How Robust are Sign and Rank Order Tests of the Heckscher-Ohlin-Vanek Theorem?

Abstract

The present paper will concentrate on four different rank order and two different sign propositions of the HOV model, which have, at various times, all been the subject of empirical analysis in the literature, and it will establish conditions, under which a given data set will support one (rank order or sign) hypothesis while at the same time rejecting another. This will shed some light on the circumstances influencing the likelihood of such non-robust test results. The paper will also examine the empirical relevance of this phenomenon of non-robustness using the multi-country data set published in Bowen et al.