Last Friday, I was honored to receive the David Kershaw Award appam.org/awards/david-n… from APPAM. At the conference, I gave a talk entitled “What Does Education Do?” In case you missed it, I am going to TWEETSTORM the main points below 1/N
As we all know, economic inequality has risen a lot over the last 40 years. Understanding why is challenging, because there are many potential explanations, but only one fact pattern 2/N
My first main point was that the fact pattern suggests education is a very important part of the story. 3/N
As Autor (2014) points out, between 1979 and 2012 the gap in HH earnings between two college-educated partners and two high-school educated partners has grown by more than $28k in real terms. 4/N 
Over that same period, the top 1% HH income share more than doubled. If we redistributed all of the increased income going to the top 1% to the other 99%, HH income would rise by about $7k. 5/N
Thus the growing earnings gap between HS and college educated households is 4 times LARGER than the income gains among the top 1% since 1980 6/N
This begs the question – do we see rising inequality in academic skills that matches the time trend in earnings inequality? And if so, when do skill gaps emerge? 7/N
The answer is NO – at least not by age 17. Looking at the NAEP long-term trend (essentially the same test given nationally by the DOE since the 1970s), we see a *compression* of the test score distribution among 17 year olds. 8/N
The basic pattern is modest gains at the bottom of the distribution with no change at the top. It looks nothing like the widening wage distribution. This pattern holds for math and reading, for ages 9, 13, and 17, and by race and parental education. 9/N
See the report for yourself: nces.ed.gov/nationsreportc… And there are other studies (cited in my slides) that find the same pattern 10/N
There are 3 possible explanations for this puzzling pattern. 1) Achievement and school quality don’t matter for earnings; 2) Skill gaps exist, but they emerge after age 17; 3) Skill gaps exist, but the NAEP and other tests don’t measure them very well. 11/N
Lots of research rules out No. 1 (see my slides for details). No. 2 suggests that postsecondary education may play a key role in creating inequality in skills; No. 3 points to the possible importance of “soft skills”. 12/N
The first point I make on 2 is that resources are distributed much more equally in K-12 than in postsecondary education. Almost all states have progressive funding formulae, so poor districts spend more per pupil than rich ones. But the differences are small in any case 13/N
Dollars don’t tell the full story, of course – but compare K-12 to college resources, where private research universities spend $75k+ per student and community colleges spend $10k or less. And of course many students get zero resources because they don’t attend college 14/N
And there are huge differences in college selectivity by family income. Students at highly selective colleges are 14x more likely to be from a family in the top 20% of the income distribution than from the bottom 20%, and 77x more likely to be from the top 1% 15/N
Also, as pointed out by Goldin and Katz (2009) and others, rising wage inequality and the rising college premium coincides precisely with the U.S. slowdown in college completion 16/N
All this suggests that if we want to reduce economic inequality in the US, we must find a way to equalize access to postsecondary resources. That’s part of what we are trying to do at the CLIMB initiative! climbinitiative.org 17/N
This is a strong circumstantial case that spending more $ on postsec ed and distributing it more equitably would lead to more growth and less inequality. But *more* is not enough – we also need to redesign education systems to meet the changing demands of the modern economy 18/N
That means designing schools to develop the skills employers want – teamwork, flexibility, problem-solving in complex and uncertain environments – and not focusing too narrowly on specific careers. Jobs change rapidly, and you want your education to be “future proof” 19/N
The main challenge – for an economist at least – is that we don’t really understand *how” education builds skills. Good design necessitates an understanding of mechanisms, and economists’ main tool for thinking about education is the human capital model. 20/N
It's not that the HC model is wrong. It's just incomplete. The idea that school enables me to produce more widgets per hour was a revolution in its time. But it doesn't really get inside the black box enough to be helpful for design and reform. 21/N
In conclusion: education "works"! Yet it’s inequitably distributed, especially at the college level. So increasing access would reduce inequality. But we don’t know why, and we need to know *why" to make colleges *better* than they currently are. 22/23
If you're interested in the details, see scholar.harvard.edu/files/ddeming/… for the slide deck (including references). Thanks for reading! 23/23
