Unveiling the Informal Economy: An Augmented Factor Model Approach
UnveilingtheInformalEconomy AnAugmentedFactorModelApproach JiaxiongYao WP/24/110 IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate. TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement. 2024 MAY ©2024InternationalMonetaryFundWP/24/110 IMFWorkingPaper EuropeanDepartment UnveilingtheInformalEconomy:AnAugmentedFactorModelApproachPreparedbyJiaxiongYao* AuthorizedfordistributionbyJamesWalsh May2024 IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement. ABSTRACT:Thispaperdevelopsanewapproachtoestimatingthedegreeofinformalityinaneconomy.Itcombinesdirectyetinfrequentmeasuresoftheinformaleconomyinmicrodatawithanaugmentedfactormodelthatlinksmacroindicatorsoftheinformaleconomytoitscauses.Weshowthattheprevailingmodelusedintheliterature,themultipleindicatorsmultiplecausesmodel,isaspecialcaseoftheaugmentedfactormodelanddepictsanincompletepictureoftheinformaleconomy.Usingtheaugmentedfactormodelapproach,weshowthatthedynamicsoftheinformaleconomyisshapedbythestrengthofoveralleconomicactivityaswellastheinterplaybetweentheformalandinformaleconomies.Contrarytopreviousworkthattypicallyfindsdeclininginformalityformostcountries,wefindthatthedegreeofinformalityhasincreasedforlow-incomecountriesforthepasttwodecades. RECOMMENDEDCITATION: JELClassificationNumbers: E26,C38,O11 Keywords: Informality;augmentedfactormodel;MIMICmodel;surveydata Author’sE-MailAddress: JYao@imf.org *TheauthorthanksShuYu,YuanLiao,JamesWalsh,andseminarparticipantsattheIMFforhelpfuldiscussions.TheviewsexpressedherearethoseoftheauthoranddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement. ContentsPage I.Introduction4 II.LiteratureReview7 III.MeasuringtheInformalEconomy:AnAugmentedFactorModelApproach8 IV.Data12 A.EnterpriseSurveyData12 B.OtherSurveyData13 C.CausesandIndicators14 V.UnveilingtheInformalEconomy16 A.ProjectedPrincipalComponentAnalysis16 B.EstimatesoftheDegreeofInformality:CountryExamples20 C.InformalEconomy:CausesandPatterns24 VI.Conclusion26 References28 Appendices A.TheMIMICModel31 B.EstimationoftheAugmentedFactorModel32 C.RelationshipbetweenWBESandOtherSurveyData33 D.AddingNighttimeLights33 E.CountryGroups36 ListofTables 1.CountryandYearCoverageoftheWBES14 2.CausesandIndicatorsoftheInformalEconomy16 3.ProjectionofIndicatorsonCauses18 4.RelationshipbetweenEstimatedFactorsandIndicators18 5.SurveyDataandEstimatedFactors20 6.EstimatedWeightsonProjectedIndicators20 7.EstimatedInformalEconomyandCauses25 8.SurveyDataandEstimatedFactorswithNighttimeLights35 9.ListofCountrieswithCauses,Indicators,andWBESData36 10.ListofCountriesinEachIncomeGroup37 ListofFigures 1.EstimatedDegreeofInformalityfromtheWBES15 2.ProjectedPrincipalComponentAnalysis19 3.DegreeofInformality:SelectedCountries21 4.DecompositionoftheDegreeofInformalitybyContributionofFactors:Se- lectedCountries22 5.DecompositionoftheDegreeofInformalitybyContributionofIndicators: SelectedCountries23 6.MedianDegreeofInformality26 7.DegreeofInformality:2002vs.202126 8.DegreeofInformalityinSurveyData:WBESvs.LaborSurveys34 9.ProjectedPrincipalComponentAnalysiswithNighttimeLight35 10.EstimatedDegreeofInformalitywithandwithoutNighttimeLight35 I.INTRODUCTION Theinformaleconomyisasmysteriousasitisimportant:mysterious,becauseitssizeisoftenunfathomable—evenitsdefinitionvarieswidelyacrossdifferentstudies;important,becauseshoulditbeformalized,itwouldpotentiallyraiseeconomicgrowth,boostgov-ernmentrevenue,andimprovesocialwelfare. Existingstudiesthatestimatethesizeoftheinformaleconomyhavebroadlyfollowedtwoapproaches.Thefirstisanempiricalapproachthatattemptstoobtainestimatesdirectlyfrommicrodata.Theobviousdownsideofthisapproachisthatmicrodataareonlyavail-ableinfrequently,makingitdifficulttoassessthedynamicsoftheinformaleconomy.Thesecondisastatisticalmodelingapproachthattreatstheinformaleconomyasalatentvari-ableandtriestoestimateitfrommacrodataofvariouscausesandindicators.Suchanap-proachisalsonotwithoutlimitations.Akeyproblemisthatthelatentvariableissubjecttomanydifferentinterpretations,renderingitsexactmeaningabstruse.Inaddition,thees-timatesofthisapproacharesensitivetomodelspecifications,benchmarkcalibration,andsamplecoverage. Inthispaper,wedevelopanewapproachtoestimatingthedegreeofinformalityinaneconomy,marryingdirectmeasuresfromsurveysandinsightsfromastatisticalmodelandcombiningmicroandmacrodata.Wecallthisapproachtheaugmentedfactormodelapproach,buildingonrecentresultsonaugmentedfactormodelsfromFan,Ke,andLiao(2021).Itconsistsoftwoparts:onepartisafactormodelofindicatorsoftheinformaleconomyaugmentedbyitsobservablecauses;theotherisafunctionthatmapstheesti-matedfactorstodirectmeasuresinsurveydata.Theaugmentedfactormodelsummarizesthechannelsthroughwhichthecausesofth
