Comparative Advantage

In the AI Story

The empirical record on comparative advantage is mixed in ways that the orthodox presentation of the doctrine systematically obscures. Trade does generate gains, in many cases substantial gains. But the distribution of those gains depends critically on whether countries can move up the value chain over time — and movement up the value chain is precisely what comparative advantage doctrine, applied without modification, makes structurally impossible. A country specializing in low-value-added activities reflecting its current comparative advantage is locked into low-value activities indefinitely, while countries that violated the doctrine through strategic intervention built capability in high-value activities.

The historical pattern is unambiguous. Every nation that successfully developed industrial capability did so by violating comparative advantage doctrine — building industries in which it had no current comparative advantage but for which strategic state action could produce future competitive capability. Britain in textiles, the United States in manufacturing, Germany in chemicals, Japan in electronics, Korea in semiconductors — each case involves the same pattern of building capability against current comparative advantage in order to produce future comparative advantage that would not exist otherwise.

The contemporary application to AI is direct. The standard prescription for developing nations facing the AI transition — adopt the leading tools, integrate them into your economy, compete on existing comparative advantage — is comparative advantage doctrine applied to AI. Following the prescription locks developing nations into AI consumption rather than AI production, into providing data for foreign training rather than building domestic models, into application development on foreign platforms rather than capability development on domestic infrastructure.

Chang's critique of comparative advantage is not advocacy for autarky or rejection of trade. It is advocacy for sequencing — the recognition that countries need to build capability before being exposed to competition that destroys nascent industries before they have time to learn. The critique is what Friedrich List articulated in 1841, what successful developers have practiced consistently, and what contemporary orthodoxy continues to deny in defiance of the historical record.

Origin

The doctrine was formalized by David Ricardo in his 1817 Principles of Political Economy and Taxation, building on Adam Smith's earlier discussion of absolute advantage. Ricardo's specific example involved English cloth and Portuguese wine — an example whose empirical inaccuracy (Portugal had been deindustrializing under the Methuen Treaty's free trade with England) Ricardo did not address.

The doctrine has remained the foundation of mainstream trade theory through subsequent elaborations by Heckscher and Ohlin (factor proportions), Samuelson (factor price equalization), and the new trade theorists of the 1980s (economies of scale and product differentiation). Each elaboration has added complexity to the original framework while preserving the core conclusion that trade liberalization produces mutual benefit.

Key Ideas

Mathematical elegance. The theory is mathematically coherent — given its assumptions, the conclusions follow rigorously.

Assumption fragility. The assumptions on which the mathematics rests (no factor mobility, fixed technology, no scale economies, full employment, costless adjustment) systematically fail in real-world conditions.

Sequencing problem. The doctrine, applied without temporal sequencing, locks developing nations into low-value activities and forecloses the capability building that would produce different future comparative advantages.

AI application. The contemporary prescription that developing nations should adopt AI tools and compete on existing comparative advantage is comparative advantage doctrine applied to AI — with the predictable consequence of locking them into consumption rather than production.

Debates & Critiques

Defenders of comparative advantage argue that the dynamic version of the doctrine — recognizing that comparative advantages can change over time — addresses the sequencing critique. Chang's response is that the dynamic version, properly understood, requires exactly the strategic state intervention that the static version is invoked to prohibit. The doctrine, in either version, fails to provide a basis for the contemporary policy advice that developing nations should liberalize before building capability.