UNDERSTANDING POLICY TRANSFER: A MULTI-LEVEL, MULTI-DISCIPLINARY PERSPECTIVE

UNDERSTANDING POLICY TRANSFER:A MULTI-LEVEL, MULTI-DISCIPLINARYPERSPECTIVE

MARK EVANS AND JONATHAN DAVIES

At the same time that comparative and international political scientists have beenconfronting the problems of analysing state behaviour under conditions of uncer-tainty, state-centred political scientists are attempting, somewhat belatedly, to dealwith the increasing complexity and uncertainty which underpins modern govern-ance. Yet despite similar research agendas these disciplines have continued to speakpast each other. This article contends that policy transfer analysis can provide acontext for integrating some key concerns of these disciplines. Further, we arguethat the process of policy transfer should be examined through a structure andagency approach with three dimensions: global, international and transnational lev-els, the macro-level and the interorganizational level. This three-dimensional modelemploys the notion of a policy transfer network as a middle-range level of analysiswhich links a particular form of policy development (policy transfer), micro-decision making in organizations, macro-systems and global, transnational andinternational systems. It is hoped that this approach will stimulate an empiricalresearch agenda which will illuminate important policy developments in domesticand world politics.

INTRODUCTION

In recent years there has been a growing interest in the study of policyconvergence (Coleman 1994), policy diffusion (Majone 1991), policy learn-ing (P. Haas (ed.) 1992) and lesson drawing (see Rose 1991). All refer todifferent forms of policy transfer – ‘a process in which knowledge aboutpolicies, administrative arrangements, institutions etc. in one time and/orplace is used in the development of policies, administrative arrangementsand institutions in another time and/or place’ (Dolowitz and Marsh 1996,p. 344). These studies encompass a host of disciplines ranging from dom-estic and international political science to comparative politics; the studyof policy transfer has a truly multi-disciplinary character. Yet what is com-monly viewed as a strength can also be identified as a weakness. As aconsequence of the diffuse nature of this field of study, policy transfer ana-lysts do not have the benefit of a common idiom or a unified theoretical ormethodological discourse from which lessons can be drawn and hypothesesdeveloped. Indeed, despite complementary research agendas, these disci-plines have continued to speak past each other. Our starting point, and also

Mark Evans is Lecturer in Politics and Jonathan Davies is a D.Phil. student in the Department ofPolitics at the University of York.

Public Administration Vol. 77, No 2, 1999 (361–385) Blackwell Publishers Ltd. 1999, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street,Malden, MA 02148, USA.

362 MARK EVANS AND JONATHAN DAVIES

our conclusion, is that policy transfer analysis can provide a context forintegrating common research concerns of scholars of domestic, comparativeand international politics. Further, that the study of policy transfer isimportant not just because of its multi-disciplinary potential, but becausewhen considered inductively, processes of globalization appear to haveincreased policy transfer. Consequently, the aim of this article is to improveour understanding of the policy transfer phenomenon and to determine thetheoretical and methodological utility of a multi-level, multi-disciplinaryapproach to policy transfer analysis.

It is important at the outset to brush aside one or two misconceptionsthat might arise from the discussion which follows. The nature of the sub-ject matter means that this article is speculative and is offered as a guideto further research. We merely seek to determine the key research questionsthat policy transfer analysis should address if it is to move towards a modelwhich has what we will term additionality. Specifically, it may be asked ofpolicy transfer, what does it tell us which we didn’t know before? MaxWeber once cautioned: ‘Consider the historical and cultural sciences . . .they give us no answer to the question, whether the existence of these cul-tural phenomena have been and are worthwhile. And they do not answerthe further question, whether it is worth the effort required to know them’(Gerth and Wright-Mills 1948, p. 145).

A further problem requires some consideration. The multi-organizationalsetting in which policy transfer tends to take place, coupled with the multi-disciplinary nature of its study, has inevitably meant that we have had tojuggle with a range of concepts in order to deepen our understanding ofthe phenomenon and sharpen the research questions we need to pose.Consequently we may justly be accused of being too concept-heavy. Indefence of this eclecticism, we hope that from our consideration of the pro-cess of policy transfer through the lenses of the ‘international’, the ‘transna-tional’, the ‘interorganizational’ and the ‘domestic’, an important researchagenda will emerge which integrates the common concerns of scholars ofboth domestic and international politics. It is also intended that the conceptswe utilize will be framed in such a way that their meaning will be precise,their domain of utility apparent and their implications clear.

The main research questions to be addressed in this article are therefore:what domestic, international and transnational circumstances are likely tobring about policy transfer and how?; what are the scope and dimensionsof policy transfer?; and which aspects of the framework should and shouldnot be pursued in empirical work.

The article is divided into three main sections. In the first section somecrucial prerequisites to modelling policy transfer are reviewed. The secondsection then outlines our own multi-level approach to policy transfer analy-sis which represents an attempt to transcend some of the problems therein.Here the article builds upon and integrates five approaches which havedeveloped independently from each other and have never been brought

Blackwell Publishers Ltd. 1999

UNDERSTANDING POLICY TRANSFER 363

together as an analytical ensemble – international structure and agency andthe epistemic community approach, domestic structure and agency, policynetwork analysis and formal policy transfer analysis. They all deal withstructures of governance and forms of agency, they have all generated fruit-ful empirical research and, hence, in combination can enlighten the scopeof our enquiry. In the third section some crucial questions of validationand evaluation are addressed.

PROBLEMS AND THE WAY FORWARD

We begin by looking at the knowledge claims of policy transfer analysis.Minimally, a theory is ‘a systematically related set of statements, includingsome law-like generalizations, that is empirically testable’ (Rudner 1966,p. 10). Given this definition, policy transfer analysis does not constitute anexplanatory theory but may be viewed as an analogical model in the sensethat it refers to the suggestion of substantive similarities between two enti-ties. For example, likening an organization to a machine or an organism.It may also be viewed as a meso-level concept which can provide a linkbetween the micro-level of analysis, which deals with the role of interestsand levels of government in relation to particular policy decisions, and themacro-level of analysis, which is concerned with broader questions con-cerning the distribution of power within contemporary society.

The 1990s have witnessed an upsurge of interest in meso-level analysisin British Political Science (see Dowding 1995). The reasons for this are bothpolitical and intellectual. Politically it has been a response to what Rhodes(1996, p. 652) has termed ‘the New Governance: Governing withoutGovernment’, which refers to policy making through multi-layered, self-organizing, interorganizational networks. Intellectually it reflects the chal-lenge of studying the New Governance. The meso-level is exalted as themost fertile level for analysing policy making in Britain for two mainreasons. Macro-level theories are often abstract and frequently applied toconcrete situations with little attention to mediating processes, while micro-level theories tend to ignore the impact of broader structural factors onmicro- decision-making settings. Hence operating at the meso-level acts asa corrective device for ensuring that policy scientists don’t lose sight of themacro- or micro- level questions, while simultaneously observing that muchpolicy making takes place within multi-layered, self-organizing, inter-organizational networks. Meso-level analysis has become the crucial ana-lytical tool for multi-level, integrative analysis.

Dolowitz and Marsh (1996) have led efforts within British politicalscience to develop a model of policy transfer. In essence, they have drawntogether a general framework of heterogeneous concepts including policydiffusion, policy convergence, policy learning and lesson drawing underthe umbrella heading of policy transfer which mainly draws on the workof Rose (1991, 1993), Bennett (1991a, b), Robertson (1991) and Wolman(1992). Policy transfer is thus used as a generic concept which encompasses



quite different claims about the nature of policy development. Rose, forexample, characterizes lesson drawing as ‘voluntaristic’ (1991, p. 9),whereas policy diffusion and policy convergence tend to be associated with‘structural’ dynamics which Rose associates with ‘technocratic determin-ism’ (ibid.). Alternatively, Wolman argues that similar policies can arisein different countries either from common process or through consciousimitation; both structure and agency. As Eyestone puts it, [d]iffusion pat-terns may reflect either the ‘spread of necessity’ or the ‘emulation of virtue’(1977, p. 441).

Dolowitz and Marsh (1996) suggest that all these phenomena occur andcan be organized into one framework as ‘dimensions of policy transfer’.Thus lesson-drawing is categorized under the sub-heading ‘voluntarytransfer’ and structured change is categorized within ‘voluntary’, ‘percep-tual’ and ‘direct’ or ‘indirect’ coercive policy transfer. Dolowitz and Marshhave provided an extremely useful framework which invites others to critic-ize and develop it – a map of the process of policy transfer which can onlyever be a representation of a reality which needs to be proved or disprovedin an objective sense. Indeed it is at best a heuristic model employed inthis case for cognitive purposes to suggest something about the propertiesand relations understood to exist within the process of transfer. Usuallysuch models are invoked to suggest something about disparate, often com-plex phenomena – characterizing group-government relations as webs,whirlpools, triangles, or iron triangles. However, it does suggest theinability of the approach to determine with precision the phenomenon it istrying to explain. As James Gregor (1971, pp. 193–4) puts it, these are ‘sus-tained metaphors’, ‘promissory notes’ on theories to emerge from withinthe analytical process. In short, policy transfer analysis does not have fullexplanation and theory status. This would require the development of acausal model based upon a series of propositions which may be validatedor falsified by evidence.

This argument does not negate the importance of the policy transferapproach. A sound model is not necessarily one that purely explains orpredicts with precision. It is one rich with implications. Brian Barry argues,

Our understanding of a subject may be advanced if concepts and pro-cesses can be translated into other terms more readily grasped and fruit-ful analogy will suggest new lines of enquiry by provoking the specu-lation that relationships found in the one field may hold, mutatismutandis, in the other as well (1975, p. 86).

Novel hypotheses may be extracted from the policy transfer approach andmust themselves be articulated in a systematic fashion and be subjected toempirical testing. But in order to make stronger knowledge claims it mustengage in theoretical and methodological pluralism and integration.

The framework developed by Dolowitz and Marsh is clearly designed toincorporate a vast domain of policy-making activity by classifying all poss-



ible occurrences of transfer, voluntary and coercive, temporal and spatial.Policy transfer is common in this scheme and processes such as the ‘rapidgrowth in communications of all types since the Second World War’ haveaccelerated the process (1996, p. 343). This is not, however, a claim thatpolicy transfer is all-pervasive and Dolowitz and Marsh make no commenton the scope for policy innovation, nor whether policy transfer falls entirelywithin the parameters of policy succession. We discuss the question ofboundaries to the study of policy transfer further below. However,although Dolowitz and Marsh have done a great service in organizing afragmented literature into a coherent whole and suggesting their own pre-ferred interpretation a number of problems remain.

At the level of global, international and transnational structuresFirst, the research is at its weakest when it considers the questions of towhat extent and why policy transfer has become widespread throughoutwestern democracies in the course of the past two decades. We need empiri-cally to examine whether recent manifestations of policy transfer are dis-tinct because they have sprung from quite unique global and domestic insti-tutional and socio-economic conditions. The two questions we need toaddress here are: has there been an increase in policy transfers? If there has,is the increase due to globalization processes or are they the most recentmanifestation of a long-term modernization process in the West.

At the level of domestic structuresThe relationship between state structures and agency remains underdevel-oped. There is a tendency towards contextualizing the macro-level factorswhich affect transfer (i.e. a failure to articulate a link between variables).This must be resisted and replaced with a proper integration of levels ofanalysis.

At the interorganizational levelThere needs to be some account of how interorganizational relationshipsdevelop and help to precipitate processes of transfer. This brings us inexor-ably to network analysis which provides us with a tool for studying theinteractive process of policy transfer, together with a well-developedcharacterization of relationships within networks which can help us to ana-lyse the effects of structure and agency at any level of governance (seeMarsh and Rhodes 1992b). In addition, the epistemic community approachprovides us with a rich source for evaluating the role of knowledge elitesas agents of policy transfer pushing for new or changed international prac-tices and institutions nationally, transnationally and internationally (see P.Haas (ed.) 1992).

At the level of policy development – the boundary questionOne cannot deduce from the literature claims concerning the inclusivenessof policy transfer as a theory of policy development. To be fair, Dolowitz



and Marsh do not concern themselves with this but rather with the pro-cesses which lead to the transfer of specific policies (1996, pp. 343–4). How-ever, as suggested above, it is necessary to classify the phenomenon byscope. Much policy development literature suggests either directly orindirectly that forms of transfer are common and on the increase. Hogwoodand Peters (1983), for example, assert that policy succession is increasinglypervasive in the sense that innovation can only be said to occur if an ideaor a programme is entirely new. There are, they argue, few, if any policyareas in which governments are not active. These arguments certainlyimply that ‘policy transfer’, at least from an organization’s past, will beextremely common. Rose’s voluntaristic approach to lesson drawing alsosuggests that unique policies and programmes are rare although he doesnot discount major change – particularly where lessons are drawn trans-nationally (1993, pp. 120–36). Rose doesn’t state whether this leads him toconclude that lesson drawing is as pervasive as Hogwood and Peters argueof policy succession. However, he argues that the first logical response ofa policy maker in attempting to deal with a problem will be to look forsimilar examples elsewhere.

These approaches beg the question: if policy transfer has become suchan every-day part of policy development, when does policy transfer analy-sis cease to be a meaningful exercise? Rose is clear that lesson drawing isan intentional exercise involving research – suggesting that personal pastexperience does not count. Thus, a policy maker who moves from organiza-tion A to organization B and then draws instinctively or deliberately uponsome fragment of his/her past experience in organization A cannot be saidto have drawn a lesson. Rose’s caveat is important for otherwise there isno boundary to what counts as lesson drawing or policy transfer. Evenpolicy innovations will be bound to rely on prior knowledge to somedegree. It would, in any case, be impossible to identify lessons at this levelof detail. No group or individual can reasonably be asked to identify allthe influences which led them to take a particular complex decision.

Thus, for the purposes of this article, policy transfer is defined in Rose’sterms as an action-oriented intentional activity. The implications of this pos-ition are discussed further below, but it is suggested that this definitionmight apply equally to the voluntary and coercive dimensions of policytransfer since intent may be ascribed both to those who seek to borrow andto those who seek to impose. This cautionary note illustrates that policytransfer must become more distinctive and purposeful if it is not simply torearticulate other ideas. It must justify itself in the theoretical domain.

Argument summaryThe survival of policy transfer as a model relies on its ability to be adaptedinto a multi-level, multi-disciplinary perspective. Presently the compositeliterature on policy transfer analysis offers us a series of dependent vari-ables which make up a definitional criteria and a check list of categories.



We need to advance beyond this and develop a conception of policy trans-fer analysis which both recognizes the importance of global, international,transnational and domestic structures and their ability to constrain and/orfacilitate policy development and also understands that policy transfer maypurely be a product of interorganizational politics. By implication, policytransfer analysis can become heuristically useful at each of these levels ofanalysis if it provides an approach which deals with the complex interactionbetween structure and agency.

Prerequisites for modelling policy transfer

A focus on the spatial domainAnalysing the scope of the policy transfer framework necessarily dependsupon assumptions about its purpose which may not be shared by itsauthors. However, we make a number of assumptions concerning the limitsof policy transfer as a tool of research and analysis. It is not an inclusivemodel of policy development and it is not a model concerned with thegeneral diffusion of knowledge among humankind. It is best concernedwith discernible and remarkable features of contemporary policy change.We are sceptical about the utility of intra-organizational policy transfer, inparticular its temporal dimension. This view is predicated on four basicassumptions.

First, we argue that policy transfer is a model of policy change. It istherefore better focused on identifying processes of change than on themeasurements of continuity and change which intra-organizational trans-fers point toward. Second, it is assumed that policy transfer must seek toidentify and classify remarkable phenomena not otherwise explained. Here,it is suggested that the day-to-day diffusion of knowledge, intentional orotherwise, at the micro-level within organizations is not remarkable in thecontext of policy transfer either in terms of process or of fact. These trans-fers are better the subject of organizational analysis or management studies.Third, when remarkable intra-organizational transfers occur it is likely tobe extra-organizational factors which are remarkable, rather than the intra-organizational process of transfer. Fourth, there is a presupposition thatpolicy transfer in international dimensions is most likely to be remarkablein falling outside other explanatory frameworks. Here, the argument is sup-ported by much of the prior and contemporary literature which focuses oninter/transnational transfers (see Wolman 1992).

This brief outline proposes that the temporal dimension of policy transferis likely to be remarkable only in the context of a spatial variable, whichwill tend to exclude most intra-organizational transfers from consideration.A qualification to this proposal is the potential for intra-organizationaltransfer analysis across spatial dimensions, although this does not seemappropriate to current institutions of governance either at the supranationalor intranational levels. By a process of elimination, it may be concluded



that the great majority of remarkable policy transfers will occur betweendistinct organizations at the international or transnational levels.

Five levels of political spatiality are commonly referred to in politicalscience: transnational, international, national, regional and local. Inter-national policy transfer might conceivably take place at and between anyof these levels. Hence, as Box 1 illustrates, theoretically it is possible toconceive of at least twenty-five transfer pathways along which transfersmight occur. The frequency with which transfers along each pathway hap-pen is evidently a subject for empirical analysis. However, the study ofsystemic context might reveal whether transfer along a particular pathwayis likely. For example, transfer between German Lander and British regionsis unlikely since governance at regional level in Britain is underdevelopedand has no formal, let alone constitutional, status.

The role of the agent(s) of transferIt has been argued above that policy transfer analysis should be restrictedto action-oriented intentional learning – that which takes place consciouslyand results in policy action. This definition locates policy transfer as apotential causal phenomenon – a factor leading to policy convergence.However, we distinguish policy transfer from policy convergence in thatthe latter may occur unintentionally – for example due to harmonizing mac-roeconomic forces or common processes. The element of intentionality inour definition of policy transfer makes an agent essential to both voluntaryand coercive processes. Intentionality may be ascribed to the originatingstate/institution/actor, to the transferee state/institution/actor, to both, orto a third party state/institution/actor. For example, if the agent of a parti-cular transfer is the state which first developed the policy, or a third partystate (Country C) seeking to make Country B adopt an approach by Coun-try A, it is likely that there are coercive processes at work. Alternatively,

BOX 1 Policy transfer pathways

International ⇒ International National ⇒ InternationalInternational ⇒ Transnational National ⇒ Transnational

International ⇒ National National ⇒ NationalInternational ⇒ Regional National ⇒ Regional

International ⇒ Local National ⇒ Local

Transnational ⇒ TransnationalTransnational ⇒ International

Transnational ⇒ NationalTransnational ⇒ Regional

Transnational ⇒ Local

Regional ⇒ International Local ⇒ InternationalRegional ⇒ Transnational Local ⇒ Transnational

Regional ⇒ National Local ⇒ NationalRegional ⇒ Regional Local ⇒ Regional

Regional ⇒ Local Local ⇒ Local



there may be a series of agents at work either simultaneously, or at differentpoints in the process. A necessary, but insufficient criterion for identifyingpolicy transfer is therefore to identify the agent(s) of transfer, to specify therole played by agent(s) in the transfer and the nature of the transfer thatthe agent(s) is/are seeking to make.

MODELLING POLICY TRANSFER

Our model has three dimensions – global, transnational and internationallevels; the macro-state level and the interorganizational level.

The global, transnational and international levelsIt is important for the clarity of our argument to affirm that we are not usingthe terms transnational and international interchangeably. We recognize asinternational those structures and processes which inform state-to-staterelations and although the extent of their influence has been exaggerated,we accept the increasing importance of non-state transnational actors inpolicy making at all levels of governance (see Risse-Kappen (ed.) 1995 andPauly and Reich 1997). The term globalization is clearly more problematic.The lack of an agreed understanding of the term is one of the most commonassertions in modern political science (see Amin and Thrift 1994, p. 1), yeteven the growing band of critics of globalization explicitly place the studyof the term at the centre of their analyses (see Hirst and Thompson 1996).Irrespective of one’s position within the globalization debate no seriousscholar would deny that patterns of increased internationalization haveoccurred and that these have posed significant constraints on the ability ofnation states to forward national economic strategies. In particular therehave been significant changes in the organization of production and pat-terns of economic power. Of course this pattern of increased internationaliz-ation and transnationalization has not been confined to the economic arena,strong and weak versions of the globalization thesis may also be foundin relation to studies of changing trends in culture (Smith 1990), politicalintegration (Camilleri and Falk 1992), geopolitics (Rosenau and Czempiel1992), technology (Dosi, Pavitt and Soete 1990), knowledge (Strange 1988),domestic governance and statecraft (Rhodes 1997) and public management(Dunleavy 1994). However, it must be maintained that as these claimsbecome stronger the need for rigorous empirical analysis becomes greater.

Evidently, globalization has had an uneven and ambiguous impact onthe international political system and indeed, the most telling contributionof this exhaustive (and often exhausting) literature is that it has identifieda series of independent and dependent variables which require furtherempirical analysis. For analytical reasons we refer to structures as inde-pendent variables and processes as dependent variables. Hence if we takeCerny’s (1998, p. 2) definition of globalization as ‘. . . the sum total of thewide range of political, economic and social processes of transnationaliz-ation and internationalization taking place in the world today’ our primary



concern becomes the analysis of those structures and processes which canshape the behaviour of international, state and non-state actors and con-strain or facilitate processes of policy transfer. For us, as well as Cerny, thekey to understanding this dimension of policy transfer is largely a questionof structure and agency.

Globalization as a question of structure and agencyIn order to comprehend the nature of policy transfer it is crucial that weplace social and political action within the structured context in which ittakes place. We must assess whether structural processes external to theprocess of transfer we are looking at have an impact (directly or indirectly)upon the context, strategies, intentions and actions of the agents directlyinvolved. Conversely we must also run the counterfactual and assesswhether the strategies, intentions and actions of agents can constrainand/or enable structures. Whether we decide to describe the structures weidentify primarily as resources enabling action or constraining action dependson how we reconcile the notions of structure and agency which for thesepurposes is largely an empirical question (see Hay 1995 for a broaderdiscussion). We adopt Wendt’s (1987) conception of structuration theory,which draws heavily from Giddens (1984). Wendt’s approach emerged ina critique of neo-realism and world systems theory as adequate concep-tualizations of international relations and provides a convincing methodfor unravelling the interaction between structural forces and forms ofagency (see Keohane 1986).

At this juncture it is important to establish why this approach isimportant to our analysis. We are not using structuration theory in orderto defend an ontological world view but to posit some important questionsrelating to anti-reductionism in policy analysis. Structuration theory allowsfor the generative and relational aspects of structuralism while simul-taneously resisting the analytical separation of these generative structuresfrom the practice of human agents. As Wendt (1987, p. 355) puts it: ‘It pro-vides a meta-theory for thinking about real world social systems, but itdoes not tell us what particular kind of agents or what particular kinds ofstructures to expect in any given concrete social system’.

Structuration theory in this formulation has five central features. Firstly,it accepts both the reality and the explanatory importance of irreducibleand often unobservable social structures which generate agents (forexample gender, race, knowledge) and strictly opposes functionalism. Sec-ondly, structures are defined in generative terms as a set of internallyrelated elements which occupy a position within a social organization (forexample agents, practices, technologies, ideologies, territories, etc.). Thirdly,agents and structures are reconciled in a ‘dialectical synthesis’ which over-comes the subordination of one to the other. Fourthly, social structures areperceived as inseparable from spatial and temporal structures, and it isargued that time and space must be incorporated directly and explicitly



into theoretical and empirical research. Finally, structuration theorists con-tend that social structures cannot exist independently of the activities theygovern as they are given essence through the practices of agents (Wendt1987, pp. 335–70). Structuration theory, then, conceptualizes agents andstructures as mutually constitutive yet ontologically distinct entities. Eachis an effect of the other. They are ‘co-determined’ through what Giddens(1979, p. 69) terms the ‘duality of structure’.

So how can we operationalize Wendt’s formulation of structurationtheory at this level of analysis? Figure 1 indicates the range of externalstructures which are central to our discussion. It is proposed that the litera-tures on globalization, internationalization, transnationalization and policytransfer may be linked in three main ways. First we argue that these pro-cesses can act as facilitators of policy transfer in the sense that they increaseopportunity structures for policy transfer (for example, globalcommunications) and secondly, at the same time, policy transfer facilitatesprocesses of globalization (for example, political integration and conver-gence in formations of governance) through the creation of further opport-unity structures, such as European Union (EU) economic development pro-grammes. It would be easy just to contextualize these factors. We thereforeneed a concept which allows us to measure the impact of global, inter-national and transnational forces on the behaviour of states. It is thereforefurther suggested that international regimes play a key role in processingpolicy ideas through epistemic communities which attempt to use theirknowledge resources to promote global awareness of certain policy prob-lems and policy options (note policy convergence in areas such as GATT,food aid, financial regulation and environmental issues through regimepolitics). Regimes are the practical functional application of governance ininternational relations; for it is regimes that articulate the principled andshared understandings of desirable and acceptable forms of state behaviour(Higgott 1996, p. 21). The key research question at this level becomes: inwhat sense do these external structures facilitate state behaviour with parti-cular regard to processes of policy transfer and how? One plausibleapproach is proposed by Cerny who argues that changes in the nature ofthese structural processes have helped to condition the emergence of a com-petition state ‘. . . with the state increasingly using new forms of economicintervention intended to marketize the state itself as well as to promote thecompetitive advantage of national industrial and financial activities withina relatively open world economy’ (1992, p. 241). For Cerny the transform-ation of the nation-state into a competition state lies at the heart of politicalglobalization. The international diffusion of new public management stra-tegies provides a telling illustration of this tendency (see Dunleavy 1994).

The notion of the competition state provides us with a further heuristicdevice – a useful metaphor for describing some key changes in the natureof the capitalist state. Economically the competition state constitutes a moveaway from the industrial-welfare state through the introduction of a distinc-



FIGURE 1 Conceptualizing structure and agency in policy transfer analysis*



tive economic project which embraces the pressures of international mar-kets through the adjustment of domestic as well as foreign economic poli-cies. The competition state pursues increased marketization in order tomake national economic activities more competitive in both internationaland national terms. This strategy has manifested itself in the reduction ofpublic spending, the control of inflation and general neo-liberal monetar-ism, combined with the deregulation of economic activities (especially fin-ancial markets). In order to complete such an ambitious economic project,new forms of statecraft have emerged and institutional structures and polit-ical practices reshaped with the aim of enhancing the steering capacity ofthe state.

The style of government intervention has altered too, represented in ashift from the macro-economy to micro-economic interventionism. Thisshift has been reflected in both deregulation policy and the movement inthe focal point of party and governmental politics away from the generalmaximization of welfare within a nation (for example full employment,redistributive transfer payments and social service provision) to the pro-motion of enterprise, innovation and profitability in both the private andthe public sectors. Ideologically, the competition state constitutes a politicaldiscourse which privileges micro-economic policy making over macro-economic policy making and combines new (for example the impact offlexible specialization and new technologies in industrial policy, deregu-lation and the marketization of public services) and old (for example lib-eralization and monetary and fiscal policy) policy initiatives.

For Cerny ‘the competition state is becoming increasingly both the engineroom and the steering mechanism of an agent-driven political globalizationprocess’ (1998, p. 13). Although we have some misgivings about the privi-leging of the political inferred in this statement, we agree that the policyagenda of the competition state is where we are most likely to locateexamples of policy transfer (Cerny and Evans, forthcoming). For example,transfers in new public management (Stevens 1995), economic (Ikenberry1990), urban (Wolman 1992), foreign (Levy 1994) and welfare policy(Dolowitz 1997). The concept of the competition state links us into themacro-level of enquiry.

Macro-level questionsAs we have already argued, policy transfer must be understood within thecontext of the relationship between structure and agency. We must assessthe impact that changes in economic, technological, ideological or insti-tutional structures have on both facilitating the space for policy transferand affecting the nature of the transfer process itself. For example, in theBritish context policy development must be analysed within the context ofrelative economic decline and attempts by the state to steer capitalismaround periodic economic crisis. Moreover, unless a policy idea fits in withthe strategic concerns of the competition state it is unlikely to be picked-



up. In this sense, the introduction of new service delivery systems in Britishpublic administration correspond directly to the strategic concerns of thecompetition state.

It is also clear that global, international and transnational forces havean impact on institutions and processes internal to the nation state. Thisendogenous process is sometimes referred to as the ‘hollowing-out’ of thestate (Jessop 1992). Rhodes (1994) has argued that there are four key inter-related trends which illustrate the scope of this process: privatization andlimits on the scope and forms of public intervention; the loss of functionsby central government departments to alternative service delivery systems,such as ‘next step’ agencies and through market testing; the loss of func-tions to European Community institutions; and the emergence of limits tothe discretion of public servants through the New Public Management. ForRhodes policy networks are central to understanding internal ‘hollowing-out’, while globalization is central to understanding a concomitant processof external ‘hollowing-out’ because it ‘posits a world of complex interde-pendencies characterized by governance without goverment’ (1997, p. 18).Unlike Rhodes we will hold onto the importance of policy networks asproviding a crucial insight into the external as well as the internal processesof ‘hollowing-out’. The justification for this argument is elaborated below.At this level, therefore, it is further proposed that in a period of majorinstitutional change and under conditions of uncertainty, new opportunitystructures for policy transfer are likely to emerge. These opportunity struc-tures may either be external to the governmental system or internal to it.

The interorganizational levelWe have established that policy transfer takes places within a multi-organi-zational setting, so the analysis of policy transfer requires a method whichwill provide us with the tools for understanding the nature of interorgani-zational politics. It must also furnish us with the appropriate tools for ana-lysing how decision makers acquire and utilize knowledge. The policy net-work and the epistemic community approaches provide us with such amethod. We argue that the relationship between the literatures on policytransfer, policy networks and epistemic communities can be integratedthrough the development of the notion of a policy transfer network whichcan operate at different levels of spatiality. It is further suggested that mostempirical examples of policy transfer tend to emphasize close-knit policycommunities (a form of policy network) which may be comprised of a con-stellation of representatives of epistemic communities, other forms of policyentrepreneur, key bureaucrats, politicians or privileged groups. Box 2 mapsout the relationship between the Marsh and Rhodes (1992b) conception ofa policy community, the Adler and Haas (1992) conception of an epistemiccommunity and our conception of a policy transfer network. There are keysimilarities between the first two characterizations in terms of membership,the nature of integration and resources. Moreover, the membership and


UN

DE

RST

AN

DIN

GPO

LIC

YT

RA

NSFE

R375

BOX 2 The characteristics of policy communities, epistemic communities and policy transfer networks

Dimension A policy community An epistemic community A policy transfer network(Marsh and Rhodes) (Adler and Haas) (Evans and Davies)

Membershipnumber of participants very limited number, some groups limited, a shared set of causal and principled beliefs very limited, the system has a bias against certain

consciously excluded (analytic and normative) act as a filter mechanism inputs, emphasis on bureaucratic and technocraticwhich precludes certain inputs elites

type of interest economic and/or professional natural and social scientists or individuals from any agents of policy transfer, affected politicians andinterests dominate discipline or profession with authoritative claims to bureaucrats

policy relevant knowledge which reside in bothnational and international organizations

Integrationfrequency of interaction frequent, high-quality, interaction of a continuous process of bargaining and negotiation within set time scale frequent, high-quality,

all groups on all matters related to takes place within and between epistemic interaction of all groups on all matters related topolicy issue communities the policy transfer

continuity membership, values, and outcomes membership and values persist over time as long as ad hoc – action-oriented networks set up with thepersist over time reputation survives specific intention of engineering policy change

consensus all participants share basic values all participants share a consensual knowledge base all participants share basic valuesand accept the legitimacy of the and a common policy enterpriseoutcome

Resourcesdistribution of resources all participants have resources; basic all participants have knowledge resources; basic all participants have resources; basic relationship(within network) relationship is an exchange relationship is an exchange relationship is an exchange relationship

relationship

distribution of resources hierarchical; leaders can deliver policy makers are dependent on the intelligence policy makers are dependent on the intelligence(within participating members gathering skills and knowledge resources of the gathering skills and knowledge resources of theorganizations) epistemic community agent of transfer and the donor organization

Power there is a balance of power among the view of policy makers ultimately determines the the success of a policy transfer network rests onmembers – although one group may influence of an epistemic community and their the ability of the agent of transfer to satisfy thedominate, it must be a positive-sum status of acceptance objective policy problem of the client, there mustgame if community is to persist be a positive-sum game if the network is to

persist

Source: Adapted from Marsh and Rhodes, 1992b and Adler and Haas 1992.

B

lackwell

PublishersL

td.

1999


values of both a policy community and an epistemic community tend topersist over time. Conversely, policy transfer networks are an ad hocphenomenon set up with the specific intention of engineering policy changeand thus no extensive process of bargaining or coalition building externalto the transfer network is usually required. Nonetheless, the epistemic com-munity approach deals directly with many of the same agents of transferwhich play a pivotal role in policy evolution within transfer networks. Epis-temic communities are comprised of natural and social scientists or individ-uals from any discipline or profession with authoritative claims to policyrelevant knowledge which reside in both national, transnational and inter-national organizations.

The application of a version of policy network analysis which incorpor-ates the strengths of the epistemic community approach is importantbecause it allows us to focus on the nature of intentional explanation withparticular reference to the role of agents in the process of policy transfer.The notion of a policy transfer network can also help us to evaluate thecognitive dimension of decision making – i.e. how decision makers acquireknowledge. Thus through its emphasis on structural (organizational rulesand imperatives) and interpersonal relationships (information and com-munication exchange) within networks, together with an acceptance ofaccounting for structural factors exogenous to the network (for exampleideology, economy, technology and resource exchange) a method is pro-vided for understanding forms of policy development within a multi-organizational setting. In this sense policy transfer networks provide a con-text for evaluating the complex interaction of state and international policyagendas forged through the interaction of state, non-state, transnational andinternational actors.

Policy transfer networks are an ad hoc, action-oriented phenomenon setup with the specific intention of engineering policy change. They exist onlyfor the time that a transfer is occurring. By implication, policy transfer net-works matter because without them other policies might be adopted. Differ-ent levels of government participate in transfer networks for a variety ofreasons. When governments (local, regional, national or supranational)engage with these networks, it reflects an interaction between (1) the needto satisfy objective policy problems, (2) gaining access to other organiza-tional networks, (3) further relevant motivating values (regime-pull, dis-course-pull, ideological factors), and (4) providing certain essential skillsand knowledge resources.

Figure 2 illustrates how the voluntary policy transfer process can bebroken down into twelve stages. It must be noted that we are making noclaims here about the rationality or otherwise of the policy transfer process.The capacity of an issue to pass through these stages is often contingenton environmental factors (for example economic conditions, changes ingovernment) and the type of agent of transfer. Moreover, the process ofvoluntary transfer can break off at any point past search and still result in



FIGURE 2 The emergence and development of a voluntary transfer network

a form of transfer (for example, the drawing of a lesson, symbolic or percep-tual transfer). The scheme which we present is thus wholly illustrative andprovides the basis for empirical research. We briefly review each of theseputative stages.

RecognitionThe potential for voluntary policy transfer exists at any spatiality of govern-ment as long as enough dissatisfaction or necessity exists and providingthat an entrepreneurial elite is ready to manipulate these grievances andestablish a transfer network. However, the emergence of a policy transfernetwork begins with the recognition by a decision-making elite, politiciansor bureaucrats, of the existence of a decision problem which requires, dueto contextual factors, pressing attention. The political motivation for policymakers to engage in policy transfer have been well documented elsewhere(see Bennett 1991a).

SearchThe absence of acceptable alternative responses or solutions may lead anagent to engage in a search for policy ideas. This is quite often an ad hocprocess characterized by trial and error. We would define searching anorganization’s past as normal policy development. The role of searchactivity as a key feature of the process of policy transfer must be emphas-ized. For as we shall see, it is within this process of search that the natureof information gathering enters new arenas.

ContactDuring the search process an organization may come across a potentialagent of transfer with specialist ‘cognitive’ and ‘elite’ mobilization skills(for example an epistemic community residing within an internationalorganization). In this context, cognitive mobilization refers to the ability ofthe agent of transfer to develop the necessary political and knowledgeresources necessary to satisfy successful policy development. ‘E

´lite’ mobil-

ization refers to the ability of the agent of transfer to gain access to knowl-



edge elites and bring their expertise into the transfer network. At this junc-ture the potential agent of transfer will only be interested in disseminatingbasic information to the potential client with the aim of seducing them intoa dependency relationship. It must be noted that for some agents, policytransfer is a lucrative business. Hence a significant deal of strategic calcu-lation will go into closing a lead. The nature of contact which takes placeis ultimately dependent on whether the agent is independent of both clientand donor organization. It is conceivable, for instance, that the agent mightbe part of the search organization itself.

The emergence of an information feeder networkIf the curiosity of the client is aroused through preliminary contact theagent will act as an information feeder network increasing both the volumeand the detail of information. At this stage the agent will be intent on dem-onstrating the quality of their access to communication and knowledge net-works and further opportunity structures for transfer.

Cognition, reception and the emergence of a transfer networkThe client will evaluate the information that has been provided throughthe information feeder network. Cognition and reception will then usuallydepend on both agents sharing a commitment to a common value system.In this sense the politics of policy transfer networks infer that this form ofpolicy development is the preserve of elite activity and that involvementin the game is wholly dependent on an agent’s resources. It is here thatsome key normative questions are raised about the legitimacy of this formof policy development such as – how accountable are policy transfer net-works?

Elite and cognitive mobilizationThe process of elite and cognitive mobilization is critical to the success ofthe transfer network. It is here that the quality of the agent’s resources areput to the test. The agent will be expected to provide detailed informationabout programmes elsewhere which have addressed a similar problem. Inthe foundational work on systems theory, Easton (1965) argued that inorder to maintain the status quo within a decision-making structure, mech-anisms exist to filter out, or exclude input which would be dysfunctional,or which would overload the system. He referred to these regulators asgatekeepers which exclude from the system excessive or unacceptabledemands. The same argument can be applied to this stage of the policytransfer process. Hence the system has a bias against certain inputs. It isin this stage of the transfer process that policy transfer networks can act asgatekeepers. Of course gate keeping of a more informal kind may also occurwithin what Lukes refers to as the (1974) third, hidden dimension of power,prior to formal filtering, if certain potential inputs are not considered at all.



InteractionThe agent of transfer will often be expected to organize forums for theexchange of ideas between the client and knowledge elites with policy rel-evant knowledge. These may take the form of representatives of an epis-temic community who have similar professional beliefs and standards ofjudgement and share common policy concerns. A context of interaction maytherefore take place through the organization of seminars, fact-finding mis-sions, conferences and the exchange of specialist policy advice documents(for example the drafting of legislation). It is through these forms of dif-fusion activity that agents of transfer can act as a channel for the develop-ment of consensual knowledge. Of course this characterization is alsodependent on the type of agent of transfer under consideration.

EvaluationOnce the client is satisfied with the degree of intelligence gathering thatthey have engaged in, a process of evaluation will commence. The evalu-ation process is critical in determining: the objects of transfer (for examplepolicy goals, structure and content; policy instruments or administrativetechniques; institutions; ideology; ideas, attitudes and concepts; policy styleor negative lessons); the degree of transfer (for example copying, adap-tation, hybridization, synthesis, inspiration or symbolic); and the prerequi-sites of transfer (for example policy feasibility and political, cultural andinstitutional conditions) which will condition the emergence of a policyidea.

DecisionThe policy transfer process is not an isolated enterprise, but an integral partof the policy process. Hence it is best understood, in John Kingdon’s terms,as one of a variety of policy alternatives competing in the ‘policy’ streamof the process. In this sense ideas for policy transfer compete with otheralternatives in what Kingdon (1984) terms the ‘policy primeval soup’. Thesuccess of a process of policy transfer may be conceptualized within King-don’s framework and its success is ultimately dependent on satisfying thesame criteria.

ImplementationThe study of policy transfer is incomplete without an implementation per-spective. Even if a policy is a faithful programmatic copy of the original,it can ultimately only be said to have been transferred if it is carried out.It is by no means certain that those implementing a programme will dowhat was intended or specified by those who formulated it. Indeed, it mayfor reasons of structural incompatibility be impossible for them to do so.Hence the analysis of policy transfer is incomplete without reference toimplementation.



Coercive policy transferHow might this schema differ in a process of coercive policy transfer? Fig-ure 3 suggests some differences in process. In overview a coercive policytransfer network is even more action oriented than a normal transfer net-work. Coercive policy transfer ‘. . . involves one government or supra-national institution pushing, or even forcing, another government to adopta particular programme’ (Dolowitz and Marsh 1996, p. 344). As such, acoercive agent is likely to be involved in the process of transfer from thebeginning. This will confine the search activities of the agent to its immedi-ate regime and it is likely that a representative of an epistemic community(the OECD, for example) will act on behalf of the agent of transfer. It isalso important to note that this process cannot break off at any point before‘action’ otherwise the process of coercion would have failed. We wouldfurther propose that inputs will be closely monitored and controlled at allstages of the process. Having said that, it is equally clear that cultural effectscan influence the evaluation and implementation stages of the transfer pro-cess.

Argument summaryAs figure 4 illustrates, a series of empirically testable hypotheses can bededuced from the above characterization of the process of policy transfer.Following Marsh and Rhodes (1992b), these may be organized into a set ofindependent and dependent variables in which structures should be viewedas independent variables (see 1–4 in figure 4) and functions as dependentvariables. Any variation in the dependent variable – function – may be theresult of variation in either the structure – independent variable – or in theintervening variable – process or mechanism. For example, exogenous ornetwork environment changes may lead to the creation of a policy transfernetwork leading to policy change. These may be economic/market, ideo-logical, knowledge/technical, or institutional effects. If economic factorsconstitute the catalyst for change, the form of the response may be influ-enced by the ideology of the competition state. It may also be deduced thatpolicy change which emerges from a policy transfer network could be theproduct of endogenous factors such as the influence of the agent of transferor the donor organization. However, as a structurationist approach dictates,

FIGURE 3 The emergence and development of coercive policy transfer networks



FIGURE 4 Multi-level policy transfer analysis

policy transfer networks are but one component of an explanation of pol-icy change.

DEMONSTRATING POLICY TRANSFER: SOMEMETHODOLOGICAL AND PRACTICAL ISSUES

Dolowitz and Marsh (1996) rightly argue for a clear scheme for measuringthe occurrence of policy transfer. To achieve this, he identifies five sourcesthrough which the existence of policy transfer might be detected: the media,reports, conferences, visits and government statements. However, these cat-egories are presented more as sources of learning than as sources of proofthat policy transfer has occurred. This section argues that the existing litera-ture does not provide adequate techniques for demonstrating policy trans-fer and it proposes criteria for this purpose. Given adequate standards ofvalidation, proof of policy transfer may be more difficult than is commonlyassumed by those arguing that it is on the increase. Much of the existingliterature rests too much on abstracting alleged perfect fit cases of policytransfer; attention must be paid to the boundary questions, which meansin an empirical sense establishing cases which are not examples of policytransfers. Here we demonstrate the need for caution in claiming the occur-rence of policy transfer and a check list of factors necessary to demonstratethat policy transfer has occurred is then proposed.



Validation in policy transfer analysisWe propose the following sequence of steps as a safeguard against exagger-ated claims about the nature and extent of particular examples of policytransfer.

Subject of analysisAt the outset, it is important to be clear about the phenomenon under study.In primary empirical work, a range of possibilities exist. One might con-sider attempts to facilitate/enforce policy transfer, the process of transferas it is occurring, or a claim that policy transfer has occurred in the past.

Who or what is identified as the agent(s) of transfer – who wants it, what dothey want from it, how are they going about effecting it, to whose benefit, andwhy?The concept of agency is discussed above. To reiterate, an agent is essentialto the voluntary and coercive dimensions of policy transfer given our defi-nition – action-oriented intentional learning. Hence, transfer must be a con-scious process, whether this is undertaken voluntarily or the subject ofcoercion.

Is there evidence of non-transfer?There are two potential dimensions of non-transfer to be taken into accountin any validation exercise. Elements of an idea or a programme which arefound to have been borrowed from domestic antecedents or which areinnovative can be described as non-transfers. Parts of an original idea orprogramme discarded or filtered out by the subject/agent are also non-transfers. Detailed comparison of the subject policy against both domesticand original settings is therefore essential if the real extent of transfer in aparticular case is to be discovered.

What is the evidence offered to support the claim? How good is it?Researchers should look for a preponderance of evidence which demon-strates or refutes a process of policy transfer. Clearly, evidence will differdepending on the nature of the subject. For example, in seeking to demon-strate whether an idea or an attitude has been transferred, a researcher willseek to examine the views and interpretations of the recipient subject(s). Ifon the other hand it is being argued that a programme has been copied,one would expect to find more concrete ‘physical’ evidence. One can onlysay that a programme has been copied by another programme if they havebeen compared. The question of whether the programme has been carriedout might then be subjected to implementation analysis to determine howfar the transfer has permeated. We suggest that a distinction can be madebetween soft transfers (ideas, concepts, attitudes) and hard transfers(programmes and implementation).

The question then follows as to how good the evidence on offer is. Ofcourse, this is a problem of empirical analysis in general, but it is of parti-cular interest here, given the different kinds of transfer and different evi-



dence which might support them. We merely caution that evidence of onething should not be taken to be evidence of another thing.

What conclusion can be drawn from the above about the nature and extent oftransfer which has taken place?Here, answers are given to questions about the subject of analysis – is itideas, programmes or implemented programmes which have/have notbeen or are/are not being transferred? If policy transfer is found to haveoccurred, one might then determine the degree of transfer – copying, emu-lation, hybridization, synthesis or inspiration.

CONCLUSION

As in all mapping exercises we have simplified a complex process in orderto provide a heuristic model which seeks to comprehend the multiplicityof factors which shape the process of policy transfer. Our framework whichestablishes the useful scope and dimensions of policy transfer defines theconcept in Rose’s terms as an action-oriented intentional activity that takesplace within a multi-organizational setting. This heuristic model can use-fully be employed through either forward or backward mapping (the latteras long as policy transfer has occurred). We thus argue that the process oftransfer must be analysed within a three dimensional frame whichencompasses – global, international and transnational levels; the macro-state level and the interorganizational level. The article further suggests avalidation sequence which will aid policy transfer analysts in demonstrat-ing that policy transfer has occurred. In short we claim that this heuristicmodel has additionality because policy transfer analysis can now tell ussomething which we did not know before. We should now have a goodidea about the domestic and international circumstances which are likelyto bring about policy transfer, the scope and dimensions of policy transferand which aspects of the framework should and should not be pursued inempirical work.

ACKNOWLEDGEMENTS

The authors are grateful to the editor Rod Rhodes, Neil Carter and parti-cularly Richard Higgott for their comments on earlier drafts.

REFERENCES

Adler, E. and P. Haas. 1992. ‘Conclusion: epistemic communities, world order, and the creation of a reflec-tive research program’ in P. Haas (ed.), ‘Knowledge, power and international policy coordination’, specialissue of International Organisation.

Amin, A. and N. Thrift. 1994. ‘Living in the global’, in A. Amin and N. Thrift (eds.), Globalisation, institutionsand regional development in Europe. Oxford: Oxford University Press.

Barry, B. 1975. ‘On analogy’, Political Studies, 1950–1975 anniversary issue, XXIII (June to Sept.), 2–3, 208–24.Bennett, C. 1991a. ‘How states utilise foreign evidence’, Journal of Public Policy 11, 31–54.—. 1991b. ‘Review article: what is policy convergence and what causes it?’ British Journal of Political Science

21, 215–33.



Bennett, C. and M. Howlett. 1992. ‘The lessons of learning: reconciling theories of policy learning and policychange’, Policy Sciences 25, 275–94.

Camilleri, J.A. and J. Falk. 1992. The end of sovereignty. Aldershot: Edward Elgar.Cerny, P. 1992. The changing architecture of politics. Structure, agency, and the future of the state. London: Sage.—. 1998. ‘Globalization and politics’, Swiss Political Science Review (forthcoming).Cerny, P. and M. Evans (forthcoming). ‘New Labour, globalization and the competition state’, Economy

and Society.Coleman, W. 1994. ‘Policy convergence in banking: a comparative study’, Political Studies XLII, 274–92.Dolowitz, D. 1997. ‘British employment policy in the 1980s: learning from the American experience’, Govern-

ance 10, 23–42.Dolowitz, D. and D. Marsh. 1996. ‘Who learns what from whom: a review of the policy transfer literature’,

Political Studies 44, 343–57.Dosi, G., K. Pavitt and L. Soete. 1990. The economics of technical change and international trade. New York:

New York University.Dowding, K. 1995. ‘Model or metaphor? A critical review of the policy network approach’, Political Studies

43, 1, 136–50.Dunleavy, P. 1994. ‘The globalization of public services production: can government be “best in world”?’,

Public Policy and Administration 9, 2, 36–64.Easton, D. 1965. A framework for political analysis. N.J: Prentice Hall.Eyestone, R. 1977. ‘Confusion, diffusion, and innovation’, American Political Science Review 71, 441–47.Gerth, H. and C. Wright-Mills. 1948. From Max Weber: essays in sociology. London: Routledge and Kegan Paul.Giddens, A. 1979. Central problems in social theory. Berkeley: University of California Press.—. 1984. The constitution of society: outline of the theory of structuration. Cambridge: Polity Press.Gummett, P. (ed.). 1996. Globalisation and public policy. Cheltenham UK: Edward Elgar.Haas, P. (ed.). 1992. ‘Knowledge, power and international policy coordination’, special issue of Inter-

national Organisation.Hay, C. 1995. ‘Structure and agency’, in D. Marsh and G. Stoker (eds.), Theories and methods in political

science. London: Macmillan.Higgott, R. 1996. ‘Beyond embedded liberalism: governing the international trade regime in an era of econ-

omic nationalism’ in P. Gummett (ed.), Globalisation and Public Policy. Cheltenham: Edward Elgar.Hirst, P. and Thompson, G. 1996. Globalization in question. Cambridge: Polity Press.Hogwood, B. and B.G. Peters. 1983. Policy dynamics. Brighton: Wheatsheaf.Ikenberry, J.G. 1990. ‘The international spread of privatisation policies: inducements, learning and policy

bandwaggoning’, in E. Suleiman and J. Waterbury (eds.), The political economy of public sector reform.Boulder: Westview Press.

James Gregor, A. 1971. An introduction to metapolitics. The Free Press: New York.Jessop, B. 1992. ‘From the Keynsian welfare to the Schumpeterian workfare state’, Lancaster Regionalism

Group, Working Paper 45. Lancaster: University of Lancaster.Keohane, R. 1986. Neo-realism and its critics. New York: Columbia University Press.Kingdon, J. 1984. Agendas, alternatives and public policies. Boston: Little Brown.Levy, J. 1994. ‘Learning and foreign policy: sweeping a conceptual minefield. International Organisation 48,

279–312.Lukes, S. 1974. Power: a radical view. London: Macmillan.Majone, G. 1991. ‘Cross-national sources of regulatory policy making in Europe and the United States’,

Journal of Public Policy 11, 79–106.Marsh, D. and R.A.W. Rhodes. 1992a. ‘New directions in the study of policy networks’, European Journal of

Political Research 21, 181–205.—. (eds.). 1992b. Policy networks in British government. Oxford: Clarendon Press.Marsh, D. and G. Stoker. (eds.). 1995. Theories and methods in political science. London: Macmillan.Pauly, L.W. and S. Reich. 1997. ‘National Structures and multinational corporate behavior: enduring differ-

ences in the age of globalization’, International Organization 51, 1–30.Rhodes, R.A.W. 1994. ‘The hollowing-out of the state: the changing nature of the public services in Britain’,

Political Quarterly 65, 2, 138–51.—. 1996. ‘The new governance: governing without government’, Political Studies XLIV, 652–67.



—. 1997. Understanding governance – policy networks, reflexivity and accountability. Milton Keynes: Open Uni-versity Press.

Risse-Kappen, T. (ed.). 1995. Bringing transnationals back in: non-state actors, domestic structures and inter-national institutions. Cambridge: Cambridge University Press.

Robertson, D. 1991. ‘Political conflict and lesson drawing’, Journal of Public Policy 11, 331–54.Rose, R. 1991. ‘What is lesson drawing?’, Journal of Public Policy 11, 3–30.—. 1993. Lesson drawing in public policy. NJ: Chatham House.Rosenau, J.N. and E.-O. Czempiel. 1992. Governance without government: order and change in world politics.

Cambridge: Cambridge University Press.Rudner, R. 1966. Philosophy of social science. Englewood Cliffs, NJ: Prentice-Hall.Smith, A.D. 1990. ‘Towards a global culture?’, Theory, Culture and Society 7, 171–91.Stevens, H. 1995. ‘Converging administrative systems: recruitment and training in the United Kingdom’,

Working Paper, The European Institute, London School of Economics and Political Science.Strange, S. 1988. States and markets. London: Francis Pinter.Stone. D. 1996. Capturing the political imagination: think tanks and the policy process. London: Cass.Wendt, A. 1987. ‘The agency-structure problem in international relations’, International Organisation 41,

335–70.Wolman, H. 1992. ‘Understanding cross-national policy transfers: the case of Britain and the United States’,

Governance 5, 27–45.

Date received 10 September 1997. Date accepted 15 June 1998.


UNDERSTANDING POLICY TRANSFER: A MULTI-LEVEL, MULTI-DISCIPLINARY PERSPECTIVE

Documents