What choice for Europe? Reflections on agency and structure in Liberal Intergovernmentalism
This article examines how the relationship between agency and structure is dealt with in Liberal Intergovernmentalism, a prevailing theory of European integration. It demonstrates that, contrary to the widespread view that it is agency-centred, Liberal Intergovernmentalism is in fact a highly structuralist theory in the issue areas it claims to explain best. In these areas integration is ultimately explained in terms of developments in economic structures, leaving no room for agency and ideas. The article also shows that, despite the importance it ascribes to changes in economic structures, Liberal Intergovernmentalism fails to theorise their possible causes.
Liberal Intergovernmentalism; Moravcsik; Agency; Structure; Integration theory
Over the past two decades Andrew Moravcsik’s Liberal Intergovernmentalism (LI) has established itself as one of the prevailing theories of European integration. Elegantly combining a liberal theory of preference formation with an intergovernmentalist theory of interstate bargains and a functional theory of institutional choice it explains European integration as the outcome of a series of intergovernmental negotiations. More than any other contemporary theory of integration LI and its application in empirical analyses has provoked discussion in the field of EU studies. Opinions are divided between those who admire LI for its parsimony and predictive power and those who feel that its account of regional integration misses out on too much of importance. Either way, hardly anyone would dispute that it continues to be a theory that it is necessary to relate to in one way or another in theoretically informed work on European integration.
If you need assistance with writing your essay, our professional essay writing service is here to help!
The purpose of this article is to critically examine the liberal intergovernmentalist explanation of integration from a meta-theoretical perspective. More precisely, it will be systematically analysed how the relationship between agency and structure is dealt with in LI. Any theory’s account of the social world, or delimited parts of it, is based on a particular, albeit often implicit, conceptualisation of the agency-structure relationship – and whether or not this conceptualisation is convincing impacts greatly on the quality of the theory’s account of social phenomena and change. There is thus much to be learned about a theory, in this case LI, from examining its underlying assumptions with respect to agency and structure. This is even more so because appearances can be deceiving: as it will be argued in this article, LI which appears and is widely assumed to offer an agency-centred account of European integration, turns out to do the opposite on closer scrutiny.
In addition to this introduction and a conclusion the article is divided into seven sections. The first two sections set the stage for later analyses by briefly introducing LI and the question of the agency-structure relationship, while also accounting for their respective significance. The following three sections examine how the agency-structure question is dealt with at each of the three stages of LI: preference formation, interstate bargaining and institutional choice. Against this background section six critically examines the liberal intergovernmentalist explanation of European integration before section seven discusses the political implications of LI.
1. Liberal Intergovernmentalism
Andrew Moravcsik’s Liberal intergovernmentalism (LI) was first presented in the early 1990s and later elaborated and applied in a string of publications of which the monumental book The Choice for Europe (1998) contains the most detailed exposition and test of the theory. LI is presented as a framework for synthesising theories into a coherent account of regional integration. The latter is explained as the result of ‘a series of celebrated intergovernmental bargains’ (Moravcsik, 1993: 473). More precisely integration is seen as the outcome of a three-stage process where: (1) national interests or goals arise in the context of domestic politics; (2) governments bargain with each other to further their national interest; and (3) governments make an institutional choice to secure credible commitment once a substantive agreement has been reached.
LI quickly became a focal point in debates on how to theorise European integration and it has subsequently kept this position. According to Moravcsik and Schimmelfennig (2009: 67), LI ‘has acquired the status of a “baseline theory” in the study of regional integration: an essential first cut explanation against which other theories are often compared’. In their view, ‘it has achieved this dominant status due to its theoretical soundness, empirical power, and utility as a foundation for synthesis with other explanations’ (2009: 67). To be sure, not everyone would agree with this latter sentiment. As alluded to in the introduction, several scholars have criticised the theory for painting a too incomplete or even misleading picture of the European integration process and the ‘empirical power’ of the resulting analyses has often been questioned (e.g. Diez, 1999; Smith, 2000; Wincott, 1995; see also Cini, 2007: 112-14 for an overview of some critiques of LI). Inasmuch as relatively few scholars besides Moravcsik appear to wholeheartedly embrace LI (Pollack, 2001; however, cf. Laursen, 2002), it is probably fair to say that it has acquired its status as a “baseline theory” as much because of its perceived weaknesses as because of its strengths.
Similar to Waltz’s (1979) neorealism LI is a parsimonious and bold theory that lends itself to accusations of neglecting or underestimating the significance of important parameters – in the case of LI for instance transnational business groups and activist supranational institutions. Indeed, LI does this deliberately, seeking ‘to simplify EU politics, stressing the essential and excluding certain secondary activities’ (Moravcsik and Schimmelfennig, 2009: 68). Hereby it follows the neo-positivist recipe for theory construction, according to which theories should take the form of simplified models that can support efforts to make generalisations by singling out as few variables as possible and account for the causal relations between, and the relative weight of, these variables. Falsifiable hypothesises are derived from such theories and subsequently tested against reliable empirical data. On the basis of such tests, theories can then be further refined or occasionally discarded. (1)
Testing LI is precisely what Moravcsik sets out to do in The Choice for Europe. Here standardised hypotheses derived from LI and competing (albeit for the most part artificial) theories are tested against an overwhelming amount of empirical data in five cases studies. Needless to say, LI comes out on top as the theory with the by far greatest explanatory power. More generally, The Choice for Europe constitutes an example par excellence of research informed by neo-positivist methods and standards. In its early pages Moravcsik informs his readership that the book ‘eschews ad hoc explanation and seeks instead to discover what is generalizable about EC history’ (1998: 2) and that it ‘is based on methods which, while far from ideal, generate more rigorous, transparent, objective, and reliable tests of competing theoretical claims about European integration than have heretofore been conducted’ (1998: 10). The bulk of studies of EC decision-making are criticised for biased data selection and for relying on ‘citations to secondary sources themselves drawn from journalistic commentary or still other secondary sources’ (1998: 10). In contrast to this, Moravcsik claims to have backed ‘potentially controversial attribution of motive or strategy … by “hard” primary sources (direct evidence of decision-making) rather than “soft” or secondary sources’ (1998: 10, see also pp. 80-84). (2)
2. Agency and structure
The question of how to conceptualise the relationship between agency and structure is arguably one of the most important questions facing social scientists (Archer, 1995: 65). This is due to the importance of agency and structures in the social world and to the fact that it is impossible to offer explanations of events in the social world without appealing to some understanding of their relationship. As mentioned in the introduction there is thus much to be learned about the nature and quality of substantive theories from examining their underlying assumptions with respect to this relationship. Yet the way the latter is dealt with is also important for political reasons, to which we will come back in section 7 below. “Agency” denotes the ability of agents, whether individuals or groups, to act upon situations and it ‘implies a sense of free will, choice or autonomy – that the actor could have behaved differently’ (Hay, 2002: 94). Agency should thus not be confused with concepts like “individuals”, “actors” or “agents”: without anticipating the conclusions of this article too much, a theory can refer to plenty of agents, while not allowing for any agency. “Structure”, on the other hand, refers to the relational context within which agents operate. Structures define the range of options available to agents.
Nowadays the vast majority of scholars agree that both agency and structure matter: phenomena and developments in the social world issue not from either one or the other but are a product of both. If this is the case then it is necessary to break with the two ways of conceptualising the relation between agency and structure that have traditionally been dominant within social theory, namely structuralism and individualism. In their pure versions these positions either picture agents as marionettes (structuralism) or as omnipotent puppet-masters (individualism) (Archer, 1995; XXXXX). However, knowing that both agency and structure matter does not in itself take us far. To make a difference the insight needs to be incorporated into substantive theories and this is by no means an easy task. This contributes to explain why many theories end up offering reductionist explanations of the specific social phenomena they are meant to render intelligible.
In the discipline of International Relations (IR) a debate over the “agent-structure problem” was initiated in the late 1980s by scholars such as Wendt (1987) and Hollis and Smith (1990). Later, and certainly no less interesting contributions to this debate included Doty (1997), Bieler and Morton (2001) and Wight (2006). The debate has done much to clarify and in many cases criticise the ontological and epistemological assumptions underlying mainstream IR theories, particularly Waltzian neorealism (Waltz, 1979). In EU Studies a similar debate has not taken place, and although in particular some constructivist scholars, have taken an interest in the agency-structure relationship (e.g. Wind, 2001), a comprehensive study of the way the most important theories of European integration and governance deal with it has yet to be published. However, it seems to be a widespread view among EU scholars that many of these theories privilege agency over structure. For instance, Risse (2004: 161) writes that the ‘prevailing theories of European integration – whether neofunctionalism, liberal intergovernmentalism, or “multi-level governance” – are firmly committed to a rationalist ontology which is agency-centred by definition’. In a similar vein, other scholars have noticed ‘the ahistorical and structure-blind assumptions’ underlying intergovernmentalism (Hix, 1994: 9) and observed that in LI ‘agents are, implicitly or explicitly, considered primary – actors ultimately determine the shape of overall structures’ (Christiansen, 1998: 103). In the next sections, the validity of this widespread view will be examined through an analysis of the way the agency-structure relationship is dealt with at each of the three stages in LI.
3. National preference formation
The first stage in explaining the outcome of intergovernmental bargains is to account for the national preferences, which are defined as ‘an ordered and weighted set of values placed on future substantive outcomes … that might result from international political interaction’ (Moravcsik, 1998: 24). This is done by means of a liberal political economy theory of preference formation, according to which national preferences arise in the context of domestic politics, where national government leaders form them on the basis of the preferences and actions of the most important societal groups. Most important among these are domestic producers: ‘The systematic political bias in favor of existing producer groups and against those, notably consumers, taxpayers, third-country producers, and also potential future producers, stems from the former’s more intense, certain, and institutionally represented and organized interests’ (1998: 36). The state is conceptualised as ‘a representative institution constantly subject to capture and recapture … by societal groups’ (Moravcsik, 1997: 518). Because governments have an interest in remaining in office, they need the support from coalitions of domestic actors. The policies pursued by governments are ‘therefore constrained by the underlying identities, interests, and power of individuals and groups … who constantly pressure the central decision makers to pursue politics consistent with their preferences’ (ibid: 518). In other words, ‘[g]roups articulate preferences; governments aggregate them’ and it is through this process that ‘the set of national interests or goals that states bring to international negotiations’ emerges (Moravcsik, 1993: 483).
To evaluate the way the agency-structure relationship is dealt with at this stage in LI it is clearly crucial to understand the origins of the preferences of societal groups. Some of the early critics of LI suggested that the theory fails to account adequately for this. For instance, it was pointed out that ‘the origins of such interests are exogenized’ (Risse-Kappen, 1996: 56) while others claimed that in LI ‘interests are not structurally derived’ (Caporaso and Keeler, 1995: 44) and even that they ‘emerge mysteriously’ (McSweeney, 1998: 101). Had it in fact been the case that LI leaves completely open the question of where the preferences of societal groups come from it would have allowed for an agency-centred perspective on preference formation. That is, preferences could have been formed on the basis of all sorts of ideas and individual inclinations. However, this would have seriously undermined the parsimony and explanatory power of the theory and hence it was in fact never left open where preferences come from. As Moravcsik has made clear, LI perceives preferences to be directly caused by structural circumstances, more precisely economic structures:
‘I employ a structural theory of those preferences. My structural approach…employs trade flows, competitiveness, inflation rates, and other data to predict what the economic preferences of societal actors – and therefore governments – should be’ (Moravcsik, 1999b: 377).
In other words, economic preferences are derived from economic structures: societal groups organise and articulate their preferences ‘on the basis of calculations of net expected costs and benefits resulting from the introduction of new policies’ (Moravcsik, 1993: 489). It follows as a logical implication that ‘shifts in preferences should follow the onset and precede the resolution of shifts or trends in economic circumstances’ (Moravcsik, 1998: 50).
The assumption that economic structures translate directly into specific preferences is made possible by the rationality assumption underpinning LI. The widespread view that LI is agency-centred is related to this assumption that individuals, groups, governments and even states are rational. This begs the question of what “rationalism” precisely entails, especially as some scholars have suggested that Moravcsik fails to spell this clearly out (Christiansen et al., 2001: 4). In a recent piece Moravcsik and Schimmelfennig (2009: 68) put it as follows:
‘Rationalism is an individualist or agency assumption. Actors calculate the alternative courses of action and choose the one that maximizes (or satisfies) their utility under the circumstances. Collective outcomes are explained as the result of aggregated individual actions based on efficient pursuit – albeit subject to the information at hand and uncertainty about the future’.
Despite the qualifications at the end of the quote it is clear that whatever this uncertainty pertains to it is not to the consequences of the actions of agents: agents are assumed to be very well-informed about these because, as Moravcsik has put it himself, in ‘a world in which the future consequences of actions are unknown … LI would make little sense’ (1995: 626). This is an important manifestation, because the more it is assumed that agents know the future consequences of their actions, the more it must also be assumed that they are fully informed about the context in which they currently find themselves. It is quite simply logically inconceivable that an agent can somehow know the future consequences of his or her actions without having perfect or very close to perfect information at hand at the moment of the action itself. Moravcsik is thus significantly underplaying the strength of his rationality assumption when stating that ‘it takes no position on whether states are fully informed, though a framework in which states are assumed to be informed generally performs well’ (1998: 23). Why not walk the plank? Surely states and other agents can safely be assumed to be blessed full information if it has already been established that no or very few unintended consequences will follow from their actions?
At the end of the day the rationality assumption boils down to the view that agents are utility-maximisers with clearly ordered preferences who are (almost?) fully informed, also about the future consequences of actions. However, it should not be concluded from this that LI is an agency-centred theory as the conventional wisdom has it. As we have seen above, preferences are derived from economic structures – not just in the weak sense that structures are important in relation to preferences but in the strong sense that they alone dictate preferences (albeit with a minor qualification to which we will return in a moment). Because the rational agents are assumed to be so well-informed their actions become predictable once their structural environment has been mapped. Indeed, only structures matter here inasmuch as ‘[p]references are by definition causally independent of the strategies of other actors’ (Moravcsik, 1997: 519, see also 1998: 24-25).
Moreover, ideas are for the most part not allowed to play any role in relation to preference formation. It is worth dwelling on this for a moment. On one hand, Moravcsik does not hesitate to acknowledge the importance of ideas, as when he proclaims that they ‘…are like oxygen or language; it is essentially impossible for humans to function without them’ (Moravcsik, 2001: 229). On the other hand, ideas do not play a very prominent role in LI, which is also recognized by Moravcsik when he writes that ‘[i]n the LI account of integration, ideas are present but not causally central. They may be irrelevant or random, or, more likely, they are “transmission belts” for interests’ (Moravcsik, 2001: 229). The only reason why Moravcsik can correctly maintain that in LI ‘[s]ome national preferences … are grounded in ideas’ (1998: 23) is because some importance is ascribed to the latter in issue areas where the material consequences of policy initiatives are more or less impossible to calculate. For instance, he mentions ‘questions of European institutions and common foreign policy’ as issues where governments/states will generally not to be under strong pressure from societal groups to pursue particular policies, which creates some room for government leaders to act on the basis of ‘ideologies and personal commitments’ (Moravcsik, 1993: 494; see also Moravcsik and Schimmelfennig, 2009: 85).
According to Moravcsik and Schimmelfennig (2009: 76), ‘LI best explains policy-making in issue areas where social preferences are relatively certain and well defined’. In the core areas, like trade, agriculture and monetary policy, ideas are not assumed to influence preference formation at all. When it comes to ‘insignificant, exceptional and speculative issues’ like those mentioned above or the Open Method of Coordination (ibid.: 85) where the preferences of societal groups are less clear and strong, and where the explanatory power of LI is thus recognised to be limited, ideas are conveniently allowed to play a role. To recapitulate, in LI no importance is ascribed to ideas in the explanation of what is (correctly) considered to be the ‘substantively important issues’ (ibid.: 85) in the European integration process: here economic structures do the job alone. (3)
4. International bargains
Once the national preferences have been formulated, national decision-makers bring them to the intergovernmental bargaining table. At this second stage LI applies an intergovernmentalist bargaining theory in order to explain the outcome of negotiations. As the primary interest of the governments is to remain in office, they have a clear incentive to defend the national interest in the negotiation. Accordingly, ‘[t]he configuration of domestically determined national preferences defines a “bargaining space” of potentially viable agreements’ (Moravcsik, 1993: 496-497). The outcome of a concrete negotiation, however, not only reflects the different national preferences but also the relative bargaining power of different states. Moravcsik defines power in terms of asymmetric interdependence: ‘Bargaining leverage stems most fundamentally from asymmetries in the relative intensity of national preferences, which reflect … the relative costs of agreements to remove negative externalities’ (ibid.: 1993: 499). This means that ‘[t]he power of each government is inversely proportional to the relative value that it places on an agreement’ (Moravcsik, 1998: 62).
How, then, is the agency-structure relationship dealt with at this second stage? Or to put it differently: how much freedom does government leaders have to pursue their own preferences or ideas (agency) and how much are they constrained by their context (structure)? First, the answer to this question depends on the issue area. As mentioned, governments are severely constrained by domestic societal groups in core areas: here they can only act within a narrow “bargaining space” which limits their freedom considerably. In more marginal (non-economic) issue areas this space widens and government leaders enjoy more freedom to pursue their own agendas. Second, any particular government is constrained by the bargaining spaces of other governments. The nature of these determines the extent to which a government leader is capable of realising national interests. Finally, the outcomes of previous bargains serve as the status quo ‘with respect to which societal actors and governments calculate preferences and alternatives to agreement’ (Moravcsik, 1995: 612).
As mentioned above, the outcome of a concrete bargain reflects the relative bargaining power of each state. As bargaining power is defined in terms of asymmetric interdependence it is, in fact, derived from the very same structures as national preferences. These structures determine how attractive a potential policy is to societal groups – and thus governments – and consequently they also determine the relative bargaining power. Relative power is thus ultimately decided at the structural level – not at the level of agents. This brings us back to the point that was raised in the previous section, namely that the rationality assumption underpinning LI does not serve to render it an agency-centred theory. To be sure, there are plenty of agents in LI, and there is no denying that the theory belongs to the tradition of “methodological individualism”. But by substituting real agents with ‘calculating machines who always know what they want and are never uncertain about the future and even their own stakes and interests’ (Risse, 2009: 147), LI effectively ends up with no notion of agency at all, at least not in its account of integration in core issue areas.
That it is apparently unnecessary to study the interaction between state representatives in order to explain the outcome of a bargain tells it all: the creativity, charisma, persuasiveness and negotiating abilities of particular agents are insignificant in LI. By assuming that agents are identical in the sense of being rational it is possible to derive the outcome of bargains simply by looking at the context in which it takes place. This makes LI a structuralist theory also in its second stage. To be sure, the structuralism of LI differs from conventional structuralism inasmuch as the former retains a focus on agents and their free choices. But the point is that the “methodological individualism” of LI and other rational choice theories does not entail a genuine notion of agency in that a free choice is neither free nor, indeed, a real choice, if it is always already given by the context in which the agent operates (see also Hay, 2002: 103-104; Tsebelis, 1990: 40). (4)
5. Institutional choice
Once governments have reached substantive agreement in a bargain, they set up institutional arrangements in order to secure it. At this third stage LI adopts a functional theory of institutional choice according to which governments pool or delegate authority in order to ‘constrain and control one another’ (Moravcsik, 1998: 9). Authority is “pooled” when governments for instance agree to take decisions in an issue area by means of qualified majority voting in the Council, whereas “delegation” refers to the transfer of authority to more or less autonomous supranational institutions (ibid.: 1998: 67). Pooling and delegation are ‘viewed as solutions to the problem of “incomplete contracting,” which arises when member governments share broad goals but find it too costly or technically impossible to specify all future contingencies involved in legislating or enforcing those goals’ (ibid.: 1998: 73).
By pooling or delegating, the credibility of the commitment to the substantive agreement that has been reached is enhanced. But by giving up authority in an issue area governments clearly run the risk of being either outvoted by other governments (pooling) or of being overruled by supranational institutions (delegation) in future cases. Hence, ‘[t]he specific level of pooling or delegation reflects a reciprocal cost-benefit analysis: governments renounce unilateral options in order to assure that all governments will coordinate their behavior in particular ways’ (ibid.: 1998: 75). LI predicts that pooling and delegation will vary across issues and countries. Again, the preferences of societal groups are crucial: ‘Governments transfer sovereignty to commit other governments to accept policies favored by key domestic constituencies’ (ibid.: 1998: 76). As accounted for above the preferences of societal groups are seen as structurally determined, at least in the core issue areas. In the end the governments’ institutional choices thus become rather mechanical, following more or less automatically from the circumstances in which they are made.
According to LI, international institutions are ‘passive, transaction-cost reducing sets of rules’ (Moravcsik, 1993: 508) that for instance serve to provide states with information ‘to reduce the states’ uncertainty about each other’s future preferences and behaviour’ (Moravcsik and Schimmelfennig, 2009: 72). Somewhat surprisingly, Moravcsik and Schimmelfennig link the existence of such institutions to “unanticipated consequences” of actions, proclaiming that LI also assumes the existence of the latter:
‘If unanticipated consequences did not exist, there would be no need for international institutions to elaborate “incomplete contracts” to begin with’. The reason for institutions is precisely to elaborate agreements and credibly lock in compliance against defection by future unsatisfied governments’ (2009: 75).
This, to be sure, is a somewhat unorthodox and problematic use of the concept. When, for instance, historical institutionalists are talking about unintended or unanticipated consequences in the context of European integration, their argument is that supranational institutions and policies tend to develop in ways not originally envisaged and subsequently not approved of by member state governments (Pierson, 1996). Due to “path dependency” and other mechanisms such institutions and the course of the integration process can become impossible for governments to control.
Moravcsik and Schimmelfennig clearly have something altogether different in mind when they talk about unanticipated consequences. In fact, what they are talking about can more accurately be denoted ‘anticipated but undesired outcomes’. These arise when rational governments anticipate that there is a risk that other rational governments will not comply with the substantive agreement that has been reached in an intergovernmental bargain. To avoid this undesired outcome governments agree on an institutional arrangement to create certainty. On this view, institutions (being ‘passive, transaction-cost reducing sets of rules’) only contribute to minimise uncertainty by eliminating the risk of undesired outcomes (see also Moravcsik and Schimmelfennig, 2009: 72) – it is unthinkable that they can develop and behave in ways not intended by governments. At the end of the day, it is not unintended consequences that LI assumes the existence of but rather the ability of governments to very accurately predict the consequences of their substantive agreement and on this basis chose the most suitable institutional agreements. This dubious assumption can obviously only be made if it is held, as LI does, that agents are blessed with more or less perfect information, also of future outcomes of their actions (see also Pierson, 2004: 115-XXX).
6. What choice for Europe?
‘the motivations and coalitions underlying national preferences in specific decisions reflected the economic interest of sectors as predicted by their structural position in global markets … any feedback must take the form of changes in economic structures, not ideas’ (Moravcsik, 1999b: 382)
The liberal intergovernmentalist explanation of regional integration ultimately comes down to developments in economic structures and it is therefore logical and appropriate that Moravcsik (1998: 501) refers to it as a ‘structural perspective’. Although the concept of “economic structures” is not defined as clearly as other LI concepts, it basically appears to denote the phenomena that economic indicators are expressions of, examples being trade flows, inflation rates, wealth and competitiveness. The method is thus to use economic indicators as expressions of the economic structures determining the preferences of agents. For instance it reads that
‘Taken together, capital mobility, trade flows, and inflationary convergence provide a prima facie explanation of the progressive shift in national preferences away from, then back toward, exchange-rate cooperation over the two decades following the collapse of the Bretton Woods. This period saw an increase in economic openness and, beginning in the late 1970s, convergence toward low inflation’ (Moravcsik, 1998: 48)
But what caused this and other shifts in economic circumstances? What explains the timing of such shifts? Considering the enormous importance it ascribes to economic structures it would be completely reasonable to expect LI to address and tentatively provide a theoretical answer to such questions. But it doesn’t! In all fairness, Moravcsik