基于MCMC模拟的贝叶斯厚尾金融随机波动模型分析.pdfVIP

第16 卷 第4 期 运 筹 与 管 理 V o l. 16, N o. 4 2007 年8 月 O PERA T IO NS R ESEA RCH A N D M A N A GEM EN T SCI ENCE A ug . 2007 MCMC 朱慧明, 李 峰, 杨锦明 , 4 10082) : , SV SV- T ) , G ibbs M CM C , DIC SV- N SV- T : , SV- T SV- N , , : ; M CM C ; SV- T ; Gibbs ; D IC : O 212 . 8 : A : 1007- 3221 2007) 04- 0 111- 05 Bayesian Modeling of Heavy-tailed Stochastic Volat ility Financial Model ZH U H ui- ming , LI Feng , YA N G Jin- ming ( School of B usiness A d ministration, H unan Univ er sity , Chang sha 410079, China) Abstract: T o so lve t he problem t hat t he ex ist ing st ochast ic v olat ilit y m odel cannot describe t he character- ist ics of par am et ers. t ime- chang ing, t his paper est ablishes t he Bayesian heav y- t ailed volat ilit y m odel. T he paper f irst ly st udies t he m odel. s st at ist ical str ucture, cho oses the param et er. s prior distr ibution , designs a M arkov chain M ont e Car lo alg orit hm procedur e w it h Gibbs sam pler to carr y o ut simulat ion a- nalysis, and co mpar es t he SV- N model and SV-T model in t he quality u sing t he DIC crit erion. T he re- sults indicat e that , in modeling t he volat ilit y in t he Chinese st ock market, t he SV- T m odel is superior t o the SV- N m odel, w hich can bet t er char act eriz e t he lept okurt ic of st ock ret urns. F urt hermo re, t he re- sults also prov e t hat t he Chinese sto ck m ar ket has hig h persist ence of vo lat ilit y. Key words: bay esian analysis ; MCM C mo deling ; SV-T model; g ibbs sampling ; DIC cr it erion , , Random Walk) , : 2003 Eng le 1982 aut ore- g ressio n condit ional het er oscedast icit