NYU Tax Policy Colloquium, week 13: Lilian Faulhaber's Lost in Translation: Excess Returns and the Search for Substantial Activities

Yesterday at the colloquium, Lilian Faulhaber presented the above-named paper (not posted online because it's an early draft). It has some interesting overlaps with my minimum tax article, which we discussed in the colloquium earlier in the semester, as to some extent we are looking at the same phenomena from complementary angles.

The paper traces a single idea that has spread rapidly from one source to another: the idea of enacting a minimum tax on foreign excess returns. Initially appearing as an Obama Administration budget proposal, it crossed the aisle to the House Republicans' international tax reform plans back in the day, and then in 2017 entered the law (of course, with substantial modifications from the original version) as GILTI. Meanwhile, it has also crossed the ocean and entered OECD deliberations, most recently in the still-evolving GLoBE proposal.

In all manifestations, this is a minimum tax on foreign source income (FSI), imposed by the residence jurisdiction on multinationals that have paid "too low" a rate of source country taxation, as determined either on a worldwide basis or (as in the Obama Administration versions) country by country. It's a tax on "excess returns" because, at least in most versions, it allows the exclusion of some sort of "normal" return, a term that the various versions do not define consistently, and that date back (in the international realm) to a 2013 paper by the late Harry Grubert and Rosanne Altshuler.

One of the paper's main arguments, which I found convincing, was that these various minimum taxes on foreign excess returns should be viewed as new efforts to take on the "substantial activities" issue, but via a rule in lieu of a standard. What I mean here by "substantial activities" is that, in response to profit-shifting by multinationals, typically into tax havens in which little actual economic activity is being conducted, countries have long sought to provide, by one means or another, that the shifted profits cannot exceed those that might conceivably or plausibly have been earned in the claimed source jurisdictions via actual "substantial activities" there.

Among the paper's nice contributions, even in this early draft, was its offering a taxonomy of different ways in which the notion of "extra-normal" returns - the ones that the global minimum tax might aim to reach - has been used. For example:

1) A well-known thread in the literature on income and consumption taxation posits that excluding "normal" returns - often (but not always) defined as meaning the risk-free rate of return - prevents the tax system to avoid distorting inter-temporal choices by taxpayers, in particular between current consumption and saving for future consumption.

2) Grubert and Altshuler argue that a multinational that is earning extra-normal returns may be facing lighter competitive pressures than those earning merely normal returns, offering more scope to tax them without competitive harm.

3) A familiar thread in the economics literature distinguishes rents from normal returns, arguing that the former can efficiently be taxed without thereby discouraging the activities that yield the rents.

4) Proponents of excess profits taxes often view such profits as the fruits of a "windfall" that could not have been anticipated, supporting the view that they can be taxed without worsening incentives. (The standard response is that this will discourage companies from seeking to anticipate changes that are not wholly unanticipatable.) Some arguments for a tax on foreign excess profits occupy similar territory, although they focus on concern about stable competitive distortions, rather than sudden "windfall" changes.

5) When an affiliate of a multinational company that resides in a low-tax jurisdiction, such as a tax haven, appears to be reaping extra-normal profits even though little goes on there, this may indicate that transfer pricing or other such games are going on within the multinational affiliated group.

6) If a multinational is reaping huge profits abroad, this may indicate that it has zero-basis intangible assets that are easily shifted to tax havens for profits-reporting purposes.

One thing about all these different theories is that all of them might be true, at least in particular cases. Yet they are using the distinction between normal and extra-normal profits in different ways, arguably inviting multiple and inconsistent definitions. They also quite clearly might motivate very different types of policy responses.

In sum, the paper identifies a very rich area in which a lot is going on, and a lot can be said. Here are four different directions in which the paper's further analysis might go - which is not to say either that it should cover all four, or that there aren't also various other fruitful lines of inquiry.

1) Rules versus standards - A familiar theme in the legal literature distinguishes between rules and standards. For example, a 65 mph speed limit is a rule; a requirement that one not drive unreasonably fast given all of the relevant circumstances is a standard. Both typically apply to us when we are driving on the highway. Each approach has its familiar set of pluses and minuses, and the use of each may be preferable under particular circumstances. 

The foreign excess profits taxes are generally straightforward rules, whereas requiring "substantial activities" in relation to the profits being claimed in a particular jurisdiction is very much a standard. The paper currently criticizes how the rule-like approach is being implemented in different settings, but it also reflects the view that, in this particular setting, a rule may be greatly preferable to a standard

2) Frictions, backflips, and economic substance - The question of why one would require "economic substance" in order for a particular taxpayer position to be accepted by the tax authorities is an old one. I've written about it extensively, and I've argued that the issue is distinct from rules versus standards, because one can write specific rules requiring particular indicia of economic substance.

What makes the case for economic substance requirements a bit perplexing or paradoxical is its seemingly gratuitously inducing taxpayers to incur extra deadweight loss as the price of avoiding their reach. Thus, consider the classic Knetsch case from 1960. Here the US Supreme Court held that a particular tax shelter transaction was a sham, and hence would be disregarded for tax purposes.

In Knetsch, the taxpayer purported to borrow $4 million from an insurance company at 3.5% interest, in order to "invest" the $4 million at a 2.5% interest rate. Thus, every year he was ostensibly incurring a $140,000 interest liability in order to earn $100,000. The annual return of negative $40,000 was well worth it, however, at least in the short term, if (as black-letter tax law at the time suggested), he could deduct the full $140K downside while deferring the entire $100K upside. Marginal tax rates at the time caused a $140K tax loss to reduce his annual tax liability by a lot more than $40K.

As I've noted, the "pure tax arbitrage" here makes this potentially an unlimited money machine, at least until one has wholly wiped out one's positive tax liability (since losses are nonrefundable). E.g., one can just as well borrow $4 billion, or for that matter $4 trillion, in order to "invest" the same amount with the same counter-party.

The transaction lost because, inter alia, it served no non-tax business purpose and had no pre-tax profit potential. But suppose that the upside (the $100K annual return) had been double or nothing, based on the equivalent of a coin toss. Then the business purpose would be "I'm feeling lucky," and each year there would have been a 50% chance of a pre-tax profit (albeit, also a 50% chance of losing $140K instead of just $40K).

Suppose we agree that taxpayer risk aversion makes this simple fix to the transaction unappealing. Then we are imposing a risk - for those taxpayers who do it and go forward - that they didn't want, this tax consideration aside, suggesting that it increases deadweight loss, to no good end, in those instances. The reason for imposing the requirement is that it avoids the transaction's bad consequences in cases where they give up rather than taking on board more risk. I've called this a way of imposing costlier (in lieu of cheaper) effective electivity, although the "fee" is paid in the form of deadweight loss.

This has come to be known as the "backflips" point, since in one of my writings on the topic I said that one might as well require taxpayers to perform backflips at the IRS Chief Counsel's office at midnight on New Year's Eve, as require them to add unwanted features to their business transactions. (But this was not a criticism of the approach, at least faute de mieux.)

A "substantial activities" requirement for claiming profits arose in a given jurisdiction can be viewed as requiring a backflip. It requires one to place enough actual economic activity (e.g., workers and physical assets) in the low-tax jurisdiction, even if this is otherwise suboptimal and lowers global pre-tax profitability. It also might be quite difficult for taxpayers to meet the test in a "small" or "pure" tax haven jurisdiction that doesn't have enough room to accommodate the trappings.

Optimizing in this sort of a choice environment is rather 97th-best. One wants to optimize the mix between reducing the tax planning one dislikes and increasing deadweight loss in cases where the taxpayers go forward anyway.

3) Alternative approaches to exempting normal returns - The paper discusses the wide range of approaches taken here. For example, the Grubert-Altshuler paper urges the use of expensing to exempt the normal return, but none of the prominent rules in this realm that have been proposed or adopted to date do so. Instead, they come up with differently defined exempt rates of returns that may be applied to different denominators (e.g., tangible assets, assets plus payroll, and with other variations as well).

Leaving aside the denominator question, I would think that current market rates - although there are multiple choices here - should be used. Why, for example, should the rigor of GILTI's 10% tax-free return effectively vary as prevailing market interest rates rise and fall?

GILTI also arguably shows what can go wrong when one makes the tax-exempt rate "too high." For example, suppose that a US multinational envisions that a given tangible asset will yield a marginal return no higher than 5 percent, no matter where it is placed. Locating it abroad - even, say, if this reduces the marginal return from 5 percent to 4 percent - might be well worth it after-tax, given that the asset's deemed normal return of 10 percent abroad might reduce the firm's GILTI liability.

Could a rule like GILTI use expensing to specify the exempted normal return? One virtue of this approach is that we don't need to decide what tax-free rate we want. Instead, it may come from the firm's own decisional margins. But using expensing to eliminate the tax on the normal return is not without defects. For example, it may cause tax rate differences between years to matter more, and it may create greater year-by-year swings between "income" under the excess returns minimum tax and the regular tax to which it's being compared. But the literature concerning cash-flow taxation addresses such issues and could be consulted for guidance regarding such issues.

4) Relevance of foreign taxes paid - In each of the rules that the paper addresses, the multinational's tax liability depends on (a) how one computes the extra-normal return, (b) what tax rate applies to this return, and (c) what foreign taxes have been paid. The last of these three factors raises what I call the "marginal reimbursement rate" (MRR question). To what extent does the country that is imposing the tax wish to let domestic tax liability decrease - by as much as dollar for dollar? - per extra dollar of foreign taxes paid, given that "we" don't get the money from foreign tax payments.

I've written about this issue for years, and I feel that I have gotten through to a degree. For example, the Obama Administration proposals and GILTI both provide MRRs below 100%, thus retaining some foreign tax cost-consciousness on the part of US taxpayers. But the issue isn't just whether the MRR should be below 100% - in a unilateral model where one favors national not global welfare, it clearly should - but also what sorts of tradeoffs should govern the choice. This question, I feel, has not attracted a ton of scholarly attention even though it is clearly significant.

The MRR question is among those potentially plaguing the OECD GLoBE efforts. 100% MRRs might be fine if a rule was being imposed from the top, but the OECD isn't quite in that position. Countries might also differ in what sort of MRRs they prefer. For example, a bigger country, such as Germany or France, might have different interests than a smaller one, such as Holland or Luxembourg.

Also, if countries are fine with 100% MRRs in the GLoBE setting because (for reasons of tax competition) they don't actually want the revenue anyway, that perception of their interests might have broader implications for the GLoBE's actual feasibility and evolution in the field.

Just to sum up, the paper we discussed yesterday has identified a rich mother lode of interesting issues, and I look forward to its further development.