Do endowments predict the location of production?: Evidence from national and international data

Examining the relationship between factor endowments and production patterns using data from Japanese prefectures and from OECD nations, we find evidence of substantial production indeterminacy. Regressions of outputs on endowments yield prediction errors six to 30 times larger for goods traded relatively freely than for non-traded goods. We argue that a compelling explanation for these results is the existence of more goods than factors in the presence of trade costs. If so, regressions of trade or output on endowments have weak theoretical foundations. Furthermore, since errors are largest in data sets where trade costs are small, we explain why the common methodology of imputing trade barriers from regression residuals has produced counterintuitive results.