Back to Annual Meeting Page		American Public Health Association 133rd Annual Meeting & Exposition December 10-14, 2005 Philadelphia, PA

Abstract #100129

Cristian Meghea, PhD, Research Dept, American College of Radiology, 1891 Preston White Drive, Reston, VA 22201, 703-648-8983, cristianm@acr.org

Motivation: The research literature had difficulties establishing causality in the relationship between income and health since income and health are jointly determined. It is critical to understand the causal relationship between elderly income and mortality, especially in the light of the dramatic demographic changes to take place in the near future, and at a time when reform of the Social Security program seems imminent.

Objective: This study uses a new natural experiment to address the causal effect of income on the mortality of retired elderly. I exploit an exogenous shift in benefit generosity for divorced retired women to estimate the impact of income on their mortality.

Data and Methods: Data comes from the New Beneficiary Data System (NBDS), a survey developed by the Social Security Administration. The first wave was administered in 1982 yielding a sample of 18,599 new Social Security beneficiaries. There was a follow-up interview conducted in 1991. Matched data from the SSA administrative records complete the NBDS. Under current law, a divorced spouse can claim Social Security spousal benefits on her ex-husband's Social Security record if their marriage lasted at least ten years. In addition, if an ex-spouse dies, a woman becomes entitled to widow benefits, roughly double the amount of spousal benefits. I use instrumental variable and treatment effect techniques to investigate the effect of income on the mortality of the elderly.

Results: To identify the causal effect of income on elderly ten-year mortality, I restrict the sample to divorced women. Instrumental variable estimation shows no effect of income on mortality. For a more precise identification I further restrict the sample to only divorced women receiving benefits based on the ex-husband working record. A treatment effect analysis on the comparison group of divorced women whose ex-husband is alive and the higher-benefit treatment group of divorced women whose ex-husband is deceased confirms the lack of an effect of income on the mortality of elderly divorced women.

Conclusions: Previous literature finds that socioeconomic status is a major determinant of health, but that seems to be not true late in life. A better socioeconomic status at an early age may lead to better health outcomes and a longer life, and that is when policies that enhance socioeconomic status are the most effective in improving health. However, if the policy interventions come late in life, the improvement in socioeconomic status is likely to have no effect on health and mortality.

Learning Objectives: