Does Knowledge about Social Security Affect Behavior?
Decisions about when to retire and claim Social Security benefits influence the economic well-being of older workers and their families for the rest of their lives. Yet getting these decisions right can be challenging, as it requires understanding the details of Social Security rules and making complex calculations that incorporate uncertainty and trade-offs over time.
An important policy question is whether there are relatively simple and inexpensive ways to provide information that can potentially improve retirement decision-making and increase well-being. This is the subject of a new working paper by Jeffrey Liebman and Erzo Luttmer, "Would People Behave Differently If They Better Understood Social Security? Evidence from a Field Experiment" (NBER Working Paper 17287). The authors conduct an intervention that provides information about Social Security to older workers and examine its effect on their retirement behavior.
The intervention designed by the authors has two components, an informational brochure mailed to participants' homes and an invitation to participate in a web-based tutorial. The brochure and tutorial both provide in-formation on longevity, the relationship between retirement age and standard of living (e.g., the fact that Social Security benefits rise if the worker delays claiming or stays in the work force longer), and the Social Security earnings test. The information is provided through vignettes featuring actual retirees as well as more standard text, figures, and statistics.
The authors use a sample of about 2,500 older workers, mostly aged 60 to 65, in their analysis, assigning half of them to receive the intervention, while the other half did not. One year after the intervention, participants are invited to take a follow-up survey to assess the effects of the intervention on their understanding of Social Security rules and on their retirement and claiming behavior.
The authors first examine the effect of the intervention on labor supply. They find that those who received the intervention were 4.2 percentage points more likely to be working for pay one year later (this compares to an overall participation rate of 74% in their sample). Interestingly, the effect is concentrated among women, who were 7.2 percentage points more likely to work if they received the intervention. The intervention also increased hours worked and earnings, again primarily for women. The authors do not detect an effect of their treatment on Social Security claiming behavior.
To better understand the results, the authors explore whether the intervention increased knowledge about different aspects of Social Security. They find that the intervention generally increased awareness of the benefits of working longer and delaying claiming. The treatment had a particularly strong effect on the perceived incentive for women to work more years, which is consistent with women's stronger labor supply response to the intervention. For men, the primary effect of the intervention on knowledge was to increase their awareness of the return to delayed claiming between ages 66 and 70. Overall, these findings suggest that one important pathway through which the intervention affected behavior was by changing respondents' perceptions of incentives.
To explain the gender differences in their findings, the authors note that while men's and women's knowledge of Social Security prior to the intervention was generally quite similar, women were much less likely to think that they got a "better deal" from Social Security if they worked longer. The authors speculate that many women believed that as secondary earners, their Social Security benefits would be determined on the basis of their husbands' earnings records. While this was true for most women in the past, about 70 percent of women claiming today receive benefits based on their own earnings record. As the authors note, "it is possible, therefore, that our intervention affected women by counteracting the notion that working women get no benefit on the margin from Social Security."
Finally, the authors explore the effect of the intervention on expected future behavior, since the time frame of their survey does not allow them to observe all participants' retirement and claiming decisions. They find that the intervention delayed the average expected claiming age by about 4 months for men, consistent with the finding that the intervention raised men's awareness of the gains from delaying claiming between ages 66 and 70.
The authors conclude that their experiment "demonstrates that a relatively mild informational intervention can have important impacts on the labor force participation of older individuals." While their results suggest that having better information about Social Security program rules was at least part of the reason for the labor supply response, they caution that further information is needed about the mechanisms through which such interventions affect behavior. For example, if the primary effect of interventions is to educate workers and allow them to make better choices, then interventions will generally improve well-being. By contrast, if the intervention's effect came from delivering a general message such as "working to older ages is better," it is less certain that the intervention will improve welfare, as this message might or might not constitute good advice for any individual worker.
The authors acknowledge financial support from the Social Security Administration through grant #10-M-98363-1-01 to the NBER as part of the SSA Retirement Research Consortium.