Does the Law of One Price Really Hold for Commodity Prices?

Abstract In spite of the empirical failure of the “Law of One Price,” it is usually assumed that commodity prices are perfectly arbitraged, at least in the long run. This paper argues that this assumption is counterfactual and that much of the empirical evidence provided to support it is flawed and affected by econometric shortcomings (spurious regressions, nonstationarity in the data, inappropriate use of first differences). An alternative methodology (co‐integration), by which a long‐run relationship between nonstationary series can be tested, is proposed. Tests of nonstationarity and co‐integration for a group of commodities show quite uniformly that the “Law of One Price,” as a long‐run relationship, fails, and that deviations from the pattern are permanent.