Spurious Tests and Sign Reversals in International Economics

Empirical tests of numerous theories in international economics, e.g., the monetary explanation of the balance of payments/exchange rate, the portfolio balance model of capital flows, purchasing power parity, and covered interest rate arbitrage, usually impose the subtractive linear constraint that corresponding foreign and domestic explanatory variables have equal but opposite effects on the dependent variable.' Rasulo and Wilford [15] investigate empirically the bias in general caused by such a restriction in monetary models of the balance of payments/exchange rate.2 What is not recognized, however, is the peculiar nature and potential severity of the bias, including sign reversals, which results from subtracting (as opposed to aggregating) explanatory variables; and the wide range of tests in international economics which may be spurious because of this type of bias.