《结构向量自回归（SVAR）模型》课程讲义 上海财经大学

《《结构向量自回归（SVAR）模型》课程讲义 上海财经大学》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、SVARModelChanglinGuoShanghaiUniversityofFinanceandEconomics........................................[DISCLAIMER]ThesenotesaremeanttoprovideintuitiononthebasicmechanismsofVARsAssuch,mostofthematerialcoveredhereistreatedinaveryinformalwayIfyoucraveaformaltreatmentofthesetopics

2、,youshouldstophereandbuyacopyofHamilton’s“TimeSeriesAnalysis”VARmodels–VARRepresentation2VARs&Macro-econometricians’jobIAccordingtoawell-knownpaperbyStock&Watson(2001,JEP)macroeconometricians(wouldliketo)dofourthings1.Describeandsummarizemacroeconomictimeseries2.Makeforecas

3、ts3.Recoverthetruestructureofthemacroeconomyfromthedata4.AdvisemacroeconomicpolicymakersIVectorautoregressivemodelsareastatisticaltooltoaddressthesetasksVARmodels–VARRepresentation3Whatcanwedowithvectorautoregressivemodels?I3variables:realGDPgrowth(Dy),inflation(p)andthepoli

4、cyrate(i)IAVARcanhelpusansweringthefollowingquestions1.Whatisthedynamicbehaviourofthesevariables?Howdothesevariablesinteract?2.WhatistheprofileofGDPconditionalonaspecificfuturepathforthepolicyrate?3.WhatistheeffectofamonetarypolicyshockonGDPandinflation?4.Whathasbeenthecontribu

5、tionofmonetarypolicyshockstothebehaviourofGDPovertime?VARmodels–VARRepresentation4WhatisaVectorAutoregression(VAR)?IGiven,forexample,a(31)vectoroftimeseriesxtwhere23Dy1Dy2...DyTxt=4p1p2...pT5r1r2...rTIAstationarystructuralVARoforder1isAxt=Bxt1+#tfort=1,...,TAandBare(33)

6、matricesofcoefficients#tisan(31)vectorofunobservablezeromeanwhitenoiseprocessesVARmodels–VARRepresentation5ThreedifferentwaysofwritingthesamethingITherearedifferentwaystorepresenttheVAR(1)Axt=Bxt1+#tIForexample,wecanalsowriteit:Inmatrixform2323232323a11a12a13Dytb11b12b13Dyt

7、1#Dyt4a21a22a2354pt5=4b21b22b2354pt15+4#pt5a31a31a33rtb31b32b33rt1#rtAsasystemoflinearequation8>:a31Dyt+a32pt+a33rt=b31Dyt1+b32pt1+b33rt1+#rtVARmodels–VARRepresentation6T

8、hestructuralinnovationsIWedefined#tasa“vectorofunobservablezeromeanwhitenoiseproces