Comparative public policy combines theories of the policy process with the study of political systems and specific issue areas. Yet, some ambiguity surrounds what distinguishes the comparative approach from other perspectives on public policy. This review brings greater clarity to the comparative policy project by emphasizing the need to be attentive to similarities and differences regarding the institutional contexts in which policymaking takes place. This attention is necessary to avoid “forcing a fit” between the empirical reality and theories and frameworks designed with specific institutional configurations in mind. While forced fit posed problems for past research, recent theoretical advancements have been devised to facilitate comparison across dissimilar institutional settings. The following discussion highlights amendments to established approaches intended to deal with problems of comparison and identifies promising new perspectives from which comparative analysis may be conducted. The latest wave of comparative policy scholarship, having accounted for institutional variation, looks beyond institutions to policy discourses in order to explain how ideas, norms, and political culture affect how policy actors maneuver within, maintain, or change the institutional environment in which they operate.
comparative public policy, institutions, policy discourse, political culture
The idea that the study of public policy both is and should be comparative is well established in the literature (Dodds, 2012; Heidenheimer, Heclo, & Adams, 1975). Yet, to the extent that comparative public policy is a unified body of knowledge, its origins were not explicitly comparative. While few would deny that there have been exemplary comparative studies over the years, they tended to be “one-off” projects (e.g., Hall, 1986; Heclo, 1974; Wilks &apm; Wright, 1987). In contrast to comparative political economy,1 the most popular approaches for studying public policy were for the most part not designed with comparison in mind; rather, many were conceived to explain policy processes in most-similar policy environments, namely those found in the United States (Sabatier, 2007a, p. 11). Many leading theories were thus originally intended to explain policy variation with institutional variables held constant. As a consequence, studies that sought to apply certain perspectives to cases outside the institutional milieu for which they were intended risked erring as a result of “forced fit,” which occurs when empirical findings are tailored to suit theoretical assumptions. Forced fit is a problem because it runs afoul of the standard procedures of scientific inquiry, specifically that theory be updated to more accurately reflect empirical realities.
Fortunately, core perspectives on public policy have since been amended to account for institutional variation (Baumgartner et al., 2009; Cairney & Jones, 2016; Weible & Jenkins-Smith, 2016). This push to make the study of public policy more comparative has coincided with the proliferation of new approaches, many of which are not sui generis but rather novel offshoots of established theories and frameworks (Schlager & Weible, 2013). Having developed means to account for institutional variation, comparative policy scholars have recently begun to look beyond institutions to explain how policy discourses and political culture affect policy outcomes (Jenkins-Smith, Silva, Gupta, & Ripberger, 2014; Schmidt, 2008; Trousset, Gupta, Jenkins-Smith, Silva, & Herron, 2015; Weare, Lichterman, & Esparza, 2014). This shift of emphasis from rules to context has involved significant efforts to bring together positivist, post-positivist, and constructivist-interpretivist themes and methods (M. D. Jones & Radaelli, 2015). By building on established literatures, new lenses on public policy seek to improve upon, rather than compete with or replace, existing perspectives (Breunig, Koski, & Workman, 2016; Cairney & Heikkila, 2014, p. 383; Howlett, McConnell, & Perl, 2016). More than ever before, the comparative project is characterized by a desire to expand the frontiers of conventional wisdom, eschewing the parsimony of most-similar comparison in the interest of getting a more systemic understanding of how variables nested at different levels of abstraction come together to affect the policy process and policy outcomes.
The foundation of any science, from the most elementary to the most complex, is a system of classification based on thoughtful comparisons (Mill, 1868). While the process of classification may be inductive in its initial stages, the epistemology of comparison is deductive: it involves determining whether two or more phenomena are alike or dissimilar according to some established criteria. As these criteria become more elaborate, the line separating facts from theory blurs (L evi-Strauss, 1966). Classification comes to rely progressively on data not directly observed but rather inferred by means of hypothesis testing (Popper, 1959, p. 76).
This practice, known as the hypothetico-deductive method, relies on replication for the purpose of falsification.2 The intention of the “research program”—as opposed to experiments in the singular—is to avoid error by performing more sophisticated tests than single experiments permit. Science is said to progress according to a “logic of discovery” premised on the systematic amendment or elimination of existing theories (Lakatos, 1976; cf. Popper, 1959). Since social scientists lack means of experimental control, comparison across multiple cases is the best available alternative for assessing the strength of hypotheses. Beyond replication, an advantage of comparative research is that biased estimation may be gradually reduced as more cases are analyzed.3
What we call cases in policy research range from the very general to the highly specific. Policy generalists typically focus on the process by which policy is made. They seek to explain not the substance of policy per se but rather the procedures and processes that produce policy outcomes (Sabatier, 2007a). By contrast, policy specialists focus on the substance of specific policies. These scholars are interested in why policy outcomes vary across jurisdictions, which involves making substantive comparisons by studying the same policy in a variety of settings (e.g., Bonoli, 2003).
Although specialist studies have led to many valuable insights, the emergence of comparative public policy as something resembling a unified field is rather recent and follows from the generalist literature (Baumgartner, Jones, & Wilkerson, 2011; Sabatier, 2007b). Insofar as there is a community of researchers engaged in continuous dialogue and the collective pursuit of theory building, the origins of this research tradition were not all that comparative. Rather, given that they stemmed principally from the study of American politics, the roots of comparative public policy were comparative primarily in the most-similar case sense (Blomquist, 2007; Sabatier, 2007a, p. 11).
Consequently, comparative policy scholars outside of the United States were until recently confronted with a dilemma. They could attempt to avoid forced fit by cautiously applying a given approach to cases it was not necessarily well suited to explain, taking care to critically assess the results and amend the theory if necessary (as per the hypothetico-deductive method). Alternatively, international comparative policy scholars could devise their own approaches better suited to comparative analysis. While the latter option is perhaps the more obvious choice, the extent to which it is more conducive to knowledge-building in a global community of scholars hinges upon whether alternative perspectives are successful in both challenging and supplanting dominant approaches (a process akin to Kuhnian notions of paradigmatic overthrow).
What has been dubbed the “network approach” is by far the most well-known alternative to core perspectives on public policy (Adam & Kriesi, 2007). Inspired by Hugh Heclo, the network approach is premised on the idea that concepts historically used to describe policy settings in the United States were “disastrously incomplete” (Heclo, 1978, p. 88). The takeaway from this literature is that policy networks are characterized by different dynamics in different political systems because of different institutional configurations (Scharpf, 1997). Following from this observation, variation in decision rules has been used to construct elaborate typologies of policy networks (Van Waarden, 1992).
Limitations of space preclude a thorough discussion of the networks literature and its contribution to comparative public policy (see, e.g., Knoke, Pappi, Broadbent, & Tsujinaka, 1996; Wilks & Wright, 1987). Suffice it to say, discussion of policy networks has waned in recent years, likely because its advocates failed to produce a widely accepted typology of networks, much less a unified theory (but see Marsh, 1998). The effort was not for nothing, however. On the issue of forced fit, policy network scholars were instrumental in challenging the assumption, popular in American political science, that institutions limit debate on complex policy issues to a single (often left–right) dimension that permits the formation of stable policy monopolies (Heclo, 1978, p. 119; cf. Baumgartner & Jones, 1993, p. 5; Shepsle & Weingast, 1981, pp. 510–11). A close analysis of this argument reveals that tendencies toward stable policy monopolies are a consequence of agenda–setting roles and majoritarian decision rules (Scharpf, 1997; Tsebelis, 2002). The implication is that alternative decision rules and policymaking processes, which are common in political systems outside the United States, do not produce such “winner-take-all” outcomes.
In sum, a desire to avoid forced fit provoked a reaction from policy scholars outside of the United States that, for better or worse, hindered the advancement of comparative public policy. Nonetheless, regardless of whether the major debates of previous decades were based primarily on genuine controversy or mere confusion, the smoke appears to have cleared (see, e.g., Börzel, 1998; Dowding, 2001). As it stands, although the concept of policy networks was coined as a reaction to the American conceptualization of policy subsystems, these two terms are now used more or less interchangeably (Cairney & Heikkila, 2014, p. 365).
The remainder of this essay chronicles the comparative turn in the policy literature. I first detail how established lenses have dealt with issues of comparison and emphasize that, notwithstanding the institutional analysis and development (IAD) approach (which was designed with comparison across institutional contexts in mind), the most popular perspectives used to study public policy have had to be consciously reconfigured in order to account more satisfactorily for institutional variation. I then discuss recent lines of theoretical inquiry, which stand out from earlier research in two main respects. First, they avoid a national-level orientation; second, they offer a more sophisticated account of policy discourse than was typical of past research. The latter trend may prove useful for not making too much of formal institutions.
As Ostrom (2011, pp. 8–9) reminds us, analytical models are too precise in their predictions to be of much use beyond most-similar comparisons. Theories and frameworks devised to explain phenomena in a general class of cases, however, can be amended or extended to explain phenomena in a still larger class.4 The four most prominent perspectives on public policy are theories and frameworks. They are the Advocacy Coalition Framework (ACF), Institutional Analysis and Development (IAD), Punctuated Equilibrium Theory (PET), and the Multiple Streams Approach (MSA). With the exception of IAD, which has been geared toward explaining the effect of institutional logics on policy outcomes since its inception, the other three approaches were developed to explain political behavior in the American institutional environment. Consequently, many time-honored perspectives for studying public policy were not designed with institutional variation in mind. Rather, in contrast to the “metatheoretical” orientation of IAD, their intent was by and large to demonstrate how common institutions—namely, those of American government—produced similar outcomes regardless of variation with respect to the policy issue under analysis (Baumgartner & Jones, 1993). Despite their origins, recent efforts to make core perspectives amenable to comparative analysis have made forced fit less and less of an issue as more precision and qualification has been built in to each approach.
In contrast to other perspectives on public policy, IAD has been both explicitly comparative and explicitly institutionalist from the outset. Since its beginnings in the early 1980s, IAD scholars have emphasized the usefulness of game-theoretic analogies to explain policy outcomes wherein institutions—which are understood as the rules of the game—are considered the most important explanatory variable (Kiser & Ostrom, 1982). As in board games, rules in political games have wide-ranging, deterministic effects on player strategies, the distribution of resources, series of moves and, ultimately, outcomes (Scharpf, 1997).
The wisdom imparted by IAD—and institutional rational choice more generally—is that causal variables are not entirely independent but rather activated by institutional rules (Immergut, 1998). Institutions often take the form of necessary but insufficient conditions for a particular outcome. Determining the precise combination of causal variables in a given situation demands that researchers assess how institutions “fit” within the explanation they wish to employ (Young, 2002).
Given that rules are neither agents nor resources but rather influence the causal process by imposing contingencies on agents’ behavior (Hurwicz, 1973), IAD relies on a complex, multilevel, seven-part typology of rules to explain how a variety of rules may combine to impact policy outcomes in a particular situation (Ostrom, Cox, & Schlager, 2014). Moreover, institutions are not the only variable of interest. Because the attributes of actors, environments, and communities also affect outcomes, policy action is seen as only “partially dependent” on rules (Ostrom, 2011, p. 17). As such, and although IAD scholars maintain that a systematic approach to the study of institutions is essential for informed comparisons, they acknowledge that a complete theory of institutions is likely beyond our reach (Kiser & Ostrom, 1982).
Considering IAD’s sophisticated account of institutions, several authors have suggested that IAD insights be imported into other perspectives on public policy for the end of improving comparative research (Basurto, Kingsley, McQueen, Smith, & Weible, 2010; Real-Dato, 2009; Schlager, 2007). While there is no obvious reason why alternative perspectives should be impervious to the tools of IAD (or other institutionalist tools, such as game theory or veto players analysis, for that matter), most have confronted the issue of cross-institutional comparison in their own ways (but see Lubell, 2013).
An outgrowth of Cohen, March, and Olsen’s (1972) “garbage can model of organizational choice,” MSA appeared at a time when social scientists were particularly interested in the organizational labyrinth that is the state (Jessop, 1990; Meltsner, 1976; Skocpol, 1985). Looking both inside the state and beyond it, Kingdon’s (1984) Agendas, Alternatives, and Public Policies accounted for the synergies between the various moving parts that interact to produce public policy.
Although the variables in MSA are many, they are collapsed into three independent “streams” representing politics, policies, and problems. Complementary conditions in the three streams are hypothesized to create temporary windows of opportunity during which “policy entrepreneurs” may effect change to the policy status quo. Although some subcomponents comprising the streams are institutional (e.g., network characteristics), MSA was not designed for cross-country comparisons but rather developed to explain variation within the macropolitical institutions of American government.
Not long after Agendas, Alternatives, and Public Policies was published, policy scholars began to amend MSA to make it more applicable to comparative analysis (Durant & Diehl, 1989). The latest MSA “meta-review,” which limited its analysis to the period between 2000 and 2013, found studies spanning 65 countries, multiple levels of government, and 22 policy areas (M. D. Jones et al., 2016). The international appeal of MSA no doubt follows from efforts on the part of Nikolaos Zahariadis to expand the framework’s range of application beyond agenda setting in the U.S. Congress to the larger policy process (Zahariadis, 2007). Chief among the amendments is an appreciation of cross-jurisdictional variability with respect to how institutionally structured the process of agenda setting is. Whereas Kingdon (1984) described the environment in which policy proposals are developed as a “policy primeval soup,” Zahariadis and Allen (1995, p. 72) point out that, outside the United States, the environment is often more “stew-like”—that is, less fluid, more bureaucratic, and at times quite scripted (see also Durant & Diehl, 1989). Others have also pointed out that agenda setting is much more structured, and thus less fortuitous and more predictable, in systems in which executives control the legislative process than it is in the United States (Howlett, 1998; cf. Light, 1999).
In short, although a central premise of MSA is that policymaking takes place within an environment of “organized anarchy,” comparative research using MSA has been careful to qualify the extent to which policy settings are organized or anarchic. It would be a mistake, however, to assume anarchy, or outcomes typically associated with it, follows from a lack of institutions. On the contrary, unpredictability in the American political system is largely due to a multiplicity of overlapping institutions and policymaking jurisdictions (i.e., checks and balances). Political arrangements that disperse authority give rise to “institutional frictions” that increase policy stability while reducing predictability with respect to major policy change. Such is the crux of PET.
PET was originally developed to explain episodic spikes in the otherwise incremental pattern of federal budget allocations in the United States (Baumgartner & Jones, 1993). PET surmises that day-to-day policymaking in the United States occurs within relatively closed, often monopolistic, policy subsystems. Owing to the institutions of American government, monopoly control over how policy is understood, formulated, and implemented is disrupted when issues rise to prominence on the legislative agenda. In contrast to subsystem politics, in which conflict is restricted by institutions limiting the number of policy-relevant issues considered at any time, the macropolitics of Congress allow the agenda space to expand. As a consequence, previously neglected issue dimensions elicit government attention. Subsystemic monopolies are challenged as the venues in which certain policy issues are handled increase in number or otherwise change. Owing to the status quo bias of American political institutions, subsystemic policymaking is characterized by stability while policymaking at the macropolitical level is periodically change oriented.
Given the American flavor of PET’s origins, it is not surprising that early observers doubted the applicability of its conclusions to nonpresidential systems (Howlett, 1997). In response, the architects of PET have since added empirical and theoretical precision to their studies, concluding that although “differences in the law’s basic parameters are country and institution specific,” PET constitutes both a “general empirical law of public budgets” and a “theory of government information processing” (B. D. Jones & Baumgartner, 2012; B. D. Jones et al., 2009, p. 856). For example, applying PET to budgetary allocations in France, Baumgartner, Foucault, and Franc ̧ois (2006) discovered a pattern of punctuations “remarkably similar” to those found in the United States. Comparing several government activities in the United States, Belgium, and Denmark, Baumgartner and colleagues went on to demonstrate that since institutional costs escalate over the course of the policy process regardless of the political system, observable policy outputs are increasingly punctuated as the policy process progresses from agenda setting to decision making (Baumgartner et al., 2009).5 Controlling for institutional difference in these studies allowed the authors to conclude that limitations of human cognition, which are the same everywhere, must be more significant than institutional factors in determining shifts in government attention that ultimately lead to policy outcomes.
The fact that PET’s premises appear to be generalizable does not mean of course that researchers can afford to eschew consideration of institutions (M. D. Jones & Jenkins-Smith, 2009). PET, after all, posits that “frictions” leading to the punctuated pattern of policy outputs are a consequence of the circumscribing effects of institutions on government attention and information processing (B. D. Jones & Baumgartner, 2005). As stressed by Baumgartner, Jones, and Mortensen (2014, p. 88) “it is critical in the future to begin to understand which aspects of policymaking are due to more general dynamics based in human cognition and organizational behavior and which are due to the particulars of the institutions under study.”
Thus, despite the apparent generality of outcomes, using PET for analysis of cases beyond the United States requires that researchers be well acquainted with institutional variation across countries. Although it may be true that policy is punctuated regardless of the policymaking system, the institutions governing the specific pattern of punctuations vary tremendously from system to system and must be accounted for (John & Jennings, 2010). Where (and by whom) the agenda is set determines how the policy process unfolds (Light, 1999). Beyond recognizing that whether a political system is presidential or parliamentary will determine if the agenda is set by the executive or legislative branch, researchers should be appreciative of subtler institutional idiosyncrasies as well (Tsebelis, 2002). At the subsystem level, two factors—executive capacity and procedures governing participant interactions—bear significantly on policy learning and, consequently, policy responsiveness to external stimuli (Breunig & Koski, 2009). Institutional variation at the subsystem level should explain differences in the amount of “bottom up” policy change observed across cases. Accurately accounting for institutional variation at the subsystem level requires that PET researchers reflect upon what rules and procedures lead policy subsystems toward monopolies in the U.S. context and consider whether the story is the same elsewhere (Scharpf, 1997). We should, for example, expect institutions designed to facilitate corporatist interest intermediation to produce subsystem coalitions completely different from those found in more adversarial policy settings (Knoke et al., 1996; Wilks & Wright, 1987). This brings the discussion to how tendencies toward policy monopoly have been treated in the explicitly subsystem-oriented ACF.
Developed in the 1980s, the ACF marked a significant departure from how public policy had previously been studied by extending the scope of analysis beyond what was conventional for political science. As Jenkins-Smith and colleagues explain, the ACF was radical in its program to (i) cast off the rigidities of the policy stages heuristic, (ii) provide means for analyzing the role of technical information in political debate, and (iii) explore policymaking dynamics as they played out away from public view in policy subsystems (Jenkins-Smith, Nohrstedt, Weible, & Sabatier, 2014, p. 184). The major theoretical challenge confronting ACF scholars was to differentiate between those aspects of policy that come about as a consequence of learning and those aspects that depend on “noncognitive factors” which establish “basic resource and normative constraints” (Sabatier, 1987, p. 651).
The ACF is not a theory but is rather an analytical framework (Jenkins-Smith, Nohrstedt, et al., 2014, p. 188). Nevertheless, Sabatier’s (1987) delineation of two general avenues by which policies remain stable or change constitutes a theoretical claim about actors’ propensity to alter their preferences in the face of evidence that contradicts prior beliefs. It premises that policymaking will only reflect the dynamics of solutions-oriented learning when differences of opinion do not threaten actors’ deeply held “core beliefs.” When core beliefs are in conflict, solutions will depend on the resources held by advocacy coalitions, how well coalition strategies fit professional norms and standards of policy forums, intervention on the part of a policy broker, or some combination of these factors (Sabatier, 1987, p. 683). Explanation in the ACF is thus part universal and part institutional. It is universal in the sense that all actors are assumed to have belief systems that are organized according to core and peripheral aspects regardless of situational setting. It is institutional in that, on one hand, brokerage roles are institutionally determined while, on the other, opportunities to marshal and utilize coalition resources are governed by institutionalized opportunity structures and professional norms (Sabatier & Weible, 2007; Weible, Sabatier, & McQueen, 2009).
Attention to the distribution of coalition resources in pluralistic policy settings reflects the fact that the ACF—like MSA and PET—was designed not only with the institutions of American government in mind but also its adversarial political culture. Despite being U.S.-centric in its initial formulation, international applications of the ACF have since grown to exceed U.S.-based studies (Weible et al., 2011). The result has been the development of variations on the ACF intended to capture the nuances of particular institutional arrangements (Weible et al., 2009).6 However, rather than allowing the framework to be overwhelmed by ceaseless typological distinctions, beginning with Sabatier (1998), the ACF’s architects have accommodated complexity while maintaining the framework’s general applicability by incorporating a limited number of intervening variables to the standard depiction of the ACF (Sabatier & Weible, 2007).7
Yet, despite efforts to build nuance into the ACF, Jenkins-Smith, Nohrstedt, et al. (2014, p. 205) note that, while the ACF is “based on implicit comparison across political-institutional systems” they are “not aware of any empirical study based on the ACF that systematically compares policy subsystems, coalition behavior and policy processes across political systems.” More recently, Weible and Jenkins-Smith (2016, p. 23) stated “one of the next steps is to document systematic differences in coalitions across different forms of governments.” Thus, while proponents of the ACF long have been cognizant of the differential impacts of institutions on policy outcomes, a systematic comparative research program has been slow to materialize.
Determining whether and how institutional configuration matters requires that institution-specific hypotheses be advanced and tested across a number of cases (e. g., Gupta, 2014). Jenkins-Smith and colleagues explain:
frameworks [e.g., the ACF] are not directly testable but provide guidance toward specific areas of descriptive and explanatory inquiry ... a framework supports multiple theories which are narrower in scope and emphasize a smaller set of questions, variables, and relationships. Theories provide more precise conceptual and operational definitions of concepts and interrelate concepts in the form of testable and falsifiable hypotheses or propositions. The theories within the framework are where students and researchers should attempt to test and develop descriptions and explanations. (Jenkins-Smith, Nohrstedt, et al., 2014, p. 189)
It should be no surprise, then, that emerging theory is not entirely sui generis but rather emblemizes offshoots and hybrids of the ACF and other established perspectives summarized here. As detailed in the discussion to follow, new perspectives advance hypotheses related to specific contingencies that the ACF in its basic formulation purposefully avoids. Seen in this light, the established literature offers, to greater and lesser degrees, context-free frameworks upon which more specific theories and models may be devised.
In contrast to the one-shot nature of many previous comparative policy studies, the latest wave largely consists of outgrowths of established theories and frameworks. The literature now comprising what could be considered comparative public policy is voluminous and growing (Schlager & Weible, 2013). Consequently, there is only space to mention a fraction of all the new and emerging perspectives currently being developed. A few themes stand out.
Whereas conventional approaches to studying public policy were premised on the study of formal, geographically situated institutions and organizations, contemporary governance demands that researchers be attentive to transboundary policy dynamics (Eisner, 1993; Jochim & May, 2010). The main insight gained from recent literature on transboundary policymaking is that patterns of collaboration and conflict have both scaled up to regional and international levels as well as scaled across individual policy subsystems, each of which may have unique procedures of its own (Martinsen & Wessel, 2014).
In contrast to earlier musings on the expected effects of globalization on public policy (e.g., Strange, 1996), recent literature on “transboundary subsystems” and “policy regimes” explains why the degree of policy convergence varies from one policy area to the next. Building on both the ACF and PET, a key finding of this research is that policymakers are just as often capable of preventing the adoption of a particular policy—or of tailoring it to their own ends–as they are powerless to resist it (M. D. Jones & Jenkins-Smith, 2009; Worsham & Stores, 2012). Properly anticipating how complex, multilevel interactions play out demands that researchers be mindful of how nested institutions combine to determine the arrangement of veto players in multilevel actor constellations. While political science is equipped with tools to do this (Scharpf, 1997; Tsebelis, 2002), branching off IAD, recent works using Norton Long’s concept of the “ecology of games” advance a framework specifically suited to the analysis of such “complex adaptive systems” (Lubell, 2013).
Capitalizing on recent advances in network science, the ecology of games approach uses exponential random graph models to explain why some elements in complex systems of actors and institutions thrive while others decay. Importantly, this evolutionary theory looks beyond mere institutional effects (“capacity constraints”) to include variables related to norms, narratives, and discourses (a “reputation institution”), the latter of which have been shown to allow higher relative payoffs for cooperators but only when information about reputations flows quickly and only when systems are characterized by a cooperative ethos (i.e., relatively more cooperators than defectors) (Smaldino & Lubell, 2014).
The ecology of games approach thus marks a very sophisticated return to systems analysis, but one that does not neglect lower order processes taking place at the policy subsystem level. Contrary to conventional systems-theoretic approaches to the study of public policy—which were critiqued for paying inadequate attention to the deliberative and decisional processes involved in policymaking (Blomquist, 2007, p. 272; Hofferbert, 1990, p. 147; Schlager, 2007, p. 313)—the ecology of games framework, policy regime theory, and other work on transboundary subsystems are attentive to how processes occurring within and across subsystemic units bring about stability or change in the larger policy system. Such a return to big picture considerations, which follows in no small part from advances in data collection and analytical methods, constitutes a major trend in the policy literature. Another major trend, related to the return to system-level analysis but not obviously so, is a phenomenological orientation that emphasizes the role of ideas, norms, and narratives as manifested in policy discourses.
Phenomenology and hermeneutics are by no means new to political social science (Dryzek, 1982; Giddens, 1984). Policy studies have particularly benefited from the influence of constructivist lenses on public policy as well as from their proponents’ critiques of conventional approaches (Durnová, Fischer, & Zittoun, 2016; Hay & Wincott, 1998; A. Schneider & Ingram, 1993). Yet, in recent years, owing in large part to advances in research methods, degrees of separation between positivist and constructivist poles of policy research have narrowed (M. D. Jones & Radaelli, 2015).
Although it has been maintained for some time that institutional discourses have “constitutive effects” (e.g., on actors’ interpretations of the action situation, institutional rules, payoff matrixes, opponents’ intentions, as well as their own interests [Blyth, 2003; Schmidt, 2008; Snyder & Mahoney, 1999]), it is only recently that research has shown, using statistical methods, whether and how discourse matters (Wueest & Fossati, 2015). Bringing discourse analytic techniques to the ACF, Leifeld (2013) demonstrates in the case of German pension policy how belief change among early adopters triggered a diffusion process whereby the balance of coalition membership shifted as a consequence of learning. In a similar vein, honing in on the concept of narratives, M. D. Jones, McBeth, and Shanahan (2014) extend the ACF to develop the Narrative Policy Framework (NPF), the purpose of which is to deduce how stories—namely as they relate to assigning roles to “heroes and villains”—are used politically by coalitions to gain support for their policy positions (cf. Stone, 1989). In this sense, NPF shares affinities with the literature on the Social Construction of Target Populations (SCTP), which similarly focuses on the persuasive power of discourse in the determination of which groups are deserving (and which are undeserving) of the benefits of redistributive policies (A. Schneider & Ingram, 1993). Although constructivist approaches have been criticized in the past for their lack of scientific rigor (Dowding, 2001), both the NPF and SCTP are noteworthy for their efforts in advancing specific, falsifiable hypotheses (Sabatier, 2007a, p. 11; Weible, 2014, p. 12).
The trend toward discourse analysis can be attributed to the post-positivist orientation of the mainstream approaches outlined earlier. For example, framing has always been central to PET because of its role in shifting actors’ limited attention toward certain information. Yet, consistent with the general trend of “taking discourse seriously” (Schmidt, 2008), the importance of framing has grown over time as PET scholars have increasingly emphasized the politics of information (Eissler, Russell, & Jones, 2016). Similarly, while IAD has always been attentive to the attributes of actors, environments, and communities, the ecology of games offshoot of IAD goes further to adopt a “realistic model of human decision making drawn from biological and cultural evolution, which recognizes how behavioral, social, and cognitive processes constrain rational choice” (Lubell, 2013, p. 539).
Not all such developments are recent, of course. Branching off of MSA, M. Schneider, Teske, and Mintrom (1995) emphasized the importance of seeing old things in new ways in their discussion of how policy entrepreneurs effect change through policy innovation, adaptation, and “arbitrage.” Likewise, learning and the communication of beliefs have always been central to ACF. But, as with PET and IAD, the role of discourse in the ACF has drawn more and more emphasis and attention over time as evidenced by Leifeld’s (2013) work and the development of NPF as an accompaniment to ACF.
Not wanting to give the impression that all constructs are liberating, it is important to stress that discourse can also be constraining. This happens when ideas are attenuated by pre-existing norms or attitudes. Coming to grips with the variables at play in such situations has led some policy researchers to (re)engage with political culture.
The sort of systemism that colors studies that take seriously discourse and narratives is also apparent in recent research that makes comparisons across political cultures. Given its emphasis on belief structures, those working within the ACF have shown a particular interest in discerning the effects of political culture on policy processes and outcomes. Borrowing insights from Douglas and Wildavsky (1982), Jenkins-Smith and colleagues build a four-part typology of cultural worldviews with the aim of bringing greater precision to the treatment of belief systems in the ACF. Whether one is biased toward an individualistic, egalitarian, hierarchical, or fatalistic worldview is hypothesized to affect how groups construct or otherwise navigate institutional structures (e.g., by entering into coalitions with culturally proximate groups) (Jenkins-Smith, Silva, et al., 2014, pp. 491–92).
Weare et al. (2014) find support for the notion that cultural biases influence whether policymaking is cooperative or adversarial. Importantly, these authors demonstrate that the impact of culture is independent from both actors’ material interests and institutional decision rules: two pillars of explanation in rational choice theory (Scharpf, 1997; Titmuss, 1972). Evidence that preferences are affected by culture is also found in recent survey-based research investigating whether cultural belief systems influence individuals’ readiness to participate in public engagement on the issue of nuclear facility siting. Using the four-part typology of cultural world views outlined above, researchers found that egalitarians and hierarchs are more likely to express a desire to participate in public engagement than fatalists and individualists (Trousset et al., 2015).
As comparative policy scholars continue to extend their gaze beyond political systems in North America and Europe, an appreciation of political culture will be all the more necessary. Although the majority of political systems worldwide are modeled on Western institutions, we should not assume that like-institutions function the same way in different cultural contexts. In game theoretic terms, player strategies are often influenced by cultural norms, both when the payoff matrix is affected by culture and when it is not (Scharpf, 1997, pp. 88, 162–68). Furthermore, as pointed out by Wong (2014), since heuristics used by actors to arrive at policy decisions vary cross-culturally, understanding local idiosyncrasies is a prerequisite to properly employing attention-based approaches (such as PET) in unfamiliar policy environments.
Although an appreciation of cultural norms has long been standard fare for IAD (Ostrom, 1990), comparison beyond North America and Europe came later for other foundational approaches (Henry, Ingold, Nohrstedt, & Weible, 2014). While limitations of space prevent a detailed discussion of the insights gleaned from empirical studies of policymaking in the developing world, it suffices to say that analytical sensitivity to variables beyond institutional rules will be integral to avoiding forced fit as it pertains to the Western cultural biases some suspect underlie contemporary approaches to the study of public policy (Wong, 2014).
Arguably, the three trends discussed above amount to a sea change in the discipline. This shift is most evident with respect to recent efforts to draw together positivist, post-positivist, and constructivist-interpretivist approaches, which have resulted in significant theoretical and methodological synergies (M. D. Jones & Radaelli, 2015; Wueest & Fossati, 2015). Not only have discourse and frame analysis been central to NPF, these concepts and methods have also been increasingly emphasized in the ACF and PET (Eissler et al., 2016; Leifeld, 2013).
Needless to say, inquiry into how institutions are sustained or subverted through cultural practices, narratives, and discourse has drawn policy studies, political economy, and political sociology much closer to one another (Streeck & Thelen, 2005). Moreover, taken together, recent developments in comparative public policy signal a return to more “systemic” explanation that is in some ways reminiscent of classic approaches to public policy (e.g., Dye, 1966; Easton, 1953). However, contrary to the state of policy studies half a century ago, attitudes and behavior of policy actors can now often be observed directly.8 Although much remains to be done, discourse-oriented multilevel models of the sort that characterize the latest research avoid many of the pitfalls that frustrated previous attempts at comprehensive theory-building and rigorous comparative analysis.
Why study comparative public policy? More specifically, what is so important about comparison and why, as I have argued, is it prudent to take stock of institutional variation and, beyond that, policy discourse and political culture? Although the detail and specificity of the case studies that populate policy research make it easy to lose sight of the larger purpose of the social sciences, the development of theories and frameworks for understanding and explaining social phenomena is a prerequisite to improving peoples’ lives.
As per the themes discussed in this essay, attentiveness to institutional context is integral to assessing which institutional configurations produce what policy results and how (Ostrom et al., 2014). Yet, institutions are not everything (Young, 2002). Rather, even when institutions do not leave “gaps” or “contingencies” that permit the exercise of agency (Ostrom, 1990; Streeck & Thelen, 2005), institutions rarely prohibit discourse outright. Consequently, institutions may be discursively navigated in order to subvert or sustain the policy status quo (Schmidt, 2008). However, just as actor constellations are nested within institutional contexts, institutions and actors are nested within still larger cultural structures which may maintain or undermine institutional authority by virtue of the relationship between culture and norms of appropriate discourse (Titmuss, 1972).
The above argument was succinctly captured in the title of a 2003 article by Mark Blyth called “Structures do not come with an instruction sheet” (Blyth, 2003). The trouble with stylized facts of the sort conveyed in Blyth’s title is that stylized facts require qualification. Some structures do in fact come with an instruction sheet, as is the case when institutional rules are codified or otherwise stated. When rules are followed, the job of the social scientist is easy. Blyth’s point, however, is that rules, if they exist, are contestable if not openly contested. When rules are contested, understanding public policy requires that researchers be attentive to the vagaries of the policy discourse. As most forcefully argued by Schmidt (2008), but equally recognized by practitioners in the field of comparative political economy, policy discourse is the primary medium through which institutions are maintained or subverted (Streeck & Thelen, 2005). To this I add, although discourse may be exogenously driven, it is always endogenously mediated by culture.
Returning to the question of purpose, in the very least, a proper appreciation of the nuances highlighted above should prevent comparative policy scholars from concluding that like-institutions will produce similar policy outcomes when transplanted from one context to another (Przeworski, 2004; Weaver & Rockman, 1993). Indeed, research on policy diffusion and transfer has often found that transplanted institutions fail to produce convergence (Steinmo, 2010; Weyland, 2005). While policy scholars may have some explanations for why policy transfer does not always go hand in hand with policy convergence, as noted by Wong (2014), the extent to which theories and frameworks developed to explain public policy in North America and Europe carry overseas remains a rather open question.
Although vexing—and at times frustrating for those hopeful for crisp, straight-forward, or “parsimonious” explanations—the complexity of the world in which we live gives rise to tough, but by no means intractable, questions. As evidenced by the empirical research and analytical perspectives summarized in this essay, students of comparative public policy are well equipped with an expanding set of tools designed with the intricacies of modern policymaking in mind.
Matt Wilder is a Ph.D. candidate in the Department of Political Science at the University of Toronto, Canada.
1 For example, Esping-Andersen (1990), Halland Soskice (2001), Steinmo (2010), and Streeck and Thelen (2005).
2 A review of recent major works in public policy suggests that the field—at least in its main stream— has stayed true to the hypothetico-deductive approach (Sabatier, 2007b, p. 327). Several contributors to the latest edition of Theories of the Policy Process, for example, reflect critically on the extent to which certain hypotheses have stood up to empirical scrutiny (Weible, 2014, p. 398). Far from espousing covering laws, however, social scientists have long recognized that many of their claims are at best probabilistic. Thus, having embraced “sophisticated falsificationism” (Lakatos, 1976), policy scholars are well aware that hypotheses are not easily rebuffed (Meier, 2009).
3 Paradoxically, comparative research may realize the benefits of experimental control by avoiding it altogether. The logic here is that, although a diversity of cases adds confounding variables, consequential variables do not remain unobserved in the long run; rather, they are either recognized by researchers as deterministic and built into models or they are identified as “noise” and treated as such (see Gerring, 2007).
4 The difference among models, frameworks, and theories hinges on the specificity of each and, correspondingly, the degree of predictive precision. Models are reserved for the study of specific phenomena (e.g., community irrigation management in the Philippines [Ostrom, 1990]); theories make assumptions about the behavior of general classes of actors under given conditions (e.g., elite information processing at times of crisis [Jones & Baumgartner, 2005]); frameworks relax assumptions about actors’ specific motivations and instead map out the relationships between relevant variables (e.g., a framework for understanding coalition formation at the policy subsystem level [Sabatier, 1987]).
5 More specifically, Baumgartner et al. (2009) demonstrate that the kurtosis of distributions measuring policy inputs and outputs increases as the analysis moves from the public agenda (i.e., public opinion, election results, media coverage) to the policy process (hearings and the introduction of bills) to budgetary allocations (the passage of money bills) (see also B. S. Jones & Baumgartner, 2005).
6 One particularly illustrative example is Gupta’s (2014) comparison of forest management and nuclear energy policy in India in which she found the level of subsystem centralization to affect whether coalitions adopt confrontational or assimilative strategies.
7 Three of these intervening variables fall under the category of “coalition opportunity structures”—including “degree of consensus required for major change,” “openness of the political system,” and “overlapping societal cleavages”—whereas one other—“institutional rules”—falls under the heading of “subsystem characteristics” (Weible & Jenkins-Smith, 2016, p. 18).
8 As lamented by Richard Hofferbert, a key proponent of earlier systems-theoretic approaches to the study of public policy, “because of the difficulties of measurement and the poverty of appropriate theory, attitudes or behavior of elites (those whose choice should be greatest) were rarely measured directly, but rather equated with the residual once other elements of the model had been correlated with a particular set of policy indicators. . . Theory was and still is light. Induction, guided by previous insights of less technically rigorous scholarship, has driven the inquiry” (Hofferbert, 1990, pp. 146–47).