证券市场跳跃现象非参数辨识方法比较研究  

Comparison of Jump Phenomenon with Nonparametric Identification Method in Stock Market

作　　者：王春峰[1] 郑仲民[1] 房振明[1] 

机构地区：[1]天津大学管理学院金融工程研究中心,天津300072

出　　处：《天津大学学报（社会科学版）》2011年第6期498-504,共7页Journal of Tianjin University：Social Sciences

基　　金：国家自然科学基金资助项目(70771076);国家杰出青年科学基金资助项目(70225002)

摘　　要：基于订单冲击效应的无限活跃跳跃是一种近期才引起学术界关注的价格行为,与信息到达引起的有限活跃跳跃一起被称为广义资产价格跳跃。文章从理论上对BNS方法与阈值多幂次变差方法(TMPV)对广义跳跃行为的辨识进行了深入的比较研究,得到结论BNS方法只适用于在能够有效鲁棒噪音的相对较低的高频数据段上对事件触发的有限活跃跳跃行为进行识别;而阈值多幂次变差方法在资产价格非连续成分框架下,对不同类型噪音、有限活跃跳跃、无限活跃跳跃行为的统计特性进行了深入挖掘,进而构造统计量对不同类型跳跃行为的存在性进行有效辨识。基于蒙特卡洛技术的模拟结果也验证了理论探讨的结论。Infinite activity jump based on the order impact is a type of price behavior,which causes scholar＇ s concern recently. Infinite activity jump and finite activity jump based on the information impact are called generalized asset price jump. This paper compares the BNS and threshold multi power variation（TMPV） methods in theory. The result shows that compared to BNS which can distinguish the finite activity jump at relatively lower frequencies among high-frequency data segment with robnst noise, TMPV can depict the character of generalized jump with different types of noise, and construct statistics to distinguish the generalized jump. The result of Monte Carlo simulation also confirms the result of the theory.

关 键 词：无限活跃跳跃 有限活跃跳跃 TMPV BNS 

分 类 号：F830[经济管理—金融学]