Bryan Caplan on the Case Against Education

Episode Summary

Executive Summary: Russ Roberts and Brian Kaplan debate the real returns to education, contrasting human capital, ability bias, and signaling. Kaplan argues much of college’s wage premium reflects credentials, sorting, and conformity rather than useful skill acquisition, though Roberts stresses education can still deepen thinking, judgment, and tacit learning in ways hard to test. They also discuss the sheepskin effect, the weak case for transfer learning, and controversial implications for child labor and school policy.

Main Topics: Three explanations for the college wage premium (Priority: 5/5): Kaplan frames the core dispute as a three-way debate: human capital (school builds skills), ability bias (more capable people select into college), and signaling (college certifies preexisting traits to employers). Human capital vs. signaling (Priority: 5/5): Roberts presses the case that education can teach reading, writing, math, reasoning, and tacit habits of mind; Kaplan counters that much coursework is irrelevant to jobs and that school mainly functions as a credentialing device. Ability bias and statistical adjustment (Priority: 4/5): Kaplan argues that once researchers control for IQ and conscientiousness, a large share of the observed education-earnings gap shrinks, implying part of the premium reflects who attends college rather than what college teaches. Transfer learning and whether school teaches people to think (Priority: 4/5): The conversation disputes whether schooling produces generalizable thinking skills. Roberts argues that learning can be subtle and cumulative; Kaplan cites educational psychology’s weak evidence for transfer and its failure to find robust effects despite decades of searching. The sheepskin effect and credential completion (Priority: 5/5): Kaplan uses the strong payoff to finishing a degree, rather than merely accumulating years of study, as evidence that diplomas matter as signals of conformity and employability more than classroom content. Classroom incentives, grading, and student motivation (Priority: 3/5): Roberts and Kaplan discuss why students cheer when classes are canceled, how grading curves and grade inflation shape behavior, and how academic incentives can obscure true learning outcomes. Child labor and the social value of work experience (Priority: 3/5): Kaplan defends moderate child labor, arguing that work can teach valuable skills and that society inconsistently tolerates boredom and coercion in school while condemning similar experiences in paid work.

Key Arguments: College graduates earn substantially more than high school graduates, but that premium does not by itself prove school raises productivity; it may reflect selection and signaling. Human capital is real: basic academic skills and some increases in IQ from schooling plausibly raise job performance. Ability bias explains part of the earnings gap because people who attend college tend to be more talented, conscientious, or otherwise advantaged before schooling. Signaling is powerful because employers use degrees to infer hidden traits such as intelligence, work ethic, and conformity under conditions of imperfect information. The sheepskin effect is difficult for pure human capital models to explain because the labor-market payoff jumps sharply at degree completion rather than accumulating smoothly over years. Educational psychology has found very weak evidence that students transfer classroom knowledge to real-world tasks; most claimed 'learning how to learn' effects are overstated. Many students do not retain basic facts from school, suggesting that if school were teaching deep generalizable skills, evidence of those skills should be easier to observe. Roberts argues that some of education’s most important effects are subtle, cumulative, and hard to measure directly, such as improved judgment, narrative understanding, and probabilistic thinking. Kaplan argues that if a claimed educational effect cannot be demonstrated on any well-designed test, then it should not be credited as a meaningful gain. The persistence of ineffective or incompetent workers in jobs supports signaling stories, because firms often retain or 'de-hire' workers rather than quickly firing them. Child labor can be beneficial when it is safe and limited, because it may provide practical skills and experience more effectively than some forms of schooling.

Data Points: College earnings premium: 73% more - Kaplan cites the common average wage gap between college and high school graduates. Ability bias share of apparent education effect: 45% - Kaplan’s estimate of how much of the observed earnings premium should be attributed to preexisting ability rather than education. Human capital share of causal effect after ability bias: 20% - Kaplan’s approximate remaining share of the causal return to education he attributes to actual skill formation. Signaling share of causal effect after ability bias: 80% - Kaplan’s approximate weighting of signaling among the non-ability portion of the education return. IQ gain from education: 1 to 3 points per year - Kaplan cites a literature review suggesting schooling slightly raises measured IQ. New meta-analysis estimate of IQ gain: 1 to 5 points per year - Kaplan notes a newer analysis that gives a somewhat larger estimate. Shrinkage of education payoff after IQ controls: about 30% - Kaplan says controlling for IQ typically lowers the estimated return to education by roughly this amount. Additional shrinkage from work ethic/conscientiousness controls: about 15% - Kaplan’s rough estimate of the contribution of non-IQ ability controls. High school time spent on reading, writing, math: under 50% for most students - Kaplan argues much of K-12 curricula is not directly job-relevant. Proportion of students who cannot name senators per state: about half of Americans - Kaplan uses this as evidence that many people fail to retain even basic civic facts taught in school. Students who work at George Mason: 80% - Kaplan says about four-fifths of his students have jobs, which he uses to discuss job performance and signaling. Degrees of program history: 7th appearance - Roberts introduces Kaplan as making his seventh EconTalk appearance.

Pivotal Quotes: "And the last story is called ability bias." — Brian Kaplan: Kaplan introduces the three explanations for why educated workers earn more. "If there's no designable test that can show that people have learned something, then they haven't learned it." — Brian Kaplan: Kaplan argues that unobservable educational gains should not be credited without empirical evidence. "Imagine the stark dilemma. You can have either a Princeton education without a diploma or a Princeton diploma without an education, which gets you further on the job market?" — Brian Kaplan: Kaplan uses this hypothetical to illustrate why the diploma itself is valuable as a signal.

Implications: The discussion suggests colleges may be overvalued as skill producers and underestimated as credentialing systems. For students, employers, and policymakers, that means scrutinizing what schooling actually teaches, how degrees are used, and whether cheaper, more targeted training could replace some of higher education.