A Weakened IRS Has Substantial Consequences

When Congress passed the Inflation Reduction Act (IRA) in August 2022, it included a $79.4 billion supplemental appropriation for the IRS spread over a decade. The investment was premised on a straightforward and empirically grounded argument that the IRS had been systematically underfunded for decades and as a result, a substantial share of legally owed taxes was going uncollected.

The Congressional Budget Office estimated the almost $80 billion in spending would yield roughly $204 billion in additional revenue over ten years, a net fiscal gain of approximately $124 billion. Those estimates, while debated, reflected a well-established literature showing that IRS enforcement spending generates high returns, especially if that spending is aimed at supporting audits of high-income filers and complex business returns. Academic researchers have found similar directional results1.  

The “tax gap”—the amount of taxes legally owed that are not paid in a timely way—currently totals about $700 billion annually as of the most recent IRS data available. Cumulatively over a decade, the gross tax gap approaches $7 trillion. Even modest improvements in efficiency of the IRS, applied to that base, generate meaningful revenue.

The IRS funding legislated in the IRA has not lasted. The Fiscal Responsibility Act of 2023 rescinded $1.4 billion and subsequent appropriations in 2024 and 2025 legislation rescinded $40.4 billion. IRS funding was reduced by an additional $11.7 billion in the most recent budget deal. Ultimately, just over 2/3 of the funding initially allocated in the IRA has been rescinded. 

The cumulative effect is a reduced IRS. The agency had begun hiring enforcement personnel and investing in technology modernization, but those efforts have not lasted. Hiring pipelines were opened and are now being partially closed. Institutional knowledge built through training is being lost to attrition and reductions in workforce.

The appropriations rescissions may prove secondary in long-run consequence to the workforce reductions that began in earnest in early 2025. Beginning in February 2025, approximately 7,300 probationary employees were terminated, followed by roughly 4,700 who accepted the administration's deferred resignation program. By the end of 2025, the total reduction in staffing amounted to 27,636 employees2. Administration officials have indicated a goal of reducing the IRS to roughly 50,000 employees, a reduction of approximately 50 percent that would return the agency to a staffing level not seen since the 1960s, when total IRS employment ranged from 47,000 to 52,000. At that reduced staffing level, IRS would need to process about 2.4 times as many returns: returns filed grew from roughly 110 million in 1969 to 267 million in 20243. These numbers understate the burden, given that the tax code has grown considerably more complex in the intervening decades. However, there has been some modernization at the IRS (e.g., electronic processing of returns) that makes this burden potentially manageable. 

The composition of the workforce reductions is at least as important as the scale. By mid-2025, the IRS had lost more than 3,600 revenue agents, representing roughly 31 percent of the agency's entire auditing staff. Revenue agents are the employees responsible for conducting complex, face-to-face examinations of high-income individual returns, partnership filings, and large corporate returns, audit categories with the high yield per staff hour and the most significant deterrence effects on non-compliance4. 

Viewed in one light, workforce reductions are a cost-saving measure consistent with broader government efficiency objectives. However, the IRS is not an ordinary expenditure center: it is the mechanism through which the federal government collects 96 percent of its revenues5. Salary savings from eliminating IRS employees are real but modest. Revenue losses from reduced enforcement capacity are likely to be substantially larger, and they compound over time through deteriorating voluntary compliance.

Estimates of how revenues respond to IRS enforcement depend on several empirical foundations worth reviewing carefully.

The most direct evidence comes from audit yield data. Data from the IRS Statistics of Income division, showed that an extra hour spent auditing a taxpayer earning over $5 million generated roughly $4,900 in recommended additional tax—compared to $650 for audits of taxpayers earning around $200,000. A subsequent 2020 paper argued that CBO's revenue estimates systematically understated the potential revenue gain from enforcement activities by ignoring deterrence effects, information-reporting improvements, and longer-run gains.

The audit-rate collapse for high-income filers over the preceding decade provides essential context. A 2022 GAO report found that audit rates for taxpayers with incomes above $200,000 fell more sharply than for any other income group between 2010 and 2019, a decline the IRS attributed primarily to reduced staffing from years of appropriations cuts. Taxpayers earning over $5 million saw their audit rate fall from roughly 16 percent in 2010 to just over 2 percent by 2019. The IRS staffing data underlying these patterns are published annually in the IRS Data Book; by fiscal year 2021, the number of full-time revenue agents and tax examiners had fallen 28 percent from 2011 levels.

The deterrence effect of enforcement is harder to measure but likely significant. Tax compliance is partly a function of perceived detection probability. When audit rates fall to levels that make non-compliance feel low-risk, some share of filers adjusts behavior accordingly (Boning, Guyton, Hodge, and Slemrod 2020; Boning, Hendren, Sprung-Keyser, and Stuart 2024). Based on the empirical literature, the Budget Lab conservatively estimated the return to deterrence to be 2.5 times each dollar recovered directly through audits. 

Income for which there is third-party information reporting provides a useful point of comparison. Wages, for which employers file W-2s, exhibit compliance rates above 99 percent. Business income reported only on Schedule C has net misreporting rates estimated at 57 percent for sole proprietors that have little or no information reporting and 12 percent for partnership and S corporation income that have some information reporting, per the IRS's own tax gap estimates for Tax Years 2014–2016. The compliance gap is overwhelmingly concentrated in income streams that the IRS cannot verify automatically—exactly the income streams that require enforcement capacity to address.

In April 2025, the IRS and DHS formalized a Memorandum of Understanding (MOU) allowing Immigration and Customs Enforcement (ICE) to request taxpayer information from the IRS for individuals under final orders of removal or subject to federal criminal investigation. Under the agreement, ICE provides the IRS with names and addresses of taxpayers, which the IRS then cross-references with its own records and returns to ICE. In August 2025, the IRS disclosed tens of thousands of taxpayer records to ICE, including personally identifying information and home addresses, after ICE had requested more than one million records. A federal court subsequently blocked further sharing in November 2025, finding the arrangement likely unlawful. 

This sharing of taxpayer data with ICE may reduce how much federal tax the IRS collects to the extent that undocumented immigrants currently filing taxes are hesitant to file taxes for fear of deportation. Undocumented immigrants pay about $66 billion in federal income and payroll taxes each year. As a result of the MOU, and depending on assumptions about taxpayer and employer responses, the federal government may lose between $147 billion and $479 billion over the next ten years, according to previous Budget Lab estimates.

Another challenge for the IRS is presented by changing (and in some cases, more complex) tax provisions of the One Big Beautiful Bill Act (OBBBA). Changes in the tax code complicate filing of taxes and put stress on IRS in terms of processing and creating new forms. The specific provisions of the OBBBA that induce new burden are the overtime, auto loan, and senior deductions as well as the revised State and Local Deduction (SALT) cap. In addition, OBBBA’s mix of permanent and temporary provisions as well as the fact they apply retroactively to the 2025 tax year adds obstacles for the IRS. For all of these reasons, we believe the estimates presented in Table 1 potentially understate the revenue effect of a reduced IRS.

Prior Budget Lab research indicates that more than $700 billion in annual taxes are owed but not paid in a timely way. This tax gap is highly responsive to the resources IRS is given for enforcement. A landmark 2022 investment of almost $80 billion in IRS capacity aimed to reduce the tax gap, but most of that investment has since been cancelled, and a significant portion of the IRS workforce has been terminated. These reductions in IRS enforcement resources have substantial fiscal consequences that are difficult to understand with prior models. 

A new Budget Lab analysis provides a richer framework for understanding effects of reductions in both IRS funding and staffing. We find that a 10-year reduction of $20 billion in the IRS budget increases the deficit (net of the IRS cost savings) by about $263 billion. Similarly, reduction of almost 28,000 in staffing—mirroring cuts that occurred in 2025—increases the deficit by almost $600 billion.  Together, these cuts increase the deficit by about $861 billion.