Effect of Retirement on Health and Mortality

Abstract

Retirement involves a reduction in work hours, and is usually associated with a reduction in household income, and both these factors can affect health. This dissertation comprises of three studies that, together, explore the (1) effect of own and spousal statutory pension eligibility on health outcomes; (2) effect of retirement and change in household income on self-reported and objectives measures of general health, mental health, and functional health; and (3) effect of retirement and change in household income on mortality. The first study analyzes the effect of pension eligibility on health using data from the Health and Retirement Study (HRS) and the Survey on Health, Ageing and Retirement in Europe (SHARE). It finds that own pension eligibility is associated with better health for both sexes. The results also suggest intra-household spillovers as spousal pension eligibility reduces the odds of poor health outcomes for men and raises the odds of depression for women. The second study estimates the effect of retirement and reduction in household income on health, controlling for the potential reverse causality from health, using pension eligibility and payment amounts as instruments that affect both retirement decisions and income in retirement. It uses longitudinal data on men and women from HRS and SHARE on retirement, income, and health, combined with data on statutory pension eligibility and payment amounts in each country. It finds that both being retired, and higher household income, are significantly associated with better health outcomes, such as lower risks of poor self-rated health, depression, or difficulty with activities of daily living. Overall, the results suggest that taken together these two effects combine to give a small improvement in health on average when people stop working and replace labor income with a lower pension. The third study analyzes the impact of retirement and household income on the mortality of men in the United States using data from the HRS. It undertakes a survival analysis using a control function approach to allow for the fact that retirement and income may reflect unobserved health conditions that affect mortality. It identifies the model using early, and normal, statutory pension eligibility as instruments that affect retirement decisions and household income but do not directly affect mortality. The study finds that retirement and higher household income both substantially lower the hazard of mortality. Pension eligibility increases the likelihood of retirement, but reduces household income, but we estimate the net effect through both mechanisms to be a small reduction in mortality hazard.