Association for Public Policy Analysis & Management

Prize-linked savings are an innovation for savings for low-income families that take advantage of the motives underlying the popularity of lotteries and other forms of gambling. Gambling, and especially lottery purchasing, is popular among lower income households (Kearney 2005). The attractiveness of combining lotteries and saving is even more directly suggested by the success of lottery-linked savings vehicles, although most of the evidence on their success comes from other countries, given legal restrictions on the offering of such accounts in the U.S. until the “American Savings Promotion Act” was passed in December 2014 (Public Law 113-251).

Perhaps the best known of such lottery-linked savings products are the “premium bonds” sold by the British government (Kearney et al. 2011). Some of the best evidence on lottery-linked savings accounts comes from a large retail bank in South Africa from 2005 to 2008. Besides robust effects on overall savings amounts in the raffle accounts, the program was associated with overall greater regular savings (Cole, Iverson, and Tufano 2014). One of the few savings lotteries run in the U.S. is the save-to-win program at credit unions in Nebraska from 2012 to 2013, which was found to reduce other forms of gambling, such as casino gambling, by up to fifteen percent relative to untreated counties (Cookson 2014).  

In a series of field experiments we tested whether saving and retention rates in a federally funded, matched savings program for low-income families – the Individual Development Account (IDA) program – can be improved through the introduction of program features inspired by behavioral economics, such as prize-linked savings. We partnered with eight IDA programs across the U.S. who agreed to randomly assign participants to different experimental conditions. Data were collected from 2008 to 2013 at the 48 implementation sites in California, Connecticut, D.C., Kentucky, Michigan, Missouri, Oregon, and Texas.

Eight hundred and ninety five program participants were enrolled in studies that tested the impact of four changes in key program features, including the introduction of a lottery-based incentive structure, whereby match rates were determined in part by a lottery at the time of each deposit. The experiments were designed to be revenue-neutral so that no additional match funds were required, and to keep administrative costs low.  

Overall, our results are disappointing. None of the interventions, not even the savings lottery, significantly increased savings, whether we examine final savings positions or progress over time. Study participants’ reaction to the experimental features from phone interviews suggest, for instance, that reminder calls to make lottery deposits were viewed as a nuisance, were distracting and hard to manage in the context of everyday life demands.

The scarcity of financial resources of program families and the remoteness of the savings goal for such families may help to explain the failure of our interventions to increase savings.  Household incomes in the study are close to the federal poverty threshold. Our interventions may have failed to improve saving because liquidity constraints, rather than cognitive biases were the primary impediment to saving.