« Department of "No"! | Main | What an Economist gets his Wife for Valentines »
Book Forum: Harford and Caplan on Statistical Discrimination
The Logic of Life contains an excellent chapter explaining statistical discrimination but does the theory hold up? Bryan Caplan says no.
...[Tim] heavily emphasizes a few experiments showing that statistical discrimination could be a "self-fulfilling prophesy." For example, he describes a resume experiment where otherwise identical fake resumes with "black names" were less likely to get a response. "High-quality applicants were more likely to be invited for an interview, but only if they were white. Employers didn't seem to notice whether black applicants had extra skills or experience." If that is how employers treat black applicants, what's the point of trying? As Tim asks, "Why bother to get a degree or work experience if you are young, gifted, and black?"
But is it really true that the market fails to reward blacks for getting more education? Is it even true that the market rewards them less? I tested these claims using one of the world's best labor data sets, the NLSY. The results directly contradict Tim's self-fulfilling prophesy story. Blacks actually get a substantially larger return to education than non-blacks! The same goes for experience, though the result is not statistically significant. The real lesson of the data is that if you are young, gifted, and black, you should get a ton of education, because it has an exceptionally large pay-off.
Why would this be so? I'm not sure, but one simple story is that counter-stereotypical behavior stands out. When my sons were young, my wife was working a lot, so I often took my kids places on my own. Funny thing: Time and again, strangers came up and said, "Wow, you're such a great dad!" But there were moms of young kids doing the same thing in plain sight, and the strangers rarely praised them. Why not? Because a dad taking care of two babies is counter-stereotypical, which grabs people's attention.
Purely anecdotal, yes. But it is consistent with the small academic literature on counter-stereotypical behavior. If you clearly violate expectations, people not only notice; they often over-react.
The upshot is that stereotypes may actually be self-reversing rather than self-fulfilling. The marginal payoff of distinguishing yourself from the pack is high if people think poorly of the typical member of the pack.
Bryan has much more on the unpleasant truths about discrimination. Read the whole thing.
Posted by Alex Tabarrok on February 14, 2008 at 07:46 AM in Books | Permalink
Comments
Caplan doesn't tell us exactly what his work on the NLSY found.
It is certainly possible that blacks get a higher return to education than whites, but are nonetheless discriminated against in the labor market. All that has to happen to make this so is that the gap between high-school and college-educated blacks has to exceed the the gap between high-school and college-educated whites.
There is no reason this can't be the case in a labor market that discriminates against blacks. Indeed, unless college-educated blacks earn more than whites the greater return suggests that less-well-educated blacks are discriminated against by comparison with their white peers.
Posted by: Bernard Yomtov at Feb 14, 2008 8:09:49 AM
Does Bryan have a paper where he fleshes out his claim to have "tested" Tim's hypothesis using the NLSY? Can someone provide a link?
If he simply observed correlations between education and pay then that is not a test since there are so many unobserved variables that could drive both variables -- correlation is not causality.
Ideally, you would want an experiment, e.g. randomized education being provided to treatment and control groups, with follow up over a period of years. The resume study that Tim references at least was a blind, randomized experiment. It showed that extra years of education on one's record had a lower pay off.
Posted by: A student of economics at Feb 14, 2008 8:17:34 AM
Small nit to pick: isn't the excerpted truth a pleasant one? So far as I'm convinced that education pays off for young blacks, that's good news. (It does seem that the rest of Caplan's post is rather unpleasant in the relevant sense).
Posted by: Justin at Feb 14, 2008 8:20:24 AM
The resume experiment was very interesting, but it only proved the existence of discrimination, not that discrimination affected wages in equilibrium.
If a store discriminates against me and treats me badly, I go to a different store.
In the case of labor markets, if some employers miss the boat and discriminate, other employers hire those workers. You could easily have an equilibrium where some firms simply hire a disproportionate share of minority workers, and wages are the same across workers.
In other words, just because there's discrimination, doesn't mean it has an effect. It certainly could, but you do need data like Caplan's, i.e. actual wage data, to see if it does.
Posted by: Keith at Feb 14, 2008 8:27:42 AM
Bernard
As I understand Caplan, he does not deny that discrimination exists, but he doubts that only the
"self-fulfilling" vicious circle works. His study shows that blacks have even higher incentives to
invest in education than whites. If it is the case, then discrimination should self-reverse.
Posted by: vic at Feb 14, 2008 9:47:33 AM
vic,
True. At least in this post, he does not address the question of whether discrimination exists.
Still, he does make the mistake of claiming that for statistical stereotyping to be stable the underlying stereotype must be statisticaly valid. This is wrong. The stereotype must be believed, true, but not all things that are widely believed are true. Didn't Caplan himself write a book making this point?
Another quibble I have is that he measures the return on education only in terms of the payoff, and not the cost. Is it possible blacks have a higher (non-financial) cost of going to college? if you come from a poor neighborhood, having gone to lousy schools, it may well be much harder to get through college, so the cost as well as the payoff is higher. And of course the investment may be riskier as well.
Posted by: Bernard Yomtov at Feb 14, 2008 10:37:12 AM
One thing to keep in mind is that fewer jobs are gotten by submitting a resume in response to a want ad than through personal connections (not even counting very indirect connections like those of a college job placement office). It may be that the return to education for African Americans comes in other places than during the 15 seconds that someone screening resumes spends assessing their qualifications while reviewing a resume. It seems reasonable to expect rules of thumb (which might include discriminatory assumptions) to be more prevalent when the vast majority of candidates are going to be unqualified (as has usually been the case in piles of resumes that I've seen).
Posted by: Telnar at Feb 14, 2008 10:55:36 AM
My understanding is that Tim and Bryan are talking about two distinctly different things.
Tim's argument suggests although not rigorously tested that blacks will make less in equilibrium. This is because blacks are discrimminated against in the labour market. As a result, they receive less opportunities than their white counterparts to gain employment.
Bryan however doesn't test Tim's argument; Bryan measures something different namely the marginal payoff (ROR) of receiving more education. While the marginal impact of increasing education is higher for blacks, it doesn't necessarily support or refute Tim's argument. If anything, one implication of Tim's argument is that blacks will have a lower *level* of income, completely independent of education!
Posted by: econometron at Feb 14, 2008 11:08:16 AM
I couldn't find any refereed article by Bryan supporting his claim that he tested the story using NLSY data, or any article or working paper at all, for that matter. All I found was this http://www.gmu.edu/departments/economics/bcaplan/e321/lab7.htm which has some correlations he apparently ran on these data. He reports that black earned $6200 less than whites, and that the difference was $5300 when he controlled for education and experience. He doesn't provide standard errors, but elsewhere he says that $900 is not a statistically significant difference. Thus, I could find no support for his claim that "Blacks actually get a substantially larger return to education than non-blacks!" [emphasis in the original].
More significantly, such correlations really don't answer the question because they are rife with problems of unobserved heterogeneity, missing variables bias and selection bias. For instance suppose the educated blacks live disproportionately in high wage, high education states. Then a correlation between wages and education in this population is not a measure of the "returns to education".
Likewise, the (published) resume experiment shows that things as trivial and subtle as one's name can affect employment prospects. This suggests numerous confounding variables that make simple inferences from NLSY data problematic. Methodologically, retrospective correlations are much less convincing than controlled experiments.
Unless Bryan or one of his defenders can point to better evidence, I'd have to conclude that his claims, while consistent with his ideology, are unsubstantiated by the evidence.
Posted by: A student of economics at Feb 14, 2008 12:29:26 PM
At every big company I've worked at, "underrepresented" groups are overrepresented in middle management and special groups (diversity groups, game-changers, leadership training programs, etc.). Why? Because upper management is so eager to show they are progressive and non-discriminatory that they pluck out any underrepresented employees who show any talent at all. So, group X is say 15% of the total population of the country, 4% of the population with requisite skills to work here, 10% of the employee population, and 20% of the special groups.
Posted by: Jim P at Feb 14, 2008 2:22:00 PM
Among the 1979 National Longitudinal Survey of Youth group of 12000 or so young people, as of 1990 when they were 25-33 years old, blacks were earning 98% as much as whites with the same scores on the military's heavily g-loaded Armed Forces Qualification Test.
Source: Herrnstein and Murray, 1994.
As a more general point, the government has been running these amazing longitudinal studies tracking thousands of people's lives for decades that can answer countless social science questions. The 1979 NLSY is particularly interesting because in 1980 the U.S military paid to have the NLSY panel of youths tested on the military's entrance exam, the AFQT, which is essentially an IQ test (it correlates at a high level with the classic Wechsler IQ test). We now have a remarkable amount of data on this nationally representative sample of people, including IQ scores for about 6000 of their children! I would encourage economists to familiarize themselves with these longitudinal resources.
Posted by: Steve Sailer at Feb 14, 2008 4:04:29 PM
As Freakonomics showed at length, first names are not a trivial indicator of who someone is. (While resumes are useful, too, they are hardly totally honest and reliable indicators.) Somebody named Dov or Ansel is likely to have come from an above average family background in terms of education and intellect. (Girls' names cycle through fashions faster, so it's harder to keep track, but in general, young women named after Jane Austen heroines, such as Emma, will come from a more responsible social class than women named after brand names, such as Tiffany.)
Imagine receiving resumes from two young men who both live in a what you know is a black part of town. One is named Michael B. Jones and the other is named D'Shqyuan Jones. What inferences can you reasonably draw about the aspirations and attitudes their respective mothers inculcated in them?
Posted by: Steve Sailer at Feb 14, 2008 4:18:58 PM
Steven Levitt and Steve Dubner wrote in Slate in "Would a Roshanda by Any Other Name Smell as Sweet?" about those super-black names that black mothers started giving their babies during the Black Pride era:
"The typical baby girl born in a black neighborhood in 1970 was given a name that was twice as common among blacks than among whites. By 1980 she received a name that was twenty times more common among blacks."
Levitt and Dubner show strikingly little sympathy toward blacks who have a harder time getting called in for a job interview because, as shown by numerous "audit studies", employers are dubious of DeShawns and Darnells.
Levitt and Dubner scoff:
"Was he rejected because the employer is a racist and is convinced that DeShawn Williams is black? Or did he reject him because ‘DeShawn’ sounds like someone from a low-income, low-education family?"
Sure, as the authors imply, a boy named DeShawn may indeed be, on average, more likely to goldbrick or to rip off his employer than a boy named, say, "Dov" (the male name with the most educated parents according to the book).
Following their Naturist inclinations, Levitt and Dubner conclude:
"And that's why, on average, a boy named Jake [the whitest common male name] will tend to earn more money and get more education than a boy named DeShawn. A DeShawn is more likely to have been handicapped by a low-income, low-education, single-parent background. His name is an indicator—not a cause—of his outcome. Just as a child with no books in his home isn't likely to test well in school, a boy named DeShawn isn't likely to do as well in life."
There aren't too many people who make _me_ sound like a diversity-sensitive multi-cultist, but sometimes Levitt is one of them! The authors could at least have a little compassion for the poor kid. DeShawn didn't ask to be given his name.
And I must point out that a recent study by economist David Figlio calls into question Levitt's assumption that DeShawns aren't hurt by prejudice. Figlio cleverly looked at siblings, and found that the ones with the blacker names tended to get rated more poorly by their schoolteachers even when their test scores were the same.
Posted by: Steve Sailer at Feb 14, 2008 4:23:55 PM
One issue that economists might want to study is: "What is the logic, economic or otherwise, behind saddling your child for his or her entire life with a first name that screams "Ghet-to!"?
Posted by: Steve Sailer at Feb 14, 2008 4:41:45 PM
If you clearly violate expectations, people not only notice; they often over-react.
Does this help explain the fact that even many non-lefties swoon over Obama? He's not Jesse Jackson. He must be ... wonderful.
Posted by: Roger Sweeny at Feb 14, 2008 5:08:13 PM
By the way, has anybody ever looked into how easy it would be to prejudice these resume-audit experiments in order to get the politically desired results? If we are assuming that employers are biased, why not hold similar suspicions about social scientists?
It might be not just deliberate sabotaging, such as smudging the resumes, but a lack of logic in the details that say the person submitting the resume is a "liar."
For example, say they send out two resumes to Chicago employers for a young lady who gives her address as posh Lake Forest, IL. They only differ in terms of names. One resume comes from "Kerri Emma Jones." The prospective employer says, "That sounds like a responsible North Shore girl raised in a solid family. Give her a call."
The other comes from "Carrion LaTreen Jones" of Lake Forest. The employer takes one look at it and says, "Nobody stupid enough to name their daughter after dead meat and a toilet can afford to live in Lake Forest. The whole resume is fake [which it is]. Throw it out."
In contrast, say the resume's address is given as 95th Street on the South Side of Chicago, where Senator Obama's church is. The employer says, "Okay, we need to hire some blacks or the EEOC will sue us. Let's see who looks like the best of the lot. Here's D'Shqyuan LaBradfurd Jones. Oh, dear. His mom was probably 16 years old when she picked a name that ridiculous out to impress her girlfriends. He sounds like trouble. His mom likely raised him to be extra ghet-to, so he'll probably sue us the first time we don't promote him. Forget him."
"Oh, look, 'Douglas MacArthur Jones' from 95th St. That's more like it! Obviously, his father was around to pick _that_ name out. His dad was probably a sergeant in the U.S. Army, so he raised him strict. Give him a call!"
Posted by: Steve Sailer at Feb 14, 2008 5:33:10 PM
The problem with Caplan's calculations is that he looks at returns to easily observed signals like education. Models of statistical discrimination that predict workers from the group facing discrimination will accumulate less human capital (e.g., Lundberg and Startz, AER, 1983) typically predict they will accumulate less hard-to-observe human capital. This type of statistical discrimination assumes groups differ in how accuately they can be evaulated. The result is that employers rely more on group averages and easily observed characteristics (like education) when forming expectations, and pay workers from groups facing this type of discrimination less by individual ability.
See my paper in the 2006 ILRR for a test of this idea, using the AFQT scores mentioned above:
http://digitalcommons.ilr.cornell.edu/ilrreview/vol59/iss2/5/
See Altonji and Pierret (2001) for a look at the "rational stereotyping" variety of statistical discrimination.
Posted by: Josh Pinkston at Feb 14, 2008 5:47:13 PM
One obvious way to ameliorate the burden of rational statistical discrimination on high IQ members of groups with lower average IQ is to repeal the 1991(?) Civil Rights Act that enshrined into legislation the 1972 Duke Power Supreme Court decision. That anti-testing law made it difficult and expensive for employers to evaluate job applicants using IQ tests and other objective measures that have "disparate impacted" on "protected" groups.
In other words, the solution for statistical discrimination is more IQ testing, not less.
Posted by: Steve Sailer at Feb 14, 2008 6:38:17 PM
So, why did the Democratic majority in Congress shove through an anti-anti-statistical discrimination bill in 1991? Because the bigger employment problem for blacks and Hispanics is not that high IQ blacks are assumed to be of the average IQ for their group, but that the group averages are lower. So, the point was to make it harder for business to find smart workers by taking objective tools away from employers.
Similarly, the outgoing Jimmy Carter threw out the elaborately validated federal civil service exam, PACE, in January 1981. It was supposed to be replaced by a test that would have predictive validity but not have disparate impact on blacks and Hispanics. That has proven impossible to develop. It's like squaring the circle. So, the federal civil service has relied on a hodge-podge of hiring techniques ever since, typically emphasizing resumes, which may explain a lot about the weakening effectiveness of the federal government compared to its mid-20th Century peak when the civil service exam was finding high potential job candidates.
In contrast, simple racial quotas in which everybody is tested and then ranked within their own race and hired according to quotas would probably be better for the economy as a whole than the government's efforts to aid minorities by, in effect, increasing the randomness of hiring decisions.
Posted by: Steve Sailer at Feb 14, 2008 6:50:03 PM
Commenter Joshua Pinkston's very interesting paper (URL above) found:
"Variables that are difficult for employers to observe, such as the AFQT score, had less influence on the wages of black men (and easily observed variables had more influence) than on the wages of white men."
The solution, of course, is to make it legal once again for all employers to give IQ-type tests (of which the AFQT is one) to all job applicants.
Posted by: Steve Sailer at Feb 14, 2008 7:08:20 PM
By the way, it would be in best interests of everybody if whites made the effort to learn the variety of subtle clues that middle class black men have evolved in recent decades so they can show other blacks that they are middle class without being accused of "acting white." For example, after Spike Lee's 1992 movie "Malcolm X," it became a fashion for college educated black men to wear glasses, especially glasses with metal frames and small lenses. And there's a distinctive black college-educated male accent that has evolved as well.
Posted by: Steve Sailer at Feb 14, 2008 7:17:03 PM
One issue that economists might want to study is: "What is the logic, economic or otherwise, behind saddling your child for his or her entire life with a first name that screams "Ghet-to!"?
By handicapping the child's ability to assimilate into higher society his/her commitment to social capital in the neighborhood is credible. They can't sell out because no one's buying.
See Acting White or Just Acting Rational
Posted by: TGGP at Feb 14, 2008 8:55:37 PM
I find many of the comments here deeply offensive.
Posted by: Yourhighness Morgan at Feb 14, 2008 10:39:10 PM
Has Sailer moved his blog into the comment section of MR?
And should we be GLAD, WELL, maybe maybe not.
Posted by: burger flipper at Feb 15, 2008 12:55:40 AM
Non-whites who I go to college with seem to have an equal, statistically insignificant, chance of obtaining a job through resume. In fact, among those who have applied to positions, I find are more likely to receive an interview than the white students. Especially a student with a non-white sounding name who has more than one expected degree. I think minorities in this manner have a high return on education, minus all the opportunity cost. Minorities are also more likely to be on financial assistance/scholarship, making the cost of education quite low. Also, during the interview process, all applicants are tested in an IQ based manner, from solving complex logistic problems to how much they know about the world. So I wonder, what kind of jobs are in question here? They must be jobs for people without college educations. So on the topic of the book, I trust the research is correct.
Posted by: Brainwarped at Feb 15, 2008 7:35:11 AM
