基于高频数据已实现GARCH-HAR模型的研究

发布时间：2018-01-06 22:37

本文关键词：基于高频数据已实现GARCH-HAR模型的研究　出处：《厦门大学》2014年硕士论文　论文类型：学位论文

【摘要】：鉴于波动率的不可观测性,一直以来波动率的估计都是金融市场的重要课题,不管是在金融产品定价、资产配置,亦或是风险管理方面,波动率估计都扮演着举足轻重的角色。随着计算机等相关技术的发展,金融高频数据越来越荣易获得,传统的GARCH族模型已不能满足高频数据的研究需要,而以通过加总高频收益率得到的已实现波动率为代表的波动率代理变量,为市场的波动提供了一个可靠的度量,因此如何将这二者的优势进行整合从而更好地预测市场波动具有重要的现实意义。本文选取上证综合指数的5分钟高频收益序列,计算出已实现波动率,并在现有文献的基础上将其分解成连续和跳跃成分以及已实现半方差的形式,然后通过GARCH模型和HAR模型对收益率序列、已实现波动率序列进行联合建模,得到一个具有混频性质的Re-GARCH-HAR模型。实证结果表明,该模型的预测效果比传统的GARCH模型要好,同时在对波动率进行中长期预测时,模型的表现也优于Realized GARCH模型,在探究该模型的具体形式时,我们发现：仅仅利用负的收益率计算出已实现半方差,并将其加入GARCH模型所构建的Re-GARCH-HAR-RS模型的预测具有最好的预测效果。
[Abstract]:Given the unobservability of volatility, volatility estimates have long been an important issue in financial markets, whether in terms of financial product pricing, asset allocation, or risk management. Volatility estimation plays an important role. With the development of computer and other related technologies, high-frequency financial data are more and more readily available. The traditional GARCH family model can not meet the research needs of high frequency data, but the volatility proxy variable represented by the realized volatility obtained by adding the total high frequency return rate. It provides a reliable measure for the volatility of the market, so how to integrate the advantages of the two to better predict the volatility of the market has important practical significance. This paper selects the 5-minute high-frequency return sequence of Shanghai Composite Index, calculates the realized volatility, and decomposes it into the forms of continuous and jump components and realized semi-variance based on the existing literature. Then the return series is modeled by GARCH model and HAR model, and the realized volatility series is modeled jointly. An Re-GARCH-HAR model with mixing property is obtained. The empirical results show that the prediction effect of the model is better than that of the traditional GARCH model. At the same time, the performance of the model is better than that of the Realized GARCH model when the volatility is forecasted in the medium and long term. We found that:. Only using the negative rate of return to calculate the realized semi-variance and adding it to the Re-GARCH-HAR-RS model constructed by the GARCH model has the best prediction effect.
【学位授予单位】：厦门大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F224;F830.9

【参考文献】