unobserved skill

Andrew Kliman Andrew_Kliman at
Thu Oct 15 12:35:07 PDT 1998

(1) I agree with what Doug, Heather, and Jim Devine have to say about the "unobserved skill" stuff, and human capital (HC) theory in general.

(2) Some questions. The HC studies used to try to attribute as large a portion of wage differences as possible to "skill" differences, and the remainder was attributed to "labor market discrimination" and perhaps other variables, including "skill differences," that they couldn't measure. The basic aim became that of finding new measures of "skill," or new ways of writing down the relationships, that increased the portion of the wage difference attributable to HC. Nowadays, it seems from what Heather is saying (please correct me if I'm wrong), that many of them don't even bother. They simply attribute *all* wage differences to HC factors, apportioning the total difference between the portion due to the "skill" variables in their equations and the remainder, which is then attributed to "unobserved skill differences." Is this right? If so, then they're making the a priori *assumption* that there's no "labor market discrimination," right? So what do they think they're proving? Or have they consciously abandoned the task of testing their theory empirically?

(3) EVEN WITHOUT the "unobserved skill" dodge, the work that tries to attribute wage differentials to HC or skill differences -- in other words, HC empirical work as a whole -- is COMPLETELY BOGUS, PURE CRAP. The basic reason is that it is impossible to divide wage differentials into a portion due to skill differentials, a portion due to discrimination, etc.

For instance, one's years of schooling is supposedly a key HC variable, one that reflects the person's level of "skill." But years of schooling also differ due to discrimination, and everyone knows it, including the HC people. So, in response, they create the nice division between "pre-market," or "societal," discrimination, on the one hand, and "labor market discrimination" on the other. Discrimination in schooling is supposedly of the former type. The point is to absolve employers, as distinct from "society," of discrimination.

But the distinction is absolutely ridiculous. Not only does the amount of schooling affect wages, but wages affect the amount of schooling. If someone realizes that, due to discrimination in the labor market itself, s/he will not benefit financially from additional schooling, s/he will tend not to "invest" in additional schooling. (How rational is it for the average kid in the ghetto to "invest" in the "skills" needed to become an investment banker?)

So "labor market discrimination" is ITSELF a cause of "pre-market" discrimination! As a consequence, differences in years of school, the pre-eminent measure of "skill" differences, in fact reflects discrimination in the labor market ITSELF! So it becomes completely futile to try to separate out the portion of wage differences that stems from "skill" differences and the portion that is due to "labor market discrimination" and other stuff. The resulting numbers are total garbage. But these are precisely the numbers that the HC people have served up, again and again.

The same thing arises in connection with other HC variables that supposedly measure post-school "skill acquisition" -- on-the-job training, etc. -- such as length of tenure in a particular job or a particular industry. If, due to discrimination in the labor market ITSELF, you're stuck in jobs in which you're continually getting laid off or fired, or in temporary jobs, then you won't "acquire" much "skill." The same is true when you decide not to stick around on a dead-end job, because it is a dead-end job. So here again, a portion of wage differentials that is supposedly due to "skill" differences, not "labor market discrimination," is indeed due to labor market discrimination, because the skill differences themselves stem from labor market discrimination.

BTW, the HC people, at least some of them, are well aware of these problems. In their papers, they make cryptic references to the problems in passing, then blithely proceed.

(4) There are other factors that also make the HC studies bogus. These boil down to the fact that the variables that supposedly measure differences in "skills" in fact measure other things. For instance, years of schooling: people with more schooling tend to get paid more, but is that because they are more "skilled"? Or because they are more "credentialed," because they have learned to do what they're told, etc? Or take "quality" of schooling as measured by achievement test scores: do the differences in the scores reflect "skill" differences or cultural bias?

(5) In sum, the "empirical results" of HC theory are such a crock, such an utter waste of time and effort from the vantage point of true knowledge, that, if it weren't for the ideological support they give to the powers that be, no one would take them the least bit seriously. But if it weren't for their ideological import, no one would have tried to foist such garbage off on us in the first place.

Andrew ("Drewk") Kliman Home: Dept. of Social Sciences 60 W. 76th St., #4E Pace University New York, NY 10023 Pleasantville, NY 10570 (914) 773-3951 Andrew_Kliman at

"... the *practice* of philosophy is itself *theoretical.* It is the *critique* that measures the individual existence by the essence, the particular reality by the Idea." -- K.M.

More information about the lbo-talk mailing list