Accounting for Geographic Variation in DI and SSI Participation

DRC Working Paper Number 2016-03
Publisher: Washington, DC: Center for Studying Disability Policy, Mathematica Policy Research
Mar 31, 2016
Authors
Jack Gettens, Pei-Pei Lei, and Alexis Henry

Background. There is wide geographic variation in DI/SSI participation among the working-age population. The reasons for the variation are not known. The geographic variation garners interest because the variation may possibly identify factors that affect DI/SSI participation that are not apparent from studies of individual-level data. The variation is comprised of two components, variation in disability prevalence (disability component) and variation in DI/SSI participation among persons with disabilities (participation component).

Objective. To account for the geographic variation in DI/SSI participation.

Methods. We use SSA administrative data and American Community Survey data to estimate the geographic variation in DI/SSI participation and the variation in the components. Descriptive statistics and thematic maps are used to describe the variation. We decompose the variance in DI/SSI participation into the two component variances. We use regression methods to examine the association between the participation component and area-level socioeconomic characteristics using exogenous predictors. Principal components analysis is used to decompose the variance of the participation component into the variance contributions of area-level characteristics.

Results. The variances of DI/SSI participation are greater than the sum of the component parts because of correlated components. At the state level, the disability component is 54% and the participation component is 21% of total variance in DI participation; and for SSI, the disability component is 35% and the participation component is 47% of total variance. The sub-state level results are consistent with the state-level results. Variance in the DI participation component is decomposed into the variance contributions of exogenous area-level characteristics as follows: demographics (17%), labor market (15%), disability types (7%), other (11%), and unaccounted (50%). Variance in the SSI participation component is decomposed as follows: demographics (20%), public assistance participation (10%), income (8%), labor market (9%), disability types (6%), other (9%), and unaccounted (38%).

Conclusions. Approximately 90% of the geographic variation in DI/SSI participation can be accounted for by the wide geographic variations in disability prevalence and socioeconomic characteristics. The accounting is different for DI compared to SSI. More of the variation in DI participation is accounted for by variation in disability prevalence and less by socioeconomic characteristics compared to SSI. Compared to DI, variation in the characteristics associated with economically disadvantaged areas accounts for more of the variation in SSI participation.