Education and Inequality

David Leonhardt had a piece about inequality at the New York Times over the weekend. Although the overall message is correct (inequality is a policy choice), but the piece has a few problems, with the education parts the most significant of them.

I.

The problems in the piece start when Leonhardt writes this:

The countries that have done a better job increasing their educational attainment, like Canada and Sweden, have also seen bigger broad-based income gains than the United States.

Clever people who paid close attention to the LIS data that the Upshot profiled at the beginning of the week (and that Leonhardt is relying upon here) will spot the issue. The LIS income data is tracking disposable household income. This refers to the amount of income households have "after" taxes and transfers, not their market incomes. But market incomes are what you want to look at to assess some education effect. It could be that market incomes also track this trend, but Leonhardt hasn't shown this and is relying upon the wrong data to make his point.

When we look at poverty rates (defined as 50% of the median income) to get some basic sense of what is going on in these three countries, we don't find a great deal of encouragement from the suggested education-driven egalitarianism of Canada and Sweden:

The market poverty rates (blue) span from 26 percent (Canada) to 28.4 percent (US). So the spread is 2.4 percentage points from lowest to highest. Not a lot of difference.

However, the disposable income poverty rates (red) span from 9.1 percent (Sweden) to 17.4 percent (US), with Canada at 11.9 percent. So the spread is 8.3 percentage points from lowest to highest. At least on the bottom end here, transfers are what's responsible for delivering the goods in Canada and Sweden.

II.

Leonhardt continues on the education point in the very next sentence:

Yet the debate over our schools and colleges tends to exist in a separate political universe from our debate over inequality. Liberals often shy away from making the connection because they worry it holds the struggling middle class and poor responsible for their plight and distracts from income redistribution.

The inequality sphere is full of people who write on education. Those who emphasize distributive policy (like me) are mainly in the background. Our policy discourse is obsessed with trying to find some education-focused solution to distributive problems.

The second sentence in Leonhardt's quote also strikes me as false. Liberals love education stuff. You can hardly get liberals more excited than when you are talking about teachers and college and K-12, especially when you combo those things with "disadvantaged" children. That is the holy grail of liberal fixations.

Nonetheless, liberals should, on the merits, be upset that educational nonsense distracts from income redistribution. Consider the graph above and then consider what Leonhardt's piece is doing. He has taken an apparent confusion about disposable income data (which doesn't speak one way or another on education effects) and created a narrative about education that does not actually tell the major story of how Canada and Sweden achieve their broad-based income distributions. This is a distraction.

III.

In closing, it is perhaps helpful to rehearse the reason why education won't solve our distributive woes. Education confers upon people some absolute advantages that are not zero-sum, but more than that it confers upon people positional advantages that are more or less zero-sum. Education puts you above others in the competition for scarce high-paying economic positions. Our joint-production economy can sustain only so many managers, accountants, engineers, bankers, lawyers, and so on. Minting new degrees does not cause matching jobs to pop into existence. We can't all be in the professional or creative class because, if we were, who then would do the work?

Education boosters make the mistake of misunderstanding the positional nature of the advantage education confers. They are people who, upon spotting that those who camp out days early get front row seats at the concert, conclude that if everyone camped out days early they'd all sit in the front row. This wont work in the context of capturing concert seats and it wont work in the context of capturing economic and social positions.

Education is good for all sorts of reasons. It also has good economic impacts insofar as it helps make people more productive. But it is not sufficient to solve our distributive woes. We'd all be better off in the world of policy if the pedagogy people stuck to pedagogy policy and the distribution people stuck to distribution policy. It's not useful to blend discussions of educational improvements and distributional improvements together, and any effort to force a choice between them should be answered back with the question: "why not both?"

Comments