基于ARMA-GARCH模型的股指波动压力测试情景设计研究  

A Study of the Design of Stock Index Fluctuation Stress Testing Scenario:Based on ARMA-GARCH Model

作　　者：侯外林[1] 

机构地区：[1]中南大学商学院,湖南长沙410083

出　　处：《当代财经》2011年第11期71-79,共9页Contemporary Finance and Economics

摘　　要：针对股指收益率时间序列某期间的异方差、尖峰厚尾以及序列自相关等特性,将ARMA模型与GARCH模型相结合,回归建模测算相关股指年度收益率VaR值,可以有效预测类似市场条件下股指的波动以及相伴概率。因此,在证券公司压力测试实践中,基于相伴概率合理设计股指下跌的压力测试情景,可以进一步提高压力测试情景设计的科学性,增强压力测试结果的现实指导意义。同时,可以将本文研究思路推广应用于利率、汇率、市场交易量等历史数据较充分的金融时间序列的实证分析,借以指导债市波动、汇市波动以及市场交易量波动等压力测试情景的设计工作。Aiming at such characteristics as the heteroscedasticity, sharp peak and thick tail, serial autocorrelation and so on during some time series of stock index returns, this paper combines the ARMA model with GARCH model to set up a regression model, so as to estimate the VaR value of the annual return rate of the related stock index, which can effectively predict the stock volatility and the accompanying probability under the similar market conditions. Therefore, during the stress testing practices of the securities companies, reasonably designed stress testing scenarios of downward stock index based on accompanied probability can further improve the scientificalness of the design of stress testing scenarios and enhance the practical guidance of the stress testing results. At the same time, the idea of this paper can be applied to the empirical analysis of financial time series with abundant historical data of interest rates, exchange rates, market trading volume etc., so as to guide the design of stress test scenarios, such as bond market fluctuations, foreign exchange fluctuations and fluctuations in market trading volumes.

关 键 词：压力测试 时间序列模型 VAR理论 ARMA-GARCH模型 

分 类 号：F830.91[经济管理—金融学]