A0790
Title: Estimation and inference for Markov-switching GARCH models
Authors: Chaojun Li - East China Normal University (China) [presenting]
Abstract: Markov-switching GARCH models have been widely used to account for distinct volatility clustering patterns under different economic conditions. It is challenging to estimate this class of models because of path dependence on regimes. An approximate maximum likelihood estimator (MLE) is proposed based on truncated regime paths. The upper bounds of the approximation errors are derived in likelihood computation and latent variable inference, investigate the asymptotic properties of the approximate MLE, and examine the performance of the approximate MLE through simulation experiments and an empirical application.
