6 VIVIEN A.SCHMIDT represent the incentive structures that reduce the uncertainties resulting from the multiplicity of individual preferences and issues(see e.g.,Hardin,1982;Ostrom, 1990).This approach has arguably been the most resistant to the turn to ideas. The ways in which DI deals with questions of interests and institutions challenges the very premises of RI about the fixed nature of preferences and the stability of institutions.Preference formation,moreover,which is at the center of DI con- cerns,was until very recently of little interest to RI scholars,while institutional change was ruled out analytically a priori.But critiques of RI-emphasizing its economic determinism,its inability to deal with institutional change endogenously, and its difficulties in accounting for preference formation (see e.g.,Mansbridge, 1990;Green and Shapiro,1994;Scharpf,1997)-have led scholars in the RI tradition in recent years to seek to 'endogenize'change. In their attempts to find new ways to account for preference formation and institutional shifts,RI scholars have mostly looked to political coalition forma- tion,although they have sometimes even sought a rapprochement with HI(e.g., Katznelson and Weingast,2005).For example,Iversen and Soskice(2006)argue that formal electoral institutions act as incentive structures that produce political coalitions which implement certain kinds of socio-economic policies,more ine- galitarian in majoritarian systems,less inegalitarian in proportional ones.The problem with such an approach is that it remains highly deterministic,as insti- tutions determine politics which determine political outcomes,and it still can't explain the origins of the institutions or of the political coalitions that created them,let alone why they might change institutions or policies over time.In contrast,Grief and Laitin (2004),in seeking to build HI insights into game- theoretic analysis,redefine the goal of institutions-from 'self-enforcing'to self- reinforcing or self-undermining institutions-and their effects-as 'parametric'in the short term,meaning exogenous and fixed for agents who act on self-enforcing beliefs,but only 'quasi-parametric'in the long run,meaning endogenous and variable as individuals are led to act in a manner that does not reproduce the associated beliefs.The problem here is that although this may better account for change over time in game-theoretic terms,we are still left with the irrationality of the choice of institutions to begin with;the deterministic trajectory of change over time,now for better or worse;and the limited rationality of these supposedly rational'actors at any given point in time. Relatively few RI scholars have turned to ideas to solve the problem of insti- tutional change.Among those who have,the most significant RI engagement with ideas began in the 1990s,although it has remained rather circumscribed. Goldstein (1993)and Goldstein and Keohane (1993)provide the classic exposi- tion of the RI approach to ideas(see also Weingast,1995).They deem ideational explanation useful only when and if explanation in terms of 'objective'or material'interests is insufficient,which may occur in one of three ways,each of which is problematic in a different way.First,ideas may come before interests, acting as 'road maps'for individual actors to clarify their goals or limit the range
represent the incentive structures that reduce the uncertainties resulting from the multiplicity of individual preferences and issues (see e.g., Hardin, 1982; Ostrom, 1990). This approach has arguably been the most resistant to the turn to ideas. The ways in which DI deals with questions of interests and institutions challenges the very premises of RI about the fixed nature of preferences and the stability of institutions. Preference formation, moreover, which is at the center of DI concerns, was until very recently of little interest to RI scholars, while institutional change was ruled out analytically a priori. But critiques of RI – emphasizing its economic determinism, its inability to deal with institutional change endogenously, and its difficulties in accounting for preference formation (see e.g., Mansbridge, 1990; Green and Shapiro, 1994; Scharpf, 1997) – have led scholars in the RI tradition in recent years to seek to ‘endogenize’ change. In their attempts to find new ways to account for preference formation and institutional shifts, RI scholars have mostly looked to political coalition formation, although they have sometimes even sought a rapprochement with HI (e.g., Katznelson and Weingast, 2005). For example, Iversen and Soskice (2006) argue that formal electoral institutions act as incentive structures that produce political coalitions which implement certain kinds of socio-economic policies, more inegalitarian in majoritarian systems, less inegalitarian in proportional ones. The problem with such an approach is that it remains highly deterministic, as institutions determine politics which determine political outcomes, and it still can’t explain the origins of the institutions or of the political coalitions that created them, let alone why they might change institutions or policies over time. In contrast, Grief and Laitin (2004), in seeking to build HI insights into gametheoretic analysis, redefine the goal of institutions – from ‘self-enforcing’ to selfreinforcing or self-undermining institutions – and their effects – as ‘parametric’ in the short term, meaning exogenous and fixed for agents who act on self-enforcing beliefs, but only ‘quasi-parametric’ in the long run, meaning endogenous and variable as individuals are led to act in a manner that does not reproduce the associated beliefs. The problem here is that although this may better account for change over time in game-theoretic terms, we are still left with the irrationality of the choice of institutions to begin with; the deterministic trajectory of change over time, now for better or worse; and the limited rationality of these supposedly ‘rational’ actors at any given point in time. Relatively few RI scholars have turned to ideas to solve the problem of institutional change. Among those who have, the most significant RI engagement with ideas began in the 1990s, although it has remained rather circumscribed. Goldstein (1993) and Goldstein and Keohane (1993) provide the classic exposition of the RI approach to ideas (see also Weingast, 1995). They deem ideational explanation useful only when and if explanation in terms of ‘objective’ or ‘material’ interests is insufficient, which may occur in one of three ways, each of which is problematic in a different way. First, ideas may come before interests, acting as ‘road maps’ for individual actors to clarify their goals or limit the range 6 VIVIEN A . SCHMIDT
Taking ideas and discourse seriously 7 of strategies to be taken-in which case ideas seem to determine interests,but we have no explanation of the selection mechanism by which certain ideas get chosen over others (Yee,1997:1024;Blyth,2002:16;Gofas and Hay,2010).Second, ideas may come after interests,acting as 'focal points'for actors to choose among equally acceptable alternatives (i.e.,multiple Pareto-improving equilibria)-in which case ideas serve at best to 'mop up'residual variance,and we still can't explain the mechanism by which the now exogenously,interest-determined ideas are picked(see Yee,1997:1025-1027;Blyth,2002:26;Gofas and Hay,2010). Third,ideas may be embedded in institutions,in which case it is the institutions rather than the ideas that really matter to the actors(Yee,1996;Gofas and Hay, 2010).A fourth-way RI scholars see ideas coming in is as after-the-fact legit- imation of actors'interest-based action,following an instrumental logic,or as 'hooks'for elite interests (e.g.,Shepsle,1985)-in which case ideas are not really taken seriously at all. In all of these approaches,then,ideas have not gone very far beyond interests, since they are little more than mechanisms for choosing among interests,focal points for switching among equilibria(see critique by Ruggie,1998:866-867),or after-the-fact justification for interest-based choices.Douglas North(1990)went farther,first by using ideas to overcome the problem of how to explain institu- tional construction,then by casting ideas as 'shared mental modes'.However,as Blyth(2003:696-697,2002:Ch.2)argues,the contradictions inherent in both such approaches may have been 'a bridge too far'.First,if ideas create institu- tions,then how can institutions make ideas 'actionable'?But second,if instead ideas are 'mental modes',then what stops ideas from having an effect on the content of interests,and not just on the order of interests,which means that ideas would constitute interests,rather than the other way around. The problem for RI scholars,then,and the reason most of them quickly abandoned the pursuit of ideas,is that they could not continue to maintain the artificial separation of 'objective'interests from 'subjective'ideas about interests, that is,beliefs and desires.Such subjective interests threatened to overwhelm the objective ones which are at the basis of the rationalists'thin model of rationality, by undermining the 'fixed'nature of preferences and the notion of outcomes as a function of pre-existing preferences.And without fixed preferences as well as neutral institutional incentive structures,RI scholars lose the parsimony of the approach and everything that follows from it,including the ability to mathema- tically model games rational actors play as opposed to those 'real actors play' (see Scharpf,1997;Rothstein,2005:Ch.1).This helps explain why the foray into ideas for most dyed-in-the-wool RI scholars was short-lived.For those who persisted,however,a whole new approach to the explanation of interests and institutions has opened up. For DI scholars engaged with the RI tradition,subjective interests replace the objective ones of RI,as ideas about interests that bring in a much wider range of strategic ideas and social norms that must be explained in terms of their meaning
of strategies to be taken – in which case ideas seem to determine interests, but we have no explanation of the selection mechanism by which certain ideas get chosen over others (Yee, 1997: 1024; Blyth, 2002: 16; Gofas and Hay, 2010). Second, ideas may come after interests, acting as ‘focal points’ for actors to choose among equally acceptable alternatives (i.e., multiple Pareto-improving equilibria) – in which case ideas serve at best to ‘mop up’ residual variance, and we still can’t explain the mechanism by which the now exogenously, interest-determined ideas are picked (see Yee, 1997: 1025–1027; Blyth, 2002: 26; Gofas and Hay, 2010). Third, ideas may be embedded in institutions, in which case it is the institutions rather than the ideas that really matter to the actors (Yee, 1996; Gofas and Hay, 2010). A fourth-way RI scholars see ideas coming in is as after-the-fact legitimation of actors’ interest-based action, following an instrumental logic, or as ‘hooks’ for elite interests (e.g., Shepsle, 1985) – in which case ideas are not really taken seriously at all. In all of these approaches, then, ideas have not gone very far beyond interests, since they are little more than mechanisms for choosing among interests, focal points for switching among equilibria (see critique by Ruggie, 1998: 866–867), or after-the-fact justification for interest-based choices. Douglas North (1990) went farther, first by using ideas to overcome the problem of how to explain institutional construction, then by casting ideas as ‘shared mental modes’. However, as Blyth (2003: 696–697, 2002: Ch. 2) argues, the contradictions inherent in both such approaches may have been ‘a bridge too far’. First, if ideas create institutions, then how can institutions make ideas ‘actionable’? But second, if instead ideas are ‘mental modes’, then what stops ideas from having an effect on the content of interests, and not just on the order of interests, which means that ideas would constitute interests, rather than the other way around. The problem for RI scholars, then, and the reason most of them quickly abandoned the pursuit of ideas, is that they could not continue to maintain the artificial separation of ‘objective’ interests from ‘subjective’ ideas about interests, that is, beliefs and desires. Such subjective interests threatened to overwhelm the objective ones which are at the basis of the rationalists’ thin model of rationality, by undermining the ‘fixed’ nature of preferences and the notion of outcomes as a function of pre-existing preferences. And without fixed preferences as well as neutral institutional incentive structures, RI scholars lose the parsimony of the approach and everything that follows from it, including the ability to mathematically model games rational actors play as opposed to those ‘real actors play’ (see Scharpf, 1997; Rothstein, 2005: Ch. 1). This helps explain why the foray into ideas for most dyed-in-the-wool RI scholars was short-lived. For those who persisted, however, a whole new approach to the explanation of interests and institutions has opened up. For DI scholars engaged with the RI tradition, subjective interests replace the objective ones of RI, as ideas about interests that bring in a much wider range of strategic ideas and social norms that must be explained in terms of their meaning Taking ideas and discourse seriously 7
8 VIVIEN A.SCHMIDT to the actors within a given'meaning context'rather than in terms of some set of universally identifiable interests.Material interests,economic in particular,which are at the basis of much of the institutional incentives in the rational choice institutionalist literature,are not ignored.But in discursive institutionalism, scholars tend to separate material interests analytically into material reality and interests rather than to conflate them,such that material reality constitutes the setting within which or in response to which agents may conceive of their interests (see Schmidt,2008). The kind of knowledge and degree of certainty agents may have with regard to their ideas about material reality may also differ,depending upon the aspect of material reality with which they are concerned.Illustrative of this epistemological observation is Wittgenstein's(1972)little-noticed distinction between the language- games based on our everyday experiences in the world,which tend to admit of few doubts or mistakes,and language-games based on our (social)scientific pictures of the world,which may always allow for doubts,mistakes,and even gestalt switches (see Schmidt,2008).The problem with RI is that it tends to develop(social)scientific pictures-games of the world that it treats as if they had the certainty of experience- games-a point also made by Taleb and Pilpel(2005;see discussion in Blyth,2010), who demonstrate that the world in which we live is a lot more uncertain than the world of risk economists and rational choice institutionalists generally assume,given the impossibility of knowing let alone statistically predicting the effects of all the forces that may have an impact on economic and political realities. As Blyth(2002:31-39,2010;see also Schmidt,2008)argues,the problem with RI is that it assumes that most phenomena are explainable in terms of 'Knightian risk',because they are part of a directly observable world that agents can perceive more or less well and in which they can calculate the subjective probability of the likely outcomes of their preferences,such as in the US Congress.Such phenomena, Blyth shows,are in actuality better explained in terms of 'Knightian uncertainty, because they are part of a world that is not directly observable,such as the global economy,in which agents are not simply unsure about how to achieve their interests but unsure of what their interests are,given that the uncertainties are too great,the moment unique,prediction impossible,and agents'interests always structurally underdetermined.Blyth (2002)illustrates this by noting that the Great Depression in the 1930s and the economic crisis following the oil shocks of the 1970s were cases of Knightian uncertainty for both the United States and Sweden,as governments,business,and labor at these critical junctures all sought to reconstitute interests through alternative 'narratives'and 'causal stories'about the reasons for the crisis,seeking to produce new political coalitions for collective action,with ideas serving as 'weapons'and 'blueprints'in the struggle to replace existing institutions. Woll (2008)demonstrates that there are also less dramatic instances of Knightian uncertainty,such as the kind faced by multinational businesses lobbying for the deregulation of international trade in services.Here,firms'ideas about which
to the actors within a given ‘meaning context’ rather than in terms of some set of universally identifiable interests. Material interests, economic in particular, which are at the basis of much of the institutional incentives in the rational choice institutionalist literature, are not ignored. But in discursive institutionalism, scholars tend to separate material interests analytically into material reality and interests rather than to conflate them, such that material reality constitutes the setting within which or in response to which agents may conceive of their interests (see Schmidt, 2008). The kind of knowledge and degree of certainty agents may have with regard to their ideas about material reality may also differ, depending upon the aspect of material reality with which they are concerned. Illustrative of this epistemological observation is Wittgenstein’s (1972) little-noticed distinction between the languagegames based on our everyday experiences in the world, which tend to admit of few doubts or mistakes, and language-games based on our (social) scientific pictures of the world, which may always allow for doubts, mistakes, and even gestalt switches (see Schmidt, 2008). The problem with RI is that it tends to develop (social) scientific pictures-games of the world that it treats as if they had the certainty of experiencegames – a point also made by Taleb and Pilpel (2005; see discussion in Blyth, 2010), who demonstrate that the world in which we live is a lot more uncertain than the world of risk economists and rational choice institutionalists generally assume, given the impossibility of knowing let alone statistically predicting the effects of all the forces that may have an impact on economic and political realities. As Blyth (2002: 31–39, 2010; see also Schmidt, 2008) argues, the problem with RI is that it assumes that most phenomena are explainable in terms of ‘Knightian risk’, because they are part of a directly observable world that agents can perceive more or less well and in which they can calculate the subjective probability of the likely outcomes of their preferences, such as in the US Congress. Such phenomena, Blyth shows, are in actuality better explained in terms of ‘Knightian uncertainty’, because they are part of a world that is not directly observable, such as the global economy, in which agents are not simply unsure about how to achieve their interests but unsure of what their interests are, given that the uncertainties are too great, the moment unique, prediction impossible, and agents’ interests always structurally underdetermined. Blyth (2002) illustrates this by noting that the Great Depression in the 1930s and the economic crisis following the oil shocks of the 1970s were cases of Knightian uncertainty for both the United States and Sweden, as governments, business, and labor at these critical junctures all sought to reconstitute interests through alternative ‘narratives’ and ‘causal stories’ about the reasons for the crisis, seeking to produce new political coalitions for collective action, with ideas serving as ‘weapons’ and ‘blueprints’ in the struggle to replace existing institutions. Woll (2008) demonstrates that there are also less dramatic instances of Knightian uncertainty, such as the kind faced by multinational businesses lobbying for the deregulation of international trade in services. Here, firms’ ideas about which 8 VIVIEN A . SCHMIDT
Taking ideas and discourse seriously 9 utility to maximize (interests),how to maximize it (strategies),and to what end (goals)explains not only changes in their identity,as they moved from seeing themselves as national champions to global players,but also changes in institutions, as trade in services was developed in a context of single-level(US)games spurred by business and multi-level (EU)games in which global liberalization was linked to attempts to spur business support for liberalization in the member states.But Woll (2008:161),as she herself notes,reifies a single corporate rationality for the sake of parsimony,and thus misses out on the negotiation processes inside firms as well as between them that make for a much more complex ideational construction of interests than the stylized one she developed. Culpepper(2005)shows how one might zoom in to consider such ideational processes in detail,in the institutional changes(or continuities)of the corporate governance systems of France and Germany in the 1990s.In the case of France,he argues that CEOs underwent a joint belief shift at a critical juncture,when a central figure in the system of cross-shareholding exited,thereby 'signalling'an idea that led to a shift in belief system when other French CEOs followed suit after an exchange of ideas.German CEOs,by contrast,did not change the rules of the game at their own critical juncture,when a major firm was taken over by a foreign company,because the outsider nature of the takeover had little effect on the CEOs'belief system. Culpepper's approach has the advantage for RI of helping to maintain the possibility of game-theoretic analysis,by positing an equilibrium state with fixed preferences and stable institutions before and after the period of ideational change.The disadvantage is that it formalizes an empirical process that was,in reality,a lot more messy and uncertain than the discussion suggests(see Schmidt, 1997,2002:Ch.6),making 'revolutionary'a process of change that was a lot more incremental.French CEOs had already voiced dissatisfaction with the sys- tem prior to the central figure's precipitating action while the Germans were a lot more dissatisfied with the system,as was evident from their partial defection once the rules were changed with regard to capital gains tax in 2002(Schmidt,2002: Ch.6;Kinderman,2005).Moreover,by limiting the importance of ideas to the period of uncertainty between the end of the old institutional 'game'and the beginning of the newly agreed institutional 'game',we have changing DI ideas within a critical moment preceded and followed by RI crystallized preferences and frozen institutions. Institutions,however,as Rothstein(2005)shows,also change over time as the ideas that infuse them change.This is because,as he argues,institutions them- selves should not be treated as neutral structures of incentives but,rather,as the carriers of ideas or 'collective memories'which make them objects of trust or mistrust and changeable over time as actors'ideas and discourse about them change in tandem with changes in their performance.Rothstein(2005:168-198) illustrates this with the case of the long-term survival of the institutions of Sweden's collective bargaining system,which became the carriers of ideas or 'collective
utility to maximize (interests), how to maximize it (strategies), and to what end (goals) explains not only changes in their identity, as they moved from seeing themselves as national champions to global players, but also changes in institutions, as trade in services was developed in a context of single-level (US) games spurred by business and multi-level (EU) games in which global liberalization was linked to attempts to spur business support for liberalization in the member states. But Woll (2008: 161), as she herself notes, reifies a single corporate rationality for the sake of parsimony, and thus misses out on the negotiation processes inside firms as well as between them that make for a much more complex ideational construction of interests than the stylized one she developed. Culpepper (2005) shows how one might zoom in to consider such ideational processes in detail, in the institutional changes (or continuities) of the corporate governance systems of France and Germany in the 1990s. In the case of France, he argues that CEOs underwent a joint belief shift at a critical juncture, when a central figure in the system of cross-shareholding exited, thereby ‘signalling’ an idea that led to a shift in belief system when other French CEOs followed suit after an exchange of ideas. German CEOs, by contrast, did not change the rules of the game at their own critical juncture, when a major firm was taken over by a foreign company, because the outsider nature of the takeover had little effect on the CEOs’ belief system. Culpepper’s approach has the advantage for RI of helping to maintain the possibility of game-theoretic analysis, by positing an equilibrium state with fixed preferences and stable institutions before and after the period of ideational change. The disadvantage is that it formalizes an empirical process that was, in reality, a lot more messy and uncertain than the discussion suggests (see Schmidt, 1997, 2002: Ch. 6), making ‘revolutionary’ a process of change that was a lot more incremental. French CEOs had already voiced dissatisfaction with the system prior to the central figure’s precipitating action while the Germans were a lot more dissatisfied with the system, as was evident from their partial defection once the rules were changed with regard to capital gains tax in 2002 (Schmidt, 2002: Ch. 6; Kinderman, 2005). Moreover, by limiting the importance of ideas to the period of uncertainty between the end of the old institutional ‘game’ and the beginning of the newly agreed institutional ‘game’, we have changing DI ideas within a critical moment preceded and followed by RI crystallized preferences and frozen institutions. Institutions, however, as Rothstein (2005) shows, also change over time as the ideas that infuse them change. This is because, as he argues, institutions themselves should not be treated as neutral structures of incentives but, rather, as the carriers of ideas or ‘collective memories’ which make them objects of trust or mistrust and changeable over time as actors’ ideas and discourse about them change in tandem with changes in their performance. Rothstein (2005: 168–198) illustrates this with the case of the long-term survival of the institutions of Sweden’s collective bargaining system, which became the carriers of ideas or ‘collective Taking ideas and discourse seriously 9