Biden invested big in more tax audits. A new IRS analysis claims it’s working.

In 2021, Americans owed about $625 billion in taxes that they never paid. This number, called the “tax gap,” represented some 13.7 percent of all taxes due, and, had it been collected, it would have reduced the deficit by nearly one-quarter.

Getting more of that number collected is naturally an obsession of many tax and budget experts. It represents a way to fund government programs without raising taxes, adding to the deficit, or making other cuts. Deals like that can feel as rare as unicorns in the tax world. Funding the IRS in an effort to reduce the tax gap was one of the key ways Democrats funded their climate subsidies in the Inflation Reduction Act (IRA) of 2022.

The question is what share of that $625 billion could be realistically recovered. Even a perfectly resourced Internal Revenue Service would, after all, miss some tax evasion and misreporting. The Congressional Budget Office estimated that the IRS funding in the IRA would raise an average of $20.3 billion a year over 10 years, a small fraction of the $625 billion gap. For years, though, some economists have been vocally arguing that better enforcement could raise much, much more than that.

A new report released by the Treasury Department on Tuesday argues that those economists had a point, that we have been understating the revenue this new IRS funding will bring in. They argue that it will bring in over $170 billion more than they previously thought over an 11-year window and that, if Congress extends this funding for the IRS, the ultimate revenue gain could be more than double initial estimates.

The takeaway is that investing in the IRS could, if this study is accurate, be a better deal than we thought.

Why Treasury thinks it underestimated revenue from IRS funding

The increased revenue projected comes from a few different sources. One of the biggest is that Treasury is finding that marginal audits in recent years have been bringing in more money than anticipated. They had assumed that additional audits would bring in less and less revenue, on the grounds that the most profitable audits are already taking place — the low-hanging fruit has been plucked, in other words.

But the new report argues this isn’t as true as they had assumed. Recent IRS examinations, it writes, have shown “much greater revenue potential than is reflected in previous estimates of the impact of marginal work.”

The report also incorporates money the IRS anticipates getting due to “specific deterrence.” That’s a fancy term for a simple idea: If you audit a taxpayer in one year, they are less likely to hide income from the IRS in the next few years because now they are more worried about getting caught. This finding is based on a study released last year by economists Will Boning, Nathaniel Hendren, Ben Sprung-Keyser, and Ellen Stuart. While Boning is at the Treasury Department, the rest are independent academic researchers. The academic paper estimated the “returns” on spending on IRS audits: How much money does $1 spent on audits yield in new revenue?

The finding was that this spending pays for itself many times over and that the return is higher the wealthier the audited person is. Audits of low-income taxpayers bring in $5 for each $1 spent, while audits of high-income taxpayers bring in $12 per $1 spent. Much of that return, they find, comes from deterrence causing audited individuals to report more money in ensuing years. The Treasury paper takes the Boning et al paper’s estimates of increased revenue from deterrence and uses them to adjust upward Treasury’s previous estimates of how much revenue the audits funded by the IRA money will raise directly.

Finally, Treasury assumes that investment in better taxpayer services (like shorter wait times for calls, simpler reporting methods, and nudges/reminders to pay estimated taxes) and in better IT systems will result in still further revenue increases.

Prior to this paper, Treasury estimated that over the 11 years from 2024-2034, and if extended past 2029 when the IRA funding is due to run out, a fully funded IRS would bring in $390.3 billion in revenue directly from audits.

$497 billion in revenue once you add in the specific deterrence and greater efficiency effects $851 billion once you include better IT and taxpayer services.

That final number is more than double the initial $390.3 billion estimate.

Should we believe this?

The methodology of the Treasury report is straightforward, and the Boning et al. research is from very respected, careful researchers who I find credible. But it’s always worth being a little skeptical of releases like this. For one thing, the Treasury oversees the IRS, and so is effectively putting out research arguing that it and its subsidiary institutions should get more money. They may very well be right, but they’re hardly a disinterested party.

Secondly, the difference between even the initial $390.3 billion estimate of revenue from Treasury and the CBO’s $203.7 billion estimate gives me a little pause. The difference between the CBO estimate and the massive $851 billion number gives me still more (though, in fairness, the CBO is not modeling the IT and other effects in the Treasury paper). Some of the discrepancy is explained by the former being an 11-year estimate and the latter being a 10-year one; another difference is the CBO estimate starts in 2022, not 2024, so the Treasury number will naturally be a little bigger because of inflation in those two years.

But there’s still a big gap after adjusting for those factors. In latter years, Treasury is projecting much more revenue, even before these new deterrent effects, than CBO did. In 2031, for instance, CBO estimates $35.3 billion in new revenue from the IRS funding, while Treasury’s prior estimate was $57.1 billion, before all the increases it’s implementing now. The baseline Treasury estimate that year being over 60 percent higher than that of Congress’s impartial budget arbiter raises some red flags for me. A Treasury official explained that the CBO assumes lower returns on audits than Treasury has, and that this disagreement predates the new report. But that still doesn’t mean Treasury has the correct side of that argument.

All that said, the core idea of the report — that deterrence effects mean that increased numbers of audits could greatly increase revenue from funding the IRS — seems sound. It’s particularly relevant because the cost of the clean energy subsidies in the Inflation Reduction Act appears to be much higher than estimated when the bill was based: about 65 percent higher, per the Joint Committee on Taxation (JCT, the CBO’s partner that estimates tax provisions).

That’s largely due to more people and businesses wanting to invest in clean energy than initially estimated. For instance, the JCT has increased its estimate of what the US will spend subsidizing electric cars by more than fivefold because of surging demand for the cars. If you like these subsidies, this is a good thing: They’re doing their job and driving demand up.

But it raises the question of whether these provisions will actually be paid for by the revenue provisions of the Inflation Reduction Act. The new Treasury report is, in effect, an argument that the IRS funding could be up to the challenge of paying for all this unexpected clean energy spending.