Why do couples retire when they do?

Abstract

This dissertation consists of three chapters studying the trends and determinants of observed retirement patterns in couples. The first chapter provides an overview of the existing literature on the retirement patterns of couples and discusses open questions and directions for future research. I divide my review into two parts – empirical and theoretical. The empirical studies (or, reduced-form) find that a significant fraction of couples, irrespective of their age differences, tend to retire within a year of each other in several countries, including the U.S., UK, Sweden, other European nations, and Australia. It also argues that economic variables alone cannot explain the observed joint retirement patterns observed in the data. Hence, it concludes,joint retirement must be due to unobservable preference parameters such as leisure complementarities. However, recent attempts to estimate leisure complementarity in the data show that the elasticity of substitution of leisure for husband and wives is greater than 1. This means that leisure is highly substitutable between the husband and the wife. The second chapter documents the labor supply and retirement patterns of couples in the U.S. across time and cohorts. As in the previous literature, I find that couples tend to retire together, irrespective of their age differences. I also show that the age gap between husband and wife is declining over time and across cohorts, which suggests that, while couples might still retire together in the future, they might do so simply because they both reach their retirement age together. I also estimate areduced-form model and show that spousal health status plays a significant role in determining the retirement behavior of couples, especially for women. The third chapter uses a rich structural model to evaluate the impact of spousal health shocks on the retirement decisions of couples. I explore the role of time lost in taking care of the spouse. To do so, I develop a rich dynamic life-cycle structural model of labor supply, earnings, savings, and Social Security for healthy and unhealthy couples. My model captures the structure of the Social Security retirement benefits, uncertainty in earnings, survival rates, wages, and health. I calibratethe model to match the observed life-cycle profiles of labor force participation for men and women by health status from the Health and Retirement Study data. The model matches the labor force participation profiles for men and women. In the model, 30.42% of couples retire within one year of each other. I find that setting the time lost due to bad health of spouse to zero reduces the coordination to 26.13%. The finding suggests that spousal health plays an important role in the retirement coordination of couples.