基于分位数回归VaR模型的股票风险实证分析

Research on Stock Risk Based on Quantile Regression VaR Model

摘要: 在险价值VaR (Value at Risk)是最近发展起来并被广泛应用的一种衡量股票风险的方法。本文收集了约两年(2016年1月~2017年12月)来自主板市场，中小板市场，创业板市场的9只股票的收益率数据，运用t-GARCH(1,1)模型和Quantile-ARCH(1)模型两种方法计算了9只股票的VaR值。并根据似然比检验和失败率检验方法得出：基于Quantile-ARCH(1)模型计算出的VaR更加精确。

Abstract: VaR (VaR at Risk) is widely used to measure stock risk in recent years. In this paper, based on the historical data of nearly recent two years (Jan. 2016-Sept. 2017) of 9 stocks from the main board market, the small and medium board market and the growth enterprise board market, we compute VaR of the 9 stocks by using t-GARCH(1,1) model and Quantile-ARCH(1) model. According to the likelihood ratio test and the failure rate test, it is concluded that the Quantile-ARCH(1) model is more accurate.