A0411
Title: Regime-specific exchange rate predictability
Authors: Joscha Beckmann - University of Hagen (Germany)
Marco Kerkemeier - University of Hagen (Germany) [presenting]
Robinson Kruse-Becher - FernUniversität in Hagen (Germany)
Abstract: Explaining exchange rate behaviour is a long-standing puzzle in international finance. The link between exchange rates and corresponding economic fundamentals has been elusive. Predictability highly depends on the choice of predictors, forecast horizon, sample period, chosen models and the applied evaluation metrics. We consider the most established benchmarks in the exchange rate literature with recent upcoming research on the relationship between uncertainty and exchange rates. We implement a threshold predictive regression framework with transition variables capturing uncertainty and a broad range of other variables. This setting is applied to nine major currencies against the US Dollar. The predictors we consider cover a wide range of measures, including uncovered interest rate parity, purchasing power parity, monetary fundamentals, Taylor rules, yield curve factors, stock markets and order flow data. In addition, expectation and uncertainty measures derived from the financial market and survey data, as well as sentiment measures, are adopted. Our results demonstrate regime dependence in the sense that predictability is only observed during times of high uncertainty, while there is no predictability under medium and low levels of uncertainty. Additionally, we observe strong co-movement between the predictability of different currencies.