Weak Science on SSI

By Mark J. Warshawsky of AEI.

"The Social Security Administration (SSA) administers the Supplemental Security Income (SSI) welfare program, providing monthly financial benefits to low-income individuals with disabilities, including children. There is little evidence on the direct anticipatory effects of these transfers. If parents expect federal benefits later on in life for their children, will they reduce investments in their children now in anticipation of those future benefits?

In a recent volume of the American Economic Review—a prominent journal in economics—Professors Manasi Deshpande and Rebeccca Dizon-Ross of the University of Chicago published a massive and costly experimental study examining this question. They present specific information to random families of teenage children receiving benefits from SSI about the modeled likelihood of those benefits continuing into adulthood, and look at whether this information influences the decisions those families make about investing in their children. They find that “reducing parents’ expectations that children will receive benefits in adulthood does not increase investments in children’s human capital” and that “this zero effect is precisely estimated.”

But rather than measure actual policy-relevant outcomes of the children, like educational performance, graduation from high school, any college education, technical training, or earnings from work, Deshpande and Dizon-Ross measure four largely inconsequential action choices: (1) completing, with assistance, the intake forms for minimal job training services from state vocational rehabilitation agencies available to all disabled beneficiaries (but not actual attendance in the program), (2) signing up for online math lessons (but not actual attendance), (3) entering a lottery for $300 worth of one-on-one tutoring, and (4) the choice of a slightly reduced honorarium for participating in the study and also getting the book, “What Color Is Your Parachute? For Teens.” (unavailable in Spanish even though almost a fifth of study participants were Hispanic). Based on the fact that, comparing the intervention and control groups, families passed on these four almost trivial actions described above, Deshpande and Dizon-Ross conclude that SSI has minimal dynamic discouragement effects and that redistribution in general is therefore not costly. This conclusion is in contrast to several other studies in the literature with results of actual policy-relevant outcomes. In fact, the only substantive measure considered by the authors—parental earnings in the year after the information intervention—showed that an increase in the prospect of losing future SSI benefits for a disabled child, most of whom stay with their families into young adulthood, encouraged parents to increase their work effort.

In terms of mechanics, methodology, structure, and resources, the study is formidable. Yet on the level of common sense, it is lacking:

• In the baseline survey of parents, most report a probability of zero that their children would be removed from SSI upon turning 18, the age at which a redetermination by SSA is automatically conducted. This result cannot be taken at face value, as Deshpande and Dizon-Ross do. Beginning with the very first benefit payment given to a family, SSA communicates about a future redetermination at age 18, and in the actual experience of the study participants, three-quarters of children had medical reviews, some lost benefits because of them, and a plurality of parents themselves are disabled and presumably receiving disability benefits themselves, and are therefore subject to continuing disability reviews and periodic redeterminations. A more reasonable interpretation of these survey results is that these low-income parents are justifying to the government-sponsored surveyors their decision to apply for and to continue to receive welfare benefits for their children, most with behavioral issues.
• Basic survey readings indicate many parents did not find the four “action” offerings helpful, and many had concerns about their children’s disabilities rendering the offerings largely unhelpful.
• The response rate to participate in the study was less than 20 percent, opening up the possibility of selection bias.
• The intervention was conducted in Fall 2021 when there was an outpouring of government aid, and schools and families were still recovering from the harmful disruptions of the pandemic.
• The authors find that parents’ responses to the information intervention reduces their intention to send their children to college and interpret this to mean that the parents can no longer afford college. But for these poor families, scholarship and grant opportunities are widespread. Rather, families are likely making the rational calculation that with no future SSI disability benefits, which would allow their children to attend college but not to work, their children are better off to skip college and to start work sooner to gain experience and practical knowledge. This is another dynamic discouragement effect not recognized by the authors.

Given the expense of this elaborate study, it is still worthwhile for the authors or SSA researchers to complete it using real outcomes for children and parents in terms of employment and education and to see if the authors’ information intervention proved to be accurate after the fact."