基于GARCH-POT-VaR模型的社保基金投资尾部风险测度研究  

Research on Tail Risk Measurement of Social Security Fund Investment Based on GARCH-POT-VaR Model

作　　者：陈国栋 王家琪 CHEN Guodong;WANG Jiaqi(School of Management and Economics,North China University of Water Resources and Electric Power,Zhengzhou 450046,China)

机构地区：[1]华北水利水电大学管理与经济学院,郑州450046

出　　处：《中国证券期货》2024年第6期32-42,共11页Securities & Futures of China

基　　金：河南省哲学社会科学规划项目“熵视阈下河南省城镇职工基本养老保险基金的投资策略及风险控制研究”(2021BJJ064)。

摘　　要：社保基金作为解决我国老龄化问题的重要保障基金,近五年投资收益波动较大,如何准确度量社保基金投资尾部风险是提高社保基金投资安全性的重要问题。在考虑到收益率序列波动特征的基础上,本文提出以GARCH族模型刻画收益率序列波动性特征,POT模型处理极端尾部数据,构建三种金融市场尾部风险度量模型:ARMA-GARCH-POT、ARMA-EGARCH-POT和ARMA-GJRGARCH-POT,将其应用于风险价值VaR的动态测度。在极值POT模型构建时采用AU2统计量确定阈值,W2和A2统计量进行尾部拟合优度检验,避免了主观性。最后对VaR进行测度及回测,结果表明:传统GARCH-VaR模型会低估极端尾部风险,结合POT模型的GARCH类模型对动态VaR的测度效果更为准确,且各模型在99.0%置信水平下能够更加准确地量化股市收益率尾部风险。Social security fund is an important security fund to solve the aging problem of our country,the investment income fluctuates greatly in the past five years,how to accurately measure the investment risk of social security fund is an important issue to improve the investment security of social security fund.On the basis of considering the volatility characteristics of the return series,the GARCH family model is proposed to depict the volatility characteristics of the return series and the POT model to process extreme tail data,and three tail risk measurement models of financial markets are constructed:ARMA-GARCH-POT,ARMA-EGARCH-POT and ARMA-GJRGARCH-POT are applied to the dynamic measurement of VaR.In the construction of extreme POT model,AU 2 statistics were used to determine the threshold,and W 2 and A 2 statistics were used to test the goodness of tail fitting,which avoided subjectivity.Finally,VaR is measured and back-tested.The results show that the traditional GARCH-VaR model will underestimate the extreme tail risk,and the GARch-type model combined with POT model has a more accurate measurement effect on dynamic VaR,and each model can more accurately quantify the tail risk of stock market return at the 99.0%confidence level.

关 键 词：社保基金 尾部风险测度 ARMA-GARCH族模型 极值理论 VaR 

分 类 号：F842.61[经济管理—保险] D632.1[政治法律—政治学] O212[政治法律—中外政治制度]