Are exchange rates ultimately tied down by economic fundamentals, or are they free to drift at random on a sea of speculation? This is of course an interesting question both on the theoretical and the empirical levels. If fundamentals play no empirically significant role in the data, then this is something that needs to be explained and taken into account when writing models. Most theoretical open economy models make strong theoretical assumptions regarding nominal exchange rates: for example purchasing power parity (PPP) holding every instant, that the current and capital accounts must be in equilibrium in the long-run and so on. If these basic theoretical restrictions are found to be absent from the data what is to be done? The papers in this Controversy focus mainly on the empirical aspects of this debate and in particular the mean and variance of nominal exchange rates. Ever since Meese and Rogoff (1983) reported that a comprehensive range of exchange rate models were unable to outperform a random walk, the role of traditional macroeconomic variables in explaining exchange rate movements has been a hotly contested issue. Some researchers have interpreted the evidence that has accumulated post Meese and Rogoff as reinforcing the original result, whereas others are more favourably inclined towards fundamentals. The issue of using fundamentals to outperform a random walk may be interpreted as relating to shortto medium-run exchange rate behaviour (usually thought of as relating to horizons of up to 36 months); however, an important controversy also arises with respect to the longer run predictability of exchange rates. The long-run fundamentals-based model which economists often first turn to when asked for a view of how close a currency is to equilibrium is that of purchasing power parity (PPP). In fact PPP has become one of the most intensively scrutinised relationships in applied macroeconomics over the last 10 years or so. Although the intensity of this research has been driven more by a desire to implement recently developed econometric methods such as multivariate cointegration and panel unit root tests than any new interest in PPP per se, this work has, nevertheless, thrown up some interesting findings. The key finding in this literature is that real exchange rates are mean-reverting (which is a key requirement of PPP), but the magnitude of such reversion is far too slow to be consistent with a traditional form of PPP such as that formulated by Gustav Cassel. Should PPP be abandoned, therefore? Some would argue in the affirmative and propose that equilibrium exchange rates can only be understood if the key real determinants of real exchange rates are