ARCH模型族的建模研究.pdf

1、天津大學碩士學位論文ARCH模型族的建模研究姓名：柯珂申請學位級別：碩士專業(yè)：技術(shù)經(jīng)濟指導教師：張世英2000.1.1AbstractTraditionaleconometricmodelingisbasedonthehypothesisofhomoskedasticityHowevermosttimeseriesofactuaIsystemaresubjecttostochasticchangesinvolatilityovertim

2、eARCHmodelpresentedbyEnglehasincreasinglygainedprominenceastheleadingmodelfordescribingstochasticvolatilityNowtheARCHfamilyofmodelshasbecomeahotresearchtopicinEconometriCSHoweverthis“ndofresearchhasonlymadeslowprogressin

3、ChineseeconometriCSresearchcircleandevensuchlirtleresearchiSfocusedonapplication。notontheoryThisthesistriestomakesomebreakthrough111efractionalintegrationtheoryandgeneticalgorithmideasareusedinmodclingtimeserieswithheter

4、oskedasticityThenewmodeJoffractionalintegratedaugmentedGARCH—ManditsrelatedestimationandtesttechniqueareputforwardThefractionalintegratedaugmentedGARCHMisprovedrationalandfeasibleanditsutilityiSvalidatedbyempiricalstudyT

5、hethesisbeginswiththeresearchbackgroundChaptertwoisliteraturesurveyofthedevelopmentinARCHmodelingandmakestheARCHfamilyofmodelsasystemforthefirsttimeChapterthreesummadzesthelongmemorytimeseriesanalysisandmodeling，introduc

6、esthelongmemorypropertyexistinginthesecondordermomentanddepictsthatthepersistenceofconditionalheteroskedasticityiSthereasonresultinginlongmemorypropertyinvarianceBasedOilthispoint，chapterfourproposedfractionalintegrateda

7、ugmentedGARCHMwhichCanembracealmostallthecurrentARCHmodelsanddescribethelongmemorypropertyThehandicapinmodelspecificationtestlongmemorytestandestimationisovercomeTheavailabilityandsuperiorityofFIAugmentedGARCHmodelisasse

8、ssedthroughasimulationstudyOnthebaseofChapterFivewhichSumsupthecurrentestimationmethodsandtestproceduresforARCHmodels，ChapterSixintroducestheprobleminestimatinglongmemoryARCHmodelsItalSOpointsoutthattheestimationmethodma

9、ynotexitforsomeARCH1110delsespeciallyafterthelongmemoryiSintrodueedintoARCHmodelingAccordingtoFIAugmentedGARCHmodelthatCan’tbedifferentialtraditionallikelihoodestimationmethodandmostparameteroptimizationmethodsareinvalid

10、TabuSearchHybridHierarchyGeneticAlgorithm(TSHHGAlwhichcanavoid“premature”toacertainextentiSproposedtoestimateFIAugmentedGARCHmodelAusefulARCHsoareiscompiledThediagnoseanalysisandvariablestructureresearchinARCHmodelsaredi

11、scussedinChapterSevenWiththecompositeindexofclosingpriceinShanghaistockmarket，thevalidityofthediagnoseresearchisverifiedKeywords：longmemorytimeseries，fractionalintegration，ARCH，persistence，fractionalintegratedaugmentedGA