The impact of media tone on dollar exchange rate volatility : GARCH applications

Abstract

Using several million news and social media articles related to US dollar, I examine the role of Media tone in predicting daily dollar exchange rate returns from 1998 to 2021. In this study, I use GARCH specifications to explore the relationship between currency Media tone and dollar exchange rate volatility. I include currency Media tone as an external regressor into both mean and variance equations of different specifications following the significance and sign of the causal relationship. The results reveal that the impact of currency Media tone to be time-dependent, source-dependent, statistically significant but economically insignificant. The inclusion of the currency Media tone into the variance equation of the GARCH specification increases the explanatory power of the model from 10,43% to 10,85%. The sign of the impact has been varying from specification to specification, however in most cases the sign of the coefficient for currency Media tone was positive in the mean equation and negative in the variance equation. This means that information about US dollar increases the dollar exchange rate returns but decreases the exchange rate volatility.