42 (2006) 1558–1572www.elsevier.com/locate/dss

Decision Support Systems

Adoption of internet banking: An empirical study in Hong Kong

T.C. Edwin Cheng ⁎, David Y.C. Lam, Andy C.L. Yeung

Department of Logistics, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong

Received 4 November 2004; received in revised form 31 December 2005; accepted 21 January 2006Available online 9 March 2006

Abstract

This study investigates how customers perceive and adopt internet banking (IB) in Hong Kong. We developed a theoreticalmodel based on the Technology Acceptance Model (TAM) with an added construct Perceived Web Security, and empirically testedits ability in predicting customers' behavioral intention of adopting IB. We designed a questionnaire and used it to survey arandomly selected sample of customers of IB from the Yellow Pages, and obtained 203 usable responses. We analyzed the datausing Structured Equation Modeling (SEM) to evaluate the strength of the hypothesized relationships, if any, among the constructs,which include Perceived Ease of Use and Perceived Web Security as independent variables, Perceived Usefulness and Attitude asintervening variables, and Intention to Use as the dependent variable. The results provide support of the extended TAM model andconfirm its robustness in predicting customers' intention of adoption of IB. This study contributes to the literature by formulatingand validating TAM to predict IB adoption, and its findings provide useful information for bank management in formulating IBmarketing strategies.© 2006 Elsevier B.V. All rights reserved.

Keywords: Technology Acceptance Model (TAM); Internet banking; Structural equation modeling

1. Introduction

The bursting of the Internet bubble in early 2001 hasgenerated numerous speculations that the opportunitiesfor Internet services firms have vanished. The dot.comcompanies and Internet players have been struggling forsurvival, and most of the related businesses are stillsuffering losses. Practicing managers and academicshave not yet reached a consensus in their debate aboutthis new technology: whether the Internet brings about a

⁎ Corresponding author.E-mail address: lgtcheng@polyu.edu.hk (T.C.E. Cheng).

0167-9236/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.dss.2006.01.002

revolutionary change in the fundamental way we dobusiness or whether it is only an evolutionary process,offering simply a new distribution channel and com-munication medium [29]. According to Brown [9], the“New Economy” or e-commerce businesses are still atthe infancy stage. Despite the crash of dot.com stockprices in March 2001, Internet usage and e-commercehave continued to grow at a fast pace.

According to eMarketer [16], the US B2C e-commerce revenues reached US$70 billion in 2002,compared to US$51 billion in 2001, i.e., a jump of 37%.It also predicted that by 2003, the e-commerce revenueswould increase by 28% to US$90 billion; another 21%increase to US$109 billion by 2004; and to US$133billion, a further 22% increase, by 2005. Compared with

1559T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

the global economic growth of less than 5%, and theforecast GDP growth of 1–2% for Hong Kong in 2003,the predicted e-commerce growth of 28% is veryencouraging. The anticipated explosive growth of onlinepurchases via the Internet will present immenseopportunities to businesses in general, and internetbanking (IB) in particular.

In this study, we use the terms IB and OnlineBanking (OB) interchangeably. IB/OB is differentfrom Electronic Banking (e-banking) in that the latteris a higher level activity that encompasses not only IB/OB, but also Telephone Banking, ATM, WAP-bankingand other electronic payment systems that are notoperated through the Internet. We focus on IB becauseit is widely seen as the most important and mostpopular delivery channel for banking services in thecyber age.

Banks can benefit from much lower operating costsby offering IB services, which require less staff andfewer physical branches. Customers will also benefitfrom the convenience, speed and round-the-clockavailability of IB services. However, IB has not takenoff in Hong Kong as spectacularly as expected.According to ACNielsen and NetRatings [1], 522,700people in Hong Kong visited an IB site from their homePC in January 2003 (out of 2,194,600 active Internetusers), representing a penetration rate of only 23.8%. Torealize the full potential of IB, banks need to developnew products and services to fully utilize the Internet'scapabilities. On the other hand, customers need to bemade aware of IB services, and feel secure andcomfortable with using such services as the new IBoperating procedures are radically different from thosethey are used to.

There is a clear need to study the factors thatinfluence customers' intention to adopt IB so thatbanks can better formulate their marketing strategies toincrease IB usage in the future. This study aims toinvestigate the behavioral intention of customers to useIB services with a focus on users' perceptions of easeof use and usefulness of IB, and of security of usingthis new technology to meet their banking needs. InSection 2, we present a review of the literature oninnovation diffusion and technology adoption, basedon which we propose a model of customers' intentionto adopt IB, and formulate the associated researchhypotheses. We discuss the research methodology inSection 3, and present the findings from the analysisof the empirical data in Section 4. Section 5 concludesthe paper with discussions of the limitations of thestudy, managerial implications and further researchdirections.

2. Literature review and model formulation

Although an abundance of studies aimed atextending our understanding of user adoption oftechnology have been conducted in the past, few ofthese studies were conducted on IB services byextending the well-established Technology AcceptanceModel (TAM). With the number of global bankinggroups offering and improving IB services rapidly onthe rise [4], it is an opportune time to study the useradoption of IB. Such a study will be of interest to bothacademics and banking executives. Specifically, thisstudy investigates individuals' perception of theadoption of internet banking for corporate purposes.In other words, our survey focused on individuals'intention to use internet banking to handle their work-related banking issues.

2.1. Technology Acceptance Models

Davis [14] developed the Technology AcceptanceModel, according to which “users' adoption ofcomputer system” depends on their “behavioralintention to use”, which in turn depends on “attitude”,consisting of two beliefs, namely Perceived Ease ofUse and Perceived Usefulness. In fact, Davis [14]developed TAM by building upon an earlier theory,the Theory of Reasoned Action (TRA) by Fishbeinand Ajzen [17]. In TRA, Fishbein and Ajzen [17]proposed that intention is “the immediate determinantof the corresponding behavior”, which is divided into(1) “attitude toward behavior”, and (2) “subjectivenorm concerning behavior”. Davis [14] posited inTAM that the two theoretical constructs, PerceivedUsefulness and Perceived Ease of Use, are fundamen-tal determinants of system use in an organization.These constructs also provide better measures forpredicting and explaining system use than otherconstructs [14].

TAM has become a widely used model forpredicting the acceptance and use of informationsystems, and has recently been applied to predictInternet adoption as well. In a recent study, Lederer etal. [26] adapted TAM to study World Wide Web(WWW) usage and found evidence to support TAM.Another study of applying TAM in the WWW contextwas conducted by Moon and Kim [30]. Theyintroduced the construct Playfulness to predict Atti-tude. Data were collected from 152 graduate studentsof management in Korea. Their findings showed thatalthough TAM-related hypotheses were all supported,the results deviated from the basic belief of TAM that

1560 T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

Perceived Usefulness is the key determinant of useracceptance of IT. The results of Moon and Kim [30]revealed that Perceived Ease of Use has a moresignificant effect on Attitude than Perceived Useful-ness in the WWW context, and Perceived Playfulness(an intrinsic motivational factor) has a more significanteffect on Attitude than Perceived Usefulness (anextrinsic motivational factor).

2.2. Theoretical model for IB adoption

IB is a new distribution channel for the deliveryof banking services. From both academic andpractical perspectives, it is interesting to understandand assess customers' intention to use IB services.We have chosen TAM as the baseline model for thisstudy because it is a well-tested model concerningusers' acceptance of technology. We augment TAMwith the construct Perceived Web Security. Specifi-cally, we hypothesize that Intention to Use isinfluenced by Attitude, Perceived Usefulness, Per-ceived Ease of Use and Perceived Web Security. Wewill test the strength of the hypothesized relation-ships embedded in the theoretical model and therobustness of the model in predicting customers'intention to adopt IB in the Hong Kong businessenvironment. The theoretical model is graphicallypresented in Fig. 1.

TAM has been used by various researchers topredict users' intention to accept or adopt a variety oftechnologies and computer systems. The technologiesinclude electronic mail, text editor, word processing

H3 Attit

Perceived Usefulness

Perceived Ease of Use

Perceived Web Security

H5b

H4a H2b

H4b

Fig. 1. Research model of customer's i

systems, and graphics software [15,14], spreadsheets[20], Database Management Systems [40], voice-mailand word processors [2,12]. We use TAM with theconstructs Perceived Usefulness and Perceived Easeof Use to assess the determinants of customers'Attitude and Intention to Use (equivalent to theconstruct Behavioral Intention in TAM). While weadopt the original TAM in this study, we useBehavioral Intention as the dependent variable andskip the construct Actual Usage. On the theoreticalfront, an abundance of research studies have reporteda strong and significant causal relationship betweenbehavioral intention and usage of technology ortargeted behavior [37,41]. It is therefore theoreticallyjustifiable to use Behavioral Intention as a dependentvariable to examine the acceptance of IB [28].Agarwal and Prasad [3] also argued that for asurvey-based research design, Behavioral Intention ismore appropriate than Actual Usage as “they aremeasured contemporaneously with beliefs” and ourstudy is survey-based research. On the practical front,it is worth noting that IB is still at an early stage ofdevelopment in Hong Kong. The percentage of usageis relatively low (less than 15% of the total number ofbank customers use IB services). Therefore, thechoice of Behavioral Intention, rather than ActualUsage, as the dependent variable is considered bothappropriate and necessary.

In addition, feeling secure in doing transactionson the Web is often cited by users as a major factorthat removes their concerns about the effective useof the Internet for making online purchases [34].

Intention to Use H1

H5a

ude

Web Security related hypothesis

H2a

TAM related hypothesis

ntention to use internet banking.

1561T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

Therefore, we include the construct Perceived WebSecurity as a predictor of Attitude and Intention toUse, as in the earlier study conducted by Salisbury etal. [34]. In their study, Salisbury et al. [34]developed a set of four items to measure PerceivedWeb Security using a 7-point Likert scale. The re-sults of their study showed that the three constructs,namely Perceived Web Security, Perceived Ease ofUse and Perceived Usefulness, have a positiverelationship with intention to purchase online.However, Salisbury et al. [34] did not include theconstruct Attitude, which is included in our theoret-ical model.

2.3. Hypothesis development

Based on the theoretical model developed above,we formulate the research hypotheses as follows. AsTAM is used as the base model, we need to test thefollowing TAM hypotheses in the context of IBadoption.

H1. Customers' Attitude positively influences theirIntention.

H2a. Perceived Usefulness has a direct positive rela-tionship with customers' Intention.

H2b. Perceived Usefulness has a direct positiverelationship with customers' Attitude.

H3. Perceived Ease of Use has a direct positiverelationship with customers' Attitude.

H4a. Perceived Ease of Use has an indirect positiverelationship with customers' Intention via PerceivedUsefulness.

H4b. Perceived Ease of Use has an indirect positiverelationship with customers' Attitude via PerceivedUsefulness.

In addition, we seek to test the hypothesizedrelationships related to PerceivedWeb Security. Accord-ing to Salisbury et al. [34], Perceived Web Security wasfound to favorably influence customers' intention topurchase on the WWW. Customers tend to increasepurchases only if they perceive that their credit card andother sensitive information are safe. Therefore, Per-ceived Web Security is expected to have a positiverelationship with customers' Intention to use IB, as wellas with their Attitude.

H5a. Perceived Web Security has a direct positiverelationship with customer's Intention.

H5b. Perceived Web Security has a direct positiverelationship with customer's Attitude.

3. Research methodology

3.1. Research strategy

There exists virtually no research examiningcustomers' behavioral intention to adopt IB servicesby extending TAM. To fill this gap, we conducted apostal survey study for hypothesis testing using theframework of the original TAM as the foundation todetermine the predictors of customers' Intention touse IB in Hong Kong. To collect data, we designeda questionnaire by adapting the instrument andscales developed for TAM. We augmented TAMby adding the construct Perceived Web Securitydeveloped by Salisbury et al. [34] and adapting theirinstrument and scale to measure this construct in ourquestionnaire.

3.2. Sample size

This study aims to investigate the self-reportedbehaviors of individual customers and their intentionto use IB services for corporate purposes in HongKong. We requested respondents to return theircompleted questionnaires either by fax or by Freepost.We collected data from bank customers in HongKong who use internet banking for their work andtargeted to obtain no less than 193 usable responses(a 19% effective response rate) from a randomlyselected sample of 1000 customers, in order toachieve 0.80 statistical power, a 0.05 statisticalsignificance criterion, and an r-value of 0.20 inaccordance with the suggestions of Cohen [13]. Wemailed the survey questionnaires to a sample of 1000names randomly selected from a total of 157,000names listed in the Yellow Pages, which is thetelephone directory of virtually all businesses in HongKong. We used Excel to generate 1000 randomnumbers, which were then used to select the namesfrom the Yellow Pages. Sathye [35] employed asimilar process to randomly select 1000 businessclients from the Yellow Pages in his study and foundthat “security concerns” and “benefits not clear” arethe main factors that explain why Australian clientsdo not adopt IB.

1562 T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

3.3. Measurement/operationalization of the constructs

In devising a useful measurement instrument for thisstudy, we adapted the instrument and scales developedand validated in the following two studies:

1. The questionnaire developed for TAM by Davis [14]—adapting the scales for Perceived Usefulness andPerceived Ease of Use.

2. The questionnaire developed by Salisbury et al. [34]—adapting the scale for Perceived Web Security.

Both studies have established the validity andreliability of their instrument, particularly the TAMinstrument, which has been replicated and widely usedin other studies. In this study, we use the adapted questionitems as the instrument to measure the respectiveconstructs below, using a 7-point Likert scale for eachitem (with 1=strongly disagree, 2=disagree, 3=slightlydisagree, 4=neutral, 5=slightly agree, 6=agree, and7=strongly agree). The constructs are defined as follows.

Perceived Ease of Use (PEOU)—defined by Davis[14] as “the degree to which a person believes that usinga particular system would be free of effort”. Fourquestion items are used to measure PEOU.

Perceived Usefulness (PU)—defined by Davis [14]as “the degree to which a person believes that using aparticular system would enhance his job performance”.Four question items are used to measure PU.

Perceived Web Security (PWS)—defined by Salis-bury et al. [34] as “the extent to which one believes thatthe World Wide Web is secure for transmitting sensitiveinformation.” It is also noted that “the adoption ofpurchasing products on the World Wide Web mayinvolve a greater degree of risk than the adoption ofother IT innovations. When one purchases productsonline, there may be a perception of risk involved intransmitting sensitive information such as credit cardnumbers across the World Wide Web.” Four questionitems are used to measure PWS.

Attitude (ATT)—refers to an individual's positive ornegative feelings (evaluative affect) about performing aparticular behavior [17]. Three question items are usedto measure ATT.

Intention to Use (INT)—refers to customers' inten-tion to use, as opposed to their actual use of, IB services.Three question items are used to measure INT.

3.4. Survey questionnaire and pilot test

Based on the hypothesized model (Fig. 1) developedthrough a detailed review of the related literature on user

acceptance of technology and new technology diffusion,we devised a 33-item questionnaire (19 items relate touser acceptance behavior and 14 are demographic/general questions) as a measurement scale for theresearch. The questionnaire was originally developed inEnglish, and subsequently translated into Chinese by auniversity graduate with special training in English–Chinese translation. A back translation was carried outby another trained translator to ensure the accuracy ofthe translation.

The bilingual questionnaire was then used in a pilottest involving 120 trainees from the Vocational TrainingCenter, who completed the questionnaire and providedvaluable comments. Based on the respondents' feed-back, we made modifications to the questionnaire toimprove its readability and ensure its accuracy andappropriateness. A factor analysis was performed on thedata collected from the pilot study. The test results weresatisfactory, with five factors corresponding to the fiveintended constructs emerging with factor loadingsranging from 0.6 to 0.9, and all Cronbach's alphavalues surpassing the commonly adopted thresholdvalue of 0.70.

4. Results and discussions

A total of 212 responses were received, of which 203were accepted as valid responses for further analysisafter removing a few erroneous (e.g., the respondentgave more than one answer to a question that expectedonly one answer) and missing items. The response ratewas 20%, which exceeded the minimum desirablepercentage (19%) for statistical power for the purposeof this research. To assess the possible existence of non-response bias, we divided the respondents into twogroups as follows. The first group of respondents(N1=143) consisted of those whose returns werereceived within 2 days of sending out the questionnaires.The second group (N2=60) included those who sent intheir returns more than 2 days after dispatching thequestionnaires and after receiving a follow-up phone callto prompt them to give their reply. The second group wastaken to represent the non-respondents. A t-test wasperformed on the responses of the two groups to see ifthey were different. The results of the t-test showed nodifferences in the responses of the two groups, suggest-ing that there was no evidence of non-response bias.

To ensure research rigor and validity of the results,we followed the procedures proposed by Koufteros [24]in applying Structural Equation Modeling (SEM) toanalyze the data. First, we developed an instrument forthe measurement scale by following a systematic

Table 1Demographics of respondents

Demographicprofile

Frequency Percent(%)

SexF Female 120 59.1M Male 83 40.9

Total 203 100.0

AgeA Under 18 1 0.5B 19–30 76 37.4C 31–45 74 36.5D 46–60 49 24.1N Over 60 3 1.5

Respondent positionA CEO 7 3.4B CFO 8 3.9C CIO 3 1.5D Owner 47 23.2E Manager 70 34.5F Clerk 58 28.6O Other 10 4.9

Respondent industryB Banking 4 2.0F Financial Institutional 9 4.4I IT related 44 21.7M Manufacturing 48 23.6O Other 29 14.3R Retail 60 29.6T Telecommunication 6 3.0U Tourism 3 1.5

Company size (employees)A Less than 50 120 59.1B 51–100 41 20.2C 101–300 24 11.8D 301–500 10 4.9E 501–1000 3 1.5F Over 1000 1 0.5N Not applicable 4 2.0

1563T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

approach and incorporating a pre-test and a pilot test toensure the appropriateness of the instrument. Second,we adopted an effective approach for data collection.Third, we performed an evaluation at the item levelusing the tests for convergent validity and itemreliability. Fourth, as a satisfactory model was derived,we carried out the analysis with an assessment of themodel fit and unidimensionality. Fifth, we evaluated thediagnostics and tests for discriminant validity, compos-ite reliability and variance extracted to gain confidencein the measurement scales. Finally, we tested thestructural model by means of Confirmatory FactorAnalysis (CFA).

4.1. Demographic characteristics of the respondents

The descriptive statistics of the respondents' demo-graphic characteristics were analyzed and presented inTable 1. Of the 203 respondents, 59% were female; 37%were in the 19–30 age group, 36% were 31–45 in age,and 24% were 46–60 in age. Few were under the age of18, or over 60. Other demographic details can be foundin Table 1.

4.2. Exploratory factor analysis

An exploratory factor analysis using SPSS wasconducted on the survey data. The rotated factor matrix,resulting from a Varimax rotated principal axis factorextraction of the independent variables using the 1.0eigenvalue cut-off criterion, is shown in Table 2, whichindicates that five factors emerged and reports theirfactor loadings.

The data were tested using the SPSS ExploratoryFactor Analysis (EFA) to evaluate the Cronbach's alpha,which ranged from 0.902 to 0.939. Each item wasevaluated individually to ensure convergent validity anditem reliability. Except for b7 and b8 (0.403 and 0.406,respectively), all factor loadings were larger than 0.5,representing an acceptable significant level of internalvalidity. The factor loadings ranged from 0.689 to 0.722for Perceived Ease of Use, 0.403 to 0.619 for PerceivedUsefulness, 0.751 to 0.901 for Perceived Web Security,0.618 to 0.710 for Attitude, and 0.501 to 0.608 forIntention. Since all factor loadings were of an acceptablesignificant level, all 19 questionnaire items wereretained for further analysis.

4.3. Confirmatory Factor Analysis

We used the hypothesized model to test model fitnessby performing a Confirmatory Factor Analysis (CFA)

on the data. The results show that the hypothesizedmodel is recursive, i.e., uni-directional (Table 3). Therewere 190 distinct sample moments (i.e., pieces ofinformation) from which to compute the estimates of thedefault model, and 45 distinct parameters to beestimated, leaving 145 degrees of freedom. Theminimum iteration was achieved, thereby providing anassurance that the estimation process yielded anadmissible solution, eliminating any concern aboutmulticollinearity effects. The results also provide aquick overview of the model fit, which includes the χ2

value (395.03), together with its degrees of freedom(145) and probability value (<0.0005).

Table 2Factor loadings (from SPSS exploratory factor analysis)

Factorloading

Cronbach alpha Varianceexplained (%)

Total cumulative % 78.194

Perceived Ease of Use (PEOU) 0.722 0.934 19.484B/b1—Using the Internet Banking (IB) service is easy for me 0.718B/b2—I find my interaction with the IB services clear and understandable 0.689B/b3—It is easy for me to become skillful in the use of the IB servicesB/b4—Overall, I find the use of the IB services easy 0.720

Perceived Usefulness (PU) 0.929 8.219B/b5—Using the IB would enable me to accomplish my tasks more quickly 0.619B/b6—Using the IB would make it easier for me to carry out my tasks 0.595B/b7—I would find the IB useful 0.403B/b8—Overall, I would find using the IB to be advantageous 0.406

Perceived Web Security (PWS) 0.939 21.568B/b9—I would feel secure sending sensitive information across the IB 0.751B/b10—The IB is a secure means through which to send sensitive information 0.825B/b11—I would feel totally safe providing sensitive information about myself over the IB 0.804B/b12—Overall, the IB is a safe place to transmit sensitive information 0.901

Attitude (ATT) 0.902 20.456B/b13—Using the IB is a good idea 0.654B/b14—I would feel that using the IB is pleasant 0.710B/b15—In my opinion, it would be desirable to use the IB 0.697B/b16—In my view, using the IB is a wise idea 0.618

Intention (INT) 0.923 8.466B/b17—I would use the IB for my banking needs 0.566B/b18—Using the IB for handling my banking transactions is something I would do 0.608B/b19—I would see myself using the IB for handling my banking transactions 0.501

Used SPSS Principal Axis Factoring extraction with Varimax rotation method.

1564 T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

4.4. Model assessment

4.4.1. Conformance to the expected parameterestimates

To begin with, it is necessary to assess the fit ofindividual parameters in the hypothesized model todetermine the viability of their estimated values.Parameter estimates are expected to exhibit the correctsign and size, and to be consistent with the underlyingtheory. Any unexpected estimate indicates either the

Table 3Fit statistics (original model)

Fit statistic Suggested Obtained

Chi-square 395.03df 145Chi-square significance p< or =0.05 <0.0005Chi-square/df [42] <5.0 2.72GFI [22] >0.90 0.84NFI [8] >0.90 0.91CFI [7] >0.90 0.94RMSEA [39] <0.05 0.09PCLOSE >0.50 0.00

model is inappropriate or the input data are missing. Theresults indicated that the estimates were quite normaland acceptable.

4.4.2. Reliability of parameter estimatesAnother test statistic is the Critical Ratio (C.R.),

which represents the parameter estimate divided by itsstandard error (S.E.). As a “rule of thumb”, the C.R.needs to be >±1.96 if the estimate is acceptable[11,19]. The results showed that all C.R. values were

Table 4Fit statistics of final re-specified model

Fit statistic Suggested Obtained

Chi-square 202.31df 141Chi-square significance p< or =0.05 0.001Chi-square/df [42] <5.0 1.44GFI [22] >0.90 0.91NFI [8] >0.90 0.95CFI [7] >0.90 0.99RMSEA [39] <0.05 0.05PCLOSE >0.50 0.65

1565T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

larger than 1.96, indicating that they were allacceptable.

4.4.3. Review of fit statisticsReviewing the fit statistics in Table 3, we noted that

not all the fit statistics indicated a well-fit. The key fitstatistics showed a value of χ2 (145 df)=395.03, GFI of0.84, NFI of 0.91, CFI of 0.94, and RMSEA of 0.09(PCLOSE of 0.000). These statistics indicated amarginal fit, instead of a very good fit, according tocommonly suggested criteria [11]. Therefore, it wasnecessary to apply the procedures to test whether anymisspecifications and/or any violation of the assump-tions existed (Table 4).

4.5. Post-hoc analyses

To take all reasonable precautions with respect to thestatistical analysis of our assertions, we performedadditional post-hoc analyses to detect multicollinearityand performed bootstrapping.

4.5.1. MulticollinearityTo determine whether any multicollinearity effects

existed, we reviewed the correlations in-depth to look

.84

PU

.77

b6

e6

.88

.71

b7

e7

.84

.69

b8

e8

.83

PEOU

.75

b4

e4

.77

b3

e3

.80

b2

e2

.78

b1

e1

.87.88.89

.88PWS

.91

b10

e10

.79

b9

e9

.95.89

.82

AT

.52

b13

e13

.81

b14

e14

.72 .90.74

.92

res1

.11

.73

b5

e5

.85

.57

.

.31

.64

Fig. 2. AMOS graphic of th

for the presence of any correlation >1.00 and to checkwhether there was any warning message produced bythe AMOS Output that signaled a problem of multi-collinearity. The results showed that there was noevidence of multicollinearity.

The potential problem of multicollinearity can befurther examined formally in the context of regressionanalysis. The variance inflation factor (VIF), whichindicates the degree to which each predictor variable isexplained by other predictor variables, is a commonmeasure of multicollinearity in regression analysis[19,36]. High multicollinearity masks the effects of anindividual predictor, and results in incorrect estimationsof regression weights [19]. A threshold VIF that is lessthan or equal to 10 (i.e., Tolerance >0.1) is commonlysuggested [5,19]. The VIFs for PU, PWS, PEOU andATT were 4.23, 1.65, 3.38, and 3.39, respectively, inpredicting INT, providing further evidence againstmulticollinearity.

4.5.2. Incorporating the construct actual useWe conducted a further analysis of an “Expanded

Model” that incorporates the construct Actual Usagefollowing the construct Intention to Use, and com-pared the results with the hypothesized model. The

.76

b12

e12

.63

b11

e11

.87.79

T

.79

b15

e15

.68

b16

e16

.83 .83

INT.30

.35

res2

res3

.66

b19

e19

.81

.88

b18

e18

.94

.87

b17

e17

.93

.50

.89

09.37

e re-specified model.

1566 T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

results of the SPSS Exploratory Factor Analysisshowed that the Total Variance Explained for the“Expanded Model” was 76.898%. Comparing with thecorresponding value of the hypothesized model of78.194% (Table 2), we see that the hypothesizedmodel has a higher percentage of Total VarianceExplained. We conclude that the hypothesized modelis better in explaining the data and representing therelationships between the variables.

4.6. Results of hypothesis testing

After establishing an acceptable measurement model,we evaluated the structural model shown in a pathdiagram in Fig. 2. The regression weights are reported inTable 5. The results of the SEM analysis show that H1,H2a, H2b, H4a, H4b, H5a, and H5b were supported,while H3 and H5b were not supported.

4.6.1. Attitude and web securityTo test the statistical significance of the parameter

estimates from SEM, the test statistic is the CriticalValue (C.R.), which represents the parameter estimate

Table 5Parameter estimates (AMOS report) of re-specified model

Regression weights

Estimate S.E. C.R. P

PU←PEOU 0.878 0.062 14.216 0ATT←PU 0.574 0.138 4.144 0ATT←PEOU 0.082 0.127 0.645 0.519ATT←PWS 0.06 0.036 1.665 0.096INT←ATT 0.476 0.191 2.484 0.013INT←PU 0.422 0.145 2.903 0.004INT←PWS 0.378 0.051 7.395 0b6←PU 1.046 0.042 25.129 0b7←PU 0.988 0.066 15.06 0b8←PU 0.985 0.067 14.736 0b4←PEOU 1b3←PEOU 1.003 0.058 17.301 0b2←PEOU 0.956 0.054 17.85 0b1←PEOU 0.929 0.053 17.49 0b12←PWS 1b11←PWS 0.917 0.046 20.05 0b10←PWS 0.967 0.047 20.526 0b9←PWS 0.946 0.053 18.013 0b13←ATT 1b14←ATT 1.266 0.1 12.687 0b16←ATT 1.238 0.106 11.653 0b19←INT 0.945 0.056 16.818 0b18←INT 1.071 0.044 24.266 0b17←INT 1b5←PU 1b15←ATT 1.26 0.1 12.544 0

divided by its standard error (S.E.). Based on asignificance level of 0.05, the C.R. needs to be >±1.96. Below this level, the parameter can beconsidered unimportant to the model. The factorloading on PEOU-ATT (Table 5) was 0.082 (with C.R.=0.645, p=0.519), which was not significant, andthus H3 was not supported. The factor loading onPWS-ATT was 0.06 (with C.R.=1.665, p=0.096),which did not support hypothesis H5b. These resultsabout Attitude are not surprising. They are consistentwith the suggestion of Moore and Benbasat [31] thatin the context of IT usage, attitude can besynthesized from perceived characteristics of innovat-ing [33]. In recent empirical studies, the revised TAMfrequently drops the construct Attitude. Futureresearch can also consider taking this alternativeapproach.

4.7. Predictive and explanatory power of the model

The results of this study provide support for ourextended TAM. The original TAM postulates thatintention is the major determinant of user behavior,and it is powerful in predicting and explaining userbehavior based on only three theoretical constructs, i.e.,Intention, Perceived Usefulness and Perceived Ease ofUse [15]. These hypotheses were supported in ourfindings. The positive influence of Perceived WebSecurity on behavioral intention in our extended TAMwas supported, too.

Regarding the individual predicting power of eachconstruct on Behavior Intention, the results indicatethat Perceived Usefulness (0.573) has slightly signif-icant explanatory power, maintaining its influentialrole. Perceived Ease of Use (0.559) and PerceivedWeb Security (0.403) have an added significant effecton Intention. However, Perceived Ease of Use doesnot impact on intention to use directly. Attitude(0.302) has a smaller effect. Hence, we review theAttitude–Intention relationship by contrasting it withTRA and TAM. Both TRA and TAM postulate thatAttitude is determined by one's relevant beliefs. TRAcombines all beliefs into a single construct, whereasTAM posits Usefulness and Ease of Use as twofundamental and distinct constructs. TRA suggests thatAttitude has an intervening effect on beliefs andintention, but TAM and alternative intention modelsprovide theoretical justification and empirical evidencefor direct beliefs–intention links such as the Useful-ness–Intention link [6,15]. In some empirical studies,it has also been found that Attitude might on someoccasion show an insignificant effect on Intention

1567T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

[15], which is consistent with the findings of thisstudy.

Our results provide support for TAM. Ourfindings are also consistent with the findings ofDavis et al. [15] in that Attitude does not fullyintervene the effect of Perceived Usefulness andPerceived Ease of Use on Intention. In conclusion,these results provide several key insights into thedeterminants of IB usage. First, Perceived Usefulnessis a major determinant of customer's intentions touse IB. Second, Perceived Ease of Use is asignificant secondary determinant of customer'sIntention. However, it is mediated through PerceivedUsefulness instead of having a direct impact onIntention. Finally, Perceived Web Security is too asignificant and direct determinant of customer'sIntention.

5. Conclusions and implications

5.1. Limitations

Since this empirical study was performed with atime constraint, as with other cross-sectional studies, itis not without limitations. The IB market in HongKong, as well as knowledge about user behavior inrelation to IB, is at the infancy stage. At a time whenrapid changes in new technologies come to marketdaily, the results of a cross-sectional study may not beperfectly generalizable. A longitudinal study to morefully investigate the pre-launch stage, the promotionstage and the post-launch stage of IB would certainlybe a significant contribution to the IB literature in thefuture.

This study only covers the construct of Websecurity. Future research may deal with the issue ofsecurity and privacy separately, as the latter isarousing increasing attention in the Web literature[21]. In addition, the measurement instrument for Websecurity could be further refined to enhance its validityin future studies.

Our study was carried out in Hong Kong. It maynot be fully generalizable to other territories andcountries, especially those with the western culturesuch as the USA or Australia. More importantly, inthe Chinese market, with rapid changes in rural areasand increasing numbers of Internet users and WAPphone applications, the IB adoption environmentcould change dramatically at any time. It is notdifficult to argue that a theory applicable to HongKong may not necessarily hold true on the Chinesemainland.

5.2. Implications for bank management

Sathye's [35] survey revealed that over 70% ofcustomers expressed concerns on Web security andreal benefits when considering internet banking,which are the highest among other factors such asdifficulty in use and no internet access. However,Sathye [35] did not propose any model to test theimpact of these factors on acceptance of internetbanking. Our research based on TAM suggests thatperceived usefulness has the greatest influence oncustomer intention to adopt internet banking. Per-ceived ease of use, however, does not have a directimpact on intention to use, although it affects theperceived usefulness of customers, which in turn leadsto acceptance of internet banking. Similar findingswere obtained by Pikkarainen et al. [32] and Chanand Lu [10], who investigated acceptance of internetbanking in Finland and Hong Kong, respectively.Both studies reached the same conclusion thatperceived usefulness is more influential than per-ceived ease of use in explaining technology accep-tance of internet banking. One interpretation is thatdifficulty in using on-line systems is becoming less ofa concern as they are increasingly user-friendly. Inaddition, since on-line systems are more prevalent andstandardized nowadays, the public becomes increas-ingly competent in its use. Accordingly, in theplanning and development of internet banking,software developers should pay attention to practicalfunctions and extend key features that are frequentlyrequired. On the marketing side, bankers shouldaccentuate the full functionality of their systems tocater for the different banking needs of the usersefficiently.

This study has also revealed that, in additional toperceived usefulness, perceived Web security has astrong and direct effect on acceptance of internetbanking, too. Perceived Web security was not takeninto consideration in previous research on TAM inthe context of internet banking (e.g., [10,25]). Wesuggest that bankers improve the security features oftheir systems and stress their system security and theprecaution functions they have implemented. Withthese functions, they could reassure their customersthat internet banking is a safe mode to performtransactions.

Banks should also consider how to shift theperceptions of their customers by emphasizing thepositive safety features in any marketing campaign.They should pass an effective message to customersthat the Web security facility now available will

1568 T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

eliminate any third-party intrusions into their IBaccount in order to turn around the negative percep-tions of their customers, thereby enabling customers tofeel secure and comfortable in using IB services.However, incidents where on-line banking suffersfrom security breaches could seriously undermineconsumer confidence. In this regard, proper trainingof customers on the safe use of internet banking couldhelp improve security and enhance their overallconfidence in long-term.

Although we did not study the robustness of ourrevised TAM model, previous research has suggestedthat the constructs of TAM are generally invariant acrossdifferent gender, age, and IT competence subgroups[25]. Accordingly, bankers could adopt this revisedTAM with reasonable confidence.

In recent years, TAM has also been adopted as atheoretical framework to study the acceptance of onlineshopping. Klopping and McKinney [23] found thatperceived usefulness is the key aspect of adoption, whileperceived ease of use has only a minor effect. Consistentwith our findings, Liu et al. [27] found that althoughperceived ease of use affects perceived usefulness, itdoes not impact on attitude to use directly. Theyexplained that since online shopping systems havebecome easier to use and users have become moretechnically savvy, variations in the perceived ease of usedimension are reduced. Similar to this research oninternet banking, Shih's [38] study on e-shoppingextended TAM by adding perceived Web security as afactor and found that high perceived Web securitydirectly increases consumer attitudes toward e-shop-ping, while privacy is not a factor. Gefen et al. [18]suggested that a belief that there are safety mechanismsbuilt into the Web site is essential.

It appears that technology acceptance for internetbanking is very similar to that of on-line shopping.Theoretically, this implies that the revised TAMwith the factor of perceived Web security worksequally well in predicting the acceptance of internetbanking and online shopping. Perceived usefulnessand Web security are key factors nowadays in bothtypes of system. However, one might predict thatwith security features properly built, Web securitymight not be a key concern and perceivedusefulness will become the sole key factor in thefuture.

5.3. Conclusions

The findings of this empirical study providesupport for the theoretical model embracing TAM

and the construct Perceived Web Security. Theresults support the view that Perceived Ease ofUse and Perceived Web Security are predictingvariables, affecting Perceived Usefulness and Atti-tude as intervening variables, and Intention to UseIB as the dependent variable. Perceived Usefulnessand Perceived Web Security have a direct effect onIntention, while Perceived Ease of Use has only anindirect impact. All hypotheses were supportedexcept for H3 (i.e., PEOU-ATT) and H5b (i.e.,PWS-ATT). Perceived Web Security influencesIntention directly, rather than passing through theintervening variable Attitude. This is consistent withthe findings of earlier empirical studies [34]. As isclear from the key fit statistics, the model testingyielded a set of fit indices with an overall well-fit,indicating that the model fitted well with the data.The results of hypothesis testing provide satisfactorysupport for the extended TAM through the SEManalysis.

5.4. Suggestions for future research

Since the study is cross-sectional in design, futureresearch could undertake a more in-depth longitudinalstudy. It would also be worthwhile to examine anyspecific new IB product/service to be launched onthe market, such as mutual funds managementservices or insurance protection. In particular, thestudy could be conducted at an early stage of servicelaunching, or at a pilot test stage in order toinvestigate customers' behavior towards the adoptionof the new IB service. Any corrective actions ifnecessary could then be made at an early stage sothat bank management can improve the new serviceand accelerate the usage rate to recoup the invest-ment costs earlier. Future research may be conductedby further extending and refining TAM and test it inthe context of other technologies. In particular,research could be carried out on technology accep-tance in an organizational setting, such as acceptanceof Enterprises Resource Planning (ERP), and inter-organizational context, such as adoption of Point ofSales (POS). Other potential factors, such as seniormanagement attitude, could be considered in suchstudies as well.

Acknowledgement

We are grateful to two anonymous referees for theirmany constructive comments on an earlier version of thepaper.

1569T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

Appendix A. The questionnaire

Part A:

I. Introductory / general questions (Please circle your ch

Internet usage habit:1. I have access to the Internet: No / yes at home / yes at w2. Months using IB : < 3M / 3–6M / 6–12M / >12M / N.A.

Strdis

3. I use IB frequently4. I encounter problem frequently inusing IB5. I use IB service frequently as sourceof information6. I use IB from more than one bankfrequently7. Besides IB, I use frequently otherbanking channels (ATM, branchteller, phone banking...)

II. Demographic characteristics of the respondent (Please

8. Respondent Position: CEO / CFO / CIO / Owner / Mana(Please state:______________ )9. Industry of corporate: Manufacturing / Retailing / IT relaTourism /other (Please state:_______________ )10. Company size: Staff no. < 50 / 51–100 / 101–300 / 301–(Circle one choice only).11. Using banking services for: Enquiry / Deposit / Loan / Equity / Corporate Finance / Private banking / Asset Manag12. Sex: Male / Female13.Age group: <18 / 19–30 / 31–45

oice of answer)

ork

onglyagree

Stronglyagree

circle yourchoice of answer)

ger / Clerk /Other

ted / Telecomm/ Banking / Financial Institutions /

500 / 501–1000 />1000 / N.A.

Export Financing/ Import Financing / Fixed Income /ement (Circle more than one choice)./ 46–60 / >60

Neutral

1 1 1 1 1

2 2 2 2 2

3 3 3 3 3

4 4 4 4 4

5 5 5 5 5

6 6 6 6 6

7 7 7 7 7

1 2 3 4 5 6 7

1570 T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

Part B:

Please respond to questions below by circling your choice (1=strongly disagree, 7=strongly agree)

Stronglydisagree

Stronglyagree

Perceived Ease of Use(1) Using the Internet Banking service is easy for me(2) I find my interaction with the use ofthe InternetBanking services clear and understandable(3) It is easy for me to become skillful at the use of theInternet Banking services(4) Overall, I find the use of the Internet Bankingservices easy

Perceived Usefulness(5) Using the Internet Banking would enable me toaccomplish my tasks more quickly(6) Using the Internet Banking wouldmake it easier for meto carry out my tasks(7) I would find the Internet Banking useful(8) Overall, I would find using the Internet Banking to beadvantageous

Perceived Web Security(9) I would feel secure sending sensitive information across theInternet Banking(10) The Internet Banking is a secure means through which tosend sensitive information(11) I would feel totally safe providing sensitive information about myself over the Internet Banking(12) Overall, the Internet Banking is a safe place to transmitsensitive information

Attitude(13) Using Internet Banking is a good idea(14) I would feel that using Internet Banking is pleasant(15) In my opinion, it would be desirable to use Internet Banking(16) In my view, using Internet Banking is a wise idea

Intention to Use(17) I would use the Internet Banking for my banking needs.(18) Using the Internet Banking for handling my bankingtransactions is something I would do(19) I would see myself using the Internet Banking forhandling my banking transactions

Neutral

1 1

1

1

1

1

1 1

1

1

1

1

1 1 1 1

1 1

1

2 2

2

2

2

2

2 2

2

2

2

2

2 2 2 2

2 2

2

3 3

3

3

3

3

3 3

3

3

3

3

3 3 3 3

3 3

3

4 4

4

4

4

4

4 4

4

4

4

4

4 4 4 4

4 4

4

5 5

5

5

5

5

5 5

5

5

5

5

5 5 5 5

5 5

5

6 6

6

6

6

6

6 6

6

6

6

6

6 6 6 6

6 6

6

7 7

7

7

7

7

7 7

7

7

7

7

7 7 7 7

7 7

7

1 2 3 4 5 6 7

1571T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

References

[1] ACNielsen and NetRatings, 2nd Quarter 2002 Report, 2003,http://www.eratings.com/news/ (retrieved on 24 April 2003).

[2] D.A. Adams, R.R. Nelson, P.A. Todd, Perceived usefulness, easeof use, and usage of information technology: a replication, MISQuarterly 16 (1992) 227–247.

[3] R. Agarwal, J. Prasad, Are individual differences germane to theacceptance of new information technologies? Decision Sciences30 (1999) 361–391.

[4] American Banker, In brief: fast growth for small-biz webbanking, American Banker 167 (20) (2002).

[5] H.B. Asher, Causal Modeling, Sage University Press, NewburyPark, 1983.

[6] R.P. Bagozzi, A field investigation of causal relations amongcognitions, affect, intentions and behavior, Journal of MarketingResearch (Nov. 1982) 562–584.

[7] P.M. Bentler, Comparative fit indexes in structural models,Psychological Bulletin 107 (1990) 238–246.

[8] P.M. Bentler, D.G. Bonett, Significance tests and goodness of fitin the analysis of covariance structures, Psychological Bulletin88 (1980) 588–606.

[9] R.D. Brown, E-commerce: customer service success factors,Futurics 25 (2001) 18–30.

[10] S.C. Chan, M.T. Lu, Understanding internet banking adoptionand use behavior: a Hong Kong perspective, Journal of GlobalInformation Management 12 (2004) 21–43.

[11] P.Y.K. Chau, Reexamining a model for evaluating informationcenter success using a structural equation modeling approach,Decision Sciences 28 (1997) 309–335.

[12] W.W. Chin, P.A. Todd, On the use, usefulness, and ease of use ofstructural equation modeling in MIS research: a note of caution,MIS Quarterly 19 (1995) 237–246.

[13] J. Cohen, Statistical Power Analysis for the Behavioral Sciences,2nd Edition, Lawrence Erlbaum Associates, Hillsdale, NewJersey, 1988.

[14] F.D. Davis, Perceived usefulness, perceived ease of use, and useracceptance of information technology, MIS Quarterly 13 (1989)319–336.

[15] F.D. Davis, R.P. Bagozzi, P.R. Warshaw, User acceptance ofcomputer technology: a comparison of two theoretical models,Management Science 35 (1989) 982–1003.

[16] eMarketer, The US B2C Review, 2003, http://www.emarketer.com/estats/081699_data.html http://www.emarketer.com/products/report.php?eglobal (retrieved on 29 August 2003).

[17] M.A. Fishbein, I. Ajzen, Belief, Intention and Behavior: AnIntroduction to Theory and Research, Addison-Wesley, Reading,Massachusetts, 1975.

[18] D. Gefen, E. Karahanna, D.W. Straub, Trust and TAM in onlineshopping: an integrated model, MIS Quarterly 27 (2003)51–90.

[19] J. Hair, R. Anderson, R. Tatham, W. Black, Multivariate DataAnalysis, Prentice Hall, New Jersey, 1998.

[20] A.R. Hendrickson, P.D. Massey, T.P. Cronan, On the test–retestreliability of perceived usefulness and perceived ease of usescales, MIS Quarterly 17 (1993) 227–230.

[21] S.L. Jarvenpaa, P.A. Todd, Consumer reactions to electronicshopping on the world wide web, International Journal ofElectronic Commerce 1 (1997) 59–88.

[22] K.G. Joreskog, D. Sorbom, LISREL 7: A Guide to the Programand Applications, SPSS Inc, Chicago, 1988.

[23] I.M. Klopping, E. McKinney, Extending the technologyacceptance model and the task-technology fit model to consumere-commerce, Information Technology, Learning, and Perfor-mance Journal 22 (2004) 35–48.

[24] X.A. Koufteros, Testing a model of pull production: a paradigmfor manufacturing research using structural equation modeling,Journal of Operations Management 17 (1999) 467–488.

[25] V.S. Lai, H. Li, Technology acceptance model for internetbanking: an invariance analysis, Information and Management42 (2005) 373–386.

[26] A.L. Lederer, D.J. Maupin, M.P. Sena, Y.L. Zhuang, Thetechnology acceptance model and the world wide web, DecisionSupport Systems 29 (2000) 269–282.

[27] S. Liu, D. Tucker, C.E. Koh, L. Kappelman, Standard userinterface in e-commerce sites, Industrial Management and DataSystems 103 (2003) 600–610.

[28] K. Mathieson, Predicting user intention: comparing the technol-ogy acceptance model with the theory of planned behavior,Information Systems Research 2 (1991) 173–191.

[29] W.W. Moe, P.S. Fader, Uncovering patterns in cyber-shopping,California Management Review 43 (2001) 106–117.

[30] J.W. Moon, Y.G. Kim, Extending the TAM for a world-wide-web context, Information and Management 38 (2001)217–230.

[31] G.C. Moore, I. Benbasat, Development of an instrument tomeasure the perceptions of adopting an information technol-ogy innovation, Information Systems Research 2 (1991)192–222.

[32] T. Pikkarainen, K. Pikkarainen, H. Karijaluoto, S. Pahnila,Customer acceptance of on-line banking: an extension of thetechnology acceptance model, Internet Research 3 (2004)224–235.

[33] E.M. Rogers, Diffusion of Innovations, 4th Edition, The FreePress, New York, 1995.

[34] W.D. Salisbury, R.A. Pearson, A.W. Pearson, D.W. Miller,Perceived security and world wide web purchase intention,Industrial Management and Data Systems 101 (2001)165–176.

[35] M. Sathye, Adoption of internet banking by Australianconsumers: an empirical investigation, International Journal ofBank Marketing 17 (1999) 324–334.

[36] S. Sharma, Multivariate Techniques, Wiley, New York, 1996.[37] B.H. Sheppard, J. Hartwick, P.R. Warshaw, The theory of

reasoned action: a meta-analysis of past research with recom-mendations for modifications and future research, Journal ofConsumer Research 6 (1988) 144–176.

[38] H.P. Shih, An empirical study on predicting user acceptance of e-shopping on the web, Information and Management 41 (2003)351–368.

[39] J.H. Steiger, J.C. Lind, Statistically based tests for the number ofcommon factors, Paper, presented at the Psychometric SocietyAnnual Meeting, Iowa City, Iowa, 1980.

[40] B. Szajna, Software evaluation and choice: predictive validationof the technology acceptance instrument, MIS Quarterly 18(1994) 319–324.

[41] V. Venkatesh, M.G. Morris, Why don't men ever stop to ask fordirections? Gender, social influence, and their role in technologyacceptance and usage behavior, MIS Quarterly 24 (2000)115–139.

[42] B. Wheaton, B. Muthen, D.F. Alwin, G.F. Summers, Assessingreliability and stability in panel models, Sociological Methodol-ogy 8 (1977) 84–136.

1572 T.C.E. Cheng et al. / Decision Support Systems 42 (2006) 1558–1572

Prof. T.C. Edwin Cheng is Chair Professorof Management in the Department of

Logistics of The Hong Kong PolytechnicUniversity. He obtained his bachelor's,master's and PhD degrees from the Uni-versities of Hong Kong, Birmingham, andCambridge, respectively. Prof. Cheng'sresearch interests are in Operations Manage-ment and Operations Research. He haspublished over 300 journal papers and co-

authored three books. He received the Outstanding Young Engineer ofthe Year Award of the Institute of Industrial Engineers, U.S.A., in1992, and the Croucher Senior Research Fellowship (the top scienceaward in Hong Kong) in 2001. He was named one of the “most citedscientists” in All Fields and in Engineering over the period 1996-2006by the ISI Web of Knowledge in 2006. One of his papers published inthe European Journal of Operational Research in 1989 was voted oneof 30 influential articles published in that journal since its inception.

Dr Y.C. David LAM is Senior VicePresident and Head of Treasury RiskManagement Department of Fubon Bank(Hong Kong) Ltd. He obtained a Master ofBusiness Administration degree from theUniversity of East Asia, Macau, a Master ofScience degree in Financial Managementfrom the National University of Ireland,Dublin, and a Doctor of Business Admin-istration degree from The Hong KongPolytechnic University. He was among the

top 50 in the world in passing the Banking Diploma examinationconducted by the Institute of Bankers, UK, two decades ago. He is aFellow Member of the Association of International Accountants, UK,and an Associate Member of the Hong Kong Institute of CertifiedPublic Accountants. He has over 35 years of banking experience andhas held senior management positions at various international banks,including Citibank, Manufacturers Hanover Trust Company (now JPMorgan Chase Bank), and BNP Paribas. When working for BNPParibas, he was responsible for the operational and technology fieldsof the bank’s Head Office in Paris, and its branches in Hong Kong,London and Singapore.

Andy C. L. Yeung is an Assistant Professor,

specializing in operations management, inthe Department of Logistics, The HongKong Polytechnic University. He receivedhis Ph.D. in operations management fromThe University of Hong Kong. His researchpapers have been published in differentacademic journals, including the Produc-tion and Operations Management, theJournal of Operations Management, andthe IEEE Transactions on Engineering

Management, among others.