It is a well-documented fact that changes in exchange rates are very difficult to explain using macroeconomic fundamentals such as, money supply, real income, interest rate, trade balance and bond supply. Forecasting models based on macroeconomic variables, tend to do no better than a random walk model in out-of-sample exercises. This phenomenon is known as the Meese and Rogoff puzzle. We re-examine this puzzle by employing commodity prices as an alternative variable. We find that changes in commodity prices have power in explaining fluctuations in commodity currency exchange rates both in-sample and out-of-sample. This relationship is linear in nature and strongest at the daily frequency. The relationship is present for all four studied economies and does not weaken when the GBP is used instead of USD as a base currency. The observed relationship is also robust to using either the recursive or rolling estimation scheme. We also find that controlling for asymmetries in changes in commodity prices does not lead to any significant improvements in the performance of the commodity driven exchange rate model. The observed relationship, however, disappears when the lagged commodity price change is used as the predictor instead of the realized change.

nhhmas