隐Markov链驱动关联性和波动性的传染分析

Contagion Analysis Based on Correlations and Volatilities Driven by Hidden Markov Chain

本文将隐Markov链对波动性和相关性的驱动分析引入DCC多元GARCH,对波动和相关分析建立起了直接的联系,进而考察次贷危机、欧洲债务危机在主要证券市场间的传染性。研究发现,高波动高相关机制为联动性提供了一种直接的表述方式,且这一机制在危机期间处于支配地位;次贷危机、欧洲债务危机具有传染性,传染期以区间的形式出现,且危机初期的市场在各机制间有较为频繁的转换,不可根据危机事件对样本进行武断地分割;同时,危机的传染在所考察的市场之间具有系统性,应对危机需要各国政策间的协调配合;另外,有证据显示美国次贷市场在2006年年中已显现出问题,有关国家贻误了深入分析和应对危机的时机。

A hidden Markov chain is introduced to drive both volatilities and correlations into dynamic conditional correlation multivariate GARCH model,which can put direct analysis to volatilities and correlations under one framework.Then the contagion resulted from American subprime mortgage crisis and European sovereign debt crisis among the major stock markets is in restigated.The findings are firstly,the regime of high volatility with high correlation,which dominates the market during crises,provides a direct way of expression to the concept of comovement.Second,the American subprime mortgage crisis and European sovereign debt crisis are contagious and emerge in the form of intervals,and show the market transfer between different regimes more frequent in the early stage of these crises.So it is arbitrary to investigate crisis contagion based on dividing sample into subsamples according to prior breakpoints.Third,it is necessary for countries to cooperate with each other because the contagion resulted from American subprime mortgage crisis and European sovereign debt crisis both are a systemic risk among the investigated markets.Finally,there is evidence that American subprime mortgage market encounters dilemma in mid-2006 and the countries involved miss the time to analyze and response to the crisis.