简介:InthispaperthelimitingdistributionoftheleastsquareestimatefortheautoregressivecoefficientofanearlyunitrootmodelwithGARCHerrorsisderived.Sincethelimitingdistributiondependsontheunknownvarianceoftheerrors,anempiricallikelihoodratiostatisticisproposedfromwhichconfidenceintervalscanbeconstructedforthenearlyunitrootmodelwithoutknowingthevariance.Togainanintuitivesensefortheempiricallikelihoodratio,asmallsimulationfortheasymptoticdistributionisgiven.
简介:Motivatedbythedoubleautoregressivemodelwithorderp(DAR(p)model),inthispaper,westudythemovingaveragemodelwithanalternativeGARCHerror.ThemodelisanextensionfromDAR(p)modelbylettingtheorderpgoestoinfinity.Thequasimaximumlikelihoodestimatoroftheparametersinthemodelisshowntobeasymptoticallynormal,withoutanystrongmomentconditions.Simulationresultsconfirmthatourestimatorsperformwell.WealsoapplyourmodeltostudyarealdatasetandithasbetterfittingperformancecomparedtoDARmodelfortheconsidereddata.
简介:Thegeneralizedautoregressiveconditionalheteroskedasticity(GARCH)typemodelsareusedtoinvestigatethevolatilityofBangladeshstockmarket.Thefindingsofthestudydemonstratethattheindexvolatilitycharacteristicschangesovertime.Thearticleshowsthatthedataaredividedintothreesub-periods:precrisis,crisis,andpostcrisis.Accordingly,theresultsofthefindingsindicatechangesintheGARCH-typemodelsparameter,riskpremiumandpersistenceofvolatilityindifferentperiods.Asignificant'low-yieldassociatedwithhigh-risk'phenomenonisdetectedinthecrisisperiodandthe'leverageeffect'occursineachperiods.Theinvestorsareirrationalwhichisbasedonassumptionofriskandreturncharacteristicsofassets.Consequently,themarketisnotasmatureasdevelopedmarket.Itisfoundinthearticlethatthethresholdgeneralizedautoregressiveconditionalheteroskedasticity(TGARCH)modelismoreaccurateforthemodelaccuracy.Additionally,statisticerrormeasurementsindicatethatGARCHmodelismoreefficientthanothersandithasalsomoreforecastingability.
简介:
简介:根据ES风险度量方法,拓展了马克维茨均值一方差资产组合模型,研究均值一ES准则下的资产组合问题。用APD—GARCH模型刻画风险资产收益率序列,以多元Copula函数描述风险资产间的相关结构信息,构造灵活的Copula—APDG—ARCH模型。利用该模型,借助MonteCarlo模拟,分别研究相关结构是多元正态Copula函数、多元t-Copula函数和多元ClaytonCopula函数的风险资产组合的均值-ES有效前沿,并进行比较。实证研究表明,在有效组合范围内,正态Copula函数明显高估了资产的组合风险;当期望收益较小时,t-Copula函数对应的风险值最小,但随着期望收益的增加,多元ClaytonCopula函数对应的有效前沿表现最好。
简介:本文利用收集到的上证指数的日数据,应用TAR-CARCH模型分析股市中信息不对称对收益波动影响的持续程度及风险与日收益的关系,发现利空消息对波动性的影响更为持久,而且政策因素与股市波动关系密切.