The strongest evidence that industrial policy can work is the East Asian miracle — Korea, Taiwan, Japan, Singapore — which industrialised at unprecedented speed under active state direction. But the lesson is widely misunderstood. This module covers what the developmental state actually did, the discipline that made it work where import substitution failed, and the hard question of whether African states can do the same.
The developmental-state thesis
The 'developmental state' literature (Chalmers Johnson on Japan's MITI, Alice Amsden on Korea, Robert Wade on Taiwan) established that the East Asian success was NOT achieved through free markets and laissez-faire (the free-market reading), but through active state GOVERNANCE of the market: the state picked strategic sectors, directed credit to them through state-controlled banks, protected and subsidised them, coordinated investment, and deliberately shaped the economy's structure toward higher-value industries. These states intervened pervasively and selectively — the opposite of the Washington Consensus prescription — and they industrialised spectacularly. The developmental-state thesis is the central empirical challenge to the anti-industrial-policy orthodoxy: here were states that did exactly what the orthodoxy said couldn't work, and it worked.
The key: disciplining rents
Why East Asia succeeded where ISI failed
The single most important lesson — and the most misunderstood — is what distinguished successful East Asian industrial policy from failed import substitution. Both PROTECTED and SUBSIDISED industries (so the difference is not intervention vs free markets). The difference was DISCIPLINE: East Asia tied its support to PERFORMANCE, especially EXPORT performance. Amsden's phrase for Korea: the state imposed 'reciprocal control mechanisms' — firms received subsidies, protection, and cheap credit, but ONLY IF they met performance targets, above all succeeding in competitive EXPORT markets. Support was conditional, monitored, and withdrawn from failures. So the protected firms faced the discipline of world-market competition (they had to actually become competitive to keep their support), achieved global scale (serving world markets, not just a small domestic one), and were forced to improve — exactly the discipline that ISI, which subsidised firms to serve protected domestic markets with no performance test, completely lacked. The lesson: subsidies and protection are not enough (and are dangerous) without discipline; the discipline — conditioning support on performance and being willing to cut off failures — is what turns industrial policy from a giveaway to inefficient rent-seekers into an engine of competitiveness. This is THE lesson of East Asia, and the one African (and other) industrial policies most need to learn: it's not whether you intervene, but whether you discipline.
Embedded autonomy
The bureaucratic condition (Evans)
What allowed East Asian states to discipline rather than be captured? Peter Evans's concept of 'embedded autonomy' (the Governance course) identifies the institutional condition: the developmental bureaucracy was EMBEDDED enough in the private sector to have the information and relationships to coordinate effectively (it understood the industries it was promoting and could work with them), but AUTONOMOUS enough not to be captured by them (it could impose discipline, withdraw support from failures, and resist the demands of particular firms in the national interest). This combination — connection without capture — is what let the state pick sectors and discipline rents without becoming a tool of the rent-seekers. It required a competent, meritocratic, professional bureaucracy (the Weberian state of the Governance course — Evans-Rauch) with a corporate identity and long-term incentives, operating under a political settlement that gave it the autonomy to act. Embedded autonomy is the institutional precondition for successful industrial policy — and its absence (a captured or incompetent bureaucracy) is why industrial policy fails. The developmental state was as much about state capacity (the Governance course) as about economic instruments.
Can it be replicated in Africa?
The hard question: can African states replicate the developmental-state model? The pessimistic view: the conditions that enabled East Asian success — embedded autonomy, a competent disciplined bureaucracy, a political settlement permitting long-horizon action, and the willingness to discipline powerful firms — are demanding and largely absent in many African states (the state-capacity and political-economy problems of the Governance area), so attempts at industrial policy risk the capture and indiscipline that produced ISI's failures rather than East Asia's success. The optimistic/nuanced view: (1) East Asian conditions were not pre-given but were BUILT (the Governance course's point that capacity is constructed); (2) some African states (or sectors, or sub-national units) have or can develop the requisite capacity (the 'pockets of effectiveness' idea); (3) the lessons (especially discipline) can be applied even imperfectly; and (4) recent African attempts (Ethiopia's state-led industrialisation push before its troubles, Rwanda's developmental ambitions) show the model being attempted, with mixed results. The honest position: the developmental-state model is demanding and not easily transplanted — its success depended on state-capacity conditions that are the binding constraint in much of Africa — but its core lesson (discipline support, condition it on performance, build the autonomous capacity to do so) is applicable and important, and building that capacity is itself the developmental task. Industrial policy in Africa is not impossible, but it is inseparable from the state-building agenda of the whole program.
Exercise
An African government, citing Korea, wants to launch an industrial policy: it will protect and subsidise selected manufacturing sectors and direct credit to them through a state bank. (1) Explain what Korea actually did that the government must replicate beyond 'protect and subsidise'. (2) Explain why the discipline of conditioning support on export performance was essential. (3) Explain the embedded-autonomy condition and why it's the binding constraint. (4) Assess whether and how this government could realistically pursue the model.