From the aDepartment of Economics, Queens College, City University of New York, Flushing, NY; bNational Bureau of Economic Research, Cambridge, MA; and cDepartment of Economics, Bentley University, Waltham, MA.
Correspondence: Inas Rashad Kelly, 300 Powdermaker Hall, 65–30 Kissena Boulevard, Flushing, NY 11367. E-mail: Inas.Kelly@qc.cuny.edu.
The pure investment form of Grossman's seminal demand-for-health model1 proposes that people invest in health to raise productivity and earnings; thus there would be little point to investing in one's health postretirement, and the health stock is predicted to experience a sharp decline. Yet, as Grossman also points out, health has consumptive aspects. One has an incentive to invest in one's health even postretirement because, as economists would say, health directly enters the utility function. Put more simply, it makes one happy to be healthy. So perhaps retirees have a reason to care about their health after all, although the net effect on health may depend on which effect dominates.
In the absence of unambiguous theoretical predictions, the question of whether retirement has an adverse effect on mental and physical health remains an empirical one, and one that several studies have recently attempted to address. The question has important policy implications as we witness a trend toward earlier retirement in most developed nations, along with rising life expectancy and higher health care costs.
In this issue of Epidemiology, Oksanen and colleagues2 carefully analyze the effect of retirement on mental health as measured by antidepressant use 4 years before and 4 years after retirement among a longitudinal cohort of Finnish employees. Finland had a low average retirement age of 59 years in 2004.3 However, the Finnish Pension Reform of 2005 made the retirement age more flexible and encouraged postponing retirement, with the increase in labor supply purported to lead to beneficial welfare effects.4 Oksanen et al suggest otherwise, at least with respect to mental health. Accounting for time-invariant factors such as socioeconomic status, sex, geographic area, and type of employer, and analyzing those who retired due to poor mental or physical health separately, the authors find some evidence that retirement may be beneficial for mental health. They also conclude that those who retired due to mental health issues showed postretirement improvements, in that they reported less antidepressant drug use. Diabetes medications, presumed not to be directly influenced by retirement, are used as a counterfactual. (No effects are found for diabetes drugs, as expected.)
The literature on this issue has been fragmented. In another study from Finland, Salokangas and Joukamaa5 also noted mental health improvements among retirees, but no clear effect on physical health. In contrast, Szinovacz and Davey6 found that depressive symptoms increases in women after retirement, especially if retirement is perceived as abrupt or forced. The effect is reinforced by the presence of a spouse with functional limitations.
Oksanen and colleagues2 acknowledge that “the reasons for the[se] inconsistencies … may be due to differences in study design.” For one thing, the ways health is measured can lead to vast differences in results. Although the use of a more objective measure such as prescription drug use in the Oksanen study may bypass some misreporting issues, there are nonetheless concerns with this measure, particularly as a proxy for underlying mental health, for which reported measures (such as the DSM-III and CESD scales of depression) are heavily relied upon. Use of antidepressants, in addition to reflecting underlying mental health, also reflects health care access, physician contact, diagnosed mental illness, and decision to seek treatment, all of which may be affected by retirement due to shifts in resource, time, and income constraints.
Another key issue relates to endogeneity—in that health is not completely independent and is likely determined within the system being modeled. Endogeneity can arise from 2 sources. Structural endogeneity, or reverse causality, can occur because poor physical or mental health may be one reason to retire. This is reflected in the data used by Oksanen et al (notably in the Table 1 statistics and in eFigures 3 and 4) by a run-up in antidepressant use prior to retirement, especially among those who retire earlier than at the statutory age (suggesting poor health as an underlying factor). Moreover, statistical endogeneity implies that unobservable factors common to an individual may affect both his decision to retire and his health trajectory. Confounding factors abound, and they complicate the relationship. What are spouses up to, and how does their behavior affect the decision to retire? Does the person work in a field she is passionate about and enjoys, or is work unstimulating and stressful? Environmental factors related to the person's geographic area of residence may affect the outcomes as well, and whether she has hobbies and social activities outside of work. The possibility that standard of living could decline after retirement must also be considered (and varies considerably by socioeconomic status and country). Then, there is the simple fact that people age as they retire, and the rate of depreciation in one's health stock tends to increase at an accelerating rate over that age range. Oksanen and colleagues account for some of the time-invariant confounders, but not the time-varying ones.
With our colleague Jasmina Spasojevic,7 we have estimated the potential effects of retirement on various measures of physical and mental health, using longitudinal US data from the Health and Retirement Study spanning 1992 through 2005. Being able to observe health, other time-varying factors, and spousal behaviors over all waves, we accounted for the potential endogeneity of retirement and for confounding factors affecting the retirement decision. Our results consistently pointed to retirement having adverse effects on both physical and mental health on average. However, it is important to underscore the considerable heterogeneity that can be masked by average population effects. For instance, adverse effects on mental health tend to be larger on average among women than among men after retirement. Among those who are married and have social support, those who remain physically active after retirement, those for whom retirement was voluntary (excluding health as a reason), or those for whom retirement was partial, there were no adverse effects on mental health.
These results may help explain the divergent findings when countries with varying social programs are analyzed. They may also highlight the importance of identifying subgroups, and what may be most beneficial for them. Although retirement may be the optimal solution for those with mental health problems, it is unclear that retirement is beneficial for mental health or for physical health in general.
These studies and the results highlighted by Oksanen et al2 have implications for social security programs in most industrialized nations. Gruber and Wise8 note that many countries structure their benefits so as to discourage work (eg, lowering lifetime benefits for people who work longer). With an aging population that is retiring earlier, and an unfunded liability facing public pensions, policy makers have pressed for several reforms including an increase in the retirement age. In the United States, social security data indicate that the retirement age for men has declined from 68.5 to 62.6 years and for women from 67.9 to 62.5 years; similar trends have been noted in Germany, Japan, and Sweden.9,10 More recently, however, 63% of workers in the United States reported wanting to delay their retirement age in light of the poor economy and related factors.11 Whether early retirement is individually or socially optimal depends on how retirement affects subsequent health status, among many other things. If retirement improves mental health, then this may lower health care spending among older adults. Evaluation of policies that seek to prolong retirement should account for this effect on health.
ABOUT THE AUTHORS
INAS KELLY is an Assistant Professor of Economics at Queens College and Faculty Research Fellow at the National Bureau of Economic Research. Her research focuses on consumer choices and economic effects related to obesity, diabetes, nutrition, and exercise. DHAVAL DAVE is Associate Professor of Economics at Bentley University and Research Associate at the National Bureau of Economic Research. His research focuses on the analysis of health and labor policy and on the economics of health outcomes and behaviors, health insurance, and human capital.
1. Grossman M. The human capital model. In: Culyer AJ, Newhouse JP, eds. Handbook of Health Economics. Amsterdam: North-Holland; 2000.
2. Oksanen T, Vahtera J, Westerlund H, et al. Is retirement beneficial for mental health? Longitudinal analysis of antidepressant use before and after retirement. Epidemiology. 2011;22:553–559.
4. Lassila J, Valkonen T. The Finnish Pension Reform of 2005. Geneva Papers. 2007;32:75–94.
5. Salokangas RK, Joukamaa M. Physical and mental health changes in retirement age. Psychother Psychosom. 1991;55:100–107.
6. Szinovacz ME, Davey A. Retirement transitions and spouse disability: effects on depressive symptoms. J Gerontol B Psychol Sci Soc Sci. 2004;59:S333–S342.
7. Dave D, Rashad I, Spasojevic J. The effects of retirement on physical and mental health outcomes. South Econ J. 2008;75:497–523.
8. Gruber J, Wise D. Social security programs and retirement around the world: fiscal implications, introduction and summary. Cambridge, MA: NBER; 2005. NBER Working Paper Number 11290.
9. Gendell M. Retirement age declines again in 1990s. Monthly Labor Rev. 2001;124:12–21.
10. Gendell M. Trends in retirement age in four countries, 1965–95. Monthly Labor Rev. 1998;121:20–30.
11. Helman R, Copeland C, VanDerhei J. The 2009 retirement confidence survey: economy drives confidence to record lows; many looking to work longer. Washington, DC: Employee Benefit Research Institute; 2009. Issue Brief 328.
© 2011 Lippincott Williams & Wilkins, Inc.