Trade, Multinational Production, and the Gains from Openness

A preliminary version of this paper can be found here.

This paper is part of a research agenda that aims to understand and quantify the gains from “openness.” By openness I mean all the different ways in which countries interact with each other. This research has so far focused on trade, multinational production and diffusion.

The first paper in this research agenda was: "Trade, Diffusion, and the Gains from Openness," November 2007 (first version November 2006, submitted). MATLAB codes and data. This paper presents a model where countries interact through trade and diffusion of ideas and uses the growth rate as a key target to calibrate the model. The main result is that the gains from openness are mostly due to diffusion and are much larger than the gains from trade.

This paper has now been absorbed by a more ambitious paper co-authored with Natalia Ramondo titled “Trade, Multinational Production, and the Gains from Openness.” A first version of this paper circulated in June 2008 under the title "The Gains from Openness: Trade, Multinational Production, and Foreign Ideas." The current version of this paper develops a model of trade and multinational production using the Multivariate Frechet distribution. The model is calibrated to match data on bilateral trade, bilateral multinational production, intra-firm trade, and growth.

Preliminary results imply that the gains from openness are much larger than the gains from trade – this is thanks to the large gains from MP. Moreover, our estimated gains from trade are significantly larger than the analogous results in Eaton and Kortum (2002) and Alvarez and Lucas (2007) – this result arises from our use of growth facts in calibrating the key parameter that determines the strength of comparative advantage in the model. Finally, for a reasonable range of parameter values our estimated model does not allow for large degrees of substitutability or complementarity between trade and MP: the different forces that we include in the model tend to cancel each other out.