Association for Public Policy Analysis & Management

Background and Purpose. Low- and moderate-income (LMI) households lack sufficient liquid assets to address unexpected emergencies and dips in income (McKernan, Ratcliffe, & Vinopal, 2009; Pew Charitable Trusts, 2015). Receiving tax refunds is an opportunity for recipients of the Earned Income Tax Credit (EITC) to build emergency savings to help cope with these financial shocks. However, prior research has shown that EITC recipients use their tax refunds for several purposes, such as reducing unsecured debt and making large purchases, and save only small portions of their refunds. The purpose of this study was to assess the effectiveness of savings messages among a large sample of EITC recipients who file their taxes online, and to assess the probability that EITC recipients would defer a portion of their refund for six months if offered savings bonuses like those in recent policy proposals (H.R. 4236, S.2797). We also determined whether these savings behaviors and preferences varied by the size of expected refunds.

Methods. Tax filers eligible for the EITC and who expected to receive a federal income tax refund during the 2015 (n = 270,891) and 2016 (n = 112,562) tax seasons were randomly assigned into one of several treatment groups that received a savings message and/or choice architecture manipulation delivered via online tax filing software, or a control group that did not receive a savings message. Treatment-control group differences in refund savings take-up and amounts were assessed in both years, and across refund quintiles among 2015 filers. Also, we used responses to three different, randomly assigned hypothetical scenarios to assess refund deferral and saving bonus preferences among respondents to a household financial survey delivered in 2016.

Results. In both 2015 and 2016, refund saving take-up was 3-4 percentage points higher among treatment compared to control group EITC recipients (p < .001; d=.09 to .14), while average amounts of refunds saved were $83 to $137 higher among treatment compared to control group EITC recipients (p < .001; d=.06 to .10). In 2015, these treatment-control group differences were table across refund quintiles. With respect to hypothetical deferral and bonus preferences, EITC recipients had 49% and 59% increased probabilities of deferring 20% of their refunds for six months when hypothetically offered 5% and 10% bonuses (p < .001), respectively, compared to an offer of no bonus. These probabilities rose with greater refund amounts.

Conclusions and Implications. Messages delivered in tax filing software that prompt filers to save using motivational messages and/or choice architecture manipulations represent an efficient and scalable strategy to encourage EITC recipients to build emergency savings. However, the overall refund saving take-up rate is low, consistent with prior research showing competing priorities for refund allocation. Nonetheless, refund saving take-up might increase considerably with the offer of modest savings bonuses, which are included in current policy proposals.