Panel Paper: Refund to Savings: Exploring the Intersection of Behavioral Economics and Asset Building At Tax Time

Friday, November 9, 2012 : 1:40 PM
Schaefer (Sheraton Baltimore City Center Hotel)

*Names in bold indicate Presenter

Michal Grinstein-Weiss1, Dan Ariely2 and Clinton Key1, (1)University of North Carolina, (2)Duke University


Federal policies have long incented savings, but few low-income families benefit from these incentives.  At the same time, a growing body of research has shown that low-income individuals desire to save, and can manage to do so if given the right tools, incentives, opportunities and infrastructure. Many projects have tested approaches to encouraging savings among low-income individuals, often focused on tax time when many people receive a large lump sum refund, and using some kind of savings match.

While demonstrating a promising opportunity for saving at the “golden moment” of tax time, this work also reveals several key gaps: 1) Few programs are easily scalable; 2) the value of each program feature (in particular the match) is not well understood vis a vis other factors; 3) there has been little experimental evidence; and 4) it is increasingly clear that “one size does not fit all,” and that interventions are more likely to be successful when tailored to attributes of the participants.

To address these gaps and build on the prior research, the UNC Social Work team partnered with Dan Ariely of Duke University and Intuit, the makers of Turbo Tax, in 2009 to launch a multi-phase randomized controlled trial of a scalable intervention using a customized blend of prompts, incentives, and opportunities to encourage savings among lower-income households. In the first phase (2011 and 2012), to better understand the motivations underlying tax time-saving, the R2S team fielded a survey among TurboTax users as they filed their taxes (results from the 2011 and 2012 surveys will be available at the time of the conference). The surveys, integrated as a module in Intuit’s customer feedback survey, investigated the influence of behaviorally informed prompts on how people planned to allocate and use their refunds. Respondents were randomly assigned to receive one of 9 prompts designed to incent saving or the unprompted control condition. They were then asked about the allocation of their refund to different types of saving, debt clearing, and spending. Survey results show how different prompts affect the allocation choices and provide increased savings and debt clearing among different individuals.

In 2012, the best performing prompts from the 2011 survey were integrated into a randomized control experiment in the TurboTax Freedom Free-File product and were presented to about 80,000 lower-income tax filers during the tax filing season. These prompts together with variations in default saving amounts (anchoring) will be tested for their impact on savings behavior.

The findings from the survey and the experiment will be used in developing subsequent phases, including a larger rollout of the experiment in 2013, as well as integrating new prompts and additional savings products and mechanisms in TurboTax. Moreover, these findings will have wide implications for the whole array of programs seeking to enhance savings for low-income households.