Bryan Caplan  

A Question of Educational Discrimination

PRINT
Why So Little Exploitation of ... Lincoln: Public Choice 101...
My favorite question from my latest Labor Economics final exam:
Some sociologists have argued that discrimination on the basis of educational credentials should be illegal.  What do the human capital and signaling models of education predict about the effect of such a law?
Your answers?  I'll honor the best responses and offer my own around Christmas.


Comments and Sharing





COMMENTS (18 to date)
mobile writes:

Under both models the law would be a boon to lawyers who graduated from top-tier programs.

Brad D writes:

The "seen" would be decreased interest in traditional colleges & universities. Learn your skills somewhere else more cheaply and more quickly.

The "unseen" would be trickier to predict. Perhaps we'd see more occupational licensing requirements. Maybe apprenticeships would make a comeback. However, other methods of signalling would emerge in the face of such a law.

Tracy W writes:

Are there any real sociologists who suggest this or is this a toy problem? And if so, do you have a link to their arguments?

Okay, under the signalling theory, people would need to find some other way of signalling their ability other than educational credentials. So, for example, volunteer work for multiple years would become more important. There would be some unemployment resulting from the disruption of existing patterns of employment, as the education system dissolves, but in the long-run economic output would be about the same.

Under human capital model, employers would have difficulty in working out which job applicants had in fact learnt useful skills. Consequently, economic output would fall.

Tom West writes:

Tracy W has it completely correct in both the sentiment to the original question and the likely outcome.

Hadur writes:

This gets me thinking about what signals could be used as proxies for educational credential signaling.

Fraternity and Secret Society membership?

On-Campus Activities?

Game Show Appearances?

Parents are willing to pay $100K+ to send their goods to a school that looks good on a resume. They would certainly find something else to spend that money on. The human interest articles in the New York Times would be hilarious as they chronicle this unironically.

Anthony writes:

In the human capital model, employers discriminate on the basis of educational credentials because they are interested in the stuff students learned in college. Under this change, they attempt to come up with ways to measure the stuff students learned that's independent of the credential. Students still go to college, because they still want to learn that stuff (although there may be a drop of some magnitude, if the ways to assess skills aren't very good.) The fall in output is dependent on the costs of measuring that stuff.

In the signaling models, employers discriminate on the basis of educational credentials because of what it sorts for. This could take two forms - either what it sorts for upon entrance or what it sorts for upon graduation (conscientiousness, etc.) With using educational credentials banned, college attendance drops. Employers have to come up with other tools to use. Again, the fall in output is dependent on the costs of measuring that stuff.

It's not possible to say which causes a higher decrease in terms of output. That depends on how good the substitutes are for assessing skills vs. assessing whatever it is education is signalling for. The big difference is the bigger drop in schooling in the signaling model compared to the human capital model.

Brian writes:

Counterintuitively, there would be no effect at all for either model if academic credentials were removed as a discriminator.

Hiring would be done exclusively on the basis of skills listed on the resume, job interviews, and personal contacts. Skills would be obtained either through schooling or through work experience. Job seekers who couldn't claim specific skills from either of those two sources would be unlikely to apply in the first place. For those claiming such skills from school (without SAYING it's from school), they would either possess those skills (human capital model) believe they do (signaling model), or are looking to fool the employer (cynical signaling model).

In the job interview, those who went to college would have an advantage over those who didn't go to college and were never previously employed in that field. By the human capital model, such interviewees would be knowledgeable about the field and would express themselves better because they had been trained to do so. In the signaling model, the same interviewees would be smarter and more conscientious (because that's what college selects for).

Finally, those who went to school, whether they learned anything or not, would have more contacts and recommenders available than those who did not. This too is independent of the model of education one chooses.

Ivan writes:

Assuming that the application of such law would be 100% effective, we would certainly expect the educational attainment (in terms of high school and college graduation) to drop, as the ROI for this signaling method would quickly diminish.

However, I predict that another signaling method, which potentially would be just as costly and complex as the current higher education system would spring up almost immediately. It could be some elaborate volunteer work scheme, or I could also see the testing companies such as ETS re-brand themselves and offer IQ tests that the job candidates can show to their potential employers. Higher education is a very inefficient way of signaling ability, but if this system was eliminated, there is no guarantee that the new model would be any more efficient.

The bottom line is that eliminating job discrimination based on education will not eliminate the demand for a strong signaling model. In the world of imperfect information employers will still need to discern their job applicants based on ability and elaborate schemes will ensue to accomplish that goal.

jure writes:

i think that it is very hard for our sociologists to determine when you actually discriminate or not. If they pass such law, and i graduate at harvard, but you graduate at No name faculty- then i get the job but you don't- how can they be sure that the reason for hiring was my harvard bachelor's ?

But i have a question for everyone. Why do we always assume that employer actually cares for your bachelor's degree? Well in a way he does, but how many employers actually DO check if your CV is one big lie?? My guess is , that almost none. If employer say's that precondition for application is a bachelor's degree- and you don't have it, but you do possess ability to sell a fog about yourself; What is the chance or risk that he is gonna get you? How many of them really checks if you ever walked through the community college or ivy league university ? And what is the cost of such decision for an applicant? Well, if you only lied about your college and never forged- it's not big deal and risk for you at all. Better this than 4-6 years in college - and out from the real world.

Consider this: In my country there were recently discovered by media many forgeries done by politicians. So many MP's had/have bought their bachelor's degree. I also know few who have nice jobs, but never actually made it by them own through the college. Simply bought degree.

Bryan, Interesting stuff for your book is to collect data about: if politicians have so many fake degree's (and media checks them severly compared to joe the plummer) how many employees have faked their degree's? Or lied. My guess is that when you succed in faking your CV- you already have a job, what means that no devil is ever gonna waste his time on checking your history.

And next interesting study can be made if you simply collect many applicants and let them try to lie ybout their CV's. By doing this you can also find out how to lie effectively/ and how not to.

Only bad thing about faking is again based on signaling effects of schooling. If you need to fake your credentials (your signals actually) and lie about CV to get nice job- it is unlikely for you to be able to handle it.(although your ability to sell yourself might signal your persuasive skills)

Philo writes:

Does "discrimination on the basis of educational credentials" mean *any* taking of educational credentials into account in hiring, promotion, etc.; or does it cover only *unjustified* taking-into-account?

Silas Barta writes:

It seems to me that they both make the same prediction: that employers will switch to the nearest legally-defensible correlate of "having gone to a good school" (and take every opportunity they can get away with to flout the law). Perhaps they'll go even *deeper* into the land of "we consider only people we've met personally", and then just get new hires from going to alumni events. (Oops, gotta ban going to alumni events for colleges you didn't attend. Rinse and repeat.)

The signaling model predicts this, of course, because college sends a strong, useful signal, and employers will still want to use it.

The human capital model, while claiming that all or most of the benefit of college comes from increasing the student's productivity, nevertheless recognizes that this is hard to identify in an interview, and so depend on the college signal at the hiring stage, and will take similar steps to recover the illegal signal.

This is just like what was proposed in the movie Gattaca: genetic discrimination will be nominally illegal, but employers will covertly test you anyway and then "mysteriously" lose interest in those that don't pass. The more the law focuses on a specific part of the search phase (interviewing, where they look, whether they hire, etc), the more employers will obscure or bypass that step.

(Of course, that's all assuming that the law is passed as an isolated intervention. In reality, it would only come about as part of change in public opinion, which would itself influence how hiring decisions are made.)

Anthony writes:

I should have been more precise before. In the pure signaling model, no one goes to college anymore if educational discrimination is banned (assuming totally effective enforcement.) In the pure human capital model, degree attainment doesn't change at all (assuming employers can perfectly measure skills.) If there continues to be any college in the pure-signaling model, it's because the law isn't being completely enforced. Whereas if college decreases at all in the pure human capital model, it's because whatever proxies employers come up with to measure skills are worse than college, either because they introduce more error or because they are more expensive. In either of those cases, college-going would decrease. But if employers come up with something that's just as good a proxy for human capital as educational attainment (in terms of both accuracy and cost), there's no shift in terms of college-going behavior (or subsequent employment). Nothing happens.

Duncan Earley writes:

In STEM fields I would expect an increase in interview stage testing and exams to confirm skills. I believe Engineering firms do this already and I know lots of software companies do. Collage attendance wouldn't change much in these fields as they provide concrete skills.

In Non-STEM fields I would expect an increase in internships and the length of them to confirm contentious and conformity. I would also expect more hiring of "who you know" or have met, e.g. some kid I met at the Harvard end of year function. You may also see more essay style tests in interviews in these fields. Collage attendance would probably drop significantly in these fields.

On last point, is it reasonable to say that the signalling from your education drops significantly after you have held a job for some time? Surely any future employer rates that much more highly than your education at that point. In other words this ban only effects recent graduates?

Jim Rose writes:

The ban is already in place in Japan and S. Korea for all practical purposes.

University diplomas and grades have no signalling value because everyone who passes the really tough university entrance exam then passes all subjects with an A!

The entrance exams are individual to each university. When they are underway, students who must move around campus are asked to make no noise.

The last year of undergraduate study is spent mostly sitting tough entrance exams that employers administer because they know university grades are of no value as information. Entrance to a high prestige university is taken as evidence of quality, but the workplace entrance exam still must be passed.

Because grades matter little, students slack off a lot because the value of whatever skills they acquire is not immediately recognised. Takes a little of the pressure of professors too! They put their feet up.

In a manner inconsistent with the human capital model, students do not work hard as a way of honing their skills for the employer exams. In a manner consistent with signalling, grades are no longer available to signal traits such as conscientiousness so why work hard?

Japanese high school students work hard and attend cramming school to prepare for entrance exams at age 13 and for university.

This is consistent with the human capital model. The skills acquired at high school have value so students work hard. Maybe schools teach the test?

Hugh writes:

Larger multinational corporations would hire graduate recruits on a short-term basis in a jurisdiction that did not have this legislation.

There they would be subjected to a battery of IQ, personality and knowledge tests.

Recruits deemed suitable would then be hired by the US mother company.

Smaller companies could band together to establish one foreign recruitment entity serving , say, 50 smaller companies to perform the same trick.

John Lewis writes:

I can think of a very clever reason to argue for that proposition.

Essentially we are in an education bubble of sorts. We have subsidized education with easy loans, but the rapidly changing technology of the day is not Hicks neutral. While this is not an obstacle across the board, it does mean that a lot of middle class folks who think that education is the pathway to sucess may find that it is a burden instead.


Basically a lot of folks have college debt that is not dischargeable in bankrupcy.... The solution is for the Federal Government to reform education in a way that will involve a revolutionary Supreme Court case answering questions concerning the status of intellectual property vis a vis the takings clause of the 5th ammendment. The tremendous payout on the class action suit would be sufficient to put us at full employment, and by releasing consumer debt would allow us to move on towards more narrowly tailored education reform and an economy once again more driven by consumer spending. It wouldn't bankrupt us, we are monetarily sovereign, we just need congress to raise the debt ceilling.

Of course the outcome would not be certain, and the result would do more to polarize(and possibly redraw political lines.) This of course would solve the Republican crisis of which way to go, while also involving a way of punting all the difficult questions to the supreme court! (It is perfect!)


In any case I think a sociologist might question what the human capital and signaling mechanisms signal not from the perspective of the agent, but from the perspective of the principal. While Lexis Nexis does provide an interesting service, a lot of this is trade secret.

What employers giving up " discrimination on the basis of educational credentials" would mean in practice would essentially be moot. No one policing the law would actually necessarily care about the policy working in this way, because it will be designed for a different set of unexpected consequences. In other words we are regulating the employer, not for the sake of regulating the employer, but for the sake of nudging the Jure's of the world towards re-appraising the value of an education. By looking like it is hitting demand(for education), the intended result is supposed to be something like what Antony suggests would happen to supply. (At least this might be the republican side of it, forcasting that such a restraint on alienation would not be a taking...not that they are so bold to stir up Kelo. Meanwhile democrats go with it assuming it will be overturned, thus effectuating a stimulus and because that is what the courts and administrative law is for, nowadays.)

But it is clear that if such a law is to pass muster constitutionally under a commerce clause and spending power justification perhaps, it would first have to be designed to avoid a potentially novel argument that this human capital or "brand"/" institutionally "aquired trademark" intellectual property should be viewed as real property, or that the "real property" distinction is arbitrary and capricious...under Monsanto the Supreme Court might recognize say the intellectual property of a firm that would replace the accreditation value of an education as protected trade secret (here I am thinking Lexis Nexis currently has a substitute already quasi-enabled)...certainly part of what is being alienated as a college education is the diploma, or certification, which has to be some form of property? (but not I think, contra some here, the whole of its value).

I am actually suspicious of trying to answer your question without understanding exactly what the status of the intellectual property in a college education is. While I suspect it isn't technically recognized as such under the Lanham act, I am thinking that both theories might be recognized in trademark law. Isn't a large part of the "Servicemark" of a lawyer or Doctor tied up in what the degree and certification signals to the customer/client about the professionalism of the individual?(in fact more narrowly such a restraint on alienation exists as a minority rule in legal ethics) So if for advanced degrees it certainly is part of Trademark, is a college degree servicemark for everyone? A designation of the source, origin and sponsorship of the person with the degree? Lawyers will always argue in the alternative but which economic theory has the most respect in a court of law?

In any case all this is a joke/rant... but in order for it to get passed as a law, all of this would have to be discussed, and a court challenge of the 5th ammendment variety is likely or foreseeable to occur.

My answer is thus the answer of Nanci Pelosi in regards ObamaCare, and my excuse is that this is a hypothetical and not one that I drafted...but even then! Once congress passed it and the president signed it, and after the Court answered just what it considers human capital to be in terms of private property, i.e. a deprivation of servicemark and thus capacity for signalling? Or is the law a species of argument about trademark/servicemark that considers the institutional servicemark of a university and seeks to prevent its dilution in commerce, by graduates who may combine the certification with a bad credit score and thus undermine the signaling power of the "brand" under a theory of dilution? Would the law be unconstitutional because it was a taking if such property interests are viewed as significant? Or could the law be a taking which would expose the U.S. to liability but still be justified by a public interest in full employment/consumer demand/shifting the national structural unemployment problem abruptly but in a clean break which removes the issue of promissory estopel? Would congress have anticipated and secretly wanted this result (because they want someone to blame) Would congress still grandstand and refuse to increase the debt ceilling or appropriate funds for such a settlement (You gave out the settlement, Supreme Court...now enforce it!) Or would the supreme Court rule that this is not real property and thus not a taking, or that a restraint on the alienation of a degree to an employer is not a complete taking, because the signaling portion of the property (i.e. its function as trademark) was taken but not the portion of that human capital that functioned more as patent(human capital)?

The answers above all seem to me to be answers directed at some fictional "long run"(It is a fatal conceit to think we can answer these questions!), which would indeed be greatly complicated by a foreseeable short run+intermediate run legal circus all of which would do a great deal to bring forward the best arguments concerning the economic value of human capital and signaling models. But quite potentially if a large enough 5th ammendment takings sum was granted and dispersed this could overshadow by a considerable amount the effects of a restraint upon the alienation of educational credentials. What sorts of standards a court case such as that might end up settling on in terms of valueing servicemark would itself become human capital in the form of precedent!

The answer if I understand what you are talking about would be that human capital would not be impacted, by a restraint on alienation (except in the long run if Jim Rose is correct). That is Jim Rose is saying that human capital is impacted because without a mechanism to signal accomplishment, concrete accomplishment, or learning for its own sake is discouraged. Potentially we could say that this is not a concern for human capital in the short run, since you would still have your education or set of skills. Maybe in fact folks would take to blogging, or entrepreneurship, as entrepreneurship (except for the problem that it does involve servicemark!) essentially is about what you can produce with your degree, not what you can signal capacity to produce. To some extent human capital is slightly naked without also having the signaling or recognition mechanism that I am interpreting the signaling theory to be speaking towards. I am sure this is an actual debate in economics to which I am ignorant, (in part why I wouldn't mind seeing it translated into a catagory of intellectual property.) While signaling would seem to fall under servicemark/trademark, thinking in these terms might actually lead you to eroneously believe that human capital is unrecognized potential, or potential incapable of being recognized(or simply not immediately or concretely given a functional use). A restraint on the capacity to alienate educational credentials would thus directly impact the signaling theory. So it might be a taking under a signaling heavy theory of human educational capital, but not a taking under a theory of human capital? My guess is that since the pro-takings theory might lean more towards a signaling approach, the human capital position if the two are legally bifurcated in this way would end up being turned towards the viewpoint that such a law would not be a taking. Albeit if human capital is what it is in labor law, I could see the argument being made that it is not readily transferable, requires considerable time lag...requires long term planning/forecasting (reinforcing a promisory estopel argument and going to public policy in terms of structural unemployment.)

Matt T. writes:

The human capital model predicts significant inefficiencies as potential employers find it more difficult to hire employees with the relevant training. Turnover and unemployment would rise among the educated for entry level jobs as firms cycle through more candidates searching for the desired set of training/skills. Interviewing techniques would probably evolve to include more direct testing of employees, possibly in large groups (picture every interview in an auditorium taking the equivalent of a final exam). Despite the difficulties in matching employee to firm, education would continue to be highly valued.

The signaling model predicts a smaller effect on inefficiency and turnover, as firms attempt to sort candidates on innate ability, conscientiousness, and conformity (more general) as opposed to specialized training (more specific). Unemployment would be lower for entry level hires than the human capital model predicts. Most importantly, the demand for education would plummet. Other employee sorting mechanisms such as aptitude tests would increase in importance. Bryan Caplan finishes "The Case Against Education" in mid-2013.

Jim Rose writes:

Error in earlier post:

Students applying to national or other public Japanese universities take two entrance examinations: a nationally administered uniform achievement test; and then an examination administered by the university that the student hopes to enter.

Applicants to private universities need to take only the university's examination.

Truancy among Japanese university students is extremely high. what is there to lose if grades do not matter?

A number of Japanese courses give 50%+ of all marks just for class attendance. 100% of marks just for class attendance are not unknown.

A colleague at the American university where I taught in Japan had a pop test at the start of every class to induce students to attend. The pop tests were in lieu of the final exam.

the american professors who doing everything they could to induce class attendance.

Graduate school enrolment is 4% of the entire Japanese university student population.

Comments for this entry have been closed
Return to top