Stock计量经济学第14章ppt课件.ppt

《Stock计量经济学第14章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、Chapter14IntroductiontoTimeSeriesRegressionandForecastingIntroductiontoTimeSeriesRegressionandForecasting(SWChapter14)2Example#1oftimeseriesdata:USrateofpriceinflation,asmeasuredbythequarterlypercentagechangeintheConsumerPriceIndex(CPI),atanannualrate3Example#2:USrateofunemploymen

2、t4Whyusetimeseriesdata?5Timeseriesdataraisesnewtechnicalissues6UsingRegressionModelsforForecasting(SWSection14.1)7IntroductiontoTimeSeriesDataandSerialCorrelation(SWSection14.2)8Wewilltransformtimeseriesvariablesusinglags,firstdifferences,logarithms,&growthrates9Example:Quarterlyr

3、ateofinflationatanannualrate(U.S.)10Example:USCPIinflation–itsfirstlaganditschange11Autocorrelation1213Sampleautocorrelations14Example:1516Othereconomictimeseries:17Othereconomictimeseries,ctd:18Stationarity:akeyrequirementforexternalvalidityoftimeseriesregression19Autoregressions

4、(SWSection14.3)20TheFirstOrderAutoregressive(AR(1))Model21Example:AR(1)modelofthechangeininflation22Example:AR(1)modelofinflation–STATA23Example:AR(1)modelofinflation–STATA,ctd.24Example:AR(1)modelofinflation–STATA,ctd25Forecasts:terminologyandnotation26Forecasterrors27Example:for

5、ecastinginflationusinganAR(1)28TheAR(p)model:usingmultiplelagsforforecasting29Example:AR(4)modelofinflation30Example:AR(4)modelofinflation–STATA31Example:AR(4)modelofinflation–STATA,ctd.32Digression:weusedInf,notInf,intheAR’s.Why?33SowhyuseInft,notInft?34TimeSeriesRegressionwith

6、AdditionalPredictorsandtheAutoregressiveDistributedLag(ADL)Model(SWSection14.4)35Example:inflationandunemployment36TheempiricalU.S.“PhillipsCurve,”1962–2004(annual)37Theempirical(backwards-looking)PhillipsCurve,ctd.38Example:dinfandunem–STATA39Example:ADL(4,4)modelofinflation–STAT

7、A,ctd.40ThetestofthejointhypothesisthatnoneoftheX’sisausefulpredictor,aboveandbeyondlaggedvaluesofY,iscalledaGrangercausalitytest41Forecastuncertaintyandforecastintervals42Themeansquaredforecasterror(MSFE)is,43Therootmeansquaredforecasterror(RMSFE)44ThreewaystoestimatetheRMSFE45Th

8、emethodofpseudoout-of-sampleforec