NYU Tax Policy Colloquium on Herzfeld, Defining Taxes in Economics, Accounting and Law

 Yesterday was our penultimate NYU Tax Policy Colloquium session. The semester has certainly gone fast. Mindy Herzfeld presented the above paper, and here are some quick thoughts that I had in response:

1. Overview

In a number of areas, legal rules distinguish between "direct" and "indirect" taxes, and treat them quite differently. Often "direct tax" means income tax (and wealth tax if pertinent) in particular, while "indirect tax" describes VATs, excise taxes, and so forth. The paper looks at 4 distinct areas in which some version of this happens:

a) US Constitution - Direct taxes, unlike indirect taxes, must be apportioned. As we discussed earlier in the semester, in connection with our discussing this paper by Jake Brooks and David Gamage), in 1895 Pollock held (contradicting prior authority) that income taxes are direct taxes for this purpose, although the 16th Amendment then intervened with respect to the need for apportionment. 

b) International trade law - VATs, as indirect taxes, can exempt exports without violating WTO rules. Income taxes, by contrast, as direct taxes, have been held to violate the WTO if they have export subsidies. The US learned this the hard way by losing WTO litigation with respect to its DISC rules, then FSC, then ETI.

c) Foreign tax credits - Under US federal income tax law, foreign tax credits are allowed only for other income taxes (or certain levies that are deemed to be in lieu of income taxes) that look sufficiently like out income tax. Recent proposed regulations make it clear that digital services taxes (DSTs) are not creditable.

d) Financial accounting - Income taxes get detailed reporting in financial statements. Other taxes show up, at best, in a single line entry for "Other Taxes," or may even be amalgamated with other above-the-line items without separate reporting - e.g., being included in the cost of goods sold.

2. Overall takeaways

I personally find "direct tax" vs. "indirect tax" too underspecified and formalistic to be of much use analytically. However, once we recognize that the distinction often cashes out as things like income taxes vs. things like VATs, one can start drawing meaningful contrasts. Consider the following polar distinctions between pure or typical income taxes, and pure or typical VATs and other consumption taxes.

a) Is the normal return to waiting being taxed? In a pure income tax the answer is yes; in a pure consumption tax (such as a VAT) the answer is no. In practice, however, things may be far more blurred. For example, the existing US income tax has provisions such as the realization requirement, expensing for various capital outlays, and lower capital gains than ordinary income rates, that may make it more consumption tax-like. For that matter, a consumption tax that uses the "traditional IRA" methodology (deduct savings today, tax them when used to fund consumption) may place a tax burden on saving if the later tax rate is higher than the earlier one.

b) Is the tax base domestic production, or domestic consumption? An income tax is typically origin-based, while VATs are generally destination-based. Thus, in theory the income tax reaches all domestic production, and a VAT all domestic consumption. Note, however, that in practice each may exempt or tax-favor a significant portion of its presumed base. Plus, actual income taxes may have destination-based features. Consider, for example, the use of sales-based formulary apportionment in US state income taxes, along with US source rules for royalties that rely on where the items is used.

c) Who is the tax "on"? - We commonly think of business income taxes as being "on" the business itself. But when they collect and remit VATs or retail sales taxes (RSTs), we think of these taxes as being "on" the consumer. This distinction, however, lies somewhere in between being an economic incidence theory and being based on the taxes' formal features.

d) Administrative timing - Income taxes often allow deferral for economically accrued income (and for such income that has accrued for financial accounting purposes). They also often have deferred resolution of reporting and audit issues. These tend not to be issues (at least, for the latter, to the same degree) with respect to consumption taxes such as VATs.

The paper performs a service by laying out such issues as these across the various contexts. But a question of interest is what main takeaways one should derive from the comparisons.

One takeaway might be that problems arise when these distinctions prove to be less binary in practice than they are in theory. However, this view was more prominent in earlier drafts of the paper than it is in the current one.

3. U.S. constitutional law

Here "direct tax" has a distinctive historical meaning. The phrase was inserted into the Constitution's text at a time when modern income taxes (even in the sense of what the US had during the Civil War) did not yet exist. Moreover, while in other areas (such as WTO law) it's clear that an income tax is viewed as a direct tax, this apparently was not the case in US constitutional law until the Supreme Court in 1895, pursuant to its right-wing members' ideological mission of combating socialism, so held in not incredibly good faith.

Today's right-wing Supreme Court majority will undoubtedly likewise so hold, if an unapportioned wealth tax comes before them. But many who are more on the left disagree, and much of the dispute comes down to the question of how much precedential weight (if any) one should give Pollock.

In my view, however one comes out in the end, this is a rather hard issue to relate to the other three that the paper discusses. The distinguishing point is that it involves the interpretation of cryptic eighteenth century text that predated the modern context.

4. Trade

Again, the WTO bans export subsidies in direct taxes such as income taxes, but it allows indirect taxes (such as VATs) to tax imports and exempt exports.

The historical and political grounds for this distinction may pre-date the WTO, reflecting earlier stages such as the GATT. But in economic models, based on idealized models of an income tax on the one hand and a VAT on the other, it makes perfect sense.

Suppose that we have, on the one hand, an origin-based income tax that reaches all domestic production, and on the other hand a VAT that reaches all domestic consumption. Then an export subsidy in the income tax, equivalently with an import tariff, violates free trade principles and ought to be struck down under a global free trade agreement. By contrast, the VAT's taxing the full value of imports and exempting exports does not violate free trade principles and should not be struck down. Without providing the full analysis here, it is simply tautologically correct under the assumed premises.

So what's the problem with drawing these distinctions in WTO law? Insofar as the analysis is correct, US resentment of the decisions striking down DISC, FSC, and ETI, and associated griping about why the VATs (which everyone else has) do better, is simply misguided.

But in fact there are two big problems with accepting the binary distinction in current WTO law:

a) The models are too simple - As noted above, it's not quite true that actual income taxes reach all domestic production and only that, whereas VATs reach all and only domestic consumption.

Starting with the income tax, suppose various income tax preferences either wholly or partially exempt production for the domestic market. (E.g., consider tax preferences for real estate). Then tax-favoring exports may not, on balance, cause the latter to be, in the aggregate, more favorably taxed than the former. Indeed, it's conceivable that export subsidies might in practice push towards leveling the playing field.

Turning to the VAT, suppose (as is generally true of existing VATs) that it exempts or effectively excludes lots of domestic consumption, and that this tends mainly to benefit domestic production. Examples might include food, financial services, the informal economy, and small businesses that are exempted. Collecting the full VAT on imports, when home production often escapes being taxed, might end up being a bit tariff-like.

b) Hybrid instruments - Recent decades have seen the birth of the proposals that occupy space in between existing income taxes and VATs. For example, the Hall-Rabushka flat tax, David Bradford's X-tax, and more recently the DBCFT put businesses on what is effectively a VAT plus wage deductions. Most legal experts view this as making them direct taxes for WTO purposes, with the consequence that they might be viewed as containing illegal export subsidies. But proponents say: Why should they be treated any differently than VATs, when the economics are the same? (Whether they are actually the same depends on how one analyzes the wage deduction, but here they note that it could be placed in a separate tax instrument from the VAT, avoiding any WTO violation even if the overall economics were now in fact entirely the same).

The conclusions I would reach are as follows. First, it's a serious question whether the application of the binary distinction to actual income taxes and VATs is more problematic than the textbook economic analysis would suggest. This is partly an empirical question: how consequential are the real world deviations from the pure models?

As to the hybrid instruments, I (and most others) tend to accept the proponents' argument that finding WTO violations for what are essentially formalistic reasons does not make a great deal of sense. But even if this is so, the rise of hybrid instruments may present real challenges for the binary WTO distinction that could get worse if and as further variants in the hybrid space emerge.

5. Foreign tax creditability

The US rules limiting foreign tax credits (FTCs) to income taxes in the US sense can lead to some odd results. For example:

--A Haig-Simons income tax, while a truer income tax than ours, would not be creditable because it would not use realization.

--Some years ago, Bolivia considered replacing its corporate income tax with a flat tax, but then backed off when it learned that this tax, unlike the instrument that it was replacing, would be non-creditable. Flat tax proponents responded, with understandable indignation, that it was hard to see why the US would seek to discourage such a tax change, rather than regarding it as an internal Bolivian decision that had no adverse impact on US interests. On the other hand, it is true that VATs (which a flat tax resembles, as it is simply a VAT plus wage deduction) are not creditable, a feature of the US rules that has not attracted significant complaint.

The big current issue in this area is the recently issued proposed regulations that make it clear that DSTs paid by US companies abroad do not get the FTC. They might already have been non-creditable because they are gross, not net, taxes, but the proposed regs also take aim at them for not following traditional international norms with respect to tax jurisdiction.

My own view about this is twofold. First, I think the US rules are logically wrong (or inconsistent) in thus distinguishing between DSTs and traditional source-based income taxes (or more specifically taxes in lieu thereof, such as gross withholding taxes). But second, the denial is at least arguably in the US national self-interest.

The logic supporting DST creditability - As Wei Cui has argued (and as I largely concurred here), multinationals often have limited marginal costs of using their digital platforms in a given country. Thus, a tax on their gross advertising revenues from exploiting a particular national consumer market may not be all that different from their net revenues. Indeed, a gross measure might even come closer in practice to capturing the true net than a purported net measure in which they got to play lots of fun and games.

Additionally, DSTs generally have a very traditional corporate income tax aim. They seek to impose some tax on income that the multinationals derive from tapping their consumer markets. This truly is domestic source income if one adopts a destination-based rather than an origin-based view of source (and actual income taxes do indeed do this to a degree), yet the companies have been able to avoid paying it because they don't need physical presence, permanent establishments in the jurisdiction, etc. In that sense, they are in substance traditional / playing defense rather than offense, however novel their means of doing this may be. So there is a powerful logic in favor of the view that they are merely in lieu of traditional corporate income taxes, and hence as a matter of logical consistency ought to be creditable just like the traditional instruments that they, in effect, aim to revivify.

National welfare grounds for denying FTC creditability - But this is not to say that the US ought to credit them. As I have written in various formats (for example, here), it generally does not make sense, as a matter of unilateral national self-interest, to offer resident companies 100% reimbursement of their foreign taxes paid. And, while this can potentially make sense reciprocally or multilaterally (i.e., we'll credit your income taxes if you credit ours), the US is not in the position of imposing DSTs that it would like other countries to credit.

More broadly, however, why should we generally limit creditability to foreign income taxes? As noted, for example, here, the policy reasons for doing this are, at best, far from obvious. Perhaps the best and only reason one can come up with is that this is the deal countries implicitly made - insofar as there ever was a deal to begin with. That is, the reciprocal practice in fact applied to income taxes and nothing else (albeit, not really to income taxes given that one could exempt foreign source income instead of crediting the source-based taxes).

6. Financial accounting

Whether or not financial accounting should provide for greater disclosure than it does with regard to non-income taxes, it is easy to understand why it requires the more extensive income tax disclosure than it does. Reason 1 is that deferral for taxable income relative to accounting income means that there are deferred liabilities to consider - not similarly an issue in the ordinary course for, say, VATs. Reason 2 is that income tax reporting positions that are uncertain to prevail may commonly linger unresolved for a significant time - again, less of an issue in other contexts (where disclosure may be required in special cases where there is such an issue).

While those two reasons are surely the operative ones in practice, there's arguably a third reason why we might be glad that financial accounting provides as much information about companies' income tax liabilities as it does (which is not to say that it currently provides enough, or does the analysis entirely right). Because the income tax planning space is so capacious, especially internationally, capital markets may derive valuable information from disclosure of how much in such taxes given companies are paying relative to their reported financial accounting income. For example, investors might like companies that are very aggressive. Or, they might alternatively fear that such companies are also aggressively managing reported earnings and/or engaging in managerial diversion of profits. Either way, they may learn more from reporting about income tax liability than from that concerning other taxes that are less planning-susceptible.

Is this binary distinction also susceptible to breaking down in practice? Maybe so. Had the US adopted the DBCFT in lieu of the existing corporate income tax, back when the House Republican leadership was briefly contemplating doing such a thing, this would have given the financial accounting authorities an interesting classification challenge. They were probably as relieved as anyone not to have to face it. But in any event, the issues that control how best to draw lines in financial accounting appear to me to be quite distinct from those that are most relevant, say, for WTO purposes.

In the end, therefore, these various disputes may all end up proceeding on separate tracks, even if to a degree they employ common language and concepts.