GENERATION IN ITSELF OR FOR ITSELF?: The conflict potential of cohorts in the German, Dutch and British pension systems compared

‘Will the Third World War be a war between generations rather than states?’ This rather provocative question was posed in a leading article of the British newspaper The Guardian in 1990.^1,2 The article dealt with a publication of the three social historians Paul Johnson, Christoph Conrad and David Thomson (1989). In this book, titled ‘Workers vs. Pensioners’, the three editors collected articles on demographic change, on the growing scarcity of financial resources in social insurance systems and on possible resulting generational conflicts arising from these developments. In the two decades since the publication, ideas of a violated generational justice and of possible generational conflicts have become commonplace in discussions about the future of the welfare state. Numerous conferences on ‘intergenerational equity’ have been held and several welfare reforms have been evaluated with respect to their ability to restore ‘intergenerational justice’. The idea of disparity between birth cohorts and possible resulting conflicts about scarce financial resources has also been taken up by the media in Western welfare states, it has become a part of public debates (Edmunds and Turner 2002: 3; Szydlik 2004: 10–11; Conrad 2003: 55; Kohli 2006: 456).

The problem of intergenerational equity is especially feared in Pay-As-You-Go (PAYGO)-Systems, most notably in pension schemes. In these systems, the currently working generations pay contributions that are diverted directly to the retired to pay for their pension benefits. Once the workers retire themselves, they can hope to be supported by the younger working cohorts. Due to the drop in fertility rates and growing life expectancy, the ratio between workers and pensioners has become disadvantageous, with ever smaller groups of working cohorts having to support ever larger groups of pensioners. This development is seen especially unfavourably in times of economic recession and welfare retrenchment since financial resources become even scarcer (Thomson 1989, 1996; Leisering 2000a,b; Bude 2003, 2005; Willetts 2010).

The main question posed by this article is whether there is empirical evidence for special cohorts to form generational interest groups because of their specific location in historical time. For a thorough analysis of this question, two sets of characteristics need to be taken into account. First, there needs to be a description of possible external factors shaping generations in the welfare state: these can be financial and non-financial resources and changing patterns of welfare institutions in general. If one takes the socio-historical concept of ‘generation’ seriously, however, a description of these external factors on its own will not suffice. It is also of vital importance whether special cohorts perceive themselves as winners or losers in the context of the welfare state. This is especially relevant in the context of possible generational conflict: generations need to be aware of their differing status in the welfare system in order to form a conflict party, at least on a latent basis. The research focuses on pension systems, since these are state-of-the-art PAYGO-schemes and are even often described as ‘generational contracts’.

In order to take different distributive logics into account, three different pension systems, those of Germany, Great, Britain and the Netherlands will be compared. First, there will be an account of the theoretical background of ‘generation’ in general and ‘welfare generation’ in particular. After a short description of the pension systems in Germany, Great Britain and the Netherlands, possible sources for the formation of pensioner generations in the three countries will be evaluated. This entails external influences as well as evidence from opinion studies. A conclusive chapter will combine the findings and evaluate the concept of ‘generation’ as a category of social inequality and a catalyst for social conflict in the welfare state.

Pension systems have always entailed a source of possible conflicts between members of different ages. Since the implementation of the Social Security Act in 1935 the United States has been debating if the comparatively generous benefits for the aged (Social Security and Medicare) should rather be invested in the young (Walker 1996: 26). In the context of the British pension system, Richard Titmuss (1963) pointed to an exaggeration of negative effects of demographic ageing in public debates as early as the 1960s. After the implementation of the German pension system in 1957, the liberal economist Friedrich von Hayek even envisaged concentration camps for the elderly, in case of a growing dependency ratio of pensioners (Hayek cited in Conrad 2003: 64). These debates point to differing conflict lines within the pension system that often get confused. There is a conflict between age groups, e.g., the young vs. the old. It is not an entire cohort that is at disadvantage but the young in general – this holds true for every cohort successively. Viewed from a life cycle perspective, every individual has equal advantages and disadvantages in the same life cycle periods respectively: everybody will belong to the young at one point in the life course and become a member of older age groups in due course as well. Therefore, an individual experiences all the positive and negative sides of the different age groups, irrespective of the year he or she was born. This is true at least for an abstract and short-term perspective in which historical variables and welfare institutions have a uniform influence on individual lives. In the long-term life cycles certainly differ historically: changing roles and institutions will affect age groups differently at distinct points in time.

The second strain of argument differs insofar as it does not entail a question of distribution throughout the life cycle. Instead, entire cohorts of individuals are believed to be at advantage or disadvantage because of a change in the population structure or specific economic developments. Due to their birth into a certain year, they are treated differently at a specific age than the cohorts before or after them. In the context of pensions, it is therefore not necessarily important if someone belongs to the old or the young. Instead, different birth cohorts are treated differently in their old age – some get generous pension benefits, others do not. From this perspective, equalisation in the course of the life cycle is not possible any more.

These different strains of argument get mixed up very often, which leads to confusion of the concepts. In order to differentiate the perspectives, it is vitally important to distinguish between the concepts of age group and cohort or generation. An age group is a group of individuals who are in the same stage of their biography, i.e., the young or the old. An individual changes age groups as he or she gets older. A cohort cannot be changed; it defines a group of people born in the same or several neighbouring years. The concept of generation is based on the cohort, but it requires further characteristics: the members of a generation have undergone a common socialisation and share common experiences, values or goals. Although the definition of generation is changing with different theoretical and empirical approaches, the focus on a common group experience always remains at the heart of every socio-historical theory of generation (Jureit 2006: 78).

It seems to be impossible – at least in the social sciences – to deal with the concept of generation without mentioning the name of Karl Mannheim. In 1928, he explained in his article ‘On the Problem of Generations’ that individuals who participate in the same socio-historical context or generational layer (‘Generationslagerung’) can form connections based on their common experiences – the generational context (‘Generationszusammenhang’). Based on these experiences, they might finally form unified groups or generational unities (‘Generationseinheiten’) who share common political goals, and it is in these unified groups that Mannheim sees the main catalyst for historical change (Mannheim 1964 [1928]).

Mannheim's account provides a range of valuable insights into the concept of generation and can be of very great help if one is concerned with a male- and youth-centred intellectual history of political change. For welfare generations – understood as a category of social structure – it lacks the description of certain characteristics. There are no clues as to who the individuals are that make up generations, to how many members a generation can have, to where the boundaries between different generations lie and how these generations set up and negotiate (financial) bonds and connections between each other. In order to research welfare generations, Mannheim's theory needs to be refined.

Norman Ryder's (1997 [1964]) cohort concept shares most of the characteristics of a socio-historical generation. Like Mannheim, he stresses the importance of a common experience: For Ryder, ‘[a] cohort may be defined as the aggregate of individuals (within some population definition) who experienced the same event within the same time interval’ (Ryder 1997 [1964]: 68). The experience of cohorts or generations can therefore change during their life course – it is not fixed at birth. A generation's experience is shaped by the common susceptibility to historical events or institutional changes at shared sections of the life cycle. In addition to a common experience, cohorts can also be described in quantitative terms. As Ryder puts it, ‘an extraordinary size deviation is likely to leave an imprint on the cohort as well as on the society’ (Ryder 1997 [1964]: 69). Since this argument is often discussed in the context of pension reform and problems resulting from demographic changes, its implications might help in operationalising the concept of welfare generations. Ryder also points to differentiations within the cohort: ‘Virtually every subsequent occurrence will depend on the societal plan for utilising characteristics present at birth: sex, race, kinship, birthplace and so forth’ (Ryder 1997 [1964]: 84).

In the context of the welfare state, it is therefore not enough to be born in the same year. The generational layer and the generational context are of importance: these are formed by the institutions of the welfare state – in this case, the pension system. In this context, changing policies and changing contributions and benefits need to be taken into account. Until the 1970s, the development of the pension system, as well as of most welfare payments, was characterised by growth: both qualitatively by larger and larger groups eligible for welfare services as well as quantitatively by rising payments. After the 1970s, welfare states mostly aimed at reducing costs as quickly as possible. All welfare states have undergone considerable reforms in the period up to now and most are still in a continuing process of shifting objectives (Thomson 1989, 1996; Leisering 2000a,b; Bude 2003, 2005). Demographic trends need to be considered as well: low fertility and high life expectancy lead to a shrinking working age population and a growing number of elderly. For PAYGO-systems this consequently means that fewer and fewer pension contributors have to support more and more pensioners.

This leads to the question how generations as groups can form collectives that are in conflict with each other. In order to illustrate the process of generation forming, Mannheim refers to the picture of Marx's process of class creation (Mannheim 1964 [1928]: 525–26). Marx distinguishes between class in itself (‘Klasse an sich’), that merely depicts a loose group of individuals exposed to the same economic status and class for itself (‘Klasse für sich’), that has identified a common status and common interests (Marx 1960[1852]: 194–207, 1972[1847]: 180–81). Ralf Dahrendorf (1959) later on incorporated this idea into his theory of classes in the industrial society. Here classes in themselves are quasi-groups that are merely theoretical constructions. In order to become fully formed interest groups, they need to have a structure, a form of organisation, a programme and a personnel of members (Dahrendorf 1959: 180). To achieve this, the group needs technical means (e.g., media of communication) as well as necessary political and social preconditions for the formation of a group. However, these preconditions are necessary, but not sufficient: even technical possibilities combined with democratic and social potential, as it characterises most Western societies today, do not necessarily lead a quasi-group to become an interest group that engages in open conflict with another.

The financial, institutional and demographic situation is first and foremost an external characteristic that hints to the existence of a ‘cohort in itself’, or a quasi-group in Dahrendorf's terms. In order to form a generational unity, especially one that is ready to engage in open conflict, the individuals within a cohort first need to identify their material situation and their relation to each other in order to become an interest group, ‘cohort for itself’ or generational unity. As long as this fact is only presupposed from the outside, the cohorts in question cannot be assumed to be fully formed generations, let alone interest groups with any conflict potential. This question will be resumed below, first a more detailed description of the institutional context of pension systems in Germany, the Netherlands and Great Britain is necessary.

In this article it is assumed that different pension systems influence experiences differently: specific national pension arrangements expose cohorts to different institutional circumstances and therefore have consequences for their specific generational layer and their generational context. This in turn might influence the formation of generational units that again may result in differing goals or aims of these units. In order to analyse this process, three different pension systems are compared. The pension systems of Germany, Great Britain and the Netherlands can stand as examples for the different welfare regimes as proposed by Gøsta Esping-Andersen (1990) (see Table 1). Although Esping-Andersen's theory has been seriously and rightly challenged in various ways, up to this point it is still regarded to be the ‘classic’ concept to categorise Western welfare states with specific regard to their policies and outcomes (for an overview of the critique see Castles 2001: 140–44; Schubert et al.2009: 4–10).

The German statutory pension system is a classical example of the conservative social insurance regime and mainly aims at preserving the status quo of the pensioners. The PAYGO-financed statutory pension has remained the most important pillar of the German pension system. For some employment sectors occupational pension plans have gained considerable importance. But since pension payments are based on contributions and the German system aimed at a life standard saving pension payment at least until 2001, the statutory system has overtaken the largest and most important function in ensuring old age provision. Since the reforms of 2001, the aim of life standard saving has been dismissed; the payments will get reduced in the years to come. The gap in provision is to be met by an increase in private pension plans to be subsidised by the government. Since private pensions are not mandatory, it is feared that especially low earners will not provide adequately for their old age and gaps between low and high earners will widen for future pensioner cohorts (Hauser 2001).

In Great Britain, the Basic State Pension (BSP) has been of comparatively lesser importance. Since it corresponds to the liberal welfare logic, its benefits are relatively meagre; introduced in 1946 it was hardly ever sufficient to meet the basic existential requirements (Thane 2006: 72). The flat-rate pension gets paid to those employed, but only if the required contribution periods are met. This is especially difficult for individuals with discontinuous working biographies, such as a considerable amount of women. Occupational and private systems have been of great relevance to British pensioners. There are several additional statutory systems that were initiated over the years in order to ensure the status quo of the living standard into the pension phase. Recent governments have tried to encourage private pension provision, but due to mis-selling scandals in the 1990s, these systems have been undermined by a considerable amount of mistrust amongst the population (Taylor-Gooby 2005: 185). In contrast to Germany, where the poverty rate of pensioners equals that of the rest of the population, the risk for British pensioners to be poor is higher (14.4%) then for the population in general (11.4%) (OECD 2005). Although the pension payments are low, at least the BSP is financially stable in the foreseeable future (Blake 2002: 317), which means that the situation is not deteriorating for future cohorts. Additionally, John Hills has analysed the pension contributions and payments for cohorts in the past decades based on computer simulations. He concludes that there are no remarkable differences in the pension accounts of different cohorts (Hills 1996, 2005).

The Dutch pension system (Algemene Ouderdomswet, or AOW) pays a universal flat-rate benefit to everybody over the age of 65 who has been living in the Netherlands in the 50 years prior to that birthday. This universal logic gets closest to a social democratic form of welfare regime. Since 1975, the pension rate for couples is the same amount as the minimum wage. It is therefore possible for nearly every pensioner to meet his or her basic existential requirements, which is mirrored in the very low pensioner's risk of poverty (1.6%), especially compared to that of the rest of the Dutch population (6%) (OECD 2005). To ensure the standard of living in retirement, occupational pensions are very important in the Dutch pension system; nearly 100% of the current Dutch population has an occupational pension plan (de Kruik 2006: 21). The provision of private pensions is rather restricted to the sectors of high wage earners (Caminada and Goudsward 2003: 77). The evolution of the spending capacity of pensioners in the Netherlands has been very positive in the recent decades, even when compared to the economically active sectors of the population (de Kruik 2006).

If one follows the logic of generational accounting theories, and the idea of ‘winner’ and ‘loser’ generations, the severity of differentiation between cohorts and the possible resulting conflict potential should be differing in the three countries. It should be highest in Germany, where the statutory pension system has taken the most central role in the system of old age provision. The redistribution between cohorts has been changed through the implementation of private provision in 2001; statutory pension benefits will be reduced in the years to come and are expected to be met by a growing amount of private pension plans. Since the German pension system aims at stabilising the status quo, intragenerational differences might be of some impact as well. In the Netherlands, there is no evidence of a great source for generational inequalities or conflicts: The treatment of Dutch pensioners has been rather equal, with even a slight betterment for younger pensioner cohorts. The equalising factor of the flat-rate pension benefit might be relativised by occupational pension payments. Great Britain shows the least potential for generational inequalities. The Basic State Pension has been of so little importance that it hardly worked as a decommodifying instrument of social security. This means that stratification is even intensified in old age; pensioners are being faced by risks of poverty even more often than the working population.

So far, the discussion of the phenomenon of welfare generations has mainly followed the general argument: the findings presented so far have focused on the distribution of financial resources across entire cohorts. The picture changes if one takes other resources into account, such as time. Working hours have decreased significantly per year, because of less weekly working time and increasing holidays. In Germany for example, the mean working time per year for employees was 1966 hours in 1970, and only 1436 hours in 2006 (DRV 2007). For Great Britain the number decreased from 1939 hours per year in 1970 to 1672 in 2006, for the Netherlands it decreased from 1540 in 1987 to 1391 in 2006 (OECD 2008).³ Overall, the working time invested in pension payments has decreased over the years – this puts younger cohorts into a more favourable position. At the same time, the longer life expectancy leads to a prolonged retirement period. John Hills concludes that, even if pension payments are reduced, ‘the average member may survive long enough to be in a better life-time position than the average member of an earlier cohort’ (Hills 1996: 68).⁴

But whether it is justified to talk of entire cohorts as collectives that share advantages and disadvantages with respect to their position as clients of the welfare state may be questionable. The public discourse hardly shows any attention to possible intragenerational differences. However, there are several empirical clues that a welfare generation is not a homogeneous group but rather made up of several subgroups that might be formed by different experiences – a fact that Norman Ryder has pointed to repeatedly. Therefore it would be difficult to assign the same advantages or disadvantages, let alone a common interest in the distribution of welfare payments across entire cohorts. The risk of unemployment, for example, is not the same for every member of a cohort. The lower the education, the higher is the risk to be unemployed, and, what is worse, the higher is the risk to be unemployed for a long period of time. That means that workers whose job situations are already characterised by low wages and therefore low pension payments are at risk of falling into even more precarious circumstances because of unemployment. This holds especially true for systems where pension benefits are very strictly connected to the contributions paid. Differentiation by job or education is merely one indicator for the heterogeneity of pension cohorts. There is also a considerable gender divide within the pension system, especially in systems that connect benefits to contributions (Stegmann 2003; Penneck and Lewis 2005; DWP 2007; Frericks 2007): women often find themselves in much poorer conditions in comparison to men. Due to the fact that they often do not work continuously in full-time jobs and are on average less well paid than men, the only reason why most women are not living in poverty in old age is because they live off their husbands’ pension.

This is just preliminary evidence pointing to the problem of intragenerational differentiation. The seemingly clear-cut picture of winners and losers in the welfare state gets more complicated the more details are taken into account. The concise and simple utilisation of the concept ‘generation’ is lost. The consequence is a rather complex picture of different strata of inequalities. With respect to winners and losers in the welfare state, it is very difficult even to try to identify boundaries between birth cohorts that would lead to clear categories of individuals that have won or lost with their status in the pension system. The collectives shift once different indicators are taken into account, the groups appear too heterogeneous to form collectives in their own right, and once non-financial indicators are taken into account, the picture gets even more complex. There are strong clues for the possibility that intragenerational differences have a much stronger influence on the formation of interest groups in the welfare state than a common date of birth. This will affect the chances for a cohort ‘in itself’ to become aware of a common interest base and become an interest group, a generation or cohort ‘for itself’.

After having described some general trends concerning the generational layer and the generational context of the birth cohorts in hand, it remains to be analysed if there is any possible latent conflict potential that leads to the development of generational interest groups or ‘cohorts for themselves’. Studies into the attitudes towards social policies show repeatedly that in all Western welfare states there is overwhelming support for the welfare state in general and for pension systems in particular (e.g., European Commission 2004; Kohli 2006; Blome et al. 2008). There has been no clear evidence, however, as to what exactly influences preferences in this context. Apart from age or cohort, also gender, education, employment status and employment sector could be influential. However, differing databases as well as differing indicators and a varying courage in explaining the evidence have led to a rather unclear picture of how these factors influence attitudes (for an overview see Ullrich 2000). Personal values might also be considered: for the formation of individual preferences a mix of self interest (as suggested by rational-choice theories) as well as general values (such as solidarity, reciprocity of justice) seems to be of importance (Blekesaune and Quadagno 2003; Mau and Veghte 2007; Rehm 2007).⁵ Iversen and Soskice (2001) have pointed to the fact that people demand insurances against the risks and shocks of the occupational structure, especially if they have made risky investments in order to acquire these skills. In this logic, there should be a general support for pension systems by all age groups, as the acquired skills might be fading with getting older.

For the study in hand, it is of special importance if age group or cohort has any influence on the attitudes towards pensions. For Germany and Great Britain⁶ Busemeyer et al. (2009) have found relatively little evidence for strong age-cleavages in the attitudes towards pension benefits. This puzzled the authors somewhat as they, relying on a rational-choice model, had expected stronger effects (Busemeyer et al.2009: 208). In a study comparing the baby-boomer generation to older cohorts in Germany and Britain, Achim Goerres also has found surprisingly little evidence of cleavage towards social policies (Goerres 2009a). When it comes to attitudes towards the pension system, there is obviously no striking disunion between age groups. In order to find out if this also holds true for cohorts or generations, a more detailed analysis needs to be undertaken. Furthermore, for the question of a possible conflict potential between generations, it might be less important what individuals think about pensions as an abstract concept. Much more important is the evaluation of their pension time to come.

A secondary analysis of opinion data shall give further insights into these questions. The data basis is the Eurobarometer survey, which has been carried out twice per year since 1973 in all member countries of the European Union. In every country a representative sample of 1000 interviewees is asked about topics relevant to the European Union. The standard questionnaire is adjoined by the so-called ‘Special Eurobarometer’, which is dedicated to varying topics (European Commission 2002). Of special interest to the study in hand are those Special Eurobarometers that deal with opinions about old age insurance. Three of those were selected: the Eurobarometer 08 of autumn 1977, the Eurobarometer 37.1 of spring 1992 and the Eurobarometer 56.1 of autumn 2001.⁷ These studies cover a time period of 34 years; they can therefore form a basis for the study of long-term trends. The study focuses on West Germany, the Netherlands and Great Britain.⁸

Any cross-sectional cohort analysis faces the problem of distinguishing between three different effects: period of the study, years of age, and date of birth (Converse 1976; Glenn 1976). The three of them are always interwoven: ‘Each is completely defined in terms of the other two’ (Converse 1976: 20). A group of individuals born in the same year will experience a special historical event in a special section of the life cycle. However, historical events shape cohorts differently as they are in different age groups. For example, a pension reform might have a completely different effect on 20-year-olds than on 50-year-olds. For the former, the reform might be less relevant, as the pension phase is still a long time ahead. Or it might on the contrary be perceived as very risky, since the long time span makes further developments highly contingent. The latter have already paid contributions into the PAYGO-system; the pension phase is much more foreseeable and therefore less risky. Furthermore, period effects such as economic developments and prevailing discourses about the security of pension systems might affect the attitudes of 20-year-olds at one point in time differently than 20-year-olds a few decades earlier or later.

Both Mannheim (1964 [1928]: 536–37) and Ryder (1997 [1964]: 81) point to the fact that younger age-groups are much more prone to formative influences on behaviour and attitudes than older ones. Although this might be contested when viewed from the perspective of theories of the life-cycle and life-long-learning, we might still expect a stronger imprint of events on younger individuals than on older ones.⁹ However, the effects of growing older may also differ between cohorts (Glenn 1976: 901). As Glenn (1976: 902) resumes: ‘[…] if the effects are not assumed to be additive, it must be assumed that age effects vary systematically among the periods and cohorts, that period effects vary systematically among age levels and cohorts, and that cohort effects vary systematically among age levels and periods’. Furthermore, the composition of cohorts changes over time: Mortality is affected by gender, income and socio-economic status (Cutler and Meara 2001; Wanless et al. 2011). Therefore, as cohorts age, they will become richer, better educated and more female. This will probably not change the picture decisively, but these drifts need to be taken into account. Considering methodological problems, it should also be added that the Eurobarometer study is a survey, not a panel study. This means the three different studies do not actually ask the same individuals at the three points in time. However, as these surveys are representative, they can still give evidence of the development in attitudes of the countries’ population.

The main question concerns the formation of generational units with clear and distinctive opinions about their old age retirement. In all three selected Eurobarometer surveys there is a question about the evaluation of the individual retirement period. This question is posed to all interviewees that have not yet retired. If the worsening economic and demographic situation has resulted in the formation of generational interest groups or ‘cohorts for themselves’, there should be clear evidence for it, despite the intermingling effects of period, age and cohort. Younger cohorts should be considerably more worried than older ones. Furthermore, the younger the cohorts are, the more worried they should get over their life course. The findings are presented in Table 2.

The first striking result is that there is no cohort in which a majority of members is worried about their pension time to come. Apart from a few exceptions, the amount of members worrying lies between 10 and 20%. However, some, albeit weak, trends can be shown with regards to the development over the life cycle of cohorts: Whereas the three cohorts born between 1920 and 1949 tend to get less anxious over time, the cohort born between 1960 and 1969 tends to get a little more worried, this also holding true for the cohort 1970–79 in Germany and Great Britain. However, these cohorts are only between 20 and 40 years old in 2001, therefore the pension phase is still a long time ahead. The youngest cohort born between 1980 and 1989 seems to be least worried in all three countries. This is probably due to a life cycle influence: the very young age (15–21 years) has not given much cause to an intensive reflection on the pension time to come.

The description of the institutional background above had already given cause for the assumption that cleavages between cohorts should be largest in Germany. In the Netherlands, the variation is much more subtle, and overall no cohort is especially anxious about their pension phase. These findings parallel the description of the historical background of the Dutch pension system – the very stable evolution of the AOW, the broad coverage of the population by occupational pensions and the quite low risk of old-age poverty do not give much reason for concerns. Intergenerational differences are hard to detect. The British old age insurance system on the other hand obviously gives much more reason for negative evaluation, but here, interestingly, cohorts seem to get less worried over time. Still, in all the countries considered the actual differences between cohorts are not striking enough to expect the formation of interest groups. Overall, the results confirm the influence of the institutional backgrounds of the three pension systems: the British pension system gives its members the strongest reason to feel insecure, the Dutch AOW the least, Germany is somewhere in the middle.

Table 3 compares five life-cycle sections over the three points in time. Together with Table 2, it shows two sides of the same coin: Instead of following different cohorts (or representatives of an unchanging group) through the three points in time, Table 3 shows the opinion of age groups (or varying representatives for ‘the young’ or ‘the old’) at different historical dates.

If the economic and demographic development did indeed have an effect on the attitude formation of different cohorts, one would expect a considerable deterioration of attitudes towards the individual retirement phase by younger age groups between 1977 and 2001. However, effects of younger age groups getting more worried can – to a small extent – only be identified in Germany; in the Netherlands and Great Britain the levels of anxiety stay about the same. Again, no age group shows a majority of members being worried about the retirement phase. Furthermore, the rather related levels of anxiety do not point to cleavages between generations being in similar cycles of the life course at different points in time. Considering the dominant media discourses since the 1990s, one would at least have expected the younger cohorts to be more anxious, if not in 1992, then at least in 2001. This is especially true for Germany, where the 2001 reforms resulted in dismissing the life standard saving component. Considering these changes, the reactions of the younger age groups in Germany are relatively subdued. With the results in hand it is hardly possible to mark them as interest groups or ‘cohorts for themselves’.

Welfare Generations have been very visible in both scientific and media debates, they seem to be posing new cleavages between social categories that are defined by year of birth. However, as soon as one tries to describe generations as social collectives, one gets confronted with a set of rather severe difficulties that hinder a clear and concise operationalisation of the concept. First of all, advantages and disadvantages between different cohorts shift once different indicators are taken into account: the successively shrinking working time invested in pension payments challenges the idea of younger cohorts being disadvantaged in comparison to older cohorts. Although different generational layers can be detected in the three countries under scrutiny, it is doubtful as to whether these result in specific generational contexts that are so influential that they become more important for the individual status in the welfare state than other contexts, especially that of income or class. There are certainly no hints to possible generational unities in the welfare state. This means that actual or even perceived advantages or disadvantages do not transfer into experiences or opinions of specific cohorts.

Being asked about the evaluation of their pension time to come, a surprisingly small amount of people seems worried about retirement, and there is relatively little variation between the cohorts. Despite a strong communication in the media and both the social and political preconditions of group forming fulfilled in the three countries, the younger cohorts have so far not engaged in group formations as cohorts or generations ‘for themselves’. These findings parallel those of other studies. As mentioned earlier, Busemeyer et al. (2009) and Goerres (2007, 2009a) have found little evidence for political cleavages between age groups and cohorts – this also contradicts the idea of a generational conflict soon to be experienced. This is apparently not due to the fact that the people in question do not follow the media debates: Carsten G. Ullrich (2008) found in a study that focused on Germany that many people fear a conflict between generations but hardly show any disposition or opinions that would mark them as members of conflicting parties. This could be due to the fact that the potential for the formation of interest groups still lies in different social categories – income, gender, occupation, social status could prove more unifying then a common date of birth. Clara Sabbagh and Pieter Vanhuysse (2010) have analysed intergenerational justice perceptions among undergraduate university students in 1996–98. This special group of respondents indeed shows a high perception of intergenerational injustice. However, this might also due to the fact that highly educated individuals are especially influenced by media reports about vague generational conflicts. We do not know if the students perceived themselves as taking part in an actual dispute. A Third World War between generations ‘for themselves’ is therefore hardly to be expected. And even as an empirical category and socio-historical concept of collective experience, generation has so far failed to prove useful.

Christina May, sociologist, is lecturer and research associate at Georg-August-Universität Göttingen. Her research interests include social policy, social theory, historical sociology, and methods of comparative research. She is author of Generation als Argument. Konflikte um die Rentenversicherung in Deutschland, Großbritannien und den Niederlanden (Frankfurt 2010).