IMF-稅收征管改革收益的量化（英）

QuantifyingtheRevenueYieldsfromTax

AdministrationReforms

HassanAdan,Jean-MarcB.Atsebi,NikolayGueorguiev,Jiro

Honda,andManabuNose

WP/23/231

IMFWorkingPapersdescriberesearchin

progressbytheauthor(s)andarepublishedto

elicitcommentsandtoencouragedebate.

TheviewsexpressedinIMFWorkingPapersare

thoseoftheauthor(s)anddonotnecessarily

representtheviewsoftheIMF,itsExecutiveBoard,

orIMFmanagement.

2023

NOV

N

Ar

R

*TheauthorsaregratefultoElenaD’Agosto,YongquanCao,SeydouCoulibaly,ElizabethGavin,JanGottschalk,AnamariaMaftei,AndrewMasters,AndrewOkello,PauloPaz,andMiguelPechofortheirinsightfulcommentsandtheusefuldiscussions.TheyalsothanktheparticipantsoftheIMF’sFiscalAffairsDepartmentseminars.WealsothankDhruvGandhiforexcellentresearch

assistance,andDáneaTrejoCarcamoandVictorMaradiagaforeditorialassistance.TheviewsexpressedherearethoseoftheauthorsandnotnecessarilythoseoftheIMF,itsExecutiveBoard,ormanagement.

?2023InternationalMonetaryFund

WP/23/231

IMFWorkingPaper

FiscalAffairsDepartment

QuantifyingtheRevenueYieldsfromTaxAdministrationReforms

PreparedbyHassanAdan,Jean-MarcAtsebi,NikolayGueorguiev,JiroHonda,andManabuNose*

November2023

IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicit

commentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseofthe

author(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.

ABSTRACT:Despitethecriticalityoftaxadministration(TA)reformsinenhancingdomesticrevenue

mobilization,fewstudieshaveattemptedtoquantifytherevenueimpactofsuchreforms.Thispaperfillsthis

gapbyestimatingtherevenueyieldsassociatedwithvarioustaxadministrationcapabilities,basedonthe

InternationalSurveyonTaxAdministration(ISORA),theTaxAdministrationDiagnosticAssessmentTool

(TADAT),andTAreformepisodesdatasets(identifiedbyAkitobyetal.,2020).ItusesaHausman-Taylor

cross-countrypanelregressionandaneventstudyforspecificTAreformepisodes.Ourresults(usingthe

ISORAdata)showthatanincreaseintheoverallstrengthofTAfromthe40thpercentiletothe60thpercentileisassociatedwithanincreaseintaxrevenueby1.8pp.ofGDP(witha95percentconfidencerangeof0.5-2.6pp.ofGDP).Similarly,theevent-studyassessmentshowsthatsustainedTAreformsledtoanincreaseintaxrevenuesbetween2to3pp.ofGDP,inlinewiththeexperienceinthreecountrycases(Jamaica,Rwanda,andSenegal).Also,therevenueyieldsareincreasingovertimetomorethan3pp.ofGDPafterthe6thyear

followingacomprehensivereform.Theanalysisalsohighlightsthesignificantimpactofspecificmeasures

including:i)strengtheningcompliancerisksmanagement,ii)enhancingpublicaccountability,iii)establishing

LargeTaxpayerOffices(LTO),iv)strengtheningaccountabilityandtransparency,andv)enhancingtimelyfilingoftaxdeclarations.

RECOMMENDEDCITATION:Adan,H.,Atsebi,J.,Gueorguiev,N.,Honda,J.,Nose,M.(2023).Quantifying

theRevenueYieldsfromTaxAdministrationReforms.IMFWorkingPaper,WP/23/231.InternationalMonetaryFund,Washington,DC.

JELClassificationNumbers:C1,E62,H2,H83,O23

Keywords:TaxRevenuesYields;TaxAdministration,ISORA,TADAT.

HAdan@IMF.org,JAtsebi@IMF.org,NGueorguiev@IMF.org,

Author’sE-MailAddress:

JHonda@IMF.org,

MNose@IMF.org

WORKINGPAPERS

QuantifyingtheRevenueYields

fromTaxAdministrationReforms

PreparedbyHassanAdan,Jean-MarcAtsebi,NikolayGueorguiev,Jiro

Honda,andManabuNose

IMFWORKINGPAPERSQuantifyingtheRevenueYieldsfromTaxAdministrationReforms

INTERNATIONALMONETARYFUND

2

Contents

I.Introduction

4

II.EstimatingtheImpactsofOverallStrengthofTaxAdministration

6

A.Data

6

B.StylizedFacts

7

C.EmpiricalMethodology

11

D.Results

13

1.OverallStrengthofTaxAdministration

13

2.SpecificComponentsofTaxAdministration:MultivariateModels

18

III.EstimatingtheImpactsBasedonReformEpisodes

19

A.DataandStylizedFacts

19

B.EmpiricalMethodology

21

C.Results

24

1.Difference-in-DifferencesinaStaggered-AdoptionDesign

24

2.SyntheticDifference-in-DifferenceswithCommonSupport

25

3.ControlledInterruptedTime-seriesAnalysisCaseStudies

27

a.Jamaica–ComprehensiveTAReforms

27

b.Rwanda–ComprehensiveTAReforms(LessLTOandSegmentation)

28

c.Senegal–ComprehensiveTAReforms(LessImprovingCompliance)

29

IV.ConcludingRemarks

30

AnnexI.DataDescriptionsandSources

32

AnnexII.DescriptiveStatisticsandStylizedFacts

33

AnnexIII.ResultsforMultivariateModels

35

AnnexIV:TheImpactofTaxAdministrationReformsinthePost-GFC

37

References

38

BOX

1.TheNewIMFFADTaxAdministration-AssessmentandYieldTool.............................................................18

FIGURES

1.ScatterPlotBetweenISORAIndexesAcrossVintages(2017vs.2014)

8

2.ComparisonsofTAStrength,Practices,andCharacteristicsAcrossIncomeGroupLevels

9

3.TaxRevenuesandISORAandTADATAggregateIndexes

10

4.ScatterPlotBetweenISORAandTADATAggregateIndexes

11

5.EstimatedRevenueYieldsfromanIncreaseoftheStrengthofTaxAdministration

14

6.Tax-to-GDPRatiobyCountryGroup

21

7.GroupTimeATT(Yieldin%)

24

8.DurationATT(Yieldin%)

25

INTERNATIONALMONETARYFUND

3

9.CountryGroupvs.ControlGroup

26

10.CITSAJamaicavs.ControlsAverage

28

11.CITSARwandavs.ControlsAverage

29

12.CITSASenegalvs.ControlsAverage

30

TABLES

1.EstimatedRevenueYieldsfromanIncreaseinTAStrengthfromthe40thtothe60thPercentile

14

2.BaselineResults,Overall,Strength,ISORA

16

3.BaselineResults,Overall,Strength,TADAT

17

4.TAReformsinEMsandLIDCsAftertheGlobalFinancialCrisis

20

5.AggregatedATE

25

ANNEXTABLES

AI.1.DataDescriptionsandSources

32

AII.1.SummaryStatistics

33

AII.2.CorrelationsBetweenISORAandTADATIndexesAcrossYears

33

AII.3.CorrelationsamongISORAandTADATIndexes,andbetweenISORAandTADATIndexes

34

AIII.1.MultivariateModels,ISORA

35

AIII.2.MultivariateModels,TADAT

36

ANNEXFIGURES

AIV.1.RevenueRatioIncreasevs.RevenueRatiopriorYear

37

AIV.2.Pre-vs.Post-GFCRevenue/GDPRatio

37

IMFWORKINGPAPERSQuantifyingtheRevenueYieldsfromTaxAdministrationReforms

INTERNATIONALMONETARYFUND

4

I.Introduction

1.Canwequantifytheimpactoftaxadministration(TA)reforms?Thispapertacklesthis

seeminglystraight-forwardbutchallengingquestionthroughasetofempiricalapproaches.Giventheneedtoenhancedomesticrevenuesforacountry’sdevelopment(particularlyindevelopingcountries),manycountrieshavebeenimplementingtaxpolicyandadministrationreforms.Criticalasthesetaxadministrationreformmeasuresare,thequantificationoftheirbenefitsinvolvessignificantchallenges.Toaddressthese

challenges,weproposeamethodologytomeasurethestrengthoftaxadministrationanditspractices,

althoughthedataontaxadministrationisquitelimitedforanalysesoveranextendedperiod.Wealsoemployanempiricalmethodologythataccountsforthepotentialendogeneityofthestrengthoftaxadministrationanditspracticestotaxrevenues.

2.VeryfewempiricalstudieshaveexploredtheimpactofTAreforms.

1

Ebekeetal.(2016),Crivelli(2019),andChangetal.(2020)conductedcross-countryanalysesontheestablishmentofLargeTaxpayersOffice(LTO),Semi-AutonomousRevenueAgency,thestrengthoftaxadministration,andtaxperformance.Further,therehasbeenanemergingmicroeconomicliteraturethatidentifiestheeffectivenessofTAreformsusinganexperimentaldesigninasingledevelopingcountrycontext,relatedtotheuseofthird-party

information(Pomeranz,2015;Carriloetal.,2017)andorganizationalreformsinstrengtheningtax

administration(Basrietal.,2021).

2

However,inacross-countrycontext,amorerobustempiricalframework(thataccountsforslow-movingfeaturesofTAindexesandtheendogeneityrelatedtoacountry’sTAstatus)isneeded.AdetailedexaminationofspecificreformepisodesisalsoneededtobetteridentifytherevenuegainsandpredicttheimpactofTAreformsovertime(i.e.,howlongafterthereformtheeffectstartsmaterializing).

3.Tofillthegapintheliterature,weemploytwodifferentmethodologiestoestimatetheyieldsfromTAmeasureswhileaddressingtheshortcomingsofearlierstudies.Theuseofthedifferent

methodologiesaimsatcross-checkingtheresults,whichmayleadtocomplementaryfindings.

.Inthefirstpart,thispaperappliesanempiricalstrategyusingcross-countrypaneldatawithslow-movingindicatorsofTAstrengththataccountsfortheendogeneityoftheTAreformimpact.

Specifically,theHausman-Taylorcross-countrypanelregressionwithendogenoustime-invariantvariable(ChatelainandRalf,2021)estimatesthecontributionofcharacteristicsoftaxagencytodomestictax

revenues,controllingforotherkeydeterminants(macroeconomic,structural,andtaxpolicychanges)asidentifiedinthepastliterature.

.Inthesecondpart,anevent-studydesignisusedtoquantifytheyieldsfrompastTAreform

episodesinLow-IncomeDevelopingCountries(LIDCs)andEmergingMarkets(EMs)withsustainedrevenuemobilizationeffortsintheaftermathoftheglobalfinancialcrisis(GFC)(Akitobyetal.,

2020).Thisanalysisemploysthreerecentextensionstothecanonicaldifference-in-differences(DID)to

1Cross-countrystudiesfocusedonmacroeconomicconditions(Tanzi,1992;Gupta,2007;IMF,2018),taxpolicy(Dabla-Norrisetal.,2020;NgomaandKrsic,2017;Akitobyetal.,2020),andgovernanceorsocio-politicalinstitutions(IMF,2019;Morrowetal.,2022)askeydeterminantsoftaxrevenues.

2OncorporationtaxationinIndonesia,Basrietal.(2021)examinedtheimpactofmore“intensive”taxadministration,throughthecreationofmedium-sizedtaxpayerofficesandincreasingtherevenuestaffing-to-taxpayersratioonactualtaxfilingsandpayments.Comparedwiththeelasticityoftaxableincomewithrespecttothecorporatetaxrate,theyfoundthatimprovedtaxadministrationcreatessignificantreturnsingeneratingrevenues(equivalenttoraisingtopCITratesonallfirmsbyeightpercentagepoints).

INTERNATIONALMONETARYFUND

5

estimatetherevenueyieldsfromTAreformsindifferentgroups,whileprovidinginsightsontheevolutionoftheyieldsintheyearsfollowingthereforms.

4.Ourfindingsarethreefold.

.First,improvementsintheoverallstrengthoftheTAareassociatedwithasignificantincreasein

taxrevenues.Specifically,wefindthatanincreaseinthestrengthofthetaxadministrationthatwould

moveacountryfromthe40thpercentiletothe60thpercentileinthedistributionofIMFmembercountriesbythisindicatorisassociatedwithanincreaseintaxrevenuesbyupto1.4percentagepoints(pp.)ofGDPforallcountriesand1.8pp.ofGDPforEMsandLIDCs.

3

ThesepointestimatesofrevenueyieldsarebroadlyconsistentwiththeestimatesinoureventstudyofspecificTAreformsaftertheGFC(i.e.,additional

revenuesbetween2tocloseto3pp.ofGDP).

4

.Second,whilesignificantTAreformsmayleadtosomeinitialgains,itcouldtaketimeforthefullresultstoberealized.Theevent-studyanalysisrevealsthattherevenueyieldsareincreasingovertimetomorethan3pp.ofGDPafterthe6thyearfollowingafar-reachingandcomprehensivereformacrossseveralpracticesoftaxadministration.

.Third,severalspecificTAmeasuresarestronglycorrelatedamongthemselvesandassociatedwithbettertaxperformance.UsingISORA,ourresultssuggestthatcomplianceriskmanagement

improvementshavethehighestrevenueyields.Inaddition,publicaccountabilityandLTOalsoplayamoresignificantroleinincreasingrevenueinEMsandLIDCs.WhileusingTADAT,wefindthataccountability

andtransparency,andtimelyfilingoftaxdeclarationsmatterthemostforraisingrevenues.Wealsofind

highcorrelationsamongISORAandTADATsub-indexes.ThissuggeststhatdifferentTAfunctionssupportandreinforceeachother.

5.ThispaperbroadenstheanalysisbyChangetal.(2020)andprovidesacomprehensive

databaseinbenchmarkingacountry’sstrengthofTAanditspracticesagainstcomparatorsacross

multipledimensionsandestimatingrevenueyields.Fromanoperationalperspective,thecomprehensivedatabaseandanalysis,aswellasanoperationaltooldevelopedinparallel

(Box1)

,helptoidentifygapsinTAandsystematicallyestimatethecontributionofpastorfutureTAreformsforIMFmembercountries.

6.Thispaperisorganizedasfollows:ThefirstpartanalyzestheroleofTAstrengthandpracticesontaxrevenuesusingtheISORAandTADATdatasets(seebelowfordatasetsdescription).Thesecondpart

studiestheimpactofTAreformsinthepost-GFCperiod.Ineachpart,SectionApresentsthedataandstylizedfacts,SectionBdiscussestheempiricalstrategy,andSectionCpresentsthefindings.Lastly,thepaper

concludesandprovidessomepolicyrecommendations.

3Inoursample,17countriesachievedanimprovementoftheirISORA’sOSindexthatishigherthananamountequivalenttothechangefromthe60thpercentiletothe40thpercentileduringtheperiod2014-17.ThosewiththehighestincreaseareGrenada,St.VincentandGrenadines,Uruguay,Greece,Ghana,Poland,andMyanmar.

4Giventhelimitedavailabilityofdata,theanalyseswithISORAandTADATdatadonotnecessarilycovertheperiodsofthereformepisodesinsectionIII(Jamaica,Rwanda,andSenegal).Basedontheepisodes,weassumethatthereformsdescribedinsectionIIIwouldbroadlycorrespondtoanincreaseinTAstrengthequivalenttothechangefrom40thpercentileto60thpercentile.

INTERNATIONALMONETARYFUND

6

II.EstimatingtheImpactsofOverallStrengthofTaxAdministration

A.Data

7.WeusetheInternationalSurveyonRevenueAdministration(ISORA)andtheIMF’sTax

AdministrationDiagnosticAssessmentTool(TADAT)datasetstomeasurecountries’overallstrength

oftheTA,aswellasthestrengthacrossthedifferentcharacteristicsandpracticesoftheTA.The

ISORAsurveyscountries’TAfeaturesthroughbothnumericalandcategoricalsurveyquestions(with982datapoints)onawidespectrumofpracticesandinstitutionalfeaturesagreedonbyfourpartnerinternational

organizations.

5

Thedataiscollectedthroughavoluntaryself-assessment,wherecountriesrespondtoan

onlinequestionnaire.Itcovers154countries,including36AdvancedEconomies(AEs),71EMs,and46LIDCs,andencompassesfourvintages.

6

However,weonlyusethefirsttwovintagesforouranalysis,ISORA2016

andISORA2018,giventhemajorrevisionofthesurveyforthesubsequentvintages.

7

TheISORAandTADATdataareincreasinglybeingusedindepth-researchonTA(seee.g.,Dabla-Norrisetal.,2020).

8.UsingtheISORAdata,wecloselyfollowChangetal.(2020)toconstructanoperational

strengthindexofTAandsevensub-indexesmeasuringthestrengthofspecificTApracticesand

characteristics.Thesevensub-indexescoveradministrativeandoperationalpracticesandstructural

foundations(laws,regulations,andpolicies)oftheTA.Theyencompass:i)complianceriskmanagement(CRM)approach,ii)theuseofthird-partydata,iii)digitalizationofservices,iv)serviceorientation,v)publicaccountability,vi)thedegreeofautonomy,andvii)theestablishmentofLTOorlargetaxpayerprogram.Inaddition,theyarecompiledastheequallyweightedaverageofthespecificquestionsrelatedtoeachsub-index.

8

,

9

Wealsocompileanoveralloperationalstrength(OS)indexastheequallyweightedaverageofthesevensub-indexes,whichmeasuresthestrengthoftheTAacrosspracticesandcharacteristics.

9.TheTADATdataprovidesinformationontherelativestrengthsandweaknessesofkey

componentsofacountry’sTAbyfocusingontheperformanceofthemajornationaltaxes.

10

Startingin2013,ithascovered86countries,with11repeatedassessmentsinthepre-pandemicperiod(i.e.,between

5TheISORAdatasetisthecollaborationoftheInter-AmericanCenterofTaxAdministrations(CIAT),theFiscalAffairsDepartment(FAD)oftheIMF,theIntra-EuropeanOrganisationofTaxAdministrations(IOTA),andtheOrganisationforEconomicCo-operationandDevelopment(OECD).Moreinformationaboutthesurvey,questionnaires,ISORAdataandrelatedpublicationscanbefoundfollowingthislink:

/?sk=BA91013D-3261-42F8-A931-A829A78CB1EC.

6TheinitialISORAdatasetcovers159countries,howeverduetosomemissingvalues,wecover154countriesinouranalysis.

7ThefivevintagesincludeISORA2016,ISORA2018,ISORA2020,ISORA2021andISORA2022coveringrespectively,theyears2014–15,2016–17,2018–19,2020,and2021.UntilISORA2021,thesurveywasrunbiennially.FollowingtheISORA2018,amajorrevisionofthequestionnairehasresultedinadiscontinuityofmanyquestions,asmallersetofquestionsaskedannually,and

additionalquestionstobeaskedeveryfouryears.ThismajorrevisionlimitsusfromusingtheISORA2020and2021vintagesforouranalysis.SeeCrandalletal.(2021)foracomprehensivedescriptionoftheresultsoftheISORA2018.

8FormoredetailsonthecomputationoftheISORAsub-indexes,pleaserefertoAnnexIinChangetal.(2020)ormodule3ofthedevelopedFADoperationaltoolTA-AYTthatprovidethedifferentquestionsusedtoconstructtheindexes.

9OtherdatafromtheISORAisalsoused.Thisincludes:i)theratiooftaxstaffoverthelaborforce;andii)theactivetaxpayersoverthelaborforce,measuringthehumanresourcecapacityandtaxpayers’baseoftheTA,respectively.

10TADATdatasetisnotpubliclyavailableforseveralcountries.DetailedinformationabouttheTADATdatasetandmethodologycanbefoundfollowingthislink:

/home#overview.

INTERNATIONALMONETARYFUND

7

2013and2019).

11

TADATassessmentfocusesonninekeyperformanceoutcomeareas(POAs)capturingthemostcriticalTAfunctions,processes,andinstitutions,including:i)integrityoftheregisteredtaxpayerbase,ii)effectiveriskmanagement,iii)supportingvoluntarycompliance,iv)timelyfillingofdeclarations,v)timely

paymentoftaxes,vi)accuratereportingindeclarations,vii)effectivedisputeresolution,viii)efficientrevenue

management,andix)accountabilityandtransparency.

12

TheseninePOAsarecompiledusingthe32high-levelindicators(HLIs),whichareintheirturnconstructedusing55measurementdimensions(MDs),followingthe

TADATaggregationM1andM2methodstoconstructthe32HLIs.

13

Wealsocompileanoverallperformanceassessmentreport(PAR)indexastheequallyweightedaverageoftheninePOAs,whichmeasuresthe

strengthoftheTAacrossallpracticesandcharacteristics.

10.Otherdataforempiricalanalysesincludethetax-to-GDPratioandstandardmacroeconomic

controls.Asadependentvariable,thetax-to-GDPratio(excludingtradetaxesandsocialsecurity

contributions)isused,whichiscompiledfrommultiplesources.Weexcludetradetaxesandsocialsecurity

contributionstoaccountforthevolatilityofcommoditypricesandexcluderesourcerevenues,andbecause

otheragenciesthantheTAmaycollectthoserevenues.Wealsouseseveralcontrolvariablescomingfrom

differentsources.Theyarediscussedinthemethodologysection,andtheirdatadescriptionsandsourcesarepresentedinAnnexI.

B.StylizedFacts

11.ThissectionpresentsseveralstylizedfactsontheTAstrength,itspractices,and

characteristicsusingISORAandTADATdatasets.ItdiscussesthedifferencesbetweenISORAand

TADATaggregateindexesandthecorrelationsamongindexesovertimeandbetweenthemselves.It

alsoprovidescomparisonsofTAstrength,practices,andcharacteristicsacrossincomegrouplevels,aswellasaforetasteoftherelationshipbetweenTAstrengthandtaxrevenues.Thesummarystatisticsoftheindexes

constructedusingISORAandTADATdatasetsarepresentedinAnnexII.Ourmainfindingsarethefollowing:

a)ThestrengthofTAgenerallychangesonlyslowlyovertime,withsomeexceptions.Lookingatthe

changesoftheISORAandTADATindexesovertime(acrossvintagesandrepeatedassessments,

respectively),weshowthatTAstrength,practices,andcharacteristicsonlyslightlychangedoverthe

years––theyareslow-movingvariables(seeillustrationin

Figure1

forISORA).ItindicatesthatittakestimeaswellasstrongandsustainedeffortstoimproveTAstrength.

TableAII.

showsthatthecorrelations

betweenISORAindexesfor2014and2017areelevated.Forinstance,eightoutofthetenISORA’sindexeshaveacoefficientofcorrelationhigherthan0.8.

14

b)ThestrengthofTA,itspractices,anditscharacteristicsvarysubstantiallywiththelevelsof

economicdevelopment.Theresultsofthecomparisonacrossincomegrouplevelsarepresentedin

11Thisisaverydynamicdatabase.Thenumberofassessmentshasnowincreased(seetheexactnumberasoftoday:

/home)

.However,sinceourfocusisontheperiod2010-2019,wedonotincludeanyoftheassessmentsafter2020(post-pandemicperiod).

12TADAT’stimelyfilingandtimelypaymentscoresfactorinthecredibilityofthedataprovided:ifregardedasnotcredible,

administrationsreceivethelowestscore.i.e.,thesearenotonlyameasureoftheactualtimelinessoffiling,buttheyarecombinedwithanassessmentofwhethertheadministration’srecordkeepingonfilingandpaymentisreliable.

13PleasefollowthislinkforthedescriptionoftheM1andM2methods:

/tadatAtAGlance.

14Onexceptionalbasis,therewereafewcountrieswhichdemonstratedsignificantimprovement.Thoseare:Grenada,St.VincentandGrenadines,Uruguay,Greece,Ghana,Poland,andMyanmar.

INTERNATIONALMONETARYFUND

8

Figure2.

TheTAstrength,asmeasuredbytheISORAandTADATindexes,isonaveragemuchstrongerinAEs,whileLIDCshavetheweakestTA(PanelA).ThesefindingspointtothepotentialscopeforLIDCsandEMstofurtherimprovethestrengthofTAacrossallpracticesandcharacteristics.

Figure1.ScatterPlotBetweenISORAIndexesAcrossVintages(2017vs.2014)

2017ISORA'sOS

1

.8

.6

.4

.2

AEsEMsLIDCs45deg.line

0.2.4.6.81

2014ISORA'sOS

Sources:Authors’calculationsusingISORA.

.ForISORA,AEshavehigherscoresthaninEMsandLIDCs,exceptfortheestablishmentofLTO.

15

EMsalsohavehigherscoresonaveragethanLIDCs,exceptforautonomyandestablishmentofLTO,

wheretheyhavealmostsimilarscores.InEMsandLIDCs,theweaknessesaremoreaccentuatedfortheuseofthird-partydataanddigitalization,andcomplianceriskmanagement.

.ForTADAT,thedifferencesinTAperformancebetweenAEs,EMs,andLIDCsaremore

pronounced.AEshavehigherscoresacrossallpracticesandcharacteristics(inparticularfortimelyfilingoftaxdeclarationsandtimelypaymentsoftaxes),whicharealsohigheronaveragethaninEMsandLIDC.Also,EMshavehigherscoresonaveragethanLIDCsacrossallpracticesandcharacteristics.However,

theyhaveweakerscoresforeffectiveriskmanagement,accuratereportingindeclarations,timelyfilingoftaxdeclarations,andefficientrevenuemanagement.LIDCshavethelowestscoresonaverageacrossallpracticesandcharacteristics,especiallyfortimelyfilingoftaxdeclarations,efficientrevenuemanagement,accuratereportingindeclarations,andeffectiveriskmanagement.

.Notsurprisingly,thestrengthofTAandtaxrevenuesarepositivelycorrelated.Specifically,wefindthatthecorrelationbetweenISORAandTADATindexesandtaxrevenues(excludingtradetaxesand

socialcontributions)ishigh,althoughthereisahighvarianceamongpeergroupcountries

(Figure3)

.

c)Therearesignificantheterogeneitiesbyincomegroups.AEstendtocollectmoretaxrevenueswithstrongerTA.Ontheotherhand,LIDCshave,onaverage,weakerTAandtendtocollectlowertax

revenues.TherearesignificantlylargevariationsinTAstrengthinEMs,wheresomecountrieshaveweakerTAsimilartoLIDCswhileothershavestrongerTAsimilartoAEs.

15TheLTOindicatorreflectsonlythepresenceofanLTO,notnecessarilytheperformanceorefficiencyoftheoffice/program.

Anotherreasonisthatsomemoremoderntaxadministratio