International trade economics is a constant disappointment to me. It still relies on the Hecksher-Ohlin model, which is static and thus cannot say anything reliable about dynamics. And the vast majority of the empirical work uses linear reduced forms and is only out there to test signs, not to find out how large an effect is. It is then refreshing to see a paper that steps out of this morass, and into the right direction.
Yu Sheng and Xinpeng Xu get back to the foundations of the Hecksher-Ohlin model and integrate some frictions that go beyond the usual iceberg costs. Specifically, they take into account that the reallocation of labor across sectors takes a while by including a Mortensen-Pissarides search framework. Even in steady-state, it turns out some classic trade theorems may not hold. For example, factor price equalization does not hold for labor, as there is unemployment and the expected wage is equalized. Endowments determine the labor force, but employment is endogenous. This allows also to explain why countries with similar endowments still trade a lot: cross-country sectoral differences in unemployment mean the factor content of trade is not a sufficient statistic. All this also means that empirics of the Hecker-Ohlin model need to be adjusted, in particular unemployment needs to be factored in.