Yesterday at the colloquium, Daniel Hemel presented the above-named article. We have now settled into our every-other-week, hybrid format, and numerous friends from outside NYC dropped by via Zoom, although most of them didn’t stay all the way through. I hope the audio is good enough that they can hear speakers from around the room during the discussion, although they can’t see them on the video feed. I probably wouldn’t stay all the way through under those conditions either, but it would certainly be great to hear from them if they have comments.
Anyway, on to the paper. My thoughts about the issues it raises can be grouped into two main headings: how tax lawyers should or do use economic theory, and particular issues discussed in the paper. While Part 1 appears directly below, I will reserve Part 2 for a separate blog entry.
1) Lawyers and Economic Theory
The paper focuses on the existence of a new(ish) economic literature out there, commonly called the New Dynamic Public Finance (NDPF), that has been flourishing for a couple of decades in the economics journals, while almost never being cited in the law reviews. (Okay, the one prominent exception to this, as it notes up front, is Daniel Shaviro, “Beyond the Pro-Consumption Tax Consensus,” 60 Stanford Law Review 745 (2007).)
The paper’s central message is that tax (and other) lawyers with academic or policy interests should read and use the NDPF literature, although it is not super user-friendly in form (e.g., highly technical models with lots of math), due to the important insights they can derive from it.
But here is an alternative framing of the same central message: In evaluating tax and other legal policy issues, it’s important to think about such things as:
1) the fact that people’s opportunities to earn $$ in the market can change unpredictability,
2) the effects that people’s expectations regarding future government policy can have on their behavior, and
3) the information that policymakers can derive from the year-to-year details of people’s earnings flows, consumption, and saving or dissaving.
The NDPF framing is catchier and more salient. However, its effect on a legal audience’s reception of the message may be complicated by a widely-known (if little-discussed) sociological fact about the tax policy world: that is, the fact, that economists as a group have more prestige within it than lawyers as a group.
By reason of this fact, lawyers who are eager to emulate economists at all costs will be excited to hear about this literature. But from others, such as lawyers who are uneasy or insecure about the economists’ reign, it might provoke hostility or resistance. These folks might therefore miss the intellectual payoff that the article actually offers them.
With either framing, a key takeaway is that a prominent branch of the public economics literature that has deeply influenced academic tax lawyers (and perhaps policymakers) for decades – the optimal income tax (OIT) literature that was founded by James Mirrlees’ classic 1971 article – has been revised or even (in its earliest forms) refuted in certain key respects, with the consequence that certain familiar conclusions that commonly are derived from it turn to be mistaken.
As a result, here is a misapprehension that prospective readers could derive from the article’s chosen framing: Rather than saying: “Lawyers must follow abstruse economic models in order to better grasp important policy issues,” it is actually saying: “Lawyers should free themselves from too narrowly and literalistically following economic models that inevitably are stripped-down and simplified relative to the reality that they seek to represent. Instead, they should embrace more complex and multifaceted, albeit inevitably open-ended and ambiguous, ways of thinking about various important issues.”
Going back to the three points that I listed above in the alternative framing, one should not think of them as being only in the NDPF literature and nowhere else. For examples from the pre- or non-NDPF OIT literature, and/or relevant legal literature, of considering their importance, consider the following:
–In the OIT literature, there is a thread (which I associate with an article by Hal Varian) that looks at the income tax (i.e., wage tax) as offering risk insurance for under-diversified human capital, which is subject to stochastic shocks. This is the point about earnings opportunities changing unpredictably, although this branch of the OIT literature does not (to my knowledge) look at the information to be gleaned from year-to-year earnings changes, a factor that is emphasized in some NDPF literature.
–The classic time consistency problem in tax and other policy was well-known in prior literature E.g., in principle there are huge efficiency gains to be derived from encouraging people to invest, then expropriating their holdings and promising never to do it again. But this can utterly break down not, just because ex post the promise may prove less than credible, but also because ex ante people may anticipate its being done. NDPF merely adds to this a richer and more varied inquiry into how expectations regarding possible future policies might play out.
–There is a huge legal and economic tax policy literature, to which I along with many others have contributed, concerning the effects of tax deferral as creating both uncertainty and optionality, if the tax rate in the year of realization might be different from that applying today. If all of us were therefore doing NDPF without realizing it, then I am reminded of the line in Moliere about the character who is thrilled to learn that he has been speaking prose all his life.
–Again, the paper kindly cites my 2007 Stanford article that actually mentions NDPF. But, as it happens, in writing that article I had already found my way to my main conclusions before learning about NDPF, and adding it to the article’s back end, from feedback at a conference where I presented an early draft.
That article sought to show, and then engage with, the point that certain simple OIT models that were in widespread use among lawyers and economists led straightforwardly, and indeed ineluctably, to the linked conclusions that (1) the tax system should employ lifetime income averaging, and (2) a consumption tax is superior to an income tax. Under these models, your welfare and marginal utility depend purely on the present value of your lifetime earnings, which, with the aid of perfect rationality and complete capital markets, you are presumed to deploy such that you choose the lifetime consumption stream that has the greatest subjective value to you (and presumably, equalized marginal utility for the last bit of consumption in each period).
In short, that article first sought to explain why an “ideal consumption tax” is such a slam-dunk intellectual winner over an “ideal income tax” within the standard OIT model’s contours. But then it moved on to (a) why the model’s simplifying departures from reality should not be forgotten, and (b) how a fuller and more realistic view ends up defeating the presumed takeaways (or at least their certainty) regarding both lifetime income averaging and the ostensible superiority of consumption tax over income taxation.
(BTW, this heresy drew a stern, albeit amiable, written response from two friends and colleagues in the biz, also appearing in Stanford, to the effect that I was all wrong in backing off as I did the standard OIT conclusions. I believe, although I suppose I would, that, when one looks back at 2007 from the perspective of 2021, my side of the debate comes off pretty well, and indeed has been decidedly favored by the movement of the field since then.)
Anyway, back to the point after this perhaps self-indulgent detour. In that paper, I got where I was going initially without NDPF. I focused mainly on such issues as incomplete capital markets – which impede, for example, borrowing against one’s reasonable expected future earnings – and limited rationality. NDPF, when I found out about it, was icing on the cake, but I had already gotten the most of the way there without it. This arguably weighs against viewing NDPF as such as being vital to the Hemel paper’s main takeaways, although it weighs in favor of viewing the 3 big issues (as noted above) that the paper foregrounds as really important.
Whatever the framing, Hemel’s paper and NDPF are above all about the relevance of time. Mirrlees’ classic 1971 set-up for balancing efficiency against distributional considerations in the structure of an optimal income tax expressly leaves time out. It’s about policy choice in a snapshot moment where people’s utility functions, capacity to earn $$ through labor supply, and labor supply elasticity are fixed. You don’t develop your “ability” in that model – it’s just there, as a random draw from the box – nor can you save given that there’s no other period.
Subsequent work in the classic OIT tradition then added time to the framework. For example, in Atkinson-Stiglitz (as applied to present vs. future consumption) and Chamley-Judd – often both commonly cited in the income vs. consumption tax debate – time is there all right, and it plays a central analytical role. But time is effectively uniform or flat. All periods are presumptively the same; they just come one after another. Thus, long-term or even infinite-horizon present value comparisons are king.
For example, it simply may not matter when within your lifespan given $$ were earned (except insofar as this affects present value) or when you or your heirs choose to consume it. Perfect capital markets shift $$ as needed between periods. And the infinite horizon perspective means that there is no difference between the government’s being committed to pay, say, $1 today or its present value equivalent in 5,000 years.
What NDPF and similarly minded work add is what one might call differentiated, rather than flat or continuous, time.
By ignoring per-period information, classic OIT effectively throws out valuable information. This made perfect sense as it was developing, as a strategy to simplify the issues being considered so that they would be more analytically tractable at a first cut. But to stick to that model even once its insights have been developed and duly absorbed, and to rule out the sorts of issues and information that NDPF emphasizes, is to handicap oneself for no good reason. And again, it can lead to one’s asserting dubious conclusions with undue self-confidence.
In sum, NDPF doesn’t tell lawyers: The economists used to be saying A, B, and C, but now they’re saying D, E, and F. Rather, it offers them a more complicated picture (or choice between alternative pictures) in which the issues are more interesting and indeterminate than they had seemed to be before. This may make things more challenging, but it is also liberating.