Arnold Kling  

Education, Assortative Mating, and Inequality

Junk Macroeconomic Science... Tyler on Signaling: Even the N...

I am going to combine comments on two papers cited by Tyler Cowen. First, he cites a paper on inequality and mobility. It argues that skill differentials are widening, and that parental education seems to be increasingly important. (Note that Peter Orszag has pointed to research suggesting inequality in longevity shows a similar pattern.)

Second, Tyler points to a paper on how firms learn worker quality. The significance of the paper is that firms can learn worker quality quickly by observing workers on the job. This implies that Bryan's favorite hypothesis, which is that firms use education as a signal for worker quality, lacks a rational basis.

Tyler writes,

In my view education is mainly about indoctrination to give you more productive habits. So yes it is learning, but not in the way they might have told you, and that is why it so often does not feel like learning.

Both of these papers support my pet theory of everything, which is that the value of college is in assortative mating. Consider the power of this theory.

1. It explains rising tuition costs. If your goal as a parent is to put your child in a milieu of affluent children, then a school with higher tuition may be more attractive to you. The demand curve will not be elastic--it may even be upward-sloping.

2. It explains rising inequality. Fifty years ago, well-educated men might have chosen wives for their ability to cook and clean. Now, they are looking for well-educated women, who in turn are looking for well-educated men. These matches serve to raise inequality by themselves. But they also increase the inequality in subsequent generations, by decreasing the share of children that come from combinations of high-SES and low-SES parents (SES = Socioeconomic Status). Instead, you see more families where both parents come from the same end of the SES spectrum.

3. It explains why education is highly correlated with income, even though the skill-value-added of college education appears to be low. Going to college increases your chances of landing a high-income spouse. That raises your income, even if your skills do not go up. Moreover, going to college is highly correlated with having affluent and intelligent parents, which in turn means that young people who go to college will themselves tend to be intelligent and start out with some financial flexibility to invest in their human capital in other ways (by traveling abroad, for example). So there will be a correlation between college and income, even if there is not much causation.

Comments and Sharing

TRACKBACKS (3 to date)
TrackBack URL:
The author at Newmark's Door in a related article titled Kling and Alchian on the value of college writes:
    Arnold Kling has an interesting post on the value of college is in assortative mating. I first heard that theory in 1978 in Armen Alchian's class. As far as I know, Professor Alchian never developed the theory in a paper, [Tracked on March 30, 2008 7:21 AM]
COMMENTS (13 to date)

Boy, I did this all wrong, then. With my Ph.D. I managed to make myself overqualified for almost every job (apparently most companies haven't gotten the news regarding the two papers that my Ph.D. primarily indicates that I have high worker quality), and though I did marry someone with a Master's Degree in Organizational Development, she is a Kindergarten teacher and doesn't really make all that much money. Of course, she came out even worse -- I have a Ph.D. . . . and no job!

Sharper writes:

I could be wrong, but I seem to recall rising tuition costs already being adequately explained by the ongoing increases in the government subsidy of education costs through grants, loans, and other "financial aide".

Those subsidies leading to the situation where the true amount paid by students and families hasn't increased much over inflation, but the price tag has gone up as the government has paid for more of it.

Or was that just someone's unverified theory?

Josh Lyle writes:

Not that I doubt your assortative mating thesis, but wouldn't the signaling model still make sense to employers that face large hiring costs or difficulty firing poor employees? Knowing that an employee is poor isn't helpful if you can't get rid of them.

Marcus writes:


I'd like to see you apply this reasoning to housing prices and perhaps even the bubble.

I've been trying to understand why market innovations don't drive the price of houses down. I think at least part of the explanation is precisely what you've hit upon here.

I've been calling it 'exclusivity' but perhaps 'assortative mating' is a better term. Basically, people are signaling information about themselves, their economic status and what they're interested in for neighbors and mates by the houses they buy.

If this is the case, or at least an important factor (though I'm not suggesting it's an exclusive factor) then all government aid for housing can do (ie. low interest loans or zero-downpayment loans) is drive up prices. And I mean more than what could simply be explained by increased demand.

In other words, as government assistance helped lower quintile people buy houses usually bought by higher quintile people this creates a pressure for those higher quintile people to buy even more expensive houses to keep themselves differentiated from lower income people.

What do you think?

BGC writes:

AK's ideas here are very interesting and seem convincing.

From what I have read, the best predictor of worker quality is IQ and the best selection mechanism is an IQ test (see Intellgence: a short introduction, by Ian Deary).

Approx equal in predictive power is a trial period of work of a few weeks, but of course that is extremely costly (and sometimes risky) for employers compared with a 15 minute test.

The second best predictor of worker quality is a higher level of the personality trait called conscientiousness (C.) (see Personality, by Daniel Nettle). However, C. is only measured (at present) using self-rating questionnaires, which can be faked.

Educational success is a matter of exams, and exams probably signal (broadly) how much 'stuff' you know. How much stuff you know depends on how quickly you learn (IQ) and application (C.) - so a lot of the selectivity of education is probably for some combination of IQ and C.

If you control for IQ and C. then _probably_ the benefit of education is probably much more modest than most people (excepy Bryan C) imagine. Of course, many top jobs require that you know a lot of specific stuff and learn specific skills - but the necessary level of attainment can probably be achieved in months rather than years.

If this is correct, then there is huge potential for shortening the period of time spent in general education, and therefore hugely increasing the efficiency of developed countries.

randy writes:

signalling pisses me off, so much.

Dan Weber writes:

If the purpose of college is to find a mate, going to a male-dominated technical school was really the wrong decision for me!

Although my wife is happy (I hope!) with the way things worked out.

Butter writes:

Interesting addition to your theory: Fraterities and Sorrities act as a collusive devise to limit competition for mates on campus. Unless you are inside of the Greek system, you can get excluded from access to parties where most mating activity occurs. This is why fraterities have "lists" for parties and only allow high quality people as members...they want to generate a certain degree of monopoly power in the mating market on campus.

Larry writes:

College's role in the mating ritual at least re marriage, has been declining for decades, as each generation further postpones and/or skips that step. Assortative mating has become much more workplace-oriented.

Instead, the notions that you can't get a good job without a degree and that a degree from a "good" school is incredibly valuable have become universally accepted.

A more logical but not widely accepted reason is that research shows that peer groups are far more influential than either parents or schools in what children become. The peer group found in colleges is for most people a significant upgrade from high school...

BGC writes:

It is interesting to speculate about long term trends.

Assortative mating - whereby high SES people marry and reproduce, and vice versa - will tend to increase inequality, and also increase the extremes of intelligence (and behaviours associated with intelligence).

IQ and personality are both substantially inherited.

However, average fertility (number of children) is inversely proportional to both IQ and years of education. So the higher the intelligence and more education, the fewer kids you will have on average; and vice versa.

In the UK the projection is that nearly half of current college educated women will not have children at all. And the more years of education the greater is this effect - presumably due to lateness of marriage and first child, and perhaps due to the lack of big families among high IQ people (which would compensate for those who reproduced at below replacment level).

This suggests that the cognitive elite will, for a while, probably get smarter, but there will be a lot fewer of them (leaving out any high IQ immigration).

My point is that the situation described by Arnold is dynamic, and might change radically within a generation (c 25 years).

david foster writes:

"firms can learn worker quality quickly by observing workers on the job"...maybe this is true for some entry-level jobs; it is not true for most management and senior professional jobs. It would generally take a minimum of 6 months to a year to assess a marketing or sales manager, for example, and the cost of a bad hire is usually a lot more than the salary & benefits paid to the individual.

Ken writes:


PhD's in homeopathic medicine is a PhD in baloney. I don't know anyone with a PhD in a credible field of learning that is too 'overqualified' to find a job (this doesn't even make sense). Sounds like you got a PhD in baloney.


The idea is not that you find a mate while in college, it's that colleges sort out those with degrees and those without degrees. Those with degrees seek mates with degrees and tend not to marry those without.

Hope this helps,

Scott writes:

In the US women outnumber men in college about 56% to 44%. If college is about assortative mating then it seems only the women are participating.

I also went to a male dominated engineering school. Employers really like to hire its graduates because they "hit the ground running." It seems that the school taught us a lot of productive habits.

Comments for this entry have been closed
Return to top