Politicians across the political spectrum tend to agree that education is a worthwhile investment. Bryan Caplan disagrees: he believes that while education pays off for individuals (though not for everyone), it’s a terrible investment for society at large.

His rationale is that education pays for individuals because it signals intelligence, work ethic, and conformity, not because it helps people develop useful skills. But because signalling is a zero-sum game, education’s benefits to individuals do not translate to gains for society at large. In this post, I want to review Caplan’s thesis, and suggest a few counterpoints that I think he has failed to address.

Education as signalling

Education raises people’s income. Even after accounting for factors such as ability, a bachelor’s degree raises earnings by 35–50% (in the US). The question is what is causing this ‘education premium’? One school of thought says that education builds human capital: it teaches students useful skills that employers are willing to pay for. The signaling school of thought, on the other hand, argues that education signals attributes employers are looking for: to graduate, students need to be smart, hard-working, and willing to conform. Employers use degrees to sceen for these qualities.

To be clear, Caplan does not completely dismiss the human capital story, but argues that signaling is by far the most important factor, accounting for ~80% of the education premium.

That education does not impact knowledge directly applicable to the job market is fairly uncontroversial. First, most of us work in jobs that are completely unrelated to our degrees — I have worked with marketeers with degrees in Classics, which is by no means uncommon. Secondly, most of us have forgotten the vast majority of things we learnt in school (and Caplan presents a lot of evidence corroborating this). So, arguing that employers value the knowledge we acquired during our studies, when most of us delete this knowledge from our brains right after our exams, and when this knowledge is irrelevant to our jobs anyway, is rather unconvincing.

Proponents of education, however, argue that the value of schools and universities is not the explicit knowledge they impart in their students, but their teaching students how to think. Caplan counters with several studies that disprove this belief:

Caplan’s most convincing argument in support of the signaling theory is that if employers valued education because of the skills it imparts, then every year of education should increase income by the same amount: a person who attended Harvard for three years would get an increase in their income equal to 3/4s of someone who attended Harvard for four years, and graduated; in reality, graduation years are worth much more — in fact, according to Caplan, ~60% of the education premium is attributed to earning a diploma. This suggests that at least 60% of the university premium is thanks to its signaling.

(If you are still unconvinced, consider this: if your son/daughter/best friend announced they would drop out of school the day before their finals, because they have already attended all lectures and tutorials and have received the education they set out to receive, would you think this is a smart move?)

Of course, the fact that education increases people’s income doesn’t mean it’s always a good deal. After taking into account things like cost, returns by major, and graduation rates — as well as additional, non-financial benefits stemming from university (job satisfcation, health, etc)— , Caplan suggests that while it’s in almost everyone’s interest to finish highschool, university education makes sense for good students only, whereas masters are almost always a bad deal.

Is education a good investment for society?

Caplan’s answer is a resounding no. After accounting for all of the benefits education can be said to provide, he finds that the returns are paltry given the cost to society. He believes the US would be better off by slashing education funding.

On the face of it, Caplan is right: if 80% of the education premium is attributed to signaling, this means that employers pay well-educated people more not because their education made them more productive, but because education helped employers identify those people who are more productive anyway. This means that education itself did not raise productivity, at least not significantly. It is therefore incomprehensible why we spend so much on it.

Some critics may dismiss Caplan as a philistine: his emphasis on whether education increases employees’ productivity is predicated on education’s sole role being to prepare people for the workforce. What about enligthenment? What about broadening people’s horizons?

Caplan’s convincing response is that education fails on these accounts too. Take enligthenment first: a vague concept, but if the claim is that it is linked to education, then a proxy for it is the number of people who appreciate the things education is all about. Yet, few people read books, much less poetry. Few people find mathematics beautiful. Few people appreciate fine art. And for those who do, I think it’s unlikely that it is formal education that inspired such appreciation.

For education to lead to enlightement, it requires three ingredients: inspiring teachers, eager students, and a good curriculum. All three are missing: most teachers are dull, most students don’t want to be in school, and standard curricula are awful.

I can attest to all three, and I’m sure everyone who reads this blog will agree if they look back to their own experience. I have attended a good school and good universities, yet most teachers were mediocre (as Caplan points out, many university lecturers consider teaching a chore, getting in the way of their research).

I was a good student, and as evidenced by the fact that I’m spending my free time reading and reviewing a book on education, I’m into learning, yet like virtually all my classmates I would cheer when classes were cancelled for whatever reason. Most kids don’t want to be in school — in fact, according to surveys Caplan cites, students dislike school slightly more than workers dislike work.

As for curricula, I think we can all agree they fail to inspire curiosity and excitement. I mean, The Catcher in the Rye may have been radical when it was published, but ‘bored rich kid gralping with angst’ isn’t exactly avante garde these days.

(Right-wing pundits roll their eyes when they hear the Left talking about ‘decolonising the curriculum’. Inflammatory and cliched language aside, left-wing activists do have a valid point here, that often gets lost among their more extreme demands: a lot of students feel disconnected from the materials they are being forced to study.

I do think literary classics are classics for a reason, and ‘the canon’ is important. But you won’t get people to genuinely appreciate it by forcing them to read the Great Gatsby; if you want to inspire a love of reading, you don’t do that by shoving incomprehensible and boring books down students’ throats. Let them start reading for fun — for the plot, for the characters, for the thrill of finding out what happens next. After they start appreciating books, start teaching them more complex books (and for the love of God, stop being snobs about what constitutes ‘high literature’).)

As for broadening horizons, teachers are usually just as narrow minded as their students, as Caplan puts it; most simply attempt to replace their students’ narrow-mindedness with their own. (When I was at school, there was a phase when students would meet at the school library to play Magic the Gathering. Teachers soon put a stop to that — why? What can be more narrow-minded than forbidding students to use school facilities to explore a game that teaches strategy, while also allowing students to make use of their imagination? Similar examples abound — teachers frowning down on comic books, lecturers dimissing arguments from different political viewpoints, etc.)

So, education does not teach people useful skills, it does not broaden people’s horizons, and it does not lead to enlightenment. Why then do we make people go through it? Why not direct more people towards vocational training? Caplan even goes to the seemingly extreme of trying to de-stigmatise child labour. ‘Seeminlgy’ because his argument is rather compelling: we instinctively recoil when we hear ‘child labour’ because we picture Dickensian scenarios of abject poverty and exploitation — but what is the reasoned objection to child labour? That it leads to exploitation through low pay? But education is completely unpaid, and less enjoyable, and less likely to develop skills employers value (and that they will therefore compensate in the future). That children are not mature enough to make such a choice? But we rely on parents to protect their children from other poor choices, so why not this one? Etc.


Compelling though Caplan’s arguments are, and although I agree with him in general, I think there are a few weak points in his thesis.

Knowing what you don’t know

As mentioned above, Caplan marshalls a lot of evidence to prove people forget most of what they have studied, and takes this to mean that education is often useless. I said above that on the face of it, this argument is rather uncontroverial — we have all at times thought the things we have been learning are useless, and have forgotten most of them.

But this isn’t quite right: there is a huge difference between knowing what you don’t know, and not knowing what you don’t know. Formal education puts you in the former camp.

Take Caplan’s book itself: I finished it a few days ago, and I have already forgotten the precise figure he gives for the education premium. But I know he has calculated it, and I know where to find it — so, having read the book isn’t a waste of time.

The same is true for a huge number of things: I don’t recall how to calculate statistical significance, but I know it can be done, so if I want to test a hypothesis, I know what to look up. I don’t know how inflation is calculated, but I remember reading that price indices are often adjusted for product quality, so I can brush up on the details.

Knowing what to look for, and where to start solving a problem are immensely valuable skills (especially when it is so easy to look things up).

Re-learning is easier than learning

I studied SQL in my second year of university, and did not use it again for 10 years. Then I joined Google, and had to rely on it to pull data; it took me a couple of days to remember it. So while Caplan is right that people forget what they have learnt, the fact is that it is easy to pick up old skills if required.

The same is true for foreign languages etc; though Caplan’s argument that most people won’t ever use the vast majority of things they learn is still valid.

Education should be improved, not scrapped

This is a common critique of Caplan’s thesis, and one that he has explicitly addressed: his response is that politicians and educators have had ample time and resources to improve education; academics have pondered what makes for a good education for ages. No-one has come up with a good solution — so, it doesn’t make sense throwing good money after bad. As he puts it, if a friend of yours tells you they have been using a $100 cream for a condition they have, and you give them proof it doesn’t work, they’d be insane to answer ‘well, I will keep using it until I find one that does work.’

Fair points, but not entirely true: there is no doubt that some countries have education systems that work better than that of the US; instead of drastically reducing the provision of education in America, why not emulate the education system of others? To use Caplan’s own analogy, if a third friend comes and says ‘yeah that cream doesn’t work, but this one does’, why would you insist on not buying the second cream?

Caplan cannot respond that this is difficult to do in practice: in practice, it is political suicide to cut education funding, so his solution isn’t any more practical than this recommendation.

(I am playing devil’s advocate a little on this one: I haven’t done serious research in alternative education systems to know whether they genuinely do develop human capital — but I think Caplan needs to address this criticism more thoroughly.)

Casting a wide net

I learnt ancient Greek in school. In practical terms, this is probably one of the most useless subjects one can study — even more so than Latin, which is, after all, the root of some of the world’s most spoken languages.

Now, Caplan is right that education is not seen as an intrinsically wortwhile good by most people — but it is by some; and I think that casting a wide net in the hopes of inspiring as many of those people as possible is a sensible idea. Most people will drudge through ancient Greek classes, and hate school all the more for it — but some will fall in love with the classics, and will dedicate their lives to studying them. Is this worth the cost of education as it stands? I don’t know — but I don’t think Caplan does either.

Similarly, if Caplan had his way, and education were subsidised to a much lower degree, then the people who would be disproportionately affected would be those whose families cannot pay for education out of their own pocket. Caplan’s response to this is that his recommendation would benefit the poorest the most, since they are the ones who are most harmed by credentialism; this is true on average, yes, but what about the geniuses who could have become professors, and were never given the chance? I don’t know that it always makes sense to base policy only on averages, which brings me to the next point…

… what about disproportionate impacts?

On average, education does not raise human productivity enough to offset its cost. But this is assuming that innovation and progress rely on average productivity. I don’t think this is the case. I think innovation progresses linearly, until a major breakthrough occurs, and then it leaps to a new phase of linear growth:

If these leaps are thanks to individuals who were given the opportunity to make them, then it’s probably worth the cost of putting a million mediocre students through university to give the opportunity or even encourage the few outstanding ones to do so too.

This is overly simplistic of course: one does not have to assume that such leaps only take place within academia, nor that cutting subsidies to education would preclude programmes to identify people with the potential and will to go to university; but discouraging academia and directing more people to vocational training may have unintended consequences. A brilliant youth may decide to become a plumber to support their family, instead of becoming an engineer who’ll develop a new energy source. If innovation depended on average productivity, it’d be worth directing one brilliant mind to a mundane job, to help a million non-academic kids find jobs they like instead of sitting in classes bored out of their minds. But if it depends on leaps, I’m less certain.

The US is not a closed system

Suppose Harvard’s value is 100% signalling. Then, while it’s partly true that Americans attending Harvard yields no net gains to America, Harvard is still beneficial to the American economy because it attracts (productive) foreign students who come for the credential, but then stay in America. In addition, the tuition fees of foreign students (which are generally not subsidised by the host country) fund research.

The reverse is true in Greece: because Greek universities are largely dysfunctional and less prestigious, many young Greeks leave to study abroad, and then stay there.

So, given that students can study elsewhere, unless the entire world stops the credentialism arms race, Caplan’s policy recommendation may backfire.

And finally…

The signal is losing its power

Caplan believes formal education’s status as a signalling mechanism is secure, because although different mechanisms (e.g. standardised test, online education) may replicate some of its signals, none can signal all three of intelligence, work ethic, and conformity.

This may be true, but I think that one of education’s signals is becoming weaker, and two are becoming less relevant. The more people who go to university, the weaker the intelligence / ability signal becomes. More credentials don’t solve that — it doesn’t take any more intelligence to do a masters degree than it takes to complete a bachelors (at least in the UK). Grade inflation is making things worse — having a First or graduating cum laude is meaningless when 30% of people do it.

Work ethic is becoming less relevant: in my experience, an excellent worker is indefinably more productive than a poor one — indefinably because whereas the former adds value to a company, the latter often detracts (by, e.g., creating non-value adding work, which takes time away from productive employees). So, I care far more about hiring a productive worker than a hard-working one. I would much rather be faced with the problem of motivating a lazy but brililant employee, than the one of trying to train a committed but incompetent one.

Similarly, conformity matters less, especially in some fields: in industries that are in constant flux, and under threat of disruption, conformity that leads to groupthink is often a liability instead of an asset. I mean, yes, you do need people who can work together and be part of a group — but choosing to drop out of uni only fails to signal conformity, it does not actively and loudly signal obnoxious rebeliousness.

(All this is not to say that things like online courses will replace formal education; if anything, in my view, people who list a plethora of online diplomas from Harvard or Stanford on their CVs send a very strong negative signal: that they care too much about signalling and less about substance. I am not alone in thinking this:


Now, Caplan has offered a bet to anyone who’ll take it: “even odds that 10 years from now, the fraction of American 18–24 year-olds enrolled in traditional four-year colleges will be no more than 10% (not 10 percentage-points!) lower than it is today”.

I am not prepared to take that bet, because there is huge inertia in the economy. For the fraction of students to drop, first companies have to realise that education is losing its signal; most companies are badly run, and will not realise this for ages. Second, not only do companies have to find a better signal / screening mechanism than education (which I think will happen, but will take time), but they must derive a significant advantage from it. This won’t happen across all sectors of the American economy in such a short timespan. And thirdly, 18–24 year olds must realise this shift is happening, and respond accordingly — which again will take time, as they will need to push back on the experience and expectations of their parents, grandparents, teachers, and family friends.

But here’s a different bet — unfortunately not as pithy as Caplan’s: I bet Caplan $100 that in 10 years, the share of employees with no university degree in the US’s top 3 venture capital companies’ non-healthcare / biotech portfolio firms will be 10% higher than today. The rationale is that these companies are the ones that are most likely to care less about the work ethic / conformity signals in education, as well as the fastest ones to adopt new signals. (The difficulty here is getting these companies to share data on their employees).

I work at Monzo. Ex Google, P&G