截面与面板数据分析-复旦2

《截面与面板数据分析-复旦2》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、AnalysisofCrossSectionandPanelDataYanZhangSchoolofEconomics,FudanUniversityCCER,FudanUniversityIntroductoryEconometricsAModernApproachYanZhangSchoolofEconomics,FudanUniversityCCER,FudanUniversityAnalysisofCrossSectionandPanelDataPart1.RegressionAnalysisonCrossSectionalData

2、Chap2.TheSimpleRegressionModel——PracticeforlearningmultipleRegressionBivariatelinearregressionmodel:theslopeparameterintherelationshipbetweenyandxholdingtheotherfactorsinufixed;itisofprimaryinterestinappliedeconomics.:theinterceptparameter,alsohasitsuses,althoughitisrarely

3、centraltoananalysis.MoreDiscussion:Aone-unitchangeinxhasthesameeffectony,regardlessoftheinitialvalueofx.Increasingreturns:wage-education(f.form)Canwedrawceterisparibusconclusionsabouthowxaffectsyfromarandomsampleofdata,whenweareignoringalltheotherfactors?Onlyifwemakeanassu

4、mptionrestrictinghowtheunobservablerandomvariableuisrelatedtotheexplanatoryvariablexClassicalRegressionAssumptionsFeasibleassumptioniftheintercepttermisincludedLinearlyuncorrelatedzeroconditionalexpectationMeaning=内生性PRF(PopulationRegressionFunction):sth.fixedbutunknownOLS

5、Minimizeuusampleregressionfunction(SRF)ThepointisalwaysontheOLSregressionline.拟合值与残差PRF:OLSCoefficientofdeterminationthefractionofthesamplevariationinythatisexplainedbyx.thesquareofthesamplecorrelationcoefficientbetweenandLowR-squaredsUnitsofMeasurementIfoneofthedependentv

6、ariablesismultipliedbytheconstantc—whichmeanseachvalueinthesampleismultipliedbyc—thentheOLSinterceptandslopeestimatesarealsomultipliedbyc.Ifoneoftheindependentvariablesisdividedormultipliedbysomenonzeroconstant,c,thenitsOLSslopecoefficientisalsomultipliedordividedbycrespec

7、tively.Thegoodness-of-fitofthemodel,R-squareds,shouldnotdependontheunitsofmeasurementofourvariables.FunctionFormLinearNonlinearLogarithmicdependentvariableAPercentagechangeiny,semi-elasticityanincreasingreturntoedu.Othernonlinearity:diplomaeffectBi-LogarithmicAaConstantela

8、sticityChangeofunitsofmeasurementP45,error:b0*＝b0+log(c1)-b1·log(c2)Bi-LogarithmicAaConst