This paper examines the determinants of the Euro-Dollar exchange rate, using the news approach to exchange rate modeling and some behavioral finance categories in the interpretation of its dynamics. A twice-daily frequency estimation was chosen, dividing the global trading day into an European Time Zone, ETZ (including also Asian trading), and an American Time Zone, ATZ. A new typology of news variables, unscheduled news, was employed, together with the traditional scheduled macroeconomic news. These former news, consisting of policy statements, market events, market beliefs, terror-related events turned out to be the main determinants of Euro-Dollar movements. Coefficient stability tests suggested to divide our 1999-2004 sample into three sub-periods roughly corresponding to the three phases of recent Euro history. The main finding of our analysis is the rejection of the semi-strong EMH once we move from the estimation over the entire sample to the three sub-periods. Here we find many lagged news variables to be significant, contrary to what EMH posits. The distribution of lagged news across time zones (ETZ and ATZ) and among the three sub-periods, indicates a substantial heterogeneity in the way news are decoded by market participants in the two trading zones and that exchange rates in ATZ react almost exclusively to American news, indicating that this zone influences the rest of the world but it is not affected by it. Scheduled news play a much bigger role in ATZ than in ETZ, especially the creation of new jobs in the US (the Non-farm Payroll). Exchange rate dynamics in ETZ is determined mostly by unscheduled news.
Read full abstract