Until the mid-1970s conceptions and studies of the welfare state largely revolved around a generic notion with little emphasis on types of welfare states. However, within the span of a few years, three works (Titmuss 1974; Mishra 1977; Furniss & Tilton 1977) containing the first efforts at constructing typologies of welfare provision initiated a trend of underlining the importance of welfare state variations. Empirical research during the past two decades has reinforced this trend. Statistical and quantitative analyses focusing on expenditures and revenues have documented the range of differences between countries in public provision of welfare (Wilensky 1975, 1976; Cameron 1978; Korpi 1980). Both comparative policy studies (e.g. Heidenheimer et al. 1975. 1983. 1990; Kamerman & Kahn 1981: Gordon 1988) and comparisons of two or three welfare states (Furniss & Tilton 1977; Ruggie 1984; Esping-Andersen 1985; Weir and Skocpol 1985: Ashford 1986; Rainwater et al. 1986; Davidson 1989) have indicated the wide variety of governmental responses in the area of social policy. Several recent anthologies comprising analyses of individual countries, utilizing a common framework (Flora 1986b, 1987. 1989; Friedmann et al. 1987; Dixon & Scheurell 1989), have provided standard descriptions of legislation on social benefits, thus enabling comparison.

Nonetheless, certain features of this empirical research have acted to
retard comparison and identification of major welfare state variations. The

Resumé

A kev (distinction in models of social policy snu tvpolosies of welfare provision has been a differentiation between residual und institutional types. However, despite the gradual elaboration models based on this distinction, there has been little effort to apply them in article initially examines two main approaches in identifying types of welfare states and major welfare state variations. It subsequently applies several dimensions of variation posited by the residua] and institutional models in a comparison of the United States, the United Kingdom, the Netherlands and Sweden. This comparison forms the point of departure for

earlier quantitative analyses centred attention on differences in size at the expense of other possible variations, and interest gravitated towards determinants of the growth of welfare states resulting in a major debate on the relative importance of socio-economic and political explanations (for summaries, see Uusitalo 1984; Trousdell et al. 1986). A further shortcoming of several comparative policy studies and comparisons of welfare states is that they have seldom indicated the theoretical criteria for the choice of countries under examination. Or when such criteria have been stipulated, they have frequently been of an ad hoc or post hoc nature.

Finally, much comparative research on social policy and welfare provision has been transfixed on the horns of a dilemma. On the one hand, comparisons social policy which examine ten to twenty countries, although useful as exploratory studies, entail several pitfalls. Because a large number of countries are included, the analysis must be pitched at a macro-level, and it risks being superficial; it also encounters practical difficulties of a single researcher or team of researchers adequately mastering information for so many countries. The inclusion of a large number of countries also means that a particular policy is often removed from its context (e.g. Gordon 1988). On the other hand, the spate of recent collections of countryspecific adopting a shared framework, are ironically characterized by a lack of comparisons between countries. Moreover, the common categories of analysis in these collections are in some instances so broad that they yield uncomparable information (cf. Flora 1986b, 1987, Vols 1 and 2). In summary, despite the profusion of cross-national studies during the past two decades, our understanding of major welfare state variations and their implications is still quite limited, and efforts to build upon the earlier typologies of welfare provision and to apply them in empirical analysis remain modest.

This article addresses the issue of which dimensions of variation are central to the comparative analysis of welfare states and social policy. In exploring this question the article looks briefly at some of the important discussions on types of welfare states and models of social policy in order to determine how distinctive features and important variations have been defined. More than the other typologies examined, models of social policy based on the residual-institutional distinction provide a clarification of possible dimensions of variation. Despite the prominence of these models in the literature, however, these dimensions of variation have not generally been utilized as the basis of a framework for an empirical comparative analysis of welfare states. In attempting to assess the fruitfulness of this typology, a major portion of the article uses the dimensions of variation elaborated by the models to compare four countries: the US, the UK, the Netherlands and Sweden. This application allows us to identify certain ambiguities in the dimensions as well as important omissions. The coneludingsection

eludingsectionsummarizes the advantages and shortcomings of using the
residual and institutional models as a framework for analysing welfare state
variations.

By concentrating on four countries the intent is to avoid the dilemma described above. At the same time the potential payoff is greater than if the comparison only focuses on two countries. An initial reason for the choice of these countries has been the desire to combine a 'most different' and 'most similar' design in gauging the utility of models based on the residual-institutional distinction. The underlying assumption of these models is the importance of the state in welfare provision. Intuitively, the models seem appropriate when one is interested in differences, e.g. in comparisons of welfare state leaders and welfare state laggards, such as Sweden and the US respectively. But how useful are the models in making comparisons between leaders or between laggards? or between welfare states which share fundamental features?

For example, Britain and the Scandinavian countries are often bracketed together as welfare states which put a much greater emphasis on universalism social citizenship compared to the welfare states on the Continent (e.g. Elmer 1975, 252-258; Flora 1986a, xix). Universalism and social citizenship are manifested in unified programmes covering the population with considerable funding through taxation. Recently, however. Ruggie (1984) has challenged the image of commonality between the British and Swedish welfare states and analyses them with the aid of contrasting ideal types distinguishing between the nature of state intervention.

The selection of the Netherlands has been predicated by the desire to include a welfare state of the continental variety. At the same time the Netherlands is also viewed as sharing features of the Scandinavian welfare states, especially having an ambitious level of welfare provision as evidenced heavy social expenditures. In analyses using a variety of quantitative the Netherlands and Scandinavia generally rank among the top countries (Wilensky 1975, 1976; Heidenheimer et al. 1990; Castles 1978;Therborn 1989).

Let us now turn to the question of how to analyse the similarities and
differences between welfare states by examining the models and typologies
of welfare provision in the literature.

Types of Welfare States and Dimensions of Variation

Two distinct approaches have figured prominently in the discussion on types of welfare states and welfare state variations, although at times the two approaches have derived inspiration from each other. The first approach identifies major differences through the construction of analytical models

based on contrasting ideal types. The second examines a specific country or sets of countries with the purpose of describing its/their distinctive features (e.g. Furniss & Tilton 1977: Rainwater et al. 1986; Esping- Andersen if Korpi 1987; Davidson 1989).

Basic to the first approach of model building has been a distinction between residual (or marginal) and institutional (or comprehensive) welfare provision. Although a distinction between residual and institutional conceptions social welfare had already been made in the late 1950s (Wilensky & Lebeaux 1958), it was not until nearly twenty years later that social scientists, using this distinction, presented the first typologies of welfare provision.' Building on this distinction, Titmuss presents the rough outlines of a typology of welfare provision which centres on the functions or tasks of social policy. He sketches three models of social policy: the residual model, the industrial achievement-performance model, and the institutional model. Along the lines of the earlier distinction, Titmuss identifies the major task of the residual welfare model as remedial but temporary aid because of the failure of other channels of welfare provision - the market and the family. In the case of the industrialachievement the function is to provide a minimum standard of social security, and public provision of welfare largely complements the market economy. As indicated by its name, the task of the third model is redistribution, redistribution focuses on 'command-over-resources through time' in order to achieve greater equality (Titmuss 1974, 30-31).

A more recent trend has been to concentrate on the residual-institutional
distinction, using it to construct contrasting polar models and adding several

new dimensions of variation (Mishra 1977; Korpi 1980). Table 1 combines
and sums up several of these dimensions.

These models were initially presented as heuristic devices to clarify analytical distinctions. Interestingly enough, there appears to have been little attempt to apply these models in empirical comparative analysis, although individual dimensions, especially those lending themselves to quantifiable measures - such as, social spending as a percentage of GDP, level of benefits and coverage of the population - have been utilized. Also the models have been loosely applied in descriptions of individual countries (Esping-Andersen & Korpi 1987; Kuhnle & Selle 1989). Consequently, the question remains largely unanswered as to how useful the models are when applied in comparing welfare states in the real world.

As noted above, the second approach starts with a particular country or set of countries and seeks to pinpoint the distinctive features of the welfare state/s in question. These distinctive features are then described as a particular type of welfare state, sometimes in general terms. Using this approach, Furniss and Tilton (1977), for example, outline three models: the positive state, the social-security state and the social welfare state.² They construct these models by distinguishing between different forms of state intervention for different purposes. In particular, Furniss and Tilton's models emphasize the aims of goals of intervention along with who benefits from a particular form of state intervention.

Several problematical aspects permeate the models, however. The first is that the dimensions of variation underlying the three models remain implicit. Nor is there any assurance that the defining properties of one particular model constitute a variable or dimension of variation pertaining to all models. Parenthetically, this latter difficulty also pertains to Titmuss's models (cf. Flora 1986a. xxi). Conversely, this approach's emphasis on the distinctive and the most salient features can easily obscure the fact that the same features exist in the other welfare states under examination but are less prominent. A final difficulty is the issue of broader applicability. As distinct from the ideal types of the first approach which are essentially logico-analytical constructs, these models are basically inductive and extracted from the experiences of three specific countries: the US. Britain and Sweden respectively. How useful are they for understanding welfare states in other advanced industrial countries, say. the Netherlands. France. Italy or Germany? Can the key features of the Dutch welfare state be analysed with the aid of these models or must a fourth model be devised?

Welfare State Variations: The US, Britain, the Netherlands and Sweden

In view of the problems associated with this second inductive approach.

the comparison undertaken here utilizes the dimensions of variation identifiedby institutional and residual models as summarized in Table 1. For practical reasons such as availability of data, the period of comparison focuses on the early 1980s. This comparison also allows us to explore the question of the usefulness of this sort of model building by revealing its strengths and shortcomings.

Proportion of National Income Devoted to Social Purposes

The proportion of public resources earmarked for social purposes provides some initial indication of the extent of a country's commitment to the welfare of its citizens.³ The four countries differ sharply with respect to the share of resources allocated to social purposes. Sweden and the Netherlands rank as top spenders, devoting around 40% of their GDP to welfare expenditures, defined in a broad sense as made clear in Table 2, whereas the US and Britain are located at the other end of the scale utilizing between roughly 20 and 25 percent of their GDPs for these purposes.

Level of Benefits

Again Sweden and the Netherlands tend to cluster together providing more adequate benefits than either Britain or the US. For example, as noted in Table 2, sickness benefits comprise around 2.5 percent of GDP in Sweden and the Netherlands. In Sweden sickness benefits are equivalent to 90 percent of one's daily wage or salary up to a certain ceiling, and in the Netherlands the figure was 80 percent in the early 1980s (Roebroek & Berben 1987, 682) but has subsequently been reduced to 70 percent (Info 1988). By contrast, British sickness benefits amounted to roughly a scant 20 percent of the average wage, although supplements could double the percentage (Parry 1987, 365). In the US there is no public provision of sickness benefits for short-term illness. For pensions, family benefits and unemployment compensation the picture is much the same.

Range of Statutory Benefits and Services

On this dimension the United States and Sweden appear as polar opposites. The US social policy system is generally less complete than those of other welfare states. Most obvious is the lack of a national sickness insurance, health insurance covering the entire population, the absence of family policy measures such as general family allowances, and a parsimonious public housing policy. To some extent the major gaps in social protection show up in Table 2 which presents the percentage of the GDP allocated to total social expenditures and spending by purpose. On the other hand. Sweden stands out in terms of the variety of policy areas to which substantial resources are allocated, and Swedish spending on health services, family policy and housing outstrips that of the other countries.

Nevertheless it is on this dimension that a clear-cut pattern between the countries starts to become blurred. At first glance, the Netherlands, largely because of its higher levels of spending, seems the most likely candidate for second place on this dimension too. In one respect, however, the pattern

of British spending challenges such a conclusion. Britain puts a good deal
more resources into public housing compared to the Netherlands.

One problem here concerns the conceptualization of this dimension of variation. By including both cash transfers and services, this dimension conceals a crucial difference between welfare states. They differ considerably respect to their reliance on transfers and the priority given to provision of services as reflected in public consumption expenditures (excluding defence). In the Netherlands more resources have been devoted to transfer expenditures, whereas in Sweden and Britain public consumption indicate that provision of services has been favoured (Kohl 1982, 312). In fact, confining the comparison to public consumption as a percentage of the GDP, we find Sweden at the top of the list, and the Netherlands near the bottom, and clearly below the OECD average in the early 1980s (Lybeck 1984, 61. Cf. OECD 1989, 38).

Coverage of the Population

In all the countries a trend towards a wider coverage of the population exists. Despite this trend, sizeable variations in coverage characterize the four countries. In part, these variations are a reflection of the range of benefits and services as well as the extent to which policies are targeted to special groups. However, even looking at programmes common to all four countries - such as in the area of income maintenance - we find substantial differences. In most countries income maintenance programmes (pensions, sickness benefits and unemployment benefits) cover the labour force, and estimates of coverage are percentages of the labour force (cf. Heidenheimer et al. 1983. 219; Zarf 1986, 128). A less common estimate is coverage in terms of the entire population, yet this sort of measure points up a major welfare state variation. In Sweden the basic pension, sickness benefits, parental benefits and an unemployment cash allowance cover groups outside labour force - housewives and students - although benefits are set at a lower level than those of the work-force except in the case of the basic pension.

Importance of Programmes Preventing Need

Table 2 also discloses a striking contrast between Sweden and the other three countries with respect to spending on social assistance - on 'welfare' in the narrow sense of the word. In this area Sweden spends an infinitesimal amount of its GDP - 0.2 percent - whereas spending on social assistance ranged from 1.2 percent to 2.8 percent of GDP for the other three countries. In large measure, the low level of spending in the Swedish case is because programmes in other policy areas alleviate needs and reduce the necessity

of resorting to public assistance. For example, income-tested housing allowances are available to ensure that low income groups have access to high-quality housing. Family policy provides fairly substantial benents to single parents who are very vulnerable and frequently comprise a major clientele group of social assistance programmes. Perhaps the best illustrationof programmes preventing need, however, are in the area of employment policy (cf. Esping-Andersen & Korpi 1987. 56-57). As distinct from the other three countries. Sweden puts the bulk of its resources into employment exchange services, retraining programmes. labour mobility allowances and public works projects. Only around 25 percent of the expenditure went to unemployment benefits. The picture is the reverse for the other countries, and especially for the Netherlands where unemploymentbenents for the bulk of the expenditures in this area (Olsson 1987. 17. 685-688. 725. Cf. Therborn 1989. 233 and OECD 1989. 42. 103).

Dominant Type of Programme

It is no easy task trying to fit the programmes in the four countries into the categories of 'selective' versus 'universal' for two very different reasons: the complexity of reality and ambiguities built into the categories. First, the sheer diversity of the programmes, which have incrementally evolved since the turn of the century, does not lend itself to classification into two broad categories. Second, the categories themselves are not as clear cut as they may appear at first glance. On the other hand, 'selective' is a notoriously category. It may be interpreted as targeted measures or as measures not covering the entire population. Accordingly, this category can include as diverse programmes as means-tested measures and social insurance programmes limited to employees. On the other hand, 'universal' programmes in principle apply to the whole population irrespective of the individual's financial needs or income. But the term also implies uniform benefits and frequently the absence of contributions.

In any event, universal non-contributory programmes are conspicuous by the near absence in the United States. Only in the areas of primary and secondary education does a universalist approach exist. Otherwise programmes are either targeted to the needy or they are earnings-related as in the case of social insurance legislation. Public services - hospitals, housing and employment programmes- frequently cater to welfare clientele groups.

The dominant type of programme in the Netherlands has been selective in the sense that many programmes were tied to one's occupation, sector of employment, branch of industry and/or enterprise. Furthermore, in the past compulsory programmes were often limited to groups under a certain

income - and health insurance providing medical benefits still is. Family allowances - typically and historically a universal programme in Britain and Scandinavia - followed the continental pattern and were differentiated until consolidated by legislation in 1980. Previously one set of provisions existed for wage-earners, another for the self-employed, a third for public employees and a fourth for residents (Roebroek & Berben 1987, 690-692). In many areas the trend has been towards consolidation and more uniform programmes covering all residents. Or, as it has been aptly described, the Beveridge approach (programmes aimed at the entire population and administered by state institutions) has tended to supplant the earlier reliance on a Bismarckian solution (concentrating on programmes for wageearnersto loss of income, financed by employees and employers and administered by corporatist bodies) (Roebroek 1989, 148-149). However, fragmentation still exists. These fragmented and particularistic programmes are hardly in accord with the ideal of uniform benefits inherent in universalism.Despite the Dutch welfare state comes close to approximating the universal ideal in terms of coverage of the population and standardizationof

In comparison with the Netherlands, Britain through its national insurance scheme has a more unified system of social security benefits, which applies to the entire labour force and their dependents. Payment of these benefits is, however, contingent upon past contributions. Furthermore, unique characteristic of most programmes is the provision of flatrate Non-contributory universal programmes are primarily limited to family allowances (now the child benefit) and the National Health Service. Selective measures in the sense of means-tested programmes constituted quite a small proportion of income maintenance programmes (in terms of spending around 13 percent) in the early 1980s. These measures are directed to individuals and families who fall below a certain minimum (Parry 1987, 359).

In certain respects, there are resemblances between Britain and Sweden, although stronger universalist elements are present in Swedish social policy. First, a unified system applies to a wider range of benefits. Sweden is one of the few countries, for example, with a uniform compulsory state programme of occupational pensions which covers the entire labour force (until 1982 the self-employed could opt out however). Second, as noted earlier, some income maintenance programmes encompass the entire adult population, not just persons on the labour market. Third, public services are widespread and based on the principle of equal access. Compared to the other countries, the universalism of the Swedish welfare state stands out. Nonetheless, universalism is complemented by selective measures and extensive earnings-related programmes, and universal programmes with

respect to cash transfers are less dominant today than they were two
decades ago (Nasenius & Veit-Wilson 1985, 146; Olsson 1986, 7-12).

Type of Financing

Of the four countries the Netherlands relies most heavily on the social insurance model which entails contributions and fees. Funding of income maintenance programmes is overwhelmingly through contributions by employees and employers. Even family allowances are based on the insurance principle. Similarly, health services are largely financed as 'fees for services' through an extensive medical insurance system. Fees apply also to secondary education. Contributions by insured persons constitute the lion's share of social security receipts, and contributions are earningsrelated. funding of social security programmes through taxation primarily limited to social assistance, extended employment benefits, housing subsidies and allowances, and exceptional medical expenses (Roebroek & Berben 1987, 688, 695, 703).

At the other extreme is Sweden where the contributions of insured persons, as a share of social security receipts, has dwindled during the past two decades, amounting to a mere 1 percent in the early 1980s. By contrast, the employers' share has increased over the years, jumping sharply in the early 1960s and again in the mid-1970s (Olsson 1987, 53). Uniquely, at least in relation to our other countries, employers fund non-contributory programmes, such as the basic pension, national dental insurance and the labour market allowance which is available to persons without unemployment and even to persons without labour force status. Funding employers does not take the form of contributions earmarked for their employees but is more in the form of a general payroll tax. Looking at total social expenditure, taxes accounted for nearly 85 percent of the revenue.

Both Britain and the US represent hybrids as far as financing is concerned. Taking the British case first, the national insurance scheme is funded primarily by the contributions of employers and employees with a small part (20 percent) coming from government subsidies (Parry 1987. 359). Non-contributory benefits are financed through taxation. Taken as a whole, slightly over half of income maintenance expenditure is financed by contributions. However, social spending is overwhelmingly financed by taxation. Contributions only amounted to approximately 25 percent of total social expenditure whereas in the Netherlands contributions comprised nearly 50 percent. In the US case, social insurance is also essentially financed through contributions of employees and employers. Funding through contributions plays a greater role compared to Britain because

social insurance constitutes a larger proportion of social expenditure than
in Britain.

The Role of Private Organizations

One of the fundamental differences between welfare states is what has been called 'stateness' (Flora 1986a, xvii) - that is, the degree of state involvement in welfare provision. To some extent, the role of private organizations can be seen as an inverse measure of stateness. According to Flora, the major implication of stateness is that involvement of the state circumscribes or delimits the arenas of activity for private organizations. An inherent difficulty in this conceptualization, however, is its global and residual nature; it leaves open all manner of diversity.

In attempting to sort out this diversity, we need to distinguish, first, between welfare provision as social benefits and as services and, second, among types of private organizations - economic enterprises (to which Flora does not allude in his discussion), religious institutions, ethnic associations voluntary organizations or networks. On the basis of these distinctions we can identify basic patterns or tendencies. Looking initially at the provision of benefits and private organizations, three variants are especially important.⁴ The first is the extent to which enterprises provide benefits for their employees in the form of occupational welfare. The second variant, which might be termed corporatist welfare, is where the organizations of employers and employees are jointly engaged in the determination and administration of benefits. The third is altruistic welfare where charitable and voluntary organizations are prime agents in the dispersal of benefits or provision of services.

The extent of occupational welfare is most pervasive in the US. In areas where the US public system of social protection is incomplete or nonexistent, and enterprises often provide benefits in connection with employment and as a means of attracting desirable labour. One telling illustration and contrast with Europe is the lack of legislated paid vacations. Instead individual employers and enterprises decide vacation benefits. Similarly, sickness and medical benefits are frequently fringe benefits related to employment. The end result is fragmentation, partial coverage and enormous inequalities in the provision of benefits - with occupational welfare tending to conform to the pattern of market distribution. Coverage is most widespread among professionals, executives and administrators and least among certain categories of workers, such as those in farming and services {Statistical Abstract 1988, 396-397). The level of fringe benefits, moreover, is often graduated so as to replicate the pay scale and place a premium on seniority.

Nonetheless, occupational welfare co-exists with public provision in all

four countries, and over the years occupational benefits have steadily grown. All countries have experienced a proliferation of benefits in kind, such as the use of a company car or equipment. partially or wholly subsidizedmeals, recreation facilities. discounts, transport subsidies, services provided by company doctors. etc. The profusion and growth of these benefits should cause us to be wary of the idea that a simple inverse relationship exists between state provision and occupational welfare.

Of our four welfare states, two - the Netherlands and Sweden - are frequently classified as prime examples of 'neo-corporatisnr or 'democratic corporatism' (Wilensky 1976, 1982; Schmitter 1981; Cameron 1984; Katzenstein Cawson 1986). This categorization rests on the degree to which the organizations of employees and employers are represented in the broad process of policy formulation and implementation. But what role do these organizations play in the provision and administration of social benefits? That is, what is the extent of corporatist welfare in these two countries? Despite shared associations with neo-corporatism, the patterns of corporatist welfare in the two countries diverge drastically, with state institutions generally playing a dominant role in Sweden. At first glance, this may seem highly paradoxical in view of the much greater strength of union organizations in Sweden compared to the Netherlands."'

In the Netherlands corporatist welfare arrangements are found in the areas of occupational pension schemes, the administration of basic pensions, unemployment and sickness insurance, and family allowances. These benefits have been administered by one of two sets of corporatist hierarchies of organizations comprised of employer and employee representatives: the Labour Councils (Raden van Arbeid), organized on a regional basis, are responsible for basic pensions and family allowances; and (2) the Industrial Insurance Boards (Bedrijfsverenmgenen), organized according to branches of industry, deal with sickness, disability and unemployment and the recently introduced supplementary benefit. At the apex of both hierarchies is the Social Insurance Council, in which the state, employers and employees are represented.⁶

The involvement of Swedish unions and employers" organizations in the administration of social benefits is dwarfed in comparison to the Dutch case. Instead, public bureaucracies administer social benefits. The major exception is unemployment insurance benefits. However, unemployment insurance is organized solely by the unions and heavily subsidized by the state - and does not qualify as a corporatist arrangement defined as bodies with joint representation of employers and employees. Corporatist structures and arrangements are essentially more important in other policy areas such as employment and labour market policy, industrial policy. labour legislation and the policy process more generally.

Turning to altruistic welfare and the role of voluntary agencies and nonprofit we find that they are generally more active in the provision of services compared to benefits. However, in the Netherlands medical benefits, under the Health Insurance Act, are administered by health insurance funds, which are non-profit organizations. In addition, Dutch non-profit organizations play a vital role in housing.

With respect to services, the UK and Sweden are similar in the marginal role of private provision in education and the health services. In both countries private provision covers 10 percent or less of the relevant population these two fields (for the UK see Parry 1986, 162).⁷ In Sweden this pattern applies to a wider variety of services, such as employment services and training programmes, day care and other social services. The two countries are also similar in the extent to which state institutions are responsible for the administration and delivery of services. They differ, however with respect to the role of local authorities. In contrast to the more centralized pattern in Britain, both local and regional government in Sweden have an independent power of taxation and these levels of go\ernment are active in providing public assistance, health care, education, social services and housing.

The Netherlands and the US represent the other end of the scale but also completeh different variants. In both countries state provision in education and health is not as extensive, although the state plays a substantial in funding. In fact, the Netherlands represents nearly a mirror ,nidge of Sweden and Britain inasmuch as only about 10 percent of all health, education and social-welfare services are provided directly by local or regional government (Kramer 1981, 19. Cf. James 1987, 399). Instead education and health services, social services, day care etc. are largely organized on a denominational basis reflecting the religious divisions or pillarization of Dutch society. More specifically, organizations often originally in one of the pillars have been major providers of education, health tare housing and social services. Collectively, these organizations tire referred to as patticulier initiatief, which is generally translated as private enterprise (Brenton 1982, 61). In certain respects, 'private enterprise a misnomer A cardinal feature of the organizations is that they ,re non-profit organizations. Further, although their legal status is private, h\ virtue of heavy funding from the state and their tasks as providers of legislated benefits and services, they de facto assume quasi-public status.

In the US, market solutions figure more prominently in the provision of emcts - education health, employment, day care and personal social M_mtcs Voluntary agencies have traditionally and are still involved in ¦ •inning personal social services (see Magill 1989), but since the 1960s the practice ot purchasing services has accelerated, increasing the element of t jhhc funding. In addition, because of the leeway provided by limits of

state action, the voluntary element is strong and the ethnic and religious
mosaic of US society is in part projected on to the delivery of services.

In sum, looking at this dimension of variation across the four countries, the role of private organizations stands out in the Dutch and US cases. In the Netherlands we encounter an intricate pattern of cross-cutting segmentation in the provision of benefits and services with state administration mainly to programmes of social assistance. Corporatist structures are heavily involved in the provision of cash benefits, while denominational and non-profit organizations with a strong element of voluntary work are active in services.⁸ Simultaneously, state legislation applying to a wide range of benefits and certain services ensures coverage of most of the population, uniformity in benefits and a supervisory framework. this way, private organizations are integrated into the public provision of welfare. By contrast, the role of private organizations in providing benefits and services in the US is more autonomous and supplemental state provision, and in many instances it is the only form of benefits and services. Nor is the profit-making capacity of private organizations

Ideology of State Intervention and Distribution Based on Needs

The philosophy and values underlying the welfare state form a crucial dimension of variation. In fact, some authors (e.g. Furniss & Tilton 1977: Ashford 1986) argue that values constitute one of the most decisive distinctions welfare states. This dimension, as formulated in Table 1. focuses on conceptions of the state and principles of distribution.

In the US several factors - the lack of a feudal past and a centralized state encompassing the Church and Crown, a revolutionary break with Britain, the enthusiastic and enduring embrace of liberalism, and the settler experience - created a unique environment with respect to the formation of the state and attitudes towards the state. These influences worked to produce a fragmented state structure and the prevailing conception that the proper sphere of state action is of a limited nature and individuals are primarily responsible for their own welfare.

In the Netherlands social doctrines embedded in Catholicism and Protestantismhave conceptions of the legimate purview of the state. More specifically. Catholics have enshrined the principle of 'subsidiarity', essentially a hierarchical principle of intervention where the state intercedes only when smaller and lower bodies have failed, while Protestants subscribedto doctrine of 'sphere sovereignty* which holds the sovereignty of God over the family, state. Church and society, each of which has its own sphere of influence (Kramer 1981. 20-21: Roebroek 1989. 148. Cf. Therborn 1989, 206-208). These beliefs have tended to curb state responsibilityand

bilityandelevated the role of church authorities and individual responsibility.On other hand, a hierarchical view of authority has placed the state above divisional interests and the state has been envisioned as a higher agent in harmonizing the conflicting interests of groups with the aim of maintaining their autonomy (cf. Wilensky 1982, 353). This formula has been applied to the religious communities themselves and to major aspects of the provision of welfare.

British notions of state responsibility in welfare provision have been broader than those in the US and the Netherlands. More generally, the state has been viewed as a vehicle for combating social evils and providing a minimum standard of life for all. Nevertheless, major currents of economic liberalism have set important limits on state intervention in terms of proper forms of action in at least two decisive ways. First, although liberal dictates on the separation of state and the economy have been eroded through its state intervention, they prevail to the extent that intervention is confined to compensating for market deficiencies rather than modifying or abolishing them (cf. Ruggie 1984, Chapter 1). Second, economic liberalism prevails in the concern attached to individual incentives in the provision of social benefits.

Three aspects of Swedish conceptions of the state have facilitated the expansion of state responsibility in welfare provision. The first is basically an instrumentalist view - that is, that the state is an instrument for achieving collective purposes and the common good. The second is a communitarian notion of the state. Solidarity has infused views of the state, and in socialdemocratic as distinct from liberal thinking, the ideas of the state and society have been intertwined (Tingsten 1973, 345). Simultaneously universalist social policies have tended to nurture these images of the state, since these benefits and services are available to everyone. The third aspect, shaped by a spirit of social trust, is that the state is less perceived as an instrument of repression than in many other societies (cf. Allardt 1986; Kuhnle & Selle 1989).

The strength of an ideology of distribution based on needs also varies in the four countries. In the US such an ideology has been overshadowed by a creed stressing that rewards should be allocated on the basis of individual initiative and achievement. However, alleviation of misery and need, along with social protection, have been motivating forces behind welfare legislation. In Britain and the Netherlands distribution based on needs has been inextricably related to provision of a minimum standard. Typical for Britain, however, is that the discussion of basic needs has been heavily influenced by studies on poverty. Equally important, the doctrine of 'less eligibility' - the principle that the resources of recipients of welfare should be less than the lowest income of the working poor - has coloured the British definition of an adequate standard. These two influences have

resulted in a narrow conception of basic needs and an emphasis on a bare minimum. This emphasis reflects a concern that social benefits might corrode individual incentives and disrupt the smooth functioning of the market. By contrast, the Dutch define a socially adequate standard in more generous terms. Religious humanitarian values have stressed the elimination of human misery as a prerequisite for living a Christian life; poverty has been viewed as robbing persons of their fundamental human dignity and preventing the needy from being able to live according to the laws of God (cf. Roebroek 1989. 148). Nor has the conception of needs been limited to material needs. These views have undermined market considerations and, unlike Britain, have brought the social minimum in line with the minimum wage (cf. Therborn 1989, 212-213). In Sweden the fulfilment of basic needs has been increasingly viewed as a right of citizenship.This with citizenship has also broadened the definition of basic needs inasmuch as citizenship consists not only of participation in politics but also the sharing of social goods.⁹ Notions of an adequate minimum have been successfully upgraded, and in the 1950s the idea of an equal standard throughout one's lifetime or a guaranteed standard replaced the minimum standard conceived as a safety net. Furthermore, distribution based on needs has been fused with an ideology of equality which underlies not only equal opportunities but equality in outcomes.

To conclude, although conceptions of the state and distribution based on needs are crucial variations, a fundamental issue is whether the ideological dimensions suggested by the contrasting ideal models based on the residualcomprehensive are sufficient to capture the important ideological differences between the welfare states considered here. In contrast with the previous dimension which was excessively broad, this dimension appears too narrow for exploring the relevant variations between welfare states.¹⁰

In summary, this comparison using the dimensions in Table 1 discloses substantial variations in the role of the state in welfare provision, the coverage of the population, the range of statutory provision of benerits and services, the level of benefits. dominant type of programme, and the nature of funding, the role of private organizations and the ideological underpinnings of welfare provision. However, it is extremely difficult and hardly meaningful to classify the four welfare states in terms of the residualinstitutionaldistinction. in terms of the contrasting types, the US comes closest to approximating the residual model - and Sweden the institutional model. Nevertheless, both countries deviate in important ways from the defining characteristics of each of the models. Furthermore, Britain and especially the Netherlands are difficult to accommodate within this scheme. Instead the four countries represent hybrids uniquely combiningvarious of each dimension. By way of conclusion, it is

fruitful to sum up the principal variations of the four welfare states, thereby
delineating the distinctive features of each.

The United States: The Minimal Welfare State

Both a residual and an institutional conception of social welfare co-exist uneasily in the US, resulting in a two-tier system of public provision of welfare. Contributory social security programmes form one tier, and these programmes enjoy overwhelming public support. During the past decades the resources allocated to them have steadily grown from roughly 35 percent of social expenditures in 1960 to around 50 percent in 1985 {Statistical Abstract 1988, 334). The second tier consists of non-contributory programmes to the poor - or more precisely categories of the poor - on a means-test basis. Expenditures for these programmes have also grown, but the increase is relatively modest compared to that in the case of socialsecurity Nor are these targeted programmes widely supported by the public.

A basic feature of the minimal welfare state is the lack of a fully developed system of social protection. Underpinning the limited nature of the welfare state is a philosophy of economic individualism, self-help and a preference for market solutions involving a minimum of state intervention. As we have seen, US social legislation is characterized by major gaps, especially in the areas of sickness insurance, health insurance, family policy, public housing and social services. In all these areas, market solutions, fragmentation, and incomplete coverage prevail. Diversity and fragmentation reinforced by federal institutions, a heavy reliance on voluntary agencies, sometimes of a religious or ethnic complexion, and the large role of occupational welfare (cf. Glazer 1986). Also in the area of public assistance, the US lacks a legislated statutory provision which applies to the entire population. Instead nationally legislated programmes provide assistance to vulnerable groups: the elderly, the blind, the disabled and families (usually women) with dependent children. The distinction between the 'deserving' and the 'undeserving poor' survives as a cornerstone of the minimal welfare state.

The Netherlands: The Segmented Welfare State Based on Transfers

Social policy in the Netherlands is characterized by a high degree of segmentation with respect to both benefits and services. Benefits are attached not merely to labour market status but to occupation, sector of employment and even place of work. Furthermore, in contrast to the other countries, a complex network of corporatist bodies representing employees and employers - and at the apex including also government representatives

resentatives- is responsible for the administration of most benefits. The segmented nature of the Dutch welfare state is also reflected in the faet that services such as education, health and personal services are mainly organized and run on a confessional basis.

An outstanding feature of the Dutch welfare state is the extent to which benefits consist of transfer payments (26 percent of GDP in 1983). The bulk of these benefits are based on contributions, and the insurance principle is widely applied. It is also striking that transfers represent a high level of spending in terms of both social insurance benefits and welfare grants (the latter was 6 percent of GDP in 1987) (OECD 1989, 39-41). This paradox probably reflects two circumstances: first, the lack of benefits not attached to one's job and to past contributions and, second, the Dutch emphasis on an adequate minimum standard for the entire population. In the Netherlands social minimum has been defined as equivalent to the net minimum wage, and in the early 1980s the minimum wage was much closer (roughly 75 percent) to the average wage than in most countries (OECD 1989: 68-70). fairly ambitious definition of a social minimum, linked to the minimum wage, also constitutes a distinctive feature of the Dutch welfare state (cf. van Amelsvoort 1984).

Britain: The Guaranteed Uniform Minimum Welfare State

The scope of the British welfare state has been ambitious in terms of its uniform state programme of cash benefits and its breadth of services in kind in health, education, personal social services and housing. It is less ambitious with respect to the level of provided benefits, and this is largely because of the principle of a uniform national minimum. This principle has also been manifested in flat-rate benefits, which have been one of the hallmarks of the British welfare state. Compared to the other welfare states, earnings-related benefits have occupied a uniquely modest position in British income-maintenance programmes." Moreover, uniformity in benefits refers not only to equal benefits but also to roughly the same level of benefits regardless of type (Rainwater et al. 1986. 160). In the early 1980s, for example, most cash benefits for a married couple ranged between slightly less than 30 percent and 40 percent of the average earnings of industrial male workers (Parry 1986, 188). In large measure, this particular pattern of benefits has reflected an underlying aim of the British welfare state: to guarantee a minimum standard and social protection for the working population. Unfortunately, benefit levels were so low that they were only modestly above the poverty line, which is conventionally defined as an income beneath the threshold for social assistance, i.e. 33 percent of average earnings for a married coupled in 1980 (Parry 1986, 188-189; 1987. 372). Lastly, the goals of a minimum standard and social protection for

workers have also shaped the provision of personal services and housing
in that they tend to be targeted to low-income groups and the socially
deprived (Ruggie 1984, 199-210; Heidenheimer et al. 1983, 92-93).

Sweden: The Comprehensive Welfare State

Among the typical features of the Swedish welfare state is its comprehensive nature with regard to the range of publicly provided benefits and services, the coverage of the population and the high levels of benefits (Heckscher 1984, 79-80; Furniss & Tilton 1977, Chapter 6; Allardt 1986, esp. 114). One effect of this comprehensive scope has been to diminish the impact of market forces on distribution, which results in less dependency upon one's market position and a decommodification of wants and needs (Esping- Andersen & Korpi 1987, 40-41). Instead medical services, education at all levels, day care, family services and transportation have largely assumed the character of public goods (Siim 1987, 3), providing a new dimension to the welfare state as a 'social service state'. Furthermore, to a greater extent than the other welfare states considered here, there is a strong emphasis on universalism, with entitlement to benefits and services based on the idea of social citizenship (Elmer 1975, 252-258; Esping-Andersen & Korpi 1987). The funding of benefits also reflects the idea of social citizenship rather than the insurance principle. Taxation constitutes the main source of revenues and contributions by the insured, and fees are minimal. A final distinction is related to a commitment to equality and solidarity as socialpolicy (Esping-Andersen & Korpi 1987). What is distinctive is that the commitment to equality is not limited to equal opportunities but extends to equality of result.¹² Solidarity is manifested in inclusive policies as a means of integrating the entire population (cf. Davidson 1989).

An Assessment of the Strengths and Limitations of the Residual and Institutional Models

One major strength of model building based on the residual-institutional distinction has been to set forth a number of crucial distinctions in a systematic fashion. Yet, as far as I know, comparisons of welfare states have failed to make the most of this asset by utilizing as extensive a battery of variables as in the preceding discussion. An application of the dimensions of the models yields a number of insights, and it has allowed us to map out the similarities and differences between our four welfare states as well as to pinpoint their distinctive aspects.

Contrary to much of the literature which emphasizes common trends and
convergence in the development of welfare states, diversity emerges from

this comparison. Nor is there a coherent pattern in the interrelationships between dimensions, conforming to the logic of the ideal types. This is clearly demonstrated by the welfare states of the Netherlands and the UK. On the one hånd, the Netherlands represents a 'weak' state where, under the banner of 'particulier initiatief, a great deal of responsibility for benefits and services has been entrusted to non-profit organizations, often of a confessional nature, and labour market "partners". In Britain state responsibilityhas and is greater by virtue of unified state programmes, public taxation comprising a larger component of funding, and state provision and administration of benefits and services. On the other hånd. state effort in the UK is less than the Netherlands in terms of social spending and levels of benefits.

Indirectly an important issue raised here concerns corporatism and the welfare state. The multiple meanings of corporatism and their application to a variety of dissimilar contexts have tended to create more confusion than clarity. In an effort to overcome the confusion, corporatist theories have proposed a number of distinctions. Very briefly, they have distinguished types of corporatism (societal versus state), variants of types (liberal and social) and levels of corporatism (macro, meso and micro) (cf. Schmitter 1974; Katzenstein 1985; Cawson 1986). In the area of welfare state theory and social policy, however, few distinctions or an inventory of uses have been made, and definitional disarray prevails. The differences between the cases of Sweden and the Netherlands indicate the necessity of clarifying the discussion of corporatism and the welfare state. This is underlined by Sweden with its statist stance and the Netherlands with its 'social partners' approach which stand out as opposites in the extent to which the organizations of employees and employers are involved in the provision and administration of benefits (cf. Therborn 1989).

Despite the usefulness of applying the dimensions of the residual and institutional models for descriptive purposes in order to establish similarities and differences between specific countries, the models are flawed in several fundamental ways. First, as we have seen, the models are unsatisfactory as a classificatory scheme. No welfare state in the real world exhibits all the defining properties of the models but is instead a combination. Further, highly different welfare states such as Britain (Mishra 1977, 91-92) and Sweden (Esping-Andersen & Korpi 1987) have been classified as being of the institutional type. An obvious problem is the overly simplistic nature of the scheme with only two alternative types. As the preceding comparison demonstrated, it makes little sense to try to subsume the diversity of the four welfare states by classifying them as one of these two types.

A second shortcoming is that ideal type models are essentially static, but
welfare states are dynamic and policies are continuously changing. A
developmental perspective has cropped up in the discussion of the two

models in that welfare states are viewed as generally moving from a residual to an institutional type. From this perspective, the theoretically interesting issue is the dynamics of the process or how welfare states have been institutionalized - not that they are institutional, which the models might help to establish. Furthermore, the rudimentary developmental perspective contained in this reasoning is unilinear and scarcely admits to a reversal in trends.¹³

A third limitation is that although the models may be useful for descriptive purposes, they provide virtually no guidance in formulating explanations. The crux of the matter is that the dimensions consist of attributes - or are conceptualized as attributes - and not as determinants and outcomes respectively. As a series of characteristics, the models compound what is to be explained and possible explanations.

A fourth difficulty inherent in a scheme based on polar ideal types is that it limits the framework of analysis to two categories conceived as opposites. This problem has been especially acute in four key areas, which deserve further comment. A major weakness of the scheme, which became apparent in attempting to establish the dominant type of programme in each of the four countries, is related to reducing programmes and policies to dual categories of selective versus universal. A more conventional, and arguably more useful, categorization is the threefold division of programmes into (1) means-tested, (2) universal and (3) earnings-related.¹⁴ As has been cogently observed, the residual and institutional models are associated with means-tested and universal programmes respectively and easily neglect contributory, earnings-related programmes which are probably the most widespread type of programme in many capitalist democracies. Even countries which are categorized as institutional welfare states have substantial programmes (Marklund & Svallfors 1987, 13-19). Ironically, however, these critics of the residual-institutional dichotomy fall victim to the same error when the advance their own model of dual welfare. In their scheme, means-tested and universal benefits are combined into a single category which fails to recognize the major differences between the two. In short, models which reduce programmes or benefits to dual categories are both inadequate and misleading.

The same difficulty also surfaces in the discussion of the funding of social policy which is restricted to dual categories: contributions and fees vs. taxation. On the basis of an analysis using the categories, Sweden and the UK are similar in a heavy reliance on taxation, but Sweden does not neatly fit into these two categories. Funding in Sweden represents an interesting and perhaps unique variant which becomes apparent when the sources of funding are included in the analysis. Despite the fact that the social insurance schemes of the other countries are based on joint contributions by insured persons and employers, Swedish employers foot a larger share

of the bill for social spending. The logic of the models tends to obscure this possibility. Nor do the models include other crucial aspects of funding, such as the form of contributions - flat-rate or a proportion of earnings - or types of taxation.

Similarly, the ideological dimension of the residual and institutional models is inadequately developed. (It is not even included in Korpi's first version (1980).) In Mishra's discussion (1977), focus is on two facets of ideology: state intervention and distribution based on needs. Because of the strictures provided by polar ideal types, examination concerns the degree of their presence and not the importance of other values or principles. Surely, the analysis of ideas, ideologies and principles as a welfare state variation in its own right and a determinant of other variations requires fuller development.

A final area where dualistic categories are deficient concerns the dimension the role of private organizations in welfare provision. As we have seen, their role is relatively more prominent in both the US and the Netherlands yet their roles are very different and the degree and pattern of interaction with the state are dissimilar. They are so dissimilar that it is misguided to put the two countries in the same bracket. Again the problem is that a simple dichotomy focuses exclusively on involvement versus noninvolvement private organizations and assumes a state-private split. The scheme fails to distinguish between various kinds of private organizations and the possibility of a state-private mix or a variety of mixes.

A fifth fundamental weakness of the residual-institutional models, at least in Mishra's version, is that they build upon a single underlying dimension: the extent of state responsibility in welfare provision. A closer examination of the main features Mishra attributes to each ideal type reveals that they stem from this basic trait (Mishra 1977. 90-91). In this way his framework largely overlooks other sorts of variations.

What types of variations are missing or are not explicitly included? I would suggest that there are at least five areas which require more attention: the form of intervention, other "sectors' of provision, the bases of entitlement, purposes or rationale of intervention, and strategies of redistribution their effects. Furthermore, these variations ought to be divorced from the dichotomous thinking inherent in polar ideal types.

The extent or scope of state intervention must be complemented by considering the nature or form of intervention. Such a consideration has not been totally missing in all models of welfare states. Korpi's version of the models (1980) hints at this sort of variation by including employment policy and programmes which eliminate need as defining features of the institutional model, and other authors have made the form of intervention the crucial variation underlying their models (Furniss & Tilton 1977; Ruggie 1984). However, much of the current dissatisfaction with welfare state

theory centres on the fact that the role of the state has not been sufficiently
differentiated and calls for explicitly defining the varieties of tasks involved
in the state's role in welfare provision.

Similarly, little attention has been given to the distinction between benefits as cash transfers and benefits in kind in the form of services. In fact, in Mishra's models the two are unsatisfactorily combined as a single dimension of variation. However, as we have seen in the cases of the Netherlands and Sweden, welfare states can differ greatly in terms of either transfers or services. Furthermore, as again strikingly demonstrated by the Netherlands, divergent principles and organization can co-exist in the provision of cash benefits and services respectively.

Since the focus of the models has been on the extent of state responsibility in welfare provision, they are poorly equipped to deal with other welfare providers or 'sectors' of provision. Recent discussions of 'the welfare mix' looking at the household, market, state and informal economy (Rose 1986, 14-17), the public-private interplay (Rein & Rainwater 1985) and sectors of provision (e.g. Johnson 1987, 1989) grapple with this inadequacy. In comparison to the residual and institutional models, these frameworks represent an advance by abandoning a dichotomous paradigm and seem more amenable to an analysis of interaction.

In their current versions, the models do not explicitly make the basis of entitlement a variation. Furthermore, the analytical construct of polar contrasting types implies only two bases of entitlement which is neither realistic nor fruitful. More commonly, the discussion on entitlement has instead tended to focus on three broad categories - needs, contributions and rights (e.g. Rainwater et al. 1986, 126-132), but this categorization fails to capture all varieties of entitlement (Sainsbury 1989. Cf. Barry 1989). In addition these three categories of entitlement have been linked to specific types of programmes, which has had the unfortunate effect of stultifying the conceptions of needs and rights. For example, too frequently needs have been associated with means-tested programmes in policy analysis.

A consideration of the purposes and rationale of social policies offers one possibility in broadening the examination of ideas and values in shaping the welfare state. This may interject the problem of intention into the analysis, but purposes and rationales of intervention or non-intervention and of policies constitute one yardstick in evaluating policy outcomes and the impact of specific welfare states.

Finally, by concentrating on attributes the models only indirectly confront the issue of effects, impact and policy outcomes. One area central to welfare state theory is missing - and that is the strategies of redistribution and their effects. It is imperative to incorporate redistribution into the analysis but at the same time avoid the limitations of a framework based on polar opposites. A crucial distinction, frequently made in the literature, is

between horizontal redistribution, alternatively defined as redistribution within classes or within the life-cycle of individuals, and vertical distribution from the better off to the worst off or between classes or income groups. Intuitively the residual model with selective measures targeted to the needy may seem more conducive to vertical redistribution, and the institutional model with universalist policies applying to the entire population would seem to promote horizontal redistribution (cf. Hadenius 1987. 160-165: Davidson 1989).

This line of reasoning entails two serious problems. In the first place, the vertical-horizontal distinction may be mutually exclusive definitionally, but it is not necessarily so empirically in terms of policy outcomes. Policy measures leading to a horizontal redistribution may result in improvements for the population as a whole, but the relative gains for disadvantaged groups are far more important inasmuch as their situation is now much better than originally (Smeeding et al. 1988, 111). In the second place, and more importantly, as the previous comparison revealed, real welfare states deviate substantially from the models. This simple fact would seem to necessitate caution concerning the usefulness of an analysis of redistribution exclusively based on integrating the vertical-horizontal distinction into the residual and institutional models. In fact, the four countries exhibit quite distinctive strategies of redistribution, and the effects of these strategies also diverge.

Is it possible to adapt the models so as to include the five neglected research concerns mentioned earlier? Obviously, a major obstacle is a framework based on opposites. Does the solution then lie in constructing ideal type models which are not polar types - say. for example, in developing Titmuss's three ideal types (cf. Kuhnle & Selle 1989; Pettersen 1985)'? Probably not. There is no guarantee that a trichotomy is superior to a dichotomy, and both variants suffer from the same weaknesses: they limit a priori the number of categories of analysis and they assume co-occurrence of certain defining properties. One possible improvement is to retain and expand the battery of fundamental variations laid down in the residual and institutional models but to make these distinctions open-ended. This is one of the main lessons of attempting to apply the dimensions empirically to real welfare states.

Annex: Components of social expenditure in Table 2.
The components of the various categories of social expenditure are the following:

---

NOTES

* This article is a revised version of a paper presented at the Annual Meeting of the Swedish Political Science Association, Umeå, 2-4 October 1989.1 would like to thank Elias Berg, Mats Dahlkvist, Lennart J. Lundqvist, the journal's referees, and especially Axel Hadenius for their valuable comments.

1. The original formulation of the distinction emphasized that the residual conception assumes that the family and market are the 'natural' channels of welfare, and that only when there is a malfunctioning in these channels should one rely on public welfare structures; and reliance should be of an emergency and temporary nature. By contrast, the institutional conception sees public welfare structures as an integral part of society providing an organized system of social services on a regular and legitimate basis to aid individuals and groups to attain satisfying standards of life and health (Wilensky & Lebeaux 1958, 138-140).

2. Furniss and Tilton do not regard the first model as representing a bona fide welfare state, since it does not meet their defining characteristic of a welfare state, i.e. the establishment of surrogate forms of poverty for all those without an adequate basis for security and self-development.

---

3. At the same time one must be aware of two drawbacks with this indicator. First, differences in demographic composition of populations (the proportion of elderly or children) can affect the size of spending "automatically'. Second, social spending as a percentage of GDP can be misleading as an indicator of welfare commitment over time. In this instance, differences in the growth of the GDP are not taken into account. For example, from the 1970s to the early 1980s social expenditures as a percentage of GDP increased rapidly in Sweden and Denmark, while the increase in Norway was sluggish, yet real growth in social expenditures in Norway was greater than in Sweden or Denmark (Johansen 1986).

4. It needs to be stressed that these variants by no means cover the range of possible variations.

5. In cross-national comparisons of density of union membership Sweden generally ranks at the top, whereas the Netherlands is at the lower end of the scale. In the early 1980s between 80 and 85 percent of Swedish employees were organized in unions, but in the Netherlands the figure was only around 40 percent (Wallerstein 1989. 482. Cf. Korpi 1983. 31) and had fallen to about 30 percent toward the end of the decade (Statistical Yearbook of the Netherlands 1988. 146).

6. A reorganization reform transformed the Labour Councils into regional offices of the Social Insurance Bank in 1988. The corporatist nature of the councils was retained in the representative composition of the regional offices at the same time as it was modified. The board of the offices consists of representatives of employees. employers and government appointed representatives with preference given to women's and pensioners' organizations (Social Security in the Netherlands 1990. Chapter 2).

7. Despite Thatcherism and increasing privatization in Britain. inroads in education and health services have been fairly modest (see Parry 1989. 13. 24-25. Cf. Papadakis & Taylor-Gooby 1987).

8. Although the voluntary element has declined over the years. it is still quite pervasive in services, such as home helps. social work and day care {Statistical Yearbook of the Netherlands 1984, 362-363).

9. For one of the clearest statements on citizenship as participation in politics and the sharing of social goods see Nils Karleby's discussion on the 'problem of taking part' (Karleby 1926. 87-102).

10. In faet. in applying this dimension. I have found it necessary to be more inclusive by. first. looking beyond conceptions of state prerogatives per se toward tangential sets of ideas and principles, such as individualism. economic liberalism. equality and security, which indirectly promote or constrain state action. Secondly. a discussion of distribution based on needs requires an examination of the breadth or scope of needs (cf. Esping-Andersen & Korpi 1987. 41) and conflicting principles of distribution.

11. Earnings-related supplements (ERS) were available in the case of unemployment and sickness benefits, and widows' pensions from 1966 to 1982. and a state superannuation scheme (SERPS) providing earnings-related benefits to supplement the Hat-rate pension being phased in. However, legislation in the mid-1980s has encouraged contracting out of SERPS and has assigned priority to private occupational pensions and personal pensions based on insurance premiums.

12. This emphasis is reflected for example. in a continuous monitoring of inequalities in Swedish society. even published in English (cf. Social Report on Inequality in Sweden 1981: Inequality in Sweden 1988).

13. For a recent effort to utilize the institutional and residual models in the analysis of welfare state retrenchment. however. see Pierson 1990.

14. Although this classilication is an improvemenl upon onc hasedon a dichotomy, it is (see Sainsbury 1988).

REFERENCES

Allardt, E. 1986. 'The Civic Conception of the Welfare State in Scandinavia', in Rose. R. i:
Shiratori, R.. eds. The Welfare State East and West. New York: Oxford University Press.

Barry, B. 1989. The Welfare State versus the Relief of Poverty', paper presented at the
ECPR workshop on Needs, Contributions and Welfare, Paris.

Brenton, M. 1982. 'Changing Relationships in Dutch Social Services', Journal of Social Policy
11, 59-80.

Cameron, D. 1978. The Expansion of the Public Economy', American Political Science
Review 72, 1243-1261.

Cameron, D. 1984. 'Social Democracy, Corporatism, Labour Quiescence, and the Representation
Economic Interests in Advanced Capitalist Society', in Goldthorpe, J. ed.,
Order and Conflict in Contemporary Capitalism. Cambridge: Cambridge University Press.

Castles. F. G. 1978. The Social Democratic Image of Society. London: Routledge & Kegan
Paul.

Cawson, A. 1986. Corporatism and Political Theory. Oxford: Blackwell.

Davidson, A. 1989. Two Models of Welfare: The Origins and Development of the Welfare
State in Sweden and New Zealand, 1888-1988. Stockholm: Almqvist&Wiksell International.

Dixon. J. & Scheurell, R. P. 1989. Social Welfare in Developed Market Countries. London:
Routledge.

Elmer. Å. 1975. Svensk socialpolitik. Lund: LiberLåromedel.

Esping-Andersen, G. 1985. Politics against Markets. Princeton: Princeton University Press.

Flora, P. 1986a. -Introduction", in Growth to Limits. Vol. 1. Berlin: De Gruyter.

Furniss. N. &Tilton. T. 1977. The Case for the Welfare State. Bloomington: Indiana University
Press.

Glazer. N. 1986. Welfare and "Welfare" in America", in Rose, R. & Shiratori, R., eds., The
Welfare State East and West. New York: Oxford University Press.

Gordon, M. S. 1988. Social Security Policies in Industrialized Countries. Cambridge: Cambridge
Press.

Hadenius. A. 1987. 'Vålfårdspolitikens dimensioner', in Festskrift till professor skvtteanus
Carl Arvid Hessler, Acta Universitatis Upsaliensis, no. 103.

Heckscher, G. 1984. The Welfare State and Beyond; Success and Problems in Scandinavia.
Minneapolis: University of Minnesota Press.

Heidenheimer. A., Heclo, H. & Adams, C. T. 1975. 1983. 1990. Comparative Public Policy.
New York: St Martin's Press.

Info. 1988. A Short Survey of Social Security. The Hague: Ministerie van Sociale Zaken en
Werkgelegenheid.

lnequality in Sweden, Living Conditions 1975-1985. 1988. Stockholm: National Central
Bureau of Statistics.

James, E. 1987. The Nonprofit Sector in Comparative Perspective', in Powell, W. W., ed.,
The Nonprofit Sector: A Research Handbook. New Haven: Yale University Press.

Johansen, L. N. 1986. 'Welfare State Regression in Scandinavia?' in Øyen, E.,ed., Comparing
Welfare States and their Futures. Aldershot: Gower.

Johnson, N. 1987. The Welfare State in Transition. The Theory and Practice of Welfare
Pluralism. Brighton: Wheatsheaf Books.

Johnson, N. 1989. 'Problems for the Mixed Economy of Welfare', paper presented at the
ECPR workshop on Needs, Contributions and Welfare, Paris.

Kamerman. S. & Kahn, A. J. 1981. Child fare, Family Benefils and Working Parents: A
Studv in Comparative Policy. New York: Columbia University Press.

Karleby, N. 1926. Socialismen infor verkligheten. Stockholm: Tiden.

Katzenstein, P. J. 1985. Small States in World Markets. Ithaca, NY: Cornell University Press.

Kohl, J. 1982. 'Trends and Problems in Postwar Public Expenditures in Western Europe and
North America', in Flora, P. & Heidenheimer, A., eds., The Development of Welfare States
in Europe and America. New Brunswick: Transaction Books.

Kommunernus finanser 1981. 1983. Stockholm: National Central Bureau of Statistics.

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

Kuhnle. S. & Selle. P. 1989. 'Integrated Dependence or Separate Autonomy? Relations
between Government and Voluntary Organizations in Norway', paper presented at the
ECPR Workshop on Needs, Contributions and Welfare. Paris.

Lybeck. J. A. 1984. Hur stor år den offentliga sektorn? Malmo: LiberForlag.

Marklund. S. & Svallfors. S. 1987. Dual Welfare - Segmentation and Work Enforcement in the
Swedish Welfare System'. Research Reports from the Department of Sociology, University of
Umeå 94.

Mishra. R. 1977. Soden- and Social Policy. London: Macmillan.

Nasenius. J. & Veit-Wilson. J. 1985. 'Social Policy in a Cold Climate: Sweden in the Eighties'.
in Jones. C. & Brenton. M.. eds.. The Year Book of Social Policy in Britain 1984-5.
London: Routledge & Kegan Paul.

OECD. 1985. Social Expenditure 1960-1990. Paris: OECD.

OECD. 1989. Economic Survey of the Netherlands 1988/89. Paris: OECD.

O'Higgins. M. 1988. 'The Allocation of Public Resources to Children and the Elderly in
OECD Countries". in Palmer. J. Letal.. TheVulnerable. Washington DC: Urban Institute.

Parry. R. 1986. 'United Kingdom". in Flora. P.. Growth to Limits. Vol. 2. Berlin: De Gruyter.

Parry. R. 1989. 'Making the Connection between Political Success. the Satisfaction of Needs
and the Provision of Services", paper presented at the ECPR workshop on Needs. Contributions
Welfare. Paris.

Pettersen, P. A. 1985. Valfar J, vårderingar, vågval. Borås: LO.

Pierson. P. 1990. Taking the Jewel from Labour's Crown9 Thatcher's Assault on the British
Welfare State", paper presented at the Seventh International Conference of Europeanists.
Washington, DC. 23-25 March 1990.

Rainwater. L.. Rein. M. & Schwartz, J. 1986. Income Packaging in the Welfare State. Oxford:
Clarendon Press.

Rein. M. & Rainwater. L. eds.. 1985. Public Private Interplav in Social Protection. Armonk.
NY: Sharpe.

Roebroek. J. M. 1989. 'Netherlands', in Dixon. J. & Scheurell. R. P.. eds.. Social Welfare in
Developed Market Countries. London: Routledge.

Roebroek. J. & Berben, T. 1987. 'Netherlands', in Flora, P. Growth to Limits. Vol. 4. Berlin:
De Gruyter.

Rose, R. 1986. 'Common Goals but Different Roles: The State's Contribution to the Welfare
Mix'. in Rose. R. & Shiratori. R.. eds.. The Welfare Suite East and West. New York: Oxford
University Press.

Ruggie. M. 1984. The State and Working Women. Princeton: Princeton University Press.

Sainsbury. D. 1989. Welfare State Variations. Women and Equality', paper presented at the
ECPR workshop on Equality Principles and Gender Politics. Paris.

Schmitter. P. C. 1974. 'Still a Century of Corporatism. Review of Politics 36, 85-131.

Schmitter, P. C. 1981. 'Interest Intermediation and Regime Governability in Contemporary
Western Europe and North America, in Berger, S., ed., Organizing lnterests in Western
Europe. Cambridge: Cambridge University Press.

Siim, B. 1987. "AComparative Perspective onthe Organizationof Care Work in Denmark and
Britain', paper presented at the ECPR workshop on Women and Democratic Citizenship,
Amsterdam.

Smeeding, T. et al. 1988. 'Patterns of Income and Poverty: The Economic Status of Children
and the Elderly in Eight Countries', in Palmer, J. L. et al., The Vulnerable. Washington
DC: Urban Institute.

Social Report on Inequality in Sweden. 1981. Stockholm: National Central Bureau of Statistics.

Social Security Statistics 1981. London: Her Majesty's Stationery Office.

Statistical Abstract of the United States. 1988. 1989. Washington, DC: Bureau of the Census.

Statistisk årsbok 1982/83. 1982. Stockholm: National Central Bureau of Statistics.

Therborn, G. 1989. '"Pillarization" and "Popular Movements" Two Variants of Welfare State
Capitalism: The Netherlands and Sweden', in Castles, F. G., ed., The Comparative History
of Public Policy. Cambridge: Polity Press.

Tingsten, H. 1973 (1941). The Swedish Social Democrats. Totowa, NJ: Bedminster Press.

Trousdell, E., et al. 1986. 'Bibliographical Essay: Organizing Principles and Issues in the
Welfare State Literature", in Furniss, N., eds.. Futures for the Welfare State. Bloomington:
Indiana University Press.

Uusitalo, H. 1984. 'Comparative Research on the Determinants of Welfare States', European
Journal of Political Research 12, 403-422.

van Amelsvoort, A. A. M. 1984. 'The Netherlands: Minimum Wage and Unemployment
Policies', in Walker, R. et al., eds., Responses to Poverty: Lessons from Europe. London:
Heinemann Educational Books.

Wallerstein. M. 1989. 'Union Organization in Advanced Industrial Democracies", American
Political Science Review 83, 481-501.

Weir, M. & Skocpol, T. 1985. 'State Structures and the Possibilitiesof "Keynesian" Responses to the Great Depression in Sweden, Britain and the United States', in Evans, P., Rueschemeyer, D. & Skocpol, T., eds., Bringing the State Back In. Cambridge: Cambridge University Press.

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

Wilensky, H. 1982. 'Leftism. Catholicism and Democratic Corporatism: The Role of Parties in Recent Welfare State Development' in Flora, P. & Heidenheimer, H., eds., The Development of Welfare States in Europe and North America. New Brunswick, NJ: Transaction

Wilensky, H. & Lebeaux, C. N. 1958. Industrial Society and Social Welfare. New York: The
Free Press.

Zarf, W. 1986. Development, Structure and Prospects of the German Social State', in Rose,
R. & Shiratori, R., eds., The Welfare State East and West. New York: Oxford University
Press.