No.071TheBostonFedStudyofConsumerBehaviorandPaymentChoice:

ASurveyofFederalReserveSystemEmployeesMarquesBenton,KristaBlair,MarianneCrowe,andScottSchuh

Abstract: Thewaypeoplepay forgoods and services is changingdramatically,but littledataandresearchon consumerbehavior andpayment choice arepublicly available.Thispaperdescribes theresultsof a surveyofpaymentbehavior and attitudes takenbyFederalReserve employees in 2004.Majorcontributionsofthesurveyarethatitasks:1)whypaymentchoicesaremade;2)whyindividualpayment behavior has changed; and 3) why individualspecific payment characteristics matter forpaymentchoice.AlthoughthesurveyisnotstatisticallyrepresentativeofU.S.consumers,andthusmaynotprovideaccurateestimatesofaggregateU.S.paymenttrends,manyresultsareconsistentwithdatafrommore representative payment surveys. For example, the data show a trend away from checkwriting and toward electronic and emerging paymentmethods, but the choice of paymentmethoddepends on the type of payment, amount of payment, and other complex factors. Also, cost,convenience,andcontrolovertimingarethemost importantcharacteristicsdeterminingrespondentsadoption and use of payment methods. We find that payment characteristics vary widely acrossrespondents,partlybecauseofinherentheterogeneitybutperhapsalsobecauseofmeasurementerror,misperception,orinadequateinformation(lackofconsumereducation).Crosssectionalevidenceshowsthat respondents tend to use paymentmethods in amanner broadly consistentwith their reportedassessmentsofthepaymentcharacteristics.

JELClassifications:D12,D14,E41,C81MarquesBentonandKristaBlairareAssistantVicePresidents,MarianneCroweisaVicePresident,andScottSchuhisaSeniorEconomistandPolicyAdvisor,allattheFederalReserveBankofBoston.Theiremail addresses aremarques.benton@bos.frb.org, krista.blair@bos.frb.org,marianne.crowe@bos.frb.org,andscott.schuh@bos.frb.org,respectively.

ThisprojectsprangfromaleadershipdevelopmentprogramattheFederalReserveBankofBoston.Theauthors thank our colleague, Geoff Tootell, whose suggestions, together with the research of ourcolleagueJoannaStavins(2001),helpedsparktheideaoftheconsumerpaymentsurveyandwhowasinstrumentalinhelpingusdeveloptheidea.WethankJeffFuhrerandSallyGreenforsponsoringthesurvey project. We also thank Pattie Allouise, David Brown, David DeRemer, Nicola Lostumbo,Caroline Theoharides, Marcella Wiegand, Lisa Wright, and Jennifer Young for providing excellentresearch, administrative, and legal assistance.We thank JimCunha, Jeff Fuhrer,GeoffGerdes, JaneLittle,andGeoffTootellforhelpfulcommentsandsuggestions,andBradHirschbein,SuzanneLorant,andTylerWilliams forvaluableeditorialassistance.SeeAppendixB forcompleteacknowledgments.Anyerrorsremaintheresponsibilityoftheauthors.

TheviewsexpressedinthispaperarethoseoftheauthorsandnotnecessarilytheviewsoftheFederalReserveBankofBostonortheFederalReserveSystem.

Thispaper,whichmayberevised,isavailableonthewebsiteoftheFederalReserveBankofBostonathttp://www.bos.frb.org/economic/ppdp/index.htm.

Thisversion:February14,2007

1

mailto:marques.benton@bos.frb.orgmailto:krista.blair@bos.frb.orgmailto:marianne.crowe@bos.frb.orgmailto:scott.schuh@bos.frb.orghttp://www.bos.frb.org/economic/ppdp/index.htm

1. Introduction

The way people pay for goods and services is changing dramatically. Evidence

includingnewresults fromaBostonFedsurveyreported in thispaperisaccumulating that

consumersaremakingsubstantiallydifferentchoicesamongpaymentmethods to settle their

transactionsthaninthepast.Apparently,consumersareshiftingawayfrompaperchecksand

currency and turning instead to electronics, an expanding array ofmethods that leverage

advances in information technology and financialmarkets to offer lowcost transactions and

improved convenience.1 Even the British version of the classicMonopoly board game has

switchedfromcashtoadebitcard.2

Perhapsthemosttangiblechangeinpaymentbehaviorisoccurringintheuseofpaper

checks,achangethatdirectlyaffectsacentralbusinessconcernoftheFederalReserveSystem.

Figure1depictstheavailabledataoncheckvolumes.Theupperpanelshowsthelatestofficial

Fed estimates of the totalU.S. volume of check payments,which are available only at four

selectedpointsintime.3Theestimatesrevealadeclineincheckvolumefrom50billionin1995

to42billionin2000andto37billionin2003(26percentoverall).Thefirstevidenceofdecline

wasannouncedwhen theestimates for1995and2000werepublished in2002byGerdesand

Walton(2002,p.360):Theexactyearinwhichcheckusepeakedisunknown,butitappearsthe

number paid began to decline sometime in themid1990s. The relative scarcity of reliable,

publicly available data on check volumes prior to 2000 made it difficult to know and

understand completely the aggregate trend, much less the underlying consumer payment

1Notethattheuseofthetermpaperchecktodistinguishtraditionalcheckwritingfromelectronicforms of checkingsuch as electronic images of paper checks, paper checks initiated electronicallythrough online bill payment, or conversion of paper checks to electronic debit via the AutomatedClearingHouse(ACH)systemillustratesourpointnicely.2SeeHenry(2006)athttp://www.boston.com/news/local/massachusetts/articles/2006/07/25/new_monopoly_version_drops_paper_money_adds_debit_card/.3Actual survey data estimates are from FederalReserve Bank ofAtlanta (1981), the FederalReserveSystem(2002,2004), GerdesandWalton(2002),andGerdes,Liu,Parke,andWalton(2005);ProjectionsarefromFederalReserveBankofAtlanta(1983).

2

http://www.boston.com/news/local/massachusetts/articles/2006/07/25/new_monopoly_version_drops_paper_money_adds_debit_card/http://www.boston.com/news/local/massachusetts/articles/2006/07/25/new_monopoly_version_drops_paper_money_adds_debit_card/

behaviordriving the trend.Therecognitionofasignificantchange in theaggregate trend isa

keyreasonwhytheFedhasbeenestimatingcheckvolumesmorefrequentlysince2000.

The bottom panel of Figure 1 shows data on check processing done by the Federal

Reserve onlyroughly 40 percent of the U.S. total processing. The total number of checks

processedbytheFeddeclinedfromabout19billionperyearin1994toabout12billionin2005

(37percent).Muchofthefluctuationintotalvolumeduringthemid1990swasassociatedwith

legal,regulatory,andoperationalchangesthatsignificantlyaffectedthevolumeofarelatively

smallproportionofcheckshandledbytheFed.4However,thecorenumberofunsortedchecks

processedby theFed trendedupwarduntil the early2000s,as indicatedby thedashed line.

Prior to theGerdesandWalton (2002)announcement inAugust2002,anobserverusing this

coremeasureofcheckprocessingasanestimateofaggregatecheckusemighthavesurmised

thatvolumewasincreasing.Butsince2002,thevolumeofunsortedchecksprocessedbytheFed

hasdeclinedsignificantlytoo.

The decline in volume of Fedprocessed checks has impacted System operations

significantly.IthascontributedtoareductioninFedemploymentfrom23,448in2001,to19,935

in2005(15percent)andareductioninthenumberofcheckprocessingoperationssitesfrom45

to22(51percent)bytheendof2006.5Notsurprisingly,thesedevelopmentshavefullycaptured

theattentionofFedandprivatepaymentsindustrymanagers,whoneedtoknowhowfastand

howfarcheckwritingwilldecline,whatpaymentmethodswillreplacechecks,andwhatroles

theFedshouldplayinthenewpaymentsenvironment.

The papertoelectronics transformation clearly has important implications for

participantsinthepaymentsindustry,includingtheFederalReserveSystem,butithasbroader

economicandpublicpolicyimplicationsthatmeritexaminationaswell.Whatisnotunderstood

well iswhy the transformation is occurring. For example,why is ithappeningnow andnot

earlier (or later)? Why are some consumers participating in it, but not others? A full

understandingofthebroaderimplicationsofthepaymentstransformationhingescriticallyon

4Thesewerechecksinfluencedbythesamedaysettlementprocess,whichaffectedthewaycheckswerepresentedandsettledamongbanks.5ThisincludestheclosingoftwocheckprocessingbranchesintheFirst(Boston)District:Lewiston,ME,in1997,andBostonin2006.

3

understandingwhy it ishappening.Unfortunately,answers to thesequestionsarenoteasily

forthcoming, largelybecausedemand forpaymentmethodsbyconsumersperhaps themost

importantendusersofpaymentmethodsisnotwellunderstood.

Although some information is available about what payment choices consumers are

making, little isknownaboutwhyconsumersmake thosechoices.Forexample,whydosome

consumers still write checks and put them in the U.S. mail when they could write an

electroniccheckviaonlinebillpaymentfasterandmorecheaply?Amongconsumerswhohave

stoppedusingchecksforretailpayments,whydosomeswitchtoadebitcard,especiallyifthe

paymentinvolvesafee,whentheycoulduseacreditcardandgetwhatis,ineffect,aninterest

freeloanandperhapsfrequentflyermilestoboot?Unansweredquestionslikethese,andmany

others,areperplexingpaymentssystemproviders,researchers,andpolicymakers.

Akeygapintheunderstandingofconsumerbehaviorandpaymentchoiceisashortage

of fundamental research on the topic. Perhaps the clearest evidence of the lack of research

appears inaveryusefulbibliographycompiledbytheFederalReserveBankofPhiladelphia.6

AlthoughthetitleofthedocumentisConsumerPaymentsBibliography,onlyabout5percent

ofthemorethan700entriesareevenindirectlyconnectedwithconsumerbehaviorandpayment

choice per se, and far fewerdirectly address the theoryof consumerpayment choice.Schreft

(2006)providesanoverviewofthissmallliteratureandadvocatesmoreresearchonthetopic.7

The shortage of research on consumer payment behavior both stems from and is

exacerbatedbyalackofdataonactualconsumerpaymentchoices.KennickellandKwast(1997)

wereamongthefirsttocitethelackofdataasanimportantpartofthereasonfortheneglectof

research on consumerdemand forpayments.They turned to the FederalReserves triennial

Survey ofConsumer Finances (SCF),which contains some information aboutwhat payment

choicesconsumersmake,althoughpaymentsarenotthemainfocus.8AsidefromtheSCF,there

are fewpubliclyavailablealternativedatasourceson thissubject;privatesectordatasources

6 See http://www.phil.frb.org/pcc/bibliography.pdf for a PDF file of the bibliography orhttp://www.philadelphiafed.org/pcc/bibliography.cfmforasearchableonlineversion.7SchreftspaperwascommissionedbytheEmergingPaymentsResearchGroupattheBostonFedandpresentedattheBanksfirstConsumerBehaviorandPaymentChoiceconferencein2005(seeFederalReserveBankofBoston2005).8ForinformationabouttheSCF,seeBoardofGovernors(2006a).

4

http://www.phil.frb.org/pcc/bibliography.pdfhttp://www.philadelphiafed.org/pcc/bibliography.cfmhttp://www.bos.frb.org/news/conf/payments2005/

aregenerallynotavailabletopolicymakersandresearchers.9Anothercorroboratingassessment

appearsinasurveyofresearchonfinancialinnovationbyFrameandWhite(2004):Astriking

featureof this literature,however, is the relativedearth of empirical studies that specifically test

hypothesesorotherwiseprovideaquantitativeanalysisoffinancialdata. [Emphasisadded.]

Thelackofempiricalstudiesontheadoptionofpaymentmethodsbyconsumersisonespecific

exampleoftheproblemcited.

NotingthetransformationinpaymentsanditsvitalimportancetotheFed,wesetoutto

makeamodestcontribution toward filling thedatagapand improving theunderstandingof

consumer behavior and payment choice. Beginning in 2002, we developed a survey of

consumerbehaviorandpaymentchoicedesignedtogobeyondaskingwhatchoicesconsumers

weremakingbybeginningtoaskwhytheymakethosechoices.10Becauseofourconnectionto

theFederalReserve,wefocusedthesurveyoncheckwritingbehavior.Butbecausethedecision

aboutwhethertowriteacheckismadejointlywiththedecisionaboutwhethertouseanyother

paymentmethod, the survey includes questions aboutmost common alternatives.The three

primary contributionsofour surveydesignare:1) the surveyaskswhy consumersmake the

payment choices they do; 2) it asks individual consumers about changes in their specific

payment behavior over time (although it is not a true longitudinal panel); and 3) it asks

consumershow they rate the relativevaluesof some fundamentalcharacteristicsofpayment

methods,inadditiontoaskingaboutactualpaymentbehavior.

Working on a limited budget, we chose a convenience sample of the most readily

available,highlymotivated,andinexpensivepotentialrespondents:FederalReserveemployees.

Wefirstconductedthesurveyin2003,withmorethan500BostonFedemployees,andthenin

2004,withmorethan5,000FederalReserveemployeesfromacrosstheSystem.11Thesesamples

havefiveserious limitations:1)FedemployeesarenotdemographicallyrepresentativeofU.S.

consumers(onaverage,Fedemployeesareolder,bettereducated,andhavehigherincomes);2)

9SeeAppendixAforabriefdiscussionofdatasourcesandtheiravailability.10Foranexampleofresearchalongtheselines,seeSchuhandStavins(2005).11ThesurveyresultsarepostedontheBostonFedspublicwebsite.Forthesurveyinstrumentwithallnumericalresultsintabularform,seehttp://www.bos.frb.org/economic/eprg/FRBBsurvey04num.pdf.Foragraphicalsummaryofmostresults,seehttp://www.bos.frb.org/economic/eprg/FRBBsurvey04.pdf

5

http://www.bos.frb.org/economic/eprg/FRBBsurvey04num.pdfhttp://www.bos.frb.org/economic/eprg/FRBBsurvey04.pdf

FedemployeesarefarmoreknowledgeableaboutthepaymentssystemthanU.S.consumers(a

bias problem); 3) Because of professional duty, reward incentives, and a general sense that

providing the information could help the Fed during a time of institutional change, Fed

employeesweremore likelythanotherU.S.consumerstorespondtothisvoluntarysurvey(a

sample selectionproblem); 4)Because Fed employeeswork in an organization that supplies

payment services, theymay answer certain survey questionsknowingly orunknowingly

fromtheperspectiveofafirmratherthanfromtheperspectiveofaconsumer;and5)Because

almost every Fed employee has a checking account, Fed employees behavior is not

representative of the behavior of all U.S. consumers, particularly those without checking

accounts.

Because the survey resultsarenot representativeofallU.S. consumers, theaggregate

statistics from our survey should not be used todraw conclusions aboutU.S. consumers in

general.However,crosssectional(individualconsumerlevel)studiesusingthesurveydatacan

providevalidanduseful informationabout individual consumerbehavior.12Furthermore, in

2006,wecollaboratedwiththeAARP(formerlytheAmericanAssociationofRetiredPersons)to

implementamodifiedversionoftheBostonFedsurveywithanationallyrepresentativesample

drawnrandomlyfromtheentireU.S.population.13

This paper provides an overview of the Boston Fed survey.14 Section 2 defines

terminology, and Section 3 describes the survey. Section 4 compares our results on what

payment choices consumers are making with results from similar surveys that are more

representative of U.S. consumers. Section 5 presents selected results on why consumers are

makingthepaymentchoicestheymakeandhowthosechoicesarechanging.Section6discusses

some survey results that bear directly on Federal Reserve payment operations. Section 7

presents a brief assessment of the issues raised by the transformation in payments and the

challenges facing policymakers. Section 8 concludes with a highlevel assessment of the

12Foranexampleofresearchalongtheselines,seeSchuhandStavins(2005).13DocumentationofthissurveyisforthcominginanotherBostonFedPublicPolicyDiscussionPaper.14KriegerandBraun(2005)describetheresultsofananalogoussurveyoflargecorporationspreferenceswithregardtopayments,conductedbytheNewYorkFed.

6

contributionsofthesurveyandsuggestionsforfutureresearchtoadvancetheunderstandingof

consumerpaymentchoice.

Although the survey design is imperfect, we were encouraged to discover general

consistencywithfindingsofother,morerepresentativesurveys(SurveyofConsumerFinances

(2001,2004),MichiganSurveyofConsumers(2003),andAmericanBankersAssociation(ABA)

andDove(2003,2005),foraspectsofthesurveythatwerealsoaddressedbyothers.

Our results include several findings that seem important and suggest the need for

furtherstudyandconsideration:

Fedrespondentsoverallratechecksworsethanotherpaymentmethodsacrossmost

fundamentalcharacteristics,andACHpaymentsappeartostrictlydominatechecks

inallcharacteristics.This findingmayhelp toexplainwhyconsumersareshifting

frompapertoelectronics,especiallyforrecurringbillpayments.

Yet,even in2004,checkswerestillthemostcommonsinglepaymentmethodused

by respondents. The survey results suggest that improvements in user control of

payment timing, overdraft protection, and consumer knowledge of ACH might

increaseuseofACHpaymentsconsiderably.

Nearly onethird of respondents indicate that their payment behavior, especially

checkwriting,issensitivetofloat.15However,evenfewerrespondentswouldordid

alter theirpayment,banking, or shoppingbehavior in response to changes in the

checkprocessing environmentparticularly the introduction of electronic check

images and conversion ofpaper checks to electronicACH transactions that affect

float.

Cost, convenience, and timing are the three most important fundamental

characteristics that determine respondents adoption and use of all payment

methods. Safety and privacy also are important formethodsmore susceptible to

consequenceslikeidentitytheftthatarecostlytoconsumers.

15Floatistheintervalbetweenthephysicalwritingofthecheck(payment)andthetimewhenthefundsaredeductedtheconsumerscheckingaccount(settlement)bythebank.

7

Respondentsareheterogeneous in theirviewsaboutanduseofpaymentmethods.

Themostimportantbarriertotheadoptionofapaymentmethoddiffersbymethod,

and most respondents use a variety of payment methods each month. When

respondentsswitchfromcheckstootherpaymentmethods,themethodtheychoose

dependsonthelocationandamountofthepayment,amongotherthings.

Respondentschoosepaymentmethods inamanner that isbroadlyconsistentwith

their reported ratings of the relative characteristics of those payment methods.

Reported characteristics could vary across respondents, even within narrow

demographicgroups,formanyvalidreasons.Moreresearchanddataareneededto

determinewhetherthereportedcharacteristicsaccuratelyreflecttheactualpayment

characteristics consumers face, and the extent to which the assessments may be

subjectiveorbiased.

When thepayments system changes, some respondents report that theywouldor

did alter their payment behavior. The extent towhich behavior changes, and the

circumstancesunderwhich itchanges,dependson the typeandmagnitudeof the

paymentssystemchange.Understandingwhetherthechangeisbeneficialforsociety

as a whole requires a much deeper understanding of consumers tastes and

preferencesregardingpaymentsthanwecurrentlyhave.Putsimply,foracomplete

understandingoftheoverallpaymentsindustry,theviewsofconsumersmatter.

Weviewthesekeysurveyresultsasanillustrativesteptowardgainingabetterunderstanding

of consumer payment behavior. We hope that they will stimulate further research, data

collection,andpublicpolicydiscussion.

Asconsumerbehaviorisunderstoodmorefully,itwillbenaturalforpolicymakerslike

theFederalReserve tobegin toask,what is best for consumerswith regard to thepayments

system? A major hurdle to answering this question is the surprisingly widespread and

persistentheterogeneityofconsumerspaymentpreferences.Also,someU.S.householdshave

nobank account and thus cannotparticipate in some electronicpayment innovations.Other

developmentsidentity theft, credit and debit card interchange fees, technology barriers,

8

financial market complexity, rapidly expanding choices, and the likehave potentially

importantimpactsonthewelfareofconsumersaswell.

2. DefinitionsandTerminology

Aswithany rapidchange insociety, thepayment transformation isengenderingnew

languageandterminologythatissometimesredundantandunclear.So,beforedivingintothe

detailsofconsumerpaymentbehavior,itwillbehelpfultodefinethetermsusedinthispaper

andtheirrelationshiptoalternativetermsusedelsewhere.

The term consumer refers to a household member who makes payment decisions

whenbuyinggoodsand services forpersonalorhouseholduse.Thus,weareall consumers

whodemandpaymentservices.However,oursurveyqueriedFederalReserveemployees,who,

aswehavenoted, arequitedifferenton average from the averageU.S. consumer.For these

reasons, we try to distinguish carefully between consumer and respondent, and the reader

shouldalwaysbearinmindthepotentiallyimportantdifferencesbetweenthem.

Thepaymentsurvey includessevenmainpaymentmethods:cash,papercheck,credit

card,debit card,AutomaticClearingHouse (ACH), storedvalue card (SVC), and onlinebill

payment.Tosomeextent,thesepaymentmethodtermsareselfevidentandhavebecomepart

oftodaysvernacular,butnoconsensusonallterminologyhasemergedyet.Thus,tobeclear

andprecise,welistanddefinethetermsweuse,plusalternativeterms,inTable1.

Threegeneral issuesconcerningpaymentmethodscauseunusualconfusionandmerit

additional brief discussion. First, there is great variety in the payment methods that are

availabletoconsumers;thesearesummarizedinTable1.Consumerpreferencesandchoicesin

thefaceofthisvarietyarediscussedfurther inSection6.Second,ACH,whichplaysapivotal

role in several other payment methods (debit cards, check conversions, and online bill

payments), is not a payment method per se but rather an electronic medium for settling

payments that can be initiated by various payment methods. However, in developing our

surveywereliedonACHpaymentsasashorthandforoneformofpaymentautomaticbill

payments initiatedbyconsumers (mortgagepayments, insurancepayments,and the like)so

thereadershouldthinkofthesewhenreadingreferencestoACH.

9

Similarly, a third issue is thatpaymentsmade over the Internet can occur in at least

threeways,anyofwhichmightbeviewedasa formofonlinebillpayment.Oneway isvia

online banking, a process that allows a consumer with a checking account to pay bills

electronically rather thanwithapaper check.Typically, thebill ispaidby thebank through

ACH, but, ironically, this method sometimes involves the banks writing a check for the

consumer.Asecondformofonlinebillpaymentoccurswhenaconsumeraccessesacompanys

websiteandinitiatesapaymenttothecompanyforthebill.Thisformofpaymentcanbemade

bycreditordebitcardorbyACH,eitherperiodbyperiodorrecurringautomatically,thelatter

requiringonlyasingleonlinetransactiontosignup.Athirdformofonlinebillpaymentcanbe

mediated througha thirdparty,suchasapaymentprovider likePayPal.This formofonline

paymentissimilartoapaymentmadedirectlytoacompany(orindividual),butthepayment

provideractsasanintermediaryandmayofferadditionalpaymentoptions.

Wedidnotdistinguishamongthesedifferentformsofonlinebillpayment,leavingthe

surveyrespondentssome leeway indefining the term.Thus,aswithACH,we treatedonline

billpayment as apaymentmethod even though it is actuallynot apaymentmethod per se;

rather, the Internet is an electronicmedium used to facilitate various paymentmethods. In

future surveys, it might be better to distinguish more clearly among the various forms of

paymentmethodsthatcanbeinitiatedonline.

Finally,weneedtoclarifytheterminologyusedinthispapertodescribevariousaspects

ofpayments themselves.Weuse the termpaymentmethod todescribe thevariousmeans

thatconsumers(includingrespondents)canusetoeffectapayment;weusethetermformin

ageneralsense,asinthediscussionimmediatelyabove;andweusethetermpaymenttypeor

typeofpayment todescribe thekindofobligation that isbeingdischarged, forexample,a

mortgagepayment,autilitybill,aretailsalespurchase,etc.)

3. TheBostonFedConsumerPaymentSurvey

Thissectiondescribesthedesign,implementation,andsamplepropertiesoftheBoston

Fedconsumerpaymentsurveyconductedin2004.

10

3.1 SurveyDesign

To begin,we conducted focus group interviews for threemonths in early 2003with

employeesof theBostonFed togainabetterunderstandingof the importanceof changes in

checkprocessingforFederalReservepaymentoperationsandofthegapsinknowledgeabout

consumers use of checks. These interviews, formal and informal, included the Banksmost

seniorofficerswithexpertiseinpaymentsandeconomicresearch. Twokeyfindingsemerged

clearlyfromtheinterviews:

Althoughmostobserverswerekeenlyawareofthedeclineincheckvolumebyearly

2003, the timingandmagnitudeof thecheckdeclinewerenot fullyanticipatedor

explained, suggesting theneed forgreaterunderstandingof theunderpinningsof

consumerdemand forchecksandotherpaymentmethods.Thisgap inknowledge

mirroredthe lackoffundamentalresearchonconsumerpaymentdemandnoted in

the introduction. It also stemmed in part from the fact that the Feds primary

customers are depository financial institutions (banks, for short), rather than

individual consumers, and neither the Fed nor banks had developed a sufficient

understanding of why consumers make their payment choices.16 Together, the

decline in check volume and the gap in knowledge about consumer payment

demandarecultivatingtheneedtodevelopgreaterinsightintoconsumerpayment

demandanditsimpactonallFederalReservepaymentoperations.17

It is impossible to design a proper survey of consumer behavior about a single

paymentmethod.Whenconsumersmakechoicesaboutbuyinggoodsandservices,

theyformrelativecomparisonsamongthesegoodsandservices,basedonpricesand

othercharacteristicsoftheproducts.Consumersdecisionsaboutpaymentsworkthe

16ThispointalsowasmadebyRichOliver,headoftheFederalReservesRetailPaymentsOffice,inhiskeynote address at the Boston Feds 2005 Consumer Behavior and Payment Choice conference. Oliveridentifiedfourstumblingblockstoeliminatingthegapofunderstandingbetweenbanksandconsumers,oneofwhichwastheneedtoaskconsumerswhattheywantintheirpayments.Infact,Olivernotedthattheconferencessurveyofparticipantspaymentpreferencesrepresentedthefirsttimein40yearsthathehadbeenaskedaboutpaymentsasaconsumer.Formoredetails,seeCrowe,Schuh,andStavins(2006).17The seriesofRetailPaymentStudies in the early 2000sdescribed inGerdes andWalton (2002) andGerdesetal.(2005)representoneimprovementinstrategicthinkingonthistopic.

11

sameway.Thedecision towriteacheck isachoiceamongoptions,andassuch is

integrallyrelatedtothechoicetopaybyalternativemethods.

Thesetwofindingsinfluencedthesurveydesignintwoways:

The survey aimed to elicit informationprimarily about checkwriting behavior to

understandhowandwhyconsumerpaymentbehaviormattersfortheFed.

Although focused primarily on checkwriting, the survey also includes questions

about other common payment methods; specifically, it asked for pairwise

comparisonsofchecksversusotherpaymentmethods.

The result of these findings anddesign considerations is apayment survey that emphasizes

checks and FederalReserve operations to a greater extent thanwould a surveydesigned to

understandU.S.consumerpaymentbehavioringeneral.18

TheBostonFedconsumerpaymentsurveycontains55questionsofvarioustypesinthe

followingcategories:19

Generalcheckwritingbehaviorandattitudes[questions18];

Adoption and use of six noncash payment methods (checks, credit cards, debit

cards,ACH,onlinebillpayment,andSVC)[917];20

Threeyearretrospectivechangesinpaymentmethodusage[18];

Reasonsforsubstitutingotherpaymentmethodsforchecks[1922];

Typesofpaymentsmadeusingdebit,ACH,andSVC[2426];

Pairwise comparisons of noncheck payment methods, relative to paper checks,

alongsevenpaymentcharacteristics[2731];

Experiencewith check conversion toACH in retail stores,alsoknownaspointof

sale(POS),andcheckconversiontoACHforrecurringbillsmailedtoalockbox,also

knownasAccountsReceivableConversion(ARC)[3242];

Miscellaneouspaymentexperiences[4348];

18 Indeed, theversionofour survey implementedby theAARP excludes someof thequestionsmostrelevantfortheFederalReserveandmostdetailedaboutcheckwriting.19SeeTable1 for completedefinitionsofallpaymentmethodsaswellasallnew terms (for example,AccountsReceivableConversion).20Theomissionofcashfromthisgroupwasinadvertent.

12

Reasonsfornothavingacheckingaccount[49];

Demographics[5054];

Openendedessayaboutcheckwritingdecisions[55].

Althoughthesurveyseeksprimarilytomeasureconsumersbehaviorandattitudesatapointin

time, it asks some retrospective questions to gain perspective on changes in attitudes and

behaviorovertime.Mostrespondentsreportedthattheytookaboutonehalfhourtocomplete

thesurvey.

3.2 SurveyImplementation

Wedistributed themainversionof thesurvey toallemployeesof theFederalReserve

SystemduringapproximatelysixweeksinJuneandJulyof2004.Itwasadvertisedasvoluntary

andrespondentswerepromisedcompleteconfidentiality.

Theprimary formof thesurveywaspostedby theBostonFedon the internalFederal

Reserve intranet,which isaccessiblebyemployeesof theBoardofGovernors,all12 regional

Banks, and all Bank branches. Each Federal Reserve District (plus the Board of Governors)

appointedarepresentative toworkwith theBostonFedsurvey team.However,each location

wasindividuallyresponsibleforpromotingthesurveyandencouragingparticipationamongits

ownemployees.Notsurprisingly,participationandenthusiasmvariedacrossFedlocationsfor

thisvoluntaryproject.Papercopiesofthesurveywerealsomadeavailabletoanextentthat

variedacrossBanks,butonly4percentofallsurveyswerecompletedinwriting.21

Althoughparticipation in thesurveywasvoluntary, itwasencouraged inat least two

ways.First,wemadeanunabashedappealtoemployeesloyaltytotheFed,citingthechanges

inFederalReserveemploymentandtheneedtounderstandthepaymentstransformation.Most

employees who provided feedback indicated that loyalty, duty, and a desire to help the

Systemespecially a desire to help employees in checkprocessing operationswere highly

motivating factors.Second, theBankalsosponsoredamodestraffledrawingas incentive; the

21Theavailabilityofpapersurveyswasanimportantissuebecausecomputeraccessisnotuniformacrossall jobswithin theFederalReserve.Employees in jobswith lesscomputeraccess tend tobeamong themostunderrepresenteddemographicgroupsoftheFeduniverse(younger,withlowerincomesandlesseducationthanothers).Thus,broaderavailabilityofpapersurveysmayincreasesamplerepresentation.However,distributionandcollectionofpapersurveys isalsomuchmorecostly(especially intimeandlabor)thanisthecasewithelectronicsurveys.

13

top prizes were valued at about $25, and participants had a 4.7 percent ex post chance of

winningaprize.22

3.3 SampleProperties

Nearly one in four Federal Reserve System employees participated in the consumer

paymentsurvey.A totalof5,344Fedemployeesresponded to thesurvey,5,110electronically

(96percent)and234bypaper,aresponserateof24percent.23Thisrateisconsiderablyhigher

thantherateobtainedbysomeprivatesectorfirmsthatconductvoluntaryconsumerpayment

surveysdesignedtoberepresentativeoftheU.S.population.24

Surveyparticipationwasunevenacross the regionsof the country. The toppanelof

Figure 2 compares the Fed survey responses with the regional distribution of the U.S.

population.25 The distribution of survey responses generally reflects the U.S. population

distributioninallbutfourDistricts:MinneapolisandKansasCityareoverrepresented,andSan

Francisco andAtlanta are underrepresented.26 If variation in respondents payment behavior

acrossDistrictsdiffersfromvariationinU.S.consumerspaymentbehavioracrossDistricts,the

Fedsurveyresultswillbesubjecttobias.

Even if thegeographicalvariation of respondents accurately reflects thegeographical

variationofconsumersintheserespects,theunequalrepresentationofDistrictsmaybeasource

of additional geographical bias. Gerdes and Walton (2002) and Gerdes et al. (2005) find 22InMayandJuneof2003,wealsoconductedaverysimilar,butslightlyshorter,versionofthesurveywithBostonFedemployeesaswellas several testversionsprior to thatwith smallgroups.Themaindifferencesbetween the two surveyswere: 1)mostof theBostonFed 2003 surveywas conductedviapapercopies(64percent)rather thanelectronically;and2) the topprize in theBostonFed2003surveyrafflewasafreevacationday,considerablymorevaluablethan$25.23Unfortunately,1040 electronic responseswere lostdue to technicaldifficulties inBoston,but 329ofthese lost responses (32percent)were replacedbyemployees taking the surveya second time.Theseemployeeswereofferedanadditionalrafflewitha19.6percentexpostchanceofwinning.Resultsfromtheemployeeswhoretookthesurveyclearlymaybebiasedforseveralreasons,butwecanidentifytheseresponsesandcontrolfortheirpotentialbiases.24Forexample,theABA/Dove(2005)surveyresponseratewasabout6percentforpapersurveys(therateforemailsurveysisnotknown).25TocreatepopulationbyFederalReserveDistrict,wemadetwosimplifyingassumptions:1)TheBoardof Governors (Washington, DC) was included in the Richmond District; and 2) population in statesservedbytwoReserveBankswasdividedevenlybetweenthetwoDistricts.26ThetoppanelofFigure2usesthetotalusablenumberofsurveystocalculatetheFedSurveynumbers(thatis,itdoesnotincludethelostsurveysdescribedinfootnote22).MostofthelostsurveyswereintheSanFranciscoDistrict;thisexplainsmuchofthatDistrictsrelativelylow(usable)responserate.

14

statisticallysignificant regionaldifferences incheckwriting; forexample, theMidwest region

writesmorecheckspercapitathantheotherthreeregionsthattheseauthorsstudied.However,

this evidence is for total checks (including those of businesses), not just for consumer

(individual) checks, and thus it is not comparable to the Fed survey estimateswe report in

Section4.Althoughwefindregionalvariationinpaymentbehavior,FederalReserveBanksdo

not provide representative sampling of the regions in which they are located and thus are

unsatisfactoryindicatorsforregionalanalysis.27

ParticipationwasalsounevenacrosstheFederalReserveSystem,inaslightlydifferent

manner.ThebottompanelofFigure 2plots theperemployee response ratesand theusable

responserates(excludinglostsurveys)byFederalReserveDistrict,aswellasthetotalSystem

responserate.Participationwasgreatest inMinneapolis (51percentrate)andKansasCity (46

percentrate),andgenerallyaboveaverageinmostMidwestdistricts.Participationwaspoorest,

intotalandrate,attheBoardofGovernorsinWashington,DC.

Inadditiontogeographicregion,fourotherdemographiccharacteristicsfromthesurvey

areavailabletogaugetherepresentativenessoftheFedsamplerelativetotheU.S.population:

age, education, income, and home ownership, which proxies for wealth. Figure 3 plots the

distributionsofeachofthesecharacteristicsforFedrespondentsandfortheU.S.population.

AGE:FedrespondentsareolderthantheU.S.population.MostFedrespondentsare

middleaged(3555)employeesfromaheavilywhitecollarindustry;asaresult,the

relatively young and relatively old are underrepresented, especially the youngest

(1825).

EDUCATION:AbouttwothirdsofFedrespondentshaveacollegedegreeorsome

highereducation,comparedwithonlyaboutonefourthoftheU.S.population;asa

result, consumers with a highschool diploma or less education are vastly

underrepresentedinthesurvey.

INCOME: In part because of these age and educational advantages, Fed

respondents tend to have much higher household incomes than U.S. consumers 27Moreover,wefindconsiderableheterogeneitywithinthebroaderregions.FiveReserveBankDistrictsare included intheMidwestregiondefinedbyGerdesandcoauthors,andourestimatesofcheckspercapitapermonthrangefrom9.3to13.2acrossthoseDistricts(theSystemaverageis10.5).

15

generally.28 About half of Fed respondents have annual household incomes of

$75,000 or more, about twice the share of the total population with such high

incomes.

HOMEOWNERSHIP(WEALTH):Fedrespondentsaremorelikelythanthegeneral

populationtoownahome(82percentvs.67percent).

Like the regional imbalances, thesedemographicdifferenceswillcontribute tobias in the

aggregate survey results if consumer payment behavior is systematically related to

demographics.

3.4SurveyQuestionTypes

Manyofthesurveyquestionsarestraightforward,singleresponsequestions,butsome

ofourotherquestionsaredesigned togainanunderstandingofwhy consumersmake their

payment choices; these latter questions readily admit multiple responses. For example, we

askedrespondentswhodonotusecertainpaymentmethodstoexplainwhytheydonot,and

onecaneasily imaginemultipleanswers to this typeofquestion.Surveydesignershave two

basicchoiceswithregard tomultipleanswerquestions:1)providea largenumberofreasons

andallowrespondentstochoosemultiplereasons;or2)askopenended,essaytypequestions.

Openendedquestionsgenerallyarepreferable foranumberofreasons; forexample, theydo

not lead respondents toanswersand theyallow for responses that thesurveydesignersmay

nothaveconsidered.However,openendedquestionsaremoredifficultandtimeconsumingto

handleindataanalysis,andtheyrequirethesurveyanalyststointerpretrespondentsanswers,

leaving open the possibility of subjective interpretation and the introduction of errors and

biases.Forthisreason,weoptedprimarilyformultipleresponsequestionswithpredetermined

answers.

Having chosenmultipleresponsequestions,we thenhad todecidewhetherornot to

design the response format to elicit the relative importance (orweight)of each respondents

multipleresponses.Oneoptionisordinalranking,whichwouldrequirerespondentstolistthe

orderof importanceofeach response (from1 to the totalnumberof responses, forexample). 28 The income data are in nominal terms (2004) and not adjusted for costofliving differences acrossregions.MostFederalReserveemployeesworkinlargemetropolitanareasthattendtohavepopulationswithmuchhigherincomesonaveragethantherestofthecountry,especiallyontheEastandWestcoasts.

16

However, ifone responsewas farmore important thanall theothers, this formatwouldnot

provide that information.Anotheroptionwas toaskrespondents torateeachresponseonan

absolute scale of importance (from 1 to 10, for example). After considering the additional

complexityandburdenofelicitingrelativeimportance,weoptedtouseunweighted,multiple

responsequestions.

In analyzing the results of these unweighted, multipleresponse questions, we

examined the data from two different perspectives. For some questions, we were most

interested in knowing the distribution of choices among respondents; for example,wemight

wanttoknowwhatpercentageofrespondentscitedconcernsaboutsafetyasabarriertotheuse

ofanelectronicpaymentmethod.Forotherquestions,weweremostinterestedintheintensity

of response to each choice. For this latter set of questions,we looked at the distribution of

responses among the possible choices, for example, what percentage of responses identified

convenienceasafactorininfluencingrespondentschoiceofpaymentmethod.Thepercentage

ofresponsesprovidesacrudemeasureoftherelative importanceofaresponse,although it is

clearlyinferiortothedirectweightingalternativesdiscussedintheprecedingparagraph.

4. WhatConsumersUseWhenTheyPay

This section reportsbasic resultsonFed respondents typicalpaymentbehavior,asof

June2004.Thesurveyaskedrespondentswhatpaymentmethods theyhadadoptedandhow

theyusedthepaymentmethodsingeneral,andinspecificpaymentsituations.Becausemuch

ofthiswhatbehaviorhasbeenmeasuredinother,morerepresentativesurveys,wecompared

ourresultswiththesebenchmarksurveystohighlightpotentialbiases intheFedsample.The

main benchmarks for the generalpayments questions are the Survey ofConsumer Finances

(2001, 2004) and the Michigan Survey of Consumers (2003). For more detailed payments

questions,thesurveysbytheABAandDove(2003,2005)arethebenchmark.29Wealsoprovide

demographicallycorrectedFedresultstocheckforbiases.

29SeeAnguelov,Hilgert,andHogarth(2004)formoredetails.

17

4.1 PaymentMethodAdoption

The survey begins with several questions (28) about checking accounts.30 Fed

respondentsareconsiderablymorelikelythantheU.S.populationtohaveacheckingaccount.

Nearlyallhaveone(99percent),asshowninTable2,comparedwithfewerthan90percentof

the general population. There are two reasons for this disparity. First, Fed employees are

employed,and theunemployedandnonparticipants in the labor forceare less likely tohave

(or need) a checking account. Second, the Fed has been particularly aggressive among

employers in requiring their employees to sign up for direct deposit of their pay; thus Fed

employeesmusthavecheckingaccounts.31Forthesereasons,demographicadjustmentstothe

Fed checking account adoption ratedo not alter the responsemuch. Interestingly, however,

only 49 percent of Fed employees with a checking account earn interest on their checking

accountbalances(question3).

Amongthosewhodonothaveacheckingaccount,theFedrespondentsaremuchmore

likely to eschew them because theydo not need one, rather than for other reasons.Table 3

shows that53percent (20of the38)Fedrespondentswithoutacheckingaccountreportedno

need for one,muchhigher than the 5percentbenchmark results.Because respondentswere

allowedtorespondmorethanonceintheFedSurvey(andonlyonceintheothersurveys),this

comparisonmay not necessarily suggest that having no need for a checking account is Fed

respondentsmost importantreason fornothavingacheckingaccount.Fedrespondentsmay

actuallyweight thereasons foreschewingchecksdifferently than theoverallU.S.population.

Forexample,theseFedrespondentsarelikelytobemorecomfortablethantheU.S.population

atlargeindealingwithbanks(presumablybecausetheyworkforone).

Unfortunately,weinadvertentlydesignedthesurveytoexcluderespondentswholacka

checkingaccount fromanswering theremainingquestions in thesurvey.Consequently,allof

theremainingstatisticsinthispaperarebasedonrespondentswhodohaveacheckingaccount. 30Thesurveydidnotdistinguishbetweentraditionaldemanddepositcheckingaccountsatcommercialbanksandmoneymarketaccountsthroughinvestmentbrokers.AnecdotalevidencesuggeststhatsomeFed employeesmay have only the latter type of checking account; the other surveyswere clearer inaskingabouttraditionalcheckingaccounts.31AsmallpercentageofFedworkerstemporary, itinerant,contractors,and the likemaybeexemptfromdirectdeposit,makingthetotaladoptionrateslightlylessthan100percent.

18

Fortunately,asnotedabove,99percentofrespondentsdohaveacheckingaccount (however,

applicationof thissurvey toarepresentativesamplewouldbeevenmoreproblematic in this

regard).

Amongthosewhodohaveacheckingaccount,Fedrespondentsaremorelikelythanthe

overallU.S.populationtoadoptalloftheothernoncashpaymentmethodsexceptstoredvalue

cards,asshownintheremainingrowsofTable2.32Respondentsaresomewhatmorelikelyto

adopt credit and debit cards and online bill payment. Demographic adjustments actually

increasethedebitcardadoptionestimatesslightly,makingthemevenmoreupwardlybiased.

Income and education adjustments, however, help to explain some of Fed respondents

relativelyhighadoptionofcreditcardsandonlinebillpayment.RespondentstotheFedsurvey

aremore likely thanrespondents to theothersurveys touse (preauthorized)ACHpayments

for somebillsbyawidemargin (71percentversus4447percent).Even though incomeand

education adjustments reduce the bias a bit, Fed respondents clearly areunusually frequent

ACHusers.This result likely reflects theirmuchgreater familiaritywith, and trust in,ACH

paymentsthanisthecasefortheU.S.populationoverall.

Themost strikingdifference betweenFed employeeswith checking accounts and the

U.S.population is in theadoptionofstoredvaluecards,whereFedrespondentsare less than

halfas likely (35percentversus73percent) toadopt the technologyevenafterdemographic

adjustments. In fact, only 5 percent of Fed respondents even use the storedvalue cards

availabletotheminternallyattheirownFedlocation(question26).Threequalificationsapply

to thisresult.First,withonlyonebenchmarkestimateofpopulationadoptionofstoredvalue

cards(2003MichiganSurvey),thisresultshouldbeviewedwithcaution.Itispossiblethatthe

responses to the twosurveysdifferbecauseof theway termsweredefinedorquestionswere

32Inthefollowingdiscussionofdemographicallyadjustedstatistics,itisimportanttounderstandthatfordemographicweightingtohaveaneffect,twoconditionsmusthold:1)thegroupmustbeundersampledbyalargeamount;and2)thegroupmusthaveverydifferentbehaviorfromtherestofthepopulation.Thus,wefindthatweightingbyeducationorincomehasamuchlargereffectthenweightingbyhomeownershiporage.SeeMester(2006)fordatafromtheSCFonpaymentmethodadoptionratesfrom19952004.

19

asked.33Second,asweexplain further inSection6, theoverallFedadoption rates forstored

valuecardsmaybeconsiderablylowerthanadoptionratesatindividualFedlocationsifsome

locationsdonothavethecards.Finally,aswithotherpaymentmethods,therecouldberegional

differencesmoregenerallyinstoredvaluecardadoption.

4.2 PaymentMethodUse

Paper checks still represented the single largest payment method in terms of

transactionsvolumeforFedrespondentsin2004,butthevolumeofallelectronicpaymentsfar

exceededcheckswritten.Checksrepresented30percentofallmonthlynoncashtransactions,as

shown inFigure4,whereaselectronicpaymentsaccounted for70percent.Because thesedata

on payment use were collected from memory rather than from observed or recorded

transactions,aconcernarisesabouttheaccuracyoftheseresponses.Fedrespondentswhowrite

checks (that is,99percentofallrespondents)stated that theywriteanaverageof10.5checks

eachmonth.Thisestimateislessthantheestimateof12.1(notshown)inferredfromABA/Dove

(2005)survey,whichalsoreliesonrespondentsmemoriesandwasobtainedayearlater.34

Whenconsideringallrespondents(maroonbars,Figure4)thatis,bothusersandnon

users of payment methods togetherthe noncheck payment methods were used less

frequentlythanchecks,butuseofdebitandcreditcardswasnearlyascommon.Debitcarduse

for all respondents averaged 9.4 transactions per month, and credit card use averaged 8.5

transactions per month. Respondents used newer emerging payment methods much less

frequently.OnlinebillpaymentandACHwere themore commonof these,averagingabout

threetransactionspermontheach,whilestoredvaluecardswereusedonaverageaboutonce

33TheMichigansurveysaid:Prepaidcardsarecardsthatcontainastoredvalue,oravaluethathasbeenpaidupfront,allowingyou touse thecardmuch likecash.Asyouuse thecard, theprepaidvalue isdrawndown.Examplesofprepaidcards includephonecards,giftcards,andstudentcards.Haveyoueverhadorusedaprepaidcardorboughtoneasagift?TheBostonFedsurveysaid:StoredValueCardAplasticcardthatstoreselectronicdataondollarvalues.Oneexampleisaprepaidphonecard.Otherexamplesincludegiftcardsfromstores likeMacys,HomeDepot,orStarbucks.IftheMichigandefinition were clearer or easier for respondents to understand, it might have increased Michiganrespondentslikelihoodofacknowledginguseofstoredvaluecards.34TheABA/Doveestimatewasobtainedfromthedataonthepercentagesofcheckswrittenforpaymentsinstore (retail),online,andbills.Although thisestimatemayundercount checkswritten if consumerswritethemformanyothertypesofpayments,theresultsmaystillgiveageneralideaofthecomparisonbetweenthetwosurveypopulations.

20

per month. Thus, checks, debit cards, and credit cards accounted for the vast bulk of

respondents paymentsabout 28 per month compared with about seven per month for

emergingpaymentmethods.Note that these selfreportedmonthlypayments imply that the

typicalrespondentismakingslightlymorethanonepaymentperday(roughly35permonth);

unfortunately,wedonothaveindependentinformationtoverifytheaccuracyofthisnumber.

Looking by payment method only at respondents who have adopted each method

(yellowbars,Figure4),we find,notsurprisingly, that the incidenceofuse issomewhatmore

evenly divided across the spectrum of payment methods. Most notably, respondents who

adoptedadebitcard tend touse itmuchmore frequently thananyotherpaymentmethod is

usedby itsadopters (14.7 transactionspermonthonaverage). Inaddition, respondentswho

adoptedonlinebillpayment tend touse itaboutseven timespermonthonaverage,arateof

useapproachingthatoftheothermostcommonmethodsbytheiradopters.Respondentswho

adoptedACHorastoredvaluecardtendtousethemsubstantiallyless(aboutfourtransactions

permonthonaverage).Thus,evenamongusersofemergingpaymentmethods,wefindthe

useofthesemethodsisstillrelativelyinfrequent.Thereasonforthelowintensityofuseamong

usersoftheseemergingpaymentmethodsisunclear.Probably,itisexplainedpartlybylow

demand by respondents andpartly by limited supply, since formany types of transactions,

thesemethodsaresimplynotavailableoptionsatthepresenttime.

4.3 PaymentChoicesbyVenue

Oneof themost striking findings,andone thatmakes it clear thatagreatdealmore

study will be needed to gain a full understanding of consumer payment choice, is that

consumers payment choices are quite heterogeneous, varying by venue (retail store or

whatevervenueisusedforrecurringbillpayment),andwithinagivenvenue,bybillpayment

typeand(forretailstorepayments)bypaymentamount.

4.3.1 PaymentChoicesatRetailStores

Although checks are themost common paymentmethod overall, at retail stores Fed

respondents most often choose a debit (36 percent) or credit (31 percent) card to make

payments,asshown inFigure5. In fact,only12percentofFedrespondentsmakemostretail

21

paymentsbywritingchecks.Comparingour resultswith themore representativeABA/Dove

survey results, we see that Fed employees tend to favor checks more than does the U.S.

population(12percentversus7percent),andtousecashless(21percentversus26percent).But

overall,theresultsaresimilar.

Interestingly, Fed respondents use of payment methods at retail stores is far from

uniformacrosspaymentamounts,asshown in theremainingbarsofFigure5.35Forpayment

amountsbelow$20 (alsocalledmicropayments),cash isstillking (51percent).Formedium

sizedpayments ($2050),debit cardsaremost common (46percent),and for largepayments

(amountsabove$50),creditcardsaremostcommon(48percent).Oneshortcomingofthestudy

isthatthesurveydidnotobtainenoughinformationtodistinguishbetweenuseofcreditcards

forcredit(borrowingwithrepaymentovertime)anduseofcreditcardsfordeferredpayment

(interestfree loanornegative float)or incentives. In any case,payment size appears tobe a

factorinrespondentspaymentchoices.

4.3.2 PaymentChoicesforRecurringBills

Checksarestillthedominantpaymentmethodforrecurringbills,asshownintheupper

panelofFigure6,butrespondentstobothsurveysalsouseothermethods.Eightythreepercent

ofrespondentstotheFedsurveyand72percentofrespondentstotheABA/Dovesurveywrite

checkstopaysomebills.ACHpaymentsarethenextmostcommonamongFedrespondents,

followedbyonlinebillpaymentandcreditcards.Comparedwith theABA/Doveresults,Fed

respondents appear to be roughly representative ofU.S. consumers in their checkwriting.36

Heterogeneity of payment choice is evident from the data on payment choice (of Fed

respondents)bytypeofbillpayment,asshowninthebottompanelofFigure6.Abouthalfof

allrespondentspaytheirhousing,telephone,cable,andinsurancebillsbycheck,butonlyone

thirdofrespondentspaytheirInternetserviceprovidersbycheck.TheuseofACHpaymentsis

relativelywidespreadforhousing,insurance,andloanpayments,butrelativelyunpopularfor

telephone and cable bills. Roughly onethird of respondents use credit cards to pay their

Internetserviceproviders,butnomorethanaboutonetenthofrespondents(andinsomecases

35Wedonothaveresultsfromarepresentativesurveyagainstwhichtocomparetheseresults.36CreditcardswereexcludedfromtheABA/Dovesurvey.

22

hardly any) use this method to pay other bills. Understanding the root causes of this

heterogeneityinconsumerpaymentsacrossrecurringbilltypesisachallenge.

5. WhyConsumersPaytheWayTheyDo

Becauseourprimarygoalistodevelopabetterunderstandingofwhyconsumersmake

thepaymentchoicestheydo,asignificantportionoftheBostonFedconsumerpaymentsurvey

contains two types of questions designed to elicit this information. One type of question

exploreschangesinthewayconsumersmakepayments,askingthemwhatchangestheymade

and why. A second type of question explores the fundamental characteristics of payment

methodsthatlikelyinfluenceconsumersdecisionstoadoptandusethemethods,andseeksto

understand whether these characteristics are consistent with actual payment behavior. The

remainder of this section examines these two important dimensions of consumer payment

behavior.

5.1 ChangesinConsumerPaymentBehavior

Ideally,changesinconsumerpaymentbehaviorwouldbemeasuredusinglongitudinal

surveysinwhichthepaymentbehaviorofthesameconsumerswasfollowedconsistentlyover

time.However,because theBostonFedpaymentsurveywasoriginally intended tobeaone

time survey, we designed some retrospective questions to try to obtain information about

changesinindividualconsumerpaymentbehaviorovertime.Themostdirectquestionofthis

kindasked respondentswhether theyuse fewer,more,or thesamevolumeofeachpayment

method compared with three years earlier in June, 2001 (question 18). Followup questions

asked respondentswho said theywerewriting fewer checks than in thepastwhatpayment

methods theyhad substituted for checks, andwhy theyhad chosen thesemethodsboth in

generaland for retailor recurringbillpayments (questions1923,4446,and48).Additional

queries asked about payment method substitution, specifically for retail and recurring bill

payments.

5.1.1 ChangesinPaymentMethods:FromPapertoElectronics

Most respondents use of checks (for all types of payments) had declined over the

preceding three years, and their use of electronic payments had increased commensurately.

23

Figure 7 shows that 64percent of respondentswhowrite checks reportedhavingdecreased

theiruseofchecks inthepastthreeyears,andonly8percentreportedhaving increasedtheir

use,withtherestreportingaboutthesameuse.37

The largest increases intheuseofelectronicpaymentsoccurred inonlinebillpayment

andtheuseofdebitcards.EightyfourpercentofFedrespondentswhouseonlinebillpayment

increasedtheiruseofthatmethod,77percentofusersofdebitcardsincreasedtheiruseofdebit

cards,andfewerthan5percentofusersofeachofthesetwopaymentmethodsdecreasedtheir

useofthemethodinquestion.JustoverhalfofFedrespondentswhouseACHreportedhaving

increased theiruseofACH,withonlyonepercent reporting reduction in theiruseofACH.

Fortythree percent of users of storedvalue cards reported increased use of those cards,

althoughhalfreportedthattheirusehadnotchanged.

Ontheotherhand,usersofcreditcards,amoretraditionalandpreviouslymorewidely

usedtypeofelectronicpayment,reportedthattheyhadnotincreasedtheiruseofcreditcardsin

the past three years. Thirtyfive percent of respondents who used credit cards reported

increasedcreditcardusage,with42percent indicatingnochange,and24percentreportinga

decline.

5.1.2 ReplacementofChecksbyElectronicMethods,byBillPaymentTypeandValue

Electronicpayments are replacing checks in retailpayments and in awidevarietyof

recurring bill payments, as shown in Figure 8. Mortgage and rent payments are the most

common recurring bills forwhich Fed respondents eliminated checkwriting completely for

each given payment type and switched to electronic payments for that payment type (39

percent).Amongrespondentswhoswitchedfromwritingchecks,about30percentswitchedto

electronicstopaytheircreditcards,utilities,andinsurance,and22percentdidsofortelephone

bills.However,ofthosewhoswitchedfromwritingcheckstoelectronicpaymentmethods,only

14percent switched for their retailpurchases. In eachof these cases, the surveydatadonot

indicatedirectlywhetheralowpercentageindicatesthatconsumersaresubstitutingelectronic

paymentsforchecksbecausetheelectronicpaymentoptionshavebecomemoredesirableand 37 The results are quite similar for retail purchases only. Sixtyfive percent of respondents reportedwriting fewerchecks for retailpayments than theydid threeyearsearlier,with31percentwriting thesameamountandonly4percentwritingmore(question45).

24

feasible,orwhetherconsumerswerealreadywritingfewerchecksforthesepaymentsthanthey

hadthreeyearsearlier.Theformerseemsmorelikelyformanytypesofbillpayments(only6

percentofrespondentswrotemorechecksforbills,accordingtoquestion48);the latterseems

more likely for retailpurchases (65percentof respondentswrite fewerchecksat retail stores

thanformerly,accordingtoquestion45).

Changes in retail payments from checks to electronic payments were not uniform:

Respondentswhowrote fewer checks for retailpayments comparedwith threeyears earlier

chosedifferentelectronicpaymentmethodsdependingonthevalueoftheretailtransaction.38

Figure 9 shows what other payment methods respondents used as their checkwriting

decreased.Results for these changes in payment behavior between 2001 and 2004 are quite

similar to the resultsdepicted in Figure 5 for respondentspayment behavior in 2004.Most

respondentswho reduced theiruse of checks primarily replaced checkswith cash for small

purchasesunder$20 (43percent), followed closelybydebit cards (41percent).Formedium

sizedpurchasesof$20to$50,nearlytwothirds(62percent)ofthosewhoreducedtheiruseof

checks primarily substituted debit cards, with 31 percent using credit cards. But for large

purchasesover$50,onlyabouthalfofallrespondentswhoreducedtheirpapercheckwriting

optedfordebitcardsinstead,whilealmostasmany(48percent)chosecreditcards.Thosewho

reducedtheiruseofchecksmostdefinitelydidnotchoosestoredvaluecardstoreplacechecks

foranysizepurchase,andtheychosecashforsmallpurchasesonly.

It isdifficult tounderstandwhyconsumerschange theirpaymentmethods ingeneral,

andevenmorecomplextounderstandwhytheshiftinpaymentchoicedependsonthetypeof

bill and the value of a payment. Still,we offer some possible explanations.With respect to

payments in general, it may be that the convenience of setting up automatic payments for

recurringbillsisseenasastrongeradvantagethantheconvenienceofpayingelectronicallyat

retailstores.Withrespecttopaymentsatretailstores,itmaybethatconsumersdonotincrease

theiruseof cash as they reduce theiruseof checks for larger (>$20)purchasesbecause they

wanttoavoidcarryingagreatdealofcashasaresultofsecurityconcernsorotherreasons.The 38Here,Respondentswhowrote fewer checks, refers to any respondentwho simply reduced checkusageforretailpurchases,whereasthediscussioninthepreviousparagraphdealswithrespondentswhoeliminatedcheckwritingentirelyforagivenpaymenttype.

25

shiftfromcheckstocreditcardsforlarger(>$20)purchasesingeneralmaybeexplainedpartly

byloyalty(reward)programs.However,thefactthatthischeckstocreditcardsshiftisgreatest

for the largest purchases (>$50) is more difficult to understand fully without separately

identifying respondents use of credit cards to borrow money to pay for these very large

purchases.Moreover,formediumsizedretailpurchases,averyhighpercentageofthosewho

reduced theiruseofcheckschosedebitcards instead (62percent).Thisraises thequestionof

whytheserespondentsdonottakeadvantageoftheinterestfreeloanandincentivefeaturesof

creditcardsforthesepurchases,too.Perhapstheyareguardingagainstoverdraftsorbuilding

updebt.Thevarietyandcomplexityofpossibleexplanationsforconsumerbehaviorpayments

behaviorhighlightstheneedforfurtherresearch.

5.1.3 WhyDidRespondentsSubstituteElectronicPaymentsforChecks?

Whatexplains thegeneral shift frompaper checks toelectronicpaymentmethods for

retail purchases and recurring bills? Respondents reported that convenience and various

elements of cost were the primary influences on their decisions to substitute electronic

payments for checks, as shown in Table 4. Note, however, that the term convenience is

somewhatvague,bothinherentlyandbecauseitwasleftforrespondentstodefinesubjectively.

Asonlyoneresponsewasallowedandbecausethedefinitionofconveniencemayvaryacross

respondentsandmayoverlapwithothersimilarterms,interpretingtheresponsesissomewhat

difficult. In future research, a more granular definition of the characteristics of payment

methodsmayshedadditionallightonwhyconsumersasawhole,andbydemographicorother

characteristics,shiftfromcheckstospecificalternativepaymentmethods.

Foronlinebillpayment,whichshowedthelargestincreaseinuse,severaldiversefactors

promptedFedrespondentswhoswitchedfromwritingcheckstoelectronicmethodstochange

theirpaymentmethod.Respondentsswitchedfromcheckstoonlinebillpaymentsbecausethey

found that it is faster (22percentof responses)andgenerallymoreconvenient (21percentof

responses) to pay bills online. In addition, the greater ability to control the timing of their

payment (20percentof responses), saveonpostage (19percentof responses), and enjoy the

recordkeepingbenefitsofonlinebillpayments(13percentofresponses)wereallinfluentialin

thedecisionsofrespondentswhomadetheswitchfromcheckstoonlinebillpayment.

26

ThereasonsforswitchingfromcheckstoACHweresimilartothereasonsprovidedfor

switching to online bill payment. Convenience, lack of mail hassles, and not having to

remember to pay the bill each month were reasons cited in 19 percent of responses of

respondentswhoswitched,whilecostsavingssavingmoneyonpostageand theabsenceof

feeswereslightlylessimportantinthiscase(16percentofresponses).

Thirtyseven percent of the responses indicated that convenience was a factor for

respondentswhoswitched fromchecks todebitcards,while18percentof responsesshowed

cost (saving money on checks) as an influencing factor on respondents decisions. Thirteen

percentofresponsesindicatedthatbetterrecordkeepingwasafactorinrespondentsswitching

from checks to debit cards.Almost onethird of responses indicated that use of debit cards

increasedsimplybecausemorestoresnowacceptthemthanpreviously(asupplysidereason).

Conveniencewas also citedmostoftenby respondentswho switched from checks to

creditcards,butseveralaspectsofcreditcardcostwerecitedrelativelyfrequentlyaswell.The

abilitytobuynowandpaylater(17percent)makesthecreditcardpaymentanimplicitinterest

free loan.Incentives(rewards),suchasmileage,points,orcashback,promotedbycreditcard

use(15percent)alsoreducethenetcostofcreditcardpayments.Theseexamplesofcostsavings

arereductions in the fulleconomiccostof thepaymentmethod,notoutofpocketcashcosts.

However,savingmoneyonpostage(11percent),anoutofpocketcashcost,alsowasimportant

tosomerespondents.

5.2 DeterminantsofPaymentChoice

A consumersdecision to adopt anduse aparticularpaymentmethod is likely tobe

basedheavilyon the fundamental characteristics embodiedby thatpaymentmethod. In this

regard,theconsumerspaymentchoiceisessentiallynodifferentfromotherchoicesconsumers

maketouse(buy,orconsume)anyothertypeofgoodorservice.However,unlikethecaseof

consumptionofgoodsandservices,verylittleattentionhasbeenpaidthusfartothestudyof

paymentcharacteristicsandtherolethattheyplayinconsumerschoicesofpaymentmethods.

This subsection offers three types of evidence on the fundamental characteristics of

paymentmethodsthatinfluenceconsumersdecisionstoadoptandusevariousmethods.First,

for respondents who do not adopt a payment method, we describe the main barriers they

27

reportedtotheadoptionofpaymentmethods.Thesebarrierssuggestimportantcharacteristics

thatdiscouragecertainconsumersfromadoptingpaymentmethods.Second,wedescribehow

respondentsrateeachpaymentmethod,relativetochecks,withrespecttosevenfundamental

characteristics. Relative characteristics of payment methods can vary widely across

respondents, even within demographic categories, because of actual heterogeneity in the

paymentcharacteristicsattheindividuallevel.However,asweexplaininmoredetaillater,the

assessments of relative characteristics reported by respondents may differ from the actual

relativecharacteristicsof thepaymentmethodsforavarietyofreasons.Third,wepresenta

relatively simple comparison of respondents actual payment behavior with their reported

assessmentsofrelativepaymentcharacteristics.

5.2.1 BarrierstoAdoptionofPaymentMethods

Many respondents are now using electronic payment methods instead of checks for

sometransactions,andtheygaveusnumerousreasonstoexplainwhytheyswitched.Butwhy

arentmorepeoplemakinggreateruseofelectronicpaymentmethods?Foreachmethod,we

askedrespondentstocitethereasons,orbarriers,fornotadoptingtheparticularmethod.(This

was a multipleresponse question where we allowed respondents to select more than one

choice.)

Barriers to adoption vary considerably across payment types for two reasons, as

illustrated inTable5.39First, the rangeofpossiblebarriers isnot the same foreachpayment

typebecausesomebarriersarenotapplicabletocertainpaymentmethods(reflectedbymissing

values in a row). Second, the relative importance of the barriers is not the same for each

payment typeor foreachconsumer.Thecolorcodingof thebarriers indicates theirdegreeof

importance,asmeasuredbythepercentageofresponsescitingthebarrier.Redbarriersarethe

mostimportant(morethan15percentofresponses),andwhitebarriers,theleast(5percentor

fewer),withorangeandyellowinbetween.Inassessingtheimportanceofbarrierstopayment

methodadoption,thereadershouldbearinmindtheadoptionrateofeachmethodfromTable

39Table5isderivedfromquestions1017.Thenumbersinthetablearethepercentagesofresponsesthatcite the barrier as a factor (row) for respondents who do not use the payment method (column).Respondentswereallowed tochoosemore thanonebarrier.Forresultson thebarriers in termsof thepercentagesoftotalrespondents,seethesurveyresultsfiles.

28

2,which(excludingchecks)variesfrom86percentforcreditcardsto35percentforstoredvalue

cards.Thus,themostimportantbarriertotheadoptionofstoredvaluecardslikelyhasagreater

effectonoverallconsumerpaymentchoicethandoesthemostimportantbarriertotheadoption

ofcreditcards.

Scanning the columns of Table 5, one can see that the most important barriers to

adoptionvaryacrosspaymentmethods.Althoughsomemethodssharesimilarbarriers,notwo

paymentmethodshavethesamemostimportantbarrier,andeachpaymentmethodhasamost

importantbarrierthattheothersdonot.Themostimportantbarrierstocreditcardusearethe

amountofdebt in thehousehold (38percentof responses)and interestand feeexpenses (32

percentofresponses)oroutofpocketcost.Fordebitcards,thebarriersaremuchmorevaried,

withfiveratedasmostimportantorsecondmostimportant(atleast11percentofresponses).

Interestingly,twoelementsofthecostofcreditcardshaveoppositeeffectsontheadoptionof

debit cards. Incentives,whichmake credit cards cheaper relative todebit cards, are themost

important barrier to the use of debit cards (19 percent of responses), but interest and fee

expenses,whichmakecreditcardsdearer relative todebit cards,are tied for thirdplacewith

theftandmisuse,amongbarriers to theuseofdebitcards (13percentofresponses).Concern

abouttrackingpayments(14percentofresponses)isthesecondmostimportantbarriertodebit

carduse.ThemostimportantbarriertotheuseofACHpaymentscitedbynonusersofACHis

concernaboutoverdrafts(20percentofresponses),withconcernaboutthelackofflexibilityin

timing ofpayments coming in second (16percent of responses).Not surprisingly, themajor

concernsrelatedtoonlinebillpaymentswereprivacyandsecurity(22percentofresponses)and

theft(19percentofresponses).

Theseresultsonadoptionofpaymentmethodssuggestthatthebarriersaresodifferent

among theelectronicpaymentmethods that it isunlikely thatanysingle improvement to the

payments system will encourage consumers to increase their use of electronic payments in

general.However,respondentsgenerallywouldbewillingtouseelectronicpaymentsmore if

thespecificbarriersthatinhibitedthemforusingeachtypeofpaymentwereremoved(seethe

oddnumberedquestionsfrom1017).Theseresultssuggestwaysthatpaymentprovidersmay

have an opportunity to increase the use of their payment methods. For example, offering

29

consumersprotection against overdrawing accounts andmore flexibility onpayment timing

mightsignificantlyincreaseuseofACHpayments.

5.2.2 RelativePaymentCharacteristics

Tounderstandbetterhowconsumersthinkabouttheirpaymentchoices,weattempted

to determine the relative importance (or value) of some fundamental characteristics of the

paymentmethods.Ourstrategywastoaskrespondentstoselfreportthequalitativevaluesof

thecharacteristicsofeachpaymentmethodrelativetothevaluesofthesamecharacteristicsofa

benchmarkpaymentmethod(paperchecks).Asisthecasefortheentiresurvey,selfreporting

by respondents is susceptible to a variety of errors,whichmight be lessened if thedata on

relativepaymentcharacteristicswerecollected fromothersources.Furthermore, thepotential

weaknessof consumer selfreporting is exacerbated forpayment characteristicsbecause they

arehardtodefineasubjectwediscussinmoredetailinthenextsubsubsection.Nevertheless,

for thissubsubsection,weproceedunder theassumption that thereportedandactualrelative

paymentcharacteristicsarethesame.

Tokeep the comparisonmanageable,we asked respondents to compare each typeof

electronicpayment,oneatatime,withpapercheckswithrespecttosevencharacteristics:cost,

convenience,safety,privacy,abilitytoresolveerrors,abilitytocontrolthetimingofpayments,

andeaseof recordkeeping.Specifically,weasked fora simple,qualitative comparisonof the

valueofeachcharacteristicbyaskingthefollowingquestion:Istheelectronicpaymentmethod

better, the same as, orworse than a check?40 For each respondent, these value resultswere

converted toadiscretenumeric format (better is1.0,worse is1.0,and same is0.0).Table6

summarizes theresultsofourqualitativecomparisonofpaymentcharacteristicsbyaveraging

thequantitativevaluesassignedtothequalitativeassessmentsacrossrespondents.LikeTable5,

theentries inTable6arecolorcoded to indicate the relativevalue reflectedby theaverages.

Red shading indicates that the electronicpaymentmethod is ratedworse than checks,green

shadingbetter,andwhiteshadingthesame.Eachcolorhastwoshades,withthedarkershade

40Amorecontinuousandquantitativeevaluationwouldbepreferable,butwedecidedthatitwouldhavebeentoodifficulttoobtainaccurateanswersofthistypeinthesurvey.Datafromthiskindofqualitativeassessmentquestionhavesomenotorietyineconomicanalysis;forexample,seethesurveysofbusinessactivitybytheInstituteforSupplyManagement(ISM),whichcanbefoundathttp://www.ism.ws/.

30

http://www.ism.ws/

indicating greater intensity. Thus, green indicates better, darker green indicates much

better,redindicatesworse,anddarkerredindicatesmuchworse.

Our assessment of relative payment characteristics is adversely affected by a serious

error in our survey design. We asked only the respondents who use a particular payment

methodtoanswerthegroupofquestionsthatcomparepaymentmethodsbycharacteristic.But

togainafull,accurateunderstandingoftherelationshipbetweencharacteristicsandpayment

choice, one needs the payment characteristic comparisons of both users and nonusers.41

Fortunately, because some respondents did not follow directions, we actually did obtain

responsesforthepaymentcharacteristiccomparisonsforarelativelysmallandprobablyeven

moreunrepresentativenumberofnonusersof eachpaymentmethod.Thus,we can report

characteristic ratings separately for respondents who used the particular payment method

(Table6,toppanel)andforthosewhodidnot(Table6,bottompanel).42

The clear andpervasivemessage ofTable 6 is that respondentswhouse each of the

electronicpaymentmethods (toppanel)generallyview them asmarkedly superior topaper

checks.Of the35possiblecomparisons, inonly threecases is theelectronicpaymentmethod

ratedworse thanchecks (negativenumbers in the toppanelofTable6),while in27cases the

electronicpaymentmethodwas ratedby itsusersasbetterormuchbetter thanchecks (light

greenordarkgreeninthetoppanel).Respondentswhouseelectronicpaymentsviewthemas

strictlybetter thanchecks (value ispositive) in fourcharacteristics:cost,convenience, timing,

and errors. Respondents think electronic payment methods are overwhelmingly more

convenientthanchecksallfivepaymentmethodswereratedasmuchbetterforconvenience

thanchecks.Respondentsalsothinkthatelectronicpaymentmethodsaresignificantlybetterin

termsofcostandtiming.

41Unlessconsumersarecompletelyillogicalorconstrainedintheirbehavior,theusersofanyparticularpaymentmethodalmostsurelyrate thatparticularmethodbetter thanothermethods,asevidencedbytheirdecisiontouseit.Thus,datagatheredonlyfrompaymentuserswillgiveabiased(morefavorable)assessmentofrelativepaymentcharacteristicsthansuchassessmentsmadebyallconsumers.42Ofcourse,itisconceivablethatrespondentswhodonotfollowdirectionshavetheirownsetofbiasesand thus induce sample selectionproblems.We have noway to test or verify thispossibility, so thereadershouldkeepthispotentialflawinmind.

31

Ofall theelectronicpaymentmethods, respondentswhoused themethod ratedonly

oneACH paymentsas uniformly dominant over checks. In all seven categories, ACH

paymentswere ratedbetterormuchbetter than checks.Overall, it seems likely thatabetter

understandingoftheavailabilityofACHpaymentswouldleadtofurtherreductionsincheck

writing.However,withrespecttothisfinding,itisimportanttorecallonceagainthatFederal

ReserveemployeesareprobablynottypicalusersofACHpayments.

Incontrast,respondentswhodonotuseaparticularelectronicpaymentmethod(bottom

panel) tend to rate theelectronicmethodsabout the same,onaverage,as checks.Evennon

users of a particular payment method view the electronic payment method as better in

convenienceacrosstheboard,andtheytendtoviewACHasessentiallyequivalenttochecksin

allcharacteristics.However,nonusersseeseveralareasinwhichcreditcards,debitcards,and

online bill payment are significantly worse than checks. Cost, safety, and privacy are key

concerns for the respondentswhodonotuse these electronicmethods andwho rated them

worse than checks. Among these nonusers, recordkeeping (for debit cards) and payment

timing(forSVCs)arealsoviewedasworsethanchecks.Interestingly,respondentswhodonot

useSVCsnevertheless rate thembetter than checkson convenienceandprivacy;apparently,

someotherunobservedfactorsdiscouragedtheserespondentsfromusingstoredvaluecards.

The important result to keep inmindwith respect to the characteristics of payment

methodsisthatFedrespondentsoverallratechecksworsethanotherpaymentmethodsacross

most fundamental characteristics.Onlyamong the smallgroupof respondentswhoarenon

usersofelectronicpaymentmethodsdidwe findsomecharacteristics forwhichcheckswere

viewedassuperior.

5.2.3 RelativePaymentCharacteristicsandConsumerPaymentBehavior

A natural question arising from the evidence on relative payment characteristics is

whethertheyplayanimportantroleindeterminingactualconsumerpaymentbehavior.Inthis

subsubsection,weperforma simple comparison to try toanswer thisquestion, at leastata

32

cursorylevel.43However,firstwemustexaminetheconceptofrelativepaymentcharacteristics

inabitmoredetail.

Economistsgenerallyassumethatconsumersactintheirownbestinteresttomaximize

their utility, which depends positively on their consumption of goods and services and

negatively on their supply of labor for work. Implicitly, economists typically assume as a

benchmark thatconsumersaccuratelyandcompletelyobserve thefundamentalcharacteristics

ofthegoodsandservicesprice,color,quality,speed,andthelikethattheyconsume,thereby

enabling them tomake fully informeddecisionsandbehave in theirownbest interest. If this

benchmark assumption is correct for relative payment characteristics, thenwe can compare

themwiththeobservedpaymentbehavioroftherespondentwhoreportedthecharacteristicsto

seewhether thebehavior isconsistentwith thecharacteristics.Forexample,respondentswho

ratepaperchecksascostlierthananotherpaymentmethodshould(allelsebeingequal)usethat

otherpaymentmethodmorefrequentlythanchecks.44

However,numerousaspectsof therelativepaymentcharacteristicsandcorresponding

consumer payment behavior may complicate this simplistic analytical conclusion. Before

proceedingtoourcomparisonofpaymentcharacteristicsandbehavior,weprovideanitemized

discussionofthesekeyissues:

ExternalheterogeneityRelativepaymentcharacteristicsmost likelyvaryacross

individualconsumers,foravarietyofreasons.Onecanimaginethatthecostofa

paymentmethodmaybedifferentfortwoconsumers,eveniftheyhavesimilar

demographiccharacteristics.Forexample,checkingaccountshavedifferent fee 43SeeSchuhandStavins(2005)foramorerigorousandthoroughinvestigationofthisissues.Theyfindevidence that relative payment characteristics are much more important than demographics ineconometricmodelsthattrytoexplainconsumerpaymentbehavior.44Evenifconsumershaveaccuratelyandcompletelyobservedtheireconomicenvironment,theymaynotmakechoicesthatareintheirownbestinterest,orrational.Arelativelynewbranchoftheprofession,calledbehavioraleconomics,challenges thebasicpremise thatconsumersconsistentlymakechoices intheir best interest and offers evidence that consumers sometimes deviate from the rational behaviortraditionallyassumedbyeconomists(seetheoverviewsbyMullainathanandThaler(2000)orCamererandLoewenstein (2004)).Consumerpayment choice is an areawheremanyof the ideas advanced inbehavioraleconomicsmaybeevident in the realworld,buta formalevaluationofwhetherconsumerpaymentbehavior is consistentwith relativepayment characteristics requiresan economicmodel andstatisticaltests,whicharebeyondthescopeofthissurveyoverview.Forexamplesofthiskindofwork,seeZinman(2005)orMeierandSprenger(2006).

33

structures, interest rates, cost of paper checks, and other pecuniary features;

likewise,paymentserviceproviders(suchasbanks)maydiffer intheirabilities

to resolve errors or payment disputes with their customers. This kind of

characteristicheterogeneityisexternaltoconsumers,thustakenasgiven.

EndogeneityOnereasonpaymentcharacteristicsmaybeheterogeneousacross

consumers is that consumers can choose many of the characteristics of their

paymentmethods simultaneouslywith the choice of themethod itself. In the

preceding example of the checking account, a consumer who chose to write

checks intensivelywould likelysearchfor(andchoose)arelatively inexpensive

checkingaccount(lowfees,highinterest)andbuydiscountchecksactionsthat

work to lower the relative cost of checks for that consumer.Likewise, among

credit card users, the cost is heavily determined by a respondents decision

whethertocarrybalancesacrossmonths(therebyincurringoravoidinginterest

charges) and by the decision whether to pay in a timely manner (thereby

avoiding or incurring late fees and perhaps an increase in the interest rate

charged on future balances). For these reasons, we cannot view the relative

paymentcharacteristicsassomethingconsumerstakeasgivenwhentheychoose

theirpaymentmethods.

Measurement error Some payment characteristics are very difficult for

consumers (and perhaps even others) to measure properly. For example,

althoughidentitytheftisbeingdiscussedandpublicizedmuchmorefrequently

thanbefore, the actualunconditionalprobabilitiesofvarious formsof identity

theftarenotwellknown.Moreover,itislikelythattheconditional(onindividual

consumerbehaviorandcharacteristics)probabilitiesareevenmoredifficult for

individual consumers to estimate. Thus, reported relative characteristics

probablycontainsomemeasurementerror.

Experience Some payment methods may exhibit the characteristics of an

experiencegood,whichisonewhosepriceandfeaturesarenotfullyknownby

the consumer until the consumer actually consumes (experiences) it. For

34

example,consumersmaynotgraspaccurately theconvenience (easeofuse)of

onlinebillpaymentuntiltheyactuallysetuptheironlinebankingfacilitiesand

begintopaybillsthisway.45

Subjectivity, internalheterogeneity,andperceptionsSomepaymentcharacteristics

may be subjective from the perspective of the consumer. For example,

convenience(easeofuse)andrecordkeeping(easeoffinancialmanagement)are

characteristics thatmaybedetermined largelyby the tastesandpreferencesof

individual consumers. If so, the relative payment characteristicswould reflect

heterogeneity in consumerutility functions,which is internal to the consumer,

rather thana fundamental featureof theeconomicenvironment that influences

consumer decisions. But subjectivitymay run even deeper. Some consumers

may form perceptions about payment characteristics that have no systematic

relationship to measurable, objective characteristics of the economic

environment.Forexample,consumerswhoreadagreatdealaboutidentitytheft

inthenewspapermayperceivethatsafetyismuchlowerforonlinebillpayment

thanisactuallythecase(eveniftheyareshownevidencetothecontrary),orthan

do consumerswho rarely read the newspaper orwho already use online bill

payment.

Eachofthesefactorsmayinfluencerespondentsreportedrelativepaymentcharacteristicsand

may drive a wedge between the reported characteristics and the actual characteristics that

prevail in the marketplace.46 Nevertheless, actual payment behavior is most likely to be

determinedprimarilyby reportedpayment characteristics.Muchmoredataand researchare

neededtodistinguishmoreclearlybetweenreportedandactualpaymentcharacteristics,andto

45SeeGoettlerandClay (2006) foranexamplepertaining to theuseofonlinegroceryservices thathassimilaritiestothepaymentchoice.Experiencegoodsarealsosusceptibletopricediscrimination.46Forseveralpaymentcharacteristics,suchasprivacy,security,andpaymenttiming,onemightwonderwhetherthedifferenceinreportedcharacteristicsacrossrespondentsandthedivergence(ifoneexists)ofreported and actual characteristics arematters ofmisinformation that could perhaps be remedied byeducation.Forexample,apapercheck,whichhasonesname,address,andaccountnumberon it,andwhich passes through many individuals hands as it makes itsway through the clearing process, isprobablynotmoresecure thanadebitcard,especially if the latter isguardedbyaPIN,but individualconsumersmaynotbeawareofthisfact.

35

understand the extent to which deviations from actual characteristics actually influence

consumerpaymentbehavior.

Proceeding under the assumption that respondents reported payment characteristics

areaccurateassessmentsofactualpayment characteristics,we seek todetermine, ina rough

way, whether there might be some obvious, firstorder deviation of respondents payment

behaviorfromtherelativecharacteristicstheyreportedforthevariouspaymentmethods.Todo

so, we devised a simple way to evaluate whether respondents assessments of relative

characteristicswere consistentwith theirpaymentmethodusage.From the results onusage

(question9),weidentifiedthreetypesofrespondents:

Check payers: Respondents who wrote checks for more than 75 percent of their

monthlypayments.

Eclecticpayers:Respondentswhowrotechecksforbetween20and75percentoftheir

monthlytransactions.

Electronicpayers:Respondentswhowrotechecks for fewer than20percentof their

monthlytransactions.

ThesethreetypesofrespondentsarerepresentedbythedifferentcoloredbarsinFigure10.

Fed respondentsappear tobehave inamannerbroadlyconsistentwith their reported

assessmentsof relativepayment characteristics,apoint illustratedbyFigure10 in twoways.

Boththeintensityofuseofelectronicpaymentmethodsandtheadoptionofelectronicpayment

methodsareconsistentwiththereportedrelativesuperiorityofelectronicpaymentmethods.

Electronic payers (third, deepest row of bars) rated electronic payment methods

consistentlybetter thanchecks. Ineachof14possiblecategoriesusers (topgraph)andnon

users (bottom) across seven payment characteristicsmore electronic payers than eclectic

payers rated electronicmethods as superior to checks, andmore eclectic payers than check

payers rated electronic methods as superior to checks. For four characteristicscost,

convenience(ease),timing,andrecordkeepingmorethan50percentofelectronicpayersrated

the electronic payment methods that they use better than checks. In contrast, check payers

conceded the superiority of electronic payment methods in only one characteristic

convenience(easeofuse);fewerthanhalfofcheckpayers(nearestrowofbars)ratedelectronic

36

payments(whethertheyusethemornot)betterthanchecksinallothercharacteristics.Eclectic

payersratedelectronicpaymentsbetterthanchecksataratethatliesbetweentheratingsofthe

othertwogroupsofpayers,asonemightexpect.Thus,theintensityofel