Tax Progressivity and Inequality

The other day, Vox published a column by political scientists Cathie Jo Martin and Alexander Hertel-Fernandez arguing that progressive, "soak-the-rich" tax systems don't address inequality. The evidence for that argument is that Sweden collects a larger portion of its tax revenue from the bottom and middle of the income distribution than does the US. Europe generally relies on the Value-Added Tax, as opposed to our heavier reliance on income taxes. Weirdly, this argument also cropped up in a New York Times Op/Ed by Ed Kleinbard that uses the same "soak-the-rich" locution.

The argument is not correct for multiple reasons. The two main problems appear in a response by Mike Konczal: first, looking at where tax revenue comes from isn't the right measure of progressivity, since that just captures income inequality. Second, there's good reason to believe that the regressification of the US tax system is precisely why inequality has grown so much. I'll add a few further points to Konczal's analysis.

Using the same OECD data Martin and Hertel-Fernandez report, Konczal regresses the income share of the top decile on "progressivity" as measured by the concentration of tax system revenue at the top. He finds they're tightly linked. The more unequally distributed is income, the more tax revenue will be drawn from the rich. But that doesn't mean that the rich are more burdened by taxes, the usual definition of progressivity.

To follow up, the OECD's measure of income is top-coded, so the top decile income share it reports is too low. I've reproduced Konczal's scatterplot and overlain one drawn from the World Top Incomes Database, which is not topcoded. Obviously all the points for the top decile's share are shifted vertically, with the increment proportional to tail inequality. One surprising thing, at least to me, is that the slope didn't get steeper as well. It does if you include capital gains, but there are only five countries that have that data in the WTID for 2005.
 

 Within each color, each point is a country. Non top-coded data come from WTID "Top 10% Income Share (2005)" without any adjustment for households in OECD data vs. tax units in WTID. Some countries are in the OECD but not the WTID. The US is the top-right-most point.

The OECD's top-coded measure of tax progressivity misses an outstanding aspect of the US tax system: that the effective tax rate is declining in income at the top of the distribution. In this country, the rich enjoy a variety of tax advantages. Those include a lower tax rate for capital gains,the famed "carried interest loophole" that lets hedge fund managers pay income taxes as though they are capital gains, and all manner of tax deductions that are more valuable if you have a higher income. That declining effective tax rate is why Warren Buffett's tax burden is lower than his secretary's.

Obviously, these facts makes the US tax system more regressive, but not as measured by the OECD.

Konczal also cites the great paper by Piketty, Saez, and Stantcheva (discussed at length in Capital in the 21st Century) arguing that the statutory decline in the top marginal tax rate is why the US has become so much more unequal in the past several decades. At the Center for Equitable Growth's annual conference last month, Saez argued that we already know how to reduce inequality at the top: progressive income taxes. He repeated that at an event at the University of Chicago this past week. In some ways, this is disappointing to those of us thinking up what to do about inequality, because of course we've known progressive taxation is effective since the publication of the Gotha Program in 1875 and the enactment of a permanent federal income tax in the US in the Progressive Era.

Finally, let's get back to the question of revenue, which is in fact where Martin and Hertel-Fernandez focus. The US income distribution post-tax-and-transfer has become more unequal since the late 1970s. But our spending programs have remained intact. They've become more expensive because of the in-kind nature of the government's healthcare commitments. They've also become more tilted toward the elderly of all income levels, as opposed to the poor, thanks to the aforementioned trend as well as the enactment of Medicare Part D. But it's fair to say that on the spending side, redistribution hasn't changed *that much.*

On the revenue side, however, we've become much less redistributive, thanks to regressification at the very top. I've plotted the post-tax income concentration over time against the statutory top marginal tax rate below.

The CBO's measure of post-tax income share of the top quintile is on the right axis, and the statutory top marginal income tax rate is on the left axis.

The time trend isn't dispositive, but it is highly suggestive. Post-tax income inequality has gone up as the tax rate has gone down. Piketty and Saez (2006) look at the declining progressivity of the US tax system in more detail and in a comparative context. Their entire body of research is what underlies Saez's argument that we need to get back to talking about tax progressivity.

Crucially, though, that's not because of revenue. Saez's argument, and Piketty's in Capital in the 21st Century, is that we should enact progressive taxes precisely because that would compress the pre-tax income distribution. It's not about increasing the government's revenue, but rather about creating a more egalitarian society. In theory that could be done through redistribution, but if Piketty, Saez, and Stantcheva are correct, the reason for high incomes in the first place is just bargaining power, and society doesn't lose anything by reducing them.

When President Obama committed himself not to raise taxes on households earning less than $250,000, some commenters decried the implication that we wouldn't be able to enact an ambitious liberal agenda without a larger tax base. First of all, thanks to tail inequality, the tax base above $250,000 is quite high if you get serious about loopholes and overseas tax havens, as the administration has to some extent. But more importantly, and contra Martin and Hertel-Fernandez, we don't need ambitious redistribution to deal with inequality.