Yesterday at the colloquium we discussed with Owen Zidar the above work (coauthored with Ole Agersnap), along with a preliminary draft of a short piece (coauthored with Natasha Sarin, Larry Summers, and Eric Zwick) entitled "Rethinking How We Score Capital Gains Tax Reform."
The work's main finding is that capital gains realizations are far less tax-elastic than the current conventional wisdom, including that of the Congressional Joint Committee on Taxation (JCT), has assumed. Hence, for example:
--The revenue-maximizing tax rate for capital gains lies between 33% and 47%, whereas currently it is just the sum of (a) 20%, plus (b) the 3.8% Medicare payroll tax if one is subject to it, plus (c) any state and local income tax liability that one faces with respect to capital gains.
--A 5% increase in the capital gains tax rate would raise $18 billion to $30 billion in annual revenue. By contrast, the JCT, which assumes far higher tax elasticity for capital gains and thus a far lower revenue-maximizing rate, might score this as being close to a revenue wash.
--Under plausible median assumptions, raising the capital gains high to the neighborhood of the top individual rate (under current law or pre-2017) might raise as much as $1 trillion over ten years, whereas the JCT would likely come up with a number in the ballpark of $80 billion.
This work obviously has major implications for how we might think about tax law changes that could conceivably be on the table in the next few years (especially if the Democrats win both Georgia Senate runoff elections). If accepted by the JCT it might also have a huge impact on the legislative process, given the important role that official revenue estimates can play.
The paper relies mainly on data over the last several decades concerning state-level capital gains rate changes, since there are far more of these than federal changes. But it also offers a set of reasons why JCT estimates may be systematically too low. For example:
--Capital gains that are locked in, rather than being currently realized by reason of a CG rate hike may be realized several years later, rather than being deferred until death (at which point they would disappear from the tax base given Code section 1014, the step-up in basis at death).
--The composition of capital gains has changed over time, reducing their tax elasticity. In particular, the rice of pass-throughs such as hedge funds that generate a lot of capital gain but have a business model requiring regular realization has had this effect.
--Substitution by taxpayers between generating capital gains and ordinary income means that higher CG rates may boost other tax revenue, as taxpayers shift away from taking advantage of the lower CG rate.
--Legal changes, such as the enactment of section 1259 (treating certain transactions as constructive sales) have made it less feasible and realistic for taxpayers holding appreciated assets to defer CG realization indefinitely, or even just until death.
I found the papers' empirical analysis and reasoning to be highly plausible. Note also that Zidar should have a lot of credibility on this potentially politically fraught topic - e.g., his work on high-end inequality tends to find less wealth concentration at the top than that of, say, Saez and Zucman. So, if he were to invoke the oft-abused metaphor of the umpire who merely calls the balls and strikes as he sees them, he would be far more believable than certain others who have invoked it in the past.
I'll close here, but add a follow-up post that mentions some broader conceptual issues that the work brought to mind.