The problems of our dysfunctional Congress are legion. This dysfunction not only affect Congress’ ability to get the basic work of government done, but it shrinks the space available to have rational conversations about improving public policy. We see this in the debate over the Earned Income Tax Credit (EITC). The EITC retains bipartisan support. Nevertheless, it regularly comes under criticism for the high error rates in the program.
At the high end, the IRS estimates the overclaim percentage at a high of 39.1%, but even the IRS’s low estimate of 28.5% represents about $14 billion a year. Conservative commentators use these estimates to argue the program is riddled with fraud. Liberals, on the other hand, stress that the complexity of the program’s eligibility requirements makes inadvertent mistakes likely and note that methodological problems may lead the IRS to overstate the improper payment rate.
As Book, Williams, and Holub note, Congress has, over the years, given the IRS a number of tools, including allowing the IRS to make unilateral changes to returns that would be prohibited on non-EITC claiming returns, imposing stiff civil penalties, and, most recently, imposing mandatory delays in refund payments for returns who claim the EITC and other refundable credits. Moreover, Congress requires the IRS to report on EITC integrity annually. Nevertheless, it’s been difficult for the IRS to reduce the error rate in the EITC.
In this context, Book, Williams, and Holub’s new paper on improving EITC administration is both refreshing and relevant. The paper is forthcoming about the error rate in the EITC and practical in its approach to solving the problem. Their suggestions require minimal Congressional action and are mindful of the importance of cost-conscious enforcement strategies given the realities of the IRS budget. In this way, they offer not only a model for improved administration of the EITC but also contribute to a growing literature on ways for the IRS to address compliance in the face of budget cuts. As importantly, they focus on ways of improving compliance that don’t threaten one of the great successes of the program, its high take-up rate among eligible beneficiaries, or reduce eligibility generally.
Because the paper seems intended for a non-technical audience, Book, Williams and Hulub begin by offering detailed background on the EITC, the problems with its administration, and the current enforcement framework. To summarize (overly) briefly their thorough explanations, three main types of errors contribute to EITC overpayments: income reporting, qualifying child errors, and filing status errors. Qualifying child errors represent roughly half of total overclaim payments, with estimates ranging from 42% to 54% of overclaims. Book, Williams, and Hulub suggest that that qualifying child errors are more likely to arise inadvertently, while income reporting errors are more likely to be intentional, though they acknowledge that some qualifying child errors are likely intentional as well.
Recognizing the limitations now faced by the IRS, Book, Williams, and Holub persuasively argue that the “IRS cannot audit itself out of the EITC compliance problem.” Nor do they think third-party reporting can help significantly. Third-party reporting can’t really address the complexity of the qualifying child test, nor may it help detect over-reporting of self-employment income. (One of the quirks of the EITC is that for taxpayers in the phase-in period of the credit, the refund increases when the taxpayer reports more income.) Moreover, increasing the penalty regime may not deter innocent mistakes, even assuming the IRS could impose the penalties with sufficient frequency to create a real deterrence threat.
At the heart of their paper is the idea of using insights from behavioral psychology to try and target compliance activities to the taxpayers most likely to be affected and at the time they are most likely to be affected. Specifically, they urge a compliance model that takes into account the different types of non-compliance present within the EITC.
There have been various proposals suggesting the IRS require EITC claimants to submit additional documentation proving eligibility. The authors suggest that rather than requiring this documentation, the IRS could simply ask the taxpayer additional questions about eligibility in situations where overclaims are more likely. Those situations include taxpayers claiming a qualifying child for the first time or reporting significant additional self-employment over the previous year’s return. Such responses would be voluntary, and the authors suggest that taxpayers who don’t provide this information shouldn’t automatically be selected for audits “to ensure that the tax system does not create a new subclass of taxpayers with differing rights.” I’m not sure how much a difference it makes whether the taxpayer is automatically selected rather than significantly more likely to be selected for audit, but it is possible that even if the failure to answer didn’t change audit risk significantly, it might improve compliance. The simple act of asking for this information may reduce inadvertent errors or discourage taxpayers from intentional misrepresentations.
Drawing on other work on behavior economics in tax compliance, the authors offer a number of suggestions to encourage taxpayer honesty. For example, they suggest the IRS could tailor the taxpayer experience to “convey that the taxpayer is known and understood” and therefore suggest it is harder to cheat. The IRS could ask specific questions like “where did you live with the child you are claiming” to raise the cognitive cost of providing inaccurate information. Given the length of the paper, I wish Book, Williams, and Holub had spent more time developing these behavior economics ideas. In part, it was occasionally difficult for me to imagine how the IRS could implement many of them without either the help of the tax software industry or regulatory power of that industry.
Nevertheless, Books, William, and Holub are modest in their goals for the paper. They note that many additional ideas may be worth exploring and that their ideas are “neither exhaustive nor fully fleshed out.” Rather, they suggest their interventions suggest areas where “further exploration . . . is warranted.” In that, their paper more than succeeds.