Trade Patterns, Income Differences and Gains from Trade

Quantifying the gains from international trade is an area of research that has been widely studied using a variety of trade models. At the same time, it has been shown that non-homotheticities are useful for matching the systematic patterns of trade present in disaggregated trade data. We bring these two literatures together to ask how non-homotheticities affect our predictions for gains from trade. To do so, we develop a N-country trade model that exactly matches bilateral trade, population, GDP per capita and within country income inequality for many countries. We include non-homotheticities to match patterns of trade between rich and poor countries that we observe in highly disaggregated trade data. We then make use of the results from Arkolakis, Costinot, and Rodriguez-Clare (2012), which gives a simple formula for gains from trade in a large class of homothetic models, including a version of our model with the non-homotheticity removed. Our main finding is that homothetic models underestimate gains from trade in countries with small populations and low productivities, and overestimate gains in countries with large populations and high productivities. The homothetic model overestimates the gains from being open to trade in the U.S. and Japan by 14% and 22%, and underestimates them in Spain and Italy by 24% and 14%.