Richard Hyman

This article is concerned with problems of comparative research and analysis in industrial relations, and in particular with cross-national comparison of trade unions. Comparison is of both practical and theoretical importance, but is fraught with difficulties, in part because of the paradox involved in attempting to generalise concerning national instances which are in so many respects unique. The author considers three different approaches to analysis, in terms respectively of institutions, functions and issues. In conclusion the article emphasises the iterative nature of research and analysis and insists that even if the goal of satisfactory cross-national comparison may be unattainable, its pursuit is both necessary and valuable. In this article I address the questions why, how – and perhaps more fundamentally, whether – one can apply cross-national comparison to the study of trade unions. This is based in part on reflection, in part on discussion of some of the literature on this theme. I deliberately focus mainly on English-language sources. I begin by suggesting that much supposedly comparative work in industrial relations fails to match its pretensions, for comparison requires the deployment of cross-national evidence for purposes of systematic analysis, with the objective of generating plausible explanation of patterns of similarity and difference. In the following section I consider some familiar controversies concerning the nature of comparative method(s), identifying rival conceptions which reflect the well-known confrontation between nomothetic and idiographic methodological traditions. Here I suggest that an iterative approach combining aspects of both traditions is necessary and possible. In the third, and longest section I discuss some of the distinctive problems of comparative research on trade unions, stemming from the emergent and contested nature of trade union identities; and I go on to review three main types of approach: institutional, functional and issue-oriented. Challenges to inherited national distinctiveness, I suggest, may provide a basis for the integration of these different approaches. Finally I consider some problems which stem from distinctive national research tradition – and from nationally-specific modes of conceptualising industrial relations processes – and call for an openness to mutual understanding, both across nations and languages and between trade unionists and academic analysts. 1] Issues in comparative industrial relations Why should students of trade unionism (or of industrial relations more generally) be concerned with cross-national comparison? There are two familiar answers to such a question, one analytical and the other practical. First, comparative analysis is necessary if we are to develop robust explanations and encompassing theories. The literature of industrial relations is littered with generalisations which are assumed to be universal but which are in fact conditioned by the circumstances of time and place. So, for example, American writers in the 1950s and 1960s often identified a dynamic of ‘union maturity’ whereby the early radicalism of labour movements became marginalised or ritualised, political objectives largely discarded, and militancy abandoned or compartmentalised, as union leaders concentrated on the routines of collective bargaining and accomplished sufficient material improvements to keep the rank and file content. Serious comparative analysis would have shown that US experience was in many respects exceptional, and conditioned by distinctive features of American trade unionism and of the economic and political environment in which it operated. (In retrospect it is also clear that the period constituted a purely temporary accommodation between ‘big capital’ and ‘big labour’, made possible by an expansionist phase of ‘Fordist’ production and by US global hegemony; without these preconditions, ‘institutionalised’ industrial relations fell into disarray.) More recently, and in a much more sophisticated manner indeed, Kochan (1980) produced an industrial relations textbook which offered many generalisations on the basis of US practice without any attempt to test their applicability elsewhere. Yet only by investigating experience in other environments can we discover whether our explanatory arguments are contextbound; if they apply in a diversity of times and places, we may assume that we have identified a robust generalisation; if not, we are led to explore which situational factors may explain the differences (Przeworski and Teune, 1970; Bean, 1987). A second rationale for comparative research is its potential role in deriving ‘lessons’ and identifying ‘best practice’. To take some simple examples: the experience of the German IG Metall in 1984-85 in its struggle for a shorter working week was closely studied by its British counterpart before initiating its own campaign at the end of the decade. The ‘Accords’ between the Australian union confederation and the Labour government in the 1980s were to a large degree modelled on European ideas of ‘social partnership’ (which contrasted radically with more militant Australian industrial relations traditions). Italian unions, and in particular the main confederation (the formerly communist CGIL), have looked to both Sweden and Germany in their search for new policy directions. In the 1990s, the British TUC and many of its member unions have been assessing the innovative organising campaigns of their American and Australian counterparts as they struggle to halt and roll back almost two decades of membership decline. (We might ask in passing: how far are the ‘academic’ and the ‘practical’ motives for comparative research compatible? If we consider the unity of theory and practice to be more than an empty slogan, they must surely be complementary. The history of industrial relations in many countries is marked by the failure of heavy-handed efforts to ‘transplant’ policies and institutions which proved successful elsewhere, without adequately understanding the contextual reasons for this success.) Comparative industrial relations is a much abused notion. Comparison surely means the systematic cross-analysis of phenomena displaying both similarities and differences. As indicated above, it both contributes to and is informed by theory and generalisation. Against this criterion, much work which is presented as comparative does not justify the title. Nonauthentically comparative industrial relations can take at least five different forms. The first is the multi-national study, which involves sequential accounts of separate national experience (in edited volumes, often by different authors) but with little or no attempt to provide comparative analysis. The popular text of Bamber and Lansbury (1987) is one obvious example. In the specific context of trade union research, the weighty studies of the ‘Harvard’ team in the early 1980s (Lange et al., 1982; Gourevitch et al., 1984) may also be viewed in this light: the lengthy country-specific accounts are barely utilised for comparative purposes. A second approach is taxonomic. Here, the aim of assigning national cases to categoric boxes overwhelms any attempt at analysis and explanation. One much-cited instance is the essay by Cella and Treu (1993), in which classification is seemingly the principal objective of the authors. This is also the case with the supposedly comparative analysis by Poole (1986), who uses a neo-Weberian conceptual framework as an organising device rather than as an explanatory instrument. A third approach is very familiar: what might be termed the ‘catalogue of diversity’. The characteristic of this genre is that national differences are identified, but with little systematic attempt at explanation. For example, Sturmthal (1972) offers an interesting overview of different national patterns of development; while Banks (1974) provides a series of contrasting pairs of countries with a focus on specific aspects of union organisation and action; but neither provides effective instruments for generalisation. Windmuller (1974), in a ‘comparative study’ introducing a set of national accounts of collective bargaining developments, documents the variety of institutions and practices, while von Beyme (1980) offers a particularly wide-ranging survey of national experience; again, if there is any underlying argument it is the extent and the persistence of variation. Of course, an intellectually defensible position is that diversity is the primordial essence of trade unionism: each national movement (or each individual union, or workplace organisation...) is a special case, subject to distinctive causal dynamics, and hence generalisation is impossible. Such an argument (and perhaps some variants of the ‘societal effects’ school could be interpreted in this way) is possible, though it seems to entail that cross-national research is pointless; but those who adopt the ‘catalogue of diversity’ approach do not explicitly assert this. A fourth variant of ‘not-really-comparative-analysis’ involves elaborate mathematical games with cross-national data sets. Trade union membership statistics offer fertile material for such exercises. The typical approach is to present complex models of the relationship between a range of variables and to subject these to regression analysis; this is followed by a very brief discussion of the results. The primary objective of such work seems more to establish statistical relationships than to explain them.