This paper examined health, demographic, and disability trends over four birth-year cohorts using data from the Health and Retirement Study (HRS) and restricted linked SSA data with the goal of understanding current and future Social Security Disability Insurance (SSDI) prevalence. The analysis included (1) identifying physical and mental health responses in the HRS most predictive of SSDI receipt and how these responses and the likelihood of SSDI receipt have changed over cohorts; and (2) a decomposition to determine what share of the changes in SSDI receipt can be attributed to differences in the effects of health and other factors over cohorts, while accounting for the changing selection into program coverage. Key aspects to be addressed in future work include analysis of the younger adult population and modeling application timing and behavior, as this work focused instead on ultimate SSDI application approval among older birth-year cohorts.
This paper found that:
- For both men and women, several health factors (e.g., back pain and heart disease) and non-health measures (e.g., share never married) that are associated with SSDI receipt have become more common.
- There has been an increase in SSDI that is beyond what changes in health and demographics alone would predict in the population studied here. For men, only about half of the increase can be explained by changes in these measures; among women the increase explained by changes in these measures is only about 35 percent.
- Because factors associated with non-coverage are also associated with receiving SSDI conditional on being covered, results are sensitive to how selection into the SSDI program coverage is modeled.
The policy implications of these findings are:
- While the SSDI incidence and prevalence rates have begun to decline in recent years, this does not seem to be due to improved health, and the trend seems unlikely to continue.
- While non-health factors matter overall more for women, within these non-health factors, job characteristics mattered somewhat more for men in the sample. Changes to the determination process related to the vocational grid may, then, have more of an impact on men than women.