Lant Pritchett on Education in Poor Countries
What this covers
Lant Pritchett discusses with Russ Roberts why massive expansions in school enrollment across poor countries have failed to produce learning gains. The conversation centers on a structural problem: governments have built education systems as spider systems—top-down hierarchies designed to manage logistics uniformly and control socialization—but these architectures systematically eliminate the bottom-up feedback and accountability that would drive actual teaching quality. Pritchett argues that this design is deliberate, grounded in the fact that socialization is not third-party contractible. Because states want to shape children's values and identities, they cannot simply finance schooling and let parents choose; they must control production itself. The result is isomorphic mimicry—systems that produce all the visible markers of education (enrollment numbers, buildings, textbooks) while systematically avoiding the metrics and incentives that would expose whether children are actually learning.
The discussion ranges across concrete evidence and structural analysis. Pritchett presents data showing that children in India completing fifth grade often cannot read a single word, that performance gaps between poor countries and wealthy ones dwarf what input-based reforms can close (roughly 10 points on a 500-point scale versus gaps of 100–200 points), and that even decades of spending increases in wealthy countries have yielded minimal measured learning gains. He argues that the Hayekian Knowledge Problem becomes acutely binding in education because teaching involves dense human interaction requiring subtle, context-specific judgment—work that cannot be centrally managed. The conversation also examines why PISA scores matter less to poor countries than they appear to, discusses how East Asian high-stakes exam systems achieve results through brutal student effort rather than system design, and considers whether decentralized or market-like mechanisms could eventually reshape incentives. A closing segment addresses whether these lessons apply to affluent countries, where the problem manifests differently but partly reflects similar accountability deficits.
Pritchett argues that the global expansion of schooling in poor countries has produced more years in classrooms but almost no gain in actual learning, because top-down 'spider' systems built for ideological socialization and logistics deliberately lack the bottom-up accountability that would let parents and local feedback drive real education.
- Spider (top-down) systems do logistics and socialization control well but destroy the bottom-up information needed for quality teaching
- Input-based reforms (class size, textbooks, money) can close at most ~10 of a 100-200 point learning gap
- Governments control production of schooling to control socialization, which is not third-party contractible
Public schools perform well only where parents have credible exit options; lack of competition is the fundamental problem.
- Affluent US public schools perform well in part or wholly because those parents could afford private schools if they chose, and many do; that credible exit option gives them effective choice and competitive leverage, so the school must respond — whereas in many American schools without such effective competition or parent control, children get neither the enrichment nor the basic learning, and the fundamental problem of the US public school system is this lack of effective competition.
“part of the reason that's true... is that those parents in those affluent suburbs can afford private schools if they choose to... they have a choice. And so the school knows that. And the school has to respond”
Schooling expansion in poor countries has not translated into increased student learning despite dramatic growth.
- Schooling in poor countries has expanded so rapidly that the average Haitian or Bangladeshi had more years of schooling in 2010 than the average Frenchman or Italian had in 1960, and the developing-world labor force has tripled its years of schooling since 1950 — yet this dramatic expansion has not increased what students actually know.
“the average Haitian and Bangladeshi had more years of schooling in 2010 than the average Frenchman or Italian had in 1960”
Class-size studies mistake context-dependent behavioral effects for universal physical laws, revealing no fixed impact exists.
- Because education involves human beings with hopes, fears, identities, and choices — not fixed physical quantities like proton mass — the impact of adding a teacher or lowering class size is determined by people's behavior and varies by context; therefore studies measuring such interventions do not reveal 'the' impact of class size but rather reveal that there is no such thing as the impact of class size.
“Therefore the impact on learning from adding a teacher to a classroom is not a fixed quantity like proton mass, but rather is determined by the behavior of people”
Education is not a public good, so the economic case for government provision fails; economists should abandon that argument.
- Education is not a public good, and even if it were, the abstract blackboard case for government schooling fails in practice — so if government does a horrible job providing it, economists should stop making the public-good argument for government provision.
“Which it's not.”