The Potential of Path Dependence in Political Studies
This article explores the difficulties with both the theoretical content and application of the concept of ‘path dependence’ in political studies, but suggests that, by combining it with insights from morphogenetic social theory, we can provide a coherent framework for its use. After providing a brief survey of the literature on path dependence, it presents a summary of the most significant criticisms made of the approach. The article then moves on to examine morphogenetic social theory and its potential to meet these criticisms before concluding by characterising the elements of a path-dependent system incorporating insights from both new institutionalism and morphogenetic social theory.
Introduction: historical institutionalism and path dependence
The central claim of historical institutionalism is that choices formed when an institution is being formed, or when a policy is being formulated, have a constraining effect into the future (Hall and Taylor, 1996; Koelble, 1995; Peters, 2001). This dynamic occurs because institutions and policies have a tendency towards inertia; once particular paths have been forged, it requires a significant effort to divert them on to another course. ‘History matters’ because formations put in place in the early stages of an institutional or policy life effectively come to constrain activity after that point (Peters, 2001; Skocpol, 1992).
Arguably, historical institutionalism's most distinctive feature is an image of social causation that is based around the notion of ‘path dependence’– the means by which the ‘historical’ gets into historical institutionalism. But, work utilising the concept appears to use it inconsistently, with little agreement on its definition or use of an explicit analytical framework (Goldstone, 1998; Gorges, 2001). Without these elements, path dependence can often appear to be a kind of metaphor for a political organisation in which ‘history matters’ (Dopfer, 1991; Greener, 2002a; Pierson, 2000a; Schwartz, 2003). This is not to say that studies utilising the concept are poor in their own terms, but without any consistency of usage the term ‘path dependence’ is in danger of becoming meaningless.
Path dependence in political studies
Path dependence has become, within a relatively short space of time, a widely-used concept in social science (see, for example, Alexander, 2001; Arrow, 2000; Berman, 1998; Bruggeman, 2002; Garud and Karnoe, 2001; Greener, 2002a and 2002b; Hansen, 2002; Hedlund, 2000; Holzinger and Knill, 2002; Mahoney, 2001; O’Brien, 1996; Pierson, 2000a; Scott, 2001; Sterman and Wittenberg, 1999; Torfing, 1999 and 2001; Wilsford, 1994), but studies often have remarkably little in common in terms of their conceptual framework or approach. A brief consideration of a few representative cases demonstrates this.
Alexander's (2001) study of democratic consolidation is based explicitly on a rational choice framework, within which actors are voluntarily bound by history because of the positively perceived trade-off between short-run costs for long-term improvements. As such, path dependence is perceived to be the result of a rational cost-benefit calculation. In contrast, Dimitrakopoulos (2001) describes the relationship between European integration and national parliaments in terms of a process of incremental change that he appears to equate with the concept of path dependence. Hedlund's (2000) study of Russian policymaking under Yeltsin has clearly-defined mechanisms specified for the exertion of history over the policy process, but within a conceptual model that is sometimes unclear. Equally, Berman's (1998) excellent study of political action in response to the depression is rich on detailed analysis of the empirical case, but its conceptual framework for using path dependence is not always explicit. These studies have a concern to introduce the role of history as a central feature in political analysis, but the means through which they achieve this are so diverse that the studies often have as much in difference as they do in common. In Alexander's case, path dependence is entered into voluntarily within a specific framework, not really requiring a mode of reproduction – the rational choice framework means it is unnecessary, whereas in Hedlund's study such mechanisms become the centrepiece of analysis. Equally, the apparent model of human action varies from the hyper-rational in Alexander's case, through to an acceptance that the very source of path-dependent behaviour might be due to technically irrational behaviour such as habit or insecurity. There is no consistent approach to the problem of agency and structure within studies making use of path dependency – a difficulty it shares with the historical institutionalist approach upon which it rests (Hall, 1998; Hay and Wincott, 1998).
As well as the substantial empirical work examined above, theoretical papers also exist that attempt to consider how we might construct a framework for specifying what elements and circumstances combine to form a path-dependent system (Goldstone, 1998), and how path-dependent systems manage to reproduce their form (Mahoney, 2000 and 2001; Pierson, 1993 and 2000a; Thelen, 1999; Thelen and Steinmo, 1992). Arthur’s work on path dependence (1989 and 1990; Arthur et al., 1983) provides a series of discussions on how positive feedback mechanisms pervade economics (in contrast to more traditional assumptions that are concerned with equilibrium concepts, and so negative feedback). Arthur is keen to demonstrate the sub-optimality that can result from possible multiple equilibria being subject to positive feedback mechanisms, and so inferior technologies becoming ‘locked in’ to dominance through little more than chance (Arrow, 2000; David, 1985 and 1997). Political studies based around Arthur's work tend to make extensive use of multiple equilibria and positive feedback, and so suggest a framework where there are initially multiple possibilities from which a sensitivity to the ‘initial conditions’ of the political situation lead to a particular policy or institution becoming ‘locked in’ through a series of contingencies, and then maintained through some kind of positive feedback mechanism (Goldstone, 1998). Mahoney (2000 and 2001) goes further than this, suggesting that maintenance might occur through either positive feedback or ‘reactive’ mechanisms of interplay between interest groups.
Criticisms of path dependence
Despite the conceptual work of Archer, Arrow, Goldstone, Pierson and Mahoney, we still appear to lack a coherent framework for delimiting what elements might comprise a path-dependent system, preventing case comparisons and the possible generation of additional theoretical insights from their use. A number of specific criticisms emerge from the literature that we must consider if we are to use path dependence as a coherent framework in political analysis.
Many of the problems with path dependence come from its intellectual roots in historical institutionalism. First, if path-dependent political processes preserve the past in their form, how do we break free from them? If history matters so much, how do we break from it? How does change occur? (Gorges, 2001; Hira and Hira, 2000). In historical institutionalism, political processes come to resemble ‘punctuated equilibria’ (Krasner, 1984), where substantial change is only possible in ‘critical junctures’ (Collier and Collier, 1991) or ‘policy windows’ (Kingdon, 1995 and 1996) before institutions and policies once again settle down on to a new path, and inertia becomes the norm. Hall and Taylor (1996, p. 942) admit that well-developed responses to the question of why critical junctures arise have not yet been formulated – change is effectively an exogenous feature of the model.
Second, what exactly is the role of ideas in path dependence (and historical institutionalism generally) (Blyth, 1997)? What is the relationship between ideas and history, and how can they combine to create continuity and resist forces for change in the past? Path dependence often attempts to locate institutions in a causal chain that gives a substantial role to the development of ideas (Hall, 1993; Mahoney, 2001; Torfing, 1999 and 2001), but that has also led to some substantial criticisms of the approach with claims that ideas are not treated in a systematic and coherent way (Blyth, 1997), and that ideas are being used as a means of ‘propping up’ the institutionalist research agenda without appropriate care.
Third, how can we characterise the feedback mechanism through which path-dependent processes prevent change? Are they subject to increasing returns and positive feedback mechanisms (Arrow, 2000; Arthur, 1990; Pierson, 2000a) or can we include negative feedback mechanisms as well (Mahoney, 2000)? If we include both kinds of feedback mechanisms, are we at risk of losing the distinctiveness of path dependence as a concept (Schwartz, 2003), so reducing its use to a loose metaphor rather than a clearly defined framework for analysis?
Finally, Pierson, one of the most significant writers on path dependence, (see, for example, Pierson, 1993, 1996, 2000a and 2000b), has commented that the diversity of studies now being published under its name risks ‘concept stretching’ (Pierson, 2000a, p. 252) occurring, and the risk of it becoming meaningless.
These are considerable problems. To attempt to resolve them, we turn to morphogenetic social theory to argue that it holds considerable complementary explanatory power, and by incorporating key insights from it, we can more clearly elaborate what we mean by path dependence and so find solutions to many of the problems raised by its critics.
Morphogenetic social theory
Morphogenetic social theory (Archer, 1982a, 1982b, 1995, 1996a, 1996b, 2000a and 2000b) provides an analytical approach based around the ontology of critical realism (Archer et al., 1998). Archer's ideas have been discussed in organisational studies and sociology (King, 1999; Reed, 1997 and 2001; Stones, 2001; Willmott, 2000), but appear to have not yet ‘crossed over’ widely into political studies. The morphogenetic approach divides analysis into three interrelated stages. The morphogenetic approach first analyses the structural and cultural ‘conditionings’ that act as an influence on human actors, and which create ‘emergent properties’ and ‘situational logics’ for their interactions with them (see below). Second, it explores how these conditioning factors influence actors within the system through their interactions with them, primarily in the form of their behaviour in vested interest groups. The third and final stage analyses the result of these interactions, and the resulting conditioning effects that will feed into the next morphogenetic cycle.
As well as having specific analytical stages, we can characterise the morphogenetic approach as having two particular ontological characteristics derived from critical realism that are especially distinctive. First, there is an analytical separation between structure and agency. Morphogenetic social theory recognises the interdependence of structure and agency, but claims that the two are analytically separable because of the additional insights that can be generated in this way. So, in stage one of the morphogenetic cycle, we analyse the structural and ideational influences present in a political system, before considering how these interact with human agency in stage two, and the result of those interactions in stage three. As such, we have a distinctive means of approaching the debate between structure and agency that has been at the heart of debates on historical institutionalism (Hall and Taylor, 1996; Hay and Wincott, 1998). In addition to this, there is a genuine sense of the role of history in each morphogenetic cycle we analyse – we see the interactions between pre-existing ideas and structures and human actors unfold before us (see Archer's extended examples in 1995 and 1996).
Second, morphogenetic social theory also specifies an analytical separation between ‘structural’ and ‘cultural’ systems. This means that we examine the realm of institutions (the structural system) separately from the realm of ideas (the cultural system), and so deal explicitly with Blyth's (1997) criticism that ideas are not treated seriously within historical institutionalism, as they acquire an analytical dimension all of their own. As with structure and agency, morphogenetic social theory does not treat ideas and structures as systems that do not interact, but instead suggests they should be treated as being separable because of the additional analytical power that this approach creates. We will specify below how this applies to the specific area of path dependence.
Examining a morphogenetic cycle, then, consists first of specifying the structural and cultural influences upon the political system that we are analysing. Morphogenetic social theory utilises two specific means of analysis in its first stage. First, we must consider the ‘emergent properties’ of the system in both structural and cultural spheres. An emergent property can be either ‘necessary’, in which relationships between those dominating the system are recognised by all the parties involved as being interdependent, or ‘contingent’, in which vested interests in the dominant faction believe they are able to work relatively autonomously from one another. As well as this, emergent properties are either ‘compatible’, in which the dominant vested interests have a considerable amount either culturally or structurally in common, or ‘incompatible’, in which case they have not. By specifying the nature of the dominant vested interests in terms of the extent of their interdependence and compatibility in both the structural and cultural spheres, we are able to assess the probability of their ability to preserve their position (we can hypothesise that necessary relationships tend to be longer lasting than contingent ones) and their likelihood of conflict (we can hypothesise that compatible relationships tend to be less conflictual than incompatible ones).
Second, we must consider the ‘situational logic’ that the emergent properties are likely to create. The combinational possibilities, along with their likelihood of path dependence, are shown in Table 1.
Table 1. : The likelihood of path dependence from structural and situational logics
|Necessary incompatibilities||Compromise||Correction leading to syncretism||High – medium|
|Contingent incompatibilities||Elimination||Choice (forcing of)||Lowest|
|Contingent compatibilities||Opportunism||Cultural ‘free play’||Low|
A morphogenetic analysis of path dependence
In morphogenetic terms we can hypothesise a path-dependent system as most likely to emerge where both structural and cultural vested interest groups are dependent upon one another to hold power (a ‘necessary’ emergent property). Where structural interest groups are dependent upon each other (‘necessary’) and their goals are compatible, this leads to a situational logic where interest groups attempt to ‘protect’ themselves. This is especially powerful where the dominant cultural ideas utilised by vested interests are also compatible – and the greater the compatibility the greater the possibility of generating increasing returns in the political system we are investigating. This combination of powerful structural interests plus mutually compatible ideas is the most stable of the ‘morphostatic’ (or stable) cycles Archer posits (Archer, 1995, pp. 308–312), and, so, is the most likely to condition path dependence.
The production of path dependence through increasing returns is therefore most likely to occur in systems where necessary complementarities exist in both the structural and cultural systems. It may also occur where there are necessary incompatibilities in either the structural or cultural system, especially when allied to a necessary complementarity in the other. Where there are contingent incompatibilities in both the structural and cultural systems (vested interests have rival structural or cultural loyalties based around differences of opinion that cannot be easily resolved), the likelihood is reduced, however. As we can see from Table 1, emergent properties with ‘contingent’ relationships (where vested interest groups are not dependent upon one another) tend to lead to situational logics that favour non-stable (morphogenetic) outcomes such as ‘elimination’ of rival groupings, and so have a low likelihood of being path dependent. The results of the conditionings from these combinations are more likely to result in change than continuity.
We therefore have a means of explaining systemic continuity, but also need a clearly specified mechanism for change (Hira and Hira, 2000). In a morphogenetic-inspired model of path dependence, forces for change can come endogenously or exogenously, or both. If we have ‘necessary’ and ‘compatible’ relations in both structural and cultural spheres, we have the most powerful force for morphostasis (continuity), with actors engaged in a logic of protection in both areas. Even this, however, can eventually lead to change because of the very limited range of legitimising ideas that are being drawn from (extreme specialism can ultimately lead to irrelevance), or through the structural vested interest groups becoming so insular that they engage in factional infighting, and war on ‘deviant’ groups or ideas to the extent that they actually cause these groupings to begin to establish separate identities and differentiated ideas (Archer, 1995, pp. 237–239). Change may also come from exogenous factors, such as a wider shift in structural societal relations at the level of international political economy (‘fiscal crises’ for example), or through the emergence of challenging ideas that are backed by vocal and powerful vested interests (this is more or less the argument as currently presented in most historical institutionalist analyses that utilise path dependency (Greener, 2002b)).
Equally, in less stable versions of path dependence, where the structural or cultural systems have a built-in incompatibility, attempts at ‘compromise’ or ‘syncretism’ can break down, and vested interest groups attempt to achieve greater power for themselves. This, again, can come about as a result of either endogenous or exogenous factors. Endogenous change would come about as a result of a significant group no longer being able to sustain the incompatibility built into the system, and so fragmentation occurring despite attempts from structural or cultural vested interests to continue to compromise. Exposure of incompatibilities would tend to result in greater mobilisation from other vested interest groups against them, and the likelihood that actors would be forced into choices about whether or not they would continue to support the dominant coalition or idea.
Situational logics do not create compulsory rules for those operating within them, but actors do have to work within the context where they prevail. As such, the opportunity cost of working against them is likely to be high – in other words, particular configurations of these ‘emergent properties’ are more likely to lead to path dependence than others, and a change in emergent properties resulting from interaction between actors and the situational logics that they face will result in the system becoming either more morphogenetic (generating change) or morphostatic (generating continuity), depending upon the new prevailing situational logic.
Towards a framework for the analysis of path dependence
From the insights generated above from both existing path-dependence literature and Archer's morphogenetic approach, we are now in a position to specify a framework for considering path dependence in political processes.
First, path-dependent processes begin with multiple equilibria situations (Goldstone, 1998; Mahoney, 2000; Pierson, 2000a). We must be able to demonstrate that a number of viable alternatives existed for the development of the policy in question, or for the development of the institutions we are examining. Leading on from this is the second element: contingent events must be shown to have played a substantial role in establishing the particular policy or institutional form that emerged.
Third, we must specify the conditions in which we would expect path-dependent systems to reproduce their form and ‘lock-in’ to occur. The use of insights from morphogenetic social theory allow us, through the analysis of the relationships between vested interests in the structural and cultural spheres, to begin to generate hypotheses about the likelihood of continuity occurring in political systems. After the period of production, a period of reproduction appears during which the policy or institution must generate feedback mechanisms that create inertia, or possibly even increasing returns to ‘lock out’ competing political ideas and vested interests. Once the logic of path-dependent policy or institution has been established, it will tend to generate an inertial force where established vested and cultural interests have a high opportunity cost for challenging the system (based on a ‘necessary’ relationship both within and between the groups). This will tend to lead to morphostasis, which is most likely to appear where ‘necessary’ emergent properties are reproduced in the policy or institution.
Critics of path dependence have, as we noted above, made great store of the lack of specificity in the model of the nature of the returns that the model is specifying. Pierson (2000a and 2000b), basing his work on Arthur (1989 and 1990), makes clear that he believes that path-dependent systems are the study of positive returns, but this approach is logically flawed. Were positive returns to dominate a system for a considerable period of time, this would eventually lead to the removal of all opposition as the influence of the dominant idea or vested interest became so widespread that there would no longer be anywhere left for notions or interests that challenged them. This would effectively take away the opportunity for change completely, short of an exogenous shock. In the world of physics, where models of path dependence consider stable equilibria generated by the selection of coloured balls from bags, this lack of internal change might be possible, but in the world of political systems we need some modification of the idea. As such, it makes sense to hypothesise, in line with the insights given above, that a path-dependent system might go through, in its creation phase, a period where increasing returns are generated. Once it enters its reproduction phase, however, it seems unlikely that anything greater than the preservation of the status quo is possible – a situation more in line with constant than increasing returns. As Schwarz notes (2003), there are considerable costs involved in keeping things the same. We are able, by analysing emergent properties in the political system and their corresponding situational logics, to locate the difficulties involved in keeping the system on its particular path, and the likely costs involved.
Finally, as we noted above, we have a mechanism for change in a path-dependent system, located not in the cultural or structural spheres, nor in human agency, but in the interactions between all three. The analytical separation of cultural and structural systems allows us to examine the role of ideas as well as structure in political analysis, and the analytical separation of both from human agency through the three-step morphogenetic cycle allows us to see the process through which change unfolds from apparent continuity before us. We have no need to make the source of change an exogenous factor in our model, as historical institutionalism tends to do.
Path dependence can be retrieved from the rather unsystematic use to which it is presently put. Its combination with morphogenetic social theory appears to be a potentially fruitful one. In providing a more coherent framework for its use, and combining it with coherent social theory and ontology, we can both deepen its analytical content as well as provide a basis from which hypotheses and comparative studies about political stability and change can be drawn. Path dependence has considerable potential for providing the basis for substantial empirical studies of the linkages between policy and organisation continuity and change and the relationships within institutions that shape the behaviour of agents and the structural and cultural conditionings that act upon them. Providing a more detailed framework than presently exists allows comparative cases to be constructed in a more systematic way, and for the complex processes making up political life to be better understood.
Thanks to the reviewers of this journal for their helpful comments on this article's improvement, to Hugh Pemberton for organising the specialist stream on path dependence at the PSA Annual Conference at Leicester University in 2003, to Fiona Ross for her encouragement and criticism at that panel, and to Herman Schwartz for his provocative and incisive critique.