Fundamental Statistical Inference WILEYSERIESINPROBABILITYANDSTATISTICS EstablishedbyWalterA.ShewhartandSamuelS.Wilks Editors:DavidJ.Balding,NoelA.C.Cressie,GarrettM.Fitzmaurice, GeofH.Givens,HarveyGoldstein,GeertMolenberghs,DavidW.Scott, AdrianF.M.Smith,RueyS.Tsay EditorsEmeriti:J.StuartHunter,IainM.Johnstone,JosephB.Kadane, JozefL.Teugels TheWileySeriesinProbabilityandStatisticsiswellestablishedandauthoritative.Itcovers manytopicsofcurrentresearchinterestinbothpureandappliedstatisticsandprobability theory.Writtenbyleadingstatisticiansandinstitutions,thetitlesspanbothstate-of-the-art developmentsinthefieldandclassicalmethods. Reflectingthewiderangeofcurrentresearchinstatistics,theseriesencompassesapplied, methodological and theoretical statistics, ranging from applications and new techniques made possible by advances in computerized practice to rigorous treatment of theoreti- cal approaches. This series provides essential and invaluable reading for all statisticians, whetherinacademia,industry,government,orresearch. Acompletelistoftitlesinthisseriescanbefoundathttp://www.wiley.com/go/wsps Fundamental Statistical Inference A Computational Approach MarcS.Paolella DepartmentofBankingandFinance UniversityofZurich Switzerland (cid:2) Thiseditionfirstpublished2018 ©2018JohnWiley&SonsLtd Allrightsreserved.Nopartofthispublicationmaybereproduced,storedinaretrievalsystem,ortransmitted,in anyformorbyanymeans,electronic,mechanical,photocopying,recordingorotherwise,exceptaspermittedby law.Adviceonhowtoobtainpermissiontoreusematerialfromthistitleisavailableathttp://www.wiley.com/ go/permissions. TherightofMarcS.Paolellatobeidentifiedastheauthorofthisworkhasbeenassertedinaccordancewithlaw. RegisteredOffices JohnWiley&Sons,Inc.,111RiverStreet,Hoboken,NJ07030,USA JohnWiley&SonsLtd,TheAtrium,SouthernGate,Chichester,WestSussex,PO198SQ,UK EditorialOffice 9600GarsingtonRoad,Oxford,OX42DQ,UK Fordetailsofourglobaleditorialoffices,customerservices,andmoreinformationaboutWileyproductsvisitus atwww.wiley.com. Wileyalsopublishesitsbooksinavarietyofelectronicformatsandbyprint-on-demand.Somecontentthat appearsinstandardprintversionsofthisbookmaynotbeavailableinotherformats. LimitofLiability/DisclaimerofWarranty Whilethepublisherandauthorshaveusedtheirbesteffortsinpreparingthiswork,theymakenorepresentations orwarrantieswithrespecttotheaccuracyorcompletenessofthecontentsofthisworkandspecificallydisclaim allwarranties,includingwithoutlimitationanyimpliedwarrantiesofmerchantabilityorfitnessforaparticular purpose.Nowarrantymaybecreatedorextendedbysalesrepresentatives,writtensalesmaterialsor promotionalstatementsforthiswork.Thefactthatanorganization,website,orproductisreferredtointhis (cid:2) (cid:2) workasacitationand/orpotentialsourceoffurtherinformationdoesnotmeanthatthepublisherandauthors endorsetheinformationorservicestheorganization,website,orproductmayprovideorrecommendationsit maymake.Thisworkissoldwiththeunderstandingthatthepublisherisnotengagedinrenderingprofessional services.Theadviceandstrategiescontainedhereinmaynotbesuitableforyoursituation.Youshouldconsult withaspecialistwhereappropriate.Further,readersshouldbeawarethatwebsiteslistedinthisworkmayhave changedordisappearedbetweenwhenthisworkwaswrittenandwhenitisread.Neitherthepublishernor authorsshallbeliableforanylossofprofitoranyothercommercialdamages,includingbutnotlimitedto special,incidental,consequential,orotherdamages. LibraryofCongressCataloging-in-PublicationDataappliedfor HardbackISBN:9781119417866 CoverdesignbyWiley Coverimages:CourtesyofMarcS.Paolella Setin10/12ptTimesLTStdbySPiGlobal,Chennai,India 10 9 8 7 6 5 4 3 2 1 (cid:2) Contents PREFACE xi PARTI ESSENTIALCONCEPTSINSTATISTICS 1 IntroducingPointandIntervalEstimation 3 1.1 PointEstimation / 4 1.1.1 BernoulliModel / 4 1.1.2 GeometricModel / 6 1.1.3 SomeRemarksonBiasandConsistency / 11 1.2 IntervalEstimationviaSimulation / 12 1.3 IntervalEstimationviatheBootstrap / 18 1.3.1 ComputationandComparisonwithParametricBootstrap / 18 1.3.2 ApplicationtoBernoulliModelandModification / 20 1.3.3 DoubleBootstrap / 24 1.3.4 DoubleBootstrapwithAnalyticInnerLoop / 26 1.4 BootstrapConfidenceIntervalsintheGeometricModel / 31 1.5 Problems / 35 2 GoodnessofFitandHypothesisTesting 37 2.1 EmpiricalCumulativeDistributionFunction / 38 2.1.1 TheGlivenko–CantelliTheorem / 38 2.1.2 ProofsoftheGlivenko–CantelliTheorem / 41 v vi CONTENTS 2.1.3 ExamplewithContinuousDataandApproximateConfidence Intervals / 45 2.1.4 ExamplewithDiscreteDataandApproximateConfidence Intervals / 49 2.2 ComparingParametricandNonparametricMethods / 52 2.3 Kolmogorov–SmirnovDistanceandHypothesisTesting / 57 2.3.1 TheKolmogorov–SmirnovandAnderson–DarlingStatistics / 57 2.3.2 SignificanceandHypothesisTesting / 59 2.3.3 Small-SampleCorrection / 63 2.4 TestingNormalitywithKDandAD / 65 2.5 TestingNormalitywithW2andU2 / 68 2.6 TestingtheStableParetianDistributionalAssumption:FirstAttempt / 69 2.7 Two-SampleKolmogorovTest / 73 2.8 Moreon(Moron?)HypothesisTesting / 74 2.8.1 Explanation / 75 2.8.2 MisuseofHypothesisTesting / 77 2.8.3 UseandMisuseofp-Values / 79 2.9 Problems / 82 3 Likelihood 85 3.1 Introduction / 85 3.1.1 ScalarParameterCase / 87 3.1.2 VectorParameterCase / 92 3.1.3 RobustnessandtheMCDEstimator / 100 3.1.4 AsymptoticPropertiesoftheMaximumLikelihoodEstimator / 102 3.2 Cramér–RaoLowerBound / 107 3.2.1 UnivariateCase / 108 3.2.2 MultivariateCase / 111 3.3 ModelSelection / 114 3.3.1 ModelMisspecification / 114 3.3.2 TheLikelihoodRatioStatistic / 117 3.3.3 UseofInformationCriteria / 119 3.4 Problems / 120 4 NumericalOptimization 123 4.1 RootFinding / 123 4.1.1 OneParameter / 124 4.1.2 SeveralParameters / 131 4.2 ApproximatingtheDistributionoftheMaximumLikelihoodEstimator / 135 4.3 GeneralNumericalLikelihoodMaximization / 136 CONTENTS vii 4.3.1 Newton–RaphsonandQuasi-NewtonMethods / 137 4.3.2 ImposingParameterRestrictions / 140 4.4 EvolutionaryAlgorithms / 145 4.4.1 DifferentialEvolution / 146 4.4.2 CovarianceMatrixAdaptionEvolutionaryStrategy / 149 4.5 Problems / 155 5 MethodsofPointEstimation 157 5.1 UnivariateMixedNormalDistribution / 157 5.1.1 Introduction / 157 5.1.2 SimulationofUnivariateMixtures / 160 5.1.3 DirectLikelihoodMaximization / 161 5.1.4 UseoftheEMAlgorithm / 169 5.1.5 Shrinkage-TypeEstimation / 174 5.1.6 Quasi-BayesianEstimation / 176 5.1.7 ConfidenceIntervals / 178 5.2 AlternativePointEstimationMethodologies / 184 5.2.1 MethodofMomentsEstimator / 185 5.2.2 UseofGoodness-of-FitMeasures / 190 5.2.3 QuantileLeastSquares / 191 5.2.4 PearsonMinimumChi-Square / 193 5.2.5 EmpiricalMomentGeneratingFunctionEstimator / 195 5.2.6 EmpiricalCharacteristicFunctionEstimator / 198 5.3 ComparisonofMethods / 199 5.4 APrimeronShrinkageEstimation / 200 5.5 Problems / 202 PARTII FURTHERFUNDAMENTALCONCEPTSINSTATISTICS 6 Q-QPlotsandDistributionTesting 209 6.1 P-PPlotsandQ-QPlots / 209 6.2 NullBands / 211 6.2.1 DefinitionandMotivation / 211 6.2.2 PointwiseNullBandsviaSimulation / 212 6.2.3 AsymptoticApproximationofPointwiseNullBands / 213 6.2.4 MappingPointwiseandSimultaneousSignificanceLevels / 215 6.3 Q-QTest / 217 6.4 FurtherP-PandQ-QTypePlots / 219 6.4.1 (Horizontal)StabilizedP-PPlots / 219 viii CONTENTS 6.4.2 ModifiedS-PPlots / 220 6.4.3 MSPTestforNormality / 224 6.4.4 ModifiedPercentile(Fowlkes-MP)Plots / 228 6.5 FurtherTestsforCompositeNormality / 231 6.5.1 Motivation / 232 6.5.2 Jarque–BeraTest / 234 6.5.3 ThreePowerful(andMoreRecent)NormalityTests / 237 6.5.4 TestingGoodnessofFitviaBinning:Pearson’sX2 Test / 240 P 6.6 CombiningTestsandPowerEnvelopes / 247 6.6.1 CombiningTests / 248 6.6.2 PowerComparisonsforTestingCompositeNormality / 252 6.6.3 MostPowerfulTestsandPowerEnvelopes / 252 6.7 DetailsofaFailedAttempt / 255 6.8 Problems / 260 7 UnbiasedPointEstimationandBiasReduction 269 7.1 Sufficiency / 269 7.1.1 Introduction / 269 7.1.2 Factorization / 272 7.1.3 MinimalSufficiency / 276 7.1.4 TheRao–BlackwellTheorem / 283 7.2 CompletenessandtheUniformlyMinimumVarianceUnbiasedEstimator / 286 7.3 AnExamplewithi.i.d.GeometricData / 289 7.4 MethodsofBiasReduction / 293 7.4.1 TheBias-FunctionApproach / 293 7.4.2 Median-UnbiasedEstimation / 296 7.4.3 Mode-AdjustedEstimator / 297 7.4.4 TheJackknife / 302 7.5 Problems / 305 8 AnalyticIntervalEstimation 313 8.1 Definitions / 313 8.2 PivotalMethod / 315 8.2.1 ExactPivots / 315 8.2.2 AsymptoticPivots / 318 8.3 IntervalsAssociatedwithNormalSamples / 319 8.3.1 SingleSample / 319 8.3.2 PairedSample / 320 8.3.3 TwoIndependentSamples / 322 8.3.4 Welch’sMethodfor𝜇 −𝜇 when𝜎2 ≠𝜎2 / 323 1 2 1 2 8.3.5 Satterthwaite’sApproximation / 324 CONTENTS ix 8.4 CumulativeDistributionFunctionInversion / 326 8.4.1 ContinuousCase / 326 8.4.2 DiscreteCase / 330 8.5 ApplicationoftheNonparametricBootstrap / 334 8.6 Problems / 337 PARTIII ADDITIONALTOPICS 9 InferenceinaHeavy-TailedContext 341 9.1 EstimatingtheMaximallyExistingMoment / 342 9.2 APrimeronTailEstimation / 346 9.2.1 Introduction / 346 9.2.2 TheHillEstimator / 346 9.2.3 UsewithStableParetianData / 349 9.3 NoncentralStudent’stEstimation / 351 9.3.1 Introduction / 351 9.3.2 DirectDensityApproximation / 352 9.3.3 Quantile-BasedTableLookupEstimation / 353 9.3.4 ComparisonofNCTEstimators / 354 9.4 AsymmetricStableParetianEstimation / 358 9.4.1 Introduction / 358 9.4.2 TheHintEstimator / 359 9.4.3 MaximumLikelihoodEstimation / 360 9.4.4 TheMcCullochEstimator / 361 9.4.5 TheEmpiricalCharacteristicFunctionEstimator / 364 9.4.6 TestingforSymmetryintheStableModel / 366 9.5 TestingtheStableParetianDistribution / 368 9.5.1 TestBasedontheEmpiricalCharacteristicFunction / 368 9.5.2 SummabilityTestandModification / 371 9.5.3 ALHADI:The𝛼-HatDiscrepancyTest / 375 9.5.4 JointTestProcedure / 383 9.5.5 LikelihoodRatioTests / 384 9.5.6 SizeandPoweroftheSymmetricStableTests / 385 9.5.7 ExtensiontoTestingtheAsymmetricStableParetianCase / 395 10 TheMethodofIndirectInference 401 10.1 Introduction / 401 10.2 ApplicationtotheLaplaceDistribution / 403 10.3 ApplicationtoRandomizedResponse / 403 10.3.1 Introduction / 403 10.3.2 EstimationviaIndirectInference / 406 x CONTENTS 10.4 ApplicationtotheStableParetianDistribution / 409 10.5 Problems / 416 A ReviewofFundamentalConceptsinProbabilityTheory 419 A.1 CombinatoricsandSpecialFunctions / 420 A.2 BasicProbabilityandConditioning / 423 A.3 UnivariateRandomVariables / 424 A.4 MultivariateRandomVariables / 427 A.5 ContinuousUnivariateRandomVariables / 430 A.6 ConditionalRandomVariables / 432 A.7 GeneratingFunctionsandInversionFormulas / 434 A.8 ValueatRiskandExpectedShortfall / 437 A.9 JacobianTransformations / 451 A.10 SumsandOtherFunctions / 453 A.11 SaddlepointApproximations / 456 A.12 OrderStatistics / 460 A.13 TheMultivariateNormalDistribution / 462 A.14 NoncentralDistributions / 465 A.15 InequalitiesandConvergence / 467 A.15.1 InequalitiesforRandomVariables / 467 A.15.2 ConvergenceofSequencesofSets / 469 A.15.3 ConvergenceofSequencesofRandomVariables / 473 A.16 TheStableParetianDistribution / 483 A.17 Problems / 492 A.18 Solutions / 509 REFERENCES 537 INDEX 561
Description: