Inequality Reduction Is Not Mysterious

Brookings is out with a rather strange four-page report purporting to show that it's hard to reduce inequality with tax-and-welfare policies. It doesn't actually show this, but it's getting a lot of play, so it's worth discussing.

What It Says

Using tax filing units, the authors determined that if you increased the top marginal tax rate on labor income from 39.6% to 50%, this would only reduce the Gini Coefficient from 0.574 to 0.571. It further finds that if you were to take that money and dole it out to the bottom 20%, you'd only knock the Gini down to 0.560.

This finding is supposed to seem shocking, but it's really not for the following reasons.

First, the top marginal tax rate only applies to tax units whose family income tends to put them at near the very top of the income distribution. This means that the tax base for this rate increase is not really that large.

Second, the simulated reform only applies to the top marginal tax rate of those affected, not the average tax rate. Thus, this is not a proposal to increase the taxes of the richest 5% by 10.4 percentage points, which some will read it as. The tax surcharge is only applied to the income made over the top tax bracket threshold. The authors don't estimate how it would change the average tax rate (or effective tax rate) of the rich people who will pay it, but the increase in that rate will necessarily be well below 10.4 percentage points.

Third, the simulated reform only applies to labor income, not capital income, but, for the very rich, capital income makes up a surprisingly large share of their personal income.

So, taken together, this simulation only increases taxes on a small percentage of top earners, excludes a big chunk of their income from the tax, and then only increases their top marginal tax rate by 10.4 percentage points, which means increasing their effective tax rate by a much lower amount. The fact that this weaksauce tax hike is not delivering monster inequality changes should be surprising to nobody.

But wait, there's more.

Fourth, the authors use a Gini Coefficient metric to determine how much their plan reduces inequality. However, anybody familiar with Gini Coefficients can tell you that a transfer plan which scrapes slightly from the top end of the distribution to deliver stuff to the bottom end of the distribution will not show up much in Gini calculations. This is a function of the way Gini works.

Here is a graphical representation of how Gini measurements work:

As you move from left to right, you move from the poorest person to the richest person. On the vertical axis, you have the cumulative share of the national income that has been collected by everyone below each given percentile on the horizontal axis. So, if everyone has the same income, the values on the horizontal and vertical axes will match. At the 10th percentile, the cumulative share of the national income collected by those below the 10th percentile will be...10%. At the 50th percentile, it will be 50%. At the 100th percentile, it will be 100% (in fact, it's always 100% at the 100th percentile for obvious reasons). Thus, in the perfectly equal world, you get the 45-degree line labeled in the above graph as the Line of Equality.

But in an unequal world, what you get is something that looks like the line labeled "Lorenz curve." The way the Gini measurement works is to find the difference between the Lorenz curve of the income distribution and the Line of Equality, which is to say to find the area of the space labeled A.

For our purposes here, what's important to note is that the vast majority of the space labeled A is not at the tails (the furthest left and right side of the graph). It's in the middle. What the proposed reform does is slightly clip income from the very far right-hand tail and move it to the left-ward tail, which leaves the overwhelming majority of the A area undisturbed.

So even if you looked beyond the rather modest nature of this tax reform, there is still the problem that the Gini metric is not well-suited to capture tail-focused distributive changes.

Of Course Tax and Welfare Works

The point of this little report, it seems, is to suggest that there is something particularly complicated and evasive about inequality reduction. This is underscored in the last paragraph, which states:

This analysis, coupled with the previous one, in turn leaves us with the open and important question: if neither a substantial expansion in education nor a big increase in the top marginal tax rate would significantly affect measured income inequality, what would?

If educational attainment and tail-focused modest tax level hikes don't work, then this is going to be a much tougher nut to crack than we thought, the authors seem to be implying.

But it's not at all a tough nut to crack. A few weeks ago I laid out a pretty standard four-point egalitarian plan (see here). But one of the best ways to see what makes the US so different is to do some cross-country comparisons, which shows the US standing out especially in two domains.

First, the US has extraordinarily low taxes.

Second, the US has very low levels of public social expenditure.

The US also manages to get remarkably little out of its public social expenditure, mainly because of its dysfunctional health care system.

If you want to get lower inequality through distibutive mechanisms, the way you do that is by increasing the tax level and increasing the social benefit level. Basically all taxes (income, consumption, wealth) pull more dollars from the better off than the worse off. And basically all social expenditures deliver vastly more benefits (relative to market income) to those who are worse off. Taken together, then, high taxes and high welfare (even universal welfare) greatly compresses the economic distribution.

Back in the day, people used to say egalitarianism was a bad idea because it would reduce growth. This is not true, but at least there was a theory behind it. It seems like Brookings is trying to take us into a new world where the opposition to egalitarianism is based on suggesting it is impossible to actually achieve. This is nuts. Other countries have done it and how they did so is not mysterious.