Dynamic Models of the Social Security Disability Insurance Application

This dissertation develops dynamic structural models--option value and dynamic programming models of the Social Security Disability Insurance (SSDI) application decision. In the first chapter, we estimate the time to application following the onset of a health condition that first begins to affect the kind or amount of work that a currently employed person can do. We use the Health and Retirement Study (HRS) augmented by linked restricted access Social Security earnings record data. We test both the in-sample and out-of-sample predictive accuracy of our models using multinomial likelihood ratio tests based on discrete periods. Based on their predictive validity, we find that our option value model is superior to both our dynamic programming model and our reduced form hazard model. We also investigate the robustness of the results by only including individuals for whom restricted access Social Security earnings record data are available in our sample. These results are virtually the same as the ones for the full sample. In the second chapter, we use the preferred option value model estimates from the first chapter to simulate application timing under alternative SSDI policies. Three types of disability policy reforms are simulated: changing the benefit amounts, changing program eligibility stringency by changing the acceptance rates, and extending employer accommodation to all individuals with work limiting health conditions. Our simulations suggest all three of these policy changes would have substantial effects on lifetime application rates and on expected spell lengths until application for SSDI, and hence on SSDI caseloads.

RDA

Social Security Disability Insurance

12782