Resumé

Claims have been raised from different perspectives that the welfare state suffers from eroding public support. From an examination of public attitudes towards taxation in Sweden by utilising data spanning the period 1981-1997, it is concluded that no long-term trend of increased tax discontent can be distinguished. There are, however, some indications that discontent taxation has increased during the most recent years. Social conflicts over the preferred distribution and levels of taxes vary in intensity over time, but are foremost structured by class, income, and, to a lesser extent, sector of employment. It is argued that the observed attitude patterns can to some extent be understood as responses to changes in tax and welfare policy.

Introduction

The recent trends of welfare state retrenchment that have occurred in Sweden and many other Western countries have revitalised the longstanding 'crisis of the welfare state' debate. Although the topics and explanations addressed by researchers show substantial variation, there are nevertheless some common targets for criticism, such as the tax system and generous welfare policies. Issues that have received much attention focus on the effects that advanced welfare statism is supposed to have on macroeconomic performance public support for the welfare state.

The prominent levels of taxes collected by the state, the critics have argued, are harmful to economic competitiveness in an increasingly global economy, since they distort market relations and weaken work motivation and entrepreneurship.' In a similar vein, claims have been put forward that heavy taxation has eroded support for the welfare state among the public at large.

The first aim of this article is to analyse whether a trend of increasing
discontent with taxation has occurred in Sweden, a country that by many is
regarded to be the epitome of a high-tax society. It has. moreover, been

---

* Jonas Edlund, Department of Sociology. University of Umeå. 901 87 Umeå. Sweden. E-mail: jonas._edlund(« soc.umu.se

argued that the specific characteristics of the Swedish welfare state may also facilitate a shift from class-based political conflicts over welfare and tax policies to other 'new' dimensions of cleavage, such as gender and sector of employment. The second aim of this paper pertains to these and other argumentsconcerned dominant dimensions of cleavage in contemporary societies, for it is to analyse along which dimensions conflicts over taxation are structured and whether conflict patterns tend to change over time.

The paper is organised as follows. The next section summarises the main arguments on which the declining welfare state support thesis is founded. Previous research on Sweden is also presented. This is followed by a description recent developments in Swedish welfare and tax policy. In the fourth section, aggregate levels of tax discontent and attitudinal dimensions are analysed. Thereafter, arguments suggesting different types of social conflict are introduced and empirically examined. In the final section, the main results are reviewed and discussed.

Declining Welfare State Support? Arguments and Previous Research on Sweden

Government responses to the 'new' economic situation in the wake of the oil crises gave birth to the first wave of the 'crisis of the welfare state' debate. Seen in retrospect, one of the most influential comments appears to be Wilensky's (1976) claim that high reliance on direct 'visible' taxes, among which the personal income tax is the most important revenue source, feeds tax resistance and undermines public welfare state support (Coughlin 1980).

Agreeing with Wilensky that the composition of taxes may evoke tax resistance, Hibbs and Madsen (1981) additionally suggest that welfare state opposition is more likely to shed its blossoms in service-intensive welfare states than in cash transfer welfare states. According to them, the electoral mobilisation of 'anti-welfare state . . . political movements' is the 'sharpest challenge to the high-taxing, high-spending state,' and they identify visible direct taxes and public services as particularly vulnerable to critique and comparatively easy targets in mobilising anti-tax sentiments into electoral support (Hibbs & Madsen 1981, 414, 434). Relying on tentative evidence from various sources, they state that 'there is little doubt that public resistance the growth of taxation and state expenditure has increased in many Western industrial societies during the last decade' (Hibbs & Madsen 1981,413).

More recently, the crisis of the welfare state has been conceptualised as
a crisis of public legitimacy for the principles on which the welfare state
rests. Commenting on the wave of regressive tax reforms which swept over

the Western world during the 1980s, Musgrave (1990, 317) suggests that
'the major factor [behind tax reforms] has been a change in political climate
toward a less egalitarian view of distributive justice.'

Inglehart (1990, 257) reaches a similar conclusion concerning public support welfare state principles, although from a somewhat different perspective. is concluded that the combination of economic prosperity and ambitious welfare policies has had important implications for changing political priorities among the public. Welfare policies, it is argued, 'have largely solved the problems they are capable of solving most readily - and have thereby reduced the demand for more of the same.' Moreover, 'the growth of the welfare state has begun to reach its limits . . . there is virtually no room for further expansion; taxation becomes massive, and the majority of the public feels the burden' (Inglehart 1990, 10).

Although the theoretical foundations upon which the above arguments are built differ, a common assumption is that the welfare state suffers from eroding public support. Focusing on Sweden, to what extent does previous research support such a claim? While some of the above authors might undoubtedly hesitate to answer yes, most of them would certainly agree that it is likely that discontent has increased. One major reason for the uncertainty the data on which they have relied, since these data suffer from various problems. While Coughlin (1980) and Hibbs and Madsen (1981) are forced to rely on scattered data from national surveys with obvious difficulties comparability, Inglehart (1990) relies on items pertaining not to welfare and tax policies per se, but rather to government intervention in the economy. Musgrave (1990) does not present any data.

Moreover, taking into account other findings, empirical support appears to be limited for the claims of increased tax discontent in Sweden. Comparing attitudes in Sweden between 1969 and 1981, it has been shown that citizens became more discontented with marginal tax rates while other attitudes related to taxes showed decreased levels of discontent (Hadenius 1986, 23-24). Research covering more recent time periods demonstrates an equally disappointing pattern for the welfare backlash thesis. In a more comprehensive analysis using time-series data for 1981, 1986, and 1992, Svallfors (1995) finds no indication of decreased support for welfare policies in terms of service delivery, financing, and spending. However, as Hibbs and Madsen (1981, 434) point out, symptoms of welfare state backlash probably have more to do with resentment of heavy taxation than of social security.

Thus, to detect whether any trend of eroding support for the welfare state has occurred in Sweden, popular attitudes towards taxation appear to be a key area to analyse. Unfortunately, for more recent time periods, little effort has been made to examine to what extent, if any, public discontent with taxation has risen. By utilising data from six national surveys conductedin

ductedinSweden between 1981 and 1997, the first aim of this article, as
pointed out above, is to fill this paucity of research by analysing whether a
trend of increasing discontent with taxation can be distinguished.

Few would deny that Sweden is a suitable case. It has been repeatedly demonstrated in comparative studies of welfare states that the configuration and effects of tax and welfare policies vary substantially between countries. Irrespective of which measurements are used for classification, Sweden ranks among the most advanced and redistributive welfare states (Esping- Andersen 1990; Fritzell 1991; Korpi & Palme 1998; McFate et al. 1995; OECD 1990; Stephens 1996).

There are some characteristics among the traits distinguishing the Swedish welfare state which, following the above suggestions, may be of particular importance for encouraging public discontent with taxation. First, the total tax revenue and the personal income taxes collected by government are substantial and located in the top bracket among OECD countries. Secondly, in comparative perspective the tax system appears to be highly progressive.² Thirdly, the Swedish welfare state is service intensive and mainly organised within the public sector.

When examining attitudes to taxation, at least two recent developments of the Swedish welfare state need to be considered. First, the tax system during the studied time period has undergone substantial changes. Second, Sweden faced an economic recession during the 19905. Rising unemployment and decreasing tax revenues put the welfare state under severe pressure. The government attempted to solve the financial problems of the state through consecutive reforms, which consisted of a combination of cutbacks in replacement levels, reduced social services, and increased taxes on income and social security. In order to understand how attitudes advance times of welfare state retrenchment, a brief description of the changes in tax legislation and welfare programmes is needed. This is the purpose of the next section.

Recent Tax System and Welfare Developments

In the late 1970s the need for tax reform was broadly endorsed within the conservative coalition government. Due to internal strains, however, the governing parties were unable to settle an agreement to reform the tax system. At the beginning of the 1980s, negotiations across party lines resulted a tax deal, subsequently called the 'Wonderful Night,' between the Centre Party, the Liberal Party, and the Social Democratic Party (Hadenius 1981). At the centre of reform was the steep statutory income tax scale running from 35 to 85 percent (Wahlund 1991).

Among those advocating reform, it was argued that the design of the

income tax scale had negative effects on tax compliance and resource allocationin of work incentives, investment, and savings. Furthermore, due to generous tax deductions on interest payments, the tax system encouragedtax and extensive consumption - factors that undermined the redistributive profile of the tax system, since high income earners could lower their taxable income substantially (Åberg et al. 1987; Wahlund 1991, 69f; Steinmo 1993, 183ff). Another problem, equally important to manage, was the consequences of bracket creeping, which affected an increasing share of citizens. Before the reform, about half of those in the labour force exceeded the 50 percent marginal tax rate (SCB 1994, 43).

The tax reform attempted to mitigate these effects by cutting marginal tax rates and limiting tax deductions. The main objective was to ensure that in 1985, when the reform was fully implemented, 80 percent of those working and 90 percent of tax payers would have a marginal tax rate below 50 percent. A thorough analysis of Wahlund (1991, 74) suggests that this objective was by and large fulfilled.

While the reform was the first successful attempt in years to bring down marginal tax rates, critical arguments concerning the tax system did not soften. The lowered marginal tax rates, which mainly affected the upper income brackets (Wahlund 1991, 78), created severe tensions within the labour movement, mainly between the Social Democratic Party, which had returned to power in 1982, and the Swedish Confederation of Trade Unions (LO). The left-wing critique, foremost articulated by LO, argued that the tax burden had been redistributed downwards and the measures introduced to circumscribe tax planning were insufficient (Feldt 1991). From the right, on the other hand, the tax system was attacked for its high revenue levels and the high marginal tax rates imposed on personal incomes.³

These critical points were by and large shared by the Social Democratic Ministry of Finance. The Finance Minister Kjell-Olof Feldt and his associates the tax system in articles and speeches and pleaded the need for further reform (Steinmo 1993, 186; Feldt 1991, 379). During the 1987 party congress, the Social Democratic elite received a mandate to examine and suggest appropriate changes in the tax system. The guidelines for tax reform offered by the appointed public official commissions (SOU 1989a: 1989b; 1989c) were indeed comprehensive and 'more far reaching than the mandate given by the party congress' (Feldt 1991, 380).

The government, firmly determined to implement the suggested changes, negotiated with LO and both agreed that the current tax system contributed to a series of negative effects. However, the proposed large cuts in income tax caused disagreement. LO argued that such large cuts would increase economic inequalities and jeopardise the financing of welfare programmes. The government argued that the sharpened capital tax. the broadened goods and services tax, and the removal of a number of tax deductions

would compensate for the lowered revenues of income taxes. Analyses of the reform's relative effects on disposable income distribution among differentincome showed that these would be relatively small.⁴ After some deliberation LO finally accepted the tax reform.

In order to carry the proposal through the Riksdag the Social Democrats invited the other political parties to negotiations. The financing of the reform major disagreements between the parties, but in late 1989, after intense deliberation in which the Social Democrats were forced to make a few concessions, the Liberals and the Social Democrats signed a tax reform deal.

Given the remarkable pace of implementing such a far-reaching reform and the avoidance of internal party debate before the tax deal was signed, the Social Democratic leadership was aware that getting support from rank-and-file members would 'demand exceptional pedagogical and agitatorial .. . since the new tax system at first glance could be perceived as a violation of every social democratic idea of how justice is done' (Feldt 1991, 385; cf. Åsard & Bennett 1992; Elvander 1994). In order to justify the reform from a left perspective, the Social Democrats, accompanied by economic experts, stressed that the reform should be viewed in its totality, and not only as a reform of marginal taxes. It was emphasised that the current tax system did not serve the national economy efficiently. The progressive and high taxes on income, the unequal treatment of different incomes, and the misdirected effects of tax deductions encouraged neither work nor saving, but they did encourage tax evasion, tax planning, and other nonproductive activities. The new tax system with simple and uniform rules would critically remove incentives for tax planning and evasion, while strongly promoting incentives to work (denoted 'dynamic effects' in public debate), save, and invest. The redistributive effects of the tax and transfer system would largely prevail, as many loopholes were closed while housing allowances and child subsidies were increased (Sodersten 1991; Finansdepartementet

The main changes in the tax system, which came into full force in 1991, included broadened, sharpened, and uniform taxes on goods and services (25 percent) and capital (30 percent), reduced tax rates on corporate income, decreased possibilities for tax deductions, and lowered income tax rates. The four-step marginal tax rate, ranging from 35 to 72 percent, was replaced with a two-bracket structure of 30 and 50 percent (Finansdepartementet 1990). The new tax system would entail that nearly 85 percent of income earners would have a marginal tax rate of 30 percent and no one would have a marginal tax rate exceeding 50 percent (Agell et al. 1995,10). Compared with the intentions inherent in the 'Wonderful Night' agreement, it is understandable this reform earned the title 'The Tax Reform of the Century.'

Simultaneously with the later stages of the tax reform implementation

process, the government introduced several measures that were designed to cool down a booming economy, contribute to a balanced budget, and decreaseinflation 1994). While some of these measures may be consideredsuccessful the government's point of view, it is nevertheless clear that in several respects the economic situation remained problematic. In 1991 general elections were held and the Social Democrats lost power with a large margin. While some argue that the 'big defeat of the Social Democrats . . . has a significant relationship with an ideological shift among the voters - from left opinions to right opinions, from socialism to market economy' (Gilljam & Holmberg 1993, 141), others emphasise that the economicpolicies by the Social Democratic government, such as the fight against inflation, interventions in wage negotiations, deregulation of financial markets, and the reformation of the tax system, gradually disillusionedtraditional voters from the party, since the government appearedto tremendous difficulties controlling the national economy (Korpi 1993; Åberg 1994).

Whether or not the tax reform contributed to the electoral defeat, empirical suggests that the Social Democratic designed reform was perceived a marginal tax reform and had its strongest supporters among professionals, managers, high income earners, and those affiliated with conservative (Edlund 1999b).⁵ It is equally true that the initiated policies did not sufficiently solve the problems of the Swedish economy. However, coincidental with the four-party conservative coalition government, signs of a brewing economic recession increased sharply and culminated in autumn 1992. Two 'crisis packages,' including cutbacks in social spending, were proposed by the government in cooperation with the Social Democrats and carried through the Riksdag (Stephens 1996).

The deep recession, accompanied by rising unemployment and decreasing revenues from an eroding tax base, put the welfare state under severe pressure. 1994, the Social Democrats returned to power. Bringing the public finances into balance topped the agenda. Compared with the cautious measures by the conservative government, the Social Democrats implemented changes in welfare programmes, such as cutbacks in replacement levels, stricter qualifying conditions, and reduced social services, all of which aimed at bringing down costs (Palme & Wennemo 1998). Concerning taxation, the most significant changes in fiscal policy include an additional 5 percent extra tax on income which was added on top of the marginal income tax rate and increased social security contributions paid by employees, a revenue source that had previously been virtually nonexistent Sweden.

In response to the implemented cutbacks and the financial crisis of the
state, some critical arguments tracked the tax reform as one of the roots
behind the worsened situation of the Swedish welfare state. The absence of

dynamic effects in the labour market along with decreasing income tax revenues, which had not been compensated for by other revenue sources, indicated, the critics argued, that the tax reform relied on inaccurate assumptionsand, was not fully financed. The tax reform's contribution to the trend of rising inequalities in disposable income distribution was another issue that received much attention.⁶

Attitudinal Trends and Patterns towards Taxation, 1981-1997

To summarise welfare state developments during the 1980s in very broad terms, the social policy changes enacted may be depicted as a 'period of continued marginal expansion of social security entitlements' (Palme & Wennemo 1998, 15). Concerning taxation, one can observe a trend of decreasing tax rates, declining tax expenditures, and expanding revenue sources. Unemployment rates were kept at a low level. The 19905, particularly after the 1994 election, may be described as a period of increased taxes and welfare policy retrenchment. Unemployment rates increased sharply during 1992-1993 and still remain at a high level.

In the realm of politics on the elite level, these welfare state developments have been bordered by political conflicts between the Social Democratic Party and the conservative parties in general and the right-wing Moderate Party in particular. Despite these conflicts, cross-party agreements for tax reform have been accomplished. However, as previously noted, these agreements other economic policies adopted by the Social Democratic Party have at times contributed to harsh friction within the labour movement.

Acknowledging these policy changes and the political context within which they were settled, we now turn to an analysis of how citizens' attitudes towards taxation have evolved over time. The data used in this paper come from six nationally representative surveys conducted in Sweden between 1981 and 1997. A brief description of the data is available in the Appendix.

In Table 1, various attitudes concerning taxation are displayed. As indicated the questions dealing with the general level of taxes, the marginal tax rates, and whether the distribution of taxes is fair, those expressing discontent clearly outnumber those who signify satisfaction with taxes. This is indicated by both the percentage distributions and the arithmetic means.⁷ The evidence for massive public discontent with taxes is, however, less than clearcut. With two exceptions, it can be observed that a majority of citizens agree that taxes are not too high considering the public benefits provided by the state, and, furthermore, that those in favour of replacing taxes with user fees are not in a majority.

While it is clear that the observed attitudes do not shift greatly over time, some differences between the years may still be discerned. Between 1981 and 1992, dissatisfaction with taxes seems to have decreased. Citizens became less critical to the fairness of the tax distribution and less willing to agree that the general tax level and the marginal tax rates were too high.

Attitudes appear to have been quite stable between 1992 and 1996. While discontent with tax levels and marginal tax rates decreased slightly, public views on the fairness of taxes and the tradeoff between taxes and benefits tended to go in the opposite direction. This latter pattern continued in 1997. and was accompanied by increasing dissatisfaction with marginal tax rates and the general level of taxes.

Continuing with an analysis of public perceptions and preferences on
the level of taxes paid in different income strata during the 19905, it is

apparent, as shown in Table 2, that most citizens were supportive of reduced taxes for low income earners, while the taxes paid by middle income earners were perceived either as legitimate or as too high. A more distinguished variation in response patterns can be observed when it comes to taxes imposed on high income earners. The belief that taxes are too low was fairly widespread, but there were sizeable proportions of respondents viewing the level of taxes either as appropriate or as too high.

Comparing marginal frequencies over time, the main impression from the table is stability. However, there was an increased tendency towards disapproval of the taxes paid by low income and middle income earners, albeit slightly, between 1996 and 1997. Furthermore, the upward trend of public willingness to increase taxes on high income earners appears to weaken in 1997.

Since these questions deal with public desires relating to both the levels of taxes and the distribution of taxes between different income groups, it is possible to construct two different tax measures, which will be used in the forthcoming analyses. The first measure is a simple additive index and attempts to measure preferences towards the level of taxes. The higher the value the more the respondent is in favour of increased taxes.⁸ The second

measure intends to capture demands for increased tax progression, which is denned as the distance between the variable values for taxes on high incomesand incomes, respectively.⁹ The higher the value the more sympathetic the respondent is towards a more progressive distribution of taxes. Both indices vary between —1 and 1.

How, then, are these indices and the other tax indicators associated with each other? Can the various items be considered as parts of one single attitudinal dimension or do they represent different dimensions of attitudes towards taxation? Furthermore, is it possible to observe any changes in correlation patterns over time? One way to answer the above questions is to examine the correlation matrix, which is displayed in Table 3.

Beginning with items 1 to 5 in the table, an inspection of the coefficients suggests that the correlation patterns between different items in most instances quite stable over time.¹⁰ Secondly, we can observe that attitudes towards the general level of taxes (item 1) correlate rather strongly with attitudes towards the marginal tax rates (item 2) and the tradeoff between taxes and benefits (item 3). While the associations between the latter two items are somewhat less pronounced, it is suggested that these three questions measure some form of general tax discontent (Hadenius 1986). The additive index construct appears in row 6.¹¹ It is furthermore decided to keep item 5 as a single indicator of tax distribution fairness, due to its either weak or irregular relationships with other items.

How do these two measures correlate with the tax progression (row 7) and tax level (row 8) indices? As shown, there are some indications that attitudes taxation are multidimensional. While general tax discontent and attitudes concerning the level of taxes are quite strongly related to each other, correlations between tax discontent and the other two indices are either weak (tax progression) or vary in magnitude over time (tax distribution

It may also be noted that attitudes concerning the fairness of the distribution taxes are foremost correlated with attitudes towards progressive taxation. The latter seems also to be associated with perceptions concerning the level of taxes. Those in favour of increased (decreased) tax progression tend both to perceive the distribution of taxes as unfair (fair) and to be less (more) interested in lowered taxes.

Except for the relationships between general tax discontent and the tax level index, correlations between the four tax measures vary between medium strong to negligible. These patterns of correlation suggest that attitudes towards are to a certain degree intertwined, but hardly as tightly as a unidimensional pattern would prescribe. In the forthcoming analysis we will distinguish between general tax discontent, tax distribution fairness, tax progression, and attitudes concerning the level of taxes.

At this point, is it possible to draw any general conclusions regarding

the ways public attitudes towards taxation have evolved on the aggregate level during the past 15 years? Admittedly, interpretations depend to some extent on expectations. Among those expecting to find well distinguished and systematic changes over time, the observed differences might be regarded as small and, thus, pointing to the stability of attitudes. Still others may anticipate the observed patterns as a pretext of attitude change.

Irrespective of whether stability or change is emphasised, it is difficult to maintain that a long-term trend of eroding support for taxation has occurred. considering the implemented tax and welfare policy changes, there are reasons to suspect that the attitudinal variations should not be dismissed as trendless fluctuations. Seen in this perspective, it can be argued that the tax reforms during the 1980s contributed to decreasing tax discontent. a similar vein, the indications during the 1990s of gradually increasing discontent with welfare benefits and the distribution of taxes may be a consequence of the enacted tax policies and cutbacks in welfare programmes.

These results evoke some questions about the relationships between attitudes towards taxation and structural conditions. First, along which dimensions of cleavage are attitudes towards taxation structured? Second, it may be of interest to analyse whether the observed attitudinal changes are related to changes in preferences among certain social groups. More specifically, tax reforms contributed to weakened or reinforced social divisions? Is it possible to link the trend of augmented tax discontent during the 19905, particularly between 1996 and 1997, to specific social groups or is it distributed randomly? In the next section we will attempt to answer these questions.

Social Divisions and Attitudes to Taxation, 1981-1997

One of the major objectives of the welfare state is to provide social protectionby various risks connected with unemployment, ageing, childcare, and illness. Viewing the labour market as the main system of stratification, it has been argued that the unequal distribution of socioeconomicresources risks is closely linked to occupational positions (Korpi 1983; Esping-Andersen 1985). Since those with more fragile positionsin labour market have fewer resources and are exposed to higher risks, they are supposed to support strategies that redistribute market outcomes.From perspective, then, the welfare state is conceived as a major arena for class conflicts, and its political support is foremost located among those belonging to the lower strata in terms of resources (Matheson 1993). However, the relevance of class in contemporary society has recently been widely debated (Lee & Turner 1996). While the issues addressed differ, most

of the critical objections, at least when it comes to themes concerned with
class and politics, can be subsumed under two headings.

According to the first set of arguments, it is generally agreed upon that the labour market is a major system of stratification, but it is suggested that the welfare state has increasingly become a prominent force in that process. Within feminist discourse it is emphasised that the structure and scope of social provisions provided by the state affect gender relations and work opportunities and structure gender-specific interests (Orloff 1993; Daly 1994). Another aspect is the changing occupational structure, which is intimately connected with the establishment of the welfare state. In Sweden these changes have led to a heavily gender-segregated labour market (Esping-Andersen et al. 1993). As women are seen as more dependent on the state in terms of social services, benefits and employment - a pattern that appears to be particularly salient in Scandinavia (Siaroff 1994) - it is suggested that they are more supportive of the welfare state than are men (Hernes 1987; Borchorst & Siim 1987).

In a comparative perspective it has been argued that substantive gender and sector conflicts over welfare policies are a likely scenario in Sweden. These are conflicts that may even supersede class divisions, since, it has been argued, universalistic social services, income-related policies, and commitment full employment benefit both the working class and the middle class (Esping-Andersen 1990). The potential source of sector conflicts is found in the public social service and taxation nexus (Hibbs & Madsen 1981). An ambitious public service sector which requires high taxes may be subjected to ideologically mounted attacks by political movements in matters of taxation and wage demands, which result in cleavages between tax payers in the male-dominated private sector and the, mostly female, tax-receiving employees in the public sector.

The main point of the second set of arguments is that the potential of class behind the creation of identities and interests has gradually lost its power. Due to forces inherent in processes of post-industrialisation and globalisation, contemporary social divisions have increasingly shifted away from class towards cleavages manifested in patterns of consumption and value commitments. This view, represented by Pakulski and Waters' (1996) description of divisions in present 'life after class' societies, argues that identities and interests tend to be reflexively self-composed, rather than structurally determined by factors such as class. The rise of a 'fluid matrix of fragile formations' has led to more complex and fragmented patterns of 'multiple status cleavages' (Pakulski & Waters 1996, 155-8). If these suggestions correct, then we should expect a gradual decline of associations between attitudes to taxation and structural factors in general and class in particular between 1981 and 1997.

Pointing at the 'silent revolution' of slowly emerging postmaterialist

value patterns, Inglehart (1990; 1997) suggests that younger generations tend to prioritise different political issues from those prioritised by older cohorts, due to their different formative experiences concerning material well-being. In the realm of politics, it is claimed that questions of redistributionand will gradually lose their significance, resulting in a diminishing relevance of the traditional left-right axis and in declining ties between political party preferences and welfare state issues (Inglehart 1990, 248-334). While Inglehart is explicit on this point, it is less clear whether levels of support for taxation are supposed to differ between materialists and postmaterialists, since it is emphasised that the materialist/postmaterialistthesis about priorities, not levels (Inglehart 1997, 108-30).

Still, if attitudes towards taxation are related to the materialist/postmaterialist it is possible that these relationships are expressed as rather distinctive differences between age cohorts. Furthermore, age may not only be relevant in terms of generational shifts in value preferences. Position in the life cycle can also affect views and preferences on taxes. From this perspective, one likely scenario is that those more dependent on the welfare state - the young and the elderly - are more in favour of taxes than are the middle-aged.

In the following we will attempt to evaluate these arguments empirically by analysing how class, education, household income, gender, age, sector of employment, and political choice are associated with attitudes towards taxation. that, however, the measurements of variables, the criteria used for classification, and the methods applied need some further description.

Class is measured with the Swedish socioeconomic classification of occupations: (SCB 1982). The SEI schema shows apparent similarities to the Erikson/Goldthorpe (EGP) classification of occupations, a class schema that has proven to be fruitful in a number of studies focused on class-related social issues, such as voting and social mobility (Erikson & Goldthorpe 1992; Nieuwbeerta 1996). Furthermore, the internal consistency of the different that the EGP concept of class is based upon has been empirically validated (Evans 1992). Six different classes are distinguished: unskilled workers, skilled workers, routine non-manuals, intermediate nonmanuals, non-manuals (professionals and managers), and the selfemployed ¹²

Two other factors that are rather strongly related to class location are education and income. Due to problems of standardisation, only three levels of education are distinguished. All respondents with compulsory schooling are coded as primary level of education. The secondary level is a rather heterogeneous category and comprises all sorts of post-compulsory schooling the university level, both academic and non-academic. The third and final category is university education.¹³

The other class-related factor, household income, is recoded into quintiles.In

tiles.Inorder to correct for differences in household size, the income registered for married and cohabiting couples' households has been divided by two. As additional information on household size is not available in some surveys, it was not possible to assess a more nne-grained weighting.

The final variable to be discussed is political party affiliation. Four categories are extracted. The label 'Left' is equivalent to supporters of the Social Democratic Party and the socialist Left Party. The category 'Centre' consists of those sympathising with either the Centre Party, the Liberal Party, the Environment Party, or the Christian Democratic Party. 'Right' refers to those affiliated with the Moderate Party. The final category is the remainder including those indicating no party affiliation or another party than those described above.¹⁴

It should be emphasised that the political party variable does not indicate a structural position, which the other independent variables do. If strong links exist between political party and attitudes to taxation, relationships between other variables rooted in structural conditions and tax attitudes may be distorted. We will therefore run separate analyses. In Tables 4 and 5, the relationships between the structural factors and each of the tax indices are displayed. Associations between political party preferences and the tax indices are shown in Table 6.

The method used in the following analysis is multiple classification analysis (MCA), a method well suited for multivariate analysis when independent variables are categorical and the dependent variable is at least on the interval scale level. Cautious interpretations of the data are, however, suggested. We should not take coefficients too literally but instead focus on broader patterns and relationships: a 'gestalt' as Scheuch (1989) puts it. The MCA output is restricted to coefficients controlled for the impact of other independent variables (beta and adjusted group deviations) (Andrews et al. 1973).'⁵

By comparing the grand mean over time in Table 4, columns 1-4, the Ucurved tax discontent pattern revealed in the previous section is also apparent here. However, while the composite measure of general tax discontent, between —1 and 1, indicates roughly similar levels of discontent 1981 and 1997, it may be acknowledged that discontent in 1981, compared with 1997, had more to do with tax rates than with the tradeoff between taxes and benefits.

The general impression from the table is that neither the class-related nor the alternative cleavage factors tend to be strongly associated with tax discontent. Beginning with the latter, it can be seen that gender and age differences are in most instances negligible. Sector divisions, on the other hand, appear to be somewhat more pronounced and in most cases they are significant. Both gender and sector associations tend to be stable over time.

The class-related divisions show more variation over time. Differences
between workers and higher non-manuals decreased between 1981 and

1992, the decade of tax reforms. During the following years, divisions seem to expand. A similar pattern can be observed for household income. Discontentincreased among those with the highest incomes. It should also be noted that education shows an unexpected pattern in that the universityeducated often less critical than those with lower education.

As noted above, the observed differences between different social groups were not particularly strong, but when they occurred, discontent with taxation foremost located among young citizens, high income earners, higher non-manuals, self-employed, and among those employed in the private sector.

Moving to columns 5-8, which deal with public perceptions of the fairness of taxes, there are some weak indications that the tax reforms implemented during the 1980s have contributed to altering attitude patterns. Between 1981 and 1992 it can be observed that workers became less willing to view the distribution of taxes as fair compared with those in non-manual occupations. This pattern, though somewhat less clear, continued during the 19905.

The other background variables show little variation between 1981 and 1992, with the possible exception of education, which shows a roughly similar but weaker pattern to class. However, the general impression from the table is that perceptions concerning the fairness of taxes were not clearly linked to specific groups, which is shown by the low proportions of variance explained.

Taking a closer look at the attitudinal developments during the implementation the 'Tax Reform of the Century' and the years following, we will examine whether the recent changes in Swedish welfare and tax policies have contributed to reinforce or weaken social conflicts over the levels and distribution of taxes.

The relationships shown in Table 5, columns 1-4, suggest that the observed in preferences for lowered taxes were comparatively less marked in 1991 and 1992 compared with more recent years. For class there is a tendency that higher non-manuals and the self-employed on the one hand and the lower classes on the other were increasingly taking different positions in this matter. Household income and sector of employment show a similar pattern.

When it comes to gender, education, and age, conflicting views on appropriate levels can generally not be distinguished. It should be noted, however, that in those cases age differences were somewhat more marked, the youngest cohort generally tending to be more positive about cutting taxes than other cohorts.

A trend of increasing demands for a more redistributive profile of the tax burden can be distinguished in the last area of tax attitudes subjected to analysis, which is depicted in columns 5-8. This pattern was, furthermore, accompanied by a parallel trend of sharpening social conflicts, which tend primarily to be structured by social class and class-related factors. The different on this issue taken by higher non-manuals and high income earners on the one hand and the lower social strata on the other hand were quite substantial.

Sector of employment was comparatively less associated with tax progression, it is apparent that public employees to an increasing extent tend to be more in favour of progressive taxes compared to those employed in the private sector.

Differences between women and men were generally small, but it can be
seen that women gradually turned more favourably towards progressive
taxation during the 19905.

Age differences were in all cases but one non-signincant. There seems to
be a consistent pattern, though, that the young were less supportive of
progressive taxation than were other age groups.

How well, then, do the above findings reflect any of the cleavage scenarios
suggested earlier? In terms of the various post-industrial claims, some receive
empirical support than others do. Let us examine each in turn.

Are there any signs indicating that the linkages between structural factors and attitudes are gradually dissolving? It is true that some of the observed relationships are weak. Whenever they do appear, however, they may in general be depicted either as parts of an overall over-time zero correlation pattern (e.g. gender and tax level index) or as indicating the beginning of an upward trend of strengthening associations (e.g. sector and tax level index). A general trend of eroding linkages cannot be traced, a finding that we will return to later when discussing the impact of class.

Furthermore, the observed gender differences are generally negligible. It is of course possible to argue that the more accentuated gender divisions that occurred in 1997 may indicate a recent shift of increased polarisation between men and women. Given the overall picture depicted by the data, such interpretations are afflicted with a large element of uncertainty, and it may be wise not to overemphasise the differences.

The suggested scenario of conflicts between private and public sector employees appears to fit the data better. Those employed in the private sector were comparatively more discontented with taxes. Although the observed relationships were not particularly strong, divergent views between and private sector employees concerning the distributive profile of the tax burden tended to be enhanced during the 19905.

The relationships between age and attitudes towards taxation vary both over time and across indices, which makes it difficult to draw any comprehensive While it perhaps is tempting to consider the observed patterns as trendless fluctuations, it can be noted that the young generally tended to stake out a more tax-critical position compared with other age groups. Given the heterogeneity of the patterns found, however, the data offer little guidance to determine whether this mirrors life cycle or generational

The impact of social class and class-related factors appears to vary over time. Between 1981 and 1991-1992 divisions tended to decrease. At the beginning of the 19905, simultaneously with the launching of the 'Tax Reform of the Century,' the observed differences were comparatively small regarding general tax discontent and the levels of taxes. One exception from this pattern was the tax-critical position taken by the self-employed.

Divisions are clearer when it comes to issues concerning the distribution
of taxes. Strata with fewer resources appear to be less positive to the fairness
of tax distribution and seem more inclined to support increased tax progressioncompared

gressioncomparedwith higher social strata. In the most recent years of observation, there were some indications that the magnitude of class conflictshas to increase, particularly over the levels and the progressivityof

These findings suggest that the overall impact of class across the different attitudinal dimensions is quite prominent. Even though some variation over time can be registered, empirical evidence supporting the 'death of class' thesis appears to be rather limited.

In Table 6, the relationships between political party affiliation - controllingfor
impact of structural factors - and the four tax indices are
displayed. It is clear that voters on the left and on the right, respectively,

have quite divergent preferences and perceptions of taxes. This is particularlyevident it comes to general tax discontent and appropriate tax levels. The similarities between these two patterns should not be surprising, since the correlation between general tax discontent and tax level preferencesis significant, as shown in Table 3. Furthermore, attitudinal differences regarding progressive taxation are somewhat less distinct but still reminiscent of the two depicted above.

Perceptions of tax distributive fairness, on the other hand, do not follow party lines clearly and the strength of relationships varies considerably over the years. Apart from those expressing no party preference, a group that regularly tend to view taxes as more unfair compared with other groups, divisions between those on the left and those on the right developed quite peculiarly during the studied time period. In the aftermath of the tax reform, those on the right were more positive than those on the left. During the 1990s left-wing supporters gradually turned more favourable while those affiliated with the right-wing party become more dissatisfied.

However, in contrast to expectations of diminishing links between political and welfare state issues, we have to admit that most of the observed relationships appear to be rather strong and persistent. In some cases, divisions even tended to increase over time. Here, it might be suggested the roots behind the increased differences may be found within the realm of public welfare state discourse. As shown by Svallfors (1996, chapter 8), the financial crisis of the state during the 1990s contributed to a shift in welfare state discourse among organised interests towards a greater emphasis on the costs and taxes required for the welfare state. This shift is particularly evident for the right-wing Moderate Party, which increasingly tended to criticise the welfare state almost exclusively from a costs/tax perspective.¹⁶

Discussion and Conclusions

In the introduction to this paper, indications of a scenario of eroding public support for redistributive policies and taxation were presented. Analysing survey data spanning the period 1981-1997, a long-term trend of increased dissatisfaction with taxation cannot be detected. Between 1981 and 1991— 1992 discontent appears to have decreased. In a shorter-term perspective, however, there are some indications that tax discontent gradually increased during the 19905, particularly between 1996 and 1997.

These findings offer interpretations suggesting that the attitudinal developmentsrevealed data can to some extent be attributed to the changes in tax and welfare policy enacted during the studied time period. Seen in this perspective, the tax visibility thesis lends empirical support. Another expianationthat

pianationthatseems plausible is Korpi's (1980) thesis that differences in public tax preferences should not be treated in isolation from other aspects of the welfare state, particularly the public benefits provided by government.

It seems likely, then, that the increased discontent observed during the 1990s may to a considerable extent be explained by the combination of implemented tax increases and cutbacks in welfare programmes. To put it crudely, citizens perceive that the public goods provided offer 'less value for money' today than they did yesterday.

When it comes to group differences, most of the various post-industrial
theses pointing to different kinds of plausible conflict patterns in contemporary
receive limited support.

First, it is difficult to find any substantial evidence suggesting a trend of
declining relationships between attitudes to taxation and structural determinants
general and class in particular.

Second, although there are some indications that the young tend to be
more negative towards taxes than other age groups, the impact of age tends
to vary in a rather non-systematic fashion.

Third, according to suggestions of emerging new divisions, it can be noted that differences between men and women are generally small. It may be acknowledged, nevertheless, that differences concerning the distribution of taxes were fairly pronounced in 1997. Whether this indicates a temporary or a more enduring gender division may be answered in future analysis.

Relationships between sector of employment, the other new dimension of cleavage, and attitudes towards taxation appear to be more persistent. Those in the private sector are more discontented with taxes and are less willing to increase tax progression compared with public sector employees. It seems, moreover, that the politics of retrenchment have contributed to growing conflicts over the distribution of the taxes.

Without denying the impact of gender and sector of employment, it may be concluded that these conflicts generally tend to be of minor importance compared with those structured by social class and class-related factors. While class divisions appear to vary across attitudinal dimensions and over time, there is little evidence in favour of claims suggesting the diminishing relevance of class. The relevance of class politics is further underpinned by the substantive divisions found between those affiliated with left-wing parties those sympathising with the Moderate Party.

If one considers the impact of class in broader terms, it appears that the tax reforms contributed to recasting class conflicts from a domain of tax discontent to a tax redistributive domain. In the early 19905, striking similaritiesbetween with fewer resources and the more affluent could be observed concerning tax discontent and demands for lowered taxes in general. In the other domain, dissatisfaction with the distribution of taxes

and tax progression revealed more strongly defined differences where the
strongest proponents for redistribution were found among workers.

In a similar vein, the retrenchment era promoted substantive conflicts in both domains. Whether or not this is an indication of the eroding support for the welfare state among the higher level salaried employees noted by Svallfors (1999), it is clear that the tax increases and cutbacks in major welfare programmes have entailed a relatively stronger disapproval of progressive and rising desires for reduced taxes among higher nonmanuals, self-employed, and high income earners compared with other strata, especially skilled workers.

Finally, in the light of the above findings, is it justified to assume that the Swedish welfare state is facing a crisis of legitimacy? While 'crisis' may be too strong an expression, the downward trend of increased dissatisfaction the benefits received for taxes paid, accompanied by rising discontent various aspects of taxation between 1996 and 1997, may nevertheless be an ominous sign from a welfare state perspective.¹⁷

To some extent this comparatively sharp increase in discontent may be explained by inertia in public opinion. The effects of retrenchment politics have gradually increased in visibility among ordinary citizens in their everyday In this respect the impact of the anti-tax rhetoric, forcefully exercised by the political right, on public opinion should not be neglected, since such allegations may gain public appeal more easily under current circumstances.

Another factor that may be of relevance is the historically high unemployment in Sweden. From 1992-1993 until the present, open unemployment rates have oscillated steadily around the 8 percent mark, with another 4 percent accounted for by active labour market policies (Åberg 1998). Continuing mass unemployment in parallel with retrenchment politics may have negative consequences for public confidence in the capabilities of the welfare state and trust in the ambitions of political parties, since it has been argued that the earlier successful strategies for full employment were the key determinants behind the broad support for the Swedish welfare state (Korpi 1983; Esping-Andersen 1985; 1990).

If this trend continues, then it is possible that the Swedish welfare state may face severe challenges, from both apparent financial problems and more widespread public discontent with taxes - a process that, reinforced by right-wing anti-tax rhetoric, eventually may lead to more accentuated patterns of withdrawal of popular support for the welfare state.

On the other hand, there are reasons to assume that these suggestions are too pessimistic. Advocates of counter-arguments can take some comfort in the finding that the most dissatisfied citizens are not found among the strata that have probably suffered most during the economic recession. Moreover, support for the welfare state in terms of service delivery, financing,and

cing,andsocial spending remains strong during the 1990s and has in some instances slightly increased. Among the best off, however, welfare state supportappears have decreased (Svallfors 1998; 1999). In summary, the conflictpatterns taxation and welfare policies suggest that further moves towards a less ambitious, less universal welfare model may cause resistance from fairly large segments of the public.

To decide whether Sweden is facing an emerging 'crisis' or a temporary downturn in public support for taxes is difficult for several reasons. We may, however, be confident that political struggles on the terrain of the welfare state will continue into the next millennium.

---

NOTES

1. For an overwiew of arguments Concedrnes with the negative effects of advances welfare statism, see Pfaller et al. (1991). Atkinson & Mogensen (1993), and Esping-Andersen (1996) and works cited therein.

2. For descriptions of the Swedish tax system, see Steinmo (1993) and Edlund (1999a).

3. In 1987, nearly 75 percent of those working full-time had passed the 50 percent marginal tax level (SCB 1994,43).

4. Works analysing the reform's effects on income distribution are summarised in Åberg (1993).

5. The term 'conservative parties' refers to right-wing and centre parties (Moderate Party, Liberal Party, Centre Party, Christian Democratic Party).

6. Official reports (SOU 1995; Agell et al. 1995) evaluating the tax reform acknowledge both that the reform constrained the redistributive effects of the tax system as well as its significance for the loss of tax revenues. Taking into account the overall redistriposable before and after the tax reform have less to do with the reformation of the tax system than with rising inequalities in factor income.

7. Although variables are on the ordinal level, arithmetic means are reported, since the displayed percentage distributions do not distinguish between different levels of approval/disapproval 'agree'/'strongly agree' or 'too high'/'much too high.'

8. The tax level measure (tlm) is defined as: tlm = (tli + tmi + thi)/3, in which tli is the variable value for taxes on low incomes, tmi is the variable value for taxes on middle incomes, and thi is the variable value for taxes on high incomes. Since the tlm is divided by 3, it varies between -1 and 1.

9. The tax progression measure (tpm) may be defined as: tpm = (thi — tli)/2, in which thi is the variable value for taxes on high incomes and tli is the variable value for taxes on low incomes. The tpm varies between -1 and 1.

10. It may also be noted that while these items do not explicitly distinguish between different types of taxes, public preferences towards the general level of taxes tend to be closely associated with attitudes towards the income tax (0.57). Relationships between the general tax level and other types of taxes are in comparison much weaker: goods and services tax (0.19); corporate tax (0.18); taxes on tobacco and alcohol (0.10); energy taxes (0.18). Coefficients are Pearson's r. These items are only available in the 1981 survey, but the tentative evidence suggests that public attitudes towards taxation are heavily influenced by perceptions of the personal income tax.

11. The tax discontent index (tdi) is defined as: tdi = (i, +i₂+i₃)/3, in which i, is the variable value for item 1, i, for item 2, and i, for item 3 (item numbers correspond to items in Table 3). The index varies between -1 and 1. The reliability coefficient (standardized Cronbach's alpha) varies between 0.64 (1981), 0.71 (1992), 0.68 (1996), and 0.69 (1997). If'don't know' responses are recoded as missing values, the reliability coefficient increases, but only slightly.

---

12. The SEI classification is assessed in the following way: 11,12= unskilled workers; 21, 22 = skilled workers; 33, 36 = routine non-manuals; 46 = intermediate non-manuals; 56, 57, 60 = higher non-manuals; 79, 89 = self-employed.

13. The primary education category in the 1981 survey deviates, however, from the general classification. Due to the coding schema used in the 1981 survey, it is not possible to separate some early types of secondary education from compulsory schooling.

14. There are two deviations from this general schema. The Environment Party category is not available in the 1981 survey. In the 1992 survey, New Democracy is classified as a right-wing party. Additional analysis in which New Democracy and the Environment Party are classified as 'other party' shows similar results to those presented in this paper.

15. Since one of the dependent variables (tax distribution fairness) consists of a single question only, MCA may not be appropriate. For this reason, additional logistic regression (Tabachnick & Fidell 1996) applying a dichotomous tax distribution fairness variable was performed. However, the results in Table 4 are very similar to those observed using logistic regression.

16. During the most recent years, variations on two themes articulated by the political right in public debate can be distinguished. The first stresses that problems in the social services are primarily a question of inefficient allocation and organisation, not financial resources (e.g. 'We have the highest taxes in the world, but do we have the best schools [healthcare, elderly care]?'). The second emphasises the necessity of reduced taxes in order to bring down unemployment. Left-wing counter-arguments have been foremost articulated by LO and the Left Party. The Social Democrats have, by and large, taken a more reserved position in public debate.

17. It should be mentioned that an examination of the questionnaires and non-response characteristics shows that it is highly unlikely that the observed differences between 1996 and 1997 can be explained by context effects and/or biased samples (cf. Svallfors 1998).

ACKNOWLEDGEMENTS

The research has been financially supported by the Swedish Council for Social Research (SFR) and by the Swedish Council for Research in the Social Sciences and Humanities (HSFR). Comments by Ola Sjoberg, Stefan Svallfors, three anonymous referees, and the editors on a previous draft are appreciated.

REFERENCES

Åberg, R. 1993. Århundradets skattereform? Stockholm: Fritzes.

Åberg, R. 1994. 'Wage Control and Cost-Push Inflation in Sweden since 1960,' in Dore, R.,
Boyer, R. & Mars, Z., eds., The Return to Incomes Policy. London: Pinter.

Åberg, R. 1998. 'Arbetsloshet och marginalisering,' in Palme, J. & Stenberg, S. Å? eds.,
Arbetsloshet och vdlfdrd. Vålfårdsprojektet. Skriftserien nr 11. Stockholm: Socialdepartementet.

Åberg, R., Selen, J. & Tham, H. 1987. 'Economic Resources,' in Erikson, R. & Åberg, R.,
eds., Welfare in Transition. Oxford: Clarendon Press.

Agell, J., Englund, P. & Sodersten, J. 1995. Svensk skattepolitik i teori och praktik. 1991 års
skattereform. Bil. 1, SOU 1995, 104. Stockholm: Fritzes.

Andrews, F., Morgan, J. & Sonquist, J. 1973. Multiple Classification Analysis. Ann Arbor:
University of Michigan.

Åsard, E. & Bennett, W. L. 1992. Regulating the Marketplace of Ideas: Tax Reform and
Election Rhetoric in Sweden and the United States. Uppsala North American Studies
Reports, No. 8. Swedish Institute for North American Studies. Uppsala University.
Atkinson, A. B. & Mogensen G. V., eds. 1993. Welfare and Work Incentives. Oxford: Clarendon
Press.

Borchorst, A. & Siim, B. 1987. 'Women and the Advanced Welfare State - A New Kind of
Patriarchal Power?' in Showstack Sassoon, A., ed.. Women and the State. London:
Hutchinson.

Coughlin, R. 1980. Ideology, Public Opinion and Welfare Policy. Berkeley: Institute of International
University of California.

Daly, M. 1994. 'Comparing Welfare States: Towards a Gender Friendly Approach.' in
Sainsbury, D., ed., Gendering Welfare States. London: Sage.

Edlund, J. 1999a. "Progressive Taxation Farewell? Attitudes to Income Redistribution and Taxation in Sweden, Great Britain and the United States,' in Svallfors, S. & Taylor-Gooby, P., eds.. The End of the Welfare State? Public Responses to State Retrenchment. London: Routledge.

Edlund, J. 1999b. 'Attitudes Towards Tax Reform and Progressive Taxation: Sweden 1991-1996,'
Sociologica 42, 337-55.

Elvander, N. 1994. 'Demokratins problem och den svenska modellen,' in Holmlund, 8., ed.,
Arbete, loner och politik. Stockholm: Fritzes.

Erikson, R. & Goldthorpe, J. H. 1992. The Constant Flux. Oxford: Clarendon Press.
Esping-Andersen, G. 1985. Politics against Markets: The Social Democratic Road to Power.

Princeton: Princeton University Press.

Esping-Andersen, G. 1990. The Three Worlds of Welfare Capitalism. Cambridge: Polity Press.
Esping-Andersen, G., ed. 1996. Welfare States in Transition. London: Sage.
Esping-Andersen, G., Assimakopoulou, Z. & van Kersbergen, K. 1993. 'Trends in Con-

Evans, G. 1992. 'Testing the Validity of the Goldthorpe Class Schema,' European Sociological
Review 8, 211-32.

Feldt, K.-O. 1991. Alia dessa dagar . . . I regeringen 1982-1990. Stockholm: Norstedts. Finansdepartementet. 1990. Århundradets skattereform. Stockholm: Allmånna Forlaget. Fritzell, J. 1991. Icke av marknaden allena: Inkomstfordelningen i Sverige. Stockholm: Almqvist & Wiksell International.

Gilljam, M. & Holmberg, S. 1993. Våljarna infor 90-talet. Stockholm: Norstedts.
Hadenius, A. 1981. Spelet om skatten. Lund: Norstedts.

Hadenius, A. 1986. A Crisis of the Welfare State? Stockholm: Almqvist & Wiksell International.

Hernes, H. 1987. "Women and the Welfare State: The Transition from Private to Public
Dependence,' in Showstack Sassoon, A., ed., Women and the State. London: Hutchinson.
Hibbs, D. A. & Madsen, H. J. 1981. 'Public Reactions to the Growth of Taxation and

Government Expenditure,' World Politics 33, 413-35.

Hjerm, M. & Svallfors, S. 1997. Asikter om politiken och den offentliga sektorn. Codebook
for machine-readable datafile. Department of Sociology, Umeå University.

Hjerm, M. & Svallfors, S. 1998. Asikter om skatter och offentliga sektorn. Codebook for
machine-readable datafile. Department of Sociology, Umeå University.

Inglehart, R. 1990. Culture Shift in Advanced Industrial Society. Princeton: Princeton University
Press.

Inglehart, R. 1997. Modernization and Postmodernization. Princeton: Princeton University
Press.

Korpi, W. 1980. 'Social Policy and Distributional Conflict in the Capitalist Democracies. A
Preliminary Comparative Framework,' West European Politics 3, 296-316.

Korpi, W. 1983. The Democratic Class Struggle. London: Routledge & Kegan Paul. Korpi, W. 1993. 'Politik och våljare bakom valutgången 1991,' Sociologisk forskning 30, 3-28. Korpi, W. & Palme, J. 1998. 'The Paradox of Redistribution and Strategies of Equality: Welfare State Institutions, Inequality, and Poverty in the Western Countries,' American Sociological Review 63, 661-87.

Lee, D. J. & Turner, B. S., eds. 1996. Conflicts about Class: Debating Inequality in Late
Industrialism. London: Longman.

Matheson, G. 1993. The Decommodified in a Commodified World. PhD Thesis, University of
New England, Armidale.

McFate, K., Lawson, R. & Wilson, W. J., eds. 1995. Poverty, Inequality and the Future of
Social Policy. Western States in the New World Order. New York: Russell Sage
Foundation.

Musgrave, R. A. 1990 'Comments,' in Slemrod, J., ed., Do Taxes Matter? Cambridge, MA:
MIT Press.

Nieuwbeerta, P. 1996. 'The Democratic Class Struggle in Postwar Societies,' Acta Sociologica
39, 345-83.

OECD. 1990. The Personal Income Tax Base. Paris: OECD.

Orloff, A. S. 1993. 'Gender and the Social Rights of Citizenship: The Comparative Analysis
of Gender Relations and Welfare States,' American Sociological Review 58, 303-28.

Pakulski, J. & Waters, M. 1996. The Death of Class. London: Sage.

Palme, J. & Wennemo, I. 1998. Swedish Social Secuhty hi the 19' MI '.s: Reftirm and Retrenchment.
Vålfårdsprojektet. Stockholm: Printing Works of the Cabinet and Ministries.

Pfaller, A., Gough, I. & Therborn, G., eds. 1991. C an the Welfare State Compete? London:
Macmillan.

SCB. 1982. Socio-ekonomisk iiulchiinv, (SEI). Meddelanden i samordningsfrågor 1982. 4.
Stockholm: Statistiska Centralbyrån.

SCB. 1994. Skatter, inkomster och avgifter. En oversikt 1994. Stockholm: Statistiska Centralbyrån.

Scheuch, E. K. 1989. 'Theoretical Implications of Comparative Survey Research: Why the
Wheel of Cross-Cultural Methodology Keeps on Being Reinvented,' International Sociology
4,147-67.

Siaroff, A. 1994. 'Work, Welfare and Gender Equality: A New Typology,' in Sainsbury, D.,
ed., Gendering Welfare States. London: Sage.

Sodersten, 8., ed. 1991. Den stora reformen. 2nd edn. Stockholm: SNS forlag.

SOU. 1989a. Reformerad inkomstbeskattning. SOU 1989, 33. Vol. I-IV. Stockholm: Allmånna
forlaget.

SOU. 1989b. Reformerad foretagsbeskattning. SOU 1989, 34. Vol. I-11. Stockholm: Allmånna
forlaget.

SOU. 1989c. Reformerad mervdrdesskatt. SOU 1989, 35. Vol. I-11. Stockholm: Allmånna
forlaget.

SOU. 1995. Skattereformen 1990-1991. En utvdrdering. SOU 1995, 104. Stockholm: Fritzes.
SSD. 1995. Skatter i Sverige 1981. Codebook SSD 0491. Swedish Social Science Data Service,
Goteborg University.

Steinmo, S. 1993. Taxation and Democracy. New Haven: Yale University Press.
Stenberg, K. & Svallfors, S. 1992a. Åsikter om den offentliga sektorn. Codebook for machinereadable
Department of Sociology, Umeå University.

Stenberg, K. & Svallfors, S. 1992b. Attitudes to Inequality - A Swedish Survey. Codebook
for machine-readable datafile. Department of Sociology, Umeå University.

Stephens, J. D. 1996. 'The Scandinavian Welfare States: Achievements. Crisis, and Prospects.'
in Esping-Andersen, G., ed.. Welfare States in Transition. London: Sage.

Sundstrom, E. & Svallfors, S. 1995. Åsikter om den offentliga sektorn och skatterna. Codebook
for machine-readable datafile. Department of Sociology, Umeå University.

Svallfors, S. 1995. 'The End of Class Politics? Structural Cleavages and Attitudes to Swedish
Welfare Policies,' Acta Sociologica 38, 53-74.

Svallfors, S. 1996. Vdlfdrdsstatens moraliska ekonomi. Umeå: Boréa.

Svallfors, S. 1998. 'Mellan risk och tilltro: opinionsstodet for en kollektiv vålfårdspolitik."
Mimeo. Department of Sociology, Umeå University.

Svallfors, S. 1999. 'The Middle Class and Welfare State Retrenchment: Attitudes to Swedish
Welfare Policies,' in Svallfors, S. & Taylor-Gooby, P., eds.. The End of the Welfare State?
Public Responses to State Retrenchment. London: Routledge.

Tabachnick, B. G. & Fidell, L. S. 1996. Using Multivariate Statistics. 3rd edn. New York:
HarperCollins.

Wahlund, R. 1991. Skatter och ekonomiska beteenden. Stockholm: EFI.

Wilensky, H. 1976. The 'New Corporatism,' Centralization, and the Welfare State. Beverly
Hills: Sage.