Dynamic correlation and equicorrelation analysis of global financial turmoil: evidence from emerging East Asian stock markets
Dynamic correlation and equicorrelation analysis of global financial turmoil: evidence from emerging East Asian stock markets
复制标题
DOI：
10.1080/00036846.2016.1145349
复制
发表时间：
2016-08-01
影响因子：
2.2
通讯作者：
Hamori, Shigeyuki
中科院分区：
文献类型：
--
作者：
Cai, Xiao Jing;Tian, Shuairu;Hamori, Shigeyuki
本文考察了东亚八个新兴股市与美国股市的动态条件相关性，并分析了这九个股市之间的动态等相关关系。我们发现，在全球金融危机的第一阶段，条件相关性和等相关性显着增加。我们将这一发现称为从美国股市向新兴东亚市场的传染。我们还发现，在全球金融危机的第二阶段，还存在一个增加相关性和等相关性(羊群效应)的显着过程。此外，我们使用了两个新的模型，即DCCX-MGARCH(一个带有外生变量的DCC多变量GARCH模型)和DECOX-MGARCH(一个带有外生变量的动态等相关多变量GARCH模型)来识别传染的渠道。我们发现，VIX指数的增加增加了条件相关性和等相关性，而TED利差的增加降低了六个新兴东亚国家与美国的条件相关性。我们通过构造损失函数来比较DCC和DCCX模型(或DECO和DECOX模型)的条件相关估计的精度。我们发现，通过从外生变量中提取额外信息，DCCX(DECOX)模型比DCC(DECO)模型提供了更准确的条件相关估计。
This study investigates the dynamic conditional correlations (DCCs) between eight emerging East Asian stock markets and the US stock market and analyses the dynamic equicorrelation among these nine stock markets. We find a significant increase in the conditional correlations and equicorrelation in the first phase of the global financial crisis. We refer to this finding as contagion from the US stock market to the emerging East Asian markets. We also find an additional significant process of increasing correlations and equicorrelation (herding) in the second phase of the global financial crisis. Further, we employ two new models, namely DCCX-MGARCH (a DCC Multivariate GARCH model with exogenous variables) and DECOX-MGARCH (a dynamic equicorrelation multivariate GARCH model with exogenous variables), to identify the channels of contagion. We find that an increase in the VIX Index increases the conditional correlations and equicorrelation, while increases in TED spreads decrease the conditional correlations of six emerging East Asian countries with the USA. We compare the accuracy of the conditional correlation estimates of the DCC and DCCX models (or DECO and DECOX models) by constructing a loss function. We find that the DCCX (DECOX) model provides more accurate conditional correlation estimates than the DCC (DECO) model by extracting additional information from exogenous variables.