基于Copula的VsR度量与事后检验

估计VaR的传统方法有三种:协方差矩阵法、历史模拟法和蒙特仁洛模拟法。通常,文献中认为刚蒙特卡洛模拟法度量VaR有很多方面的优点。但是,本文通过实证检验发现,使用传统蒙特卡洛模拟法估计的VaR偏小,事后检验效果很不理想。本文引入Copula函数来改进传统的蒙特卡洛模拟法。Copula函数能将单个边际分布和多元联合分布联系起来,能处理非正态的边际分布,并且它度量的相关性不再局限于线性相关性。实证检验表明,基于Copula的蒙特卡罗模拟法可以更加准确地度量资产组合的VaR。

VaR Measurement and Backtesting Based on Copula

There are three traditional methods of estimating VaR: Covariance matrix,History simulation and Monte Carlo simulation.Generally,many advantages of Monte Carlo simulation are introduced in literature.But our Empirical test shows that VaRs estimated by traditional Monte Carlo simulation are small and the relevant backward tests are not good.In this paper,we use Copula function to modify the traditional Monte Carlo simulation.Copula function can link between marginal distribution and joint probability distribution.It not only can deal with non-Gauss marginal distribution,but also can describe non-linear correlation.Empirical test shows that VaR of an asset portfolio estimated by the new Monte Carlo simulation based on Copula is more accurate.