The paper “Trade Policy in a Ricardian World with International Knowledge Diffusion” shows that, although the “non-discrimination” principle is optimal from a static perspective, when dynamic knowledge diffusion is taken into account, countries will set trade policy discriminatively so as to reorient trade towards more technologically advanced trading partners for faster economic growth. The author builds upon a heterogeneous-firm Ricardian model (Eaton and Kortum, 2002) with international knowledge diffusion (Buera and Oberfield, 2020). While goods are traded and producers interact, technology is transmitted. International knowledge diffusion is linked to international trade in that the probability of learning from an origin is positively correlated with trade volume. Thus, a country’s trade pattern can alter its law of motion of technology, and trade policies can influence economic growth through adjusting trade barriers. In order to enjoy a higher growth rate by engaging more with advanced economies, countries tend to discriminate against relatively under-developed trading partners when setting trade policies. This implies that the non-discrimination principle advocated by WTO is suboptimal, as it requires countries to impose a uniform tariff on all trading partners. To study the optimal trade policies of countries in different development levels, the paper considers three cases of the Home country — 1) at the 5th percentile if the Home Country is under-developed, 2) at the 50th percentile if middle-income, and 3) at the 95th percentile if most-developed.

This paper shows that the optimal tariff is discriminatory based on trading partners’ development levels. The discriminatory nature of optimal tariff is a result of the dynamic interaction between international trade and knowledge diffusion. In comparison with non-discriminatory tariff schedules, the ability to tax imports based on their origins enables countries to obtain a higher living standard. With free trade being the baseline, an under-developed country can promote welfare by 2.14% if discrimination is allowed, compared with 1.98% achievable by a non-discriminatory tariff schedule. In this case, welfare gains from trade policy increase by 7.87%. However, welfare gains from trade policy is only 0.08% for a developed Home country, indicating that the WTO Most-Favored-Nation (MFN) principle is plausible for developed countries, and yet is costly for under-developed countries. This paper calls for a certain level of flexibility with regard to the “non-discrimination” principle for under-developed countries, since quantitative simulations show that the ability to impose differential trade policy is significantly welfare-improving if the Home country is under-developed.

In the long run, however, countries’ technology levels converge in the absence of external shocks. Thus the incentive to influence diffusion weakens. As a result, trade policies converge to balanced growth path policies, which is equivalent to the trade policy preferred by a myopic social planner under the parameters assumed by the paper’s benchmark simulations. The optimal trade policy in the long term, therefore, is non-discriminatory. This paper contributes by adding nuance to the discussion on the appropriate trade policy that countries should pursue, and does not take non-discrimination for granted.

Moreover, without concerns of retaliation or coalition, even in the long run countries will charge positive tariffs on imports. In doing so, countries can lift their own terms of trade and shift the distribution of welfare gains from trade across countries in favor of themselves.

By CHOW Yi Lin Dawn

Researcher: Chi ZHANG