Work affects both the time available for non-market activities and the times at which those activities are performed – and therefore work-induced constraints on time use may influence retirement decisions. We analyze these effects by combining new data from the American Time Use Survey with information on retirement in the Health and Retirement Study.
We find that the propensity to engage in three types of non-work activities – household production, leisure, and tertiary activities (eating, sleeping, grooming) – are substantially altered by work. Moreover, the ways in which the timing of these activities are distorted differ across ten different job types (industry-occupation combinations) that we examine in the ATUS. We use the resulting measures of time distortions as control variables in multinomial logit retirement models that we estimate in the HRS. Older workers in jobs with greater distortions to the quantity and timing of leisure activities have an increased propensity to leave those jobs, either for new jobs or for retirement. On the other hand, workers in jobs with greater distortions to household production have a reduced propensity to leave their jobs, and distortions to tertiary activities raise the propensity to take new jobs but reduce the propensity to retire.