考虑收益非正态性的资产配置模型及应用

An Asset Allocation Model Incorporating Market Return Non-normality and Its Applications

为更准确评估尾部风险,将资产收益的非正态性引入资产配置过程中,使用"反平滑"、半参数广义帕累托分布方法和连接函数等统计技术,获取更准确的资产收益分布,并以条件在险价值为风险测度,建立了一个考虑收益非正态性的资产配置框架。基于2008年1月～2017年12月间股票、债券、另类投资等6类资产的月度收益数据,结合一个养老基金的案例,验证了在资产配置决策时考虑非正态性的必要性,并展示如何将该模型应用于资产配置实践中。结果表明:忽略非正态性将会显著低估组合的下行风险,而考虑非正态性则有利于降低风险,改善风险收益比。所用方法和所得结论对于养老基金等长期投资者的资产配置和风险管理具有现实意义。

To evaluate tail risk more precisely,this paper introduces the non-normality of asset returns into asset allocation process.By employing a series of statistical techniques,including"anti-smoothing",semi-parametric General Pareto Distribution method and copulas to model non-normally distributed asset returns,using conditional value-at-risk as risk measure,the paper establishes an asset allocation decision framework that considers non-normality.Based on monthly market return series of six asset classes,including equity,bonds and alternative investments,from January 2008 to December 2017,it also verifies the necessity of and how incorporating non-normality into asset allocation decision-making process,by analyzing an empirical case of pension fund.The results indicate that ignoring non-normality would significantly underestimate the downside risk of the portfolio,while incorporating non-normality may help to reduce the portfolio’s volatility and improve its risk-return profile.The methods and conclusions are of great significance for asset allocation and risk assessment of long-time horizon investors such as pension funds.