Public opinion and the reform of the pension systems in Europe: the influence of solidarity principles

Abstract

The demographic changes that have occurred in European countries in recent decades have made the policies of the public pension system one of the most debated issues of the welfare state. In this paper, I focus on preferences for three pension policy reforms with different distributive consequences: raising contributions, raising the age of retirement, and allowing free choice between public and private pension plans. I use multilevel models to analyse how individual attachment to different solidarity principles (universalistic, conservative, liberal and familistic) affects attitudes toward pension system reforms while controlling for institutional factors. The empirical results strongly support the hypothesis that solidarity principles have a significant influence on individual preferences. I find that individuals who adhere to universalistic or conservative principles are more in favour of increasing contributions in order to maintain the level of pensions, whereas they oppose a postponement of retirement age. In contrast, those who adhere to liberal or familistic principles are against increasing contributions and prefer extending retirement age. The findings at least partially support the ‘regime hypothesis’, as a more generous pension system appears to increase support for raising contributions while decreasing support for a raise in the age of retirement.

Keywords

pensions preferences retirement solidarity principles welfare regime

Introduction

Demographic changes in European countries in recent decades have made the public pension system policy one of the most debated issues of the welfare state (Bonoli, 2003; Taylor-Gooby, 2002). Several institutions and scholars argue that the aging process caused by growing life expectancy and low fertility rates make a pay-as-you-go (PAYG) public pension system unsustainable in the long term (Whitehouse and Queisser, 2007). It is often argued that pension increases as well as an increase in retirement years will cause the mature welfare systems to become unsustainable at some point in the course of this century (Bonoli, 2003; Pierson, 1998). Thus, as fiscal pressure on an ever smaller active population increases, the intergenerational contract will be increasingly in danger of collapse (Pierson, 1998). The debate has moved into the arena of public opinion and substantial reforms have taken place in several European countries since the early 1990s (Taylor-Gooby, 2002; Whitehouse and Queisser, 2007). These reforms have primarily affected the age of retirement (both eligibility and incentives to delay retirement) and the introduction of complementary private pension plans, whether mandatory or encouraged by means of incentives (Whitehouse, 2006; Whitehouse and Queisser, 2007).

In spite of the considerable volume of research on attitudes toward social policies (see a survey, for instance, in Alesina and Giuliano, 2010), the topic of preferences for different pension systems has received little attention in the literature. On the one hand, there is evidence that reforms have generated significant resistance in the countries where they have occurred (Boeri et al., 2002; Bonoli, 1997b; Pierson, 1996), even though demobilization protest strategies have facilitated reforms in some cases (Béland and Marier, 2006). At the same time, some studies have found that preferences concerning the pension system tend to be inconsistent. In a study of public opinion, Boeri et al. (2002) indicate that citizens in Germany and Italy are aware that current pension systems are not sustainable in the long term and recognize the need to reform them, yet they tend to ignore or underestimate the costs of the system as a whole. Likewise, Janky and Gál (2007) argue that, despite awareness of the impact of demographic problems on the pension system, Europeans continue to oppose the reforms, although such resistance is not uniform.

This research focuses on the analysis of public opinion toward the different reform alternatives present in the public debate in the European context: changes in Social Security contributions, postponement of retirement age, and allowing free choice between public and private pension plans. I garner insights from self-interest and ideological approaches, and institutional theories, which argue that the solidarity principles that inspire different models of the welfare state at the macro-level have an impact on individual preferences (Arts and Gelissen, 2001; Jakobsen, 2011; Mau; 2004). However, departing from previous studies, I take into account how individual commitment to different solidarity principles (universalistic, conservative, liberal and familistic), affects attitudes to pension policy reforms, after controlling for institutional factors. To assess the impact of the explanatory variables, I estimate multilevel models, using data from the Eurobarometer for EU-15 countries. The rest of the paper is organized as follows. The next section presents theoretical approaches that explain pension policy preferences. The following section introduces some operational definitions and the hypotheses. I then present the data and methods, followed by a discussion of the main findings. The paper ends with a section of conclusions along with open questions for further research.

Theoretical approaches

The literature on social policy preferences has evolved along three theoretical lines: the self-interest approach, the ideological approach and the institutional theories. The self-interest approach, inspired by the economic models proposed by Meltzer and Richard (1981) and Browning (1975), holds that welfare programmes generate redistributive effects, which make net beneficiaries favourable to these policies, whereas net contributors oppose them (Galasso, 2006; Lynch and Myrsylä, 2009). In particular, mature welfare states show a distinct cleavage between the working-age population (the net contributors) and the retired population (the beneficiaries of the public pension system). For this reason, Weaver (2003) indicates that cuts in the pension system are especially risky for governments. First, the losses are strongly felt by the beneficiary group and, second, the elderly population has a high electoral participation in many countries. Likewise, the paradigm of retrenchment is based on the idea that cuts in social policies will be opposed mostly by the groups that benefit from the policies (Pierson, 1996). Furthermore, Pierson argues that the beneficiary groups (including retired people) are more important in the ‘new politics’ of welfare state reform than the wide range of public interest organizations. In a similar vein, Campbell (2003) argues that the development of social security can only be understood by taking into account the support dynamics generated by the beneficiary groups.

It is not always easy, however, to derive specific theoretical predictions concerning preferences for different pension systems. Thus, the effect of the aging process on the welfare state has been interpreted in radically opposed ways by two rival theories within the self-interest approach: the theory of ‘elderly power’ and the theory of ‘fiscal leakage’ (Tepe and Vanhuysse, 2009). According to the former, the aging process creates a demand for more generous pension systems, which is proportional to the increase in the number of potential beneficiaries (Galasso, 2006; Galasso and Profeta, 2000, 2007; Persson and Tabellini, 2000). According to the second theory, however, low fertility rates and growing life expectancy reduce the value of public pension systems, since today’s workers expect difficulties in collecting their pensions at retirement age due to the potential collapse of the system (Breyer and Stolte, 2001; Razin and Sadka, 2007). Faced with this situation, the working-age population loses confidence in the implicit generational agreement that sustains the public pension system. They oppose presently contributing, as they think they will not become beneficiaries in the future (Tepe and Vanhuysse, 2009). Given that, in spite of the aging process, most of the population is not retired, thus there will probably be a demand to reduce contributions to the public pension system along with a shift to private pension systems.

In spite of the predominance of the self-interest approach in the literature on redistribution, empirical results are not always in line with theoretical expectations (Sears and Funk, 1990). Thus, the ideological approach states that attitudes toward social policies must be explained in terms of values and basic political orientations (Feldman and Zaller, 1992; Jacoby, 1994). The debate remains open. Taylor-Gooby (2001) shows that labour market position and gender are fairly weak predictors of support for cuts in welfare programmes. Furthermore, Tepe (2006), and Groezen et al. (2009) have found that preferences for different pension systems cannot be explained only by self-interest, since ideological factors are also significant predictors. In turn, Boeri et al. (2002) encounter mixed evidence and conclude that both economic factors and ideology have a significant effect on pension system reform preferences.

Finally, the institutional approach states that attitudes toward social policies depend on macro-level factors such as the welfare regime. The concept of the welfare regime is made of ‘formalized social policy arrangements, a shared history of class mobilization, institutionalized solidarity, and social justice beliefs’ (Jakobsen, 2011: 329) and it defines who deserves to be helped and under what circumstances (Larsen, 2006). According to Jakobsen (2011), solidarity principles reflect value systems that are deeply rooted in history and tradition because there is an interplay between welfare state institutions and people’s basic set of values. Each welfare model follows a distinctive path and attitudes toward particular policies can only be understood within their cultural context. Therefore, we should expect an isomorphy between institutions, value systems and attitudes. For instance, Velladics et al. (2006) show that Eastern Europeans tend to favour welfare reforms in which the benefits depend on the number of children, because of the predominance of traditional values promoted during the Communist period.

Welfare regimes are inspired by different distributive justice criteria (Mau, 2004), and Arts and Gellisen (2001) argue that each ‘ideal type’ of welfare state will produce different patterns of solidarity. The liberal type is based on equality of opportunities, which implies that each person is responsible for his or her own welfare and thus that the role of government is to guarantee equal conditions of access to vital opportunities without concerns about equality of results. The conservative type is based on a hierarchy of class and status, which produces a segmentation of welfare, based on corporate lines of division. Finally, the social democratic model is inspired by universalism and egalitarianism, which leads to an inclination for equality of results, with significant emphasis on income redistribution.

Regarding the specific realm of pension policies, Gelissen (2001) argues that the principles inspiring welfare regimes translate into different models of pension system organization. Gelissen predicts that the liberal model will attribute the greatest importance to institutions of private provision, with a modest role for the state and a relatively important role for employers and employees. In the corporate or conservative model, private pension funds will have a modest role, whereas great importance will be given to the participation of employers and employees, the role of the state being relatively important. In the social-democratic model, the role of the state will be very important, while pension funds and the role of employers and employees will be of modest importance. In the Mediterranean model both the state and private funds have marginal importance, with a relatively significant role for employees and employers. Empirical results support the idea that preferences for different models of pension systems depend on the welfare regime (Gelissen, 2001), although there are important differences between countries belonging to the same welfare regime.

Several debates concerning the regime argument are still open. First, empirical evidence has failed to provide conclusive support for the link between welfare regime and attitudes (Andreβ and Heien, 2001; Arts and Gellisen, 2001; Svallfors, 1997). Some studies conclude that attitudes toward welfare policies ‘are only to a very limited extent structured by differences in welfare regimes’ (Svallfors, 2003: 190), although Jakobsen (2011) and Mehrtens (2004) find partial support for the ‘regime hypothesis’ and Jaeger (2006; 2009) argues that methodological limitations in the operationalization of the independent variables has prevented the finding of more robust results. Second, there has been an important debate on alternative classifications of welfare regimes (Arts and Gelissen, 2002; Jakobsen, 2011), and some authors have proposed the use of independent continuous variables, such as social spending, as an alternative to categorical classifications (Jaeger, 2006). Finally, the literature on the ‘welfare regime hypothesis’ has focused on the effect of solidarity principles that structure welfare institutions at the national level. In doing that, an important point is missed: solidarity principles are not homogeneously distributed among the population.

Even if two individuals live in the same country and interact with the same institutions, we cannot assume that they share the same commitment to the solidarity principles that structure national institutions. Arts and Gelissen (2001) point out that it would be ingenuous to think that individual preferences are entirely determined by contextual factors. Welfare regimes are inspired by generally accepted principles of solidarity, but they also generate differences between social groups as a result of the different redistributive conflicts produced in each regime (Svallfors, 1997). On the other hand, different welfare regimes are a product of the social cleavages produced in the transition from the industrial to the post-industrial society (Esping-Andersen, 1990). This leads Pettersen (1995) and Gelissen (2001) to predict that attitudes toward public sector intervention on pension provision would also depend on social categories. In terms of Jakobsen’s (2011) cultural argument, we could argue that distributive conflicts around norms of justice are deeply rooted in national histories, and individuals define their own identities in terms of the norms and solidarity principles of the social groups to which they belong.

Solidarity principles and preferences for pension reforms

The main argument of this paper is that, in order to understand individual choices when facing policy alternatives, we need to take into account individual commitment to solidarity principles as well as the solidarity principles that structure welfare institutions at the national level. Following standard theoretical approaches, I propose the following definitions. First, I understand that an individual subscribes to a universalistic principle in the realm of pension policy if he or she considers that pension benefits constitute a universal social right for every citizen. Second, an individual adheres to a conservative principle if he or she subscribes to the idea that the goal of pensions is to maintain status, or one’s pre-retirement income. Third, an individual subscribes to a liberal principle if he or she believes that the pension each person receives should depend exclusively on his or her individual contributions. However, welfare regimes and solidarity principles need to be expanded in order to take into account the Mediterranean model (Gelissen, 2001), also known as the Southern Model (Bonoli, 1997a; Ferrera, 1996; Moreno, 2000) or Latin Rim (Leibfried, 1992). This model is distinguished from the others by a strong implementation of familistic values. Individuals who adhere to the principle of solidarity based on familistic values are expected to value the importance of inter-generational solidarity and assistance to the elderly within the family. Thus, I understand an individual to subscribe to a familistic principle if he or she believes that the family is obligated to guarantee the protection of retired people. It is important to note that, from the previous definitions, an individual does not adhere to one principle of solidarity exclusively. Rather, each individual has different degrees of commitment to each of the principles proposed.

Since each policy reform will have different distributive consequences, it is expected that each solidarity principle will induce different preferences for pension system reforms. First, we should expect a correlation between the universalistic principle and greater support for increasing contributions, since the public pension system has both redistributive and insurance effects (Tepe, 2006). At the same time, we can predict that the universalistic principle is against free choice between public and private plans, because that might cause a displacement of resources from public schemes to private plans, reducing the capability of the public system to fulfil its goals. The universalistic principle is also in contradiction with a rise in retirement age because it undermines the redistributive impact of the pension system. Other things being equal, raising the retirement age means that high-income workers can increase their savings for retirement more than low-income workers, while redistribution through public pensions will take place during a shorter period of time. The liberal principle will be correlated with lower support for increasing contributions and higher support for allowing free choice between public and private pension plans, since increasing contributions is against free individual choice. On the other hand, those who adhere to the liberal principle would prefer a raise in retirement age because that allows individuals to increase their private savings for retirement and reduces the percentage of pensionable population. That would alleviate the fiscal burden of the public system, which would prevent future increases in taxes or contributions. For those who adhere to the conservative principle, their interest lies in the preservation of pre-retirement status thus they will prefer an increase in contributions in order to maintain that status after retirement via the public pension system transfers. For those who adhere to the familistic principle, opposition to an increase in contributions is to be expected since that will reduce private resources to take care of the elderly within the family. The following hypotheses summarize these expectations:

Hypothesis 1a: The universalistic principle increases support for raising contributions and reduces support for raising the retirement age and allowing private plans.

Hypothesis 1b: The liberal principle reduces support for raising contributions and increases support for raising the retirement age and allowing private plans.

Hypothesis 1c: The conservative principle increases support for raising contributions.

Hypothesis 1d: The familistic principle decreases support for raising contributions.

Lastly, according to the self-interest approach, socio-economic status is expected to have an impact on preferences for pension reforms (Boeri et al., 2002; Groezen et al., 2009). Given that contributions to the social security are generally proportional to income, those with higher incomes tend to oppose an increase in contributions. At the same time, the wealthy voters will favour an allowance of the choice of private schemes instead of public ones, since that will produce a higher income during retirement. Conversely, the poor will oppose private schemes, since their savings for retirement will be low. The increase in the retirement age will produce a loss in welfare for every individual. However, those with higher incomes will be compensated by an increase in private savings. Thus, the support for increasing retirement age will be lower among low-income earners.

Hypothesis 2: Attitudes toward pension system reforms depend on socio-economic status. The higher the socio-economic status, the greater the preference for increasing retirement age and favouring private systems will be, and the lower the support for increasing contributions.

Data and methods

Data and variables

The data used in this research come from Eurobarometer 56.1 (European Commission, 2003) carried out in 2001. The reasons for using this database are twofold. Although several pension reforms have taken place in different EU countries since the data were collected, the main goal of this research is to explain how solidarity principles affect preferences when individuals must choose among policy alternatives with different distributive consequences.¹ On the other hand, more recent Eurobarometers do not contain information about the key explanatory variables that measure individual commitment to solidarity principles in the realm of pension policy. I analyse three dependent variables that refer to different pension policy alternatives. In this survey, the interviewees were asked whether they agree with the following statements: (i) ‘Contributions should not be raised even if this means lower pensions’; (ii) ‘The age of retirement should be raised so that people work longer’; and (iii) ‘The government should allow people to put their contributions into a private pension fund or life insurance policies of their choice’. The available responses for these variables were: (1) ‘Strongly agree’, (2) ‘Slightly agree’, (3) ‘Slightly disagree’, and (4) ‘Strongly disagree’. However, I reversed the order of the response categories of the last two statements to facilitate their interpretation.

Explanatory variables include individual and national factors. Individual variables measure commitment to different solidarity principles based on different models of the welfare state, socio-demographic variables and ideology. The universalistic principle is measured by agreement with the statement: ‘A guaranteed minimum pension should be a basic social right of every citizen.’ The conservative principle is measured through agreement with the statement, ‘A good pension system should allow everybody to maintain an adequate standard of living relative to their income before retirement.’ The liberal principle is measured through agreement with the statement, ‘The amount of one’s pension should be strictly based on the amount of contributions one has paid into the pension scheme’; and the familistic principle is measured through agreement with the statement, ‘There should be a legal obligation for children to financially support their elderly parents if they don’t have enough income of their own.’ The scale of response to these variables was: (1) ‘Strongly agree’, (2) ‘Slightly agree’, (3) ‘Slightly disagree’, and (4) ‘Strongly disagree’. However, the order of the response categories has been reversed to facilitate their interpretation.

The socio-demographic variables include those that reflect differences in income sources, which correlate with redistributive preferences in previous studies (Alesina and Giuliano, 2010). This group of variables includes: gender (0 = ‘Male’, and 1 = ‘Female’), age and age-squared in order to capture a possible non-linear effect of this variable, education, which measures the respondent’s age at the end of his/her full-time education² (1 = ‘Up to 14 years’, 2 = ‘15–20 years’, and 3 = ‘More than 20 years’), and marital status (0 = ‘Married or living in a couple’, 1 = ‘Single’, and 2 = ‘Divorced, separated or widowed’). I was unable to include other relevant variables in the analysis such as occupation or income level due to the high number of missing values in the national samples. Ideology is measured through a ten-point scale where 1 means extreme left and 10 extreme right.

National variables measure institutional differences between welfare regimes. Following Arts and Gellisen (2001), Ferrara (1996) and Jakobsen (2011), I distinguish between the social-democratic model (Sweden, Denmark, Norway, Finland), the conservative (Germany, France, Belgium, Netherlands, Luxembourg and Austria), the liberal (United Kingdom and Ireland) and the Mediterranean model (Italy, Spain, Portugal and Greece). As an alternative way to measure the institutional features of the welfare state I use the decommodified security ratio (DSR) computed by Menahem (2007), which is based on the previous work of Esping-Andersen (1990). It measures the economic security provided by the welfare state by taking into account income replacement, social benefits and economic insecurity.³ Finally, in order to consider the institutional characteristics of the pension system, I use an index of pension generosity, which is computed as the expenditure in pensions as percentage of GDP (Eurostat, 2012) divided by the percentage of people aged ≥65 years (Eurostat, 2012). Data are always taken for the year before the survey data were collected. A descriptive analysis of the variables included in the analysis is reported in Table 1.

Table 1.

Descriptive statistics.

Variable	Mean	Standard deviation	Minimum	Maximum
Raising contributions	2.773	0.953	1	4
Raising the retirement age	1.873	0.958	1	4
Allowing private choice	3.252	0.792	1	4
Female	0.523	0.500	0	1
Age	45.205	18.376	15	99
Education
15–20 years of education	0.514	0.500	0	1
>20 years of education	0.289	0.453	0	1
Marital status
Single	0.184	0.388	0	1
Divorced, separated, or widowed	0.220	0.414	0	1
Ideology	5.197	1.916	1	10
Universalistic principle	3.552	0.636	1	4
Conservative principle	3.399	0.689	1	4
Liberal principle	2.879	0.931	1	4
Familistic principle	2.248	1.047	1	4
Decommodification index	16.302	7.639	1.700	27.800
Pension generosity index	0.480	0.118	0.146	0.654

Source: Eurobarometer 56.1 (2001).

Statistical methods

Given that the dependent variables are ordinal categorical, I use ordinal logistic regressions to estimate the effect of the explanatory variables (McKelvey and Zavoina, 1975). However, since individuals are nested within countries having different institutional characteristics, a multilevel setup is used to estimate the models reported in the next section. The main advantage of using multilevel models in comparative research is that they account for variations in the response across countries and enable an estimation of the effect of national variables on individual responses (Raudenbush and Bryk, 2002). In this case, I use random intercepts models to take into account the fact that the popularity of each pension reform varies between countries. That amounts to the assumption that individual responses within the same country share a common source of variation due to the effect of unknown country level characteristics.

Estimation of multilevel models with categorical dependent variables involves significant computational problems. There are two main approaches for tackling this problem in the literature: quasi-likelihood methods (Goldstein, 2010) and quadrature methods (Rabe-Hesketh et al., 2005). While the methods of quasi-likelihood are less computationally demanding, their main disadvantage is that, in the presence of large variances between countries or dependent variables that deviate greatly from the normal distribution, estimates would be biased. For these reasons, I have used the quadrature method proposed by Rabe-Hesketh et al. (2005). The models were estimated using the GLAMM routine for Stata.⁴

Findings and discussion

It is useful to begin the analysis by reviewing the average national support for each type of pension reform shown in Table 2: support for raising contributions, raising the retirement age and allowing free choice between public and private pension plans. It should be pointed out that each individual’s response to these reforms refers to the pension system in place in each country, which introduces a certain problem for comparison of the raw percentages between countries. For example, when an individual indicates a preference for raising the retirement age, he or she expresses something different in the different countries if the retirement ages differ from country to country. However, the basic aim of this research is to analyse attitudes toward reform of the pension system with regard to the status quo in each country. The data in Table 2 show that in all the countries a majority of public opinion support allowing free choice of private pension plans, whereas raising the retirement age is the least preferred option in all of the analysed countries. That suggests a framing affect, according to which reforms expressed in positive terms (such as giving freedom of choice) receive greater support (Janki and Gál, 2007).

Table 2.

Attitudes toward pension reforms by country.

	Raising contributions	Raising retirement age	Allowing private choice
Austria	65.3	27.4	82.9
Belgium	59.2	23.9	90.5
Denmark	68.0	29.1	92.8
Finland	63.9	26.8	88.4
France	58.4	26.3	85.7
Germany	60.9	16.8	85.4
Greece	60.8	14.0	88.7
Ireland	58.9	48.7	94.7
Italy	59.3	31.2	88.3
Luxembourg	66.2	19.0	88.4
Netherlands	75.7	26.0	86.5
Portugal	49.8	31.1	92.1
Spain	67.5	22.2	76.6
Sweden	67.2	13.8	71.5
United Kingdom	73.9	25.8	93.2

Notes: Cells display percentage of individuals who ‘strongly agree’ or ‘slightly agree’.

Source: Eurobarometer 56.1 (2001).

Support for the allowance of private fund plans choice ranges from 71.5 percent in Sweden to 94.7 percent in Ireland. The greatest support for increasing contributions occurs in the Netherlands, the United Kingdom, Denmark, Spain, Sweden and Luxembourg, where more than two-thirds of the population is in favour of raising contributions. In contrast, the greatest resistance to this reform is in Portugal, where only about a half of the population supports this measure. The greatest resistance to the increase in retirement age occurs in Sweden, Greece, Germany and Luxembourg, where the proportion of individuals who would accept this proposal is less than one-fifth of the population. In contrast, almost half of the population of Ireland would be willing to accept a retirement-age postponement. Apart from the common pattern in preference ordering, data also reveal striking differences between countries. The relevant question is whether these differences can be explained by solidarity principles. To answer this question, I report the results of multilevel analysis for each dependent variable in Table 3.

Table 3.

Preferences for pension reforms: multilevel ordered logit.

	Raising contributions			Raising retirement age			Allowing private choice
	1	2	3	4	5	6	7	8	9
Individual-level variables
Female	–0.004	–0.003	–0.003	–0.020	–0.020	–0.020	–0.078*	–0.077*	–0.078*
Female	(0.039)	(0.039)	(0.039)	(0.039)	(0.039)	(0.039)	(0.040)	(0.040)	(0.040)
Age	0.014**	0.014**	0.014**	–0.017**	–0.017**	–0.017**	–0.006	–0.006	–0.006
Age	(0.007)	(0.007)	(0.007)	(0.007)	(0.007)	(0.007)	(0.007)	(0.007)	(0.007)
Age²	<–0.001	<–0.001	<–0.001	<0.001**	<0.001**	<0.001**	<0.001	<0.001	<0.001
Age²	(<0.001)	(<0.001)	(<0.001)	(<0.001)	(<0.001)	(<0.001)	(<0.001)	(<0.001)	(<0.001)
Education (ref: ≤14 years)
15–20 years	0.137**	0.139**	0.140**	–0.028	–0.023	–0.024	0.025	0.032	0.033
15–20 years	(0.061)	(0.061)	(0.060)	(0.062)	(0.062)	(0.062)	(0.063)	(0.063)	(0.063)
>20 years	0.224***	0.222***	0.226***	0.232***	0.237***	0.233***	0.087	0.092	0.093
	(0.068)	(0.068)	(0.067)	(0.069)	(0.069)	(0.069)	(0.071)	(0.071)	(0.071)
Marital status (ref: married or living as a couple)
Single	0.065	0.066	0.066	0.032	0.031	0.032	–0.159**	–0.159**	–0.160**
Single	(0.066)	(0.066)	(0.066)	(0.066)	(0.066)	(0.066)	(0.068)	(0.068)	(0.068)
Divorced or widowed	0.113**	0.112**	0.112**	–0.013	–0.013	–0.013	–0.085	–0.084	–0.084
	(0.050)	(0.050)	(0.050)	(0.050)	(0.050)	(0.050)	(0.052)	(0.052)	(0.052)
Ideology	–0.033***	–0.033***	–0.033***	0.046***	0.046***	0.046***	0.081***	0.081***	0.081***
Ideology	(0.010)	(0.010)	(0.010)	(0.010)	(0.010)	(0.010)	(0.011)	(0.011)	(0.011)
Universalistic principle	0.271***	0.271***	0.272***	–0.339***	–0.340***	–0.340***	0.272***	0.273***	0.273***
Universalistic principle	(0.033)	(0.033)	(0.033)	(0.033)	(0.033)	(0.033)	(0.034)	(0.034)	(0.034)
Conservative principle	0.124***	0.125***	0.125***	–0.274***	–0.273***	–0.273***	0.360***	0.361***	0.361***
Conservative principle	(0.030)	(0.030)	(0.030)	(0.030)	(0.030)	(0.030)	(0.031)	(0.031)	(0.031)
Liberal principle	–0.229***	–0.228***	–0.229***	0.157***	0.157***	0.157***	0.255***	0.255***	0.255***
Liberal principle	(0.022)	(0.022)	(0.022)	(0.022)	(0.022)	(0.022)	(0.023)	(0.023)	(0.023)
Familistic principle	–0.262***	–0.261***	–0.262***	0.262***	0.261***	0.263***	0.048**	0.046**	0.046**
Familistic principle	(0.021)	(0.021)	(0.021)	(0.021)	(0.021)	(0.021)	(0.021)	(0.021)	(0.021)
Country–level variables
Welfare regime (ref: social democratic)
Conservative	0.054			0.136			–0.086
	(0.129)			(0.212)			(0.183)
Liberal	0.225			0.540**			0.471*
	(0.171)			(0.265)			(0.274)
Mediterranean	–0.034			0.071			–0.283
	(0.152)			(0.208)			(0.208)
Decommodification		0.006			–0.018			–0.003
		(0.007)			(0.012)			(0.011)
Pension generosity			0.008*			–0.014*			–0.006
			(0.005)			(0.007)			(0.008)
Threshold μ₁	–1.478***	–1.393***	–1.110***	–1.177***	–1.599***	–1.964***	0.044	0.050	–0.179
	(0.258)	(0.266)	(0.322)	(0.273)	(0.315)	(0.421)	(0.262)	(0.328)	(0.461)
Threshold μ₂	0.058	0.143	0.426	0.210	–0.213	–0.577	1.177***	1.184***	0.954**
	(0.257)	(0.265)	(0.322)	(0.272)	(0.314)	(0.421)	(0.261)	(0.326)	(0.459)
Threshold μ₃	1.798***	1.882***	2.165***	1.657***	1.234***	0.870**	3.441***	3.447***	3.218***
	(0.258)	(0.266)	(0.323)	(0.274)	(0.316)	(0.421)	(0.263)	(0.328)	(0.461)
Variance (thresholds)	0.318***	0.310***	0.316***	0.400***	0.407***	0.401***	0.380***	0.407***	0.406***
	(0.040)	(0.038)	(0.039)	(0.062)	(0.070)	(0.061)	(0.040)	(0.060)	(0.056)
No. observations Level 1	9136	9136	9136	9603	9603	9603	9499	9499	9499
No. observations Level 2	15	15	15	15	15	15	15	15	15

Notes: ***, **, and * indicate significance level at 1%, 5%, and 10%, respectively. Standard errors are shown in parentheses.

Source: Eurobarometer 56.1 (2001).

The results show that solidarity principles at the individual level have a highly significant effect on preferences for pension reforms. Adhesion to universalistic and conservative principles increase support for raising contributions and reduce support for raising the retirement age, which is consistent with H1a and H1c (since the universalistic and conservative principles are expected to support policies that guarantee a public protection of incomes after retirement). Conversely, liberal and familistic principles increase support for raising the retirement age and reduce support for raising contributions, which is consistent with H1b and H1d (given that the liberal and familistic principles are expected to oppose policies that imply a greater government control over private incomes). Finally, the effect of solidarity principles on the preference for free choice of private pensions is significant but positive for all of the four principles. This is only partly consistent with the previous hypotheses, since adhesion to the liberal principle was expected to increase support for private choice but adhesion to the universalistic was expected to reduce support for that measure. Thus, that could be the consequence of a framing effect, since interviewees might prefer a measure that is expressed in positive terms (Janki and Gál, 2007), as in the case of expanding the possibility of choice.⁵

An additional question is whether preferences for pension reforms depend also on institutional factors at the national level. Since the Level 2 sample size is small (only 15 countries), three different models have been estimated for each dependent variable. In each model a different country-level fixed effect is included to control for national factors. The first set of models includes a dummy variable for each welfare regime, taking the social democratic model as the reference category. The second set of models includes the decommodified security ratio proposed by Menahem (2007), which measures how social security and social benefits safeguard ‘the individual against the risk of poverty due to lack or loss of income or an inadequate income’ (Menahem, 2007: 70). And the third set of models includes a specific measure of pension generosity (see definition above).

Although the significance of country-levels predictors is rather modest due to the small sample size at Level 2, results are mostly consistent with expectations about the effect of solidarity principles at the national level. That is, the effect of solidarity principles seems to tend toward the same direction both at the individual and the aggregate level. Regarding the effect of the welfare regime, living in a liberal country increases support for raising the retirement age and allowing private choice of pension plans (although the last effect is only significant at p<0.10) but there is no significant effect of the welfare regime on support for raising contributions. The effect of the decommodified security ratio has the expected sign (positive for raising contributions and negative for raising the retirement age and allowing private choice of pension plans), but it is far from being significant. However, if we focus more specifically on the features of the pension system, the generosity index has a significant influence on two dependent variables. The effect is positive for raising contributions and negative for raising the retirement age (although only significant at p<0.10). All in all, results suggest that welfare states providing more generous pension systems foster support for raising contributions and reduce support for raising retirement age. Furthermore, adding country-level controls does not affect the size and significance of solidarity principles at the individual level, which remain unchanged in the three different specifications.

There is still one additional concern about the findings’ robustness. Since respondents are supposed to express their preferences regarding the pension system in their own country, it is important to check whether the effect of solidarity principles on the dependent variables still hold after controlling for the pension policies in place. To do that, I have run an additional set of multilevel models controlling for different characteristics of the pension systems at country-level. In the model for raising contributions, I control for social security contributions as a percentage of GDP (Eurostat, 2012). In the model for raising the retirement age, I control for the effective age of retirement (Eurostat, 2012). And in the model for allowing private choice of pension plans I control for the private pension fund assets as a percentage of the GDP (Palacios and Pallarès-Miralles, 2000). Nevertheless, the size and significance of the effect of solidarity principles remain unchanged.⁶

To understand the substantial impact of the solidarity principles on support for pension reforms, I used estimates from the multilevel models in Table 3 to simulate the probabilities of supporting each reform at different levels of the variables of interest. In the three plots in Figure 1, each line represents the probability of support for each reform (‘strongly agree’ and ‘slightly agree’). The simulated probabilities are computed for a representative individual, male, with between 15 and 20 years of education, married or living as a couple, while holding all the continuous variables at their means. As already said, universalistic and conservative principles induce support for a rise in contributions, while liberal and familistic values lead to the rejection of it. However, the change in probabilities caused by the universalistic principle (from 0.52 to 0.71) is greater than the change associated with the conservative principle (from 0.62 to 0.70). On the other hand, liberal and familistic respondents are more in favour of raising the retirement age, whereas the universalists and conservatives oppose it. In addition, the change in probabilities induced by the universalistic principle (from 0.34 to 0.15) is greater than that of any other variable. In contrast to other reforms, the probabilities of supporting private choice increase for every solidarity principle, as previously explained.

Figure 1.

Support for pension reforms.

Finally, the effect of socio-demographic variables is rather modest though, at least, partly consistent with H2 (those who have better chances of earning high incomes prefer raising the retirement age and allowing free choice of private plans rather than increasing contributions). Gender is only significant in the case of attitudes toward free choice of private pension plans: women are less supportive of that measure. Also consistent with the self-interest approach, age has a positive effect on support for increasing contributions and a negative effect on support for raising the retirement age. However, the effect of age on the last variable is not linear, so we can say that the support for raising the retirement age tends to increase in advanced ages, which is consistent with the fact that the elderly are typically absent in the labour market. Last, more educated individuals are more supportive of increasing contributions and raising the retirement age, which is inconsistent with the self-interest thesis that predicts that the effect of education in both equations would be the opposite. As an alternative explanation, most educated individuals may be more willing to accept reforms if they are better informed about pension system problems (Boeri et al., 2002) or if they are more concerned about the viability of the pension system because of their value orientations.

Conclusions

Pension system reform has been on the political agenda in most European countries for at least two decades. However, although there are several studies on the relation between welfare regimes and preferences for different social policies (Arts and Gelissen, 2001; Jaeger, 2006, 2009; Mau, 2004), the topic of attitudes toward reforms of the pension systems has hardly been addressed in the literature (Boeri et al., 2002). The main goal of this paper has been to provide an explanation of the attitudes toward three different proposals of reform: increasing contributions, raising retirement age and granting individual freedom of choice between private and public pension plans. The central hypothesis has been that preferences for specific reforms are determined, among other factors, by the principles of solidarity that inspire different welfare regimes. However, following Pettersen (1995) and Gelissen (2001), I have argued that these principles are not homogenous within countries, since individuals in different socio-economic positions are expected to embrace different solidarity principles.

Given that each pension reform will have different distributive impacts, I hypothesized that support for a particular reform depends on attachment to different solidarity principles. The empirical findings are consistent with the hypotheses. Multilevel analysis on a sample of EU countries show that individuals who adhere to universalistic or conservative principles are more in favour of raising contributions to maintain the level of pensions, whereas they oppose a postponement in retirement age. Conversely, those who adhere to liberal or familistic principles are against increasing contributions and prefer raising retirement age. The results also support the ‘regime hypothesis’, as they suggest that more generous pension systems increase support for raising contributions and decrease support for raising retirement age. In the end, the likelihood of public opinion resistance to a particular policy does not depend only on the welfare regime but also on the mobilized segments of the population’s commitment to each solidarity principle.

These findings also leave open some important questions for future research. First, further investigations should address the origin of these solidarity principles at the individual level and, more specifically, whether they depend mostly on self-interest motivations or on cultural orientations learned through the socialization process. Second, we need a better understanding of how these solidarity principles at the individual level are related to institutional variables at the aggregate level, as well as whether the distribution of solidarity principles within a particular country informs the design of its welfare institutions, or whether causality runs in the opposite direction. Finally, a more fine-grained knowledge is necessary in order to conclude whether the effect of these principles can be translated to other areas of social policy.

Footnotes

Acknowledgements

I am grateful to Belén Barreiro, Inés Calzada, Elisa Chuliá, Eloísa del Pino and Alberto Penadés for their helpful suggestions on earlier versions of this paper. The usual disclaimer applies.

Author’s note

An early version of this paper was presented at the Seminar at the Centro de Investigaciones Sociológicas (CIS) in Madrid, Spain (2010).

Funding

The author gratefully acknowledges financial support from the Spanish Ministry of Economy and Competitiveness (CSO2011-29346) to conduct this research.

Notes

References

Alesina

Giuliano

(2010) ‘Preferences for Redistribution’, in Benhabib

Jackson

M.O.

Bisin

(eds) Handbook of Social Economics, Vol. 1a. The Netherlands: Elsevier.

Anderson

K.M.

(2001) ‘The Politics of Retrenchment in a Social Democratic Welfare State. Reform of Swedish Pensions and Unemployment Insurance’, Comparative Political Studies 34(9): 1063–91.

Andreβ

H.-J.

Heien

(2001) ‘Four World of Welfare State Attitudes? A Comparison of Germany, Norway, and the United States’, European Sociological Review 17: 337–356.

Arts

Gelissen

(2001) ‘Welfare States, Solidarity and Justice Principles: Does the Type Really Matter?’, Acta Sociologica 44(4): 283–99.

Arts

Gelissen

(2002) ‘Three Worlds of Welfare Capitalism or More? A State-of-the-Art Report’, Journal of European Social Policy 12(2): 137–58.

Béland

Marier

(2006) ‘Avoiding Protest: Labor Mobilization and Welfare State Politics in France’, Mobilization 11(3): 297–311.

Boeri

Börsch-Supan

Tabellini

(2002) ‘Pension Reforms and the Opinions of European Citizens’, American Economic Review 92: 396–401.

Bonoli

(1997a) ‘Classifying Welfare States: A Two-Dimension Approach’, Journal of Social Policy 26(3): 351–372.

Bonoli

(1997b) ‘Pension Politics in France: Patterns of Cooperation and Conflict in Two Recent Reforms’, West European Politics 20(4): 111–24.

10.

Bonoli

(2003). ‘Two Worlds of Pension Reform in Western Europe’, Comparative Politics 35(4): 399–416.

11.

Breyer

Stolte

(2001). ‘Demographic Change, Endogenous Labor Supply and the Political Feasibility of Pension Reform’, Journal of Population Economics 14: 409–24.

12.

Browning

(1975). ‘Why the Social Insurance Budget is Too Large in a Direct Democracy’, Economic Inquiry 13: 373–88.

13.

Browne

Draper

(2006) ‘A Comparison of Bayesian and Likelihood-Based Methods for Fitting Multilevel Models’, Bayesian Analysis 1: 473–514.

14.

Campbell

(2003) How Policies Make Citizens: Senior Political Activism and the American Welfare State. Princeton: Princeton University Press.

15.

Esping-Andersen

(1990) The Three Worlds of Welfare Capitalism. Oxford: Polity Press.

16.

Esping-Andersen

(ed.) (1996) Welfare States in Transition. London: Sage.

17.

European Commission (2003) ‘Eurobarometer 56.1:Social Exclusion and Modernization of Pension Systems’, September–October 2001. European Commission.

18.

Eurostat (2012). Social Protection (ESPROSS). http://epp.eurostat.ec.europa.eu/portal/page/portal/social_protection/data/database.

19.

Feldman

Zaller

(1992) ‘The Political Culture of Ambivalence: Ideological Responses to the Welfare State’, American Journal of Political Science 36(1): 268–307.

20.

Ferrera

(1996) ‘The ‘Southern’ Model of Welfare in Social Europe’, Journal of European Social Policy 6(1): 17–37.

21.

Galasso

(2006) The Political Future of Social Security in Aging Societies. Boston: MIT Press.

22.

Galasso

Profeta

(2000) ‘The Political Economy of Social Security: a Survey’, European Journal of Political Economy 18: 1–29.

23.

Galasso

Profeta

(2007) ‘How Does Ageing Affect the Welfare State?’, European Journal of Political Economy 18: 1–29.

24.

Gelissen

(2001) ‘Old-Age Pensions: Individual or Collective Responsibility? An Investigation of Public Opinion Across European Welfare States’, European Societies 3(4): 495–523.

25.

Goldstein

(2010) Multilevel Statistical Models. New York: Wiley.

26.

Groezen

Kiiver

Unger

(2009) ‘Explaining Europeans’ Preferences for Pension Provision’, European Journal of Political Economy 25: 237–46.

27.

Jacoby

W.G.

(1994) ‘Public Attitudes toward Government Spending’, American Journal of Political Science 38(2): 336–61.

28.

Jaeger

M.M.

(2006) ‘Welfare Regimes and Attitudes toward Redistribution: The Regime Hypothesis Revisited’, European Sociological Review 22(2): 157–70.

29.

Jaeger

M. M.

(2009). ‘United but Divided: Welfare Regimes and the Level and Variance in Public Support for Redistribution’, European Sociological Review 25(6): 723–37.

30.

Jakobsen

(2011). ‘Welfare Attitudes and Social Expenditure: Do Regimes Shape Public Opinion?’, Social Indicators Research 101: 323–340.

31.

Janky

Gál

R.I.

(2007) ‘Public Opinion on Pension Systems in Europe’, European Network of Economic Policy Research Institutes Research Report No. 36.

32.

Larsen

C.A.

(2006) The Institutional Logic of Welfare Attitudes: How Welfare Regimes Influence Public Support. Aldershot: Ashgate.

33.

Leibfried

(1992) ‘Towards a European Welfare State? On Integrating Poverty Regimes into the European Community’, in Ferge

Kolberg

J.E.

(eds) Social Policy in a Changing Europe. Frankfurt am Main: Campus Verlag.

34.

Lynch

Myrskylä

(2009) ‘Always the Third Rail? Pension Income and Policy Preferences in European Democracies’, Comparative Political Studies 42(8): 1068–97.

35.

Mau

(2004) ‘Welfare Regimes and the Norms of Social Exchange’, Current Sociology 52: 53–74.

36.

Menahem

(2007). ‘The Decommodified Security Ratio: A Tool for Assessing European Social Protection Systems’, International Social Security Review 60(4): 69–103.

37.

Menahem

(2008) ‘How can the Decommodified Security Ratio Assess Social Protection Systems?’. Madrid: Instituto de Estudios Fiscales. P.T. 11/08.

38.

McKelvey

R.D.

Zavoina

(1975) ‘A Statistical Model for the Analysis of Ordinal Level Dependent Variables’, Journal of Mathematical Sociology 4: 103–20.

39.

Mehrtens

F.J.

(2004) ‘Three Worlds of Public Opinion? Values, Variation, and the Effect on Social Policy’, International Public Opinion Research 16(2): 115–43.

40.

Meltzer

Richard

(1981) ‘A Rational Theory of the Size of Government’, Journal of Political Economy 89: 914–27.

41.

Menahem

(2007) ‘The Decommodified Security Ratio: A Tool for Assessing European Social Protection Systems’, International Social Security Review 60(4): 69–103.

42.

Menahem

(2008) ‘How can the Decommodified Security Ratio Assess Social Protection Systems?’. Madrid: Instituto de Estudios Fiscales, P.T 11/08.

43.

Moreno

(2000) The Spanish Development of Southern Welfare, in Kuhnle

(ed.) Survival of the European Welfare State. London: Routledge.

44.

Palacios

Pallarès-Miralles

(2000) International Patterns of Pension Provision. Washington, DC: World Bank.

45.

Persson

Tabellini

(2000) Political Economics. Cambridge: MIT Press.

46.

Pettersen

P.A.

(1995) ‘The Welfare State: The Security Dimension’, in Borre

Scarbrough

(eds) The Scope of Government: Beliefs in Government (vol. 3), pp. 198–233. Oxford: Oxford University Press.

47.

Pierson

(1996) ‘The New Politics of the Welfare State’, World Politics 48(2): 143–179.

48.

Pierson

(1998) ‘Contemporary Challenges to Welfare State Development’, Political Studies XLVI: 777–94.

49.

Rabe-Hesketh

Skrondal

Pickles

(2005) ‘Maximum likelihood estimation of limited and discrete dependent variable models with nested random effects’, Journal of Econometrics 128: 301–23.

50.

Raudenbush

S.W.

Bryk

A.S.

(2002) Hierarchical Linear Models. Applications and Data Analysis Methods. Thousand Oaks: Sage.

51.

Razin

Sadka

(2007) ‘Aging Population: The Complex Effect of Fiscal Leakages on the Politico-Economic Equilibrium’, European Journal of Political Economy 23(2): 564–75.

52.

Sears

Funk

(1990) ‘The limited effect of economic self-interest on the political attitudes of the mass public’, Journal of Behavioral Economics 19: 247–71.

53.

Svallfors

(1997) ‘Worlds of Welfare and Attitudes to Redistribution: A Comparison of Eight Western Nations’, European Sociological Review 13: 233–304.

54.

Svallfors

(2003) ‘Welfare Regimes and Welfare Opinions: A Comparison of Eight Western Countries’, Social Indicators Research 64: 495–520.

55.

Taylor-Gooby

(1999) ‘Policy Change at a Time of Retrenchment: Recent Pension Reform in France, Germany, Italy and the UK’, Social Policy and Administration 33(1): 1–19.

56.

Taylor-Gooby

(2001). ‘Sustaining State Welfare in Hard Times: Who will Foot the bill?’, Journal of European Social Policy 11: 133–47.

57.

Tepe

(2006) ‘Individual Preferences for Old Age Pensions. Evidence from the International Social Survey Program’, paper presented at the FNA Graduiertenkolloquium der Deutschen Rentenversicherung, Berlin, Germany.

58.

Tepe

Vanhuysse

(2009) ‘Are Aging OECD Welfare States on the Path to the Politics of Gerontocracy? Evidence from 18 Democracies, 1980–2002’, Journal of Public Policy 29(1): 1–28.

59.

Velladics

Henkens

Van Dalen

H.P.

(2006) ‘Do Different Welfare States Engender Different Policy Preferences? Opinions on Pension Reforms in Eastern and Western Europe’, Ageing and Society 26: 475–95.

60.

Weaver

R.K.

(2003) ‘The Politics of Public Pension Reform’, Working Paper 2003–06, Center for Retirement Research, Boston University.

61.

Whitehouse

E.R.

(2006) Pensions Panorama: Retirement-Income Systems in 53 Countries. Washington: World Bank.

62.

Whitehouse

E.R.

Queisser

(2007) Pensions at a Glance: Public Policies Across OECD Countries. Paris: OECD.