The dynamic effects of retirement on well-being

ABSTRACT

In this paper we examine whether and to what extent people anticipate their retirement and adapt to it over time. For identification we use an instrumental variable approach that exploits the eligibility ages for state pension as a source of exogenous variation in retirement status. We show that retirement increases satisfaction with the amount of leisure time up to two years prior to retiring. Upon retirement, individuals reported higher satisfaction with their income and leisure. For leisure the well-being boost did not dissipate over time, but for income satisfaction adaptation was complete. These findings provide new empirical evidence on the causal link between retirement and well-being and suggest that such link may be better understood as a three-stage process occurring before, during and after retirement.

KEYWORDS:

• Retirement
• life satisfaction
• well-being
• BHPS

JEL CLASSIFICATION:

• D01
• I31
• J26

Acknowledgments

We would like to thank Mark Taylor (the Editor) and two anonymous referees. The Understanding Society data used in this publication were made available through the UK data archive at the University of Essex. Understanding Society is an initiative by the Economic and Social Research Council (ESRC), with scientific leadership by the Institute for Social and Economic Research (ISER), University of Essex, and survey delivery by the National Centre for Social Research and TNS BRMB. Neither the original collectors of the data nor the Archive bear any responsibility for the analyses or interpretations presented here.

Disclosure statement

The authors declare that they have no conflict of interest.

Notes

¹ A growing body of research in gerontology has proposed a life-course, ecological perspective for understanding the retirement transition (Elder 1995). This view suggests that retirement is a dynamic process and, as such, it should be placed in a larger content of ever-changing roles and social relationships. A related observation is that lives are inter-dependent in that other people’s health and retirement status are often important drivers of one’s own retirement decision.

² Atchley (1976) has also suggested that retirees potentially enjoy their retirement during a ‘honeymoon phase’, a period of freedom to do things they did not have time for before retirement.

³ Studies have also recognized that post-retirement activities may play an important role in retirement satisfaction, with those who engage, for instance, in volunteer work being more likely to achieve higher well-being in retirement (Haski-Leventhal 2009; Wang and Hesketh 2012).

⁴ In a related study, Heckhausen and Schulz (1993) integrated Baltes’ model with the life span theory of control. Combining the concepts of primary and secondary control with those of selection and compensation, four types of life-management strategies are suggested, which in turn are governed by a so-called higher-order process of optimization.

⁵ Wave 11 of the BHPS did not include questions on satisfaction with life and domains of life. However, it provides information on the respondents’ retirement status which we use to track the number of years leading up to and following retirement.

⁶ According to Van Praag, Frijters, and Ferrer-i-Carbonell (2003), satisfaction with overall life is a global conception of well-being. Satisfaction with specific domains of life, such as those of income and leisure, contribute to one’s satisfaction with life overall.

⁷ The base year is 2010.

⁸ The home price reflects the price that an outright homeowner would expect to sell his/her home at the time of the interview.

⁹ This variable is constructed from the employment rate of those aged 50–65 and the employment rate of those aged over 65, both measured by region and year.

¹⁰ During our sample period, there were two types of state pension: the Basic State Pension and an earnings-related state pension (known as SERPS/S2P) that only some people had eligibility to. For our IV estimation, we use the Basic State Pension as it applied to all individuals.

¹¹ The UK basic state pension can be claimed at the state pension age, which was 60 for women and 65 for men until 2010. There is no earnings test or means test for eligibility. Since April 2010, the women’s qualifying age for state pension started rising, something that we are able to account for in our IV approach.

¹² Our IV estimates identify a Local Average Treatment Effect (LATE) for compliers, namely those who would retire if they were eligible for basic state pension (versus those who would not retire if they were not eligible); see e.g. Angrist, Imbens, and Rubin (1996). Clearly, for those compliers retirement is an anticipated life event because they are entitled to receive the basic state pension as soon as they reach the qualifying age.

¹³ This was implemented in Stata using the xtivreg2 command of Baum, Schaffer, and Stillman (2010).

¹⁴ Since we observe a small decrease in the number of working hours in the years prior to retiring (Appendix Table A2), we are unable to pinpoint the exact source of these anticipation effects – increased optimism in view of people’s upcoming retirement or a reduction in the actual time spent working. The reduction in the number of working hours, however, is rather small and does not appear to spill-over on satisfaction with one’s life overall. Thus, we believe that the reduction in the number of working hours is not a leading candidate in explaining our anticipation results.

¹⁵ We also estimated our model by including at the same time controls for health status and household income, and found that the results were consistent with the model that did not condition for both variables in the estimation.

¹⁶ Similar to satisfaction with the domains of income and leisure, job satisfaction is measured on a 7-point scale that ranges from 1 (not satisfied at all) to 7 (completely satisfied). Note that using here the random effects estimator allows us to keep in job satisfaction which is measured at baseline.

Additional information

Funding

This work was supported by the University of the Thai Chamber of Commerce under its grant scheme.