Abstract: Against the background of high price fluctuations in today's economic environment, it is significant to analyze the deep reasons for the portfolio volatility by measuring the risk contribution of each asset. Historical data measure method has been widely used in risk contribution which is mainly suitable for the situation of larger data sample and short duration up to now. Generally, the risk contribution in extreme situation is decided by the feature of the tail data. Therefore, it is difficult to make sure of the accuracy of estimation in this case. In this article, we optimize the saddle point approximation to solve the above problem, and make an empirical analysis of these methods by using Chinese stock data. Compared with the traditional methods, the advantages of the saddle point approximation are listed as follows: the distribution function of the portfolio is more concise and the calculation has a higher accuracy and efficiency. Moreover, the stress test results show that the saddle point approximation model has more stability. Therefore, it is likely that the saddle point approximation will be helpful for early warning of risk and the risk prevention for investment portfolio.

Key words: risk Contribution, value at risk, saddle point approximation

摘要： 在当今金融市场资产价格高波动的背景下,度量投资组合中各资产对总体风险的风险贡献度对探析投资组合风险波动不定的深层次原因有重要意义。关于风险贡献度的测算,目前运用较广泛的是历史数据法,其主要适用于存在大量数据样本且持续期较短的情况。特别地,极端情况下的风险贡献度估计主要由处于分布尾部的少量观测值决定,因此历史数据法估计的准确性此时较难保证,为此,本文对鞍点逼近模型优化并考察上述情形。通过对中国股市进行实证分析发现,与传统历史数据法相比,鞍点逼近模型呈现下列优点:投资组合分布函数简洁、风险贡献度计算效率和准确性较高,压力测试表明该方法具有较好的稳健性。因此该方法有望对投资组合的风险预警与防范起到决策支持作用。

关键词: 风险贡献度, 风险价值, 鞍点逼近