In this prospective study of Finnish public sector employees, several characteristics of social relationships in private life, such as marital status, high social participation, as well as financial difficulties predicted short and long extension of employment beyond the pensionable age. In men and women, having a working spouse was associated with long-term extension. In addition, women who were active in consumptive social participation (e.g., attending cultural and religious activities and studying) or in formal social participation (e.g., being members of club and societal activities) were more likely to extend their employment for at least one year. Interestingly, those women who actively participated in handwork and collecting hobbies, playing an instrument, singing, photographing, painting, physical activity, or outdoor activities, were less likely to extend their employment.
In agreement with previous studies (Gonzales and Nowell 2017; Scharn et al. 2017; Virtanen et al. 2014), we observed that men were more likely than women to extend their employment. It may be that women more often prioritize private-life activities, for example taking care of grandchildren and parents, meeting friends, and sharing time with a spouse when deciding about retirement timing (Szinovacz et al. 2001). Our findings that being alone (Gonzales and Nowell 2017; Virtanen et al. 2014) or having a working spouse was associated with extension of employment accords with previous studies (Gonzales and Nowell 2017; Scharn et al. 2017), and may indicate the principle of ‘linked lives’ (Morrow-Howell et al. 2014). According to that principle, individuals’ lives are bound to the lives of others, and for example, transition to retirement may largely be shaped by close social relationships.
To assess social engagement type and frequency, we used similar concept as in the Maastricht social participation profile with different categories: consumptive social participation, formal social participation, informal social participation, and other social participation, reflecting different types of activities (Mars et al. 2009). To our knowledge, there are no previous longitudinal studies assessing specifically these social activity characteristics with regard to extended employment. We found that women who were active in consumptive social participation (e.g., cultural and religious activities and studying) or in formal social participation (e.g., club and societal activities) were more likely to extend their employment for at least one year beyond the pensionable age. Supporting these results, previous studies have reported an association between volunteering and working after retirement (Carr and Kail 2013; de Wind et al. 2016; Gonzales and Nowell 2017). This suggests that those who are active in leisure time are inclined to actively participate in work life, and vice versa (Carr and Kail 2013; Morrow-Howell et al. 2014). Engaging in different social activities foster also interactions with a wider social network of individuals in heterogeneous social groups, which in turn can offer resources (such as information exchange, social capital, social support) needed to remain active and hold to one’s work (Carr and Kail 2013; Gonzales and Nowell 2017). In addition, engagement in volunteering and other social activities associate with better health and reduced depressive symptoms (Carr and Kail 2013; Gonzales and Nowell 2017), and those, in turn, who have better health are more likely to have more role involvements both in volunteering or other social activities and in work life (de Wind et al. 2016; Morrow-Howell et al. 2014). In our analyses, however, significant association between engagement in social activities and extension of employment remained after adjusting for self-rated health suggesting independent effect of those activities. However, high frequency of social engagement in activities such as handwork and collecting hobbies, playing an instrument, singing, photographing, painting, physical activity, and outdoor activities (listed as other social participation in our study) were associated with a lowered likelihood of long extension of employment among women in our study. There is a lack of previous studies assessing specifically the association of these activities with extension of employment beyond the pensionable age. The study by Tuisku et al. (2016) showed that cultural leisure activities were associated with beneficial recovery experiences and work engagement among hospital employees. These features could be expected to associate with higher likelihood to extend employment. Although we do not have a conclusive explanation to our finding, the difference could be attributed to the type of study. In the aforementioned study, they checked the cross-sectional association compared to the longitudinal predictions used in our study.
In addition to social relationships, we and other studies have observed financial difficulties to predict extension of employment in women (de Wind et al. 2016; Sewdas et al. 2017; Szinovacz et al. 2001). Since women have in general lower average pension accrual than men (Eläketurvakeskus 2018), they may need to continue working for economic reasons more often than men do. However, not all studies have supported the evidence that financial resources are associated with employment after retirement (Dingemans et al. 2016). The reason could be that some people might continue working due to financial difficulties and some other could continue on the same job due to other motivational aspects. On the other hand, it has been suggested that higher income is associated with extended employment (Scharn et al. 2017). We were unable to provide information on those whose financial situations were very good because we did not have information related to income of the participants in our study. Therefore, it is possible that the perception of one’s financial situation is conceptually different from actual value of income, and thus, the results differ (de Wind et al. 2016).
We did not find significant association between care giving and extended employment either in men or in women although previous studies have found that providing high intensity of informal care is associated with higher likelihood of being retired or out of the work force (Bolin et al. 2008; Jacobs et al. 2014). On the other hand, it has been suggested that probability of being employed depends on the type of care provided. Care provided within the household (co-residential care) has been shown to significantly reduce the employment probability, while extra-residential care has not been shown to have an impact on the labor market participation (Heitmueller 2007). In our study, the number of participants who reported to provide informal care giving was small, which may at least partly explain the observed nonsignificant associations. For the same reason, it was not possible to assess whether the association would have been different according to different intensities or type of provided care. On the other hand, providing informal care may be associated with higher likelihood to continue working especially when expenses of care arrangements require that. Moreover, working may provide a coping strategy for dealing with the stress of being a caregiver (Burr et al. 2005).
The use of repetitive yearly measurements among a large representative sample of an occupationally established cohort and date-based information about individual pensionable date and timing of the actual retirement are the strengths of this study. We covered different characteristics of social relationships assessed when still in employment. All participants were still in employment, when first contacted. This ‘healthy worker effect’ implies that those with major health problems, an unlikely population for a lengthy extension of employment, had retired earlier, and were not in our study cohort. Thus, the participants in the groups of no extension, short extension, and long extension were relatively homogeneous suggesting that health related selection bias is not a major concern in our study.
The study has also some limitations, which warrant discussion. First, the lack of information about earlier mid-life experiences, e.g., educational investments, job changes, late transitions into parenthood, or late divorces, which have been shown to associate with intentions to retire later (Damman et al. 2011), is the salient limitation of this study. However, we did have information about occupational and marital status, which reflect some of these earlier experiences. Likewise, we did not take into account the factors associated with early health-based retirement because our sample includes those who continued until old-age pension and retired on their individual pensionable age or worked beyond that. In addition, although our findings may be generalizable to employees in Finland and other Nordic welfare countries, they may not be generalizable to countries with different pension systems. Moreover, there were relatively few men compared to women, due to which not all the associations among men were found statistically significant, despite tending a direction toward an association. This type of gender distribution (83% women in our study) is typical in Finnish public sector occupations (Statistics Finland 2016).
In conclusion, social relationships in private life could predict the choice to extend employment beyond the pensionable age. While having a working spouse was a contributing factor for both sexes, several other factors, such as living alone and high social engagement frequency, were additionally related to extended employment among women. The examination of potential mechanisms through which different characteristics of social relationships contribute to extended employment is warranted for future studies.