Wednesday, March 12, 2014

NYU Tax Policy Colloquium, week 7: Stephanie Sikes' "Cross Country Evidence on the Relation Between Capital Gains Taxes, Risk, and Expected Returns"

One of our aims in the Tax Policy Colloquium is to be interdisciplinary.  In addition to law professors plus several economists, we try to invite people from several other disciplines as well.  For some years now we've annually had a paper by an accounting professor, and it's also nice when we learn of suitable on-topic papers by political scientists or philosophers. We've had both in the past though not regularly, and I hope to again in the future.

We also often have papers by tax practitioners.  Indeed, we ought to do this annually, but sometimes I fall short on the invitation front because, while I know a number of practitioners who periodically write papers that I'd love to include, it's often hard to know sufficiently far in advance who has current plans to write such a paper in the right time frame for us.

Anyway, what prompted this opening digression now is the fact that yesterday Stephanie Sikes, from the Accounting Department at Wharton, presented "Cross Country Evidence on the Relation Between Capital Gains Taxes, Risk, and Expected Returns."  This is an empirical study, following up on a previous theoretical paper, that finds evidence against the tax capitalization scenario in which raising capital gains tax rates would tend to lower prices for capital assets, such as newly issued corporate stock.

The logic behind the standard tax capitalization scenario is that, if I'd have to pay higher taxes upon selling an appreciated asset, that reduces the amount I'd be willing to pay for it.  Note that the capital gains tax rate does not apply to all capital income - e.g., it has no bearing on the tax rate for interest income - so one might expect taxpayers to demand equal after-tax returns across differently taxed assets.

Suppose that we are thinking of raising the capital gains rate - say, for distributional reasons.  If the standard scenario does indeed hold, that scenario may reasonably be advanced in opposition to the rate increase, on the ground that lowering issue prices for new securities and raising the required pretax return (i.e., the hurdle rate) will reduce subsequent investment and growth.  Clearly a relevant argument, though not dispositive standing by itself given the multiple effects, both at various margins and on distribution, that the enactment might have.

But the theoretical paper suggests, and the empirical paper finds evidence supporting, the view that this will not necessarily happen because higher capital gains taxes have an insurance function.  Suppose that there is a single capital gains rate applying symmetrically to gains and losses.  (Given the capital loss limitation in U.S. tax law and that of various other countries, this may rest on the taxpayer's expecting to have other capital gains.)  Then the higher the tax rate, the lower the taxpayer's after-tax risk

In short, a symmetric capital gains tax provides insurance that is actuarially fair if the risk-free rate is zero.  It becomes actuarially adverse if that rate is positive, since the risk-free return is included in the tax base, but still the insurance may have net value to taxpayers if not otherwise available.

One of the big questions raised by the paper is why, as the evidence appears to suggest, taxpayers would actually value the risk-sharing properties of the capital gains tax.  After all, they presumably have some ability to optimize risk levels on their own, such as by diversifying and hedging.  This might reflect limitations in capital markets, or else the government's superior ability to handle systematic risk (although, in that case, one must ask where the risk that the government at least nominally absorbs in fact ends up).  There is always a possibility of alternative explanations for the paper's findings, although the authors took reasonable steps to address the main contenders.