Effectiveness of the Ticket to Work program in supporting employment among adults with disabilities

Abstract

Background: Returning to work benefits many people with disabilities, as it supports personal financial independence and provides opportunities for greater societal contributions. The U.S. Social Security Administration's Ticket to Work (TTW) program offers expanded support services to help disability beneficiaries achieve financial independence through gainful employment. SSA has continuously sought new ways to identify those who would most benefit from using a Ticket.

Objective: To identify factors contributing to TTW participation and assess its impact on benefits forgone for work.

Methods: We conducted a cohort study using SSA administrative data to predict TTW participation and its impact on benefit cessation. The study sample included beneficiaries with a physical or mental residual functional capacity assessment from 2016. We applied a frequentist propensity score matching estimate and a doubly robust Bayesian hierarchical model-based estimate.

Results: The study included 172,640 beneficiaries (52.7% male, average age 52 years) with a range of qualifying conditions: musculoskeletal disorders (45.04%), mental disorders (29.10%), neurological disorders (9.82%). Both analytic methods yielded consistent results, showing that TTW participation is effective even after controlling for confounding factors. Personal characteristics (e.g., sex, age, education, race/ethnicity), health and functional status (e.g., work cessation due to health issues, need for alternate sitting arrangements, and limitations in understanding and memory) and environmental factors (e.g., region of residence) influenced Ticket participation.

Conclusions: Our findings offer valuable insights for identifying potential TTW participants and estimating benefit savings for SSDI/SSI programs. Future research should explore available support services and barriers to access to improve TTW outcomes.

Keywords: Bayesian Doubly Robust Estimation; Career support services; Causal inference; Disability; Propensity Score Matching; Return to work; Ticket to Work program.

Figure 1.

Distribution of functional limitations in each of four categories of mental function.

Figure 2

Frequentist logistic regression model for the instrument of ticket utilization. Shown are odds ratios and 95% confidence intervals for the top eighteen most-impactful predictors (ranked by absolute log odds ratios) by model.

Figure 3.

Bayesian logistic regression model coefficients (exponentiated into odds ratios) for ticket participation prediction. Mean and middle 95% credible region are shown.

Figure 4

Frequentist logistic regression model for the primary outcome of benefits cessation due to work. Shown are odds ratios and 95% confidence intervals for the top eighteen most-impactful predictors (ranked by absolute log odds ratios) by model.

Figure 5.

a) Bayesian logistic regression model coefficients (exponentiated into odds ratios) for predicting benefits cessation due to work. b) Bayesian doubly robust estimates of Ticket to Work average treatment effects conditional on SSI/SSDI status and Physical/Mental RFC presence. Mean and middle 95% credible region is given.