I welcome the opportunity to respond to Vivien Schmidt's thoughtful views on my article.Footnote 1 This is an important debate about institutional theory and how we explain institutional change. Yet, foregrounding agents and their ideas in such change processes, as Schmidt does, whilst leaving institutions (and wider structures) almost as residuals – as merely ‘meaning contexts’ – leaves too much out of the explanatory equation, in my view. I am sure we agree that agents, ideas and institutions all matter. The goal, then, should not be to privilege certain of these elements theoretically, but to theorize and study their interactions empirically – indeed, their mutually shaping or dialectical interactions over time. This is what drives institutional change.

I will return to these central theoretical questions presently, but first I will respond briefly to Schmidt's claims that I have misread the work of the initial targets of critique in my article. I invite interested readers to make their own assessments of my reading of Colin Hay and Mark Blyth. In relation to Schmidt's work, I am accused of slicing and dicing her ‘complex set of arguments’, but in reality my article uses extensive quotes from her work to establish that she foregrounds agents and their ideas and relegates institutions to a residual, to entities which are derived from the ideas of agents. As she herself explains, institutions become ideational constructions, as arenas that ‘frame the discourse’.Footnote 2 ‘The “institutionalism” in discursive institutionalism’, she writes, is not about ‘the external rule following structures of the three older institutionalisms’, but is instead reduced to the ‘constructs of meaning which are internal to “sentient” (thinking and speaking) agents’.Footnote 3 For Schmidt, institutions appear as a ‘meaning context’, as providing ‘background information’, or as ‘contingent (the result of agent's thoughts, words and actions)’.Footnote 4 Schmidt also argues that the ‘deliberate nature of discourse, allows agents to ‘conceive of and talk about institutions as objects at a distance, and to dissociate themselves from them’.Footnote 5 In more recent work, Schmidt argues that constructivist institutionalists ‘insist that institutions are ideas.’Footnote 6

As I argue in my article, the constructivists in question oscillate between several varieties of extant constructivism, varying from post-modern accounts, where ideas, inter-subjective meanings and discourse are primitive and wholly define or constitute social and institutional life, to more ontologically realist accounts, which admit that institutions and wider structures can have real effects.Footnote 7 Here, institutions and wider structures take on a congealed reality of their own and cannot simply be reduced to real-time sets of ideas or inter-subjective meanings held by agents, as I elaborate below. On the one hand, when Schmidt states that institutions are ideas, she is in post-modern mode. On the other hand, she is in a more realist mode when she worries that her approach ‘risks appearing highly voluntaristic unless structural constraints derived from the three older institutionalisms are included’.Footnote 8 In her response, Schmidt states that she ‘precedes’ her account of ‘the politics of ideas and discourse’ with a historical institutionalist (HI) account of ‘crisis/driven and/or incremental changes in rules and regularities’.Footnote 9 Again, the latter appears more realist. In any case, and in my view, a historical institutionalist analysis should not precede but should be fully integrated with a suitably tailored constructivist analysis of how agents use ideas in institutional settings and how institutions shape agency.

Schmidt responds that I think that the trio of constructivists I critique ‘understate the flexibility of the institutions developed in existing institutionalist scholarship’. Yes, I do. The constructivist institutionalists in question correctly critique what I call the ‘sticky’ version of HI theory, of the kind that focuses on heavy institutional constraint, a limited account of agency, crisis as a key mode of change and constraining notions of change such as path dependency. I discuss the work of Krasner, Pierson and other sticky HI theorists in this regard in my article. The constructivists have devised their own escape route here with their emphasis on agency and ideational construction. But, as I point out in detail in my article, the constructivists in question completely overlook a second school of HI scholars that stretch way back at least to the early 1990s, who have emphasized a degree of institutional flexibility, allowing agents scope for change and innovation, either during crisis moments or through ongoing incremental change processes. Schmidt is incorrect to say that these more flexible approaches only came on stream from about 2005. True, such ideas became more pronounced then, especially the agency-driven models of incremental change invoked by Kathleen Thelen and others.Footnote 10 Yet as far back as 1992, Steinmo and Thelen were emphasizing the role of strategic actors in shaping and being shaped by institutions, as I point out in detail in my article.Footnote 11 So, it is simply not true that ‘historical institutionalism on its own has little room for agency’, as Schmidt has recently claimed.Footnote 12 Early HI scholars in this mould were emphasizing agency and were even talking about the role of ideas in shaping the interests and preferences of agents. This discussion is quite explicit in Steinmo and Thelen's Reference Thelen, Steinmo and Longstreth1992 discussion.Footnote 13 Other early HI scholars also theorized the role of ideas in a similar fashion, as in Virginia Hattam's work in Reference Hattam1993, when she added an ‘interpretive leg’ to her analysis by insisting that ‘only by considering the meaning and significance that workers brought to their particular institutional environments can we begin to decipher the distinctive pattern of working-class formation in the United States’.Footnote 14

In her response to my article, Schmidt then critiques the incremental change theory of Streeck and Thelen and Mahoney and Thelen,Footnote 15 arguing that they categorize change but cannot explain it because they do not deploy or emphasize ideas. This is not correct, because these scholars do talk about the way in which actors interpret their reality and seek to carve out discretionary spaces in institutional settings. For example, Streeck and Thelen talk about actors engaging in ‘interpretive struggles’ over institutional rules, whilst Mahoney and Thelen talk about how institutional rules and expectations are often ambiguous and subject to ‘interpretation, debate and contestation’.Footnote 16 These writers could go further down this track, especially in more fully theorizing constructivist elements, as called for by Schmidt,Footnote 17 but the more important point is that Schmidt's critique overlooks or underplays a crucial element that these scholars do deploy within a HI framework – agency. This is what drives incremental change in their approach and it is what frees their approach from the sticky version of HI theory. Ideas will certainly shape the change process and help define agency, but agents themselves and their capacity to act strategically in institutional settings is a certainly a prime mover in institutional change processes.

Although seeing it as ‘intuitively appealing’, Schmidt also questions my argument that institutions can have both constraining and enabling effects. ‘For Bell, institutions constrain, until they don't, when they become resources’.Footnote 18 This is not my position. I simply argue that institutions can have constraining and enabling effects on agents. This view is now commonplace in many institutional accounts.Footnote 19 I can think of many ways the School I work in has such effects. Schmidt seems to think that if institutions provide agents with resources, then ‘their institutional effect, their ability to structure behaviour, falls away’.Footnote 20 But why? Surely institutions can have constraining and enabling effects, limiting some forms of agency and enhancing other forms, in a contingent manner. Schmidt claims that such institutional enablement is tantamount to ‘using institutions to change institutions’, something she labels as a ‘deep endogeneity problem’.Footnote 21 Yet, agents routinely use institutional resources to help win battles and forge change. In the empirical case in my article, I show, for example, how Governor Fraser of the Reserve Bank of Australia (RBA) appealed to the Bank's dual-goal legislative mandate (read institutional resource) to help him win internal battles over monetary policy and the direction of institutional change. Schmidt's attempt ‘to get over’ her self-constructed problem here involves considering ‘how agents get beyond their institutional constraints, with ideas conveyed through discourse’. But why should we be suddenly bracketing off or trying to ‘get beyond’ the issue of institutional constraints? Surely, this is an important part of any institutional analysis?

In my view, my approach helps us to move beyond several theoretical dualisms that hinder attempts to explain institutional change endogenously. For some reason Schmidt reconstructs the two dualisms I isolate and critique, especially in sticky HI accounts (the distinctions between change/stasis and exogenous/endogenous change processes) into a quite different set, which I find mystifying. In any case, moving beyond such dualisms is important. I fully endorse Streeck and Thelen's arguments about how the theorization of incremental change processes can help move us beyond the change/stasis dichotomy.Footnote 22 Using an ‘institutionally situated agency’ approach, as part of a more flexible HI analysis, can also help overcome the distinction between exogenous and endogenous change processes. As I argue in my article, even a supposedly exogenous account of change – such as the impact of a crisis – cannot in any reasonable sense be thought of as just that. Only agents operating within institutions can produce change; hence, it is hard to see how exogenous and endogenous change dynamics can be separated. Schmidt, however, questions collapsing the change/stasis dualism, worrying that such a move will either privilege incremental change and/or make it more difficult to distinguish between incremental and more rapid forms of change. Agents can produce change incrementally or perhaps more rapidly, depending on the circumstances. Why collapsing the change/stasis dualism theoretically privileges incremental change is not really explained by Schmidt, and, on the second issue, surely telling the difference between incremental and more rapid change is an empirical issue.

Contrary to Schmidt, my approach is not ‘derived’ from Margaret Archer's dialectical analysis (or from Wendt or Culpepper, whose work I hardly mention). Instead, I endorse Archer's approach as a way of theorizing the mutually shaping relations between agents and institutions over time. Schmidt thinks that dialectical analysis of this kind ‘restates the problem: it does not solve it’. I disagree. Agents and structures are mutually shaping over time. In my view, Archer's approach provides a very useful analytical account of these dynamics. Indeed, I agree with a recent analysis by Keith Dowding where he argues that the agency/structure problem has essentially been ‘solved’.Footnote 23

Schmidt also questions my claims that institutions are ontologically prior and also emergent. This is the whole point of the temporal approach embedded in Archer's dialectics. At any given time, agents will confront existing institutions. Over time, change processes produce new institutional forms, essentially new emergent forms born of the change process. So, I agree with Schmidt's claim that institutions cannot be ‘causally prior, and emergence all at once’. But this is not what Archer or I argue. The ongoing temporal change process is the key here. Schmidt claims that ‘an institution cannot emerge from itself’, but this is exactly what institutional change, as an endogenous dynamic process, is all about.

Schmidt applauds the constructivist elements of my approach, saying ‘I could not agree more’. In her critique of my empirical case study, she attempts to shoehorn me more fully into the constructivist camp. On the way she questions my use of Hay's concepts of ideational crisis construction and ideational path dependency. But surely, being critical of Hay's particular attempt to build a new constructivist institutionalism does not preclude me from referring to some of Hay's better ideas? In any case the main thrust of my approach is to locate ideationally active, interpretive agents within institutional settings. I use the example above of how Governor Fraser strategically used the Bank's legislative mandate to help win internal struggles. In her shoehorning effort, Schmidt wants to collapse this legislation, which I think most analysts would be happy to call an ‘institution’, into a ‘norm’, something that is ‘ideational, constructed’! In my view, legislation is a good example of a piece of hard or ‘black letter’ institutionalization. True, legislation embodies ideas, but legislation is more than just a set of ideas held by given agents. I return to this issue below.

I also distinguish between institutions and wider structures, things that often get conflated in institutional theory. A crucial part of my empirical story argues that key agents within the RBA were empowered by a broader set of structural changes in the economy, namely Australia's radically improved terms of trade during the 1990s as well as a more flexible, resilient economy in the wake of extensive microeconomic reforms. These structural changes were important. They helped empower and embolden the Bank's leaders as they confronted an assault from financial market speculators that had already wrecked havoc elsewhere during the Asian financial crisis. The structural changes meant that the economy and the RBA could more easily withstand a currency sell off. Yet Schmidt uses the word ‘vague’ to refer to these factors.

Schmidt also thinks that the Asian crisis was ‘exactly about a moment of uncertainty when Australian policy makers ignored what their institutions told them and tried a new policy idea’. But this was not the case. The Bank's leaders were primarily guided by their institutional mandate, one that stressed the importance of not only controlling inflation but also maintaining growth and employment, and thus, in this case, not appeasing the currency markets with higher interest rates, potentially threatening a domestic recession. True, the response was not exactly orthodox central banking. But as I have explained elsewhere, a key factor assisting such expansionary policy choices was that the RBA's leaders could appeal to the growth and employment mandate in the Bank's legislation, as set out in brass letters against black marble in the Bank's foyer in Martin Place, Sydney.Footnote 24 Schmidt also claims that my theoretical emphasis on dialects ‘inform, as far as I can see, none of the case study’. But dialectical processes of mutual shaping are exactly what is going on in terms of the relations I depict between my key agents and the changing institutions and structures in which they are embedded.

Schmidt thinks that ‘Bell's own analysis is really much more constructivist than it is institutionalist’. I find constructivist insights very useful, as evidenced in my article, and in a more recent paper dealing with how ideas mediated the structural power of financial markets.Footnote 25 However, I reject Schmidt's implied notion that there is some kind of contest between constructivist and institutionalist analysis. As she states in a recent paper, ‘historical and discursive institutionalism are nevertheless two very different analytical enterprises’.Footnote 26 I reject this claim. The flexible HI approach I elaborate begins its analytical investigation with agents and works outwards to explore how agents interact with institutions. Discursive analysis is very important in how we understand the ideational drivers of agent's behaviour. But there is no need to draw a distinction between discursive analysis and the kind of HI analysis I advocate. This was the whole point of my article, to show how constructivist insights can help build rounded accounts of agency within institutional settings. Certainly, ideas matter, but so does the way in which interpretive agents interact with institutions. The aim is to theorize and to illustrate empirically the mutual shaping of agents and institutions over time and how this shapes change processes.

In my view, the nub of this debate boils down to how we see institutions and how constitutive we take ideas to be. Schmidt distinguishes between scholars who ‘talk about “ideas as institutions” and those who talk about “institutions and ideas”. Schmidt is in the former camp, whilst I am in the latter. I think institutions are more than just real-time ideational constructs. Schmidt approvingly quotes Seale's definition of ‘ “Institutional facts” as those things that exist only through collective agreement about what stands for an institution’.Footnote 27 The problem with this formulation is that whilst institutions are shaped by agents, institutions also confront agents with sets of what Searle calls ‘constitutive rules’Footnote 28 and with other constraining and enabling conditions that can become, as Adler puts it, ‘diffused and consolidated’Footnote 29 over time and which shape behaviour. It may be possible under extraordinary (say revolutionary) conditions for agents collectively to overturn or deny institutions, but, more ordinarily, institutions confront agents in the here and now as embedded, already structured terrains. Institutions are thus ontologically prior to the individuals who populate them at any given time. This is what gives institutions causal properties and why at the bottom we pursue ‘institutional’ analysis. The same logic applies the analysis of structures. Yet, on Schmidt's reading, if we are to reduce institutions to ideas, does not the same apply to structures? The implication here is that we redefine a major structural shift, such as Australia's changing terms of trade in the 1990s, as simply ideational. True, such events are always subject to interpretation by agents, but to wholly collapse such events into the realm of the ideational goes too far in my view.

Ultimately, the approach I advocate does not give primacy to agents, institutions, structures or ideas, but instead holds each to be mutually constitutive in a dialectical manner. Agents, as Archer puts it, thus confront institutions and structures as a ‘distinct strata of reality’,Footnote 30 which must be dealt with in the here and now and perhaps changed over time. In other words, institutions are more than just real-time ideational artefacts. Moreover, ideas do not operate in a vacuum but are instead ‘embedded in a historical context and need institutional support to be effective.’Footnote 31 Hence, agent's choices are not made on a tabula rasa. Agents and environments interact and mutually shape one another over time. When Hay and Rosamond argue that what matters is the ‘ideas that actors hold about the context in which they find themselves rather than the context itself’,Footnote 32 they adopt either/or thinking and overlook this crucial dialectical interaction.

Hence, at one level, institutional or structural environments are analytically distinguishable from agents and are ‘out there’, exerting real (though always interpreted) effects by imposing costs or benefits on agents, by shaping (though not determining) actor's interests and preferences, the scope of agent's ‘bounded discretion’ in institutional settings and the resources and opportunities that are available to agents. Schmidt worries about theorizing institutions as ‘external (formal) institutions’ because she assumes this will radically limit agency. She depicts the approach of Robert Lieberman, whom she casts as an ‘institutions and ideas’ scholar, as creating the dilemma ‘of how to explain how agents conditioned by the institutions that limit their ideas are able to transform those very same institutions, since they are seemingly caught within them.’Footnote 33 Schmidt seems fixated on the sticky version of HI and fails to fully appreciate that the more flexible version of HI allows greater room for interpretive agency. In this latter view, there is no reason why we should not be integrating suitably tailored constructivist accounts into a flexible HI approach. That way we would get the best of both worlds.