A Range-Based Multivariate Model for Exchange Rate Volatility

In this paper we present a parsimonious multivariate model forexchange rate volatilities based on logarithmic high-low ranges ofdaily exchange rates. The multivariate stochastic volatility modeldivides the log range of each exchange rate into two independentlatent factors, which are interpreted as the underlying currencyspecific components. Due to the normality of logarithmic volatilitiesthe model can be estimated conveniently with standard Kalman filtertechniques. Our results show that our model fits the exchange ratedata quite well. Exchange rate news seems to be very currency-specificand allows us to identify which currency contributes most to bothexchange rate levels and exchange rate volatilities