Is It Possible to Beat the Random Walk Model in Exchange Rate Forecasting? More Evidence for Brazilian Case.

The seminal study of Meese et al. (1983) on exchange rateforecastability had a great impact on the international finance literature.The authors showed that exchange rate forecasts based on structural modelsare worse than a naive random walk. This result is known as theMeese--Rogoff (MR) puzzle. Although the validity of this result has beenchecked for many currencies, studies for the Brazilian currency are notcommon. In 1999, Brazil adopted the dirty floating exchange rate regime.Rossi (2013) ran an extensive study on the MR puzzle but did not analyseBrazilian data. Our goal is to run a “pseudo real-time experiment” toinvestigate whether forecasts based on econometric models that use thefundamentals suggested by the exchange rate monetary theory of the 80s canbeat the random model for the case of the Brazilian currency. Our work hasthree main differences with respect to Rossi (2013). We use a biascorrection technique and forecast combination in an attempt to improve theforecast accuracy of our projections. We also combine the random walkprojections with the projections of the structural models to investigate ifit is possible to further improve the accuracy of the random walk forecasts.However, our results are quite in line with Rossi (2013). We show that it isnot difficult to beat the forecasts generated by the random walk with driftusing Brazilian data, but that it is quite difficult to beat the random walkwithout drift. Our results suggest that it is advisable to use the randomwalk without drift, not only the random walk with drift, as a benchmark inexercises that claim the MR result is not valid.