The Case Against Expensing R&E

On January 16 Senate Finance Committee Chair Ron Wyden, D-Ore., and House Ways and Means Committee Chair Jason Smith, R-Mo., introduced a bipartisan tax package that would revive the child tax credit and modify three provisions in the Tax Cuts and Jobs Act.¹ Under the proposal, the 30 percent limit on interest deductions would be relaxed by including depreciation and amortization in calculating deductible interest, expensing of corporate investment would be extended, and expensing of research and experimentation would be restored in lieu of amortization.

All the business provisions are problematic. The limitation on corporate interest deductions is necessary because some foreign-source income is exempt under section 245A, and allowing an interest deduction plus exempting income translates into negative tax rates.² Expensing investment means that the normal return on capital is also exempt, the exempt amount is larger when interest rates are higher, and there is no evidence that expensing increases capital investment. These provisions primarily benefit large corporations and their rich shareholders and top management.³

The most interesting provision is expensing R&E. Arguably, R&E produces positive externalities because people can move from one corporation to another, carrying with them knowledge resulting from R&E. But expensing R&E is the wrong way to subsidize research.⁴

R&E Subsidies

There are three major tax subsidies for R&E, and two of them are unaffected by the tax package.

First, R&E results in patented and copyrighted items, which in turn earn royalties. When these royalties are foreign source, they increase the foreign tax credit limitation. But it is clear that the source rule for royalties is wrong because royalties are entirely sourced to where the underlying patents or copyrights are protected (that is, the demand jurisdiction), but their economic source is also where they are created by R&E (the supply jurisdiction). That is why, for example, both the United States and the EU should have taxed Apple Ireland’s profits before the TCJA (and not Ireland, which did not contribute to earning those profits).⁵ The source rule for royalties is a subsidy. In addition, foreign-source royalties are eligible for the lower foreign-derived intangible income rate (13.125 percent instead of 21 percent), and that, too, is a subsidy. Both subsidies are de facto (and de jure, in the case of FDII) contingent on exports and therefore are a violation of WTO rules (which cannot, however, be enforced because the United States has deliberately sabotaged the WTO dispute resolution body).⁶

Second, the source rule for R&E expenses under section 864(g) is to apportion 50 percent of R&E conducted in the United States to domestic sources and to apportion the remaining 50 percent (at the annual election of the taxpayer) on the basis of gross sales or gross income (except that the amount apportioned to income from sources outside the United States must be at least 30 percent of the amount that would be so apportioned on the basis of gross sales). This is a subsidy because under the general rule for allocating and apportioning deductions they must be allocated to the category of income they produce (royalties, in the case of R&E), and then apportioned based on the source of that income. This means that under the general rule, if a taxpayer conducts R&E in the United States and all the resulting royalty income is foreign source under the royalties source rule, 100 percent of the R&E must be apportioned to foreign-source income, decreasing the FTC limitation. Thus, the section 864(g) rule is an export subsidy, and it is enhanced by the ability to partially source R&E based on income (subject to the taxpayer’s control) and not just sales.

So R&E of multinationals is arguably sufficiently subsidized without expensing. But the problems of expensing run deeper than that.

First, under normal tax rules, a deduction that produces income beyond the year it is incurred must be amortized over the period over which the income is generated: 15 or 20 years for a patent, and over 70 years for copyrights. Allowing amortization over five years as provided in the TCJA is itself a large subsidy. Expensing is a much larger subsidy.

Second, expensing means that the normal return to capital is exempt, and that is higher when interest rates are high. As noted above, it is hard to justify exempting income of large corporations that benefits rich shareholders and top management. Above-normal returns (rents) are not exempt but should be taxed at a higher rate than 21 percent given that taxing them is efficient (does not change behavior).⁷

Finally, expensing is primarily beneficial to large corporations with a lot of income and a lot of R&E. But in many cases this R&E does not produce positive externalities because it is just an update of existing technology and the knowledge it creates is already widespread. Creating a new edition of the iPhone does not really result in important positive externalities.⁸ The taxpayers that do create important positive externalities through R&E are start-ups, but start-ups typically do not have enough income in the early years to benefit from expensing; their R&E just produces losses.

This is why the correct way to subsidize R&E is not through source rules (start-ups do not have foreign-source income) or expensing, but rather through refundable credits (nonrefundable credits are useless to start-ups that only have losses). Current section 41 (the R&E credit) should be made refundable instead of expensing.

Refundable Credits

An important advantage of a refundable credit is its treatment under the UTPR (formerly known as the undertaxed payments rule) of pillar 2 of the base erosion and profit-shifting project 2.0. Under pillar 2, any country that a large multinational operates in has the right to levy a top-up tax under the UTPR if another part of the same multinational is subject to an effective tax rate below 15 percent. Thus, if expensing R&E reduces a U.S. multinational’s ETR below 15 percent, which is very plausible in many cases, an EU member state — or any country that adopts the UTPR — that the multinational operates in could apply the UTPR to its part of the multinational, negating the benefit of expensing.

Under OECD guidance, a qualified refundable tax credit (QRTC) does not trigger a reduction in the numerator of the ETR formula. Instead, it just counts as income that increases the denominator, and because the numerator is smaller than the denominator, its effect on the ETR is much smaller.⁹

A refundable credit costs money. But that is the point of a tax expenditure, and Congress approved a refundable credit for domestic microchip manufacturing and refundable green credits in 2022. Moreover, refundability for OECD purposes is measured over four years, which reduces the outlay. It is not clear that a refundable R&E credit will be more expensive than expensing, and it is the only way to benefit start-ups. In addition, Congress could allow the credit to reduce payroll taxes, which is also permitted by the OECD.

The Purpose of Corporate Tax

Many observers have criticized the OECD refundability rule.¹⁰ Why should refundability be the dividing line?¹¹

To answer this question, it is worthwhile to take a step back and ask what the purpose of the corporate tax is. There are two good reasons to tax publicly traded corporations: (1) The tax burden is likely to fall on rich shareholders or top management, and (2) doing so limits and regulates large corporations.¹²

First, most of the profits of the largest multinational enterprises are economic rents derived from their quasi-monopoly status, and taxing them on the rent is a good supplement to antitrust enforcement. This taxation of rents is efficient and will not cause the MNEs to change their behavior, except perhaps to discourage anti-competitive acquisitions of their potential rivals. I have argued that for a country like the United States that is the home of many of the largest MNEs, the corporate tax should be progressive and, at the top, apply at much higher rates on a worldwide basis.¹³

Moreover, as Kimberly Clausing has argued, when large corporations are monopolies that benefit from market power, their shareholders are more likely to bear the burden of the tax on their excess profits. These behemoths are already undertaking the profit-maximizing wage and price decisions, so if they could have shifted burdens to workers and consumers earlier, they would have already done so (that is, they are already paying profit-max wage and charging profit-max price). Therefore, a tax on pure profits falls on shareholders, except perhaps to the extent that rents are shared with top management who can negotiate for very large compensation packages.¹⁴

Positive capital income is far more concentrated than labor income, and 70 percent of all shareholders are either foreign or retirement account holders. Foreign shareholders benefit from U.S. public goods and are hard to tax other than at the corporate level, and tax-exempt shareholders are overly subsidized through hard-to-claw-back tax preferences. Thus, the corporate tax falls primarily on disproportionately well-off individuals at home or abroad who otherwise escape taxation.¹⁵ It also falls on the richest Americans, whose wealth is largely in the form of unrealized appreciation in corporate stock, but in their case, a 21 percent corporate tax is not an adequate substitute for taxing them on a mark-to-market basis, as both the Biden administration and Sen. Wyden have recently proposed.¹⁶

Second, the corporate tax is a good vehicle to regulate corporate behavior. Corporations, and especially the largest MNEs that are the target of pillar 2, are very important economic actors, and the government should be able to provide incentives for them to invest, for example, in clean energy, microchip manufacturing, or building affordable housing. Having a robust corporate tax enables the government to grant tax breaks for positive-externality-producing activities. From this perspective, ensuring that large MNEs pay 15 percent everywhere enables the government to reduce their tax rate at least from 21 percent to 15 percent on activities it deems worthy.¹⁷

But the question at hand is: In which cases should pillar 2 allow for reductions below 15 percent? This is what happens with QRTCs, qualified flow-through tax benefits (QFTBs), and the substance-based income exclusion.

The theoretical answer can be derived from the above analysis. Tax incentives like tax holidays, special economic zones, and patent boxes are designed to attract investment from other countries and therefore undermine the corporate tax base of those countries. It is those types of incentives — ones that just relate to the geographic location of income-producing activities (spatial tax expenditures) — that are the proper target of pillar 2 and should not be allowed to reduce the 15 percent minimum rate. They just shift the corporate tax base and therefore neither increase taxation of rents nor address market failures.

However, tax incentives to address market failures (market failure tax expenditures) like green energy credits, microchip manufacturing credits, low-income housing credits, and R&E credits should be acceptable because their main purpose is not to redirect investment that will happen anyway in some other location (a tax reduction that would be a windfall to the MNE and a cost to other countries) but rather to inspire investment that would not have happened at all but for the credit.¹⁸

What’s Refundability Got to Do With It?

Ideally, it would have been better to investigate the purpose of each tax expenditure and establish whether it has a legitimate link to a market failure. But that may be hard from an administrative perspective. Instead, drawing the line at refundability can be a reasonable proxy.

Tax holidays, special tax zones, and patent boxes involve a reduction of the tax rate, frequently to zero. The same applies to expensing and accelerated depreciation that reduce the ETR. They are typically not refundable but just involve the government refraining from collecting tax revenue.

On the one hand, those spatial tax expenditures could be made refundable, but most developing countries that engage in this type of harmful tax competition cannot afford refundability because it requires a cash outlay. Rich democracies can afford refundability, but then there are political constraints on doling out actual cash to large MNEs that do not apply to less visible tax expenditures.¹⁹ China may be an exception because it is both rich and nondemocratic, but even China recently has not wanted to directly subsidize foreign MNEs (as opposed to its domestic electric vehicle, solar panel, and microchip manufacturers).

On the other hand, the market failure tax credits are frequently refundable, and as we have seen, that is even true in the United States because credits created under both the Inflation Reduction Act and the Creating Helpful Incentives to Produce Semiconductors and Science Act (the CHIPS Act) are refundable, and other credits are typically QFTBs. The major exception in the United States is R&E, which traditionally has involved expensing and is not refundable, but that could and should be made refundable, as discussed above.

Thus, I believe the line drawn by pillar 2 is defensible. What is not defensible is the substance-based income exclusion, because that involves spatial tax competition for tangible investment (depreciation) and jobs (payroll), and it should never have been carved out of pillar 2 (or the global intangible low-taxed income regime, for that matter). The fact that it can be satisfied by a single highly paid manager driving a Porsche (because it is based on depreciation of tangible assets and payroll of employees) makes matters worse.

Nevertheless, refundability has its limitations as a proxy for the distinction between spatial tax expenditures and market failure expenditures. The refundability criterion relies on lack of resources (so it is biased against lower-income countries) and on political visibility (so it is biased against democracies). Rich nondemocratic countries like China benefit because they can afford refundable credits and do not care about public opinion.

It would be better if the OECD were explicit about the line it is drawing, saying that the relevant distinction is between spatial and market failure tax expenditures. This would clearly leave some expenditures like tax holidays on the wrong side of the line, and it would clearly leave green credits on the right side of the line.

Countries could try to manipulate this line, like any line in tax law (including refundability), but that is precisely what the peer review process in the inclusive framework is designed to prevent, by enabling any one of the 140-plus member countries to investigate tax expenditures in another member country that are suspicious and report the findings to the inclusive framework so that it can decide. Peer review worked well in the WTO to detect violations of the trade agreements (for example, on export subsidies, when peer review forced small countries to give up on tax holidays that de facto were all about exports). Peer review can work in this case as well, but only if it has a good criterion to work with, and the line between spatial and market failure tax expenditures seems to me a better theorized one than refundability.

Pillar 2 would treat the IRA and the CHIPs Act credits as QRTCs. Many other market-failure-oriented credits are QFTBs. QRTCs and QFTB would not significantly reduce the ETR for pillar 2 purposes. But what about other U.S. tax incentives? The main ones are R&E and FDII. In both cases, the UTPR could neutralize U.S. tax benefits under current law, even though the corporate alternative minimum tax is likely to reduce the frequency of such an impact.

As noted above, the FDII regime is an export subsidy that grants corporations that export goods, services, or intangibles from the United States a reduced 13.125 percent tax rate. The UTPR could result in the effective rate on FDII being raised to 15 percent (but only until 2026, when the FDII rate increases to 16.406 percent). But that is exactly the type of effect pillar 2 is supposed to have. FDII is a blatant spatial tax competition incentive designed to encourage U.S. and foreign MNEs to shift export income to the United States (without necessarily making any real U.S. investment, because all that is needed is to shift intangibles used to produce foreign income). As such, it is a violation of the WTO rule against export subsidies, but no country can sue the United States in the WTO because the United States has, as noted above, sabotaged the WTO dispute resolution mechanism.²⁰ Thus, imposing UTPR on FDII is perfectly justified by the need to establish a level playing field. And it is not inconsistent with the intentions of Congress, as evidenced by the scheduled rise of the FDII rate above 15 percent in 2026 and by the adoption of the corporate AMT to neutralize the overuse of tax expenditures by U.S. MNEs to the extent that they reduce their ETR below 15 percent.

As argued above, R&E is a market failure tax expenditure because it can generate positive externalities. It could and should be made refundable.

Congressional Options

It can be argued that Congress could also substitute a refundable credit for the FDII regime, and the same can be said for other patent boxes abroad with which FDII competes. The revenue cost of R&E and FDII over the 2023-2032 budget window is about the same ($111 billion each). But in both cases, the cash outlay would be quite visible, and it would be much easier for members of Congress to justify an R&E refundable credit that goes primarily to activities the public supports (for example, inventing new drugs) and that are frequently carried out by small start-ups than a refundable credit that goes primarily to the biggest MNEs. The main benefits of FDII go to U.S. MNEs bringing back intangibles that they shifted offshore before 2017, and it is much harder to justify a tax break for that, given that no jobs are created and that the intangibles should not have been shifted offshore in the first place. Finally, the benefit of FDII compared with a 15 percent rate disappears in 2026 (and perhaps earlier because of the corporate AMT), while the benefits of a refundable R&E credit should be made permanent.

The fundamental logic behind pillar 2 is to create a level playing field. From a U.S. perspective, by enacting the corporate AMT, Congress decided in 2022 that the ETR of U.S. MNEs should not be below 15 percent. The world made the same decision for all large MNEs in backing pillar 2. Even if the United States never adopts pillar 2 (although it came within one vote in the Senate of doing so), it benefits from it because U.S. MNEs will have less of an incentive to shift profits out of the United States if their ETR is 15 percent both within and without the United States (as opposed to 21 percent within and 10.5 percent without, as under U.S. law before the corporate AMT and pillar 2). And the global application of pillar 2 means that no U.S. MNE will be put at a competitive disadvantage by having competitors subject to an ETR below 15 percent, regardless of whether the United States adopts pillar 2.

Given these goals, it makes sense to distinguish between those tax expenditures that should be allowed to reduce the ETR below 15 percent and those that should not. Tax expenditures that are designed to attract mobile businesses like FDII should be discouraged, but tax expenditures to address market failures should be encouraged. While not perfect, refundability is a reasonable proxy for that distinction, although I would prefer to see the OECD clarify this explicitly and perhaps directly investigate the rationale behind each tax expenditure through the pillar 2 peer review process.