Long-term Fiscal Spillovers from Social Security Income Shocks

Jacob Berman, University of California, Los Angeles

Social Security is the largest social insurance programs in the United States. It affects nearly every American over 62 and its spending will soon exceed 5 percent of GDP. Given the size and complexity of the program, developing models that forecast the budgetary cost of policy reforms has been a challenge. One difficultly is modeling how changes to Social Security benefit rules indirectly affect expenditures in other federal programs. In particular, any causal relationship between income and health would imply that changes in Social Security benefits could indirectly affect Medicare spending. If higher income improves underlying health, then Medicare spending may decline as healthier beneficiaries reduce their demand for health care services. Conversely, if higher income reduces beneficiaries’ sensitivity to cost-sharing, they may demand more services and overall spending will increase. Finally, if Social Security income impacts mortality, then short-term changes to benefits may have long-term demographic consequences. Given the level of Social Security expenditures, even modest spillovers could generate large fiscal externalities.

This paper attempts to estimate the direction and size of these spillovers. Specifically, I measure the causal effect of Social Security income on Medicare spending and life expectancy. To estimate this effect, I exploit a feature of the Social Security benefit formula which creates discontinuous changes in benefit levels that vary by exact date of birth. These shocks differ in sign and magnitude for persons born near January 2 of specific years. Using a regression discontinuity design, I examine payments in Medicare administrative data for covered health services. I start by analyzing aggregate spending for all Part A and B services, and then decompose the effect within utilization categories. I also explore whether benefit shocks lead to changes in diagnoses for various chronic conditions. Finally, I test for changes in mortality over a five-year period. My preliminary results suggest that increases in Social Security income are offset by declines in long-term Medicare spending.