【摘要】 在基于pair-Copula高维建模方法的藤Copula理论框架下,构建了藤结构Copula-POT模型,并采用C藤和D藤结构分解下的Gaussian Copula和Frank Copula函数来研究外汇资产间的尾部相依结构.该模型考虑了单个资产的尾部特征,且克服了传统Copula的"维数灾难"问题,能更好地描述资产尾部间的相依结构.基于四种外汇资产(美元、欧元、日元和港币)的实证结果表明D藤能更好的对外汇资产间尾部相依结构进行描述,且D藤分解模式下的Frank Copula能更准确反映外汇资产尾部间的相依性,并针对外汇储备投资,给出相应的建议.更多还原

【Abstract】 Under the vine Copula theory framework based on high dimension modeling method of pair- Copulas,the vine Copula- POT model was built. Gaussian and frank Copulas based on Canonical vine( C- vine) and D- vine structure decomposition methods were used to study the tail dependence structure among foreign exchange assets. The model not only considered the tail characteristics of the individual asset,but also overcome the "dimension disaster"caused by traditional Copulas. The empirical results based on foreign exchange assets showed that the tail dependence structure could be described more accurately by D-vine,especially by the frank Copula based on D- vine structure decomposition method. At last,some suggestions are presented for the foreign exchange investment.更多还原