Three Worlds of Welfare Capital - Arts, Wil

Three worlds of welfare capitalism or more?A state-of-the-art report

Wil Arts and John Gelissen*, Tilburg University, The Netherlands

Article

Journal of European Social Policy 0958-9287 (200205)12:2 Copyright © 2002 SAGE Publications, London, Thousand Oaks and New Delhi, Vol 12 (2): 137–158; 023114

Summary This paper surveys the debateregarding Esping-Andersen’s typology ofwelfare states and reviews the modified oralternative typologies ensuing from thisdebate. We confine ourselves to the classifica-tions which have been developed by Esping-Andersen’s critics in order to cope with thefollowing alleged shortcomings of his typol-ogy: (1) the misspecification of the Mediter-ranean welfare states as immature Continentalones; (2) the labelling of the Antipodeanwelfare states as belonging to the ‘liberal’regime type; (3) a neglect of the gender-dimension in social policy. We reconstructseveral typologies of welfare states in order toestablish, first, whether real welfare states arequite similar to others or whether they arerather unique specimens, and, second, whetherthere are three ideal-typical worlds of welfarecapitalism or more. We conclude that realwelfare states are hardly ever pure types andare usually hybrid cases; and that the issue ofideal-typical welfare states cannot be satisfac-torily answered given the lack of formal theo-rizing and the still inconclusive outcomes ofcomparative research. In spite of this conclu-sion there is plenty of reason to continue towork on and with the original or modifiedtypologies.

Key words comparative social policy analysis, typology, welfare state regimes,worlds of welfare capitalism

Résumé Nous présentons un état des lieuxdes débats qui ont entouré la typologie desEtats providences proposée par Esping-Andersen ainsi que des typologies modifiées oualternatives qui ont été présentées par la suite.Nous nous limiterons aux classifications quiont été proposées par les critiques du travaild’Esping Andersen qui visaient à dépasser les«prétendues» limitations de sa typologie.1)une mauvaise spécification des Etats Providenceméditerranéens comme des Etats providencescontinentaux inachevés 2) la labelisation desEtats providences des Antipodes commeappartenant au régime de type «libéral» 3) lanon prise en compte de l’effet «genre» dans lespolitiques sociales. Nous avons reconstruit dif-férentes typologies d’Etats Providences afind’établir, tout d’abord si la réalité des EtatsProvidences est unique ou si certains sont fortsemblables et ensuite s’il existe trois idéal typed’Etats providences ou davantage?

Nous sommes arrivé à la conclusion que:premièrement les Etats providences réels sontrarement des cas purs mais constituent bienplus des hybridations. En deuxième lieu, on nepeut répondre de manière satisfaisante à la ques-tion de savoir le nombre d’idéal types d’EtatsProvidences du fait de l’absence de théorisa-tion adéquate et parce que les résultats desrecherches comparatives ne permettent pas deconclure dans un sens ou un autre. Troisième-ment, en dépit de cette conclusion, il existe demultiples raisons pour continuer à travailler suret à partir des typologies originales ou modifiées.

* Author to whom correspondence should be sent: John Gelissen, Department of Sociology, TilburgUniversity, PO Box 90153, 5000 LE Tilburg, The Netherlands.

http:\\www.sagepulications.com

Introduction

When Esping-Andersen (1990) published hisThree Worlds of Welfare Capitalism it wasreceived with applause (Offe, 1991; Cnaan,1992; Hicks, 1991; Kohl, 1993). After a fewyears, the book even became a modern classic.Nevertheless, Esping-Andersen (1993; 1994;1996; 1997; 1999) has been forced time aftertime by both his critics and his adherents toelaborate on his original arguments. The tenetof Esping-Andersen’s treatise of the welfarestate was that, for a long time in both the the-oretical and empirical literature, too littleattention had been given to cross-national dif-ferences in welfare state structures.1 In spite ofthis putative lack of attention he could – theo-retically and empirically – stand ‘on the shoul-ders of giants’. Theoretically, the work ofMarshall (1950; 1963; 1965; 1981) andTitmuss (1958; 1974) laid the foundations forEsping-Andersen’s typology (Boje, 1996: 19).Empirically, he could profit from the compar-ative research by, among others, Wilensky(1975), Flora and Heidenheimer (1981),Mommsen (1981) and Flora (1983; 1986). Heargued that we are entering better timesbecause ‘the most intensive activity of welfarestate theorising at the moment has becomeidentifying diversity, specifying welfare statetypologies’ (Esping-Andersen, 1994: 715).Research has to follow theory’s lead because‘only comparative empirical research will ade-quately disclose the fundamental propertiesthat unite or divide modern welfare states’(Esping-Andersen, 1990: 3). In his ‘seminal’book he suited the action to the word by con-structing today’s best-known and most fre-quently used typology of welfare states, andby empirically testing whether distinct welfarestates that resemble his ideal-types can beobserved.

For accomplishing this feat, he not onlyreceived wide critical acclaim and constructivecritism, but also some negative criticism. Themore amicable critics argue that his typologyhas merits but is neither exhaustive nor exclu-sive and therefore needs revising. Others refer

to theoretical and methodological shortcom-ings (cf. Lessenich and Ostner, 1998). Themore hostile critics feel that typologies as suchhave no explanatory power and, therefore, hisscheme does not contribute to proper theoriz-ing about what is happening with and withinwelfare states (cf. Baldwin, 1996).

In this paper, we give an overview of thedebate about Esping-Andersen’s typologyduring the last decade and review the state ofthe art of typifying welfare states at the turnof the millennium. The pivotal questions areas follows: How and why has the welfare statedeveloped? How and why do national welfaresystems differ from one another – or are theysimilar to each other? Do welfare statescluster into different regime types and, if so,how and why? It is not our intention to raisenew questions. Our objective is to settleaffairs, for the time being, by giving anoverview of what we think is the gist of thediscussion, by weighting the most importantarguments and taking stock of the modellingbusiness. To achieve this goal, we elaborate onearlier overviews by Abrahamson (1999), Artsand Gelissen (1999) and Kohl (2000). First,we review the debate with respect to Esping-Andersen’s typology and modified versions ofit. Second, we try to establish whether thereare three ideal-typical worlds of welfare capi-talism or more. Third, we outline whichauthors identify which national states asbelonging to a particular type of welfare state.This means that we will not only look at idealworlds of welfare capitalism, but also at realones.

Ideal-types

Do typologies based on ideal-types have theo-retical and empirical value as Esping-Andersenassumes? The conclusion emerging from thephilosophy of science literature is clear: not ifideal-types are goals in themselves, but only ifthey are the means to a goal; namely, the rep-resentation of a reality, which cannot (yet) bedescribed using laws (Klant, 1984). This

138 Arts and Gelissen

Journal of European Social Policy 2002 12 (2)

means that typologies are only fruitful to anempirical science that is still in its infancy. Incontrast, a mature empirical science empha-sizes the construction of theories and not theformulation of typologies. There are goodreasons to argue that the comparative macro-sociology of welfare states is still in statunascendi. Therefore the formulation of typolo-gies could be useful. Before answering thequestion of whether welfare state typologiesbased on ideal-types are not only useful butalso have explanatory value, we first considerwhat Esping-Andersen himself says about themethodological status of his typology.

He addresses the question of whether thewelfare state is merely the sum total of anation’s social policy repertoire, or whether itis an institutional force above and beyond agiven policy array. His answer is straightfor-ward: the welfare state cannot be regarded asthe sum total of social policies, it is more thana numerical cumulation of discrete programmes(Esping-Andersen, 1994: 712). Therefore, inthe relation between state and economy, hedefines welfare state regimes as a complex oflegal and organizational features that are sys-tematically interwoven. Esping-Andersen(1990: 3, 26, 32) boldly suggests that whenwe focus on the principles embedded inwelfare states, variations are not linearly dis-tributed around a common denominator. Theyare clustered around three highly diverseregime-types, each organized according to itsown discrete logic of organization, stratifica-tion, and societal integration. Therefore, wecan identify three models, or ideal-types ofwelfare states: conservative, liberal and social-democratic. These ideal-types owe theirorigins to different historical forces and theyfollow qualitatively different developmentaltrajectories. Contrary to the ideal world ofwelfare states, the real world is likely toexhibit hybrid forms. There are no one-dimen-sional nations in the sense of a pure case.Today, every country presents a system mix.Esping-Andersen (1997: 171) argues thatdespite this it is fruitful to construct ideal-types for the sake of economy: to be able to

see the forest rather than the myriad of uniquetrees. However, he warns of the danger thatthe resulting forest may bear little resemblanceto reality.

Looking for a more detailed answer to thequestion of the theoretical status of ideal-typesand typologies, we must return to the locusclassicus: Weber’s methodological essays.Weber (1949; 1968 [1922]) deals with twodifferent kinds of ideal-types, individualisticand holistic ones (cf. Hempel, 1965 [1952];Rogers, 1969; Watkins, 1969 [1953]). Esping-Andersen’s ideal-types of welfare state regimesare holistic. They propose to give a bird’s eyeview of the broad characteristics of a social orhistorical situation. The ideality of such typeslies in their simplification and aloofness fromdetail. They emphasize the ‘essential’ featuresof a situation considered as a whole. By com-paring an impure welfare state with an ideal-typical one – both considered as a whole – thedeviations of the former from the latter arethrown into relief. It is the simultaneousknowledge of both the ideal-type and the real-type that enables holistic ideal-types to beused ‘as conceptual instruments for compari-son with and measurement of reality’(Watkins, 1969 [1953]: 458–9).

From a logical point of view (von Kempski,1972), the general term ‘welfare state’ is alabel for a certain class of democratic indus-trial capitalist societies, characterized bycertain properties (i.e. social citizenship or thefact that more or less extensive welfare provi-sions are legally provided, or, in still otherwords, the fact that the state plays a principalpart in the welfare mix alongside the market,civil society, and the family). Welfare stateshave seldom been established as a result of bigplans or big fights, but mostly as results ofcomplex processes and successive steps ofsocial and political engineering in the historyof democratic industrial capitalist societies. Inspite of the largely incremental emergence ofwelfare states, Esping-Andersen is of theopinion that this class of societies does notconsist of a great number of unique cases, but that they cluster together in three distinct

Three worlds of welfare capitalism or more? 139


subclasses. Each of the three types he identi-fies with a deep tradition in political mobiliza-tion and political philosophy (conservatism,liberalism and socialism respectively) whichthen link to particular features of contempo-rary social policy (and broader politicaleconomy) configurations.

To determine the characteristics of thesesubclasses without going back in history, twoindicators are crucial: decommodification andstratification. Together they define a two-dimensional property-space. Although realwelfare states are most of the time not unique,they certainly are never completely similar.This means that they are almost alwaysimpure types. The consequence is thatalthough they cluster together in three sub-classes it is not always easy to classify all casesunambiguously. In practice it is possible thatdifferent judges assign a particular welfarestate to different subclasses. Interjudge valid-ity can be accomplished by assessing whichideal-type – the extreme limiting cases in thisordering – they approximate best. By compar-ing impure real welfare states to ideal-types,the deviations of the ‘impure’ real-types arecontrasted with the ‘purity’ of the ideal-type.This simultaneous recognizability of both theideal and the real-type make it possible to useholistic ideal-types as conceptual instrumentsfor comparison and for the empirical determi-nation of reality (Watkins, 1969 [1953]). Butaccurate ordering is not enough. After all,ideal-types are also instruments for providingexplanations. If we use them to satisfy thisobjective, they should not only be understoodas a conceptual system but also as a system of theoretical statements. These shouldencompass testable, general hypotheses or, atleast, provide a framework for interpretation(Hempel, 1965 [1952]).

To what conclusion does the precedingreflection lead? In reply to Esping-Andersen’sfiercest critics, we can say that their criticismis unjust if certain conditions are met. Thefirst condition is that the typology is a validand reliable instrument for classifying welfarestates. Whether this condition holds will be

tested later on in this paper. The second con-dition is that the typology is a means to anend – explanation – and not an end in itself.Esping-Andersen uses the regime types notonly as dependent variables but also as inde-pendent variables to explain cross-nationalvariations in dependent variables such associal behaviour and social attitudes. He alsouses the typology to postulate and explain theoccurrence of positive feedback loops. Accord-ing to him, existing institutional welfarearrangements heavily determine, maybe evenover-determine, national trajectories (Esping-Andersen, 1999: 4). This suggests path-dependency. This is because the stratificationoutcomes of particular arrangements shapeclass coalitions, which tend to reproduce theoriginal institutional matrix and welfare out-comes. This means that policies provide incen-tives that encourage individuals and groups toact in ways that lock in a particular path ofpolicy development (cf. Pierson, 1993).

A third condition for accepting typologizingas a legitimate endeavour is that theory con-struction on welfare states is still in an earlystage. This is also true. Boje (1996: 18) arguesthat the present ‘state of the art’ in mostwelfare state research is characterized by alack of theory. Few theoretical alternatives areavailable. At the most, one can think ofmarxist explanations (see, for exampleTherborn, 1995) or de Swaan’s (1988) synthe-sis of rational-choice theories and figurationalsociology. The construction of ideal-types canbe fruitful under the condition that these willeventually lead to theories. We will return tothis issue in the final section.

Three worlds of ‘welfare capitalism’ . . .

The central explanatory questions Esping-Andersen (1990: 4, 105) asks are: Why is theworld composed of three qualitatively differ-ent welfare-state logics? Why do nations crys-tallize into distinct regime-clusters? Thesequestions demand a theoretical answer. Since



he is of the opinion that the existing theoreti-cal models of the welfare state are inadequate,reconceptualization and retheorization arenecessary. In answering these questions hestarts with the orienting statement that historyand politics matter. Or, more specifically: ‘Thehistorical characteristics of states, especiallythe history of political class coalitions as themost decisive cause of welfare-state variations,have played a determinate role in forging theemergence of their welfare-statism’ (1990: 1).

What are the historical and political forcesbehind the regime differences? According toEsping-Andersen (1990: 29), three interactingfactors are significant: the nature of classmobilization (especially of the working class),class-political action structures, and the his-torical legacy of regime institutionalization.The provisional answer to his central ques-tions is therefore: If you look at the history ofso-called welfare states you find three ideal-typical trajectories, a liberal, a conservativeand a social-democratic one. Fortunately, onedoes not have to go back in history, however,in order to typify ‘real’ welfare states. We cancharacterize them, as we have mentionedbefore, by looking at their positions on twofundamental dimensions of welfare statism:

1. The degree of decommodification, i.e. thedegree to which a (social) service is ren-dered as a matter of right, and the degreeto which a person can maintain a liveli-hood without reliance on the market.

2. The kind of social stratification and soli-darities, i.e. which social stratificationsystem is promoted by social policy anddoes the welfare state build narrow orbroad solidarities?

What are the characteristics of the three dis-tinct regime-types to which the historicalforces lead? To answer this question, Esping-Andersen (1990: 73) argues that although thebefore-mentioned dimensions are conceptuallyindependent, according to his ‘theory’ hewould expect that there is sufficient covaria-tion for distinct regime clusters to emerge. In

accordance with this theoretical expectation,he succeeds in empirically identifying threeclosely paralleled models – ideal-types – ofregime-types on both the stratification and thedecommodification dimension. There appearsto be a clear coincidence of high decommodi-fication and strong universalism in theScandinavian, social-democratically influencedwelfare states. There is an equally clear coinci-dence of low decommodification and strongindividualistic self-reliance in the liberalAnglo–Saxon nations. Finally, the ContinentalEuropean countries group closely together ascorporatist and etatist, and are also modestlydecommodifying (Esping-Andersen, 1990: 77).

In spite of anomalies such as theNetherlands and Switzerland, the overallpicture is convincing, at least at first glance.This empirical success permits a more exten-sive description of these three worlds ofwelfare capitalism. First, there is the liberaltype of welfare capitalism, which embodiesindividualism and the primacy of the market.The operation of the market is encouraged bythe state, either actively – subsidizing privatewelfare schemes – or passively by keeping(often means tested) social benefits to amodest level for the demonstrably needy.There is little redistribution of incomes withinthis type of welfare state and the realm ofsocial rights is rather limited. This welfareregime is characterized by a low level ofdecommodification. The operation of theliberal principle of stratification leads to divi-sion in the population: on the one hand, aminority of low-income state dependants and,on the other hand, a majority of people ableto afford private social insurance plans. In thistype of welfare state, women are encouragedto participate in the labour force, particularlyin the service sector.

Second, there is a world of conservative-corporatist welfare states, which is typified bya moderate level of decommodification. Thisregime type is shaped by the twin historicallegacy of Catholic social policy,2 on the oneside, and corporatism and etatism on theother side. This blend had three important



consequences in terms of stratification. In thefirst place, the direct influence of the state isrestricted to the provision of income mainte-nance benefits related to occupational status.This means that the sphere of solidarityremains quite narrow and corporatist.Moreover, labour market participation bymarried women is strongly discouraged,because corporatist regimes – influenced bythe Church – are committed to the preserva-tion of traditional family structures. Anotherimportant characteristic of the conservativeregime type is the principle of subsidiarity: thestate will only interfere when the family’scapacity to service its members is exhausted(Esping-Andersen, 1990: 27).

Finally, Esping-Andersen recognizes asocial-democratic world of welfare capitalism.Here, the level of decommodification is high,and the social-democratic principle of stratifi-cation is directed towards achieving a systemof generous universal and highly distributivebenefits not dependent on any individual con-tributions. In contrast to the liberal type ofwelfare states, ‘this model crowds out themarket and, consequently, constructs anessentially universal solidarity in favour of thewelfare state’ (Esping-Andersen, 1990: 28).Social policy within this type of welfare stateis aimed at a maximization of capacities forindividual independence. Women in particular– regardless of whether they have children ornot – are encouraged to participate in thelabour market, especially in the public sector.Countries that belong to this type of welfarestate regime are generally dedicated to fullemployment. Only by making sure that asmany people as possible have a job is it possi-ble to maintain such a high-level solidaristicwelfare system.

. . . Or more?

In the Introduction, we indicated the tremen-dous impact of Esping-Andersen’s work oncomparative social policy analysis. Since then,several authors have developed alternative

typologies or added one or more types toexisting classifications for greater empiricalrefinement. From this vast array of welfarestate typologies we have selected six classifica-tions, which we think draw attention to inter-esting characteristics of welfare states notdirectly included in Esping-Andersen’s classifi-cation. All these typologies and their maincharacteristics are summarized in Table 1.

These alternative classifications relate tothree important criticisms of Esping-Andersen’s classification (for these and otherpoints of critique see Schmidt, 1998; Gough,2000b).3 First, the misspecification of theMediterranean welfare states; second, labellingthe Antipodean welfare states as belonging tothe ‘liberal’ welfare state regime; and finally,the neglect of the gender-dimension in socialpolicy. In the following sections, we willdiscuss these criticisms in more detail andpresent some of the alternative classificationsdeveloped by his critics.

The Mediterranean

One important criticism of Esping-Andersen’sclassification is that he did not systematicallyinclude the Mediterranean countries.Specifically, in The Three Worlds of WelfareCapitalism Italy belongs – according to him –to the family of the corporatist welfare stateregimes, whereas Spain, Portugal and Greeceare not covered by his typology. Although headmits that these countries have some impor-tant characteristics in common – i.e. aCatholic imprint (with the exception ofGreece) and a strong familialism (Esping-Andersen, 1997: 180) – he seems to includethem in the continental/corporatist model. Hisomission of a systematic treatment of theMediterranean has brought about a livelydebate about the existence of a ‘Southern’ or‘Latin Rim’ model of social policy. Forexample, Katrougalos (1996) supportsEsping-Andersen’s position by arguing thatthe Mediterranean countries ‘do not form adistinct group but rather a subcategory, a



Three w

orlds of welfare capitalism

or more?

143

Journal of European Social P

olicy 2002 12 (2)

Table 1 An overview of typologies of welfare states

Types of welfare states and their characteristics Indicators/dimensions

Esping-Andersen (1990) 1. Liberal: Low level of decommodification; market-differentiation of welfare • Decommodification2. Conservative: Moderate level of decommodification; social benefits mainly dependent on • Stratification

former contributions and status3. Social-democratic: High level of decommodification; universal benefits and high degree of

benefit equality

Leibfried (1992) 1. Anglo–Saxon (Residual): Right to income transfers; welfare state as compensator of last resort • Poverty, social insurance and tight enforcer of work in the market place and poverty policy

2. Bismarck (Institutional): Right to social security; welfare state as compensator of first resort andemployer of last resort

3. Scandinavian (Modern): Right to work for everyone; universalism; welfare state as employer offirst resort and compensator of last resort

4. Latin Rim (Rudimentary): Right to work and welfare proclaimed; welfare state as a semi-institutionalized promise

Castles & Mitchell (1993) 1. Liberal: Low social spending and no adoption of equalizing instruments in social policy • Welfare expenditure2. Conservative: High social expenditures, but little adoption of equalizing instruments in social • Benefit equality

policy3. Non-Right Hegemony: High social expenditure and use of highly equalizing instruments in • Taxes

social policy4. Radical: Achievement of equality in pre-tax, pre-transfer income (adoption of equalizing

instruments in social policy), but little social spending

Siaroff (1994) 1. Protestant Liberal: Minimal family welfare, yet relatively egalitarian gender situation in the • Family welfare orientationlabour market; family benefits are paid to the mother, but are rather inadequate

2. Advanced Christian-democratic: No strong incentives for women to work, but strong incentives • Female work desirabilityto stay at home

3. Protestant Social-democratic: True work–welfare choice for women; family benefits are high and • Extent of family benefits always paid to the mother; importance of Protestantism being paid to women

4. Late Female Mobilization: Absence of Protestantism; family benefits are usually paid to thefather; universal female suffrage is relatively new

Ferrera (1996) 1. Anglo–Saxon: Fairly high welfare state cover; social assistance with a means test; mixed system • Rules of access (eligibility)of financing; highly integrated organizational framework entirely managed by a public administration

continued over

144A

rts and Gelissen


olicy 2002 12 (2)

Table 1 continued

Types of welfare states and their characteristics Indicators/dimensions

2. Bismarck: strong link between work position (and/or family state) and social entitlements; • Benefit formulaebenefits proportional to income; financing through contributions; reasonably substantial socialassistance benefits; insurance schemes mainly governed by unions and employer organizations

3. Scandinavian: social protection as a citizenship right; universal coverage; relatively generous • Financing regulationsfixed benefits for various social risks; financing mainly through fiscal revenues; strongorganizational integration

4. Southern: fragmented system of income guarantees linked to work position; generous benefits • Organizational–managerial without articulated net of minimum social protection; health care as a right of citizenship; arrangementsparticularism in payments of cash benefits and financing; financing through contributions and fiscal revenues

Bonoli (1997) 1. British: Low percentage of social expenditure financed through contributions (Beveridge); low • Bismarck and Beveridgesocial expenditure as a percentage of GDP model

2. Continental: High percentage of social expenditure financed through contributions (Bismarck);high social expenditure as a percentage of GDP

3. Nordic: Low percentage of social expenditure financed through contributions (Beveridge); high • Quantity of welfare state social expenditure as a percentage of GDP expenditure

4. Southern: High percentage of social expenditure financed through contributions (Bismarck);low social expenditure as a percentage of GDP

Korpi & Palme (1998) 1. Basic Security: Entitlements based on citizenship or contributions; application of the flat-ratebenefit principle

2. Corporatist: Entitlements based on occupational category and labour force participation; use of • Bases of entitlementthe earnings-related benefit principle

3. Encompassing: Entitlement based on citizenship and labour force participation; use of the flat- • Benefit principlerate and earnings-related benefit principle

4. Targeted: Eligibility based on proved need; use of the minimum benefit principle • Governance of social5. Voluntary State Subsidized: Eligibility based on membership or contributions; application of the insurance programme

flat-rate or earnings-related principle

variant of the Continental model. They aremerely underdeveloped species of theContinental model, welfare states in theirinfancy, with the main common characteristicsbeing the immaturity of the social protectionsystems and some similar social and familystructures’ (Katrougalos, 1996: 43). However,according to other commentators (Leibfried,1992; Ferrera, 1996; Bonoli, 1997; Trifiletti,1999) it seems logical to see the SouthEuropean countries as a separate cluster. They have developed classifications ofEuropean welfare states which try to show theexistence of a separate ‘southern model’ ofsocial policy.

Leibfried (1992) distinguishes four socialpolicy or poverty regimes within the countriesof the European Community: the Scandina-vian welfare states, the ‘Bismarck’ countries,the Anglo–Saxon countries and the Latin Rimcountries. These policy regimes are based ondifferent policy models – modern, institu-tional, residual and rudimentary – in whichsocial citizenship has developed in differentand sometimes incomplete ways. Within thesepolicy regimes, welfare state institutions havea different function in combating poverty.However, it is particularly important thatLeibfried adds a fourth category – the ‘LatinRim’ countries – to Esping-Andersen’s originalclassification. He emphasizes as an importantcharacteristic of these countries the lack of anarticulated social minimum and a right towelfare.

Ferrera (1996) also argues explicitly for theinclusion of a ‘Southern model’ of socialpolicy (1996: 4–7). He concentrates on fourdimensions of social security systems: the rulesof access (eligibility rules), the conditionsunder which benefits are granted, the regula-tions to finance social protection and, finally,the organizational-managerial arrangementsto administrate the various social securityschemes. Based on these dimensions, he makesa distinction between the Scandinavian,Anglo–Saxon, Bismarckian and Southerncountries. The Scandinavian countries arecharacterized by universal coverage for the

risks of life. The right to social protection isattributed on the basis of citizenship. TheAnglo–Saxon family of welfare states is alsocharacterized by a highly inclusive social secu-rity coverage, but only in the area of healthcare can one speak of fully universal risk cov-erage. Also flat-rate benefits and means testingplay an important role. In the third group ofcountries, the relationship between socialsecurity entitlements, a person’s labour marketstatus and role within the family (breadwinneror not) is still clearly visible. Contributionsplay an important role in financing thevarious schemes. Almost everybody has socialinsurance coverage through their own orderived rights. Finally, the social protectionsystems of Southern countries are highly frag-mented and, although there is no articulatednet of minimum social protection, some bene-fits levels are very generous (such as old agepensions). Moreover, in these countries healthcare is institutionalized as a right of citizen-ship. However, in general, there is relativelylittle state intervention in the welfare sphere.Another important feature is the high level ofparticularism with regard to cash benefits andfinancing, expressed in high levels of clien-telism. The most important features of eachtype are summarized in Table 1.

Bonoli (1997) uses the Mediterranean coun-tries – among others – to develop the finalclassification we wish to discuss in thissection. He is especially critical of the decom-modification approach. According to him, itdoes not allow one to discriminate effectivelybetween the Bismarckian and the Beveridgeanapproaches to social policy. As an alternative,he combines two approaches to the classifica-tion of welfare states. One concentrates on the ‘how much’ dimension (emphasized in theAnglo–Saxon literature) and the other on the‘how’ dimension of social policy (emphasizedin the Continental-European or French tradi-tion). As an empirical indicator of the firstdimension, he uses social expenditure as a pro-portion of GDP, and of the second dimensionthe percentage of social expenditure financedthrough contributions. These indicators lead



him to identify four types of countries: theBritish countries, the Continental Europeancountries, the Nordic countries and theSouthern countries, thus giving credit to theproposal of a ‘Southern model’.

Upon examining the combined argumentsof Leibfried, Ferrera and Bonoli, as presentedin Table 1, it appears that a strong similarityexists among their first three types and thoseof Esping-Andersen. However, all threeauthors add a fourth – Mediterranean – typeof welfare state regime to the original Esping-Andersenian classification. Using empiricalevidence, they argue that this is a prototyperather than a subcategory of the continental/corporatist model.

The Antipodes

Esping-Andersen also discusses the Antipodeancountries (i.e. Australia and New Zealand) asrepresentatives of the liberal welfare stateregime. This is because of their marginal com-mitment to public welfare and strong relianceon means testing. However, according toCastles (1998), Australia and New Zealandhave a more particular and a more inclusiveapproach to social protection than the stan-dard liberal form. Thresholds are set at com-paratively high levels, so that a large part ofthe population receives some means-testedbenefits. The result is that the Antipodesexhibit the world’s most comprehensive systemsof means-tested income support benefits.Redistribution has been traditionally pursuedthrough wage controls and employment secu-rity rather than social programmes. Incomeguarantees, realized by using market regulationthus play an important role in the institutionalset-up of these welfare states. It therefore seemsthat the Antipodean countries represent a sep-arate social policy model. It led Castles andMitchell (1993) to question whether ‘socialspending is the only route to greater incomeredistribution’, implying that there may be otherways than income maintenance by which statesmay mitigate the effects of market forces.

In a discussion of their study, Hill (1996)points out that Castles and Mitchell’s critiqueof Esping-Andersen’s work essentially followstwo lines. In the first place, they draw atten-tion to the fact that political activity from theLeft may have been introduced into thosecountries in the achievement of equality inpre-tax, pre-transfer income rather than in thepursuit of equalization through social policy.Second, they argue – again about Australiabut also with relevance to the United King-dom – that the Esping-Andersen approach dis-regards the potential for income-relatedbenefits to make an effective contribution toredistribution. Australian income maintenanceis almost entirely means-tested. It uses anapproach that neither concentrates on a liberal-type redistribution to the very poor, norresembles the more universal social-democraticand hierarchical solidaristic conservativeideal-types highlighted in Esping-Andersen’sstudy (Hill, 1996: 46). This is the reason whyCastles and Mitchell develop an alternative,four-way classification of welfare states:Liberal, Conservative, Non-Right Hegemonyand Radical. This utilizes the level of welfareexpenditure (i.e. household transfers as a per-centage of GDP); average benefit equality; andincome and profit taxes as a percentage ofGDP.

Other evidence for the exceptional positionof the Antipodean countries, specificallyAustralia, is found when countries are classi-fied according to the typology developed byKorpi and Palme (1998). This is based oninstitutional characteristics of welfare states.They try to investigate the causal factorswhich influence the institutional aspects of thewelfare state on the one hand, and the effectsof institutions on the formation of interests,preferences and identities – as well as on thedegree of poverty and inequality in a society –on the other hand. They argue that institu-tional structures can be expected to reflect therole of conflicts among interest groups, whilethey are also likely to form important frame-works for the definitions of interests and iden-tities among citizens. They can thereby be



expected to influence coalition formation,which is significant for income redistributionand poverty. As the basis of their classifica-tion, Korpi and Palme take the institutionalstructures of two social programmes – old agepensions and sickness cash benefits – whichthey consider to lie at the heart of the welfarestate. The institutional structures of the twoprogrammes are classified according to threeaspects: the bases of entitlements, the princi-ples applied to determine benefit levels (towhat extent social insurance should replacelost income), and the governance of a socialinsurance programme (whether or not repre-sentatives of employers and employees partici-pate in the governing of a programme). Basedon these three aspects, they discriminateamong five different ideal-types of institu-tional structures: the targeted (empiricallyexemplified by the Australian case), voluntarystate subsidized, corporatist, basic securityand encompassing model. In Table 1, theseideal-types and their most important featuresare delineated. Again, the Esping-Andersenmodel stands. However, a number of countriesare no longer considered to belong to a sub-category of his three prototypes, but to a newprototype.

Gender, familialism and late femalemobilization

By explicitly incorporating gender, severalauthors have tried to reconceptualize thedimensions of welfare state variation. Sub-jecting the mainstream welfare state typolo-gies to an analysis of the differential places ofmen and women within welfare states would,according to them, produce valuable insights.This does not mean, however, that the charac-teristics used to construct the typologies areexhaustive (Sainsbury, 1996: 41). Genderanalysis suggests that there are whole areas ofsocial policy that Esping-Anderson simplymisses. What seems to be particularly lackingis a systematic discussion of the family’s placein the provision of welfare and care. Not only

the state and the market provide welfare, butalso families. A further omission is that thereis no serious treatment of the degree to whichwomen are excluded from or included in thelabour market.4 Instead of employing the all-or-nothing words ‘inclusion’ and ‘exclusion’to gender differences, it seems sensible tostress the importance of partial citizenship(Bulmer and Rees, 1996: 275). Womenobtained full civil and political rights a consid-erable time ago, but with regard to socialrights, women are still discriminated against,sometimes formally, and nearly always infor-mally because of different labour market posi-tions, linked to different gender roles.According to many feminist authors, it is thesexual division of paid and unpaid work –especially care and domestic labour – thatneeds incorporating in the typology (Lewis,1992; O’Connor, 1993; Orloff, 1993;Sainsbury, 1996; O’Connor et al., 1999).

With respect to another issue, social care,Daly and Lewis (2000: 289) argue that differ-ent styles of social policy have incorporatedthe key element of social care differently. Theyidentify certain tendencies concerning care inspecific welfare states. For example, theScandinavian countries form a distinct groupin that they have strongly institutionalizedcare for both the elderly and children. In theMediterranean welfare states, care tends to beprivatized to the family whereas, in Germany,it is seen as most appropriately a function ofvoluntary service providers. In France, astrong distinction is made between care forchildren and for the elderly, with a strong col-lective sector in the former and little voluntaryinvolvement. Another form is found in theBeveridge-oriented welfare states – GreatBritain and Ireland – where a strong distinc-tion is also made between caring for childrenand caring for (elderly) adults. In the former –as opposed to the latter – little collectivizationhas taken place. Although they do not reallyclassify welfare states into actual clusters,Daly and Lewis make a strong case for usingsocial care as a critical dimension foranalysing variations.



As far as the gender gap in earnings is con-cerned, Gornick and Jacobs (1998) found thatEsping-Andersen’s regime-types do captureimportant distinctions among contemporarywelfare states. Their results showed that the size of the public sector, the extent of the public-sector earnings premium and theimpact of the public sector on gender differen-tials in wages all varied more across regimesthan within them. In this way, they showedthe fruitfulness of emphasizing the gender per-spective in Esping-Andersen’s classification ofwelfare states. Moreover, Trifiletti (1999)incorporated a gender perspective into Esping-Andersen’s classification by showing that asystematic relationship exists between thelevel of decommodification and whether thestate treats women as wives and mothers or asworkers. The latter is also an importantdimension identified by Lewis (1989).

Finally, Siaroff (1994) also argues that theexisting literature does not pay enough atten-tion to how gender inequality is embedded insocial policy and welfare states. In order toarrive at a more gender-sensitive typology ofwelfare state regimes, he examines a variety ofindicators of gender equality and inequality inwork and welfare. He compares the work–welfare choice of men and women (i.e.whether to partake in the welfare state or toengage in paid labour) across countries. Thisallows him to distinguish among a Protestantsocial-democratic, a Protestant liberal, anAdvanced Christian-democratic and a LateFemale Mobilization welfare state regime.Although the labels suggest otherwise, thistypology also shows a strong overlap with theEsping-Andersenian classification. Only thelatter type – the Late Female Mobilizationwelfare state regime – is an addition, whichresembles the previously distinguishedMediterranean type of welfare states.

Ideal and real-types

In Table 1, we ordered the types discussedabove broadly in accordance with the worlds

of welfare capitalism as defined by Esping-Andersen. For example, Bonoli’s Continentaltype is very much like Esping-Andersen’sConservative type; in both types contributionsplay a rather important role. Equally, Castlesand Mitchell’s Non-Right Hegemony typeshows a large amount of congruence withEsping-Andersen’s Social-democratic type,because of the high degree of universalism andequalization in social policy. We could go on,but we would like to raise another issue.

One may wonder whether, if the relation-ship among the different typologies is asstrong as we assume, this close correspon-dence of types will also be apparent in theactual clustering of countries. Although notevery classification developed by these authorscovers the same nations, there is a rather largeoverlap which makes it possible to answer thisquestion. For that purpose, Table 2 shows theextent to which the ideal-types – constructedby Esping-Andersen’s critics – coincide withhis own ideal-types. We then added the ideal-types, proposed by these critics, placingrelated ideal-types, when possible, under oneheading. This results in five – instead of theoriginal three – worlds of welfare capitalismand answers our original question. Next, inTable 2 we arranged the real-types accordingto the different ideal-types, thereby followingthe suggestions of the different authors.

It appears that, even when one uses differ-ent indicators to classify welfare states, somecountries emerge as standard examples,approximating certain ideal-types. The UnitedStates is, according to everyone’s classifica-tion, the prototype of a welfare state whichcan best be denoted as liberal (with or withoutthe suffix: Protestant, Anglo–Saxon or basicsecurity). Germany approaches theBismarckian/Continental/conservative ideal-type and Sweden approximates the social-democratic ideal-type (Scandinavian/Nordic).

However, consensus seems to end here. Forexample, according to some, Italy can best beassigned to the second, corporatist/continen-tal/conservative type, but belongs, accordingto others, along with Greece, Spain and



Three w

orlds of welfare capitalism

or more?

149


olicy 2002 12 (2)

Table 2 Classification of countries according to seven typologies

Type

I II III IV V

Esping-Andersen Liberal Conservative Social-democratic(Decommodification)

• Australia • Italy • Austria• Canada • Japan • Belgium• United States • France • Netherlands• New Zealand • Germany • Denmark• Ireland • Finland • Norway• United Kingdom • Switzerland • Sweden

Leibfried Anglo–Saxon Bismarck Scandinavian Latin Rim• United States • Germany • Sweden • Spain• Australia • Austria • Norway • Portugal• New Zealand • Finland • Greece• United Kingdom • Denmark • Italy

• France

Castles & Mitchell Liberal Conservative Non-Right Hegemony Radical• Ireland • West-Germany • Belgium • Australia• Japan • Italy • Denmark • New Zealand• Switzerland • Netherlands • Norway • United Kingdom• United States • Sweden

Siaroff Protestant Liberal Advanced Christian-democratic Protestant Social-democratic Late Female Mobilization• Australia • Austria • Denmark • Greece• Canada • Belgium • Finland • Ireland• New Zealand • France • Norway • Italy• United Kingdom • West-Germany • Sweden • Japan• United States • Luxembourg • Portugal

• Netherlands • Spain• Switzerland

Ferrera (Europe only) Anglo–Saxon Bismarckian Scandinavian Southern• United Kingdom • Germany • Sweden • Italy• Ireland • France • Denmark • Spain

• Belgium • Norway • Portugal• Netherlands • Finland • Greece• Luxembourg• Austria• Switzerland continued over

150A

rts and Gelissen


olicy 2002 12 (2)

Table 2 continued

Type

I II III IV V

Bonoli (Europe only) British Continental Nordic Southern• United Kingdom • Netherlands • Sweden • Italy• Ireland • France • Finland • Switzerland

• Belgium • Norway • Spain• Germany • Denmark • Greece• Luxembourg • Portugal

Korpi & Palme Basic Security Corporatist Encompassing Targeted• Canada • Austria • Finland • Australia• Denmark • Belgium • Norway• Netherlands • France • Sweden• New Zealand • Germany• Switzerland • Italy• Ireland • Japan• United Kingdom• United States

Note: Underlined countries indicate a prototype.

Portugal to a distinctive Mediterranean type.The same holds for Australia which mayeither be classified as liberal or is the proto-type of a separate, radical welfare state.Nevertheless, as far as these countries andtypes are concerned, consensus is strongerthan was initially assumed. One must,however, recognize that discussions are mainlyconcerned with whether certain types ofwelfare states are either separate categories orare subgroups of certain main types.

Hybrid cases are a bigger problem. TheNetherlands and Switzerland are clear exam-ples of this. If we take, for example, a closerlook at the Dutch case, we see that Esping-Andersen (1990) originally assigned theNetherlands to the social-democratic type,whereas Korpi and Palme see it as liberallyoriented; the basic security type. However,most authors place the Netherlands in thesecond category of corporatist/continental/conservative welfare states. This is also thechoice of Visser and Hemerijck (1997),perhaps the foremost specialists on the Dutchwelfare state. Curiously enough, this is doneusing Esping-Andersen’s work as a constant,positive reference. If we have another look atEsping-Andersen’s work, this is not as surpris-ing as one would expect. It is true that theNetherlands is rated relatively high on social-democratic characteristics, but not exception-ally low on liberal and conservativecharacteristics. Recently, Esping-Andersen hascalled the Netherlands the ‘Dutch enigma’because of its Janus-faced welfare regime(1999: 88). The Netherlands is indeed more ahybrid case than a prototype of a specificideal-type. If one attaches more importance tocertain attributes than to others – and addsother characteristics or substitutes previousones – then it is easy to arrive at different clas-sifications.

Empirical robustness of the three-wayclassification

Esping-Andersen claims that if we rate real

welfare states along the dimensions of degreeof decommodification and the modes of strati-fication, three qualitatively different clusterswill appear. Alongside the more fundamentalcriticism of his three-way classification – thatEsping-Andersen employs faulty criteria todemarcate a regime – the empirical fit of histhree-way classification has also been ques-tioned. Several authors have tested the good-ness-of-fit of the three-way regime typology.In the following, we discuss their findings,which are presented in Table 3.

In an effort to evaluate the possible extentto which quantitative techniques – OLSregression and cluster-analysis – suggest thesame conclusions as alternative qualitativeapproaches – such as ‘BOOLEAN’ compara-tive analysis – Kangas (1994) found somesupport for the existence of Esping-Andersen’sdifferent welfare state regimes. Specifically,cluster-analyses of data on characteristics ofhealth insurance schemes in OECD countriesin 1950 and 1985 corroborated his conjec-tures. However, the results also showed theexistence of two subgroups within the groupof liberal welfare states, which largelyaccorded with the classification of Castles andMitchell (1993).

Ragin (1994) also tested Esping-Andersen’sclaim of a three-world classification. By apply-ing a combination of cluster-analysis and‘BOOLEAN’ comparative analysis to charac-teristics of pension systems, he determinedwhich, if any, of Esping-Andersen’s threeworlds of welfare capitalism each countryfitted best. His cluster analysis suggested theexistence of a social-democratic cluster, a cor-poratist cluster and, finally, a rather large‘spare’ cluster, which accommodates cases thatdo not conform to Esping-Andersen’s threeworlds. On the basis of his findings, Raginconcludes that the three-worlds scheme doesnot capture existing diversity as adequately asone would wish.

Shalev (1996) applied factor analysis to 14pension policy indicators collected by Esping-Andersen, to test for the presence of liberal,social-democratic and corporatist regime-



152A

rts and Gelissen


olicy 2002 12 (2)

Table 3 Empirical robustness of the three-worlds typology

Number of clusters and cluster assignment Method of analysis

Kangas (1994) 1. Liberal: United States, Canada Cluster analysis2. Conservative: Austria, Germany, Italy, Japan, the Netherlands3. Social-democratic: Denmark, Finland, Norway, Sweden4. Radical: Australia, Ireland, New Zealand, United Kingdom

Ragin (1994) 1. Liberal: Australia, Canada, Switzerland, United States BOOLEAN comparative analysis2. Corporatist: Austria, Belgium, Finland, France, Italy3. Social-democratic: Denmark, Norway, Sweden4. Undefined: Germany, Ireland, Japan, the Netherlands, New Zealand,

United Kingdom

Shalev (1996) 1. Liberal: United States, Canada, Switzerland, Japan Factor analysis2. Conservative: Italy, France, Belgium, Austria, Ireland3. Social-democratic: Sweden, Norway, Denmark, Finland4. Undefined: Germany, the Netherlands, United Kingdom, Australia,

New Zealand

Obinger & Wagschal (1998) 1. Liberal: United States, Canada, Japan, Switzerland Cluster analysis2. European: Belgium, Germany, Finland, Ireland, United Kingdom,

the Netherlands3. Conservative: France, Italy, Austria4. Social-democratic: Denmark, Norway, Sweden5. Radical: Australia, New Zealand

Wildeboer Schut et al. (2001) 1. Liberal: United States, Canada, Australia, United Kingdom Principal component analysis2. Conservative: France, Germany, Belgium3. Social-democratic: Sweden, Denmark, Norway4. Undefined: the Netherlands

types. This factor analysis showed that theintercorrelations among these social policyindicators were dependent on two dimensions.The first factor measured the level of social-democratic features, whereas the seconddimension measured corporatist features ofwelfare states. Based on the assignment offactor scores to individual nations, Shalevconcluded that his findings were in close cor-respondence with Esping-Andersen’s charac-terizations of the three welfare state regimes.He admitted, however, that some countries aredifficult to classify.

Using cluster-analysis, Obinger and Wagschal(1998) tested Esping-Andersen’s classificationof welfare state regimes using the stratifica-tion-criterion. After a detailed re-analysis ofEsping-Andersen’s original data on stratifica-tion, they concluded that these data are bestdescribed by five regime-clusters. In additionto Esping-Andersen’s conservative, liberal andsocial-democratic types, they distinguish aradical and a hybrid European cluster.

The most recent attempt to empirically cor-roborate Esping-Andersen’s classification hasbeen undertaken by Wildeboer Schut et al.(2001). This study examined the actual simi-larities and differences among the welfarestate regimes of the countries originallyincluded in Esping-Andersen’s classification.For these countries, 58 characteristics of thelabour market, tax regime and social protec-tion system at the beginning of the 1990s werecollected. These were submitted to a non-linear principal component-analysis. Theresults largely confirmed the three-regimetypology of Esping-Andersen.

Summing up, Esping-Andersen’s originalthree-worlds typology neither passes theempirical tests with flying colours, nor dis-mally fails them. The conclusion is, first, thathis typology has at least some heuristic anddescriptive value, but also that a case can bemade for extending the number of welfarestate regimes to four, or even five. Second,these analyses show that a significant numberof welfare states must be considered hybridcases: no particular case can ever perfectly

embody any particular ideal-type (Goodin etal., 1999: 56). Third, if one looks at othersocial programmes than the ones applied byEsping-Andersen, it becomes clear that theydo not conform so easily – if at all – to hiswelfare regime patterns (Gough, 2000a: 4).

Conclusion and discussion

Before we reach our conclusions, let usexamine how Esping-Andersen himself hasreacted to the various attempts to amend histypology. The problem is that after consider-able discussion it seems impossible for him tomake up his mind once and for all. Initially,Esping-Andersen (1997) reacted, for example,positively to Castles and Mitchell’s proposalto add a fourth type – a radical welfare stateregime – to his typology. He recognized thatthe residual character and the matter of ameans test are just one side of the coin of theAntipodean welfare states. However, he feltthat a powerfully institutionalized collectionof welfare guarantees, which operate throughthe market itself, could not be neglected. Lateron, however, he argued that the passage oftime is pushing Australia, Great Britain andNew Zealand towards what appears to beprototypical liberalism (Esping-Andersen,1999). At first he also partially supported theproposal to add a separate Mediterraneantype to his typology (Esping-Andersen, 1996:66; 1997: 171). He acknowledged the – some-times generous – benefits which are guaran-teed by certain arrangements, the near absenceof social services and, especially, the Catholicimprint and high level of familialism. Fromthe feminist critics he learned not so much theoverarching salience of gender as the analyti-cal power that a re-examination of the familycan yield. Recently he argued that the acid testof a distinct Mediterranean model depends onwhether families are the relevant focus ofsocial aid, and whether families will fail justas markets and states can fail (Esping-Andersen, 1999: 90).

All in all, Esping-Andersen is very reluctant



to add more regime-clusters to his originalthree. Against the benefits of greater refine-ment, more nuance and more precision, heweighs the argument of analytical parsimony,stressing that ‘the peculiarities of these casesare variations within a distinct overall logic,not a wholly different logic per se’ (Esping-Andersen, 1999: 90).

The answer to the question of whetherEsping-Andersen’s three-type or a derivativeor alternative four, or five-type typology ispreferable depends, however, not only on par-simony and verisimilitude. It also depends onwhether these typologies lead to a theoreti-cally more satisfying and empirically morefruitful comparative analysis of welfare stateregimes. As far as theory construction is con-cerned, Baldwin (1996: 29) has argued thatwhen asking about typologies, whether ofwelfare states or anything else, we must asknot just what but also why. Esping-Andersen’stentative answer to the question of why threedifferent welfare state regime types emergedhas been sketched earlier on in this paper.Different welfare regimes are shaped by differ-ent class coalitions within a context of inher-ited institutions. This answer is embedded in apower-resources mobilization paradigm. Thetentative answer to the question of whyregime shifts are scarce is that a national statecannot easily escape its historical inheritance.Institutional inertia is one factor why differentwelfare state regime types persist, and pathdependency is another (Kohl, 2000: 125;Kuhnle and Alestalo, 2000: 9). Korpi andPalme – and some feminist authors – work inthe same power-resources mobilization tradi-tion as Esping-Andersen. It would be worth-while to develop a theoretical reconstructionof the different contributions of this paradigm(for an initial impetus to such an endeavour,see Schmidt, 1998: 215–28). Only then couldthe explanatory value of the typology becomeapparent.

Whether there is, within the welfare model-ling business, an alternative available toEsping-Andersen’s power-resource mobiliza-tion cum institutional inertia/path dependency

theory is difficult to determine. The work ofsome of the other authors we discussed in thisoverview has a strong empiricist flavour.However, if we are searching for an underly-ing theoretical notion, it can be found in therather general statement that similar causeshave similar consequences. Considering thelabels these authors have put to the prototypesthey distinguish, which are predominantlygeographical and ideological in nature, themost important causes are seen to be the pres-sure of functional exigencies and the diffusionof innovations (Goldthorpe, 2000: 54). Thefirst factor could be translated into a ‘chal-lenge response’ hypothesis. The challengesproduced by the force of similar circumstances(characteristics of pre-industrial social struc-tures, political institutions, degree of homo-geneity of population, culture, problemperceptions and preferences) lead to compara-ble welfare state regimes (responses) (Kuhnleand Alestalo, 2000: 7). The second factorcould be put in terms of learning effects inpolicy making. New ideas, new solutions areoften a product of a diffusion process. Theyhit political systems and societies at differentpoints in ‘developmental time’. As far as thislatter factor is concerned, Boje (1996: 15)argues that the fact that most welfare statesare confronted with huge social problems hasnecessitated politicians finding alternativeprocedures, which may solve these problemsmore efficiently. Politicians have come torealize that much may be learned from otherwelfare states.

Castles (1993; 1998) too underscores theimportance of both factors. He argues that itis likely that policy similarities and differencesamong welfare states can be attributed to boththe force of circumstances and to diffusion. Asfar as the latter factor is concerned we can dis-tinguish the institutional arrangements andculture of prototypical welfare states and theirtransmission and diffusion to other countries.Regarding the former factor, we can observethe immediate impact of economic, politicaland social variables identified in the contem-porary public policy literature. Whether these



very general ‘challenge response’ and ‘diffu-sion’ hypotheses will be further developedremains to be seen. For the moment, we canconclude that, given the empiricist nature ofthe work of the authors who provided alterna-tive typologies, there should be hardly anyobjection – for the time being – to the incor-poration of their findings into a power–resources mobilization paradigm.

Finally, we arrive at the empirical fruitful-ness of the typology. In his overview,Abrahamson (1999) concludes that as anorganizing principle for comparative studiesof welfare states the typologies have proven tobe a very robust and convincing tool. Withinthe power–resources mobilization paradigm(Korpi, 1983; Esping-Andersen and Korpi,1984; Esping-Andersen, 1990) it has beenproposed that the nature of the welfare stateregime would decisively influence support forcertain forms of social policy. A type that ischaracterized by universalism would generatethe strongest support, whereas arrangementswhich apply only to minorities would notsucceed in winning the support of majorities.Tests of this hypothesis (Papadakis and Bean,1993; Peillon, 1996; Gelissen, 2000; Gevers etal., 2000) have shown some empirical support,but the evidence is not really encouraging.More encouraging were the results of an effort(Gundelach, 1994) to explain cross-nationaldifferences in values with respect to welfareand care using the Esping-Andersenian welfarestate regimes. Also, Svallfors’s (1997) and Artsand Gelissen’s (2001) tests of the hypothesisthat different welfare state regimes matter forpeople’s attitudes towards income-redistribu-tion, were strongly endorsed. What especiallymatters to us here is that Svallfors distributivejustice and solidarity had included not onlyEsping-Andersen’s regime-types, but alsoother types and Arts and Gelissen.

It is more difficult to draw a conclusionconcerning the influence that welfare stateregimes have on social behaviour and theireffects. Much of this research has a bearing onthe distributive effects of welfare state regimes.Because they are often described in terms of

their intended social stratification, a tautologi-cal element easily sneaks into the explana-tions. Positive exceptions are Goodin et al.(1999) and Korpi and Palme (1998). Usingpanel-data, Goodin et al. (1999) show thatwelfare state regimes do not only haveintended results, but also generate unintendedconsequences. As intended and expected, thesocial-democratic regime succeeds best in real-izing its fundamental value: minimizing inequal-ity. But this regime is also at least as good inpromoting the goals to which other regimesostensibly attach most importance. Specifically,the social-democratic regime also does verywell in reducing poverty – a goal which is pri-oritized by the liberal welfare state regime –and in promoting stability and social integra-tion, which is the home ground of the corpo-ratist welfare state regime. Korpi and Palme(1998) find that institutional differences leadto a paradox of redistribution: the more bene-fits are targeted at the poor and the more thecreation of equality through equal public trans-fers to all is a matter of priority, the less povertyand equality will be reduced. Thus, institu-tional arrangements characteristic of certainwelfare state regimes not only have unintendedconsequences, but even perverse effects.

All in all, these conclusions provide suffi-cient impetus to continue the work concerningthe resulting welfare state typology. A betterformulation of the theory on which it is baseddeserves priority. Only then can predictions belogically – instead of impressionistically –deduced from theory. Only then is a strict testof the theory possible and only then will the heuristic and explanatory value of thetypology become apparent.

Acknowledgements

We would like to thank Ian Gough and fouranonymous reviewers for valuable comments.

Notes

1 This is an exaggeration. Schmidt (1998),Abrahamson (1999) and Arts and Gelissen



(1999) have convincingly argued that the evi-dence is otherwise.

2 The importance of Catholicism is emphasized byvan Kersbergen (1995) in his discussion includ-ing Christian-democratic nations such asGermany, Italy and the Netherlands in main-stream welfare state typologies.

3 For reasons of conciseness we refrain from thedebate regarding Esping-Andersen’s classifica-tion of Japan as a liberal welfare state. For areaction to this critique see Esping-Andersen(1997; 1999). Becker (1996) and Goodman andPeng (1996) are even of the opinion that Japanbelongs to a sixth prototype of welfare stateregimes, the so-called East-Asian welfare states.We acknowledge the importance of these argu-ments, but cannot engage with them here(Gough, 2000a; 2000b).

4 Gornick and Jacobs (1998: 691) point out thatEsping-Andersen himself argues that eachregime-type is associated with women’s employ-ment levels. Specifically, he (Esping-Andersen,1990) expects that women’s employment rateswill be highest in social-democratic countries,whereas in liberal welfare states, moderate levelsof female employment will be found. The lowestlevels of women’s employment will be found inthe conservative welfare states.

References

Abrahamson, P. (1999) ‘The Welfare ModellingBusiness’, Social Policy & Administration 33 (4):394–415.

Arts, W. A. and Gelissen, J. (1999) ‘Verzorgings-staten in soorten: Op zoek naar ideaal- en reëletypen’, Mens & Maatschappij 74 (2): 143–65.

Arts, W. A. and Gelissen, J. (2001) ‘Welfare States,Solidarity and Justice Principles: Does the TypeReally Matter?’ Acta Sociologica 44 (4):283–300.

Baldwin, P. (1996) ‘Can We Define a EuropeanWelfare State Model?’, in B. Greve (ed.)Comparative Welfare Systems: the ScandinavianModel in a Period of Change, pp. 29–44.London: Macmillan.

Becker, U. (1996) ‘Over de typologie van welvaarts-bestellen: Esping-Andersens theorie in discussie’,Beleid & Maatschappij 23 (1): 19–30.

Boje, T. (1996) ‘Welfare State Models inComparative Research: Do the Models Describethe Reality?’, in B. Greve (ed.) ComparativeWelfare Systems: the Scandinavian Model in aPeriod of Change, pp. 13–27. London:Macmillan Press.

Bonoli, G. (1997) ‘Classifying Welfare States: aTwo-dimension Approach’, Journal of SocialPolicy 26 (3): 351–72.

Bulmer, M. and Rees, A. M. (1996) ‘Conclusion:Citizenship in the Twenty-first Century’, in M.Bulmer and A. M. Rees (eds) Citizenship Today.The Contemporary Relevance of T. H. Marshall,pp. 269–84. London: UCL Press.

Castles, F. G. (1993) ‘Introduction’, in F. G. Castles(ed.) Families of Nations: Patterns of PublicPolicy in Western Democracies. Aldershot:Dartmouth.

Castles, F. G. (1998) Comparative Public Policy:Patterns of Post-war Transformation. Cheltenham:Edward Elgar.

Castles, F. G. and Mitchell, D. (1993) ‘Worlds ofWelfare and Families of Nations’, in F. G. Castles(ed.) Families of Nations: Patterns of PublicPolicy in Western Democracies. Aldershot:Dartmouth Publishing Company.

Cnaan, R. A. (1992) ‘Three Worlds of WelfareCapitalism’, Acta Sociologica 35 (1): 69–71.

Daly, M. and Lewis, J. (2000) ‘The Concept of SocialCare and the Analysis of Contemporary WelfareStates’, British Journal of Sociology 51 (2): 281–98.

de Swaan, A. (1988) In Care of the State: HealthCare, Education and Welfare in Europe and the USA in the Modern Era. Cambridge: PolityPress.

Esping-Andersen, G. (1990) The Three Worlds ofWelfare Capitalism. Oxford: Polity Press.

Esping-Andersen, G. (1993) ‘The ComparativeMacro-sociology of Welfare States’, in L. Moreno(ed.) Social Exchange and Welfare Development,pp. 123–36. Madrid: Consejo Superior deInvestigaciones Científicas.

Esping-Andersen, G. (1994) ‘Welfare States and theEconomy’, in N. J. Smelser and R. Swedberg(eds) The Handbook of Economic Sociology, pp.711–32. Princeton/New York: PrincetonUniversity Press/Russel Sage Foundation.

Esping-Andersen, G. (1996) ‘Welfare States withoutWork: the Impasse of Labour Shedding andFamilialism in Continental European SocialPolicy’, in G. Esping-Andersen (ed.) WelfareStates in Transition, pp. 66–87. London: Sage.

Esping-Andersen, G. (1997) ‘Hybrid or Unique?The Japanese Welfare State between Europe andAmerica’, Journal of European Social Policy 7(3): 179–89.

Esping-Andersen, G. (1999) Social Foundations ofPost-industrial Economies. Oxford: OxfordUniversity Press.

Esping-Andersen, G. and Korpi, W. (1984) ‘SocialPolicy as Class Politics in Post-war Capitalism’,in J. Goldthorpe (ed.) Order and Conflict in Con-temporary Capitalism. Oxford: Oxford UniversityPress.

Ferrera, M. (1996) ‘The “Southern” Model ofWelfare in Social Europe’, Journal of EuropeanSocial Policy 6 (1): 17–37.



Flora, P. (1983) The Growth of Mass Democraciesand Welfare States, Vol. 1. Frankfurt am Main:Campus Verlag.

Flora, P. (1986) Growth to Limits: the WesternEuropean Welfare States since World War II, 5vols. Berlin: de Gruyter.

Flora, P. and Heidenheimer, A. J. (1981) TheDevelopment of Welfare States in Europe andAmerica. New Brunswick, NJ: TransactionBooks.

Gelissen, J. (2000) ‘Popular Support forInstitutionalised Solidarity: a Comparison amongEuropean Welfare States’, International Journalof Social Welfare 9 (4): 285–300.

Gevers, J., Gelissen, J., Arts, W. and Muffels, R.(2000) ‘Public Health Care in the Balance:Exploring Popular Support for Health CareSystems in the European Union’, InternationalJournal of Social Welfare 9 (4): 301–21.

Goldthorpe, J. H. (2000) On Sociology: Numbers,Narratives, and the Integration of Research andTheory. Oxford: Oxford University Press.

Goodin, R. E., Heady, B., Muffels, R. and Dirven,H-J. (1999) The Real Worlds of WelfareCapitalism. Cambridge: Cambridge UniversityPress.

Goodman, R. and Peng, I. (1996) ‘The East AsianWelfare States: Peripatetic Learning, AdaptiveChange, and Nation-building’, in G. Esping-Andersen (ed.) Welfare States in Transition, pp.192–224. London: Sage.

Gornick, J. C. and Jacobs, J. A. (1998) ‘Gender, theWelfare State, and Public Employment: aComparative Study of Seven IndustrializedCountries’, American Sociological Review 63 (5):688–710.

Gough, I. (2000a) ‘Welfare Regimes in East Asiaand Europe: Comparisons and Lessons’, retrieved9 March 2001 from the World Wide Web:http://www.bath.ac.uk/Faculties/HumSocSci/IFIPA/GSP/. University of Bath: Institute forInternational Policy Analysis.

Gough, I. (2000b) ‘Welfare Regimes: on Adaptingthe Framework to Developing Countries’,retrieved 9 March 2001 from the World WideWeb: http://www.bath.ac.uk/Faculties/HumSocSci/IFIPA/GSP/. University of Bath: Institute forInternational Policy Analysis.

Gundelach, P. (1994) ‘National Value Differences:Modernization or Institutionalization?’, Interna-tional Journal of Comparative Sociology 35(1–2): 37–58.

Hempel, C. G. (1965) [1952] ‘Typological Methodsin the Natural and the Social Sciences’, in C. G.Hempel (ed.) Aspects of Scientific Explanationand Other Essays in the Philosophy of Science,pp. 155–72. New York/London: The FreePress/Collier-Macmillan.

Hicks, A. (1991) ‘The Three Worlds of WelfareCapitalism’, Contemporary Sociology 20 (3):399–401.

Hill, M. (1996) Social Policy: a Comparative Ana-lysis. London: Prentice Hall/Harvester Wheatsheaf.

Kangas, O. E. (1994) ‘The Politics of Social Security:On Regressions, Qualitative Comparisons, andCluster Analysis’, in T. Janoski and A. M. Hicks(eds) The Comparative Political Economy of theWelfare State, pp. 346–64. Cambridge:Cambridge UP.

Katrougalos, G. S. (1996) ‘The South EuropeanWelfare Model: the Greek Welfare State in Searchof an Identity’, Journal of European Social Policy6 (1): 39–60.

Klant, J. J. (1984) The Rules of the Game: theLogical Structure of Economic Theories.Cambridge: Cambridge University Press.

Kohl, J. (1993) ‘Der Wohlfahrtsstaat in vergleichen-der Perspective’, Zeitschrift für Sozialreform 39:67–82.

Kohl, J. (2000) ‘Der Sozialstaat: Die deutscheVersion des Wohlfahrtsstaates – Überlegungen zuseiner typologischen Verortung’, in S. Leibfriedand U. Wagschal (eds) Der deutsche Sozialstaat:Bilanzen- Reformen-Perspectiven, pp. 115–52.Frankfurt/New York: Campus.

Korpi, W. (1983) The Democratic Class Struggle.London: Routledge & Kegan Paul.

Korpi, W. and Palme, J. (1998) ‘The Paradox ofRedistribution and Strategies of Equality: WelfareState Institutions, Inequality and Poverty in theWestern Countries’, American SociologicalReview 63 (5): 661–87.

Kuhnle, S. and Alestalo, M. (2000) ‘Growth,Adjustments and Survival of European WelfareStates’, in S. Kuhnle (ed.) Survival of theEuropean Welfare State, pp. 3–18. London/NewYork: Routledge.

Leibfried, S. (1992) ‘Towards a European welfare state? On Integrating Poverty Regimesinto the European Community’, in Z. Ferge and J. E. Kolberg (eds) Social Policy in a ChangingEurope. Frankfurt am Main: Campus Verlag.

Lessenich, S. and Ostner, I. (1998) Welten desWolhlfahrtskapitalismus: Der Sozialstaat in ver-gleichender Perspektive. Frankfurt/Main: CampusVerlag.

Lewis, J. (1989) ‘Lone Parent Families: Politics andEconomics’, Journal of Social Policy 18 (4):595–600.

Lewis, J. (1992) ‘Gender and the Development ofWelfare Regimes’, Journal of European SocialPolicy 2 (3): 159–73.

Marshall, T. H. (1950) Citizenship and Social Classand other Essays. Cambridge: Cambridge UP.

Marshall, T. H. (1963) Sociology at the Crossroadsand other Essays. London: Heinemann.



Marshall, T. H. (1965) Social Policy. London:Hutchinson.

Marshall, T. H. (1981) The Right to Welfare andother Essays. London: Heinemann.

Mommsen, W. J. (1981) The Emergence of theWelfare State in Britain and Germany. London:Croom Helm.

Obinger, H. and Wagschal, U. (1998) ‘DasStratifizierungskonzept in der ClusteranalytischenÜberprüfung’, in S. Lessenich and I. Ostner (eds)Welten des Wohlfahrtskapitalismus: DerSozialstaat in vergleichender Perspektive, pp.109–35. Frankfurt/Main: Campus Verlag.

O’Connor, J. S. (1993) ‘Gender, Class andCitizenship in the Comparative Analysis ofWelfare State Regimes: Theoretical andMethodological Issues’, The British Journal ofSociology 44 (3): 501–18.

O’Connor, J. S., Orloff, A. and Shaver, S. (1999)States, Markets, Families. Gender, Liberalism andSocial Policy in Australia, Canada, Great Britainand the United States. Cambridge: CambridgeUP.

Offe, C. (1991) ‘The Three Worlds of WelfareCapitalism’, American Journal of Sociology 96(6): 1,555–7.

Orloff, A. S. (1993) ‘Gender and the Social Rightsof Citizenship: a Comparative Analysis of GenderRelations and Welfare States’, AmericanSociological Review 58 (3): 303–28.

Papadakis, E. and Bean, C. (1993) ‘Popular Supportfor the Welfare State: a Comparison amongInstitutional Regimes’, Journal of Public Policy13 (3): 227–54.

Peillon, M. (1996) ‘A Qualitative ComparativeAnalysis of Welfare State Legitimacy’, Journal ofEuropean Social Policy 6 (3): 175–90.

Pierson, P. (1993) ‘When Effect becomes Cause:Policy Feedback and Political Change’, WorldPolitics 45 (4): 595–628.

Ragin, C. (1994) ‘A Qualitative ComparativeAnalysis of Pension Systems’, in T. Janoski and A.M. Hicks (eds) The Comparative PoliticalEconomy of the Welfare State, pp. 320–45.Cambridge: Cambridge UP.

Rogers, R. E. (1969) Max Weber’s Ideal TypeTheory. New York: Philosophical Library.

Sainsbury, D. (1996) Gender, Equality and WelfareStates. Cambridge: Cambridge UP.

Schmidt, M. G. (1998) Sozialpolitik in Deutschland:Historische Entwicklung und internationalerVergleich. Opladen: Leske & Budrich.

Shalev, M. (1996) The Privatization of Social

Policy? Occupational Welfare and the WelfareState in America, Scandinavia and Japan.London: Macmillan.

Siaroff, A. (1994) ‘Work, Welfare and GenderEquality: a New Typology’, in D. Sainsbury (ed.)Gendering Welfare States, pp. 82–100. London:Sage.

Svallfors, S. (1997) ‘Worlds of Welfare andAttitudes to Redistribution: a Comparison ofEight Western Nations’, European SociologicalReview 13 (3): 283–304.

Therborn, G. (1995) European Modernity andBeyond. The Trajectory of European Societies1945–2000. London: Sage.

Titmuss, R. (1958) Essays on the Welfare State.London: Allen & Unwin.

Titmuss, R. (1974) Social Policy. London: Allen &Unwin.

Trifiletti, R. (1999) ‘Southern European WelfareRegimes and the Worsening Position of Women’,Journal of European Social Policy 9 (1): 49–64.

van Kersbergen, K. (1995) Social Capitalism: aStudy of Christian Democracy and the WelfareState. London: Routledge.

Visser, J. and Hemerijck, A. C. (1997) A DutchMiracle: Job Growth, Welfare Reform andCorporatism in the Netherlands. Amsterdam:Amsterdam University Press.

von Kempski, J. (1972) ‘Zur Logik derOrdungsbegriffe, besonders in den Sozialwissen-schaften’, in H. Albert (ed.) Theorie und Realität.Ausgewählte Aufsätze zur Wissenschaftslehre derSozialwissenschaften, pp. 115–38. Tübingen: J.C. B. Mohr (Paul Siebeck).

Watkins, J. W. N. (1969) [1953] ‘MethodologicalIndividualism and Non-Hempelian Ideal Types’,in L. J. Kimmerman (ed.) The Nature and Scopeof Social Science: a Critical Anthology. NewYork: Appleton-Century-Crofts.

Weber, M. (1949) The Methodology of the SocialSciences. Glencoe, IL: Free Press.

Weber, M. (1968) [1922] Gesammelte Aufsätze zurWissenschaftslehre. Tübingen: J. C. B. Mohr(Paul Siebeck).

Wildeboer Schut, J. M., Vrooman, J. C. and de Beer,P. (2001) On Worlds of Welfare. Institutions andtheir Effects in Eleven Welfare States. The Hague:Social and Cultural Planning Office of theNetherlands.

Wilensky, H. L. (1975) The Welfare State andEquality: Structural and Ideological Roots ofPublic Expenditures. Berkeley/Los Angeles:University of California Press.



Documents

Three Worlds of Welfare Capital - Arts, Wil