带多个跳跃因子GARCH模型对中国股市捕捉能力研究

发布时间：2018-01-10 17:07

本文关键词：带多个跳跃因子GARCH模型对中国股市捕捉能力研究　出处：《南京财经大学》2013年硕士论文　论文类型：学位论文

更多相关文章： GARCH-JUMP HongLi 多个跳跃因子 VaR

【摘要】：金融风险的主要来源是金融资产价格的波动，即金融资产收益率的不确定性，而有关金融资产的波动性研究则一直是国内外学者研究金融风险问题的重点。近年来，受新一轮国际金融危机的影响，金融风险变得越来越复杂，这便推动国内外学者提出更优秀的模型来完美刻画实际金融市场价格波动的动态变化趋势。基于此，文章结合国内外金融时间序列的最新成果，提出一种新型的时间序列模型来捕捉中国股票市场的收益率与波动率变化特征。本文将双指数跳跃因子引入传统的GARCH族模型，从而利用双指数跳跃因子来拟合中国股票市场收益率序列呈现的非对称性和波动杠杆效应等特征，实证结果显示，引入双指数跳跃因子的GARCH模型与一般GARCH-JUMP模型相比，能够更好的拟合中国股票市场收益率与波动率的动态变化过程，此外模型的HongLi检验结果也表明前者的模型设定更准确。但是，HongLi检验结果表明带双指数跳跃因子的GARCH模型仍然没有通过检验，所以我们在新模型的基础上引入方差跳跃因子，从而提出带多个跳跃因子GARCH模型。实证结果表明，新模型能够更好的拟合中国股票市场收益率与波动率变化过程中存在的尖峰厚尾性、波动聚集性、波动的杠杆效应以及非对称性等特征；模型的HongLi检验结果也表明带多个跳跃因子GARCH模型要比带双指数跳跃因子GARCH模型更准确，能够更好的刻画中国股票市场收益率与波动率的动态变化趋势。本文还基于两类新的模型对中国股市作了VaR风险度量分析，结果显示带多个跳跃因子的GARCH模型预测的VaR值的准确性要高于带双指数跳跃因子GARCH模型的预测结果，前者的失败率更接近于我们选取的置信水平所对应的失败率，更能够准确的反映两类综合指数的风险情况。我们认为双指数跳跃因子GARCH模型预测VaR没有通过检验的原因可能是其一定程度上高估了风险，从而导致其预测VaR的预测区间更广，，预测误差更大，进而一定程度上减少了预测的失败率。
[Abstract]:The main source of financial risk is the fluctuation of financial asset price, that is, the uncertainty of financial asset yield. The volatility of financial assets has been the focus of domestic and foreign scholars on financial risk. In recent years, due to the impact of a new round of international financial crisis, financial risk has become more and more complex. This prompted scholars at home and abroad to put forward a better model to describe the dynamic trends of the real financial market price volatility. Based on this, the article combines the latest results of domestic and foreign financial time series. A new time series model is proposed to capture the characteristics of volatility and yield in Chinese stock market. In this paper, the double index jump factor is introduced into the traditional GARCH family model, and the double index jump factor is used to fit the characteristics of asymmetric and volatility leverage effect in the return series of Chinese stock market. The empirical results show that the GARCH model with double index jump factor can better fit the dynamic process of the return and volatility of Chinese stock market compared with the general GARCH-JUMP model. In addition, the HongLi test results of the model also show that the model setting of the former is more accurate, but the results of the HongLi test show that the GARCH model with double exponential jump factor has not passed the test. So we introduce the variance jump factor on the basis of the new model, and then we put forward the GARCH model with multiple jump factors. The empirical results show that. The new model can better fit the characteristics of the Chinese stock market in the process of the change of return and volatility, such as peak and thick tail, volatility aggregation, the leverage effect of volatility and asymmetry. The results of HongLi test also show that the GARCH model with multiple jump factors is more accurate than the GARCH model with double exponential jump factors. It can better depict the dynamic trend of the return and volatility in Chinese stock market. This paper also makes the VaR risk measurement analysis of Chinese stock market based on two new models. The results show that the accuracy of VaR predicted by GARCH model with multiple jump factors is higher than that of GARCH model with double exponential jump factors. The failure rate of the former is closer to the failure rate corresponding to the confidence level we selected. We think that the double index jump factor GARCH model can not pass the test of VaR may be because it overestimates the risk to a certain extent. Therefore, the prediction range of VaR is wider, the prediction error is larger, and the failure rate of prediction is reduced to a certain extent.
【学位授予单位】：南京财经大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F832.51;F224

【参考文献】