1 Introduction

In the eyes of the general public, some needy groups deserve more generosity than others, i.e., they are considered to be more deserving of welfare support. Existing research on popular deservingness opinions has identified the needy groups that are considered more and less deserving, the underlying criteria in welfare granting, as well as the individual-level determinants of such opinions (see for example Coughlin 1980; Van Oorschot 2000).

Despite the progress made, most of the empirical studies in the field have a static perspective, analyzing cross-sectional data for single years in time. There is hardly any research on longer-term trends in deservingness opinions and how these may be influenced by changes in contextual factors (but see Becker 2005; Soede et al. 2009). The lack of a dynamic perspective and analysis is unfortunate because welfare opinions generally, and deservingness opinions among them, most likely react to changing socio-economic, political and institutional developments in society (e.g. Blekesaune 2007; Erikson et al. 2002). Knowledge about such influences is essential for understanding the social context and processes of welfare opinion formation in a field in which the individual determinants of such opinions are increasingly known (Svallfors 2007). Importantly, at present we do not know how public opinion reacted to the ‘politics of austerity’ (Pierson 2001) that followed the end of the ‘golden age’ of welfare state expansion after the oil crises in the 1970s.

This leads us to two general research questions. The first is descriptive: How did welfare deservingness opinions change, if at all, in The Netherlands—our country case—during the period studied (1975–2006)? This general research question involves several sub-questions: Did popular welfare opinions coincide with general welfare retrenchment policies, becoming less supportive of granting welfare rights to needy groups? Or did welfare solidarity remain stable or—as a reaction to welfare retrenchment—even increase? How do these opinions fluctuate in the shorter term? Do possible long-term development or short-term changes in deservingness opinions hold for all needy groups, or are there differences in deservingness trends for different needy groups? Our second general research question is explanatory: To what extent can possible long-term developments and short-term fluctuations in deservingness opinions be attributed to changes in the economic, political, or institutional context? The contextual changes we investigate are economic changes in GDP and unemployment rate; changes in the political climate; and changes in specific welfare policies for target groups.

We investigate these trend questions with data from 22 repeated cross-sectional Dutch surveys, collected between 1975 and 2006. In addition to data availability, The Netherlands is an interesting country to study. During this period, a series of welfare reform measures were taken that focused strongly on stricter entitlement criteria for benefits to reduce the number of recipients and to emphasize people’s individual responsibility (Van Oorschot 2006; Yerkes and Van der Veen 2011). During the same period, The Netherlands was hit with two economic recessions, the first in the early 1980s, when The Netherlands experienced extraordinarily high unemployment in 1983–1984, and the second in the early 1990s, with periods of strong recovery in between. These developments make questions on trends in deservingness opinion and the influence of contextual-level factors relevant.

Our analyses focus on deservingness opinions regarding the target groups for five different benefits: the disability pension, old age pension, unemployment benefits, social assistance benefits and sickness benefits. Of the target groups, the old, the disabled and the sick are considered highly deserving, the unemployed less deserving, and people on social assistance least deserving (see for example Van Oorschot 2000). The different benefits offer different entitlements. For a proper understanding of our findings, some basic information on the character of the benefits is necessary. The Dutch old age pension is a universal, flat rate benefit at subsistence level, paid to all citizens 65 years of age and older, often topped by additional occupational pension and/or rent income. The social assistance scheme offers at most the same basic benefit amount as the old age pension scheme, but with the important difference that social assistance is means-tested and aimed at the poorest households, which do not qualify for any other benefit scheme. Unemployment, sickness and disability benefits are collectively organized workers’ insurance schemes. Unemployment insurance pays out non-means-tested, earnings-related benefits at 70 % of the previous wage. For those with short work records and for those whose earnings-related benefit duration has expired, the benefit is at a non-means tested flat rate subsistence level. The same is true for the disability pension, with age categories specifying the level of benefits received instead of work record. The sickness benefit has a statutory benefit level of 70 % of the wage, but in nearly all collective labor agreements this is topped up to 100 %, implying that being on sick leave has little or no negative income consequences for the sick employee. Sick pay can last up to 2 years, after which it is replaced by disability benefits if the employee is still unable to work. When claimants reach age 65, all other benefits expire and are replaced with the old age pension. Job seeking obligations apply to all persons who claim either unemployment benefit or social assistance.

We investigate opinions on these five benefits separately because contextual effects may depend on the aforementioned differences. For example, a higher unemployment rate may have a different effect on opinions towards needy groups that have a job-seeking obligation than other needy groups. Similarly, actual levels of deservingness may differ in relation to differences in the replacement rate. Yet, there may also be crossover effects: for example, policies affecting a specific benefit group may affect opinions towards other welfare benefits as well. Such crossover effects could indicate that the public views single policies in a broader social context. In brief, in this study we describe trends in Dutch deservingness opinions and relate fluctuations to changes in contextual factors.

2 Previous Welfare Opinion Trend Research

2.1 Long-Term Trends

The literature on welfare opinion trends is scarce and mostly concerns Scandinavian countries, The Netherlands, Great Britain, and the US. For the Scandinavian countries, researchers have put forth contrasting expectations about the long-term development of welfare opinions in the past few decades. Some expect them to go downwards, with the traditional high welfare support withering away due to the increase in individualistic values in society (Pettersen 1995; Wilensky 1975), while others expect welfare support to remain stable, because large groups in the Scandinavian countries have a vested interest in the comprehensive welfare state (Goul Andersen et al. 1999). Empirical studies in these countries have asked people about their support for government regulated income redistribution and whether the welfare state and specific benefits should be expanded, reduced, or maintained as they are. Findings support the expectation of stability: data from Sweden, Denmark, Norway, and Finland from the 1960s or 1970s to the mid-1990s, show a fairly stable or even slightly increasing long-term trend in support of welfare (Goul Andersen 1993; Goul Andersen et al. 1999; Martinussen 1993; Pettersen, 1995; Sihvo and Uusitalo 1995). In addition to long-term trends, this literature focuses on explaining the short-term fluctuations in opinions, an issue that we will address later on.

For the Dutch case, Becker (2005) analyzes the same longitudinal data that we use in this article, but he only analyzes the long-term trend, ignoring the numerous, substantive year-to-year fluctuations that we will focus upon. Becker (2005) finds that people increasingly feel that a number of benefits is insufficient. He interprets this finding as an increasing feeling of solidarity with the needy. As in Scandinavian countries, support for the welfare state and its benefits did not erode in a period of overall welfare retrenchment.

The situation seems to be different in Great Britain. Data from the British Social Attitudes survey show stability in the proportion of Brits who prefer welfare expansion between 1974 and 1987 (Pettersen 1995). However, for the subsequent period between 1987 and 2000, Hills (2002) analyzes the same survey and finds that the balance of people who agree that ‘Government should spend more on welfare benefits for the poor’ steadily dwindled. Hills also finds a drop in the percentage of people who support income redistribution, and argues that these opinion changes are in line with so-called ‘redistribution by stealth’, i.e., the implementation of policy measures that favor lower incomes but avoid the term ‘redistribution’. Furthermore, Hills argues that the British public’s increasing feeling that welfare suffers from fraud and creates disincentive to work is in line with the stricter activation policies that were put in place.

In the US, one of the most stable elements of public opinion is the unpopularity of welfare. Based upon published public opinion polls, MacLeod et al. (1999) and Weaver et al. (1995) show that between 1938 and 1995, a majority of the American public believed that the government spent too much on welfare. On the other hand, in this same time period, a steadily increasing percentage of Americans felt that it was the responsibility of the government to provide for the truly needy, although this number eroded slightly after 1987. However, a growing percentage of the American public also felt that welfare recipients were to blame for their poverty and, therefore, were not deserving of governmental support (MacLeod et al. 1999; Weaver et al. 1995).

To sum up, there is little information on longer-term trends in welfare opinions and even less information on opinions of the deservingness of specific target groups; however, it seems that trends differ between countries or types of welfare state. The latter implies that our findings from The Netherlands need to be put into perspective. We return to this point in the discussion section.

2.2 Short-Term Opinion Fluctuations and Contextual Factors

While changes in welfare opinions are at most modest in the long run, in the short run, changes seem to be much stronger. While the public may feel quite supportive of needy groups 1 year, this support may have changed substantially the next. In public debates and empirical research, short-term fluctuations are usually related to contextual factors that also fluctuate over the years. The factors cited include economic, institutional, and political. In this section, we will explain these relationships further and formulate hypotheses.

Starting with the relationship between economic context and welfare opinions, there is a debate in the literature regarding the direction of this relationship. Some authors who focus on general welfare state opinions suggest that economic downturn is associated with decreasing support (Becker 2005; Goul Andersen 1993). The proposed reason is people’s self-interest: when economic problems arise, people lose their sense of security, causing them to focus on themselves and to give less weight to the concerns of the disadvantaged (Durr 1993). Or, as Alt (1979) states, when people’s own economic situation is likely to decline, they become less altruistic, because ‘people are as generous as they can afford to be’ (Alt 1979, p. 184).

However, other authors hypothesize that the public is less confident about individual responsibility and more in favor of governmental support during economic down times (Blekesaune 2007). Additionally, during economically difficult times, which affect many, the general need for support becomes more obvious to all (Sihvo and Uusitalo 1995), and people would be less likely to blame benefits claimants for their situation (Bryson 1997; Fridberg and Ploug 2000; Hills 2002), which leads to increasing welfare state support (Shaw and Shapiro 2002; Soede et al. 2009). We note that this line of reasoning is especially evident in studies examining opinions on the deservingness of the unemployed, using the unemployment rate as a measure of the economic situation.

To us, the debate suggests not only that there is no full consensus about the direction of a possible economic effect on welfare and deservingness opinions at present but also that findings may be affected by the type of economic measure used. A similar remark is made by Erikson et al. (2002) who states that people’s opinion are affected by a ‘mix of potential economic maladies’ (p. 231), which can have opposite results. Therefore, for our analyses of the effects of the economic situation on Dutch deservingness opinions, we include two measures of the economic situation, economic growth and the unemployment rate, and formulate separate hypotheses for each.

With regards to the more general measure of the state of the economy, economic growth, we hypothesize that the self-interest perspective is correct: during economic downturns, people are faced with higher income risks and job risks and therefore focus more on their own self-interest and deservingness than on the deservingness of specific needy groups. We assume that people favor more attention being paid to economic recovery than to the particular needs of groups. In contrast, when the economy is strong, people’s own situation and their perception of it is likely to be better as well, allowing generosity towards others (Alt 1979; Durr 1993), which implies that economic growth makes people more likely to consider needy groups to be deserving of support (see also Becker 2005).

Our expectation regarding the effect of the unemployment rate on opinions is that when the unemployment rate rises—all else being equal—needy groups that are dependent on the labor market are more likely to be considered to be deserving of more. This may also be understood from a self-interest perspective. In times of high unemployment, the odds of people becoming unemployed themselves increases, making it in their own interest to consider needy groups deserving of more, especially (or exclusively) those who are unemployed (Fraile and Ferrer 2005). An alternative explanation from deservingness theory points in the same direction. This theory states that needy groups are considered more deserving when target groups are seen as less in control of their neediness (Van Oorschot 2000). With increasing unemployment, people may be less likely to blame the unemployed for being out of work and, therefore, may be more sympathetic towards their troubles (Bryson 1997). In addition, as Maassen and De Goede (1989) point out, when unemployment is high, people are more likely to have family members and friends that are out of work, making it easier to identify with jobless individuals and to understand their need. We assume that a rise in the unemployment rate increases opinions about the deservingness of groups that are regarded as part of the working population and that have a job-seeking obligation in particular. In the Dutch case, these groups include the unemployed and the social assistance beneficiaries (e.g., Soede et al. 2009). As old age pensioners and the (fully) disabled are not expected to find work and those on sickness benefit are still employed, we assume that their deservingness is less or not at all related to the unemployment rate.

The general political climate is another factor that researchers have suggested explains fluctuations in welfare attitudes (Pettersen 1995; Weaver et al. 1995). The argument is that in times when liberal ideas and right wing parties are stronger, popular ideologies are more liberal and thus put more emphasis on personal responsibility. Deservingness theory, in turn, suggests that, when needy people are seen to a greater extent to be responsible for their situation, their popular deservingness is lower. Empirically, researchers find that general dissatisfaction with welfare spending increases when the strength of right wing political parties increases. This relationship has been found in studies using American (Wlezien 2004), British, Danish, Norwegian and Swedish data (Goul Andersen et al. 1999; Pettersen 1995). However, the specific effect on deservingness opinions has not been studied yet. Our hypothesis is that needy groups are more likely to be considered to be deserving of support when there is a more leftist political climate in the country, while a more rightist political climate is related to needy groups being considered to be less deserving of support.Footnote 1

Finally, we take into account institutional factors, by which we mean specific policy developments. We have already seen that long-term trends in welfare opinions differ by country in the decades after the golden age of the welfare state. We describe the specific Dutch trend later in the results section. For now we concentrate on the effect that specific policy changes may have on the popular deservingness of specific target groups. That is, we relate opinion fluctuations to particular policy events. Although the literature generally assumes that policy events affect people’s opinions on related benefits and target groups (Hills 2002), empirical analysis is very scarce.Footnote 2

Soede et al. (2009) analyze how Dutch public opinion about unemployment and social assistance benefits reacted to restricting reforms targeting these benefits. They found that the tightening of benefit levels in 1985 was especially associated with higher numbers of people considering the benefits to be insufficient, but other reforms were only modestly or not at all associated with opinions. In a Danish study, Goul Andersen (1993) found that the freezing of benefits in the 1980s was associated with more positive attitudes towards the unemployed and social assistance beneficiaries. Therefore, the research suggests that reforms reducing the rights of beneficiaries are associated with increasing popular opinions about the deservingness of the related target groups.

We take this as our general hypothesis, stating that when there is a downward reform—i.e., a specific policy reform that makes a benefit less accessible and/or less generous—there is an upswing in popular opinion regarding the deservingness of related needy groups. This can be understood from deservingness theory: welfare retrenchment makes the public more aware of the needs of the welfare beneficiaries and the hardships they face [possibly due to increased media attention given to the events and the consequences thereof for the beneficiaries (Zaller 1992)], which increases the public’s perception of welfare deservingness. We do not expect an effect on opinions when a reform does not involve a clear change in the accessibility or generosity of benefits.

3 Data and Methods

3.1 Data

To answer our research questions, we use 22 waves of the Cultural Changes in The Netherlands (CCN) data, collected between 1975 and 2006. The CCN survey is a national representative survey of the Dutch public aged 16 and over, commissioned by The Netherlands Institute for Social Research. Each wave consists of approximately 2000 respondents; our pooled sample of analysis consists of between 27,002 and 38,594 respondents, depending on the analysis (see SCP 2010 for more information on the dataset).

3.2 Deservingness Opinions

Our main variable of interest is whether specific needy groups are considered to be deserving of more as measured by the following question: ‘I will give you a list of social benefits. Could you tell me for each of these if you think they are sufficient or insufficient?’ The listed social benefits are the Dutch disability benefit for workers, universal old age pension, the unemployment benefit, the social assistance benefit and the sickness benefit (or: sick pay for workers). Respondents were given the option ‘sufficient’ and ‘insufficient’ as answer categories. We consider the ‘insufficient ‘answer as indicating that the relevant target group is seen as ‘deserving of more’.Footnote 3 The response code ‘too good’ was offered when the respondent refused to choose between the sufficient and insufficient category. Because only few respondents choose this response code and because it indicates low rather than high welfare solidarity, we included this code in the ‘sufficient’ category.Footnote 4

3.3 Contextual Factors

The OECD Stat Extracts provides data on real GDP growth (i.e., the annual growth at constant prices in percentage) to measure economic growth, and unemployment rates (i.e., the number of people unemployed as percentage of the entire labor force, both employed and unemployed) for all survey years. For real GDP growth we used the growth at t = 0. Although information on real GDP growth for a certain year only becomes available when that specific year has ended, we expect the public to have a feel of the economic situation due to media coverage and prognoses. For the unemployment rate, we used the rates at t − 1 because we expect there to be a short delay before people are aware of the labor market situation and because the consequences are also often not immediately visible. To check our assumptions, we also carried out analyses with the t = 0, t − 1 and t − 2 scores on these two variables. These additional analyses did not change our main findings. We note that the correlation between real GDP growth (t = 0) and the unemployment rate (t − 1) is moderate (r = .433, p < .001), so that these economic measures measure two different things.

To measure political climate, we aggregated an individual level variable from the data asking respondents to indicate whether they see themselves as politically left or right on a scale of 1 (very leftist) to 5 (very rightist).Footnote 5 Because this question was not asked in 1979 and 1981, we imputed the average political climate for those years, and added a dummy variable to the analyses (1 = missing information; coefficients not shown in the tables). The correlation between (right-wing) political climate and the economic measures is moderate (with real GDP growth, the r = −.269, p < .001; with unemployment rate, r = 0.331, p < .001). We measure institutional factors, i.e., policy events, using a series of dummies indicating whether a specific policy reform took place in a specific year or not. Table 1 gives an overview of all reforms in the time period examined and the effect on the entitlements of beneficiaries (positive, negative or neutral).

Table 1 Policy events 1975–2006
Full size table

3.4 Micro Characteristics

In addition to contextual-level factors, individual-level factors are included in our analyses to control for composition effects. We include those factors that are likely to affect people’s deservingness opinions (age, sex, education, income, work status, left–right orientation). Such factors are commonly understood to relate to people’s structural position and life cycle, indicating the personal interest they have in welfare benefits and provisions (Svallfors 2007). However, the results of the individual-level factors are not reported (results are available upon request) because the focus of this research is on the context factors and no hypotheses concerning the individual characteristics are formulated. The descriptive statistics for the independent variables for the pooled sample can be found in Table 2.

Table 2 Descriptive statistics of the independent variables
Full size table

3.5 Methods

After a descriptive analysis of the long-term trends in deservingness opinions for five different benefits and related target groups, we examine the effect of contextual factors on short-term opinion fluctuations by carrying out two explanatory analyses for each benefit. First, we use multilevel logistic regression analyses to examine the relation between different contextual factors and the deservingness of needy groups. The two levels that are distinguished in this model are individuals and survey years. In the second analyses we examine to what extent policy events are related to deservingness opinions. We use logistic regression analyses and include a dummy variable for each policy event year (cf. Soede et al. 2009), while controlling for the individual and economic contextual variables. We include this dummy variable (event year = 1, other years = 0) in the analysis for the specific benefit for which there was a policy event but also in the analyses for benefits that did not have a policy event that year. That way, we can test whether the event year of one benefit only affects the corresponding opinion, or whether there is crossover effect of the policy event, i.e., whether a reform also affects opinions on other benefits. The analyses of institutional effects are restricted to the period 1985, when the reforms started, until 2006.

4 Results

4.1 Long-Term Trends

Figure 1 shows the percentage of the Dutch public that feels that recipients of disability pensions, old age pensions, unemployment benefits, social assistance benefits and sickness benefits are deserving of more support than they receive for the time period between 1975 and 2006.

Fig. 1
figure 1

The percentage of people who believe that recipients of benefits are deserving of more, 1975–2006

Full size image

When looking at the entire time period, Fig. 1 shows increasing popular deservingness as a long-term trend. Whereas during the late 1970s and early 1980s of the past century between 5 (sickness benefit) to 35 % (old age pension) of the Dutch population thought needy groups deserved of more support, during the mid-1990s, these numbers were, respectively 25 (sickness benefit) and 65 % (social assistance). This suggests a strong overall increase in welfare deservingness opinions.

However, closer inspection of the trend lines reveals that the increase was not steady but sudden. Notably, a tipping point seems to exist in the early 1980s. Before this time, deservingness opinions were rather steady (and even decreasing in case of old age pensions), but in the early 1980s, at the height of the (oil-price-shock induced) recession in The Netherlands, a change took place, lifting the opinions to a new and overall higher level. The opinions then more or less stabilized on that higher level from the mid-1980s until the end of our time line in 2006. Therefore, we conclude that welfare deservingness opinions remained rather stable during the period investigated, with the exception of an upward ‘shock’ in the early 1980s making opinions more generous.

We analyzed the long-term trend using multinomial regression analysis, including a linear independent variable measuring the year of the survey and controlling for individual characteristics (results not shown). For the entire period, we find a significant positive year effect for all benefits (b = 0.039 to b = 0.064 depending on the benefit). However, when we exclude the years before 1985, there is a small positive effect left for the disability benefit (b = .015), the elderly benefit (.019), and the sickness benefit (.046), indicating only a slightly increasing long term trend, and insignificant effects in the case of the unemployment benefit (b = −.002) and a very small negative effect for the social assistance benefit (b = −.005).

Furthermore, Fig. 1 shows that deservingness is especially high for the high deserving target groups of pensioners and disabled workers, as well as for social assistance claimants. The latter seems to reflect that the Dutch population is aware of the means-tested minimal character of these benefits, whereas the other benefits are non-means-tested (being either flat rate, in the case of pensions, or wage-related in the other cases). Sickness benefits are least likely to be seen as insufficient, which may be related to the fact that in most cases the statutory benefits of 70 % of the wage is topped up to 100 % by collective labor agreements.

4.2 Short-Term Opinion Fluctuations and Contextual Factors

In addition to the (shock wise) long-term trend, Fig. 1 displays considerable short-term fluctuations in the deservingness opinions between 1975 and 2006. Before we attempt to explain these fluctuations, we first assess which proportion of variance is accounted for by the survey years (the group level) using intraclass correlation (Snijders and Bosker 1999).Footnote 6

The results (see Table 3) show that between 7 and approximately 12 % of the variation in opinions (depending on the benefit at issue) is related to the specific year of interviewing, and thus possibly to context factors. We also see that the variation in pension deservingness is least dependent on year-related context factors (6.0 %), while social assistance deservingness is most affected by it (11.7 %). Now we turn to the analysis of context effects.Footnote 7

Table 3 Multilevel logistic regression analyses, deserving of more versus sufficiently deserving: effects of context factors, 1975–2006
Full size table

4.2.1 Economic Changes

How can economic, political, and institutional change in The Netherlands account for fluctuations in deservingness opinions? Table 3 reports the results of multilevel analyses, in which contextual effects are estimated controlling for individual-level determinants.Footnote 8 When examining our first economic indicator, economic growth, the results in Table 3 (model 1 and, net of the unemployment rate, model 3) show that higher economic growth increases the odds of finding the related needy groups deserving of more support for all five benefits. This confirms our hypothesis, which suggests that, when the economy is prospering, people are more generous towards the less well-off, while in times of economic downfall people seem to be more worried about themselves and restrict the deservingness of specific target groups. This relationship is also clearly visible in Fig. 2, where the real GDP growth is added to the previous graph showing the percentage of the Dutch public that is of the opinion that the various needy groups are deserving of more support.

Fig. 2
figure 2

The percentage of people who believe that recipients of benefits are deserving of more, 1975–2006 (left axis), and the real GDP growth (right axis)

Full size image

However, the effects of economic growth are diminished when taking into account the political (right) climate in model 5 (note that the correlation between both context variables is −.269), and, in the case of unemployment and sickness benefits, causing into lose significance. This indicates that the effect of economic growth is partly ‘political’.

Our second economic indicator, the unemployment rate, shows effects that run counter to GDP growth rates. When unemployment is higher, the popular deservingness of unemployed people and social assistance beneficiaries is higher, as indicated by the positive unemployment rate effects in Table 3 (model 2). In model 3, the effect of the unemployment rate is hidden by economic growth, but it is again visible when political climate is taken into account in model 5. Both the unemployment benefit and social assistance benefit are work-related benefits to which job seeking obligations are attached. An explanation of the finding could be that in times of higher unemployment the public at large is more aware of the fact that unemployed and social assistance claimants have more difficulty in finding jobs, and therefore are less to blame for their inability to escape their neediness. In addition, the public may also be more aware of the level of need that they face, because, for example, family or friends may experience unemployment. This interpretation is strengthened by the lack of effects of the unemployment rate on the deservingness opinion regarding disability, pension and sickness benefits.

4.2.2 Political Climate

The second contextual factor that may affect welfare deservingness opinions is the political climate. We expect needy groups to be considered less deserving when the political climate is more right wing. To control for a possible composition effect, we control for individual political left–right orientation. Table 3 reports that for four out of five benefits, the political climate significantly affects the likelihood of considering needy groups to be deserving of more. This means that in a more right-wing political climate, both left-oriented and right-oriented people are less likely to be generous about the needs of these groups, indicating they are less deserving of more support. This is true for all but the old age pension, for which no significant effect was found. Apparently, the level of the Dutch old age pension and the related deservingness of pensioners are beyond ideological divide. This may be due to its universal nature, covering all Dutch citizens regardless of income and status, in contrast to the other benefits, which cover mainly the working population and poor sections of the population. The finding also corresponds with the earlier mentioned consistent high ranking of old aged people as highly deserving needy group.

4.2.3 Policy Events

To see if the opinion fluctuations seen in Fig. 1 are a reflection of social policy reform measures, we look at the specific reforms that were put in place for the different benefits. Table 1 presents an overview of all reforms over the years (see Van Oorschot (2006) for a detailed discussion of each of the changes and the socio-economic contexts in which they were implemented). As the table shows, most changes in entitlements for Dutch social security benefits have been retrenchments. However, not all retrenchment reforms have been equally substantial. Reforms of the disability pension and unemployment benefit have significantly curtailed the level and duration of these benefits, and as such, these reforms have been very visible in the public debate. Reforms of old age pension and social assistance have had less drastic effects on entitlements, but they have had a high public visibility. In the case of old age pension, this is related to the fact that pensioners are seen as a highly deserving group, which ensures that any change to their benefits receives significant media attention. In the case of social assistance, attention in the public debate is often large because the benefit is regarded as a subsistence minimum, where any curtailment may have important effects on poverty rates.

Comparing Table 1 with Fig. 1 makes it clear that there are more opinion fluctuations than policy events, which, as a first observation, implies that opinions can fluctuate in the absence of specific policy reforms. However, some reforms may have had an effect. To statistically test this, we analyze the effects of policy reform events on opinions about the relevant benefit as well as their possible effect on the opinions regarding the other benefits (crossover effects). The results, presented in Table 4, show whether public opinion is different from the average of the other years in a policy event-year (as represented by the bold coefficients), which may indicate a policy effect. Two general observations can be made as to the patterns seen in Table 4. First, only half of the event years have a significant effect on the opinions when one was expected. For instance, for the disability benefit, there were 8 years in which a policy event took place (in bold). For two of these event-years (1993 and 2006), the results are as expected: There was a downward reform (see Table 1), and the target group are considered to be deserving of more than the average in the entire time period examined here (although the effect was only marginal in 2006, see Table 4). In 2002 (non-directional reform) and 2004 (both upward and downward reforms), we did not expect an effect, and none was found. In the remaining event years for the disability benefit, an effect was either found but not expected, or vice versus. Similar descriptions can be given for the other benefits and event-years: At most half of the policy reforms have the predicted effect. Yet, when effects are found, they are mostly in the expected direction (downward reform and an upswing in deservingness).

Table 4 Logistic regression analyses on deservingness opinions: coefficients event years, 1985–2006, deserving of more versus sufficiently deserving
Full size table

Second, policy events seem to have crossover effects. That is, a policy event directed at a specific target group may affect opinions towards target groups that were not the focus of the specific policy event. For instance, stronger work record requirements were put in place for the unemployed in 1987. It seems that the public opinion responded to this policy event, and people were more likely to feel that the unemployed were deserving of more. However, in the same year there was also an increase in the deservingness of social assistance and sickness benefit claimants even though there was no policy change for these groups in 1987. The event-year 1996 provides another example. That year, there was a policy event for only the social assistance (simplified rates) and the sickness benefits (privatization), but significant relations are found for all five benefits. The existence of these crossover effects, together with the finding that almost half of the policy events do not affect opinions on specific target groups despite expectations to the contrary, lead us to conclude that policy effects are mostly modest and if present often complex.

5 Discussion and Conclusions

Our analyses of the dynamics of welfare deservingness opinions in The Netherlands have shown some interesting findings. First, when looking at the long-term development of deservingness opinions in the period 1975–2006 (regardless of the short term fluctuations), we find a rather stable long-term trend. This long-term stability is remarkable given the welfare retrenchment that occurred in The Netherlands. Similar to the findings in the Scandinavian countries, a downward trend in public policy is not clearly reflected in distinct increasing or decreasing deservingness opinions. The exception to this long-term stability is the shift towards more generous attitudes in the early 1980s. It is tempting to attribute this shift to the economic recession that was at its height in The Netherlands in that period, yet other changes during this period—political and institutional—may also have been responsible for the sudden shift in opinions. In addition, it is unclear whether it was the state of the economy—as measured by GDP—or the level of unemployment that affected this shift.

Second, despite the long-term stability, we find considerable year-to-year fluctuation in deservingness opinions. The public could be quite lenient in 1 year and radically change their opinions the next. These year-to-year fluctuations in deservingness opinions provided us with the opportunity to test effects of changes in the economic and political climate on deservingness opinions, as well as the effect of policy reforms. Our third general finding arises from these explanatory analyses. We find the predicted economic and political effects on deservingness opinions, but policy effects are limited and often occur in an unpredicted way. With regards to the economic climate, we find that economic growth makes opinions more generous, increasing the perceived deservingness of the disabled, the elderly, and social assistance beneficiaries. This can be understood from a self-interest perspective: During economic downfall, people are faced with higher income and job risks and therefore may focus more on their own self-interest and their own deservingness instead of on the deservingness of specific needy groups, whereas, during economic good times, people’s own situation and their perception of it is likely to be better as well, allowing more generosity towards others (Alt 1979; see also Becker 2005; Durr 1993). However, a higher unemployment rate increases the perception of needy groups as deserving. This holds only for the needy groups that are most dependent on a good labor market, the unemployed and social assistance beneficiaries. Because we control for many of the self-interest variables (education level, work status, income), we interpret this findings from deservingness theory: with an increasing unemployment rate, the extent to which the jobless are seen as responsible for their needy situation decreases, and the public can more easily identify with these individuals and the needs they face, leading to more generous deservingness opinions. As to the political climate, we find that when the political climate is more right-wing, people are less likely to consider needy groups as deserving of more support. As we controlled for individual political orientations, this contextual effect means that both left- and right-wingers shift to less generous deservingness opinions in more right-wing times.

As to the policy events, we found only limited indications that these events were responsible for fluctuations in deservingness opinions. There are more fluctuations in the opinions than could be explained by these events alone, and statistical analyses of the effect of policy events showed that deservingness opinions differed from other years in only a little over half of the event years. In addition, when a policy effect was found, it often effected not only opinions about the needy groups directly concerned by the policy event but also opinions about other needy groups. That is, specific policy events had crossover effects, affecting opinions on needy groups that were not the target of the policy.

The limited indications we found of direct relations between specific policy events and specific opinions may be explained by people’s lack of knowledge about policy changes. Alternatively, policy events may affect deservingness opinions in a different year from the year in which it occurs; perhaps effects instead appear the year after or before a policy event. However, additional analyses with different time specifications (event change in year t − 1, t − 2, t + 1) did not change findings. This underlines that the effects of policy reforms on welfare opinions are limited—just as the stable long-term trend in these opinions taught us—and that its effect must not be overrated. Additional evidence for this conclusion is found in the crossover effects of policy. The existence of these effects implies that future welfare opinions research should not only focus on policy directly affecting the welfare group investigated but also policy reforms affecting other groups. Additionally, future studies could measure opinion changes over shorter time spans than we are able to here. For instance, it might be the case that opinions change temporarily in the month of the reform, when it is given a lot of media attention, but return to the status quo soon thereafter.