Monetary Models of Dollar/Yen/Euro Nominal Exchange Rates: Dead or Undead?

Abstract

the predictions of the models were based on ex-post-realised values of the explanatory variables. In other words, the standard exchange rate models we estimated exhibited extremely poor out-of-sample fit.2 At the time, these nihilistic results seemed quite radical, since there was widespread optimism about the potential of monetary models, especially sticky-price monetary models, to explain how flexible exchange rates function. After all, the main cornerstones of monetary exchange rate models were relatively noncontroversial: long-run purchasing power parity (or some variant of PPP), and a presumption that there is a strong long-run connection between money growth and inflation. Today, the Meese and Rogoff (1983a) results no longer seem quite so crazy. Despite longer data sets on modern floating rates, and the application of more sophisticated econometric techniques, researchers have continued to find it very frustrating to firmly demonstrate any systematic relationship between exchange rates and macroeconomic fundamentals, at least for the cross rates between the dollar, DM (euro) and yen. It is true that researchers have occasionally found particular sub-samples where certain models seem to perform noticeably better than the random walk model but, as a rule, these results wilt under sustained out-of-sample testing. Surveying the evidence in their survey for the Handbook of International Economics, Frankel and Rose (1995) conclude that numerous attempts to overturn the Meese and Rogoff (1983a) results have failed.