The Society for Social Work and Research

2014 Annual Conference

January 15-19, 2014 I Grand Hyatt San Antonio I San Antonio, TX

Refund to Savings: Building Contingency Savings At Tax Time

Saturday, January 18, 2014: 5:00 PM
HBG Convention Center, Room 001A River Level (San Antonio, TX)
* noted as presenting author
Michal Grinstein-Weiss, PhD, Associate Professor and Associate Director, Washington University in Saint Louis, St. Louis, MO
Dan Ariely, PhD, James B. Duke Professor of Behavioral Economics, Duke University, Durham, NC
Clinton Key, MA, Research Manager, University of North Carolina at Chapel Hill, Chapel Hill, NC
Krista Holub, MA, Program Manager, Washington University in Saint Louis, St. Louis, MO
Background and Purpose:

Low levels of contingency savings is a challenge for millions of American families and is a big contributor to family stress, vulnerability, and instability. With currently available resources, about half of households are economically vulnerable. This vulnerability has implications for social work.

Research on small pilot programs to help families prepare for financial emergencies using the tax refund have shown promise. These programs are not scalable. The Refund to Saving (R2S) Initiative builds an experimental savings intervention into the widely used TurboTax platform to offer a scalable, testable opportunity for households to build economic security and prepare for contingencies.


During the 2013 tax-filing season, all TurboTax Freedom Edition users owed a federal tax refund (n~1,000,000) were randomly assigned. The intervention, built into the software, seeks to increase the rate at which people split their tax refund, channeling money directly into existing savings accounts or Series-I U.S. savings bonds and the amount saved.

The randomly assigned treatment conditions vary the language used, the suggested savings amount, and the presentation of the choice. For treatment, splitting the refund is displayed as the default, and a user must choose not to save.

Data are collected administratively on users’ interactions with the software, their tax characteristics, and their savings choices. These data are analyzed to determine each treatment’s relative efficacy in generating savings.

In addition, some users responded to a longitudinal survey on household finances. The baseline data from this survey, collected in a web-based survey immediately after tax filing, is used to describe the study population in detail. A second wave of data will be collected 6 months after tax filing to assess the effect of savings generated by the intervention.


Descriptive data from the first wave of the longitudinal study (n=21,030) show that TurboTax Freedom edition users are similar to the lower-income population generally. Over 60% of the population could not come up with $2,000 in 30 days for an emergency. Over 80% experienced material hardship in the past year, including over half who experienced food insecurity. Data on the pilot year of the intervention suggest that it has a significant positive effect on the rate and amount of saving (p<0.01). Those assigned to a treatment group are more likely to save part of their refund and save more. Among those who split their refund, those in the treatment groups saved between $200 (25% anchor groups) and $900 (75% anchor groups) more than those in the control group. Data from this year’s implementation (n~1,000,000) are being cleaned and will be ready for analysis in early May.

Conclusions and Implications:

The economic health of families is at the core of policy and social work practice. These findings show how contingency savings might be increased and families might be made more economically secure. The initiative also models a cost-effective, scalable approach to delivering needed services and possible public-private partnerships. Further research will explore the role of contingency savings generated by the program in households’ lives.