Copula方法及函数估计在企业风险分析中的应用

Enterprise Risk Analysis Using Copula and Function Estimation

风险分析是风险管理的重要内容,核心任务是预测主要指标的概率分布。通过新兴的Copula方法拟合影响指标的概率分布,采用随机梯度Boosting、支持向量机、自适应样条等函数估计方法,得到主要指标同影响指标之间的函数关系。利用蒙特卡罗方法预测未来时间主要指标的概率分布,在此基础上计算风险度量——方差、VaR和条件风险值。结合机械企业的实证分析,结果表明,该方法在风险分析中具有较好的可行性和有效性。

Risk analysis is the core of risk management. Its objective is to predict the probability distribution of the main index. The probability distribution of influential indexes is fitted by Copula. And the function between the main index and the influential indexes is estimated by such function estimation methods as stochastic gradient Boosting, support vector machine, and multivariate adaptive regression splines. The prdbability distribution of the main index in the future is predicted by Monte Carlo approach based on previous results. Risk measurement indexes, such as variance, VaR and conditional VaR, are calculated from predicted distribution. The example in a mechanical enterprise verifies the validity and effectiveness of the proposed method.