Table of contents(17 chapters)
Choosing Varieties of Capitalism as the title of their 2001 edited volume, Peter Hall and David Soskice monopolized a label that was much too broad for the project they were actually reporting. Their project was in line with a style of research, which may be called “bringing yet another factor back in”. That term stems from another pioneering edited volume emerging – like Hall and Soskice's volume – from the Harvard circuit: Evans, Rueschemeyer, and Skocpol's (1985) Bringing the State Back In. Following that volume, a number of other factors were “brought back in”: classes, geopolitics, finance and so on.
In their well-known contribution to the “varieties of capitalism” debate, Peter Hall and David Soskice (2001, Ch. 1) highlight the distinction between a “coordinated market economy” as exemplified by Germany and a “liberal market economy” as exemplified by the United States. Under the heading, “Liberal Market Economies: The American Case”, Hall and Soskice (2001, p. 27), argue:Liberal market economies can secure levels of overall economic performance as high as those of coordinated market economies, but they do so quite differently. In LMEs, firms rely more heavily on market relations to resolve the coordination problems that firms in CMEs address more often via forms of non-market coordination that entail collaboration and strategic interaction. In each of the major spheres of firm endeavor, competitive markets are more robust and there is less institutional support for non-market forms of coordination.
This paper is ambitious. Its central purpose is to examine how a number of developed economies, plus the largest developing economy, vary in terms of corporate governance: USA, Japan, Germany, UK, France, Italy, South Korea, Taiwan, Sweden, Switzerland and mainland China. We understand corporate governance in a very broad sense, descriptive not prescriptive: as who controls and influences firms, and how. We are thus dealing very much with varieties of capitalism. In a sense, we shall be seeking to characterise national systems of corporate governance, but we must stress that our concern is always with the situation of the individual firm. We shall find it convenient most of the time to give one label to a country's whole economy, but this will always be an approximation, which conceals variations among that country's firms. At other points, we shall distinguish types of firm and indicate the rough proportions of each type in a particular economy.
An Early Approach to the Varieties of World Capitalism: Methodological and Substantive Lessons from the Senghaas/Menzel-Project
Recurrent “methodological disputes” have haunted the social sciences, again and again polarizing the case-oriented quest for specification against the natural science inspired quest for general, high-level theory. As a consequence, too much social science research is captured in either one of two vicious circles: ever more highly specified monographic case studies or preoccupation with periodically shifting general theories. The interaction of these two circles increases the risk of widespread amnesia: as social scientists are either bogged down in a stream of cases or flying high with the most recent grand (meta-)theories, social science forgets the actual empirical knowledge that is being meticulously created, maintained and revised in the daily handicraft carried out by a growing mass of researchers.
There is little doubt that in terms of speed and scale, China's economic transformation is without parallel in the past. Never has the world seen a major economic power emerge in such a short time span and attain such a weight in the total world economy. Intriguingly, few social scientific analyses have explicitly interpreted the massive socio-economic changes taking place within China as associated with the emergence of a capitalist political economy.
Institutions underpin the operation of national economies. These differ significantly between countries reflecting varying historical paths, policy choices and national cultures. Moreover, they need to be understood systemically as an ensemble of relations between their component parts: financial systems, corporate governance, industrial relations, patterns of state intervention, etc., have evolved together so that their operation and effects tend to reinforce each other. Different countries faced by common exogenous changes will tend to evolve along different lines rather than converge. National institutions matter: they significantly affect economic performance and distribution.
The difficulties that MR poses for comparativists were anticipated 40 years ago in Sidney Verba's essay “Some Dilemmas of Comparative Research”, in which he called for a “disciplined configurative approach…based on general rules, but on complicated combinations of them” (Verba, 1967, p. 115). Charles Ragin's (1987) book The Comparative Method eloquently spelled out the mismatch between MR and causal explanation in comparative research. At the most basic level, like most other methods of multivariate statistical analysis MR works by rendering the cases invisible, treating them simply as the source of a set of empirical observations on dependent and independent variables. However, even when scholars embrace the analytical purpose of generalizing about relationships between variables, as opposed to dwelling on specific differences between entities with proper names, the cases of interest in comparative political economy are limited in number and occupy a bounded universe.2 They are thus both knowable and manageable. Consequently, retaining named cases in the analysis is an efficient way of conveying information and letting readers evaluate it.3 Moreover, in practice most producers and consumers of comparative political economy are intrinsically interested in specific cases. Why not cater to this interest by keeping our cases visible?
The first thing most people learn in statistics is that correlation is not causation, and that inferring causation from statistical results requires that there is a theoretical model (a good reason to think there is a causal effect), not just a statistical one. Usually, this implies a theory with some “mechanisms” that may also be subject to investigation. Except in some quite limited senses of the term, almost no one thinks that any MR results justify a causal claim (Goldthorpe, 2001).
For comments on a previous draft, I wish to thank Mary O’Sullivan, Michael Shalev and Bruce Western.
Michael Shalev has turned his attention, once again, to the bad methodological habits that social scientists – like myself – often adopt. As always, he presents us with thoughtful, rigorous, and penetrating criticism, but also with a generous dose of constructive prescription. His target is the widespread use of regression techniques in cross-national comparative research. The gist of the argument is that multiple regression (MR) is a far too blunt instrument if our aim is to arrive at a robust identification of crucial causal mechanisms. MR, as he puts it (p. 42), renders the cases invisible and, hence, precludes researchers from having any dialogue with them. The case becomes a set of scores; the causal mechanisms are reduced to correlation coefficients. As a result, analytical power is sacrificed rather than gained. Shalev advocates simpler ‘low-tech’ approaches such as tabular representations, tree diagrams, or clustering techniques either as substitutes for, or as companions to, regression analysis.
Shalev's third suggested path for progress consists of using tables, graphs, and tree diagrams to examine causal hierarchy and complexity and to identify cases meriting more in-depth scrutiny. This should be viewed not as (or at least not solely as) a substitute for regression but rather as a critical component of regression analysis. All of us were (I hope) taught in our first regression course that it is not enough to simply get the data, estimate some regression equations, and then draw conclusions. It also is necessary to get a feel for the data, in large part by examining descriptive statistics and looking at bivariate and/or multivariate patterns. Too many macro-comparativists, I suspect, either do not do this at all or do not do it sufficiently carefully.
I am grateful to the editors of this journal to be given the possibility to comment on Michael Shalev's article. Although I have some minor disagreement with his general argument, I am also grateful to Michael Shalev for taking up what I think is an important question in comparative social science. I find myself in the curious position of being a target of a general critique that I mostly agree upon, namely that too much energy is going into sophisticated methodological techniques at the expense of substantive knowledge about individual cases and theoretical reasoning about causality. However, and probably not surprisingly, I find Shalev's critique of my particular venture into this area far from convincing.
Perhaps one of the most important points Shalev makes is that Przeworski and Tuene's (1970) admonition to comparative political analysts to replace the proper names of cases with concepts and variables, or to pursue what Ragin (1987) dubs variable-oriented research, has gone too far. In Shalev's view, cases (typically nation states for the purposes of this discussion) have become all but invisible. This is particularly troublesome in Shalev's mind because, at least as far as comparative analysis of developed democratic capitalist systems is concerned (and we can say the same for political research on Latin America, Africa, or Asia), the cases are few enough to know quite well and bring to the forefront of sophisticated analysis.1 In addition, Shalev makes the distinct point that the theories we seek to test in comparative political research entail complex and often non-linear causal sequences: causes of particular political outcomes are commonly contingent on the presence of other forces, or conjunctural with temporally and spatially bound forces and contexts. In fact, in comparative theory, it is fair to argue (as Shalev does) that causal explanations of important political outcomes are often put forward in terms of complex configurations of multiple factors. Moreover, in theory and in practice, we are often confronted with the prospect of multiple configurative paths of causation of the same outcome. In the end, Shalev believes that the linear and additive logic of general MR analysis, as well as the more sophisticated versions with non-linear specifications and interaction terms, cannot adequately test our complex theories.
Shalev's (2007) critique of the use of multiple regression in comparative research brings together and synthesizes a variety of previous critiques, ranging from those focusing on foundational issues (e.g., the persistent problem of limited diversity), to estimation issues (e.g., the unrealistic assumption of correct model specification), to narrow technical issues (e.g., the difficulty of deriving valid standard errors for regression coefficients in pooled cross-sectional time-series models). Broadly speaking, these concerns can be described as epistemological, theoretical, and methodological, respectively. While the distinctions among these three are not always clear-cut, the tripartite scheme provides a useful way to map the different kinds of critiques that may be directed at the use of regression analysis in comparative research. In the first half of this essay we build upon Shalev's discussion to clarify the conditions under which regression analysis may be epistemologically, theoretically, or methodologically inappropriate for comparative research. Our goal is to situate Shalev's specific critiques of the use of multiple regression in comparative work within the context of social research in general.
At first sight the reader may find it odd that I have grouped the commentaries by Esping-Andersen and Rubinson/Ragin together, given that while the former proposes revamping the use of MR in comparative research, the latter offers a radical alternative. Nevertheless, while suggesting different solutions, both largely agree with my diagnosis of the problems. Also, my principal response to both is that their practical proposals look promising, yet are difficult to judge. That would require, (a) a user-friendly guide to implementing the advocated techniques; and (b) side-by-side comparison of the results obtained by their favored methods, the conventional MR method, and my own suggestions. Let me hasten to add that given (a), I would be ready to undertake the work involved in generating (b).