Too old to work,or too young to retire? The pervasiveness of age norms in Western Europe

Abstract

The ageism debate has pointed to the persistence of negative age stereotypes that hinder the prolongation of working lives. However, the actual holders of discriminatory norms have remained largely anonymous because there is limited understanding of the pervasiveness of age norms. This article discusses arguments derived from life course and social norms theory regarding the degree of internalization of age-related norms. The focus is on individual differences in terms of social class and gender. Using 2006 data from the European Social Survey, the article gathers empirical evidence on attitudes towards the timing of retirement in 14 Western European societies. A set of tobit models examines the determinants of retirement age norms for men and women. The results suggest that social class has a strong impact on retirement age norms. Moreover, the analysis reveals a complex pattern of gendered norms concerning the timing of retirement.

Keywords

ageism class Europe gender life course retirement age social norms work orientations

Introduction

Once popular early-exit arrangements have become discredited among analysts and decision-makers in the face of population ageing. As a result, there is today widespread consensus that in order to contain public spending, employment among older workers needs to be stimulated (OECD, 2006). Many European welfare states have thus strengthened the bonus-malus system in their public pension schemes by means of actuarially defined adjustment factors (European Commission, 2010: 18). Others have decided to increase pension eligibility ages – either by gradually equalizing pensionable ages between women and men, or even (in countries such as Germany, Denmark and the UK) by introducing reforms that over the long term will raise the statutory pension age to 67 or 68 years, abandoning the standard age of 65 years that has endured for more than a century (European Commission, 2010: 17).

But whereas policy changes aim to replace an alleged ‘early-exit culture’ with a model of ‘active ageing’ (Ebbinghaus, 2006), negative age stereotypes have persisted among both employers and employees (Loretto et al., 2000). Not only are older workers often regarded as less productive than their younger peers, but the majority of Europeans also seem to prefer early retirement for themselves (Esser, 2005). Such entrenched attitudes are likely to hinder the prolongation of working lives. Without public support, the efficacy of pension reforms aimed at deferring retirement ages seems questionable (Litwin et al., 2009). Yet, the actual holders of discriminatory norms have remained curiously anonymous in the debate over ageism because there is limited understanding of the degree of pervasiveness of social norms of retirement.

Consequently, this article discusses arguments derived from life course and social norms theory regarding the degree of internalization of age-related norms. Drawing on recent micro data from the European Social Survey, it gathers empirical evidence on attitudes towards the timing of retirement in 14 Western European societies. The data analysis focuses on individual-level differences. In particular, it examines the extent to which retirement age norms are gender-graded and thus potentially hamper the increase in older women’s labour market participation. Moreover, the article tests the claims of the individualization theorem, according to which lifestyle preferences today are unaffected by social class membership.

Age norms, gendered ageism and retirement

Age-related social norms are a foundational theme within life course research. The pioneering study by Neugarten et al. (1965) was the first to draw attention to ‘systems of norms which refer to age-appropriate behaviour’ (Neugarten et al., 1965: 710). Age norms are seen as a basis for self-assessment since people compare themselves to others in search of guidelines for the timing of life transitions. This idea soon became conceptually embedded in a broader argument concerning the institutionalization of the life course and the temporal order of biographical events (Scherger, 2007; Settersten and Mayer, 1997). The life course literature showed that the institutional framework of modern societies establishes normative age thresholds for the central life transitions, most notably for entry into and exit from employment (Kohli, 2007). More recent findings corroborate the existence of age norms of retirement (Esser, 2005; Han and Moen, 1999; Moen et al., 2005; Settersten, 2003).

An often neglected aspect that adds to the salience of retirement age norms is the unavoidable uncertainty related to the concomitant features of post-working life. Despite being reversible in principle, for the great majority of people, retirement de facto puts a definitive end to working life (particularly in continental Europe, where returning to work after having retired is much less common than in Anglo-Saxon countries). Unlike employment decisions in mid-adulthood, retirement is a genuine biographical milestone loaded with social significance and is often infused with great expectations (Vickerstaff and Cox, 2005). However, for the individual actor it is virtually impossible to anticipate the whole range of consequences that this transition will have on his or her living circumstances. Uncertainty complicates cost-benefit calculations, making actors sensitive to significant signals from their social environment. Age norms and standardized transition patterns can thus serve as blueprints for decision-making. Because age norms probably have an influence on actual behaviour (Billari and Liefbroer, 2007), furthering understanding of the ways in which age-graded norms are sustained and reproduced can inform policies aimed at increasing the employment rate of older workers (Litwin et al., 2009).

With respect to social heterogeneity in age norms, ironically, among the individual characteristics that have received most attention is age itself (Neugarten et al., 1965). However, research has struggled to disentangle age from cohort effects. Moreover, existing studies emphasize gender differences in age norms. Specifically, age norms typically define an earlier retirement age as appropriate for women than for men (Esser, 2005; Settersten, 2003). Duncan and Loretto (2004) speak of ‘gendered ageism’ in reference to their finding that older women experience more age discrimination in employment than older men. Women’s early retirement has also been related to the desire to retire at the same time as their spouses (Moen et al., 2005). Joint retirement implies a younger retirement age for wives than for husbands due to the characteristic age difference between spouses.

Comparative retirement research has focused on the relationship between the welfare state and work exit dynamics (Blöndal and Scarpetta, 1998; Blossfeld et al., 2006; Ebbinghaus, 2006; Gruber and Wise, 2004). By all accounts, the idea that age norms differ between national societies is quite uncontroversial (Kohli et al., 1991; OECD, 2006). In this vein, the term ‘age cultures’ has been coined to describe ‘social norms, values, ideals or perceptions in society that structure the ideas of the age-work relationship’ (De Vroom, 2004: 8).

While the driving forces of norm evolution at the societal level are beyond the scope of this article, there are important issues attached to the question of how institutions shape the understandings of age-appropriate retirement behaviour. The regulatory effect of public policies is of special interest here (Kohli, 2007). For instance, Han and Moen (1999: 193) posit that the convergence of retirement ages at 65 during the 1950s and 1960s was ‘the result of a multitude of [economic driving] factors coalescing and coinciding around a specific age, which in turn became defined as normative post hoc’. Similarly, the subsequent emergence of an ‘early-exit culture’ in the 1970s is frequently attributed to the incentives established by early retirement policies (De Vroom, 2004; Mortimer et al., 2005).

Work orientations and retirement

Gruber and Wise (2004: 13) note that ‘since age sixty-five is the normal retirement age, many employees may think that sixty-five is the age to retire’. However, in the standard economics approach to retirement, preferences are assumed to be homogeneous (Blöndal and Scarpetta, 1998; Gruber and Wise, 2004). The guiding idea is that homo economicus only works to earn an income, the final aim always being consumption. On the basis of the assumption that ‘if they aren’t paid, people don’t work’ (Gruber and Wise, 2004: 1), the retirement decision is modelled as a strictly pecuniary calculation. By this measure, one would expect to observe a universal affinity to early exit from work, since late retirement only occurs as a consequence of budget restrictions.

In contrast to this account, sociological approaches usually assume that there is an intrinsic value to work. Accordingly, work is not merely pursued as a source of income, but also as a means of self-realization, social recognition and social contacts (Doherty, 2009; Riach and Loretto, 2009). The intrinsic value of work should be reflected in a resistance to leaving the job prematurely. In fact, this hypothesis is supported by the frequent occurrence of involuntary retirement (Szinovacz and Davey, 2005; Vickerstaff and Cox, 2005). Interestingly, financial considerations are rarely mentioned among people’s motives for wanting to remain employed. Rather, typical reasons are either a strong work attachment or a ‘fear of giving up work and losing the social contacts and routine associated with it’ (Vickerstaff and Cox, 2005: 89).

However, the notion that work has an intrinsic value has been questioned by the ‘end of work prophets’ (Doherty, 2009), such as Beck (1983), who posit the demise of the conventional work ethic that characterized the industrial age. Accordingly, the increase in non-standard work meant that employment has largely lost its capacity to convey identity and shape lifestyle preferences in post-industrial society. Although this argument has been challenged on both theoretical and empirical grounds (Doherty, 2009), it still figures prominently in the discourse on employment and work orientations (Crompton, 2010).

Theoretical issues in research into social norms

Age norms are age-related social norms. Therefore, advances in research on age norms hinge on the same complex issues that have been at stake in the long-standing theoretical literature on social norms. Social norms are usually conceived as the formal and informal rules that prescribe or interdict particular actions and that are acknowledged by the members of a social group (Boudon, 2003). However, there exist fundamentally divergent theoretical claims about the ways in which social norms operate within society (Etzioni, 2000).

It is helpful to distinguish two broad schools of thought which rest on opposed presumptions regarding the enforcement mechanism of social norms. The rational choice tradition postulates that social norms regulate individual choices by imposing (positive or negative) sanctions on particular forms of behaviour (Coleman, 1990; Hechter and Opp, 2001). In this view, social norms disincentivize socially undesirable courses of action by establishing additional costs, e.g. through the loss of social prestige for the norm violator. Hence, social norms can enhance the aggregate well-being of the group. From the perspective of the individual actor, however, social norms represent constraints on self-interested action (Anderson, 2000).

By contrast, the ‘socioeconomics’ paradigm points to internalization as an alternative norm enforcement mechanism (Bicchieri, 2006; Elster, 1989; Etzioni, 2000). Rather than mere external constraints on individual preferences, social norms from this perspective are capable of shaping actors’ ‘intrinsic predispositions’ (Etzioni, 2000). In other words, norms are adopted by individual actors as goals in their own right. Notably, if social norms are internalized, external sanctions are not a necessary condition for norm conformity (Bicchieri, 2006: 24). Instead, conformity is motivated by ‘the recognized legitimacy of mutual normative expectations’ (Bicchieri, 2006: 25). Rather than induced by fear of punishment, the potential loss in self-respect can be sufficient to bring about norm compliance (Elster, 1989: 100).

The question of whether constraints or internalization act as norm enforcement mechanisms has far reaching implications, for example for the relationship between gender and age norms. According to the rational choice model, norms are an anonymous property of social context. As a consequence, women’s early retirement may be attributed to gender discrimination despite women preferring later retirement. By contrast, from the perspective of the internalization model, the picture becomes more complex, as women themselves might endorse gender-graded age norms.

Social class and individualization

Qualitative studies on ageism have shown a substantial influence of organizational characteristics on attitudes towards older workers (Loretto et al., 2000). However, social class has received little attention in research on age norms thus far. In one rare exception, Esser (2005) could not find any significant class differences in terms of attitudes towards ideal retirement age. In the ageism literature, the absence of a class perspective is partly due to the initial concern to distance the concept from other forms of discrimination. With age today widely acknowledged as a genuine category of social inequality, increasing attention is paid to the intersection of class and age. Duncan and Loretto (2004) find pronounced differences with respect to the perceived age-related productivity decline of blue-collar and white-collar workers respectively. Moore (2009) similarly concludes that age and gender discrimination are closely interwoven with class inequalities.

Although neo-Weberian class theory abandoned the notion of an immanent class consciousness link, occupational class is often regarded as key to individual attitudes, especially to political orientations and work-related issues (Crompton, 2010; Scott, 2002; Svallfors, 2006). There are various theoretical mechanisms that plausibly link social class and norms. As Grusky and Sørensen (1998) point out, preference-driven self-selection leads individuals with similar predispositions to enrol in similar occupational tracks in the first place. In addition, lengthy formal training and social closure at the workplace foster occupational socialization.

In the case of attitudes towards retirement, occupational pension plans and typical career trajectories can also be expected to play a major role in rendering age norms class-specific (cf. Radl, 2012). Moreover, a high incidence of disabilities, which for hazardous occupations (e.g. miners or firemen) normally corresponds with lower statutory pension ages, probably leads to greater normative approval of early exit from work. In detail, one expects the degree of specificity of human assets to be positively correlated with the preferred age for exit from work. Employers and the self-employed generally should have a high level of affinity towards late withdrawal from the labour market because they enjoy a maximal degree of work autonomy. Further, as a consequence of differences in work strain one would expect manual workers to adhere to lower retirement age norms than white-collar service employees.

Like the notion of gendered age norms, the hypothesis of class-graded age norms brings forward the possibility of differences between target groups. The fact that social consensus is a precondition for the existence of social norms does not necessarily imply uniformity across all social strata. Instead, social norms can vary according to the target actor’s situation (gender, job, family situation, etc.). In Bicchieri’s terminology, age norms are understood as local ‘insofar as their content and recommendations are context-dependent’ (Bicchieri, 2006: 80).

Based on the observation of an ongoing pluralization of lifestyles, the individualization approach questions the usefulness of social class as a sociological category (Beck, 1983) and especially the impact of class on attitudes (Atkinson, 2007; Higgs and Gilleard, 2006; Scott, 2002). It has been argued that social class has allegedly lost its power to shape identities and lifestyles in later life due to a decoupling of working and post-working life: ‘Variability in the material and social outcomes of individual lives can no longer be seen as resulting from stable, fixed social positions’ (Higgs and Gilleard, 2006: 231). The retirement decision, which certainly marks a profound change in lifestyle, should thus be unaffected by class membership.

Data and methodology

Data

In its third round, the European Social Survey (ESS) included a module on ‘The Timing of Life’, which contained questions on retirement. Interviews were conducted between August 2006 and December 2007. The sample was restricted to respondents in the 14 participating Western European countries: Austria, Belgium, Denmark, Germany, Great Britain, Finland, France, Ireland, the Netherlands, Norway, Portugal, Spain, Sweden and Switzerland. Eastern Europe was not considered because the lack of longitudinal information was especially problematic due to the profound labour market adjustments experienced in the 1990s. Only respondents over 50 years of age were selected because previous research has shown that attitudes towards retirement are not well-defined among young and middle-aged persons (Moen et al., 2005). The final sample comprised respondents born between 1930 and 1957. A gender-sensitive split-ballot procedure was used in the ESS: one randomly selected half of the respondents were asked about their attitudes concerning men’s retirement and the other half on women’s retirement.

Three survey questions are analysed, the first of which reads: ‘In your opinion, what is the ideal age for a man/woman to retire permanently?’ In case of doubt, retirement is specified by the interviewer as giving up paid work. In the words of Settersten and Mayer (1997), this item assesses an optimal age norm as it comprises notions about the ‘best’ or ‘preferable’ age to experience a particular life event. Critically, the authors suggest that optimal age norms are not necessarily closely linked to actual behaviour, and that prescriptive and proscriptive age norms may be more relevant, as they entail ‘shared expectations about when certain transitions “should” or “should not” occur’ (Settersten and Mayer, 1997: 242). Empirical research should thus strive to identify the age range that social norms of ageing define as appropriate for the transition from work to retirement. The proscriptive norm that defines the age before which one should not retire is captured by the following question: ‘Before what age would you say a man/woman is generally too young to retire permanently?’ The upper threshold is addressed correspondingly: ‘After what age would you say a man/woman is generally too old to be working at least 20 hours a week?’ The multivariate analyses concentrate on this latter ‘age deadline’ because it taps into the crucial issue of postponing retirement.

This pair of survey questions seems adequate for measuring internalized age norms as defined above: it explicitly asks for the respondent’s own subjective view (as opposed to that of the generalized other) while also making it clear that the object of the question consists in men’s or women’s retirement in general (as opposed to the respondent’s personal situation). It is particularly convenient that the survey methodology differentiates between respondents’ attitudes towards men and respondents’ attitudes towards women. However, it is unfortunate that no distinction is made between target groups in terms of occupation or class.

Statistical model

The data analysis uses the two-limit tobit censored regression model. Censoring arises from non-trivial non-response, notably by persons stating that one is ‘never too old to work’. The assumption behind the tobit model is that there exists a true, unobserved value corresponding to the censored cases. In this case, responses are treated as point data in the interval age 40 to 80. All ‘never’ answers are treated as either censored from below (‘should never work’) or from above (‘never too old to work’). At the same time, this method implies a straightforward handling of the small number of outliers. As the overall share of censored cases is still relatively low (<20%), the interpretation of the regression results is undertaken by reference to the underlying variable. Standard errors are corrected according to the clustered sample design and weights are applied to correct for unit non-response. Missing values have been imputed using multiple imputation (Royston, 2004).

Independent variables

Social class is measured according to the European Socio-economic Classification (ESeC) (Rose and Harrison, 2007). A further development of the Erikson-Goldthorpe schema (EGP), the main sorting criterion in the ESeC is the type of employment relationship, but unlike the EGP it has been purposefully designed for cross-national studies within the European context. The coding of social class is based on the characteristics of the main job for employed respondents; otherwise, the basis is the last job.

Education is usually one of the strongest predictors of attitudes of any kind. In terms of retirement age, it is conceivable that particularly time spent in higher education modifies people’s worldviews. Education might also affect retirement norms through its relationship with employment position, but this dimension should be captured more adequately by class membership. Independent from occupation, there might be timing effects related with later entry into paid employment. Therefore, the measure years of full-time education rather than level of educational attainment is used for operationalization.

Employment status is a self-reported measure. From a revealed preferences perspective, one should expect unemployed persons as well as homemakers to have lower work attachment than employed persons. For retirees, in turn, reported attitudes probably depend on the experience of retirement.

Other covariates. Age is implemented in both linear and quadratic form in order to account for possible non-linearities. The model controls for the father’s level of education and for the mother’s employment status when the respondent was 14 years old as proxies for family effects. Health status is captured through a dichotomous variable indicating whether or not self-perceived health is less than good. The model also controls for whether the respondent has experienced any periods of unemployment that lasted 12 months or longer. Household income is measured in terms of the natural logarithm using an equivalence scale of the square root of the household size. Further controls are the number of grandchildren, trade union membership and family situation, measured as a set of dummies that simultaneously captures marital status and partner’s employment status (if present). Civil partnerships are treated as equivalent to marriage. The employment status of the partner distinguishes between retired, active (working or unemployed and actively looking for job) and inactive (unemployed and not looking, homemaker or other).

Results

Descriptive findings

The data reveals widespread approval of early exit from work, albeit within certain limits. Whereas the mean ideal age for retirement is 61 years and four months for men, it is two and a half years lower for women. As for the lower retirement age boundary, the mean response is 52.8 years among men but only 50 years among women. Women on average are held to be too old to work already at age 60, while the mean response for men is 64.5 years. In short, retirement age norms in Western Europe prescribe earlier retirement for women than for men. It is also noteworthy that after passing the ideal age to retire, women are considered too old to work after only one more year. Given the fact that old-age pensions rarely allow for retirement before age 60, it appears that for women there is a particularly narrow age corridor in which retirement is both practically feasible and socially accepted.

The modal response on the latest age for retirement is 65 years. More than 40 per cent of respondents do not consider a man over 65 too old to work, whereas only one out of four do not consider a woman over 65 too old to work. Notwithstanding variation in the ages cited by respondents, then, the hypothesis that there are retirement age norms, and that there is a firm core to such norms, is supported by the fact that four out of five agree that someone is too old to work when he or she is over 70 years old.

Despite the openness of the survey questions, there is a significant clustering of answers at quinquennial ages. This pattern underlines the fundamental insight that social ageing is a non-linear process. Moreover, on all three items the variability of answers is larger when the focus is on women’s retirement and non-response/indifference is higher too. This indicates a lower degree of normative standardization of the female life course.

Taking a comparative perspective, Figure 1 looks at the possible influence of country-specific institutions and specifically at the age thresholds of the public pension system (Queisser and Whitehouse, 2006). It shows considerable variation in age cultures across Western European societies. Moreover, there seems to be an association between a country’s ‘normal’ and early pension age, on the one hand, and the average age at which men and women are considered too old to work on the other. Age cultures are most favourable of late retirement in Norway, where the statutory pension age is 66 years, whereas early retirement is popular in France, where a full pension could be drawn at 60 years of age. There are two interpretations to the correlation between pensionable ages and age norms. First, it might reflect the impact of economic incentives inherent in pension legislation. Accordingly, ‘age norms may arise in response to institutional regulatory schemes and to the rewards and punishments that are linked to compliance’ (Mortimer et al., 2005: 177). Second, there may be a symbolic significance to institutionalized age boundaries (Kohli, 2007), which leads them to exert ‘independent normative effects on the retirement decision’ (Esser, 2005: 30).

Figure 1.

Age norms of retirement and statutory pension ages by country and gender

Modelling strategy

Models are estimated separately for age norms targeting men and women. The first model includes the main variables of interest – sex of the respondent and social class – alongside most control variables including country dummies. In an attempt to cope with the potential endogeneity of employment status, the corresponding set of dummies is entered in a second step. The third specification aims to disentangle the direct class effect from characteristics such as education or unemployment experience. A second set of models on the subsample of employed respondents additionally includes covariates related to work satisfaction and the quality of the job.

Estimation results

Table 1 displays the results from the multivariate two-limit tobit regression model concerning the age-appropriate retirement behaviour of men. Judging by the goodness of fit, the predictive power of the regression models is high.

Table 1.

Tobit regression results: retirement age deadline for men

Age too old to work	Respondents born 1930–57			Only employed
	1	2	3	4	5	6
Female	.826***	.724***	.624**	.832**	.753**	.774**
Age	.439	.607	.585	−1.446	−1.406	−1.373
Age squared	−.003	−.004	−.004	.013	.013*	.013
Social class (ESeC) (Reference: routine workers)
Higher salariat	1.673***	1.557**	1.038**	2.185**	1.722**	1.734**
Lower salariat	1.652***	1.655***	1.268***	1.800***	1.538**	1.560**
Intermediate occupations	.462	.442	.188	.242	.189	.217
Petite bourgeoisie	.829*	.674	.327	.101	−.235	−.216
Agricultural self-employed	1.810*	1.654	1.229	1.130	.868	.897
Higher-grade blue-collar	−.233	−.151	−.206	−1.100	−1.083	−1.057
Lower sales and service	−.156	−.107	−.227	.627	.503	.479
Skilled manual	−.439	−.322	−.327	−2.227**	−2.230**	−2.178**
Employment status (Reference: employed)
Retired	–	−1.519***	−1.277***	–	–	–
Unemployed	–	−1.512**	−.861	–	–	–
Homemaker	–	−.200	−.208	–	–	–
Years of education	–	–	.031	–	.025	.027
Ever unemployed >12 months	–	–	−.682**	–	−.422	−.435
Log HH-income	–	–	.360	–	.306	.310
Trade union member	–	–	−1.071***	–	−1.360	−1.385
Health less than good	–	–	−.792	–	−.636	−.629
Job satisfaction	–	–	–	–	–	−.029
Job interesting	–	–	–	–	–	.084
Job stressful	–	–	–	–	–	−.443
Job stressful squared	–	–	–	–	–	.064
Family situation (Reference: married. Spouse employed)
Married. Spouse retired	−.872***	−.549**	−.544**	−1.593***	−1.632***	−1.679***
Married. Spouse inactive	−.503*	−.300	−.241	−1.063*	−1.016	−.999
Divorced	.074	.337	.635	.226	.382	.351
Never married	−.353	−.058	.247	2.071***	2.298***	2.270**
Widowed	−1.138	−.919	−.736	−.982	−.970	−.923
No. of grandchildren	−.142	−.138	−.116	−.151	−.109	−.114
Mother working (age 16)	.294	.298	.358	.947*	.999*	1.006*
Father’s education (Reference: less than primary)
Primary	.209	.197	.073	−.107	−.222	−.224
Lower secondary	.918	.858	.735	−.648	−.765	−.772
Upper secondary	.819	.772	.566	−.399	−.563	−.592
Tertiary	.960	.883	.674	−.003	−.173	−.148
Country Dummies	YES	YES	YES	YES	YES	YES
Constant	49.542***	43.689***	42.597***	105.236***	102.809***	102.279***
McKelvey and Zavoina’s R²	.506	.515	.525	.551	.558	.558
N	5213	5213	5213	2069	2069	2069

Sources: European Social Survey 3 2006/07; own calculations.

p< .1 **p< .05 ***p< .01

The estimates show that, after controlling for other characteristics, women situate the moment at which a man becomes too old to work about nine months later (on the latent dependent variable) than do men themselves. Female respondents hold retirement age norms targeted at men that are significantly higher than those of male respondents.

In terms of the effect of class, members of the service class (i.e. the upper and lower salariat) are much more supportive of late retirement for men than routine workers. Also farmers and the petite bourgeoisie show an affinity towards late exit from work. Yet the approval of early exit from work is not restricted to blue-collar jobs; it is equally pronounced among the intermediate administrative occupations as well as lower service employees.

In the second model, the employment status variable is shown to be correlated with retirement age norms. Specifically, the non-employed are more supportive of early retirement than the employed. Retirees and unemployed persons rate at one and a half years below older workers with respect to their taste for late exit. These effects are in line with a revealed preferences perspective, yet they should not be interpreted causally as selection effects are likely to be at work. The effects of class and gender are hardly affected by the inclusion of employment status.

According to the third model, having experienced unemployment lasting more than 12 months significantly lowers respondents’ support for a long working life among men. Trade union membership is likewise related to greater approval of men’s early exit from work. By contrast, the number of years of education per se seems to be irrelevant.

The second set of regressions exclusively analyses the retirement attitudes of employed respondents. This yields even larger class effects. For instance, skilled manual workers situate the age at which men become too old to work at more than two years below the age at which routine workers believe men become too old to work. In agreement with previous evidence (Moen et al., 2005), one also finds an increased preference for early exit among married persons with a retired spouse.

Surprisingly, the job satisfaction variables included in the last model are not statistically significant. Moreover, the class effects remain unaltered. Although working conditions certainly vary according to class position, differences in reported job satisfaction are virtually non-existent. It can be argued that this ambiguity between objective and subjective job quality is due to adaptive expectations (Svallfors, 2006).

The regression results regarding attitudes towards women’s retirement are shown in Table 2. Class effects are markedly stronger than in relation to male norms. For example, large employers and professionals think that women become too old to work more than three years later than do unskilled workers. In contrast to the male results, one finds significant effects for the intermediate occupations and higher-grade blue-collar workers who support longer working lives of women than routine workers do.

Table 2.

Tobit regression results: retirement age deadline for women

Age too old to work	Respondents born 1930–57			Only employed
	1	2	3	4	5	6
Female	1.504***	1.782***	1.988***	2.082***	2.251***	2.333***
Age	.435	.622*	.553*	.178	.243	.254
Age squared	−.003	−.004*	−.004	−.000	−.001	−.001
Social class (ESeC) (Reference: routine workers)
Higher salariat	3.318***	3.295***	2.069**	2.657**	1.607	1.694
Lower salariat	2.927***	2.856***	1.943***	2.846***	2.101**	2.253**
Intermediate Occupations	1.526*	1.534*	.890	.619	.220	.183
Petite bourgeoisie	3.945***	3.785***	3.440***	3.518***	3.282***	3.102**
Agricultural self-employed	.959	.705	.666	−1.311	−1.398	−1.015
Higher-grade blue-collar	1.635***	1.601**	1.239*	1.156	.952	.988
Lower sales and service	.970	.978	.711	.862	.881	.922
Skilled manual	.766	.840	.908	1.058	1.236	1.081
Employment status (Reference: employed)
Retired	–	−1.223**	−.796*	–	–	–
Unemployed	–	.963	1.473	–	–	–
Homemaker	–	−2.028***	−1.500***	–	–	–
Years of education	–	–	.133	–	.096	.088
Ever unemployed >12 months	–	–	.213	–	−.591	−.483
Log HH-income	–	–	.693*	–	.552	.513
Trade union member	–	–	.186	–	.282	.345
Health less than good	–	–	−.962***	–	−1.166	−.923
Job satisfaction	–	–	–	–	–	.184
Job interesting	–	–	–	–	–	.109
Job stressful	–	–	–	–	–	−.093
Job stressful squared	–	–	–	–	–	−.060
Family situation (Reference: married. Spouse employed)
Married. Spouse retired	−1.640***	−1.492***	−1.391***	−1.737**	−1.667**	−1.771**
Married. Spouse inactive	−1.234	−1.182	−.852	−1.840*	−1.602*	−1.540
Divorced	−.635	−.715	−.255	−.308	.005	.010
Never married	−.710	−.679	−.284	−1.565	−1.223	−1.271
Widowed	−1.466**	−1.410**	−.905	−2.347**	−1.956*	−1.967*
No. of grandchildren	−.108	−.101	−.078	.024	.067	.057
Mother working (age 16)	.160	.168	.195	.387	.398	.405
Father’s education (Reference: less than primary)
Primary	.526	.513	.194	.401	.139	.108
Lower secondary	1.244**	1.175**	.848	2.105**	1.871**	1.974**
Upper secondary	1.491***	1.418**	.731	2.164**	1.554	1.617
Tertiary	2.316***	2.226***	1.395*	2.996***	2.368**	2.629**
Country dummies	YES	YES	YES	YES	YES	YES
Constant	46.338***	39.657***	35.189***	50.435	43.854	43.792
McKelvey and Zavoina’s R²	.483	.491	.497	.521	.524	.538
N	5361	5361	5361	2145	2145	2145

Sources: European Social Survey 3 2006/07; own calculations.

p<.1 **p<.05 ***p<.01

That women are much less supportive of women’s early retirement than men is noteworthy; the estimated effect is as high as two years. This finding runs counter to the common idea that women would have a weak attachment to work. Additional information on the gender nexus is conveyed by the second model, which shows that homemakers (who are mostly female) endorse extremely low retirement age norms for women. There is, hence, a high degree of heterogeneity among women regarding retirement preferences. Furthermore, household characteristics are more powerful predictors of attitudes towards women’s exit from work than of attitudes towards men’s exit from work. The presence of a retired husband has a greater effect on wives’ retirement age norms than vice versa. This is in line with the recurrent finding that women more often adjust their work exit decisions to their spouses’ retirement timing than men (Moen et al., 2005).

In contrast to attitudes towards men’s retirement, the father’s level of education matters a great deal for attitudes towards women’s retirement. The effect is linear and positive. All else being equal, children of fathers with higher education think that women are too old to work more than two years later than do children whose father had less than primary education (some 18% of respondents). Evidently, the family of origin still plays an important role for gender-graded age norms even after age 50.

Retirement age norms and social class

Social class is of central relevance for retirement age norms. This finding leads to a rejection of the individualization hypothesis of a disappearance of traditional class divides in social norms. Members of the service class and the self-employed consistently prefer later retirement than do working class respondents. By and large, class disparities in age norms correspond to the usually observed regularities in retirement behaviour (cf. Blossfeld et al., 2006; Radl, 2012). This novel result implies that rather than having lost salience in regard to lifestyle preferences, class remains a crucial determinant of attitudes towards work and retirement, which are deeply rooted in occupational cultures. Moreover, the analyses have revealed that class disparities in attitudes towards retirement can be partly explained by differences in union membership rates and the risk of experiencing unemployment. The fact that skilled manual workers display the strongest approval of men’s early retirement suggests that firms’ labour shedding practices promote an entrenched early-exit mentality. The marked taste for late exit among the self-employed is probably not only related to their greater degree of work autonomy, but also to the lack of access to public early-retirement schemes. It can be concluded that policy instruments aiming at the prolongation of working lives need to be tailored to the varied needs and predispositions of workers situated in different positions within the occupational hierarchy.

The finding of pronounced class disparities stands in some contrast with previous research on work as a source of identity. Doherty (2009) reports that even respondents who worked in routine jobs displayed high levels of work attachment. Similarly, the older unemployed workers interviewed for another qualitative study invariably preserved their former work identity despite redundancies or health problems (Riach and Loretto, 2009). The contrast with the evidence of pronounced class differences reported here can be explained with the specificities of retirement as status configuration. Unlike unemployment or disability, retirement is not associated with social stigma. Rather, retirement is recognized in society as the final of three major life stages established by the institutionalized life course (Kohli, 2007). For older workers, the normative exigency of being economically active, both in terms of external demands (as a social norm) and as a source of identity (as internalized work ethic), ceases after reaching the ‘appropriate’ age for retirement. Although work is an important source of identity for all workers, retirement age norms can therefore differ across occupational categories.

Unfortunately, data restrictions have limited the scope of the analysis with respect to disentangling subject and object in the relationship between age norms and social class. The established class disparities may either reflect class-graded approval with generalized social norms of ageing, or indicate the locality of age norms in terms of differences across target groups. Nevertheless, the results point to the empowering capacity of age norms. In agreement with the ‘socioeconomics’ approach to social norms, internalized temporal scripts for retirement entail not only duties but also entitlements: on the one hand, retirement age norms amount to the social obligation to work until reaching the age threshold; on the other, age norms also legitimize the eventual end of the working career. Moreover, by being flexible enough to approve an earlier exit of the working class, they facilitate the psychological adjustment to retirement of occupational groups with high risks of involuntary early retirement.

Discussion and conclusions

Social norms regarding retirement are of both theoretical and practical interest. However, empirical evidence is scarce, not least in relation to Europe. The presented analyses have shown that age norms of retirement are highly pervasive among Western Europeans over 50. Significantly, unlike as suggested by postmodernist writers, most individuals ascribe an intrinsic value to work by disapproving of very early retirement. In this sense, the presented findings support the claims of life course scholars who view age norms as blueprints for biographical decision-making. In addition, the article has made a case for placing the age norms concept on a sounder theoretical basis. Accordingly, it is essential to distinguish between external constraints and internalization as two alternative norm enforcement mechanisms, which imply markedly different understandings of the relation between social norms and individual actors.

In terms of policy implications, the ongoing debate about the financial sustainability of the welfare state has raised the question of whether an entrenched ‘early-exit culture’ may undermine pension reforms aimed at deferring retirement. The presented findings indeed show that a majority adheres to social norms against working beyond the age of 65. The trend towards early exit from work during the last decades makes itself noticeable in these attitudes, which are likely to hinder the postponement of retirement ages in Europe. The efficacy of policies which rely only upon economic disincentives to reverse the early exit trend is called into question by widely internalized age norms which are incompatible with later retirement. Yet, as Etzioni (2000) points out, social norms are not entirely predetermined after early socialization, but are susceptible to modification by means of ‘persuasion’. If prolonging working lives is the aim, European employers and employees thus need to be convinced of the individual and social benefits of active ageing; instead of merely being faced with monetary punishment for early retirement, actors need to be persuaded, for example by means of targeted political and media campaigns, that older workers are more productive than often assumed. On the contrary, using public money to finance large early retirement programmes in former state-owned companies (as recently in the case of Post and Telekom in Germany or Radio Televisión Española in Spain) could well lead to undesired consequences in the long term by sending the wrong signals to younger colleagues and the wider public.

The relationship between ageism and gender inequality is complex. Early exit from work is generally seen as more appropriate for women than for men. At the same time, the empirical analyses presented here have shown that women, like men, have internalized gendered age norms of retirement. Moreover, there is substantial heterogeneity among older women with respect to attitudes towards retirement, with work attachment being especially low among homemakers. All else being equal, women show a greater level of approval of late retirement than men. While the majority of women thus consider early retirement appropriate for themselves, among work-oriented women gendered age norms probably act as external constraints on their late-career employment perspectives. With a view to equalizing employment opportunities between the sexes and reducing discrimination based on gendered ageism, the decision of policy-makers in several European countries to unify pension ages between men and women should therefore be welcomed.

Besides, for women more so than for men, age norms represent a legacy of the past. Individual perceptions as to what is the ‘right’ age for women to retire strongly depend on the father’s education. By contrast, social origins do not significantly affect male retirement norms. This discrepancy in the intergenerational transmission of retirement age norms resonates with the notion that ageist attitudes are inextricably intertwined with gender and class inequalities (Duncan and Loretto, 2004; Moore, 2009).

To conclude, this study shows that the content of age norms differs across population groups. At the same time, the presented evidence also reveals marked cross-national differences in retirement age norms. This suggests a multi-level structure of age norms, which are dependent on context conditions at the country level as well as at the level of occupational classes. While the central concern of this article has been with individual differences at the micro level, social norms of ageing are without doubt embedded in country-specific institutions such as the pension system and labour market regulation. In order to better understand international variation in attitudes toward retirement, future research needs to tackle unresolved questions regarding the relationship between age norms, the welfare state and the timing of biographical events.

Footnotes

Acknowledgements

The author wishes to thank the Editors and three anonymous referees for invaluable feedback and suggestions. I am also indebted to Fabrizio Bernardi, Klaus Haberkern, Jan Paul Heisig, Andrea Ichino and Martin Kohli for their insightful comments and criticism.

Funding

The author acknowledges with thanks the support of the Spanish Ministry of Science and Innovation (Projects CSO2010-21881 and CSO2010-21004).

Jonas Radl is a postdoctoral researcher at the Spanish Open University (UNED) in Madrid. His research focuses upon social stratification and the life course. He also has a strong interest in family sociology, social participation and pensions.

References

Anderson

(2000) Beyond homo economicus: new developments in theories of social norms. Philosophy and Public Affairs 29(2): 170–200.

Atkinson

(2007) Beck, individualization and the death of class: a critique. The British Journal of Sociology 58(3): 349–66.

Beck

(1983) Jenseits von Stand und Klasse? In: Kreckel

(ed.) Soziale Ungleichheiten. [Soziale Welt, Sonderband 2] Göttingen: Schwartz, 35–74.

Bicchieri

(2006) The Grammar of Society: The Nature and Dynamics of Social Norms. New York, NY: Cambridge University Press.

Billari

Liefbroer

(2007) ‘Should I stay or should I go?’ The impact of age norms on leaving home. Demography 44(1): 181–98.

Blöndal

Scarpetta

(1998) The Retirement Decision in OECD Countries, Economics Department Working Paper (98)15. Paris: OECD.

Blossfeld

H-P

Buchholz

Hofäcker

(eds) (2006) Globalization, Uncertainty and Late Careers in Society. Abingdon: Routledge.

Boudon

(2003) Beyond rational choice theory. Annual Review of Sociology 29: 1–21.

Coleman

(1990) Foundations of Social Theory. Cambridge: Belknap.

10.

Crompton

(2010) Class and employment. Work, Employment and Society 24(1): 9–26.

11.

De Vroom

(2004) Age-arrangements, age-culture and social citizenship: a conceptual framework for an institutional and social analysis. In: Maltby

. (eds) Ageing and the Transition to Retirement. Burlington: Ashgate, 6–17.

12.

Doherty

(2009) When the working day is through: the end of work as identity? Work, Employment and Society 23(1): 84–101.

13.

Duncan

Loretto

(2004) Never the right age? Gender and age-based discrimination in employment. Gender, Work and Organization 11(1): 95–115.

14.

Ebbinghaus

(2006) Reforming Early Retirement in Europe, Japan, and the USA. Oxford: Oxford University Press.

15.

Elster

(1989) The Cement of Society. A Study of Social Order. Cambridge: Cambridge University Press.

16.

Esser

(2005) Continued Work or Retirement? Preferred Exit-Age in Western European Countries, Arbetsrapport, Institutet för Framtidsstudier 9. Stockholm: Institute for Future Studies.

17.

Etzioni

(2000) Social norms: internalization, persuasion, and history. Law and Society Review 34(1): 157–78.

18.

European Commission (2010) Interim EPC-SPC Joint Report on Pensions. Brussels: European Commission.

19.

Gruber

Wise

(2004) Introduction and summary. In: Gruber

Wise

(eds) Social Security Programs and Retirement around the World: Micro Estimations. Chicago, IL: The University of Chicago Press, 1–40.

20.

Grusky

Sørensen

(1998) Can class analysis be salvaged? American Journal of Sociology 103(5): 1187–1234.

21.

Han

S-K

Moen

(1999) Clocking out: temporal patterning of retirement. American Journal of Sociology 105(1): 191–236.

22.

Hechter

Opp

K-D

(2001) Introduction. In: Hechter

Opp

K-D

(eds) Social Norms. New York, NY: Russell Sage Foundation, xi–xx.

23.

Higgs

Gilleard

(2006) Departing the margins: social class and later life in a second modernity. Journal of Sociology 42(3): 219–41.

24.

Kohli

(2007) The institutionalization of the life course: looking back to look ahead. Research in Human Development 4(3): 253–71.

25.

Kohli

Rein

Guillemard

A-M

Von Gunsteren

(eds) (1991) Time for Retirement. Cambridge: Cambridge University Press.

26.

Litwin

Achdut

Youssim

(2009) Who supports delayed retirement? A study of older workers in Israel. Journal of European Social Policy 19(3): 245–57.

27.

Loretto

Duncan

White

(2000) Ageism and employment: controversies, ambiguities and younger people’s perceptions. Ageing and Society 30(3): 279–302.

28.

Moen

Sweet

Swisher

(2005) Embedded career clocks: the case of retirement planning. In: Macmillan

(ed.) The Structure of the Life Course: Standardized? Individualized? Differentiated? Amsterdam: Elsevier, 237–65.

29.

Moore

(2009) ‘No matter what I did I would still end up in the same position’: age as a factor defining older women’s experience of labour market participation. Work, Employment and Society 23(4): 655–71.

30.

Mortimer

Oesterle

Krüger

(2005) Age norms, institutional structures, and the timing of markers of transition to adulthood. In: Macmillan

(ed.) The Structure of the Life Course: Standardized? Individualized? Differentiated? Amsterdam: Elsevier, 175–203.

31.

Neugarten

Moore

Lowe

(1965) Age norms, age constraints, and adult socialization. American Journal of Sociology 70(6): 710–17.

32.

OECD (2006) Live Longer, Work Longer. Paris: OECD.

33.

Queisser

Whitehouse

(2006) Comparing the pension promises of 30 OECD countries. International Social Security Review 59(3): 49–77.

34.

Radl

(2012) Labour market exit and social stratification in western Europe: the effects of social class and gender on the timing of retirement. European Sociological Review. Available at: http://esr.oxfordjournals.org/content/early/2012/03/28/esr.jcs045.short?rss=1 (published online March 29 2012).

35.

Riach

Loretto

(2009) Identity work and the ‘unemployed’ worker: age, disability and the lived experience of the older unemployed. Work, Employment and Society 23(1): 102–19.

36.

Rose

Harrison

(2007) The European Socio-economic Classification: a new social class schema for comparative European research. European Societies 9(3): 459–90.

37.

Royston

(2004) Multiple imputation of missing values. Stata Journal 4(3): 227–41.

38.

Scherger

(2007) Destandardisierung, Differenzierung, Individualisierung: Westdeutsche Lebensläufe im Wandel. Wiesbaden: VS Verlag.

39.

Scott

(2002) Social class and stratification in late modernity. Acta Sociologica 45(1): 23–35.

40.

Settersten

(2003) Age structuring and the rhythm of the life course. In: Mortimer

Shanahan

(eds) Handbook of the Life Course. New York, NY: Kluwer, 81–98.

41.

Settersten

Mayer

(1997) The measurement of age, age structuring, and the life course. Annual Review of Sociology 23: 233–61.

42.

Svallfors

(2006) The Moral Economy of Class: Class and Attitudes in Comparative Perspective. Stanford, CA: Stanford University Press.

43.

Szinovacz

Davey

(2005) Predictors of perceptions of involuntary retirement. The Gerontologist 45(1): 36–47.

44.

Vickerstaff

Cox

(2005) Retirement and risk: the individualisation of retirement experiences. The Sociological Review 53(1): 77–95.