基于长记忆的中国期货市场实证研究  

The Empirical Analysing for China Futures Markets Based on Long Memory

作　　者：杨桂元[1] 刘坤[1] 

机构地区：[1]安徽财经大学数量经济研究所,蚌埠233030

出　　处：《价值工程》2009年第8期152-157,共6页Value Engineering

基　　金：教育部人文社会科学研究资助项目(08JA630003);安徽财经大学研究生科研创新基金资助项目(ACYC2008043)

摘　　要：通过某期货公司研发部编制的期货指数,基于长记忆研究方法,运用经典的R/S分析、修正的R/S分析,分别建立了研究长记忆的ARFIMA模型、FIGARCH模型和ARFLMA-FIGARCH模型。并运用这些模型对我国上海期货交易所的铜、铝、大连期货交易所的大豆、玉米、豆粕以及郑州期货交易所的小麦的收益率序列进行相关研究和分析,得出它们的收益率序列以及收益率波动序列均存在长记忆性,且ARFLMA(0,d1,0)-FIGARCH(1,d2,0)模型的预测效果较好。This paper introduce this method by index futures which is draw up by some futures company. To use of the classic R/S method,revision R/S method and build up the long memory method ARFIMA model,FIGARCH model,ARFIMA-FIGARCH model, at the same time, the authors make some researches on China futures markets copper, aluminum, soybean, wheat, corn yield sequences related researches and analysis through these methods. As a result ,they are all being long memory and find that the models make more precise in forecast.

关 键 词：期货 长记忆 ARFLMA模型 FIGARCH模型 ARFIMA-FIGARCH模型 

分 类 号：F713.35[经济管理—产业经济] F019.3