Online-banking-performance-evaluation-using-data-e

Computers&OperationsResearch36(2009)1835--1842ContentslistsavailableatScienceDirectComputers&OperationsResearchjournalhomepage:∗aInstituteofElectronicCommerce,NationalChungHsingUniversity,TaiwanbRiskLab,UniversityofToronto,19BordenST,Toronto,ON,CanadaARTICLEINFOABSTRACTAvailableonline22May2008Keywords:InternetbankingElectroniccommerceDataenvelopmentanalysis(DEA)Principlecomponentanalysis(PCA)WebmetricThispaperpresentsahybridapproachtoconductingperformancemeasurementsforInternetbankingbyusingdataenvelopmentanalysis(DEA)andprincipalcomponentsanalysis(PCA).Foreachbank,DEAisappliedtocomputeanaggregatedefficiencyscorebasedonoutputs,suchaswebmetricsandrevenue;andinputs,suchasequipment,operationcostandemployees.The45combinationsofDEAefficienciesofthestudiedbanksarecalculated,andusedasarankingmechanism.PCAisusedtoapplyrelativeeffi-cienciesamongthebanks,andtoclassifythemintodifferentgroupsintermsofoperationalorientations,i.e.,Internetbankingandcostefficiencyfocusedorientations.Identificationofoperationalfitnessandbusinessorientationofeachfirm,inthisway,willyieldinsightsintounderstandingtheweaknessesandstrengthsofbanks,whichareconsideringmovingintoInternetbanking.©2008ElsevierLtd.Allrightsreserved.1.IntroductionSincetheintroductionofInternetbanking,andduetoitsgrowingconvenienceformostindividuals,theInternetmarkethasgrownintoaprofitablecompetitivevenueforthebankingindustry.ImprovingtheefficiencyofInternetbankingisnowconsideredtobeimportanttothebankingindustryandthisresearchaimstohelpclarifytheissuebyproposinganewmethod.In2000,itwasreportedbytheEconomistthatonlinebankingcostsfirmsapproximately$0.01pertransactionandaccountedforonly1%ofallbankingtransactions.Thisisasituationwhichischang-ingrapidly.InareportbyJupiterResearchin2004,anestimated30millionfamiliesintheUSusedInternetbanking.ItwasforecastedbyJupiterthatin2008,thosenumberswillreach56million[1].Inter-netbankingcanreducebusinessoperationcostsandhasdecreasedthedistributionofbranchlocations.However,lackofresearchhasstymieditsgrowth.Lookingatthedataalreadybeinggatheredonwebefficiencywithinthewideranalysisofe-commerceisthefirststepinad-dressingthisproblem.Electroniccommerce(EC)istheprocessofelectronicallyconductingallformsofbusinessbetweenentitiesinordertoachievetheorganization'sobjectives[2].Forthedotcommarket,non-financialinformationnormallytakestheformofweb∗Correspondingauthor.E-mailaddresses:[e-mail](C.-T.BruceHo),[e-mail](D.DashWu).0305-0548/$-seefrontmatter©2008ElsevierLtd.Allrightsreserved.doi:10.1016/j.cor.2008.05.008metrics,whichcanbeobtainedfromInternettrafficmeasurementreports,orwebsitessuchasAlexa.com.TheimpactofInternetfirmscanbeassessedbyexamining“uniquevisitors”whoaccesstheweb-sitesorbythenumberof“pagehits”[3].DemersandLevstatedthatwebtrafficmeasurementshavebecomestandardperformancebenchmarksthatarenowcommonlyreportedinthebusinesspress[4].Theycitedthreedimensionsofwebtrafficperformance,“reach”,“stickiness”,and“customerloyalty”,respectively,asperformancemeasurementsforevaluatingafirms'Internetprofile.“Reach”isthoughttobeimportantwhenthefirmobtainsrevenuesfromadver-tising.“Stickiness”isconsideredrelevantfordotcomfirmswhoserevenuedependsonthetimethattheuserspendsinthewebsite;thisisoftenanunusualwayofgeneratingincome.“Customerloyalty”considersrepeatedsalestoaparticularcustomer.Onlineperformanceassessmenthasbeenanemergingresearchareaforbothresearchersandpractitioners.Whilescopeoftheex-istingliteratureisgenerallythoughttobefinancialperformance,thefindingsofarecentstudyfromSerrano-Cincaaetal.suggestthatwithoutthehelpofsupportingnon-financialinformationfinancialinformationmightnotalwaysbesufficienttojudgeanonlinebusi-ness[3].Itispossibleforvariousperformanceevaluationmodelstobeconstructedbycombiningallthepossibleinputsandoutputsthatabankmightuse.Traditionallyusedratioanalysiscanprovidearelativelyinsignificantamountofinformationwhenconsideringtheeffectsofscaleeconomies,theidentificationofbenchmarkingpoli-cies,andtheestimationofoverallperformancemeasuresoffirms.Asaresult,theremaybeincentivetoutilizemoreeffectivemethodsinevaluatingtheoverallperformanceofbanks[5].1836C.-T.BruceHo,D.DashWu/Computers&OperationsResearch36(2009)1835--1842Studiesonbankperformanceevaluationinliteraturemainlyem-ployfivedifferentapproaches,i.e.,DEA(e.g.,[6,7]),freedisposalhull(e.g.,[8]),stochasticfrontierapproach(e.g.,[9]),alsocalledecono-metricfrontierapproach(e.g.,BergerandHumphrey[9]),thickfron-tierapproach(e.g.,[10]),anddistributionfreeapproach(e.g.,[11]).Theseapproachesprimarilydifferinhowmuchrestrictionisim-posedonthespecificationofthebestpracticefrontierandtheas-sumptionofrandomerrorsandinefficiency[12].Comparedtootherapproaches,DEAisconsideredabetterwaytoorganizeandanalyzedatasinceitallowsefficiencytochangeovertimeandrequiresnopriorassumptiononthespecificationofthebestpracticefrontier.Thus,DEAhasbecomealeadingapproach