Does capitalism (still) come in varieties?

Abstract

That capitalisms vary and that these capitalisms neatly resolve themselves into distinct and discrete ‘varieties of capitalism’ is an almost foundational claim of contemporary comparative political economy. Yet it is far from evident that it is true. In this article, I return to the varieties conjecture, assessing the degree to which the claim might be warranted. In the process, I argue for the importance of differentiating clearly between ideal types and real types and for the value of heeding Weber’s advice about the dangers of confusing one for the other. I suggest that although capitalisms do not really come in varieties it is sometimes useful to proceed on the basis that they do, particularly if we think of such varieties as potentially dystopic. I suggest that such an acknowledgement is crucial in sensitizing us to the potential biases of varietal thinking, drawing out the implications for the positing of capitalist varieties in the period after the global financial crisis.

Keywords:

• Capitalism
• varieties of
• crisis
• political economy
• institutionalism
• globalization

The starting point for comparative political economy is the claim that capitalism varies—over time, certainly, but more significantly, between spatially discrete political economic units: that there is, in short, not one capitalism but different capitalisms.¹ Indeed, were this not the case there would be no need for a comparative political economy. In a sense, then, the claim that capitalism varies is an ontological antecedent of comparative political economy. But it is typically accompanied by another: that capitalism comes in varieties, and still does so even in a world of globalization in which there might be held to be strong pressures for convergence (Hall & Soskice, 2001; see also Amable, 2003; Garrett, 1998; Schmidt, 2002).

It is in this context that I wish to situate the present contribution. In what follows I seek to explore a potentially heretical claim: that, even if it is accepted that capitalisms vary (that capitalisms come in different forms), it does not necessarily follow that capitalism comes in varieties. Yet we typically proceed on the basis that it does. We infer, I suggest problematically, that if capitalisms vary then capitalism must take varietal forms. In pointing to the potential problems of such a move, my hope is as much to open a debate about the status of the claims we typically make when we propose a varietal solution to the problem of the patterned diversity of capitalisms than it is to resolve these issues once and for all. In the process I suggest that we lack an ontology of capitalist diversity and, more significantly, that such an absence has arguably contributed to our collective failure to see the global financial crisis coming and to our enduring failure to make much sense of it in terms of existing analytic approaches. My hope is to help us retain the analytic traction offered by ‘capitalist varietalism’ whilst sharpening our capacity to use it to make sense of a world in which capitalist crises are not only possible but present.

The argument is simply stated. It is ontological in kind, but its importance lies in its analytic and methodological implications. Although capitalisms vary, capitalism does not really come in varieties. But it is sometimes useful to proceed on the basis of the assumption or heuristic that it does. Crucial here, I suggest, is that we acknowledge the analytic status of the claim and, precisely in so doing, the inevitability of a certain distorting bias in the positing of varieties which follows from that acknowledgement. We need to be careful, in other words, to remember what we are doing in positing varieties of capitalism—and, accordingly, not to reify or ontologise our analytical convenience (see also Ahlquist & Breunig, 2012, p. 108). In the process I suggest that we would do well to learn (or recall and re-learn) a thing or two from Weber about ideal types and their use and misuse in political/economic analysis.

But that is not all. For there is a particular salience to these issues now. I suggest that, especially in the wake of the global financial crisis, we need to rethink our approach to the dynamic patterned diversity of actually existing capitalisms (see also Bruff & Horn, 2012; Hall, 2014; Streeck, 2014). And we need to do so in the light of the predictable (but no less palpable) failure of the varieties of capitalism perspective to anticipate and arguably even to illuminate retrospectively the crisis issuing from the heart of its core ideal type (the liberal market economy).

However, here above all, there is a certain need for clarity. For the argument that I will make about the positing of capitalist varieties does not rest on the global financial crisis. Crucially, it could have been made before it. As I have suggested elsewhere, the capacity to anticipate a crisis is not a good (nor, indeed, a legitimate) test of social scientific theory (Hay, 2019). As such, the problem with current capitalist varietalism is not that it failed to furnish us with a prediction of that specific crisis itself. That is not my argument. Rather, I suggest, the problem with it is that its analytic parsimony gives it no conceptual resources to make sense of that crisis, or of crises more generally, even after the fact. As I hope to demonstrate, crises (let alone the specific character of this global financial crisis) are simply not intelligible in its terms. The presence of crisis, of course, makes us more aware of that now, just as we might be more aware that a political economy incapable of anticipating crisis is inadequate by virtue of that incapacity. But the conceptual problem itself predates the global financial crisis and is independent of it.

Does capitalism come in varieties?

In analytic terms, the positing of varieties is little more than a simplifying device. The re-description in varietal terms of the patterned diversity exhibited by real cases is a simplification of a more complex world. So how might that claim be justified? What kind of claim could we be making (and might those who make such a claim credibly be making) when we (or they) claim that capitalism comes in varieties? What, in other words, is the universe of credible claims here?

In the context of the existing literature, I suggest, the claim that capitalism comes in varieties is invariably one (or more) of the following:

1. The statement of a self-evident truth or truism (which itself requires no justification) and where the task of the analyst is to establish the number and nature of the varieties in question where that is not also deemed self-evident (Calmfors, Driffill, Honkapohja, & Giavazzi, 1988; Esping-Andersen, 1990; Hall & Soskice, 2001; Schneider & Paunescu, 2012; Witt & Jackson, 2016);

2. A hunch, intuition or feeling derived from some prior experiential or, conceivably, ethnographic process of induction (Albert, 1993);

3. A theoretically-derived proposition and deductive inference of the kind ‘if capitalism is like this, then it follows that it should be expected to come in varieties of the following kind’ (Amable, 2003, 2016; Hall & Soskice, 2001);

4. The description of an empirically-established regularity through inductive inference—either through the tracing of a common historical path or trajectory (Streeck & Yamamura, 2001; van Kersbergen & Manow, 2009) or by mapping over time the position of political economies in a multi-dimensional space (Boyer, 2004; Hall & Gingerich, 2009);

5. The positing of an ideal type in the Weberian sense (Hall & Soskice, 2001; Hall & Thelen, 2008);

6. An analytic convenience, an aid to analysis, a convenient simplification (Crouch, 2005; Garrett, 1998; Hall & Soskice, 2001);

7. A purely heuristic device, a thought experiment (that might prove analytically useful);

8. A false premise, a fiction, a distortion, an error of political economic categorization (as, implicitly, in neo-classical economics).

Of these eight possibilities, it is not difficult to associate authors with at least seven. The exception, of course, is the penultimate position (the positing of varieties as an heuristic device)—predictably perhaps that closest to my own (though see also Crouch, 2005, p. 151). It is important to note that a number of these positions are far from mutually exclusive—positions 5 and 6 and, indeed, 6 and 7 might almost be seen as interchangeable. Similarly, position 1 might easily be reconciled with position 2 and positions 3 and 4 might, just as easily, be seen as further elaborations of either. But note, for positions 1–4, varieties of capitalism are real or at least potentially real (varieties exist independently of our knowledge of them; the question is merely whether we are capable analytically of capturing their real world facticity correctly). By contrast, for positions 5, 6 and 7 this is not the case. Here, varieties are theoretical abstractions with, presumably, a rather different ontological and analytic status (since they have no direct real world referent with which we might seek to make them correspond). The point is that, if this is accepted, it should be difficult to hold any of positions 1–4 simultaneously with any of positions 5–7. Or so one might think.

For what is striking is the extent to which key authors in the debate seem ostensibly to hold multiple and incommensurate positions on this question. Take, for instance, Hall and Soskice, so often the intellectual guardians of and guides to ‘capitalist varietalism’. Without ever really addressing the question directly, it is not difficult to see that they inhabit, sometimes in the same work, sometimes in different pieces, sometimes together, sometimes apart and with a range of co-authors, at least 5 and probably 6 of the positions identified above. These include all bar one of the positions for which the claim that capitalism comes in varieties in an empirical fact and at least two of the positions for which it is a purely analytic construct. This is certainly confusing and, as I will suggest, itself the product of an unfortunate confusion with important and wider implications.

It is instructive to examine their position in some detail, not least because it remains the paradigm-defining contribution in comparative political economy (Blyth, 2003; Howell et al., 2003).

Ideal types and real types

But to get all of this right, we need to turn to Weber on ideal types. For much of the mystique and mystery—and arguably much of the confusion too—of Hall and Soskice’s take on varieties of capitalism arises from the claim that the twin varieties to which they appeal are ideal typical. That claim, I want to suggest, is confused, confusing and ultimately problematic in ways which bear directly on the question of the status and value of the varieties of capitalism posited and in ways clearly anticipated by Weber and underscored by the wider literature on ideal types to which he has given rise.

To be clear, I am not the first critic of capitalist varietalism to return to Weber (see, most recently, Hodgson, 2016). But I do so in a distinct way. The conventional Weberian critique of the varieties of capitalism perspective is that it is based on a misunderstanding of Weber’s ideal types—in particular, that it treats Weber’s ideal types as holistic (in the sense that they might refer to an entire variety of capitalism) when they are in fact far from holistic (referring, separately and in a stylized and hence idealized way, to dimensions of capitalism which exist in actual cases only in combination). In short, Hall and Soskice’s mistake is to think of real cases as ‘belonging’ to one or other ideal typical variety (see, for instance, Becker, 2007, pp. 263-5, 276-7; Crouch, 2005, p. 26, 71; Goodin, 2003; Streeck, 2010, p. 28). A more consistent application of Weber’s method would, then, posit not ideal/idealized varieties of capitalism (to which real world cases might ‘belong’) but ideal/idealized modes of capitalist coordination present in combination in actually existing capitalisms.

This is a valid critique, to whose implications we will return presently. But it is not the critique that I advance here. For, more important still is that, even understood holistically, Weberian ideal types stand in a markedly different relationship to the patterned diversity of real cases than Hall and Soskice imply. It is this, I contend, that ultimately leads the varieties of capitalism perspective to its unwarranted and problematic expectations about the dynamic patterned diversity of real cases considered over time.

Weberian ideal typicity

As the literature on the subject, much of it dating from the 1950s, makes clear, Weber used the term ideal type in two rather different senses. But both derive from a common understanding. As Werner J. Cahnman puts it, ideal types are, for Weber, ‘an accentuation or an enhancement (Steigerung) of actually existing elements of reality to the point of their fullest potentiality, amounting to the image of a utopia’ (1965, p. 269)—or, one might add, a dsytopia (a point to which we return presently). As such, and as he goes on to explain, ‘the ideal type is not the description of a concrete reality nor even of the essential features of such a reality (eigentliche Wirklicheit); it is not a hypothesis; it is not a schema under which a real situation or action is subsumed as one instance; it is not a generic concept or a statistical average’. Rather, it is ‘an ideal limiting concept with which the real situation or action is compared’ (p. 269; Weber, 1922 [1949], pp. 190, 191, 192, 194).

This general conception leads Weber (albeit at different stages in his writing) to make the case for the use of ideal types and ideal typical reflection in two different forms. These Watkins (1952) refers to as individualistic ideal types (agential or motivational ideal types) and holistic ideal types (structural or, indeed, institutional ideal types).

In the former (IT1), the ideal type is constructed at the level of the individual actor, as in neo-classical economic theory and its many derivatives (Watkins, 1952: p. 24-5; Weber, 1947). It is the actor that is idealized (or, indeed, rendered dystopic) in that the real world actor is replaced in the ideal typical reflection by a stylized simplification. This pared down account of the individual and his or her motivational disposition is used to derive an idealized model—the analysis, in effect, proceeds by way of ‘as if reasoning’ (if the actor were of this kind then, logically, s/he would act as follows …). The aim is not to describe the real world, but to open an analytic window on it by developing a stylized/idealized model of it. This might then be compared with real world scenarios in the hope of gaining some additional insight or analytic traction (on ‘as if realism’ see Hay 2014; Parsons 2015).

In the second (arguably more familiar) type of analytical strategy (IT2), the ideal type is constructed at the level of the system in question (Watkins, 1952, p. 25-6; Weber, 1922[1949], p. 87, 93, 96). It is holistic, but similarly stylized. Such an ideal type (the protestant ethic, the spirit of capitalism, the liberal market economy) is, again, a theoretical construct and a (potential) aid to analysis. In it, the analyst seeks to capture, clarify and convey precisely through abstraction certain features of the whole to which attention is being drawn in a pure (and hence idealized) way. In both cases (IT1 and IT2), the positing of the ideal type is an heuristic devise, analytically useful insofar (and only insofar) as it aids comparison.

To IT1 and IT2, which come directly from Weber, we might usefully add a third type of analytic strategy, which does not. This also seeks to develop holistic types at the level of the system or institutional complex in question. But it does so not deductively (on the basis of a theoretical hunch or conjecture as to the importance of a factor or process whose pure logic might be captured in the form of an ideal type), but inductively (from an analysis of the patterned dispersion and diversity of actually existing cases). This is not an ideal type, but a real type (RT1). It is, invariably, an inductively inferred statistical archetype or aggregate held to represent (or stand in for) an actually existing empirical cluster (Gerhards, 1995; Spiethoff, 1952; Watkins, 1952; see also Ahlquist & Breunig, 2012; Amable, 2003, 2016; Arts & Gelissen, 2002; Boyer, 2004).

The conflation of ideal and real-typicity

Armed with the distinction between IT1, IT2 and RT1 as different analytic strategies which might credibly be applied to the question of capitalist variation and variety, we can now return to Hall and Soskice. Their claim is that the account of varieties of capitalism that they offer—and, specifically, of liberal and coordinated market economies within it—is ideal-typical (2001, p. 8). Since the categories they deploy are clearly holistic, it would seem obvious that their analysis is ideal typical in the second sense above (it is, in effect, a form of IT2). Obvious, perhaps; but wrong. Ironically, perversely even, Hall and Soskice’s construction (in effect, deduction) of the liberal and coordinated market economies is much more clearly consistent with ideal typical analysis of the first kind identified above (IT1). For even although its aim is to produce an holistic typology, it is ideal-typical, first and foremost in the assumptions at the level of the individual (the ‘micro-foundations’, in standard economic parlance) that it deploys and on which its account is explicitly premised. But, as this perhaps already serves to suggest, things are rather more complicated than this implies. For, in effect, Hall and Soskice derive IT2 holistic ideal types from an IT1 mode of analysis, most clearly in their seminal introductory essay to the Varieties of Capitalism collection itself (2001).

Weber never countenances such a move. He sees IT1 and IT2 as mutually exclusive strategies, as alternative and incompatible means of generating analytical purchase. Indeed, part of the justification for IT2 is to expose the limitations of IT1-type reflection. For Weber, as indeed, in the wider literature on ideal types, questions like ‘does capitalism come in varieties and, if so, how many and which varieties?’ cannot be answered in this way (on the general point, see Papineau, 1976).

Yet, at best, this is only half the problem. For when Hall and Soskice and their various co-authors go through the details of showing us how the holistic types to which they refer are constructed, more often than not they reveal not a deductive logic (the derivation of twin Pareto optimal solutions in a game theoretic scenario, for instance) but an inductive one (see, particularly, Hall & Gingerich, 2009). And that suggests that their holistic ‘ideal’ types are not in fact ideal types (IT2) at all, but real types (RT1). They are, typically, inductively-derived statistical archetypes or aggregates used to capture, in effect, the median position of a (demonstrably) discrete cluster of actually-existing capitalisms when mapped in a multi-dimensional empirical space. Though this is undoubtedly an exercise in typology and typography, it is not ideal typical in Weber’s sense at all. The claim that it is so is misleading—not least because it conflates and confuses IT1, IT2 and RT1. Consequently, it leaves almost completely unresolved the question of the status of the resulting typology (and the categorical distinction between the varieties of capitalism posited). For IT1, IT2 and RT1, as distinct (and arguably incommensurate) analytic strategies, generate very different expectations about whether (and to what extent) we would expect the world of actually existing capitalisms to correspond to any varieties that we might identify.

This might not seem particularly important, but I want to argue that it is in fact crucial to the political economy of contemporary capitalism, above all in the period following the crisis.

Deductively derived ideal typicity

IT1 as an analytic strategy deploying stylized assumptions at the level of the individual is entirely neutral with respect to the question of whether capitalisms come in two or more varieties. Ideal-typical analysis of this kind is, of course, invariably deployed by neo-classical economists for whom the claim that capitalism comes only in one variety is an almost ontological article of faith. But, as authors like Calmfors and Driffill (from whom Hall and Soskice clearly draw much inspiration) in effect show, it is perfectly compatible with the claim that capitalism comes in two (or more) forms (each with a distinct institutional form which structures the games idealized actors might play to produce different anticipated outcomes). For Calmfors et al. (1988), the assumption that the relevant institutional settings for the game they seek to model vary and come in two varieties is externally generated—it cannot be derived from the idealized assumptions about human behavior on which their model is premised. Of course, Hall and Soskice imply that it can—in appealing, implicitly, to some kind of capitalist coordination game in which, for firms at least, there are two Pareto optimal solutions—one liberal and one coordinated (2001, p. 45). But they never show us the details of the game. Yet even assuming that they could, there are at least two objections to this as a strategy for deriving the number of actually existing varieties of capitalism (and the defining features of each). The first is that a different specification of the game would almost certainly produce a different solution—more (or, conceivably, fewer) varieties and/or varieties of a different kind. But second, and arguably more fundamentally, it is just not credible to think that a form of political economic analysis so self-consciously parsimonious at the level of its basic assumptions about human nature (to say nothing of its privileging of the firm as the core actor in capitalist coordination) could be used to derive expectations about the number and nature of varieties of capitalism which we would expect to be able to pick up in the clustering of real world cases.²

In short, it might well be useful, heuristically, to explore the value of the simplifying idea that capitalisms come, not in one, but in two varieties and to do so using game theoretical techniques. But it is surely to misunderstand the value of such a hypothetical mode of reflection to assume that it is capable of generating usefully testable propositions about the patterned diversity of actually existing capitalisms. Yet that appears to be precisely what Hall and Soskice (and their many followers in this respect) assume (amongst the many ‘tests’ of this type performed, see Schneider & Paunescu, 2012; Witt & Jackson, 2016).

IT2 as an analytic strategy proceeds very differently. Here the ideal types are, in effect, parsimonious categories which the analyst constructs in order to explore in a pure or purer form some logic or logics at the level of the system as a whole (here, a variety of capitalism as a political economic order). Such ideal types are, in effect, abstracted extrapolations from the real world, designed to capture a (putative) essence, held to be defining of the variety in question, which might warrant further and more focussed reflection. The logic here is neither purely inductive nor purely deductive. Typically, the analysis proceeds from an inductively inferred hunch treated as an heuristic which is then developed deductively in the fleshing out of the characteristics of the ideal type. The ideal type is, then, at best a stylized extrapolation from processes or logics credibly (and putatively) present in real cases. Its aim is to aid reflection and to sharpen the capacity for comparison. But the crucial point is that there is absolutely no expectation that real cases should correspond at all to the holistic ideal types developed and deployed in this kind of analysis. As Weber himself suggested, such ideal types are limiting extremes (1922[1949], p. 93).

Moreover, and crucially for the varieties of capitalism perspective, that one can posit two discrete, distinct and analytically separable coordinating logics within contemporary capitalism and develop ideal types based on a pure expression of each is not to generate the expectation that real world cases should resolve themselves (or will resolve themselves, under the competitive challenges of globalization or austerity, say) into one or other type.

Crucially, that a categorical distinction can be posited at the level of the ideal type should not lead us to expect a real world bifurcation of cases (as, for instance, in Hall & Gingerich, 2009; Walker, Brewster, & Wood, 2014). For categorical distinctions are inherent in the very nature of ideal types (see also Becker, 2007, p. 277). Put slightly differently, we are almost bound to see things in (artificially) categorical terms if we think in terms of ideal types. This, arguably, is an occupational hazard from which we need to protect ourselves—and from which we need to protect ourselves rather better than we do now (a point to which we will return presently).

Inductively derived real typicity

Finally, we come to the analytical strategy associated with RT1, the positing of real types. Here the question becomes, for the first time, a purely empirical one. Whether capitalism comes in discernible varieties or not can only be established empirically, inductively and, invariably, statistically (through cluster analysis, for instance). As with IT1, the approach itself is agnostic as to whether things like capitalisms come in varieties—it is, in end, for the statistics to tell us whether a varietal solution to the problem of patterned diversity is a good one and, if so, how many varieties we should identify (though, of course, we have to insert some of the parameters—how much of the variance are we prepared to sacrifice for a more parsimonious solution, how many varieties are too many and so forth).

There are two important points to be made about the RT1 strategy. First, the types produced are clearly and consciously acknowledged to be convenient analytical/statistical constructs. Even if they quite closely approximate real world cases, they are not themselves real—their properties are (statistical) artefacts determined by the (statistical) parameters we choose. Their value is in simplifying a more complex world than we can credibly deal with in one go—and to do so with as little distortion as possible. Consequently, if a two varieties solution is best and we have good grounds for seeing in the cases which together comprise each variety the expression of a conserved logic, then we have established a firm empirical basis from which to draw a categorical distinction between these two varieties of capitalism.

Second, and crucially, the relationship between the types identified by this kind of approach and the patterned diversity of actually existing cases is very different from that of IT2. In the former, the ideal types are polar extremes, well beyond the range likely to be exhibited by actually existing cases (there is, in effect, a reductio ad absurdum logic in play); in RT1, by contrast, insofar as clear clusters of cases can be identified the real types to which they give rise are cluster medians. In short, IT2 ideal types are outside of the pack (and they exist in effect in a parallel universe of the analyst’s construction) whilst RT1 real types are in the middle of the pack. As such, whilst the properties of the former are ‘other-wordly’, the properties are the latter are ‘real-worldly’ (and may correspond quite closely to actually existing cases).

Putting this together, it is clear to see the compound conflations and confusions which continue to characterize the varieties of capitalism literature. Here, as ever, the problems of Hall and Soskice’s account are merely indicative of a wider failure to think through the very notion of a variety of capitalism and what it means to posit varieties of capitalism as ideal types. The result is that: (i) they deduce holistic ideal types from individualistic ideal types; (ii) they anticipate a patterned diversity in the real world that corresponds with the holistic ideal types thus derived; (iii) they fail to see that the categorical distinction between the twin varieties they identify is a product of the use of an ideal type analytical strategy in the first place (which can only deal with continuous variables by turning them into categorical distinctions); (iv) they use statistical techniques inductively to identify real types which they (mis)-label as ideal types; (v) they take the fact that their inductive and deductive logics end up with equivalent types as a verification of the former strategy; and (vi) they then proceed to treat real world cases (the US and Germany, respectively) as, in effect, hybrid ideal-cum-real types—despite the fact that they can be neither (in that ideal types are theoretical abstractions and extrapolations and real types are statistical aggregates not outliers).

This is all very well but it remains rather abstract. How might such conceptual problems lead us astray? To see that, it is important to turn from the static treatment of the problem of capitalist variation to the rather more dynamic form that it usually takes.

But before doing so, it is perhaps important to consider, albeit briefly, how and to what extent the preceding critique might apply to other approaches to capitalist diversity. To what extent is the analysis that I have developed more broadly applicable beyond the work of Hall and Soskice, their various co-authors and their many followers?

Beyond ideal typical capitalist varietalism

Limits of space prevent an exhaustive treatment of the now extensive literature. Instead, it is perhaps useful to pick two seminal authors within this debate whose work both posits varieties of capitalism and explicitly uses the term, yet whose approaches might be seen as distinct and as distinctly different from that of Hall and Soskice: namely, Bruno Amable (2003, 2016) and Robert Boyer (2004).

Of the two approaches it is that of Boyer that more successfully avoids the pitfalls of varietal thinking described above. It makes no reference to ideal types, exploring only the theoretically-informed hunch that there may be more than one capitalist institutional configuration conducive to technology-led growth. From this insight it proceeds inductively, using a form of cluster analysis to map the patterned diversity of real cases with respect to variables chosen in response to a specific research question. Its varietal solution is, moreover, not assumed time-invariant and, consequently, the typology to which it gives rise (a three-varietal solution) is not used to generate expectations about the dynamic distribution of cases over time. The approach is far more limited and less ambitious than that of Hall and Soskice; but in all other respects, this is an exemplary form of substantive capitalist varietalism largely immune to the critique developed above.

Amable’s rather more ambitious approach is something of a half-way house between Hall and Soskice’s deductivism on the one hand and Boyer’s inductivism on the other. Predictably, perhaps, it is not immune to the above critique; but it is certainly less floored by it. Though its analysis is, indeed, couched in ideal typical terms in a manner difficult to reconcile with Weber, its ideal types are derived not from stylized game-theoretic reasoning, but from a careful theoretically-informed account of both the key institutional domains likely to influence capitalist economic performance and also of their potential interaction in real world cases—an institutionalist theory of capitalism, in effect (see also Streeck, 2010). This leads Amable to posit a five-varietal solution to the problem of capitalist institutional diversity that he then proceeds to test inductively. Though he is clear that his ideal types are not expected to match real world cases (thereby avoiding the potential conflation of ideal types and real types), he does nonetheless generate expectations about the patterned diversity of real cases from an avowedly ideal-typical reflection. That is clearly problematic in the terms set out above. But much of what is problematic about it could arguably be resolved by the rather more cautious appeal to the language of ideal types. In short, Amable retains much of the ambition of the varieties of capitalism approach whilst avoiding many of the pitfalls of varietal thinking already identified. The remaining problems are arguably less to do with the analysis itself than the manner in which it is presented.

Does capitalism still come in varieties?

Thus far we have confined ourselves to the (static) problem of mapping capitalist institutional diversity at any given moment in time. But what is clear is that the varieties of capitalism perspective is perhaps best known for its treatment of this question dynamically. This brings to the surface a whole host of potential new analytic problems.

What Hall and Soskice and other proponents and followers of the varieties of capitalism perspective typically do is to import their preferred solution to the varieties problem into a no less stylized (though very familiar) account of the competitive environment in which political economies now find themselves (see for instance Amable, 2003; Garrett, 1998; Kitschelt, Lange, Marks, & Stephens, 1999). The assorted challenges of that environment are usually bundled together in the rather generic term ‘globalization’—and globalization, as in a much wider literature, is here essentially synonymous with the claim that nations and national economies must compete successfully to survive and that the rigours of such competition are far more intense today than they have ever previously been.

The varieties of capitalism perspective’s contribution to this literature is two-fold: the idea that there is more than one way to make this work (two ways, in fact); and that the key to understanding how capitalisms can make this work is the economic return, in effect, that they might be seen to gain from their institutional investment or endowment (Hall & Soskice 2001, pp. 36–44, call this ‘comparative institutional advantage’; see also Schneider & Paunescu, 2012; Whitley, 1998, 1999). In short, in an ideal typical account of capitalism’s adaptation to an equally stylized account of the competitive challenges posed by globalization, there are held to be two Pareto optimal solutions. Liberal and proto-liberal market economies need to consolidate their comparative institutional advantage by, presumably, either reinforcing or augmenting their market liberalism; coordinated and (conceivably) proto-coordinated market economies need to shore up their comparative institutional advantage by, presumably, resisting the inexorable march to liberalization and by defending their institutional endowment. The result is a predicted dual convergence, in which capitalisms, at pain of poor economic performance, gravitate towards the liberal and coordinated market economic poles, hollowing-out a dysfunctional middle ground in the process (see also Garrett, 1998).

It is not difficult to see how the problems that we have already identified in the varieties of capitalism literature lead to problems here—and how those problems, in turn, might fail to provide us with the analytical resources to make sense of the period of crisis in and through which we now acknowledge ourselves to be living.

The first problem is that Hall and Soskice’s failure to see holistic ideal types as Weber sees them leads them to treat as real ontologically (and also as categorical) the distinction between liberal and coordinated market economies—and to generate illegitimate expectations about the real world on this basis. What, in fact, they show (and this is no mean feat) is that the highly stylized modelling typically used by neo-classical economists to demonstrate the superiority of pure market liberalism can, if subtly altered to give even some basic consideration to the possibility of non-market-based modes of coordination, generate more than one Pareto optimal solution. But that should not lead us to expect that real world cases will line-up neatly in conformity with the revised model they produce (and its predictions of dual convergence).

This leads directly to a second problem. For it is only on the basis of this epistemic and ontological error (the expectation that what the stylized model elucidates should find a direct real world referent) that good economic performance is associated with proximity to one or other of the hypothesized ideals (Crouch, 2005; pp. 47–55; Deeg & Jackson, 2007, p. 157; Streeck, 2010, p. 30; Witt & Jackson, 2016: pp. 782, 797). Dual convergence follows directly from this, in that the new rules of the game (those prevailing under conditions of globalization) are held to increase the economic returns to institutional investment and endowment. The result is to increase the costs of what Garrett (1998) calls ‘incoherence’ (non-conformity with an ideal type), generating the expectation of a strong ‘centrifugal’ force driving real world economies inexorably towards the idealized poles (though Garrett identifies the poles rather differently to Hall and Soskice, the analysis takes essentially the same form).

A range of additional problems follow from this. First, Hall and Soskice in particular offer us no account of change other than the centrifugal cleansing of institutional dysfunction and no account of institutional dysfunction other than the lack of conformity to a stylized ideal type. This precludes the very possibility that liberal (or, indeed, coordinated) market purism, if taken to its logical extreme, may well sow the seeds of its own downfall. As a number of critics, before and after the crisis, have suggested, the wider problem here is that the varieties of capitalism approach tends to posit as one of its ideal types an understanding of market liberalism that comes straight out of a neo-classical economics textbook (see, for instance, Coates, 2000; Crouch, 2005; Howell et al., 2003; Streeck, 2010). As a consequence, it tends both to present market liberalism, somewhat naively, as institutionally-light (a tendency reinforced by the positing of CMEs as its polar opposite) and, partly as a consequence, as stable precisely by virtue of its disposition to leave markets essentially to regulate themselves (Witt & Jackson, 2016). Tellingly, Hall and Soskice do not discuss regulation at all in their field-defining introduction to the perspective—and neither do they have anything to say about financial markets. After the crisis these seem like egregious omissions—and they are; but arguably they were no less egregious before. Yet more egregious still is that even that which they do consider, they do so in narrowly equilibrium terms.

But in a sense the problem is a more general one. For although (as widely noted) the varieties of capitalism perspective’s account of LMEs is particularly thin (see especially Howell et al., 2003), what applies to LMEs also applies to CMEs. The point is that, in their desire to present each in stylized ideal-typical terms, all potential sources of endogenous stress, tension and contradiction are excluded from consideration (as, indeed, is the possibility that, through interdependence, crises might prove contagious between LMEs and CMEs) (see also Baccaro & Pontusson, 2016; Streeck, 2010).

What we are offered, in effect, is a highly abstract depiction of a liberal and a coordinated market economy in a benign condition of perfect equilibrium (it is only economies that fail to conform to either ideal that are in disequilibrium). The problem here is simple: highly parsimonious accounts of equilibrium are far more dangerous analytically than similarly parsimonious accounts of disequilibrium. And, for now at least, varieties of capitalism perspectives are invariably examples of the former. Yet there is nothing inherent in the positing of capitalist varieties that guarantees this. But understood thus, the perspective leads us to assume that as long as real world political economic systems are recognizable variants of one or other ideal type, we should assume that they will prove stable. A similarly parsimonious account of the (ideal) typical pathologies, tensions and contradictions of LMEs and CMEs might be no less problematic theoretically. But, in both alerting us to the possibility of disequilibrium and crisis and in giving us at least some warning signs to look out for, it is far less dangerous politically. This, I think, is the chastening lesson of the crisis for the varieties of capitalism perspective.

In sum, varietal thinking, at least in its current form has the potential to lead us astray—theoretically, analytically and, above all, empirically. Put simply, it has led us to—and is still in danger of leading us to—expectations about actually existing cases and their developmental trajectories that are unwarranted, unfounded and/or simply wrong. Among these are the following (with, in parentheses, those who have shown these assumptions to be wrong):

1. That proximity to an analytically derived and ideal typical variety of capitalism is an index of good economic performance—at any given moment in time and, above all, prospectively (Ahlquist & Breunig, 2012; Schneider & Paunescu, 2012; Walker et al., 2014; Witt & Jackson, 2016);

2. That, by intensifying the struggle for competitiveness, globalization selects for ideal typical conformity (Bohle & Greskovits, 2009; Hay & Wincott, 2012; Streeck, 2010);

3. That institutional purity (the coherence rather than the hybridity of one’s institutional configurations) is the best way to ensure continued good economic performance (Amable, 2016; Crouch, 2005; Deeg & Jackson, 2007; Witt & Jackson, 2016);

4. That economic performance is a token of institutional functionality (achieved under selective competitive pressures) and that, as a consequence, we have no need to fear path dependent yet destabilizing dynamics (such as the inflation of asset-price bubbles or growth- and risk-enhancing speculative dynamics in deregulated financial markets) (Hay, 2016; Streeck, 2014);

5. That real world economies are not subject to contagion effects since the highly stylized and ideal typical models we construct of them, even in a context of acknowledged globalization, have yet to deal with their interdependence (Baccaro & Pontusson, 2016; Hall, 2014; Streeck 2010, p. 27-8).

Implications and conclusions

As this suggests, there is much to put right in the varieties of capitalism perspective. But, to reiterate, there is nothing inherent in the positing of capitalist varieties to preclude that revision. That said, and as I have argued throughout, to get our treatment of capitalist variation right entails being rather clearer about the status of the claim that capitalism does, indeed, come in varieties in the first place—and following through the implications for our expectations about the real world of actually existing capitalisms.

I conclude with ten simple implications which arise more or less directly from seeking to pose and to clarify as best I can at least some of these issues. These lessons in fact come in three groups relating, respectively, to: (i) the strategy for identifying capitalist varieties; (ii) the attempt to correct the equilibrium-bias in much current varietalist thinking; and (iii) credible extensions to nationalist varietalism that allow us better to acknowledge the systemic interdependence of capitalist economies.

Identifying capitalist varieties

1. Capitalisms do not really come in varieties, even if it is sometimes useful to proceed on the basis that they do. Such varieties, however derived or inferred, are analytical/theoretical constructs lacking direct real world referents. They are best seen as heuristic devices to aid the analysis of multiple cases simultaneously. Their parsimony comes at an analytic price—a price almost certainly worth paying, but a price to which we need to be extremely sensitive.

2. Such ‘varieties’ are best generated inductively (as in Boyer, 2004) on the basis of the exhibited clustering of cases with respect to relevant variables, factors or dimensions (which might themselves be established more deductively). Indeed, varieties of capitalism cannot credibly be derived deductively; the generation of holistic ideal types from individualistic ideal types is a deeply problematic analytic strategy, certainly if it leads to the expectation that the patterned diversity of actually existing cases should correspond in some way to the ideal types posited.

3. There is nothing inherently wrong with the use of game theoretical techniques to model and thereby to explore in a stylized way the coordination problems and dynamics of contemporary capitalism (and capitalisms). But this is not a legitimate basis from which to generate expectations about the patterned diversity of real world cases treated holistically.

4. A varietal solution to the problem of capitalist variation needs to be gauged, judged and evaluated not in terms of its correspondence to the world of actually existing capitalisms and the patterned diversity they exhibit, but in terms of its analytical utility for the task in hand (as, again, in Boyer, 2004). There is potential analytic value to be gained by positing capitalist varieties, but there are also costs. The balance between the two can only be adjudicated on a case by case basis, just as the varieties one might posit (and the means by which one might determine their number and nature) will vary depending on the research question.

5. The analysis of capitalist economies as belonging to distinct varieties should preclude the privileging, above all in the models used to derive such varieties, of a sub-category of relevant actors, such as the firm. Varieties of capitalism cannot credibly be established on the basis of a consideration of the stylized coordination games that firms play, or are seen to play, alone. Finance, in particular, needs to be fully integrated into any reflection on capitalist varieties.

From equilibrium to disequilibrium and ideal to dystopic typicity

• Parsimonious accounts of equilibrium (such as might be generated by a reflection on the properties of ideal types) are dangerous in that they may lead us to the fallacious assumption that as long as real world cases are recognizable as variants of a stable ideal type, they are themselves likely to prove stable. Parsimonious accounts of disequilibrium, though no less problematic theoretically, are far less dangerous in that they sensitize us to the possibility of crisis and are suggestive of at least some of the warning signals.

• As this suggests, the varieties of capitalism approach might most usefully proceed through the development of what might be called ‘dystopic types’ to be set alongside and counter-posed to its more familiar ideal types. These might build from a more general reflection on the crisis-prone character of capitalist accumulation over time (see also Hodgson, 2016). Thus, we might complement Hall & Sockice’s idealized account of the LME, for instance, with a dystopic account of its characteristic contradictions and pathologies (its tendency to leave inadequately regulated the markets it venerates, its tendency to undermine the political stability on which it relies through its various inequality multipliers and, associated with that, its tendency to become ever more reliant on private debt to fuel consumption).³

• Path dependence is no guarantee of dynamic stability. Path dependent processes (such as the inflation of asset-price bubbles or the related herding effects in the behavior of financial market actors or the over-exploitation of natural resources) can be cumulatively destabilizing just as much as they can be cumulatively stabilizing. Moreover, if there are economic returns to institutional investments and endowments and institutions are interdependent, then incremental change may well be the norm but it can also give way to domino effects—as institutional change in one domain may serve to reduce the economic return from institutional endowments elsewhere.

Dealing with systemic interdependence

• The analysis of capitalist variation and variety cannot be limited to a consideration of the institutional configurations that might be seen to characterize particular national models (and the complementarities and dis-complementarities they might be seen to exhibit) alone. It must also consider the more historically variable and politically contingent growth models in and through which any exhibited pattern of comparative institutional advantage and disadvantage is likely to be manifest (see, especially, Baccaro & Pontusson, 2016).

• As well as a consideration of endogenous sources of disequilbrium, it is equally important to consider the interdependence (though trade, finance and shared environmental externalities, for instance) of varieties of capitalism and the growth models to which they give rise. Crises are likely to prove contagious between putative varieties insofar as, for instance, the demand satisfied by a growth model in one variety is generated in and through the growth model of another. In short, it is not sufficient for the stability of a putative variety of capitalism that its endogenous conditions of reproduction are met. For all national capitalisms exist in a relationship of dynamic interdependence with other (national) capitalisms and global capitalism more systemically. This interdependence is a characteristic blindspot of current capitalist varietalism.

In short, viewing capitalist diversity in varietal terms distorts our object of analysis—and the fewer the varieties (the more parsimonious the varietal solution to the variation problem) the greater that distortion. Varietal thinking shines an analytic torch on some aspects of political economic reality at the expense of others. And the varieties we have tended to construct distort principally in that they have led us to anticipate stability, equilibrium and the cleansing of institutional dysfunction to produce dual convergence (particularly in a context of globalization)—when, if differently specified, they could have led us to anticipate disequilibrium, the path dependent accumulation of destabilizing contradictions, and the factors most likely to precipitate crisis. That would have been no less distorting, but it would have been less damagingly distorting—and it would give us more resources to make sense of the world in which we now acknowledge ourselves to live.

My aim has not been to bury the varieties of capitalism perspective but rather to reflect seriously on what is needed to restore it; the phoenix may well rise one more from the ashes. But there is quite a lot of conceptual work required before it will find the wings to fly again.

Notes on contributor

Colin Hay is Professor of Political Science at Sciences Po, Paris and Director of SPERI at the University of Sheffield, UK. He is the author of many books including, most recently, Dictionnaire de l’economie politique (Presses de Sciences Po, with Andy Smith), Diverging Capitalisms (Palgrave, 2019) and The Coming Crisis (Palgrave, 2017, with Tom Hunt). He is lead editor of New Political Economyand founding co-editor of Comparative European Politicsand British Politics, and a Fellow of the UK Academy of Social Science.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

Notes

1 Earlier versions of at least parts of the argument developed here have been presented at L’Institut d’études politiques de l’Université de Bordeaux, Sciences Po, Paris, and at the Institute of Political Science at the University of Tübingen. I am grateful to participants on each occasion for their probing comments and thoughtful suggestions, as I am to both the referees and editors of the Review of International Political Economy. The paper was finished whilst a visiting fellow at Pembroke College, Oxford, whose generosity and warm hospitality I gratefully acknowledge. The usual caveats apply.

2 As David Papineau puts it, “ideal type concepts cannot be used in the formulation of testable generalisations whose acceptance enables the explanation and prediction of actually existing phenomena” (1976, p. 137). To assume that they can is to misunderstand the role Weber assigned to them.

3 Since the focus of my own research has typically been on cases regarded by Hall and Soskice as LMEs, it is perhaps easier for me to imagine a dystopic account of Anglo-liberal capitalism (Hay, 2013); but the logic of the argument is that the development of an equivalently dystopic account of CMEs is no less important.