NB20-05: The Effects and Channels of Early-Life Removal from the Disability Rolls: Evidence from SSI Children

Supplemental Security Income (SSI) provides cash welfare to the families of 1.4 million low-income children with disabilities. Davies et al. (2009) find that children receiving SSI benefits have low employment rates in adulthood. Deshpande (2016b) finds that removing youth from the SSI program at age 18 results in large income losses in adulthood.
These findings of poor life outcomes among SSI youth have two potential interpretations. The first is that SSI is a well-targeted program: when these highly disadvantaged children are removed from the program, they do poorly in the labor market, as expected. An alternative explanation is that SSI itself contributes to poor outcomes. These explanations have opposite policy implications, yet research has thus far failed to distinguish between them.
I propose to distinguish between these two hypotheses—selection into SSI versus treatment effects of SSI—and study the channels through which these effects operate. To distinguish between these two hypotheses, I will study the effect of removing children from SSI earlier than age 18 on their adult earnings. If the first hypothesis is correct, then we would expect that children have the same or worse outcomes when removed sooner from the program. If the second hypothesis is correct, then we would expect that children have better long-term outcomes when removed at a younger age. I will use variation in continuing disability reviews for children between the ages of 3 and 12, originally used in Deshpande (2016a) to study effects on parental earnings, to study the effect of removal of young children on their long-term outcomes. After adjusting for observable differences, I will test whether children removed at younger ages have better or worse outcomes than the children in Deshpande (2016b) who were removed at age 18. I will also investigate the channels through which SSI could affect outcomes, including the income transfer and incentives created by program rules. I will study the effect of the 18-year-old’s removal in Deshpande (2016b) from SSI on the adult outcomes of siblings, since siblings who do not receive SSI benefits should be affected by the 18-year-old’s removal through a pure income effect. Specifically, I will do the following:
• Use variation from Deshpande (2016a) to estimate effects of removing children at a young age from SSI on adult earnings
• Develop a matching procedure in the Numident to identify siblings of children included in the SSR
• Use variation from Deshpande (2016b) to estimate effects of removing 18-year-olds on younger sibling earnings in adulthood