Jump to ContentJump to Main Navigation
The Scope of Government$

Ole Borre and Elinor Scarbrough

Print publication date: 1998

Print ISBN-13: 9780198294740

Published to Oxford Scholarship Online: November 2003

DOI: 10.1093/0198294743.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 17 September 2019

The Welfare State: The Security Dimension

The Welfare State: The Security Dimension

Chapter:
(p.198) 8 The Welfare State: The Security Dimension
Source:
The Scope of Government
Author(s):

Per Arnt Pettersen

Publisher:
Oxford University Press
DOI:10.1093/0198294743.003.0008

Abstract and Keywords

The basic question for analysis in this chapter is whether people prefer that security benefits such as state pension schemes, public health services, and unemployment benefits should expand, continue as they are, or diminish. The chapter reviews the historical development of welfare and benefit provision from the self‐confident 1960s, through the stagnation of the 1970s, through to the austerity of the 1980s. It then examines which demographic and social groupings are likely to support or oppose the expansion of security provision by the welfare state, and the nature of the connections between public perceptions and party‐political positions on welfare policies.

Keywords:   pension schemes, public health services, security benefits, unemployment benefits, welfare state, Western Europe

Whatever the historical origin and nature of the welfare state, there are programmes to protect people against misfortune, especially loss of income, ill health, or unexpected expenditure. These are some of the means of pursuing the goal of social equality, which was analysed in the previous chapter. Among the most common and visible programmes are old age pensions, public health services, and unemployment benefits. The most expensive, but also probably the most popular, include the various systems of old age pensions and health services. These two programmes are the cornerstones, and the least contested aspects, of the welfare state. Old age comes to everyone (who does not die young) and sickness is fairly randomly distributed through a population, so no one can be certain that they will not need these services at some stage in their life. Hence, the widespread support for old age pension schemes and health services found in most studies of the welfare state.

Still, even for these two programmes, there is not universal agreement that the government should be the sole provider. Some feel that private savings and private insurance should provide economic security in later life; some prefer the real costs of health services to be paid for by individuals or private insurance within a health services market. However, the basic legitimacy of old age pensions and health services is derived from the universality of ageing and the risks of illness.

Unemployment, by contrast, is distributed in a systematic way. The young and old, the less educated and unskilled, people living in areas of declining economic activity—all have higher than average risks of (p.199) unemployment. However, for people who work, unemployment benefits might be viewed as an alternative to wages and salaries, as another form of income. If their earnings are low, the employed might even question whether they are better off than those living on the ‘dole’. They may wonder if all the unemployed are deserving individuals compared to wage earners. These kinds of doubts may reduce support for unemployment benefits.

The basic question for analysis in this chapter is whether people prefer to see state pension schemes, public health services, and unemployment benefits expand, continue as they are, or reduced. Is spending to be increased or are services, pensions, and benefits considered adequate? These questions concern ‘how much’, or the degree dimension of the scope of government. The range dimension of attitudes towards social security was discussed in Chapter 4.

In the next section, we analyse how preferences for increased security and protection have developed and varied from the economically confident period of the 1960s, which saw the maturing of the welfare state, through the economic stagnation of the 1970s, and the fiscal austerity of the 1980s. The second section deals with who—which demographic and social groups—is more likely to support or oppose the expansion of the programmes for old age pensions, public health services, and unemployment benefits.

Developmental Theories

Two theoretical traditions are explicitly committed to explaining developments in popular support for the welfare state. One we might call ‘resistance among the well‐to‐do’, the other ‘new values theory’. Both positions argue that the original support which the welfare state enjoyed during the maturing years of the 1960s is fading away.

Typical of ‘resistance’ arguments is Rose and Peters's (1978) claim that support for welfare programmes will decline as economic recession produces falling real disposable incomes for wage earners. Although varying in severity, economic recession is a common experience in West European countries; indeed, an important factor in the growth of welfare expenditure. The burdens on the wage earners are increased by the ‘invisible hand’ of inflation and progressive taxation which together reduce the purchasing power of incomes (Rose and Peters 1978; Rose 1984)—or, at least, obstruct expectations about rising real (p.200) incomes. Rose and Peters predict common trends of reduced support for the expansion of public spending in every West European country, with opposition being most prominent where economic stagnation has been most apparent, as in Britain, or where taxation levels have been most extravagant, as in the Scandinavian countries. Thus, if Rose and Peters's developmental theory is correct, a reduced willingness to expand spending on welfare programmes should be evident among the employed, especially among the more heavily taxed upper income groups. These trends should be apparent regardless of how support is measured.

One version of the ‘new values’ theory is suggested by Wilensky (1975), who argued that the advanced welfare states were heading for a ‘welfare backlash’. Rather than the impact of economic recession, Wilensky was concerned with the impact of economic growth on the transformation of modern society. The message was that education, new industries, and economic development breed individualistic values. An increasingly complex division of labour multiplies the number of occupations requiring specific skills, necessitating individual and discriminating compensation. Specialization is produced through education and occupation, which subsequently promote social mobility. Since this evolution affects mostly the young, Wilensky argued that—in particular—the well‐educated younger generation will abdicate as supporters of the welfare state. But even the working‐class, often the original target group of welfare programmes, will eventually forsake the welfare state; career patterns and salaries are bound to become individualistic, leading to the abandonment of collective wage negotiation between labour unions and employer's organizations. The ‘new’ social category, according to Wilensky, is the ‘middle mass’, which will refuse to pay the necessary taxes, since they perceive no benefits from the welfare state. In common with Rose and Peters, Wilensky suggests continuously falling support for welfare arrangements, including old age pensions, but, by contrast, he suggests that wage earners, especially as income increases, will lose faith in the welfare state. Some may argue that the ‘yuppies’ of the 1980s precisely fit with Wilensky's predictions of ‘economic individualism’ (Wilensky 1975: 32–9).

Another version of the ‘new values’ theory, similarly predicting falling support for the welfare state, can be derived from Inglehart's theory of generations (Inglehart 1977). A simple and direct interpretation of Inglehart's arguments implies that people who grew up before the Second World War—in societies suffering from unemployment, (p.201) precarious wage agreements, and sparse welfare arrangements—were the advocates of the welfare state and, later, its most reliable defenders. Thus, Inglehart predicts, people of those generations will tend to be more materialist in outlook than the generations who grew up after the war. By contrast, the post‐war generations have known material security and moved towards giving priority to postmaterialist values which emphasize the quality of life (Pöntinen and Uusitalo 1986). Even though the implications of ‘postmaterialist’ values are not entirely clear, it is hard to avoid the conclusion that they imply a continuous decline in support for the welfare state as ‘materialist’ generations are replaced by new cohorts of ‘postmaterialists’.

For Wilensky, too, changing levels of support are related to generations: younger people are less inclined to support the welfare state than pre‐war generations. Inglehart also concurs with Wilensky in suggesting that education promotes a new value system. But postmaterialist values are rooted in socialization experiences and encouraged by an awareness of new problems and inventive priorities, not by current occupational opportunities (Inglehart 1977). However, an alternative to the two generation thesis is a ‘three or four’ generation classification: the pre‐war generation, the early post‐war generation, the generation which saw the welfare state mature, and the youngest ‘post‐welfare state’ generation. The latter two, or just the last generation, is thought to oppose welfare programmes (Svalfors 1989). However, whatever the details, the predictions made by Wilensky and Inglehart, and the classification offered by Svalfors, agree in focusing on the impact of generational differences; in brief, that the expansion of the modern welfare state is opposed by the young and well‐educated. Thus, the successive replacement of old cohorts by new young cohorts will lead to declining support for the welfare state.

Note that these theoretical positions, although they rely on different assumptions, are not exclusive. Rather, they may reinforce each other and simultaneously yield accurate predictions. Rose and Peters's assumption that taxpayers constitute an opposition to welfare state expansion may be correct in itself and be in agreement with the predictions of both Wilensky and Inglehart that well‐educated younger people oppose expansion. Moreover, such findings might be consistently related to changing value systems. It is even possible that part of the new generation adheres to the values of ‘economic individualism’ while others have become ‘postmaterialists’.

A quite different theoretical position argues that looking for linear (p.202) trends in support for the welfare state is futile (Downs 1972). The assumption is that people's attitudes on political issues, including the welfare state, are cyclical, both with regard to issue attention and support. The opinion formers and leaders in a society are politicians and the mass media in juxtaposition (Lazarsfeld, Berelson, and Gaudet 1948). The key condition for popularity cycles is the introduction of political novelties, new ideas, or the revival of old ones. Exciting personalities, dramatic presentation and campaigning, conflicts and fierce struggles seize the attention of the mass media. This is reinforced by the tendency of the mass media to concentrate on those with unconventional convictions. Yet political ideas compete for attention, and even the most charismatic individuals and fascinating ideas lose their impact when constantly repeated. Downs's theory of ‘issue attention cycles’ seems appropriate to analysing attitude change during the 1970s and 1980s when the ‘new right’ movement was initiated and grew in vitality.

Operationally, this kind of argument suggests that disapproval of the welfare state increases as conservatives celebrate their political victories. We might even anticipate that attitude fluctuations were exposed suddenly, abruptly, and most visibly as the conservative moods of the 1970s and 1980s broke through in various countries. Moreover, it may be the case that stable attitudes have been preserved best where no neo‐liberal party (or party leadership) emerged or where no conservative parties won increased support from the electorate. This position, of course, implies that as the conservative mood fades away, so does opposition towards welfare state expansion.

Finally, we might consider that, even if predictions of a linear decline in support should prove to be wrong, there might be tendencies which would save at least parts of these positions. Inside every wave of conservative mobilization, the front runners in opposing welfare state expansion, and the last to drop their opposition as the political mood changes again, might be Wilensky's ‘middle mass’, Inglehart's ‘post‐materialists’, or Rose and Peters's ‘abused tax‐payers’. None the less, the hypotheses implied by these theories engage the assumption that particular groups will become more pronounced opponents of the welfare state as we move from the economic growth of the 1960s, through the stagnation of the 1970s, and the crises of the 1980s.

(p.203) Support for the Welfare State

The theories discussed so far have been concerned with declining support for the welfare state, and predictions about who constitutes the opposition. Other theoretical approaches suggest that large segments of the citizenry have acquired strong and lasting confidence in the welfare state. If this is so, the welfare state has the necessary political strongholds to survive and even to develop.

Pöntinen and Uusitalo (1986) argue that at least three groups in society have vested interests in the cultivation and expansion of the welfare state. The consumers, or prospective consumers, of welfare transfers are obvious candidates; the oldest cohorts, in particular, are most likely to support the expansion of old age pensions.1 Secondly, people who perceive that they incur high tax burdens without consuming any services or receiving transfers, now or in the future, are likely to develop negative attitudes. This implies that people in middle and lower income groups would offer firm support for welfare programmes if the perceived protection and benefits were to surpass tax burdens. This assumption is linked to Rose and Peters's theory of linearly declining support but is distinct in so far as levels of support are expected to hold fairly constant, as between income categories, over the thirty‐year period under study. Thirdly, those employed in the public sector, often in welfare state occupations, are likely to develop positive attitudes, because the growth of the welfare state enhances their employment prospects and occupational prestige (Kolberg and Pettersen 1981; Dunleavy 1989). In addition, it is reasonable to assume that public sector employees identify with the values associated with welfare state institutions (Edgell and Duke 1991; Dunleavy and Husbands 1985).

Next, we suggest that people identifying with the lower or working class are most inclined to support the expansion of the welfare state—if only because they are influenced by political rhetoric about the welfare state and on account of patterns of political support among the working class. In most countries, the working class has been thought to be the target group for welfare state reforms, even though empirical evidence suggests that the better off often derive greater benefits from many public services (Olson 1990). Where the connection between the working‐class vote and welfare state expansion has been close, we expect class differences in support to be most apparent. Obviously, this also implies that supporters of socialist parties, perceived as the designers of the welfare state, constitute the cornerstone of political support for the (p.204) welfare state. The Scandinavian countries have been identified as paradigms of the close ties between the working class, labour and social democratic parties and support for the welfare state (Korpi 1981; Esping‐Andersen 1985). Britain should fit this picture as well. But evidence of the weaker association between class and party preference over the last thirty years (Worre 1980; Franklin and Mughan 1978) suggests that variation in welfare state support among voters for different parties might have disappeared.

Suggestions that women have increasingly benefited from welfare state programmes have also been advanced (Hernes 1987; Waernes 1987). Women's traditional obligations to family care have been relieved by welfare arrangements, and welfare programmes have reduced women's dependence on their husband's income, especially in cases of divorce and widowhood. Simultaneously, the welfare state has become a major arena for women's employment. We anticipate that all these factors promote positive attitudes towards welfare programmes among women in general. However, the major programmes benefiting women, as well as the advancement of women in the labour force, have appeared only during the last twenty years or so. Thus, we assume that younger women are more favourably disposed to the welfare state than their mothers and grandmothers. This tendency should increase as we move from the 1960s to the 1980s. Yet younger or working women may have higher expectations of the welfare state, leading them to be less satisfied and more critical than older, and possibly less demanding, women.

Analytic Strategy

Our theoretical assumptions focus on who opposes or supports welfare state expansion. However, some of our questions are implicitly concerned with why some people support and some people oppose welfare state expansion. We can say little about this because we do not have comparable survey data on attitudes towards the welfare state which might be correlated with opinions about welfare state expansion. A broader picture of the attitude structures which underpin support or opposition has to wait until we have a series of surveys over several years which ask the appropriate opinion and value questions.

Comparing popular opinions towards welfare state programmes at one specific time point, let alone over time, involves a multitude of (p.205) obstacles and possibilities for misinterpreting the data. Survey questions are not identical, or even similar, between national surveys; question and answer categories in national surveys are modified over the years. The specific character of national welfare programmes also means that we have to be cautious about interpreting findings since institutional variations may lead people to react differently even when the survey questions are similar.

Even so, although the survey questions are not identical and programmes differ, we may still assume that the different preferences of various social groups about expansion are sufficiently robust and stable to be revealed by multivariate analysis. Drawing on the theoretical arguments outlined above, we consider similarities of support and opposition across a number of countries. In the next section, we focus on similarities in the development of attitudes across countries, starting by introducing trends in Britain as a comparative baseline. The British Election Study and the British Social Attitudes surveys constitute the longest time series available. Even more to the point, Britain has endured the economic, social, and political changes most relevant to our theoretical propositions.

Development of Preferences

Longitudinal data about the evolution of popular support for old age programmes are available for six countries: the four Nordic countries, Britain, and the Netherlands. There are two basic variants of the relevant question; both of them are included in the British data. One concerns preferences for increasing old age pensions, which has also been asked in Finland and the Netherlands. In Britain and the Netherlands, the wording of the question is virtually identical. The second, and comparable, question relates to programmes of social security. A question of this kind has been asked in Denmark, Norway, Sweden, and Britain. As transfers to the older generation constitute the major part of the programme in these countries, respondents are likely to associate the questions with old age pension schemes. However, this kind of question is, in fact, more general in nature than those specifically about old age pensions.

Overall, we consider these questions as functional equivalents—in the sense that they should reveal the same tendencies. That is, according to our initial theoretical arguments, we anticipate either a linear decline (p.206) in support or politically based cycles of support. We also assume that the differences in question formulation are of minor importance when analysing social variation in the structure of preferences.

First, then, we examine the moods of the British over the last twenty years. The period was marked by major political shifts, from the Conservative government of the early 1970s, the Labour government of the mid‐1970s, through the landslide victories of Margaret Thatcher and towards her less popular years of the late 1980s. This gives us both a preliminary overview of developments in attitudes, and an initial test of the theoretical arguments. In Table 8.1 we show, for four elections, whether people in Britain thought that social services and benefits had gone too far and should be cut back, should stay as they are, or be expanded.

Three main trends are evident. First, there is a basic stability over the period in the proportions who prefer expansion; approximately 30 per cent preferred an increase in social services and benefits in 1974, 1983, and 1987. But, secondly, in 1979 the proportion who preferred expansion fell to less than 20 per cent. The 1979 election was the first of Mrs Thatcher's victories, when the Conservative Party won its largest share of the vote since the 1950s (44 per cent). The notable corollary is the increase in the proportion of the electorate wanting a reduction in services and benefits. The balance of opinion did not shift from expansion to the status quo, but exhibits almost a 20 percentage point increase in the proportion of people preferring reduced levels of services and benefits. Thirdly, by 1983 the proportion wanting to

Table 8.1. Support for Expansion or Reduction of Social Services and Benefits in Britain, 1974–87

1974

1979

1983

1987

Expansion

31

19

28

34

As now

33

26

49

39

Reduction

32

50

19

24

Don't know

4

5

4

3

N

2,431

1,859

3,933

3,810

Notes: Entries are percentages. The questions are formulated similarly for 1974, 1979, 1983, and 1987: ‘Now, we would like to ask what you think about social services and benefits. Which of these views comes closest to your own view: social services and benefits have gone too far and should be cut back; social services and benefits have gone somewhat too far and should be cut back a bit; social services and benefits should stay as they are; more social services and benefits are needed?’

Sources: British Election Study (1974, 1979, 1983, 1987).

(p.207) reduce provision fell to a lower level than in 1974. Whereas 32 per cent opted for reductions in 1974, this figure dropped to 19 per cent in 1983, then climbed back to 24 per cent in 1987. This pattern implies that over the years an expanding proportion felt that the level of benefits and services was adequate and satisfactory. Note also that the proportion preferring expansion peaks at the 1987 election, while opposition to expansion is lowest at the 1983 election.2

The only explanation which fits this pattern of fluctuations within basic stability is political preference cycles stimulated by political campaigns. The breakthrough of the right wing of the Conservative Party in the late 1970s stimulated a general feeling of dissatisfaction with the over‐expansion of social policies in Britain. Neither the generation theories of Wilensky and Inglehart, nor the tax burden theory of Rose and Peters—all predicting a linear decline in preferences for expansion—contain the possibility of generally stable support along with opposition cycles. Whether the specific groups predicted by these theories constitute the new opposition will be analysed later. Now we look to see if the same pattern holds for support for old age pensions in Britain and our other West European countries.

Comparing support for old age pensions in Britain, shown in Table 8.2, with general preferences for the expansion of benefits and services (Table 8.1), immediately reveals that schemes specifically related to old age pensions are more widely supported. The only real disagreement is between those who want expansion to continue and those who feel that provision is satisfactory. Hardly any opposition exists.

However, popularity cycles are again evident. In the early 1960s in Britain, 77 per cent wanted spending on old age pensions to increase, the figure drops to 55 per cent in the mid‐1960s, and to 43 per cent in the late 1960s. Then a slight increase occurs in 1970. Unfortunately, the British Election Study did not include the question about old age pensions in 1974 and 1979, so we cannot tell if support for increasing old age pensions fell in the 1970s as it did for benefits and services. But the increase in support for old age pensions during the 1980s is clear, reaching the same levels as in the early 1960s.

The Netherlands, too, shows declining support for increasing old age pensions through the 1970s, followed by a resurgence of support during the second half of the 1980s. That is, support for expansion followed a cyclical pattern. The low proportion of expansionists in 1980 coincided with increased support for the liberal parties and a drop in support for the socialist parties.3 In the following election, support for the liberal (p.208)

Table 8.2. Support for Social Security and Old Age Pensions in Britain and the Netherlands, 1964–87

Expansion

As now

Reduction

Don't know

N

Britain

1964

77

20

3

1,769

1966

55

41

4

1,874

1969

43

44

8

5

1,114

1970

57

36

4

3

1,334

1983

81

16

3

3,941

1985

74

23

2

1

1,474

Netherlands

1970

59

28

13

1,901

1975

33

58

1

8

1,975

1980

22

65

1

12

1,977

1985

33

54

1

12

1,954

1986

35

49

1

15

2,012

1987

37

47

1

15

1,985

Notes: Entries are percentages. The British question for 1964, 1966, 1969, 1970, and 1983 is: ‘Do you feel that the government should spend more on pensions and social services, or do you feel that spending for social services should stay as it is now?’ For 1985: ‘Do you think that the government should spend more or less on old age pensions?’ The Dutch question is: ‘Now, I mention some social services. Will you please tell me for every one of them whether it is, in your opinion, sufficient or insufficient at the moment? What is your opinion of old age pensions?’

Sources: British election studies; British Social Attitudes Study (1985); Cultural Change in the Netherlands.

parties dropped back again, accompanied by a revival of support for more spending on pensions, although not reaching the same levels as in the late 1960s. Finally, it is particularly striking that there was virtually no support for reducing old age pensions in either Britain or the Netherlands during this period.

Does support for social security follow the same pattern in the Nordic countries? These are often portrayed as consensus nations, where there was negligible conflict among parties and voters about the development of the welfare state. The common picture is also of basic similarities in their welfare state institutions. This picture reflects both self‐perceptions among the Nordic peoples, as well as descriptions provided by foreign observers (Graubard 1986; Kolberg 1992).

The distributions in Table 8.3 reveal similarities as well as differences in the evolution of attitudes towards social security schemes and reforms. As in Britain and the Netherlands, none of the Nordic countries (p.209) reveals basic and long‐term preferences for reducing old age pensions. However, the proportion in favour of expansion varies considerably over time and between the four countries.

The pattern of support in Denmark and Norway conform nicely to the popularity cycles revealed in Britain and the Netherlands. These tendencies are less obvious in Sweden, but they are not at all visible in Finland. Indeed, only Finland fits the assumption of a linear decline in support. However, as we have few data‐points for Finland, interpretations have to be cautious. Although after 1975 there was altogether a 20 percentage point decline in the proportion supporting the expansion of pensions, there was still, in 1985, close to a majority preferring expansion. Moreover, as these were Finland's most prosperous years, economic recession or fiscal crises do not provide credible explanations for the decline. Again, Finland is neither among the most heavily taxed countries, nor did Finland experience any major increases in the level of taxation during this period (Lane, McKay, and Newton 1991: tables 5.1–5.4). At face value, it looks as if either Wilensky's or Inglehart's generation theory might fit the pattern of attitudes in Finland. However, closer inspection of the data reveals that it is actually among the younger generation that the welfare state enjoys most popularity, even if the differences are rather modest (Pöntinen and Uusitalo 1986: 54).

The popularity cycle is most visible in Denmark. During the 1960s, at least 60 per cent supported the expansion of social security; in the early 1970s, this proportion dropped to around 40 per cent; by the late 1970s, there was again a majority preferring expansion; and by the 1980s firm majorities for expansion, reaching 80 per cent in 1987, had returned. By 1990 there was decline again, with 66 per cent supporting expansion. The ups and downs of the Danish popularity cycle are closely related to the rise and fall of the populist—anti‐tax and anti‐government—Progress Party. The Progressives had an immediate and surprising success during the 1970s, garnering around 15 per cent of the votes in three elections. In 1984, however, support dropped to a meagre 3 per cent.

In Norway, preferences for expanding social security peaked with the National Insurance Scheme, introduced in 1966 by a non‐socialist government. The first major drop follows closely after 1973, when, as in Denmark, the Progress Party was established and won its first seats in parliament.4 Ties to the cycles of political popularity are clear again in 1981, when there was a further decline in support, but this time linked to the victory of the Conservative Party, which achieved its highest (p.210)

Table 8.3. Attitudes to Social Security, Social Reforms, and Old Age Pensions in the Nordic Countries, 1960s–1990s

Expansion

No opinion

No expansion

Don't know

N

Sweden

1960

33

8

49

10

1,466

1964

33

63

4

2,849

1968

51

42

7

2,866

1970

35

59

6

2,743

1973

32

60

8

2,421

1976

32

61

7

2,683

1979

27

67

6

2,674

1980

25

68

7

1,449

1985

30

62

8

2,671

1988

30

61

9

2,529

1991

24

68

8

2,478

Denmark

1964

62

22

16

3,104

1969

67

5

23

5

707

1973

42

58

1,850

1974

39

7

47

7

3,915

1978

55

7

32

6

2,085

1979

56

7

31

6

1,973

1984

63

6

24

7

1,029

1987

80

4

15

1

1,022

1988

73

4

20

3

670

1990

66

5

26

3

1,002

Expansion

As now

Reduction

Don't know

N

Norway

1965

42

36

9

13

1,619

1969

(68)

(25)

( 7)

1,162

1973

18

49

22

11

1,218

1977

33

38

7

22

1,727

1981

22

51

16

11

1,596

1985

38

39

5

18

2,180

1989

27

56

11

6

2,195

1993

21

63

10

6

2,194

Finland

1975

68

23

4

5

1,011

1980

58

37

2

3

976

1985

47

46

4

3

961

Notes: Entries are percentages. ‘No opinion’ in the Swedish and Danish data means that the responses are not classifiable according to the categories used. The Swedish question, formulated similarly for each year, is: ‘Social reforms have gone so far in this country that the government should reduce rather than increase social support and benefits in the future. What is your opinion on this issue? Do you entirely agree, agree somewhat, have an ambivalent position, disagree somewhat, or disagree entirely?’ The ambivalent category was used only in 1960. The ‘entirely’ and ‘somewhat’ responses are collapsed in the table. The Danish question varies. In 1964, the question was: ‘What do you think of the current size of the old age pension?’ In 1969, 1974, 1978, 1979, and 1984, the question was: ‘Which of these statements fits your own opinion: (a) Social reforms have gone too far, people should manage without social support and benefits from government; (b) The social reforms which have been introduced in Denmark should at least be maintained at the present level. Which do you agree with?’ In 1973, 1987, 1988, and 1990, the question is similar to that asked in Sweden. The Norwegian question, except in 1969, is: ‘Many people think that by now we have more than enough social security benefits, and that we should attempt to limit them in the future, while others claim that we should maintain our social benefit programmes and if necessary extend them. Do you have an opinion on that? What is your opinion?’ In 1969 the question was: ‘Social reforms in this country have come so far that the government must reduce social payments and support to people instead of increasing them.’ Those disagreeing are classed as expansionists; those agreeing wanting reduction. The Finnish question is: ‘In the following I list some cash benefits mainly financed by taxes. Please tell me, in each case, whether taxes should be used more than now, the same as now, or less than now to finance these benefits.’ (Pöntinen and Uusitalo 1986).

Sources: national election studies.

(p.211) post‐war vote: from 17 per cent in 1973, Conservative support rose to 32 per cent in 1981. The 1985 election was virtually an election on the welfare state, which revitalized the popularity of social security expansion. The drop yet again in 1989 is associated with the best ever result achieved by the Progress Party. However, both the Conservatives and Progressives did rather poorly in 1993, while support for expanding social security dropped even further. But this decline is more than compensated for by the highest ever proportion expressing satisfaction with the current level of provision, while opposition remained stable.

Sweden reveals more stable patterns, although some variation is evident. Swedish support peaked during the late 1960s, when about 50 per cent of the electorate preferred increased social support and benefits. This corresponds to the best ever electoral result for the Social Democrats, who won more than 50 per cent of the vote. Thereafter, during the 1970s, support fell back to the levels found in the early 1960s, and, in parallel, opposition to expansion grew until, in 1980, expansion was supported by only about a quarter of the electorate and opposed by about two‐thirds. This pattern broadly continued through to 1991, when the Conservative vote increased to 23 per cent (from 18 per cent in 1988) and New Democracy—a neo‐liberal, populist party similar to the Progress Party in Denmark and Norway—won almost 7 (p.212) per cent of the vote in its first election. Even so, note that the variation in support for expanding social security is only slightly above 20 percentage points (based on the deviant case of the 1968 election), whereas in Denmark the variation was about 40 percentage points. Generally, Swedish support levels have stayed at around the 30 per cent level for a period of twenty‐five years, with the exception of the electoral successes of the Conservative Party in 1980 and 1991. This pattern clearly supports the theory of popularity cycles.

The overall picture reveals some common patterns in support for pensions and social security. During the 1960s, when most countries introduced major social reforms, their popularity seems unquestionable. This conforms to our initial theories about attitudes towards the welfare state. During the 1970s, welfare systems continued to enjoy support, but with growing division between those wanting continued expansion and those wanting only to preserve, or even dismantle, what had been established. This mood of non‐expansion or opposition seems to be shared in a number of West European countries from the 1970s through to the 1990s. These changes in attitudes are closely linked to the political campaigns of either new populist, neo‐liberal parties, or changes in the ideology and policies of old liberal or conservative parties. Common to their campaigns was opposition to welfare state expansion, even proposals to dismantle established programmes in some cases. In Britain and Scandinavia, where these shifts have been most evident, shifts in attitudes have also been most pronounced. Yet, according to our evidence, bouts of opposition to the welfare state were short‐lived exceptions. During the 1980s, as economic recession became evident, the popularity of the welfare state was revitalized—except in Finland.5

Another major finding is the lack of hard core opposition to the welfare state. Where the survey question explicitly offers the option, the proportions preferring a reduction in social security are largest in Norway. However, over the entire period for which we have data, Norwegians were quite satisfied with the levels of social security provision. This was also true for the Dutch and the Swedes. By contrast, the demand for expansion was pronounced in Britain and Denmark, perhaps because, for most people, pensions are not income related.6

Finally, we examine the popularity profiles of three major programmes relating to the security dimension of the welfare state. From the ISSP surveys in 1985 and 1990 we have identical questions about whether or not the government should spend more on old age pensions, (p.213) health services, and unemployment benefits. For Germany, Britain, and Italy the questions were asked in both years, but only in 1985 for Austria and 1990 in Norway. The distributions are presented in Table 8.4. Before embarking on this analysis, however, we should keep in mind Beate Huseby's finding (Chapter 4) that people tend to respond negatively to general questions about increasing government spending but to react more positively to questions about spending in specific policy areas such as pensions and health services.

As we noted earlier, in Britain and the Netherlands, with unambiguous questions about which programmes to expand, support for both old age pensions and health services is relatively high compared to general support for social security and benefits. Nowhere do we see any sizeable opposition to government spending on either programme. In every country, except for pensions in Germany in 1985, support levels for increased spending on pensions and health services are well above 50 per cent. There is no sign of reduced willingness to spend on either programme; on the contrary, although small, the changes between 1985 and 1990 show the increasing popularity of both programmes. This is most prominent in Germany, where there is a ten‐point increase in support for spending more money on pensions, and a twenty‐point increase in favour of spending more on health services. The same tendencies are evident in Italy; the changes are smaller but preferences for more spending in 1985 are much higher than in Germany.

Unemployment benefits, on the other hand, do not enjoy the same level of popularity. Austria even demonstrates that in 1985 a majority favoured less spending on unemployment benefits. Still, in countries where the question is repeated, the basic profile is one of stability. Only in Britain do we find a noticeable change, a drop of five points among those who want to spend more.

An intriguing pattern is revealed, however, when we compare these countries. Only in Italy is there a majority—just—in favour of more government spending on unemployment benefits. In Germany and Britain, this proportion drops to 35 per cent. In Norway, it drops to less than 20 per cent; one in four even wants unemployment benefits cut. This clearly undermines the idea that the Scandinavian welfare states are rooted in stronger and more stable popular support than in less affluent countries which spend less on welfare. This is even more striking since unemployment, which was of no social or economic significance in Norway during the 1960s and 1970s, had become one of its political embarrassments by the late 1980s. (p.214)

Table 8.4. Support for Increased Spending on Old Age Pensions, Health Services, and Unemployment Benefits in Five West European Countries, 1985 and 1990

Programme

1985

1990

Old age pensions

Health

Unemployment

Old age pensions

Health

Unemployment

Austria

More

55

62

13

As now

41

32

35

n.a.

Less

4

6

52

N

1,471

1,474

1,475

Germany

More

46

52

35

55

73

37

As now

50

41

52

43

24

50

Less

4

7

13

2

3

13

N

1,021

1,019

1,006

2,724

2,756

2,719

Britain

More

75

88

41

82

90

36

As now

24

11

40

17

9

47

Less

1

1

19

1

1

17

N

1,484

1,497

1,465

1,167

1,172

1,144

Italy

More

75

79

56

80

85

52

As now

21

14

26

17

11

31

Less

4

7

18

3

4

17

N

1,539

1,551

1,511

969

976

959

Norway

More

72

83

19

As now

n.a.

27

16

57

Less

1

1

24

N

1,447

1,461

1,416

Notes: Entries are percentages. The question is: ‘Listed below are various areas of government spending. Please show whether you would like to see more or less government spending in each area. Remember that if you say “much more”, it might require a tax increase to pay for it.’

Sources: ISSP (1985, 1990).

(p.215) Social Protection and Social Position

Our next question concerns who supports the welfare state, and what differences we find between expansionary periods for the welfare state and recessionary times. The comparisons across countries and years are made using the variables and definitions described in the Appendix. The questions are from the British election studies (Table 8.1), the Cultural Change Study for the Netherlands (Table 8.2), and several national election studies for Denmark, Norway, and Sweden (Table 8.3). For Norway and Sweden, we can examine the demographic correlates of support for expanding social security as far back as the 1960s. Denmark, Britain, and the Netherlands provide such evidence for the 1970s and 1980s. The results are shown in Tables 8.5 and 8.6.

In both Norway and Sweden support for expanding old age pensions is clearly class related. From the 1960s to the 1980s people who identify with the working class are invariably more in favour of increased spending on pensions and benefits than almost any other group. No other variables have such consistently significant correlations. However, close inspection of the relationship indicates that the effects are weakening: there is a steady fall in the size of the coefficients as we move from the 1960s to the 1980s. But age, surprisingly, has an opposite impact in the two countries. Elderly people are more favourable towards expansion in Norway, although the relationship is significant only in 1973. In Sweden, it is the younger generation which supports expansion, and the impact seems to grow over the years. Also, contrary to our initial assumptions, education tends to induce support for spending. Indeed, in the mid‐1970s the well educated are among the most positive about expanding social security.

Two other groups show a similar willingness to support spending. People employed in the public sector are more favourable to expansion than those working in the private sector, which fits with the assumption of socialization in ‘public policy’ values. Again, employed people are consistently more hostile to social security expansion than those not in the labour force, which conforms with Rose and Peters's expectations. However, there is not the anticipated opposition as income increases. In view of the strong relationship between income and pensions in Norway and Sweden, few Scandinavians would be surprised that the well‐to‐do are rather satisfied. Lastly, note that Swedish women tend to be more positive about increasing pensions than men, although the relationship is statistically significant only in 1968. (p.216)

Table 8.5. Support for Old Age Pensions in Norway and Sweden, and for Increased Social Benefits in Denmark

Norway: old age pensions

Sweden: old age pensions

Denmark: social benefits

1965

1973

1985

1968

1973

1988

1979

1981

1984

1987

Gender

−0.006

0.032

−0.028

−0.090**

−0.038

−0.013

−0.008

0.058

−0.018

0.012

0.004

0.028

−0.011

−0.067**

−0.038

0.039

−0.013

0.058

−0.006

0.030

Age

0.017

0.106**

0.008

−0.076**

−0.043

−0.171**

−0.120**

−0.128*

0.058

−0.056

0.010

0.086**

0.015

−0.041

−0.016

−0.106**

−0.096**

−0.129*

0.038

−0.053

Income

−0.010

−0.020

0.044

−0.039

−0.007

−0.040

−0.042

−0.109

−0.036

−0.042

−0.023

−0.032

−0.071*

−0.037

0.014

−0.086

Education

0.007

0.093**

−0.015

0.048*

0.094**

0.028

0.130**

0.117*

0.096

0.017

−0.050*

0.015

−0.044

−0.004

0.065**

0.045

0.059

0.040

0.008

0.041

Employed v. non‐employed

−0.018

−0.023

−0.090**

−0.060*

−0.057*

−0.169**

−0.068

0.015

0.063

−0.052

0.002

−0.042

−0.066**

−0.028

−0.029

−0.022

0.014

0.055

−0.004

−0.005

Publicly employed

0.075**

0.028

0.055*

0.168**

0.034

0.076*

0.063**

0.025

0.050*

0.129**

0.028

0.065

Working‐class identification

0.170**

0.129**

0.124**

0.207**

0.155**

0.101**

0.246**

0.286**

0.159**

0.156**

0.105**

0.118**

0.192**

0.130**

0.076**

0.207**

0.238**

0.136**

R 2

0.028

0.021

0.019

0.049

0.028

0.049

0.065

0.081

0.020

0.004

N

1,334

1,058

2,023

2,398

1,917

1,811

930

453

465

993

(*) p < 0.05

(**) p < 0.01

Notes: Entries in the upper line are beta coefficients; in the lower line, Pearson's r. A dash indicates that relevant data are not available. The definition of variables is given in the appendix. The dependent variable is based on questions reported in Table 8.3. R 2 is adjusted.

Sources: national election studies.

(p.217) Looking at support for social benefits in Denmark reinforces the picture of a common pattern in Scandinavia. Again, identification with the working class is consistently the best predictor of support for spending, and during the late 1970s and early 1980s education has a positive impact on willingness to spend. Moreover, as in Sweden, it is the younger generation which wants to expand benefits, even if this inclination fades during the 1980s. We also find public sector employees consistently more supportive than those employed in the private sector. But, in contrast to the other Scandinavian countries, Denmark reveals no differences between the employed and people not in work. Note also that, again, there is no significant relationship to income. Considering that Scandinavians are among the most heavily taxed in Western Europe, this suggests that the Scandinavian model of income‐related pensions and benefits does not provoke opposition to tax‐financed social security systems.

Britain's social benefits are basically flat rate payments; any additional benefits come from private insurance or employers' schemes. Thus, there is not the direct connection with income—the higher the earnings, the higher the benefits—as in Scandinavian systems. Looking at Table 8.6, Rose and Peters's assumption about the effect of income appears to have some merit. Although data on family income are available only for 1979 and 1987, the negative relationships are quite striking. For those years for which we lack data on income there are negative relationships with education, although weaker than for income. Given the strong relationship between education and income in Britain,7 this suggests less willingness to fund social benefits among the well‐to‐do. The consistently negative attitude among tax payers in general, with the employed being more sceptical than those not in the workforce, reinforces an image of the British as reluctant taxpayers so long as they perceive that benefits are distributed to the less fortunate. This patterning of attitudes is well in line with the tax burden assumptions of Rose and Peters.

However, these preferences are consistently challenged by the younger generation. Contrary to the theories of both Wilensky and Inglehart, the young are not abandoning the welfare state, but, rather, demanding its expansion. Together with the young, and in opposition to the well‐off and well educated, the British working class look for larger pensions and benefits. This portrays British welfare politics as even more related to class differences than is the case among Scandinavians.

The data for the Netherlands reveals a third and new picture. For the (p.218)

Table 8.6. Support for Increased Social Benefits in Britain, 1974–87, and Old Age Pensions in the Netherlands, 1970–85

Britain: social benefits

Netherlands: old age pensions

1974

1979

1983

1987

1970

1975

1980

1985

Gender

−0.022

−0.035

−0.008

−0.023

−0.052

−0.027

−0.004

0.085*

0.023

0.017

−0.022

0.001

0.063*

−0.018

−0.001

0.062

Age

−0.167**

−0.098*

−0.197**

−0.205**

−0.004

−0.050

−0.054

0.008

−0.101**

−0.013

−0.135**

−0.120**

−0.020

−0.049*

−0.028

0.007

Income

−0.149**

−0.159**

−0.033

−0.094**

−0.061

−0.051

−0.180**

−0.120**

−0.045*

−0.081**

−0.071*

−0.105**

Education

−0.054*

−0.039

−0.038*

−0.044

0.037

−0.051

0.027

−0.055**

−0.077*

−0.039*

−0.012

0.005

−0.063

−0.050

Employed v. non‐employed

−0.134**

−0.080*

−0.106**

−0.048*

−0.009

0.042

0.000

−0.069**

−0.106**

−0.145**

−0.034

−0.018

0.035

−0.011

Publicly employed

Working‐class identification

0.115**

0.078**

0.154**

0.055**

0.057

0.015

0.004

0.114**

0.123**

0.114**

0.160**

0.095**

0.054*

0.036

0.054*

0.127**

R 2

0.041

0.044

0.055

0.048

0.005

0.007

0.006

0.018

N

2,215

1,418

3,450

3,084

1,345

1,324

1,380

1,238

(*) p < 0.05

(**) p < 0.01

Notes: See notes to Table 8.5.

Sources: British election studies; Cultural Change in the Netherlands.

(p.219) Dutch, there is little discrimination regarding preferences about spending on pensions. Indeed, for nearly all the variables, the relationships are less consistent and weaker than for both the Scandinavian countries and Britain. The upper‐income groups are reluctant about spending more, but the association tends to drop to non‐significance as we introduce statistical controls. Social class is significantly related to expansion only in 1985, the last year for which we have data. In that year, however, we find gender emerging as significant, with women being slightly more in favour of expansion than men. On the whole, the Netherlands seem to be without persistent tension in the field of old age policies. At the same time, no social group emerges as the political stronghold of welfare state support.

Expanding the Security Dimension

So far we have analysed only the structure of support for old age pensions, or policies which include social security as a major part of the programme. The three questions in the 1985 and 1990 ISSP surveys which we looked at earlier (Table 8.4) allow us to extend the analysis. The question about spending on pensions enables us to validate the inquiry so far, by bringing our data for Britain and Norway up to 1990 and introducing three additional countries: Austria, Germany, and Italy. Moreover, as the questions are identical in each country, the data allow us both to examine within‐country group differences and to compare such differences across countries. In Table 8.7 we report the structure of support for old age pensions in Germany, Britain, Italy, and Austria in 1985; and for Germany, Britain, Italy, and Norway in 1990. The multivariate results (beta coefficients) are shown in the upper row of each entry, the bivariate results (Pearson's r) in the lower row.

Some common trends are evident among these countries in 1985. Although the relationships are sometimes weak, they correspond broadly with generation theory as well as theories of tax frustration. In each country, older people are more likely to support pensions, especially in Britain; in Germany, only the bivariate relationship is significant. The positive impact of age in Britain, compared to Table 8.6, is best explained by the change from a general question about support for ‘social benefits and services’ to explicit support for ‘old age pensions’ in Table 8.7.

At the bivariate level, increasing education and income induce (p.220)

Table 8.7. Support for Old Age Pensions in Western Europe, 1985 and 1990

Austria

Germany

Britain

Italy

Norway

1985

1985

1990

1985

1990

1985

1990

1990

Gender

0.041

−0.003

0.042

−0.008

−0.032

0.027

0.046

0.086*

0.068

0.035

0.067*

0.020

−0.016

0.072*

0.068

0.113**

Age

0.117**

0.038

0.059*

0.147**

0.106**

0.105**

0.080**

−0.089*

0.160**

0.080*

0.120**

0.222**

0.193**

0.155**

0.133*

0.047

Income

−0.047

−0.012

−0.088**

−0.073*

−0.174**

0.034

−0.068

−0.083*

−0.112*

−0.080*

−0.170**

−0.252**

−0.243**

−0.090*

−0.139**

−0.139**

Education

−0.066

−0.071

−0.084**

−0.120**

−0.080*

−0.100**

−0.060

−0.012

−0.098*

−0.138**

−0.186**

−0.263**

−0.182**

−0.173**

−0.131**

−0.072

Employed v. non‐employed

−0.057

−0.079

−0.021

−0.031

0.003

−0.065*

−0.015

−0.076*

−0.144**

−0.115**

−0.109**

−0.177**

−0.135**

−0.130**

−0.072*

−0.115**

Publicly employed

−0.045

0.003

−0.004

0.021

0.049

−0.002

−0.085*

−0.063

−0.050

−0.007

−0.008

−0.040

Working‐class identification

−0.005

0.091*

0.132**

0.141**

0.089**

0.075*

0.126**

0.051

0.139**

0.206**

0.214**

0.143**

0.132**

0.158**

R 2

0.034

0.032

0.067

0.114

0.072

0.049

0.034

0.047

N

793

885

1,975

1,230

967

1,224

855

853

(*) p < 0.05

(**) p < 0.01

Notes: See notes to Table 8.5.

Sources: ISSP (1985, 1990).

(p.221) scepticism about the continued expansion of pensions, although education is significant only in Italy and Britain after statistical control. And in Britain income remains important for support. Furthermore, in line with Rose and Peters, we find that in all four countries employees are more opposed to expanding pensions than people not in work. Except in Austria, social class is relevant but gender has almost no impact. Where we have data on occupational sector, there is not the anticipated support among public employees towards public pension schemes.

Much the same pattern is reproduced when we analyse the corresponding set of variables for 1990. Increasing age continues to have an impact in all four countries. Note, however, that in Norway, after controlling for other variables, the impact changes, with younger generations being the most supportive of pensions. Increasing education and income continue to induce opposition to expanding pensions. Both variables are significant in Germany and Britain. Norway conforms to the other countries in exhibiting opposition from higher income groups. Lack of employment, of course, correlates with low income and—except in Norway—is significant only at the bivariate level. Moreover, for the second time, gender makes a difference in Norway where women are slightly more favourable towards spending on pensions than men. Finally, belonging to the working class continues to stimulate preferences for increased spending on pensions.

Some comments on our initial theoretical propositions are appropriate here. No single theoretical position alone can explain the pattern of support we found. Rather, several theoretical positions have to be taken into account, but all of them with weak explanatory capacity, as revealed by low correlations and rather low R 2 values. For some countries, the consistent connection with age, income, and/or education, gives some indication that both generational differences and tax frustration may be involved. So does the bivariate relationship for support among the employed—‘the general tax payer’. Both the correlations with age and employment could be explained by an ‘interest’ argument: people not in work receive the smallest pensions, whether the system is income related as in Scandinavia, work related as in Germany, or is a flat rate system, with insurance adding to pensions, as in Britain.

The correlations between class and spending preferences are obviously the strongest and most consistent, which supports the theoretical arguments about welfare politics as a matter of class differences. When gender is significant, Norway and the Netherlands being the clearest cases, women are slightly more supportive than men. Being (p.222) employed in the public sector makes the least impact, which leaves Scandinavia (see Table 8.5) as the only case giving credibility to the theory that people who work in public agencies have distinctive values and preferences.

Public health services provide security against unexpected, and sometimes very large, costs. In most countries, relying on public health services means free or inexpensive services, but traded off against a waiting period for treatment other than in emergencies. This is often thought to encourage people who can afford it to prefer private health services, and they are likely to be less concerned with continuous growth in the public sector. Our analysis of support for public spending on health services, presented in Table 8.8, tends to confirm this.

In 1985, both increasing income and education reduces support for spending on health services. Both variables have a significant impact in Austria, whereas education is the more prominent factor in Germany and Italy, while income is the better predictor in Britain. Social class continues to be related to support for public spending on health, but only in Germany and Britain does the relationship hold up in the multivariate analysis. In Britain, people employed in the public sector were rather more in favour of public health services, but in Germany these attitudes were more characteristic among employees in the private sector.

Five years later, in 1990, a very similar attitude structure persists. Income continued to hold back support for public spending on health services in Britain, while education had the same effect in Germany and Norway. Class identification has much the same impact as in 1985 in Germany, but rather less in Italy. Unfortunately, there are no data for class identification for Britain in 1990. The only new trend, which was just visible in 1985, is that women are more in favour of public spending on health services than men. This was most evident in Germany and Norway.

While the security problems related to old age and health affect everyone, unemployment involves only fractions of the population. Thus, attitudes towards unemployment benefits constitute a test of solidarity, since most people neither wish nor expect to be a beneficiary. The social profile of support for unemployment benefits in 1985 and 1990 are shown in Table 8.9.

In 1985, the most striking difference compared to support for pensions and health services is that the impact of age has changed completely. Now it is the younger generation in Britain and Austria which prefer increased spending. Another new aspect is the stronger and more (p.223)

Table 8.8. Support for Public Spending on Health in Western Europe, 1985 and 1990

Austria

Germany

Britain

Italy

Norway

1985

1985

1990

1985

1990

1985

1990

1990

Gender

0.030

0.039

0.075**

0.028

0.058

0.028

0.032

0.137**

0.056

0.074*

0.086**

0.044

0.070*

0.049

0.044

0.144**

Age

−0.033

0.053

−0.015

−0.044

−0.077*

−0.037

−0.018

−0.116**

0.007

0.067

0.043

−0.019

0.034

0.015

0.010

−0.058

Income

−0.086*

−0.045

−0.009

−0.085*

−0.148**

−0.037

−0.066

−0.035

−0.103**

−0.105**

−0.074**

−0.117**

−0.105**

−0.096*

−0.081*

−0.085*

Education

−0.081*

−0.162**

−0.136**

0.004

−0.012

−0.147**

0.002

−0.115**

−0.092*

−0.230**

−0.170**

−0.068

−0.047

−0.165**

−0.026

−0.098*

Employed v. non‐employed

−0.043

−0.026

−0.017

0.006

0.038

0.006

−0.029

−0.033

−0.070*

−0.045

−0.042

−0.047

−0.018

−0.052**

−0.032

−0.074*

Publicly employed

−0.041

0.042

0.080**

0.054

0.047

0.001

−0.096*

−0.007

0.056

0.037

0.014

−0.018

Working‐class identification

−0.050

0.086*

0.087**

0.166**

0.031

0.043

0.017

0.004

0.175**

0.146**

0.189**

0.105**

0.065

0.070*

R 2

0.014

0.059

0.039

0.042

0.018

0.027

0.003

0.036

N

804

885

2,000

1,242

970

1,231

862

863

(*) p < 0.05

(**) p < 0.01

Notes: See notes to Table 8.5.

Sources: ISSP (1985, 1990).

(p.224)

Table 8.9. Support for Unemployment Benefits in Western Europe, 1985 and 1990

Austria

Germany

Britain

Italy

Norway

1985

1985

1990

1985

1990

1985

1990

1990

Gender

−0.022

0.029

0.020

−0.029

−0.005

−0.005

0.111**

0.039

0.027

0.075*

0.044

0.035

0.047

0.051

0.165**

0.048

Age

−0.106**

0.007

−0.122**

−0.107**

−0.113**

−0.031

−0.067

−0.121**

−0.015

0.016

−0.062*

0.004

0.009

0.026

0.019

0.080*

Income

−0.097*

−0.067

−0.047

−0.197**

−0.179**

−0.007

−0.207**

−0.044

−0.132**

−0.118**

−0.130**

−0.290**

−0.198**

−0.102**

−0.216**

−0.116**

Education

0.009

−0.008

−0.004

−0.025

−0.047

−0.033

−0.024

−0.061

−0.048

−0.074*

−0.075*

−0.166**

−0.107**

−0.107**

−0.081*

−0.117**

Employed v. non‐employed

−0.120**

−0.037

−0.079**

−0.109**

−0.096*

−0.117**

−0.127**

−0.027

−0.111**

−0.084*

−0.074**

−0.206**

−0.145**

−0.137**

−0.129**

−0.059

Publicly employed

−0.011

−0.011

0.063*

0.075*

0.144**

0.025

−0.052

−0.055

−0.004

0.041

0.032

0.060

Working‐class identification

0.098**

0.094*

0.195**

0.194**

0.119**

0.007

0.203**

0.117**

0.112**

0.209**

0.292**

0.148**

0.089*

0.240**

R 2

0.031

0.017

0.059

0.134

0.050

0.032

0.076

0.068

N

784

872

1,972

1,217

948

1,202

846

838

(*) p < 0.05

(**) p < 0.01

Notes: See notes to Table 8.5.

Sources: ISSP (1985, 1990).

(p.225) consistently significant difference between the employed and non‐working population. In all four countries, those who work are less inclined to support increased spending on unemployment benefits. The impact of income and education is much the same as for old age pensions and health services: support for spending declines as income and education increases. But in each case, support is related to social class; working‐class self‐identifiers are more willing to support spending on unemployment benefits than those identifying themselves as middle class.

As we move to 1990 little changes. Including Norway in the analysis only confirms that the younger generation is more inclined to support unemployment benefits. But, in contrast, there is no such tendency in Italy in either year. We also continue to see a lack of solidarity towards the unemployed among the employed. Income and education are still negatively related to support in all four countries at the bivariate level, but at the multivariate level they are significant only in Italy and, as usual, Britain. Finally, class endures as an important determinant of welfare state support, although the impact in Italy weakened between 1985 and 1990. However, Italy reveals some exceptional impressions in 1990: women seem to be more in favour of unemployment benefits than men, and so are people employed in the public sector compared to those working in the private sector.

Considering our initial theoretical arguments, the generational explanations proposed by Wilensky and Inglehart cannot account for the patterns of support for unemployment benefits. On the contrary, the youngest age groups are the most concerned about security against unemployment, as well as those who—on circumstantial evidence—seem to have a sense of solidarity. However, Rose and Peters's arguments about tax burdens and frustration accord well with the general patterns of attitudes for both 1985 and 1990. As for other welfare programmes, social class seems to be the most consistent catalyst in generating support for increased spending. This tendency is evident from the mid‐1960s through until 1990 for at least three different policy areas related to the security aspects of the welfare state.

Welfare Politics

Today the welfare state is a common political project. Conflicts about the welfare state rarely relate to basic principles but to priorities about which needs have to be met and the extent of coverage. However, (p.226) historically the welfare state was associated with political parties which were concerned to protect workers against the fluctuations of markets, the dangers of the work place, and illness or infirmity. These were generally socialist, social democratic, or labour parties. Even if welfare programmes have been supported by other parties, and liberal as well as conservative parties have introduced welfare reforms, the notion of the welfare state as primarily a project of democratic socialism is well established.

Our final analysis is concerned with the relationship between party support and attitudes towards the welfare state: are supporters of social democratic—or labour—parties still more in favour of public spending on welfare than supporters of centrist or conservative parties? For this analysis, we divided voters into three groups: conservatives, including supporters of populist, ‘new right’ parties; centrists, which consisted of supporters of liberal, Christian, centre, and agrarian parties; and left voters, which embraced supporters of social democratic, socialist, left socialist, labour, and communist parties. This party preference variable was then introduced into the regression equations used for the ISSP 1985 and 1990 data. The results are shown in Table 8.10.8 Only the effect of party adherence is entered in the table since the effects of controlling for the other independent variables are displayed in the two previous tables.

Three main findings emerge. First, the effect of party politics is more important in some countries than others. The strongest relationship between party preferences and willingness to spend more on welfare programmes appeared in Britain, but there are similar effects in Germany. In Norway and Austria, party adherence is relevant only for spending on health services and unemployment benefits; supporters of different parties seem to agree about spending on old age pensions. Secondly, the importance of party politics varies across different welfare programmes. Support for unemployment benefits is most clearly connected to party preferences in all four countries. Then follows support for health services, and lastly backing for old age pensions.

Thirdly, whereas the earlier models resulted in fairly small R 2s, the introduction of party preference increases the explained variance quite substantially—even after controlling for class identification and social stratification variables such as income and education. Most prominent is the increase in the R 2 for Britain, especially on support for unemployment benefits in 1985 where the explained variance exceeds 20 per cent. (p.227)

Table 8.10. Party Adherence and Support for Increased Spending on Old Age Pensions, Health Services, and Unemployment Benefits, 1985 and 1990

Austria

Germany

Britain

Norway

1985

1985

1990

1985

1990

1990

Pensions

Beta

0.065

0.165**

0.069*

0.166**

0.211**

0.024

Pearson's r

0.074

0.176**

0.166**

0.225**

0.228**

0.060

R 2

0.027

0.056

0.075

0.141

0.111

0.055

N

619

689

1308

1094

846

667

Public health

Beta

0.085*

0.187**

0.148**

0.279**

0.258**

0.090*

Pearson's r

0.094*

0.192**

0.159**

0.324**

0.278**

0.122**

R 2

0.018

0.084

0.069

0.114

0.084

0.045

N

624

668

1323

1104

850

673

Unemployment

Beta

0.137**

0.232**

0.170**

0.306**

0.309**

0.165**

Pearson's r

0.142**

0.239**

0.162**

0.390**

0.334**

0.242**

R 2

0.053

0.061

0.085

0.207

0.139

0.102

N

609

661

1,304

1,083

832

655

(*) p < 0.05

(**) p < 0.01

Notes: The beta coefficients are controlled for gender, age, education, family income, employed v. non‐employed status, public employment v. private employment, and subjective class. The definition of the party adherence variable is given in the appendix. R 2 is adjusted.

Sources: ISSP (1985, 1990).

This suggests some conclusions relevant to our theoretical arguments. Even if class politics and political preferences do not present us with a single and conclusive explanation of support for the welfare state, this dimension—often overlooked in the last decade of political science analysis—is clearly the best predictor of welfare preferences, although still fairly modest.

Finally, we investigate whether the impact of party adherence has changed since the 1960s. Here, we have to go back to the questions asked in national election studies, so we have to allow for the slight variations in the questions asked. Not surprisingly, the countries for which we have the appropriate data are those where class politics and class parties have been most prominent: Britain, Denmark, Norway, and Sweden.9 Interestingly, these countries have all experienced some (p.228) attenuation of the relationship between class position and party preference (Worre 1980: 318; Franklin and Mughan 1978: 523–34). Among voters who would be classified as working class—according to their education, income, and occupation—fewer were voting for left‐wing parties by the late 1980s. However, socialist parties are often supported by young and well‐educated middle‐class employees in the public sector. The impact of party adherence over time is shown in Table 8.11.

Norway and Sweden provide the longest time series, and reveal slightly different patterns. In both countries, voting for left socialist or social democratic parties increases support for spending on old age pensions, but much more so in Sweden than in Norway. Indeed, there is remarkable stability in the relationship between party adherence and support in Sweden. If anything has changed it is towards the more pronounced importance of party adherence. This is explained by the fact that Social Democratic governments have introduced the major

Table 8.11. Party Adherence and Support for Increased Spending on Old Age Pensions in Denmark, Britain, Norway, and Sweden

Beta

Pearson's r

R 2

N

Denmark

1979

0.315**

0.348**

0.141

824

1981

0.349**

0.401**

0.186

392

1984

0.413**

0.380**

0.144

414

1987

0.268**

0.276**

0.074

857

Britain

1974

0.287**

0.306**

0.113

1,878

1979

0.297**

0.317**

0.120

1,186

1987

0.330**

0.364**

0.149

2,599

Norway

1965

0.191**

0.207**

0.044

1,023

1973

0.043

0.045

0.029

655

1985

0.197**

0.214**

0.045

1,256

Sweden

1968

0.334**

0.347**

0.138

2,314

1973

0.306**

0.286**

0.096

1,684

1987

0.366**

0.344**

0.162

1,605

(*) p < 0.05

(**) p < 0.01

Notes: See notes to Table 8.10.

Sources: National election studies.

(p.229) welfare reforms in Sweden, while in Norway the major programme—the National Insurance Scheme—was carried through by a non‐socialist coalition which included centrist parties (Liberal, Christian, and Agrarian) and the Conservative Party. In Norway, the impact of party is non‐significant in 1973 as the political consensus on welfare policies still existed. By 1985, however, as conflicts over welfare spending or tax relief became salient issues, political polarization reappeared.

Exactly the same picture emerges for Britain and Denmark for the 1970s and 1980s. Party adherence encourages welfare support, and there is no sign of the relationship weakening, except at the 1987 election in Denmark. And even then, the relationship is much stronger than most of the significant relationships revealed in this chapter. Although several theoretical arguments suggest hypotheses on the processes which determine expansionist or restrictive attitudes towards the welfare state, the results shown in Tables 8.10 and 8.11 indicate that welfare policies have been and remain a matter of ideological and political preferences. The conflicts and their solutions depend on party strength among the electorate and the popularity cycles of ideological positions.

Conclusion

At the outset of this chapter we presented a number of theories which have enjoyed prominence in attempting to explain the development of support for the welfare state. These theories allowed us to anticipate who are the supporters and the opponents of welfare programmes. One of our main conclusions is that no single theory alone provides reliable explanations or interpretations of the support patterns revealed. All the theories have some merit, and some findings are consistent with more than one theory. These observations suggest caution about accepting those arguments uncritically. We have to be aware of the shifting relevance of different theories as a consequence of national politics and the changing nature of the welfare state.

None the less, several conclusions seem indisputable. First, there is no evidence that welfare states, or specific welfare programmes, are generally losing support over time, as some theories predict. Rather, there are ups and downs in their popularity. The ‘downs’ seem to be connected to general changes in the national political mood. Vigorous conservative campaigns, sometimes leading to electoral victories, are (p.230) accompanied by a decline in the proportion of the public demanding expansion of the welfare state, or even an increase in the proportion supporting reductions in spending. But such campaigns appear not to create lasting opposition. Rather, the opposition created by conservative politicians tends to fade away as the novelty of their messages becomes yesterday's news and the public have direct experience of poorer services.

Secondly, to explain which groups promote or contest the welfare state, a complex relationship has to be considered. Several of the theoretical arguments are relevant but each explains only part of the process leading to different attitudes. Certainly, theories predicting a continuous decline in support for the welfare state failed when confronted with our evidence. None the less, the groups which are supposed to be primarily responsible for challenging the welfare state—the well educated and the rich—are, indeed, among the most sceptical about the expansion of old age pensions, public health services, and unemployment benefits.

The younger generation is also more sceptical than the older generation about old age pensions and health services, which accords with generation theories. But this pattern of support is also consonant with an interest interpretation: the older generation is particularly supportive due to their vested interests in old age pensions and ‘free’ health care. We also noticed that the young behaved quite differently over unemployment benefits, where they might be thought to have vested interests.

Finally, class identification and party adherence are stronger correlates of attitudes towards social security than any other variables. There is no substantial and systematic reduction in the impact of class identification on attitudes as we move from the mid‐1960s through to 1990. This indicates that welfare policies continue to be a matter of class politics, suggesting that theories of class formation provide a superior explanation for people's spending preferences on welfare programmes over the entire time period analysed. This conclusion is reinforced by the impact of party adherence. Electors who vote for leftist or working‐class parties are more inclined to support spending on welfare programmes than those voting for centrist or conservative parties, even after controlling for class position. This relationship, too, is remarkably stable from the 1960s until the early 1990s, and is by far the best predictor of spending preferences. In people's minds, there are differences in the welfare policies pursued by parties of different (p.231) political colour, and their preferences among the parties reflect their preferences about the development of the welfare state.

Appendix

The variables used in Tables 8.5–8.11 were coded as follows.

The dependent variable differed slightly according to the data set used. For the ISSP data, the question is: ‘Listed below are various areas of government spending. Please show whether you would like to see more or less government spending in each area. Remember that if you say “much more”, it may require a tax increase to pay for it.’ Three items were used: more or less spending for old age pensions; more or less spending for public health services; more or less spending for unemployment benefits. The codings are: spend much more = 4; spend more = 3; spend the same as now = 2; spend less and spend much less = 1. In national elections surveys, if possible an approximation to the four‐value classification in the ISSP is used. In some cases only a three‐fold classification was possible. The questions shown in the notes for Tables 8.1–8.3 indicate the codings used in each case.

  • Gender:

    Women = 1; men = 0.

  • Age:

    18–24 years = 1; 25–34 years = 2; 35–44 years = 3; 45–54 years = 4; 55–64 years = 5; 65 years and older = 6.

  • Income:

    The income distribution is divided into quartiles. Family income is used, except for Germany where only individual respondent's income is available.

  • Education:

    The education variable in the ISSP data is specific to each country. To allow cross‐national comparisons, three broad categories were imposed: basic education and incomplete secondary education = 1; secondary education completed and all other post‐school education, excluding a degree = 2; degree level = 3. The education data from the national election studies were recoded keeping as close as possible to the ISSP categories: basic education = 1; more than basic education but no college or university education = 2; college or university education = 3.

  • Employment Status:

    Employed = 1; unemployed or pensioners, housewives, students = 0.

  • (p.232)
  • Employment Sector:

    Works for the government or nationalized industries = 1; other occupations and unemployed = 0.

  • Class Identification:

    In the ISSP data, the coding was: upper middle class and upper class = 1; upper working class and middle class = 2; working class and lower middle class = 3; lower class = 4. In Italy, class position is based on an estimation of the socio‐economic position of the respondent by the interviewer. In the national elections studies, the coding was: self‐reported middle class = 1; forced‐answer middle class = 2; forced‐answer working‐class = 3; self‐reported working‐class = 4.

  • Party Adherence:

    In the ISSP data, the question is: ‘If there were a general election tomorrow which party do you think you would most likely support?’ The national election studies ask for vote intention just prior to the election or actual vote just after the election, depending on whether the survey is a pre‐ or post‐election survey. The coding is: parties classified or named as conservative or neo‐liberal/populist = 1; as liberal, Christian, agrarian or peoples parties = 2; as left‐socialist, socialist, social‐democratic or leftist = 3.

(p.233)

Notes:

(1.) With synchronic survey data it is difficult to discriminate, empirically, between theories claiming increased opposition among the younger generation and theories advocating increasing support among the oldest generation. The distribution of support would look similar if one or both sets of theories are correct. Only if genuine panel surveys were available could changes in individual attitudes be followed through different life‐cycles and circumstances.

(2.) The opposition to expansion at the 1979 election is confirmed by a slightly different question. The 1974 (October) and 1979 British Election Study included a question asking people if they thought social services had gone too far, were about right, or if welfare reforms had not gone far enough. In October 1974, 34 per cent thought that social services had gone much too far or a little too far (Taylor‐Gooby 1985: 27). In 1979, the responses were: ‘much too far’ and ‘a little too far’, 49 per cent; ‘about right’, 32 per cent; ‘not quite far enough’ or ‘not nearly far enough’, 17 per cent; don't know, 3 per cent. N = 1,864.

(3.) Support for the liberal parties increased by 5 percentage points to reach 28 per cent; support for the two socialist parties fell from 35 to 30 per cent. See Lane, McKay, and Newton (1991: tables 7.6 and 7.10). Note that the Netherlands does not have a conservative party in the British or Scandinavian tradition, but rather a multitude of religious, liberal, and socialist parties.

(4.) The very low level of support for expansion in 1973 was linked to a major debate about the abuse of social security rights. This debate also revealed strong suspicions inside the Labour Party that the disability pension scheme was being inappropriately used (Kolberg and Pettersen 1981).

(5.) However, we only have three data points for Finland. In view of the fluctuations over time in the other countries, we would have to be cautious about interpreting the 10 percentage point drop every fifth year between 1975 and 1985 as establishing a clear trend.

(6.) Except for people employed by the government in the public sector, state pensions are not income related in Britain and Denmark.

(7.) According to the ISSP data, the Pearson correlation between education and income in Britain is 0.45 in 1985 and 0.46 in 1990.

(8.) Italy is not included in Table 8.10 as the question about voting was not asked.

(9.) Finland is not included in the analysis as we were unable to secure the Finnish election survey data. The Netherlands is not included because the nature of the Dutch party system makes it difficult to use data from the Dutch Cultural Change Study to classify the parties in a way comparable to the other countries.