Elsevier

Economics Letters

Volume 103, Issue 1, April 2009, Pages 8-11
Economics Letters

Retiring to the good life? The short-term effects of retirement on health

https://doi.org/10.1016/j.econlet.2009.01.015Get rights and content

Abstract

We estimate the impact of retirement on three subjective and two objective measures of health using a regression discontinuity design. The results indicate that retirement increases an individual's sense of well-being and their mental health, but not necessarily their physical health. Specification tests suggest that the results are robust.

Introduction

With continued growth in life expectancies and the aging of the baby boom cohort, many OECD countries are facing large increases in the proportion of the population that are eligible to receive a publicly provided retirement pension. In response, a number of policy suggestions have been made to encourage aging workers to postpone retirement, including increasing the official retirement age. For example, in England it has recently been proposed by the independent Pensions Commission that the state pension age increases from 65 to 68. However, the success of such policies in reducing total government expenditure will depend, in part, on the potential health impacts of delayed retirement. For example, if retirees feel less stress and are able to spend more time on healthy pursuits, then increasing the retirement age will likely lead to increased expenditure on health. On the other hand, if retirees' lives lack purpose and are more sedentary than the lives of non-retirees, then increasing the retirement age may decrease health expenditures.

Given the importance of understanding the relationship between retirement and health, surprisingly few studies attempt to estimate the relationship, and the few that do, find both positive and negative results (e.g., Mein et al., 2003, Dave et al., 2008). It is likely that the scarcity of studies and the contrasting results are due to the difficulties in identifying the causal impact of retirement on health. First, unobservable factors likely influence a person's retirement decision and their health. For example, differences in discount rates will affect both investment in health and attachment to the labour force. Moreover, unobservable factors will often vary over time. For example, the contraction of a serious illness by a person's spouse may contemporaneously impact their mental health and their retirement status. Second, numerous studies have shown that an individual's health status is an important determinant of retirement (e.g. McGarry, 2004). The presence of time-varying confounding factors and reverse causation implies that simple cross-sectional estimators and fixed-effect estimators are biased.

In this paper, we estimate the impact of retirement on health for a sample of English men. To identify the causal effect, we use a regression discontinuity design (RDD), which entails comparing the health of males that are aged around 65. This RDD identification approach is similar to the identification approach used by Charles (2004) in his study of the relationship between retirement and mental health for US males. In accordance with age specific retirement incentives in the US, Charles uses binary variables indicating whether the person is at least 62, 65, 70 or 72 years old as instrumental variables. His two-stage least squares (2SLS) estimates suggest that retirement improves the psychological well-being of males.

Section snippets

Data

Our data source is the Health Survey for England (HSE), an annual cross-sectional survey. The unit of analysis in the HSE is the household, with information collected from both adults and children living in England through a combination of face-to-face interviews, self-completion questionnaires and medical examinations. In this paper, we pool data from the 1997 to 2005 surveys and omit men that have a university degree. This restriction is made because degree-holders are much more likely than

Method

A RDD arises when treatment status depends upon an observable characteristic and there is a known value for this characteristic such that the probability of treatment changes discontinuously. To identify the causal impact of retirement on health, we use a RDD that is based on a discontinuity at age 65 in the probability of retirement. This method appears appropriate as the probability of retirement significantly increases at age 65 in England because of financial incentives created by state and

Results

The results are displayed in Fig. 1b to f and in Table 1. The figures display local linear regression estimates with 95% confidence intervals of the relationship between age and health for ages 60 to 64 and for ages 65 to 70. These plots graphically demonstrate any discontinuities in the relationship at age 65. The top panel in Table 1 displays the corresponding estimated LATEs and standard errors using bandwidth values of 1, 2 and 3 years.

Using a bandwidth value of 2 years, Fig. 1b, c and d

Conclusion

Using a regression discontinuity design, we determine that retirement has a significantly positive impact on three self-reported health measures, but no impact on two objective health measures. These findings are not overly sensitive to the bandwidth values used in the empirical analysis and specification tests indicate that the results are not driven by other factors. That retirement seems to have no impact on objective health suggests that health expenditures would not be greatly affected by

References (8)

There are more references available in the full text version of this article.

Cited by (103)

View all citing articles on Scopus

This research was based on the Health Survey for England 1997-2005, produced by the Joint Health Surveys Unit of Social and Community Planning and University College London, sponsored by the Department of Health, and supplied by the UK Data Archive. The data are Crown copyright.

View full text