Housing Consumption in Late Life: The Role of Income, Health Shocks, and Marital Shocks

Past research has shown that income from the Social Security program has contributed to trends towards smaller households, greater residential independence—the tendency to live alone rather than with others—and a greater prevalence of home ownership late in life. However, the mechanism through which these associations operate has remained relatively unstudied. This paper addresses the possibility that Social Security income mediates the consequences of adverse events, whether “health shocks” such as a stroke or hip fracture, or “marital shocks,” principally the death of a spouse. We measure housing consumption using the ratio of dwelling unit size (number of rooms) to number of household members—i.e., “rooms per person.” We use panel data from the Health and Retirement Survey to model current housing consumption in relation to Social Security income as well as the occurrence of health and marital shocks over the two-year period preceding each survey year (1995; 2000; 2002; 2004; and 2006). We also use an instrumental-variables approach to deal with omitted-variables bias in the housing consumption equation, and include an additional control for selective loss from the sample due to entry into a nursing home. We find no effects of Social Security income on housing consumption once we control for selection and endogeneity, a result that contrasts sharply with past research findings. We also fail to find any evidence that Social Security mediates either health or marital shocks. However, we do find that health shocks, and to a lesser extent marital shocks, are strongly associated with both nursing home entry and changes in housing consumption.