【摘要】 区别于目前基于信号分析的货币危机预警模型,本文采用时变概率马尔可夫状态转换模型来构建货币危机的预警模型。相对于信号模型中主观设定危机的定义和阈值,该模型将危机的识别内生于模型估计中,并通过汇率剧烈波动期的发生概率对货币危机预警,使预警系统更客观。本文的主要贡献为在Bauwens等(2007)基础上改进了Griddy-Gibbs取样法的使用效率,并应用此MCMC方法估计了多个马尔可夫转换模型。对东南亚金融危机的研究证实了状态转换模型的预警能力,并且时变概率马尔可夫转换-GARCH模型揭示了关于汇率波动的更多特性。更多还原

【Abstract】 Instead of the normal Signal Models,this paper constructs early warning models of currency crisis via time varying Markov Switching models.Comparing to the subjective definition of crisis and threshold setting in Signal Models,Markov Switching models endogenize crisis identification in model estimation,thus are more objective.This paper improves Griddy-Gibbs Sampling Method of Bauwens etc(2007),and estimates Markov Switching models using this method.The result shows that time varying Markov Switching model is able to reveal crises signs successfully.Moreover,MS-GARCH models reveal more interesting features about exchange volatility.更多还原