The number of people receiving Social Security Disability Insurance (DI) benefits has doubled over the last two decades, and there is strong interest in promoting employment among DI beneficiaries. However, there remain important gaps in knowledge about how employment rates among DI beneficiaries relate to individual and environmental factors. We follow a sample of working-age Social Security Disability Insurance (DI) program beneficiaries for five years after their first benefit award to learn how certain factors help or hinder return-to-work outcomes.

We measure the return-to-work outcomes as the achievement of four important milestones: first enrollment for employment services provided by a state vocational rehabilitation agency or employment network, start of trial work period (TWP), completion of TWP, and suspension or termination of benefits because of work. We use linked administrative data from the Social Security Administration and Rehabilitation Services Administration for the period covering 1996 to 2009. We use a linear probability model to estimate the relationship of achieving the return-to-work milestones with age at award, impairment type, other beneficiary characteristics, and national and state-level economic conditions at the time of award.

We find that younger beneficiaries are more likely than older beneficiaries to achieve the return-to-work milestones within five years after award and that the likelihood of achieving the milestones varies substantially across impairment types. In addition, the probability of achieving the milestones is increased by having a greater number of years of education or being black and if there is lower state unemployment at the time of award. The probability of achieving the milestones is reduced by having a higher DI benefit amount at award, an award decision made at a higher adjudicative level, and if the beneficiary is receiving Supplemental Security Income or Medicare benefits at the time of DI award.

Finally, we find large variation in the relationships between state of residence and return-to-work outcomes and between award month (January 1996 to December 2004) and return-to-work outcomes, even when accounting for observed beneficiary characteristics and state unemployment rates. We attribute these variations to unobserved factors at the state level, policy changes over time, and trends in the beneficiaries’ unobserved ability to work.