NYU Tax Policy Colloquium, Edward Fox on banks vs. credit unions and corporate tax incidence

Yesterday, in the last session of the 2023 NYU Tax Policy Colloquium, Edward Fox co-presented his paper (co-authored by Benjamin Pyle), Who Benefits From Corporate Tax Cuts? Evidence from Banks and Credit Unions Around the TCJA. (I'm not linking it here because it's a preliminary draft that the authors plan to post when it's a bit further along.)

But just as initial background first, I've now completed 28 (!) years of running the colloquium, which for the first 25 years I always did with a co-convenor. Next year I'll be doing it solo again, and also with just 6 rather than 13 public sessions, but there is going to be a time change. Owing to NYU Law School's changing around the course scheduling blocks for reasons that are not germane here, it's highly likely that we will be meeting on Mondays, rather than Tuesdays. The time of the sessions will also change, either to 4:45-6:45 pm or to 2:35-4:35 pm, depending in large part on feedback that I am seeking from people who are on my email distribution list, or indeed from any prospective attendees. (If that includes you, please feel free to email me about your scheduling preferences.) BTW, if I had my choice, I'd still use the old time slot (going back a few years) of 4-6 pm, but that unfortunately is not available, as law schools need as a practical matter to coordinate their class meeting times across the curriculum.

Okay, enough of those not so fascinating prelims, on to the paper. It is motivated by the quest for a natural experiment that would offer insight regarding what one could broadly call the "incidence of the corporate tax" - or, more narrowly and carefully, as the paper recognizes, on the short-term incidence of the particular corporate tax cuts that Congress enacted in 2017. The natural experiment that it finds is in the small to medium-sized banking sector, in which taxable C corporations (among other taxable players) appear to compete with credit unions. The point of the comparison being: the taxable banks that meet this description are subject to the corporate tax, the rate of which was lowered in 2017 from 35% to 21% (among other changes adopted in 2017). Whereas, credit unions are federally tax-exempt, whether you look before or after 2017. So there is presumably some sort of competitive equilibrium in the sector, which then gets disrupted by the tax change for wholly exogenous reasons. One can then pursue a difference-in-difference analysis to see what happened, after duly checking on what else might have been happening at the same time (including, but not limited to, elsewhere in the TCJA), and with the look forward running only through 2019, given the pandemic's disruption of everything. With all that said, I'll offer responsive comments in 3 buckets.

1) Credit unions versus banks: The paper, while noting relevant literature, doesn't attempt to deeply theorize the competitive equilibrium between banks and credit unions, for a logical reason. However it works, why would the 2017 tax act change it? But this equilibrium is peculiar enough that I think it's worth a brief look.

Credit unions and similarly-sized banks have similar business models, except that the latter make relatively more business as opposed to consumer loans. That said, each tries to make money by holding deposits that pay comparatively low interest, and making investments (chiefly loans) that earn comparatively high interest. The spread, of course, reflects the payment and other services that they provide. Depositors typically get fixed returns on their deposits, and these are generally insured by the FDIC for banks, and the NCUA for credit unions. But the business inevitably has a variable return, depending on how well it does (including via pertinent legal changes, such as the 2017 corporate tax rate change for banks). For banks these variable returns go to shareholders. For credit unions, they at least in principle (but often not so proximately in practice) go to depositors, such as through patronage dividends or eventual liquidation claims.

One big difference, therefore, is that only the banks have shareholders. The second is that credit unions are tax-exempt, rather than facing (in the case of C corp banks) the federal income tax rules for C corporations. Let's briefly further consider each of these two differences:

     a) Why have shareholders? Normally, shareholders in a business supply a cushion that protects debt-holders against downside risk, thus benefiting the latter (who might be averse to such risk if they are seeking fixed returns). But in the case of banks and credit unions, FDIC/NCUA deposit insurance may substantially mitigate this downside risk. So, what are the shareholders "for"? Or, to put it more clearly, how do debt-holders, such as depositors, benefit from ceding both upside and downside variability to someone else, when the downside is already covered by another institutional arrangement. Or, if two otherwise identical financial institutions were offering prospective depositors otherwise identical terms, how would the institution with shareholders compensate the depositors for having taken away the variable upside risk? (Note that, with complete markets, the depositors could simply sell this upside variable return for a fixed amount reflecting its expected value, to counterparties whose preference for such variability exceeded their own - but this may be unfeasible in the credit union setting.)

Possibly, shareholders "pay" for themselves by improving corporate governance, which can be an especial problem for credit unions if the depositors are less able than shareholders to monitor management. And even with NCUA protection, credit union default might be a tough blow for depositors if that process is costly, protracted, unpredictable, etcetera. But still, it is not as obvious as it would be in the absence of deposit insurance why an equity tier is commercially useful and valuable.

For that matter, might governance concerns go both pro- and anti-credit union? For example, absent the shareholder class, might credit unions be actually and/or perceptually less likely - despite defects in managerial oversight - to play little tricks such as using hidden fees to extract $$ from depositors? 

     b) Tax exemption: Suppose that a bunch of depositors could band together (or pay an entrepreneur to assemble them) in such a way that the banking business they could fund by pooling their deposits could be either taxable or tax-exempt. Obviously, they would prefer the latter, so that the business's pretax returns, which presumably are expected to be positive (and even if negative, there are issues of nonrefundability) would not be reduced by payments to the federal income tax authorities. Thus, if being classified as a tax-exempt "credit union" was purely an election - requiring neither a lack of shareholders nor that one qualify under relevant legal criteria, such as those which generally require a "common bond" between members - then presumably all banks would happily elect to be called credit unions. So the existence of taxable banks reflects limits on (or costs to) the practical availability of the undoubted benefit of being tax-exempt rather than taxable. Credit unions have in theory a competitive advantage in attracting depositors, but practical factors, including both the (comparative) governance issue and the other legal requirements, apparently prevent them from conquering the field.

The bottom line here is that this is a complex, interesting, and distinctive business sector. How banks and credit unions ought to and do compete with each other is not pellucid, and will depend in practice on matters of institutional detail, although it is true that the 2017 tax act did not obviously change or disrupt any of this, other than via the C corporation rate changes (along with its other changes, which the paper argues tend not to have enormous, or at least direct, implications here).

2. The paper's main results: There are two principal ones. First, credit unions were paying depositors higher interest rates both before and after the 2017 act. But after the act, the gap narrowed, with banks increasing relative interest rates paid to depositors by 0.8 basis points.

Second, 74% of the banks' gain from the tax cut went to capital holders - 52% to shareholders, and 22% to depositors. The paper recognizes that it is ambiguous how we should classify the depositors, given that they are both capital holders and customers. But other players, such as borrowers (another class of customer) and employees, apparently don't gain anything. There is also some evidence of the tax cuts' leading to increased physical capital investment, such as in real estate, although with too low a confidence interval to support reporting this as a "result."

3. Interpreting the results (including as evidence regarding the broader incidence of the U.S. corporate tax): I find both of these results intriguing, but their import is not entirely clear. Starting with the first, as noted above it is challenging to try to understand the competitive equilibrium between banks and credit unions. How does one compete when offering lower interest, especially given the deposit insurance in both cases? Do the banks offer more in other ways, such as ATM availability and other payment services? If so, then why? Does this relate to governance? But, in any event, why would the banks increase the relative interest paid just because the equity tier has gained from a tax cut? True, the depositors are literally capital-holders, in that the $$ they deposit are then lent out to yield profits from the positive interest rate spread. But if they just have a fixed $$ stake that they can easily move from one bank to another at any time, without this depending on the business's distinctive features or variable returns, then why wouldn't the "customer" relationship predominate economically?

As for corporate tax incidence more generally - or even just the incidence of gain from a very particular corporate tax rate cut that was embedded in a complex system and combined with other (on balance, unfunded!) tax law changes - we clearly have a relevant datum here, pertaining to the banks within the survey. But I wonder about generalizability, even just as to the 2017 act.

One point is that (as I discussed here in re. Kim Clausing's paper earlier this semester), the presence or absence of rents seems likely to play an important role in corporate tax residence. Small to medium-sized banks seem unlikely to have significant rents - although, I suppose who knows if they hold market power in geographically segmented local markets?

A second point is that this is pretty much a snapshot story, looking only through 2019 given the pandemic's motivating curtailment of the study period. There may often be a significant gap between transition and long-term incidence, and we may have reason to care about both. As an example, suppose one believed that workers bore the incidence of the corporate tax, via its affecting investment levels that in turn affect productivity that in turn affect wages. This effect might take years to emerge. In such a case, the surprise enactment of a corporate rate cut (which itself is not exactly what happened in 2017) would be expected to give shareholders a transition gain, dissipating only over time as capital deepening lowered marginal pretax rates of return and increased wages.

As transition versus long-term incidence goes, maybe banks versus credit unions would settle out relatively fast? E.g., if it is mainly a matter of deposits moving around (at least, assuming attentive depositors!). Certainly, it might perhaps operate a bit faster than a mechanism in which big new factories are being built or a lot of fancy machines acquired and placed in service. But again, one needs a lot of knowledge about institutional detail in order to game all this out. So, while this is a valuable and informative case study in any event, the question of its broader relevance requires greater reflection and knowledge than I was able to bring to its being one of this semester's colloquium papers.

As a final word, thanks to my students, who were great and very much involved and committed at all times, and to the authors plus the public sessions' attendees. The NYU Tax Policy Colloquium is an institution that I care about, but insofar as it succeeds my efforts can only be a very small piece of the reason why.

NYU Tax Policy Colloquium, Ajay Mehrotra on the lack of a U.S. VAT

 Yesterday, at our penultimate public session for 2023, we discussed a pair of items (one already published, and one an early partial draft) by Ajay Mehrotra discussing U.S. tax history, and in particular our distinctive fiscal character.

As the papers note, the U.S. tax system, considered in isolation, is unusually progressive by peer country standards, but also unusually small. So Americans are not "over-taxed" (as some liars like to assert) by peer standards. But the broader U.S. fiscal system is unusually lacking in progressivity.

Nearly everyone but us has a national-level VAT or something quite similar, and peer countries almost invariably have more generous social spending. The US is unusually unequal, both before and after considering taxes and transfers.

How might one explain the correlation between our lack of a VAT and of generous social spending? Without more, the causal relationship between the two could be explained in any of the following ways:

a) No VAT -> less generous social spending, as the latter therefore lacks adequate funding.

b) Less generous social spending -> no VAT, as the latter therefore isn't fiscally needed.

c) Exogenous factors -> no VAT and less generous social spending.

It's plausible that each of these has some degree of truth. Be that as it may, the broader project's aim is ask why there is no US VAT, using case studies to look at different periods, and reflecting a normative as well as descriptive interest in both sides of the ledger. By using case studies, it explores questions of fortuity and path dependence, on the view that there may have been particular periods when the U.S. tax policy Overton window was potentially open to VAT (or VAT precursor) adoption, although it never ended up happening.

I see 3 topics of particular interest in responding to the papers, set forth as follows below.

1) American exceptionalism: Consider the following two alternative views of American history:

a) America is unique, due to factors X, Y, and Z. Therefore, of course we have no VAT and relatively low social spending.

b) It's all a matter of historical contingency. On several historical occasions, a VAT could have happened, but it just didn't for one very particular reason or another. For example - relying on the case studies' details - suppose TS Adams had happened to find a powerful sponsor in 1921 in his efforts to enact a proto-VAT. Or suppose that, in 1942, FDR had happened to want a national consumption tax (as Treasury Secretary Morgenthau was suggesting) rather than just a huge expansion of the income tax to help fund World War II. Or suppose that in Richard Nixon's first term, when his Administration was studying and floating proposals to enact a VAT to replace state and local property taxes in funding public education, he had managed to make a deal to this effect with Congressional Democrats. Then our country would be at a very different place fiscally today.

Evaluating these very different views is made difficult by the fact that history only happens once. But one can plausibly have a theory that mixes American exceptionalism with contingency such that, if we could run history forward 100 times (with the butterfly effect allowing it to proceed independently each time), one might predict that the US would end up with a VAT more rarely than other countries, but not necessarily zero times.

In effect, one might think of US enactment of a VAT as requiring the poker equivalent of drawing an inside straight, rather than merely having two of a kind. So perhaps we'd get a VAT 10 or 20 times out of 100, rather than 0 or (obviously) 100. Meaning that, if one accepts this view, American exceptionalism and contingency are both at work here.

That said, I have in mind the X, Y, and Z factors that help to explain why a US VAT was ex ante so unlikely (even if not impossible). I'd characterize them as follows.

X is the power of white supremacy in American history. Fueled by our history of slavery (followed by apartheid) and Native American genocide, and kept bubbling as well by our history of immigration by people who were socially coded as non-white, we have had a lack of social solidarity that undermines political support for social spending. White voters don't want to fund (as they see it) large benefits that they view as going to "them" rather than to "us."

Y is America's anti-state, anti-tax, and individualist tradition, reflecting Revolutionary War era resistance to British rule along with the frontier experience. 

Z is our having a political system with multiple choke points, biased towards inertia. Whereas in a parliamentary system the ruling majority can pretty much do what it likes, here one has the president, House, and Senate (not to mention the courts), each potentially run by a different party, and with lots of outside players and divergent views that can impede policy innovation even when all are at least nominally controlled by the same party.

A particularly keen example of Z in the case studies is the Nixon Administration's apparent interest, during the Trickster's first term, in enacting a VAT that would replace state and local property taxes in funding public education. This had a lot of underlying causes, including political and constitutional challenges to divergent levels of local funding, and the era's highly controversial disputes over busing. But its enactment would have required some sort of a deal between the Administration and the Democratic leadership in the House and Senate. Plus, it was apparently effectively vetoed early on by state and local governments' opposition, reflecting that by this time (as opposed to 25 years earlier) they had significant state and local sales taxes that they viewed as being threatened by a national VAT.

2. "BIASED" FISCAL DESIGN: The published paper argues that the US fiscal system is designed as if it had been intended to make taxes as visible and salient as possible, and benefits as invisible as possible. Hence, voters are predisposed to hate taxes and to believe that they don't purchase one anything.

Key words here are "as if ... intended." The claim is not one of intentionality but of end result. It's based on the facts that:

(a) on the tax side, the federal income tax is highly visible and salient, with April 15 filing, unaided by the likes of Ready Return. VATs, by contrast, are collected piecemeal and arguably less salient.

(b) on the social policy side, benefits are frequently delivered via tax expenditures and through private-public partnerships that make the government's role in benefit provision (e.g., for healthcare and housing subsidies) relatively invisible. 

I agree that our tax and "spending" institutions generally have this character. But an exception worth noting is the employer contribution to payroll taxes, which could hardly be less visible to workers than it is.

Also, tax expenditures' role on visibility and salience is a bit complicated. Yes, they hide the "spending" a bit. But they also hide the "tax" that funds the "spending" a bit. Thus, suppose we are comparing the the home mortgage interest deduction to a system in which the government first collected the forgone revenue through the tax system, then paid out explicit cash subsidies to existing law's beneficiaries from the deduction. This would cause, not just "spending," but also "taxes" to be optically higher than in the (by hypothesis) substantively identical system that we actually have.

The paper notes the following comparison between US and peer countries' taxes as a percentage of GDP. US taxes are typically at about 25.5% of national GDP, as compared to 33% in the UK, 38% in Germany, and 45% in France.

But our tax expenditures are 5.8% of GDP. Adding them to the 25.5%, we'd now be at 31.3%, bringing us significantly closer to the other 3, although we'd still rank below them. True, their percentages would rise as well if their tax expenditures (admittedly a fraught category to define) were added to collected taxes. But, if they use tax expenditures less than we do, which I believe to be the case, we'd still be closer after making the adjustment than before.

3. VATs and overall progressivity: Suppose one wants to make the US fiscal system more progressive. While enacting a VAT would do so, if the revenues therefrom were used to fund enhanced social spending (e.g., for education and healthcare) that sufficiently helped lower-income individuals, one could make the overall system more progressive still by using progressive taxes to fund the same thing.

Just to illustrate this, I've seen estimates suggesting that a 10%, fairly broad-based VAT would raise about $3 trillion over 10 years. By way of comparison:

--The Warren and Sanders wealth taxes - if allowed by the Supreme Court, which would require a change in the Court's membership - would raise $4 trillion over 10 years according to the proponents, and $2 trillion over 10 years according to the Tax Foundation.

--According to a 2019 article by Lily Batchelder and David Kamin, one could raise $4.5 to $5 trillion over 10 years through such tax changes, directed at the top 1 percent, as the following: higher individual and corporate rates, higher capital gains rate, realization at death, and higher rates plus lower exemption amounts under estate and gift taxes.

So it's not just about the revenue. That said, there are 2 main types of arguments for using a VAT in lieu of (or in addition to) the above alternatives. The first are arguments about efficiency and economic growth, while the latter are arguments about political feasibility.

I myself, if made the temporary fiscal tsar with some hope that my choices would persist, would add a VAT to the mix, for reasons of both (relative) efficiency and long-term political economy. But I would also revisit the existing and other proposed taxes in order to ensure that overall progressivity (giving due weight to concerns about efficiency and growth) was at the level, or achieved the set of tradeoffs, that I considered best. But, since no fiscal tsar job appears to be on offer for me at present, I don't see the need fully to decide and specify exactly what I would hypothetically do.

National Tax Association, 116th Annual Meeting

I enjoyed the National Tax Association's 116th Annual Meeting, which concluded in Denver yesterday. Due to the pandemic plus last year's hurricane threat in Miami, it was actually the first live NTA Annual Meeting since 2019 (!). This added to the pleasure this time around, as did (for me) the fact that I was awarded the Daniel M. Holland Medal for "lifetime achievement in the study of the theory and the practice of public finance."

I've posted the talk I gave at the award ceremony here. I also gave a talk about my recently posted article on medical expense deductions, and watched as my co-author, Daniel Hemel, did the honors regarding our joint work in progress, entitled Two-Level Games in International Taxation. (Not yet available for download, as we are still writing it.)

One of the more interesting presentations that I saw at NTA was one by Jim Hines, regarding an article (not yet posted in its current form) that appears to be a descendant of this one, which was then entitled "Evaluating Tax Harmonization." It's clever, ingenious, creative, and provocative - but, I believe, founded on premises that are simply wrong. That said, however, one might plausibly argue for its conclusions on very different, less "scientific," grounds.

As presented at NTA - but perhaps not to the same degree in the earlier or perhaps merely related draft that I have posted here faute de mieux - its basic premise is the following: nations that are setting their tax policies should be analogized to individuals who are making commodity choices that reflect their underlying preferences.

If you grant Jim this first step, it's all over and he wins. Since I was viewing the presentation at a conference hotel to which I had traveled from the airport, I was thinking of it in terms of one of those airport buses that you can sometimes board at, say, a resort hotel. Once you board the bus, it's done. You're going to the airport, whether you like it or not. But that doesn't mean you should board the bus to begin with.

Why (pace Jim) shouldn't we think of nations that are making policy choices in the same way that we think about individuals who are making commodity choices in a standard neoclassical economic framework? Point 1 is that, in the framework of collective choice by a given country's political actors who represent distinct interests, none of the actors has the incentive to pursue national welfare. Each is presumably pursuing her own interests or preferences or vision of national welfare. So there's a pervasive agency problem. Point 2: I think all political scientists would agree that nations' institutions for collective political choice do not generally produce the "correct" outcome in terms of their people's underlying preferences or welfare. This has been well-known for decades, if not centuries. And then Point 3: the nations themselves are not sensate. So if there are relevant "consumers" here, they are the people who reside in these nations. And not only do they not individually all share the utility functions or interests that Jim is effectively imputing to the collectivities, but they also differ in such dimensions as population. For example, the Turks and Caicos Islands' population is under 50,000, while China's exceeds 1.4 billion. (And indeed both are among the signatories of the OECD Inclusive Framework.)

So, in my view at least, the paper's basic framework is simply wrong. And, at least at the talk, I was hearing about how its analysis ostensibly reveals truths about worldwide welfare and efficiency and such things. It doesn't.

But now let's see what happens if we grant the paper's premise that nations setting, say, their corporate tax rates are like individuals making commodity choices in light of their preferences and underlying utility functions. As claimed by the paper, this does indeed lead to the firm conclusion that global tax instruments, such as the Pillar 2 global minimum tax, both (a) inefficiently reduce worldwide welfare, and (b) would need to have either very low rates (such as 4 percent) or very high ones (such as 27 percent) in order to be plausibly welfare-enhancing.

The paper very plausibly (in light of its underlying assumptions) models a given country as setting its corporate tax rate in the following way. To use my own simple notation, suppose that X% is the tax rate that it would prefer in the case of autarky. But it is subject to tax competition for mobile resources, even if countries aren't interacting strategically. Suppose then that it sets its tax rate at Y%, which is lower than X%, in light of tax competition and other countries' actual rates. 

In this scenario, there seems to be hypothetical room for collective gain, and indeed even perhaps for a Pareto improvement. Starting with just two countries, suppose that each interdependently agreed to raise its Y a bit, bringing their new Y's closer to their X's. This not only would bring each country closer to its autarkic optimum, but would be a subjective improvement even taking into account tax competition, since one's optimal Y goes up as other countries' tax rates rise.

The analysis therefore supports the view that all countries might gain if each raised its tax rate a little bit. And indeed, presumably only transaction costs and collective action problems (albeit, potentially very grave ones) stand in the way of this solution's emerging spontaneously.

But now let's turn to an actual global minimum tax like that in Pillar 2, which is set at a very intermediate 15%, rather than being very either very low or very high. Now what happens, given the heterogeneity of prior corporate rates that are presumed to reflect very heterogeneous preferences, is that some countries have to raise their rates a lot (even from 0%), while other countries don't have to raise their rates at all.

The analysis now deploys a standard, and very plausible, assumption in microeconomic rational consumer models. If I am forced to depart from my preferred choices by just a little, the welfare cost to me is likely to be very slight. By contrast, if I am forced to depart from it by a lot, the welfare cost to me is likely to be very high. Moreover, the welfare cost may grow (say) quadratically rather than just linearly. So, if the amount by which I am forced to depart from my preferred choice (say) doubles, my loss of welfare more than doubles. E.g., perhaps it quadruples.

It should now be clear why the intermediate global minimum tax rate in Pillar 2 creates so much welfare loss in the paper's model. Previously low-tax countries (again, modeled as if they were individual consumers) must accept enormous departures from setting the corporate rates that they prefer (with or without there being other low-tax countries). Meanwhile, those that already had rates of at least 15% need not change their rates at all (and indeed may now feel emboldened to increase them in the direction of X%). So the global minimum tax requires huge adverse departures from some, and no adverse departures whatsoever (and indeed, perhaps allow a small increase in welfare) for others. But with the more than linear welfare losses as one is increasingly forced away from acting as one prefers, the net welfare gain is large, compared either to doing nothing or having everyone agree to a small increase in their previous Y's.

By contrast, a global tax rate that was either very low or very high would impose either little pain on anyone, or at least some pain on everyone. (This is just the intuitive version; the paper specifically derives the alternative optimal solutions based on its model, its data, and, ahem, a second order Taylor approximation.)

Okay, so the bus is now at the airport. So, if you boarded it back at the hotel, you now must agree that Pillar 2 is terrible because the global minimum tax rate must either be very low or very high - but definitely not in the middle - in order to create overall welfare gain.

But what if you didn't board the bus to begin with, because (as I do) you rejected the basic analogy between (a) countries making collective political choices based on their politics, and (b) rational consumers making individual choices based on their preferences? Then nothing whatsoever has been convincingly demonstrated.

Still, the paper's conclusion admittedly has some intuitive force. Suppose we think that, as a general matter, the people in what we observe as low-tax countries do indeed collectively benefit (relative to alternatives) from those countries' having a low-ish rate, while those in high-tax countries collectively benefit from those countries' high-ish rates. After all, it seems clearly true that the genuinely optimal corporate tax rate (all things considered) would be higher in the US than in the Turks and Caicos Islands. After all, we're a much bigger country with more market power. So, despite all the pervasive political choice failures (at least, from an ideal standpoint) everywhere, perhaps there is some (or even a lot of) plausibility to the claim that a Pillar 2-like instrument hurts some countries a lot and others not at all. 

To be engaged in a true welfare analysis, we would still need to be looking at individuals, not collective national entities. So the above doesn't immediately translate into even a softer version of the paper's basic claims.

But one may also be intuitively inclined (despite the departure from a rigorous welfare analysis) to think of national-level actors as having moral duties to deal "fairly" with each other. E.g., suppose one subscribes to the norm of inter-nation equity. From this standpoint, one might indeed find the optimal scenario that the paper envisions more intuitively appealing than imposing large departures from preferred policies on some countries, along with no such departure at all from others. And one might conceivably object to the Pillar 2 project as something that big countries are imposing on small ones because they have greater global political power. Then one might reasonably applaud the paper's conclusions, despite rejecting its particular analysis.