文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > MCMC Methods for Estimating Stochastic Volatility Models with Leverage Effects Comments on
侵权投诉

MCMC Methods for Estimating Stochastic Volatility Models with Leverage Effects Comments on

In this note we represent the well known discrete time stochastic volatility (SV) model with a leverage effect and the SV model of Jacquier, Polson and Rossi (JPR) (2002) using Gaussian nonlinear state space forms with uncorrelated measurement and transiti

nonlinearstatespaceform: X=σtut,t

lnσ2=α+φlnσ2+ρσσ 1X+σ ,vttvttt 1(2.4)

2wherewtisiidN(0,1)andcorr(ut,wt)=0.Itisapparentinequation(2.4)thatlnσtisrelatedtoXtinamorecomplicatednonlinearway.De ne

222 1F(lnσt,Xt)≡lnσt α φlnσt 1 ρσvσtXt σv1 ρ2wt.

Theimplicitfunctiontheoremimpliesthat

2 lnσt F/ Xtρσv/σt= =. Xt F/lnσt1+0.5ρσvut(2.5)

Similarlywecanshowthat

2 lnσtρσvσt+1/σt+1=. Xt1+0.5ρσvut+1(2.6)

Whilethenumeratorsin(2.5)and(2.6)arealwaysnegativewhenρ<0,thedenomina-torscanbeeitherpositiveornegative.Asaresult,thereisnoguaranteethatthepartialderivativewillalwaysbenegativeandhencetheleveragee ectisnotwarranted.13EstimationofA-SVmodels

MCMCestimationoftheconventionallyspeci edA-SVmodelcanbedonebyusingthestatespacerepresentation(2.3)whichleadstoalog-concavefullconditionaldistribution.Inconsequence,onecanemploytheadaptiverejectionsamplingalgorithmofGilksandWild(1992).Alternatively,onecanmakeuseoftheallpurposeBayesiansoftwarepackageBUGS,basedonadi erentrepresentationintroducedinMeyerandYu(2000).Thefullconditionaldistributionbasedonthisalternativerepresentationis,however,nolongerlog-concaveandhenceaMetropolis-Hastings(MH)updatingstepisneeded.Anadvantageofthelatterapproachisthatitcanbeeasilymodi edtodealwithJPR’sspeci cation.ThisalternativerepresentationoftheA-SVmodelisobtainedbyspecifyingthestateandobservationequationsasfollows:

22ht+1|ht,α,φ,σv~N(α+φht,σv), ρht/22Xt|ht,ht+1,α,φ,σv,ρ~Ne(ht+1 α φht),eht(1 ρ2).σv

Givenempiricallypossiblevaluesofρandσv,however,itappearsmuchmorelikelyforthedenom-inatortotakeanegativevaluethanapositivevalue.Forexample,thelargestleveragee ectreportedintheliteratureis-0.66(seeHarveyandShephard(1996))andthelargestestimateofσvis0.85(seeMahieuandSchotman(1998)).Toensurealeveragee ect,uthastotakeavaluelargerthan-3.565andthisoccurswithprobability0.9998.1

上一页第5页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email