Research & Analysis by Kajal Lahiri
A Structural Model of Social Security's Disability Determination Process
We estimate a multistage sequential logit model reflecting the structure of the disability determination process of the Social Security Administration (SSA), as implemented by state Disability Determination Services (DDS) agencies. The model is estimated using household survey information exactly matched to SSA records on disability adjudications from 1989 to 1993. Information on health, activity limitations, demographic traits, and work is taken from the 1990 Survey of Income and Program Participation. We also use information on occupational characteristics from the Directory of Occupational Titles, DDS workload pressure, and local area economic conditions from unpublished SSA sources. Under the program provisions, different criteria dictate the outcomes at different steps of the determination process. We find that without the multistage structural approach, the effects of many of the important health, disability, and vocational factors are not readily discernible. As a result, the split-sample predictions of overall allowance rates from the sequential model performed considerably better than the conventional approach based on a simple allowed/denied logit regression.
Modeling SSA's Sequential Disability Determination Process Using Matched SIPP Data
We model the Social Security Administration's (SSA's) disability determination process using household survey information exact matched to SSA administrative information on disability determinations. Survey information on health, activity limitations, demographic traits, and work are taken from the Survey of Income and Program Participation (SIPP). We estimate a multistage sequential logit model, reflecting the structure of the determination procedure used by State Disability Determination Services agencies. The findings suggest that the explanatory power of particular variables can be appropriately ascertained only if they are introduced at the relevant stage of the determination process. Hence, as might be expected by those familiar with the process, medical variables and activity limitations are major factors in the early stages of the process, while past work, age, and education play roles in later stages. The highly detailed administrative information on outcomes at each stage allows clarification of the roles of particular variables. Planned future work will include policy estimates, such as the number of persons in the general population eligible for the disability programs, as well as analysis of applications behavior in a household context.