Tax policy colloquium on Auerbach's "Understanding U.S. Corporate Tax Losses"

Yesterday, with the help of Bill Gentry (Williams College/Columbia Law School) as guest commentator, we discussed co-convener Alan Auerbach's paper (with Rosanne Altshuler & two Treasury economists), "Understanding U.S. Corporate Tax Losses." The paper analyzes a puzzle: why corporations had so many more tax losses during the relatively mild recession of 2001-02 than during the previous and otherwise comparable recession ten years earlier. At the end of the paper, the puzzle is left standing, but corpses of potential explanations that failed are strewn around the landscape. E.g., the greater losses were not caused by changes in the composition of firms, whether by age or size or industrial segment, nor were they caused by particular changes in the tax rules that may have had anomalous one-time effects on reported taxable income, e.g., the dividend tax holiday or temporary bonus depreciation. An obvious potential explanation, that divergence in C corporations' economic outcomes had increased, also bites the dust. Instead, it turns out that the key change was that companies' mean rate of return dropped, leaving more of those on the lower end of the spectrum with a return below zero (i.e., a loss).

The paper leaves us with the question of whether this reduced mean rate of return pertained just to taxable income, or instead to economic income. One way to try to get at this would be to look at financial statement income for the same period. But this would require examining a smaller universe of companies, since the data set for this paper went well beyond the publicly traded sector. Plus, the book-tax gap (ratio of book income to taxable income reported by the same companies) swung wildly all over the place in the early 2000s especially.

If the reduced mean rate of return, leading to lots of losses, pertained only to taxable income, then there is no need as a policy matter to do anything about it, except that one would want to take note of the fact that companies are apparently doing lots of tax sheltering (presumably leading to overkill when economic income is unexpectedly low). If the pattern pertains to economic income, however, the upshot would be that C corporation income appears now to vary more with the stages of the business cycle than it used to. An alternative interpretation, that the economic return to the corporate sector has dropped generally, is contradicted by the steep return to profitability in 2004. (Needless to say, results for 2008 and 2009 are likely to be gruesome.)

Steeper corporate income fluctuations, in turn, would raise the possibility that the asymmetry resulting from loss non-refundability is becoming socially costlier than previously, with the implication that perhaps it needs to be rethought (e.g., longer carrybacks, interest on NOL accounts, or the return of safe harbor leasing so companies can effectively sell their unused deductions). But that in turn may not be a big problem if the business cycle means that lots of companies promptly get the losses back from subsequent profitability, in a period when low interest rates mean that the deferral of recovery doesn't cost much in present value terms.

The sentiment in the room (mine but also others') was quite unsympathetic to the current proposal to extend the carryback period for NOLs from 2 years to 5. As per an earlier post here, for existing losses this is a one-time giveaway without favorable anticipation effects. And while rationalized as stimulus, it's a bizarre form thereof in which only companies that have been losing lots of money get federal handouts in order to increase liquidity. Mightn't it be better, if we're giving handouts to (presumably cash-constrained) businesses to promote new investment, to use a selection technique other than targeting companies that have lost money recently (or at least reported tax losses)? It's like having a prize competition, to stimulate productive activity, in which only proven losers are allowed to apply.

Another point that came out forcefully in the discussion was that NOLs are in some respects a bad way to reduce the asymmetry that otherwise results from nonrefundability. The problem is their being dribbled out over time (with a 20-year carryforward). This turns loss companies into zombies that people want to keep alive, stuffing them full of profit-making activities (if the metaphor isn't too disgusting) so that the income from those new activities won't be taxed. This can lead to significant efficiency costs if the zombies otherwise ought to be put out of their misery. Better, perhaps, to say that NOLs expire in 3 years going forward if they aren't used first, but that permissible use includes selling them to someone else who can use them in the 3-year window. That would limit the zombie problem to three years going forward. Of course, applying it to preexisting losses raises the same sort of transition problem (after-the-fact betterment of incentives) as extending the carryback period to 5 years. Plus, as a move towards effective full refundability, it raises the concern about excessive ability to make use of tax shelter losses. But the basic design seems better than what we have now, assuming it could be adjusted to be comparably generous rather than more so.

The best defense I heard of the 5-year NOL proposal was that other stimulus proposals to give business tax breaks are likely to be even worse (as well as costlier over the long run). The NOL proposal's current budgetary cost would in one respect exceed its long-term cost, given that some of the losses it permits to be used today would otherwise have been used in some future year. Better a moderately bad proposal, the argument went, than something likely to be long-term costlier and no more stimulative.

One of the best things about the session was the vigorous participation from around the room, including from numerous students in the class. I personally felt sluggish at the start (perhaps from having gone to a Knicks game the night before), but the audience promptly livened things up. Students played a big role in the discussion, even though we hadn't reviewed the paper in the morning session (as we were completing a review of basic public econ ideas). This was great to see, and if it continues we will have a great semester.