Careful thinking about capital gains taxation
I've been following closely a roiling debate going on right now in the econoblogosphere over the taxation of income from capital investments, dividends, and carried interest. I can't say that I come down strongly on any particular side of the argument -- basically, the debate is whether such income should be taxed at all -- but I think everyone would benefit from a review of the positions and evidence, and probably a deep breath.
The first thing I looked at was the existing state of capital gains taxation around the world. Given the vehemence coming from the anti-CGT side, one would have thought that the U.S. and the few other nations taxing capital income were holdouts whose policies run contrary to the direction of the rest of the world. (Indeed, that is true for the corporate income tax.)
What I found was that the clear majority of nations, by simple counting and by an informal weighting of their importance in the global economy, tax capital. The importance of this should not be understated -- when the economic scholarship runs so contrary to the real-world situation, it should raise questions, because only one side can have it right.
This is not an idle question: in 1990, as capital controls were coming down and a global market for investment was forming, UMich economist Robert Gordon asked "can capital income taxes survive in open economies?" Although Gordon does not appear to take a side, he leaves us with the impression that an international community of nations, all of which tax capital income at a non-zero rate, does not represent a stable equilibrium -- and that a race to lower rates is highly probable.
I haven't had the chance to explore the literature and see if there is a sufficient response to Gordon, but we should note here that the changes in capital gains taxation since 1990 have been ambiguous on a quick review of the evidence -- there has been no obvious downward race. In fact, a number of countries raised their CGT rates in the aftermath of the financial crisis.
Broadly speaking, there are several approaches to capital gains taxation around the globe. First, some nations have no tax preference for capital income. Second, some nations tax capital income at a flat rate, usually below their tax rates for other forms of income. Third, some nations limit their taxation of capital gains to certain forms -- usually on short-term gains, on shares and stock given as compensation, on non-primary residence residential investment, on non-retirement savings, etc. Fourth, some nations have tax bracket structures for capital gains according to the magnitude of the gains or of the broader income of the taxpayer. Fifth, some nations tax capital income at some fraction of the taxpayer's marginal income tax rate. Sixth, some nations tax only real capital gains.
This is just to give you a taste of how complex the taxation of capital gains is. I agree with Gilles Saint-Paul, who wrote recently in The Economist that the "subtle" nature of the debate causes popular debate to oversimplify it -- I'm thinking of politicians make the case for capital gains taxes to "punish the rich bankers." Unfortunately, I fear that Saint-Paul and Scott Sumner, who also weighs in, both proceed to oversimplify the discussion.
The argument of Saint-Paul and Sumner boils down to the idea that tax systems should not tax future consumption -- investment -- at a rate higher than present consumption, as such a decision will restrain long-run growth. In other words, both of them worry about the incentive effects, and only that.
Again, I note here that I don't have a solid opinion in this debate. But my criticism of their position would be that they miss two major issues which compromise their (admittedly strong) case: (1) capital income does not necessarily imply prior saving in the real world as it does in their model, (2) capital income is unevenly substitutable with regular income depending on the level of gross income.
Let's start with (1). The way that capital income is taxed in the real world bears little resemblance to models which suppose that all investment is taxed at the capital gains tax rate, and all income is taxed at the normal income tax rate. This is a point excellently demonstrated by Greg Mankiw in The New York Times, who guides us through a number of the fine gradations between what counts as capital income and what doesn't in the US tax code. Out of all the economists I cite in this post, I think my position is closest to Mankiw's: "it is not obvious what the best approach would be. Not all problems have easy answers." In many cases, the point that a capital gains tax necessarily discourages future consumption seems weaker, as in many of Mankiw's scenarios there is arguably no prior saving. Uwe Reinhardt also notes that the income tax already hits some forms of capital, namely investments in human capital, and that the elimination of the capital gains tax would further worsen the distortions between forms of capital investment. (Steven Landsburg responds here, raising some valuable points about why we should at least try to favor saving and investment. I've also made myself somewhat familiar with the supply-side argument for low capital gains taxes and the "Laffer curve"-style impact on revenues and with the work of Alan Reynolds.)
My point in (2) is something that I haven't seen anyone argue yet, but I think it could be the most important reason for the continuation of a capital gains tax. It is that an individual's ability to pass off some of their income as capital income is not constant with the level of income or type of employment. The very wealthy and those in certain occupations, in other words, are substantially more able to gain preferential tax treatment than those of lesser means or in other occupations, or even of the merely somewhat wealthy. This is why I don't agree with Sumner's point that the income tax code can be made more progressive in lieu of a capital gains tax. I think under his plan, you'd end up soaking the "merely somewhat wealthy" while the very, very top end of the income distribution is able to find shelter, despite the higher income tax rates. I suppose what I'm saying is that I don't buy the idea that killing the capital gains tax will not have distributional consequences.
If I sense myself agreeing with any particular position in this debate, it's this: the capital gains tax requires substantial reform to be optimal, but that optimum is probably not a zero rate. Tax only real gains, do it as some fraction of the taxpayer's marginal income tax rate to minimize the distributional effects, rewrite the code to favor only and all of the kinds of investment which require saving, and keep the rate low to minimize incentive effects. But that is the most tentative and uneasy of conclusions. I await better argument from either side.
PS. I weighed in on this same topic in January at much less depth, and I don't think my views have changed much. I definitely don't think we should raise the capital gains tax rate.