The Markov-switching vector error correction model: dynamics, bayesian inference, and application to the spot and forward Swiss Franc, US Dollar exchange rates

Standard economic theory implies a close link between spot and forward exchange rates: in an efficient market with risk-neutral agents, the forward rate is an unbiased predictor of the future spot rate. However, it is very difficult to find this link in the data: on the contrary, the spot rate seems to move in the opposite direction than predicted by the forward rate, a phenomenon termed the ”forward discount anomaly”. For the Swiss franc, the anomaly is not only a short-run phenomenon, but is also present in the long-run. Many studies revealed persistent appreciation and depreciation periods in exchange rates. Such regime switching behavior may cause a „peso problem“, a situation arising if the ex post frequencies of the regimes differ from the ex ante probabilities. Peso problems are a possible explanation for the forward discount anomaly. This thesis examines a Markov-switching cointegrated vector error correction model of the Swiss franc/US dollar spot and forward exchange rates, in which the adjustment to long-run equilibrium is subject to regime switching. Estimation is done following a Bayesian approach, using Gibbs sampling in combination with data augmentation to explore the posterior distribution of the model parameters. Bayesian inference confirms the conclusion of classical tests that spot and forward rates are cointegrated. The results show that the presence of regime switching may cause a peso problem. However, the adjustment to equilibrium cannot be reconciled with the unbiasedness hypothesis.