资源描述:
《a comparison of extreme value theory approaches for…》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、NOTICE:thisistheauthor’sversionofaworkthatwasacceptedforpublicationintheJournalofEmpiricalFinance.Changesresultingfromthepublishingprocess,suchaspeerreview,editing,corrections,structuralformatting,andotherqualitycontrolmechanismsmaynotbereflectedinthisdocument.Changesmayhavebee
2、nmadetothisworksinceitwassubmittedforpublication.AdefinitiveversionwassubsequentlypublishedintheJournalofEmpiricalFinance,12.2(2005),DOI:10.1016/j.jempfin.2004.01.0042AComparisonofExtremeValueTheoryApproachesforDeterminingValueatRiskbyC.Brooks,A.D.Clare,J.W.DalleMolleandG.Persa
3、ndDecember2003AbstractThispapercomparesanumberofdifferentextremevaluemodelsfordeterminingthevalueatriskofthreeLIFFEfuturescontracts.Asemi-nonparametricapproachisalsoproposedwherethetaileventsaremodeledusingtheGeneralisedParetoDistributionandnormalmarketconditionsarecapturedbyt
4、heempiricaldistributionfunction.Thevalueatriskestimatesfromthisapproacharecomparedwiththoseofstandardnonparametricextremevaluetailestimationapproaches,withasmallsamplebias-correctedextremevalueapproach,andwiththosecalculatedfrombootstrappingtheunconditionaldensityandbootstrappi
5、ngfromaGARCH(1,1)model.Theresultsindicatethatforahold-outsample,theproposedsemi-nonparametricextremevalueapproachyieldssuperiorresultstoothermethods,butthesmallsampletailindextechniqueisalsoaccurate.Keywords:Bootstrap,ValueatRisk(VaR),GeneralisedParetoDistribution,Parametric,Se
6、mi-nonparametricandSmallSampleBiasCorrectedTailIndexEstimators,GARCHmodels.JELClassifications:C14,C15,G13WewouldliketothanktheEditor,FranzPalm,anAssociateEditorofthisjournal,andananonymousrefereeforusefulcommentsonapreviousversionofthispaper,andSalihNeftciforusefulconversation
7、s.Theusualdisclaimerapplies.Brooks(correspondingauthor)iswiththeISMACentre,UniversityofReading,WhiteknightsPark,POBox242,Reading,RG66BA,UnitedKingdom,tel:(+44)1189316768;fax:(+44)1189314741;e-mail:C.Brooks@rdg.ac.uk;ClareiswithLegalandGeneralAssetManagement,London;DalleMolleisf
8、romMasseyUniversity,NewZealand;andPersandiswiththeUniv