The predictive power of regime-switching models for stock market returns.

Abstract

Nowadays, nonlinearity plays an important role in predicting economic and financial variables. The objective of this paper is to investigate the out-of-sample forecasting performance of a regime-switching model for three stock markets. We examine whether the nonlinear regime-switching model improves predictions compared to linear models. We first evaluate all the models by computing forecast error criteria, we then check for statistical significance and in the end we implement a robustness test. Eventually, the regime-switching model outperforms only the random walk and has equal predictive power to the autoregressive model.