2019年第10章 序列相关性ppt课件.ppt

《2019年第10章 序列相关性ppt课件.ppt》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、第10章　序列相关性SerialCorrelation/AutocorrelationMainContentsWhatisSerialcorrelation(Autocorrelation)?TheconsequencesofserialcorrelationHowtodetecttheserialcorrelation?CorrectionsforserialcorrelationWhatisSerialcorrelation(Autocorrelation)?Theassumptionthaterrorscorrespondingtod

2、ifferentobservationsareuncorrelatedoftenbreaksdownintime-seriesstudies.Whentheerrortermsfromdifferent(usuallyadjacent)timeperiodsarecorrelated,wesaythattheerrortermisseriallycorrelated.Thatis,Cov(ui,uj)0,i.e.E(ui,uj)0forij.PatternsofserialcorrelationReasonsofserialcorre

3、lationInertiaorsluggishnessModelspecificationerrors(omittedvariables)WhatisSerialcorrelation(Autocorrelation)?Inthischapter,weonlydealwiththeproblemoffirst-orderserialcorrelation,inwhicherrorsinonetimeperiodarecorrelateddirectlywitherrorsintheensuingperiod.Forexample,ut=ru

4、t-1+vtSecond-orderserialcorrelationwillbeut=r1ut-1+r2ut-2+vtTheconsequencesofserialcorrelation(Autocorrelation)OLSestimatorswillbestillunbiasedandconsistent.takethesimpleregressionasanexampleY=b0+b1X+uWeknowtheOLSestimatorofb1isTheconsequencesofserialcorrelation(Autocorrel

5、ation)TheR2andadj-R2arestillconsistentifthetimeseriesisstationary(that’sr<1).Orelse,fornon-stationarytimeseries,theR2andadj-R2maybeinvalid.Theconsequencesofserialcorrelation(Autocorrelation)OLSestimatorswillnotbeefficient.ThevarianceofOLSestimatorswillbebiased.Theconsequen

6、cesofserialcorrelation(Autocorrelation)t-statisticsandF-statisticwillbemisleadingwhenthereareserialcorrelationinerrortermsut.Thevarianceandstandarderrorofthepredictedvaluewillbeinvalid.Howtodetecttheserialcorrelation?Time-sequenceplotRunstestDurbin-WatsontestTimesequencepl

7、otExample:Realwagesandproductivity(Example10-1)RunstestFirst,getthesignoftheresiduals,et,forexample,(---------)(++++++++)(----)(++)(---),thatis,thereare9negativesigns,followedby8positivesignsandsoon.Thesamesignsintheparenthesesarecalledarun.LetNisthenumberofobservations,an

8、dN1isthenumberofpositivesignsoftheresiduals,andN2isthenumberofnegativesigns.Andk’isthenum