ABSTRACT
This article analyses the variation in public attitudes towards welfare recipients according to economic cycles. This question is of great importance at the time of recent fiscal crisis across Europe. Previous research has indicated that economic recession leads to declining welfare state support, although some findings show stronger support during times of increased financial strain. Finland provides ample opportunity to investigate the effects of situational factors, as it experienced two severe economic downturns during the period of examination. Unlike many previous studies, we also considered whether the association between economic cycles and attitudes was dependent on socioeconomic status. Analyses reveal that high unemployment and economic downturns diminish public support for welfare provisioning. Furthermore, the results show that both individual and spousal unemployment, as well as social class position, are associated with people's attitudes towards welfare recipients. These associations remain when the national unemployment rate and GDP growth rate are taken into account. Overall, the working class has stricter opinions of welfare recipients during economic downturns, whereas the opposite is true for the service classes and the self-employed. Moreover, men have stricter attitudes towards welfare recipients during economic recessions.
1. Introduction
Claims about the pervasive effects of welfare benefits on individual initiative have existed for as long as the benefits themselves. Inevitably, the debate zeroes in on whether those who apply for welfare are truly needy or merely lazy and unenterprising. During economic up- and downturns, this discussion on public support for welfare policy and for those relying on welfare benefits may become particularly current. At present, this debate about the appropriate response to the banking and fiscal crisis is of enormously essential across Europe.
It is reasonable to assume that economic cycles do not only affect people's life circumstances but also have an impact on people's viewpoints and attitudes towards state welfare. Empirical evidence on whether and how public attitudes supporting welfare policies change over time, especially during economic ups and downs, has so far been limited. Interestingly, prior research provides conflicting results. On the one hand, the literature claims that economic recession leads to declining support for welfare policies (e.g. Alt 1979; Sihvo and Uusitalo 1995). On the other hand, some findings indicate stronger support for welfare during times of higher unemployment and increased financial strain (e.g. Blekasaune and Quadagno 2003; Blekesaune 2007).
This article tests hypotheses on the relationship between economic development and public support for welfare policies by analysing changes in attitudes towards welfare recipients (i.e. people receiving social assistance benefits) in Finland from 1995 to 2010. The Finnish case provides a good opportunity for studying the influence of macro-economic and social factors on these attitudes. Over the past 20 years, Finland has experienced two severe economic crises: the Great Depression in the 1990s and the still-unfolding global financial crisis that began in 2008. Crises were the most drastic ones among the OECD countries.
Social assistance is a means-tested benefit targeted at low-income families. With regard to selective welfare programmes, earlier studies have revealed a consistent picture: marginal policies targeted solely at low-income families are less popular than comprehensive ones (e.g. Coughlin 1980). This finding is also accurate for Finland (see e.g. Kangas 1995), and therefore we should expect to observe more contrasted variation in these attitudes than in other welfare policy-related opinions. In relation to other countries the Finns exhibit stronger preferences for the state responsibility than the EU countries on average (Pfeifer 2009), but in comparison to the other Nordic countries the Finns are more likely to endorse individualistic explanations of poverty (van Oorschot and Halman 2000; Albrekt Larsen 2006) and have more individualistic views on the unemployed (Kallio et al. 2013).
Public views on the welfare recipients are important not only on academic grounds but also in terms of policy-making. There is longstanding evidence on public attitudes influencing policy-making and vice versa (e.g. Page and Shapiro 1983). If the public support for welfare recipients weakens during a recession, it can lead to cuts in benefit levels and tighter eligibility rules. This can in turn have negative implications for the recipients’ individual welfare but also potentially for the economy, if cuts in welfare reduce demand at the same that the economy is at its weakest. As if the public support strengthens, it may result in more effective anti-poverty policy and foster recovery both at individual and aggregate levels. Understanding these dynamics is crucial in the current political and economic climate.
This article follows more recent research on welfare attitudes, which have increasingly turned to contextual, aggregate explanations and aims to investigate to what extent economic development shapes public support. The following question is posed: Do attitudes towards welfare recipients swing with economic cycles? Particular interest is given to socioeconomic differences, to which earlier studies have paid less attention. The risk of unemployment and economic strain varies according to social class during economic downturns (e.g. Goldthorpe and McKnigh 2006), and therefore, it is reasonable to assume that the association between economic cycle and attitudes also varies according to socioeconomic status. More precisely, we posit our research questions as follows: (1) How are individual-level factors associated with attitudes concerning welfare recipients? (2) How are changes in GPD and the unemployment rate associated with these attitudes? (3) How are individual- and macro-level factors associated?
The data for this research are drawn from cross-sectional surveys compiled by the Department of Social Policy at the University of Turku over five subsequent years, starting in 1995. In the survey, different claims on welfare recipients were posed to the respondents. They were asked whether welfare recipients are truly in need of support, whether applicants make fraudulent claims, and if many who depend on social assistance are lazy and lacking the willingness to solve their own problems. A summed index measuring public support was constructed on the basis of the three claims. Our study began with descriptive analyses of trends in attitudes. Thereafter, the study was completed in two stages. First, individual-level analyses were done to indicate what individual characteristics are associated with certain attitudes towards welfare recipients. Second, the analyses were extended to include data on economic development (GDP annual change) and the country-level unemployment rate to assess changes in support over time.
2. Theoretical background and previous research
2.1. Individual-level versus country-level explanations
Attitudes towards welfare state policies have traditionally been studied at the level of individuals. In these studies, individual-level characteristics are usually viewed as indicators of self-interest. The self-interest argument states that those who are more exposed to social risks are more likely to hold positive attitudes towards welfare state policies than those who are less likely to receive benefits. The theory holds that the demand for state support reflects individuals’ socioeconomic positions and their exposure to social risks (see e.g. Hasenfeld and Rafferty 1989; Linos and West 2003). Social assistance benefits, which are targeted at the poor, are favoured more by those with lower levels of education, lower income and weaker labour market positions (see e.g. Pfeifer 2009).
Social assistance is closely linked to poverty, and therefore, public support for social assistance echoes public attitudes towards the poor and is linked to the discussion about deservedness (see e.g. van Oorschot 2000). The differences in public perceptions of poverty are also explained via the self-interest argument. However, the literature holds contradictory findings. Some studies have found that people experiencing financial hardship consider the poor more favourably (e.g. Kluegel 1987) and that poor people are more inclined to support structural explanations for poverty (e.g. Niemelä 2008), whereas other research reveals that claimants of public welfare and those in lower socioeconomic positions have more negative views of people receiving benefits (Golding and Middleton 1982; Schneider and Ingram 1993; Bullock 1999).
Alternative to self-interest, the variations in the attitudes on welfare have also been explained via ideology differences (e.g. Jæger 2006). This theory states that people's political values and beliefs guide their opinions on welfare policies. Due to the lack of proper measures, these differences lie beyond the scope of the current paper. Regrettably, the survey we used did not ask for the respondents’ political leanings. However, it can also be argued that self-interest explains both political and welfare policy attitudes.
More recently, research has turned from individual-level explanations to country-level explanations. This line of research is driven by the reasoning that public attitudes are affected by collectively experienced factors. Collective attitudes are viewed as products of the institutional characteristics of welfare policies (e.g. Papadakis and Bean 1993; Arts and Gelissen 2001), but as regime theory has provided inconclusive results, studies in this field have adopted an alternative, country-level approach using other macro-sociological factors – typically economic factors related to economic cycles or income dispersal (Blekesaune 2013). This article concentrates on economic factors: what impact, if any, economic cycles (measured via GDP change) have on welfare state support, particularly on attitudes towards welfare recipients. Additionally, we consider whether unemployment has an independent effect of its own. Naumann et al. (2016) found that job loss led to an increased support for public provision of unemployment benefits.
2.2. The impact of economic cycles
Existing studies show that the state of the economy has, on many occasions, had an effect on people's attitudes towards welfare policy. These studies indicate that contextual economic factors may work in two directions, although a few studies found no effect of economic factors on people's attitudes towards welfare policies (e.g. Kluegel 1987; Jaeger 2013).
On the one hand, researchers have claimed that public support for the welfare state tends to rise during ‘bad’ economic periods of low employment (e.g. Cutright 1965; Lipset 1968). Recent empirical work lends support for this hypothesis, as the results indicate that economic growth is negatively correlated with a stronger demand for state involvement and redistribution (e.g. Dallinger 2010; Dion and Birchfiled 2010; Jaeger 2013) or with individualistic explanations for poverty (Albrekt Larsen 2006; Kallio and Niemelä 2014). These results are in line with predictions from the governmental protection hypothesis, which argues that in times of economic decline, people tend to believe that governments should shoulder more responsibility for economic provisioning (Blekesaune 2007). Furthermore, these studies agree with the self-interest explanation that the overall level of social risks in one's country also affects public attitudes towards welfare policies. Based on the existing literature, it can be hypothesised that economic growth is negatively associated with public support for social assistance.
H1: Economic decline is positively associated with attitudes towards welfare recipients (the government protection during economic decline hypothesis).
On the other hand, the literature suggests that altruism declines in times of economic hardship. Alt (1979) argues that people tend to be less generous when they experience a decline in their economic fortunes. Durr (1993) states that economic worries cause people to focus on self-interest and thus give less weight to the concerns of the disadvantaged, presumably because people realise that their taxes are being spent to benefit people other than themselves. These arguments underline that economic declines cause people to reconsider the expenditure side of the welfare state, owing to the rising costs of welfare expenditure and the increased tax-burden. This hypothesis (here named the economic growth hypothesis) has also been bolstered by empirical support; Sihvo and Uusitalo (1995), among others, found that economic crisis reduced support for the welfare state, while more affluent periods gave rise to increased support (see also, e.g. Blomberg and Kroll 1999; Forma 1999). The following hypothesis can thus be presented:
H2: Economic growth is positively associated with attitudes towards welfare recipients (the economic growth hypothesis).
2.3. The socioeconomic differences in unemployment
Previous studies concerning welfare state support and economic cycles seem to present contradictory results. We argue that these mixed results could be explained by including other factors that distinguish periods of growth and decline from each other. One of these is the size of the population at risk of unemployment, which may vary according to both the time period and the socioeconomic groups.
Blekesaune and Quadagno (2003) note that the level of unemployment influences public opinion; in countries with higher unemployment rates, public support for welfare policies is generally higher. They assume that high unemployment increases public empathy for the unemployed through several mechanisms. First is the risk of becoming unemployed; the higher the unemployment level, the more people who are confronted with the possibility of becoming unemployed and thus the more people who are supportive of welfare state policies. Second is the concern for those who are unemployed; when people have unemployed friends, relatives and other peers, they are more likely to be concerned about the economic welfare of others. Third, in situations of high unemployment, politicians tend to place unemployment on the national political agenda, which is likely to improve public attitudes towards welfare policy. Other studies have provided similar results (Fraile and Ferrer 2005; Blekesaune 2007; Pfeifer 2009; Dallinger 2010).
Thus, we provide a second version of the government protection hypothesis, which does not concern the economic cycle in general but rather unemployment specifically (Blekesaune and Quadagno 2003):
H3: Attitudes towards welfare recipients positively correlate with the unemployment rate (the government protection during unemployment hypothesis).
The previous hypotheses only consider macro-level explanations influencing opinions. At the same time, however, explanations for the opinions are also observed at the individual level. The hypotheses also utilise different mechanisms: the government protection hypotheses point to empathy, whereas the economic growth hypothesis stresses altruism.
Regarding the government protection hypotheses, we assume that those most likely experiencing unemployment during economic downturns, especially in the working class, should be the most supportive towards welfare policy, and the difference should become smaller as the unemployment rate falls. We expect that if the hypothesis finds support, it may be due to self-interest based on one's unemployment risk and one's empathy towards the unemployed. Support is expected to vary similarly according to the rate of unemployment among one's family members and peers. Moreover, we expect gender difference to be associated with unemployment in a similar manner. Men tend to be more likely employed in the private sector and thus more likely to face unemployment during economic downturns. Thus, we posit an additional hypothesis emphasising individual-level experiences:
H4: One's own unemployment or the unemployment of one's family and peers positively correlates with one's attitudes towards welfare recipients (the empathy hypothesis).
With regard to the economic growth hypothesis, we expect socioeconomic variation in public support. The literature suggests that this is explained by reducing altruism because of one's own economic hardship. Thus, we expect the working class and men to be less positive towards redistribution during times of economic downturn. Earlier studies reveal that workers tend to be more morally rigid than other social classes (Svallfors 2006) and that manual workers are among those with the highest support for individual blame for poverty (Kallio and Niemelä 2014). A study by Gloding and Middleton (1982) showed that those who were socioeconomically closest to welfare beneficiaries and the poor had the strongest negative beliefs about those groups. Their explanation for this result is that such people regard the poor as direct competitors in their own economic and cultural life chances and thus develop strong negative feelings about the poor. Thus, our final hypothesis is:
H5: Experiencing economic hardship negatively correlates with one's attitude towards welfare recipients (the altruism hypothesis).
3. Research questions, data and methods
Most empirical studies that analysed the effect of situational factors used cross-national comparisons. On the contrary, this article investigates changes in public attitudes within a single country, Finland, from 1995 to 2010. Finland's exceptional economic rise and fall during the period under examination provides a fruitful ground for investigating the effect of situational factors on attitudes towards welfare recipients. Earlier studies on the effect of macro-economic factors were based mainly on a single cross-sectional point. Here, our focus is on attitude development over time.
Unemployment rate, share of social assistance recipients and GDP annual changes in Finland 1990–2010.
Unemployment rate, share of social assistance recipients and GDP annual changes in Finland 1990–2010.
While economic growth and lower unemployment often go hand-in-hand, the Finnish case provides an opportunity to distinguish the effects. The change in GDP was actually less drastic in the early 1990s recession than in 2008, but the unemployment rate changed much more during the earlier downturn.
3.1. Data
The datasets used in this study are cross-sectional surveys, which were compiled by the Department of Social Policy at the University of Turku in 1995 (n = 1859), 2000 (n = 2400), 2005 (n = 2391) and 2010 (n = 2068). The surveys included a broad set of questions dealing with material standards and economic resources, attitudes towards redistribution and the welfare state, and people's views on their own circumstances. Every year, random data samples were collected from the population register, and the datasets represent the total Finnish population between the ages of 18 and 70. In 1995, the sample size was 3000. In 2000, the sample sizes were increased to 4001. The effective response rates were reasonable: 65%, 62%, 60% and 52%, respectively. A non-response analysis did not reveal any systematic bias associated with the traditional background variables of age, education, gender or socioeconomic position. As a rule, unprivileged groups such as the unemployed are underrepresented in the survey data. The distribution of the background variables of the datasets was proportional to the distribution of the whole population, and the proportion of unemployed in the data was relatively acceptable: it was nearly the same as in the whole population (e.g. in 2000, 10% of the respondents reported being unemployed, while the unemployment rate was 9.8% according to official statistics). We included only the working-age population (aged 18–60), as, generally speaking, those aged 60 and older were more or less retired.
We chose three claims from the survey having to do with social assistance recipiency: the belief that most social assistance recipients are truly needy, the belief that many use deceit in their applications and the belief that those dependent on social assistance are lazy and unwilling to look for other solutions (for the precise claims, see Appendix 1). The respondents marked their favoured alternatives on a 4-level scale: (1) fully agree; (2) somewhat agree; (3) somewhat disagree and (4) fully disagree. We constructed a summed index of the items, with the scale of the first one reversed. A higher value indicated stronger support for welfare recipients. The index was Z-standardised to have a mean of zero and a standard deviation of one (minimum value −1.9, maximum 1.7, Cronbach's alpha 0.72; see Appendix 1). It should be kept in mind that while the attitudinal variables applied here are often used to analyse attitudes towards the welfare recipients, they may also reflect broader secular shift in opinions about the welfare state.
The individual variables included age (categorical), gender, relationship (single/couple), unemployment, unemployment of spouse, social class (higher service/lower service/working class/self-employed/not in the working force) and self-reported income (top/bottom quintile, linear effect of quintile). As macro-level variables, we used the unemployment rate and economic growth, measured by GDP change.
We tested various alternative operationalisations for these factors. For instance, we also considered the effect of experiencing unemployment during the past five years and found that only the current unemployment rate mattered. In the case of the macro variables, we tested several versions of lagged and multiple year averaged unemployment rate and GDP growth variables. In these cases, the association with our outcome was weaker than with the variables included to the reported models and did not matter at all when included in the same model.
The data were derived from the Finnish Statistical office for the same years as the cross-sectional survey data. As our study included four rounds of data, we used linear fixed effects (least squared dummy variable) models with standard errors clustered according to year.
4. Results
4.1. General trend
The percentage of those agreeing with the claims by year (fully or partially).
Currently, more tolerant claims gain more support than in the mid-1990s. The distribution of responses still reflects the spectrum of generally held beliefs: recipients of social assistance are seen both as truly needy and as cheaters. Interestingly, Finnish public opinion is still split into two opposites. Previous research also indicates that social assistance faces an ambivalent legitimacy (Hasenfeld and Rafferty 1989).
4.2. Individual-level and macro-level effects
We now turn to the individual-level and macro-level effects and analyse how the factors are associated with attitudes towards welfare recipients. Table 1 presents estimates for the individual-level variables in a joint multivariate linear model.
. | Estimate . |
---|---|
Age | |
18–25 | −0.21* |
25–34 | −0.10* |
35–44(ref.) | 0 |
45–54 | −0.01 |
55–60 | 0.03 |
Female | 0.13* |
Single | 0.06* |
Unemployed | 0.19** |
Spouse unemployed | 0.10** |
Social class | |
Higher service (ref.) | 0 |
Lower service | 0.03 |
Working class | −0.15* |
Self-employed | −0.29** |
Not in workforce | −0.04 |
Lowest income quintile | 0.07** |
Highest quintile | 0.01 |
Quintile (linear) | −0.01 |
Year | |
1995 (ref.) | 0 |
2000 | 0.15*** |
2005 | 0.21*** |
2010 | 0.23*** |
Intercept | 0.04 |
. | Estimate . |
---|---|
Age | |
18–25 | −0.21* |
25–34 | −0.10* |
35–44(ref.) | 0 |
45–54 | −0.01 |
55–60 | 0.03 |
Female | 0.13* |
Single | 0.06* |
Unemployed | 0.19** |
Spouse unemployed | 0.10** |
Social class | |
Higher service (ref.) | 0 |
Lower service | 0.03 |
Working class | −0.15* |
Self-employed | −0.29** |
Not in workforce | −0.04 |
Lowest income quintile | 0.07** |
Highest quintile | 0.01 |
Quintile (linear) | −0.01 |
Year | |
1995 (ref.) | 0 |
2000 | 0.15*** |
2005 | 0.21*** |
2010 | 0.23*** |
Intercept | 0.04 |
Note: N = 6998.
*p < .05.
**p < .01.
***p < .001.
The results are consistent with earlier findings: women are more positive towards welfare recipients than men, and the youngest age groups are set apart from the others. As far as social class is concerned, both the working class and the self-employed differ from the other classes, as they feel most negatively about those on welfare. Both findings on gender and class are in line with the altruism hypothesis – those most likely to face economic hardship are the least positive towards welfare recipients.
However, when concentrating on the actual experience of economic hardship, the results show that both being unemployed and having an unemployed spouse lead to stronger support, although one's own unemployment status is more meaningful than one's spouse's unemployment. The effect of unemployment lends support for government protection due to the unemployment hypothesis, which states that those most likely to receive benefits are also more likely to support such schemes. The effects of household income confirm this further: the households with the lowest income had the most positive attitudes towards those on welfare benefits. However, the finding that one's spouse's unemployment also has an independent effect suggests that the empathy hypothesis also matters. As the effect remains significant when income is controlled, this is not likely simply a secondary effect of a reduced economic well-being of the household.
The additional, unreported analyses show that the controlling of unemployment in the same model does not affect estimates for class. Moreover, class and unemployment do not interact. Thus, the status differences in support do not seem to be explained by first-hand experience of unemployment.
Next, we conducted analyses on the macro-level and investigated how the changes in the economic cycle and the unemployment rate are associated with attitudes towards those receiving last-resort social assistance. These analyses are presented in Table 2 in step-wise built models. Model 1 is a simplified version of the model presented in Table 1, excluding the controls for income (the estimates for age were included in the model but omitted from the table). In Model 2, we replaced the year dummies with macro-level variables and a linear effect for the survey year to control for a potential longer term secular shift in attitudes mentioned above. The associations found at the individual-level in Model 1 remained unaltered, suggesting that simplifying the model by using macro controls does not cause omitted variable bias.
. | Model 1 . | Model 2 . | Model 3 . | Model 4 . | Model 5 . |
---|---|---|---|---|---|
Individual level var. | |||||
Female | 0.13* | 0.13** | 0.13** | 0.13** | 0.13** |
Unemployed | 0.22*** | 0.22*** | 0.22*** | 0.22*** | 0.19** |
Spouse unemployed | 0.12** | 0.11** | 0.11** | 0.12** | 0.10** |
Social class | |||||
Higher service | 0 | 0 | 0 | 0 | 0 |
Lower service | 0.02 | 0.02 | 0.01 | 0.02 | 0.03* |
Working class | −0.14* | −0.14* | −0.13* | −0.14** | −0.15** |
Self-employed | −0.28** | −0.28** | −0.33*** | −0.29*** | −0.30*** |
Not in workforce | −0.02 | −0.02 | −0.08 | −0.02 | −0.04 |
Lowest income quintile | 0.07** | ||||
Quintile (linear | −0.01 | ||||
Fixed/macro level var. | |||||
Year | |||||
1995 (ref.) | 0 | ||||
2000 | 0.16*** | ||||
2005 | 0.22*** | ||||
2010 | 0.24*** | ||||
Year (linear) | >0.01*** | >0.01*** | >0.01*** | >0.01*** | |
Unemployment rate | −0.02*** | −0.02*** | −0.02*** | −0.02*** | |
GDP yearly change | 0.03* | 0.03* | 0.02 | 0.01 | |
Female*GDP ch. | −0.05** | −0.05** | −0.05** | −0.05** | |
Unemp. Rate * Hgh service | 0 | ||||
Unemp. Rate * Low service | 0 | ||||
Unemp. Rate * working class | 0 | ||||
Unemp. Rate * selfemp. | −0.01 | ||||
Unemp. Rate * no class | −0.01 | ||||
GDP ch. * Hgh service | 0 | 0 | |||
GDP ch. * Low service | 0.01* | 0.01 | |||
GDP ch. * working class | 0.04* | 0.04* | |||
GDP ch. * selfemp. | −0.04* | −0.04* | |||
GDP ch. * no class | −0.01 | 0 | |||
Intercept | −0.14* | −0.14* | −0.13* | −0.14* | −0.09 |
. | Model 1 . | Model 2 . | Model 3 . | Model 4 . | Model 5 . |
---|---|---|---|---|---|
Individual level var. | |||||
Female | 0.13* | 0.13** | 0.13** | 0.13** | 0.13** |
Unemployed | 0.22*** | 0.22*** | 0.22*** | 0.22*** | 0.19** |
Spouse unemployed | 0.12** | 0.11** | 0.11** | 0.12** | 0.10** |
Social class | |||||
Higher service | 0 | 0 | 0 | 0 | 0 |
Lower service | 0.02 | 0.02 | 0.01 | 0.02 | 0.03* |
Working class | −0.14* | −0.14* | −0.13* | −0.14** | −0.15** |
Self-employed | −0.28** | −0.28** | −0.33*** | −0.29*** | −0.30*** |
Not in workforce | −0.02 | −0.02 | −0.08 | −0.02 | −0.04 |
Lowest income quintile | 0.07** | ||||
Quintile (linear | −0.01 | ||||
Fixed/macro level var. | |||||
Year | |||||
1995 (ref.) | 0 | ||||
2000 | 0.16*** | ||||
2005 | 0.22*** | ||||
2010 | 0.24*** | ||||
Year (linear) | >0.01*** | >0.01*** | >0.01*** | >0.01*** | |
Unemployment rate | −0.02*** | −0.02*** | −0.02*** | −0.02*** | |
GDP yearly change | 0.03* | 0.03* | 0.02 | 0.01 | |
Female*GDP ch. | −0.05** | −0.05** | −0.05** | −0.05** | |
Unemp. Rate * Hgh service | 0 | ||||
Unemp. Rate * Low service | 0 | ||||
Unemp. Rate * working class | 0 | ||||
Unemp. Rate * selfemp. | −0.01 | ||||
Unemp. Rate * no class | −0.01 | ||||
GDP ch. * Hgh service | 0 | 0 | |||
GDP ch. * Low service | 0.01* | 0.01 | |||
GDP ch. * working class | 0.04* | 0.04* | |||
GDP ch. * selfemp. | −0.04* | −0.04* | |||
GDP ch. * no class | −0.01 | 0 | |||
Intercept | −0.14* | −0.14* | −0.13* | −0.14* | −0.09 |
Note: N = 6998.
*p < .05.
**p < .01.
***p < .001.
The results show that both macro-economic factors are associated with attitudes. There is evidence of a negative effect of unemployment, which is to say that attitudes towards those on welfare benefits are more positive when unemployment is low. This result is in contrast with the governmental protection hypotheses (both unemployment and economic decline). Further, the results suggest that higher economic growth has a positive effect on attitudes. In favourable economic times, people see welfare recipients as being more deserving and perceive their need for assistance as real. This finding lends support for the economic growth hypothesis.
Support according to GDP yearly change and gender. Predictive margins from Model 2 in Table 2.
Support according to GDP yearly change and gender. Predictive margins from Model 2 in Table 2.
In the theoretical discussion above, it was assumed that support could also vary according to class and macro variables. The effects were assumed to be opposite in nature. In the case of the government protection hypothesis, those having the highest probability of unemployment during the downturn should be most positive towards income transfers, whereas the situation should be opposite according to the economic growth hypothesis: the attitudes of the working class should be particularly harsh during the downturns but come closer to those of other classes during economic growth. To test this prediction, Model 3 introduces interaction terms between social class and unemployment, and Model 4 introduces interaction effects between social class and GDP change. Contrary to expectations in relation to the government protection hypothesis, there is no significant interaction effect between social class and unemployment. However, we find a significant interaction effect between social class and economic growth, suggesting a more nuanced pattern of attitudes.
Support according to GDP yearly change and social class. Predictive margins from Model 4 in Table 2.
Support according to GDP yearly change and social class. Predictive margins from Model 4 in Table 2.
The working class is set apart from the other classes. Although we might think that the working class has a higher risk of being in receipt of social assistance than other social classes and that, therefore, they particularly would have more positive views of those on welfare, our results show the opposite: the working class has more positive views of welfare recipients during prosperous times, whereas in times of lower economic growth, they hold more negative views.
Finally, in Model 5, we include the income variables first introduced in Table 1. The other estimates do not change when income is controlled for, suggesting that the effects we observed in the previous models were not just signalling for income differences between men and women and different classes.
The findings reported above are robust to excluding the small but statistically significant linear year effect (the estimate was below 0.005, thus rounded to zero at two-digit level). While other estimates would remain the same, excluding it would make the main effect of unemployment rate more negative by 0.01. Although this is change is small, it is nonetheless also statistically significant.
In addition to the results reported here, we also tested the importance of economic inequality of society measured as Gini-index, using both current and lagged versions. There are a number of previous studies suggesting that economic inequality, especially when measured in this way, should be associated with the variation in the preferences towards redistribution and welfare state attitudes more generally (e.g. Dion and Birtchfield 2010; Jaeger 2013). This did not have a significant association with attitudes towards welfare recipients.
Finally, we also considered the influence of the childhood experience on welfare assistance and found no association. Thus, the results do not support theories about dependency culture or the culture of poverty (Lewis 1966). Attitudes of those who have childhood experiences of welfare were as positive or as negative as those of other respondents, and it appears that childhood experiences of welfare did not cause the individual to acquire a mind-set that made them feel different from others. This variable was also dropped from the subsequent analyses.
5. Discussion
In this article, we analysed the variation in public attitudes towards welfare recipients according to economic cycles in Finland from 1995 to 2010, during a period that covered two severe economic crises. The recession of the early 1990s was geographically more restricted (mainly to Finland and Sweden), but the latter recession has swept throughout all of Europe. This recession has made the topic of swinging public support towards welfare recipients a key question for the whole continent. The public support towards welfare recipients is often found to be lower than support towards other, more universal forms of welfare benefits, thus making it potentially more exposed to external economic shocks.
Existing research holds contradictory findings, with some finding that economic recession leads to declining welfare state attitudes while others indicate stronger support at times of increased financial strain. We studied whether the attitudes towards welfare recipients varied according to economic change and unemployment. In contrast to the previous studies, we considered whether this was dependent on socioeconomic status and on two individual-level mechanisms, empathy and altruism.
Our analyses of the Finnish case reveal that high unemployment and less advantageous economic growth diminishes public support for welfare recipiency, that is, public holds more negative views on those receiving last-resort benefit. Thus, our findings do not give support for either version of the government protection hypothesis (unemployment or economic decline). Instead, the results indicate that the economic growth hypothesis applies better. Although the results are in contrast with many other studies underlining the importance of the government protection hypotheses, arguing that they do not matter at all would be an oversimplification. Despite the macro-level effects supporting economic growth assumption, at the individual level, both one's own unemployment and the unemployment of one's spouse were positively associated with the support for welfare recipiency. These findings point to self-interest and suggest that empathy most likely comes into play, but this is limited to those having first-hand experiences of unemployment.
Economic downturn reduces the altruism of those who are facing the risk of greatest economic hardship, especially in the case of the working class and men. Because of this, one could assume that there would be a stronger link between unemployment and class. However, we only found an interaction between GDP growth and class and between GDP growth and gender. This was because the service classes and especially the self-employed showed the opposite pattern, suggesting increasing empathy during economic downturns. A similar pattern can be observed in the interaction between GDP growth and gender: men, also facing a greater risk of unemployment in general, have stricter attitudes towards welfare recipients during economic recessions, whereas the opposite holds true for women.
Although we control for the secular trend in attitudes with the linear year effect in our models with macro variables, it may still be that the results also reflect the broader attitude changes towards the welfare state. This may particularly apply to the findings on the negative interaction between GDP growth and the attitudes of the self-employed.
Our results provide a reminder for policy-makers that people in different social classes do not have the same policy expectations and that these expectations may change in very different ways across the social lines. While the working-class representatives have stricter views of welfare recipients during economic downturns and thus are more likely to prefer a tougher policy, the opposite is true for the service classes and the self-employed. It is also hard to imagine consistent welfare politics that could appeal especially to working-class voters during both economic up- and downturns. This could be a valuable lesson for politicians.
Future research should more thoroughly investigate variation in attitudes between socioeconomic groups in economic cycles, and country-specific socioeconomic features need to be better taken into account. Large, cross-national studies can easily hide peculiar developments of a single country, such as the divergent pattern among socioeconomics and gender.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes on contributors
Susan Kuivalainen is the Head of the Research Department at the Finnish Centre for Pensions. Prior, she has worked as a research professor at the National Institute for Health and Welfare and has hold the professorship in the Social Insurance Expert Training Programme at the University of Turku. She acquired her PhD in Social Policy from the University of Turku in 2004. Her research interest includes socio-political systems, poverty and inequality and welfare policy.
Jani Erola is Professor of Sociology at the University of Turku. He was awarded an ERC Consolidator Grant in 2013 for his project Intergenerational Cumulative Disadvantage and Resource Compensation (INDIRECT, ERC-2013-CoG-617965) and has been editing the Acta Sociologica – journal since 2014. He acquired his PhD in Economic Sociology from the Turku School of Economics in 2004. His research interests include social inequality, family formation, intergenerational social mobility and sociological research methods.
References
Appendices Appendix 1. Average inter-item correlation and Cronbach's alpha for the latent ‘Support’ variable.
. | . | . | If item deleted . | . |
---|---|---|---|---|
Item . | N . | Sign . | av. Inter-item corr. . | Alpha . |
‘Those dependent on social assistance are lazy and unwilling to look for other solutions’ | 6947 | + | 0.4099 | 0.5814 |
‘Most social assistance recipients are truly needy’ | 6981 | − | 0.5578 | 0.7161 |
‘Many use deceit in applying’ | 6931 | + | 0.4214 | 0.593 |
Test scale | 0.4629 | 0.7211 |
. | . | . | If item deleted . | . |
---|---|---|---|---|
Item . | N . | Sign . | av. Inter-item corr. . | Alpha . |
‘Those dependent on social assistance are lazy and unwilling to look for other solutions’ | 6947 | + | 0.4099 | 0.5814 |
‘Most social assistance recipients are truly needy’ | 6981 | − | 0.5578 | 0.7161 |
‘Many use deceit in applying’ | 6931 | + | 0.4214 | 0.593 |
Test scale | 0.4629 | 0.7211 |