Pension Policy Literacy and Retirement Expectations: A Cross-Country Survey Experiment.

Abstract

This study reports the findings of the first cross-national survey experiment on the effects of information on the expected retirement age. Given the drawbacks of unrealistic retirement expectations, the study examines the impacts of nonpartisan information about future demographic aging and forecasted pension benefit levels. An online survey experiment was conducted in the United States, Germany, and Spain in 2018 using an internet access panel. We assigned respondents to 2 random treatments: one citing the change in the projected share of the population older than 65 years (demographic treatment) and another citing the projected change in pension replacement rates (benefits treatment), both for 2015-2040. Treatment effects on the expected retirement age are reported. The benefits treatment has a strong influence on retirement expectations. In the United States, respondents informed of the expected decline in pension replacement rates expect to retire 2 years later than respondents not informed of the decline. In Spain, this treatment leads to an approximately 9-month postponement of expected retirement, while no significant effect is found in Germany. In addition, the demographic treatment does not affect retirement expectations in the countries studied. Respondents in all countries informed of future population aging do not show different expected retirement ages than respondents not given this information. People's retirement expectations are sensitive to information on future changes in pension generosity but not to information on population aging. The results suggest information campaigns focused on declining pension replacement rates may help extend working lives.