基于藤Copula分组模型的金融市场风险度量研究

传统未分组的藤Copula模型可用于刻画金融资产间的相依性，但其存在将所有不同行业资产视为一个整体的问题。本文在充分考虑金融市场中各机构所属行业不同的基础上，提出了藤Copula分组模型，给出了该模型算法的具体步骤，并证明了算法的收敛性。最后通过返回检验方法，对比研究了藤Copula分组模型和未分组的藤Copula模型对银行业、证券业和保险业间VAR估计的精度差异，结果表明藤Copula分组模型的预测效果更准确且更有效。

A Study on Risk Measurement of Financial Market Based on Vine Copula Grouped Model

The traditional Vine Copula model can be used to depict the dependence among financial assets, but there is an issue, i.e., taking the assets from different industries as a whole. Taking the different industries of financial institutions in the financial market, this paper proposes a grouped Vine Copula model, presents concrete algorithmic steps for it and proves the convergence of the algorithm. Finally, by means of the backtesting analysis, a study and comparison is done on the precision of VaR estimation among banking, securities and insurance industries using the grouped and the traditional ungrouped Vine Copula models. The result shows that the grouped one predicts more accurate and efficient outcomes.