understanding the internet banking adoption a unified theory of acceptance

Understanding the Internet banking adoption A unified theory of acceptance and use of technology and perceived risk application V U a C a b a A R R A A K U t P I I 1 m r z a i 2 i d c 2 c i 2 t s t.Understanding the Internet banking adoption A unified theory of acceptance and use of technology and perceived risk application V U a C a b a A R R A A K U t P I I 1 m r z a i 2 i d c 2 c i 2 t s t.

jo u r n al h om ep ag e :w w w e l s e v i e r c o m / l o c a t e / i j i n f o m g t

Viewpoint

Carolina Martinsa, Tiago Oliveiraa, Aleˇs Popoviˇca,b,∗

a Universidade Nova de Lisboa, ISEGI, Lisboa, Portugal

b Faculty of Economics, University of Ljubljana, Slovenia

a r t i c l e i n f o

Article history:

Available online 23 July 2013

Keywords:

Perceived risk

a b s t r a c t

UnderstandingthemaindeterminantsofInternetbankingadoptionisimportantforbanksandusers;our understandingoftheroleofusers’perceivedriskinInternetbankingadoptionislimited.Inresponse,we developaconceptualmodelthatcombinesunifiedtheoryofacceptanceanduseoftechnology(UTAUT) withperceivedrisktoexplainbehaviourintentionandusagebehaviourofInternetbanking.Totestthe conceptualmodelwecollecteddatafromPortugal(249validcases).Ourresultssupportsome relation-shipsofUTAUT,suchasperformanceexpectancy,effortexpectancy,andsocialinfluence,andalsothe roleofriskasastrongerpredictorofintention.ToexplainusagebehaviourofInternetbankingthemost importantfactorisbehaviouralintentiontouseInternetbanking

© 2013 Elsevier Ltd All rights reserved

1 Introduction

In recent years the Internet hasbeen growing and offering

manyWeb-basedapplicationsasanewwayfororganizationsto

retaincustomersandofferthemnewservicesandproducts(Tan

&Teo, 2000).In order for both parties (customersand

organi-zations)totakeadvantageof theseapplications,it is crucialto

analyzethegenuineperceptionandmainreasonsofpeople’s

will-ingnesstoadopt thesetechnologies(Lee, 2009;Liao &Cheung,

2002)

Internet banking has emerged as one of the most

prof-itable e-commerce applications (Lee, 2009) Most banks have

deployed Internet banking systems in an attempt to reduce

costs while improving customer service (Xue, Hitt, & Chen,

2011).DespitethepotentialbenefitsthatInternetbankingoffers

consumers, the adoption of Internet banking has been

lim-ited and, in many cases, fallen short of expectations (Bielski,

2003)

Whileearlierresearchhasfocusedonthefactorsinfluencing

theend-userITadoption,thereislimitedempiricalworkwhich

simultaneouslycapturesthesuccessfactors(positive)and

resis-tance factors(negative) that drivecustomers toadopt Internet

∗ Corresponding author at: Faculty of Economics, University of Ljubljana, Slovenia.

E-mail address: ales.popovic@ef.uni-lj.si (A Popoviˇc).

banking(Lee,2009).Buildinguponthepremisethatpurchasing Internetbankingservicesisperceivedtoberiskierthan purchas-ingtraditionalbankingservices(Cunningham,Gerlach,Harper,& Young,2005),thisstudyintroducestheperceivedriskfactor Draw-ingfromperceivedrisktheory,thisstudycouplesspecificperceived riskfacets (Featherman&Pavlou,2003)– namelyperformance, financial,time,psychological,social,privacy,andoverallrisk–with unifiedtheoryofacceptanceanduseoftechnology(UTAUT)to pro-poseanintegratedmodeltoexplaincustomers’intentiontoadopt anduseInternetbanking

Our research merges an existing and empirically validated theoreticalmodelwitha perceivedriskfactor, whichis alsoan importantconstructthatwillbetestedontheadoptionofInternet bankingforthefirsttime.Thus,thisstudymayhelpbanksto under-standthedeterminantfactorsthatinfluenceusersandtocreatethe rightpoliciesandactionstoattractcustomerstousethisservice Additionally,itisinthebanks’andclients’interesttodirecttheir communicationfrombankbranchestoonlinechannelsinorderto

bemoreproductiveandcost-effectiveforbothparties

Thestructureofthepaperisasfollows.Inthenextsectionthe conceptofInternetbanking,thecurrenttheoriesthatexplain cus-tomers’acceptanceoftechnology,thedefinitionofperceivedrisk, andearlierresearchonthistopicarepresented.Theresearchmodel

isthenconceptualized.Thesecondpartofthepaperpresentsthe researchdesign,methodology,andresults.Finally,theresultsare discussed,includingtheimplicationsfortheoryandpractice,and furtherpossibleresearchdirectionsareoutlined

0268-4012/$ – see front matter © 2013 Elsevier Ltd All rights reserved.

2 Theoretical background

2.1 TheconceptofInternetbanking

Concerningtheincreasinginnovationandurgentneedof

up-to-date,convenientandreliabledata,informationsystems(IS)have

gainedhighimportanceintheorganizationalcontext.Againstthis

background,agreatdependencybetweentheorganizations’

per-formanceandtheirISisemerging.Organizationscannowprofit

fromtheevolutionofnewtechnologiesandadapttothe

emerg-ingwaysofinteractingwiththeirclients.Thebankingsectorhas

beenusingISnotonlytoruninternalbusinessactivitiesandto

promoteproducts,butalsotoprovidemainservicestotheir

cus-tomers.Thedematerializationofcustomerrelationships, thatis,

thebetteruseofthenumerousnewISavailableinthemarket,isa

topicalchallengefacingthissector.Adjustingtothischallengewill

allowclientstosatisfyalmostalltheirbankingneedswith

mini-mumhumanintervention(Jayawardhena&Foley,2000;Tan&Teo,

2000)

Internet banking is defined as the use of banking services

throughthe computer network (the Internet), offeringa wider

rangeofpotentialbenefitstofinancialinstitutionsdue tomore

accessibilityand userfriendlyuseofthetechnology(Aladwani,

2001;Yiu,Grant,&Edgar,2007).Literaturesuggestsmanyconcepts

toidentify Internet banking, namelyelectronic banking, online

banking, and e-banking With Internet banking, customers can

perform,electronically,awiderangeoftransactions,suchas

writ-ingchecks,payingbills,transferring funds,printingstatements,

andinquiringaboutaccountbalancesthroughthebank’s

website-bankingsolution.Furthermore,Internetbankinghasasignificant

impacton e-payments, offering a platform to support many

e-commerceapplications,suchas onlineshopping,onlineauction,

andInternetstocktrading(Aladwani,2001;Lee,2009;Tan&Teo,

2000)

WhenInternet banking becamepopular,it wasusedmainly

toprovide informationfor marketingtheproductsandservices

onthebank’swebsite,butwiththetechnologicaldevelopment

ofsecuredelectronictransactions,more bankshave beenusing

italsoasatransactionalframework(Tan&Teo,2000;Yiuetal.,

2007).Recently,onlinebankshavebeenexpandingtheirpresence

in themarket (including thePortuguese market) and adopting

other channels, such as call centres, but their impact on the

wholebankingsectorhasbeenlimited(DECO,2010;Tan&Teo,

2000)

Pikkarainen,Pikkarainen,Karjaluoto,andPahnila(2004)

high-lightedtwomainreasonsforthedevelopmentandproliferation

ofInternetbanking.First,thecostsavingsbythebankscompared

withthe traditional channels; second, thereduction of branch

networksand, therefore,thecostswithstaff.Jayawardhenaand

Foley (2000) also identified the benefit of increasing the

cus-tomerbase,becauseusingmultipledistributionchannels(branch

networks,Internetbanking,mobilebanking,etc.)amplifiesmarket

coveragebyenablingdifferentproductstobetargetedatdifferent

demographicsegments.Witha largercustomerbase,bankscan

profitfrommarketingandcommunication,withthepossibilityof

masscustomizationforeachgroupofclients,offeringinnovative

products.Thisisanimportantissuebecausemanyorganizations

todayaresaturatedwithmassautomationandhomogenized

prod-uctsandservices.Inthecustomerview,thereisanincreaseinthe

autonomy,withlessdependencyonthebranchbankingand,

conse-quently,lesstimeandeffort.Recently,thePortugueseAssociation

ofConsumerDefence(DECO)performedastudyaboutcostsand

benefitsofInternetbankingusageandconcludedthatuserscan

savemorethanD300peryeariftheyusetheseservicesinsteadof

thetraditionalones(DECO,2012).OntheInternetplatform,users

canbenefitfromfinancialproductsthatareonlineexclusive,and

thesemayhavehigherreturnsthanthoseinthetraditional chan-nelsofbanks

RegardingtheprofileofInternetbankingcustomers,theyhave

anincreasedbankingactivity,acquiremoreproducts,and main-tainhigherassetandliabilitybalances,demonstrating thatthey aremorevaluablethanthetraditionalones(Hitt&Frei,2002;Xue

etal.,2011).Additionally,customerswhohavegreatertransaction demandandhigherefficiency,andresideinareaswithagreater densityof onlinebankingadopters, arefaster toadoptInternet banking.Theseadoptersalsohavealowerpropensitytoleavethe bank

LookingatthecurrentsituationinPortugal,weseethatthere aremanyInternetplatformsavailableinalmostallleadingbanks Since 2005 theuse of Internet banking servicesby Portuguese banking consumers has increased by 82%, while personal and telephone contacts have decreased approximately 17% (Grupo Marktest,2011,2012).Despitethisrecentsurgeintheuseof Inter-netbankingservices,manybankingusers(approximately70%)are not comfortable withthis channel and prefer tousethe tradi-tionalones(AutomatedTellerMachine–ATM,personalcontact,and telephonecontact).GrupoMarktesthasalsoundertakena char-acterizationofInternetbankingadoptersandconcludedthatthey aremen,young(25–34years),andfrommedium/upperclassesof society.Regardingthetypeofjob,theyfoundthatmedium/upper managementhaveanadoptionrate2.5timesabovetheaverage, with74%ofthemusingit

Despitetheincreaseinadoptionofthesekindsofservice, con-sumersstillshowsomereluctancetowardsthem,duemainlyto riskconcernsandtrust-relatedissues(Lee,2009)

2.2 Adoptionmodels TheacceptanceanduseofITsystemshasbeenthesubjectof muchresearch,andinrecentyearsseveraltheoriesthatoffernew insightshaveemergedatboththeindividualandorganizational levels,focusedonacountryorasetofcountries(Im,Hong&Kang,

2011).Eachoftheseveralmodelsthathavebeenproposedinthe literaturehasthesamedependentvariable,useorintentiontouse, butwithvariousantecedentstounderstandacceptanceof technol-ogy

Themostwell-knowntheoreticalmodelsattheindividuallevel thathavesoughttoexplaintherelationshipbetweenuserbeliefs, attitudes,andintentionsincludeTheoryofReasonedAction(TRA– Fishbein&Ajzen,1975),TheoryofPlannedBehaviour(TPB–Ajzen,

1991),andTechnologyAcceptanceModel(TAM–Davis,1989).TAM wasdesigned topredictinformationtechnologyacceptanceand useonthejob,inwhichperceivedusefulnessandperceivedease

ofusearethemaindeterminantsoftheattitudes(Davis,1989) TPB ismore focusedontheperceivedbehavioural control,that

is,the perceivedease ordifficultyof performingthebehaviour (Ajzen,1991).BothmodelswerebasedonTRA,whichproposesthat beliefsinfluenceattitudesthatinturnleadtointentionsandthen consequentlygeneratebehaviours(Fishbein&Ajzen,1975).Itisa modeldrawnfromsocialpsychology,andisoneofthemost impor-tanttheoriesofhumanbehaviour.Accordingtotheresearchers, attitude(attitudetowardsperformingbehaviour)andsubjective norms(socialpressurestoperformbehaviour)areconsideredas thedeterminantsofbehaviourinTRA

Venkatesh, Davis, Davis, and Morris, (2003) provide a com-prehensive examination of eight prominent models and derive

a Unified Theory of Acceptance and Use of Technology (UTAUT), which can explain as much as 70% of the variance in inten-tion The eight models studied by these researchers are TRA, TAM, Motivational Model (MM – Davis, Bagozzi, and Warshaw,

1992),TPB,ahybridmodelcombiningconstructsfromTAMand TPB(C-TAM-TPB –Taylor&Todd,1995), ModelofPCUtilization

Fig 1.Research model of Venkatesh et al (2003) investigation.

(MPCU–Thompson,Higgins,andHowell,1991),Innovation

Diffu-sionTheory(IDT–Moore&Benbasat,1996),andSocialCognitive

Theory(SCT–Compeau&Higgins,1995).TheUTAUTmodel(Fig.1)

postulatesthatfourconstructsactasdeterminantsofbehavioural

intentions and use behaviour: (i) performance expectancy, (ii)

effortexpectancy,(iii)socialinfluence,and(iv)facilitating

condi-tions.Inaddition,UTAUTalsopositstheroleoffourkeymoderator

variables:gender,age,experience,andvoluntarinessofuse

Sinceitsinceptionin2003,researchershaveincreasinglyturned

totesting UTAUTtoexplain technologyadoption It wastested

andappliedtoseveraltechnologies,suchasonlinebulletinboards

(Marchewka,Liu,&Kostiwa,2007),instantmessengers(Lin&Anol,

2008),andWeb-basedlearning(Chiu&Wang,2008).Forinstance,

theadoptionfactorsofInternet bankingand mobilebankingin

Malaysiawere investigatedby Tan,Chong, Loh,and Lin (2010)

withtheuseofthissamemodel;Im etal.(2011)undertookto

discoveriftheUTAUTconstructswereaffectedbytheculture,

com-paringthemp3playerandInternetbankingtechnologiesinKorea

andtheUS;andYuen,Yeow,Lim,andSaylani(2010)testedthe

UTAUTmodelintwogroupsofculturallydifferentcountries,i.e

thedeveloped(USandAustralia)anddeveloping(Malaysia)

coun-tries

Muchresearch has addressed Internet banking adoption, as

showninTable1.Therewefindthemainconclusionsofeach

inves-tigationanditspredictivepowerinexplainingintentionanduseof

Internetbankingservices,byther-square(whenavailable)

2.3 Earlierresearchonperceivedrisk

AccordingtoBauer (1960)and Ostlund (1974),thenegative

consequences that may arise from consumers’ actions lead to

an important well-establishedconcept in consumer behaviour:

perceivedrisk.Manyauthorshavestudiedtheimpactofriskonthe

adoptionofInternetbankingandsomeofthemwillbediscussed

Kuisma,Laukkanen,andHiltunen(2007)investigatedthe

resis-tance to Internet banking and their connections to values of

individualsandconcludedthatbothfunctionalandpsychological

barriersarisefromservice,channel,consumer,and

communica-tion.ATMservicesarestillpreferredbycustomers,becauseoftheir

oldroutineandtheInternet’sinsecurity,inefficiency,and

inconve-nience.Besidesthefearofpossiblemisuseofchangeablepasswords

andthelackofproofprovidedbyanofficialreceipt,theyfoundthat somecustomersseemtoperceivenoperformance-to-pricevalue duetothehighpurchasingcostsofacomputerandInternet connec-tion.Additionally,non-usersalsocomplainaboutthelackofsocial dimension,thatis,theabsenceofaface-to-faceencounter,asata branch

Inasimilarway,RotchanakitumnuaiandSpeece(2003) inves-tigated howcorporate customersperceivebarrierstousing the InternetbankingprovidedbyThaibanks.Thefindingswerethat trustandsecurityarethemostcriticalissues,especiallyamongst non-users who havehigher levelsof worry,do nothave confi-dencetomakeanyfinancialtransactionsviatheWeb,andhave

nointentionofadoptingInternetbankingservices

AccordingtoFeathermanandPavlou(2003),perceivedriskis definedas“thepotentialforlossinthepursuitofadesired out-comeof usingane-service” Thepurposeoftheirresearchwas

todiscoverhowimportanttheriskperceptionsaretothe over-alle-servicesadoptiondecision,integratingTAMwithperceived risk (research model in Fig 2) They identified seven types of risks, namely (i) performance risk, (ii) financial risk, (iii) time risk, (iv) psychological risk, (v) social risk, (vi) privacy risk, and (vii) overall risk The authorsstated that it wascrucial to includeameasureofperceivedriskintoTAMbecauseconsumers identifyandvalueriskwhenevaluatingproducts/servicesfor pur-chase/adoption,whichmaycreateanxietyanddiscomfortforthem Therefore, regardingperceivedrisk, theytested (i)ife-service’s

Table 1

Technology acceptance model (TAM) and self-efficacy as

one of the antecedent variables such as risk, Internet

experience, facilitating conditions

• Self-efficacy plays a prominent role in influencing the intention to use Internet banking in South Korea

• 32.3% of intention explained by experience, perceived usefulness, and perceived ease of use

• 4.8% of use explained by intention

Lee and Chung (2011)

• Attitudinal (relative advantage, compatibility with respondent’s values, experience, needs, trialability, and risk) and perceived behavioural control factors as the

Tan and Teo (2000)

• 12.4% of intention explained by the model

Pikkarainen et al (2004)

important control variables

• Perceived usefulness and perceived ease of use, resistance to change, trust, age, gender, education, and

online banking use Attitudes towards use explain 83% of the variance in intention

Al-Somali et al (2009)

perceived risk

• Perceived usefulness is the strongest predictor of

perceived ease of use and perceived risk

Yiu et al (2007)

• 80% of intention explained by security risk, financial risk, perceived behaviour control, subjective norm, attitude,

Lee (2009)

(UTAUT), trust, awareness of service, output quality,

• All constructs contributed to explain intention and use of internet banking, except social influence

important to explain intention

Riffai et al (2012)

cognitive theory (SCT)

indirectly play significant roles in influencing the intention

• Perceived ease of use has a significant indirect effect on intention to adopt/use through perceived usefulness, while its direct effect on intention to adopt is not significant

Chan and Lu (2004)

attitudinal factors (features of the web site and perceived

control factor (external environment), but not by

Bussakorn and Dieter (2005)

perceivedriskreducestheirperceivedusefulnessandadoption;

(ii) if perceived ease of use of e-service significantly reduces

perceivedrisksofusage;(iii)ifperceivedeaseofuseinfluences

e-service’sadoption.Asseenbelow,perceivedriskhasbeen

mod-elledasacompositevariableanddecomposedintoitstheorized

sub-facets

3 Research model

Asseenabove,theUTAUTmodelisabletoexplain70%ofthe

varianceinusageintention,whichisasubstantialimprovement

over anyof theeight original modelsused tobuild it Thus, it

demonstratesthatUTAUTisthemostcompletemodeltopredict

informationtechnologiesadoption,anditisthereforeusedinthis

investigation.Accordingtothismodel,threeconstructsare

signif-icantdirectdeterminantsofintention(performanceexpectancy,

effortexpectancy,andsocialinfluence).Facilitatingconditionsand

intentionexplainusebehaviour.Regardingthemoderatingeffects,

both experienceand voluntarinessof uselie outside thescope

of this research Experience is not evaluated because only one

momentintime is beingobserved;voluntarinessof useisalso

notfeasiblebecausenooneisobligedtouseInternetbankingin

thiscontext.Asgenderandagemayhaveaconsiderableinfluence

onusers’acceptanceofInternetbanking,bothwillbeconsidered

(Wang,Wang,Lin,&Tang,2003)

As our investigation merges two sensitivesubjects, namely

moneyandInternet, thereisalwaysariskfactor thatis

impor-tanttobemeasuredintheprocessofInternetbankingadoption

Usersalways fear losing money withtransactions, losing pass-words,makingerrorsontheplatform,etc.Wethereforepropose

totesttheUTAUTonInternetbanking,addingariskfactortothe model.Inthissection,wedefineeachofthedeterminantsofUTAUT andriskfactorandspecifytheroleofkeymoderators

Performanceexpectancy(PE)reflectsuserperceptionof perfor-manceimprovementbyusingInternetbankingontasks,i.e.,itisthe degreetowhichanindividualbelievesthatusingInternetbanking willhelptoattaingainsinperformingbankingtasks(Venkatesh

etal.,2003).Itreflectsuserperceptionofperformance improve-mentbyusingInternetbanking,suchasconvenienceofpayment, fastresponse,andserviceeffectiveness(Zhou,Lu,&Wang,2010) Accordingtotheauthors,itissimilartotheperceivedusefulness

ofTAMandtherelativeadvantageofIDT.Effortexpectancy(EE)is thedegreeofeaseassociatedwiththeuseofInternetbanking.Itis equivalenttotheperceivedeaseofuseofTAMandthe complex-ityofIDT.AccordingtoUTAUT,effortexpectancypositivelyaffects performanceexpectancy.Whenusers feelthatInternetbanking

iseasy touseand doesnotrequiremuch effort,theywillhave

ahighexpectationtowardsacquiringtheexpectedperformance; otherwise,theirperformanceexpectancywillbelow(Zhouetal.,

2010).Social influence (SI)reflects the effect of environmental factorssuchastheopinionsofuser’sfriends,relatives,and supe-riorsonuserbehaviourandissimilartosubjectivenormofTRA (Venkatesh etal., 2003).Theiropinionswillaffect user’s inten-tiontoadoptInternetbankingservices.Facilitatingconditions(FC) reflecttheeffectoforganizationalandtechnicalinfrastructureto supporttheuse ofInternet banking, suchasuser’s knowledge, ability, and resources (Venkatesh et al., 2003) It is similar to

perceivedbehaviouralcontrol ofTPB Internetbanking requires

userstohavecertainskillssuchasconfiguringandoperating

com-puters,andconnectingtotheInternet.Inaddition,usersneedto

bearusagecostssuchasdataserviceandtransactionfeeswhen

usingInternetbanking.Ifusersdonothavethesenecessary

finan-cialresourcesand operationalskills, theywillnotadoptor use

Internetbanking(Hong,Thong,Moon,&Tam,2008;Zhouetal.,

2010)

Therefore,andaccordingtotheUTAUTmodel,itcanbe

postu-latedthat:

H1. TheinfluenceofPerformanceExpectancy(PE)onBehavioural

Intention(BI)willbepositiveandmoderatedbyageandgender,

suchthatitwillbestrongerforyoungerindividualsandmen

H2. TheinfluenceofEffortExpectancy(EE)onBehavioural

Inten-tion(BI)willbepositiveandmoderatedbyageandgender,such

thatitwillbestrongerforyoungerindividualsandwomen

H3. TheinfluenceofSocialInfluence(SI)onBehaviouralIntention

(BI)willbepositiveandmoderatedbyageandgender,suchthatit

willbestrongerforolderindividualsandwomen

H4. The influence of Facilitating Conditions (FC) on Usage

Behaviour(UB)willbepositiveandmoderatedbyage,suchthat

itwillbestrongerforolderindividuals

Tomaintainconsistencywiththeunderlyingtheoryforallofthe

intentionmodels,itisexpectedthatbehaviouralintentionwillhave

a significantpositiveinfluence ontechnologyusage(Venkatesh

etal.,2003).Itcanbehypothesizedthat:

H5. Behavioural Intention (BI) willhave a significant positive

influenceonUsageBehaviour(UB)

AccordingtoFeathermanandPavlou(2003),(i)performance

riskisdefinedasthepossibilityoftheresultsnotbeingasthey

weredesignedtobeandthereforefailingtodeliverthedesired

ben-efits;(ii)financialriskreflectsthepotentialmonetarylossfromthe

initialpurchaseoftheproductanditssubsequentmaintenance;

(iii)timeriskoccurswhenuserslosetimebymakingpoor

pur-chasingdecisions,withresearchingandmakingthepurchase,and

learninghowtouseit;(iv)psychologicalriskisdefinedastherisk

thattheperformanceoftheproductwillhaveanegativeeffecton

theconsumer’speaceofmindandthepotentiallossofself-esteem

fromthefrustrationofnotachievingabuyinggoal;(v)socialrisk

reflectsthepotentiallossofstatusinasocialgroup,asaresultof

adoptingaproductorservice;(vi)privacyriskisthepotentialloss

ofcontroloverpersonal information,suchaswhen information

aboutanindividualisusedwithoutthatperson’sknowledge;(vii)

finally,overallriskisageneralmeasurewithallcriteriatogether.All

theseperceivedriskscomprisetheperceivedrisk,beingasecond

orderfactorofthem,andinfluencingtheintentionnegatively.Itis

expectedthatthemoretheuser’saversiontotheriskconcernsare

lowered,themores/heislikelytoadoptInternetbankingservices

(Bussakorn&Dieter,2005)

Thus,perceivedrisk hasbeenmodelledboth asa composite

variableanddecomposedintoitstheorizedsub-facets,andwecan

postulatethat:

H6. PerceivedRisk(PCR)isasecondorderfactorofsevenrisks

H6a. PerceivedRisk(PCR)willpositivelyinfluencePerformance

Risk(PFR)

H6b. PerceivedRisk(PCR)willpositivelyinfluenceFinancialRisk

(FR)

H6c. PerceivedRisk(PCR)willpositivelyinfluenceTimeRisk(TR)

H6d. PerceivedRisk(PCR)willpositivelyinfluencePsychological

Risk(PSR)

H6e. PerceivedRisk(PCR)willpositivelyinfluenceSocialRisk(SR)

H6f. PerceivedRisk(PCR)willpositivelyinfluencePrivacyRisk (PR)

H6g. PerceivedRisk(PCR)willpositivelyinfluenceOverallRisk (OR)

H7. Perceived Risk (PCR) will negatively influence Behaviour Intention(BI)

Regardingtheeffectsofperceivedusefulnessandperceivedease

ofuseintheapproachofFeathermanandPavlou(2003),whenwe focusontheresearchofVenkateshetal.(2003),theequivalent constructsareperformanceexpectancy(PE)andeffortexpectancy (EE).Itisexpectedthatonlyindividualswhoperceiveusing Inter-netbankingasalowriskundertakingwouldhaveatendencyto perceiveitasuseful(Chan&Lu,2004).Also,itisexpectedthatonly thosewhoperceivelowefforttouseInternetbankingwouldhave

atendencytoperceiveitasanotriskyservice.Asaresults,wecan postulatethat:

H8. PerceivedRisk(PCR)willnegativelyinfluencePerformance Expectancy(PE)

H9. Effort Expectancy (EE)will negativelyinfluence Perceived Risk(PCR)

FromthesehypothesestheconceptualmodelshowninFig.3 emerges

4 Methods

4.1 Measurementinstruments Allmeasurement items wereadapted, with slight modifica-tions, from theliterature – PE,EE, SI, FC and BI wereadopted fromVenkateshetal.(2003)andDavis(1989);UBfromImetal (2011); perceived risk constructs from Featherman and Pavlou (2003).TheitemsforallconstructsareincludedintheAppendix

A ThequestionnairewasinitiallydevelopedinEnglish,basedon theliterature,andthefinalversionwasindependentlytranslated intoPortuguesebyaprofessionaltranslator.Thequestionnairewas putontheWebthroughafreeWebhostingservice

Most items were measuredusing seven-point Likert scales, rangingfromtotallydisagree (1)tototallyagree (7).Behaviour Intention(BI)wasmeasuredbyasking respondentsabouttheir intentions and plans to use the technology during the next months.To evaluateUsageBehaviour (UB),oneitem measured users’actualfrequenciesofInternetbankinguse(havenotused, once a year,oncein six months,once inthree months,once a month, once a week, oncein 4–5 days, once in 2–3days, and almosteveryday).Wealsoincludedtwodemographicquestions relating toageandgender.Age wasmeasuredinyears Gender wascodedusinga0or1dummyvariablewhere1 represented women

4.2 Datacollection Firstly,apilotsurvey(with100answers)wasconducted(inApril

of2012)inordertorefinethequestionsandgainadditional com-mentsonthecontentandstructure.Themostimportantchange wasintheitemsofUsageBehaviour(UB),whichinitiallywerefrom Venkateshetal.(2003).Thesegeneratedmisunderstandingsand thesimulationofthePLSestimationwithafewsamplesgave sta-tisticallypoorresults.Theitemswere“Iintendtousethesystem

inthenext<n>months.”,“IpredictIwoulduseInternetBanking

inthenext<n>months.”and“Iplantousethesysteminthenext

Fig 3.Research model.

<n>months.”.Thepossibleanswerswerefrom1to+12.Internet

bankingusersunderstooditastheperiodthattheyeffectivelywill

useInternetbanking(andthereforeanswered+12)andothersas

thenearestmonththattheywilluseit(thatis,nextmonth,with

1asresponse).TheseitemswerereplacedbyonefromImetal

(2011),alreadyusedinthiscontext.Regardingtheotheritems,a

numberofsuggestionsweremadeaboutthephrasingandthe

over-allstructureofthequestionnaire.Thesuggestionswerediscussed

andsomechangesweremade.Thedatafromthepilotsurveywere

notincludedinthemainsurvey

Atotalof726studentsandex-studentsfromauniversitywere

contactedbye-mailinMayof2012andprovidedwiththe

hyper-linkof thesurvey,fromwhich 173responseswerevalidated.A

seconde-mailwasthensenttothosewhohadnotrespondedafter

twoweeks,andfinally,aftertherefiningprocess,atotalof249

validcaseswereanalyzed(34%responserate).Totestfor

nonre-sponsebias,wecomparedthesampledistributionofthefirstand

secondrespondents groups We usedtheKolmogorov–Smirnov

(K–S)testtocomparethesampledistributionsofthetwogroups

(Ryans, 1974).The K–S test suggests that the sample

distribu-tionsof the two independent groups do not differ statistically

(Ryans,1974).Thismeansthatnonresponse biasis notpresent

Further,we examinedthecommonmethodbias byusing

Har-man’s one-factor test (Podsakoff, MacKenzie, Lee, & Podsakoff,

2003).Thesetestsfoundnosignificantcommonmethodbiasin

ourdataset

Themajorityofrespondents(63%)respondedthat theyused

Internetbankingservicesonceaweek.Fourteenpercentadmitted

thattheyarenon-usersandofthese,almostallweremenwithan

averageageof27,andcharacterizedbylowlevelsofeducation

Concerningdemographicdata(Table2 59%oftherespondents

aremaleandtheaverageageis30years.Theireducationlevelis

elementaryandhighschoolfor47%ofindividuals;theothershave

anundergraduatedegreeormore

5 Results

Structuralequationmodelling(SEM)isastatisticaltechniquefor testingandestimatingcausalrelationsusingacombinationof sta-tisticaldataandqualitativecausalassumptions.Carefulresearchers acknowledgethepossibilitiesofdistinguishingbetween measure-mentandstructuralmodelsandexplicitlytakemeasurementerror intoaccount(Henseler,Ringleand,&Sinkovics,2009).Thereare twofamiliesofSEMtechniques:(i)covariance-basedtechniques and(ii)variance-basedtechniques.Partialleastsquares(PLS)isa variance-basedtechniqueandisusedinthisinvestigationsince: (i)notallitemsinourdataaredistributednormally(p<0.01based

onKolmogorov–Smirnov’stest);(ii)theresearchmodel hasnot beentestedintheliterature;(iii) theresearchmodel is consid-eredascomplex.SmartPLS2.0M3(Ringle,Wende,&Will,2005) wasthesoftwareusedtoanalyzetherelationshipsdefinedbythe theoreticalmodel

Inthenexttwosubsectionswe(first)examinethemeasurement modelinordertoassessinternalconsistency,indicatorreliability, convergentvalidity,anddiscriminantvalidity,and(secondly)test thestructuralmodel

5.1 Measurementmodel Firstly,inordertoanalyzetheindicatorreliability,factor load-ingsshouldbestatisticallysignificantandpreferablygreaterthan 0.7 (Chin, 1998; Hair&Anderson,2010; Henseleretal., 2009) Means,standarddeviations,loadings,andt-statisticvaluesfrom itemsmeasuredareinTable3.Thet-statisticobtainedfrom boot-strapping(250iterations)showsthatallloadingsarestatistically significantat1%.FC4itemwasexcludedduetoitslowloadingand lackofstatisticalsignificance.Allotheritemswereretained Fur-thermore,itispossibletoconcludethatallitemshaveloadings

Table 2

Accord-ingtoHairandAnderson(2010),CRquantifiesthereliabilityand

internalconsistencyofeachconstructandtheextenttowhichthe itemsrepresenttheunderlyingconstructs.Additionally,CRtakes intoaccountthatindicatorshavedifferentloadings(andCronbach’s alphadoesnot),andisthereforemoresuitableforPLS,which prior-itizesindicatorsaccordingtotheirindividualreliability(Henseler

etal.,2009).AsseeninTable4,CRandCAforeachconstructare abovetheexpectedthresholdof0.7,showingevidenceofinternal consistency

Performance expectancy (PE)

Effort expectancy (EE)

Social influence (SI)

Facilitating conditions (FC)

Perceived Risk

Performance risk (PFR)

Financial risk (FR)

Time risk (TR)

Privacy risk (PR)

Overall risk (OR)

Behaviour intention (BI)

Table 4

Means, standard deviations, correlations, and reliability and validity measures (CR, CA, and AVE) of latent variables.

EE 5.65 1.29 0.96 0.94 0.78 *** 0.92

Diagonal elements are the square root of the average variance extracted (AVE).

PE, performance expectancy; EE, effort expectancy; SI, social influence; FC, facilitating conditions; PCR, perceived risk; BI, behavioural intention; UB, usage behaviour; NA, not applicable.

* p < 0.05.

** p < 0.01.

*** p < 0.001; all other correlations are insignificant.

In order to assess convergent validity, average variance

extracted(AVE)wasused.TheAVEistheamountofindicator

vari-ancethatisaccountedforbytheunderlyingitemsofconstructand

shouldbegreaterthan0.5,sothatthelatentvariableexplainsmore

thanhalfofthevarianceofitsindicators(Hair&Anderson,2010;

Henseleretal.,2009).AsisalsoseeninTable4,AVEforeach

con-structisabovetheexpectedthresholdof0.5,ensuringconvergent

validity

Finally, to grant discriminant validity, the square root of

AVE should begreater than the correlations between the

con-struct (Henseler et al., 2009) This is also reported in Table 4

for all constructs We conclude that all the constructs show

evidence of discrimination Additionally, another criterion that

assesses discriminant validity is the cross loadings, which

should be lower than the loadings of each indicator (Hair &

Anderson,2010).Thiswasalsoanalyzed andwe foundthatno

indicatorhasloadingswithlower values than theircross

load-ings

5.2 Structuralmodel Finally, as the assessment of construct reliability, indicator reliability, convergent validity, and discriminantvalidity of the constructsaresatisfactory,itispossibletoanalyzethestructural model.ThemodelstestedwereUTAUTandperceivedrisk(PCR) (UTAUT+PCR–themainmodel)withinteractioneffects(D+I)and withoutthem(D)tounderstandifageandgenderhadinfluence

ontheintentionand usage.Then,we alsotested UTAUT (with-outperceivedrisk(PCR))andalsowithdirecteffectsonly(D)and addinginteractioneffects(D+I).Table5showspathcoefficients andr-squares foreach modeltested Chin(1998)stated that r-squares of the structural model should beabove 0.2, which is demonstratedbothinintentionandusageandinallmodels esti-mated,asseenin Table5.Comparisonoftheestimatedmodels revealsthatonintention,moderatingeffectsalwayshaveanimpact

onr-square,increasingit(0.52vs.0.56inUTAUTand0.56vs.0.60

inUTAUT+PCR).Inasimilarway,whenweaddperceivedriskto

effects (D + I).

Behaviour intention

PE, performance expectancy; EE: effort expectancy; SI: social influence; FC: facilitating conditions; PCR: perceived risk; BI: behavioural intention; UB: usage behaviour.

* p < 0.05.

***

Fig 4.Structural model (UTAUT + PCR–D + I) with path coefficients and r-squares.

theUTAUTmodel,r-squarealsoincreases(0.52vs.0.56withdirect

effectsonlyand0.56vs.0.60withdirectandinteractioneffects).On

theotherhand,whenweobserveusage,neithermoderatingeffects

norperceivedriskhaveanyimpactonit,becausether-squareis

alwaysthesame(0.81)

Withthesefacts,itispossibletoconcludethatourmodelthat

addedperceivedrisk(PCR)totheUTAUTmodel,withtheir

moder-atingeffects,explainstheintentionbetterthanalltheothers.We

nowfocusouranalysisonthemainmodel,thatis,UTAUT+PCRwith

moderatingeffects.Pathcoefficientsandr-squaresofthismodelare

inFig.4

Wealsocalculatedt-statisticsderivedfrombootstrapping(250

iterations).Mostdirecteffectsarestatisticallysignificant,suchas

performance expectancy( ˆˇ =0.32;p<0.001),effortexpectancy

( ˆˇ =0.33; p<0.001), social influence ( ˆˇ =0.09; p<0.05), and

perceived risk ( ˆˇ =−0.20;p<0.001) over intention To explain

usage,facilitating conditions is not statisticallysignificant ( ˆˇ=

0.03;p>0.05),andbehaviourintentionisstatisticallysignificant

( ˆˇ=0.89;p<0.001).Insummary,allofthedirecteffectsare

sta-tisticallysignificantfor intention,and forusageonlyfacilitating

conditionsisnotstatisticallysignificant

Noneoftheinteractioneffects arestatisticallysignificant,as

seeninTable5.Onlythedirecteffectofageonintentionis

sta-tisticallysignificant( ˆˇ =0.11;p<0.05)

6 Discussion

6.1 Theoreticalimplications

Theoretically,ourresultssuggestthatperceivedriskincreases

the predictive power of the UTAUT model in explaining

intention.Whileperformanceexpectancy(PE),effortexpectancy (EE),andsocialinfluence(SI)explainnearly56%ofthevariance

of behaviourintention(BI), bycouplingperceivedrisk (PCR) to UTAUT,thesevariablescontributedtoanincreaseof4p.p.of vari-ance explained, thereby providing a better explanatory power Furthermore,theproposedjointUTAUT+PCRmodelexplained81%

ofusagebehaviourvariance.Comparedwithotherinvestigations exploringInternetbankingadoption,ourstudypresentsastronger predictivepowerthansimilarstudies.For instance,Pikkarainen

et al (2004) usedTAM and explained 12.4% of intention, with perceivedusefulnessandinformationonthewebsiteasthemain determinants;LeeandChung(2011)alsoappliedTAMandadded self-efficacyasoneoftheantecedentvariablessuchasrisk, Inter-netexperience,andfacilitatingconditionsinSouthKorea’susers, withintentionbeingexplainedby32.3%throughInternet experi-ence,perceivedusefulness,andperceivedeaseofuse;andusage presentedanr-squareof4.8%,whichisconsiderablylowerthan theoneinthisstudy

Table6presentstheoutcomesofhypothesestested.Theresults

ofthemodelshowedthat,contrarytoourexpectations,theeffect

offacilitatingcondition(FC)constructfromUTAUToverusage(UB) wasnot significant.Thissuggeststhat ourrespondents arenot concernedaboutthesurroundingenvironment(necessary infras-tructures, knowledge,capabilities, etc.)toinfluence theirusage

ofInternetbanking.Asobservedinsomeotherresearch(e.g Al-Somali,Gholami,&Clegg,2009;Lee&Chung,2011;Riffai,Grant,

&Edgar,2012),theeffectsofPEandEEoverBIweresubstantial, meaningthatindividualscareabouttheoutcomesofusing Inter-netbankingandthenecessaryefforttoexpendinordertouseit Withalowmagnitude,SIalsoshowedaneffectonBI,meaningthat ourrespondentsareconcernedaboutenvironmentalfactorssuch

astheopinionsofuser’sfriends,affectingtheirintentiontoadopt

6

Internetbanking.TheimpactofBIonusagebehaviour(UB)wasalso substantial,whichindicatesthatInternetbankingusersaremore likelytousethesystemiftheyhadtheintentiontouseit Regardingtheperceivedriskpartofthemodel,ithas demon-strated evidence for a second-order composite perceived risk variable.Performance,financial,time,andprivacyrisksprovedto

bethemostsalientconcernsforperceivedrisk,thatis,theones relatedwithperformance.Socialandpsychologicalriskswereless salient.ThenegativeeffectsofPCRoverBIandPEwerealso demon-strated

Concerning theinteraction effects, we foundnosupport for eitherofthosetested,similarlytoRiffaietal.(2012)findings.We concludethatageexplainsbehaviourintentionofInternetbanking service( ˆˇ=0.11;p<0.05;inthemainmodel).Thismeansthatif respondentsareolder,theywillhavemoreintentiontouseInternet banking

6.2 Managerialimplications Thefindingsofthisstudyrevealthatperceivedriskisan impor-tantfactoraffectingend-user intentiontouseInternetbanking Therefore,managersneedfirstofalltoensurethatanInternet bank-ingplatformistechnicallysound,withgoodsecuritypracticesput

inplacetominimizetherisksfortheendusers.Thefocus,as previ-ouslynoted,shouldbeonperformancerisks,namelytime,financial, performance,andprivacy.Managersshouldadvertisetopotential usersthatInternet bankingisnot ariskyservice,bypromoting informationofsecurityandtrustontheplatform.Theyshouldalso preventuserconcernsaboutcomputercrimes,invasionofprivacy, andoverall, attempttoprovidetransactions withouterrors and allocatesufficientresourcestocorrectit,ifnecessary.Theuseof

asecurechannelfromtheconsumer’spersonalcomputertothe bankserverandhandlingofsessionswithkeyencryptionaretwo importantissuesthatinstitutionsshouldensurethatusersknow Additionaleffectiverisk-reducingstrategies mayinclude money backguaranteesandprominentlydisplayedconsumersatisfaction guarantees,sothatconsumersfeelmorecomfortableandsafewith thesystem

InrealizingthatInternetbankingplatforms’performanceand Internetbankingplatforms’easeofusearetwootherfactorsthat affectintention,institutionsneedtopromoteclarification work-shops,toteachpeopletousetheplatformandexplainthemain benefitsofInternetbanking(Bussakorn&Dieter,2005)

Lastly,bothInternetbankingmanagersanduserscantake finan-cialadvantagefromtheadoption.Withtheself-serviceconsumer software-basedserviceviaInternet,bankscandecreasecostswith branches,byencouragingandsupportingtheuseoftheplatforms Userscanalsodecreasetheircosts,bynotpayingfortransactions, benefitingfromonlineexclusiveproductswithhigherprofits,etc Additionally,Internetbankingcanprovideconsumerswithutility gainsmeasuredinconvenienceandefficiency

6.3 Limitationsandfutureresearch Whileourstudyaddstotheexistingbodyofknowledge,wealso acknowledgeitslimitations,mainlyconcerningthesampling.The respondentsweremostlyyoung,highlyeducatedpeoplewhose behaviourmight differsomewhatfrom thepopulationaverage Theyare generallymore innovativeand quicker toacceptnew technologies,andthismayhavebiasedtheresults.Itislikelythat elderlyandlesseducatedconsumersorthosewhopossessreduced computing/Internetskillswouldperceivegreaterdifficultyinuse

ofInternetbankingandhigherinherentusagerisks

Futureresearchcanbebuiltbasedonthisstudybytestingthis modelindifferentagegroups.Furthermore,itcouldbeinteresting

toapplythemodeltoothercountriesandalsoothercontexts.Next,