Seeking ways to revive the economy, President Donald Trump on May 5 glancingly referenced the possibility of eliminating all taxes on capital gains.
Speaker Nancy Pelosi (D-Calif.) was less than enthusiastic about the idea in remarks two days later. But some reports suggest that the President’s tweeted remark was an errant spark from serious ongoing discussions within the Administration. This is something we will likely be hearing more about in the coming weeks.
Under current law, when an investor buys low and sells high—having waited at least a year in between—the “long-term capital gain” is taxed at a substantially reduced rate. While the tax rate on top-bracket salary income is 37%, the rate on long-term capital gains is only 23.8%. That’s already a 35% discount. Now, the Administration appears to be toying with the idea that the discount should be 100%.
The obvious objection is that the people who will benefit most from lower rates on capital gains are the people who, well, have capital gains. These tend to be the wealthiest taxpayers. And the wealthy, almost by definition, are well buffered against losses and are thus least in need of expensive government assistance.
But proponents of lower taxes on capital gains have always had a politically potent counter-response—one that has won the day repeatedly over the last several decades and explains the existing 35% discount. Part of its power comes from the folksy resonance of the caution that things are not always what they seem. The rest of the power comes from an appeal to rigorous mathematical logic in explaining how things actually are—that lowering taxes on investment income turns out to be the best way to help those who don’t have any.
Things are not, in fact, always what they seem. And that goes double for the set of mathematical models offered to prove that the tax rate on capital gains should be zero.
With the debate on capital gains taxation likely to heat up over the next several weeks, now might be a good time to clear up a few misconceptions concerning what the mathematics does and does not say about how investment income should be taxed.
1. Taxing investment income does not necessarily reduce investment.
The most common argument against taxing investment income is an appeal, not to some complex model, but to basic economic logic: Taxing investment reduces investment, which in turn reduces wages and employment.
But that’s not what basic economic logic actually says.
Basic economic logic says that when the price of gas goes up, people buy less gas. But as everyone knows, this does not mean that people spend less on gas. It’s a race of uncertain outcome between the reduction in gallons purchased and the increase in price per gallon.
The story is similar with investment. When people invest they are effectively purchasing future consumption, if not for themselves, then for their heirs. To invest is to spend on those future uses. When investment income is taxed, future uses become more expensive and people reduce their purchases. Whether people spend more or less on future uses—that is, whether they invest more or less—is a race between the reduction in “gallons” of future uses purchased and the increase in the price per “gallon” of future use.
2. The data provide no clear answer.
In principle, the outcome of that race—and so whether taxing investment increases it or decreases it—might be resolved by the careful statistical analysis of good data. In practice, the data are messy, and the question remains very open.
3. Complex model number one.
Two highly mathematical arguments appear to produce stunningly sharp outcomes out of the smoky cloud of ambiguities of the kind just mentioned. Both are problematic on close inspection.
The first is the Atkinson and Stiglitz model from the mid-1970s. The punchline of this model is that the most efficient way to design a progressive tax system (mind you: progressive) is to base how much tax each person owes solely on wage and salary income. The implication is that capital income, inclusive of capital gains, should not be taxed.
But it turns out that this result is baked into the model. The model assumes that “leisure is weakly separable.” The assumption sounds harmlessly technical. But a bit of mathematical mining reveals that what it actually means is that, when it comes to progressive taxation, anything capital taxation can do, labor taxation can do better. In other words, the model’s assumption means its conclusion.
4. Complex model number two.
Then there is the Judd model from the mid-1980s—a model that focuses on economic growth over an infinite horizon and concludes that the optimal tax rate on capital in the long-run is zero—even when the tax system is specifically designed to help those who only have labor income. There are two separate problems with the Judd model. First, as was just recently noticed by Straub and Werning, Judd implicitly, perhaps unknowingly assumes a particular winner for the race mentioned above: he assumes that taxing investment lowers investment. Plugging the opposite assumption into the same model produces, roughly speaking, the opposite result. Second, the model itself, in which investors live forever, is based on an artificial quirk of infinite-horizon reasoning that render its results in either direction essentially inapplicable.
All else the same, it would be best if all taxes were zero. But when a certain amount of revenue must be raised, lowering the tax on one type of income means raising it on another. The resulting tradeoffs are still only roughly understood. So while it is good to be skeptical of the way things seem, it is also good to be suspicious of the pat-answer outputs of opaque models.
This column does not necessarily reflect the opinion of The Bureau of National Affairs, Inc. or its owners.
Chris William Sanchirico is the Samuel A. Blank Professor of Law, Business and Public Policy at the University of Pennsylvania Carey Law School.