When is it Acceptable to Redefine Income?
February 26, 2015
Earlier this month there was a little argument between the Treasury Department and the Tax Policy Center. The argument was over a particular provision: the White House’s proposal to tax accrued (but unrealized) capital gains at death.
The Tax Policy Center’s analysis showed that this proposal would affect some middle-income families, and the White House insisted that it wouldn’t. The Tax Policy Center’s Len Burman rounds up the story well here.
I believe this little argument was more significant than it looked. While the subject of the argument was mostly a small tax proposal that won’t become law, the issues highlighted by the dispute have greater significance to tax policy.
Realized vs Accrued Gains
On balance, most of the tax world measures realized gains, not accrued ones. There are some good reasons for this.
For one thing, measuring people’s accrued gains instead of their realized gains is mostly impossible because that information isn’t reported to any centralized authority. (And why would it be? The IRS taxes realized gains. It has no interest in your accrued gains.)
For another, though, even if that data were available, it would result in dramatic fluctuations in the measured incomes of older Americans with substantial amounts of money invested in equities. In a poor year for the market, you’d see a deeply negative income for a retiree who owns a bunch of stock. Equity-owning retirees, as a whole, are actually in pretty good shape in America, but if you measure their accrued income each year, you’re going to see them doing really well in some years, and really poorly in others.
Rather than having them fly up and down the income distribution every year depending on market returns, it’s probably a little bit better in some sense to use realized capital gains in order to smooth that income out.
However, there is some reason to believe that accrued gains works better as a measure. The “on-realization” method of counting them results in big spikes in peoples’ incomes, as they sell assets or businesses they may have spent a lifetime accumulating.
Neither measure is good for all purposes. When we make distributional tables, what we’re really trying to do is approach the platonic ideal of “tell us who is rich and who isn’t.” But that’s a surprisingly hard thing to measure.
Ad-Hoc Changes to the Definition of Income
It is obviously necessary for lawmakers to be able to change the definition of taxable income. That’s an essential part of governance. But making ad-hoc changes to the definition of income for distributional analysis is quite another thing.
The White House wanted to add accrued gains at death to taxable income. That is certainly a law that could be passed. But they also wanted to add those gains to the distributional analysis.
This is a troubling precedent.
First, it creates a definition in which we measure realized gains most of the time, but accrued gains in one particular circumstance. The consistency of that standard is shaky at best.
Second, it means that promises not to raise taxes on people in particular income brackets (a promise the President has made) are essentially without substance. You can then define such people to have more income – as the Treasury did here – and then try to tax them anyway.
Last, it does violence to the idea of distributional analysis in general, and progressivity more specifically. If you levy a tax by defining new income (or conversely, repeal a tax by excluding a source of income) then your changes are guaranteed to look more progressive than they actually are. Of course the people you’re taxing are rich! Look at all that income they have, now that you’ve defined them to have it! (Or alternatively, look at all that income they don’t have, now that you’ve defined them not to!)
This isn’t an ironclad, unshakable argument for the precise methodology that Tax Policy Center uses. But rather, it’s an argument for consistency. At Tax Foundation, we treat all tax plans to the same assumptions when we model them. Tax Policy Center does likewise. And if we change our models, we do so because we think it improves the quality of the results, not because lawmakers asked us to.
I believe that Tax Policy Center, like Tax Foundation, prides itself on analysis that is principled and independent. And of course, any analysis whose assumptions change at the request of lawmakers is neither principled nor independent.
Len Burman was right to stand his ground on this point.