Caplan, Hanushek, and my own views on education

A reader asked me to comment on the debate between Bryan Caplan and Eric Hanushek on the extent to which education confers real skills or is merely a signal. I thought that the only point that Hanushek scored was when he produced data showing that the sheepskin effect is smaller than in other studies.

As a proponent of the Null Hypothesis, I am not the one to defend the human capital view. Where I differ from Bryan is that I am inclined to put even more weight on an ability-bias story, leaving less room for signaling. For example, my understanding is that the differences in earnings between people who are accepted to Ivy League schools and similar people who are not accepted ends up being pretty small. If it were mostly signaling, then losing out on the brand-name seal of approval should be more costly.

12 thoughts on “Caplan, Hanushek, and my own views on education

  1. In thinking about this issue, I always end up with calling it a Excel circular function that the best and brightest in general understand the employers pay more for education and therefore excel at it. And knowing how signal in society is aa very important skill. (A circular function means you input variables end up be changed by the function.) So people seeing success around educated people will follow suit.

    In terms of education:
    1) Americans being over-educated has been with us for decades. I remember in 1992 we heard Americans are over-educated or I believe there was literature in Depression that women should not go to High School and should work and get married.
    2) In terms of education debates, I really don’t understand why there is not more discussion of labor demand and market. If employers did not value education or the sheepskin effect then they would not pay extra for it. And why do employers offer education assistance?
    3) The libertarian economist who fight against education the most don’t appear to follow the same thing with their children. They have great reasons for this.
    4) I can’t help but think Bryan Caplan dream society was Victorian urban living where the working class died by 45.

    • There are far cheaper ways, the employee/employer combo, to determine skill of a prospective employee, and there are far cheaper ways to enhance the skills of an employee you want to advance, but none of these ways are practical in their legality. This is why employers farm this out to higher education- it shifts the cost to the employee before he gets the job, and afterwards, it provides partial legal cover to career advancement decisions that might be challenged by others.

      • Which ways are there? I know IQ test in general but what else? And IQ test only solves for certain things.

        • Any testing of applicants, even though ostensively legal, exposes the employer to costly litigation. Most businesses are not in business to defend hiring practices. To be successful, they need to minimize the cost of hiring, that includes the cost of lawsuits, frivolous or not.

    • In the early ’70s I read a book–I think a reworking of a Stanford PhD dissertation–that set out to test the idea that “jobs require more education now days.” He took (I think Department of Labor) requirements for various jobs, weighted them by their prevalence in the economy at two periods, and came to the conclusion that it just wasn’t true. Around the same time, A. Michael Spence was publishing papers and finally a book Market Signaling: Informational Transfer in Hiring and Related Screening Processes (1974) on the book’s title. Ironically, though Spence shared an Economics Nobel in 2001, the book is now unavailable and the idea that education is more about signalling than human capital formation hasn’t made much headway in the profession. I suppose there’s a cynical economic explanation for that.

      Charles Peters, longtime editor of the old Washington Monthly, hated credential inflation. He was concerned that it hurt poor people. Pretty consistently, children of poor people do worse in school than children of rich people. Requiring many years of schooling to get a job is as unjust as requiring skill at playing polo. He commissioned a number of articles on developing better, more focused, more creative, tests for jobs (he wasn’t a fan of paper and pencil tests for anything other than academic subjects). But he gave that up after Griggs.

      Griggs v. Duke Power was a Supreme Court case, decided in early 1971, interpreting the Civil Rights Act of 1964. It held that ANY condition of employment that had an “adverse impact” (today we would probably say “disparate impact”) on blacks was prima facie illegal. You were guilty of a violation unless you could prove that the condition of employment was necessary and a very good predictor of success in the job. There was an obvious catch-22 here. No company wanted to take the risk (both legal and publicitywise) of being found in violation, so no company was likely to be in a position to gather the data to make the proof.

      The two qualifications struck down in Griggs were scores on an IQ test and a high school diploma. North Carolina had only recently desegregated its schools so it was pretty easy to see the latter as a continuation of the system of white supremacy. But schooling requirements now are almost always accepted by courts while IQ tests are pretty much never.

  2. From reading his blog and listening to interviews, I don’t think Caplan says that education (learning things) is worthless, but rather that our current education system is very costly and inefficient. That should be obvious to anyone who has earned any degree from a college or university.

    The exception might be for those who end up becoming college professors. Their college experience is probably quite applicable.

    I have long argued with friends and acquaintances about how wasteful and costly a college degree is, even while I acknowledge that for many professional fields that still use the degree as a barrier to entry, there is almost no other way around it.

    I become more confident that I’m right when most of the arguments I get in return are, “well, that class taught me how to think” or “those classes make us better citizens” or “even though it was useless, it taught me how to learn”. Bah.

  3. I agree that Hanushek did not really move the conversation much, most of it was just asserting that portions of Bryan’s arguments had to be wrong on their face. He only scored a *potential* point on the sheepskin effect, but he seemed to ignore the weight of the literature here (and strangely seemed to have never heard of GSS) and Bryan’s reply was strong.

  4. I proposed an experiment. After 10 years or so, permit student loan debt to be discharged in bankruptcy. The only caveat is that the individual can no longer claim to have the credential, nor credentials earned because they had the one defaulted on. Nor are the universities permitted to acknowledge anything other than enrollment for the defaulter. No transcripts, no degrees conferred. The defaulter is permitted to retain and use any knowledge they gained.

    Then see how many people take up the offer. Debt discharged, but sheepskin burned.

  5. I vaguely recall a study that indicated that there was little difference in life outcomes between people who attended prestigious schools and people who were admitted to the same schools but declined to attend.

    • I remember the same study. At the high end it is less about what we do to you than who shows up.

  6. A subtlety: Ability counts twice in Bryan Caplan’s analysis, if I have understood correctly.

    1) Ability counts as “ability bias,” which explains roughly half of the college “education premium,” or the average earnings difference between two sets of equally able individuals, if one set has completed college and the other set has completed high school (and no college).

    2) Ability counts also through “signaling.” (Caplan estimates that signaling explains roughly 80% of the part of the college education premium that “ability bias,” defined in no. 1 above, does not explain.) The college degree degree signals three qualities: intelligence, conscientiousness, and conformity to social expectations. Each of these three qualities may plausibly be construed as ability.

    (A separate question is, What explains ability? One story is genetic, but evidence about heritability of intelligence, conscientiousness, or conformity is contested, and genetic mechanisms have not been established.)

    Insofar as intelligence, conscientiousness, and conformity are abilities, rather than skills imparted by schooling, then, to that extent, the “ability bias plus signaling” story (Bryan Caplan) and the Null hypothesis (Arnold Kling) are roughly equivalent.

    If I have understood correctly!

  7. My comment needs rephrasing! (The matter is indeed subtle.) Here is another try.

    A subtlety: Ability counts twice in Bryan Caplan’s analysis, if I have understood correctly.

    1) Ability counts through “ability bias,” which explains roughly half of the average earnings premium of college graduates. Research about ability bias tries to estimate the effects of pre-existing abilities; namely, intelligence and a a variety of other traits that correlate with educational attainment and earnings. For example, one technique is to compare earnings of people who have different credentials (e.g., college vs. high school) but similar pre-existing abilities.

    2) Ability counts also through “signaling.” (Caplan estimates that signaling explains roughly 80% of the part of the college earnings premium that “ability bias,” defined in no. 1 above, does not explain.) A schooling credential signals a bundle of three qualities: the average intelligence, conscientiousness, and conformity of people who have the credential. Each quality may plausibly be construed as ability. The credential then signals a degree of ability that the job market otherwise would fail to recognize and reward.

    (A separate question is, What explains ability? One story is genetic, but evidence about heritability of intelligence, conscientiousness, or conformity is contested, and genetic mechanisms have not been established.)

    Insofar as intelligence, conscientiousness, and conformity are abilities, rather than skills imparted by schooling, then, to that extent, the “ability bias plus signaling” story (Bryan Caplan) and the Null Hypothesis (Arnold Kling) are roughly equivalent.

    If I have understood correctly!

Comments are closed.