Information Processing of Foreign Exchange News: Extending the Overshooting Model to Include Qualitative Information from News Sentiment

Abstract

In a globalized world, the volume of international trade is based on both import and export prices, thereby making a country's economy highly dependent on exchange rates. In order to study exchange rate movements, one frequently exploits the so-called Dornbusch overshooting model. However, the model is controversial from a theoretical point of view: it presupposes the processing of information, though this is not directly reflected by the underlying variables. As a remedy, this paper investigates a potential information bias by including textual news content, thus adjusting for information dissemption. As such, we perform a multivariate analysis to compare the classical overshooting model with an extended variant that includes news sentiment. Our results show that news has a substantial explanatory power of 11% of the exchange rate forecasting error variance. In addition, we also find statistical evidence that a shock in news sentiment could lead to overshooting.

Keywords: Information processing, economics of IS, decision analysis, country-level analysis, behavioral science, econometric analyses, cognitive bias, text mining

JEL Classification: D51, E47, E40, E43, F31, C13, C32, C51,C52, G14