Wednesday, April 13, 2016

NYU Tax Policy Colloquium, week 11: Lily Kahng's "Who Owns Human Capital?"

Yesterday at the colloquium, Lily Kahng presented the above paper.

The paper makes two main arguments.  First, recent legal developments that augment business owners’ ability to “propertize” intellectual capital created by their workers have been increasing high-end inequality.  Second, these distributional effects are further increased by the U.S. tax law’s unduly (a) favorable treatment of intellectual property to business owners, and (b) unfavorable treatment of human capital investment (and expenses of earning labor income) to workers.

Each of these arguments raises issues that are worth addressing separately.


1) The Piketty issue: Is rising high-end inequality a function of capital vs. labor, or high-earners vs. low earners, or both?

Central to the paper’s analysis is the so-called Piketty issue.  Piketty gives “r>g” a dominant role in explaining recent and predicted future increases in high-end inequality.

An alternative emphasis would focus on the role played by rising wage inequality, especially in the U.S.  Here the story is less about “capital,” and more about corporate governance, the financial sector, winner-take-all tournaments between entrepreneurial start-ups, etcetera.

The paper’s analysis of propertization fits within the Piketty story, as it is about “capital” winning relative to “labor.”  Indeed, propertization could conceivably fit within the narrative arguing that keeps r so high is not just conventional market forces but the legal and political choices that countries make.  But this requires, among other things, determining the incidence of benefit from propertization.  If businesses can increasingly claim property rights in intellectual capital created by their employees, then even if this increases the businesses’ gross revenues, the incidence as between capital and labor depends on how it affects wages.

However one ends up resolving these questions – both the relative significance of the “capital vs. labor” story and the incidence question re. benefits from propertization – I think it’s important to keep in mind the high-wage aspect.  “Workers” are heterogeneous and include some people at the very high end, and the people who hold capital are often highly compensated “workers” – whether this be owner-employees at the helm, a la Mark Zuckerberg, or hot-shot techies who really are employees rather than having controlling interests, but who get compensated with stock options.  Wage inequality and “worker” heterogeneity (plus overlap with capital owners) at a minimum complicate the story.

2)  Should propertization be expected to increase wages? 

Business owners have long been able to create patents or copyrights that give them legal ownership of intellectual capital created through someone’s labor.  But the paper notes that the effective equivalent of such ownership may arise through other mechanisms as well – for example, via covenants not to compete, nondisclosure agreements, and trade secrets laws.

Let’s focus for simplicity on a non-compete that is indeed effectively enforceable.  Say the equilibrium wage for a given job would be $X for year without the non-compete, but then the non-compete becomes standard.  Employers are getting more than previously, so their demand for labor at a salary of $X should increase.  Employees are giving more than previously (e.g., suppose they are in effect giving away expected future wages), so their supply of labor at $X should increase.  In a very simple supply-demand model, the implication is that the wage increases to something above $X.  (One could also view the “real” wage as not changing – it’s just that there’s a rearrangement of terms so that more of the value is expressed by the current year wage.)

Here the non-compete merely affects the exact terms of the bargain, so there’s no reason to think it affects one side relative to the other side.  If the non-compete also increases expected gross revenues for the business, then we’d need to know more about the supply and demand curves in order to predict how this gain is split.

Why wouldn’t this hold, and contradict the paper’s suggestion that propertization is benefiting capital relative to labor?  Well, there certainly are possible explanations.  Labor markets are distinctive in many ways.  One could have, for example, a model in which the parties differ in far-sightedness or time preference, and “capital” ends up winning.  Or one could see it as a bilateral monopoly bargaining game in which relative market power controls the outcome, and in which “capital” not only has more market power to begin with, but further increases the power imbalance via propertization.  So the paper’s story could be true, but one would need to explain why the simple standard model doesn’t hold.

Heterogeneity may also figure here.  Highly paid, highly skilled workers with market power (and stock options) may fare differently than lower-paid workers who lack distinctive skills and equity-based compensation.  So the story is most plausible for workers who are subject to propertization yet who are not the stereotypical Silicon Valley hotshots.

3)  Under propertization, is “capital” capturing returns to “labor”?  Alternatively, should we adopt a “joint venture” view of income that is jointly produced by capital owners (with or without their labor) and non-owners’ labor?

Much of the paper argues that, if the gains from propertization are being captured by capital rather than labor, this is unfair in a particular sense.  Given the top end-skewed concentration of capital ownership, one would certainly expect propertization, in this scenario, to be increasing high-end inequality.  But the paper further argues that it involves capital owners claiming labor income that was produced by workers other than the capital owners themselves.

This could be viewed as applying something like the labor theory of value (and exploitation as the term is used in Marxian economics), as distinct from a market theory of value.  I myself tend to be skeptical of viewing the fruits of production by capital plus labor as inherently belonging to one or the other of the players.  I agree that market power can influence who gets what, and my policy preferences may cause me to prefer more equal distributional outcomes to less equal ones, but the view that some portion of the value in some sense “really” was created by one party rather than the other doesn’t do much work for me normatively.

The paper argues the diversion of workers’ labor income to the owners, where it gets labeled as capital income, weighs against taxing capital income at the lower rate of the two.  From page 37: “On an intuitive level, it seems irrational and unfair that the tax disparate treatment should turn on the identity of the person who is deriving income from the labor.”  But I would note that we’re fine with applying different marginal tax rates to income, depending on to whom it accrues.

The paper also explores an alternative view under which owners (whether they work or not) and worker non-owners would be viewed as joint venturers in economic production.  It suggests that, if we can’t disaggregate their true contributions (as distinct from how much each side ends up getting to take home), this similarly supports taxing labor income and capital income at the same rate.  Leaving aside the disaggregation problem as between the two parts of income (which fits into part 2 below), I tend to take the same view of this as of the diversion issue.  That is, I care about distribution but this wouldn’t as such affect my view as to the tax rates I wanted to apply to particular individuals or types of income.  But one might reach a different conclusion than I do if one’s distributional views partake more of notions of entitlement and distributive desert (associated with the act of economic production) than mine do.


1) What’s the difference between these two types of income in theory?

One doesn’t need to have a tax system in order to want to distinguish conceptually between labor income and capital income.  For example, when Piketty discusses r > g, he needs to (and tries to) relabel capital gains that are actually better viewed as labor income.  For example, suppose Mark Zuckerberg pays $1,000 into the newly created Facebook in exchange for all of its stock, and subsequently sells the stock for $1 billion, having paid himself zero salary in the interim because that would merely be transferring funds from one pocket to another.  The salary he would have gotten at arm’s length is labor income.

Suppose he has gains above that, e.g., because he enjoyed a gigantic windfall gain (from good luck) that exceeded reasonable ex ante expectations.  Maybe we should call this capital income, but the label is only going to do so much work once we turn to the tax policy analysis.

Let’s start by taking risk out of the story.  So all we have is normal returns at the regular interest rate (without regard to risk premia ex ante, and actual risky outcomes ex post), plus extraordinary or inframarginal returns that may well represent labor income, as in the case of Zuckerberg’s unpaid salary.  But now I don’t particularly care what he would have paid himself at arm’s length – just about the distinction between the normal return and  the extra return.  (Call it rent or whatever you like – in practice, of course, it’s easiest to make sense of this part of the story if we also have risk in the picture, but I’ll bring it back in a moment.)

Suppose one favors a consumption tax, under which “capital income” is taxed at a zero rate.  Under any reasonable normative view, that support for exemption extends only to the normal rate of return – not to the extra return.  So it may be semantically convenient to say: Ah, capital income is just the normal rate – the rest is actually labor income.  But semantics aside, that exempting of the normal rate, with taxation of the extra return, is all that the reasonable arguments for consumption taxation support doing.  So if we view the semantics differently, and say that some of the extra return is “really” capital income after all, it makes no difference to the normative analysis.

Suppose we stupidly adopt yield exemption, without an arm’s length rule, for investments such as Zuckerberg’s hypothetical $1,000 stake in Facebook.  Then we get a bad result – newly minted billionaires are tax-exempt on their earnings, for reasons that make no sense distributionally even under the sort of lifetime model that leads most naturally to support for consumption taxation.

Suppose instead that we use expensing to achieve consumption taxation.  For simplicity, suppose that Zuckerberg holds the FB stock for just one year, and that the normal rate of return is 10%., while the tax rate is 40%.  Had he just earned the normal rate of return, and sold the stock in a year for $110, he would have gotten a $40 refund in Year 1 and paid $44 of tax in Year 2.  This exempts the normal return, in the sense that the present value of $40 immediately equals that of $44 in a year at a 10% discount rate.

But if he sells the stock for $1 billion in a year, he gets a $40 refund followed by $400 million of tax liability.  These rather obviously are not equal in present value at a 10% discount rate, which is why we say that expensing exempts only the normal rate of return.

There’s a long literature and debate about whether we should exempt the normal rate of return.  Most of this is about income taxation vs. consumption taxation, although there is also an extensive literature about taxing the normal rate of return at a rate between zero and the full tax rate or rates for the stuff that both systems reach. 

Wherever one comes out in the end, there clearly are significant arguments for taxing the normal rate of return – “capital income” if one likes, but without regard to the semantics of how one ends up defining the term for other purposes – at a lower rate than labor returns and extra-normal returns.  My own views on this issue are influenced by rising high-end inequality, but are not influenced by issues of joint venture or diversion of labor income to the holders of capital.

What about risk?  This blogpost is already long enough (or too long), so I’ll just say that both income taxes and consumption taxes nominally reach risky returns (other than in special cases such as the use of yield exemption), but the extent to which one actually ends up taxing risk depends on how portfolios end up being affected by the existence of this formally (whether or not substantively) mandatory insurance.  I like the insurance feature except insofar as people don’t need it because they already can and do optimize via their portfolio choices, in which case they should be able to get rid of some or all of the tax on risk and there is thus less of an issue.
2) Is it easy or hard to distinguish between capital income and labor income?  How/why might one try to do this?

It’s obviously extremely difficult, at best, under the approach of current U.S. income tax law – with partnership taxation and carried interest, sweat equity, the formal definition of capital gains, etcetera.

But in principle it’s not that hard.  Again, a well-designed consumption tax is just one example of an instrument that distinguishes between the normal return via expensing and everything else.  Another example is the dual income tax that Scandinavian countries have deployed from time to time, and that Ed Kleinbard has written extensively about.  Here the idea is that the corporate income tax base consists of normal returns plus rents, underpaid owner-employee compensation, etc.  So, if you want to tax the normal return at a lower rate than everything else, you adopt a low corporate rate, but apply that rate solely to returns that are “normal” in relation to basis.

In sum, as things go I consider it relatively easy, rather than hard, to distinguish between capital income and labor income, so long as one defines capital income as the normal return to capital.  (Perhaps the risk-free return, but that raises further issues for another day.)  That is generally the right way to define capital income, where the issue is how broadly to apply a rule treating it more favorably than labor income – whether or not it is semantically the most appealing view of “capital income” in any other context.

3) How important to high-end inequality are the “anti-INDOPCO” regulations?

The paper also discusses particular rules in the U.S. income tax that might exacerbate high-end inequality in relation to propertization and the tax treatment of human capital investment.  It notes that the INDOPCO case, decided by the U.S. Supreme Court in 1992, suggested that capitalization, in lieu of immediate deduction, might apply for a very wide range of outlays that create or enhance future value, including a whole lot of intellectual capital that has become firm-owned via propertization.  But then, for a variety of reasons that include the effects of politics, interest group lobbying, and a set of lower court defeats, the Treasury issued INDOPCO regulations (which, as has been noted, should properly be called the anti-INDOPCO regulations, that reversed much of what INDOPCO appeared to imply beyond its particular facts.  Lots of long-term value creation now gets expensing.  This is “incorrect” from an income tax standpoint, and it creates inter-asset biases if other items that are associated with long-term value get capitalized rather than being expensed.  It also might, if sufficiently widespread, but also depending on the various factors that affect economic incidence, result in r’s being higher after-tax than it would have been had the INDOPCO approach prevailed and been extended.

Did this materially increase high-end inequality?  This is an empirical question about significance and incidence.  But when only the normal return is at issue, and that is presumably fairly low in an era of relatively low nominal and real interest rates, it seems unlikely to be of first-order importance.

4)  Would more favorable treatment of workers’ human capital investments (and costs of producing labor income) increase progressivity?

The paper notes that various costs of developing human capital (e.g., education expenses) and of producing labor income often are treated more favorably when the business makes them on workers’ behalf than when the workers make them directly.  The doctrinal reason for this is that there’s no issue of personal consumption by the owner of a business that pays money for these purposes with respect to a non-owner employee.  (There is, however, a question of whether this yields taxable income to the employee, just like paying cash salary.)

In addressing how this may affect high-end inequality, it’s important to remember the high-wage workers again.  If greater deductibility were allowed at the worker level, the benefits would depend on the particular worker’s marginal tax rate and amount spent.  These may both often be higher for higher-income than for lower-income employees.

No comments: