Is implied correlation worth calculating? Evidence from foreign exchange options and historical data

This paper examines the performance of implied correlations in forecasting subsequently realized correlations between exchange rates. Implied correlations are derived from sets of implied volatilities on the three exchange rates in a currency trio. We compare the forecasting performance of the implied correlations from two currency trios with markedly different characteristics over two forecast horizons (one month and three months) against a set of alternative correlation forecasts based on time-series data. ; For the correlations in the USD/DEM/ JPY currency trio, we find that the option-based forecasts are useful in predicting subsequently realized correlations. Specifically, they tend to be more accurate than the simple forecasts based on time-series data (i.e., historical correlations and exponentially weighted moving average correlations) and contain useful information that is not present in the other forecasts. However, since correlation forecasts based on a bivariate GARCH(1,1) model improve the performance of implied correlations, we reject the hypothesis that the implied correlations fully incorporate all the information in the price history. ; For the correlations in the USD/DEM/CHF currency trio, the option-implied correlation forecasts are less useful in predicting realized correlations. For two of the three correlations, implied correlations are not as accurate as the forecasts based on time-series data and provide no additional information. For the third correlation, the implied correlations do contain useful information, but the economic benefits of using these implied correlations may be small due to this correlation's low level of variability.