This paper proposes a flexible regime-switching framework to model the dynamics of foreign exchange (FX) rates.
In their approach, the FX rate dynamics may switch between two different models. Specifically, this paper consider the regime-switching Ornstein-Uhlenbeck (RSOU), regime-switching Brownian Motion (RSBM) and regime-switching Brownian Motion – Ornstein-Uhlenbeck (RSBMOU) models, where the model parameters are modulated by a hidden Markov chain over time. This paper apply an Expectation–Maximization (EM) algorithm along with filtering and smoothing techniques.
Using over two decades of historical FX data, this paper show the effectiveness of our approach and illustrate how the regime varies over time through economic cycles and major financial crises.
This paper consider a RSOU model and a RSBMOU model to study the behavior of FX rates. To calibrate model parameters, this paper apply the EM algorithm along with time series filtering and smoothing techniques. To their knowledge, parameter estimation via the EM algorithm has not been applied to both models. This paper proceed to compare the outcomes obtained from both models and this paper show that both models are good fits for FX rate data.
