Regime-Switching Models(体制转换模型).pdfVIP

Regime-Switching Models －Foundation and Application Fang Long Content • Background and Evolution • Markov Switching Models • Model Estimation • Hypothesis Testing • Application: A Study of Taiwan’s Business Cycles Background and Evolution • Linear time series models e.g. AR(p)/MA(q)/ARMA(p, q)/ARIMA(p, q) Limitation: not applied to nonlinear dynamic patterns such as asymmetry/amplitude dependence /volatility clustering. • Nonlinear time series models e.g. ARCH(p)/GARCH(p, q)/TAR/STR Limitation: ①nonlinear optimization algorithms may be inefficient for parameter estimation, e.g. local optimum. ②inflexible and usually applied to certain nonlinear patterns of data. ③arbitrary identification of variables or parameters, e.g. threshold variable/threshold values Background and Evolution • Regime-Switching Models ①i.e. Markov-Switching Models, nonlinear time series models. ②A novel feature: switching mechanism controlled by an unobservable state variable which follows a first-order Markov chain. ③Suitable for describing correlated data which exhibit distinct dynamic patterns during different time periods. • Evolution ①Quandt (1958). The estimation of the parameters of a linear regression system obeying two separate regimes. ②Goldfeld and Quandt (1973). A Markov model for switching regressions. ③Hamilton (1989). A new approach to the economic analysis of nonstationary time series and the business cycle. ④Hamilton and Susmel (1994). Autoregressive conditional heteroscedasticity and changes in regime. ⑤Krolzig (1997). Markov Switching Vector Autoregressions: Modelling, Statistical Inference and Application to Business Cycle Analysis. Markov Switching Models  A Simple Model • A Markov switching model is