Families, Markets and Social Insurance

Published: 2000
Project ID: UM00-09


We have found modest effects of widowhood events on loss of health insurance. There are also modest effects of widowhood on labor supply, which we have not as yet attempted to attribute to insurance demand. Even new widowhood events, however, are not random with respect to initial conditions. Both initial health insurance status and risk of future widowhood are related to basic characteristics observed when married at baseline. When these confounding variables are controlled for in models of the effect of widowhood events on uninsurance, there is no longer statistical evidence of an independent effect of husband’s death on risk of losing insurance. Part of the reason why the measured independent effect of widowhood appears small is that there are events within marriage that can also affect insurance coverage, such as retirement or health events. Even though the number of uninsured women whose lack of coverage can be attributed to widowhood is therefore small, and not a distinct major policy motive for changes in age of eligibility for Medicare, uninsurance rates overall among the near elderly, and the potential public burden of cost-shifting from years just before 65 to years just after gaining Medicare coverage, suggest that Medicare eligibility policies should be a focus of continued research.