67+ and still working
The importance of earlier job situation and retirement
plans for extended careers
- Side: 95-108
- DOI: https://doi.org/10.18261/issn.2464-4161-2020-02-03
- Publisert på Idunn: 2020-12-17
- Publisert: 2020-12-17
- Creative Commons (CC BY-NC 4.0)
The decision to raise the employment protection age limit in Norway (from 70 to 72) evoked considerable debate, with both employers’ associations and most trade union confederations opposing the change. The arguments set forth revealed a need for more knowledge about the oldest workers, and factors contributing to a late exit from the labour market. In this article, we use panel data from the Norwegian Life course, Ageing and Generation Study (2007, 2017) to explore previous work history of those who end up with careers extending beyond typical retirement age (i.e. 67). Our findings indicate that men and women who are still working when aged 67–75 have a history of high work engagement and work effort. Compared to their non-working peers in 2017, they were more likely to consider work as very important in life, perceive their job motivation as stable or improved, work long hours, be self-employed, and either have planned a late exit or made no retirement plans ten years earlier (2007). All in all, a strong inner drive for work seems to be central for a prolonged career; although among women, some may have to remain in the labour market due to financial reasons.Keywords: Older workers, work exit, retirement, gender
Due to population ageing and concern for future social welfare costs, prolonging working life has become a key priority for governments across European welfare states (e.g. Furunes et al., 2015; Hernæs et al., 2016; Wahrendorf et al., 2017). In Norway, a new pension system incentivising longer careers was introduced in 2011. Before the reform, statutory pension age was 67, but the majority had the possibility of exiting earlier by opting for a voluntary early retirement scheme (or by retiring with a disability pension due to health problems). Furthermore, a significant proportion of workers were subject to lower occupation-specific retirement age limits (particularly in the public sector) (Midtsundstad, Mehlum & Hilsen, 2017). With the introduction of a flexible pension age, all workers were given the opportunity to start drawing their pension from age 62 (provided sufficient pension accrual) (Hernæs et al., 2016). The reform also allowed for combining full work with full pension income, at the same time making it profitable to postpone both work exit and pension uptake.
In order to improve older workers’ opportunities to remain gainfully employed, the government decided in 2015 to raise the age limit for employment protection in the private sector from 70 to 72 years (and from 67 to 70 for firms with a lower company-specific age limit) (Pedersen, 2015). The age limit, or mandatory retirement age, implies that when employees reach this age, employers may terminate the employment contract without further justification. In the public sector, the limit remained unchanged (70 years – or lower in case of occupation-specific limits).
The decision to raise the age limit was controversial – both employers’ associations and most trade union confederations opposed the change (Pedersen, 2015). However, many of the arguments were based on assumptions and revealed a lack of knowledge about the oldest workers (i.e. those who work until – or beyond – statutory retirement age). Until recently, this age group has tended to be excluded from research on work and retirement: the focus has typically been on younger older workers and factors associated with early exit (e.g. Barrett & Sargeant, 2015; Wahrendorf et al., 2017). Thus, there is so far little knowledge on what may contribute to extended careers and work exit after normal retirement age.
In the existing literature, older workers are often defined as 50/55 and older (Loretto, Vickerstaff & White, 2007) or even as young as 40+ (Ng & Feldman, 2008), with 65 commonly the upper age limit. In a number of European countries, 65 is the statutory retirement age (Komp, 2018), but mean effective exit age is typically 1–2 years lower (OECD, 2017). Even in Norway, where the retirement age used to be 67 and where the labour force participation rate among older age groups is comparatively high (ibid.), less than 50 per cent are gainfully employed when they reach 65. The figure drops to 30 per cent among 67-year-olds (Statistics Norway, 2020). However, the participation rate has increased considerably in recent decades – in particular from 2011 among 62–66-year-olds (Bjørnstad, 2019). Over the last couple of years (2017–19) there also seems to be a tendency to prolong working life among the oldest workers (aged between 67 and 75), with an increase of 2–3 percentage points (Statistics Norway, 2020). Similar trends have been witnessed in other European countries (OECD, 2018; Komp, 2018).
Given governments’ wish to extend working life, there is a need for more knowledge about the oldest workers and characteristics that may contribute to a late exit. Thus, the purpose of this article is to gain a better understanding of what characterises men and women who extend their career beyond normal retirement age in Norway. We ask: What distinguishes the oldest workers from their non-working peers? To what extent do previous job situation and retirement plans matter for extended careers? And are patterns different for men and women? To answer these questions, we employ longitudinal data from The Norwegian Life course, Ageing and Generation Study.
Older workers and labour market exit
Work exit before or until typical retirement age
Extensive research on older workers and retirement behaviour has uncovered a number of factors influencing the timing of labour market exit. A common practice is to group such factors into categories, with ‘push’ and ‘pull’ being most often employed (Solem, 2012). Both terms explain early work exits (e.g. Jensen, 2005; Shultz, Morton & Weckerle, 1998): ‘push’ factors are negative circumstances that induce exit, such as poor health or adverse working conditions, whereas ‘pull’ factors are positive considerations that attract workers to leave the labour force, such as generous pensions or the wish to engage (more) in leisure activities or voluntary work. The latter has also been referred to as ‘jump’ factors (e.g. Jensen, 2005).
Circumstances leading to early exit do not necessarily have the same impact, inverted, on late exits (e.g. de Wind et al., 2016). In fact, the addition of ‘stay’ and ‘stuck’ categories, introduced by Snartland and Øverbye (2003), was in recognition of the need for a separate set of explanations for why workers continue until typical (or statutory) retirement age. Some stay because they enjoy working, they feel appreciated by their employer and co-workers, and their workplace facilitates prolonged careers. Others are stuck – they may have financial responsibilities that make income reduction difficult or they may have to continue working in order to accrue enough pension. These factors are not mutually exclusive. For most workers, several interact to influence the decision to remain or leave the labour market, which can complicate untangling the significant mechanisms. Furthermore, their impact is context-dependent, meaning that they may play out differently in various settings and for different social groups (Solem, 2012). In addition, factors explaining why some continue to work until normal (or statutory) retirement age may differ from those contributing to extended careers, i.e. beyond typical retirement age.
Who works beyond normal exit age?
With governments’ recent aim to extend working lives, several scholars have turned their attention to employment after retirement age (e.g. Pleau, 2010; Dingemans, Henkens & van Solinge, 2016; Hokema & Scherger, 2016). A frequent topic in this literature is ‘bridge employment’, which has been variously defined. Some scholars only consider jobs with a new employer, possibly in a new sector (Pleau, 2010; Alcover et al., 2014); some add self-employment (van Solinge, 2014); others consider reductions in work effort or hours as bridge jobs (also referred to as ‘phased’ retirement). It has also been argued that bridge employment involves all work undertaken while receiving a pension (Dingemans, Henkens & van Solinge, 2017). In our review of earlier findings, we will employ a broad approach and consider all research concerned with the oldest workers, irrespective of whether or not they are in a new job, have reduced their work hours and/or are receiving a pension.
Recurrent findings regarding characteristics of workers who are employed beyond typical retirement age include good health, higher education and self-employment (Hokema & Scherger, 2016; Anxo, Ericson & Herbert, 2017; Dingemans et al., 2017; Wahrendorf et al., 2017). The importance of gender is considerable – men are far more likely than women to prolong their career (de Wind et al., 2016; Hokema & Scherger, 2016; Anxo et al., 2017). According to Dingemans et al. (2016), one explanation is that women face greater difficulties in accessing bridge employment. Another reason may be that men are more likely to be self-employed, and self-employment is strongly correlated with an extended working life (van Solinge, 2014). A third reason, at least in the Norwegian context, may be that women are more often employed in the public sector where occupation-specific age limits are relatively widespread (Midtsundstad et al., 2017), and where employment beyond age 67 has been proven to be financially disadvantageous (Pedersen, 2017).
Most studies involving the oldest labour force participants are based on cross-sectional data comparing workers with non-workers. In order to investigate factors that matter for an extended career, panel data are advantageous as they allow for following the same individuals over time. A useful theoretical framework to build on here is the life course perspective, with its stress on the importance of earlier circumstances and behaviour for priorities and choices later in life (Elder, Johnson & Crosnoe, 2004). In line with this approach, previous studies have demonstrated how employment experiences accumulated over the years matter for work exit timing (Carr et al., 2016; de Wind et al., 2016). Based on data from England, Carr et al. (2016) concluded that earlier experiences of high job control, together with feelings of deserved recognition for the job performed, are important for remaining in the work force later on. De Wind et al. (2016) employed Dutch data and found previous work engagement (including both work dedication and vigour) to have an impact on working beyond retirement. Job autonomy, opportunities for learning and interesting work tasks were also important factors.
Central to the life course approach is the concept of ‘linked lives’ – the notion that individuals’ behaviour and decisions are influenced by those of significant others (Elder et al., 2004). Of particular relevance here is partner’s work status. Earlier research has demonstrated the importance of partner’s retirement decisions for early exit plans and behaviour (Syse et al., 2014). This also seems to hold for late exit (Lain, 2015).
So far there is little insight regarding the importance of gender other than the observation that men are far more likely than women to prolong their career. In most analyses, gender is only controlled for. Pleau (2010) went further and included interaction terms with gender. The results revealed a significant gender difference in the importance of marital status, with married women less likely to work beyond retirement age compared to not-married women. For men, marital status was not important. In line with this finding, she found that older US women with lower wealth were more likely to work compared to women with higher wealth. Her findings may imply that some (divorced) women are ‘stuck’ in the labour market due to financial difficulties. For men, wealth did not seem to matter. The importance of marital status is confirmed by Dingemans and Möhring (2019), who did separate analyses for men and women. Based on data from SHARE, they concluded that divorced and widowed women were considerably more likely than married women to work after reaching retirement age (which might have to do with their financial situation).
As for work history, Dingemans and Möhring’s (2019) analyses showed that for women, the more years in employment (both full and part time), the greater the propensity to work after retirement age. For men, only the number of years in part-time employment mattered. However, the authors note that as only very few men work part time, the result should be treated with caution. More important for men was previous occupational status – the higher the status, the greater the likelihood of extending the career.
Despite the recent increase in research addressing oldest workers and factors associated with late exit from the labour market, few studies include respondents in their seventies. Thus, there is a shortage of relevant knowledge about workers in the age group affected the most by the recent changes in mandatory retirement age in Norway. Another research gap concerns the importance of gender. The fact that far more men than women continue to work beyond normal exit age indicates that available knowledge is mainly based on male respondents (since most studies only control for gender). Performing separate analyses for men and women may improve our understanding of gender differences in extended careers. Finally, basing the analyses on panel data where the same individuals are interviewed in multiple waves enables an exploration of how previous job situation impact on current work status, which may be valuable for better understanding prolonged careers.
Data and methods
The analyses build on data from the Norwegian Life course, Ageing and Generation Study (NorLAG). So far, the study contains data from three waves (2002, 2007, 2017), including telephone interviews and self-administered questionnaires, as well as annually updated register data. The data is collected by Statistics Norway. In order to address our research questions, we use information collected in the last two waves. Due to a large refresher sample, the wave 2 sample is considered nationally representative (N=8,977 individuals aged 40 to 79, response rate 62%) (Bjørshol, Høstmark & Lagerstrøm, 2010). In wave 3, Statistics Norway recontacted all previous participants aged 50 and older (in 2017); 68 per cent responded (N=6,099). In both waves, three out of four telephone interview respondents also returned the self-administered questionnaire (for more information about sampling procedures, etc., see Bjørshol et al., 2010; Torsteinsen & Holmøy, 2019).
In order to examine the oldest workers and factors contributing to a late exit, we base our analyses on a subsample consisting of wave 2 and wave 3 participants aged 67 to 75 in 2017 (n=1,450). Although 67 is no longer the statutory retirement age in Norway, it remains important because it is the age when all citizens are entitled to a minimum pension regardless of their pension accrual. Furthermore, at age 67, unemployment benefit is withdrawn, sickness allowance reduced, and disability pension is automatically substituted by old age pension; 75 is chosen as the upper limit as the new pension system allows for accumulating pension entitlements until this age. Since our focus here is on earlier job situation and retirement plans, we select men and women who were employed in wave 2 (when aged 57–65, n=936). As some of the questions about work (wave 2) were asked in the self-administered part, the sample is restricted to respondents who returned the postal questionnaire (n=840).
In most surveys, particularly longitudinal studies where the same individuals are followed over time, a challenging issue is non-response and selection bias. In NorLAG, two-thirds of the respondents aged 57–65 in wave 2 participated in wave 3 (then aged 67–75). If we compare the whole sample of wave 2 respondents aged 57 to 65 with the two-time participants (wave 2 and 3), who in addition completed the self-administered questionnaire in wave 2 (a prerequisite for being part of our sample), the latter have on average a higher educational level, were more often in paid work, and perceived their health status and financial situation to be better in wave 2. However, the selection bias is quite modest – the differences are between five and seven percentage points. If we only consider individuals in this age group who were working in wave 2 (i.e. the focus of our analysis), more than 70 percent took part in wave 3. Hence, the selection bias for our analytical sample is even smaller than for the whole sample (analyses available upon request).
Dependent and independent variables
Our dependent variable is being in paid work or not at the age of 67 to 75. The variable is measured in 2017 through the following questions: ‘Did you do paid work for at least 1 hour last week?’ and ‘If no: Are you in paid work that you were temporarily absent from last week?’. A ‘yes’ answer to either question is considered as being in paid work.
Gender is included in the overall regression. In addition, we perform separate analyses for men and women. Age (2017) is included as three dummy variables: ‘69–70’, ‘71–72’ and ‘73–75’, with ‘67–68’ as reference category. Education is the respondent’s highest level of education (1 = college/university education). The variable is based on register data from 2007. Due to the age of our sample, few respondents have changed value between the two waves, and the variable is thus considered to be non-varying.
As we are interested in the importance of earlier job situation and retirement plans for current work status, the majority of our independent variables stem from wave 2 (2007). These consist of both objective and subjective aspects of the respondents’ job situation, as well as exit plans. Apart from gender, age and education, we control for health status and household economy in 2007. In addition, we include changes in the respondent’s health status between 2007 and 2017, plus partner’s work status in 2017. For the regression analyses, all independent variables are made into binary variables.
Objective aspects of the respondents’ job situation in 2007 include sector affiliation and work hours, while subjective aspects consist of job satisfaction, perception of changes in work ability and work motivation, and work’s importance in life. Sector affiliation is measured by combining a question on employment/self-employment with a question about sector. In the analyses, we include two dummy variables: ‘private sector’ and ‘self-employed’, with ‘public sector’ as reference category. Work hours refers to the number of hours per week the respondents usually worked in their main occupation in 2007, including overtime. The variable is recoded into three categories: ‘normal hours (31–40)’ and ‘long hours (41+)’, with ‘short hours (1–30)’ as reference category. Job satisfaction is based on the following question: ‘On a scale from 0 to 10, where 0 means “not satisfied at all” and 10 means “completely satisfied”, how satisfied are you with your current job?’. For the binary variable, ‘1’ equals the three highest values (8–10). Work ability and work motivation is the respondent’s own evaluation of changes in his/her ability and motivation to work in 2007 compared to ten years earlier. The five response categories range from ‘much better’ to ‘much worse’, with ‘no change’ being the middle category. For the binary variables, ‘1’ equals stable or positive change and includes the three highest values (‘no change’, ‘better’ and ‘much better’). Importance of work is the respondent’s perception on how important work was in their life in 2007 (very important/quite important/somewhat important/not important). For the binary variable, ‘1’ equals ‘very important’.
Plans for retirement/work exit is measured by combining questions on whether or not the respondents have thought about retirement timing, and if yes, if they have decided when to exit. In the analyses, plans for retirement in 2007 is included as ‘no plans’ and ‘plan to exit at 67 or later’, with ‘plan to exit before 67’ as reference category. Due to complex filter procedures, 5 per cent of the sample was by mistake not asked about retirement plans. To avoid omitting these cases from the analyses, they are included in a dummy for missing information.
Financial situation is the respondent’s own assessment of how hard or easy it was for the household to make ends meet in 2007. The answers are based on a 6-point response scale (ranging from ‘very hard’ to ‘very easy’). For the binary variable, ‘1’ equals ‘easy’ and ‘very easy’. Subjective health is the respondent’s evaluation of own general health in 2007 (excellent/very good/good/fair/poor). For the binary variable, ‘1’ equals the two highest values: ‘very good’ and ‘excellent’. Health change is the change in the evaluation of general health between 2007 and 2017. For the binary variable, ‘1’ equals no change or a positive change. Finally, partner’s work status (in 2017) is measured in the same way as for the respondents (see above). The variable is included in the regression analyses as two dummy variables: ‘no partner’ and ‘partner still working’, with ‘partner not working’ as reference category.
The multivariate regression analyses are performed with the full sample, and separately for women and men to uncover whether or not the independent variables work differently for the two genders. We use linear probability models (LPMs). An important advantage is that such models provide results that are easier to interpret compared to logistic regression, which is commonly employed when the dependent variable is binary. The statistical objections to using linear regression when the dependent variable only have two values have been shown to have little practical significance (Hellevik, 2009).
In our sample of men and women aged 67 to 75 (in 2017), who were all in paid employment in 2007, 23 per cent are still working in 2017. In this group, 9 out of 10 also receive a full pension. The descriptive statistics (Table 1) confirm the gender difference reported in earlier studies – 18 per cent of the women are still in paid work, compared to 28 per cent of the men. The age distribution of our sample is skewed towards younger ages (67–70). The reason is our criteria for being eligible for this study, which was having a paid job when interviewed ten years earlier (in 2007). At that time, the respondents were aged 57–65, and some of the oldest had already left the labour market. Seemingly, a larger share of the women in our sample has a higher educational level compared to men, but the gender difference is not significant (5% level).
Turning to our main variables – job characteristics and retirement plans – we note that most women worked in the public sector in 2007 (62%), whereas most men were in the private sector (51%) or self-employed (19%). Self-employment was less common among women (8%). Women were more likely to work part-time – 38 per cent worked 30 hours our less per week, compared to 16 per cent of the men. Men, on the other hand, were more inclined to work long hours – 31 per cent worked more than 40 hours per week, compared to 13 per cent of the women. As for subjective characteristics of the respondents’ job situation ten years earlier, the majority of both men and women (around 75%) were highly satisfied with their job. For the importance of work – a larger share of women (39%) than of men (30%) regarded their job as very important in life in 2007. Identical shares of men and women (74%) reported a positive change, or no change, in their work motivation during the previous ten years; somewhat fewer (62–65%) said that their work ability had changed in a positive direction or remained the same. Men and women in our sample had similar plans regarding future work exit. Forty per cent of the women and 45 per cent of the men had made such plans in 2007 – respectively 26 and 28 per cent had scheduled to leave before, whereas 14 and 17 per cent had planned to leave after turning 67. Approximately half of the sample (50% of the men and 54% of the women) had not made any plans.
Regarding the respondents’ financial situation in 2007, more men (72%) than women (64%) reported that they found it (very) easy for their household to make ends meet. For health status, on the other hand, there are no significant gender differences: 56 per cent of the women and 53 per cent of the men considered their health to be very good or excellent in 2007, and 63 per cent (both genders) reported no changes or an improvement in health between 2007 and 2017. Finally, partner’s work situation: 21 per cent of the men have a partner who is still working in 2017 and 17 per cent of the women, a non-significant difference. Men are, however, far more likely to have a partner than women are: 81 compared to 64 per cent.
|In paid work (2017)||18||**||28|
|Education – high (university/university college)||45||39|
|Sector affiliation (2007)|
|Work hours per week (2007)|
|Short hours (1–30)||38||**||16|
|Normal hours (31–40)||50||53|
|Long hours (41+)||13||**||31|
|Job satisfaction - high (2007)||74||76|
|Importance of work – very important (2007)||39||**||30|
|Work motivation – stable or positive change (2007)||74||74|
|Work ability – stable or positive change (2007)||65||62|
|Plans for retirement/work exit (2007)|
|Exit before 67||26||28|
|Exit at 67 or later||14||17|
|No plans made||54||50|
|Financial situation – (very) easy to make ends meet (2007)||64||*||72|
|Subjective health – very good/excellent (2007)||56||53|
|Health change – stable or positive (2007–2017)||63||63|
|Partner’s work situation (2017)|
|Partner not working||47||**||60|
|Partner still working||17||21|
** p<0.01, * p<0.05 (p-values in a two-sided test for the differences between women and men in our sample).
ª Missing as a result of filtering errors.
Table 2 displays the effects of our independent variables on the likelihood of having paid work in 2017 when aged 67 to 75 for the whole sample, and separately for men and women. The model for the full sample confirms the gender difference in current labour market participation among the oldest workers, with men being somewhat more likely than women to work after 67 (in 2017). In this age group (67–75), only the oldest (aged 73–75) are less likely to have a paid job compared to the youngest (67–68). The lack of a clear gradual decline in participation with age may indicate that men and women who are still working after turning 67 are likely to continue for some more years. Education has surprisingly little importance. We find only a modest association for women, but in the opposite direction to what earlier research has reported. Having a college or university education reduces the probability of being employed after 67 for women.
Turning to the association between earlier job situation and current work status, the importance of self-employment is confirmed. Men who were self-employed in 2007 are considerably more likely to still be working in 2017 (when aged 67–75) compared to public sector workers – and also compared to those who worked in the private sector (not shown in the table). For women, self-employment does not seem to matter; what is important is work hours. Women who used to work long hours (41+ per week) in 2007 are more likely to continue working after turning 67 compared to those who worked normal hours (31–40 hours) or part-time. The pattern is somewhat different for men, as those who worked long hours are more likely to work after 67 than workers with normal hours, but they do not differ significantly from part-time workers (results available upon request).
Looking at subjective dimensions of the respondents’ work situation in 2007, we do not find job satisfaction to be associated with labour market status ten years later. The importance of work in life, however, does matter. Considering work as very important (in 2007) has a significant effect on current work status for both women and men. Deeming one’s work motivation to be stable or improved in 2007, compared to ten years earlier, also has a positive influence on the likelihood of having a paid job in 2017 (but the coefficient is only significant for men). Perceived change in work ability has no impact on current employment for either women or men. Moving on to retirement plans, we see that planning for a late exit (67+) or having no plans about when to leave (in 2007) increases the likelihood of still working ten years later for men, compared to having planned for an early exit (before 67).
Our analyses also include respondents’ perceived financial situation and health status in 2007. Women who considered it easy or very easy for their household to make ends meet are less likely to be working after turning 67 (in 2017). Considering one’s health to be very good or excellent, on the other hand, increases the likelihood of continuing working. Here, we take account of both respondents’ subjective perception of their health in 2007 and the change between 2007 and 2017. The associations are only significant for women, but the coefficient for subjective health in 2007 is close to being significant also for men (p=0.064). Finally, having a working partner increases the probability of still working after turning 67 among men.
The full model for the whole sample presented in Table 2, explains 15.1 per cent of the variability (adj. R2) in work status for the 67–75 age group. A model including only the control variables (not shown in the table) reach an adjusted R2 of 0.052, meaning that earlier job situation and retirement plans account for as much as two thirds of the explained variance.
In addition to the variables displayed in Table 2, we ran models including a number of psychosocial work characteristics in 2007, such as being asked for advice by co-workers and having opportunities to learn new things. As these variables did not have significant effects on labour market participation ten years later, they were removed from our final analyses.
One could perhaps argue that within the 67–75 age group, there are cohorts with different experiences in institutional conditions due to recent changes in both the pension system and the employment protection age limit, and thus, characteristics associated with continued work for those aged 67–69 could differ from those applying to 70+ workers. However, additional analyses separating workers below and above 70 revealed the same main findings (with the exception of health, which seemingly only matters for workers aged 70+). Furthermore, including respondents who did not return the postal self-administered questionnaire in 2007, did not alter our main results (analyses available upon request).
|Age (2017) (ref.: 67–68)|
|Education (1=higher level)||-0.048|
|Sector affiliation (2007) (ref.: public sector)|
|Work hours per week (2007) (ref.: 1–30 hours)|
|Normal hours (31–40)||0.006|
|Long hours (41+)||0.178|
|Job satisfaction (2007) (1=high)||0.024|
|Importance of work (2007) (1=very important)||0.118|
|Work motivation (2007) (1=stable or positive change)||0.106|
|Work ability (2007) (1=stable or positive change)||-0.005|
|Retirement/exit plans (2007) (ref.: exit before 67)|
|Exit at 67 or later||0.128|
|No plans made||0.087|
|Financial situation (2007) (1=(very) easy to make ends meet)||-0.058|
|Subjective health (2007) (1=very good/ excellent)||0.097|
|Health change (2007–2017) (1=stable or positive)||0.060|
|Partners work situation (2017) (ref.: partner not working)|
|Partner still working||0.133|
|Explained variance (Adj. R2)||0.151||0.101||0.187|
*** p<0.001, ** p<0.01, * p<0.05
ª Missing as a result of filtering errors.
Discussion and conclusion
The decision to raise the mandatory retirement age in Norway stirred considerable debate. Both employers’ associations and most trade union federations opposed the change. Their arguments included a fear that employers would have to keep a growing number of less productive older workers for a longer period, that workplace conflicts would be more common, and that the risk of disgraceful work exits among older employees would increase. Underlying such assumptions is a conception of older workers as unmotivated, unproductive and unprofitable, a conception based on scant empirical evidence (Ng & Feldman 2008; 2012). Thus, the debate revealed a need for more knowledge about the oldest workers and factors contributing to late work exits.
Our analyses have provided some insights that may challenge the opposition to the change in mandatory retirement age in Norway. All in all, we find that men and women who continue to work after 67 are characterised by a history of high work engagement and work effort. When ten years younger (57–65 in 2007), they were more likely to work long weeks (more than 40 hours), consider work as very important in life, and deem their job motivation to be stable or improved during the previous decade compared to their same-age peers who have left the labour market in 2017. They were also more likely in 2007 either to envision a late retirement (67+) or to have made no specific plans for the timing of retirement. The lack of a clear gradual decline in workforce participation with increasing age within our specific age group suggests that those who are still working when they turn 67 are prone to continue for at least some more years, another indication of a strong inner drive for work. All things considered, it seems reasonable to assume that many of those still working after turning 67 are doing so because they enjoy their work.
Men are more likely than women to continue working when aged 67 to 75. Our analyses reveal some similarities across gender, but also differences, with financial situation being the most important. For women, we find that having a more difficult financial situation in 2007 increases the likelihood of remaining in the labour market beyond normal retirement age compared to being in a situation where it is easy to make ends meet, suggesting that some women may have to continue working for financial reasons.
Our study confirms the importance of self-employment for extended careers reported in international studies. Older workers are more inclined to be self-employed compared to younger workers, the more so the older they become. Although promoting senior entrepreneurship was listed as one of the goals of the European Year of Active Ageing and Solidarity between Generations (EU, 2012), one may question the value of widespread self-employment in late careers if it is in response to a lack of opportunities provided by employers.
Extending working life is a key goal in ageing societies. Alongside increasing population ageing, there is a growing number of skilled and healthy older workers who are eager to delay their work exit. Raising the mandatory retirement age has removed one hindrance to extended careers. Another obstacle is employers’ lukewarm attitudes towards older workers (Conen, Henkens & Schippers, 2012). A recent scoping review provides sound evidence of widespread ageist stereotypes and perceptions of older workers that obstruct extended careers (Harris et al., 2018). Such perceptions are also present in the Norwegian debate about mandatory retirement age (Solem, 2018). The question of how to alter these deep-rooted attitudes to older workers is a timely one that needs further attention – in both research and policy – in the coming years.
The research presented here was carried out with ﬁnancial support from the Research Council of Norway (Grant N° 254786 and 236997). The NorLAG survey data collections have been financed by The Research Council of Norway, four Norwegian Government Ministries, The Norwegian Directorate of Health, The Norwegian State Housing Bank, Statistics Norway and NOVA at Oslo Metropolitan University. NorLAG data (DOI: 10.18712/norlag3_1) are part of the ACCESS Life Course infrastructure funded by the National Financing Initiative for Research Infrastructure at the Research Council of Norway (Grant N° 195403 and 269920).