世界銀行-數(shù)字支付與新冠肺炎沖擊：現(xiàn)有條件在銀行、基礎設施、人力能力和數(shù)字監(jiān)管中的作用（英）

PublicDisclosureAuthorizedPublicDisclosureAuthorized

PolicyResearchWorkingPaper10603

DigitalPaymentsandtheCOVID-19Shock

TheRoleofPreexistingConditionsinBanking,

Infrastructure,HumanCapabilities,andDigitalRegulation

RobertCull

VivienFoster

DeanJolliffe

DanielLederman

DavideSalvatoreMare

MalarvizhiVeerappan

WORLDBANKGROUP

DevelopmentResearchGroup

DevelopmentDataGroup&

MiddleEastandNorthAfricaRegionNovember2023

ReproducibleResearchReository

Averifiedreproducibilitypackageforthispaperisavailableat

,click

here

fordirectaccess.

PolicyResearchWorkingPaper10603

Abstract

Treatingdatacollectedpre-andpost-COVID-19asaquasi-experiment,thispaperexaminestheimportanceofpresumedenablersandsafeguardsindrivingtheobservedexpansionofdigitalpaymentsanddigitalfinancialinclu-sion.Theanalysisinteractsdriversofdigitalpaymentusagewithacountry-specificproxyoftheseverityoftheCOVID-19shock,leveragingvariationinboththedriversandthequasi-treatment(theCOVID-19shock)toidentifytheparameters.Althoughregulationofbanksanddigitaleconomicactivitywerecorrelatedwithdigitalpayments

beforeandduringthepandemic,thecapabilitiesofusersandconnectivity(toelectricity,theinternet,andmobiletelephony)wereresponsibleforincreaseduseofdigitalfinancialservicesinresponsetotheshock.Aninterpreta-tionisthatgovernmentsandtheprivatesectorwereabletoovercomeunderdevelopedbankingsystemsandweakregulationofthedigitaleconomy,butonlywheretherewasadequatedigitalinfrastructure,connectivity,andahighshareofthepopulationthatunderstoodandcouldmakeuseofdigitalpayments.

ThispaperisaproductoftheDevelopmentResearchandDataGroups,DevelopmentEconomicsandtheOfficeoftheChiefEconomist,MiddleEastandNorthAfricaRegion.ItispartofalargereffortbytheWorldBanktoprovideopenaccesstoitsresearchandmakeacontributiontodevelopmentpolicydiscussionsaroundtheworld.PolicyResearchWorkingPapersarealsopostedontheWebat

/prwp.Theauthorsmaybecontactedatrcull@

.Averifiedreproducibilitypackageforthispaperisavailableat

,click

here

fordirectaccess.

ThePolicyResearchWorkingPaperSeriesdisseminatesthefindingsofworkinprogresstoencouragetheexchangeofideasaboutdevelopmentissues.Anobjectiveoftheseriesistogetthefindingsoutquickly,evenifthepresentationsarelessthanfullypolished.Thepaperscarrythenamesoftheauthorsandshouldbecitedaccordingly.Thefindings,interpretations,andconclusionsexpressedinthispaperareentirelythoseoftheauthors.TheydonotnecessarilyrepresenttheviewsoftheInternationalBankforReconstructionandDevelopment/WorldBankanditsaffiliatedorganizations,orthoseoftheExecutiveDirectorsoftheWorldBankorthegovernmentstheyrepresent.

ProducedbytheResearchSupportTeam

DigitalPaymentsandtheCOVID-19Shock:

TheRoleofPreexistingConditionsinBanking,Infrastructure,HumanCapabilities,

andDigitalRegulation

RobertCull,VivienFoster,DeanJolliffe,DanielLederman,DavideSalvatoreMare,and

MalarvizhiVeerappan

JELCodes:G5HouseholdFinance;G53FinancialLiteracy;H12CrisisManagement;C26;InstrumentalVariables(IV)Estimation

Keywords:digitalpayments,financialinclusion,COVID-19,GlobalFindexDatabase

2

1.Introduction

TheCOVID-19pandemicwasashockthatprovidedimpetusforincreasedusageofdigitalpaymentservices,andperhapsspurreddeeperfinancialinclusion.Infact,between2017and2021,theshareofadultsreportinghavingusedtheinternettomakeorreceivedigitalpaymentsinthepreviousyearrose,onaverage,by7.7percentagepoints.However,thestandarddeviationofthesechangesisalsohigh,at8.5percentagepoints,indicatingsubstantialvariationacrosscountries.Becauseitwasunanticipatedandbecausecountrieshaddifferingresponsesintermsofincreaseduseofdigitalpayments,theepisodeprovidesanopportunitytoidentifythedriversofdigitalpaymentsanddigitalfinancialinclusion.Usingthe2021roundoftheGlobalFindexsurveyandothernewdatasetsdescribingbankregulationandsupervisionandlegalandregulatoryenablersandsafeguardsfordigitalcommerceand(personal)datasharing,wemeasuretheinfluenceondigitalpaymentsofdimensionsthathavenotbeenquantifiedwelltodate,whilealsomeasuringtheinfluenceofdriverssuchashumancapabilities(educationalattainment),infrastructure

(electricity,internetusage),andtheusageofmobilephones.

Theliteratureontheeffectsofdigitalpayments,anddigitalfinancialservices(DFS)ingeneral,isgrowingandsurveysofitwereavailablepriortotheCOVID-19pandemic(Karlanetal.,2016;AbbasiandWeigand,2017)andinthecontextofthepandemic(Aguretal.,2020).Keythemesthatemergefromtheliteratureonimpactevaluation,muchofwhichisfocusedonfieldexperiments(i.e.,randomizedcontrolledtrialsor‘RCTs’)arethat:(1)digitalpaymentsandtransferscanstrengtheninformalrisk-sharingnetworks,providingasourceofresiliencetounanticipatedshocksthatcancontributetohigherhouseholdsavingsandconsumption(Leeetal.,2021;JackandSuri,2011,2014,2016;Surietal.,2012);(2)digitalsystemsareprovingacost

effectiveandsecuremethodforgovernmentstodisbursefundstocitizensinmultiplecontexts

3

(Akeretal.,2016;Banerjeeetal.,2015)andforprivatefirmstopaytheirworkers(Breza,Kanz,andKlapper,2020;Blumenstock,Callen,andGhani,2018;Blumenstocketal.,2015);

1

and(3)DFSalsoholdpromisefordeliveryofsavingsandcreditproducts(Karlanetal.,2016),thoughevidenceonnon-digitaldeliverymechanismsindicatesthatcommitmentdevicesand/orpromptsthatguideclientsintheirfinancialdecision-makingmayberequiredtospurhouseholdsavings(Ashrafetal.,2010;DupasandRobinson,2013;Beamanetal.,2014;Dupasetal.,2018;Prina,2015;Bruneetal.,2016)andtheexperimentalevidencehasnotshownsubstantialimpactsofmicrocreditonhouseholdincome,consumption,orentrepreneurialactivities(Banerjee,Karlan,

andZinman,2015).

AsmallbutgrowingliteraturehasexaminedtheeffectsofDFSonfirmgrowthandprofitabilityusingnon-experimentalempiricalmethods,thoughresearchpriortothepandemicfocusedonDFSprovisionthroughbankstotheexclusionofmobilenetworkoperatorsandfintechfirmsandtendedtocoveralimitedrangeofissues(AbbasiandWeigand,2017).

2

Duringthepandemic,greateremphasiswasplacedonDFSasaquick,effectivewayforgovernmentstodisbursefundstothoseinneedandforhouseholdsandfirmstoaccessonlinepaymentsandfinancing.WorldBank(2022)reportsthatmanydevelopingcountrieswereabletousedigitalcashtransfersduringthepandemictomitigatetheadversedistributionaleffectsonpoverty.Asoneexample,theynotethatBrazilremarkablymanagedtoreducepoverty,inpartthroughtargeted

digitalcashtransfers,eveninthepresenceofeconomiccontraction.Similarly,helpedbydigital

1RecipientsofdigitalwagepaymentsinBangladeshandAfghanistanalsosavedmorethanthosewhoreceivedwagesincash,likelyduetogreatersecurityandprivacythatdigitalwagepaymentsprovide,especiallytowomen(Breza,Kanz,andKlapper,2020;Blumenstock,Callen,andGhani,2018).AndthereissomeevidencesuggestingthatreceivingdigitalpaymentscanprovideanonrampforrecipientstolearnhowtobetteruseDFSingeneral.Forexample,inBangladesh,factoryworkerswhoreceivedtheirwagesdirectlyintoanaccountalsolearnedtousetheiraccountwithoutassistanceandtoavoidillicitwithdrawalfees(Breza,Kanz,andKlapper,2020).

2AbassiandWeigand(2017)notethat“newerresearchers[inthisarea]oftenignorepastliteratureandinvestigatethesameissues.”

4

transfers,Indiawasabletoprovideaidduringthepandemicthatreachedthemajorityofurbanandruralhouseholds.Despitedocumentedsuccesses,observershavecautionedagainstrapidscalingupofDFSwithoutadequateregulationsandsafeguardsandhaveacknowledgedthepotentialfor

DFStoincreaseinequality(Aguretal.,2020).

RegardingthedriversofDFSpriortoCOVID-19,evidencefromsmallersamplesofcountriesthantheoneusedhereindicatesthatmultiplefactorswereatwork.

3

Focusingondigitalcreditprovidedbyfintechfirmsforasampleof61countries,Claessensetal.(2018)findthatfintechcreditvolumeswereincreasingwithpercapitaincomeanddecliningwiththecompetitivenessofthebankingsector,presumablybecausefintechfirmsfoundithardertocompetewithbanks.Fintechcreditvolumesalsodeclinedwiththestringencyofbankingregulation,perhapsbecause“fintechregulationsarealsomoreliberalinjurisdictionswherebankingregulationismoreliberal,”orbecausenewlendingactivitiesmaybehardertoundertakeincountrieswithstrictprudentialandbanklicensingregimes.

4

Forasampleof52countriesin2014and2017,Sahayetal.(2020)findthattheprevalenceofdigitalpaymentswaspositivelylinkedtoGDPgrowth.Basedontheexistingliteratureandinterviewswithpolicymakers,regulators,fintechcompanies,andbanks,thoseauthorsalsopositthatthesafedevelopmentofdigitalfinancialinclusionisinfluencedbythefinancial/digitalliteracyofDFSusersand

regulationsanddigitalsystemsthatinstilltrustthroughdigitalidentification,consumerprotection,

3EffortstoidentifythedriversoffinancialinclusionmorebroadlywereforerunnersofthesemorerecentstudiesondriversofDFS.Forexample,Rojas-Suárez(2016)andRojas-SuárezandAmado(2014)showthatpoorinstitutionalquality(asreflectedinweakruleoflawandalackofreliablecontractenforcement)anduncompetitivebankingsectorswereassociatedwithlessfinancialinclusioninLatinAmerica(measuredasowninganaccountataformalfinancialinstitution).UsingGlobalFindexdatafor2011and2014,Dabla-Norrisetal.,(2015a,b)alsofindfinancialinclusiontobepositivelylinkedtopercapitaincome,ruleoflaw,andcompetitioninbanking.Similarly,usingFindexdatafrom2014,Deléchatetal.(2018)showthatindividualsweremorelikelytobefinanciallyincludedincountrieswith

highlevelsofpercapitaincomeandgeneralfinancialdevelopment.

4Claessensetal.(2018),p.38.

5

andcybersecurity.

5

Weattempttocontrolforthefactorsfromthesestudiesintheregressions

presentedbelow.

AsmallerliteratureexploresthedriversofDFSgrowthduringtheCOVID-19pandemicitself.BasedonahouseholdsurveyinIndia,thosewhoswitchedtousingdigitalpaymentsduringthelockdownweresignificantlymorelikelytobeawareofdigitalpaymentsmodes,morehighlyeducated,andwithaccesstosmartphonesanddebitcards(Saroyetal.,2022).ThesearesimilartocharacteristicsofDFSusersinstudiespriortothepandemicwhowerefoundtobericher(CohenandRysman,2013;FujikiandNakashami,2019),bettereducated(Koulayevetal.,2016),and

moreawareofdigitalpaymentsandgenerallymorefinanciallyliterate(Wyman,2017).

6

OtherdriversofDFSadoptiondifferedacrossthepre-pandemicandpandemicperiods.Forexample,andperhapsnotsurprisingly,theincreaseindigitalpaymentsduringthepandemicdependedonsocialdistancingnormsandcontainmentmeasures(De’etal.,2020,AlberandDabour,2020).WhilestudiesthatpredatedthepandemicfoundthatyoungerrespondentsweremorelikelytobeDFSusers,Saroyetal.(2022)foundthatduringthepandemicmiddle-agedsurveyrespondentsinIndiawerethemostlikelytohavebecomeDFSusers,andthatmanyofthemhadearliertriedbutsubsequentlyabandoneddigitalpayments.Similarly,Jonkeretal.(2022)foundthatdigitalpaymentsincreasedmoresharplyduringthepandemicforolderagegroups,

whichhadtypicallybeenslowertoadoptDFS.

5OnapproachestoregulatingDFS,seeStaschenandMeagher(2018),GutierrezandSingh(2013),andTaraziandBreloff(2010).GutierrezandSingh(2013)alsoprovideevidenceofapositiveassociationbetweenmobilemoneyusageandaspectsoftheregulatoryregimerelatedtoe-contracting,consumerprotection,andinteroperability,amongotherfactors,forasampleofthirty-fivecountries.Onthebenefitsofdigitalidentificationsystems,andbiometricidentificationinparticular,incontextswherehumancapabilities(illiteracyandlowlevelsoffinancial/digitalliteracy)constrainfinancialinclusion,seeMuralidharan,Niehaus,andSukhtankar(2016).

6IndividualeducationandincomewerealsosignificantlyassociatedwithbroadermeasuresoffinancialinclusioninstudiesthatusedFindexdata(see,e.g.,Allenetal.,2016;Deléchatetal.,2018).

6

ThisdemographicshiftinDFSusagewasfurtheredbygovernmentprogramstotransfercashtosupportstrugglinghouseholdsduringthepandemic.Forexample,thosedependentonIndia’sDirectBenefitTransfer(DBT)-basedincomesupportprogramswerecompelledtouseDFStoaccesstheirentitlements(Saroyetal.,2022).Moregenerally,Gentilinietal.(2020)reportthatnearly17%oftheworld’spopulationwascoveredbyatleastonegovernmentCOVID-19-relateddigitalcashtransferschemein2020-2021.Becausegovernmentsneededtodepositthesetransfersintoaformalaccount,increaseduseofdigitalpaymentswasalsoassociatedwiththepresenceofformalfinancialinstitutions.IntheIndiancontext,forexample,proximitytobrick-and-mortarbankingestablishmentswassignificantlylinkedtodigitalpaymentadoptionduringthepandemic(Saroy,2022).AndtheFindex2021datarevealthat39percentofadultsindevelopingeconomiesopenedtheirfirstfinancialaccountforthepurposeofreceivingadirectgovernmentpayment(suchasawageorbenefitpayment)orawagepaymentfromaprivate-sectoremployer,andthatmostofthoseaccountswerewithbrick-and-mortarfinancialinstitutions(Demirguc-Kuntetal.,2022).Finally,thereisalsoevidencethatthepandemic-inducedincreaseinuseofdigitalpaymentspersistedaftertheinitialwavesofCOVID-19hadsubsided(Jonkeretal.,2022;Ardizzietal.,

2020).

DespitetheproliferationofstudiesontheroleofdigitalpaymentsduringCOVID-19,whathasbeenlackingissystematicglobaldataondriversofDFSusageand,untilthereleaseofthe2021GlobalFindexdata,onDFSusageitselfduringthepandemic,andaccompanyingempiricalanalysistomeasuretherelativeimpactsofthosedrivers.

7

Thispaperisanattempttobeginfillingthisresearchgap.Wefirstestablishthewidevariationacrosscountriesduringthe

COVID-19periodintheincreaseindigitalpaymentsandusageofdigitalfinancialservices.Our

7Sahayetal.(2020)notethat,“tothebestofourknowledge,therearenocomprehensiveglobalstudiesonfintechandfinancialinclusion,reflectinginpartthelimitedavailabilityofcross-countrydata.”

7

estimationapproachisthenpredicatedontheassumptionthattheshockwasunanticipatedandthatitseffectsondigitalparticipationdependedheavilyonconditionsontheeveofthepandemiconthedimensionsmentionedabove–bankingdevelopmentandactivityrestrictions,legalandregulatoryenablersandsafeguardsfordigitalcommerceanddatasharing,humancapabilitiesof

DFSusers,infrastructuredevelopment,andmobilephoneusage.

Wefindthatbankingsectordevelopmentandlegalandregulatoryenablersandsafeguardsforparticipationinthedigitaleconomyareassociatedwithdigitalpaymentusage.However,thesefactorsarestronglycorrelatedwiththelevelofDFSusageinboth2017and2021,ratherthanwiththechangeinDFSusageduringtheCOVID-19period,andtheassociationwithbankingsectordevelopmentissignificantlyweakerincountrieswhereregulationsonbanks’activitiesaremorerestrictive.Bycontrast,proxiesforinfrastructureandconnectivity(forexample,theshareofthepopulationwithaccesstoelectricity)andmobilephoneusagearebothcorrelatedwiththeincreaseinDFSusageduringthepandemic,andlessstronglyassociatedwiththelevelofDFSusagebeforeorafterthepandemic(i.e.,in2017or2021).HumancapabilitiesasproxiedbymeanyearsofschoolingarestronglyassociatedwiththelevelofDFSusage(in2017and2021),butalsosignificantlycorrelatedwiththechangeinDFSusageduringthepandemic.Similarly,theshareofpopulationthatusestheinternet,whichislikelyreflectiveofbothconnectivityandhumancapabilities,issignificantlylinkedtoboththepre-andpost-pandemiclevelsofDFSusageandto

theincreaseinDFSusageduringthepandemic.

BecausetheCOVID-19shockwasunanticipated,weviewthefactorsthataresignificantlycorrelatedwiththechangeinDFSusageasmorelikelytosummarizecausalrelationshipsthanthosemorestronglyassociatedwithpre-andpost-pandemiclevelsofDFSusage.However,and

attheriskofreadingtoomuchintoourestimates,thepatternssuggestthatlonger-rundevelopment

8

ofbankingandthelegalandregulatoryenvironmentfordigitaleconomicactivityhavebeenimportantcontributorstoDFSusage,butthatthefactorsmostcloselyrelatedtothecapabilitiesofusersandconnectivity(toelectricity,theinternet,andmobiletelephony)wereprimarilyresponsibleforincreasingDFSusagebecauseoftheCOVID-19shock.Oneinterpretationisthatgovernmentsandtheprivatesectorwereabletoworkaroundunderdevelopedbankingsystems,andpoorlawsandregulationsofthedigitaleconomy,butonlyiftherewasanadequatelevelofdigitalinfrastructureandconnectivityandahighshareofthepopulationthatunderstoodandcould

makeuseofdigitalpayments.Wereturntothisdiscussionlaterinthepaper.

Therestofthepaperisorganizedasfollows.Section2describesthedatasourcesforDFSusageanditsdriversandthespecificvariablesthatweuseintheempiricalanalysis.Section3describesourestimationapproachingreaterdetail,whilesection4summarizesourempiricalresults.Section5discussestheempiricalpatternsthatwefindandpotentialpolicyimplications.

Section6offersconcludingremarksandlaysoutpotentialdirectionsforfutureresearch.

2.Data

Fortheempiricalanalysis,werelyoncountry-leveldatasetswithwideglobalcoveragethatsummarizeDFSusage,bankingregulation,humancapitaldevelopment,andregulationofdigitalcommercialactivityanddatasharing.Inthissection,webrieflydescribethesourcesof

thosedataandthespecificvariablesthatwerelyonfromeachsource.

GlobalFindexDatabase

TheFindexdatabaseprovidesinformationonhowadultsaroundtheworldusefinancial

services,frompaymentstosavingandborrowing,andhowtheymanagefinancialevents.

8

The

8FordetailsonFindex,includinglinkstothedataandquestionnaires,anddocumentationonmethodology,see

/en/publication/globalfindex/Data

.

9

mostrecenteditionwascompletedin2021for123economies(Demirguc-Kuntetal.,2022).ItincludesindicatorsofaccesstoanduseofformalandinformalfinancialservicesthathavebeencollectedinpreviouseditionsofFindex,aswellasindicatorsoftheuseofcards,mobilephones,andtheinternettomakeandreceivepayments,includingdigitalmerchantpaymentsandpaymentsfromthegovernment.Inadditiontowidercountrycoveragethanotherdatasources,thekeyadvantageoftheFindexdatasetisthatitprovidesnationallyrepresentativedemand-sideindicatorsofDFSusage.Incontrast,studiesthathaveusedsupply-sideindicatorsnotonlyhaveuseddataforasmallersampleofcountries(inwhichadvancedeconomiesweremoreheavilyrepresented),

butalsoreliedonindicatorsforasubsetofDFSproviders.Forexample,Claessensetal.(2018)

relyondatacollectedfromsurveysoffintechplatformsandotherpublicsourcesbytheCambridgeCenterforAlternativeFinance(CCAF)thatexplicitlyexcludesdigitalcreditprovidedbycommercialbanks,focusesonthelargerfintechplatforms,andexcludesotherprovidersofdigital

creditsuchasonlinemortgagelenders,bigtechfirms,ande-commerceplatforms.

9

Intheempiricalanalysisthatfollows,thedependentvariableindicatestheshareoftheadultpopulationinacountrythatmadeorreceivedadigitalpaymentin2020-2021usinganytype

ofprovider(seeTable1foradetailedvariabledescription).

BankRegulationandSupervisionSurveyDataset

TheBankRegulationandSupervisionSurvey(BRSS),collectedbytheWorldBankGroup,providescomparableeconomy-leveldataonhowbanksareregulatedandsupervisedaroundtheworld.Themostrecenteditionwascompletedfor160jurisdictionsin2019(Angineretal.,2022;WorldBank,2019).

10

Intheanalysisthatfollows,theBRSSindicatoristheindexof

overallrestrictionsonbankingactivities.FollowingBarth,Caprio,andLevine(2013),itis

9SeeClaessensetal.(2018),p.31,BoxA.

10FormoredetailsonthesurveyitselfandtheBRSSdataset,see:

/en/research/brief/BRSS.

10

computedusingquestionsthatcapturethedegreeofstringencyofbankingregulationinengaginginsecurities,insurance,realestate,andnonfinancialbusinessactivities.Theindexrangesbetween

3and12,withhighervaluesindicatingstricterregulation.

GlobalDataRegulationDataset

ProducedaspartofWorldBank(2021),theGlobalDataRegulationDiagnosticsummarizesthelawsandregulationsconcerningkeyaspectsofadataeconomyfor80countries.FollowingChen(2021),theempiricalanalysisbelowreliesonanindexofenablersthatfacilitateuse/reuseofdataandanotherindexofsafeguardsthatprotecttherightsofmarketplayersinthedataeconomy.Theenablersindexcoversthreedimensions:e-commerce,dataproducedforpublicpurposes(e.g.,publicpolicy),anddatacollectedbytheprivatesectorforbusinesspurposes.Thesafeguardsindexcoversfourdimensions:personaldata,nonpersonaldata,cybersecurity,andcross-borderdata.Theindexesrangefrom1to17withhighervaluesindicatingthatacountryhasmoreoftheregulationsthatenabledatause/reuse,

11

orsafeguardagainstitsmisuse.Indexscoresareaggregatedacrosssurveyresponses.Forexample,countriesreceivedonepointiftheyhadalaworregulationthatexplicitlygovernse-commerce/e-transactions.Thefulllistofthesurvey

questionsusedtoscorecountriesisfoundinAnnex2ofChen(2021).

HumanDevelopmentIndex

CollectedaspartofUNDP(2022),theHumanDevelopmentIndex(HDI)isasummarymeasureofaverageachievementonthreekeydimensionsofhumandevelopment:alongandhealthylife,beingknowledgeable,andhavingadecentstandardofliving.

12

Thehealthdimension

ismeasuredusinglifeexpectancyatbirth;knowledgeusingexpectedyearsofschoolingandmean

11However,thehighestpossiblevalueontheenablersindexis16.

12UNDP(2022).Dataavailableat:

/data-center/documentation-and-downloads.

11

yearsofschooling;andlivingstandardsaresummarizedbygrossnationalincomepercapita.BecausetheknowledgedimensionisthemostrelevantaspectofhumancapabilitiesforusingDFS,werelyonlyonmeanyearsofschoolingamongadultsasanexplanatoryvariableintheregressionspresentedbelow.However,theempiricalfindingsarequalitativelysimilarifweusethefullHDIasourexplanatoryvariablesummarizinghumancapabilities,orifwereplacetheHDIwiththe

HumanCapitalIndexdescribedinWorldBank(2020).

WorldDevelopmentIndicators

Theexplanatoryvariablesthatweusetodescribeconnectivity(shareofpopulationwithaccesstoelectricity,thatusestheinternet,orhasamobilecellularsubscription)aretakenfromWorldDevelopmentIndicators(WDI),asarecontrolsforgrowthinGDPpercapita,inflation,and

population.MoredetaileddescriptionsofthevariablesusedintheanalysisarefoundinTable1.

3.EstimationApproach

Ourestimationapproachisdesignedtoreduceomittedvariablebiasandendogeneitybias.Byfocusingondigitalpaymentsin2021,wellaftertheinitialshockofCOVID-19,wearguethattheestimatedparametersfortheexplanatoryvariableswillsufferlessfromendogeneitybiasthanamodelinwhichdigitalpaymentsandtheexplanatoryvariablesarecontemporaneous.Morespecifically,serialcorrelationislikelyinacross-countrymodelofdigitalpaymentsbasedonfactorssuchasregulatoryenablers,digitalinfrastructureandhumancapabilities,butexplainingvariationindigitalpaymentsafterexperiencingtheunexpectedCOVID-19shockshouldreduceconcernsaboutthissortofbias.Toaddressomittedvariablebias,weestimatethelevelofdigitalpaymentsin2021bothwithandwithoutacontrolforthelevelofdigitalpaymentsin2017,thelastpre-pandemicyearforwhichwehavedigitalpaymentsdata.Inthemodelthatcontrolsforthe

levelin2017,weareessentiallymodellingthechangeinthelevelofdigitalpaymentsduringthe

12

pandemic.Presumablytherearemanyunobservable,time-invariantvariables(thatis,invariantoverthetimeframeweareexamining),thatexplaindigitalpayments.Byincludingthelevelofdigitalpaymentsjustpriortothepandemic,wecontrolforthesetime-invariantunobservable

determinantsofdigitalpaymentsin2021.

Equation1relatestheshareoftheadultpopulationthatmadeorreceivedadigitalpaymentincountryiin2021(DigPayi2021)tothefactorsdescribedabove.Weincludethetrendindigitalpaymentusage,measuredasthechangeintheshareoftheadultpopulationthatmadeorreceivedadigitalpaymentin2017versus2014,tocapturethebaselinetrendinDFSadoptioninacountrypriortothepandemic.

13

Xi2019representscontrolsforGDPpercapita,inflation,andcountrypopulation.WeexpectGDPpercapitatobepositivelylinkedtodigitalpaymentusage(andotherDFSvariables);wedonothavestrongpriorsaboutthelinksbetweenDFSvariablesandinflation

andpopulation.

(1)

Digpayi,2021=a+β1?Digpayi,2014?2017+β2xi,2019+β3Bankingi,20