How anticipating benefit loss shapes family choices

In this post, J-PAL staff sit down with Manasi and Rebecca to discuss the results from their randomized evaluation on the effects of anticipating social safety net benefits on family choices, forthcoming in the American Economic Review. These results have implications for the impact of social safety net benefits on childrens’ long-term outcomes.

Tell me about your study. What bodies of evidence did the team hope to contribute to?

Supplemental Security Income (SSI) is a social safety net program that provides monthly cash payments to low-income children and adults. It is a crucial income source for families with a child with a disability, often comprising about half of total household income. Many parents are unaware that about forty percent of all children, and nearly seventy percent of those with behavioral conditions like ADHD, lose their SSI benefits when they turn 18 due to stricter eligibility requirements for adults. In our randomized evaluation, we provided half of the roughly 6,000 parents of 14 to17-year-old children receiving SSI in our study with information on the likelihood that their child will lose SSI benefits at age 18. We wanted to understand whether this information would affect parents’ beliefs about expected income and how this new information would shape investment in children’s education and job preparation—often referred to collectively in economics as “human capital” investments.

Our study aimed to contribute to a few main bodies of evidence. First, we wanted to contribute to the evidence base on how the social safety net affects behavior. Past research has explored the effect of the social safety net on human capital, but primarily focused on how receiving a government benefit today affects investments today. 

Instead, we aimed to understand the “anticipatory” effect of government benefits on human capital investments. What is the effect of parents anticipating that their child will (or will not) receive SSI income in the future (in adulthood) on investments made in children today (in childhood)?

We also wanted to contribute to the body of evidence about which interventions improve life outcomes for children receiving SSI benefits. Specifically, we wanted to understand if youth who lose SSI benefits in adulthood would have better adult outcomes if their families anticipated their loss of benefits.

What challenges did you face in implementing the intervention? 

We ran pilots of our intervention so it could be effectively implemented. At first, the intervention was implemented via mailings to families, and surveys were sent out over multiple months. In the final intervention, we delivered information to parents through videos and with both baseline and endline surveys administered in a single session.

J-PAL North America’s Research Management Support (RMS) was very helpful as we were getting ready to launch. We received feedback on our research design and help connecting with researchers who had undertaken similar projects who could give advice on specific aspects of the design. The feedback and connections from RMS saved us time and money when piloting our design, improved the design of our study, and helped us anticipate questions that journal reviewers might have.

What were the results from this study? In what ways did the research design contribute to ensuring that results were informative?

We found that the information we provided affected parents’ beliefs. Parents who received the information increased their perceived likelihood that their child would lose SSI benefits by 20 percentage points, expressed greater demand for a hypothetical insurance product to insure them against the loss of SSI benefits, and were 9 percentage points more likely to make plans to work more themselves in the future if already employed. 

Despite these strong changes in beliefs due to the information, parents did not invest more in opportunities that could increase their child’s future earnings. Specifically, parents who received the information were no more likely to take-up resources such as tutoring and job training that we offered through the experiment. After seeing that our intervention had no impact on parents’ investments, we decided to field a second survey to better understand what factors were driving these null results. 

We found three main explanations for parents' behavior regarding SSI benefits and investments in their children. First, many parents plan to compensate for lost SSI income by increasing their own work. The data showed that parents who received the information and were already working both planned to work more and actually did earn more after the study compared to parents who did not receive the information. Second, parents didn’t solely base their decisions on financial returns; many prioritized helping their child "realize their potential" over financial goals. For example, they may want a certain level of educational achievement for their child because they think education is important in its own right or confers status. Third, many parents may have already reached the limit of the investments they can make given time, money, and bandwidth constraints. The vast majority of parents reported that they are already doing “all they can do” to help their child succeed.

Where do you see this work going in the future?

We think our findings are encouraging news for the social safety net. If anticipating government benefits in adulthood had in fact reduced human capital investment in childhood, this would imply that SSI benefits are less helpful to children than previously believed because parents invest less in their children due to receiving benefits. Instead, our findings suggest that this is not happening. Even when parents anticipate losing benefits, they do not change investment choices in the present, which suggests that social safety net programs are not having unintended harmful effects on children.

That said, one disappointing aspect of our findings from a policy perspective is that providing information about the availability of SSI in adulthood is unlikely to improve the outcomes of children receiving SSI. Further research is needed to identify effective interventions for children receiving SSI. 

Our findings suggest several questions for future research. First, households that receive SSI are more likely to face constraints on investment than the average household. While many social safety net programs serve families facing serious resource constraints, it would be useful to study similar questions in other contexts to determine whether families that are less constrained have stronger anticipatory effects. Additionally, our study evaluated the effect of reduced expectations about future government benefits; it is an open question as to whether increasing expectations about the availability of future government benefits would have symmetric effects. Lastly, future research can continue to study why changing expectations about future benefits does not affect parents’ investment in their children’s human capital, but does change parents’ own work effort.