The Big Question About The Inflation Reduction Act: How Well Will Added Tax Enforcement Work?
Democrats say they'll raise $124 billion over a decade by beefing up the IRS. Larry Summers thinks it will be more like $1 trillion.
It’s highly likely that Democrats will lose control of at least one house of Congress in elections this November. That means they’ll soon need to work together with Republicans to set the country’s fiscal course. If you remember the last time a Democratic president had to do a lot of this with a Republican House — from 2011 roughly through 2013 — it was not a pleasant or sensible process.
Very Serious is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
The next time will be different in at least one important way: The demands Republicans are sure to make for fiscal austerity will make more economic sense than they did last time, because inflation is so much higher and the economy is so much closer to capacity. This time around, deficit reduction will be an actual economic imperative, not just a political one.
And that I think is the most important thing to understand about the revenue side of the so-called Inflation Reduction Act: Assuming it becomes law, it will entail Democrats seizing their last chance to ensure that higher taxes on corporations and more aggressive tax enforcement are two key components of the inevitable fiscal consolidation that is to come over the next few years. The more fiscal austerity that comes out of those channels, the less that will need to come from channels that are unappealing to Democrats, like entitlement cuts. This alone is a major win.
But the sheer magnitude of the win will depend on the answer to a difficult question: How much money will those added tax enforcement measures actually raise, anyway?
A brief summary of what the Inflation Reduction Act would do
The IRA proposes to cut the federal budget deficit by cutting the rates the federal government will pay for Medicare prescription drugs (expected to save $288 billion over a decade) and by raising more tax revenue in three ways:
Imposing a new minimum corporate tax based on financial statement income ($313 billion of added revenue over 10 years, with that revenue coming especially from large manufacturing companies)
Adding resources for IRS enforcement (estimated by Senate Democrats at $124 billion net after enforcement expenses, but they may be being modest — more on that below), and
Narrowing the carried interest loophole for investment fund managers’ income ($14 billion)
By Senate Democrats’ official estimates, those provisions would produce $739 billion in added revenue over a decade. The law would also spend $64 billion to extend enhanced ACA insurance subsidies for three more years and $369 billion over a decade on energy and climate programs, leaving a little more than $300 billion for deficit reduction.1
Larry Summers thinks added enforcement will actually raise $1 trillion alone
It’s always hard to estimate how much a new law will actually affect the federal budget over a decade or more, but estimates of the result of added tax enforcement are especially speculative. How effective will the IRS be at using new funds to collect more taxes? How will taxpayers respond to an environment of tougher enforcement — will they be deterred into paying more out of fear of audits? Unlike straightforward changes to tax rates or benefit formulas, these questions are not squarely in the wheelhouse of the experts at the Congressional Budget Office or the Joint Committee on Taxation, which nonetheless have to render educated guesses in response to them anyway.
Senate Democrats’ stated estimate that the provisions will raise $124 billion on net over a decade is based on a calculation done last year by the Congressional Budget Office, which found approximately $80 billion in added IRS enforcement spending would lead to approximately $200 billion in added tax collections over a 10-year period, for a net gain in the $120 billion range. This estimate is much more cautious than figures Democrats were throwing around last year during the construction of Build Back Better. Last fall, the Biden administration was saying it could wring $400 billion in added revenue out of added enforcement spending. And economists close to the Biden administration continue to be clear that they think they will get more than $124 billion out of enforcement.2
The economist Larry Summers, widely cited in the news this week as having helped convince Sen. Joe Manchin to support this legislation on the grounds that it will reduce inflation and the budget deficit, is especially bullish on the effectiveness of enforcement measures. And he thinks $124 billion (i.e., added gross revenues of about $200 billion) is a major underestimate.3
“I think one is completely safe in saying above $500 billion, and my honest guess is closer to $1 trillion,” he told me this week, when I asked him how much added revenue he thought the added IRS spending contemplated in the reconciliation deal would actually bring in.
Summers has specific critiques of how the CBO analysts reached their more modest estimate, some of which he advanced in a Washington Post op-ed last fall. He complains that the CBO focused on the direct effects of audits (where a taxpayer faces an audit and settles or is forced to pay more tax) and did not sufficiently account for added tax revenue through deterrence — for instance, taxpayers burned once by an audit taking a persistently less aggressive approach to calculating future tax liabilities, or unaudited taxpayers (and their preparers) making more cautious tax calculations out of fear of an audit. He also notes the CBO did not account for how the IRS could use additional funding to improve its technology and keep better track of what funds it is owed, even from taxpayers who are not audited.4
Jason Furman, who like Summers is an alumnus of the Clinton and Obama administration economic teams, told me he’s inherently wary of claims that efforts to combat “waste, fraud, and abuse” will have significant effects in improving revenues or reducing spending. Claims about free and near-free lunches in public policy tend to be too good to be true. But he also has found Summers’ arguments about why added enforcement could be expected to raise so much revenue to be compelling.
“I’ve had several arguments with Larry about him assuming such strong returns from added enforcement,” he told me. “And every time we’ve had the argument, Larry has won.”
The CBO, for its part, has expressed its reasons for being cautious in its assumptions about both how much revenue the IRS could generate from added enforcement and also about how quickly it could generate that added revenue. They note historical experience: Changes in IRS enforcement effort have been slow to feed through into broad changes in voluntary taxpayer behavior, and while specific taxpayers do tend to report more income in the years after themselves being audited, CBO says that behavior does not extend to high-income taxpayers and corporations.5
There is also the question of how quickly the IRS could take a larger budget and actually turn itself into a much larger and more fearsome operation. You may have noticed that it’s a challenging time to hire. Suppose the IRS goes out and tries to hire many thousands of skilled new enforcement agents, at the same time that private industry also starts wanting to hire a lot more tax experts in order to respond to a more aggressive IRS. That could be a real challenge to scale up. Indeed, the IRS has already been having trouble hitting its existing hiring targets. And then how long will it take for taxpayers to notice that more of their friends and associates are being audited and change their behavior? There are plausible reasons to think that, even if added IRS enforcement will add a lot of revenue in the long run, it could be many years before most of the behavioral effects phase in.
Whether the law would cut inflation depends on how well the tax enforcement works
The law is called the Inflation Reduction Act, but the economists behind the Penn-Wharton Budget Model don’t believe it will actually materially reduce inflation. And if you take the numbers Senate Democrats have circulated at face value, they have good reason to reach that conclusion.
Cutting the budget deficit will tend to reduce inflation, but the IRA simply isn’t designed to cut the budget deficit that deeply, especially in the near term. In fact, the plan is likely to modestly increase budget deficits through 2026, with deficit reduction starting in 2027 as both IRS enforcement and cuts to Medicare drug prices ramp up.6 Because of this chronology, the Penn-Wharton researchers believe the plan would modestly increase inflation in the near term — perhaps by a few hundredths of a percentage point — with modest decreases in inflation to come in the out years. They also note that these inflation estimates are “not statistically different than zero, thereby indicating a very low level of confidence that the legislation will have any impact on inflation.”
But if Summers is right, those estimates could look materially different, because the law would have a much larger deficit-reducing effect. The effect on inflation still wouldn’t be very large — primary responsibility for reining in inflation lies with the Federal Reserve — but it would have a clearer negative sign.
Of course, that greater revenues from enforcement would lead to much more significant deficit reduction is also important in its own right — the more impressive the actual results of enforcement are, the less necessary it will be in coming years to increase tax rates or reduce spending.
The other big revenue raiser in the IRA isn’t my favorite, but I can live with it
I want to leave you by talking about the other big revenue-raising part of the IRA: a new minimum tax on corporations based on their book income (i.e., the income they report to shareholders, not their taxable income reported to the government.)
While the regular corporate tax rate is 21% of the income that companies count as taxable based on IRS rules, companies would also face a 15% tax calculation based on an adjusted version of the profit they report to shareholders, and they would pay whichever of the two is higher. That’s expected to raise about $300 billion over a decade, in large part by reducing the ability of firms to take advantage of tax deductions associated with depreciation, foreign-taxed income, and deferred compensation. In practice, the burden of the new tax would fall disproportionately on manufacturing firms, because those firms enjoy the greatest benefits from the accelerated depreciation provisions that the new tax would rein in.
I don’t love this idea. I think if you want to change the base of the corporate income tax, you should probably just change the taxable income rules directly, instead of creating a parallel corporate tax code based on a separate set of accounting standards. I would also prefer a tax that raises burdens more proportionally across the whole corporate sector instead of falling especially on manufacturers, a category of business that both political parties have expressed a desire to encourage to invest and expand more in the US. But a recurring theme you hear from experts talking about the minimum tax is that it is a “second-best” policy.
That is, your first choice might be to raise the corporate income tax rate, or to abolish certain corporate income deductions and credits to broaden the tax base, or to implement tax structures specifically aimed at preventing the undue offshoring of US corporations’ profits. But those things are all hard, especially since Sen. Kyrsten Sinema has been clear she won’t vote for a higher corporate tax rate, while this policy is available, and it amounts to a kludgy way to achieve something similar with less political resistance. It’s an idea people can settle for — and I talked with several experts who said they favored the plan even though, if they could write it all by themselves, they would have done something different.
“A minimum tax is rarely anyone's first-best, but if your constraint is that a pivotal senator is really focused on rates and is more accepting of the number 15 than a number starting in the 20s, then a minimum tax is what you do,” said Daniel Hemel, a tax law professor at New York University.
All that said, do you know who is actually quite enthusiastic about the minimum-tax idea? Larry Summers.
“I don't want to oversell this and say it's the world most fantastic and the first tax reform I would do, but it would be under-claiming to say it was just a crummy second-best because Sinema wouldn't accept higher tax rates,” he told me, noting that he’s been advocating for something similar since 2016 — long before he could have known there would be a 50-50 Senate to contend with.
While accountants worry a tax based on book income will encourage companies to fiddle with their books to report lower profits, Summers hopes that a minimum tax will mean both higher tax revenues and more accurate corporate financial statements. Corporations are incentivized to report high profits to shareholders and low profits to the government; tying the two figures more closely together could, in theory, cause those pressures to meet in the middle and promote accuracy. He also sees the plan as a way to achieve some of the objectives of Treasury Secretary Janet Yellen’s efforts to coordinate minimum corporate taxes in countries around the world and to rein in undue tax preferences related to deferred employee compensation.7
I don’t really buy the idea of neatly offsetting distortions on income reporting — it’s a little cute — but I’m also satisfied with the idea that the rate on this tax is low enough that it’s unlikely to have important negative economic effects compared to a straightforward increase in the corporate income tax rate or adjustment of the base. It might distort corporate behavior a little bit, but all taxes do that, and the distortions are not likely to be large. Even Kyle Pomerleau, an economist at the conservative American Enterprise Institute who’s written critically of book-minimum taxes, told me he did not think what he sees as the design flaws in the plan would have a large effect on GDP.
This tax plan is good enough
Which leads me to my final takeaway: Stronger tax enforcement and the book minimum tax are both reasonably equitable ways to raise additional tax revenue without doing much to reduce incentives to work and invest. They are a pretty good approach, even if they are not the first-best approach. More vigorous enforcement, in particular, is clearly a good idea, and the only question is how good an idea it is. Neither policy should have large effects on inflation, but especially effective tax enforcement could modestly reduce inflation. And economic growth effects should mainly come from the spending side of the bill — if the energy and climate reforms are effective at raising productivity, or stabilizing energy prices, or mitigating the negative effects of climate change, those should improve growth. But probably not very quickly.
Overall, the plan is good enough, and Congress should pass it.
I’ll come back around to more specific views on the spending side in August, but my broad view is that the slimmed-down package represents a significant improvement over last year’s bloated Build Back Better package. The old package was a grab-bag of interest-group driven social spending proposals, some of them very half baked (especially the plan for child-care subsidies that were likely to wreak further havoc on an already-dysfunctional child-care industry). The new package is focused on funding a proven health care program (enhanced subsidies that make it easier for people to get insurance under the Affordable Care Act) and investing in what I see as the most important policy area for improving productive capacity (energy, including but not limited to clean energy).
When Congress tries to move deficit-neutral legislation through the budget reconciliation process, every dollar counts, and they have reason to fight for the most aggressive possible interpretation of how much revenue a provision like IRS tax enforcement will raise. But the IRA is not supposed to be deficit-neutral, and the specific CBO estimates are therefore less important — squeezing more dollars out of the enforcement estimate doesn’t let them spend more, because they’re not trying to spend as much as possible. All of which is to say, I would not mistake Senate Democrats’ official acceptance of the $124 billion estimate as a sign they actually believe the provision will raise only $124 billion. For legislative purposes, $124 billion is good enough to pencil, and they can figure out how much the IRS actually raises from enforcement when it’s actually doing the enforcement.
To put it in context, the IRS estimates that the federal government collects only five out of every six dollars it is legally owed in taxes, meaning that over a decade there will be about $7.5 trillion in taxes that should get paid but won’t be. To collect another $200 billion requires shrinking that gap by about 2.5%.
From the CBO’s analysis last fall, producing the $120 billion ballpark estimate: “The estimate reflects CBO’s expectation that the increased enforcement activities would change the voluntary compliance rate—that is, the share of taxes owed that are paid voluntarily and on time—only modestly. The magnitude of that effect is highly uncertain, however, and the empirical evidence about the effects of audits on taxpayers’ behavior is inconclusive. Research about such deterrence finds varying responses, depending on the type of taxpayer. People generally increase their reported income in the years following an audit, but people with higher income generally do not, and neither do corporations.”
The IRA also includes just a three-year extension of enhanced ACA insurance subsidies — a violation of Manchin’s prior insistence that he wouldn’t support a spending plan where programs intended to be permanent weren’t actually financed through the whole 10-year budget window. As shown in the Penn-Wharton calculations, if you assume those subsidies will eventually be extended for the full 10 years, the deficit reduction in the out years would be reduced — but actual deficit reduction would nonetheless commence in 2027, because the cost of the extended subsidies would still be outweighed by IRS enforcement and Medicare cost cuts.
As an aside, I think Summers has to feel pretty good about the role he’s played on economic policy over the last two years. The narrative as Biden came into office was that Summers was on the outs — shut out of policymaking in the administration, hated by progressives, and widely viewed as a skunk at the garden party (or even as acting out of spite) when he warned the American Rescue Plan would be inflationary. Not only was he directionally correct about the ARP and inflation (or at least, more correct than most other observers), he preserved credibility to convince Manchin that other Biden administration fiscal initiatives could push downward on inflation. Now, two revenue policies with which he is particularly associated — IRS enforcement and the minimum tax — have survived to be the core of what’s likely to be Biden’s last major partisan fiscal bill, even as every other major Democratic tax increase proposal has gotten shot down by one veto-point senator or another.