This paper analyzes the effects of a large randomized field experiment, carried out with H&R Block, offering matching incentives for IRA contributions at the time of tax preparation. About 15,000 H&R Block clients, in 60 offices in predominantly low- and middle-income neighborhoods in St. Louis, were randomly offered a 20 percent match on IRA contributions, a 50 percent match, or no match (the control group).
A significant share of low- and middle-income American families appears to be saving little, either for retirement or for any other purpose. Families with income below $40,000 have low rates of coverage under employer-provided pensions, are extremely unlikely to contribute to Individual Retirement Arrangements (IRAs), and in 2001 had median net financial wealth outside of retirement accounts of just $2,200. Researchers and policy makers alike have long considered ways to raise saving
among these families. The conventional approach to subsidizing saving through 401(k) plans and traditional IRAs provides tax deductions for contributions and tax deferral on account earnings. This approach has not enticed low- and middle-income families to contribute very much to retirement accounts, in part because the value of tax preferences is modest since these families face low marginal income tax rates.
In contrast, the provision of matching contributions could potentially represent a much more promising way to bolster incentives for low- and middle-income households to save. Matching contributions can be independent of the individual's marginal tax rate, and thus provide a significant incentive for saving even to people in low marginal tax brackets. Little is known, however, about whether such matching contributions would in fact induce a meaningful fraction of low- and middle-income families to save for retirement.
The Saver's Credit offers one example of matching contributions provided by the Federal government. Enacted in 2001, the credit provides an income tax reduction of up to 50 percent of funds contributed to a 401(k) or IRA by qualified filers. The credit as enacted may not accomplish its full potential, however, in part due to its low income thresholds for eligibility, in part because it is not refundable, and in part, perhaps, because it has not been effectively advertised or explained to tax filers.
Matching contributions are also present in many employer-sponsored 401(k) plans. Previous studies have found mixed evidence of the effects of match rates on 401(k) participation and contributions. A significant concern in these studies, furthermore, is that the match rates offered by firms in 401(k) plans may not be independent of worker characteristics at the firm, which makes it difficult to disentangle the independent effect of matching rates on contributions. Another concern is that the results apply mainly to relatively affluent households in the presence of automatic payroll deduction and the other workplace features associated with 401(k) plans, which could affect contributions. Therefore, the results may not be directly applicable to low- and middle-income families or to policy interventions that occur outside the workplace.
This paper reports evidence from the first large-scale, randomized field experimentever conducted regarding the effects of matching rates on the willingness of low- and middle-income families to contribute to IRAs. By randomizing the matching rate across tax filers, we are able to identify not only the impact of the presence of a match, but also how variations in the matching rate affect both take-up and contribution levels. Unlike the Saver's Credit, the match provided in this experiment is available
in full to (virtually) all tax filers, has a simple structure, is explained to potential account holders in a straightforward manner, and is deposited directly into an IRA rather than reducing income tax liability. Unlike 401(k) matching contributions, the match provided is independent of individuals' characteristics and of the workplace environment.
The field experiment was conducted in conjunction with H&R Block, the largest tax preparer in the country, and coorganized with the Outreach & Business Development Group at H&R Block. H&R Block paid the direct costs associated with implementing the experiment, including for the matching contributions, advertising materials, and the training of tax professionals. The experiment was run in 60 H&R Block tax preparation offices in the St. Louis metro area from March 5th to April 5th, 2005. The experiment was built around the Express IRA (X-IRA) product offered by H&R Block, which allows clients to make IRA contributions at the time of tax preparation and to fund those contributions with part or all of their federal income tax refunds or from other sources. In effect, the X-IRA allows the client to “split” their anticipated refund between contributions to a retirement account and other uses.
Each client preparing a tax return in one of the 60 offices during the period was randomly assigned to one of three match rates for X-IRA contributions: zero (the control group), 20 percent, or 50 percent. Contributions were matched up to $1,000, a limit that applied separately for each spouse for married tax filers. Each client, including those in the control group, received a waiver of the $15 setup fee for opening an X-IRA. The minimum X-IRA contribution is $300.
The evaluation generates two overarching sets of conclusions. First, match rates can have large effects on IRA participation and contributions. Take-up rates are 3 percent, 10 percent, and 17 percent, respectively, for the control group, the 20-percent match group and the 50-percent match group. Conditional on take-up, average contribution levels (excluding the match) are $860, $1,280, and $1,310, respectively. With the match included, average IRA deposits were $860, $1480, and $1,870 respectively. Taking participation rates into account and including the matching contributions, average IRA deposits with the 20-percent and 50-percent matches were 5 and 11 times higher than with no match. The effects are particularly large for married tax filers, with take-up rates of 3 percent, 13 percent, and 25 percent in the three groups. These effects are substantially larger than those found in the context of 401(k) matches.
The second broad conclusion is that, long a number of dimensions, observed taxpayer behavior is not consistent with simple economic models that posit fully informed and fully rational consumers. First, take-up rates varied significantly across different tax preparers. Second, the IRA take-up response to the experimental match was significantly larger than taxpayers' responses to equivalent changes in incentives embodied in the existing Saver's Credit. Third, take-up rates were significantly less than 100 percent even in the presence of significant matches, which suggests that clients were not attempting to game the system by contributing to the IRA and receiving the matching contribution, and then cashing out the accounts (with minimal penalties) as soon as possible afterwards. These findings are consistent with a view that information and simplicity affect saving choices.
Taken together, the two sets of results suggest that a combination of financial incentives, tax preparer assistance, the opportunity to use part of an income tax refund to save, and easily accessible saving vehicles could generate substantial increases in both the efficacy of federal tax incentives and the willingness of households to contribute to retirement saving accounts.
The remainder of the paper is organized as follows. Section 2 describes the experimental design and data. Section 3 analyzes the effects of the experiment. Section 4 compares the results to evidence from the Saver's Credit. Section 5 concludes.