Number of regimes of the Volatility in Muscat Security Market Index in Oman using the Markov Switching GARCH models

Different GARCH-specifications are by far the most popular model based approach used for volatility study. This paper evaluates the forecast accuracy of some specific GARCH-models; Markov Switching-GARCH (MS-GARCH). First, we investigate which adequate number of regimes in MS-GARCH can describe the corresponding the conditional variance for Muscat Security Market index (MSM). For this, We compared several MS-GARCH models using the Akaike Information Criterion (AIC), provided by Maximum Likelihood Estimator (MLE). The results show clearly that the selected models have at least two regimes whatever the GARCH specification is. A predictive test using the Value At Risk (VAR) confirms that the selected process provide accurate volatility forecasts.