Applied Econometrics 3rd Edition - GBV

AppliedEconometrics3rd EditionDimitrios AsteriouProfessor in Econometrics, Hellenic Open Universily, CreeceStephen G. HallProfessor of Economics and Pro-Vice Chancellor,Universily of Leicester, UKpalgrave

ContentsList gementsxxxPart IStatistical Background and Basic Data Handling11Fundamental ConceptsIntroductionA simple exampleA Statistical frameworkProperties of the sampling distribution of the meanHypothesis testing and the central limit theoremCentral limit theoremConclusion34467810132The Structure of Economic Data and Basic Data HandlingLearning objectivesThe structure of economic dataCross-sectional dataTime series dataPanel dataBasic data handlingLooking at raw dataGraphical analysisSummary statistics14141515151617171719Part II3The Classical Linear Regression ModelSimple RegressionLearning objectivesIntroduction to regression: the classical linear regression model (CLRM)Why do we do regressions?The classical linear regression modelix272929303030

X Contents4The Ordinary Least Squares (OLS) method of estimationAlternative expressions for ßThe assumptions of the CLRMGeneralThe assumptionsViolations of the assumptionsProperties of the OLS estimatorsLinearityUnbiasednessEfficiency and BLUEnessConsistencyThe overall goodness of fitProblems associated with R2Hypothesis testing and confidence intervalsTesting the significance of the OLS coefficientsConßdence intervalsHow to estimate a simple regression in EViews and StataSimple regression in EViewsSimple regression in StataReading the Stata simple regression results outputReading the EViews simple regression results outputPresentation of regression resultsEconomic theory applicationsApplication 1: the demand functionApplication 2: the production functionApplication 3: Okun's lawApplication 4: the Keynesian consumption functionComputer example: the Keynesian consumption functionSolutionQuestions and 9505050515252535358Multiple RegressionLearning objectivesIntroductionDerivation of multiple regression coefficientsThe three-variable modelThe /(-variables caseDerivation of the coefficients with matrix algebraThe structure of the X'X and X'Y matricesThe assumptions of the multiple regression modelThe variance-covariance matrix of the errorsProperties of multiple regression model OLS R2 and adjusted R2General criteria for model selection6262646464656667686969697070707273

Contents xiMultiple regression estimation in EViews and StataMultiple regression in EViewsMultiple regression in StataReading the EViews multiple regression results OutputHypothesis testingTesting individual coefficientsTesting linear restrictionsThe F-form of the Likelihood Ratio testTesting the joint significance of the XsF-test for overall significance in EViewsAdding or deleting explanatory variablesOmitted and redundant variables test in EViewsHow to perform the Wald test in EViewsThe t test (a special case of the Wald procedure)The Lagrange Multiplier (LM) testThe LM test in EViewsComputer example: Wald, omitted and redundant variables testsA Wald test of coefficient restrictionsA redundant variable testAn omitted variable testComputer example: commands for StataFinancial econometrics application: the Capital Asset Pricing Modelin actionA few theoretical remarks regarding the CAPMThe empirical application of the CAPMEViews programming and the CAPM applicationAdvanced EViews programming and the CAPM applicationQuestions and exercisesPart III5Violating the Assumptions of the CLRMMulticollinearityLearning objectivesIntroductionPerfect multicollinearityConsequences of perfect multicollinearityImperfect multicollinearityConsequences of imperfect multicollinearityDetecting problematic multicollinearitySimple correlation coefficientR2 from auxiliary regressionsComputer examplesExample 1: induced multicollinearityExample 2: with the use of real economic dataQuestions and 10112115

xii Contents67HeteroskedasticityLearning objectivesIntroduction: what is heteroskedasticity?Consequences of heteroskedasticityfor OLS estimatorsA general approachA mathematical approachDetecting heteroskedasticityThe informal wayThe Breusch-Pagan LM testThe Glesjer LM testThe Harvey-Godfrey LM testThe Park LM testCriticism of the LM testsThe Goldfeld-Quandt testWhite's testComputer example: heteroskedasticity testsThe Breusch-Pagan testThe Glesjer testThe Harvey-Godfrey testThe Park testThe Goldfeld-Quandt testWhite's testCommands for the Computer example in StataEngle's ARCH testComputer example of the ARCH-LM testResolving heteroskedasticityGeneralized (or weighted) least squaresComputer example: resolving heteroskedasticityQuestions and exercisesAutocorrelationLearning objectivesIntroduction: what is autocorrelation?What causes autocorrelation?First- and higher-order autocorrelationConsequences of autocorrelationfor the OLS estimatorsA general approachA more mathematical approachDetecting autocorrelationThe graphical methodExample: detecting autocorrelation using the graphical methodThe Durbin-Watson testComputer example of the DW testThe Breusch-Godfrey LM test for serial correlationComputer example of the Breusch-Godfrey testDurbin's h test in the presence of lagged dependent variablesComputer example of Durbin's h 157158159159160162162162164166167168170171

Contents xiiiResolving autocorrelationWhen p is knownComputer example of the generalized differencing approachWhen p is unknownComputer example of the iterative procedureResolving autocorrelation in StataQuestions and exercisesAppendix8Misspecification: Wrong Regressors, Measurement Errors and WrongFunctional FormsLearning objectivesIntroductionOmitting influential or including non-influential explanatory variablesConsequences of omitting influential variablesIncluding a non-influential variableOmission and inclusion of relevant and irrelevant variablesat the same timeThe plug-in Solution in the omitted variable biasVarious functional formsIntroductionLinear-log functional formReciprocal functional formPolynomial functional formFunctional form including interaction termsLog-linear functional formThe double-log functional formThe Box-Cox transformationMeasurement errorsMeasurement error in the dependent variableMeasurement error in the explanatory variableTests for misspecificationNormali ty of residualsThe Ramsey RESET test for general misspecificationTests for non-nested modelsComputer example: the Box-Cox transformation in EViewsApproaches in choosing an appropriate modelThe traditional view: average economic regressionThe Hendry 'general to specific approach'Questions and ExercisesPart IV9Topics in EconometricsDummy VariablesLearning objectivesIntroduction: the nature of qualitative 3195197199202202203204207209209210

xiv Contents10The use of dummy variablesIntercept dummy variablesSlope dummy variablesThe combined effect of intercept and slope dummiesComputer example of the use of dummy variablesUsing a constant dummyUsing a slope dummyUsing both dummies togetherSpecial cases of the use of dummy variablesUsing dummy variables with multiple categoriesUsing more than one dummy variableUsing seasonal dummy variablesComputer example of dummy variables with multiple categoriesFinancial econometrics application: the January effect inemerging stock marketsTests for structural stabilityThe dummy variable approachThe Chow test for structural stabilityFinancial econometrics application: the day-of-the-week effect in actionQuestions224227227227228230Dynamic Econometric ModelsLearning objectivesIntroductionDistributed lag modelsThe Koyck transformationThe Almon transformationOther models of lag structuresAutoregressive modelsThe partial adjustment modelA Computer example of the partial adjustment modelThe adaptive expectations modelTests of autocorrelation in autoregressive 124111 Simultaneous Equation ModelsLearning objectivesIntroduction: basic definitionsConsequences of ignoring simultaneityThe Identification problemBasic definitionsConditions for IdentificationExample of the Identification procedureA second example: the macroeconomic model of aclosed 43244245245245246247247

Contents xvEstimation of simultaneous equation modelsEstimation of an exactly identified equation: the ILS methodEstimation of an over-identified equation: the TSLS methodComputer example: the IS-LM modelEstimation of simultaneous equations in Stata24824924925025312 Limited Dependent Variable Regression ModelsLearning objectivesIntroductionThe linear probability modelProblems with the linear probability modelDi is not bounded by the (0,1) rängeNon-normality and heteroskedasticity of the disturbancesThe coefficient of determination as a measure of overall fitThe logit modelA general approachInterpretation of the estimates in logit modelsGoodness of fitA more mathematical approachThe probit modelA general approachA more mathematical approachMultinomial and ordered logit and probit modelsMultinomial logit and probit modelsOrdered logit and probit modelsThe Tobit modelComputer example: probit and logit models in EViews and StataLogit and probit models in EViewsLogit and probit models in Stata25425425525525625625 art V273Time Series Econometrics13 ARIMA Models and the Box-Jenkins MethodologyLearning objectivesAn introduction to time series econometricsARIMA modelsStationarityAutoregressive time series modelsThe AR(1) modelThe AR(p) modelProperties of the AR modelsMoving average modelsThe MA(1) modelThe MA(g) modelInvertibility in MA modelsProperties of the MA modelsARMA 5

xvi Contents1415Integrated processes and the ARIMA modelsAn integrated seriesExample of an ARIMA modelBox-Jenkins model selectionIdentificationEstimationDiagnostic checkingThe Box-Jenkins approach step by stepComputer example: the Box-Jenkins approachThe Box-Jenkins approach in EViewsThe Box-Jenkins approach in StataQuestions and ling the Variance: ARCH-GARCH ModelsLearning objectivesIntroductionThe ARCH modelThe ARCH(l) modelThe ARCH( j) modelTesting for ARCH effectsEstimation of ARCH models by IterationEstimating ARCH models in EViewsA more mathematical approachThe GARCH modelThe GARCH(p, q) modelThe GARCH(%i) model as an infinite ARCH processEstimating GARCH models in EViewsAlternative specificationsThe GARCH in mean or GARCH-M modelEstimating GARCH-M models in EViewsThe threshold GARCH (TGARCH) modelEstimating TGARCH models in EViewsThe exponential GARCH (EGARCH) modelEstimating EGARCH models in EViewsAdding explanatory variables in the mean equationAdding explanatory variables in the variance equationEstimating ARCH/GARCH-type models in StataAdvanced EViews programming for the estimation of GARCH-typemodelsApplication: a GARCH model of UK GDP and the effect ofsocio-political instabilityQuestions and 0311312313316316317318319319320326330Vector Autoregressive (VAR) Models and Causality TestsLearning objectivesVector autoregressive (VAR) modelsThe VAR modelPros and cons of the VAR models333333334334335322