Sentiment and exchange rate movements: Much cry and little wool?

Exchange rates and their movements could be explained and forcasted by established economic theories and by many empirical exchange rate models. However, these traditional empirical models do not serve as a good means of properly testing and forecasting exchange rate movements nowadays because, in many cases, large exchange rate swings could be better explained by institutional, behavioural, and other determinants, and by market expectations, than by economic fundamentals. In this way, we join the strand of research using the concept of nowcasting as a method, based on using the data provided by social media.

The objective of the paper is to examine the short- and long-term co-movements between the exchange rate of the euro and the US dollar, and sentiment represented by news posted on 20 sources, mainly Reuters, Bloomberg, WSJ, LA Times, CNBC, Forbes, Business Insider, and Yahoo Finance. We cover the minute data of five major currency pairs: EUR/USD, USD/JPY, GBP/JPY, and USD/CHF. We employ wavelet coherence with phase shift to identify the causality in Granger sense. We apply the Monte Carlo method to estimate the significance of results and edge effects. Our results indicate that the impact of sentiment prevails in explaining high frequency movements of exchange rates.

We confirmed the significant impact of sentiment on the movements of the exchange rate of the euro currency pairs. In contrast, we did not confirm a stable and robustness impact of sentiment on US dollar and other currencies. In addition, we identified nonlinearity in our models. Negative news influences currency movemements particularly hard in comparison with positive news.