn aided the rise of rightist factions advocating union-management cooperation on production-related issues in the steel workers' unions, while weakening the leftist faction adopting a more independent and confrontational stance toward management.

In addition, state policies influenced the general development of the postwar labor movement by creating institutional conditions for unions in the public sector that were different from those for unions in the private sector (most notably, the prohibition of public sector workers to engage in strikes). These institutional arrangements in turn led to different policy orientations between public and private sector unions, creating a deep division between the union movements in the different sectors.

The second empirical claim is that management policies had far-reaching effects on the development of union-management relations and the internal politics of unions, particularly in the steel industry. Studies in industrial relations (including those in the area of neo-corporatism) up to the end of the 1970s paid little attention to the influence of management policies on unions, largely because they presupposed a relatively stable system of industrial relations under Fordist production arrangements and Keynesian macro economic policies. It was only in the 1980s that industrial relations researchers began to pay more attention to workplace and enterprise level industrial relations and to the impact of management's productivity improvement policies (e.g., Kochan, et. al., 1986). Some researchers pointed out the possibility that the involvement of union members in production-related issues promoted by management would facilitate identification of union members' interests with those of the firms that presently employ them, which would have significant effects on the internal politics of unions (Streeck, 1984, 1988; Turner, 1991). In the case of the steel industry in Japan, the impact of policies taken by management, particularly personnel policies, on unions and their internal politics was felt much earlier. While the policies of managements of many firms in Western countries were mainly what may be called "demand-side labor policies" until the late 1970s, management in the Japanese steel firms (and in many other private sector firms in Japan) adopted what may be called "supply-side labor policies" from the early 1960s. Here, demand-side and supply-side labor policies are those concerned with distributional issues and those with production issues, respectively. The latter policies reformed seniority-based work groups into more flexible ones and introduced a wage and promotion system based on individual workers' contributions (as evaluated by workplace leaders) to improvement of productivity and to the competitiveness of firms. In the absence of strong industry or national-based union organization, these policies promoted the depoliticization of internal politics within the enterprise unions of the major steel firms, because the boundary between management and unions was blurred, and because the influence of management became stronger in the definition of union members' interests.

Why did the militant and politicized labor movement common in the early postwar period decline, and why did cooperative economic unionism become the prevalent feature of the labor movement in large private sector enterprises by the mid 1960s? Why did cooperative economic unionism remain stable, even when the conditions of the Japanese economy changed in the 1970s and 1980s? By "cooperative economic unionism", I mean enterprise unions which pursue economic goals (e.g., wage increases and the improvement of working conditions) through labor-management cooperation in improving productivity and market competitiveness of the enterprise. The institutionalization of cooperative economic unionism in private sector enterprises raises another question: why cooperative economic unionism failed to be institutionalized in public sector enterprises, particularly JNR.

Previous studies suggest that there are two theories explaining the transformation of the Japanese labor movement from the militant to the moderate phase: one explains the transformation by the "moral economy" of rank and file union members, the other by "rational or strategic choice" of unions, particularly that of union leaders. In the first theory, both the rise of militant unionism in the early postwar period and the subsequent moderation of the labor movement can be explained by the moral economy of workers, i.e., a strong demand of workers "to be treated as full members of the enterprise" (Gordon, 1993: 378). This demand of workers had historically developed because of their indignation over discriminatory treatment by employers between blue collar workers and white collar staff in the prewar days.⁷ Workers demanded that they should be recognized by management as full members of an enterprise, and that their ability and efforts should be evaluated by a fair standard (i.e., by a standard applicable both to blue collar workers and to white collar staff) (Nimura, 1994: 60). In the early postwar period, the labor movement demanded and won the abolition of the blue collar/white collar status distinction within the enterprise, but the reluctance of management to radically reform the system, especially with respect to its wage structures, as demanded by unions often provoked militant unionism (Gordon, 1993: 379, Gibbs, 1990: 85-6). However, by the early 1960s, most large enterprises, partly in response to the pressures from unions and partly acting on their own initiative, abolished the distinction in wage structures between blue collar workers and white collar staff and began to apply the system of monthly salary and regular raise to all regular employees including blue collar workers (Nimura, 1994: 68-9). As a result, the potential power of the labor movement was weakened because the moral indignation of blue collar workers, which had functioned as the driving force behind the labor movement, largely disappeared. Moreover, many blue collar workers accepted individualized competition under an equalized promotion system, and started to compete among themselves for promotion to lower managerial positions. Thus, blue collar workers lost their independent identity as union members, because their desire for full membership was met through their participation in competition for promotion in an enterprise, rather than through participation in a union movement (Ibid.: 69, 72; Kurita, 1994: 30-1). In sum, this theory argues that the historically-developed sentiment of Japanese workers predisposed the labor movement to be enterprise-oriented, whether it was radical or moderate, and that enterprise unions became cooperative with management when blue collar workers became willing to contribute to market competitiveness of the enterprise under the equalized promotion system as full members of the enterprise.

The second theory, in contrast, focuses on policy choices made by unions and argues that the spread of cooperative economic unionism in large private sector enterprises from the mid 1950s was a result of strategic or rational choices made by "realistic" union leaders and the support of rank and file members for the choices made by the union leaders. According to Shimada Haruo, many rank and file union members realized, through their experiences of defeats in labor disputes in the 1950s and their growing awareness of exposure of their enterprises to increasingly severe market competition, that militant and politicized union strategies under the leftist leadership only aggravated labor-management relations and did not bring economic benefits to workers. Thus, they became skeptical of the radical slogans and ideologies of leftist union leaders, and began to demand realistic union policies which would address their daily concern. A new generation of union leaders "sensitively captured" the desire of many rank and file members for "realistic" union policies, advocated a new line of union policies based on economic unionism and labor-management cooperation, and replaced leftist union leaders who had lost the support of union members (Shimada, 1983a: 246; Shimada, 1988: 207).

Kume Ikuo makes clear the behavioral assumption of unions implicit in the analysis of Shimada, and argues that the maximization of economic benefits is the principle underlying actions of unions. According to him, private sector unions became moderate not because they became weak, but as a result of choices made by unions in their "political exchanges" with management. In the case of the steel industry in which market competition was intense, the political exchange between unions and management took the form of "politics of productivity," and unions chose to cooperate with management in rationalization efforts in exchange for fair shares of benefits and some influence on management decisions through labor-management consultation (Kume, 1993: 159, 173, 178).

Although the two theories are not concerned with the issue of continued militancy of public sector unions such as Kokuro, they suggest reasons that the union movement in the public sector did not become moderate as in the case of the union movement in private sector enterprises. Based on the "moral economy" theory, the militant orientation of the unions of JNR workers such as Kokuro may be explained by the fact that the status differentiation between white and blue collar workers of JNR continued to exist in the postwar period. Although the status differentiation did not provoke much militancy of JNR workers (Aoki, 1991: 97-99), the status system, particularly the existence of "career ceilings" of blue collar workers, created a feeling of constraint and alienation and a sense of "them and us" among blue collar workers, which in turn formed the basis of the militant union movement (see Inagami, 1979: 5-9; Noguchi, 1990: 124-5). The militant orientation of the unions of JNR workers may also be explained by the "rational/ strategic choice" theory. The theory suggests that Kokuro took militant union policies because JNR was not exposed to strong market pressures. Kume, in his comparative study of union movements in the steel and coal-mining industries, argues that the different intensity of market competition of the two industries largely determined how the union movements developed. While the unions in the steel industry responded quickly to the intense market competition and adopted "rationalization strategy," the unions in the coal-mining industry did not adopt this strategy and sought the protection of the industry from the government (Kume, 1993: 178-9). Kokuro's behavior can be seen as being analogous to the coal miners' union, because both the coal-mining industry and the nationalized railway were not exposed to strong market pressures.

The two theories present contrasting explanations/interpretations of how cooperative economic unionism became prevalent and how the potential power of unions was affected by the spread of this type of unionism. Both theories, however, suggest that the spread of cooperative economic unionism was inevitable because of the historical and economic forces, respectively. The first theory suggests that cooperative economic unionism spread necessarily, because the moral economy of blue collar workers had been formed since the beginning of industrialization of Japan in the late nineteenth century and determined largely the subsequent development of the labor movement. The second theory also suggests the inevitable spread of cooperative economic unionism, because it assumes that policy choices of unions were largely determined by economic conditions of industries in which unions organize workers.

The two theories on the spread of cooperative economic unionism (and the failure of the spread of such unionism to the public sector unions), however, do not capture the whole picture of the development of the labor movement in postwar Japan, particularly its political aspect. The two theories tend to treat a trade union as a unitary actor, and do not pay sufficient attention to the internal political process within the union: i.e., conflict and compromise among factions or between union leaders and union members over union policies. ⁸ They see union policies as reflection of the moral economy of blue collar workers or of the intensified market competition, and assume that interests represented by unions directly reflect social values or socioeconomic structures. As discussed above in the section on the theoretical framework, however, this assumption has been criticized by those who take the theoretical approach that politics mediates between socioeconomic structures and interests represented by unions. If we take the latter approach, we can see union policies as a result of political processes, and explain the different policy orientations of Tekkororen and Kokuro in terms of different ways that the internal politics of the unions developed. We can also suggest "suppressed alternatives" ⁹ of the union movement in the steel industry and in the national railways, i.e., the possibility that policy-orientations of the unions would have become different, if the internal politics of the unions had developed in different ways. As the subsequent historical analysis will show, the union movement in the steel industry had at least some possibility of developing along the line of industrial unionism, and industrial relations in JNR had at least some possibility of becoming similar to those in private sector enterprises.

As shown above, cooperative economic unionism became prevalent by the mid 1960s in large private sector enterprise. Cooperative union-management relations remained stable in industrial relations in the private sector in the 1970s and 1980s, when the Japanese economy moved from the period of rapid growth to the period of low growth and recession. What explains the stability of cooperative union-management relations? A common explanation attributes the stability to so-called "Japanese-style management" by characterizing enterprise-based unions as an integral part of the management system. Aoki Masahiko, one of major proponents of this approach, argues that management of large private sector firms in Japan (what he calls "J-firm") acts as "an arbitrator between the body of stockholders and the body of employees, rather than as an agent of the former" (Aoki, 1988: 182). In this model of J-firm, "quasi-permanent employees" and stockholders are "essentially in symmetrical position," and provides J-firm with "indispensable informational and financial assets." The two bodies engage in "the bargaining game" about distribution of the revenue of J-firm, and the "rational equilibrium outcome" of the game is that J-firm seeks a long-term growth (rather than short-term maximization of stock price), restrains the size of employment by heavy capital investment, provides job security to "quasi-permanent employees" in exchange for their higher "effort expenditure," and prefers work sharing to layoffs even during recession (ibid. 154-181, 305-6) The enterprise-based union constitutes "a substructure of the J-firm," because it engages in the bargaining game by aggregating and representing the interests of "quasi-permanent employees" (ibid.: 93).

Aoki's theory of J-firm, especially relations between management and employees of J-firm, draws heavily on the study of Koike Kazuo on skill-formation of blue collar workers in large Japanese enterprises (Koike, 1988). Koike argues that blue collar workers in large Japanese enterprises acquire a high level of skills through job rotation at the shopfloor level and through a long-term participation in an internal labor market of their enterprise (or what he calls "white-collarization" of blue collar workers or "deep internalization of career"). Thus, blue collar workers develop a strong desire to participate in the decision-making of management because of the high level of skills acquired through the "deep internalization of career." Enterprise unions, representing union members' strong desire to participate, exercise substantial influence over the decision-making of production-related issues at the enterprise level. Although the unions do not have a strong presence in workplaces, "semi-autonomous" groups of workers led by foremen make decisions on issues internal to the groups, such as job rotation and allocation of workers, without much control of management. He points out that enterprise unions and their members exercise such influence over management, because management depends on the high level of skills of employees for a good economic performance of firms, and because the unions or their members can withhold the skills of union members as sanctions against management (Koike, 1988: 129-133, 123-4, 257-263; see also Aoki, 1988: 49-51, 93-4). The theories of J-firm and the skill formation of Aoki and Koike, respectively, suggest that the cooperative economic unionism was a stable institution even during downturns of the Japanese economy in the 1970s and 1980s, because the enterprise-based unions became an integral part of the management system, and contributed to the competitiveness of Japanese firms by representing committed and high skilled employees.

As in the case of the theories explaining the spread of cooperative economic unionism, the theory of "Japanese-style management" tends to treat enterprise unions as unitary actors, and to assume relative homogeneity of union members' interests. Aoki argues that "the smooth mediation and representation of diverse interests among employees by a single enterprise-based union is crucial for the efficient working of the information-production system of the J-firm" (Aoki, 1988: 93). While he acknowledges the existence of "diverse interests" among members of the enterprise union, he assumes that the leadership of the union gains the support of union members through consensus building, and that the consensus-building among union members is achieved relatively smoothly, except under extreme circumstances such as large-scale discharges (ibid., 92-3). The smooth consensus-building in the enterprise union in turn depends on the incentive scheme of J-firm by which management evaluates and rewards individual employees in terms of the long-term development of their "contextual skills"¹⁰ and the application of the skills in cooperative ways (ibid.: 50). This is because the incentive scheme promotes homogenization of employees' interests in J-firm (therefore members of the enterprise union). The interests of employees of J-firm are homogenized in two ways: first, under the incentive scheme, employees of J-firm compete among themselves for acquisition of the contextual skills and for better evaluation and reward by management; and second, the incentive scheme has a function of "weeding out" those employees who do not meet standards in acquisition of the contextual skills by discouraging them to stay in J-firm (ibid.: 51, 72-73). In the model of J-firm, the enterprise-based union can build up the consensus of union members relatively smoothly, because it aggregates and represents the interests of only those employees who are willing to participate in the competition for acquisition of the contextual skills. Thus, the theory of J-firm assumes away the possibility that the enterprise-based union is politicized.¹¹

Two criticisms can be raised against the theory of J-firm. The first criticism concerns the issue of representation of union members' interests. The theory assumes that the internal politics of enterprise-based unions is usually depoliticized, because the unions organizes only those employees who respond positively to the incentive scheme of management and have relatively homogeneous interests. However, it may be argued that this view of enterprise-based unions is too restrictive, because members of the union include not only the "core" employees, but also those who are less enthusiastic in responding to the incentive scheme of management and even those who are critical of the incentive scheme. Once the diversity among union members in their attitudes toward management, especially toward the competition for acquisition of the contextual skills, is recognized, we can see at least theoretical possibility that different groups of employees engage in politics over policies of enterprise unions. Given the theoretical possibility of politicization, if the internal politics of enterprise unions is depoliticized as in the case of the steel workers' unions after the mid 1960s, we need to ask why and how the depoliticized state of the enterprise unions was maintained.

The second criticism concerns the assumption of the theory of J-firm that enterprise-based unions are independent from the control of management, because of the supposedly strong bargaining power of the unions. The theory of J-firm assumes that the bargaining power of enterprise-based unions is based on the high level of skills (or the contextual skills in Aoki's terminology) acquired by blue collar workers through job rotation, and that the unions or their members have a strong voice in the management system, because they can withhold the high level of skills, on which management depends for a good economic performance of firms (Aoki, 1988: 93-4; Koike, 1988: 262-3). However, two questions can be raised about the presumed bargaining power of enterprise unions. The first question is whether blue collar workers of large private sector firms really acquire the high level of skills. Some researchers question Koike's theory on the skill-formation of blue collar workers. Nomura Masami, for example, challenges Koike's theory by arguing that blue collar workers are multi-skilled only in the sense that they can do different low skill jobs, and were not as highly skilled as claimed by Koike (Nomura, 1992). Thus, the argument that enterprise unions have strong influence on the decision-making of management based on the skills of their members is being questioned.¹² Second, even if we assume that they acquire the high level of skills, we need to ask whether blue collar workers can threaten to withhold their skills in order to exercise their bargaining power vis-a-vis management. Aoki argues that a collective bargaining between enterprise union and management sets "the general framework for individual competitions for the intrafirm careers," and that the terms of employment of individual workers are specified sequentially within the general framework based on management's evaluation of their contextual skills. He further argues that individual workers have "potential bargaining power over the evaluation" based on their potential threat of the withdrawal of their contextual skills (Aoki, 1988: 93-4). We may wonder if the "threat of noncooperation" is a realistic option for individual workers, because making such a threat is exactly what workers who have acquired the contextual skills are not supposed to do (note that the contextual skills include not only technological elements, but also attitudinal and social skills, such as "motivate to learn, cooperativeness, and diligence"), and because workers who make such a threat will be evaluated negatively in the incentive scheme.

The review of the previous studies on the spread and stability of cooperative economic unionism suggests that these studies tend to treat unions as a unitary actor and to see interests represented by unions as reflecting competitive economic contexts (as suggested by the "rational/ strategic choice" theory), internal labor markets (as suggested by the theory of "J-firm") or the social values of blue collar workers (as suggested by the "moral economy" theory). Thus, these studies do not pay much attention to politics over the definition of interests represented by unions, which is played within unions as well as among unions, management, and the state. On the basis of the theoretical framework discussed above that the definition of union members' interests do not directly reflect socioeconomic structures but is mediated by politics, I argue that the cooperative economic unionism spread and remained a stable feature of the labor movement through politics within unions as well as among unions, management, and the state.

It is rather difficult to conduct empirical research on the internal political processes of unions, not only because internal politics often takes place in informal ways and is not recorded in official documents of unions, but because investigation of internal politics touches politically sensitive issues. In conducting research on the internal politics of Tekkororen (and its member unions at the major steel firms) and Kokuro, I try to describe the development of internal politics as objectively as possible by using documents written from different perspectives.

In doing research on the two unions, I relied on document sources much more than on interviews. I mainly relied on numerous periodicals in the field of labor studies that included articles on the internal politics of unions and interviews with union leaders and union officials. Previous studies of Japanese industrial relations have under-utilized the rich information available in these periodicals, because these journals were seen as being ideologically biased (Hazama, 1975: 111). I believe that the problem of ideological bias can be overcome, at least partially, by careful reading of materials. Besides these periodicals, I used union and management documents, results of surveys conducted by the unions, newsletters published by a leftist minority group in the steel workers' unions, and reports of union movements annually published by the Ministry of Labor. Document analysis is supplemented by several interviews with former union officials and a policy advisor.

The dissertation analyzes the development of the union movements in the steel industry and in the national railways from the early 1950s through the mid 1980s, and shows the process through which the internal politics of Tekkororen (and its member unions) and Kokuro became under- and over-politicized, respectively. Chapter 2 through Chapter 4 examine the development of the internal politics of Tekkororen and its member unions in the major steel firms and the contexts of the union movement, such as the market conditions of the industry and personnel policies adopted by management. Chapter 2 examines the internal politics of the union movement in the steel industry in the 1950s and 1960s, and shows how the internal politics of the unions, which had been active in the 1950s due to conflict between leftist and rightist factions over union policies, became depoliticized in the 1960s as the rightist factions established the dominance in the union movement. Chapter 3 and Chapter 4 examine the rightist-dominated union movement in the 1970s and in the first half of the 1980s. Chapter 3 analyzes how the rightist leadership of Tekkororen and its member unions maintained their legitimacy in relations to union members, although policies of the union leadership (such as cooperation with rationalization and wage restraint) entailed some sacrifices for union members. And Chapter 4 describes relations between the rightist union leadership and leftist minorities, and shows how the union leadership, in cooperation with management, repressed activities of the leftist minorities in order to prevent the repoliticization of the internal politics of the unions.

Chapter 5 through Chapter 7 analyze the development of the internal politics of Kokuro, the inter-union politics between Kokuro and other unions of JNR workers, and policies of the government and management. Chapter 5 examines the union movement in JNR in the late 1940s and 1950s, and shows how restrictive labor policies of the state toward Kokuro provoked militant reactions of the union and intensified its internal politics. Chapter 6 and Chapter 7 analyzes the internal politics of Kokuro from the early 1960s through the mid 1980s by examining the impact of three organizational crises of the union on its internal politics and the dilemma of the union leadership between mobilization and control in coping with the organizational crises. Chapter 6 examines the internal politics of Kokuro in the 1960s and in the 1970s. It describes the first two organizational crises of the union (the secession of the rightist faction from Kokuro in the early 1960s due to the split of Japan Socialist Party between the leftist and rightist wings in 1960, and the massive defection of Kokuro members to the rival union due to the anti-Kokuro policy taken by management in the early 1970s), and shows how mobilization-based strategies taken by the union to cope with the crises increased the influence of the opposition factions calling for more political-oriented and/or radical strategies, and how the union leadership coped with the unintended rise of the radical factions. And Chapter 7 examines the internal politics of Kokuro in the increasing hostile contexts in the 1980s. It describes the third organizational crisis of the union in conjunction with the restructuring of JNR (the drastic decline of union membership in the mid 1980s) in the context of the hardline policies adopted by management toward the union in preparation for the privatization and division of JNR and the increasing isolation of Kokuro in the labor movement as a whole. Because of the magnitude of the crisis, the internal politics of the union intensified to such an extent that the union was eventually split between the moderate and radical factions.

Chapter 8 concludes the study by summarizing the historical analysis of the union movements in the two industries and by discussing implications of the historical analysis for the theoretical questions laid out in Chapter 1.

¹For example, see different ways that Lipset, et. al. and Sabel analyze unions. While Lipset, et. al. treats labor unions as an independent political system, Sabel treats them as being embedded in a broader context of industrial relations (Lipset, et. al., 1956; Sabel, 1981).

²This theoretical assumption is also used in the theories of the "J-firm" and the "white-collarization" of blue collar workers proposed by M. Aoki and K. Koike, respectively. In their view, the participation of blue collar workers in the in the internal labor markets of large private-sector firms narrows their interests so that their interests lie mainly in the acquisition of firm-specific skills that will advance their position inside the firm. Thus, union behaviors, reflecting the interests of these workers as defined by their position in the structure of labor market, are enterprise-centered and cooperative with management in production-related issues. And since they assume that the interests of members of enterprise unions are defined by the economic structure and therefore relatively homogeneous, they do not address the possible divergence of interests between leaders and members of enterprise unions (Aoki, 1988; Koike, 1988; see also below).

³Tekkororen is a loose federation of enterprise and plant unions in the steel industry that does not have strong authority over its member unions. Although the membership of Kokuro was large, the union was enterprise-based in the sense that it organized only employees of JNR and did not include workers in private and municipal railways in its membership (the latter had their own unions).

⁴The ITU (the International Typographical Union) was the major exception to the "iron law of oligarchy." Lipset, et. al. point out the existence of occupational community and the small gap in status between union officials and rank and file members as factors contributing to the union democracy of the ITU.

⁵One may argue that the interests of these union members were after all defined directly by economic conditions, because the internal politics was greatly influenced by policies taken by management. It may be counter-argued that policies taken by management are not necessarily driven by market imperatives. In the historical analysis of the union movement in the steel industry, the dissertation attempts to show that, while management of the steel firms adopted personnel policies for an economic purpose (to make work organization adaptable to technological changes), the way that such policies were implemented was often political (with policies used to tighten control on individual employees and to eliminate workers opposed to cooperative union-management relations).

⁶G. Marks, in his comparative study of labor movements in the U.S., Britain, and Germany in the nineteenth and early twentieth centuries, also pointed out that state repression strengthened the influence of political parties on labor unions, and that the close ties between unions and political parties made unions politically-oriented (Marks, 1989). However, he stressed that repressive state policies promoted a sense of unity among workers, and did not discuss the possibility that political parties had divisive effects on labor unions.

⁷For the historical account of how blue collar workers in the prewar period developed their sense of moral indignation over the discriminatory treatment between blue collar workers and white collar staff and began to demand their recognized status within the enterprise, see Smith (1988), Chapter 10, and Gordon (1985), Part Two.

⁸The two theories acknowledge factional conflict between leftist and rightist union leaders in unions but do not see such conflict as an important factor in the development of union movements. The "rational/ strategic choice" theory emphasizes the ideological and unrealistic nature of policies advocated by leftist leaders (Shimada, 1988: 207). According to the "moral economy" theory, the value of blue collar workers who identified themselves with their workplaces rather than with unions preordained the decline of leftists which called for a union movement independent from enterprises (Kurita, 1994: 190-1).

⁹The concept of "suppressed alternatives" is used by J. Fulcher in his comparative study of labor histories in Britain and Sweden. He argues that "[a]n explanation of the different outcomes [of the labor movements in the two countries] must...account for the failure of alternatives to materialize as well as for the line of development that actually prevailed" (Fulcher, 1991: 6). And he explains why the Swedish labor movement failed to develop along craft line, and the British labor movement along corporatist line.

¹⁰By "contextual skills," Aoki means a wide range of skills which are required for employees of J-firm. The contextual skills include not only technological elements (an ability to do a wide range of jobs), but also attitudinal and social skills ("motivate to learn, cooperativeness, and diligence") (Aoki, 1988: 50, 54).

¹¹Aoki acknowledges the possibility of a split of the enterprise union and the bitter inter-union rivalry under extreme circumstances, but he assumes that the internal politics of the enterprise union is depoliticized under normal circumstances (Aoki, 1988: 93).

¹²Note that the presumed bargaining power of Japanese workers based on their high level of skills is also questioned in the recent debate over "lean production" (see Womack, et. al., 1990; Jugens, 1992; and Berggren, 1992, Chapter 2).