The most powerful objection to industrial policy is 'governments can't pick winners' — they lack the information to know which industries will succeed, and they will pick badly or be captured. This module takes the objection seriously and presents the modern response, which reframes industrial policy from picking winners toward a disciplined process of discovery, and provides the tools (growth diagnostics) to do it better.
The critique
The 'governments can't pick winners' critique (rooted in Hayek's information argument and the public-choice capture critique) holds that: the state lacks the dispersed, tacit knowledge that markets aggregate, so it cannot know which industries or firms will succeed; its choices will therefore be wrong (backing losers); and even if it could know, the process would be captured by rent-seekers (the winners 'picked' would be the politically connected, not the economically promising — the Political Economy course). This is a serious objection, and it correctly describes much of what went wrong with old-style industrial policy. The modern defence does not deny it — it reframes industrial policy to address it.
Process, not picking
Hausmann-Rodrik: the reframing
The modern response (Hausmann-Rodrik, building on the self-discovery argument): industrial policy is NOT about a government confidently selecting winners it somehow knows will succeed — it CAN'T do that, and shouldn't claim to. It is about supporting a PROCESS of discovery and self-correction: support a PORTFOLIO of experiments in new activities (correcting the discovery externality — module 1), expect MANY to FAIL, scale up support behind the ones that succeed, and — crucially — CUT OFF the failures. 'The right way to think about industrial policy is as a discovery process — one where firms and the government learn about underlying costs and opportunities and engage in strategic coordination.' The key is not the government's prescience (which it lacks) but its WILLINGNESS TO LET LOSERS GO — to embrace failure as part of experimentation and not prop up failures indefinitely. This reframing answers the critique: the government doesn't need to pick winners (it picks experiments, and the market reveals the winners), and the discipline of cutting failures (the developmental-state lesson) prevents the indefinite support of losers that the critique rightly fears. Industrial policy as a self-correcting discovery process, not an omniscient winner-picking exercise, is the central modern conception.
Growth diagnostics
Finding the binding constraint (Hausmann-Rodrik-Velasco)
A complementary tool for doing industrial policy better is growth diagnostics (Hausmann, Rodrik, Velasco, 2005). Instead of applying a generic checklist of reforms (the Washington Consensus 'do everything') or picking sectors arbitrarily, growth diagnostics asks: what is the SINGLE most BINDING CONSTRAINT on growth and investment in THIS country right now? It works like a decision tree: is the problem low returns to investment (and if so, is it poor infrastructure, low human capital, or bad governance?) or is it high cost of finance (and if so, poor domestic savings or bad financial intermediation?)? By identifying the one constraint that is actually binding — the one whose relaxation would do the most for growth — policy can FOCUS scarce capacity and resources where they matter most, rather than dissipating them across a long reform list (Grindle's 'good enough governance' selectivity, the Governance course). Growth diagnostics is a disciplined, country-specific, problem-driven approach (the PDIA spirit of the Governance course) that improves industrial and development policy by targeting the binding constraint rather than guessing or doing everything. It is one of the most practically useful frameworks in development economics.
Comparative advantage: follow or defy?
The Lin-Chang debate
A central debate in industrial-policy design: should a country FOLLOW its current comparative advantage or DEFY it? Justin Lin's 'New Structural Economics' argues industrial policy should FOLLOW the country's LATENT comparative advantage — promote industries consistent with the country's current factor endowments (a labour-abundant poor country should promote labour-intensive manufacturing, not capital-intensive heavy industry), because industries that fit the country's comparative advantage will be VIABLE and competitive without permanent subsidy. Defying comparative advantage (promoting industries the country isn't yet suited for) produces non-viable firms needing endless support (the ISI failure). Ha-Joon Chang (and others) counter that successful catch-up often REQUIRED defying current comparative advantage — Korea promoted steel and shipbuilding when it had no comparative advantage in them, building the advantage through deliberate effort (the infant-industry/learning argument); following current comparative advantage might trap a country in low-value activities forever (a country good at growing cocoa stays a cocoa grower). The debate is unresolved and important: Lin's view counsels caution and viability (don't defy your endowments); Chang's view counsels ambition and transformation (build new advantages). The pragmatic synthesis: defy comparative advantage only SLIGHTLY and with discipline — promote activities just beyond current capabilities (where learning can build advantage) rather than wildly beyond them (where you get permanent non-viable dependants), and condition support on the path to viability. The tension between viability (follow) and transformation (defy) is at the heart of industrial-policy strategy.
Exercise
A low-income, labour-abundant African country is debating its industrial strategy. One adviser says, 'Pick high-tech industries like the Asian tigers did — electronics, autos.' Another says, 'Government can't pick winners; just liberalise and let the market decide.' (1) Critique both positions. (2) Explain how to reframe the strategy as a discovery process. (3) Apply growth diagnostics to focus the strategy. (4) Use the Lin-Chang debate to advise on which industries to target.