Unlike other social sciences that have taken an interest in institutions only recently, it has long been understood in industrial relations (IR) that institutions "matter." By this we mean that practices and outcomes in IR are shaped in critical ways by the institutional environment.

     The institutional approach of IR has not been without its problems. In the United States especially, mainstream neoclassical economists were able to marginalize the field because its institutional perspective was supposed to indicate a lack of theoretical ambition (and an excessive attachment to unions). Yet empiricism was no savior, as the field's emphasis on analyzing existing (formal) industrial relations institutions and practices (rather than the tenuous context within which those institutions emerged and operated) left it totally unprepared for their eventual (but in retrospect unsurprising) disintegration.

     The tendency to see stable, static systems of institutions instead of dynamic local circumstances and real contextual decision making has marred virtually all institutional social science until recently. However, recent advances in the historical and sociological study of institutions, including a new emphasis on how institutions develop and change over time, offer the possibility of a richer institutional approach to industrial relations. The basic idea is to extend the focus of institutionalism beyond explaining periods of stability to explaining periods of institutional formation and change as well. The dissertation this paper is based on examines the creation of a paradigmatic institutional form in American industrial relations—the American industrial union. This paper explores an early false start in that process, the case of the explosive growth of unionization in Wisconsin metalworking in 1933Ð1934 (prior to the passage of the Wagner Act and the formation of the Congress of Industrial Organizations [CIO] and its member unions), to examine the respective roles of social structure, institution, and worker agency in the emergence of this institutional form. The paper tries to (re)direct attention to two areas relatively neglected by postwar American IR: the importance of timing, in particular the sequence in which events occur, and the dynamic relationship between social structure, institutional environment, and human agency.

The New Institutionalism

     At the risk of oversimplifying, the new institutionalism in the social sciences arose as an alternative to the unsatisfying accounts of human history offered by neoclassical economics on the one hand and Marxism on the other. With neoclassical economics, the overly simplistic assumptions, the inattention to social units other than "individuals," and the general disinterest in possibly complex social influences that may be difficult to model has yielded a theoretical world in which issues of (actual) time and space have no place, power differentials are masked or ignored, and the net effect of unions on society always turns out to be negative. In contrast to this hyper-individualized account, Marxism offers a view of society in which individuals have virtually no role. Classes are the social actors that matter, and the class to which any particular individual is attached is not a matter of choice but a function of one's social "location." By recognizing social influences larger than individuals but smaller than classes, institutionalism offers a more realistic worldview than either the undersocialized world of neoclassical economics or the oversocialized world of Marxism.

        Moreover, it turns out that the undersocialized world of neoclassical economics leaves no more room for real agency than Marxism does. In the neoclassical model the need for individuals to have stable, rank-ordered preferences that are exogenous (that is, they arise "outside" the model and thus not explainable by it) makes all decision making perfectly predictable and unchanging. As with Marxism, there is no tolerance for human agency as a real person would understand it: actual choice between real existing alternatives that cannot be pre-determined or changing one's mind as a result of learning or personal development.

     Until recently, however, the institutional view on agency has been nearly as weak. Most institutional accounts explain individual and collective choices and outcomes as strongly conditioned by the institutional environment (Thelen 2003). For example, the general weakness of American unions compared to unions in other advanced industrial democracies has been variously attributed to the absence of a labor party or culture of socialism (Lipset and Marks 2000), an adverse legal environment (Hattam 1993), and exceptionally high levels of managerial opposition to unionism (Kochan, Katz, and McKersie 1986) to name just three institutional explanations among many. Despite their different claims, each of the above-mentioned researchers share the view that institutional structures shape individual and collective decisions and that these decisions in turn reinforce institutional structures. Agency and institutional structure become self-reinforcing.

        The problem with this view is that there are periods when events directly contradict it. Though the following account will only touch on select aspects of the story, many of the alleged "truisms" of self-reinforcing structure and agency simply do not hold for the 1933Ð1934 period: socialism in many locations was both strong and respectable, with local and state union officials openly opposing the national American Federation of Labor (AFL) policy of voluntarism, voicing strong opinions in favor of government intervention in the economy and for industrial unions over the craft form; workers flocked to unionize despite extreme managerial opposition, the absence of union structures that matched their interests, and the inability and disinterest of most existing union leaders in organizing them; the institutional structures into which these workers were organized, federal labor unions (FLUs), were both ambiguous and contested, though as it eventually became clear that the AFL would not allow FLUs to evolve directly into industrial unions, enthusiasm for industrial unionization ground to a halt in late 1934. In sum, for this brief period, the claim that American workers had little interest in joining weak, nonradical unions that their employers nonetheless violently opposed is directly contradicted. How did this happen?

Industrial Unionization of Wisconsin Metalworking, 1933Ð1934

     The data on U.S. union density from 1930 through the turn of the century are well known. From a low starting point of about 12 percent of the workforce in 1930, unionization surges from 1933Ð1947, peaks in 1954 at about 34 percent, and declines continuously thereafter.

     While the data since the 1950s support conventional institutional accounts of union weakness and managerial oppositionism, the dramatic upsurge from 1933Ð1947 fits less well. Indeed, it requires an alternative explanation, as does the major institutional development associated with this upsurge, the American industrial union.

     Again oversimplifying, the conventional institutional account of the rise of American industrial unionism is automobile-centric, emphasizing Taylorist factory organization, the Depression, and the Wagner Act. Taylor's factory system deskilled work, making the industrial union the "natural" or even "necessary" form of union organization, regardless of AFL preference for the craft form. The Depression delegitimized business in the eyes of workers and voters, leading to an unusual willingness by workers and politicians to constrain corporate discretion and challenge managerial authority. The Wagner Act provided the legal infrastructure for plant-based industrial unionism.

     This account has much to recommend it. But it is both almost completely ad hoc and at odds with the overall institutional view of U.S. IR as a consistent story of dominant business and weak, fragmented unions. Furthermore, close inspection of the historical record emphasizes a number of complicating factors. First, the upsurge in unionization begins in 1933, two years prior to the passage of the Wagner Act and the formation of the CIO. Second, in many locations, including Wisconsin, industrial workers in fact unionized twice: first in 1933Ð1934 into FLUs under the auspices of the AFL, and again, more successfully, in 1935Ð1937 into industrial union locals (mostly under the auspices of the CIO but also, as with the machinists, under the AFL). Third, at least in Wisconsin metalworking in 1933Ð1934, the workers organized the unions, not the other way around. Unionization was not directed or even underwritten by the national AFL, and in Wisconsin the state federation, central labor councils, and local unions were too financially stressed to be able to hire organizers. Unlike in coal or clothing, there were no charismatic leaders of national stature leading the charge to unionize Wisconsin's diversified metal manufacturers; nor were there existing industrial unions for metalworkers to be organized into. Fourth, the AFL was no voluntarist monolith. In Wisconsin many of the local and statewide AFL unionists who witnessed and participated in the organizing upsurge of 1933Ð1934 were avowed socialists who favored the industrial form of union organization and promoted FLUs as the intermediate vehicle for their quickest attainment. Paying close attention to the issues of timing and the relationship between worker agency and institutional structure in the case of Wisconsin metalworking in 1933Ð1934 uncovers a more complex chapter in the story of the development of American industrial unions.

     Unlike steel in Pennsylvania or autos in Michigan, no single industry dominated Wisconsin metal manufacturing in the 1930s. Instead, the state was home to a wide variety of metal manufacturing firms, including a number of large, diversified, vertically integrated firms located mostly in Milwaukee but also in fifteen or twenty other, much smaller industrial cities. Industrial employment in Wisconsin (outside food processing) bottomed out at 136,000 in January 1933, barely half the pre-Depression peak of 258,000 reached in August 1929. From this low point, industrial employment rose steadily, reaching 180,000 in August 1933, topping out at 200,000 early in 1935, and not falling again below that level until the "Roosevelt recession" of 1938 (all figures from Wisconsin Industrial Commission 1940).

        Roosevelt was inaugurated in March of 1933 (the AFL had again chosen not to endorse a candidate for president), but it was the passage in of the National Industrial Recovery Act (NIRA) in June of that year that provided the immediate spur to industrial unionization (Bernstein 1970: 27). While IR scholars have tended to emphasis the language in Section 7(a) that explicitly legalized union activity for the first time, in a practical sense it was Section 7(b) that motivated workers to form independent unions. Section 7(b) called for the creation of industry codes of fair competition and, more importantly, allowed labor-management agreements on maximum working hours and minimum wages to be incorporated into these codes. The whole point was to utilize every means available to stabilize wages and prices. However, Section 7(b) was ambiguous on what kind of labor-management agreements would be respected. Afraid that employers would try to circumvent Section 7(b)'s intent by entering into sham agreements with company-dominated unions, thousands of Wisconsin metal industry workers almost immediately sought to join independent unions. Bereft of funds and organizers, local and state AFL leaders in Wisconsin did not "organize" these workers (Ozanne 1984: 61). They did encourage the national AFL to grant interested groups of workers charters to form FLUs (WSFL 1934), the only form of unionization offered by the AFL that was flexible enough to accommodate the new horde. To the craft unionists who dominated the AFL leadership in Washington, FLUs were seen as temporary unions, holding pens for disparate groups of workers (albeit generally of a single employer) unionized all at once who ultimately would be parceled out to the "appropriate" craft unions. To the socialists in the Wisconsin State Federation of Labor (WSFL), though, FLUs were simply an available tool, a mechanism by which industrial unionism, seen by them as inevitable, could most quickly be brought about.

     From July 1933 to June 1934, even as the drive to organize autoworkers in Detroit stalled, more than thirty new FLUs were chartered to organize metalworkers in Wisconsin, encompassing most of the largest employers in the industrial centers of Milwaukee, Racine, and Kenosha but also large firms in smaller cities such as La Crosse, Eau Claire, Manitowoc, and Sheboygan (WSFL 1934). And it did not take long for the nascent FLUs to act. The FLU at Nash Motor in Kenosha struck (and was subsequently locked out) in November 1933. The FLU at Nash in Racine struck in February 1934, followed quickly by the FLUs at Nash in Kenosha and at the related Seaman Auto Body in Milwaukee. Perhaps most notably (but least successfully), the FLU at Kohler Co. outside Sheboygan began what ultimately proved to be a losing seven-year strike in July 1934 (Ozanne 1984: 64Ð69).

     However, as the need for clarification on the ultimate status of FLUs became evident and the AFL refused to offer an unconditional endorsement of industrial unionism, interest flagged. The issue of the long-term future of the FLUs was successfully evaded at the AFL convention in San Francisco in October 1934 (Bernstein 1970: 362Ð368). In Wisconsin new FLUs were continued to be chartered in the second half of 1934—at A. O. Smith in Milwaukee and at the Chevrolet and Fisher Body plants in Janesville. However, at each of these locations most skilled workers were already members of various craft unions, and the new FLUs expressly excluded them. Only a handful of new charters for plant-wide FLUs were granted from July 1934 to June 1935, and by June 1935 paying membership in the FLUs had plunged from a year earlier (WSFL 1935). Ultimately, many of the FLUs would seek new identities as locals in the newly formed United Auto Workers (UAW) and have to reorganize all over again.

     To summarize, despite the absence of union organizers, charismatic labor leadership, or even a coherent organizational structure, Wisconsin metalworkers flocked to FLUs in 1933Ð1934. They joined together in FLUs over the opposition of their employers because these were immediately available vehicles for the creation of independent unions. They wanted independent unions because they feared employers would create sham unions to evade the promise of NIRA's Section 7(b). Once it became clear that the AFL had no enthusiasm for allowing FLUs to transition into plant-wide industrial unions (and the promise of joint labor-management sectoral economic governance under the NIRA waned), workers left the FLUs as quickly as they had joined. But the rapid rise and fall of the FLUs apparently did not quell the interest of these workers in unionization. Soon after, the original pioneer industrial unionists and many more workers enlisted in the industrial unions of the CIO (and their AFL counterparts).

Conclusion

     Until recently, institutionalism in the social sciences has yielded accurate but mostly ad hoc explanations of important historical events and developments. Institutional accounts have tended to do a better job of explaining the persistence of institutions over time than explaining their origins. With greater attention to issues of timing (see Pierson and Skocpol 2002) and to the interaction of agency and institutional structure, the new historical institutionalism offers the promise of better explanation of periods of institutional development as well.

     More accurate explanation is of course valuable in itself. In the case of U.S. unionization, a more nuanced understanding of the rise of American industrial unionism provides a useful counter to simple functionalist explanations. In the context of contemporary American IR, however, close attention to issues of timing and the interaction of agency and institutional context takes on added importance. With private sector union densities now lower than they were in the Depression, any effort to rebuild collective worker representation in the United States will require a sharp increase in worker mobilization (agency) in a context of union (institutional) weakness. At this point, whether or not this could or should be done via existing institutions of worker representation or, even more basic, what the appropriate level(s) of worker organization (worksite, firm, locality, sector, region, nation, or some transnational level) and/or degree(s) of articulation between levels might be are open questions.

     The openness of the current environment poses difficulties for all the social sciences, even those not suffering from the identity crisis that has crippled American IR. In retrospect, we now see that the American IR "systems" perspective illuminated only one (temporary) regime of industrial relations, not the entire substantive field. Yet tossing the institutionalist baby out with the systems bathwater makes no sense. Institutionalism still offers more convincing explanations of real world events than either neoclassical economics or deterministic, "structural" social theories such as Marxism. Increased attention to issues of timing, mechanisms of institutional development and change, and the relationship between institutional structure and human agency raises the possibility of an institutional social science that can not only explain the past but also offer insight (though no road map) about possible futures.

References

Bernstein, Irving. 1970. A History of the American Worker 1933Ð1941: Turbulent Years. Boston: Houghton Mifflin Co.

Hattam, Victoria. 1993. Labor Visions and State Power: The Origins of Business Unionism in the United States. Princeton, NJ: Princeton University Press.

Kochan, Thomas, Harry Katz, and Robert McKersie. 1986. The Transformation of American Industrial Relations. New York: Basic Books.

Lipset, Seymour M., and Gary Marks. 2000. It Didn't Happen Here: Why Socialism Failed in the United States. New York : W.W. Norton.

Ozanne, Robert W. 1984. The Labor Movement in Wisconsin: A History. Madison: The State Historical Society of Wisconsin.

Pierson, Paul, and Theda Skocpol. 2002. "Historical Institutionalism in Contemporary Political Science." In Ira Katznelson and Helen V. Milner, eds., Political Science: The State of the Discipline. New York: W.W. Norton, pp. 693Ð721.

Thelen, Kathleen. 2003. "How Institutions Evolve: Insights from Comparative Historical Analysis." In James Mahoney and Dietrich Rueschmeyer, eds., Comparative Historical Analysis in the Social Sciences. Cambridge: Cambridge University Press, pp. 208Ð240.

Wisconsin Industrial Commission. 1940. Statistical Release #95. Production Workers in Wisconsin Manufacturing Industries. [Data Revised December 15, 1940.] Madison: State of Wisconsin.

Wisconsin State Federation of Labor (WSFL). 1934. Proceedings of the 42nd Annual Convention of the Wisconsin State Federation of Labor. Racine, July 17Ð21.

Wisconsin State Federation of Labor (WSFL). 1935. Proceedings of the 43rd Annual Convention of the Wisconsin State Federation of Labor Appleton, July 16Ð20.