Yesterday we had our final NYU Tax Policy Colloquium of the year, with Alan Auerbach to discuss his paper (coauthored by William Gale), Tax Policy Design With Low Interest Rates.
Before getting to it, a couple of notes about next year. NYU Law School is now shifting permanently to 13-week semesters. Also, I will be repeating this year’s experiment of teaching the colloquium solo, and hence having public sessions only every other week. But next year, rather than having an extra public session, I will have an extra “private” session with just the enrolled students. So there will be just 6 public sessions. These will almost certainly still be on Tuesday afternoons, running from early September through the end of November.
The sessions will still be hybrid (i.e., in-person plus Zoom), unless the law school moves away from use of the Zoom component. But any institutional decision about that lies in the future. Also, while the sessions ran from 2:15 to 4:15 pm this year, I will move them back to something like 4 to 6 pm (depending on how the law school sets up its scheduling blocks), if the state of COVID permits small group dinners afterwards by that time.
Anyway, on to the very interesting paper.
1. Background / overview – The paper focuses on the implications of two important facts: (1) the absolute decline in recent years of interest rates on both safe assets and riskier assets, and (2) the decline of the former relative to the latter. So we are now living in a low interest rate environment, which appear likely to continue although of course one never knows for sure.
One could reasonably say that there have been 3 distinct eras in this sense, in the history of the US federal income tax: a low interest rate here from its inception until some time in the 1960s, then a high interest rate era from then through somewhere in the 1990s, and since then a low interest rate era. These eras have had an enormous effect on how people contemporaneously think about the issues. Thus, for example, in Era 1 Henry Simons regarded the deferral of unrealized gain as only a trivial concern (since the time value benefit from deferral was fairly low). Then in Era 2 William Andrews called the realization requirement the “Achilles heel of the income tax” (reflecting that the time value benefit from deferral was now quite high).
The paper also notes a third fact that initially seems in tension with low r (the term I will henceforth use here for the low-interest-rate phenomenon). This is the rise of capital income’s share of national income in official reports. All else equal, low r ought to lead to a lower, not higher, national income share for reported capital income. But the paper notes the main explanations in the literature for this seeming anomaly: (a) greater rents, (b) reduced worker power in wage-setting, (c) labor income’s being misreported as capital income (e.g., when it is earned through a passthrough entity), and (d) rising values for housing.
The paper then proceeds to discuss low r‘s implications in a number of areas:
2. Income versus consumption taxation – If these 2 systems differ in the abstract solely in how they treat the safe return (in theory, included by the former but excluded by the latter), then r‘s trending towards zero reduces the distinction’s significance.
That much is pretty intuitive, but the paper also discusses how low r affects the transition effect that a switch from income to consumption taxation (depending on how it is executed) can have. If we think of this as a one-time capital levy, then low r reduces its value to the government (which now will be earning low rather than high r), except insofar as it conveys a right to share in rents. But the latter right may actually gain value in a low-r environment, e.g., because future returns have greater present value (from a lower discount rate) if r is low rather than high, again all else equal.
I have questioned in past work the extent to which the capital levy is conceptually a proper part of the income to consumption tax transition. For example, one can make the switch in system without the capital levy, or else impose a capital levy without the switch in system. But there may be some tendency for the two to occur together, and its being a byproduct of “fundamental tax reform” may affect perceptions of whether it is likely to occur again.
3. Wealth taxation – The paper notes that the tax burden imposed by a wealth tax varies with r. Suppose, for example, that the wealth tax rate is in all cases 1%. If r is 5%, then this resembles a 20% capital income tax. But if r is 1%, it resembles a 100% capital income tax. Note that r has the potential to sink still lower, and perhaps even to turn negative.
A side implication here relates to Thomas Piketty’s claim that rising high-end inequality has been driven by r > g (the economy’s growth rate). Suppose r really is the driver of high-end inequality, as Piketty posits. Then seemingly low r should be solving the problem, all by itself. But if the problem is driven by rents and rising high-end labor income inequality, as I and many others find more plausible, then low r might even make things worse, e.g., by making rents more valuable as the discount rate drops.
Low r would weaken the case for a wealth tax on wholly separate grounds, if it indicated that the problem of high-end inequality was on a path to disappear on its own. But even absent that apparently false implication, the point about higher tax burdens (in capital income tax equivalent terms) if the wealth tax rate remains the same but r is low rather than high remains valid.
Why might one nonetheless favor a wealth tax, in an environment where r is low but high-end inequality is nonetheless remaining bad or even growing worse? Here are two quick thoughts:
a) In principle, one does not need a wealth tax to address the other causes if one can direct suitable tax instruments at rents, high-end labor income, rising but untaxed housing values, etc. For example, a VAT or other such instrument for rents, a labor income tax that better addressed mislabeling, and a Henry George site value tax insofar as housing’s rise reflects site value appreciation, might all be better directed. But political or administrative limitations to the use of those instruments might make a wealth tax a suitable second-best fallback.
b) Suppose high-end wealth concentration has negative externalities. Then it might be the very thing we want to address, and a wealth tax might in that sense be rationalized in Pigouvian terms as hitting the very thing of interest. But suppose, as I discuss here, that the harms result, not from wealth itself but particular uses. Then one might instead want to tax or regulate, e.g., wealth’s political impact, its use in high-end consumption that generates “consumption cascades,” its transmission to heirs if this triggers dynasty problems, and so forth. But limitations to those efforts might nonetheless support a case for using a wealth tax as a fallback,
c) Wealth taxes are presumably levied annually. Even if they are then repealed, prior years’ revenues presumably won’t be rebated, given the limited use in practice of nominal retroactivity. By contrast, some instruments with similar effects at some margins (e.g., an estate or inheritance tax) may have deferred collection that increases the political risk of their being repealed first.
4. Capital gains deferral – The paper notes that low r reduces the economic value of deferral from the non-taxation of unrealized appreciation. So we might not need to worry nearly so much about deferral, except insofar as the basis step-up at death permits taxpayers to escape ever paying tax on the gain.
This is certainly true. But one unfortunate political byproduct of low r is that it raises the value of basis step-up at death to politically powerful taxpayers, thus perhaps making its repeal even more difficult than before.
Also, even if time value considerations no longer motivate deferral to anything like the same degree as in a high-r environment, consider deferral’s other great advantage to taxpayers. It has option value regarding the tax rate that will apply when one realizes the gain. Arguably, this has gone up (from the rising political volatility of statutory capital gains rates) even as r has gone down. So lock-in today might conceivably be as great as it ever was, even if its main cause now is option value rather than reducing the liability’s present value.
5. Investment incentives – As r declines, the present value difference between expensing and economic depreciation also declines. So timing-based tax preferences for one type of business outlay relative to another may likewise lose significance. But this does not necessarily imply any reduction in Congress’s ability to direct more favorable treatment to industries that it likes. It simply requires a shift in means, e.g., to the use of special tax rates for tax-favored industries.
6. Carbon taxes – Suppose one is setting the optimal carbon tax rate, based on a computation of the present value of the expected harms. If one uses market r rather than some sort of a social discount rate to make the PV computation, then low r greatly increases the present value of the harm and thus the level at which the carbon tax ought to be set.
I am on the side of those who believe that market r does not really tell one much (at least directly) about the proper social discount rate. For example, future generations’ interests may matter as much as ours even if they have not been born yet. But market r does affect the computation of tradeoffs.
Suppose, for example, that one thinks of the carbon tax as aimed at reducing present consumption, with the consequence that more will be invested, and hence greater consumption will be possible in the future. With low r, one needs to reduce consumption more today in order to increase it (commensurately to under high r) in the future.
One thought that occurred to me in this regard is that it implies a balanced budget view of the carbon tax. Suppose, however, that the government uses the carbon tax revenues to fund greater present consumption. Then it is possible that current consumption might not decline after all. But I gather that the models generally do employ a balanced-budget estimate, e.g., with carbon tax revenues merely replacing other taxes on business activity that did not distinguish between activities based on the degree to which they increase carbon emission.
Once one is thinking in these broad terms about consumption in different periods and by different age cohorts, the needed analysis inevitably broadens. Thus, consider the following 2 arguments:
–Because future generations will (we hope) be richer than we are, their marginal utility of consumption might be lower than ours. Hence, we should leave more of the costs to them than if this were not true.
–On the other hand, suppose that, by reason of continuing technological advances, future generations have consumption opportunities that we lack. An example might be the development of new medical technologies that, while very costly, can save and extend lives ,while also improving life quality by more effectively fighting chronic pain, lost physical capacity from illness, etc. Then future generations might actually have a higher marginal utility of consumption than ours, even if they are wealthier, by reason of their having these opportunities that we lack to do great things with extra resourcs. This would push towards our leaving less of the costs to them than if this were not true.
In sum, while market r is surely a relevant input to our thinking about climate change and the acceptance of greater immediate costs today in fighting it, its level is just one piece of a much larger set of issues. But that said, we of course know beyond any possible doubt that we are doing far too little today to address climate change. Indeed, current policy cannot seriously be explained without making ample room, not just for collective action problems and myopia, but also for deliberate climate change denialism that reflects the same depravity as that which we see in a major US political party’s decision to side vehemently with the COVID virus and against humanity.