Most issues in international economics require recourse to some theory of purchasing power parity (PPP), but establishing the empirical validity of any of the available theories has proved very difficult in practice. There is general agreement that PPP does not hold in the short run and recent evidence, at least for the floating rate period of the 1970s and 1980s, suggests that it may not hold even in the long run. Economists have responded to this dilemma in a number of different ways. Some argue that this result is an artefact of the recent period; studies of longer periods provide stronger evidence for long-run PPP, in particular during the floating rate period of the 1920s. Others maintain that real exchange rates are determined, like other relative prices, by `fundamentals' such as productivity and demand factors; if these are non-stationary, PPP may appear to be violated over short time-periods. Yet others have sought to replace traditional PPP with an ex ante or `efficient markets' variety of the theory, in which the real exchange rate follows a random walk, arguing that the `traditional' test of PPP a regression of domestic price inflation on foreign inflation (expressed in the domestic currency) is inappropriate because it introduces a simultaneity bias.

This theory of PPP has been studied for more than a decade but it is generally ignored in discussions of international finance, perhaps because it is not based on explicit microfoundations. In Discussion Paper No. 635, Research Affiliate Michael Moore develops a heuristic argument that interprets deviations from PPP in terms of risk in a conventional Lucas-style asset-pricing framework. He develops an intertemporal model of a representative consumer whose income includes production from a number of different countries and returns from bonds denominated in the same countries' currencies and whose utility derives from these countries' goods over an infinite time-horizon. He finds that the real exchange rate between two countries is the instantaneous marginal rate of substitution between consumption of their goods, so the stochastic properties of the real exchange rate can be reduced to those of the associated marginal rate of substitution, which he uses to derive an orthogonality condition with which to test for ex ante PPP.

Moore tests this condition using the generalized method of moments for monthly data from the G7 countries for 1973-88. His results for exchange rate, price level and consumption data indicate that all the series on nominal exchange rate changes are stationary, but the stationarity of consumption growth and inflation are seriously in doubt for at least four and possibly six of the countries. Although these parameters are not very precisely estimated, the evidence in favour of this orthogonality result is strong.

Covered Purchasing Power Parity, Ex Ante PPP and Risk Aversion
Michael J Moore

Discussion Paper No. 635, April 1992 (IM)