The broad goal of this proposed research is to understand how living arrangements evolve from retirement to death, how they depend on the initial socio-economic status of the elderly at the beginning of their retirement years, and how they might change under different economic and family conditions in the future. We will use five waves of AHEAD data to construct trajectories of living arrangements, health status and wealth, together with income, marital status, family links and labor force status. Based on these trajectories, we will estimate joint transition probabilities of moving between states defined by living arrangement, health status and wealth, taking account of observable covariates and persistent individual-level heterogeneity. We will use these transition probabilities to shed light on behavioral hypotheses and to simulate future living arrangements of the cohort reaching age 70 in the year 2000 under alternative assumptions about future Social Security and Medicare benefits, number and proximity of children, and female labor force participation and similar socio-economic conditions.
