Dani Rodrik on Premature Deindustrialization and Why the World is Second Best, at Best

Dani Rodrik — the Harvard economist whose work on trade, industrial policy, and the limits of globalisation made him the discipline’s most rigorous dissenter — joins Tyler Cowen in Ep. 4, recorded live at George Mason, to argue that today’s poor countries are losing their factories before they ever grow rich, that economics is a science of knowing which model applies when rather than a body of universal laws, and that the honest answer to most policy questions is second best, at best. One of the earliest Conversations with Tyler.

Key ideas

1. Premature deindustrialisation closes the fast lane to wealth. Deindustrialisation means an economy’s manufacturing share of jobs and output falling. It happened in rich countries only after they grew rich; it now strikes poor ones at far lower income levels — they shed factory work before building a broad middle class. Rapid industrialisation was the one engine that produced 4–5 per cent annual per-capita growth for decades in Japan, South Korea, Taiwan, and China. Automation and Chinese competition have shut that engine, so sub-Saharan Africa faces slower, harder convergence built on the patient accumulation of skills and institutions.
2. Manufacturing is special, and the reasons are concrete. Factory technology travels across borders far more easily than agriculture or services, whose productivity depends on local soil, weather, or domestic demand. Manufacturing absorbs unskilled labour straight off the farm — eye-hand coordination, not literacy, ran the early footwear plants — and trebles a worker’s productivity. And it is tradable: you can import inputs and export outputs, so one segment can master world markets without waiting for the whole economy to grow.
3. The globalisation trilemma: pick two of three. Rodrik’s best-known idea holds that deep economic integration (hyperglobalisation), national sovereignty, and democracy cannot all coexist in full — you may combine any two but never all three. Applied to the eurozone, the bind is stark: keep democracy and the choice is either fuller political integration to back the existing economic union, or loosening that union, beginning with the common currency. The euro crisis, Rodrik argues, was sold as a moral failing of profligate southerners rather than a shared crisis of interdependence, and that framing made it nearly impossible to resolve.
4. Economics is a science of models, not laws — the craft is knowing which one applies. No single model is universally true; the discipline holds a library of them, each illuminating a different situation. The skill economists most lack is diagnosis — picking the model that fits the case in front of you. Rodrik’s verdict on randomised controlled trials follows from this: they buy strong internal validity (clean causal estimates here) at the cost of external validity (no guidance on whether the effect travels), and have crowded out the conceptual and theoretical work that says why and where else.
5. The world is second best, at best — so distrust best practice. Second-best reasoning, from his teacher Avinash Dixit, holds that when one part of an economy is already distorted, fixing another in isolation can make things worse; the textbook first-best optimum is rarely available. From this Rodrik rejects ‘best practice’ — the lazy faith that a recipe that worked somewhere can be copied wholesale. Industrial policy, properly understood, is not picking winners but a pragmatic government asking the private sector what it can do to unlock activity that would not otherwise start, accepting that some bets will fail.

Content

Premature deindustrialisation and the future of Africa

Cowen opens on the theme he calls the most interesting in Rodrik’s work: a mix of automation and trade with richer nations may stop today’s poor countries from industrialising along the South Korean or Taiwanese path. Rodrik does not read this as a bleak future for sub-Saharan Africa, but he does think the very rapid convergence of East and Southeast Asia is now close to impossible to replicate. Every country that made the leap to advanced status — Japan from the late nineteenth century, then South Korea, Taiwan, and China — did so through rapid industrialisation, the one route that delivers 4–5 per cent per-capita growth decade after decade. With that route narrowing, growth must instead come from the slow, unglamorous accumulation of human capital, skills, and institutional improvement, which does not produce headline numbers.

Pressed on whether such countries are condemned to a ‘ramshackle’ existence — Lagos always looking like Lagos, never like Seoul — Rodrik shifts the question from the quantitative to the qualitative. What built liberal democracy in early-industrialising Britain was not the growth rate, which by modern emerging-market standards was contemptible, but the transformation of social structure: the spread of bourgeois liberal values and the restraints placed on the state. Rapid growth, he notes, can aggravate as much as soothe; the Arab Spring erupted not in the least-developed Arab states but in Tunisia and Egypt, the ones held up as models. Invoking Albert Hirschman’s possibilism, he resists deterministic stage-theories of development: roughly half is structure inherited from geography and history, half is agency — the room to manoeuvre around the constraints you are handed.

Why manufacturing is special

Cowen asks for the microfoundations: what exactly distinguishes manufacturing from agriculture, given Rodrik’s own data showing faster productivity convergence in industry? Rodrik gives three concrete reasons. First, manufacturing technology is portable — a Toyota plant runs more or less the same in Japan, the United States, or South Africa — whereas agricultural technology carries a large non-traded component that demands local adaptation to soil and weather, as the Green Revolution showed. The same is true of most services: you cannot simply transplant an advanced country’s hospital or school system. (Pushed on whether this makes the difference cultural, Rodrik declines the bait — he is unconvinced the recent revival of culture in economics adds much we did not already know.)

Second, traditional manufacturing absorbs vast numbers of unskilled workers. A Chinese entrepreneur told him that her early footwear factory required not literacy or numeracy but basic eye-hand coordination; she could put thousands of people from the countryside into the plant and treble or quadruple their productivity, an opportunity few other sectors offer. Third, manufacturing is tradable: you import inputs and export outputs, so you need not develop a whole domestic industrial complex or wait for an economy-wide productivity revolution to create customers. Mastery of one segment at a time is enough. Services, by contrast, mostly sell at home, so they need every sector’s income to rise together — otherwise the successful service sector simply runs into deteriorating domestic terms of trade.

The new, more modest case for industrial policy

Given premature deindustrialisation and Richard Baldwin’s account of supply chains fragmenting across the globe, Cowen asks whether we should now be more sceptical of industrial policy than when Singapore began. Rodrik reframes: not more sceptical, but expecting a lower return, because the old feat of putting 30–35 per cent of the labour force into manufacturing will not recur. He draws two implications. The term itself should change — better ‘structural transformation policy’, oriented to moving up the economy generally, including into the genuinely tradable services. And the criterion of success should change: low- and middle-income countries fixate on output, exports, R&D, and patents, when the right measure is employment. A policy that generates profits and investment while shrinking jobs is not high-return for the economy as a whole. Rodrik says he now goes around advocating forceful industrial policy considerably less than he once did.

The globalisation trilemma and the eurozone

Cowen introduces Rodrik’s trilemma — democracy, national sovereignty, and global economic integration are mutually incompatible in full; you can have any two. Asked what the eurozone should do, Rodrik first plays the economist’s straight bat (the job is to lay out trade-offs, not choose), then answers as a citizen. Subject to keeping democracy, the choice is inescapable: either more political integration or less economic integration. Before and well into the crisis his preference was full steam ahead with political integration — federalise Europe, add fiscal union, build a pan-European political space to back the economic union already in place. Several years on, and very reluctantly, he now thinks at least some countries need a loosening of monetary integration, however hard a graceful exit from the euro would be to engineer. His deeper diagnosis is political: European integration was always an elite project, and the euro crisis was sold to German voters as a morality tale of profligate southerners rather than what it was — a crisis of interdependence in which German banks were as implicated as Greek borrowers.

Trilemmas around the world: China, capital controls, and immigration

The trilemma frame travels. On China’s defence of the yuan, Rodrik counsels against anything hasty on capital controls and rejects the financial-press narrative of mismanagement: this is the country that engineered history’s most miraculous poverty reduction, and outsiders should be slow to second-guess it. The 300 billion dollars of reserves spent is, to him, a drop in the bucket and part of optimal adjustment. The real risks lie elsewhere — a vast political transformation (eventually opening up and democratising, with no template for how) and an economic transformation (rebalancing toward domestic consumption) that China has barely begun. On Milton Friedman’s older dilemma between high immigration and the welfare state, Rodrik starts from the premise that borders carry moral significance, then runs the calculation: to justify keeping a migrant out, you would have to value a co-national more than five times as much as an outsider — a weighting he finds excessive. Today’s migration regime, he argues, sits roughly where the trade regime sat in the 1950s, with large gains still available from liberalisation; he points to Germany’s successful absorption of Turkish migrants in the 1960s and 1970s as the apt precedent.

Turkey, liberal democracy, and the limits of the median voter

Cowen, noting Rodrik was born and raised in Turkey, turns to liberal democracy — Rodrik’s then-current research preoccupation. Does he trust the median voter in religiously conservative Konya? Yes, as much as any country’s median voter — but democracy is not only about reflecting the median voter’s views; it is equally about not letting the majority trample everyone else. The world now has more electoral democracies than autocracies, yet most are merely electoral: free and fair voting that still lets winners ride over the rights of the losing minority. The missing element is the liberal one — rule of law, non-discrimination, the restraints on what the majority may do. Rodrik faults the political-economy literature on democracy for modelling elections and the median voter while ignoring these restraints. He illustrates with pre-1975 Lebanon’s consociational order, where so many cross-cutting cleavages meant no group could expect to rule forever, so all had repeated-game incentives to compromise — a balance the influx of Palestinian refugees ultimately upset.

Structure, agency, and the underlying current

Cowen presses for the current tying Rodrik’s work together, reading even the deindustrialisation papers as versions of a single question: why has Turkey never become a fully free and modern society? Rodrik calls this perceptive — it has motivated him since his first foray into social science. His boiled-down answer is 50 per cent structure, 50 per cent agency: Turkey began on the wrong end of the global division of labour, its craft industries decimated by British imports in the nineteenth century, then defeated in war; Atatürk’s top-down, brutal modernisation made it secular and middle-class but pushed out the religious conservatives and Kurds whose exclusion still festers. Pushed on the sharp east-west line in European development, Rodrik reaches for Barrington Moore: the decisive variable was how agriculture was commercialised — peasant smallholding and landlords who diversified into urban commerce (as in Britain) produced one path; large estates wedded to labour-repressive farming and allied to the state produced another. Geography matters, but as a conduit for ideas and benchmarking — France, he suggests, is the democracy it is largely because it sat next to Britain and the Netherlands rather than in the east.

Method: RCTs, best practice, and second best

In the lightning round and the audience Q&A, Rodrik’s methodological commitments surface plainly. Randomised controlled trials he calls today overrated — not a bad tool but a crowding-out one, prized for internal validity (clean causal identification) while saying nothing about why an effect occurs or whether it would recur elsewhere. Asked about best-practice institutions, he is blunt: he hates the notion, calls it harmful, and quotes his teacher Avinash Dixit — ‘the world is second best at best.’ Best practice rests on the fantasy that something that worked somewhere can be lifted and dropped elsewhere; it breeds the lazy scorecard mentality of World Economic Forum rankings. Successful societies instead distil outside experience through the lens of their own local knowledge. The same pragmatism shapes his closing word on India: never recommend anything unconventional, expect lower returns to industrial policy now, but keep a government that asks the private sector what it can do to unlock activity — accepting that some attempts will fail, because having tried and failed often beats never having tried.

Related

• What Makes Economies Grow — theme; Rodrik supplies the microfoundations of the structural-transformation camp and the premature-deindustrialisation warning, plus the ‘second best, at best’ epistemics
• Dani Rodrik — speaker; Harvard economist of trade, industrial policy, and the globalisation trilemma
• Tyler Cowen — host
• Daron Acemoglu — fellow Turkish-American economist; shares the institutions-over-geography framing and the structure-versus-agency debate Cowen presses here
• Joe Studwell on Africa, Asia, and What Development Actually Requires — companion on development economics; Studwell and Rodrik converge on manufacturing and industrial policy as the engine of convergence
• Abhijit Banerjee on Theory, Practice, and India — counterpoint on development method; the RCT approach Rodrik judges overrated here
• Brad DeLong on Intellectual and Technical Progress — related Conversations with Tyler on long-run growth and economic history