Forecasting exchange rates: do banks know better?

Abstract

The current paper has dual objectives (i) we first evaluate the performance of major commercial banks in forecasting future spot exchange rates, using the random walk model as the benchmark; (ii) we then investigate the sources of forecast errors and the forecasting tendencies of banks. Our analysis has been conducted using the forecasts made for four major currencies, i.e., the British pound, German mark, Swiss franc, and Japanese yen, over four forecast horizons, i.e., 3-,6-, 9-, and 12-month. Key empirical findings include First, the random walk model outperforms all sample banks in forecasting the yen exchange rate at all horizons. For the remaining currencies, however, a few banks sometimes provide more accurate forecasts than the random walk model especially at longer horizons Second, imperfect correlation between predicted and actual exchange rate changes is found to be the dominant source of prediction errors of the banks Third, the home-country bank generally forecasts the country's currency rate more accurately than other banks, suggesting a degree of information asymmetry Fourth, the majority of sample banks are found to be momentum forecasters, tending to extrapolate the recent currency changes. Interestingly. a "contrarian" bank is found to outperform other banks. Furthermore, all the banks often either over-predict or under-predict future exchange rates, suggesting a herding behavior.