On his first day in office, President Donald Trump signed an executive order making official the Department of Government Efficiency (DOGE). Members of the initiative have reportedly already been visiting federal agencies, including The Treasury Department and the Internal Revenue Service, to inquire about opportunities for cutting costs and red tape. According to the Washington Post, they’ve encountered a “slew of obstacles,” including skeptical and wary civil servants.

I get it. As a long-time Treasury economist, I was often skeptical and wary of outsiders.

But Treasury and the IRS should take this opportunity to brag: Their research and development (R&D) have led to improvements in taxpayer service and administrative efficiency. Here are four examples:

1. Direct File is in use and improving

Last year, then-designated DOGE co-leads Elon Musk and Vivek Ramaswamy reportedly brainstormed about a mobile app that would enable taxpayers to file their taxes for free. Well, the future is already here, thanks to ongoing R&D by the IRS.

Direct File, an IRS online tool, allows for filing tax returns electronically—from computers, tablets, or smartphones—at no cost to eligible taxpayers. Favorable findings from last year’s evaluation of the pilot have led the IRS to expand both the scope and number of people eligible for Direct File this year. 

Last week, despite calls from some House Republicans to end the program immediately, Treasury Secretary nominee Scott Bessent committed to maintain Direct File through the current tax season. 

This fits the ideal R&D model: Define the problem, estimate potential costs and benefits of various options, run a pilot, and—based on lessons learned from the preceding steps—roll out the new initiative at a pace that allows for mid-course corrections. 

2. EITC outreach is reaching its target group

For years, the IRS sent notices promoting the Earned Income Tax Credit (EITC) to encourage taxpayers who appeared eligible for the credit to claim it. But many taxpayers never took up the opportunity. Was it because they didn’t want to deal with the IRS or didn’t understand the notice?

Research by TPC non-resident fellow Day Manoli and Carnegie Mellon University’s Saurabh Bhargava, in collaboration with IRS staffers, tested why nonresponse was high. They found that straightforward notices that included estimates of an individual’s potential credit were most effective at spurring claims. The IRS is now including that information in notices.

3. Data analytics are improving the efficiency of audit selection

After the Inflation Reduction Act’s massive ten-year budget boost in 2022, Secretary of the Treasury Janet Yellen directed the IRS to focus its new enforcement funding on wealthy taxpayers, complex partnerships, and big corporations. But selecting the right individuals or businesses from that group—those who are truly noncompliant—to audit can be challenging, particularly among taxpayers with very complicated financial arrangements. For example, partners owe taxes on their share of the business’s income, but identifying the partners is especially difficult when shares of the business are owned by a spiderweb of other partnerships, corporations, or trusts. 

To overcome that challenge, the IRS has committed to greater reliance on data analytics. Research presented at the annual IRS-TPC joint research conferences on tax administration shows that the IRS is making good on that commitment. IRS researchers are collaborating with outside experts in tax law and artificial intelligence to develop applications of machine learning to tax enforcement. And that research is being translated into action: As of the end of 2023, the IRS had used artificial intelligence and advanced analytics to open audits on 76 of the largest partnerships in the United States.

4. Audits are better targeted for fairness 

In 2024, a team of IRS, Treasury, and academic researchers (including TPC-affiliated scholar Jacob Goldin) found that Black taxpayers were 3 to 5 times more likely to be selected for audits than non-Black taxpayers. 

The study revealed the enforcement of the EITC was the main driver of those disparities, for two main reasons.

First, the Social Security Administration (SSA) had incomplete data on the parents of children claimed for the EITC. Second, return preparers with a disproportionately high propensity to file erroneous returns for their clients happened to be concentrated in minority communities. 

The IRS has taken steps to reduce racial disparities in audit selection. The agency is working with SSA to fill in the blanks linking parents and children, introduced a new scoring system to evaluate the risk of EITC errors or noncompliance, and initiated pilots to determine which machine learning models both improve audit accuracy and reduce racial disparities.

R&D matters today and can improve outcomes tomorrow

Nearly three years after the IRA was enacted, cutbacks in those funds—combined with freezes in the annual appropriations that finance the IRS’s base activities—have dampened expectations for transformative change in the agency. 

DOGE’s stated intent to scale back government could further exacerbate concerns about how the IRS will improve taxpayer service and enforcement. But DOGE also emphasizes increasing the efficiency of the government. Continued investment in IRS R&D presents a relatively low-cost opportunity for big efficiency gains.