The theories of absolute advantage (Smith) and comparative advantage (Ricardo) remain the starting point for every AS/AL account of why trade happens. They are nevertheless subject to significant limitations when they are read as a complete explanation of real-world trade patterns.
Core assumptions that may not hold. Both theories assume two countries and two goods, constant returns to scale, perfect factor mobility within a country but immobility between countries, no transport costs, no trade barriers, no externalities and static technology. Each assumption is a simplification that may be violated in the real world.
Limitation 1 — increasing returns to scale and intra-industry trade. A striking feature of real-world data is that most trade is between similar economies trading similar goods (Germany-France in cars, Japan-USA in electronics) rather than between dissimilar economies specialising in different goods. Classical comparative advantage struggles to explain this: when countries have similar factor endowments, their OC ratios are similar and the predicted gain is small. The modern 'new trade theory' explanation is increasing returns to scale and product differentiation: firms exploit scale economies by selling to the world market, and consumers in each country enjoy varieties they would not produce domestically. This is a limitation of the classical theory rather than a refutation — the OC logic still runs underneath — but it means the classical theory is incomplete.
Limitation 2 — transport costs, tariffs and other barriers. The theories assume frictionless trade. In practice, transport costs, tariffs, non-tariff barriers (standards, quotas, licensing), customs delays and currency-exchange costs all eat into the predicted gain. For some goods (heavy, low-value, perishable), these costs are large enough that no trade takes place even where comparative advantage exists.
Limitation 3 — unfavourable terms of trade and primary-commodity exporters. The theories do not say how the gain is shared between countries — only that there is a gain. If the world price ratio sits close to one country's domestic OC ratio, that country captures almost none of the gain. The empirical worry for primary-commodity exporters (the Prebisch-Singer concern) is that the long-run terms of trade move against them, so even a textbook-correct application of comparative advantage leaves them with a small share of the gain.
Limitation 4 — infant industries, strategic sectors and externalities. Static comparative advantage ignores learning-by-doing and dynamic scale effects. An infant industry that would be world-competitive after a period of protection would never develop under immediate free trade. Comparative advantage is also silent on strategic sectors (defence, food security, critical minerals) and on externalities — the environmental damage of specialising in energy-intensive extractive sectors is not deducted from the measured gain.
Limitation 5 — distributional and political-economy constraints. The theories deliver an aggregate gain but say nothing about who inside the country gains and who loses. Workers in displaced industries can lose jobs faster than new tradable sectors can absorb them, raising inequality and generating political backlash against trade. The 'perfect factor mobility within a country' assumption is a strong one, especially for low-skilled workers.
Reasoned evaluation. These limitations are real, but they are refinements of the classical theory rather than refutations of it. The OC-based framework correctly identifies why trade is positive-sum and why specialisation along comparative advantage can enlarge the consumption possibility set. Modern trade theory (scale economies, product differentiation) is a layer added on top, not a replacement. The right judgement is therefore 'useful but incomplete': the theories remain the first-order explanation for why trade happens and why it benefits both parties, but a complete account of real-world trade patterns needs scale economies, transport costs, terms-of-trade distribution, infant-industry dynamics and the distributional impact inside each country. For policy, the lesson is that free trade should be the default, with carefully targeted interventions where one of the identified limitations bites — rather than wholesale rejection of comparative advantage as a guide.