Alan D. Viard
The METR computation is based on the before-tax rate of return that a hypothetical marginal investment must yield to cover its tax liability while paying (only) the minimum after-tax return demanded by investors. The METR is the difference between this marginal investment's before-tax rate of return and the required after-tax return, expressed as a fraction of the before-tax return.
Consider a $100 investment that yields a payoff next year and nothing thereafter. Assume that investors are willing to invest if, and only if, they receive a 5 percent after-tax annual return or more. Without taxes, investors undertake the project if, and only if, it yields a payoff of $105 or greater.
Under a 35 percent income tax that allows a deduction for true economic depreciation, the required minimum before-tax payoff rises to $107.69. The investor then pays $2.69 tax on $7.69 taxable income ($107.69 payoff minus $100 depreciation), leaving an after-tax payoff of $105 that provides the required 5 percent return. So, the before- tax net-of-depreciation return is 7.69 percent. The METR is 35 percent, because the difference between the before-tax and after-tax returns is 2.69 percentage points, which is 35 percent of the before- tax return.
The results are quite different under a 35 percent consumption tax that allows expensing. The minimum before-tax payoff is then $105, unchanged from the no-tax world. The investor claims a $100 expensing deduction and thereby receives a $35 tax saving, reducing the after-tax investment cost to $65. The $105 payoff is fully taxed, with no depreciation deduction, at 35 percent, leaving an after-tax payoff of $68.25. That net payoff provides the required 5 percent after-tax rate of return on the $65 net investment (68.25/65=1.05). Since the before-tax and after-tax rates of return are the same, the METR is zero; the consumption tax does not reduce the rate of return on this marginal investment.
As Neubig emphasizes, however, investments in unique intangible assets often yield returns above the required minimum. So let's consider how the 35 percent consumption tax affects a $100 investment that yields a $150 before-tax payoff and hence a 50 percent before- tax rate of return. Under the consumption tax, the after-tax investment cost falls 35 percent to $65 and the after-tax payoff falls 35 percent to $97.50. Once again, the after-tax return equals the before-tax return, in this case 50 percent (97.5/65=1.5).
Neubig observes (pp. 962-964) that this investment bears a positive tax burden, despite its unchanged rate of return. As explained below, his observation is correct, but it does not invalidate the zero-METR result.
The tax on this excess-return investment is indeed positive. Although the investor earns an unchanged 50 percent rate of return, he now does so on only $65 rather than on $100; the government effectively captures this lucrative return on the remaining $35. While expensing does give the investor a $35 up-front tax saving that he can invest elsewhere, any such investment would presumably yield only the required 5 percent return, since excess returns are not replicable. Neubig rightly notes that the gap between an investment's before-tax and after-tax rates of return does not correctly measure its tax burden when the payoffs cannot be reinvested at the same rate of return as that offered by the investment.
Fortunately, however, the zero-METR result is not based on the mistaken premise that excess-return investments face zero tax burdens. Accordingly, Neubig's rebuttal of that premise leaves the zero-METR result unscathed. Indeed, METR computations do not consider excess-return investments at all; as discussed above, they look at hypothetical marginal investments that yield only the required minimum return. The zero-METR result is based on the fact that the $105-payoff investment faces zero tax; it makes no statement, mistaken or otherwise, about the tax burden on the $150-payoff investment. The METR computation assumes, quite correctly, that the payoffs from a marginal investment can be reinvested at the marginal investment's rate of return; it makes no corresponding assumption for excess-return investments.
The METR's focus on marginal investments is appropriate because investment incentives depend on the treatment of such investments. In the above example, all investments that yield $105 or more will be undertaken with the consumption tax, just as they would have been without taxes; no investments are discouraged by the tax. Although the excess-return investments bear positive tax, the tax is not large enough to deter such investments, since investors retain their required minimum return, plus 65 percent of the excess return. When the METR is zero, investment disincentives are also zero. So, although Neubig is correct that expensing imposes positive taxes on excess-return intangible investments, he is mistaken when he denies (p. 959) that expensing of tangible investments "would level the playing field and increase economic efficiency" and when he asserts (p. 965) that conventional models "overstate the incentive effects" of expensing.
The zero-METR result arises from a basic property of expensing. With expensing, the government becomes a silent partner in the firm's investment decision, sharing the same percentage of the costs as it does of the payoffs. The determination of whether payoffs are greater than costs is unaffected by a uniform proportional reduction in both payoffs and costs.
Thanks to this silent-partner property, the zero-METR result remains fully valid under uncertainty. Contrary to Neubig's assertion (pp. 961-962), the result does not rely in any way, shape, or form on assumptions that gains and losses are equally likely or that risky investments have expected returns equal to the risk-free Treasury bill rate.
For example, consider a $100 investment with a 90 percent chance of a zero payoff (complete loss) and a 10 percent chance of a positive payoff. Suppose that risk-averse investors demand a 20 percent minimum expected return, reflecting a risk premium of 15 percentage points above the 5 percent risk-free rate. In the absence of taxes, the marginal investment, which remains the focus of the METR computation, would require a before-tax payoff of $1,200, if and when the payoff occurs, to yield the minimum expected payoff of $120. With expensing, this required payoff remains unchanged. The after-tax cost of the investment falls 35 percent to $65 and the after-tax payoff falls 35 percent to $780, if and when the payoff occurs. The expected payoff is then $78, yielding the required 20 percent expected rate of return (78/65=1.2). The METR is zero, because the before-tax expected return and the after-tax expected return on this marginal investment are both 20 percent.
The above analysis clarifies the advantages of consumption taxation over other tax systems. Unlike income taxation, consumption taxation does not discourage investment because it applies a zero effective tax rate to marginal investments. Unlike wage taxation, consumption taxation imposes significant taxes on investors who earn excess returns (and those with investments in place when the tax is introduced), thereby raising revenue from a generally affluent group without undermining incentives. In short, consumption taxation offers a better combination of efficiency and equity than alternative tax systems.
Alan D. Viard is a resident scholar at AEI.
1. Tom Neubig, "Expensed Intangibles Have a Zero Effective Tax Rate . . . NOT!" Tax Notes, Sept. 10, 2007, p. 959, Doc 2007-19296, 2007 TNT 176-51.
2 I am grateful to Alex Brill, Jason L. Saving, and Cindy Soo for helpful comments.
3. See Congressional Budget Office, Computing Effective Tax Rates on Capital Income, Dec. 2006, pp. 1-2, and U.S. Department of the Treasury, Treasury Conference on Business Taxation and Global Competitiveness: Background Paper, July 26, 2007, p. 35.
4. As with most economic statements, there are some exceptions. If firms are subject to borrowing constraints, the taxation of excess returns may depress investment by reducing the firm's cash flow. Also, under an origin-based tax system with imperfect transfer pricing, the taxation of excess returns may affect the international location of investment.
5. Neubig rejects (p. 963) the silent-partner analogy, arguing that expensing is instead an interest-free loan. Unfortunately, he uses the wrong reference point. Relative to the no- tax world, expensing makes the government a silent partner, which is why expensing leaves investment incentives unchanged from the no-tax world. Relative to the income-tax world, expensing is an interest- free loan rather than a silent-partner investment, which is why expensing does not leave investment incentives unchanged from the income-tax world (and a good thing, too).
6. For a discussion of the distributional effects of the taxation of excess returns, see R. Glenn Hubbard, "Would a Consumption Tax Favor the Rich?" in Toward Fundamental Tax Reform, eds. Alan J. Auerbach and Kevin A. Hassett (Washington: AEI Press, 2005), pp. 81-94. Of course, the taxation of excess returns and investments already in place explains why expensing does not always appeal to business firms, as Neubig previously documented in "Where's the Applause? Why Most Corporations Prefer a Lower Rate," Tax Notes, Apr. 24, 2006, p. 483, Doc 2006-7026, 2006 TNT 79-41.