A generalized design for bilateral trade flow models

This paper suggests a full interaction effects design to analyze bilateral trade flows. This is illustrated with an unbalanced panel of bilateral trade between the triad (EU15, USA and Japan) economies and their 57 most important trading partners over the period 1986–1997. Our full interaction model finds empirical support for the New Trade Theory and Linder’s hypothesis. We show that the omission of one or more interaction effects can result in biased estimates and misleading inference.