Dynamic Cleavages of ‘Welfare Rights and Duties’ in Public Attitude Towards Old-age Pensions: A comparative study

In an era of economic hardship, welfare states are unsustainable without a sound level of public support. Consequently there have been an increasing number of studies about public attitudes towards the welfare state in recent times. In this study, we focus on public attitudes towards public pension programs, which have been regarded as key pillars of the welfare state. In fact, they have become one of the most visible examples of the development of social rights and a symbol of public achievement in most welfare states. Pension programs are associated with relatively high levels of benefits and serve a broad range of recipients. The elderly have long been considered as ‘deserving’ welfare recipients whom social policies must protect (Coughlin 1980, in Larsen 2008: 149). Old-age pension policies have proved to be widely supported amongst welfare programs (van Oorschot 2000). This seems true even during the current era within which many nations are facing fiscal problems and are thereby struggling with cutback politics.

There is a series of plausible explanations for public support for pension programs, each cast from competing theories of welfare rights development. The irreversibility thesis argues that the maturation of public welfare programs has resulted in strong political support for the welfare state itself (Mishra 1990). As this paper is being written, the world is witnessing people marching in the streets of European cities, protesting against pension reforms involving cutbacks. This phenomenon can be explained also by the famous formalisation of the median voter model approach. The aged and the soon-to-be-aged possess vested-interests in old age programs, and therefore they tend to support public policies devised to enhance older peoples’ rights (Persson and Tabellini 2000; Galasso 2006). This is a vivid example of political cleavage derived from peoples’ rational choices in relation to the enhancement of their rights. However, the interest group thesis claims that widespread and strong pensioners’ groups tend to stand by their own interests, supporting their given welfare rights in spite of the fiscal problems associated with the maintenance of these rights. Indeed, older people seem to possess distinct interests in the face of pension politics. In sum, all these existing theoretical explanations emphasise the same message: public pensions receive overwhelming support especially from older people. As we become older it is natural for us to support our own pension rights.

Meanwhile, changes in the demographic structure combined with generous benefit levels have brought about serious financial pressure on welfare states. When we look into the spending structure of welfare states, it is evident that public pension provision is the most expensive welfare program, with the largest portion of social spending being dedicated to this post in every country. In the mass media a series of debates over related issues appears daily. Repeatedly government officials or politicians issue stern warnings of inevitable future increases in tax burdens due to population ageing, resulting in intergenerational equity problems. This is one of the most significant reasons for why public pension programs have been the major target of rollback reforms among other welfare programs. This is a relatively new issue of the welfare state in that duties, not only rights, are being emphasised when grappling with the issue of maintaining the welfare programs.

To capture both sides the coin, a new theoretical framework based on an integration of perspectives, in which costs as well as benefits of the welfare state are considered at the same time is needed. Hence, one of the major purposes of this study is to introduce a useful theoretical framework to explain recent changes in the characteristics of welfare political cleavages. Since Ahn (2000, 2002) proposed a new welfare political research framework, namely the ‘welfare status model’, a series of empirical studies have proved the model to be useful in understanding the new politics of the welfare state, especially with respect to political changes taking place after the oil crisis of the 1970s (Ahn 2003, 2009; Ahn and Jung 2010; Kim and Ahn 2010). While this framework can be helpful in understanding welfare politics in general, we will apply the ‘welfare status model’ to pension politics in this study.

In addition to the theoretical integration of welfare rights and duties, this study also tries to bridge the gap between micro and macro orientations of public attitude research. While much research looking into public attitudes towards the welfare state has examined political cleavages in an empirical manner, there has been a clear division between ‘macro and micro’ approaches.

However, micro analyses of public preference or voting behaviour have focused on political cleavages according to self-interests or political ideology within a single country (Svallfors 2004, 2008; Jaeger 2006; Blekesaune 2007; Bonoli and Hausermann 2009). These studies have reached a similar conclusion, namely that not only class cleavage but also other factors such as age, gender, and the like affect public attitudes towards the welfare state. More recently, the relative importance or inter-relationship between class- and age-based cleavages has aroused a growing interest. Several single-country studies on public attitudes towards age-biased policies underscore the impact of age-related cleavages rather than class ones (Svallfors 2008; Bonoli and Häusermann 2009). Svallfors (2008) raised a question regarding the relationship between class and age. Further empirical evidence is needed before his argument can be confirmed.

However, comparative analyses of public attitudes have tended to address the effects of welfare regimes on citizens’ preferences, which supports what the logic of institutionalism suggests (Papadakis and Bean 1993; Svallfors 1997; Bean and Papadakis 1998; Andreß and Heien 2001; Arts and Gelissen 2001; Jaeger 2006; Larsen 2008). Although such studies offer meaningful insight into the relationship between the institutional structures of the welfare state and citizens’ attitudes towards the welfare state, most of them have failed to provide any generalised explanations about the underlying mechanisms of welfare politics. Some studies choose only three or four cases that are close to the ideal types of each welfare regime (Evans 1996; Svallfors 1997; Heien and Hofäcker 1999; Andreß and Heien 2001; Taylor-Gooby 2001), and as such it seems imprudent to expand the results and apply them to broader cases. Other research has found little significant evidence for an institutional effect of a particular welfare regime on public attitudes (Bean and Papadakis 1998; Gelissen 2000; Arts and Gelissin 2001).

Some recent studies have tried to overcome this kind of macro- and micro-division in empirical studies. For instance, Gelissen (2001) investigated the impact of structural characteristics of pension systems as well as social characteristics of individuals on public preferences using logistic regression. Multilevel models have gained increasing popularity in the literature on welfare state attitudes (Blekesaune and Quadagno 2003; Jaeger 2006) as they enable the researcher to distinguish between within country (or between-individual) and between-country sources of variation at the same time.

In line with this approach, our study takes a comprehensive stance in that it assumes that public attitudes towards a welfare policy, in this case pensions, is systemically formulated by the structure of the welfare states on the one hand and by individuals’ socio-economic positions on the other. For example, if tax levels are high in a country at the same time benefits, especially for older people's welfare, are generous, the country may undergo a hard time when it faces the need to address other social risks, such as child care. When such budget conflicts arise according to age issues, an age-biased spending structure often results in a ‘generational equity’ argument, and in turn generally affects individuals’ attitudes towards pension expenditures especially where there is a high tax burden. In this sense, macro-level theories and micro level explanations of welfare politics must be linked together to capture the reality of public support for welfare states. All in all, this study develops an encompassing methodological strategy to combine national- and individual-level approaches by employing multi-level models as an analytic tool.

So far we have suggested a theoretical and methodological integration for addressing public attitudes towards pension politics. Let us now discuss the given themes in more detail. First, we will introduce the welfare status model, an integrated theoretical framework addressing rights and duties, as the underlying mechanism for political cleavages on pensions. Thereafter we will introduce operational hypotheses to be tested in the empirical analysis. In the third chapter, we will present our data, operational definitions of variables, and an analytic method for applying the multi-level model. Then we will show and discuss the results of the multilevel analysis. Finally, we will conclude by highlighting the main findings and discussing their implications.

Voluminous works on political approaches to social policy have been produced to address the various aspects of welfare-related interest cleavages. Traditionally, class has been regarded as the major source of cleavage in the study of the welfare state since Marshall (1950) lectured on the development of citizenship rights, and Korpi (1983) presented the power resources theory of democratic class struggle. However, those scholars who have recognised the decline or even demise of class cleavage have emphasised a series of alternative sources of social division such as generation, gender, race, ethnicity, employment sector, consumption, etc. (Ahn 2000).

In this context, a few mainstream arguments on welfare states have contributed to reveal the more complicated nature of current welfare politics. The new politics theory claims that the retrenchment politics of the welfare state differ from expansion politics. Among others, Pierson (1996, 2001) suggests that welfare politics operate as a combination of various political cleavages rather than being drawn along a simple class-based line. More recently, the new social risk theory has argued that the transition into post-industrial societies involved newly emerging social risks of care or employment in changing family and labour market structures as well as the perseverance of old risks such as the ageing issue (Taylor-Gooby 2004; Bonoli 2007). This more recent theoretical tradition convincingly draws attention to new cleavages in relation to the distribution of limited public resources to support welfare. Existing studies from this approach provide valuable insights on other-than-class cleavages of welfare politics such as age (Hammarström 1998; Svallfors 2008; Bonoli and Häusermann 2009) or gender (Svallfors 1997; Kim and Ahn 2010) even though some research still underscores the significance of class-based cleavages (Brooks and Svallfors 2010; Häusermann 2010).

However, neither traditional theories of class-based cleavage nor new approaches of other-than-class cleavages seem to pay serious attention to seeing the whole picture of welfare politics. They all share a crucial blind spot as they focus on just one side of social citizenship – rights. They have paid attention to ‘who gets more (or less) from a welfare state’ to explain cleavages in welfare politics, but have overlooked ‘who pays more (or less) for it’. Since the burden on the working generation has been presented as an important rationale for welfare cutbacks (Walker 1996), a theoretical focus exclusively on ‘rights’ cannot adequately grasp the dynamics of the current welfare retrenchment process (Ahn 2000). This is especially true in the case of pension reform.

The welfare status model (Ahn 2000, 2002), to the contrary, underscores that social citizenship necessarily entails duties as well as rights. Especially in terms of the welfare-related aspect, Ahn (2000) suggests that the structure of social citizenship comprises two categories: status of welfare rights and status of welfare duties. In simple terms, status of welfare rights has to do with the status of a welfare beneficiary, while the status of welfare duty has to do with the status of the taxpayer or welfare burden-taker. The status structure becomes the source of divergent political cleavages shaped by the self-interests of individuals in relation to the nature of welfare programs, and accordingly, shapes personal attitudes towards welfare programs and eventually influences future politics of the welfare state (see Figure 1).

The application of the welfare status model of rights and duties in this study enables us to incorporate the various theoretical notions of the major cleavages into a single welfare-relevant framework. First, age or generation have commanded growing attention in welfare politics. ‘Intergenerational equity’ is often argued as the rational reason for pension cutbacks, of which an unintentional result is the risk of the growth of intergenerational conflict (Walker 1990). The ‘crowding out effect’ thesis suggests that more generous public spending on the elderly with increasing population aging would result in less developed policies for the younger generations (Bonoli and Reber 2010). This implies the emergence of an interest conflict between generations. In line with these arguments, there is some empirical evidence of generational cleavage. Examining voting behaviours in democratic referendums in Switzerland, Bonoli and Häusermann (2009) concluded that age is the most influential line of conflict on issues related to pension and labour market policies. Similarly, Hammarström (1998) found that age differences are the most obvious determinants of attitudes towards public and private financing of home help services for the elderly in Sweden.

In this study we argue that generational cleavage occurs due to the gap between the rights of the elderly to receive benefits and the duties of the younger to pay for these benefits. Discord or even competition over limited public resources among generations can be explained as a conflict between the different statuses of welfare benefits as well as the tension between the different statuses of welfare duties. Although an expectation of future benefits could make the younger generation inclined to support old-age policies to some extent, a sense of current burden usually supersedes a sense of future benefit.

Second, class-based cleavage can be seen to derive from interest conflicts over social rights as welfare beneficiaries (or anticipated recipients) and social duties as those paying the cost. Public pensions generally include redistributive elements to protect low income groups, which are paid by the higher-income earners. Therefore, when a government spends more on a pension scheme, lower income groups will benefit more from it, which implies that higher income groups have to pay for it either through taxes or pension contributions. A few earlier studies report that the occupation-based class division matters in public attitudes towards the welfare state in general and pensions in particular (Svallfors 1995, 1997, 2004; Brooks and Svallfors 2010; Häusermann 2010), while Bonoli and Häusermann (2009) found that education and income level are less important than age in explaining individual preferences. Svallfors (2008) pointed out that age and class are competing sources of welfare cleavages.

Third, gender cleavage seems to occur in pension politics as well as in other welfare conflicts because women are more likely to hold the status of welfare beneficiary. Programs designed as redistribution or compensation measures for unpaid work, e.g., care credit, also benefit women due to their disadvantaged status in the labour market. Furthermore, as women tend to live longer than men, on average they will receive pension benefits for much longer (Piven 1985; Hernes 1987; Wennemo 1994). Some researchers point out that gender issues are closely linked to public and private sector cleavages (Esping-Andersen 1990; Crompton 1993; Svallfors 1997; Ahn 2000). As state-employed service providers are mostly women, women are likely to be more reliant on welfare state provisions as employees of the state. In contrast, men usually possess a higher status in the labour market and therefore it is more likely they will be burdened by welfare programs.

All in all, political cleavages in the political battle field of welfare states is summarised as interest conflicts between the status of welfare rights and the status of welfare duties. Based on the theoretical framework of welfare rights and duties, this study will examine whether people actually express welfare-political interests according to class, age and gender. Among other sites of cleavages, this study sheds light especially on pension politics. Additionally, we will examine the relative weight of these three sources of cleavage. Studies to date dealing with public attitudes towards age-related policies seem to suggest that age is more significant than class or gender.

When we statistically test the above-implied hypotheses of micro-level interest calculations based on welfare rights and duties, we need to take macro-level confinement into consideration as well. This is because people's interest calculation or judgement of welfare rights and duties tend to be influenced by the nature and structure of the welfare state where they live (Ahn 2000). For instance, a young worker who lives in a country where the tax burden is high and welfare programs are relatively generous for the aged, would be in a position of double disadvantage as he or she experiences a greater burden and smaller immediate benefits. In contrast, a young worker who lives in a different country where child care programs as well as public pensions are evenly developed and the tax burden is comparatively low faces rather different circumstances. In this second country young people are likely to regard their welfare state as fair. Controlling for all other factors, the first worker will make a negative judgement about public pensions while the second worker will have a more positive attitude towards pensions. This example illustrates so-called budget competition within welfare programs. In the first case we can say that the welfare structure is old-age biased and the younger generation has to pay for their elders while they gain very little from the welfare state. In other words, the first country is a vivid example of the potential for conflicts between the elderly and the young. In such cases, we can expect that people, especially the younger generation, will be more likely to have negative attitudes towards spending on elders. In the same vein, spending level or tax ratio, for example, might be an additional confinement that has an impact on individual attitudes towards welfare programs. If the first country has a very high level of total spending and therewith a high tax burden, the above-mentioned phenomenon of budget competition becomes much more severe. We also need to remember that impact may vary according to the existence of other factors, and that sometimes the direction of impact might change from positive to negative.

In addition to the above mentioned theoretical points, there is at least one other situation in which the impact of age and class are interlocked. Some scholars have noted that class-based interests may offset generational interests or vice versa (Pampel 1994; Kohli 2006; Svallfors 2008), but this has not been fully examined yet. Our welfare status explanation can also subsume these notions of interaction between class and age. A higher income in old-age may imply less dependence on public benefits; that is, a weaker status of welfare benefits, and greater ability to satisfy needs through personal ownership or private pension. Therefore, higher incomes will decrease differences resulting from age interests. We will also test this hypothetical relationship of interaction between age and class.

The hypothetical relationships between conceptual variables discussed earlier can be summarised as follows: being old or being female is associated with a strong status of welfare benefits and weak status of welfare duties; the higher the social class the more negative the status of welfare benefits and positive status of welfare duties; the higher the pension levels are, the more citizens’ and especially elders’ statuses of welfare benefits are enhanced; higher tax ratios bring about stronger statuses of welfare duties (Table 1).

Based on the above, the hypotheses we will examine are summarised as follows:

National level hypotheses

• The greater the pension benefit, the more support there will be for pension efforts (welfare rights thesis).

• The greater the tax burden, the less support there will be for pension efforts (welfare duties thesis).

• The older a person is, the more they will support pension efforts (generational cleavage thesis).

• The higher a person's income or education level is, the less he or she will support pension efforts (class cleavage thesis).

• Women will support pension efforts more than men (gender cleavage thesis).

• The higher a person's income, the less age will affect their support (age–class interaction thesis).

This article reconstitutes data in order to analyse macro and micro factors which affect public attitudes towards pension efforts. All the micro-level data originate from the International Social Survey Program (ISSP) dataset, particularly ‘the role of government’ module surveyed in 2006. Our national-level variable data come from OECD (2005, 2010). The analysed data come from 21 welfare countries (Australia, Canada, Czech Republic, Denmark, Finland, France, Germany, Hungary, Ireland, Israel, Japan, The Netherlands, New Zealand, Norway, Poland, Portugal, Spain, Sweden, Switzerland, the United Kingdom and the United States). Among the 25 countries for which ISSP data exists, Chile, Slovak Republic, and Slovenia are not included due to lack of availability of macro-level data. South Korea, an outlier case where the public pension system has a very short history, is also excluded.

To address both micro- and macro-level factors and test our hypotheses, we carry out a multi-level analysis. Our analytical dataset, in which individual unit surveys of a few countries are combined, implies that individuals within a country show stronger similarity than those between countries. Therefore, it is difficult to confirm the assumption of independence among individuals. In this case, multi-level analysis is a more suitable method than the traditional regression approach (Raudenbush and Bryk 2002).

Following a study by Svallfors (2008) which used a single item rather than multi-item summarised measures to examine attitudes towards age-specific policies, the model in this study also took a single dependent variable of attitudes towards pension spending. It was measured by the ISSP questionnaire asking ‘whether you would like to see more or less government spending on old age pensions (remember that if you say “much more”, it might require a tax increase to pay for it)’ with 5 scale measurement of ‘spend much more’, ‘spend more’, ‘spend the same as now’, ‘spend less’ and ‘spend much less’. As we make a reverse coding of this for convenient interpretation, the larger the score, the more support it represents in our analysis.

There are two different levels of independent variables. At the national level, we test the welfare statuses of rights and duties using two variables. With the limited number of cases at the national level (n=21), only a few national-level variables can be included in the model. A useful proxy variable for welfare rights in the present study is the pension benefits level. Among indicators of the pension benefits level, we employ the weighted average relative pension level proposed by OECD (2005), defined as the average of the pension entitlement using the earnings-distribution weights expressed as a percentage of economy-wide average earnings. By considering the different earning levels and condensing these into a single figure, we arrive at an appropriate indicator of the scale of the pension promise for the current working aged population (OECD 2005).

The other national-level variable representing the scale of welfare burden is the tax ratio calculated as the total tax revenues expressed as a percentage of GDP (OECD 2010). The tax ratio includes social security contributions as well as all other taxes such as personal income tax, consumption taxes, etc. One might argue that only pension-related taxes should be included in the analysis of public attitudes towards pension spending. However, this is not a simple matter because each country finances spending according to the specific structure of its pension system. For example, in some countries general revenues rather than contributions are the main source of financing of their pensions. In other countries, small portions of consumption taxes or eco-taxes are channelled into pension budgets. Given the complexity of financing and data availability, using the total tax ratio is considered reasonable. For these two national level variables, we use 2005 figures in order to fix a point of time for public attitudes data.

At the individual level, age, education, subjective social class and gender are considered as our main independent variables, representing generational, class-based, and gender interest, respectively. Since our concern is not the actual age but rather the age groups relevant to welfare cleavages, we deal with age as an ordinal variable grouped by 20s, 30–40s, 50s and over 60s. As for gender, men are set as the reference to allow us to examine whether women are more inclined to support pension spending than men.

As for class-related variables, there are data constraints. Income and education are usually used together as a proxy of class-based interest (Bonoli and Häusermann 2009). In the ISSP dataset, education, an ordinal variable measured with a 6-point scale, is a comparable and also available across the countries. However, incomes are hardly comparable as country-specific earnings were surveyed according to income sections defined according to different intervals in different countries. Alternatively, subjective social class measured by top bottom self-placement with a 10-point scale can be applicable. The participants were asked the following question (here quoted from the Finnish questionnaire): ‘There are different classes in our society. Some social classes are higher and some lower. On a scale from 1 to 10 where 10 is the highest class and 1 is the lowest, where would you place yourself?’ This variable is also problematic because no UK case is available. Therefore, we analyse two different models to supplement each other: one including education only (model 1), and the other including subjective social class and education at the expense of the UK cases (model 2).

Finally, the interaction term between age and education or social class is respectively included to test the interaction effect between generational and class-related cleavages.

We examined the fixed effect of national and individual level factors affecting attitudes towards pension spending. The results from the multi-level analysis are shown in Table 2.

Controlling for the tax ratio, we see that the pension level, a proxy variable of pension rights, shows a positive effect on public attitudes towards pension spending at the 0.10 level of significance in all the models. This result coincides with our prediction that a higher pension level enhances the status of welfare rights. With a generous pension promise, people may expect that the public pension will guarantee a decent standard of living in their later lives.

At the same time, the tax ratio proves to negatively affect public support for pension spending when the pension level is controlled. This means that people living in a welfare state where welfare burdens are higher tend to be less supportive of pension efforts. This provides explicit evidence that the status of welfare duties plays a significant role in pension cleavages at the macro-level. This is in line with findings by Gelissen (2001), who found that the citizens of the mature welfare states prefer less state intervention in pensions as they may incur high taxes on their incomes, and have a stronger preference for private responsibility or shared employer/employee responsibility.

In sum, welfare statuses of rights and duties proved to influence pension politics in welfare states at the national level. The coefficient of the pension level is smaller than that of the tax ratio, however. Recent pension reforms imply that the current working generation cannot expect current levels of pensioners’ benefits that they have to pay for. This generational inequity may undermine the current younger generation's status of welfare rights, and enhance their status of welfare duties in the shaping of their attitudes towards pension spending.

To make a long story short, generation-, gender- and class-related cleavages seem to work together in the pension politics of welfare states at the micro-level, which confirms the findings of previous research. As for generational cleavage, the study results statistically substantiate the fact that age has a positive effect on support for pension spending as expected. The elderly want more spending on pensions than do younger people. In terms of the welfare status thesis, we can expect younger generations to tend to be unwilling to make additional payments to increase pension spending. Their ‘expected’ status of rights a few decades later would incline them to support the government's spending on pensions to some degree. But their support proved to be much weaker than that of older generations who currently receive benefits or expect them in the near future. In other words, younger people seem to be inclined towards status of duties rather than rights in the face of pension politics. Svallfors (2008) also found that older people are more supportive of older people.¹

Evidence of class-related cleavages in relation to pension efforts is indicated across all the models analysed. People with a higher education level and social class tend to be less supportive of the government's pension efforts as they experience higher tax burdens. Gelissen (2001) also revealed that people with higher education are more likely to prefer either individual or shared collective responsibility for pension provision over state responsibility.

However, judging from the standardised coefficient values in Table 2, class-related variables seem to have relatively less effects in comparison to age. The result is analogous to the findings of Svallfors (2008) in that age differences are larger than class differences on average.

Thirdly, the importance of gender cleavages on public attitudes towards the government's pension effort is strongly confirmed. Our argument that women are more favourable to state provisions than men is not new, as many authors already insisted that women tend to be more supportive of welfare policies in comparison to men (Svallfors 1997; Ahn 2000; Gelissen 2001; Blekesaune and Quadagno 2003; Iversen and Rosenbluth 2006; Bonoli and Häuserman 2009). What is interesting here is the fact that, in terms of relative weight, the gender effect is stronger than age or class-related effects, especially in the models without the interaction term (models 1-1 and 2-1). This is in contrast to some previous single-country studies showing that age is the most significant cleavage line in welfare politics other than gender, employment sector or class (Hammarström 1998; Bonoli and Häusermann 2009). The welfare status thesis gives us a plausible explanation for women's strong inclination towards state responsibility, particularly regarding social policies for the aged. As we mentioned earlier, women live longer than men on average, and are exposed to economic insecurity more severely and for longer periods than men. In addition, while men are more likely to be covered by occupational pensions provided by employers in addition to their public pensions, public programs run by the state are the main or sometimes the only source of old-age income for most women. In other words, women tend to have far stronger positions as welfare beneficiaries with regard to public pensions. Furthermore, as women are generally relatively more disadvantaged in the labour market, they thereby end up with an unstable labour market status and lower wages, and are apt to have weaker positions as taxpayers than men.²

Finally, the interaction term between age and education (or social class) was tested in Models 1-2, 2-2 and 2-3 to examine whether the education (or social class) effect differs according to age and vice versa. The significant and negative coefficients of the interaction terms in Model 1-2 (β=−0.021, p<.05), Model 2-2 (β=−0.008, p<.10) and Model 2-3 (β=−0.008, p<.05) illustrate that the age effect decreases as education or social class increases. In other words, the age difference in the attitude towards pension spending is much smaller in the groups with higher education or social class and bigger in the other groups.

Figure 2 clearly shows that this interaction effect results in different pictures according to the different combinations of the two variables. The degree of support for pension spending decreases as social class increases (social class effect) with certain gaps among different age groups (age effect). The gaps among age groups narrow as social class becomes higher (age–social class interaction effect). The mechanism by which class-related interests offset generational interests can be explained as follows. Older generations belonging to higher social classes tend to be less dependent on public pensions compared to lower social class groups as they usually have other sources of income such as private pensions or savings.³ This condition seems to make them less supportive of public pensions compared to their lower class counterparts.

By looking even deeper into the interaction effect between age and class-related variables, we can conclude that age and class are not only competing sources of welfare cleavage, they jointly influence the overall picture of welfare politics (Pampel 1994; Kohli 2006; Svallfors 2008).

This study investigated the dynamics of public attitudes towards pension efforts in welfare states. It dealt with three major issues that previous research in this field has not covered. The first was the theoretical modification of political cleavages on welfare, especially pension politics by employing the welfare status thesis of rights and duties. To overcome the bias of existing rational choice theses based on self-interest assumptions, which have focused on ‘who gets more from welfare states’, we suggested a balanced approach by adding the complimentary aspect of ‘who pays more for it’. Based on our empirical findings we confirmed that the welfare status model of rights and duties is a powerful theoretical framework providing simple and clear predictions on political cleavages on pension spending. The validity of the model proves that not only the benefit dimension of welfare states but also the cost dimension is at play in the underlying mechanisms of political cleavages in the battlefield of welfare states.

Second, this study developed an empirical model encompassing macro- and micro-level factors operating behind individuals’ attitudes in a single analytical approach. We found that the structural features of welfare states – tax ratio as well as generosity of benefits – are systematically related to citizen's attitudes towards pension spending. Controlling for these macro factors, three micro factors of gender, class and age were statistically confirmed to be the main sources of cleavage on pension politics. Generational cleavage seems to be more crucial than class division in explaining support for state spending on pensions. People with higher education or social class are less likely to prefer state responsibility for pension provision, which indicates the significance of burden politics and a possible preference for private solutions. Gender-based interests seem to be more significant than, or at least as significant as, generational interests in determining public attitudes towards pension efforts. It implies a growing concern that enhancement of the private element in pensions or development of a multi-pillar system may have clear gender implications.

Third, one of the interesting findings of this study must be the dynamic interplay between age- and class-related variables. Although age seems to be a more influential factor behind pension politics, class-related factors also dynamically change the impact of age. The findings reveal the fact that class-based interests slightly offset age-based interests in higher social classes, while the age effect is more pronounced in lower class.

Using large scale data available from Western industrial societies, this study has sought to draw conclusions about welfare political cleavages that can be applicable to more welfare states. To capture the distinctiveness of pension politics, we may need to understand similarities and differences between political cleavages on pension and other policies. As such, this topic should be the focus of further research. Additional in-depth investigation on the possible interplay between macro structures and micro statuses might be another issue that can be dealt with in another study.

Sang-Hoon Ahn is Professor of Social Policy at the College of Social Sciences, Seoul National University, Seoul, South Korea. He has been involved in a series of international as well as national research projects on the comparative welfare state, especially in terms of political interpretation of recent changes.

Soo-Wan Kim is Assistant Professor of College of Social Welfare at Kangnam University, Gyeonggi, South Korea. She has recently done research on work and welfare, especially in the fields of pension policies and labour market issues.