基于ARIMA-GARCH族模型对余额宝收益率特征的实证研究

发布时间：2018-05-11 07:46

本文选题：余额宝 + ARIMA-GARCH族模型　；参考：《青岛大学》2015年硕士论文

【摘要】：众所周知,基于互联网支撑的第三方支付平台的“余额宝”深受社会各界关注,对传统商业银行有较大冲击。作为金融市场的一个方面,其收益率具有一定的波动性。文章就余额宝收益率的波动性及预测进行了探讨。以余额宝2013年6月3日到2015年4月10日期间的七日年化收益率为研究对象,取ARIMA模型建立均值方程,并利用GARCH)1,1(模型、GJR-GARCH)1,1(模型、EGARCH)1,1(模型和APARCH)1,1(模型建立方差方程来进行数据分析。结果表示,余额宝序列是非平稳的,具有自相关性,并且其一阶差分序列存在尖峰厚尾现象、波动集聚性以及条件异方差性。继而,分别对基于波动性特征建立的ARIMA-GARCH族混合模型参数和预测评价指标进行比较,发现ARIMA-GJR-GARCH混合模型拟合效果更好。最后,通过余额宝七日年化收益率序列与万份收益序列的协整、格兰杰检验、及脉冲响应图发现,两个序列之间存在协整联系,且互为因果关系,七日年化收益率对万份收益始终处于明显的正向冲击状态。
[Abstract]:As we all know, "Yu'e Bao", which is based on the third party payment platform supported by the Internet, is deeply concerned by all walks of life and has a great impact on the traditional commercial banks. As one aspect of financial market, its yield has certain volatility. This paper discusses the volatility and forecast of Yu'e Bao's yield. Taking the seven-day annual rate of return of Yu'e Bao from June 3, 2013 to April 10, 2015 as the research object, taking the ARIMA model to establish the mean equation, and using the model GJR-GARCH1 (model GJR-GARCH1) to analyze the data. The results show that the Yu'e Bao sequence is non-stationary and autocorrelation, and its first-order differential sequence has the phenomena of peak and thick tail, fluctuation agglomeration and conditional heteroscedasticity. Then, the parameters of ARIMA-GARCH family mixed model based on volatility characteristics and the prediction evaluation index are compared, and it is found that the ARIMA-GJR-GARCH hybrid model has better fitting effect. Finally, through the co-integration, Granger test, and pulse response diagram of Yu'e Bao's seven-day annualized yield series and 10,000 profit series, it is found that there is a cointegration relationship between the two sequences, and they are causality with each other. Seven-day annualized rate of return on 10,000 returns has always been in an obvious positive impact state.
【学位授予单位】：青岛大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F224;F832.2

【参考文献】