The Heckscher-Ohlin-Vanek (HOV) model of factor service trade is a mainstay of international economics. Empirically, though, it is a flop. This warrants a new approach. We test the HOV model with international and Japanese regional data. The strict HOV model performs poorly because it cannot explain the international location of production. Restricting the sample to Japanese regions provides no help, inter alia giving rise to what Daniel Trefler calls the "mystery of the missing trade." However, when we relax the assumption of universal factor price equalization, results improve dramatically. In sum, the HOV model performs remarkably well.