基于藤Copula方法的巨灾风险条件VaR预测
Catastrophe Risk Conditional VaR Forecasts Based on Vine Copula Method
为准确预测巨灾风险的条件VaR, 应用藤Copula方法刻画巨灾损失变量间的相关结构, 进而得到损失变量间的联合分布和 条件分布函数, 最终实现对条件VaR的估计. 对全球洪水的损失数据进行实证分析, 利用核密度估计检验法从常用多元Copula中选出 最优的Copula作为比较对象, 回测检验结果表明:准确刻画相关结构是精确估计条件VaR的关键, 藤Copula方法对巨灾风险条件VaR的 预测能力要优于常用多元Copula方法.
Aiming at enhancing catastrophe risk conditional VaR forecasting performance, this paper captures the variate dependency structure of catastrophe losses by vine Copula. Also the joint and conditional distributions could be estimated, and finally the conditional VaR is thus estimated. Selecting global flood losses data to carry out an empirical analysis and choosing the best Copula type as a compare object from traditional multivariate Copula by kernel density estimated test, the backtesting results show that the accurate characterization of the correlation structure is the key to accurately estimate conditional VaR and vine Copula approach is superior to the traditional multivariate Copula in catastrophe risk conditional VaR forecasting.
藤Copula方法 / 巨灾风险 / 条件VaR / 核密度估计检验 / 回测检验. {{custom_keyword}} /
/
〈 | 〉 |