Book Review: Bryan Caplan’s “The Case Against Education” (2018)


By the time you finish an undergraduate degree, you’ll likely have spent over 15 years in schooling – and according to The Case Against Education (2018) by Bryan Caplan, these years were largely wasted. In his book, Caplan argues that education fails in its titular task: educating. Instead of developing abilities, knowledge, and skills, schools and universities let students signal their underlying productivity to potential employers. Given this signalling benefits the individual rather than general society, Caplan recommends drastic educational austerity rather than government spending on frivolous, and ultimately worthless, education credentials. While the core of this thesis, that signalling plays a critical role in education and labour market outcomes, is well-argued and substantiated by evidence, Caplan’s extensions to this central idea are largely unconvincing.


Signalling is hardly a new idea, with Capan acknowledging its popularisation by economic titans like Kenneth Arrow. Caplan’s unique contribution is in developing a compelling theory of why education credentials are highly valued by employers even if education doesn’t make you a better worker. Employers are risk-averse in their hiring practices because firing workers is costly. There are legally mandated costs like severance pay, but equally firing workers can negatively impact the morale of staff who weren’t fired. If firms unwittingly hire a bad worker, they can either grit their teeth and accept a poor return on investment, or face the high costs of firing. To avoid this dilemma, firms pay a premium for workers whose competency they can trust.


Unfortunately, it’s difficult to determine a priori who among a pool of potential employees will be a good worker. The ideal employee is hard-working, clever, and socialises well with coworkers, but there is no cost-effective way to determine if a candidate meets this description. Consider intelligence; you could make candidates complete an IQ test, but that wouldn’t tell you if they were lazy or antisocial. Additionally, these results might reflect simple variance or a willingness to study for IQ tests. Firms can’t correctly gauge a potential employee’s intelligence. These problems extend to testing how hard-working or amiable a worker is, meaning employers have no effective way to determine who is best for the job, but face high costs for choosing poorly.


This dilemma is why firms value the signal education sends, because doing well in school or university reflects a package of three desirable traits: intelligence, conscientiousness, and conformity – Caplan’s holy trinity. For the majority of students, earning a high mark requires a combination of the three. You need intelligence to do well on exams, conscientiousness to work hard and study, and a willingness to conform to social expectations to invest in education. Firms want someone clever, hardworking and willing to adhere to workplace convention; the more years of education you have, the more likely you fit this description. Crucially, because of the drawn-out nature of education, it’s difficult to fake being clever, hard-working, and conformist. Someone smart but lazy could plausibly work hard for a year then stop, but to get a good degree you need to be working hard from sixth form to your final year. (Admittedly, this is more relevant to the American education system which uses cumulative grade point averages.) Firms don’t care if you understand the IS-LM model, they want to know if you were able and willing to study enough for a good mark.


When firms believe education sends a good signal, they’ll want to pay a premium to people who did well in their course. Clever young people, desiring higher future earnings, pursue high marks. This improves the predictive power of academic success on labour market success and the cycle repeats. This is the foundation of Caplan’s argument, that employers pay higher wages to people who’ve done well in school or university because of the signal that success sends, not because the course content is relevant to the industry.


Outside of this foundation, however, The Case Against Education is often unconvincing and runs the risk of misrepresenting research. Some key logical principles lack sufficient justification and opposing theories aren’t given a fair trial. While Caplan substantiates his arguments with empirical evidence, the reader is advised to be cautious about interpreting evidence in the book. Though in many places, there is a good-will effort to honestly explore prior research, some key research is misrepresented and evidence which might weaken the thesis is omitted.


Flaws in The Case Against Education begin to appear once it moves from arguing that signalling is important to prescribing policy based on this conclusion; prominently Caplan never establishes that signalling is wasteful. Instead, he simply states that signalling does not meaningfully help society, an assertion which clashes with his own argument on the importance of signalling. Firms value education credentials because they are a uniquely effective signal of a worker’s competency. By relying on these signals, firms are better able to hire the right person for the job. This means firms are more productive because they are less likely to misallocate labour and also face lower costs, since the probability of making bad hiring decisions and the associated firing costs are lowered. Caplan argues that this is particularly important as firms are often slow to learn the ‘true productivity’ of their workers. He estimates it could take months or years to realise a worker isn’t right for the job. If signalling makes firms more productive and helps them avoid lengthy and costly adjustment costs, it’s unclear why it has no social value.


Signalling actually helps the labour market as a whole. The labour market is characterized by information asymmetry – workers know more about their productivity than firms, and have incentives to appear more productive than they actually are. Barring educational certification, there is no effective way for potential employees to credibly disclose their ability. These are the conditions for Akerlof’s “market for lemons” (1978), where information asymmetry causes adverse selection and market degradation. Avoiding this outcome would certainly be socially desirable, so signalling is an entirely necessary element of a functioning labour market.


However even if one accepts that signalling is largely wasteful, this still leaves the question of to what degree schools and universities educate students versus merely providing them with an opportunity to signal. Caplan argues that only a fifth of education is about learning and the rest is signalling. Schools teach irrelevant material, students only learn the material for the test and forget it shortly after, and adults fail to retain knowledge later in life. He supports this with empirical evidence that knowledge fades out quickly and that adults are generally unable to remember even basic topics from their high-school education. The vast majority of American adults studied geometry at some point in their life but only a sliver remember the content. Furthermore, education seemingly fails to provide students with soft skills like critical thinking or learning to learn. For example, experiments have found that students are unable to apply logical skills to different contexts regardless of their education. If schooling truly is so impotent, Caplan’s case that signalling is the driving force of education becomes much more plausible.


The evidence on this topic is mixed, which makes it somewhat concerning that Caplan’s empirical evidence is not always a faithful representation of current research. To support his argument that schools can teach little, he predominantly uses data and experiments targeted at university and high-school students. He justifies this as follows:

“Since almost everyone in the United States at least starts high school, there is no clean way to calculate the selfish return for earlier grades. Nationally representative data on K-8 curricula are also wanting, so there’s no clear way to ballpark its signalling share – or estimate its social return.”

This is false. Researchers have extensively studied the Perry Preschool Program, taking a sample of 123 disadvantaged black children (age 3), where half were given a high-quality preschool education and half were left as a control. Those who received the treatment were more likely to graduate high school, enjoyed higher average incomes, committed less crime, and were more likely to be employed than the control group. While the program has been criticised for having a low sample size and problems with its experimental randomisation, the effects hold even when adjusting for these issues (See Heckman et al., 2010 for further discussion.) Yet despite the program’s flaws, Caplan still cites it as evidence but only discusses a particular finding of the program which supports his claim and omits the rest.


Another good piece of evidence on early education is Project STAR, a three-year experiment which randomly assigned teachers to classrooms and followed the students of those classrooms through their life using administrative government data. The most complete analysis of the data comes from Chetty et al. (2011) and their famous findings that good kindergarten classrooms have substantial impacts on labour market outcomes later in life. Caplan acknowledges this study and argues that their findings are consistent with his signalling theory. Good kindergarten teachers simply instil students with an appreciation for the value of sending good signals. Putting aside that this is inconsistent with his assertion that students retain nearly nothing, he omits findings within the study that don’t conform to his theory. Caplan argues that school doesn’t provide students with non-cognitive abilities, but Chetty et al. find that kindergarten classroom quality has large and persistent effects on non-cognitive abilities (e.g. initiative or not disrupting class.) It’s somewhat concerning that Caplan addresses the findings of this study which are arguably consistent with his thesis, but omits findings which are in more direct tension.


The omission of conflicting evidence is further highlighted in Caplan’s discussion of the scientific consensus surrounding the controversial nature or nurture debate.

“This approach, called 'behavioral genetics,' consistently finds strong, pervasive effects of nature, and weak, sporadic effects of nurture. […] The genes your parents give you at conception have a much larger effect on your success than all the advantages your parents give you after conception.”

Behavioural genetics has not only struggled to reach meaningful conclusions because of endogeneity in the adoption process, but what conclusions have been found in the research are not so decisive as Caplan claims. In one of the most influential behavioural genetics papers, noted for its quasi-experimental methodology, Sacerdote (2007) finds that both genetics and family environment have large impacts on future earnings, though genetics are somewhat more important. Conversely, in an adoptee study where data on both adopted and genetic parents are available, Bjorklund et al. (2006) find family environments better predict future income, though genetics still matter. A more accurate depiction of the behavioural genetics literature is that both nature and nurture matter, though which effect dominates is unclear.


What further muddles the debate around family circumstances and genetics is that the distinction between nature and nurture is becoming increasingly unclear. The degree to which genes express themselves is not an immutable fact. Genes change in response to environmental factors. Developmental neuroscience has found an abundance of evidence suggesting socioeconomic status influences the development of the brain. (Hackman and Farah, 2009.) Modern scientific understanding suggests that genetics are intertwined with family circumstances, that nurture and nature are complements rather than opposing forces.


This disconnect between The Case Against Education and the academic literature is reinforced by the unfortunately shallow depiction of competing theories of education. The theory which is most opposed to signalling theory, and the theory Caplan spends the most time rebutting, is human capital theory. In short, human capital theory posits that the tendency for wages to increase with age is explained by individuals acquiring more experience, skills, and abilities. As people age, they become wiser and more productive, earning commensurately higher wages. In this theory, education makes people more productive, contrary to Caplan’s argument that it does nothing much. While human capital and signalling theory are diametrically opposed, most economists accept that education has a signalling as well as human capital component. The question isn’t which theory is true, but rather how much of education is signalling versus human capital.


Unfortunately, The Case Against Education does not address this question. Instead, Caplan offers extensive rebuttal to “pure” human capital theorists who believe education is exclusively about the building of skills and knowledge, leaving no room for any signalling. While his examples do demonstrate that such a theory would be untenable, e.g. if the value of university education is knowledge not signals why don’t people sneak into prestigious university lectures, it’s unclear any economists are human capital purists. Read for example the introduction to the inaugural issue of the Journal of Human Capital; even authors zealous enough to start a journal dedicated to human capital theory never claim it exists in a pure form (Ehrlich and Murphy 2007.) The Case Against Education never acknowledges the nuances of human capital theory and thus struggles to meaningfully rebut it.


The Case Against Education falls short of its potential. It builds a strong case for the role of signalling in education and the labour market, but ultimately overstates the case for a pure signalling model of education. Humans are complex – old paradigms like nature against nurture are breaking down as we gain a more nuanced picture of human development. Both signalling and human capital play a role in education, but Caplan omits rather than acknowledges this fact. Society would perhaps be better off with more critical and nuanced perspectives about the value of education, and hopefully this book can someday contribute to one.



References:

Akerlof, G.A., 1978. The market for “lemons”: Quality uncertainty and the market mechanism. Uncertainty in Economics, pp. 235-251.


Björklund, A., Lindahl, M. and Plug, E., 2006. The origins of intergenerational associations: Lessons from Swedish adoption data. The Quarterly Journal of Economics, 121(3), pp.999-1028.


Caplan, B., 2018. The case against education: Why the education system is a waste of time and money. Princeton University Press.


Chetty, R., Friedman, J.N., Hilger, N., Saez, E., Schanzenbach, D.W. and Yagan, D., 2011. How does your kindergarten classroom affect your earnings? Evidence from Project STAR. The Quarterly Journal of Economics, 126(4), pp.1593-1660.


Ehrlich, I. and Murphy, K.M., 2007. Why does human capital need a journal?. Journal of Human Capital, 1(1), pp.1-7.


Hackman, D.A. and Farah, M.J., 2009. Socioeconomic status and the developing brain. Trends in cognitive sciences, 13(2), pp.65-73.


Heckman, J.J., Moon, S.H., Pinto, R., Savelyev, P.A. and Yavitz, A., 2010. The rate of return to the HighScope Perry Preschool Program. Journal of Public Economics, 94(1-2), pp.114-128.


Sacerdote, B., 2007. How large are the effects from changes in family environment? A study of Korean American adoptees. The Quarterly Journal of Economics, 122(1), pp.119-157.


Turkheimer, E., Haley, A., Waldron, M., d'Onofrio, B. and Gottesman, I.I., 2003. Socioeconomic status modifies heritability of IQ in young children. Psychological Science, 14(6), pp.623-628.


Image Credits: The Blue Diamond Gallery