Decision Support Systems 24 Ž1998. 29–43

An empirical study on consumer acceptance of products in electronic markets: a transaction cost model
Ting-Peng Liang ) , Jin-Shiang Huang

1

Department of Information Management, National Sun Yat-sen UniÕersity, Kaohsiung, Taiwan

Abstract
Electronic commerce is gaining much attention from researchers and practitioners. Although increasing numbers of products are being marketed on the web, little effort has been spent on studying what product is more suitable for marketing electronically and why. In this research, a model based on the transaction cost theory is developed to tackle the problem. It is assumed that customers will go with a channel that has lower transactional costs. In other words, whether a customer would buy a product electronically is determined by the transaction cost of the channel. The transaction cost of a product on the web is determined by the uncertainty and asset specificity. An empirical study involving eight-six Internet users was conducted to test the model. Five products with different characteristics Žbook, shoes, toothpaste, microwave oven, and flower. were used in the study. The results indicate that Ž1. different products do have different customer acceptance on the electronic market, Ž2. the customer acceptance is determined by the transaction cost, which is in turn determined by the uncertainty and asset specificity, and Ž3. experienced shoppers are concerned more about the uncertainty in electronic shopping, whereas inexperienced shoppers are concerned with both. q 1998 Elsevier Science B.V. All rights reserved.
Keywords: Electronic commerce; Transaction cost theory; Internet marketing

1. Introduction
Electronic commerce is a modern business practice associated with the buying and selling of information, products, and services via Internet w13x. The rapid proliferation of Internet and the world-wide web ŽWWW. has created a fast growing electronic channel for marketing. A major reason for this is that the growing popularity of Internet attracts businesses to invest in online presence w17x. According to re-

)
Corresponding author. Tel.: q886-7-531-4567; fax: q886-7531-3565; e-mail: liang@mis.nsysu.edu.tw
1
E-mail: jshuang@mis.nsysu.edu.tw

ports from the Internet Society, over 186 countries have Internet access. The total number of users was approximately 30 million by the end of 1996 w2x.
More than half US Internet users purchased merchandise online, and sales totalled US$500 million in
1996. The cases in other countries are similar. For instance, over 42% of Internet users have visited at least one electronic store in Taiwan w16x. This figures are very much underestimated, as they do not reflect users who consulted the Internet before purchasing offline. As the Internet continues to grow, analysts predict US$100 billion in retail Internet sales by
2000 w10x.
Given the enormous business potential, the number of electronic stores increases in an unprecedented

0167-9236r98r$ - see front matter q 1998 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 7 - 9 2 3 6 Ž 9 8 . 0 0 0 6 1 - X

30

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43

speed. Research issues in electronic commerce have also attracted much attention. For instance, Hamilton w10x and Hoffman w11x examined the commercial opportunities and challenges of the web. Tennebaum, et al. w22x proposed an architecture for Internet commerce. Kambil w14x studied potential electronic business processes. Bhimani w1x and Denning w4x discussed the security in electronic commerce. Panurach w19x and Neuman w18x studied electronic payment systems. Gupta, et al. w8,9x studied the economic issues in electronic commerce. Liang and Lai w16x developed guidelines for designing electronic malls.
Since Internet may have significant impact on the way we do business, the economic issue will play a key role in future development of electronic commerce. These include pricing network services, offering proper products and services, governance structures of electronic trading, marketing and consumer behavior on the electronic market, and so on.
As a new channel for marketing, the web is capable of accommodating many different kinds of products and services. Due to the nature of electronic channels, however, not all products and services are suitable for marketing electronically. Failed cases on the Internet are as common as successful ones.
Therefore, the key is not whether to market, but what to market on the web. Although some practical guidelines are available, theoretically sound research is highly desired.
The purpose of the research is to develop a consumer choice model based on the transaction cost theory, which would allow us to answer the questions of what products and serÕices are suitable for electronic markets and why they are suitable.
Transaction cost economics ŽTCE. w3,24–27x examines the appropriate governance structures to conduct transactions. It argues that transaction costs are the major concern when a company is choosing between producing internally and acquiring over the market. TCE has been applied to analyze many issues such as strategic impact of information systems, resource allocation, and outsourcing decisions.
Since doing business on Internet can be considered as an alternative channel to the traditional channel,
TCE is a viable theory to explain the acquisition decision in electronic commerce.
The paper is organized as follows. First, the characteristics and product acquisition cycle of electronic

commerce are introduced. Then, TCE is applied to build a consumer choice model for electronic commerce. This is followed by an empirical test of the model. The results indicate that the TCE-based model performs well in explaining the sample surveyed in the study. Specifically, whether a consumer would buy a product from electronic markets is determined by the perceived transaction costs, which is in turn determined by Ž1. product and process uncertainty, and Ž2. the specificity of the product, which includes site specificity, physical asset specificity, human asset specificity, and brand name specificity. An interesting observation is that, for experienced electronic shoppers, the effect of asset specificity diminishes.
The perceived uncertainty becomes the sole factor that determines the consumer choice.

2. Transaction process and transaction cost theory
There are many factors that may affect a customer’s decision to purchase from electronic stores.
For example, one may like the design of a particular shop or the discount offered by a shop. From the economic point of view, we focus on the economic factors for a consumer to purchase over the web instead of a traditional store. One particular aspect that we examine is the costs associated with the transaction process. In other words, given that all other factors are equal, a customer will go with a channel that has lower transaction costs. In this section, we review customer transaction processes and costs associated with these processes.
2.1. Customer transaction processes
When customers purchase a product from a seller, they must go through a process. This is called a transaction process. A typical consumer decision process includes five stages: problem recognition, search, alternative evaluation, choice, and outcome w5x. Another approach, called the customer resource life cycle ŽCRLC., divides the linkage between a company and its customers into four major stages: requirement, acquisition, stewardship, and retirement w12x. Each stage is composed of a few tasks as listed in Table 1. Another mercantile model decomposes

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43

31

Table 1
Comparison of several customer decision models
Customer decision process

CRLC

Problem recognition

Establish requirement
Specify attributes

Choice

Select source
Authorize and pay
Acquire
Test and accept
Integrate
Monitor
Upgrade
Maintain
Transfer or dispose
Account for

the process into three stages: purchase determination, purchase consumption, and postpurchase interaction w13x. In fact, these models propose a similar process by which a customer interacts with a seller. The major difference is their emphasis. As we see in Table 1, if we use the customer decision model as a base,
CRLC emphasizes more on the activities after making a choice, whereas the mercantile model is more balanced. For the purpose of this research, we define the transaction process as a seven-step process Žas listed in the fourth column in Table 1.:
1. Search: search for relevant product or service information. 2. Comparison: compare prices or other attributes.
3. Examination: examine the products to be purchased.
4. Negotiation: negotiate terms, e.g., price, delivery time, etc.
5. Order and payment: place an order and pay for it.
6. Delivery: delivery of products from the seller to the customer.
7. Post-service: customer service and support.
2.2. Transaction cost theory
A transaction is a process by which a good or service is transferred across a technologically separa-

Transaction process

Search
Comparison

Search
Alternative evaluation

Mercantile model

Bargaining
Order
Authorize payment

Search
Comparison
Examination
Bargaining
Order
Payment

Receipt of product
Post-service

Delivery
Post-service

ble interface w24,27x. In the classical economic theory, it is assumed that information is symmetric in the market. Since both buyers and sellers are assumed to have the same amount of information, the transaction can be executed without cost. In reality, however, markets are often inefficient. In order to proceed with a transaction, consumers must conduct activities such as searching for information, negotiating terms, and monitoring the on-going process to ensure a favorable deal w3x. The costs involved with such transaction-related activities are called transaction costs.
TCE theoretically explains why a transaction subject chooses a particular form of transaction instead of others. The basic principle of TCE is that people like to conduct transactions in a way that minimizes their transaction cost. TCE has been successfully applied in many domains. Among the well-known examples are TCE applications to facilitate the make-or-buy and business integration decisions. The market mechanism is popular when the transaction cost of buying outside is lower than manufacturing inside, while organizational hierarchy is used otherwise. In automobile and petrochemical industries, forward integration is common because transaction costs of forward-integrated companies are much lower than those of non-integrated ones. Detailed review of TCE applications is available in Refs. w20,21x. T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43

32

minimizes their transaction costs and transaction costs are affected by transaction characteristics such as uncertainty and asset specificity, the model is composed of four constructs, as shown in Fig. 1.
This model allows us to generate two sets of propositions concerning marketing products or services on the web.
Fig. 1. Research framework.

Since purchasing from electronic stores can be considered as a choice between the web and traditional stores, it is reasonable to assume that the consumer will go with the channel that has a lower transaction cost Žassuming other costs being equal..
Given our previous description of transaction steps, the transaction costs that affect the decision of whether to buy from an electronic store include costs for searching information, comparing attributes, examining products, negotiating terms, paying for products, deliÕering products, and post-sales serÕices.
The transaction cost may be affected by several factors including uncertainty and asset specificity.
Uncertainty refers to the cost associated with the unexpected outcome and asymmetry of information.
A higher level of uncertainty generally implies a higher transaction cost. Asset specificity refers to
‘durable investments that are undertaken in support of particular transactions, the opportunity cost of which investment is much lower in best alternative uses or by alternative users’ w27x. Transactions that are supported by high levels of asset specificity should be governed by hierarchical structures, whereas transactions that require only general-purpose investments will most efficiently be conducted over markets. For electronic commerce, therefore, we infer that the uncertainty and asset specificity of the product offered on the web will affect the transaction cost and, in turn, affect the customer decision of whether to buy electronically.

Proposition 1: Transaction costs of marketing a product through the electronic channel are determined by the uncertainty and asset specificity of the product and the process.
Proposition 2: The decision to purchase on the web is determined by the uncertainty and asset specificity of the product to be marketed on the web and the transaction cost.

4. Research method
To test the transaction cost model presented above, an empirical study was conducted. Five products, book, shoes, toothpaste, microwave oven, and flower, were chosen to assess the customer acceptance. A questionnaire was designed to collect information from 85 subjects who were familiar with Internet

Table 2
Operationalization of model constructs
Constructs

Operational measures

Acceptance

Perceived acceptance of electronic channel
Search cost
Comparison cost
Examination cost
Negotiation cost
Payment cost
Delivery cost
Post-service cost
Product uncertainty
Process uncertainty
Site specificity
Physical asset specificity
Human asset specificity
Brand name specificity
Temporal specificity

Transaction cost

Uncertainty

3. Research framework
Based on TCE, we build a framework for assessing the marketability of different types of products on the web. Since customers choose the channel that

Asset specificity

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43

33

Table 3
Statistics of survey data
Variables

Mean

Standard deviation Cronbach alpha Acceptance
Search cost
Comparison cost
Examination cost
Negotiation cost
Payment cost
Delivery cost
Post-service cost
Product uncertainty
Process uncertainty
Site specificity
Physical asset specificity
Human asset specificity
Brand name specificity
Temporal specificity

3.797
3.722
3.990
4.635
4.090
4.306
3.979
4.644
4.838
4.824
3.393
3.948
2.868
3.760
3.670

1.972
1.828
1.650
1.781
1.372
1.426
1.713
1.699
1.619
1.702
1.682
1.768
1.688
1.733
1.756

0.800
0.735
0.734
0.705
0.722
0.739
0.732
0.717
0.709
0.708
0.708
0.709
0.719
0.726
0.763

Ži.e., potential customers for electronic markets.. The data were then analyzed to evaluate the model.
4.1. Operationalization of constructs
The model has four constructs: customer acceptance, transaction cost, uncertainty, and asset specificity. Since it is difficult to obtain the actual cost data incurred in a transaction, the study focuses on the perceived information of the customer. This is appropriate because it is, in fact, the perceived information that leads to customer decisions Ždue to the bounded rationality.. The operationalization of the constructs is listed below.
Ž1. Acceptance of the electronic channel. The web is an alternative to the traditional channel. Different

customers may have different levels of acceptance for different products. A product with a higher customer acceptance level will have a higher likelihood of being successful when marketing on the web. In the study, we use a seven-point Likert-scale, ranging from absolutely no to absolutely yes, to assess the subject’s intention of purchasing a particular product electronically. Ž2. Transactional costs. Transaction cost is the cost involved in the transaction process. As we decompose a transaction process into seven stages
Žsee Table 1., the overall transaction cost is measured by the cost associated with each of these stages, as follows:
Ø Search cost: the perceived cost incurred at the stage of finding relevant product or process information, such as finding the source.

Table 4
Consumer acceptance of different products
Product

Overall

Experienced

Inexperienced

Mean
Book
Shoes
Toothpaste
Microwave oven
Flower

Standard deviation Mean

Standard deviation Mean

Standard deviation 5.576
2.381
2.670
3.294
5.600

1.169
1.231
1.809
1.667
1.516

6.107
2.370
2.750
3.179
5.222

0.832
1.182
1.974
1.588
1.450

5.316
2.386
2.632
3.351
4.982

1.227
1.264
1.739
1.716
1.552

Note: The center of the measurement scale is 4.0.

34

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43

Table 5
Difference between experienced and inexperienced customers
Sources

d.f.

MSE

F

P

Experience
Background
Interaction
Error

1
1
1
81

11.759
2.194
2.487
1.214

9.69
1.81
2.05

0.003
0.183
0.156

Ø Comparison cost: the perceived cost incurred at the stage of comparing alternatives based on their attributes Že.g., price. or comparing alternative channels. Ø Examination cost: the perceived cost incurred at the stage of examining the products to be purchased, such as browsing a book and trying shoes.
Ø Negotiation cost: the perceived cost incurred at the stage of negotiating terms with the seller.
Ø Payment cost: the perceived cost incurred at the process of ordering and paying for the product.
Ø Delivery cost: the perceived cost incurred at the process of receiving products, such as delay in receiving time or higher transportation costs.
Ø Post-service cost: the perceived cost incurred after receiving a product, such as maintenance and customer support.
Comparing to the traditional channel, the electronic one may have higher costs in some categories and lower costs in others. For instance, the electronic

channel may provide a search engine that lowers the search and comparison costs by helping customers find product sources and allowing customers to compare different prices, but increases the examination cost by not allowing the customer to physically examine a product.
Ž3. Uncertainty. In a transaction, there are two kinds of uncertainties. First, the received product may not meet the customer’s expectation at ordering.
This is called product uncertainty. Then, the customer may not have a complete confidence in the transaction process. This is called process uncertainty. The overall effect of uncertainty on the perceived transaction cost will be the aggregated effect of these two kinds of uncertainties.
Ž4. Asset specificity. Asset specificity refers to the durable investment to support particular transactions. In Ref. w26x, asset specificity is decomposed into several aspects, as follows: Ža. site specificity. A product may be more efficient to be traded in a

Table 6
Results from the confirmatory factor analysis
Measure specificity

Construct
Transaction cost

Search cost
Comparison cost
Examination cost
Negotiation cost
Payment cost
Delivery cost
Post-service cost
Product uncertainty
Process uncertainty
Site specificity
Special asset specificity
Human asset specificity
Brand name specificity
Temporal specificity

Uncertainty

Asset

0.63
1.18
0.75
0.41
0.47
0.39
0.37
0.27
–
–
–
–
–
–

y0.25 y0.84 –
–
y0.17
–
0.26
0.58
0.86
–
–
–
y0.24
–

– y0.02 –
0.15
–
–
–
–
–
0.75
0.68
0.63
0.79 y0.05 Note: the significance level was set at 0.05.

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43

Fig. 2. The empirical model.

particular location. Higher site specificity may favor the traditional channel. Žb. Physical asset specificity.
Transactions of certain products may need special physical equipment. A product with a higher physical asset specificity may favor the traditional channel. Žc. Human asset specificity. Transactions of certain products may need special human expertise.
A product with a higher human asset specificity may favor the traditional channel. Žd. Brand name specificity. Sometimes customers prefer buying a particular brand-name product, such as having dinner in a well-known restaurant. A product with a higher brand name specificity may favor the traditional channel.
Že. Temporal specificity. The timing of transactions may affect the transaction cost. Its exact effect varies.
In the research, asset specificity is measured based on these aspects. Table 2 summarizes the operationalization of the constructs.

35

sentatives in the questionnaire. They were book, shoes, toothpaste, microwave oven, and flower. These products were selected carefully to represent different types of products. For example, books are information-rich. Shoes have special needs of physical trial. Toothpaste is a convenient goods that most consumers buy without much thinking. Microwave ovens need maintenance. Buying flowers may have temporal considerations.
The research was conducted in Taiwan. Subjects were volunteers who had experience in using Internet. A total of eighty-six subjects was recruited.
Twenty-eight of them had actual web shopping experience, while the other fifty-seven did not. Thirtyseven had work experience ranging from one to twenty years. The other fourty-nine were undergraduate and graduate students.
The procedures for data collection were: Ž1. designing the questionnaire, Ž2. pre-testing and revising the questionnaire, Ž3. recruiting survey subjects, Ž3. giving instructions to the subject, Ž4. performing the survey, and Ž5. verification of the collected data. The
Cronbach alpha of the collected data is 0.873 for raw variables and 0.871 for standardized variables. The
Cronbach alpha for each item is also high, as listed in Table 3. This indicates that the instrument is reliable. 4.2. Data collection

5. Data analysis and results

Based on the operationalization of model constructs, a questionnaire of fifteen questions, as shown in Appendix A, was designed to collect data for testing the model. The questionnaire was tested and modified before using for data collection. Five products of different characteristics were chosen as repre-

The collected data were analyzed to see Ž1. whether products with different characteristics have different levels of customer acceptance in electronic commerce and Ž2. whether the transaction cost model in Fig. 1 is valid for explaining the consumer choice between the electronic and traditional channels.

Table 7
Coefficient matrix of the structural model
Transactional
Cost
Transactional cost
Uncertainty
Asset specificity
R2

Acceptance

–
0.40 Ž0.0001.
0.25 Ž0.0001.
0.32

y0.62 Ž0.0001.
0.11 Ž0.035.
0.09 Ž0.0631.
0.29

Note: the numbers in the parentheses are the significance levels.

36

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43
Table 8
Coefficient matrix of experienced shoppers
Transactional
Cost

Fig. 3. The experienced model.

Transactional cost
Uncertainty
Asset specificity
R2

Acceptance

–
0.61 Ž0.0001.
0.04 Ž0.6237.
0.38

y0.72 Ž0.0001.
0.25 Ž0.0065. y0.004 Ž0.9520.
0.36

Note: the numbers in the parentheses are the significance levels.

Structural equation modeling is used to test the model. Structural equation modeling is an approach to assessing a model that involves multiple constructs with multiple observation items. It evaluates the structural component and the corresponding measurement component simultaneously. The approach is a powerful second generation multivariate analysis technique for studying causal models. It is superior to traditional regression and factor analysis because the measurement model is assessed within the context of the theoretical structural model w7x. In this research, LISREL 8 was used to analyze data.
5.1. Difference in customer acceptance
A simple statistical analysis shows that Ž1. overall, customers prefer traditional markets over the web. The average customer acceptance level is 3.797 in Table 3, which is less than 4.0 Ž t s 2.15, p 0.05., and Ž2. different products do have different levels of acceptance for selling on the web. The results listed in Table 4 show that books and flowers are more acceptable to consumers Ži.e., means are higher than the indifference level of 4.0, p - 0.01., whereas shoes, toothpastes, and microwave ovens are less acceptable Ži.e., means are smaller than the indifference level of 4.0, p - 0.01.. The One-way
ANOVA results in an F-value of 77.94 Ž p - 0.0001..
If we decompose the data into two groups: experienced web shoppers and inexperienced shoppers, then the experienced shoppers show a sharper discrepancy in terms of what they would buy over the web. The means of experienced shoppers are 6.107 for books and 5.222 for flowers, compared to 2.370,
2.750, and 3.179 for shoes, toothpaste, and microwave ovens. For inexperienced shoppers, the means are 5.316, 4.982, 2.386, 2.632, and 3.351, respectively. A further examination of the subjects having experience in buying books, we find that a statistically significant difference exists between those who have prior purchasing experience and those who do not have Žsee Table 5..
5.2. Testing the measurement model
Our measurement model consists of 14 items used to measure four constructs Žsee Table 2.. We need to test the relationship between the constructs and their respective measurement items. We used the confirmatory factor analysis to assess their discriminant validity. Discriminant validity is the degree to which items differentiate between constructs, or measure different constructs. A simple test of discriminant validity is to verify that each item loads more highly on its associated construct than on any other construct w23x. The resulting output from LISREL under the significance level of p s 0.05 is shown in Table
6. All factors except the temporal specificity are significant statistically. All factors have higher significant loads on their associated constructs than on any other construct. The fit between the data and the theoretical model is also acceptable Ž x 2 s 55.94, p s 0.29.. Therefore, we conclude that the measurement model discriminated adequately between the constructs. Fig. 4. The inexperienced model.

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43
Table 9
Coefficient matrix of inexperienced shoppers
Transactional
Cost
Transactional cost
Uncertainty
Asset specificity
R2

Acceptance

–
0.34 Ž0.0001.
0.29 Ž0.0001.
0.30

y0.64 Ž0.0001.
0.04 Ž0.5419.
0.28 Ž0.0001.
0.32

Note: the numbers in the parentheses are the significance levels.

5.3. Testing the model
Following the confirmation of the measurement model, we proceeded to examine the structural model.
The evaluation consisted of the evaluation of the size, sign, and significance of the standardized path coefficients. Fig. 2 shows the results of the structural model. Table 7 shows the related coefficients. All paths except the linkage between asset specificity and acceptance are significant at least at the 5% level. The linkage between asset specificity and acceptance is also very close to significance Ž p s
0.063.. Thirty-two percent of the variance in transaction costs and 29% of the variance in consumer acceptance are accounted for by the model. The percentages of variance explained by the model are greater than 10%, which implies a satisfactory and significant model w6x.
The result allows us to accept our two propositions. In order to examine whether the experience and inexperience shoppers in electronic commerce may have different considerations, we ran two submodels. The results for experienced shoppers are shown in Fig. 3 and Table 8, whereas those for inexperienced shoppers are shown in Fig. 4 and
Table 9.
It can be seen from the models that uncertainty is the most significant construct for experienced shoppers, whereas asset specificity is the most significant for inexperienced shoppers. The difference may be due to a learning effect to be discussed later.

6. Discussion and conclusions
The above results show some interesting findings.
First, they conform our initial argument that some

37

products are more suitable for marketing on the web than others. In our study, books and flowers are more likely to be ordered by consumers than shoes, toothpaste, and microwave ovens. This is primarily due to the difference in their perceived transaction costs. The higher the perceived transaction costs
Žcompared to those of the traditional channel., the less likely a product will be purchased electronically.
In general, the electronic commerce lowers the search cost but raises the examination, payment, and post-service costs. Table 10 shows that books take most advantage of the web by reducing the search cost, whereas shoes have troubles when customers would like to examine products. Therefore, products that need detail examination or trial before purchase
Žsuch as shoes. and post-sales services Žsuch as microwave ovens. are considered to be less appropriate for electronic markets. The payment method also, to some extent, bothered the customer.
The TCE model gives us guidelines regarding what products are more suitable for electronic markets. Does the result imply that we cannot sell shoes, toothpaste, or microwave ovens on the web? The answer is no. What the result indicates is that they are less acceptable to customers at least at their current forms. They may become marketable, however, if their packaging or other related attributes change. For instance, customers may not want to order toothpaste over the web when they are out of it early in the morning, but may order a dozen pack of toothpaste electronically once in a while and put in the closet for convenience if the price is good.
Similarly, buying microwave ovens over the web may become acceptable if convenient post-sales services are available Že.g., through contracts with local

Table 10
Transaction costs by types of transaction and product
Transaction Book Shoes Toothpaste Microwave Flower Mean cost oven
Search
Comparison
Examination
Negotiation
Payment
Delivery
Post-service

3.07
3.21
4.04
3.54
3.80
3.24
4.02

4.55
4.89
6.16
4.74
4.60
4.42
5.35

3.62
4.06
3.36
3.52
4.55
4.14
3.82

4.12
4.26
5.29
4.75
4.45
4.55
5.66

3.25
3.54
4.32
3.89
4.13
3.53
4.36

3.72
3.99
4.64
4.09
4.30
3.98
4.64

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43

38

service stores.. What our TCE model indicates is that, as long as the seller can find ways that effectively reduce the overall customer transaction cost on the electronic market, any product is marketable on the web.
Another issue is the learning effect in electronic shopping. The data in Tables 8 and 9 show that inexperienced and experienced consumers have different considerations while purchasing electronically.
Both asset specificity and uncertainty affect the decision process of inexperienced consumers, whereas only uncertainty affects the experienced consumers.
This may be because the customer gets used to the process once they have experience. The learning process eliminates the problem associated with asset specificity. This suggests that the customer acceptance of electronic shopping may be increased by giving the customer incentives at their first-time purchase. Finally, the transaction cost model is confirmed by the data. That is, the customer acceptance decision is affected by the transaction cost, whereas the transaction cost is affected by the uncertainty and asset specificity. In order to market a product successfully on the web, we should manage the product and process uncertainty, and the asset specificity involved in the electronic transaction.
The contribution of the research is two-fold. We have built and tested a transaction cost model for studying what products are more appropriate for marketing on the web and why they are more appropriate. The findings can be used to help select proper

products and modify product characteristics to make them more acceptable in the electronic market. For instance, refrigerators are similar to microwave ovens in the need for post-sales services. Based on the model, we can predict that refrigerators will not be popular on the web unless certain arrangement can be done to reduce the uncertainty and the need for post-services. Similarly, we can predict that, just like shoes, clothing would not sell well, unless certain arrangement is done to reduce the examination cost.
Although the findings have been encouraging and useful, the research is not without limitations. First, the transaction costs we measured were perceived costs. In other words, we did not actually measure the cost. Rather, we asked the subject to assess the costs. This subject assessment could lead to biases.
Second, the modeling approach adopted in the research was structural equations. Although the method is good in confirmatory study, it has some inherited limitations that may reduce the validity and reliability of the findings w15x. Finally, whether the findings are generalizable to all products and services on the web is unclear. Further research is necessary to verify their generalizability.

Acknowledgements
The research was supported in part by a grant to the first author from National Science Council of
ROC, under the contract of NSC87-2416-H-110-031N6.

Appendix A. The questionnaire

Web stores are electronic stores that run on the web. Please answer the following questions by checking the proper answer. For example, if you think it is very likely that you will buy bread on web stores, then check in the I within the column, ‘ very likely,’ as follows:

Buy bread

Absolutely not I

Very unlikely I

Unlikely
I

Hard to say I

Likely
I

Very likely I

Absolutely yes I

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43

39

1. Please indicate the likelihood that you would buy the following goods from web stores:

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely not I
I
I
I
I

Very unlikely I
I
I
I
I

Unlikely
I
I
I
I
I

Hard to say I
I
I
I
I

Likely
I
I
I
I
I

Very likely I
I
I
I
I

Absolutely yes I
I
I
I
I

2. The first step for buying merchandise is often to collect information such as where to buy, prices, and others’ comments. Compared with buying in traditional stores, what is your assessment of the time and effort spent in searching relevant information when buying the following items from web stores?

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low
I
I
I
I
I

About the same I
I
I
I
I

High
I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

3. After collecting information, we often want to evaluate products based on various attributes such as prices and quality. Compared with buying in traditional stores, what is your assessment of the convenience of product evaluation when buying each of the following items from web stores?

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low
I
I
I
I
I

About the same I
I
I
I
I

High
I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

4. Sometimes, buyers may want to examine the product. The web stores usually do not allow potential buyers to physically examine the product. Compared with traditional stores, what is your assessment of the trouble due to the lack of physical examination of products when buying the following items from web stores?

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low
I
I
I
I
I

About the same I
I
I
I
I

High
I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

40

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43

5. It is usually not allowed to negotiate terms on web shopping. Compared with traditional stores, what is your assessment of the trouble due to the lack of bargaining support when buying the following items from web stores? Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low
I
I
I
I
I

About the same I
I
I
I
I

High
I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

6. Web shopping requires that the order be placed on the web and prices be paid by credit card or money orders.
Compared with traditional stores, what is your assessment of the trouble due to placing orders and paying on the web, when buying the following items from web stores?

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low
I
I
I
I
I

About the same I
I
I
I
I

High
I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

7. Web stores usually deliver the merchandise you ordered by mail or other means, which is different from traditional stores where you pick up what you buy immediately after payment. Compared with traditional stores, what is your assessment of the trouble due to delivery of products, when buying the following items from web stores? Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low
I
I
I
I
I

About the same I
I
I
I
I

High
I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

8. After receiving the merchandise, it may need to be returned or some other post-purchase services. Compared with traditional stores, what is your assessment of the trouble in post-purchase services, when buying the following items from web stores?

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low
I
I
I
I
I

About the same I
I
I
I
I

High
I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43

41

9. Comparing to traditional stores, what degree of uncertainty may be involved in the process of purchasing the following items from web stores?

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low
I
I
I
I
I

About the same I
I
I
I
I

High
I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

10. Compared with traditional stores, what degree of product uncertainty Ži.e., the product you receive may not be exactly what you want. may be involved when purchasing the following items from web stores?

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low
I
I
I
I
I

About the same I
I
I
I
I

High
I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

11. To what extent do you insist on purchasing the following items from a particular store?

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low

Medium

High

I
I
I
I
I

I
I
I
I
I

I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

12. To what extent do you insist on purchasing the following items from a particular store due to its special asset Žsuch as location, special equipment.?

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low

Medium

High

I
I
I
I
I

I
I
I
I
I

I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

42

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43

13. To what extent do you insist on purchasing the following items from a particular store due to particular persons working there?

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low

Medium

High

I
I
I
I
I

I
I
I
I
I

I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

14. To what extent do you insist on particular brand names when purchasing the following items?

Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low

Medium

High

I
I
I
I
I

I
I
I
I
I

I
I
I
I
I

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

15. To what extend do you insist on purchasing the following items due to a particular chance or other temporal considerations? Books
Shoes
Toothpaste
Microwave oven
Flowers

Absolutely low I
I
I
I
I

Very low I
I
I
I
I

Low

Medium

High

I
I
I
I
I

I
I
I
I
I

I
I
I
I
I

References w1x A. Bhimani, Securing the commercial Internet, Communications of the ACM 39 Ž6. Ž1996. 29–35. w2x L.A. Chen, Current Status of Applications and Prospect of
Internet in 1997, http:rrwww.psd.iii.org.twrretro.htm, 1997. w3x R.H. Coase, The nature of the firm, Economica 4 Ž1937.
386–405.
w4x D.E. Denning, Internatioal encryption policy, in: R. Kalakota,
A.B. Whinston ŽEds.., Readings in Electronic Commerce,
Readings, Addison-Wesley, MA, 1997, pp. 105–118. w5x J.F. Engel, D.T. Kollat, R.D. Blackwell, Consumer Behavior,
3rd edn., Holt, Rinehart and Winston, New York, 1978.

Very high I
I
I
I
I

Absolutely high I
I
I
I
I

w6x R.F. Falk, N.B. Miller, A Primer for Soft Modeling, The
University of Akron Press, Akron, OH, 1992. w7x C. Fornell ŽEd.., A Second Generation of Multivariate Analysis Methods, Vol. 1, Praeger, New York, 1982. w8x A. Gupta et al., Economic issues in electronic commerce, in:
R. Kalakota, A.B. Whinston ŽEds.., Readings in Electronic
Commerce, Readings, Addison-Wesley, MA, 1997, pp. 197–
227.
w9x A. Gupta, D.O. Stahl, A.B. Whinston, An economic approach to network computing with priority classes, Journal of
Organizational Computing and Electronic Commerce 6 Ž1996.
71–95.
w10x S. Hamilton, E-commerce for the 21st century, Computer 30
Ž5. Ž1997. 44–47.

T.-P. Liang, J.-S. Huangr Decision Support Systems 24 (1998) 29–43 w11x D.L. Hoffman et al., Commercial scenarios for the web:
Opportunities and challenges, in: R. Kalakota, A.B. Whinston ŽEds.., Readings in Electronic Commerce, Readings,
Addison-Wesley, MA, 1997, pp. 29–53. w12x B. Ives, G.P. Learmonth, The information system as a competitive weapon, Communications of the ACM 27 Ž12. Ž1984.
1193–1201.
w13x R. Kalakota, A.B. Whinston, Frontiers of Electronic Commerce, Readings, Addison-Wesley, MA, 1996. w14x A. Kambil, Doing business in the wired world, Computer 30
Ž5. Ž1997. 56–61. w15x B. Lee, A. Barua, A.B. Whinston, Discovery and representation of causal relationships in MIS research: a methodological framework, MIS Quarterly 21 Ž1. Ž1997. 109–136. w16x T.P. Liang, H.J. Lai, Development of electronic shopping in
Taiwan, Proceedings of the APEC Conference on Computer
Literacy, Tokyo, Japan, 1997. w17x C. Liu, K.P. Arnett, L.M. Capella, R.C. Beatty, Web sites of the fortune 500 companies: facing customers through home pages, Information and Management 31 Ž1997. 335–345. w18x B.C. Neuman, A flexible framework for network payment, in: R. Kalakota, A.B. Whinston ŽEds.., Readings in Electronic Commerce, Readings, Addison-Wesley, MA, 1997, pp. 229–243. w19x P. Panurach, Money in electronic commerce: digital cash, electronic fund transfer, and Ecash, Communications of the
ACM 39 Ž6. Ž1996. 45–50. w20x A. Rindfleisch, J.B. Heide, Transaction cost analysis: past, present, and future applications, Journal of Marketing 61 Ž4.
Ž1997. 30–54. w21x H.A. Shelanski, P.G. Klein, Empirical research in transaction cost economics: A review and assessment, Journal of Law,
Economics, and Organization 11 Ž2. Ž1995. 335–361. w22x J.M. Tenenbaum et al., Eco system: an internet commerce architecture, Computer 30 Ž5. Ž1997. 48–55.

43

w23x R.L. Thompson, C.A. Higgins, J.M. Howell, Personal computing: toward a conceptual model of utilization, MIS Quarterly 15 Ž1. Ž1991. 125–143. w24x O.E. Williamson, Markets and Hierarchies: Analysis and
Antitrust Implications, Free Press, New York, 1975. w25x O.E. Williamson, Transaction-cost economics: the governance of contractual relations, Journal of Law and Economics 22 Ž2. Ž1979. 233–261. w26x O.E. Williamson, The economics of organization: the transaction cost approach, American Journal of Sociology 87 Ž3.
Ž1981. 548–577. w27x O.E. Williamson, The Economic Institutions of Capitalism,
Free Press, New York, NY, 1985.
Ting-Peng Liang is Professor of information system at the Department of Information Management, National Sun Yat-sen University in Kaohsiung, Taiwan. He was Dean of the College of
Management and Director of the Institute of Information Management of the University. Prior to joining the university, he had been on the faculties of Purdue University and University of Illinois at
Urbana–Champaign. Professor Liang has served on the editorial boards of eight professional journals and program committees of many international conferences. His papers have appeared in journals such as Management Science, Operations Research, Decision Support Systems, MIS Quarterly, Journal of MIS, Decision
Sciences, IEEE Computer, among others. His research interests include electronic commerce, decision support systems, and intelligent systems.
Jin-Shiang Huang is Assistant Professor of Information Management at the Ming-chuan University ŽTaiwan, ROC.. He received his Doctoral and master degrees in MIS for National Sun Yat-sen university and his Bachelor in MIS from Tamkang University
ŽTaiwan, ROC.. His current research interests include economic issues in electronic commerce, group decision support systems, and workflow management.