Sweden: Early retirement has positive health effects
The effect of earlier retirement on health outcomes is examined in a new report that looks at the retirement plans of different two age groups of Swedish army officers. The study finds that earlier retirement positively affected health: those who retired earlier had significantly fewer days of inpatient care and reduced mortality than those who retired later.
Demographic projections shows that the populations of OECD countries are growing older, resulting in an increased proportion of retirees per taxpayer. This imposes a burden on national budgets: consequently, measures to postpone retirement have been widely discussed and enacted.
One aspect of a prolonged working life that has not been as widely discussed are the health effects of postponed retirement. The study Early retirement and post retirement health from the Swedish Institute for Evaluation of Labour Market and Education Policy (IFAU) attempts to answer this question using data on the different retirement plans of two age cohorts of Swedish army officers. Due to the need to simultaneously cut costs and rejuvenate the officer corps, the Swedish defence bill of 1992 introduced measures to promote early retirement. Previously, officers had typically retired at age 60. Starting in 1992, however, officers aged 55 and older were offered leave of absence with full pay if they chose to retire early.
Methodology and data
To estimate the effect on health of early retirement, the study compared officers who were offered early retirement (those born in 1938–1939) with officers who had already retired by the time the defence bill was implemented (those born in 1931–1932).
Since there is a six-year gap between the group of officers affected by the bill and the control group, it is possible that there were other differences between the two cohorts that affected health outcomes (such as changes in health care technology or differences in upbringing – so-called 'confounding' factors). To account for that possibility, the study uses a statistical approach called 'difference-in-difference'. By comparing each age group to non-military (male) civil servants of the same age, it was possible to separate the effect of confounding factors and the early retirement plan under the assumption that there were no other trend breaks for officers relative to the civil servants in 1992 besides the change in retirement age.
Health outcomes were measured using micro data available from administrative registers maintained by Statistics Sweden. The data contain information on the number of days of in-patient care (nights spent in a hospital) for the entire relevant population (all male civil servants) in the period 1985–2010, as well as other demographic characteristics that were also controlled for (such as residence and income). The data also records incidence of death.
The analysis shows that the number of military employees entering retirement at age 55 increased from between 5% and 10% in the group born in 1931–1932 to over 60% for the group born in 1938–1939 when the new regulations were implemented. For non-military civil servants, there is no corresponding break between age cohorts, which seems to validate the construction of the study.
Using the difference-in-difference method outlined above, the study found that the number of days in inpatient care were reduced by 35% between the ages of 56 and 70 years for those who had been offered early retirement. Broken down into five-year periods, it appears that early retirement reduced the amount of inpatient care for those aged 56–60 years by two days; for those aged 61–70 years, it reduced the amount of care by 4.7 days.
If earlier retirement had a negative health effect, it could be that the fewer days of inpatient care were due to subjects dying earlier and hence not using the health care system as extensively as later retirees. By controlling for mortality, the study finds that this is not the case. On the contrary, the rate of survival is higher for the group of officers who retired earlier – about four percentage points higher at the end of the period (2010). This corresponds to a 26% reduced probability of dying before age 70. The effect is largest just after early retirement (before the age of 66). This implies that the study might in fact under-estimate the effect on days of inpatient care of early retirement.
Moreover, the data shows that the reduction in mortality is mainly due to a reduced risk of heart attack, heart disease and cancer. This is somewhat compensated for by an increased risk of dying from strokes or external causes for those who retired earlier; however, these latter effects remain smaller. In particular, the risk of heart attack is significantly reduced before age 65 in those who retired early.
The positive health effects were found to be relatively greater for those onlower incomes than for those on higher incomes. The same was true for education. This might be because those with lower incomes and and a lower level of education had more physically demanding jobs and were already more worn out (and hence more susceptible to further health damage) than those with higher incomes and level of education. There is another possibility: if reduced income in itself has negative health-effects, it could be that since lower-income workers suffered a smaller (absolute) loss of income after retiring, the positive health effects they experienced were larger than those of their high-income peers.
To summarise, the study found that retiring early improves health and reduces mortality and especially so in the years just following retirement. The effect was most pronounced in those with lower incomes and less education.
Other sources of data
The results of the IFAU study were in line with the most recent literature on the subject. Earlier research using simple cross-sectional analysis has generally found a negative health effect from earlier retirement; however, these results are of questionable value since retirees are likely to self-select into early or late retirement based on their health status. In recent years though there has been a growing literature that uses the same kind of quasi-experimental approach as the study reviewed. These studies typically exploit some kind of event not correlated to underlying personal characteristics but rather correlated to retirement as an instrument to estimate the health effects of early retirement. These studies – for example, Coe and Lindeboom (2008) and Bloemen et al (2013) – typically find a positive effect on health from retirement. The main contribution of the IFAU study study is that instead of using self-reported health, the study uses an objective, quantitative measure of health in the form of days of inpatient care and mortality.
The results are interesting from a policy perspective. In April 2013, the inquiry on retirement age commissioned by the Swedish government proposed that the age of retirement should be successively raised from 65 years today to 68 in 2034. Moreover, the final report stated that there is no clear evidence that early retirement has a positive health effect and that while a substantial body of literature suggests that retirement and self-reported health correlates 'these correlations do not say a lot about causality' (in Swedish, 4.7 MB PDF, p. 138).
In light of this, the reviewed study should be of great interest for policymakers since here there is indicated a clear correlation between health and earlier retirement that is not self-reported. To the degree that the health outcomes of officers can be extrapolated to the larger working population, the results of the study strongly suggest that earlier retirement results in better health; it does not only cause early retirees to claim better health to (for example) vindicate their decision to retire earlier.
If health is affected negatively by later retirement and these negative health outcomes constitute an economic burden on national finances, there is a risk that the positive economic effects of a longer working life will be overestimated by policymakers while the overall social impacts – including on quality of life and longevity – are underestimated.
Bloemen, H., Hochguertel, S. and Zweerink, J. (2013), The causal effect of retirement on mortality: Evidence from targeted incentives to retire early, Tinbergen Institute Discussion Paper 13-119/V.
Coe, N.B. and Lindeboom, M. (2008), Does retirement kill you? Evidence from early retirement windows, CentER Discussion Paper Series No. 2008–93.