kitchen table math, the sequel: The Race redux

Thursday, December 22, 2011

The Race redux

Two summers ago, I wrote a number of posts on The Race between Education and Technology, a book I found revelatory. Via Greg Mankiw's blog, I've just come across economist Daron Acemoglu's recommendation of The Race as one of the top 5 books to read on inequality:
This is a really wonderful book. It gives a masterful outline of the standard economic model, where earnings are proportional to contribution, or to productivity. It highlights in a very clear manner what determines the productivities of different individuals and different groups. It takes its cue from a phrase that the famous Dutch economist, Jan Tinbergen coined. The key idea is that technological changes often increase the demand for more skilled workers, so in order to keep inequality in check you need to have a steady increase in the supply of skilled workers in the economy. He called this “the race between education and technology”. If the race is won by technology, inequality tends to increase, if the race is won by education, inequality tends to decrease.

The authors, Claudia Goldin and Larry Katz, show that this is actually a pretty good model in terms of explaining the last 100 years or so of US history. They give an excellent historical account of how the US education system was formed and why it was very progressive, leading to a very large increase in the supply of educated workers, in the first half of the century. This created greater equality in the US than in many other parts of the world.

They also point to three things that have changed that picture over the last 30 to 40 years. One is that technology has become even more biased towards more skilled, higher earning workers than before. So, all else being equal, that will tend to increase inequality. Secondly, we’ve been going through a phase of globalisation. Things such as trading with China – where low-skill labour is much cheaper – are putting pressure on low wages. Third, and possibly most important, is that the US education system has been failing terribly at some level. We haven’t been able to increase the share of our youth that completes college or high school. It’s really remarkable, and most people wouldn’t actually guess this, but in the US, the cohorts that had the highest high-school graduation rates were the ones that were graduating in the middle of the 1960s. Our high-school graduation rate has actually been declining since then. If you look at college, it’s the same thing. This is hugely important, and it’s really quite shocking. It has a major effect on inequality, because it is making skills much more scarce then they should be.

Do Goldin and Katz go into the reasons why education is failing in the US?

They do discuss it, but nobody knows....It’s not that we’re spending less. In fact, we are spending more. It’s certainly not that college is not valued, it’s valued a lot. The college premium – what college graduates earn relative to high-school graduates – has been increasing rapidly. It’s not that the US is not investing enough in low-income schools. There has been a lot of investment in low-income schools.


Goldin and Katz’s book shows that the college premium was higher in the early 1900s than it was in the 1940s and 1950s. Then it remains stable for several decades, and then it starts increasing again in the 1980s.
Daron Acemoglu on inequality
in a nutshell:
  • the US education system has been failing terribly at some level
  • the cohorts that had the highest high-school graduation rates were the ones that were graduating in the middle of the 1960s
He goes on to say that "What’s missing from the Goldin and Katz book is that they really don’t look at all at what’s going on in the top 10%," a point I think I recall Allison making (although Allison may have been talking about the the top 1 or 0.5%).


Bostonian said...

In the middle 1960s the U.S. changed its immigration policy to allow to more immigration from Third World countries. The Hispanic proportion of the population has grown a lot, and Hispanics have lower graduation rates than whites. This lowers the overall graduation rate.

Has the high school graduation rate of whites declined since the 1960s? I doubt it.

Catherine Johnson said...

Offhand, I don't know the answer to that - though I wouldn't be at all surprised to learn that white kids are graduating in reduced numbers, too. Look what's been happening to boys.

High school graduation rates of non-whites elsewhere beats ours.

Bostonian said...

A paper by Heckman and LaFontaine, "The Declining American High School Graduation Rate: Evidence, Sources, And Consequences" is consistent with my theory. Quoting a summary :

"After adjusting for multiple sources of bias and differences in sample construction, we establish that: 1) the U.S. high school graduation rate peaked at around 80 percent in the late 1960s and then declined by 4-5 percentage points; 2) the actual high school graduation rate is substantially lower than the 88 percent estimate; 3) about 65 percent of blacks and Hispanics leave school with a high school diploma, and minority graduation rates are still substantially below the rates for non-Hispanic whites. Contrary to estimates based on the status completion rate, we find no evidence of convergence in minority-majority graduation rate Exclusion of incarcerated populations from some measures greatly biases the reported high school graduation rate for blacks.

These trends are for persons born in the United States and exclude immigrants. The recent growth in unskilled migration to the United States further increases the proportion of unskilled Americans in the workforce, apart from the growth attributable to a rising high school dropout rate.

As others have shown, and we confirm, the most significant source of bias in estimating graduation rates comes from including GED recipients as high school graduates. GEDs are high school dropouts who certify as the equivalents of ordinary graduates by passing an exam. Currently 15-20 percent of all new high school credentials issued each year are GEDs. In recent years, inclusion of GEDs as high school graduates has biased graduation rates by upwards of 7-8 percentage points. A substantial body of scholarship summarized in our 2008 book 9 shows that the GED program does not benefit most participants, and that GEDs perform at the level of dropouts in the U.S. labor market. The GED program conceals major problems in American society.

The decline in high school graduation is of interest in its own right as a measure of the performance of American schools. It has important implications for interpreting a wide variety of educational statistics. The slowdown in the high school graduation rate accounts for a substantial portion of the recent slowdown in the growth of college educated workers in the U.S. workforce. This slowdown is not due to a decline in rates of college attendance among those who graduate high school."

Catherine Johnson said...

h.s. graduation for white students is still 80%?

Catherine Johnson said...

Currently 15-20 percent of all new high school credentials issued each year are GEDs.

racial breakdown for GEDs?

This paper may support your theory, but this summary doesn't seem to include the relevant data.

(I say 'doesn't seem' because I get major eye strain these days reading large chunks of prose online...)

Bostonian said...

Look at this graduation rate graph . The trend graduation rate by race is flat or up. Overall graduation rates are not rising because of Simpson's Paradox, discussed in an Education Week article U.S. Graduation Rate Continues Decline

"Though the national graduation rate dropped slightly from 2006 to 2007, the EPE Research Center’s closer examination shows that each major racial and ethnic group posted at least a marginal gain in that period. This seemingly contradictory finding poses any number of questions, including: How can that be? The answer actually lends an important insight into the nature of the challenges inherent in tackling the dropout crisis. And that answer, to a large extent, is: Simpson’s Paradox.

A familiar concept within statistical circles, but rarely part of mainstream discussions, Simpson’s Paradox observes that there are circumstances in which disaggregated trends (such as graduation rates among minority groups) may not track closely with aggregate trends (for example, the nation’s overall graduation rate). There even can be times when aggregate and disaggregated trends run counter to one another. In such cases, some initially unnoticed factor usually accounts for the non-intuitive findings.

Shifting demographic patterns are the likely explanation in the case of graduation rates. Over time, the public school population has come to consist of proportionally fewer traditionally higher-performing white students and more members of historically underserved groups, most notably Latinos.

All else being equal, population growth among groups with low average graduation rates will tend to suppress improvements in the overall graduation rate. Pertinent to the case of high school completion: The size of the Latino student population, whose graduation rate currently lags 21 percentage points behind that of non-Hispanic whites, has grown by 50 percent in the past decade alone.

Put simply, the challenge of improving high school graduation rates is analogous to swimming upstream against a rapid and generally unfavorable demographic current. Many observers would argue that there is room for considerable improvement across the entire student population. The seemingly paradoxical findings noted here, however, would further suggest that targeting intervention efforts intensively on rapidly growing and low-performing student groups will be a precondition for driving meaningful change in the graduation rate at a national level."

Catherine Johnson said...

I was talking to Barry about this ... and I realized that we're not factoring in what I believe to be the pretty significant watering-down of the high school diploma.

Not sure how to assess this (though I assume people have taken a stab at it).

If you have the same percentage of each group graduating BUT the final degree signals less achievement, then....that's not good.

I think I mentioned on another thread the 6th grade papers from the 1960s my neighbor's husband read at her dad's funeral service.

We were all marveling at how well written they were.

One after another, these were perfectly composed little themes.

6th grade.