Regime Selection as an Alternative to the Grubel-Lloyd Index

Abstract This paper contends that the regime selection probability can provide an “empirically justifiable†alternative to the GL index as a measurement of intra-industry trade. The regime selection probability is an average of sample selection probability that each trade flow data comes from inter-industry trade (or equivalently from intra-industry trade). It measures the extent to which a given set of observations is explained by the inter-industry trade regime (or the intra-industry trade regime). The Heckman two-step procedure is used: a probit model is fitted first to a separation indicator, and then the estimated parameters are incorporated into the unconditional probability equation for the second stage estimation. The sample selection model requires an auxiliary separation indicator to partition the sample, so that each of its elements may match up with either comparative advantage or product differentiation with scale economies. As it is a selection criterion based on “empirical regularities†for separating intra-industry trade from inter-industry trade, the regime selection probability can improve upon the GL index as a relevant indicator of intra-industry trade. It can also be used to infer a transition in trade patterns, since a change in its value indicates how a shifting of explanatory power from comparative advantage to product differentiation takes place.