资源描述：
《bootstrapping unit root tests for autoregressive time series》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。
1、EconomicsLetters72(2001)1±10www.elsevier.com/locate/econbaseOnbootstrapinferenceincointegratingregressionsZachariasPsaradakis*BirkbeckCollege,UniversityofLondon,SchoolofEconomics,MathematicsandStatistics,7±15GresseStreet,LondonW1P2LL,UKReceived11September2000;accepted21December2000Abst
2、ractThispaperconsiderstheconstructionofbootstraphypothesistestsandcon®denceregionsfortheparametersofcointegratingregressions.Wesuggestusingasievebootstrapschemebasedonresamplingresidualsfromanautoregressiveapproximationtotheinnovationprocessdrivingthecointegratedsystem.Simulationsdemon
3、-stratethesmall-sampleeffectivenessofthisbootstrapmethodinthecaseoftwocommonlyusedestimatorsforcointegratingregressions.Ó2001ElsevierScienceB.V.Allrightsreserved.Keywords:Autoregressiveapproximation;Cointegratingregression;SievebootstrapJELclassi®cation:C12;C221.IntroductionThispaperad
4、dressestheproblemofinferenceincointegratingregressionsinvolvingI(1)variables.Thecoef®cientsofsuchregressionsmaybeestimatedbyavarietyofdifferentmethodswhichproduceestimatorsthatareasymptoticallymixedGaussianandmedianunbiasedandwhichallowinferencetobecarriedoutusingasymptoticnormalorchi-
5、squaredtestcriteria(see,interalia,PhillipsandHansen,1990;Saikkonen,1991;Park,1992;StockandWatson,1993;PesaranandShin,1999).MonteCarlostudieshaverevealed,however,thatconventionallarge-sampletheoryoftenprovidespoorapproximationstothedistributionsofassociatedteststatisticsinsmallandmodera
6、telysizedsamples.Onewayofimprovingthequalityof®nite-sampleinferenceinthesesituationsisbyusingthebootstrap.Instationarysituations,thisresamplingtechniqueprovidesapproximationstothesamplingdistributionsofestimatorsandteststatisticsthatareoftenmoreaccuratethanthoseobtainedby®rst-orderasym
7、ptotictheory,andhasbeenfoundtoperformimpressivelyinavarietyofsettings(seeShaoandTu,1995).*Tel.:144-20-7631-6415;fax:144-20-7631-6416.E-mailaddress:z.psaradakis@bbk.ac.uk(Z.Psaradakis).0165-1765/01/$±seefrontmatterÓ2001ElsevierScienceB.V.Allrightsreserved.PII:S0165-1765(01)00410-42Z.P
