Do as I Say (because I’m Similar to You): Gender Similarity, Message Framing, and the Decision to Save for Retirement: An Abstract

Abstract

Social information about the behaviors of one’s peers typically has a positive impact on one’s propensity to save for retirement, yet a recent study by Beshears et al. (2015) suggests that peers’ savings behavior reduces savings rates. The framing of the savings message may also impact savings rates, but results are mixed in terms of whether descriptive (“what others do”) or injunctive (“what one should do”) messages are more effective (Bailey et al. 2004; Croy et al. 2010). The present study aims to reconcile these disparate findings while integrating the literature on peer effects and message framing in the context of retirement savings. We propose that when the message provider is similar to the recipient, interest in enrollment in a retirement savings plan will be greater when the message is descriptive. When the message provider is dissimilar to the recipient, injunctive messages are more effective. We conducted a series of three sequential experiments to test our hypotheses.

First, we evaluated our study design and measures with a pilot study using participants from Amazon Mechanical Turk. Next, we distributed a survey to public employees eligible for supplemental retirement savings. Results indicated that 55% of participants provided their email when felt similar to the message sender and the message was descriptive. When the message was injunctive and the message sender was similar, 73% of participants elected not to provide their email address, suggesting that similarity and descriptive messages are more persuasive. Interestingly, the interaction effect was only significant when the message sender and receiver were similar in terms of race, but not gender or age.

Finally, we partnered with the State of Oregon and the office of the Oregon Savings Growth Plan (OSGP) to conduct a widely distributed field study that ultimately allowed subjects to enroll in a supplemental savings plan, thus allowing us to measure real behavior. The preliminary results from this study support the hypothesis that descriptive messages, especially when paired with a message provider who is similar in gender to the participant, are more effective in building interest in supplemental savings. We found no evidence that ethnic similarity between the message provider and the study participant had an influence on interest in supplemental retirement savings.

References Available Upon Request