Four years ago, I read Bryan Caplan’s The Case against Education: Why the Education System Is a Waste of Time and Money, a polemic about the form of education in the U.S. I disagreed strongly with a decent fraction of it, agreed strongly with a larger fraction of it, and overall found it stimulating — one of the rare books that can change how you view the entirety of a system, not just its parts. (I should post sometime a list of eye-opening books…) Like Theodosius Dobzhansky’s famous quote that “Nothing in biology makes sense except in the light of evolution” [1], I’d argue that little in higher education makes sense except in the light of signaling. By signaling, Caplan means that the major outcome of the education system as currently structured isn’t that students gain skills or knowledge, but rather that they demonstrate that they’re the right type of person to fit into jobs and society. The acquisition of skills and knowledge isn’t a negligible part of education, but it’s not the primary aim. This isn’t a new idea — even I was aware of it before reading Caplan — but I don’t think it’s ever been expounded so thoroughly.
Accepting this framework of signaling, many aspects of how education works make sense, in a sad sort of way, and its human and economic tolls become clearer. As mentioned, Caplan doesn’t make an airtight case; I’ll describe below some strengths and weaknesses. Why, you may ask, should I write a book review four years late? As always, I’m mainly writing for my own benefit, but I’m also spurred by Andrew Gelman’s recent posting of his thoughts on Caplan’s book, six years after writing them; my four years seems speedy in comparison. Also, the book continues to spur a lot of discussion, often involving people who haven’t read it and who (often severely) misunderstand its claims, so noting some key points may be useful. The post isn’t a substitute for Caplan’s book itself which, despite its flaws, I strongly recommend to anyone in academia.
Before starting, here’s my customary plug for my own book! Will it dramatically change your views of the living world? I don’t know, but I was happy to see yesterday that it made it onto “Your Ultimate Book Guide to the Biggest Questions in Biology”. Here are the usual links: My description, Publisher, Amazon.)
Signaling
Caplan contrasts a signaling model of education with the commonly held “human capital” model. Human capital refers to actual skills; your psychology degree reflects your understanding of the mind and your ability to asses and alter human behavior. Signaling considers the utility of the degree to be, instead, a signal of other attributes, especially conscientiousness, intelligence, and social conformity. These traits that may have no relation to what you nominally learned, rather like a peacock’s tail serves as a signal of general health and fitness and not as a useful tool in itself.
Notably — and this is a point that’s often egregiously misunderstood — Caplan doesn’t claim that education is all signaling. He argues that it’s at least one-third signaling (in terms of time spent, for example), and probably 80%. There’s a big space between one-third and four-fifths, and this isn’t something readily quantifiable anyway, but the key point is that it’s neither 0% nor 100%. Caplan states that any sizeable signaling fraction points to a huge inefficiency, not just in wasted time and money now, but in the continued escalation of of waste as the signal becomes more common. I’ll return to this later.
A signal is valuable. This point is also often misunderstood. Caplan quite clearly lays out how much more money college graduates make compared to high school graduates: 73% more (Chapter 3). The reason for this correlation is what’s at issue. College graduate bartenders, for example, earn 60% more than bartenders without a bachelor’s degree, and it’s hard to imagine any other reason than that those hiring them view the degree as an indicator of reliability. Or, putting it the other way around, the high school graduate who isn’t hired is rejected not because of his lack of technical knowledge, but because he lacks the implicit character reference that a college degree provides.
At the level of individuals, a college degree can be lucrative. At the level of society, as we encourage almost everyone to take this track, is it beneficial? Are there cheaper, more humanistic, and more enriching ways to accomplish our goals (whatever they may be)? More on this later…
Caplan provides a variety of further evidence for the ubiquity of signaling in education, for example the uselessness of almost completing a degree program, despite having almost all the skills associated with the full program. More interesting, however, is the inadequacy of the human capital model to make sense of so many features of modern education. Why, for example, doesn’t forgetting the content of a course have the same consequences as failing the course? Why do so many students seek out the easiest classes (or majors), rather than the ones from which they’ll learn the most? Along similar lines, a quote from Steven Pinker:
A few weeks into every semester, I face a lecture hall that is half-empty, despite the fact that I am repeatedly voted a Harvard Yearbook Favorite Professor, that the lectures are not video-recorded, and that they are the only source of certain material that will be on the exam. I don’t take it personally; it’s common knowledge that Harvard students stay away from lectures in droves, burning a fifty-dollar bill from their parents’ wallets every time they do.
Which model explains this?
Objections to objections
A common reponse to the idea that education is mostly signaling is that being a physics, or biology, or computer science major isn’t mostly signaling — there’s actual physics, biology, or computer science that one learns. This is true. However, most students aren’t physics, biology, or computer science majors. The most common major here at the University of Oregon (and a common one everywhere) is business, a hodgepodge of stuff whose rigor I’ve never heard anyone defend. A side-effect of teaching a lot of courses for non-science majors is that I chat with students about the range of things they’re actually majoring in — business, sociology, public relations, … It’s this that one should consider if thinking about the median signaling component of an undergraduate degree. This is touched on, though not enough, by Caplan.
The other common objection is that the aim of education is to enrich the soul, not to provide specific, practical skills. I am very sympathetic to this objection, being fond of art, science, and learning in general for their own sake. Caplan claims that he is too, and devotes a whole chapter to this topic (Chapter 9) that includes a long quote from Malcolm X on the wonder of reading. The main counter-objection is that culture is more accessible than it has ever been, and doesn’t need the intermediary of university classes, even if they were effective in imparting refined tastes. Chapter 9 is a mix of good and poor arguments, but I’ll leave it at that. I’ll add, though, that I’d be happier with the cultural objection if universities weren’t so keen on watering down any requirements that might make one numerate or well read. Again, the student to keep in mind is not you, but the median student, a business major taking the History of Rock and Roll as his humanities requirement.
The book’s flaws
Despite my overall fondness for it and its arguments, The Case Against Education has significant problems.
The biggest flaw is that its claims span K-12 education and higher education. The case against the current higher education system — which really is bloated, expensive, wasteful, and cruel to “below average” students — is fine, and most of the book’s evidence is arrayed against this target. Applying these arguments to high school, however, is unconvincing. Not only are basic skills in writing, math, etc., that open doors to many paths developed there, but the relative uniformity of the curriculum provides a shared background for everyone. Caplan complains about the constrictions of school with “I love education too much to accept our Orwellian substitute” [p. 260], missing the irony that most of us know what Orwellian means thanks to mass high school education. We can certainly argue that high school isn’t doing its job of developing skills well, but that’s a separate issue.
Another flaw is that the presumed aims of education are phrased almost completely in economic terms, as preparing people for employment. As noted above, Caplan argues that culture and general enrichment are important, but argues against the education system being necessary or important for promoting that. It does not follow from this, though, that preparing people for jobs should be the goal of education; Caplan, with no elaboration, seems to see it as self-evident that job placement should be the ultimate aim.
Third, even if one accepts the signaling model, the question arises as to whether education improves the traits that are being signaled. Caplan writes, “The labor market… pays you for the preexisting traits you reveal…” [p. 13] Are these really pre-existing, or are they developed by study? Caplan does address this (Chapter 3), but rather minimally, noting that there is little research on whether education itself improves things like conscientiousness. (It would be hard to untangle from simple age: is a 22-year-old who finishes college more conscientious than a 18-year-old high school graduate simply because she’s 22?) This isn’t a fatal flaw: even if education helps develop non-cognitive skills like organization, it begs the question of whether the current system is a sensible way to implement this.
Finally, I’ll note that the book is very repetitive. While well written, albeit sometimes infuriating, it could easily be trimmed by at least 1/3.
What to do?
Caplan’s overall recommendation is that we should spend less on education. Not only is it wasteful in itself, but if much of education is signaling, the “subsidies [we provide] raise the correlation between educational attainment and employability” [p 214]. The current system is “a dysfunctional game, but if you refuse to play, the labor market brands you a loser.” This escalates, as the signal grows more common and its value as a signal declines. Already, we’re seeing the proliferation of Master’s degrees, because a Bachelor’s in itself isn’t distinctive. The cost of education, spiraling ever upwards, falls especially hard on the poor. Further subsidizing education — or worse, forgiving loans, our recently enacted policy that will serve as a giant green light to every university to increase its tuition — further dilutes the signal, and the benefit to the student. This is especially harsh for the group that it’s easy for academics to ignore: the average or below-average students.
Aside from lowering our spending, Caplan doesn’t have many concrete suggestions. He envisions a world in which there is more free play, library time, etc., for young people. He also proposes education as a tour of activities and occupations to connect students with the “real world,” but this is rather vague. The lack of solutions, or a proposed path from where we are to where we’d lke to be, is a flaw of the book.
Caplan wants to be provocative, and he needlessly overstates his case. This is most evident in the title, which should be something like “The case against higher education, as it currently is” rather than “The case against education.”
Still, the book is full of ideas worth thinking about, and it helps frame many contemporary issues in education.
Today’s illustration: A rufous hummingbird.
[1] Theodosius Dobzhansky, The American Biology Teacher, Vol. 35, No. 3 (Mar., 1973). Link
— Raghuveer Parthasarathy; October 13, 2022
The Eighteenth Elephant 
You probably don’t want to hear from me…but…
> The most common major here at the University of Oregon (and a common one everywhere) is business, a hodgepodge of stuff whose rigor I’ve never heard anyone defend.
Having worked with a lot of MBA students, participants in executive management programs, etc.
I’ve never heard many of them defend the content per se. Mostly they come from having been in industry positions, and say that what they’re learning in business school is what they were learning years earlier in the field.
But as a general rule, they say that the networking they do in business schools is invaluable. Also, the ability to learn from other students about other companies in their industry, or the opportunity to glean lessons from students in other industries.
I’ve worked with many execs from Wharton’s advance management program who have said, much to my surprise, that they greatly valued their experience despite the cost of tens of thousands for just a few weeks. I would say their evaluation was made mostly (but not entirely) based on a financial scale
I’d say that there’s a similar “networking” gain in other areas of study as well. Or from seeing teachers model all kinds of learning tasks. I don’t know if Caplan addresses those issues – I’d imagine he must to some extent. But I would suspect that his (self-professed) politically oriented slant would lead him to underestimate the value of those aspects from schooling at pretty much any level.
Don’t worry — I’m happy to read your comment! This is a good point, but I wonder about the networking benefits for a professional MBA versus an *undergraduate* business degree — the latter is what I was referring to. I doubt that the networking effects of hanging out with 2000 other undergraduate business majors are very strong. Still, I agree that it’s not clear where networking fits into a human capital vs. signaling picture.
Another thing that’s missing here is that there’s a problem with the claim that a lot of the information taught is available online*: in particular, I seem to remember that online courses turned out to be far far worse than even badly-attended in-person courses. That is, that there are very few people who actually succeed in completing an online course**. (That may have improved over Covid, but the general agreement seems to be that even “live” online courses with a schedule and feedback left a lot more people behind than regular pre-Covid courses.)
* And this is ignoring the point that being able to use online resources effectively is probably far rarer in non-college grads than in college grads.
** For example, I have good intentions of doing the MIT linear algegra course, but reviewing my 50-year old rusted-out pre-calc and calc is going very slowly; there’s a good chance I may not even get there. Sigh.
Yes and no — I agree that most people don’t complete online courses on their own, and wherever you are with the linear algebra course, you’re likely ahead of the vast majority of people! However, I also know of people who do learn things well online, especially “bite-sized” pieces like Khan Academy nuggets. Personally, I really enjoyed doing the course described here (https://eighteenthelephant.com/2015/10/23/learning-about-machine-learning-part-i/).
However, you’re right that in general it’s harder to learn independently / online as compared to an in-person course. A big challenge is that there’s no one to answer one’s questions, so one is stuck unless one can figure out everything on one’s own. I’ve often thought that a great way to structure a university would be for the “content” to be from online lectures, like MIT’s, with the role of faculty as tutors / answerers-of-questions / coaches. That probably deserves a post of its own!
Personally I’m more convinced that people fail to complete the online course because it doesn’t serve as a good signal, not because the course quality is much lower than an in-person course. I suppose it depends on the subject, but these days many professors (even at top schools) post their class syllabus online, so you can know the exact roadmap to follow pretty easily. Many professors even post their lecture slides online, if they don’t just outright record their lecture videos. More and more textbooks are freely available via the professors who write them, and if they’re not they’re very easy to find for those motivated. If you’re working through an undergraduate math textbook, you’ll likely find the answers to many questions (particularly the harder ones) on StackExchange, and frequently these are high quality answers. Courses on MOOC platforms like EdX and Coursera also contain message boards where you can ask questions and get answers from other learners or from some of the course TAs, however I can’t vouch for the quality of these personally. The number of podcasts/blogs/substacks/youtube videos covering concepts one learns in university is higher now than it ever has been.
Don’t you think it’s just more likely that people would rather do other things with their time than learn some of this content? I’ve had the Feynman Lectures bookmarked for about 7 months now. Every now and then I’ll go and read through a lecture, or more frequently, remind myself of what I had been reading last time I checked in. Part of me wants to have a better understanding of quantum mechanics, but if I’m being honest most of the time I’d rather spend my free time doing other things (like reading blog posts). Now if you attached a financial reward to the completion of the course, I think I’d be more motivated to get through it.
“Don’t you think it’s just more likely that people would rather do other things with their time than learn some of this content?” Absolutely! Caplan states this as well. I think I’m not understanding your point, though. The fact that most people *don’t* seek out / complete courses on their own is, again, an indication that the skills themselves are not as sought after as one might like to believe. My point was that among the subset of people who *are* trying to gain these skills on their own, a bit of interaction in addition to the online content may help remove roadblocks.
@Raghuveer,
Sorry my reply was meant to be for David, I think you and I are in agreement. Totally agreed on the interaction aspect, and I’m a big fan of things like flipped classrooms or a more tutor-oriented instructional style (something like St. John’s Liberal Arts program). I think MOOCs are trying to innovate there as well, but of course it’s a bigger challenge for them. Just in case anyone’s interested, I think UCSD’s Data Science series on EdX is a good example of an online course https://www.edx.org/micromasters/uc-san-diegox-data-science. I believe they mention making videos shorter and adding in more opportunities for interaction to boost engagement
Raghuveer –
Sure, the networking benefits would he less for undergrads, but I do still think even there, the more general benefits from the human interaction part aren’t at all trivial (especially in the internships that many undergrads have as a part of their business school curriculum). Students learn, socially and emotionally, from interacting with peers and with teachers as mentors. Indeed, not all of those interactions are positive – say considering the high drop out rates, the high rates of suicides among freshman, etc. And alternative activities at replacement environments, such as they might have in a work environment, would also be valuable. But I wouldn’t just make a blanket assumption that replacement experiences would be equally beneficial or even better; for example, there are plenty of generalizable negative interactions in work environments as well.
I’m a big proponent of career-based experiences being built into education programs for students in their late teens or early twenties – as I think from a developmental psychology perspective those students are very much in need of opportunities to explore adult roles and develop adult identity – and schools in many ways create an artificial environment that delays that developmental opportunity. But it’s the replacement aspect of Caplan’s argument (what I’ve gleaned without actually reading the book) that I’m highly skeptical of. Caplan hits on many criticisms of out traditional educational paradigm that I’ve been talking about (as an educator) for decades, but I have a reflexive reaction to what I see as a weaponizing of those criticisms in a libertarian campaign to create a “free market” model as a replacement.
You might want to look over this piece:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1928831
It’s written by a rather well-known climate change “skeptic” who also runs a pretty well-known program of election modeling. Anyway, it’s been a while since I read it, but as I recall, it’s from a rather similar tale as Caplan’s. And my reaction was similar there also. As a Wharton professor, he offers a critique of our educational paradigm that touches on what I think are very mportant aspects to consider in the process of reforming education. But his conclusions and overall political slant, in my view, is actually pretty dangerous in its simplistic foundation: that because these problems are real and worthy of reform, we should just basically chuck the whole thing out. I think doing so, and removing any federal or governmental role in ensuring educational opportunities, would basically be a disaster – especially for people who have fewer resources to rely on. And I say that as someone who basically sees our whole process of schooling as effectively a mechanism that in the end, functions to reinforce existing class status quo, even as it’s also often seen as the critical driver of class mobility.
Interesting post! I disagree with this point though: “College graduate bartenders, for example, earn 60% more than bartenders without a bachelor’s degree, and it’s hard to imagine any other reason than that those hiring them view the degree as an indicator of reliability.”
I think some part of that 60% must be due to a selection effect on the part of the workers. College graduates will on average have more attractive non-bartending job options, and so will have a greater propensity to pass up jobs at lower-paying dive bars. In other words, you’d expect a substantial wage gap even when college graduates are no more attractive to bar owners than non-college graduates.
But even if bar owners do have an average preference for degree holders, I don’t see how this data shows that the degree is causal. I would suspect that many of the desirable qualities that the degree is a signal of are themselves also partially observable.
If I had to guess, I would bet that selection, non-degree signs of desirable traits, degree signalling, and improvements in non-academic traits caused by college all play a role in the bartender wage premium, probably in that order of magnitude.
I don’t know enough about bartending economics, or the hiring process for bartenders, to reply. The selection idea seems odd, though: why wouldn’t the dive bar bartender apply for the higher-paying position? About non-degree, but observable, signs of desirable traits: a lot hinges on “observable,” since for many jobs, the initial screening is of a resume, not a flesh-and-blood candidate. In any case, it would be great to see work that tries to quantify all the causal routes you established — I think you’re right that there are more than two!
Imagine that all individuals have two attributes: their bartending quality X, and their uncorrelated non-bartending quality Y. X and Y are measured by the salaries that can be earned as a bartender and as a non-bartender, respectively. College grads and non-grads have the same distribution for X, but grads have a higher Y distribution. Each individual picks the career that gives them the higher salary. Then even though grads and non-grads have exactly the same X distribution, the salary distribution among those who choose to become bartenders will be higher for grads. Does that seem reasonable?
Daniel: Thanks — I think I see your point now! If people with some (irrelevant) trait (Y) avoid low-paying jobs, it will look like a wage premium for that trait. Interesting — I hadn’t thought about this! It leads to the question of why they’re able to avoid the low-paying jobs, which could be due to skills relevant to the other high paying jobs, signaling relevant to the other high-paying jobs, etc. I wonder how one would analyze this without considering the whole “system” of jobs. I’m sure economists have thought about this…