Theorizing policy diffusion: from a patchy set of mechanisms to a paradigmatic typology

ABSTRACT

In recent decades, we have witnessed the diffusion of policy diffusion studies across many sub-disciplines of political science. Four mechanisms of policy diffusion—learning, competition, emulation and coercion—have become widely accepted as explanations for how policymaking processes and policy outcomes in one polity influence those in other polities. After pointing to major shortcomings of this inductively gained set of mechanisms, we present a theoretically more coherent typology that draws on key concepts from International Relations and Policy Studies. The four mechanisms we lay down consider rationalist and social constructivist approaches equally and they incorporate symmetric and asymmetric constellations. By further distinguishing between processes confined to one policy field and those arising from links across policy fields, we present a typology of eight theoretically consistent pathways of policy diffusion. Our framework enables the aggregation of knowledge and contributes to conceptual coherence in multi-methods research.

KEYWORDS:

• Mechanisms
• mixed methods
• policy change
• policy diffusion
• policy transfer
• typological theory

Introduction

The insight that processes and outcomes of policymaking are not only shaped by domestic factors within polities (nation-states, subnational or supranational jurisdictions) but also by processes and outcomes of policymaking in other polities has made many inroads in political science. Researchers from different subfields have started contributing to a research programme on policy diffusion that is rooted in studies of federalism (e.g., Walker, 1969), has substantial overlaps with policy studies, and whose concepts have been significantly shaped by scholars of International Relations (IR) (e.g., Simmons et al., 2006).¹ Even scholars of comparative politics—who have traditionally presumed that the processes and outcomes within their units of analysis are independent—have begun to take ‘Galton’s problem’ seriously (Braun & Gilardi, 2006).

Yet, the ‘diffusion of policy diffusion research in Political Science’ (Graham et al. 2013, 673) has taken place much more in the empirical than in the theoretical realm, and comparably few scholars have made steps toward a comprehensive theory of policy diffusion.² More recently, we have observed a renewed interest in the theoretical foundations of policy diffusion processes. Gilardi and Wasserfallen (2019, 1246) highlight that we need more theoretical contributions to understand better the ‘politics of policy diffusion’. And most recently, Kuhlmann et al. (2020) argue that the diffusion literature has not paid sufficient attention to ‘power (a)symmetries’.

In this article, we take up these calls and scrutinize a typological theory of policy diffusion. We define policy diffusion as a process in which policymaking and policy outcomes in one polity influence policymaking and policy outcomes in other polities (see Graham et al., 2013, 675; Maggetti & Gilardi, 2016, 89). Our typological theory considers rationalist and social constructivist approaches equally; it does the same regarding symmetric and asymmetric constellations among the units of analysis (polities in rationalist accounts, policies in constructivist accounts). Finally, it combines the basic alternatives within the two dimensions in a configurational manner, which leads to four ideal-typical mechanisms of policy diffusion. We label the four ideal-types according to the kind of motivation that animates agency within the corresponding structures and call them interest-, rights-, ideology-, and recognition-driven processes of policy diffusion.

Our typology of motivational mechanisms provides an alternative to the four mechanisms accepted widely after an initial burst of concepts and terms: learning, competition, emulation,³ and—slightly more contested⁴—coercion (Simmons et al., 2006). ‘Learning’ is cast as a cognitive process that draws on the success or failure of policies and is often equated with an innovative or efficient solution to a public problem. ‘Competition’ refers to processes in which polities adjust their policies because they compete with other polities for attracting mobile factors of production. ‘Coercion’ is supposed to occur when powerful states or international organizations (IOs) force less powerful polities to adopt specific policies through threats and promises. Finally, ‘emulation’ serves as a residual category that includes all processes in which actors are guided by perceptions and standards of appropriateness (Gilardi & Wasserfallen, 2019).

We are convinced that our typological theory holds advantages for analysing the processes of policy diffusion over the established mechanism for two major reasons. First, in contrast to the latter, our typology consists of mutually exclusive concepts, since as ideal-types they are defined to maximize internal coherence and external distinctiveness. Doing so lengthens the odds that these concepts will be specified and operationalized in a consistent way when applied in various research designs and methods and across different policy fields and polity levels. The latter, in turn, is a precondition for the accumulation of knowledge in a field of research. Second, our deductively derived typology makes the dimensions and categories used to characterize and differentiate basic concepts much more transparent than an inductively derived set of mechanisms. Our claim that our typology contains those dimensions and categories that maximize exhaustiveness within the field of study and connectivity to neighbouring fields of study is getting transparent and open for critique. Such transparency, in turn, is a precondition for fruitful theoretical reflections and debates.

The remainder of the article proceeds as follows. In the first part, we show that the dominant set of inductively derived mechanisms applied in most empirical studies of policy diffusion are not defined—and by consequence not specified and operationalized—in a systematic and consistent manner. In the second part, we present our alternative. We briefly sketch our understanding of typological theorizing. Afterwards, we present the paradigmatic presumptions on which our ideal-typical mechanisms are based. Next, we scrutinize distinct pathways of policy diffusion. These pathways specify our ideal-typical mechanisms by combining them with a further criterion (whether the diffusion process is taking place within a policy field/subsystem or whether linkages/interferences across policy fields play a central role) and by connecting the ideal-typical mechanisms to prototypical actors conceptualized in neighbouring fields of study (i.e., Theories of the Policy Process, IR). Finally, we show how our typological theory can lay the groundwork for coherent multi-methods research.

Problems of induction: inconsistent individual mechanisms and biased set of mechanisms

In this section, we critically review the conceptualization of mechanism in the diffusion literature. We argue that each of the four established mechanisms draws on quite diverse strands of literature that propagate a specific view of policymaking processes. This conceptual ambiguity has been reduced in a stylized model of the core presumptions, which underlies most empirical studies of policy diffusion (Gilardi & Wasserfallen, 2019). We show that the inductive reduction to this stylized model has, however, resulted in an imbalance of the overall set of diffusion mechanisms regarding the principled alternatives of two core dimensions: rationalist versus constructivist accounts and symmetric versus asymmetric constellations. Specifically, the mechanisms are skewed towards a rationalist understanding of structure and agency and constellations in which polities stand in symmetric relationships, paying insufficient attention to constructivist conceptualizations and asymmetric relationships between polities. We believe that an important consequence of these conceptual inconsistencies and the inductive reduction is that empirical studies have used diverse and overlapping sets of indicators to measure the four diffusion mechanisms (Maggetti & Gilardi, 2016).

Individual mechanisms: inconsistencies and overlaps

We observe considerable heterogeneity regarding the concept of ‘learning’ in diffusion studies. Simmons and colleagues (2006, 795–799) have already indicated that the conceptualization of ‘learning’ can be based on different theoretical foundations. For instance, they point to the role of ‘epistemic communities’ of experts in learning processes. This approach relies on the creation of intersubjectively held (causal) beliefs, and on the joint establishment of valid knowledge claims.

Influential accounts in policy diffusion, which are reflected in the stylized policy diffusion model, have mostly discarded this notion of learning. Instead, they take an objectivist and individualist view and conceptualize learning as a process of Bayesian updating, in which individual actors add new information to prior knowledge and beliefs and revise their behaviour accordingly (e.g., Meseguer, 2006). Relying on the individualism implied in Bayesian updating, the conceptualization of learning in the standard model has side-lined the role of collective actors in learning processes, which constructivist accounts highlight. Yet, sophisticated conceptual analysis provides strong arguments that policy learning is a multi-dimensional concept, which includes not only rationalist but also constructivist approaches (Dunlop & Radaelli, 2013). If that is the case, ‘learning’ functions poorly as a specific policy diffusion mechanism because it lacks conceptual distinctiveness.

When we look closely at the theoretical literature, we find that scholars have also conceptualized the mechanism ‘competition’ in much broader terms than depicted in the stylized model. In fact, Braun and Gilardi (2006, 308–309) call the mechanism ‘competitive and cooperative interdependence’. They show that the core element of such a mechanism is the external effect that a policy in one polity has on the effectiveness of the corresponding policy in another polity. In contrast to most diffusion studies, they also make clear that there are not only incentives to start a competitive regulatory race to attract mobile financial or human capital and to discourage perceived burdens like welfare recipients or asylum seekers but also incentives and opportunities to establish a joint standard through cooperative action.

However, the prevalent conceptualization of this mechanism equates a specific stimulus (i.e., policy externalities) with a certain kind of reaction (i.e., competition). It thereby emphasizes one basic structuration of present-day politics to the detriment of another basic structuration. It highlights the existence of a plurality of self-interested polities and presumes that policymaking and diffusion result from the fact that these polities compete for factors of production. It thereby relegates competition between groups that pursue policies in line with their specific identities and values, such as parties, a secondary role. If diffusion studies focus on parties, they are conceptualized as an intermediary factor that facilitates or filters the diffusion of policies (e.g., Butler et al., 2017). Only recently, parties have been assigned more fundamental roles in policy diffusion processes in as much as it is presumed that policy diffusion processes are linked to the formation of transnational party families (Senninger et al. 2021; Wolkenstein et al., 2020).

As with ‘learning’, the conceptual consistency of ‘coercion’ suffers from the fact that it has been defined by referring to diverse theoretical foundations. When Simmons and colleagues (2006, 790–791) describe the mechanism of coercion, they point to material capacities and resources alongside organizational and ideational ones as preconditions on which coercive processes of policy diffusion rely. These elements refer to different paradigms in the field of IR, which makes it unlikely that coercion is specified and operationalized consistently across empirical applications.

To overcome these inconsistencies in the conceptualization of ‘coercion’, most scholars describe this mechanism in a narrower sense as a process in which hegemonic states like the US use mainly informal ways to impose their policy preferences on weaker countries. Furthermore, they depict IOs like the IMF as channels of influence that powerful states instrumentalize to pursue national goals (Simmons et al., 2006). This description of coercion puts the mechanism apart from the other mechanisms and makes its inclusion into the set of diffusion mechanisms questionable. In fact, some argue that the difference is that coercion refers to an ‘involuntary’ form of policy adaption whereas the other mechanisms would reflect a ‘voluntary’ form of policy change (Maggetti & Gilardi, 2016). Not by accident, even authors who had been critical towards the inclusion of coercion into the set of diffusion mechanisms have overcome their scepticism by conceptualizing it in a slightly different way. Gilardi and Wasserfallen (2019, 3) point to conditionality as ‘the classic example of coercion’. A conditionality-based process of policy diffusion occurs, for instance, when the European Union makes membership or access to the common market dependent on the requesting states adjusting their policies to reflect those the members have already agreed. The main difference between the US pressure referred to by Simmons and colleagues (2006) and the EU demand is that the latter does not represent the stronger actor’s arbitrary interests, but the functional necessities of a political union or a common market. Furthermore, conditionality implies a formal relationship based on law and legal agreements. We argue that a theory of policy diffusion cannot exclude processes occurring in situations when an asymmetric constellation of influence characterizes the relationships among the involved polities. Nevertheless, at least when focusing on policy diffusion in and between liberal, constitutional democracies, it should conceptualize the asymmetry between polities as one that is grounded in law.

Finally, when it comes to ‘emulation’ (Gilardi & Wasserfallen, 2019, 4–5; Simmons et al., 2006, 799–801), we observe similar problems. Earlier conceptualizations that led to more differentiated and precise understandings were side-lined when the dominant understanding took hold. Braun and Gilardi (2006, 310–312) differentiate three mechanisms that are usually connected to sociological reasoning. For example, they presume that ‘common norms’ create shared beliefs about the effectiveness of specific means to reach a sought-after end, whereas ‘imitation’ works in altering the payoffs associated with different ends.

In an effort to reduce complexity, the dominant conceptualization has concentrated on a specific understanding of ‘emulation’ that draws heavily on the World Polity approach (Meyer et al., 1997). This approach points to the spread of Western culture across the globe and foregrounds one motivation when explaining this spread. Non-Western countries copy Western policies since they want to be recognized as modern polities. Very often, this leads to a superficial implementation and/or dysfunctional applications of these policies. This description pays less attention to diffusion that results if policies become widely legitimized through social practices and communicative procedures. For a more differentiated and precise specification of diffusion mechanisms that consider such processes, we can draw on conceptual developments in neighbouring fields like IR and Policy Studies, as we will show later on.

Consequences

The inconsistencies in the conceptualizations of diffusion mechanisms and the inductive reduction to the stylized model have consequences for their operationalization and measurement in empirical diffusion studies. As Maggetti and Gilardi (2016) reveal, the explanatory factors that we find in many diffusion models (e.g., geographic proximity, joint membership/interaction, success of policies, structural equivalence, numbers of previous adopters, trade flows) show no correspondence to the four mechanisms that are supposed to guide the explanatory endeavours. For example: ‘Geography cannot be linked straightforwardly to any of the […] mechanisms. It is a catch-all indicator that usually cannot discriminate between them’ (Maggetti & Gilardi 2016, 93). All mechanisms have been operationalized with different observable indicators. Moreover, most explanatory factors applied in diffusion models have been associated with different mechanisms.

We concur with proponents of the stylized model of policy diffusion that there is a need for reducing complexities and inconsistencies at the level of concepts. However, to establish consistent operationalizations and measurements of diffusion mechanisms, we argue that this reduction must follow a systematic approach that equally considers alternative options within core dimensions along which diffusion mechanisms can be structured. The existing set of diffusion mechanisms is unbalanced in two important dimensions (see Table 1).

Table 1. Inductively gained and patchy set of diffusion mechanisms.

	Rationalism	Social constructivism
Symmetric constellations	Learning Competition	Emulation
Asymmetric constellations	Coercion

We locate ‘learning’, ‘competition’, and ‘coercion’ at the boundary between rationalist and social constructivist approaches in order to signal that despite their strongly rationalist conceptualization in the dominant model of policy diffusion, they can also be (and have been) conceptualized with reference to social constructivist presumptions. Furthermore, we locate ‘emulation’ at the boundary between symmetric and asymmetric constellations in order to signal that this mechanism is usually conceptualized on the presumption of a symmetric constellation among the involved polities. Nevertheless, when we take the presumptions of social constructivist accounts serious (see the following section) and focus on the constellation among policies, ‘emulation’ would have to be placed rather in the asymmetric category.

It contains three mechanisms that are based on rationalist presumptions (learning, competition, coercion) and just one that corresponds to a social constructivist worldview (emulation). Furthermore, three mechanisms presume a symmetric relationship among the involved polities (learning, competition, emulation), and just one an asymmetric constellation (coercion). In the next section, we will systematically develop a typology of diffusion mechanisms that overcomes these asymmetries.

Foundations of a typological theory of policy diffusion

In order to overcome the conceptual inconsistencies that we have identified among the established mechanisms, we apply what we call ‘typological theorizing’. Typological theorizing has two implications. First, it entails a deductive and systematic approach towards differentiating the subtypes of a concept (in our case: mechanisms of policy diffusion) by reflecting explicitly on the dimensions and categories used to characterize and differentiate these subtypes.⁵ Second, it refers to an ideal-typical understanding of a concept, which implies that the various elements of a subtype (e.g., presumed context, motivations of involved actors as well as their behaviours) form a coherent whole. For example, we will argue in the following that it is theoretically coherent to presume that strategic actors consider primarily the impact of a policy change in respect to the domestically determined interests when they act in a context in which the polities they represent are embedded in a system of sovereign states. If the context is characterized by a legal order exhibiting vertical or horizontal stratification (through IOs or unions) among the involved polities, it seems consistent to presume instead that strategic actors focus on the existing and expected rights (and duties).

To overcome the biases and asymmetries that we diagnose within the established set of diffusion mechanisms, and in order to provide bridges to larger neighbouring fields of studies, we propose to take the following two fundamental dimensions and two principled alternatives in each dimension into account when we develop a typology of diffusion mechanisms:

1. rationalist as well as social constructivist understandings of structure and agency,

2. symmetric as well as asymmetric constellations among the basic units of analysis.

In the following, we show how a configurational combination of these dimensions and its principled alternatives leads to four paradigmatic foundations for understanding and explaining processes of policy diffusion (see Table 2).

Table 2. Paradigmatic presumptions of distinct forms of policy diffusion

	Rationalism	Social constructivism
Symmetric constellations	Interest-driven policy diffusion Structure: –Multiple (competing) polities legally established as equal jurisdictions with formal decision-making autonomy Agency: –Calculating the policy consequences for satisfying domestically determined self-interests	Ideology-driven policy diffusion Structure: –Multiple (competing) policies legitimized as necessary expressions of divergent values within pluralist societies Agency: –Interpreting the appropriateness of divergent policies for realizing internalized values/identities
Asymmetric constellations	Rights-driven policy diffusion Structure: –Polities legally established as unequal jurisdictions with distinct rights and duties Agency: –Justifying policy positions regarding established/aspired rights and duties in inner-polity and inter-polity realm	Recognition-driven policy diffusion Structure: –Policies legitimized through reflective or reflexive processes and actors Agency: –Identifying appropriate policies for the creation or consolidation of an established/aspired reputation for being responsible or responsive

The first dimension: rationalist versus social constructivist accounts

A rationalist account of policy diffusion focusses on relations between polities as basic units of analysis. Polities are formally constitutionalized as corporative actors through legal acts. They are represented by designated institutions (governmental organizations) and individuals (e.g., ministers). These polity representatives react to policy changes in other polities by evaluating potential policy changes in their own polity according to expected consequences for the goals, which have been derived through domestic political processes (self-interests). In other words, rationalist accounts are characterized by individualism (liberal nationalism), an instrumental view on policies, and a focus on corporate/legally constitutionalized actors.

In social constructivist accounts, in contrast, relations between policies, which form the basic units of analysis, take centre stage. Policies contain specific presumptions about problems, means and ends. These normative-cognitive features not only provide motivation and orientation for individual/corporative actors but the processes by which they are created and contested play a constitutive role in forming collective actors. Consequently, when such a worldview lies at the heart of a theory of policy diffusion, the formation of boundary-spanning (cross-polity, transnational) collective actors plays a major role. These boundary-spanning collective actors facilitate or filter the diffusion of policies across political boundaries by influencing the domestic will-formation and decision-making processes. At the same time, they are constituted by these processes of political contestation. In sum, social constructivist accounts exhibit a relational ontology, a constitutive view on policies, and a focus on collective actors, which are informally held together by shared norms and perceptions.

The second dimension: symmetric versus asymmetric constellations

In both rationalist and constructivist accounts, we consider diffusion processes that are structured by symmetric constellations of equal entities and those that are embedded in a symmetric constellation of unequal entities. In rationalist accounts, (a)symmetrical constellations among the involved polities are explicitly codified in legal texts (constitutions, laws, contracts, etc.). In social constructivist accounts, (a)symmetrical constellations among contested policies emerge as a result of deliberative procedures or social practices.

Configurations: structures and agency in four paradigmatic approaches

Suppose we apply a rationalist account in a symmetric context. In that case, we presume that all relevant polities are formally constitutionalized, through national or international law, as having equal status, and the competences to decide (de jure, but not necessarily de facto) autonomously on which policy to pursue in a specific policy field. In their problem definition and policy formulation processes, these polities take influences and information from other polities into account (so that policy diffusion can take place). However, their decisions are solely based on calculations about the consequences for their domestically derived self-interest. Such an account is compatible with the ‘pluralist’ presumption that the polity’s self-interest derives from the aggregation of organized interests in the domestic realm.

If we combine a rationalist account with an asymmetric constellation, we presume that legal regulations stipulate or imply a non-equal status among the involved polities. Within these structural contexts, it seems no longer consistent to assume that the polities’ representatives take only internally derived self-interests into account when they gauge their potential reactions to external demands for policy change. Instead, their considerations focus on existing and expected formal rights and duties in the inner- and inter-polity realms. Actors within the executive and legislative branches of government choose and justify their positions primarily regarding the expected (access and participation) rights and duties that come with an externally demanded policy change. Actors in the judicial branch of government assess the legal adequacy of the external demands and the domestic adjustments based on existing legal regulations.

A symmetric constructivist approach starts by emphasizing that the political order—at least within the Western liberal-democratic world—is not only characterized by a plurality of polities, but also by a plurality of groups that pursue competing policies in line with their specific identities and values. Within a liberal-democratic order, divergent policies are not only recognized as corresponding options for addressing a public problem. The plurality of proposed policies, as such, is crucial for the legitimacy of a democratic policymaking process since it allows the members of a polity (citizens) and their representatives to choose among alternatives.

A pluralist social and political structure is consistent with an understanding of agency in which the perception of problems and the formulation of policies are strongly influenced by the identities that actors have established and the values they have internalized. Actors draw on their particular identities and internalized values when comparing a current domestic policy’s appropriateness with policies pursued in other polities. In the context of diffusion processes within and among liberal democracies, it seems adequate to assign political ideologies the role of providing interpretative coherence (Gerring, 1997).

Asymmetric social constructivist approaches, in contrast, build on the presumption that even within pluralist societies and liberal democracies, a particular policy’s dominance over other policies can not only be established (at least for a certain time) but also legitimized. This can take place through social practices and/or communicative procedures. ‘Reflexive’ (in the sense of reflex-like) social practices can play a crucial role in policy diffusion when a shift in public attention suddenly give a certain policy a prior place on the political agenda (Jones & Baumgartner, 2005). In such situations, policymakers, who aim to be perceived as ‘responsive’ to the will of the people, try to find ‘quick-fixes’ and are open to importing promising solutions from other polities or policy fields. The extensive, intensive, and systematic forms of information processing and argumentative exchanges, which are supposed to occur among scientists and experts, characterize ‘reflective’ communicative procedures. Policy diffusion occurs when policymakers—who aim to be perceived as ‘responsible’ problem solvers—search for sound solutions and import policies, which have been evaluated by experts and identified as ‘best practices’.⁶

Prototypical pathways of policy diffusion

Having laid out the fundamental presumptions on which our ideal-typical forms of policy diffusion are based, we proceed by specifying two prototypical pathways of policy diffusion for each ideal-typical mechanism. Methodologically speaking, we go down the latter of abstraction and move from the four ‘generic’ mechanisms of policy diffusion to more specific subtypes (Dunlop & Radaelli, 2013, 600). In line with the systematic account of typological theorizing, we do this by applying the same criteria in all four generic types. We distinguish between policy diffusion pathways focused on a particular policy field and those pathways where interdependencies and interferences between policy fields play a crucial role. This distinction refers to a major aspect of how theories of the policy process differ. The Advocacy Coalition Framework (ACF) is an example of a theoretical approach that presumes the former, whereas the Theory of Punctuated Equilibrium (Baumgartner et al., 2018) highlights the latter. This distinction is also at the forefront of debates located at the intersection of policy studies and IR (e.g., Blatter 2009; Lavenex, 2014). Our attempt to specify the ideal-typical mechanisms through prototypical pathways not only includes such a structural element but also leads to a stronger specification of agency.

Furthermore, the prototypical actors that we take as concrete expressions of the various social constructivist mechanisms build bridges to the mentioned neighbouring fields. Nevertheless, we would like to emphasize that our own interpretation structures the alignment of these prototypical actors to our pathways. We do not claim that the pathways correspond to entire theories or frameworks. For example, proponents of the ACF argue that their ‘model of the individual’ takes both the ‘logic of appropriateness’ and the ‘logic of consequences’ into account to overcome the classic conflicts between economists and sociologists. Nevertheless, they position their model clearly and explicitly in opposition to rational choice theory (Sabatier & Weible, 2007, 194). Furthermore, when we try to grasp what lies as the heart of the ACF and what makes this framework distinct compared to other policy process theories, it is the presumption that collective actors are formed by shared ‘normative beliefs’ (Sabatier & Weible, 2007, 194). In consequence, for us, advocacy coalitions correspond to a social constructivist account. Nevertheless, whereas advocacy coalitions correspond to a specific pathway, the ACF as a comprehensive framework for the study of policy change does not. The ACF comprises three levels of beliefs. In contrast to the ACF, we assign two of them to distinct pathways. Advocacy Coalitions are held together by ‘policy core beliefs’ (Sabatier & Weible, 2007, 195), which makes them perfect prototypical actors within a pathway that is centred around a policy field. ‘Deep core beliefs,’ in contrast, come close to what we call ‘ideologies.’ In consequence, in our theory, this level of beliefs corresponds to a pathway that is not confined to policy fields and to a distinct prototypical actor – (transnational) party families.

Interest-driven pathways: externalities and information exchanges

The two interest-driven pathways share the presumption that governments are the pivotal actors and that these actors take their strategic decisions based on domestically derived self-interests.

A first prototypical pathway puts the exchange of information among state governments at the heart of the diffusion process. Such a process is triggered by the increased problem pressure a government faces due to an external event (e.g., an international economic or humanitarian crisis), which affects many other states as well. In this pathway, it is crucial that governmental or regulatory agencies, which are tasked to formulate potential reactions to the increasing problem pressure, collect information about the related policies of other polities. This matters not only for distinguishing it from other pathways, but also from ‘non-diffusion’ (the situation in which polities adopt similar policy responses simultaneously, without the exchange of information). Afterwards, the ministry selects the policy that is perceived to serve the polity’s best interests and adapts it to fit its particular circumstances. This pathway has affinities to the established understanding of the ‘learning’ mechanism. However, it is much more precise, especially because—in contrast to the externalities pathway—it locates the moment of cross-border influence in the policy formulation stage and stipulates that the information exchange occurs within a specific policy domain. The latter implies that particular agencies or policy-field specific ministries are the prototypical actors within this pathway.

We call the second interest-driven process of policy diffusion an externalities-based pathway. It starts with a policy change in one polity and presumes that potentially mobile non-state actors (e.g., investors or asylum seekers) compare policy across polities and adjust their ‘movements’ accordingly. If the policy change produces ‘externalities’ in other polities in the sense that it undermines policy effectiveness, the latter governments react by adjusting their policies. A crucial difference between the first and second interest-driven pathway is that the policy outcome of a polity influences the problem definition of other polities within the latter pathway, whereas in the former, it influences the evaluation of policy solutions. Furthermore, in contrast to the established definition of ‘competition’, it does not mix up stimulus and response. It makes us aware that the principal decision that reacting governments have to make is whether to respond in a cooperative or a competitive manner. Consequently, it is very likely that this decision is strongly influenced by the general relationship between the polities and not by specifics of the policy domain. This, in turn, implies that the entire government, including the Ministry of Foreign Affairs, plays a crucial role within this explanatory pathway.

Rights-driven pathways: hierarchy and conditionality

The two rights-driven pathways share with the interest-driven pathways the presumption that governmental representatives of polities are the pivotal actors and that they act strategically. However, they presume that these actors take and justify their positions and decisions not concerning socio-economic interests, but regarding established and expected rights and duties.

The hierarchy-based pathway draws on the rights and duties that polities have assigned through a legal framework (constitution, law) to different levels of government within a specific policy domain. A full-fledged approach would include processes in which existing policies on a lower level influence policy formulation on a higher level. However, due to space restrictions, we limit our sketch to processes that start when governments on a lower level are faced with (or have to reckon with) a legally backed demand to adjust their policy to common standards. In its most prototypical form, this demand is formulated by a superior court, which rules that the policy in a subordinated polity does not comply with the joint standard.

In parallel, in the policy field of the inner-polity realm, specific veto players (actors who must approve any policy change under domestic constitutional rules before it can proceed, according to Tsebelis, 2002) formulate their positions in respect to how (much) the polity should comply with the ruling by referring to established rights and duties. Ministries—which are the prototypical actors within this pathway and are confined to a particular policy domain—then seek to adjust policies such that they can simultaneously convince the superior court that they now sufficiently align with the joint standard and the domestic veto players that changes deviate from their positions only as much as legally necessary.

In contrast, the conditionality-based pathway starts when a state asks for membership in an IO. In response, the organization demands (among other things) a change in a particular policy as a precondition for membership so that the policy fits the joint standards. In contrast to the hierarchy-based pathway, a larger spectrum of veto players is involved. They formulate their positions not only regarding rights and duties in the policy field, but also vis-à-vis all kinds of expectations that can be aligned to membership (policy changes in divergent policy fields with the associated costs and benefits, but also rights and duties for the individual or corporative members as well as for the representatives of member states). Consequently, it is not a specific ministry that plays a central role, but the entire government. A major policy change is possible within such a pathway if the broader considerations outweigh opposition within a policy field.

Ideology-driven pathways: principled beliefs and policy beliefs

Ideology- and recognition-driven pathways differ from interest- and rights-driven ones in as much as those collective actors that transcend the boundaries of polities play crucial roles. Ideology-driven pathways presume that these transnational collective actors are held together by internalized beliefs.

Transnational Party Families (TPFs)—often loosely organized in (con)federational associations—are the prototypical actors in pathways that assign principled beliefs a crucial role. The meetings and institutionalization of party families not only strengthen the ideological coherence among their national members—which leads to a similar positioning (in respect to the left-right dimension) in many policy domains—but also stabilizes the ideological cleavages that separate the members of different party families in the national realm (Camia & Caramani, 2012). In their meetings, TPF members exchange ideas about aims and means and help to bring the aims and means of their members in line with their shared principles (Wolkenstein et al., 2020). In other words, TPFs function as hinges for the transfer of policies across polities, but they also deepen the party cleavages within polities. Consequently, a change in government that brings a party from a different TPF into power is a necessary precondition for policy diffusion along this pathway.

A pathway that puts shared policy beliefs centre stage has to focus on Transnational Advocacy Coalitions (TACs), which are different kinds of actors who share particular interpretations of a policy problem, its causes, and appropriate policy instruments (Jenkins-Smith et al. 2018, 140). Combining crucial presumptions from the Transnational Advocacy Network (TAN) approach (Keck & Sikkink, 1999) and the ACF, such a pathway presumes that both the transnational formation of advocacy coalitions and the rivalry between antagonistic coalitions are necessary conditions for policy diffusion. Transnational meetings and exchanges allow TAC members to familiarize themselves with policies applied in other countries and interpret them as (in)appropriate. Such interpretative and evaluative exchanges, in turn, help the members of TACs to be successful in the domestic realm by bolstering their framings of policies accordingly. In consequence, contrary to the pathway based on principled beliefs, within this pathway policy diffusion is not dependent on exchanging decision-makers (parliamentary majorities and governments through elections) but relies on a switch in the frame that dominates the discourse within a policy domain.

Recognition-driven pathways: expertise and attention

While in ideology-driven pathways transnational collective actors are connected through specific substantial beliefs, in recognition-driven pathways, they share convictions regarding the procedures by which policies should become recognized solutions, or regarding the promises that are associated with particular solutions.

Within a pathway that builds on policy expertise, Transnational Epistemic Communities (TECs) take centre stage. TECs differ from TACs in that their members are not united by sharing substantial policy beliefs, but a belief in the merits of ‘reflective procedures’ of knowledge generation (Farquharson, 2003, 83). Their influence does not rely on their framing capacities, but on their recognition as authoritative sources of scientific knowledge or expertise in a specific policy field (Haas, 1992). In contrast to an information-based pathway, which presumes a rising ‘objective’ problem pressure as a starting point for the process of policy diffusion, an expertise-based pathway applies an ‘intersubjective’ understanding of knowledge. Such a process is triggered when a TEC has accumulated enough knowledge to overcome uncertainties about the problem’s relevance. An information-based pathway assumes that individual polities collect information about potential solutions by screening the reactions of similar polities. In contrast, the expertise-based approach assigns agency to TECs, who collect and compare existing policies and systematically evaluate and rank these policies. Policy diffusion occurs when governments decide to adopt a policy that has been singled out as ‘best practice’ by a TEC. Especially in those cases where individual governments do not have the capacity and the time to evaluate the effectiveness or efficiency of policies as it is the case in information-based processes, but nevertheless want to present themselves as ‘responsible’ problem solver they are open to follow the prescriptions of TECs.

Within a second pathway that focuses on popular attention, the formation of Transnational Instrument Constituencies (TICs) is the corresponding condition for making policy diffusion across polities possible. TICs differ from TECs in as much as their members are not held together by the belief in reflective/scientific procedures but by the expectations and promises associated with a particular kind of policy instrument (Simons & Voss, 2018, 21–22). Within an attention-based approach, a policy issue is set on the national political agenda through the media system (and opinion surveys), indicating to political decision-makers which problems the public perceives as the most pressing. It is theoretically consistent to assume that the policy formulation stage is not characterized by ‘reflective procedures’—as occurs with the information- or expertise-based approaches—but by ‘reflexive practices.’ This is especially true in contexts in which a policy issue suddenly achieves high political salience and a prominent place on the political agenda (which may or may not coincide with objective problem pressure or the provision of scientific evidence). When political decision-makers want to show that they are ‘responsive’ to their constituency, TIC represents prototypical actors since they offer quickly available solutions. Thereby, TICs can refer to successful implementations of their proposed policies in other polities to convince decision-makers to support importing these policies, making them central hinges for this kind of policy diffusion.

Table 3 sums up all paradigmatic pathways and highlights prototypical actors for each pathway. Please note that these actors are not the only actors that play a role within these pathways, but they function as differentia specifica for distinguishing the pathways. In their specific pathway, they play a crucial role, whereas that is not necessarily the case within the other pathways.

Table 3. Paradigmatic pathways and prototypical actors of policy diffusion.

Diffusion mechanisms	Paradigmatic pathways	Prototypical actors
Interest-driven	Exchange of information	Agencies in the policy field
	Externalities	Ministries of Foreign Affairs
Rights-driven	Hierarchy	Ministries in the policy field
	Conditionality	Prime Ministries/Entire Governments
Ideology-driven	Principled beliefs	Transnational Party Families
	Policy beliefs	Transnational Advocacy Coalitions
Recognition-driven	Policy expertise	Transnational Epistemic Communities
	Popular attention	Transnational Instrumental Constituencies

Consistent operationalization in and across research designs and methods

The systematically derived typology of diffusion mechanisms and pathways not only facilitates the accumulation of empirical knowledge across divergent policy fields and various kinds of polities. It also helps to develop conceptually consistent research designs and programs that include different kinds of methods. To provide some first evidence for the latter claim, we show how the eight policy diffusion pathways can be operationalized in three different research methods. We indicate how they help a) to determine indicators that are applied in a statistical model of policy diffusion, b) to formulate informational manipulations in experimental designs, and c) to deduce expectations that are traced by using within-case analytic methods like Process Tracing and Congruence Analysis. Due to space restrictions, we present these operationalizations in an online appendix.⁷ Together, the three operationalizations illustrate how the paradigmatic mechanisms and the specified pathways help to make multi-method research conceptually consistent. Furthermore, the expectations that we deduce for within-case analysis illustrate how their conceptual foundations help to deduce a plurality of internally coherent and externally distinctive expectations across the entire process of policy diffusion.

Summary and conclusion

There is no doubt that research on policy diffusion has generated significant insights. In our contribution, we argue, however, that the inductively derived set of diffusion mechanisms that has taken hold in diffusion studies needs to be theoretically revisited. For the four established mechanism—learning, competition, emulation, and coercion—we find quite different characterizations and theoretical foundations. This has led to inconsistent operationalisations and measurements. Stylized models of policy diffusion have been able to reduce the conceptual ambiguity of individual mechanisms, but at the price that the overall set of diffusion mechanisms takes neither rationalist and social constructivist theories, nor symmetric and asymmetric constellations equally into account.

In order to overcome these problems, we develop, in a systematic way, a typology of paradigmatic mechanisms of policy diffusion. We start by making explicit the dimensions and categorisations that are used for defining and differentiating the individual mechanisms. Building on that, we distinguish between four mechanism—interests-, rights-, ideology-, and recognition-driven diffusion processes—which can be further specified into more fine-grained pathways. This typology of prototypical diffusion mechanisms does not only overcome the identified biases of the existing set of mechanisms. We show that it also leads to individual mechanisms, which are specified in an internally coherent and externally distinct way. This, in turn, facilitates their consistent operationalization within and across research designs and methods.

Our typology could also be helpful to bring together the divergent strands of research. It might be able to build bridges to scholars who approach the spread of policies with different labels and theoretical foundations (e.g., ‘policy mobilities’). This hope is based on the fact that our typology includes social constructivist accounts to policy diffusion in a differentiated and sophisticated way, without giving up the presumption hold by positivists and critical realists that explanatory approaches can be systematically tested by empirical analysis (the latter is quested by those constructivists who insist that their approach is based on a distinct epistemology). Furthermore, the typology might be a stimulus for reflecting more systematically on the conceptual overlaps that exist between the studies of policy diffusion and neighbouring fields like International Relations and Policy Studies. Policy diffusion studies has established itself as a vibrant field of research on its own. Yet, we are convinced that establishing new fields of research should not be accompanied by ‘silo building’, but by reflective exchanges with the larger fields of studies from where they have been emerging.

Finally, we would like to reflect on applications and scope conditions of our typological theory of policy diffusion. The theory highlights the role of transnational and national actors in diffusion processes. Yet, we are convinced that the mechanisms and pathways that we have outlined can directly be translated to explain diffusion processes at the subnational level of federal states as well as diffusion processes that take place across cities. It is, however, a theory of policy diffusion within and among liberal democracies. This is evident in our specification of diffusion mechanism: in rights-driven processes of policy diffusion, legal constitutions or contracts are at the core; the dominance of policies in recognition-driven processes results from procedures that lead to responsible and responsive policy-making; and in ideology-driven diffusion, we highlight the role of Transnational Party Families. For studying diffusion processes that involve autocracies, a coherent adaptation would presume, for example, that diffusion results from (competitive) formation of international alliances among states (based on their shared regime identity) rather than Transnational Party Families. We hope that our theory thus paves the way for a reflective and careful adaptation for studying processes of policy diffusion among countries with different regime types.

Acknowledgements

We would like to thank Angie Gago, Daniel Ghezelbash, Martino Maggetti, Andreas Tunger-Zanetti, and two anonymous reviewers, as well as participants of the nccr – on the move Research Day in October 2019 and participants of the Annual Meeting of the Swiss Political Science Association 2020 for helpful advice and feedback. We appreciate Simon Watmough’s excellent support in editing the text. This research was supported by the National Centre of Competence in Research nccr – on the move, funded by the Swiss National Science Foundation (grant number: 51NF40-182897).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung: [Grant Number 182897].

Notes on contributors

Joachim Blatter

Joachim Blatter is a professor of political science at the University of Lucerne, Switzerland and a member of nccr – on the move, National Center of Competence in Research (NCCR) for migration and mobility studies, Neuchatel, Switzerland.

Lea Portmann

Lea Portmann is a postdoctoral researcher at the University of Lucerne, Switzerland and a member of nccr – on the move, National Center of Competence in Research (NCCR) for migration and mobility studies, Neuchatel, Switzerland.

Frowin Rausis

Frowin Rausis is a doctoral student at the University of Lucerne, Switzerland and a member of nccr – on the move, National Center of Competence in Research (NCCR) for migration and mobility studies, Neuchatel, Switzerland.

Notes

1 In using the label ‘policy diffusion’, we actually refer to the entire field of studies analysing the processes that drive the spread of policies across polities, including research that applies the concept of ‘policy transfer’ (e.g., Dolowitz & Marsh, 1996), but also work of scholars who prefer to talk about ‘policy mobilities’ (e.g., Peck & Theodore, 2010), or ‘policy translation’ (e.g., Mukhtarov, 2014).

2 Major exceptions are the contributions of Braun and Gilardi (2006) and Weyland (2007).

3 Sometimes emulation is called ‘imitation’ (Shipan & Volden, 2008, 840), or ‘socialisation’ (Graham et al., 2013, 690).

4 Several scholars (e.g., Braun & Gilardi, 2006, 309; Elkins & Simmons, 2005) exclude ‘coercion’ as a diffusion mechanism because they equate policy diffusion with processes of ‘horizontal’ or ‘uncoordinated interdependence’.

5 Dunlop and Radaelli (2013) do not only exemplify such an approach when they systematize the various forms of ‘policy learning’, but they also point to the underlying methodological literature on explanatory typologies.

6 ‘Reflective procedures’ correspond to what Kahneman (2011), a psychologist, calls a slow mode of thinking; ‘reflexive practices’ correspond to a fast mode. Nevertheless, we focus not on the individual, but on the social level.

7 The appendix can be found at: https://www.tandfonline.com/doi/suppl/10.1080/13501763.2021.1892801.