journal of european social policy - wordpress.com · journal of european social journal of european...

http://esp.sagepub.comPolicy

Journal of European Social

1999; 9; 231 Journal of European Social PolicyJon Kvist

conformity to ideal typesWelfare reform in the Nordic countries in the 1990s: using fuzzy-set theory to assess

http://esp.sagepub.com/cgi/content/abstract/9/3/231 The online version of this article can be found at:

Published by:

http://www.sagepublications.com

can be found at:Journal of European Social Policy Additional services and information for

http://esp.sagepub.com/cgi/alerts Email Alerts:

http://esp.sagepub.com/subscriptions Subscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

© 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. at Socialforsknings Institut on July 14, 2007 http://esp.sagepub.comDownloaded from

http://esp.sagepub.com/cgi/alerts

http://esp.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://esp.sagepub.com

Summary

This article uses a new method for policyanalysis, fuzzy-set theory, which is aframework that allows for a precise opera-tionalization of theoretical concepts. Fuzzy-settheory is used to assess the conformity of theNordic countries to a pre-conceptualized ideal-typical Nordic welfare model. This permits usto assess recent welfare reform and judgewhether changes are of a qualitative or quanti-tative nature, i.e. whether reform amounts todifferences in kind or degree. Comparing thedevelopment of benefits in kind and cash withinthree welfare areas (families, the unemployedand the elderly) during the 1990s and acrossthe Nordic countries gives us an opportunity toassess patterns of welfare reform. The patternsof welfare reform are complex, but fuzzy-settheory permits the study of diversity. Despitenumerous changes, all the countries still belongto the Nordic welfare model, although tovarying degrees. Generally, Finland andSweden have implemented more cut-backsthan Denmark and Norway, and all countrieshave both expanded and contracted welfareprogrammes. Resilience at the national levelthus masks a differential development betweenwelfare areas and within welfare programmes.Tentatively, it seems that welfare policiesoperate within upper and lower limits which inturn are likely to vary over long time periodsand among different types of welfare states; themost generous programmes are liable to cut-backs and the least generous programmes toimprovements.

Résumé

Cette article recoure à une nouvelle méthode

pour guider une analyse politique, la théoriedes ‘ensembles flous’ (fuzzy-set). Cettedernière constitue une démarche particulière-ment adaptée lorsque l’on souhaite rendreopérationnel des concepts théoriques. Cettethéorie est utilisée ici pour évaluer la confor-mité des pays nordiques à l’idéal-type dumodèle social Nordique. Cela nous permetd’apprécier les réformes sociales récentes et dejuger si ces changements sont de nature quali-tative ou quantitative, c’est-à-dire si cesréformes reflètent des différences en termes detypes de modèle social ou en termes de niveauxde réalisation. La comparaison de l’augmenta-tion des avantages en nature et en espèces pourtrois catégories (les familles, les demandeursd’emploi et les personnes âgées) au cours desannées 1990 au sein des différentes paysnordiques nous offre l’occasion d’évaluer latendance des réformes sociales. Ces dernièressont complexes, mais la théorie des ‘ensemblesflous’ permet d’étudier leur diversité. En dépitdu nombre de changements, tous les pays par-ticipent encore du modèle social nordique bienque ce soit de manière plus ou moins franche.D’une façon générale, si la Finlande et la Suèdeont opéré davantage de coupes que leDanemark et la Norvège, tous les pays ontengagé des programmes sociaux plus étendusqu’auparavant. Les différences au niveaunational reflètent ainsi des évolutions dif-férentes entre catégories et au sein desprogrammes sociaux. Ainsi, il apparaît que lespolitiques sociales opèrent dans certaines lim-ites qui, elles-mêmes, varient au cours dutemps et en fonction des situations carac-térisant les différents pays: les programmes lesplus généreux conduisent à des marchesarrière et les programmes les moins généreuxamènent à des améliorations.

ARTICLE

WELFARE REFORM IN THE NORDIC COUNTRIES IN THE 1990s:USING FUZZY-SET THEORY TO ASSESS CONFORMITY

TO IDEAL TYPESJon Kvist, The Danish National Institute of Social Research, Copenhagen, Denmark

Journal of European Social Policy 0958-9287 (199908)9:3 Copyright © 1999 SAGE Publications, London, Thousand Oaksand New Delhi, Vol 9 (3): 231–252; 009139

009139 15/7/99 9:21 31


http://www.sagepub.co.uk/

http://www.sagepub.co.uk/journals/details/j0023.html


Introduction

Historically, welfare reform has been charac-terized by expansion in coverage andgenerosity. The rich literature on welfare statedevelopment suggests that welfare reform isdriven by economic development or politicsand looks at how institutions, in particularstate agencies, mediate economic and political pressures (see Ashford, 1986; Esping-Andersen, 1990; and Wilensky, 1975; or, foran overview, Huber et al., 1993). Accordingly,welfare reform is less expansionary, if notactually contracting, in countries with respect-ively poor economic performance, witheringpolitical mobilization of the Left and/orCatholic resources, and/or weak state capaci-ties. Whether indeed economics, politics orinstitutions matter in the same way and to thesame extent for recent welfare reform isanother issue since both the goals and contextof welfare reform have changed (see Pierson,1996).

The Nordic countries provide fertile soil forexamining this issue as they are relativelysimilar with regard to important institutionalfeatures such as comprehensive welfare states– illustrated by the term ‘Nordic welfaremodel’ – but different with regard to economicand political development in the 1990s. TheNordic model of welfare has long beenregarded as among the most modern andmature expressions of the welfare state andtherefore subject to interest beyond its geo-graphical borders (Einhorn and Logue, 1989).Recent economic, political and social develop-ments, however, have put a question markover the sustainability and desirability of thismodel. Indeed some advocate the trimming, oreven dismantling, of the Nordic model. Othersare of the opinion that the political changes inthe welfare state in the 1990s have alreadymade the Nordic model history. This article isrelated to these debates and aims to shed somelight on the actuality of change through anempirical investigation of welfare reform inthe Nordic countries in the 1990s.

The Nordic model of welfare

Within academic research, no consensus hascrystallized around what constitutes theNordic model of welfare. Indeed, its theor-etical conceptualization and empiricalexistence vary between areas such as health,voluntary work, and labour markets (see,respectively, Alban and Christensen, 1995;Klausen and Selle, 1995; Wadensjö et al.,1996). Most social policy researchers, how-ever, agree that the Nordic model ischaracterized by common objectives of wel-fare policy such as the promotion of solidarityand equality (Esping-Andersen, 1990). In turn,these objectives are achieved through compre-hensive and universal policies of a high quality(Esping-Andersen and Korpi, 1987). In thisway both means and ends are used to describethe constitutive elements of the Nordic welfaremodel.

We will define some main features of theNordic welfare model as:

• Comprehensiveness: the scope of publicpolicy is broad; the state has a larger rolevis-a-vis the market and civil society than isthe case in other countries.

• Full employment: policies are committed tocontributing to full (read: more) employ-ment and/or preventing unemployment,particularly long-term unemployment.

• Equality: policies are committed to con-tributing to equality between groups basedon gender, age, class, family situation, eth-nicity, religion, region and so forth.

• Universality: right to basic social securitybenefits (in cash and kind) in a wide range of social contingencies and life situ-ations.

• High-quality benefits: services are of a highquality, and provided by welfare pro-fessionals (see Kohlberg, 1991).

• Generous benefits: cash transfers are gen-erous, in particular for low-income groups,to allow for a ‘normally’ accepted standardof living.

232 JON KVIST

Journal of European Social Policy 1999 9 (3)

009139 15/7/99 9:21 Page 232



Needless to say, this definition is of the ideal-typical Nordic welfare model. What is more,these various aspects are thought to interactand reinforce each other: only together do theyconstitute the whole that we may describe asthe Nordic welfare model (Esping-Andersenand Korpi, 1987). Accordingly, to shed lighton the pattern of recent changes to the Nordicwelfare states, one should adopt a holisticview and simultaneously examine the direc-tions, scope and interconnections of allwelfare-related change.

We have chosen three welfare policy areaswhich are essential for an understanding of thenature of the Nordic welfare model and thechanges taking place, namely child and familysupport, unemployment measures and the wel-fare of the elderly (but see also Kautto et al.,1999). At the same time, these areas are ofparticular interest when looking at the press-ures on the Nordic welfare states.

Child and family support is an area wheresocial care services, rather than income trans-fers, can be said to be the ‘key to the Nordicwelfare model’ (Sipilä, 1997). All westernEuropean welfare states have comprehensivecash benefits systems, although their coverageand income compensation differ considerably,but few have developed publicly supportedsocial care services to the extent seen in theNordic countries. In particular, their child andfamily support policies serve to promoteemployment and gender equality (but seeLeira, 1992). At the moment, there is alsointense discussion in all four countries as tothe role of the family and other parts of civilsociety vis-a-vis the welfare state. Popularpressure is mounting for adult members offamilies to be able not only to opt for work,but also to care for their own offspring, and tofind better ways of reconciling work andfamily life.

Unemployment measures constitute an areawhere cash benefits and services, in principle,work closely together to provide a safety netor income compensation, and help facilitateentry or re-entry into the labour market. InScandinavia, due to an emphasis on labour

market insertion through publicly supportedjob and education programmes, the schemeshave been said to differ from those in mostother countries. The sustainability of theNordic model is also dependent on low unem-ployment to increase revenue and reduceexpenditures. In recent years, much talk hasbeen about how to remove work disincentives,reduce structural unemployment and preventlong-term unemployment, in particular amongyoung people.

Welfare for elderly people is the last and inmoney terms most expensive area. As inother industrialized countries, the Nordiccountries’ populations are ageing. Increasingcosts for old-age pensions combined withmore elderly people getting older and techno-logical advances in health and social caremean that there is much pressure for theirshare of social expenditure to grow further.Perhaps due to risk-averse policymakers, thisissue has not been as vigorously discussed asthe other two welfare areas. Nevertheless,together with unemployment measures, wel-fare for the elderly makes up more than halfof public social budgets, making themnatural if not unavoidable targets for budgetcuts.

The following section sets out the methodand material used. The next three sectionsconceptualize the Nordic model within a wel-fare area and analyse the Nordic countries’conformity in the 1990s. We discuss theoverall development, and conclude, more gen-erally, by discussing the potential for usingfuzzy-set theory in policy analysis, and in com-parative studies in particular.

Method and material

The distinction between qualitative andquantitative methods is conventional withinsocial science and also in comparative welfarestate research. The strength of qualitativemethods is their in-depth understanding of


WELFARE REFORM IN THE NORDIC COUNTRIES 233

009139 15/7/99 9:21 Page 233



cases whereas their Achilles’ heel is the limitedrepresentativeness and thus unsuitability fordrawing generalizations from one case toanother. In contrast, quantitative methods aregood for making generalizations, but do nottend to provide informed understanding of thecases in hand (Ragin, 1994). Perhaps becauseof these inherent features, neither method hasbeen successful in assessing the recent changeof welfare states.

Here we propose a method that allows us toexamine qualitative and quantitative aspectssimultaneously. The method has been devel-oped by Charles Ragin to study social diversityin comparative studies (Ragin, forthcoming).Ragin advocates using fuzzy-set theory in thesocial sciences as it has a number of advan-tages over conventional quantitativevariable-oriented and qualitative case-orientedresearch. At the core of fuzzy-set social scienceis a perception of cases as configurations ofaspects so that a difference in one aspect mayconstitute a difference in kind and not just indegree. At the same time, the fuzzy-setapproach allows partial membership of a casein a given configuration. Consequently, usingthe fuzzy-set approach allows us to study dif-ferences in both kind and degree at the sametime – what is known as ‘diversity’. Amongother things, this makes it possible to evaluatecases relative to their membership of specifiedideal types. The ideal type is – in a Weberiansense – an analytical construct that serves as ayardstick for social researchers to determinethe extent to which real empirical phenomenaare similar and how they differ from some pre-defined measure (Weber, 1949). For ourpurposes, it means we can measure the con-formity of countries to the Nordic welfaremodel and changes therein.

Using fuzzy-set social science to study theconformity of specific national systems to idealtypes is quite straightforward. It entails fourbasic steps:

1. Informed by theoretical and substantiveknowledge, identify aspects (equal to sets) ofthe ideal type leading to the construction of a

useful property space. This step is prior tousing fuzzy-set social science.2. Specify the cases’ membership scores in thesets comprising this property space, i.e. scoresreflecting the degrees to which cases are in orout of sets where 0 is fully out, 1 is fully in,and 0.5 is the cross-over point, being neithermore in nor more out.3. Compute the membership of each case inthe ideal-typical model, i.e. the given crisplocation in the property space, using the prin-ciples of fuzzy-set theory.4. Evaluate the homogeneity of cases by usingthe information from the previous step tomeasure the conformity of each case to theideal-typical instance.

From theories we may be able to select theattributes, conditions or aspects that consti-tutes the ideal type, here the Nordic welfaremodel. These aspects are in turn transformedinto sets. For example, previous research hasidentified ‘universality’ and ‘generosity’ asconstitutive aspects of the Nordic welfaremodel so we make two sets, one on ‘univer-sality’ and one on ‘generosity’. Moreover, thepossible combinations of sets form the so-called ‘multi-dimensional property space’. Inour example, we have a two-dimensionalproperty space where * = and (also known as‘set intersection’) and ~ = not: universal*gen-erous; universal*~generous; ~universal*generous; ~universal*~generous, e.g. ‘~uni-versal*generous’ reads ‘not universal andgenerous’. These combinations of sets are cor-ners in the property space, and we will refer tothem as ‘ideal-typical locations’, as they can beunderstood as expressing ideal types sincecountries will rarely have full membershipscores in two or more aspects. The number ofideal-typical locations is given by 2k, where kis the number of sets (Ragin, forthcoming).

The membership of a case – here a country– can vary from being fully in to fully out ofthe sets. By allowing for partial membership,sets become ‘fuzzy’ in contrast to ‘crisp’. Incrisp sets cases are either in to a value of 1 (oryes) or out to a value of 0 (or no). Like crisp

234 JON KVIST


009139 15/7/99 9:21 Page 234



sets, fuzzy sets operate with 0 and 1 as quali-tative assignments. In fuzzy sets 0 indicatesfully out and 1 equals fully in. The reality formany cases, however, is somewhere inbetween. The majority of European countries,for example, may neither qualify to be fullyout of generosity nor fully in. In fuzzy-setsocial science, such cases are partial in a setand are given values between 0 and 1 to indi-cate their degree of membership. In this waythe use of fuzzy sets permits us to study bothqualitative and quantitative variations simul-taneously. In our example, fairly generousbenefits in a country may qualify it to be in theset of generous countries, but less so thancountries with more generous benefits.Through the combination of qualitative andquantitative assessment fuzzy sets allow us tostudy complex diversity.

The principle of negation is a useful prin-ciple in fuzzy-set theory. If, for example, acountry scores 0.6 in the generosity set its cor-responding score in the non-generosity set is0.4, which is found by subtracting the gen-erosity score from 1. This follows the logic ofpartial membership, i.e. that cases can bepartly in and partly out of a given aspect at thesame time. To the extent that a country is notfully generous it is somewhat non-generous.

Countries’ membership scores, so-called‘crisp locations’ in the property space, reflecttheir membership of ideal types. These mem-bership scores are computed according to theprinciples of fuzzy-set theory (see Ragin, forth-coming). According to the minimum principlein fuzzy-set theory, the conformity of a case toan ideal-typical location in a property space isgiven by the minimum value of scores in thesets involved. This goes against the frequentuse made of averages, medians and similarprinciples in social science. But in fuzzy-setlogic a high degree of, for example, univer-sality does not lead to higher degrees ofgenerosity; the lowest score dictates the mem-bership in the set of universal and generouscountries. The chain is no stronger than itsweakest link.

The fuzzy-set approach provides us with

some advantages in relation to traditionalcase-oriented methods and the variable-oriented methods (Ragin, 1994). Imaginehaving two generous welfare states, where oneis universal and the other not. According tomost variable-oriented research this may beseen as two examples of the same case, gen-erous welfare states with varying degree ofuniversality. For case-oriented research twosuch cases will often be seen as distinctly dif-ferent, but their similarities and differenceswould be unquantifiable. Like the case-oriented research, the fuzzy-set approach seesthese two welfare states as different in kind,but it can formalize statements about theirsimilarities and differences. This is done byexamining their crisp locations, one case beingequal to the Nordic welfare model (the ideal-typical location universal*generous) and theother resembling another model (~universal*generous) which perhaps, in turn, could besaid to be an expression for the Bismarckianwelfare model.

Since the fuzzy-set approach allows us toassess degrees of membership in sets,researchers can use this information to eval-uate the way cases are located in relation tothe ideal-typical locations and in relation toeach other (Ragin, forthcoming). Thereby, wecan evaluate whether, for example, the Nordiccountries are comparable as instances of theNordic welfare model in an ideal-typical sense.Due to the different degrees by which theyadhere to the Nordic model of welfare, we cansee if this ranking makes sense. By makingassessments for different points in time, wemay investigate the patterns of welfare reformand their impact on the extent to which coun-tries belong to the Nordic model.

Fuzzy-set theory offers us a tool to deter-mine whether changes are marginal in natureby amounting to a difference in degree, or fun-damental by amounting to a difference inkind. Thus, if a country’s new crisp locationremains closest to the same ideal-typicallocation as before, it constitutes a difference indegree. In contrast, when it gets closer toanother ideal-typical location than before, it



009139 15/7/99 9:21 Page 235



amounts to a difference in kind. In our case,the question concerns the position and changeof position of the Nordic countries in relationto the ideal-typical location expressed in theNordic welfare model and in relation to theposition of each other.

Let us illustrate the basics necessary for ourpurpose by using the example above of theNordic model of welfare which had beenidentified in the literature as being constitutedby universality and generosity. For the sake ofsimplicity, Table 1 gives the fuzzy membershipscores in the set of countries with respectivelyuniversality and generosity for four hypothet-ical countries; Equallygal, Transitionstan,Meritius, and Dollarland. These scores varybetween 0 (fully out) to 1 (fully in). Scores inbetween 0 and 1 indicate partial membershipin the following way: scores between 0.83 and0.99 indicate that the case is almost fully in theset of countries with universality/generosity;0.67 to 0.82 indicate fairly in; 0.51 to 0.66 aremore or less in; 0.5 is the cross-over pointwhere the case is neither more in nor moreout; 0.33 to 0.49 are more or less out; 0.17 to0.32 are fairly out; and 0.01 to 0.16 arealmost fully out. This nine-category fuzzymembership score will be used throughout thearticle to help us translate interval fuzzy mem-bership scores into verbal concepts.

In Table 1, Equallygal is fully in the set of countries with universality, and fairly in the set of countries with generosity. Trans-itionstan is almost fully universal, but it isfairly out of generosity. The case of Meritius isnearly the opposite, i.e. more or less out ofuniversality, but almost fully generous.Dollarland is both the least universal and leastgenerous of the four countries; it is fully out ofgenerosity and almost fully out of universality.

Moreover, Table 1 indicates how closelythese hypothetical countries approximate tothe pre-conceptualized Nordic welfare model(universal*generous) by using the minimumprinciple in fuzzy-set theory. Using the min-imum principle, Equallygal has a score of 0.71in the Nordic model, as its lowest membershipin the two sets is 0.71, thus being fairly close

to the Nordic model. In contrast, Dollarland isfully out of the Nordic model, as its scores 0.Meritius and Transitionstan are, respectively,more or less out and fairly out of the Nordicmodel.

Of course, the conformity of cases to otherideal types can be assessed using the principleof negation and the minimum principle. Forexample, Meritius has a score of 0.64 in theBismarckian model (~universal*generous),Transitionstan a score of 0.72 in what may bedescribed as an expression of the Beveridgemodel (universal*~generous), and Dollarlanda score of 0.86 in the residual model (~uni-versal*~generous).

Scores are sensitive to the context andranges chosen for the fuzzy scores in the sets.Thus, in the remainder of this article we willnot witness the same variation in proximity tothe Nordic model as in the hypothetical coun-tries. The Nordic countries can be expected tobe more in than out in most aspects of theNordic model. Of course, however, the aim ofthis article is precisely to shed more light onwhether this is indeed the case, and whetherwe may identify patterns of welfare reformover time and between countries, areas, andprogrammes.

The weakness of the fuzzy-set approachmay be that it demands a very high degree ofcorrespondence between the ideal types andthe fuzzy membership scores. This, in turn,necessitates close attention to the analyticalconstruction of the ideal type, the empiricalindicators of its constitutive elements, the

236 JON KVIST


Table 1 Using fuzzy-set theory to assessproximity of four hypothetical countries to theideal-typical Nordic welfare model

Universality Generosity Nordic(U) (G) welfare

model(U*G)

Equallygal 1.00 0.71 0.71Transitionstan 0.84 0.28 0.28Meritius 0.36 0.88 0.36Dollarland 0.14 0.00 0.00

009139 15/7/99 9:21 Page 236



empirical evidence used to indicate member-ship, and the criteria used to establishqualitative breakpoints as well as the transla-tion of data to fuzzy interval scores and verbalqualifiers (see, also, Ragin, forthcoming).Doing this, the researcher cannot rely on aver-ages or other relative benchmarks common inconventional social science, but must beinformed by theory and substantive knowl-edge. Averages are sample specific: ideal typesare not. The direct and active involvement ofthe researcher leaves him/her open to criticismfor choosing the wrong aspects, empirical indi-cators, empirical evidence and qualitativebreakpoints, and of not properly translatingdata into verbal qualifiers. To enrich scientificdialogue it is salient that the researcher is asopen as possible about the choices made in thecourse of investigation as these impact on theresults – just like explicit and implicit choicesmade in any other type of research. This isaccentuated by the fact that social research isnot characterized by the same rich and abun-dant evidence as the natural sciences. Incross-national analysis, scarcity of relevantcomparable data sometimes make second-bestdata the only option. This should be madeexplicit by the researcher.

An example can illustrate some of thesepractices and issues. From theoretical and sub-stantive knowledge we have identified aspectsof the Nordic model within each of three wel-fare areas (see below and the followingsection). For the first area, child and familysupport, three constitutive aspects are ident-ified, i.e. generous cash benefits, universality ofchild care and high-quality child care. This hasled to the identification of empirical indi-cators, criteria for qualitative breakpoints andfor pegging data to fuzzy interval scores, allinformed by theory and substantive knowl-edge along the following lines:

• Generosity is measured by the impact ofFamily allowances on family income aftertax. Based on three stylized family types(with differing number and age of children)the average increase per child in net dispos-

able income due to family allowances wasfound. If family allowances in a countryincrease this income by 6 percent or morewe judge this country to be fully in the set ofcountries with generous benefits whereas wejudge increases of 1.4 percent or less astrivial in relation to the cost of raising chil-dren and thus fully out.

• Universality is measured by the proportionof children in child day care and family daycare whether public or private. The some-times widespread use of child care in theinformal sector is not covered by themeasure, but this is also not directly relevantfor the conceptualized ideal type. However,there are still many other schemes for chil-dren which may potentially obstruct anymeasure of child-care coverage. In par-ticular, for children below 3 years of age,the existence of maternity and paternityallowances, parental leave and relatedschemes undermines any coverage measure.Children above the age of 6 are oftenattending school or pre-school schemes.Therefore, to arrive at the least biasedmeasure of child-care coverage, we focus onchildren aged 3–6. The universality aspect isalso not as straightforward as it mayinitially appear. The goal is not – and hasnever been – to place all children in childcare. Instead of taking an interval scale from0 to 100 percent coverage, we have there-fore set a qualitative breakpoint at 80percent for being fully in and 20 percent forbeing fully out. The relatively high cut-offpoint of 80 percent aims to take intoaccount the comparatively high labourmarket participation rates of Nordicmothers and grandmothers, who were tra-ditionally full-time carers of children.

• Quality of child care can be measured invarious ways. American studies have ident-ified the number of children per staffmember as one of the crucial parameterswhich can be said to impact on children’swell-being and later performance (Howes,1997; Peisner-Feinberg and Burchinal,1997). Other quality measures, such as staff



009139 15/7/99 9:21 Page 237



238JO

NK

VIS

T

Journal of European Social P

olicy1999 9

(3)

Table 2 Specification of empirical indicators and the translation of data to fuzzy score ranges and verbal qualifiers

Area Empirical indicator Fully in Almost Fairly in More or Neither more More or less Fairly out of Mostly out Fullythe set fully in the set less in the in nor more out of the the set of the set out of

the set set out of the set set the set1.00 .83–.99 .67–.82 .51–.66 .50 .33–.49 .17–.32 .01–.16 0.00

Families Generosity measured by average �6.00 5.20–5.99 4.40–5.19 3.60–4.39 3.50–3.59 2.80–3.49 2.10–2.79 1.40–2.09 <1.40increase in net disposable incomecaused by family allowances (%)Universality measured by share of >80 71–80 61–70 51–60 50 40–49 30–39 20–29 <20children aged 3–6 in child day care orfamily day care (%)Quality measured by child–staff ratio <3.00 3.00–3.99 4.00–4.99 5.00–5.99 6.00 6.01–7.00 7.01–8.00 8.01–8.99 �9.00in child day care

Unem- Index of accessibility to unemployment >86 75–86 68–74 51–62 50 40–49 30–39 20–29 <20ployed insurance benefita (0–100)

Generosity measured by average net >85.0 75.0–85.0 63.0–74.9 51.0–62.9 50.0–50.9 40.0–49.9 30.0–39.9 20.0–29.9 <20.0replacement rate of unemploymentinsurance benefit (%)Index on the quality of <12.0 12.1–18.0 18.1–26.0 26.1–36.0 36.1–37.0 37.1–49.0 49.1–61.0 61.1–73.0 >73.0employment policiesb

Old Universality of old-age pensionsc Universal – – – Selective – – – ResidualGenerosity measured by average net �65.00 60.00–64.99 54.00–59.99 46.00–53.99 45.00–45.99 37.00–44.99 28.00–36.99 18.00–27.99 <18.00replacement rate of national old-agepensions (%)Index on extensiveness of social care �45.0 40.0–44.9 34.0–39.9 27.0–33.9 26.0–26.9 19.0–25.9 13.0–18.9 8.0–12.9 <8.0for elderlyd

Notesa Composite index based on the coverage and allocation criteria as well as re-entitlement requirements for unemployment insurance.b Composite index based on the relative and absolute unemployment rates of young people.c Universality assessed according to the coverage and allocation criteria for national basic and national supplementary old-age pensions, resulting in atrichotomy (universal, selective and residual).d Composite index based on the share of elderly receiving home help services or living in institutions for the elderly and service-flats.

009139 1

5/7/99 9

:21 P

age 238

©

1999 SA

GE

Pu

blicatio

ns. A

ll righ

ts reserved. N

ot fo

r com

mercial u

se or u

nau

tho

rized d

istribu

tion

. at S

ocialforsknings Institut on July 14, 2007 http://esp.sagepub.com

Dow

nloaded from


education, were not available to theresearcher for all the countries in the wholeperiod in question. The specific child–staffratio for good and bad-quality child caredepends on the age and other characteristicsof the children. On average for childrenaged 3–6, however, less than 3 children perstaff member is definitely a sign of high-quality child care and more than 9 childrendistinctly low-quality (Brazelton, 1992).Less than 6 children per staff member is asign of good quality.

Table 2 sets out these empirical indicators andtheir pegging to the fuzzy interval scores forchild family support as well as for the twoother areas of unemployment measures andwelfare for the elderly (for documentation, seeKvist, 1999).

Most of the material stems from two majorongoing cross-national studies of social pro-tection, ‘Social Security in the Nordic

Countries’ (NOSOSCO/NOSOSKO 1992;1995; 1996; 1997; 1998), and ‘Elements ofSocial Security’ (Hansen, 1997; 1998).

Family support

Universal family allowances are not unique tothe Nordic countries, so we will not take uni-versality of cash benefits as a defining featureof the Nordic model, but only their generosity.The most distinct feature of the Nordic modelof child and family support, however, is to befound on the service side of social protectionfor families – the way child care is provided or,rather, is ideal-typically described. The ideal ofchild care in the Nordic countries could bedescribed as one of public support for – or per-haps even a guarantee of – universal child careof a high quality based on solidarity. The ideal



Table 3 Using fuzzy-set theory to assess the conformity of Nordic countries to the Nordic model ofchild and family support, 1990–7

Country Year Generosity of Universality of Quality of Nordic model ofcash benefits child care child care child and family(G) (U) (Q) support

(G*U*Q)

Denmark 1990/91 .56 .94 .53 .531993/94 .64 .94 .70 .641996/97 .65 1.00 .70 .65

Finland 1990/91 – .63 .66 .631993/94 1.00 .51 .62 .511996/97 .82 .70 .62 .62

Norway 1990/91 – .61 .74 .611993/94 – .67 .74 .671996/97 – .67 .74 .67

Sweden 1990/91 .88 .72 .70 .701993/94 .79 .70 .62 .621996/97 .50 1.00 .62 .50

Notes: Unfortunately, no data are available on generosity for Norway, generosity in Finland 1990/91 andthe quality of child care for 1996/97. Nevertheless, it is most likely that the absence of data on generosity inNorway has no impact on the proximity to the Nordic model of child family support as the minimum isprobably given by the scores in universality of child care. The 1996/97 data on the quality of child care hasbeen assumed to be at the same level as in 1993/94. This is a heroic assumption given the massive increasein child-care places which is likely to have involved some trade-off with quality as measured here bychild–staff ratios.

009139 15/7/99 9:21 Page 239



of universality in services encompasses theideal that every citizen, irrespective of need ormerit, is able to access child care. Solidarity, inthis case, comes to mean arrangements makingit feasible for low-income groups to access theservices. And by high quality is meant that, forexample, child-care institutions are not justcar parks for children during parents’ workhours, but also have strong elements of social-ization.

Expressed in fuzzy-set terms the Nordicmodel of family support is the ideal-typicallocation: generosity*universality*quality.Table 2 above shows the empirical indicatorsand their pegging to fuzzy membership scores.Table 3 above reports the Nordic countries’membership in the various aspects – or sets –of the Nordic model of family support and inthe model as such. For the sake of simplicitywe have chosen three points in time; 1990/91,1993/94 and 1996/97.

Generosity of cash benefits

In 1990, Denmark was the least generous ofthe Nordic countries. In 1991, familyallowances were increased and infant benefitsextended from 0–3 years old to 0–6 years old.Other improvements were made during the1990s, and as a result Denmark was close tofairly generous by 1997, and she surpassedSweden in 1996. This, however, had more todo with cuts in Swedish benefits than improve-ments in Danish ones.

In 1994 Finland was fully generous and byfar the most generous of the Nordic countriesfor families. Family allowances were notaffected by the general cuts in social protectionof the early 1990s. In 1994 benefits were actu-ally increased quite substantially by some 40percent as the cash support for familieschanged from a tax allowance to a directallowance with an extra supplement for singleparents. The family allowance remained nom-inally unchanged from 1994 to 1995 leading

to a slight decrease in generosity. In 1996 ben-efits were reduced significantly and notindexed the following year. Although fallingfrom fully generous to fairly generous, Finlandremained the most generous Nordic countryfor families.

Sweden was almost fully generous to fam-ilies in the early 1990s. Although the mainelements of the so-called ‘crisis packages’ werebudget cuts in social programmes, familyallowances were actually increased twice aspart of the tax/benefit reform of 1991. Until1995, however, the basic allowance for chil-dren remained nominally the same atSEK9000 annually per child. Supplements formultiple children were reduced in 1994,severely cut in 1995 and closed for new recip-ients in 1996. In 1997, as a result of thesecut-backs, Sweden was neither more in normore out of the set of generous countries forfamilies.

Universality of child care

In 1990, care of children aged 3–6 inDenmark was almost fully universal; and in1996 it had become fully universal. However,waiting lists have persisted through the 1990s,especially for infants and toddlers, despite top-level political priority and rhetoric. This ispartly due to the fact that child care inDenmark, as in the other Nordic countries, isthe province of local government, the munici-palities. Due to central governmental threatsand promises of extra subsidies, and, perhapsmore importantly, to attract potential tax-paying families or avoid their migration,municipalities have made tremendous progressand enrolled some 68,000 more children aged0–6 from 1990 to 1996. Nevertheless,increased fertility and lower unemploymenthave resulted in extra demand for child carewhich is only to some extent countered byimproved leave-of-absence schemes intro-duced in 1994 (though reduced gradually

240 JON KVIST


009139 15/7/99 9:21 Page 240



since). The waiting list for child care has thusonly decreased from 16,000 in 1996 to 12,000in 1998 despite considerable expansion ofchild care and the introduction of child-careguarantees in the majority of municipalities.

In Finland, there has been a child-care guar-antee for some time as municipalities have hadan obligation to offer child care for childrenbelow 3 years of age. Finland was, however,only more or less universal in 1990. As unem-ployment soared in the following years, theproportion of children in child care and thusuniversality plummeted. This was partly dueto unemployed people taking care of their ownchildren to cut expenses, and partly due tomunicipalities barring unemployed parentsfrom using child care in order to save moneyin times of austerity. Excluding the unem-ployed from making use of social services runsstraight in the face of the idea of solidarity,ideal-typically associated with the Nordic wel-fare model. This phenomenon has beenobserved not only in Finland, but in manymunicipalities in the Nordic countries.

Paradoxically, it is perhaps the work ofeconomists on work incentives and unemploy-ment traps which has helped increase thedegree of universality and solidarity in Finnishchild care. At least, they have advocatedincreasing the scope and flexibility of childcare, and action followed the words. In 1996,the child-care guarantee was extended toencompass all children under 7 years of age,resulting in a considerable increase in thenumber of children in child day care, some27,000 more from 1995 to 1996. Familiesmay also be given a home child-careallowance, which 55 percent of childrenbetween 9 months and 3 years made use of in1996, despite a 23 percent reduction in thehome child-care allowance the same year. In1997, reform aimed to increase the use andflexibility of child care further. Partly as aresult of these policy changes, universalityincreased steeply so that Finland was fairlyuniversal by 1997.

In Sweden, from 1990 to 1997 no less than45 percent more, or 145,000 extra children,

aged 0–6 became enrolled in child care. Thedramatic increase is a result of a substantialexpansion of child-care institutions and placestogether with fewer restrictions on privatechild minding. Sweden’s degree of universalityof child care has also increased significantlyfrom being fairly universal in 1990 to beingfully universal in 1997. Like Finland, Swedenhas given parents rights to child day care ifthey work or study, although working andunemployed parents may be treated differentlyas a result of local autonomy.

In Norway, the number of children aged0–6 in child care has increased considerablyby some 38 percent from 1990 to 1996, par-ticularly among the infants and toddlers.Nevertheless, the share of children enrolled inchild day care remains comparatively low.Today, Norway can be described as beingfairly universal. This can be explained by anumber of factors. High fertility rates have ledto extra demand whereas the extension ofparental leave schemes and the introduction ofa so-called ‘time account’ (allowing forreduced working time supplemented byreduced maternity leave for 6 to 29 weeks ontop of the 42 weeks with full wage) havedecreased the demand. Another factor may bethat the process of switching from childrenbeing taken care of by their mothers or grand-mothers to institutionalized care or family daycare has not gone as far as it did in Denmarkand Sweden in the 1960s and 1970s.Moreover, in 1998 a home child-careallowance was introduced for parents notmaking use of child day care. In sum, Norway,perhaps the most traditional of the Nordiccountries, has undertaken quite substantialmeasures giving parents better opportunitiesfor caring for their own children and forplacing them in child or family day care.

Quality of child care

None of the Nordic countries can be said tobelong fully to the set of countries with child



009139 15/7/99 9:21 Page 241



care of a high quality when measured by thechild–staff ratio. Denmark moved from beingmore or less in the set of countries with high-quality child care in 1990 to being fairly in by1993, as the ratio of children per staff fellfrom 5.8 to 4.7. In the same years, Finland’sposition worsened, as the child–staff ratiowent from 5 to 5.2 children. Although onlyfairly in the set of countries with high-qualitychild care, Norway had the highest degree ofquality in child care of the Nordic countries at4.5 children to staff in both 1990 and 1993.Sweden declined from being fairly in to beingmore or less in the set of countries with high-quality child care. Unfortunately, there is nodata for 1996/97, but it seems unlikely that thequality of child care as measured by child–staffratios would have improved in times of suchmassive expansion of enrolment. Rather, itseems that in striving for more child-careplaces, the objective of universality in somecases has been paid for by compromising onan already less than satisfactory degree ofquality.

Conformity to the Nordic model offamily support

In the 1990s, considerable changes have takenplace within child family support in theNordic countries with respect to the generosityof family allowances and the universality andselectivity of child care. The changes describedabove, however, have not resulted in anycountry being excluded from belonging to theNordic model of family support, althoughSweden is now a borderline case after havingbeen the ‘leader’ in 1990. Sweden andDenmark have therefore changed places.Norway has shown more stability whereasFinland managed to catch up after first beingseverely affected by the economic crisis. Noneof the Nordic countries, however, is close tothe ideal Nordic model.

Many of the Nordic countries have faced an

uphill struggle in the area of child care.Substantial expansion of day care places hastaken place during the 1990s in all four coun-tries, but – due to a combination of increasingfertility rates, unemployment and new orextended parental leave schemes and homeand private care allowances – this has notresulted in corresponding increases in univer-sality. Nevertheless, universality is the onlyaspect where all countries improved, andDenmark and Sweden are now fully universalin the area of child care. The price forexpanding child care, however, may be poorerquality of child care. Denmark is the onlycountry that has slightly improved thechild–staff ratio, but at the same time it is alsothe country with the smallest share of trainedstaff – another quality indicator. Clearly, allfour countries have some improvements tomake before they can be considered fully in theset of countries with high-quality child care.

Overall, the patterns of welfare reform arenot identical in the four countries. Despitesubstantial cuts in family allowances Finlandis still the most generous of the Nordic coun-tries. Generally speaking, Denmark and – lessso – Norway have improved their child familysupport systems. The most severe and visiblecuts were introduced by Sweden, and, less so,by Finland, in particular in family allowances.However, all four countries have also intro-duced new or improved care allowances andexpanded the universality of child-care places.In other words, both retrenchment and expan-sion can be identified in the four countries.Finally, we can note that the patterns of wel-fare reform have not resulted in any clear signsof either convergence or divergence betweenthe national family support systems in gen-erosity, universality or quality.

Unemployment measures

Ideal-typically, the Nordic model of unem-ployment measures has been associated with

242 JON KVIST


009139 15/7/99 9:21 Page 242



accessible and generous cash benefits in com-bination with an emphasis on employmentpolicies such as active labour market pro-grammes (ALMP) of a high quality. Solidaritywith the jobless underpins this model. Theunemployed should not be further victimizedthan they already are, which leads to access-ible and generous benefits. The helping handof the state should aid them into the labourmarket via ALMP and policies with similarobjectives. Similarly, efficiency arguments canbe said to underpin all four aspects, as, to putit briefly, the generosity of the Nordic modelcannot be sustained without high levels ofeconomic activity. There is no consensus onthe desirability or effectiveness of this modelof unemployment measures nor of the par-ticular programmes.

For our purposes, however, it is sufficient toidentify three constitutive features of themodel: accessible and generous cash benefitsas well as high-quality employment policies.Expressed in fuzzy-set terms the Nordic modelof unemployment measures is the ideal-typical

location: accessible*generous*quality. Table 4reports the Nordic countries’ membership inthese three sets and in the model.

Accessibility of cash benefits

All the Nordic countries had fully accessibleunemployment insurance schemes in 1990,except Finland, which, however, was almostfully accessible. By 1997, Denmark was onlyfairly accessible because of the new ‘activationline’ in Danish unemployment policy initiatedin the late 1980s and given new impetus by thelabour market reforms of 1994 and sub-sequently. Since 1994, participation intraining and job offers no longer qualifypeople for re-entitlement to unemploymentinsurance. Previously there was a virtual recy-cling mechanism as the unemployed at risk oflosing their benefits were made an offer that inturn requalified them for a new benefit period.



Table 4 Using fuzzy-set theory to assess the conformity of Nordic countries to the Nordic model ofunemployment measures, 1990–97

Country Year Accessibility of cash Generosity of cash Quality of Nordic model ofbenefits benefits employment unemployment(A) (G) policies measures

(Q) (A*G*Q)

Denmark 1990/91 1.00 .88 .87 .871993/94 1.00 .87 .97 .871996/97 .74 .80 1.00 .74

Finland 1990/91 .92 – .87 .871993/94 .65 .79 .43 .431996/97 .72 .76 .54 .54

Norway 1990/91 1.00 – .70 .701993/94 1.00 – .67 .671996/97 .96 – .64 .64

Sweden 1990/91 1.00 1.00 .81 .811993/94 1.00 .97 .59 .591996/97 1.00 .85 .69 .69

Notes: Unfortunately, no data are available on generosity for Finland 1990/91 and for Norway in general.However, it is unlikely that this will affect the conformity of Norway to the Nordic model, as its minimumis probably not given by the generosity of cash benefits, but rather the quality of employment policies.

009139 15/7/99 9:21 Page 243



Thus, the stipulation of a seven-yearmaximum period in 1994 was actually areduction of the renewable 2.5-year periodpreviously in force. From having perhaps themost relaxed work requirement amongwestern welfare states, things changed in 1997when 52 weeks of work instead of 26 weekswithin a three-year period became a require-ment.

Unemployment insurance in Finland wasalmost fully accessible in 1990, despite theFinnish self-employed being excluded fromcoverage. In 1992, it became more or lessaccessible as participation in active labourmarket programmes was no longer a qualifica-tion for a renewed benefit period. Theself-employed became covered in 1995. And in1997, the minimum contribution period forunemployment insurance was strengthenedfrom six to ten months, and the work require-ment from 26 weeks to 43 weeks of workwithin a two-year period. As a result,Finland’s unemployment insurance is nowfairly accessible.

Nothing much changed to alter the fullaccessibility of the Norwegian unemploymentinsurance system, although in 1997 there wasan increase in the amount of previous earningsto be made before becoming eligible. Swedenretained a fully accessible unemploymentinsurance system. This may come as more of asurprise than the Norwegian case against thebackdrop of the very different economic devel-opment in the two countries. In 1994, thework requirement was raised from four to fivemonths of work within a year. The work con-cept applied in the Swedish system, however,is still very broad and includes participation inlabour market programmes and leaveschemes.


Denmark was almost fully generous in both1990 and 1993, and fairly generous in 1997.

The slight differences in net replacement ratesover time are mainly a function of real-wagedevelopments and changes in the tax system,although the benefit level for young recipientswas generally reduced to half in 1996. Finlandand Sweden have been cutting benefit levelsmore consistently. In Finland this was invisiblethrough the lack of indexation and byreducing the earnings base for calculating thebenefit. Nevertheless, Finland was fairly gen-erous in both 1994 and 1997. Swedish cutswere much more visible with the gradualreduction of the earnings-replacement ratefrom 90 to 75 percent (and now 80 percent).Fully generous in 1990, Sweden fell to almostfully generous in 1997.

Quality of employment policies

The quality of these policies has been mea-sured here by success in keeping youthunemployment low, both in absolute termsand in relation to general unemployment.Measured in this way, the quality of Danishpolicies has increased from almost fully high in1990 to fully high in 1997. This is primarilybecause of a relatively low ratio of youngunemployed to general unemployed (around1.4:1, that is 1.4 young unemployed aged16–24 for every general unemployed) and tothe reduction of youth unemployment from 14percent in 1993 to around 10 percent since1994. This can in part be ascribed to the 1994and subsequent labour market reform adjust-ment targeting the young for ‘specialtreatment’.

The quality of Finnish policies was dealt asevere blow by the crisis of the early 1990s.From being of an almost fully high quality in1990, they deteriorated to more or less out ofhigh quality in 1993. Although they haveimproved since, they are only more or less inby 1997. This is, of course, due to theimmense pressure created by soaring unem-ployment. Despite a new emphasis on active

244 JON KVIST


009139 15/7/99 9:21 Page 244



labour market programmes this has been morethan an ordinary uphill struggle. By 1996,however, the scope of programmes towardsyouth had grown with 11.8 percent of youthaged 16–24 in ALMP compared to 10.9 per-cent and 4.2 percent in, respectively, Denmarkand Norway. Similarly, the ratio of young togeneral unemployed fell from 2:1 in 1990 to1.7:1 in 1996.

Norwegian employment policies were offairly high quality in 1990 but deteriorated tomore or less high quality in 1997. This is par-ticularly due to the increasing ratio of young togeneral unemployed. The position of Norwayshould be seen in the light of its new emphasison what is called the ‘work line’. According tothis, priority should be given to combatingunemployment and non-employment byincreasing the efforts of ALMP and by makingthe so-called ‘passive’ doling out of cash ben-efits conditional on work or some other type ofactivity. The similarity to the Danish ‘activeline’ is striking. So far, however, theNorwegian work line has not enjoyed the samedegree of success as the Danish, possiblybecause the remedies applied have been more inwords than actions. The 1994 reform is amongthe main initiatives, and this only entitlespeople under the age of 20 to training and edu-cation, supplementing the Youth Guaranteeoffering special labour market measures for theremainder of the young.

In general, Sweden has a longer tradition ofactive labour market policies than her Nordicsisters. Nevertheless, the quality of Swedishemployment policies also suffered a blow fromthe crisis in the early 1990s – moving from afairly high quality to only more or less high asyouth unemployment jumped from a recordlow of 3.7 percent in 1990 to 18.4 percent in1993. While youth unemployment decreasedslightly to 15.7 percent in 1996, the ratio ofyoung to general unemployed decreased from2.5:1 in 1990 to 1.9:1 in 1996. As a result thequality of employment policies improved tofairly high by 1997.

Conformity to the Nordic model ofunemployment measures

In the 1990s considerable changes have takenplace with respect to the accessibility and gen-erosity of cash benefits and the quality ofemployment policies. Most notably, Finlandlost her place in the club of Nordic modelcountries as her unemployment measures werecaught off guard by the severe economic crisisin the early 1990s. Today, however, she hasmore or less regained membership. Swedenalso experienced some turbulence from itseconomic crisis, but is now fairly in the Nordicmodel. Norway and Denmark demonstrated amore stable membership. Nevertheless, all theNordic countries experienced a reduction intheir conformity to the Nordic model over theperiod.

Sweden and Norway had fully and almostfully accessible cash benefits throughout theperiod, but the labour market reforms of 1994and since, in particular a strengthening of thework concept and of work requirements, ledDenmark to be fairly accessible by 1997.Introducing a stricter interpretation of thework concept, and moving more in line withthe other Nordic countries by extending cov-erage to the self-employed, Finland was asfairly accessible as Denmark by 1997.

The generosity of cash benefits decreased ingeneral (no data for Norway). Nevertheless,cash benefits are still almost fully generous inSweden, and fairly generous in Denmark andFinland. The crises in Finland and Swedenfaced their employment policies, in particularfor young people, with an insurmountable chal-lenge. They were not able to retain their qualityas measured by the ability to combat high rela-tive and absolute unemployment for particulargroups. Denmark is the only Nordic countryshowing an improvement, and its experiencecould point towards some type of compromiseor relationship between the accessibility of cashbenefits and the quality of employment policies,perhaps necessitating a rethink of the Nordicmodel of unemployment measures.



009139 15/7/99 9:21 Page 245



Welfare of people elderly

Old-age pensions can be considered thebackbone of the Nordic welfare model as theyare typically the first schemes to be imple-mented, take up the lion’s share of publicsocial expenditures and potentially affect thewhole population. The universality of old-agepensions coupled with relatively generous ben-efits is supposed to be characteristic of theNordic model. Similarly, social care for theelderly should be accessible to everybody inneed irrespective of previous work and contri-bution record. Social care renders autonomyto elderly people and their relatives. In sum,the Nordic model of welfare for elderly peopleis the ideal of generous, universal old-agepensions coupled with extensive social careservices. This can be expressed in fuzzy-setterms as the ideal-typical location: uni-versal*generous*extensive. Table 5 reports theNordic countries’ membership in theseaspects.

Universality of cash benefits

All the Nordic countries have retained uni-versal cash benefits for elderly people. Allcitizens are covered by a guaranteed minimumpension today as they were at the beginning ofthe 1990s. This resilience, however, masks aninternal restructuring in national pension sys-tems between the roles of the basic andsupplementary types of pensions.

In Denmark, the tax/benefit reform of 1994,among other things, introduced an income-testagainst earnings on the basic amount of thenational pension. More importantly, thereform increased the relative size of thenational pension supplements subject toincome-test against other pension income,thereby building in a claw-back mechanism topartially defuse the so-called ‘coming demo-graphic time bomb’. Future pensioners withmajor income from supplementary pensionbased on collective agreements will have theirnational pension reduced. Recent expansion ofthese collective agreement pensions is alsolikely to affect the structure of the futureDanish pension system.

246 JON KVIST


Table 5 Using fuzzy-set theory to assess the conformity of Nordic countries to the Nordic model ofwelfare for the elderly, 1990–7

Country Year Universality of Generosity of cash Extensiveness of Nordic model ofcash benefits benefits social care welfare for the(U) (G) (E) elderly

(U*G*E)

Denmark 1990/91 1.00 .73 .96 .731993/94 1.00 .73 .89 .731996/97 1.00 .63 .91 .63

Finland 1990/91 1.00 .82 .80 .801993/94 1.00 .82 .62 .621996/97 1.00 .72 .56 .56

Norway 1990/91 1.00 – .81 .811993/94 1.00 – .81 .811996/97 1.00 – .85 .85

Sweden 1990/91 1.00 .83 .64 .641993/94 1.00 .80 .58 .581996/97 1.00 .69 .66 .66

Notes: Unfortunately, no data are available on generosity for Norway. Generosity for Finland 1990/91 hasbeen set at 1993/94 level.

009139 15/7/99 9:21 Page 246



The Finnish pension system consists of anational minimum old-age pension and a so-called ‘employment’ (supplementary) pensionwithout any benefit ceiling. By introducing anincome-test of the whole national pension, thePension Reform of 1996 reinforced the role ofthe national old-age pension by providing onlya minimum pension for people with belowaverage income from the employment pensionwhich, in turn, has the role of securing accus-tomed standards of living.

The guaranteed minimum pension inNorway consists of a national basic pensionwith a special addition for people with no orinsignificant income from the supplementarypension. There are also additions for spouseand children. Eligibility to the earnings-relatedsupplementary pension requires at least threeyears of insurance coverage.

The Swedish pension system consists of abasic and a supplementary pension. The basicpension consists of a basic amount and a sup-plement, both expressed in relation to theso-called ‘basic amount’ in Swedish socialinsurance. The supplementary pension is earn-ings-related. However, the Pension Reform of1994 will gradually change the Swedish pen-sion system to become a defined contributionsystem rather than a defined benefit system(see Palmer, 1998).


The generosity of old-age patterns also followsstrikingly similar patterns across the Nordiccountries as they become somewhat less gen-erous towards the end of the period studiedthan at the start, at least when measured ashere. Danish pensions went from fairly gen-erous in 1990 to more or less generous in1997. The main reason is not so much changesin legislation as increases in real wages whichare not fully paralleled by benefit increases dueto the Danish indexation mechanism. Indeed,to make the national basic old-age pensions de

facto taxable, current pensioners were morethan fully compensated in the Tax/benefitReform of 1994.

Finnish pensions also became slightly lessgenerous today than in 1994, but at both timesthe benefits can be described as fairly gen-erous. This development can largely beexplained by the Pension Reform of 1996,which reduced the generosity of pensions infour ways. First, it changed the indexation ofemployment pensions in payment from beingbased 50:50 on wages and prices to be 20:80,thereby gradually diminishing generosity rela-tive to real-wage increases. Second, the settingof employment pension benefit levels will takeinto account the last 10 years of incomeinstead of 4 years, with the possibility of dis-regarding a maximum of one-third of theseyears with less than half average earnings.Third, additions to the national pension aregradually being phased out. Fourth, asdescribed above, the national pension hasbecome fully means-tested against other pen-sion income. The true impact of these changes,however, will only become apparent over timeso that the Finnish pension system at the startof the next century, coinciding with the baby-boomers retiring, will not be as generous as inthe 1990s.

In Norway, the additions for spouse andchildren became means-tested in 1991. Thefollowing year, an income-test was introducedaccording to which pension and earningscould not exceed previous income. At the sametime, the benefit formulae of the supplemen-tary pension were altered so as to make themaximum possible pension less generous.Despite lack of indicators, these changescannot be considered retrenchment on anymajor scale. They are likely to be more thancounterbalanced by the exceptional 18 percentor NOK1000 increase of the national pensionin 1997.

Sweden had almost fully generous pensionsin 1990. Since 1993, however, the basic pen-sion is no longer based upon 100 percent, but98 percent of the basic amount. As a partialcompensation, the means-tested supplemen-



009139 15/7/99 9:21 Page 247



tary part of the basic pension was increased.Net replacement rates did not decreaseimmediately, however, as there was also anominal decrease in the earnings of theaverage production worker from 1992 to1993. Indexation of pensions is inverselyrelated to the public budget deficit. For thisreason, in 1995 and 1996, the basic amountwas not indexed fully, but only to 60 percentof price increases. However, this 60 percentindexation continued in 1996 and 1997despite improved public budget deficits. As aresult of all these changes, Sweden no longerhas almost fully generous pensions, but ratherfairly generous. In 1999, the basic pension willbe based on 99 percent of the basic amountand in 2000 on 100 percent, the same asbefore the crisis hit Sweden.

Extensiveness of social care

During the 1990s, Denmark has had almostfully extensive social care for elderly people asmeasured by the proportion of elderly peopleliving in service flats and institutions for theelderly or in receipt of home help services.From 1990 to 1997 the proportion of elderlypeople aged 80+ living in service flats andinstitutions for the elderly declined slightlyfrom 24.6 percent to 22.6 percent, whereas theproportion of elderly aged 67+ in receipt ofhome help remained stable at a little over 19percent. Under the 1980s slogan ‘as long aspossible in your own home’, there have beensome significant changes not revealed by thesefigures. In particular, recipients of the varioustypes of social care have become older andfrailer, and there has been a shift towardshome help and service flats rather than nursinghomes.

Finland experienced the most dramaticdecline in the extensiveness of social care forelderly people of all the Nordic countries.From having close to an almost fully extensivesystem of social care in 1990, it fell to being

only more or less so by 1997. This is mainlydue to a 50 percent drop in the extent of homehelp services, from 21.4 percent to 11.5 per-cent received by people above 65 years of age.

In Norway there was also a drop in homehelp services, but less marked, from 19 percentin 1990 to 15.6 percent in 1996. However, theproportion of people aged 80+ living in insti-tutions increased by more as it went from 20.7percent to 25.1 percent. As a result, Norwayimproved its social care from being fairly toalmost fully extensive. As in Denmark, there isa move towards service flats and similar typesof accommodation for elderly people. In 1997,the Norwegian parliament decided on con-siderable investment in social care for theelderly. Starting in 1998, the state will grantmunicipalities money to build more serviceflats and earmarked subsidies for social carework.

Sweden started and ended the period withmore or less extensive social care for elderlypeople. This masks an increasing share ofelderly people living in service flats and insti-tutions, and a decline in home help. As part ofa general decentralization of social care from1990 to 1993, municipalities now have theoverall responsibility for the long-term serviceand care of elderly people. One consequencehas been a decrease in recipients of home helpaged 65+ from 17.4 percent in 1990 to 11.3percent in 1993, a level it has stayed at since.

The development in Denmark of fewerpeople in nursing homes, more people in ser-vice flats and more home help is reflecting ade-institutionalization process. This processcannot be observed to the same degree in theother Nordic countries, which have all experi-enced increasing shares of the elderly 80+ inservice flats or institutions, partly because theywere starting from a lower level thanDenmark. But these countries have also seenreductions of home help services to levelswhich in Sweden and Finland approach halfthe level of Denmark. And this may, in turn,be partly explained by local governmenthaving difficulties in finding the necessaryfunding for social services in times of austerity.

248 JON KVIST


009139 15/7/99 9:21 Page 248



Conformity to the Nordic model ofwelfare of the elderly

During the 1990s a series of changes havetaken place with regard to the universality andgenerosity of pensions and to the extensivenessof social care. However, there is no retrench-ment leading to the exclusion of any countryfrom belonging to the Nordic model of welfarefor the elderly as conceptualized and measuredhere. Hence, the changes in this welfare areahave amounted to differences in degree andnot in kind. The elderly in Finland experiencedthe biggest decline as their system went frombeing fairly in to more or less in the set ofcountries with a Nordic model of welfare forelderly people. Denmark also experienced adecline, but on a much smaller scale. Norwayand Sweden, in contrast, moved slightly closerto the Nordic model. Today, Norway has byfar the most Nordic model of welfare forelderly people of the Nordic countries.

Universality has remained a distinctive fea-ture of the Nordic pension systems. EveryNordic citizen or resident is guaranteed a min-imum old-age pension. For people with workand/or considerable income from pensions andother sources, however, there has been a shifttowards more reliance on the national sup-plementary pension and less on the basicpension. This has happened primarily throughchanges to the basic pension such as the intro-duction or expansion of means-testing and/orgreater relative weight given to pension sup-plements. Whether this claw-back mechanismwill affect the political economy of univer-sality is an open question.

This shift towards more reliance on sup-plementary pensions has, at the same time,resulted in somewhat reduced degrees of gen-erosity of pensions (though there are no dataavailable for Norway, which may run counterto this trend). Today, pensions in Sweden andFinland are fairly generous, whereas Danishpensions are more or less generous. If theDanish supplementary pensions based on col-lective agreements had been included in the

analysis, there might have been even closerintra-Nordic convergence than expressed here(see Øverbye, 1996). Convergence, however,is not what we see when we shift focus to theservice side of the Nordic model of welfare forthe elderly. Looking at the extensiveness ofsocial care, we consistently find big differencesbetween the Nordic countries, both in leveland development.

The instruments of change are numerous.Less than full indexation of pensions has beena common way of reducing benefit generosity;another is the introduction or expansion ofincome-tests and/or increasing the relative sizeof income-tested pension supplements. Suchtechnical devices are invisible to the layperson,have little immediate impact and so pain, andare time-consuming to explain (Pierson,1996). Hence, they do not attract much atten-tion, despite saving the public budgetsubstantial amounts of money in the mediumand long term. The universality of pensionshas not been touched as – in contrast to thetechnical devices – this would have been highlyvisible, painful for limited groups and easy totrack back to the responsible policymakers.Within social care, one way of saving moneyhas been decentralizing – or keeping – auth-ority of social care at the local level, inparticular in the crisis ridden economies ofFinland and Sweden. In more prosperousNorway and Denmark, more money has beenflowing to the local level to improve the exten-siveness and quality of social care for elderlypeople.

If Finland wishes to improve upon its mem-bership of the Nordic model it should look tothe extensiveness of social care where there isplenty of room for improvement. The sameadvice is valid for Sweden, although the scopefor improvement is not as great. In Norway,some things can still be done concerning theextensiveness of social care, although it islikely that the quality of social care, as inDenmark, is – or will soon become – the‘hottest potato’ in social care for elderlypeople. For example, Denmark has recentlyintroduced Elderly Advisory Boards at



009139 15/7/99 9:21 Page 249



regional level to monitor and decide on certaintypes of activities for elderly people with theexplicit purpose of not only increasing levelsof democratic participation, but also degreesof quality.

Discussion

The application of the fuzzy-set theoryanalysis above provides an empirical basis fordiscussing the overall development of nationalwelfare policies at the same time as the pat-terns of welfare reform between the threechosen areas. Table 6 indicates the overallconformity of the countries to the Nordic wel-fare model over time, showing the harmonicmean of fuzzy membership scores in the threeareas (Column 6). It also sets out the fuzzymembership scores of the countries in theNordic models for families with children, theunemployed and older people (Columns 3–5).

Nordic countries fared differently duringthe 1990s in terms of their conformity to theNordic welfare model. Finland experiencedthe most dramatic developments as she movedfrom being fairly in to more or less in the

Nordic welfare model; Sweden’s deteriorationwas less significant; and Norway andDenmark remained stable. This pattern, how-ever, masks variations in development acrosswelfare areas. For example, the situation offamilies with children in Denmark hasimproved, but not that of unemployed andelderly people. In Sweden, welfare for theelderly has been stable, but families and theunemployed became worse off. In otherwords, national economic success or failuredoes not automatically translate into corre-sponding expansions or cut-backs in welfarepolicy. We cannot even identify similar pat-terns for countries experiencing, respectively,good or bad economic times.

Perhaps the most striking feature of welfarereform in the Nordic countries in the 1990s isthe extent of resilience and restructuring.None of the countries has left the club ofcountries belonging to the Nordic welfaremodel despite the adverse national economicperformance in Finland and Sweden and thesignificant changes in the surrounding world.

The results of the empirical analysis suggestnot only that the Nordic countries demon-strate a high degree of resilience in theirmembership of the Nordic model but also thatwelfare areas and programmes with a high

250 JON KVIST


Table 6 Nordic countries’ conformity to the Nordic model of welfare, 1990–7

Country Year Child family Unemployment Welfare for Nordic model ofsupport measures people elderly welfare(C) (U) (O) (harmonic mean of

C, U and O)

Denmark 1990/91 .53 .87 .73 .701993/94 .64 .87 .73 .741996/97 .65 .74 .63 .67

Finland 1990/91 .63 .87 .80 .761993/94 .51 .43 .62 .511996/97 .62 .54 .56 .57

Norway 1990/91 .61 .70 .81 .701993/94 .67 .67 .81 .711996/97 .67 .64 .85 .71

Sweden 1990/91 .70 .81 .64 .711993/94 .62 .59 .58 .601996/97 .50 .69 .66 .61

009139 15/7/99 9:21 Page 250



degree of membership are also most liable tocut-backs, and areas and programmes with alow degree of membership tend to be mostliable to expand. This shows how impossible itis to reach the ambitious goals set out by theNordic welfare model, and that very generousareas are likely to be trimmed, or even severelycut, in economic hard times. But importantly italso shows that there are limits to cut-backsbeyond which politicians and populations areunwilling to go. Policies move within upperand lower limits. This raises the issue ofwhether there are not only limits to growth,but also to cut-backs.

Curiously, the pendulum mechanism canalso be identified at the level of nationsbelonging to the same ideal type when wecompare welfare reform over a longer timespan. Whereas Denmark and Norway faredwell in the 1990s, this was not the case adecade earlier; and vice versa for Finland andSweden (Marklund, 1988). This role reversalnot only suggest that the Nordic countries aredifferent with regard to the timing of welfarereform, but also that they follow the same tra-jectory depicted by the Nordic welfare model.

Whether welfare reform in the 1990s has ledto greater overall intra-Nordic convergencecan be examined through the sum of extremevalue differentials for countries’ fuzzy mem-bership scores. This is a measure of thedifference between the maximum and min-imum values for countries’ fuzzy membershipscore in the Nordic model of welfare in therespective years. Using this measure we findthat the start of the period is characterized bysignificant divergence as a result of Finland’smarked decrease in membership from 1990 to1993. However, as Finland improves its mem-bership in the second period and Denmarkdecreases hers, convergence characterizes theperiod from 1994 to 1997. Overall, Nordiccountries are slightly less similar at the end ofthe period than they were in the start of the1990s.

Conclusion

The comparison of policies and the assessmentof the impact of changes have been made poss-ible by using fuzzy-set theory as advocated byRagin (forthcoming). We hope that we havedemonstrated a series of advantages to thisapproach with regard to traditional qualitativecase-oriented methods and quantitative vari-able-oriented methods. First, informed bytheory and substantive knowledge, fuzzy-settheory demands an explicit definition of thesubject under investigation (here the Nordicwelfare model), thereby stimulating theexchange of ideas and knowledge accumu-lation. Second, looking at combinations orconfigurations of aspects, rather than seeingthem as existing independently of each other,allows for a holistic view of cases not presentin conventional statistical methods. In thisstudy, these aspects were seen as the constitu-tive elements of the Nordic model. Third,cases in fuzzy-set theory can have partial mem-bership of the various aspects and so betterapproach and convey the diversity of the realworld than dichotomies of yes/no assignmentscommon in case-oriented approaches. Thismakes it possible, fourth, to investigate theconformity of cases to ideal-typical locationsand to evaluate the homogeneity of cases. Inparticular, fuzzy-set theory gives us the oppor-tunity to compare diversity – differences inkind and degree – across countries and overtime in ways that were not possible before.The main contribution of fuzzy-set theory inthe examination of ideal-types is thus thebridging of the gap between case-oriented andvariable-oriented approaches.



009139 15/7/99 9:21 Page 251



Acknowledgements

Cross-national research draws on the data,knowledge, and inspiration of many people. Inparticular, I would like to thanks CharlesRagin, Stein Kuhnle, Peter Werther Andersen,Dorothy and Adrian Sinfield, Hans Hansen,Mikko Kautto, Jens Alber, Axel WestPedersen, and Maja Iben Nielsson.

References

Alban, A. and Christensen, T. (eds) (1995) The NordicLights. Odense: Odense University Press.

Ashford, D. E. (1986) The Emergence of the WelfareStates. Oxford: Basil Blackwell.

Brazelton, T. B. (1992) Touchpoints: Your Child’sEmotional and Behavioral Development. Reading,MA: Addison-Wesley.

Einhorn, E. S. and Logue, J. (1989) Modern WelfareStates: Politics and Policies in Social DemocraticScandinavia. New York: Praeger.

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

Esping-Andersen, G. and Korpi, W. (1987) ‘From PoorRelief to Institutional Welfare States: TheDevelopment of Scandinavian Social Policy’, in R.Erikson, E. J. Hansen, S. Ringen and H. Uusitalo(eds) The Scandinavian Model: Welfare States andWelfare Research, pp. 39–74. New York: M. E. Sharpe.

Hansen, H. (1997) Elements of Social Security in 6European countries. Copenhagen: The DanishNational Institute of Social Research.

Hansen, H. (1998) Elements of Social Security. AComparison Covering: Denmark, Sweden, Finland,Germany, Great Britain, Canada. Copenhagen: TheDanish National Institute of Social Research.

Howes, C. (1997) ‘Children’s Experiences in Center-based Child Care as a Function of TeacherBackground and Adult: Child Ratio’, Merrill-PalmerQuarterly, 43 (3):, 404–25.

Huber, E., Ragin, C. and Stephens, J. D. (1993) ‘SocialDemocracy, Christian Democracy, ConstitutionalStructure, and the Welfare State’, American Journalof Sociology, 99 (3): 711–49.

Kautto, M., Heikkilä, M., Hvinden, B., Marklund, S.and Ploug, N. (eds) (1999) Nordic Social Policy.Changing Welfare States. London: Routledge.

Klausen, K. K. and Selle, P. (eds) (1995) Frivillig organ-isering i Norden. Copenhagen: Jurist- ogØkonomforbundets forlag.

Kohlberg, J. E. (ed.) (1991) The Welfare State asEmployer. New York: M. E. Sharpe.

Kvist, J. (1999) Documentation: Welfare Reform in theNordic Countries in the 1990s, mimeo. Copenhagen:The Danish National Institute of Social Research.

Leira, A. (1992) Welfare States and Working Mothers.Cambridge: Cambridge University Press.

Marklund, S. (1988) Paradise Lost? The NordicWelfare States and the Recession 1975–1985. Lund:Arkiv.

NOSOSCO (1992) Social Security in the NordicCountries. Oslo: NOSOSCO.

NOSOSCO (1995) Social Security in the NordicCountries. Helsinki: NOSOSCO.

NOSOSCO (1996) Social Security in the NordicCountries. Copenhagen: NOSOSCO.

NOSOSKO (1997) Social tryghed i de nordiske lande.Copenhagen: NOSOSKO.

NOSOSKO (1998) Social tryghed i de nordiske lande.Copenhagen: NOSOSKO.

Øverbye, E. (1996) ‘How do “Organizational Designs”Influence Welfare Politics? A Study of the NordicPension Systems’, Scandinavian Political Studies 19(3): 227–55.

Palmer, E. (1998) The Swedish Pension Reform,mimeo. Stockholm: National Insurance Board.

Peisner-Feinberg, E. S. and Burchinal, M. R. (1997)‘Relations between Preschool Children’s Child-careExperiences and Concurrent Development: the Cost,Quality, and Outcomes Study’, Merrill-PalmerQuarterly, 43 (3): 451–77.

Pierson, P. (1996) ‘The New Politics of the WelfareState’, World Politics, 48 (2): 143–79.

Ragin, C. (1994) Constructing Social Research.Thousand Oaks: Pine Forge Press.

Ragin, C. (forthcoming) Fuzzy-set Social Science.Chicago, IL: University of Chicago Press.

Spiliä, J. (ed.) Social Care Services: The Key to theScandinavian Model. Aldershot: Avebury.

Wadensjö, E., Björklund, A. and Erikson, T. (eds)(1996) The Nordic Labour Markets in the 1990s.Amsterdam: North-Holland.

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

Wilensky, H. L. (1975) The Welfare State and Equality.Berkeley: University of California Press.

252 JON KVIST


009139 15/7/99 9:21 Page 252



journal of european social policy - wordpress.com · journal of european social journal of european...

Documents