The Impact of Retirement on Health Behaviours

Abstract

The objective of the analysis is to identify the causal effects of being retired for at least one year on health behaviours. To estimate the effect of retirement, biannual data from the 2004 to 2014 Health and Retirement Study (HRS) is used. The endogeneity of retirement is acknowledged in our model by using an instrumental variable approach with the social security eligibility ages of 62 and 65 as instruments. Additionally, the longitudinal nature of the data gives us the ability to estimate an individual Fixed Effect (FE) model to control for unobservable time-invariant individual heterogeneity. The retirement parameter estimates of the FE-2SLS model show that being retired for at least one year is associated with a 26 percentage point increase in the likelihood of meeting the physical activity guidelines and a 17 percentage point increase in the likelihood of getting a flu shot, but a 9 percentage point decrease in the likelihood of getting a cholesterol blood test. Thus retirement is associated with some improvement in health behaviours.