All posts by Janet Byrd

NYU Tax Policy Colloquium, week 6: Katherine Pratt’s “The Curious State of Tax Deductions for Fertility Treatment Costs”

As we reached the 42.9 percent point (6 weeks out of 14 now in the books), it was nice to have an abrupt topic shift – one of the things I like best about the colloquium – in this case, from international tax the prior two weeks, to Katherine Pratt’s The Curious State of Tax Deductions for Fertility Costs.

Narrowly speaking, this paper responds to a phenomenon we will surely be growing increasingly familiar with in coming years: a pompous, confused, biased, ignorant, and gratingly self-satisfied opinion by a Trump judge, although this individual may have higher than average judicial qualifications among the lot. The opinion at issue is Morrissey v. United States, in which an Eleventh Circuit judge purported to offer a “primer on the science of human reproduction,” sharing with lucky readers such insights as: “It is a biological fact that, unlike some lower organisms … human beings reproduce sexually …. Critically here, within the human reproductive process, the male and female bodies have different roles and purposes.”

The judge’s motive for so generously sharing these insights with us is that “the circumstances of the case – and the parties’ competing contentions” made it seem necessary to him. More specifically, he viewed the plaintiff’s contention that an unmarried gay man could claim medical deductions for assisted reproductive technology (ART) expenses (such as for egg donation and fertilized egg surrogacy) as contrary to nature, at least when backed up by (of course) statutory “plain meaning.”

Needless to say, the plain meaning is not so clear as our judge-lecturer thinks. Under Code section 213(d)(1), medical expenses include “amounts paid .. for the purpose of affecting any structure or function of the body.” I would think it clearly affects a “function of the body” for the taxpayer’s sperm to be successfully inseminated in a fertile egg that may may eventually grow into a child who is biologically his. Isn’t it a “function” of the male body, in producing sperm, to make possible one’s having biological children? That end result is certainly the evolutionarily selected function served by the male production of sperm (if we want to talk Nature).

This is not to say that “plain meaning” resolves the case in the taxpayer’s favor. Rather, it shows that other interpretive methods are needed. For example, the fact that the taxpayer would not have needed to incur ART expenses in order to have a biological child, had he been heterosexual and with a fertile female partner – a point of central importance to the judge – clearly is relevant to the analysis. But now we are engaged in a more complex statutory interpretation exercise than he seems to realize is necessary, and one as to which his tediously trying-to-seem-bemused lectures about Nature are unilluminating. The judge’s evident belief that section 213 shouldn’t be available when people don’t seek to produce children in what he thinks is the natural way – and, admittedly, thereby incur costs that could have been avoided had they used his preferred method – involves statutory interpretation via purposive analysis. That’s fair enough, methodologically speaking, but it isn’t “plain meaning.”

Anyway, one reason Pratt wrote the article, in the aftermath of Morrissey, is that the 11th Circuit opinion carelessly (to put the best face on it) misstates actual IRS practice, insofar as one can glean it from limited evidence, with respect to different types of ART-related expenses incurred by differently situated taxpayers. So a bit of clean-up crew was needed in this area, on which she’s been writing for a while, in addition to her wanting to address a set of very interesting issues around the tax and other treatment of ART expenses. With that said, let’s turn to a few of those issues.

1) Infertility and dysfertility – Even after Morrissey, there are strong grounds supporting the conclusion that section 213 medical deductions are allowable for certain ART expenses, at least in the following scenario. Say that a heterosexual married couple establishes that one of them is infertile, and they therefore pay $$ for fertility treatments of the infertile party. That appears to be deductible. There is also an IRS letter ruling (admittedly, not constituting precedent) allowing various expenses incurred in connection with egg donation to be deducted under section 213 by a woman who was unable to conceive a child using her own eggs.

Morrissey does not contradict allowing ART deductions with respect to the taxpayer’s own infertility (or that of “his spouse,” as section 213 puts it). But the taxpayer’s argument, which the court rejected, would add dysfertility to infertility as grounds for ART deductibility. As per Pratt’s paper, dysfertility “refers to individuals who cannot bear children because they are single or in a same sex relationship.”


Let’s switch from statutory interpretation to policy – since, whether Morrissey is right or wrong as to the former, rejecting it as to the latter would call for legal change. I would argue that there is a compelling case for mandatory government-provided insurance coverage for ART for dysfertility, as well as infertility.


This obviously starts from mid-conversation, so far as government-provided health insurance coverage is concerned. I state it as generally as that to extend the sphere past income tax deductions for medical expenses to the realm of Medicare, Medicaid, single-payer, the mandated scope of universal health insurance coverage, etcetera.


I’d put the starting point for the argument as follows. From behind the veil regarding one’s own particular identity within our society, one should recognize that it’s very common to have a very strong desire for a biological child, or as close to it as possible, as a joint project with one’s partner . One should further recognize that, again from behind the veil, one doesn’t know whether one will be one of the people for whom this turns out to be relatively easy – i.e., a fertile heterosexual individual with a fertile heterosexual partner. From this perspective, both infertility and dysfertility are risks – partly converted by modern technology into financial risks, insofar as one can pay a lot of money for workarounds that did not always exist – against which it is rational to want to insure.


But why mandatory government insurance? This requires a market failure that the government can address better than private insurance firms. The most obviously relevant one here is adverse selection – widely recognized now as a key reason for favoring government provision in the healthcare arena, whether it be via the current ramshackle U.S. methods, single payer, or something else. It’s very plausible that the government’s superior ability to address adverse selection with respect to dysfertility – the risk of which gets affected by knowing one’s own sexual orientation and partner preferences – creates a strong enough case to support the intervention.


The other classic issue in government vs. private insurance analyses is moral hazard. Here there is a difference as well. From the perspective of a private insurer, moral hazard is at work if people use ART coverage to do things that they wouldn’t have done if forced to bear the full freight financially. But for the government this is not so clear. To a benign government, it’s not just financial “waste” if motivated prospective parents get to have children with a particular desired relationship to themselves. (Yes, the fate of children in need of adoption is also part of the larger policy picture, but it’s not clear that shutting off other routes to raising a child is part of the optimal response to that.)


2) Beyond dysfertility? – What about ART expenses incurred, say, by heterosexual couples that are neither infertile nor dysfertile? An administrative advantage of not requiring infertility or dysfertility is that one need not demand an inquiry into either when people incurred ART expenses. But this admittedly would come at a fiscal cost that would raise issues of moral hazard, insofar as one thinks these expenses might have been incurred out of a preference for neither going through pregnancy (even when feasible) nor adopting. But I’d be inclined to say the expenses should be covered even without either infertility or dysfertility. Going through pregnancy is a pretty huge thing, with significant potential health costs, impact on one’s career and long-term earnings, etc. But I am leaving aside here questions of whether we are uneasy (e.g., for Handmaid’s Tale-type reasons) about the extent to which surrogacy becomes a go-to. These are certainly beyond the scope of my expertise.


3) Why run ART expenses through section 213? – As a long literature discusses, income tax deductions for medical expenses are a bizarre way to increase overall government health insurance coverage, relative to what it would be if one simply repealed the provision without changing anything else in the legal and policy landscape. Medical deductions come with an adjusted gross-income related deductible, followed by a marginal tax rate-related co-pay, without its being obvious why this form of government insurance is being interacted with the income tax’s general provision of ability insurance / under-diversified human capital insurance.


Even taking the use of section 213 as given, I suppose one might want to treat ART and adoption expenses in the same bucket, although to really do this right one might also want to throw in the marginal healthcare expenses that result from being pregnant, if these could be separately identified. And conceivably one would treat things in this bucket differently than other stuff, just as a well-designed health insurance schemes might vary the deductible, the co-pay, the cap if any on covered outlays, etc., based on the characteristics of particular areas.


4) The paper’s proposed solution – Rather than discussing statutory changes that might directly address ART expenses, the paper proposes modifying section 213 in more general terms, so that it focuses on “inherently medical services” and on allowing taxpayers to “restore or approximate typical human functioning.” This raises further interesting issues, although I won’t explore them here, regarding, for example, the technology-driven rise of “inherently medical” procedures that might allow people to go way above the median (a la the use of human growth hormone by someone of average height who wants to become 6’5″).

If only Descartes had met him

Sylvester, who nearly died this summer of a thymoma (along with a nearly disastrous general anesthesia episode that left him blind for 2 days) is now, many weeks post-surgery feeling well enough to meow plaintively, endlessly, and if I dare say so a tad annoyingly for food, to be followed by more food (perhaps more than his stomach will accept in so short a period).

He’s so expressive of his often churning inner states that I’m sure Descartes would have realized how wrong he was to think animals are unfeeling automata, if only he had met the little fellow.

This week’s Tax Policy Colloquium paper (by me)

This week at the NYU Tax Policy Colloquium, the paper we discussed was mine – i.e., my work-in-progress, entitled Digital Service Taxes and the Broader Shift From Determining the Source of Income to Taxing Location-Specific Rents.

It was a very fruitful session, that has inspired my making a few expansions and clarifications of the analysis in the version that I have posted.

Alas, I’m a bit handcuffed with regard to including a lot of content here. I’m not going to re-post the link to it, although you can readily find it on SSRN, as I’d rather have people read the revised version than the currently posted one. But I’m not quite ready to post that, since (1) I have some readers’ comments that I want to look at closely first, and (2) I’ll be presenting it at U Toronto Law School’s tax policy workshop in a couple of weeks, and might as well take advantage of that feedback as well.

I generally don’t discuss the colloquium sessions here, lest they lose their off-the-record status, with potential inhibiting effects on discussion. And discussing it here as if it were someone else’s paper doesn’t seem quite the thing to do either.

So why don’t I settle for posting the abstract, which at present goes something like this:


In recent decades, a number of fantastically successful, mainly American, multinational entities (MNEs) have risen to global economic hyper-prominence. While their market capitalizations and profits are high, reflecting that they earn substantial rents or quasi-rents, their aggregate global taxes are generally quite low, reflecting their ability to create stateless income.

Often, these MNEs are technology companies – but not always. Starbucks, for example, enjoys high global profits and low taxes despite its following a classic brick-and-mortar retail business model. This reflects that, like its more obviously high-tech peers, it relies on valuable intellectual property that helps it in creating both global pretax profitability and stateless income.

Such MNEs’ rise has placed substantial pressure on existing corporate income tax models. While the existing models might perhaps be significantly improved, this would still leave market countries (where the MNEs’ consumers are located) well short of being able to tax, as fully as they might like, the location-specific rents that these companies earn by interacting with their residents.

Market countries that use novel tax instruments, such as properly designed digital services taxes (DSTs) to expand their capacity to reach such location-specific rents, are not acting unreasonably, as judged within existing (and fairly lax) norms for constraining and channeling countries’ self-interested behavior. DSTs also have the potential (although whether it will be realized is uncertain) to improve, rather than worsen, global efficiency and distribution. Whether they prove permanent or merely transitional, DSTs look like harbingers of a new era in which entity-level corporate taxation rightly focuses more on immobile factors and locational rents, and less on decades-old doctrinal and semantic debates concerning the supposedly “true” source of economic income and value creation.

Talk at Reed College last week on my forthcoming literature & inequality book

As noted previously, last week, while in Portland, I gave a talk at Reed College concerning my forthcoming book, currently called Dangerous Grandiosity: Literary Perspectives on High-End Inequality Through the First Gilded Age.

I have now posted a PDF version of my slides for the talk, which you can find here.

The book is expected out in April 2020, with Anthem Press. Its title may change, however, as I gather that common practice counsels having the lead portion of the title (i.e., before the semicolon) give greater notice than it currently does of what the book actually is, and is about.

I am making final textual changes now, after having gotten anonymous reviewer letters. While ( think it’s fair to say that the reviewers were highly favorable, they did some editing changes. Worse still (as I like to put it), they made GOOD suggestions, suggesting that I mainly ought to follow them and indeed will.

I’m certainly available for talks regarding the book, especially in 2020.

Tax policy colloquium on transfer pricing regulations

Yesterday at the colloquium, Li Liu of the IMF presented her paper (co-authored with her colleague Ruud de Mooij), At a Cost: The Real Effects of Transfer Pricing Regulations. This paper finds interesting and significant results regarding the play-out when certain EU countries, between 2006 and 2014, adopted tougher and more fully specified rules regarding transfer pricing by multinational companies (MNCs). I found the results credible, both intuitively and in terms of how the paper reached them, although the authors would be the first to agree that, because this is a fairly novel research area, more work is needed to see if similar results arise using different periods and data sets.

I’ll offer comments on it here in three buckets: the model it uses, several of its main empirical results, and international tax policy takeaways.

1. THE MODEL

Suppose an MNC is deciding whether to add marginal real investment (e.g., buildings, factories, offices, or stores, whether owned or rented) in a given country where it already has some sort of physical presence. In a standard economic model, this depends on the marginal investment’s expected after-tax return (treating, let us say, the MNC as risk-neutral).

If productive inputs would be needed with respect to using the marginal real investment, then one choice the MNC faces is whether to acquire such inputs (a) in-house, or (b) from third parties. Absent tax considerations, this might be a fairly straightforward analysis. But once we have source-based taxation that makes use of transfer pricing under the arm’s length standard, things complexify a bit. Use of a cross-border affiliate permits the MNC to engage in profit-shifting, via the use of a transfer price that is either “too high” (if the source country’s tax rate exceeds that of the affiliate’s country) or “too low” if it’s the other way around.

The more aggressive the transfer price – at least, if it is too high, and one is profit-shifting OUT of, not into, the particular source jurisdiction – the greater the chance that it will be challenged by the local tax authorities. In the model, this triggers two kinds of costs: (a) the chance of a penalty, including via the ultimate selection of a transfer price less favorable than the MNC could have gotten away with had it been less aggressive, and (b) its triggering extra compliance costs, such as from mass document production or paying experts to support one’s position. So there is in principle an optimal level of aggressiveness here.

Suppose that the country initially just had a vague and general “arm’s length” requirement in its substantive law. But then it adopts more detailed substantive transfer pricing regulations (TPR) that specify particular methods, etc.

Focusing for now just on the outbound profit-shifting, the MNC may now find that the optimal level of aggressiveness in its transfer pricing has declined. So it will now expect to face a higher domestic tax liability (including from compliance costs) than before TPR was adopted. It might even switch from using in-house to third party suppliers for some of the productive inputs, but this would only mitigate the expected cost, not make things as wonderful as they were pre-TPR (given that it had previously been using affiliates).

By lowering the expected after-tax return from the marginal investment (or equivalently in the model, increasing the cost of capital), the adoption of TPR would be expected to reduce marginal real investment in the jurisdiction.

As we will see, an important further aspect of the model is that TPR might also constrain inbound profit-shifting from higher-tax jurisdictions. Even though the source country’s tax auditors are unlikely to pose angry challenges to inbound profit-shifting, which increases domestic tax revenues, I presume that the idea here is that, by specifying the applicable rules, TPR restrains what the MNC is able to do in this dimension, consistently with following the new rules. Therefore, and I’ll return to this below as it’s an important point, TPR is assumed to have what I call two-sided, not just one-sided, effects. (It would be one-sided if it only restricted outbound profit-shifting.) The data appear to confirm that this is actually happening.

With two-sided transfer pricing effects, the adoption of TPR need not be a revenue-raiser even if the MNC’s real activity choices are unaffected thereby. But with the adverse effect on the expected after-tax return from the marginal real investment (which might be increased by reduced ability to use it for inbound profit-shifting from higher-tax jurisdictions), one would expect the rate of new real investment in the jurisdiction by the MNC to decline. The paper’s main, but not only, research question is: Does it indeed decline, and if so, then by how much?

Why would the country care if marginal real investment by the MNC declines? While leaving for further discussion below the question of what the country’s full policy aims might be, the paper’s answer is that this might negatively affect positive productivity spillovers that would accrue to the benefit of the country’s residents.

This answer appears to invite (although it does not require) thinking in terms of a conventional two-factor production model (capital plus labor) in which real investment increases local workers’ productivity, and thereby their wages. But often talk of inbound MNC investment focuses on potential knowledge spillovers. More on that distinction shortly.

In terms of thinking about the model’s real-world application, I would add or emphasize the point that often, the relevant MNCs are earning global rents or quasi-rents (from now on, I’ll just say “rents,” for concision) by reason of their valuable intellectual property (IP). Given that the IP already exists, marginal real investment in the jurisdiction is likely only to earn a normal or routine return, at least in terms of its own productive contribution to the process.

Consider, for example, a pharma company producing one of its hot new products at a production facility in a tax haven. The value lies in the patent, and likely not much (if at all) in any of the production facility’s (or its managers’ and workers’) distinctive attributes. Thus, in a true arm’s length negotiation between the pharma company and a third party producer, the latter would likely only be able to capture a normal or routine return.

Ditto for Starbucks operating coffee shops in a given country (which I discuss a bit in my new paper that is nominally or partly about digital service taxes). Whatever large profits the Starbucks global group may be able to rake in by reason of its having, say, a shop every 50 feet (it can almost seem) in London, the value isn’t being added by the local drudges who manage and work in the facility. If Starbucks mainly used franchising in lieu of company-owned stores (and could do this without hurting the brand), you can bet they would extract what they could from the third party franchisees, leaving the latter with only normal or routine returns.

But once we have transfer pricing between affiliated entities, and in the absence of any global consensus that MNC profits should be taxed purely in the true production countries – which is not necessarily where the MNCs have located their IP for tax purposes! – the countries in which these routine production facilities have been placed may have an opportunity to capture some of the rents, under the (probably false) premise that the MNC would have had to share these with the local affiliate, despite its likely quite modest contribution to true value and profitability.

Back to the model and the MNC’s marginal real investment decision: adding such investment in the country might often reasonably be viewed as (a) offering only a modest pretax boost to profitability, since the real assets are likely to earn just a normal return so far as their contribution is concerned, (b) allowing the MNC through aggressive transfer pricing – like Starbucks’ in the UK – to claim zero or even negative taxable income in the country, if it likes, and (c) allowing the country, if it more assertively takes on the transfer pricing challenge, to use it as a vehicle for taxing some of the MNC’s global rents.

With TPR allowing the government to do some of this that is tied to the level of the MNC’s real investment in-country, I think it can usefully be thought of as akin to the case in which local tax profits depend partly on the use of property-based formulary apportionment (FA). Property-based FA operates somewhat like a property tax, except that the local percentage of the MNC’s global property is then applied to global profits that include rents, wherever one thinks they were “actually” created or realized.

Just like TPR as modeled in the paper, property-based FA is two-sided. It can shift profits either into the country or out of it, compared to a method in which true productive contributions were being taxed in the places where they “truly” happened. And TPR as modeled in the paper, just like property-based FA, can mean that inbound real investment carries a property tax-like domestic tax price in the form of having greater domestic tax profits that are not necessarily linked to the marginal real investment’s true (and presumably merely normal or routine) contribution to the MNC’s global profits.

2. MAIN EMPIRICAL RESULTS

The paper is based on studying EU data from the years 2006-2014. During this period, 6 countries introduced transfer pricing regulations (TPR) within the paper’s criteria. All of these countries were fairly small: Bosnia & Herzevogina, Finland, Greece, Luxembourg, Norway, and Slovenia. This raises questions regarding the findings’ more general applicability when larger EU countries – say, the UK (for the moment), France, and Germany – or other countries that are one or both of large and non-EU make similar changes.

That said, the paper’s main empirical results include the following:

a) The average TPR adoption reduced MNCs’ marginal rate of inbound investment by 11 percent.  What is more, the MNCs appear to be investing just as much worldwide as they would have absent TPR’s adoption, suggesting that marginal investments are being shifted to other countries, not reduced. The paper further finds that purely domestic businesses do not appear to be offsetting the reduction in inbound MNC investment by picking up the slack themselves.

Comment: This lives up to the paper’s title – suggesting that TPR comes “at a cost” even if one wants to raise revenue and reduce profit-shifting. But I will return below to the question of how one might think about the relationship between marginal real investment and positive productivity spillovers, given that this may depend on knowledge inflows, as much or more as on the application of a standard capital plus labor production model.

b) TPR’s effects on marginal real investment, as well as on other things such as reported domestic profits, are much greater if the adopting countries also have thin capitalization rules (TCR) than if they don’t.

Comment: This is both plausible and interesting. It suggests that MNCs use transfer pricing and the sorts of intra-group interest flows that TCR can address as substitutes for each other. Hence, countries that want to address profit-shifting need to consider broad-ranging, integrated approaches – as to which, of course, they also need to think about the full range of likely effects.

c) TPR appears significantly to reduce reported domestic tax profits, and thereby tax revenues (under constant statutory corporate income tax rates) in low-tax countries. In high-tax countries, there was a smaller finding of reduced profits and tax revenues, but it lacked statistical significance.

Comment: An issue about the paper’s distinction between high-tax and low-tax countries concerns the small group of 6 (themselves small) EU countries that adopted TPR during the period under study. This implies just 3 countries in each group (although a given country’s status as high-tax or low-tax may have differed as between years), and one wonders also about the classification metric. Luxembourg, for example, has a reputation for being lower-tax in practice, and deliberately so, than one might have thought from merely consulting its statutory rate. And even if the other 5 countries weren’t trying as hard to be accommodating to MNCs, issues of their (possibly either low or heterogeneous) capacity to audit effectively might affect what status they really ought to have in this regard.

Explaining this result – especially given finding (d), which I’ll discuss next – appears to require placing some weight on the two-sided character of TPR changes. Low-tax countries would have in particular the prospect of revenue loss from reduced inbound profit-shifting. And that’s one of only two mechanisms to explain the finding – the other, of course, being reduced inbound investment by reason of TPR’s adoption.

If we fully accept this result, it’s potentially pretty decisive in its impact on the question of whether, from a unilateral national welfare standpoint, a country like the 6 in the study ought to push ahead and adopt TPR. Losing revenue while ALSO marginally discouraging inbound real investment does not sound like a great step forward.

d) Adopting TPR causes what the paper calls the “TPR-adjusted corporate income tax rate” to be 23% higher. This is not a finding about post-TPR effective tax rates. Rather, it concerns MNCs’ sensitivity to the statutory rate. Apparently, for such purposes it’s as if this rate has increased by 23% of its pre-TPR level, presumably given that outbound profit-shifting (if desired) is now harder to accomplish than it had been previously.

Thinking about this finding in conjunction with finding (c) suggests to me that TPR may significantly increase expected compliance costs. After all, why act as if the rate were now 23% higher if one isn’t actually going to be reporting more profits and paying higher taxes? The relationship I suggest here may not be entirely certain, but would make it easier to explain why an MNC might be reporting lower profits, paying lower taxes, and yet being more averse than previously to the unchanged statutory rate.

3. MAIN POLICY TAKEAWAYS

Two in particular occur to me:

a) The “at a cost” framing and empirical takeaway (from reduced inbound real investment) fits well into my general normative framework these days for thinking about international tax policy – which is that countries face a host of what I call “Goldilocks issues.” The little girl in the Three Bears fairytale liked her porridge best if it was “just right,” rather than too hot or too cold. Likewise, countries that are setting their policy at a number of different margins – e.g., how to tax inbound investment from MNCs that may be more tax-elastic than purely domestic businesses, how to tax outbound investment by resident MNCs, and how much outbound profit shifting by MNCs to tolerate – often don’t have a clear right answer at a given 1-or-zero pole. Rather, tradeoffs place them somewhere in the middle, although unfortunately “just right” is likely to be harder to find than it was in the fairytale for Goldilocks herself.

Revenue-raising TPR that reduces inbound real investment would very plausibly raise exactly these sorts of tradeoffs and intermediate solutions. So I welcome empirical research that may help us in evaluating tradeoffs, and indeed in understanding what they are.

b) Adopting TPR looks like it was a bad idea in the particular circumstances that the paper examines. But that might crucially depend on its two-sided character, rather than on the inherent challenges (real though they are) of responding to tax competition unilaterally.

What might be a plausible unilateral national welfare objective for a country, like the 6 that adopted TPR in the survey data? It might be: revenue-maximization from MNC taxation, adjusted for any negative effect on positive productivity spillovers.

Two considerations make the revenue maximization piece especially plausible here. First, at least for the likes of those 6 countries, the MNCs are likely to be almost 100 percent owned by foreign shareholders. Thus, taxes on the MNCs that those shareholders bear economically are likely to transfer wealth from foreign individuals to the domestic Treasury. Second, the extent to which the MNCs are earning global rents raises one’s confidence that their shareholders will bear the MNC taxes economically – a prospect that might be dim indeed if the MNC was merely earning normal returns that it could also earn elsewhere. So the presence of rents, and the country’s capacity to reach them, not just the marginal local returns to real investment, is crucial to this conclusion.

Once one is actually losing revenue from TPR, things don’t look so great from it. But this is easy to understand in the framework where we think of TPR as akin to increasing property-based FA. So the lesson for me is to find other ways of getting at rents. As my DST-et-al paper discusses (and thus, as will be a center stage issue at next week’s NYU Tax Policy Colloquium), this inquiry might lead one in very different directions. It requires that one think further about such options as (1) sales-based FA and its more sophisticated sibling, residual profit allocation, and (2) novel tax instruments such as digital service taxes or diverted profits taxes.

Off to Portland

I am flying today to Portland, OR (not ME), where tomorrow, at Reed College, I will give a talk concerning my book project on literature and high-end inequality, as described here.

The book, entitled Dangerous Grandiosity: Literary Perspectives on High-End Inequality Through the First Gilded Age, should be appearing in print and as an e-book, perhaps as soon as the first half of 2020. The publisher is Anthem Press, and I have found them both pleasant and professional to work with. All my prior books were published by university presses, or in one case a D.C. think tank press. But (if I may say so) the book’s originality, refusal to stay within a single academic silo, and lack of any precedent for exactly this sort of thing, seems to have made some of the usual suspects among editor types uneasy and uncertain about how to proceed, even when they recognized (as several did) the book’s merit and commercial promise.

I will link here to the slides for my talk at Reed once I am back in NYC next week.

Tax policy colloquium on Schanzenbach & Hoynes

Yesterday at the NYU Tax Policy Colloquium, Diane Schanzenbach presented her paper (coauthored with Hilary Hoynes), Safety Net Investments in Children. This important contribution to research regarding the recent evolution of the U.S. safety net has three main findings:

1) Providing safety net benefits to households with children, especially when they are very young, offers society a large future payoff and could thus properly be described as investment. The payoff includes measurably better health and economic outcomes for the children when they grow up. Indeed, given the positive fiscal effects down the line of these improved outcomes, the present value fiscal cost of increasing such benefits should be significantly less than the current cost.

2) Since 1990, per capita federal spending on children has been fairly flat. By way of contrast (although one could debate the comparison’s cogency), per capita spending on seniors has risen significantly (driven in the main, I presume, by rising healthcare costs).

3) Since 1990, virtually all gains in social safety net spending on children have gone to families with earnings, and to families with income above the poverty line.

The first of these three findings speaks for itself. Obviously, the partial fiscal offset to increased safety net spending to benefit children in poor households, by reason of its investment character, is not the only reason for doing it.

The second and third findings clearly reflect the two big policy pivots of the 1990s with respect to the safety net. The first was the significant expansion of the earned income tax credit (EITC) in 1993. The second was the enactment of welfare reform in 1996. This featured in particular the replacement of Aid to Families with Dependent Children (AFDC) with Temporary Aid to Needy Families (TANF). The latter not only was much smaller – especially as its block-granting to the states played out over time – but had work requirements and time-limited scope.

The net result, so far as the design and character of the safety net was designed, was a significant pivot away from what I call need-based aid and towards work-based aid. The underlying design question – should safety net spending be more need-based, or more work-based? – lies at the core of U.S. safety net policy debate. So, herewith a few words about each.

Need-based – Even without an investment rationale as to the children, the case for need-based aid rests on beneficence towards our fellow citizens. (As I’ll discuss below, however, at least as to adults among the poor one could reasonably debate the extent to which its beneficence is rightly directed.)

Although AFDC was a key mechanism for providing need-based aid until its repeal (along with Food Stamps and Medicaid), its modern descendant in policy debate is universal basic income (UBI). While the former provided income-conditioned aid, and the latter would be structured as giving everyone, poor or not, the specified amount, they can be identical in practice, since income-conditioning aid creates an implicit marginal tax rate that could instead be explicit.

To illustrate, suppose that $20,000 was the poverty line, and that we wanted to make sure that everyone gets at least $10,000. (I’m ignoring here household and child issues.) Suppose also that the income tax had a $20,000 exemption amount. An AFDC-like income-conditioned grant might provide that the grant was ratably scaled down from $10,000 to zero as one’s income increased from 0 to $20,000. This would be arithmetically identical to enacting a $10,000 UBI grant and raising the tax rate on one’s first $20,000 of income from zero to 50 percent.

The case for need-based aid rests on the view that, to meet basic and urgent needs, we should be willing to tolerate adverse income effects on labor supply (that is, reduced “incentive” to work not in the economist’s sense of marginal returns to labor, but in the urgency of needing to satisfy what might otherwise be unmet needs). In practice, it may also require tolerating high at least implicit marginal tax rates in low income ranges, if for whatever reason benefits must and will be rapidly phased out.

Work-based – The view that aid should be at least substantially work-based, rather than need-based, underlies support both for welfare reform and for the relative shift towards the EITC. It rests on claims of positive externalities and internalities with respect to poor people’s entering the job market, accompanied (for some proponents) by claims about moral desert.

These beliefs may support viewing basic aid’s adverse income effects on labor supply as pernicious, rather than as innocuous. (Such considerations aside, income effects on people’s behavior merely reflects their optimizing their choices given their circumstances.) They also may support heightened concern about high marginal rates’ substitution effects against labor supply, although this would also be relevant under a need-based view.

Absent the work-based claims, the EITC’s positive wage subsidy could only make sense as an offset to high marginal rates from phasing out income-conditioned benefits. If we view labor supply as a mere commodity choice no different than any other (e.g., chocolate versus vanilla ice cream), then one wouldn’t want to distort choice towards it, by causing people to work even when the wage on offer was below their reservation wages.

Even granting strong arguments in favor of a work-based, EITC-like approach, it constitutes anti-insurance so far as involuntary unemployment is concerned. A and B both seek  work, only A exceeds, so only A gets the wage subsidy. This feature is essentially important during economic downturns, when jobs are more scarce.

The need-based versus work-based debate is a longstanding one,  and unlikely to be settled conclusively any time soon. Moreover, the differences in view between reasonable people are often a matter of degree – i.e., how far to one side or the other should one go, rather than which view one should exclusively adopt.

The Schanzenbach-Hoynes paper does not intervene in that debate, so far as the optimal safety net treatment of adults (considered in isolation from any children in their households) is concerned. But it suggests that, even insofar as one leans towards the work-based view as to the adults, one should take a need-based, UBI type of view with respect to the children in poor households.

Things would be simpler if we could easily combine significant work-conditioning of benefits for adults with need-based aid to children who live in the same households. This, however, is not so easy, given that households tend to pool and allocate the members’ combined resources based on their own internal decision structures.

The problem is not just that adults in a poor household might “steal” the kids’ benefits for themselves. Even targeted, in-kind aid to the children may free up discretionary resources for the adults, or reduce their need-based “incentive” (in the income effects sense) to work. Insofar as one can’t succeed in combining a more work-based approach with respect to the adults in a household with a more need-based approach with respect to the children, one may have to trade off and compromise between them, leaving each sub-optimal relative to what one would have done if one could set the two sets of policies entirely separately.

How might the tradeoff be mitigated, in the direction of increasing one’s policy to run both policies at the same time despite their reaching the same households? The possibilities include at least the following:

1) Use suasion rather than incentives to encourage work.

2) Provide specific aid that is directed to the needs of working parents, such as for quality childcare.

3) Expand jobs programs. This might involve, not just job training (the efficacy of which may vary), but also actual provision or even guarantees of employment, if this can be done sufficiently well.

4) Use in-kind benefits for children that are not entirely cash-equivalent in practice. Obvious examples include SNAP (aka Food Stamps), Medicaid, and direct or indirect aid for children’s education.

Apart from its contributions to empirical knowledge, the Schanzenbach-Hoynes paper reminds us of how fecklessly short-sighted (even when not deliberately cruel) current U.S. policy is in many respects. I suppose it’s no surprise that a country that has lately been pursuing policies that appear consciously designed to increase global warming would also manifest so little concern about the investment, as well as the beneficent, rationales for seeking to improve the lives of children in poor households.

Tax policy colloquium on Zwick et al, part 3

Here and here above, I discussed the Zwick et al paper on high-end wealth concentration from yesterday’s colloquium session, as well as the relationship between a semantically correct measure and issues of particular interest (such as wealth tax revenues and negative externalities).

But while wealth’s meaning seems semantically fixed if one is counting up the dollar bills in a strongbox (meant, of course, in a broader metaphorical sense), there are instances in which one’s underlying normative or empirical concerns might more plausibly affect how one does and/or uses the computation. Here are three examples:

1) Human capital – As pretty much all the leading empirical researchers agree, even if one has an intellectual warm spot for Piketty-esque beliefs about “capital” and r > g as key drivers of rising high end inequality, one simply can’t ignore, at least in the U.S. data, labor income’s central role over the last few decades. Rising high-end wage inequality (defining “wage” to include, e.g., the economic gains of a Bezos or a Zuckerberg from their companies’ outsize success) has been the drama’s lead actor.

We can reasonably use the term “human capital” (though Piketty, as I recall, disparages it) to connote the present value of people’s expected future earnings. Given the role of labor income at the top – not all of which has been formally converted into seeming “capital income” via the sale of shares to the general public – one simply can’t ignore this component of rising high-end inequality. The really big players here can use their human capital to help them buy (and prudently rationalize to themselves buying), for example, big houses and political influence. So even if one semantically considers human capital to be outside the permissible meaning of “wealth,” it needs to be added to the analysis of how much high-end inequality has increased. I would guess that it’s a far greater factor in this regard in the U.S. in 2019 than in the 1960s, helping to explain how much things appear to have changed even if technical pure wealth measures don’t show as much growth.

2) Vertically heterogeneous returns – Let’s return now at last to vertically heterogeneous rates of return, which, as I noted in Post 1 of this 3-part series, is one of the main refinements to Piketty-Saez in the Zwick et al paper. Recall the hypothetical in which, if both the plutocrat and the peasant are earning $5 per year from a given financial instrument, we plausibly assume that the former has a $100 bond (based on a 5% assumed rate of return), while the latter has a $125 bond (based on a 4% assumed return). As I noted earlier, if the rich have “better money” (in that they earn higher rates of return), does that mean we should respond by lowering our estimate of high-end inequality?

As a purely technical matter, perhaps the answer is Yes. After all, if we think of the exercise (at least metaphorically) as asking what dollar amount we would see on the instrument of we could look inside the strongbox, $100 vs. $125 are indeed the amounts that follow from the analysis. Yet the fact that some are getting higher returns from such wealth as they have than others is surely relevant as well.

One takeaway might be that we should measure income inequality as well as wealth inequality, and regard them as complementary inputs to measuring the degree of high-end concentration. But even just within the wealth measure, it may be illuminating to ask how it could possibly be that the wealthy have higher returns (and also higher expected returns, if there are positive risk premia) on the same “thing,” i.e., wealth. Here are a few possible explanations, with a few words regarding what one might make of each of them:

a) Segmented markets – Suppose that the peasants lack the minimum investment capital, contacts, or knowledge that are needed to buy higher-yielding assets. So they are stuck with low-interest bank accounts in lieu of bond funds with internal diversification and low per-dollar management fees. In this scenario, even among assets with equal risk they can’t buy the high-yielding ones, and the plutocrats don’t touch the low-yielding ones. I think of this scenario as pushing towards the view that one should use uniform capitalization rates, rather than vertically heterogeneous ones, to get a handle on retention (as distinct from sales) value.

b) Better connections, etc., as a distinct intangible asset – Say that you can earn 5 percent, while I only earn 4 percent, because you know people who will put you onto a good thing. That is in effect a distinct intangible asset that you have and I don’t, which we might indirectly include in our balance sheets by using uniform, rather than vertically heterogeneous, capitalization rates.

c) Higher-risk assets – It’s generally agreed that wealthier people tend to choose riskier investments than poorer ones. This need not reflect distinct utility functions regarding risk – a person with a very conventional utility function will be more risk-tolerant with a large cushion than a small one.

With this in mind, suppose we revise the earlier bond hypothetical to involve a pair of $100 bonds, the plutocrat’s paying $5 and the peasant’s $4 because it’s riskier. And suppose we re-conceptualize the latter bond as equivalent to an $80 bond that likewise pays 5% (i.e., $4), plus a $20 insurance contract that pays zero if nothing bad happens (i.e., what it does is top off the $4 in bad states of the world, including those in which there’d otherwise be a huge negative return).

Suppose we further agree that both the plutocrat who buys the bond paying $5, and the peasant who buys the bond-insurance combo paying $4, are (i) choosing rationally, and (ii) getting $100 of value. We might still conclude that the plutocrat is better-off in a way not captured by crediting them with $100 of value each, via the application of vertically heterogeneous capitalization rates.

The point I have in mind here is that, while the insurance is worth $20 to the peasant, it’s worth much less to the plutocrat (and again, this may be due to divergent circumstances, rather than diverse utility functions). Can this be analogized – I’m still thinking this through, which is why I’m formulating it as a question – to the case where you rationally buy $20 of health insurance, and I rationally don’t, because my health is better? Might I be ($20?) better-off than you in this respect, because my circumstances meant I didn’t need to incur this cost, while yours meant that you did?

Tax policy colloquium, Zwick et al, part 2

I closed the prior blog post by noting a conundrum seemingly presented by the Zwick et al adjustment to top wealth shares that relates to applying capitalization heterogeneously. Again, suppose that both a person at the top of the income distribution and one much further down is observed to be reporting $5 of interest income. And suppose we have established that the former individual typically earns 5% returns, while the latter typically earns only 4% returns. We would therefore infer $100 of wealth for the former, and $125 for the latter, reducing the finding of high-end wealth concentration.

Hence we seemingly get the lesson – albeit, not one that Zwick et al are peddling; they’re simply being scientists here – that things aren’t as bad as all that. Whereas an alternative take would be: Mightn’t vertically heterogeneous returns be a part of the story around rising high-end inequality, rather than a ground for adjusting downward our estimates of it? After all, ignoring for the moment the question of why there might be vertically heterogeneous returns, one might be inclined to say, or at least snark: I like the $100 bills that earn 5% returns more than those that merely earn 4%. Please gimme some of the former, not the latter.

Before going more deeply into this particular “why” question, however, I believe that it’s useful to zoom out, for a moment, rather than in, in the sense of going meta, as follows

What is wealth, and why does it matter? – The paper starts by noting that the measurement of high-end wealth concentration is relevant to addressing “public concern over rising inequality, whether the distribution of resources is fair, and how policy ought to respond.” Might the nature of these concerns affect how we ought to measure it, or adjust / apply the definition insofar as it’s already become semantically specified without full reference to them? Let’s turn to a couple of main issues here.

1) Relevant to wealth tax revenue estimates – While I’ve certainly been interested by public debates about this, such as the recent Summers versus Zucman twitter war, as a practical matter it may not be the most important piece. After all, even if one were to assume that Senator Warren is elected president with both House and Senate majorities, I’d still place a very low subjective probability on the likelihood that the proposed wealth tax will be enacted, much less allowed to take effect.

But if it were to pass, then the Zwick et al adjustment for vertically heterogeneous returns seems  logically correct, as a matter of predicting the actual wealth out there that would be subject to the tax, assuming successful implementation in spite of both avoidance and evasion opportunities. One could put the point this way, in terms of my very simplified hypothetical:

Suppose again that both a plutocrat and a peasant (so to speak) are reporting $5/year of interest income from particular financial instruments that they have placed in strongboxes on their desks. The estimator’s job is to guess correctly what’s in the strongbox, which the authorities will get to look inside once the tax is in place. If the plutocrat earns 5 percent while the peasant earns 4 percent, this suggests that in fact the former’s strongbox contains a $100 bond, while the latter’s contains a $125 bond. Any thought that the former’s money is actually “better” because it earns a higher rate of return is ruled out semantically by the question we have asked ourselves, which is how much wealth the strongbox holds with respect to this particular financial instrument. (I’m assuming, of course, that each bond actually is worth, and would sell, for the indicated amounts despite their different returns, surely reflecting other differences between them such as in riskiness, term, or ready marketability.)

Insofar as one thinks that a wealth tax might in fact be implemented in the U.S., notwithstanding my skepticism regarding its likelihood, the question of how much a reduced revenue estimate should affect a potential proponent’s normative assessment depends on the reasons for favoring it. Reduced revenue might lower one’s assessment under a standard revenue and distributional gains versus efficiency cost assessment (unless the lower wealth subject to the tax commensurately reduces expected deadweight loss from behavioral responses). But if one views the wealth tax as akin to a pollution tax, reducing negative externalities associated with high-end wealth concentration, revenue might be a smaller part of the motivation even though still relevant.

2) Relevant because wealth distribution can have important societal effects – Wealth tax or no, we should be interested in the degree of high-end wealth concentration because it can have important societal effects.

Saez and Zucman assert in their latest piece that high-end wealth concentration undermines democratic institutions and corrodes the social contract. There’s a vast literature, in both the hard and soft social sciences further exploring how it might have concrete adverse consequences. But these will vary with the broader social context, suggesting that two societies with identical wealth distributions might have very different negative externalities (both qualitatively and quantitatively) from their shared degree of high-end wealth concentration.

Going back to the strongbox metaphor for an illustration, suppose that in each of two societies the super-rich have large amounts of cash locked up in the vaults behind their desks. But in the first society – let’s call it the 1960s U.S. – the money tends to just stay there. It’s used surprisingly little for giant, ostentatious houses, Lear jets, Caribbean islands, and so forth, and it also is not much used to buy political influence or drive propaganda campaigns in such media as there then were. Indeed, this extra cash almost mightn’t be there, so far as its discernible effects on daily of behavior are concerned – e.g., its owners don’t seem to be smoothing lifetime or dynastic consumption, so much as, in Keynes’ famous phrase, “satisfy[ing] pure miserliness.”

In the second society – let’s call it the current U.S. – the strongbox funds are being used a whole lot to pay for conspicuous consumption, conspicuous leisure, and other such Veblenesque display, as well as to fund political and media dominance (without thereby depleting high-end wealth, since using it politically results in its replenishment).

High-end inequality seems likely to cause far greater problems, of multiple kinds, in the second society than the first. Yet the two societies might conceivably have identical high-end wealth concentration, as measured within the term’s apparent semantic limits.

It’s only through the reasoning suggested by this sort of hypothetical example that I can find it at all plausible that high-end wealth concentration HASN’T in fact vastly (not just marginally) increased in the U.S. since the 1960s. Or to put it another way, if wealth concentration DID uniformly translate into the social ills that I have in mind, I would be inclined to conclude that Saez and Zucman just had to be closer to the truth than proponents of downward adjustment, however technically appealing the latter’s arguments about particular measurement questions might be. But given the lack of a precise relationship, the evidence of one’s own eyes is not dispositive here.

In this example, the definition of wealth for measurement purposes was too fixed semantically to allow for adjusting it to better track our reasons for caring about high-end wealth inequality. But there are also instances in which the purpose might influence the definition, at least insofar as we were using a particular measure to help us get a handle on the underlying social problem. I will address  this aspect in a third and final blog post concerning the Zwick et al paper, to follow shortly.

Tax policy colloquium, week 2: Eric Zwick on U.S. top wealth shares, part 1

Yesterday at the colloquium, Eric Zwick presented his article (coauthored by Matthew Smith and Owen Zidar), Top Wealth in the United States: New Estimates and Implications for Taxing the Rich. It enters the fray regarding recent Saez-Zucman work on the percentage of U.S. wealth held by the top 0.1 percent, with implications as well for evaluating the controversial Saez-Zucman revenue estimate of Senator Warren’s wealth tax proposal.

Based on a set of thoughtful adjustments that appear to me clearly motivated by a scientific desire to get it right, rather than by any ideological or other axe to grind, Zwick et al find that the top 0.1%’s wealth share is somewhat smaller, and has grown less substantially in the last 50 years, than Saez and Zucman find. But I’d say that the basic outlines of the story – rising high-end wealth inequality- don’t change in any fundamental way.

One implication that could in principle matter, however, pertains to Zwick et al’s “mechanical tax revenue calculation” for the proposed wealth tax. As they emphasize, this is not an actual revenue estimate, as it omits behavioral responses and enforcement issues. It is based on asking, if all the actual wealth above the proposal’s threshold amounts were actually taxed at the indicated rates, how much revenue would be raised? So it offers an upper bound that is likely substantially to exceed the actual revenue yield if the wealth tax were enacted. Zwick et al find this amount to be significantly lower than Saez and Zucman had concluded in their letter to Senator Warren (which was self-presented as a revenue estimate, albeit based on a mechanical calculation under their figures plus a downward adjustment for tax planning effects et al).

Certainly the most interesting, and perhaps the most important, of the ways in which Zwick et al compute top wealth shares differently than Saez-Zucman pertains to capitalization rates. Suppose (as is in fact the case here) that one is using income tax data to infer wealth holdings. To illustrate with a very simple example, suppose that IRS data show I am reporting $5/year of interest income. The wealth question is what wealth holdings (e.g., a bond that I might own) are generating this income flow.Under a capitalization approach, if we assume that my savings earn 5% a year, it must be a $100 bond, so we include that amount in an estimate of my wealth. But if we assume that I earn only 4% a year, then it’s presumed to be a $125 bond. In short, the lower the presumed interest rate, the more wealth I must have to generate the observed income flow.

I gather that prior work (such as Saez-Zucman) assumed uniform capitalization rates for lower- and higher-income individuals. But Zwick et al note evidence suggesting that higher-income individuals earn higher rates of return than lower-income individuals. So they apply heterogeneous capitalization rates. In this example (retaining or increasing its hyper-simplification for expositional purposes), we might assume that it had to reflect $125 of savings if in the hands of a non-plutocrat, but only $100 if in the hands of a plutocrat.

Unsurprisingly, this adjustment reduces the estimate of top wealth shares – although Zwick et al also make adjustments that increase it. (Again, this is a fair-minded effort, however one comes out on all of the estimating issues.)

Why would higher-wealth individuals earn higher rates of return than lower-wealth individuals? There are a number of possible reasons. Savings by the latter are very substantially placed in checking and savings accounts in banks, which offer liquidity but very low returns. The upper-tier folks are far more invested in bonds that pay higher rates but that are riskier – at a minimum, by reason of not being federally insured. (They’re also far more in the stock market, but this is in a different computational bucket as it doesn’t yield reported interest income.)  They may have greater risk tolerance, less need to sacrifice expected returns for liquidity, greater access to information and opportunities that permit them to find higher-yielding investments, and so forth.

But here’s a peculiar aspect to the adjustment that calls for rumination outside the four corners of the computational debate itself. One of the aspects of the wealthy’s superior position in our society is that they can earn greater returns. So it’s peculiar and paradoxical, however logically consistent with the underlying computational enterprise, to say: Because they can earn a higher rate of return than the rest of us, therefore we will lower our estimate of how much high-end inequality there is. Isn’t that disparity a part of the broader story, rather than an indication that things are less askew than we thought?

But I will place reflections on that question in a separate blog post, to follow shortly.