Executive Summary

Electronic delivery of tax refunds through direct deposit to bank accounts is used by most upper-income families, but has so far eluded many low- and moderate-income families. One of the benefits of electronic delivery versus paper checks is lower administrative cost for the federal government—roughly one-tenth the cost of a paper check. Beyond this, electronic delivery provides tax filers faster, safer and more reliable access to tax refunds, and can improve low- and moderate-income tax filers’ access to mainstream financial services, a goal of Title XII of the Dodd-Frank Wall Street Reform and Consumer Protection Act.
Many low- and moderate-income working families receive a sizable federal tax refund, primarily through refundable tax credits such as the earned income tax credit (EITC). Many families opt to receive a check or turn to alternative financial services to access their refunds or cash their refund checks, which may undermine financial access and asset-building goals. At least 17 million U.S. adults are unbanked and about 43 million are underbanked (FDIC 2009);1 access to mainstream financial services would bring these adults federally insured accounts to build savings, cash checks, pay bills, and avoid expensive alternative financial services.
The U.S. Department of the Treasury’s pilot program delivered tax refunds electronically and increased access to the mainstream financial sector for a small sample of the low- and moderate-income population. Given the importance of properly designing an account for its target consumer, the Tax Time Account pilot was designed to evaluate card features and messaging for low- and moderate-income unbanked tax filers. The results of the pilot will help ensure that the potential effects of certain account design features are well understood before evaluating future options that could, for example, integrate an account option into the tax-filing and refund process.
Specifically, the pilot offered low- and moderate-income tax filers a safe, low-cost account for the electronic delivery of their federal tax refunds. The account can continue to be used for multiple purposes, including ongoing direct deposit of earnings and cash loading, point-of-sale transactions, safe storage of funds, ATM withdrawal, and bill payment. The U.S. Department of the Treasury selected Bonneville Bank as the financial agent for the Tax Time Account pilot. Bonneville Bank selected Green Dot to provide card processing services for the Visa branded card. The card may be used at any point-of-sale terminal that accepts the Visa card world-wide.
Electronic delivery of tax refund payments may have many benefits for the government and taxpayers. Tax refund payments are made once a year and many are for significant dollar amounts. However, simply converting a once-a-year check payment to a once-a-year plastic card may not
1 Unbanked adults have no checking or savings account. Underbanked adults have a checking or savings account but rely on alternative financial services such as (nonbank) check-cashing services, refund anticipation loans, payday loans, pawn shops, money orders, and rent-to-own agreements.
The FDIC study can be found at http://www.economicinclusion.gov/household_survey.html
ES-2
provide savings for the government. The expense of providing a taxpayer with a plastic card annually for this singular purpose would be greater than disbursing a payment via a paper check.
Instead, Treasury is exploring approaches that would provide low- and moderate-income tax filers the ability to choose a financial account that could be used for deposit of their tax refunds, and also function as an ongoing financial account for depositing other income, paying bills, making in-person and online purchases, and withdrawing cash, among other purposes. For two reasons, this approach could be more economical than a paper check—for the taxpayer as well as for the government. First, recipients would be more likely to retain a financial account that can be used in an ongoing way, making it possible to send them electronic refunds in future years. Second, the economics of an ongoing account appear to be more viable for financial providers, potentially making it possible to furnish the account at a low or no cost to consumers if done on a wide scale.
Under the pilot, roughly 800,000 individuals were randomly selected from a population of more than 8 million adults who were likely to be low- and moderate-income (under $35,000 in household income) and live in an unbanked or underbanked household.2 The 808,099 people were then randomly assigned to one of eight treatment groups. Pilot participants in each treatment group were mailed an offer to sign up for the MyAccountCard. The eight treatment groups differ along three dimensions: (1) no monthly fee versus low monthly fee ($4.95), (2) linked savings account versus no linked savings account, and (3) convenience-focused messaging versus safety-focused messaging. Based on these three categories, the eight treatment groups are as follows:
Group
Monthly Fee
Savings Account
Message
Treatment 1
Treatment 2
Treatment 3
Treatment 4
Treatment 5
Treatment 6
Treatment 7
Treatment 8
No fee
No fee
No fee
No fee
Fee
Fee
Fee
Fee
Yes
Yes
No
No
Yes
Yes
No
No
Safety
Convenience
Safety
Convenience
Safety
Convenience
Safety
Convenience
The pilot evaluation measures the impact of the different prepaid card offers on the MyAccountCard sign-up rate, subsequent card use, and accumulated account balances. For example, people assigned to receive a card with no monthly maintenance fee are compared with those who received a card offer with a monthly maintenance fee to gauge the impact of the fee on take-up and use. Similarly, people who received a card offer without a linked savings account (or safety messaging) are compared with those who are offered a savings account (or convenience messaging).
Pilot participants were on average 46 years old and almost evenly split between males/females and white/nonwhite participants. All participants had estimated annual household incomes below
2 These adults were identified using commercially available data from Experian Marketing Solutions Inc.
ES-3
$35,000, with roughly a third in each of three income categories (<$15,000, $15,000–$24,999, and $25,000–$34,999). The participants’ average underbanked score of 5 falls in the middle of the 1–9 range, where 1 indicates households most likely to be unbanked/underbanked and successively higher values indicate a lower likelihood of being unbanked/underbanked.3
Overall, 1,967 people (0.3 percent) who received a MyAccountCard offer applied for this prepaid card, of which 1,933 people (98.3 percent) were issued the card. This 0.3 percent take-up is in the 0.3 percent to 0.8 percent range of credit card direct mail take-up rates in recent years (Kiviat 2010; Woosley 2007). Although credit cards and prepaid cards are different products, the credit card take-up rates provide a benchmark for the take-up rate of card products offered through direct mail. Importantly, the card take-up rate was significantly higher for people who were most likely in unbanked households. Pilot participants identified as most likely to be unbanked had a take-up rate of 0.8 percent. This take-up rate, which is nearly three times higher than the take-up rate for the full pilot population, is at the higher end of the expected range and exceeds recent first quarter take-up rates. Only a subset of cardholders directly deposited their tax refunds into the card accounts. Sixteen percent of all cardholders and 48 percent of active cardholders did so. The analyses show that the timing of the card offer was also important for take-up—the findings indicate that earlier is better. People mailed the offer in mid-January were 85 percent more likely to apply for the card than those mailed the offer in early February. This provides helpful direction for any future efforts and suggests that information about the prepaid card program should be distributed and made available before the tax-filing season begins.
Some elements of the pilot likely lowered the take-up rate. These elements include timing of the offer letters, quality of the Experian mailing list (addresses and target characteristics), a multistep process for receiving one’s tax refund into the account, and absence of a comprehensive informational “surround sound” campaign.4 Also, the MyAccountCard cannot be used to pay tax preparation fees, which likely reduced its usefulness for low- and moderate-income tax filers that use paid tax preparers and who did not have the $150–$400 tax preparation fee upfront.5 While some of these items were apparent at the start of the pilot, they were accepted as part of its design, but would be rectified in any future pilot or program focused on enhancing take-up of a card product.
The pilot was designed to measure the effect of different aspects of card offers on take-up, not overall card take-up. Any future efforts would likely offer the card through a different mechanism—likely directly in the tax filing and refund process rather than through the direct mail offer used for
3 The household likelihood of being unbanked or underbanked is a variable constructed by Experian Marketing Solutions Inc. and is based on a statistical model that produces an underbanked score, where values between 1 and 9 identify households likely to be unbanked or underbanked.
4 An informational surround sound campaign is an effort to use earned and paid media, community partners, and population relevant organizations to make the public receiving the offer aware of its existence and the process by which to take action.
5 Fees for tax preparation services vary by provider. Average fees at H&R Block and Jackson Hewitt ranged between $185 and $210 in 2010 (Perez 2010), but have been found to reach more than $400 (Wu and Fox 2011).
ES-4
the pilot. The results of the pilot provide helpful data about the potential effects of certain card design features to inform future potential options, including those that would integrate card offers in the tax filing and refund process.
One of the tested features—card cost—stands out as influencing the behavior of pilot participants in a significant and consistent way. Charging a $4.95 monthly maintenance fee (versus no monthly maintenance fee) decreased MyAccountCard applications and issuance by 42 and 43 percent, respectively. The estimated own-price elasticity of demand implies that a 10 percent increase in the monthly cost of the MyAccountCard reduces card applications and issuance by 2.6 percent. Analyses of different subpopulations find that people in households likely to be unbanked are somewhat less sensitive to the price of the card.6 Unbanked households may be more likely to pay a monthly fee for the prepaid cards because they have fewer banking options.
Card use also decreased with the $4.95 monthly maintenance fee. The likelihood of using the card within the first six months of the pilot was 47 percent lower for people offered a card with the monthly fee. Measures of longer-term card use are also lower among those who face the monthly fee. People offered a card with the $4.95 monthly fee were 55 percent less likely to actively use their MyAccountCard six months after the pilot was launched and 52 percent less likely to directly deposit a tax refund into their accounts.
Outcomes for pilot participants offered and not offered the linked savings account are similar. Adults offered the linked savings account (versus not offered the linked savings account) were not significantly more likely to apply for or be issued a MyAccountCard, nor did they have higher account balances. The savings account feature tested in the pilot required additional cardholder action to activate the savings account (i.e., online activation) and deposits could not be made directly into the savings account (i.e., only transfers from the transaction account were allowed); different implementation could have produced different results. Product messaging (safety versus convenience) did not significantly influence pilot participant behavior.
The results of the Tax Time Account direct mail pilot offer lessons for moving forward with a similar account offer by Treasury. The pilot established proof of concept for offering a card account in conjunction with tax time. Offering and issuing cards, and delivering tax refunds to those accounts, proved feasible for Treasury operationally. In addition, the pilot established that there is a market and a demand for such a product. That direct mail take-up rates are, in absolute terms, low, but not appreciably different from similar direct mail efforts, serves primarily to underscore the importance of Treasury delivering the card with both different enrollment methods and in a different marketing context at scale. Streamlining the card’s delivery, for example, by offering the card directly in the tax filing and refund process, would reduce barriers to application and likely encourage more take-up. Publicizing such a card broadly and before tax season begins should increase filers’ familiarity with the product, and thus, its take-up and use.
6 These results are only marginally statistically significant (at the 10 percent level).
ES-5
This pilot produced a set of valuable lessons that could inform any potential future use of the tax refund delivery process as a means to reduce paper checks and to help expand access to mainstream financial products. Tests focusing on the card’s features suggest that individuals are price sensitive with respect to monthly fees, and that linked savings accounts (at least as designed in this pilot) were not perceived as valuable. Implications for future efforts suggest that Treasury direct its efforts primarily toward offering an account with a monthly fee as low as possible (zero if possible), even at the expense of additional card features such as a savings account.
In sum, the federal government’s creation of an option for tax filers to receive refunds directly onto a low-cost, account-linked card, as tested in this pilot, is a concept with promise. Such a card can reduce costs to Treasury, by reducing the number of costly paper checks used to deliver refunds. And such considerations are only one dimension of the potential benefits. Such a product could also reduce low- and moderate-income tax filers’ use of expensive alternative financial service outlets to cash their refund checks, as well as reduce the use of high-cost tax return options such as refund anticipation checks, especially if the product enables users to pay for tax preparation. And for those without a bank account, such a product could bring the benefits of access to mainstream financial services.
However, the pilot was not designed to measure the demand for an account-linked card or the costs of delivering such a card in a real-world setting. Thus, the net benefit of offering such a card cannot be determined by the pilot and hence is unknown.