The effect of retirement on health.

Summary A pay-as-you-go pension system requires a balance between the population that contributes to benefits and the population that receives them. In most Western countries, the number of elderly people (beneficiaries) is increasing faster than the number of working people (contributors). This observation has prompted various governments to implement reforms aimed at increasing the retirement age in order to reverse this trend. The study of the effects of these reforms on health is of twofold interest from a public policy point of view. Indeed, the arduousness or the risks taken in certain professions may be the source of costs in terms of health, which may be accentuated by an extension of the length of the career. It therefore seems essential to measure the social repercussions of such reforms and to have a precise idea of their effects on health inequalities. These reforms, which increase the length of working life, could also have an impact on health care expenditure, with the transfer of costs from the pension branch to the health branch of the social security system, which would consequently attenuate the expected benefits of these reforms. Epidemiological studies reveal a negative correlation between retirement and health (being retired is associated with poorer health). However, it is difficult to deduce that retirement worsens health. Indeed, some individuals remain in employment longer precisely because their health status allows them to do so. In this review, we present a comparative analysis of the results obtained in the literature, focusing on the causal effect of retirement on health, and not the reverse1 (→). We thus distinguish three elements: retirement, i.e., the transition from employment to retirement . the increase in the retirement age . and the reduction in the retirement age2. We will thus examine the mechanisms that can explain the link between retirement and health and will detail the tools for identifying the causal effect. We will then present empirical (or observational) results from studies examining this causal link.(→) See T. Barnay Forum, m/s No. 10, October 2016, page 889.