Happy Together or Home Alone: A Structural Model of The Role of Health Insurance in Joint Retirement

by Dina Guo, University of Virginia

As the baby-boom generation approaches retirement age, most families facing retirement today are dual-worker couples who coordinate retirement choices. Numerous sources show that a significant portion of spouses retired within a year or two of each other. Although several papers have identified factors that affect household joint retirement decisions, the influence of health insurance has received little attention. The availability of retiree health insurance and its possible coverage of a spouse is a crucial factor that can influence the timing of couples’ retirement decisions. By reducing the risk of catastrophic medical expenditure, retiree health insurance can induce people both a covered worker and his spouse indirectly to retire, or it can keep the spouse of someone needing care at home in the labor to maintain coverage. This project uses a discrete choice dynamic programming model to answer two research questions: How do couples make joint retirement decisions? Through what channels does health insurance affect joint retirement decisions?

I set up a dynamic structural model of older couples’ decision-making given their health
insurance and retirement benefit options. Households choose through a cooperative bargaining process when each member will retire, how much the household saves, and which among the available health insurance plans to be covered by, giving uncertain medical expenditure. Unlike other papers, I allow households to jointly and endogenously choose total medical expenditure, given observed health, and I control for the initial condition associated with the observed eligibility for employer-provided health insurance (EPHI). Also, instead of considering each spouse’s health transition probability separately, I show that their health transitions are correlated and calculate joint health transition probabilities.

The primary source of data that I use to estimate my model is the Health and Retirement Study (HRS), which is a detailed panel survey of individuals over age 50 and their spouses. It collects extensive information about household characteristics, labor supply, health insurance coverage, health transitions, income, assets, pension plans and health care expenditures. I use reported decision-making power in the HRS to identify the distribution of bargaining power within a household. Another data set I use is the Medical Expenditure Panel Survey (MEPS), which provides up-to-date and precise information on household total medical expenditure and information on characteristics (e.g., deductible, premium, and coinsurance rate) of available EPHI plans, which cannot be observed in the HRS data. I use the MEPS to predict household
total medical expenditure and to impute missing information about characteristics of some health insurance options in the HRS.

I will use the dynamic structural model to analyze the effects of policies that change health insurance options, including raising the age of Medicare eligibility; the Affordable Care Act of 2010, which helps to make health insurance independent of employment status; and altering retirement benefits, such as the ongoing increase in the Social Security retirement age.

Back to 2012 Dissertation Fellowship Recipients