Posted by on February 25, 2017

I know that it sounds weird but I think I am becoming a Bryan Caplan fanboy.

Caplan is the George Mason University economist who has appeared before in this blog, notably for his excellent book The Myth of the Rational Voter: Why Democracies Choose Bad Policies.

Recently, I have heard Caplan debating the libertarian case for a universal basic income (“UBI”).  He was against the proposition and the debate was strictly Christians versus lions, with Caplan in the role of the big cats.  If there is a libertarian, or any other, case for UBI, it remains unproven.

Even better was his participation in a recent panel discussion of “Is College Worth It?” sponsored by the Cato Institute, a  libertarian think tank.  Caplan, who is writing a book on the subject, analysed the question from two perspectives, that of the individual (the private returns) and that of the country and its taxpayers (the social returns).  Spoiler alert:  Caplan concludes that “if you are a smart student, college is a great investment.  If you are a smart taxpayer, college subsidies are a waste of money.”

Looking at the private returns, Caplan starts with the much-quoted figure that college elevates lifetime earnings by something like 83%.  But you can’t take this number at face value.

The obvious first distortion is that the college-bound population already benefits from characteristics that lift earnings, such as higher IQs.  Adjusting for these, Caplan thinks that the real average improvement is closer to 45-50%.

But these average returns vary predictably.  One very important factor, as this blog has noted before (see Sub-Prime Student Loans here), is the course of study.  Controlling once again for other factors, Caplan finds that engineering students experience about a 60% uplift in lifetime earnings from a degree, but at the cost of taking a hard subject that leaves little room for college playtime.  At the opposite end of the spectra on both earnings and hardness, education majors only see about a 20% increase[1].

However, these are the results for someone who successfully finishes college.  It is also necessary to figure in the chances that this won’t happen.  Here, Caplan finds another source of consistent variability.  The likelihood of a good student, as determined by high school math scores in the upper quartile[2], finishing a college degree is about 73%.  This probability falls monotonically through the remaining quartiles, hitting a dismal 11% with the bottom group.

Hence, the first part of Caplan’s conclusion.  College is a “good deal for good students” – who take a hard and lucrative major, and have a good chance of finishing the program – “a mediocre deal for mediocre students and a bad deal for bad students.”  Caplan is not a believer in sugar-coating his messages.

But these are the private returns on investment.  The argument for government subsidies for college education is that it produces social returns that are even larger.  This requires a different calculation.

There are two dominant schools of thought on the returns to education.  The traditional view is that college builds valuable human capital, which translates into elevated earnings.  The alternative view derives from game theory and holds that successfully graduating college is just an elaborate “signal” to the employment market that you have the right stuff for a high-paying job[3].  Caplan presents strong arguments for the latter.

He starts by noting that there is very little vocational that is taught in schools.  This is self-evident for majors like Medieval History and English Literature, but research shows that even students who take classes seemingly applicable to their careers, such as advanced mathematics, only use them about 25% of the time.  My personal experience bears this out.  Having spent my career in a pretty quantitative field, I can report that very accurate lower math was pretty much all that was required.

There is also the “sheepskin effect.”  This is the well-known fact that the returns to a college education do not accrue evenly over the four-year program.  Something like 70% of the earnings increase comes from finishing the last year and getting the diploma, a fact that is inconsistent with the linear build implied by the human capital theory.

Caplan also cites the returns that accrue to an entire country from higher education, divined by looking at countries that have ramped up college expenditures for example, and finds that these are small compared to the private gains.  If college education were building valuable human capital, we would not find this disparity between the aggregate and individual returns.

Finally, Caplan notes that student and professorial behavior is much more consistent with signalling than skill building.  Students famously rejoice when classes are cancelled due to teacher illness or snowstorms, which is a bizarre reaction for consumers which are ostensibly being shortchanged.  Students sweat exams but make only minimal efforts to retain the material after the grades are handed out.  Professors make almost no effort to confirm that only registered students attend classes; interlopers are tolerated because, although they can take the learning, they will not get the diploma without paying fees.  None of these phenomena would exist if the real effect of college was building skills and not just signalling employability.

After his review of the literature, Caplan concludes that roughly 80% of the earnings effect from college comes from signalling, with only 20% the result of skill building.  Put this together with his earlier observations about the private returns to college education, along with its exploding cost, and Caplan thinks that the social returns are negative.  The policy implications of this will come as very bitter medicine for friends of Bernie Sanders.

The first implication is that the great bulk of the money spent by all levels of government on college education is wasted in a diploma “arms race.”   Skills are not being built and the signal loses its strength if everyone has a diploma.  Caplan makes the classic analogy of standing at a concert to get a better view of the performers.  Once everyone does it, no one can see better but everyone gets tired legs.  This is what is happening with college degrees. Masters degrees are now becoming the educational equivalent of standing.

Making college a mass product has other predictable negative results.  The quality of degrees has been systematically diluted.  There is plenty of evidence that, for example, students today spend many fewer hours hitting the books than their predecessors.  Although the data are limited here, they also appear to attain less: an assessment of adult literacy done in 1992 and 2003, and unfortunately not repeated since, showed sharp drops for all levels of advanced education.  Grades inflate, even at places like Harvard, which has become like Lake Wobegon where all the kids are above average.  And more and more college graduates and post-grads find themselves doing jobs that don’t require a degree.

The solution to this arms race is, of course, to de-escalate.  Cut or eliminate government subsidies to college and instead focus on changes, notably through greater competition, that will improve the quality of K-12 education.  We should also downplay college as a universal objective, particularly because this further stigmatizes the vocational training that is a much more effective route for many students.  In other words, as is so often the case, adhere to the first rule of holes: stop digging.

These results also point to the very useful role that private lending could play in funding college education.  As I pointed out in my brief discussion of Climb Credit (see Teaching Math and Science here), the fintech startup that is using data mining of career results to guide its student lending, a real private market in student loans would provide very useful information to prospective students.  The student in the last quartile with a hankering for gender studies would get an unambiguous message from private lenders.

Finally, this all points to one of my pet beliefs: there is a fortune to be made by someone who finds a better way of educating students and “signalling” their employability.  This would economize on a huge amount of time, effort and money.  A business that captured just a tiny percentage of these savings would be very valuable indeed.

But Caplan just can’t resist raining on this parade, too.   According to him, one of the most important signals that a college degree sends is the willingness to conform and take orders.  He thinks that it will be tough to come up with a non-conforming way to signal conformity.  Even for a fanboy like me, I think that this is Caplan taking his incendiary contrariness a bit too far.

A Brief Note on Taxes

I am surprised that there has been so little comment about one aspect of the Republican tax reform plan, which is the elimination of federal tax deductions for state and local taxes.  I have always wondered why the Republicans have never gunned for this deduction before, from which residents of the high-tax, bi-coastal blue states are the biggest beneficiaries (with California residents saving $97 billion in federal taxes and New York residents saving $68 billion).  Imagine the fun of watching liberal Democrats defending the right of the high-income itemizers in their profligate home states to be subsidized by the white-supremacist losers in flyover America!


Roger Barris

Weybridge, United Kingdom



The Trump administration correctly deems transgender bathroom rights to be an issue for the states, but then threatens to enforce federal drug laws in states that have legalized marijuana.  WTF?


[1] I am pleased to report that Caplan tells his students that economics is the highest earning of all the easy subjects.  Love to hit that sweet spot.

[2] Here, Caplan’s analysis was constrained by  the lack of data on other potential predictors.

[3] For a further exposition of the opposing arguments, I recommend the “Econ Duel” video between two of Caplan’s colleagues, Tyler Cowen and Alex Tabarrok, entitled “Is Education Skill Building or Signalling?”

Read Offline:
Posted in: Economics, Policy

Leave a Reply

6 Comments on "College: Capital or Signal?"

newest oldest most voted
Notify of
Neil Winward

On the hypocrisy of states’ rights I completely agree.

On the value of college education, look at the following study by Case and Deaton:

Rising morbidity and mortality in midlife among white non-Hispanic Americans in the 21st century

Dave Anderson
Re the tax cost of capping itemized deductions — its a complex analysis, at least for high income taxpayers, since Alternative Minimum Tax already reduces the benefit of itemized deductions. I haven’t done the analysis, but AMT offset certainly undercuts the dollar-for-dollar analysis that appears in some forums. Another potentially offsetting factor is that, as a thought experiment, imagine that federal taxes were reduced to zero, and all government services were provided by state level taxation. This would mean that no taxes raised in California would go to Kansas, for example. This is a bigger deal than usually imagined, because… Read more »
Dave Anderson
Good points. Maybe our combined point is that attacking the state tax deduction is a perfect political move — lots of headlines, but the people involved aren’t actually that much affected. Suitable for our current political climate. Let me extend by commenting on “the general issue of geographical fiscal redistribution” you refer to — I’ve been interested that the progressive tax structure combined with higher cost of living in California, for example, means that lower income people, especially, in California pay more federal tax per economic dollar of income than a taxpayer in Florida. I once calculated that as I… Read more »