The predicted shortfall in the Social Security trust fund will require policy changes such as reduction in retirement benefits, an increase of the payroll tax rate and/or of the “wage ceiling” (the upper limit on SS taxable income). It is thus important to understand the potential behavioral impacts of such policies.
Though the actual elasticities for saving, labor force participation and retirement depend on many factors, one important element is how policy-changes affect the expectations of Social Security benefit receipt. To some extent, people are aware of the predicted shortfalls in the Social Security system and thus already expect to receive lower amounts than those prescribed under current rules. Therefore, changes in the rules need may have no effect on expectations. For example, a reduction in social security benefits could have no impact on the expectations of those who had already internalized such reduction.
Understanding how people’s expectations would change in response to policy levers would be a useful input into studies of the likely behavioral responses of these policy changes. For instance, Kapteyn, Prados & Yoong (ongoing, KPY from now on) are currently studying how people make decisions on wealth accumulation and decumulation over the life span and how this may deviate from an “optimal” path of wealth over the life span (in terms of an individual’s own goals and expectations). The results of these combined studies would allow tracing back the behavioral effects of policy changes on private savings through their effects on subjective expectations. Furthermore, understanding whether these changes affect subjective expectations by themselves would be worthy on its own. For instance, Luttmer and Samwick (2017) show that the uncertainty of Social Security in itself reduces welfare. Thus, an important question is whether policy changes would reduce such uncertainty.
We field a survey on a internet-based representative panel, as a follow-up to respondents of an earlier survey. We present respondents with a series of “policy scenarios” where the government enacts alternative policy changes (such as increase in the tax rates, reduction of the benefits, and other). The order of the policy scenarios randomly differs across respondents. In each of these scenarios we measure respondents’ expectations. We compare expectations across the scenarios (within and across respondents) to present an estimate of how alternative policies would affect expectations. In addition, we randomize respondents into one of three treatment arms, that alter the baseline levels of information that each respondent has. The purpose of this randomization to test whether the impact of the scenarios on benefit expectations differ by the level of information of the respondents. However, the randomization will also be used in a separate paper to test for the impacts of information.