CONSUMER ADOPTION OF INTERNET BANKING

Abstract

The rapid changes in Malaysian banking system have created a new dimension in the banking industry with the emergence of Internet banking. The study also explores the consumers adoption behaviour driven by the evolution of new, banking technology in Malaysia. The result implies that a social norm effect dominates Internet banking adoption

This paper presents the profile of the internet banking adopter in Malaysia based on a large scale survey. A logistic model is used to estimate the probability of a bank customer adopting internet banking. The profile of the adopter is constructed using demographic, social economic and technological capacity indicators. The method is a very simple one, which can be used by financial institutions to gain better understanding of their own internet banking customers.

Introduction

!anks have traditionally been in the forefront of exploiting technology to improve their products, services and efficiency. The banking industry has seen many technological changes in the last forty "#$% years, which has shaped it from manually intensive industry into one that is highly automated and technologically dependent.

Internet banking is both a process and product electronic innovation "&hang '$$#%. It enables customers to handle their banking transactions online, without physical visits to the bank. In general, however, new innovations particularly technological ones, are not readily accepted and adopted by everyone. (ogers ")**+% classifies innovation adopters into five broad categories,

")% innovators,

"'% early adopters,

"-% early ma.ority,

"#% late ma.ority, and

"+% laggards.

Innovators are the first adopters, who are interested in technology itself and possess positive technology attitudes. /n the other hand, early adopters are also interested in technology and are willing to take risks. The International 0ata &orporation investigated urban internet users in six Asian countries "I0& '$$'% and found some important profiles of internet adopters. 1arly adopters of wireless internet were usually young and male tech savvy users at an average age of '2 years. 3#4 of them were male. In contrast, the early ma.ority category consisted of mainly young working adults. There was a larger proportion of females and working population within the late ma.ority category. 5inally, laggards were found to be predominantly older people.

In Malaysia, a large scale survey was conducted in '$$# to examine selected individual characteristics of over 2$$ bank customers. This paper reports some ma.or findings of the Malaysian study. 6sing data obtained from the sample, an attempt is made in this paper to construct the profile of a 7typical7 customer who has adopted internet banking, and to estimate marginal effects of general individual attributes on the predicted probability of the internet banking adopter.

Methodology

A large scale survey was conducted on the island of 8enang in Malaysia. The

sub.ects consisted of bank customers outside of bank branches and shoppers who maintained bank accounts. /ut of the total of 2-) 9uestionnaires collected, :3- were used for the purpose of data entry, processing and analysis. Among others, the study attempted to explain the probability that a person will adopt internet banking. 5or this purpose, the explanatory variables were confined to individual attributes, which are divided into three categories, vi;., demographic, social economic standing, and technological competency<access. Theses factors are then examined to see how they affect the dependent variable, which is the adoption of internet banking. A logit specification is adopted= the parameters in the discreet choice models are transformed to yield estimates of the marginal effects, i.e. the change in predicted probability associated with changes in the explanatory variables. The model reported in this paper is therefore a very simple one, aiming only to provide an initial estimate of the probability of a person adopting internet banking, and explain how marginal changes in each explanatory variable affect the probability of adoption.

The logit model is written as follows,

88>og ) ? @)A @'B'A @-B-A@#B#A@+B+A@3B3A@:B:A@2B2A

@*B*A@)$B)$A@))B))A@)'B)'A@)-B)-Ae

Chere, 8 ? the probability of respondents adopting

Internet banking facilities

0ependent variable, D ? ) if adopt Internet banking facilities

$ otherwise "preferring other alternatives%=

1xplanatory variables, B ? hypothesi;ed to influence probabilities of

independent variables

@ ? coefficients of the explanatory variables=

e ? stochastic disturbance term.

8rofile of the 7typical7 Internet !anking Adopter

Table ) shows the mean values of the explanatory variables as found in the survey sample "column '%. 6sing the mean values, we can construct the profile of a 7typical internet banking adopter.7

In order to calculate the marginal effects of probability, the mean values of age, income and fre9uency of visits obtained from the sample date to be calculated with the probability of the Etypical, respondent. Fubstituting these values for the typical respondent into the estimated logit e9uation yields the log of odds of adopting Internet banking to be $.:2-3). The probability of adopting Internet bank is computed as $.3*.

The survey shows that a 7typical7 Malaysian internet banking adopter is a &hinese male aged '*. Ge has had tertiary education, works in the private sector or is self employed, drawing a monthly income of (M''++. The typical internet

banking adopter is married and resides in urban areas. 8roficient in the 1nglish language which is the most common medium of communication in the internet= he is also computer literate and has ready access to the internet. In general, he visits the bank three times a month and is a risk averter.

Table ), 8rofile H Marginal 1ffects on the 8robability of Adopting Internet !anking

Independent Iariables

Fample

Marginal 1ffects

1thnicity

&hinese

$.$2$$:

Jender

Male

$.$+2)2

Age

'*

$.$$+*:

1ducation

Tertiary

$.))2+3

/ccupation

8rivate Fector<self employed

$.))3+#

Income "Malaysian ringgit%

''++

$.$$$)*

Marital Ftatus

Married

$.$#$'#

(esidence

6rban

$.$2:3*

>anguage

1nglish

$.))3))

5re9uency

$.'$'-$

(isk

(isk Averter

$.)$$$)

&omputer >iteracy

&omputer >iteracy

$.'-*3#

Internet Access

Internet Access

$.':#-3

Marginal 1ffects on 8robability of Adopting Internet !anking

A significant advantage of using the logit model is the opportunity provided by the model for us to examine the marginal effects. The estimated logit coefficients generated from our sample data show how individual characteristics affect the log odds of adopting internet banking. Jiven the estimated coefficients and the changes in odds, we can compute, using the binary choice logit model, the predicted probability of the adopter for any change in any of the explanatory variable. The marginal effects shown in the last column of Table ), which are associated with respective explanatory variables, have been computed from the estimated coefficients of and the changes in the log of odds.

Table '. Marginal 1ffects on the /dds of adopting Internet !anking

Independent Iariables

1stimated &oefficient

Ftandard

1rror

K Ftatistic

8robability Ialue

&onstant

'.:)3+*2

$.:-$+$3

-.:)2:2*

$.$$$'

(ace

$.-:#-3$

$.')*3'2

).:$#+):

$.$22-L

Jender

$.':)*:3

$.')++33

).'3)32#

$.'$:)

Age

$.$':2*2

$.$)3332

).3:-3*'

$.$*#'L

1ducation

$.++#'*-

$.')3*3+

'.++#:+2

$.$)$3LL

/ccupation

$.+##2'*

$.')*-2-

'.#2-#3$

$.$)-$LL

Income

$.$$$*$:

$.$$$))2

:.33-'2#

$.$$$$LLL

Marital

$.)22)'-

$.')3+#$

$.232::$

$.-2+$

(esidence

$.#$**-*

$.'-:*'#

).:''*2)

$.$2#*L

>anguage

$.+#'2))

$.'):2##

'.#*):#+

$.$)':LL

5re9uency

$.*#+:+$

$.)'3'*+

:.#22#$-

$.$$$$LLL

(isk

$.#3:+3:

$.')'3:+

'.)*2+$#

$.$':*LL

&omp >iteracy

).)'$-+2

$.'':3$#

#.*''#$*

$.$$$$LLL

Internet Access

).'2'3#*

$.'#'-:'

+.'*'$3#

$.$$$$LLL

>ikelihood ratio ")-df%

-:$.-2*3

8robability ">( Ftat%

$.$$$$$$

LLL Fignificant at ) percent level of significance

LL Fignificant at + percent level of significance

L Fignificant at )$ percent level of significance

Adoption of Internet banking is expected to be positively related to race, gender, education, occupation, income, residential area, language, risk, computer literacy and Internet access but negatively related to age, marital status and fre9uent visit to the bank branch. All the variables have the expected signs, although not all the variables are statically significant.

Table -. Marginal 1ffects on the /dds of Adopting Internet !anking

Independent Iariables

1stimated &oefficient

&hange in /dds

.@

(ace

$.-:#-3$L

).#+#)

Jender

$.':)*:3

).-)'3

Age

$.$':2*2L

$.*:'+

1ducation

$.++#'*-L

).:#$:

/ccupation

$.+##2'*L

).:'#-

Income

$.$$$*$:L

).$$$*

Marital Ftatus

$.)22)'-

$.2'2+

(esidence

$.#$**-*L

).+$3:

>anguage

$.+#'2))L

).:'$2

5re9uency

$.*#+:+$L

$.-22#

(isk

$.#3:+3:L

).+*3)

&omp >iteracy

).)'$-+2L

-.$33$

Internet

).'2'3#*L

-.3$3'

L Ftatistically significant variables at various levels

Income of the respondents is positively correlated with adoption of Internet banking. As income increases the tendency to adopt, Internet banking also increases. Chen income increases by (M).$$, the log of odds of adopting Internet banking increases by $.$$$*. Gigher income households tend to react more positively by adopting Internet banking because they come from the more affluent part of the society.

The log of odds for respondents who are fre9uent visitor to the bank branches, adopting Internet baking is $.*#+2 times lower than respondents who are not fre9uent visitors to bank branches. &eteris paribus, when the number of visits to the bank branches increases by ) time, the odds of adopting Internet banking decreases by 3).'4.

The relationship between the decision to adopt Internet banking and computer literacy is statistically significant at $.$) level of significance. The result

suggests that the log of odds for computer literate respondents adopting Internet banking is ).)'$# times higher than respondents who are computer illiterate. Those who are computer illiterate significantly have a lower likelihood to adopt Internet banking. This result also suggest that, ceteris paribus, the odds of adopting Internet banking increases by -.$$3$ times when a respondent is computer literate compared to one who is computer illiterate.

The relationship between decision to adopt Internet banking and Internet access is one of the most important variables. Internet access is positively related with the likelihood to adopt Internet banking, at $.$) level of significance. The log of odds for respondents having Internet access adopting Internet banking

is ).'2'3 times higher than respondents without Internet access. The findings indicate that, ceteris paribus, the odds of adopting Internet banking increase by -.3$3' when a respondent has Internet access.

The log of odds of a tertiary educated respondent adopting Internet banking increases by $.++#- compared to non tertiary education. This study also indicates that, other things being e9ual the odds of adopting Internet banking increases by :#4 when a respondent has tertiary education compared to those who do not possess tertiary education.

/ccupation is positively and statistically significant related to adopting Internet banking. The relationship between occupation and adoption of Internet banking is statistically significant at $.$+. The log of odds for private sector or self employed respondent adopting Internet banking is $.+##2 times higher than public sector respondents. The outcome also indicates that, ceteris paribus, the odds of adopting Internet banking increases by :'4 when a respondent is a private sector employee or self employed compared to public sector employee. The other significant difference appears to be in the average level of monthly income in the private sector or self employed is higher compared to public sectors.

The impact of selected marginal effects on the predicted probability of the typical internet banking adopter is shown in Table #.

To start with, &ase no. ) in Table # represents the profiles of the 7typical internet banking adopter7 discussed earlier, which has been constructed from the mean values of the variables. The predicted probability of $.3* simply means that bank customers with such individual profiles have a 3*4 chance of being an internet banking adopter. Gowever, what happens to the predicted probability if the profile of the typical adopter is changed marginally each timeM 6sing data from Table ', we shall look at marginal effects arising from demographic and social economic factors below.

Marginal effects of 0emographic 5actors

Fuppose the bank customer happens to be a non &hinese, but all other characteristics of the typical adopter remain the sameM &ase no. ' shows that the predicted probability diminishes by $.$2 compared to the original typical adopter. As a result, the predicted probability of such a person is now $.3$. This means that a non &hinese customer identical in all other attributes will be 24 less likely to adopt internet banking. The next two cases indicate the changes in the predicted probability conse9uent to marginal changes in other demographic factors "vi;. gender and age%. It shows that an older person, ceteris paribus, is less likely to adopt internet banking. If the age of the typical adopter is increased to #+, "case no. #%, the predicted probability of internet banking adoption

diminishes considerably to $.+2.

Table #, Marginal 1ffects on 8robability of Adopting Internet !anking

No

&haracteristics

8redicted 8robability

&hinese male age '* with tertiary education, working in private sector< self employed, monthly individual income of (M''++, married, residing in urban area, ability to communicate in 1nglish language, average of visiting bank times in a month, risk averter, literacy in computer and have access to internet

$.3*

'

Non &hinese male, age '* with tertiary education, working

in private sector< self employed, monthly income of (M''++, married, residing in urban area, ability to communicate in 1nglish language, average of visiting bank - times in a month, risk averter, literacy in computer and have access to internet.

$.3$

&hinese female, age '* with tertiary education, working in private sector< self employed, monthly income of (M''++, married, residing in urban area, ability to communicate in 1nglish language, average of visiting bank - times in a month, risk averter, literacy in computer and have access to internet.

$.3-

&hinese male, age #+ with tertiary education, working in private sector< self employed, monthly income of (M''++, married, residing in urban area, ability to communicate in 1nglish language, average of visiting bank - times in a month, risk averter, literacy in computer and have access to internet.

$.+2

&hinese male, age '* with non tertiary education, working in private sector< self employed, monthly income of (M''++, married, residing in urban area, ability to communicate in 1nglish language, average of visiting bank - times in a month, risk averter, literacy in computer and have access to internet.

$.+3

&hinese male, age '* non tertiary education, working in private sector< self employed, monthly income of (M)$$$, married, residing in urban area, ability to communicate in 1nglish language, average of visiting bank - times in a month, risk averter, literacy in computer and have access to internet.

$.#)

&hinese male, age '* with tertiary education, working in private sector< self employed, monthly income of (M-$$$, married, residing in urban area, ability to communicate in 1nglish language, average of visiting bank - times in a

$.2)

month, risk averter, literacy in computer and have access to internet.

&hinese male, age '* with tertiary education, working in private sector< self employed, monthly income of (M''++, married, residing in rural area, ability to communicate in 1nglish language, average of visiting bank - times in a month, risk averter, literacy in computer and have access to internet.

$.+*

&hinese male, age '* with tertiary education, working in private sector< self employed, monthly income of (M''++, married, residing in urban area, ability to communicate in 1nglish language, average of visiting bank - times in a month, risk averter, non literate in computer and have access to internet.

$.#'

)$

&hinese male, age '* with tertiary education, working in private sector< self employed, monthly income of (M''++, married, residing in urban area, ability to communicate in 1nglish language, average of visiting bank - times in a month, risk averter, literacy in computer and have no access to internet.

$.-2

Ce can similarly estimate marginal effects of social economic factors on the predicted probability of the typical adopter. &ases + to 2 in Table ' show some worked out examples. Income appears to be a ma.or determinant of adoption. A

slight increase in the person7s monthly pay to (M-$$$ enhances the probability from $.3* to $.2) "&ase :%. /n the other hand, if the typical person has only (M)$$$ in monthly income, the predicted probability declines to $.#) "&ase 3%O (esidence can be classified as either demographic or social economic factor. Table ' shows that residing in rural areas reduces the predicted probability of adopting internet banking to $.+*. Among social economic environmental limitations, non availability of internet services in the rural areas is definitely a significant constraint, reducing the predicted to a mere $.-2 "&ase )$%.

&onclusion

This short paper attempts to show that using the logistic model, it is possible to 9uickly come up with a profile of bank customers who adopt internet banking. Most of the data on the explanatory variables, such as the demographic and social economic indicators can be readily found in the data base of customers. Gence financial institutions will be able to construct the profile of their own customers. Pnowledge of the profiles or more generally the individual characteristics of customers is useful in many ways. 19uipped with information on customers who already adopt internet banking or likely to do so, banks will be able to identify the market segments that should be targeted. They can then

introduce banking products and services that better suit the needs and wants of the customers in the segment.

(eferences

Abdullah, K. ")*2+%, A critical review of the impact of ATMs in Malaysia. !ankers Qournal Malaysia. April )*2+. pp. )- )3.

&hang, D.T. "'$$#%, 0ynamics of !anking Technology Adoption, An Application to Internet !anking. http,<<www.uea.ac.uk<R.)$3<I!.pdf, accessed 2 Qan. '$$+.

I0& "'$$'%, I0& predicts youths and road warriors to lead wireless data adoption. http,<<www.idc.com.sg<8ress<'$$'<A8 8( unwiring.html, accessed )* March '$$'.

Moore, &. and !enbasat, I. ")**)%, SS0evelopment of an instrument to measure the perceptions of adopting an information technology innovationT, Information Fystems (esearch, Iol. ' No. -, pp. )*' '''.

(ogers, 1. ")**+%, 0iffusion of Innovations , #th ed., The 5ree 8ress, New Dork, ND.

Fathye, M. ")***%, TAdoption of Internet banking by Australian consumers, an empirical investigationT, International Qournal of !ank Marketing, Iol. ): No. :, pp. -'# -#.

Teo, T. "'$$)%, T0emographic and motivation variables associated with Internet usage activitiesT Internet (esearch, 1lectronic Networking Applications and 8olicy, Iol. )) No. ', pp. )'+ -:.