A NON-LINER REGIME SWITCHING MODELS IN FINANCIAL SERIES WITH TWO REGIMES

A NON-LINER REGIME SWITCHING MODELS IN FINANCIAL SERIES WITH TWO REGIMES

ABSTRACT

In this study, two economic series which have changes in regimes were considered. Models considered for the two series are Simple Switching Mixture (SSM) model and Markov Switching Autoregressive (MS-AR) model.  Predictions of future transition regime probabilities were performed using the Hamilton filter of m-period transition matrix for MS-AR model, while, the two state ergodic m-step ahead transitions probabilities for SSM model. Subsequently, forecast evaluation measures for the two models were carried out with Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE).  Consumer Price Index (CPI), had a better forecast with SSM model while, Nominal Effective Exchange Rate (NEER) had a better forecast with the MS-AR model.