石油期货价格的收益率及波动率的长记忆性研究

金融市场的长记忆性问题研究一直吸引着众多研究者的目光,但是期货作为一种金融衍生工具,其长记忆性的研究相对于股票来说却很少。国内外大多数的研究都是针对股票市场,运用的主要方法是R/S分析,修正的R/S分析法,GPH分析法等,大部分都是用ARFIMA模型对长记忆时间序列进行分析。本文介绍了时间序列的一些模型,介绍了单位根检验方法,采用样本自相关函数法,传统的R/S分析法,GPH分析法,运用MATLAB,EVIEWS,Oxmetirc4环境下的G@ARCH42软件包等编程,检验了美国原油期货(NYMEX)市场样本数据的长期记忆性,并得出如下结论:石油期货收益率序列的长记忆性不明显,而波动率的替代指标绝对收益率序列呈现较强的长期记忆性。同时本文通过比较描述收益率序列以及绝对收益率序列模型的优劣性,发现收益率序列可用GARCH(1,1)模型来描述,而FIGARCH(1,d,1)模型相对GARCH(1,1)模型来说,能更好的描述波动率的替代指标——绝对收益率序列。

Analysis on Long-term Memory in Petroleum Futures Price Returns and Volatilities

Long-term memory characteristic is a main subject of financial study.Compared to stock market,long memory researches on futures is less.Lots of researches applied rescaled range(R/S)analysis,revised rescaled range and GPH methods to testing the long-term memory in stock market.And the majority of the researches use ARFIMA model to analyze the time series.This paper introduced different kinds of time series models and unit root method to test the stability of time series.With rescaled range(R/S)analysis and GPH methods,we test the long-term memory of the crude petroleum futures in NYMEX market using EVIEWS,MATLAB,OXMETRICS applied software.We conclude that petroleum futures market's return series doesn't exhibit long-term memory or persistence,but the substitute index(absolute return series)of the volatilities exhibit strong long-term memory.And at the same time,we compared the models which will be more appropriate to describe the return series and volatilities,and we can use GARCH(1,1)model to describe the return series.However,A fractional integration model,FIGARCH(1,d,1)performs significantly better than a traditional volatility model,GARCH(1,1),in modeling the substitute index(absolute return series)of petroleum futures price volatility.