competition in an internet mall: a strategic analysis of a ... · competition in an internet mall:...

52
Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He * Washington University in St. Louis Preliminary draft, comments welcome September, 2001 * I wish to thank Ambar Rao, Chakravarthi Narasimhan, Amit Pazgal, and Erik Durbin for their many valuable comments.

Upload: others

Post on 25-May-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Competition In An Internet Mall: A Strategic

Analysis of A New Marketing Venue

Chuan He∗

Washington University in St. Louis

Preliminary draft, comments welcome

September, 2001

∗I wish to thank Ambar Rao, Chakravarthi Narasimhan, Amit Pazgal, and Erik Durbinfor their many valuable comments.

Page 2: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Abstract

Retailers strive to differentiate themselves from competitors to avoid commoditizationand consequent price competition, using tools such as store location, store layout, andproduct assortment. However, such tools become ineffective on the Internet, wherecompetitors are a few clicks away, where web page design can be easily imitated, andwhere online shoppers can browse a variety of product categories at ease, making pricecomparisons using search engines. Thus, an integrated online shopping environmentsuch as that provided by an Internet mall would seem to be particularly unattractiveto retailers. Yet, Internet malls have emerged as an important electronic shoppingenvironment that, according to a recent Forrester Research report, may evolve into adominant force in Internet marketing. The purpose of this essay is to provide an expla-nation for this puzzle. We show that an Internet mall reduces price competition amongonline stores by leveraging two notable features: the search engine and the featuredstore. The search engine facilitates price comparisons but requires consumers to spendtime going through pages of search results. In contrast, consumers are directed to thefeatured store with a single click. Consumers trade off price reductions with reduc-tions in search time, leading to higher prices in the featured stores and a consequentsoftening of price competition.

We investigate a homogeneous product market with heterogeneous consumers. Weassume that consumers differ in their sensitivity to price and to the cost of time.Search costs are the product of a category specific search time and unit time cost. Tochoose between search engine driven comparison shopping and featured store shopping,consumers form price expectations that help them compare the reduction in priceobtained by searching with the cost of the search itself. We find that competitionis reduced. In addition, we find that as search costs increase, the price differentialbetween the featured stores and non-featured stores becomes larger, as is also the casewhen more shoppers purchase through the mall. We also find that a mall should featurestores for free when more shoppers purchase through it and when consumer search costsare substantial. Conversely, it should charge fees to featured stores when fewer peoplepurchase there and search costs are low. Finally, while a large mall should chargepercentage fees to participating stores, a small Internet mall may optimally offer freeparticipation to online stores. Preliminary empirical analysis supports our conclusions.

Keywords: E-Business, business models and strategies, changing customer expecta-tions, effective e-tailing approaches, game theory.

Page 3: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

1 Introduction

Marketers usually strive to differentiate themselves from competitors to avoid commodi-

tization and consequent price competition. Retailers are no exception: store location,

store layout, and product assortment are some of the tools they use to create differen-

tiation. However, such tools become ineffective on the Internet. Competitors are just

a few clicks away, even the best web page design can be easily imitated, and online

shoppers can browse a variety of product categories at ease. And, customers can com-

parison shop using search engines. Thus, an integrated online shopping environment

such as that provided by an Internet mall would seem to be particularly unattractive

to retailers. Yet, Internet malls have emerged as an important electronic shopping

environment that may evolve into a dominant force in Internet marketing. According

to a Forrester Research report (March, 2001), Internet malls will drive almost half of

online retail sales by 2005.1 The purpose of this paper is to analyze this puzzle: we

investigate whether Internet malls can improve the profit potential of online stores and

how such profits (if any) can be achieved.

Internet malls are subsidiaries of major Internet portals, which attract millions of vis-

itors through news, games, and many other forms of entertainment. Yahoo! Shopping,

eShop at msn.com, or Shop@AOL at aol.com are some of the leaders. Similar to con-

ventional malls, Internet malls have numerous affiliated stores, thrive on their ability

to attract consumers, and derive their revenue streams by collecting fees from their

member stores. However, Internet malls differ from conventional malls in at least two

major ways. First, Internet malls have an open structure, i.e., even when an online

store is affiliated with an Internet mall, consumers can still access that store directly.

For example, the popular electronic store JandR is a member of Yahoo! Shopping. A

1It is predicted that US online retail sales will reach $104 billion in 2005.

1

Page 4: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

consumer who wants to shop at JandR can find it at the Electronics section of Yahoo!

Shopping. Alternatively, this consumer can simply go to www.jandr.com. If a brick and

mortar store is located in a conventional mall, then consumers have to go to the mall

to shop at that store. By contrast, online stores have virtual locations and thus can be

accessed directly, regardless of whether they are affiliated with an Internet mall or not.

Second, Internet malls distinguish themselves by the presence of search engines and

featured stores. The search engine allows a customer to compare prices for a product

across stores more easily than in a conventional mall, where a consumer needs to have

a sharp memory and endure several shopping trips in order to comparison shop. A

featured store has its logo embedded in a large icon that is placed in a prominent spot

on the Internet mall’s web page.2 With a single click, a consumer is directed to the

featured store. Note that when using the search engine, consumers take the initiative

in determining where they shop, whereas they are guided to the featured store by the

Internet mall. We contend that the search engine and featured stores lead to consumer

shopping behavior that is significantly different from that in conventional malls.

In this paper, we consider an Internet mall with on-line stores offering a homogeneous

product. The mall offers a menu of contracts to stores to join and to be featured.

If and when a store decides to join the mall, it decides whether to be featured or

not. It then formulates its pricing strategy based on the fee structure of the Internet

mall, on whether it is featured or not, and on consumer segmentation. Consumers are

heterogeneous, varying in their sensitivity to price and to search costs, which are the

product of a category specific search time and an individual’s unit time cost. Within

this environment, we study the following questions:

2An Internet mall’s web pages consist of shopping main page, which lists a number of productcategories, and a page for each category. The featured store icon is placed on the category specificpage the store is associated with.

2

Page 5: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

• How does the presence of search engines and featured stores affect price compe-

tition between online stores?

• If customers become sophisticated and their decision to shop at the featured

store or to use the search engine and comparison shop is endogenous, rather than

exogenously specified, what happens to the profitability of the stores and the fee

structure of Internet malls?

• When are online stores better off joining Internet malls? How does an owner of

an Internet mall determine the optimal fee structure?

We demonstrate that the presence of featured stores mitigates competition that might

otherwise be accentuated by the Internet mall environment. A featured store, because

it is easily reached, delivers extra convenience to shoppers and can therefore charge a

higher price on average. Customers can trade off this cost of convenience with the search

costs they would incur if they use the search engine to find the lowest price store. Those

who are more price sensitive and less sensitive to search costs engage in comparison

shopping, while those who are less price sensitive and more sensitive to search time opt

for the convenience of shopping at the featured store. We first examine the case where

customer segmentation is exogenously specified and it is known which customers shop

at the featured store and which ones comparison shop using the search engine. Next, we

endogenize the segmentation. We consider more sophisticated customers who make the

choice between shopping at the featured store and using the search engine to comparison

shop by forming price expectations that help them to compare the gain from search

with search costs. In both cases, we find that the featured store charges a higher price

on average than the non-featured store, thus mitigating competition. Thus Internet

malls, serving as an intermediary between online stores and consumers, improve the

profit potential of online stores. From the perspective of the mall, we find that while a

3

Page 6: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

large Internet mall charges fees that are a percentage of sales to participating stores, a

small Internet mall may optimally offer free participation. In terms of pricing, as more

shoppers purchase through an Internet mall, the price at the featured store becomes

higher, and both online stores and the Internet mall become more profitable. And, as

search costs increase, the price differential between the featured stores and non-featured

stores become larger. Preliminary empirical analysis supports these analytical results.

The rest of the paper is organized as follows: section 1.1 provides an overview of related

research, section 2 presents our model. Section 3 provides a summary and concludes.

1.1 Literature Review

Two streams of research are relevant to our study, one concerns why stores collocate,

and the other is the emerging literature on Internet institutions. Dudey (1990) argues

that firms collocate to attract more consumers by facilitating price comparisons, but

clustering increases the intensity of local competition. He shows that in the presence

of positive search costs, there exist conditions under which firms agglomerate in equi-

librium. Wernerfelt (1994) contends that when buyers incur evaluation costs for search

goods, they may refrain from incurring them for fear of later opportunism on the part

of sellers since these costs are sunk at the time of transaction. He demonstrates that

seller collocation can alleviate this problem. Messinger and Narasimhan (1997) observe

the proliferation of product assortments in grocery stores and supermarkets and at-

tribute the growth in one-stop shopping to consumer’s economizing on shopping time.

They offer theoretical and empirical evidence in support of this conjecture. Using a

Hotelling framework with vertical differentiation, Fischer and Harrington (1996) show

that firms selling similar products tend to collocate when 1. products exhibit suffi-

cient heterogeneity, 2. consumer search costs are positive, and 3. there is free entry.

4

Page 7: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Under a similar framework, Iyer (2001) shows that when travel costs are low, even

firms selling homogeneous products (gas stations) may choose to cluster and that leads

to quality differentiation. In sum, the extant literature offers many insights on why

firms collocate. However, collocation on the Internet has distinct features due to its

open structure (Recall the JandR example). In the presence of general purpose search

engines, online stores compete with each other regardless of whether they are in or out-

side of Internet malls. Therefore, additional explanations are required to understand

why online stores join Internet malls.

Internet and e-commerce have aroused considerable enthusiasm among marketing aca-

demics. Bakos (1997) recognizes the role of the Internet in reducing search costs. He

argues that the Internet provides consumers easier access to both price information

and product information. Assuming that consumers are homogeneous in their ability

to obtain information, he shows that electronic marketplaces will lead to greater al-

location efficiency at the expense of seller profits. Lynch and Ariely (2000) conduct

an experimental study to explore the implications of reduced search costs for Internet

marketing. They find that reduced search costs for price information increases price

sensitivity, whereas lowering the cost of search for quality information reduces price sen-

sitivity. Lal and Sarvary (1999) investigate the conditions under which Internet may

soften price competition. They categorize product attributes into digital attributes

and nondigital attributes. While digital attributes are amenable to the Internet and

can be evaluated easily online, nondigital attributes require physical inspection. This

classification is similar to the distinction between search goods and experience goods.

They demonstrate that given consumers’ favorable prior evaluation of nondigital at-

tributes, the low cost of evaluating digital attributes on the Internet enhances loyalty

and hence increases seller profits. In a conceptual paper, Alba et al. (1997) exam-

ine the impact of e-commerce on the entire marketing channel. They point out that

5

Page 8: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

online shopping reduces the importance of location and may replace or complement

traditional retail formats. Balasubramanian (1998) models competition in a multiple

channel environment where consumers may purchase from traditional retailers and di-

rect sellers, including catalog and Internet marketers. Assuming that consumers have

complete information about product and price information in all channels, he identifies

the conditions under which traditional stores may accommodate the entry of direct

sellers. Additionally, he shows that the level of information disseminated by direct sell-

ers have strategic implications. Specifically, high market coverage may depress profits.

Iyer and Pazgal (2001) explore the strategic implications of Internet shopping agents

(ISA), they show that when the reach of the institution is endogenous and when the

traffic at the ISA confers complementary side-benefits, there exists a unique number of

retailers who will join the institution. Furthermore, they find that the ISA will have

the incentive to share the side benefits with the inside retailers. Various other aspects

of Internet and e-commerce have been addressed in recent research. Novak, Hoffman

and Yung (2000) offer an approach to measure customer experience in an online en-

vironment, Hauble and Trifts (2000) study the effects of interactive decision aids in a

controlled experiment, Bradlow and Schmittlein (2000) model the performance of In-

ternet search engines, Bakos and Brynjolfsson (2000) provide insights to the bundling

of and competition among information goods.

This research adds to the extant literature in that it shows that an Internet institution

can generate a positive externality in a commodity market.3 In particular, an Internet

mall softens price competition between online stores and as the same time offers extra

convenience to consumers. These objectives are accomplished by leveraging two notable

features of Internet malls—the search engine and featured stores. Since consumers differ

3Although products are differentiated, online retailers’ product assortments have significant over-laps. In many cases, their product bundles can be considered roughly homogeneous.

6

Page 9: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

in their ability and willingness to search, the search engine facilitates price comparisons

for those who are price sensitive, while featured stores provide one-stop shopping to

those who are price insensitive.

2 The Model

Our model analyzes the strategic behavior of competing stores in an Internet mall,

henceforth referred to as the Emall. To capture the essential features while keeping the

analysis tractable, we investigate the interactions among the Emall and two competing

stores in a representative category. It is assumed that the Emall stores offer identical

products, with the only differentiating dimension being their pricing policies. We

assume that all consumers buy one unit of product in any given period and have a

common reservation price r. Without loss of generality, we normalize the costs of the

online stores, such as payment to wholesalers/manufacturers and selling costs, to be

zero.

Consumers are divided into two broad groups: those who shop at the Emall, and those

who don’t. The Emall shoppers have two options; they can either shop at the fea-

tured stores or shop using the Emall search engine. To make a shopping decision, an

Emall shopper compares the expected price difference between the featured and non-

featured stores with the expected time and cost he has to invest in searching. If he

finds that the benefit from search is greater than his search cost, he will shop using the

search engine, otherwise, he will shop at the featured store. The non-Emall shoppers

are either loyal customers or price shoppers. The loyal customers always shop at their

favorite stores. In contrast, the price shoppers are deal prone, they are either unaware

of the Emall or prefer to use shopping bots or general-purpose search engines. The

patterns of consumer behavior discussed above imply four distinct consumer segments:

7

Page 10: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

non-Emall price shoppers (A segment), Emall comparison shoppers (B segment), fea-

tured store shoppers (C segment), and loyal customers (D segment). Normalizing the

total consumers to a unit mass, we have A + B + C +∑

i αi = 1, where∑

i αi is the

size of the D segment, the subscript i is a store index, i = 1, 2.

Joining the Emall is not without cost, each store needs to pay a portion of its rev-

enues generated through the Emall as royalties. Consider the JandR example we used

earlier. If a consumer purchases a VCR from JandR by going to www.jandr.com, this

consumer is a non-Yahoo! shopper, and JandR does not pay anything to Yahoo! for

this consumer’s purchase. Alternatively, if the consumer goes to Yahoo! ’s Electronics

section and then buys from JandR using the links provided, JandR has to pay a portion

of such revenues to Yahoo!.

We assume that the Emall, the stores, and consumers are all utility maximizers with

rational expectations. The sequence of actions is as follows (see figure 1). First, the

Emall announces its fee structure for the featured and non-featured stores. Second,

the online stores simultaneously decide whether to join the Emall or not and whether

they want to be featured or not; they then simultaneously formulate their pricing

policies. Third, correctly anticipating the stores’ pricing strategies, consumers make

their purchase decision. In particular, Emall shoppers decide whether to use the search

engine or shop at the featured store.4 We adopt subgame perfect equilibrium as the

solution concept throughout the paper. To facilitate a smooth flow of our paper, we

have kept the mathematical details at a minimum. For technical notes and proofs,

readers are referred to Appendices A and B.

4A search engine facilitates price comparisons by looking up prices of the same item and similaritems from different stores but often requires a consumer to patiently go through pages of searchresults. By contrast, a consumer can make a purchase at a featured store with a few clicks.

8

Page 11: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

The Emall

Online Stores

EmallShoppers

fee structure(two-part tariff)

join or not, whether want to be featuredor not

price setting

in-inin-outout-out

featured store shopping

search

outcome

featured store

non-featured store

Figure 1: The decision process

9

Page 12: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

2.1 Base Model

The competition between the two stores is contingent upon their decisions in two stages.

In the first stage, the stores decide whether to join the Emall or not. In the second

stage, they determine their pricing policies. We begin our analysis from the second

stage.

2.1.1 The Setting

Assume that the two stores are symmetric with equal loyal segment of size α, con-

sumer segments are fixed and their sizes are public knowledge, and the Emall charges

a two-part tariff to its member stores.5 We begin our analysis assuming that the

Emall charges fixed fees. Later in this section we will consider percentage fees and

the implications of the two alternative fee structures. In formulating its pricing deci-

sion, each store faces three possible scenarios: 1. both stores are Emall stores (in-in),

2. one store joins the Emall, the other stays out (in-out), and 3. both stores are

non-Emall stores (out-out). We discuss each of these scenarios in turn.

2.1.2 Results

in-in Without loss of generality, we assume that store 1 is the featured store.6 The

fees that the stores pay to the Emall is sunk cost, therefore, it is not included in the

equilibrium analysis. Under this setting, it can be established that only mixed strategy

equilibrium exists.7 Let the price support of the two stores be [p, r], and Fi(p) the price

5Among a variety of linear and non-linear pricing schemes, we assume that the Emall chooses thesimplest schemes that guarantees full extraction of economic rents—two-part tariff.

6We will show later that featuring both stores is not optimal for the Emall. In equilibrium theEmall will charge a fee such that the featured store and the non-featured store expect the same levelof profits. Thus, the two stores are indifferent to being featured or not. If there is a tie, that is, bothstores want to be the featured store, the Emall randomly features one of the stores.

7Rigorous proofs are developed in Varian (1980) and Narasimhan (1988). More details are givenin Appendix A.

10

Page 13: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

distribution function for store i. In a mixed strategy equilibrium, store i is indifferent

to charging any price in its price support. If the featured store charges the reservation

price r, it will be patronized by its loyal customers and featured store shoppers, the

resulting profit is (α+C)r. If the featured store charges a lower price than that of its

competitor, it will get the business from comparison shoppers in addition to its loyal

customers and featured store shoppers, the resulting profit is (α+A+B +C)p. Since

the featured store can make at least (α + C)r by charging the reservation price r, the

lower bound of the price support is given by

p =(α + C)r

α + A+B + C(1)

Note that store 1 always gets the business from featured store shoppers (C segment),

whereas store 2 never gets their business. On the other hand, the store that charges

a lower price will be patronized by comparison shoppers (A and B segments). The

expected profits for store 1 and store 2 are π1 = (α + C)r and π2 = p(α + A + B),

respectively. Given the setup specified above, the two stores’ equilibrium strategies are

characterized by the following price distribution functions:

F1(p) =

0, p <(α+C)r

α+A+B+C,

(α+A+B)[(α+A+B+C)p−(α+C)r](α+A+B+C)(A+B)p

,(α+C)r

α+A+B+C≤ p < r,

1, p ≥ r.

(2)

and

F2(p) =

0, p <(α+C)r

α+A+B+C,

(α+A+B+C)p−(α+C)r(A+B)p

,(α+C)r

α+A+B+C≤ p ≤ r,

1, p ≥ r.

(3)

Proposition 1. F1(p) first order stochastically dominates F2(p), that is, the featured

store charges a higher price on average. The expected price at the featured store and

11

Page 14: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Emall profit are increasing in the number of Emall shoppers.

Proof: All proofs are in Appendix B.

The competitive environment is moderated by the size of the loyal segment: the

larger the loyal segment, the less intense is the competition between the stores. In

addition, the degree of competitiveness is also affected by the segmentation among

Emall shoppers, which is driven by the features of the Emall—search engine and fea-

tured store. The stores compete more vigorously when more shoppers engage in search

and less so when more shoppers shop at the featured store.

in-out Assume without loss of generality that store 1 is an Emall store but store 2

is not. In this situation, store 1 is the featured store by default. The expected profits

before fees to the Emall for store 1 and store 2 are π1′ = (α+C)r and π2′ = p(α+A+B),

respectively, where p = (α+C)rα+A+B+C

. Note that when no switchers shop at the Emall,8

i.e., (B+C)→ 0, the in-out scenario converges to the situation analyzed in Narasimhan

(1988).

out-out When both stores are non-members, their expected profits are π = αr. Be-

side their loyal customers, the two stores vie for the business of comparison shoppers.9

The payoffs for the two stores in the three alternative scenarios are summarized in

table 1. Note that all payoffs are gross profits before payment of fees to the Emall , if

any.

8Of course, if that is the case, the Emall will face extinction.

9When both stores are outside of the Emall, Emall comparison shoppers (B segment) and featuredstore shoppers (C segment) are not defined. In that case, consumers in B segment use all means tolook for the lowest price, while consumers in C segment may drop out, search, or become loyal. Weassume that all consumers in C segment search or drop out in this case. The two stores’ customerbase consists of their loyal customers and comparison shoppers of size A+B+C. Comparative staticsshow that our results hold when some consumers in C segment become loyal. If all consumers in Csegment become loyal, then no online store should ever join the Emall, which contradicts reality.

12

Page 15: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Table 1: Payoffs for in-in, in-out, and out-out under fixed fees

in out

in , ,

out , α r , α r

Store 2

Store 1

)()(

BACBA

rC +++++

+ αα

α

)()(

BACBA

rC +++++

+ αα

αrC )( +α )(

)(BA

CBArC +++++

+ αα

α

rC )( +α

rC )( +α

Featuring one store vs both stores We have assumed earlier that the Emall features

one store only, will the Emall be better off featuring both stores? To answer this ques-

tion we need to look at the Emall’s revenue base (the largest possible revenue upon

which the Emall may collect fees) R under the two scenarios. The revenue base for

the Emall when one store is featured is simply the sum of the equilibrium revenues of

the two stores, R1 = (α + C)r + p(α + A + B) = (α + C)r + (α+C)rα+A+B+C

(α + A + B).

Since the two stores are identical, featured store shoppers are equally likely to shop at

either store when both stores are featured. Thus, the revenue base for the Emall when

both stores are featured is R2 = 2[(α + C2)r]. Since R1 > R2, the Emall will feature

one store only.

Percentage fee Rather than charging fixed fees, the Emall may charge percentage

fees to its member stores. It is not intuitively clear which fee structure yields higher

profit to the Emall and we need to identify the conditions under which one fee struc-

ture dominates the other. Assume that the percentages of revenue the Emall collects

from the featured store (store 1) and the non-featured store (store 2) are d1 and d2,

respectively. It should be noted that the percentage fee only applies to the purchases

driven by Emall traffic. In the in-out scenario, the Emall can only impose a percentage

fee of d on the store that chooses to join. The payoffs for the two stores in the in-in,

in-out, and out-out scenarios are summarized in table 2.

13

Page 16: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Table 2: Payoffs for in-in, in-out, and out-out under percentage fees

in out

in , ,

out , α r , α r

Store 2

Store 1

)()1) ((

) ]1([BA

dCBArdC ++−+++

−+ αα

α

)()1) ((

) ]1([BA

dCBArdC ++−+++

−+ αα

α rdC )]1([ −+α

rdC )]1([ 1−+α rdC )]1([ −+α)]1([)1)((

)]1([2

1

1dBA

dCBArdC −++−+++

−+ αα

α

Optimal fee structure of the Emall With the two stores’ pricing decisions in

mind, we now examine their decisions regarding whether to join the Emall or not in

the first stage. Although each store is better off joining the Emall individually, it is

not necessarily so when both stores join simultaneously. The intuition is as follows.

When at least one store joins the Emall, some consumers10 forego search in favor

of the featured store. The featured store shoppers are willing to pay more for the

convenience11 and as a result, the featured store is less inclined to attract comparison

shoppers by cutting price.12 Consequently, the featured store softens competition and

both stores enjoy higher profits. However, the non-featured store can stay out to avoid

the Emall’s fee and still reaps the benefit of the improved competitive environment.

This free-riding limits the Emall’s ability to collect fees because as long as the non-

featured store can make a higher profit by not joining the Emall, both stores will choose

to stay out in equilibrium. To see this point, consider the payoffs for in-in, in-out, and

out-out under fixed fees, as shown in table 1. In order to attract both stores, the

Emall has to charge a fixed fee such that each store makes at least the same level of

profit as that of the outside store in the in-out scenario, (α+C)rα+A+B+C

(α + A+ B), which

is larger than the profit a store will make without the Emall, αr.

10These are the consumers who are aware of the Emall and have no loyalty to either store.

11As we shall see in the next section, the convenience is measured by the reduction in search costs.

12Technically, the lower bound of the price support is higher when one store is featured by theEmall.

14

Page 17: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Proposition 2. The optimal strategies of the Emall are characterized as follows:

(i) For all fee structures both stores join the Emall in equilibrium.

(ii) Under fixed fees, the Emall charges f1 = (α+C)rCα+A+B+C

to the featured store and f2 = 0

to the non-featured store and its profit is Rf = (α+C)rCα+A+B+C

.

(iii) Under percentage fees, the Emall charges d1 = Cα+CA−αB(α+A)C

to the featured store and

d2 = CA−αB(α+A)C

to the non-featured store and its profit is Rp = Crd1+[α+C(1−d1)]r

α+A+(B+C)(1−d1)Bd2.

(iv) Percentage fees dominate fixed fees when CB> α

A.

Proposition 2 says that given the number of loyal customers (α segment) and those

who are unaware of the Emall (A segment), percentage fees are more profitable to the

Emall when the size of the featured store shoppers segment (C segment) is sufficiently

large relative to the Emall comparison shoppers segment (B segment). The superiority

of percentage fees vs fixed fees crucially depends on their ability to control free-riding,

which occurs when one store joins the Emall and the other store competes from outside

the Emall. In that case, the outside store reaps the benefit of the better competitive

environment in the presence of the Emall but avoids paying fee. The comparison be-

tween fixed fees and percentage fees is similar to that between lump-sum tax and ad

valorem tax. While ad valorem tax is easy to implement, it introduces distortion and

results in welfare loss. Fixed fees have no impact on the stores’ pricing strategies.

Under percentage fees, however, an online store’s pricing strategy is affected by the

Emall whether it is inside the Emall or out. The larger the size of the featured store

shoppers segment, the more influential are the Emall’s percentage fees on both stores’

pricing strategies, thereby effectively deters free-riding and improves the Emall’s prof-

itability. The drawback of percentage fees is that the higher percentage charged to

the featured store depresses the average price level and thus hurts the profits of both

stores as well as the Emall. To take advantage of percentage fees’ ability to control free

riding and ameliorate their downward pressure on price and profits, the Emall can use

15

Page 18: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

a combination of fixed fee and percentage fees. Specifically, the Emall should charge

a uniform percentage fee d to both stores and a fixed fee to the featured store in the

amount of [α + C(1 − d)]r − [α+C(1−d)]rα+A+(B+C)(1−d)

[α + A + B(1 − d)]. Under this pricing

scheme, a pure fixed fee is simply a special case when the percentage fee is 0.

Lemma 1. A two-part tariff is superior to pure fixed fees and pure percentage fees.

2.1.3 Discussion of Base Model Results

In the base model we analyzed the impact of the Emall on the competitive strategies

of the online stores and the Emall’s optimal fee. We show that the featured store

plays a competition reducing role and that the Emall is not only profitable but also

beneficial to both online stores and consumers. Our analyses demonstrate that per-

centage fees are superior to fixed fees when the featured store shoppers segment is

relatively large. This implies that the Emall should adopt fixed fees in its embryonic

stage and percentage fees as it becomes well established. These insights conform to

reality. For example, Excite shopping, a minor Internet mall, offers free participation

to online stores and charges fixed fees for advertising (note that feature is one form of

advertising). By contrast, Yahoo! Shopping charges 2% revenue share on sales driven

by Yahoo! Shopping to participating stores. In addition, it is widely publicized that

Yahoo! reached agreements with many of its featured stores in the form of payment

contract or mutual advertising agreements, which are equivalent to fixed fee.13 We

predict that the featured store charges a higher price on average, and that the more

shoppers purchase through the Emall, the higher the price of the featured store. These

predictions are supported by the data we collected.

13The pricing information of Excite shopping and Yahoo! Shopping were collected from their websites in February 2001.

16

Page 19: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

2.2 Segmentation among Emall shoppers endogenized

We have shown that Internet mall is a profitable business model that benefits online

stores as well as consumers when customer segmentation is determined exogenously.

Will this result hold in a setting where consumer behavior is endogenous? In this

section we relax the assumption that the Emall comparison shoppers segment and

featured store shoppers segment are fixed and allow Emall shoppers to choose their

segment membership endogenously. Assume that the total search costs are the product

of search time and unit time cost, i.e., t = γs, where the time cost (s) is uniformly

distributed, s ∈ U [0, 1], and search time (γ) is a category specific constant. This

assumption embodies the idea that total search costs are stable for each individual but

are heterogeneous across consumers within each category, and that total search costs

for each individual differ across categories. We may interpret unit time cost (s) as

consumer type. Search time (γ) is determined by category characteristics such as the

complexity of product attributes and usage frequency. In inexpensive categories with

simple and salient attributes, such as CDs and books, search time is shorter. Such

categories are likely to be associated with small γ. The opposite is true for categories

which are more expensive and contain products with sophisticated and ambiguous

attributes, such as apparel and computers.

The choice between shopping at the featured store and using the search engine to find

the least expensive product depends on the pricing policies of the stores as well as

search costs. If customers believe that the difference in price between the featured and

non-featured stores is relatively small, consumers who have moderate search costs will

shop at featured store and thus become a member of the C segment. Conversely, if

the price differential is expected to be substantial, even consumers with high search

costs may use the search engine and become a member of the B segment. In order to

17

Page 20: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

determine the size of the Emall comparison shoppers (B) and featured store shoppers

(C) segments, we need to consider consumers’ expectations. Recall that consumers

correctly anticipate the pricing policies of the stores. In the base model, we have shown

that the featured store charges a higher price on average. Therefore, featured store

shoppers expect to pay a higher price, but the price differential should be no greater

than the savings in search costs they would incur if they used the search engine. By

contrast, Emall comparison shoppers purchase from the low price store, which can be

either the non-featured store (more frequently) or the featured store (less frequently).

2.2.1 The Setting

As in the base model, we assume that store 1 is the featured store. Comparison shoppers

(B segment) expect to pay the average minimum price. By contrast, featured store

shoppers (C segment) expect to pay the average price (p) of store 1. The probability

distribution functions for pmin and p are as follows:

Prob{min(p1, p2) < p} = (1−Q)[1− (1− F c1 (p))(1− F2(p))] +QF2(p) (4)

and

Prob{pf = p} = Qr + (1−Q)F c1 (p) (5)

whereQ denotes the mass point of store 1’s price distribution function at the reservation

price r, Q = C(1−d)α+A+(B+C)(1−d)

, and F c1 (p) denotes store 1’s conditional price distribution

function for p ∈ [p, r).14

14Store 1’s price distribution function F1(p) as shown in the base model is not a proper distributionfunction because F1(r) < 1. To calculate the average price of the featured store, we derive F c

1 (p)conditional on p1 < r. More details are given in Appendix A.

18

Page 21: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Define disutility as any loss in utility. Emall shoppers incur different magnitudes of

disutility in comparison shopping versus featured store shopping. In both cases, con-

sumers suffer a utility loss for the price they pay. However, comparison shoppers

sustain a further utility loss by incurring higher search costs, which is measured by

the product of the time cost (s) and search time (γ). Without loss of generality, we

normalize the search costs for featured store shoppers to 0. The disutility for con-

sumers in comparison shopping (B segment) and featured store shopping (C segment)

are dui = pmin + γs and duf = p, respectively. In choosing their segment membership,

Emall shoppers will use the search engine when they expect dui < duf ; by contrast,

they will shop at the featured store when they expect duf < dui. In equilibrium, s∗

solves dui = duf , where s∗ = B

B+C. That is, consumers whose time cost is s ∈ [s∗, 1]

shop at the featured store, whereas consumers whose time cost is s ∈ [0, s∗] use the

search engine. Equilibria are obtained by backward induction and the derivation in-

volves the following steps: 1. The Emall decides on optimal fee structure, 2. Online

stores decide whether to join or not, 3. Knowing that prices affect segment sizes, online

stores formulate pricing strategies taking the Emalls fee structure as given, 4. Given

online stores pricing strategies, Emall shoppers form expectations and decide whether

to search or shop at the featured store. In equilibrium, the sizes of B and C segments

are such that all optimality conditions in steps 1, 2, and 3 are satisfied.

2.2.2 Equilibrium

As shown in Figures 2 and 3, two types of equilibria emerge. Each figure contains three

graphs. The graph on the top illustrates how the sizes of Emall comparison shopper

segment (B) and featured store shopper segment (C) change as search time (γ) varies;

the graph in the middle demonstrates how the average price at the featured store (pf )

and the average minimum price (pmin) change as search time (γ) varies; the graph at

19

Page 22: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Figure 2: A portrayal of the stable equilibrium

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 0.2 0.4 0.6 0.8 1 1.2γγγγ

Seg

men

t Siz

e

B

C

00.20.40.60.8

11.21.41.61.8

2

0 0.2 0.4 0.6 0.8 1 1.2γγγγ

p

Pf

Pmin

0

0.1

0.2

0.3

0.4

0.5

0.6

0.2 0.4 0.6 0.8 1 1.2

γγγγ

π πππ π1

π2

20

Page 23: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Figure 3: A portrayal of the unstable equilibrium

0

0.1

0.2

0.3

0.4

0.5

0.6

0 0.2 0.4 0.6 0.8 1 1.2γγγγ

Seg

men

t Siz

e

B

C

2.35

2.4

2.45

2.5

2.55

2.6

2.65

2.7

0 0.2 0.4 0.6 0.8 1 1.2γγγγ

p

Pf

Pmin

0

0.5

1

1.5

2

2.5

0 0.2 0.4 0.6 0.8 1 1.2

γγγγ

π πππ

π1

π2

21

Page 24: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

the bottom shows how profits of the featured store (π1) and the non-featured store (π2)

change as search time (γ) varies.15 In one type of equilibrium (the stable equilibrium),

as search time (γ) increases, the number of featured store shoppers (C segment) in-

creases,16 the average price at the featured store (pf ) and the average minimum price

(pmin) both increase, and the profits of the featured store (π1) and the non-featured

store (π2) both increase. By contrast, in the other type of equilibrium (the unstable

equilibrium), as γ increases, the number of featured store shoppers (C segment) de-

creases, the movement of pf and pmin diverges with pf monotonically increasing and

pmin monotonically decreasing, π1 declines and π2 roughly stays put. The first type of

equilibrium is intuitively appealing because everything else being equal, more shoppers

shop at the featured store and the featured store enjoys higher profit as search costs17

increases. However, the other type of equilibrium is indeed subgame perfect. The

Emall’s profitability and whether the profitability is invariant to the degree of con-

sumer sophistication critically depend on which type of equilibria prevails. We show

that the first type of equilibria (the stable equilibrium) dominates the other using an

intuitive refinement.

Proposition 3. There is a unique subgame perfect Nash equilibrium that survives the

refinement. In that equilibrium, the price differential between the featured and non-

featured stores widens as the size of the featured store shopper segment (C) increases.

In Figure 4, the difference between p and pmin (diffp) is drawn against C. The stable

equilibrium is located in the increasing segment of the graph, whereas the unstable

15The following parameter values are used to calculate equilibria for each given value of γ: r=3,αi = 0.1, A = 0.08, B + C = 0.72.

16Note that the sum of B and C segments is fixed, so increasing number of featured store shoppers(C segment) implies declining number of Emall comparison shoppers (B segment).

17Recall that search costs are the product of time cost (s) and search time γ.

22

Page 25: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

−2

−1.5

−1

−0.5

0

0.5

1

1.5

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

C

diff

p

a stable equilibrium

an unstable equilibrium

Figure 4: An intuitive refinement

equilibrium is located in the decreasing segment of the graph. Suppose that the two

stores’ strategy sets are fixed, as γ increases, more shoppers will shop at the featured

store (C increases). Now, since the stable equilibrium is in the increasing segment

of the graph, as C increases, the gap between p and pmin (diffp) widens—the gain

from using the search engine becomes larger, which counterbalances the increase in

C. Thus, the stable equilibrium is a fixed point. By contrast, since the unstable

equilibrium is in the decreasing segment of the graph, as C increases, the gap between

p and pmin (diffp) becomes narrower, which leads to further increase in C. Thus, an

irrelevant disturbance causes the collapse of the unstable equilibrium, in other words,

the unstable equilibrium does not constitute a fixed point.

2.3 Empirical support

In this section we test the following model predictions:

23

Page 26: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

1. Featured stores charge higher prices on average.

2. A featured store at a large Internet mall charges a higher price than that of a

featured store at a small Internet mall.

2.3.1 Featured stores charge higher prices on average

We track the prices at a number of Yahoo! Shopping featured stores and non-featured

stores for the following product categories: PDAs, printers, digital cameras, phones,

camcorders, DVD players, small appliances, DVDs. Within each category, we check

the price of an item at a featured store and then search the same item using Yahoo!

Shopping search engine. For every product, there are a few featured stores but nu-

merous non-featured stores. We record the prices of a product at all stores. Yahoo!

Shopping rates merchants on a five-star scale. One-star means poor, five-star means

excellent. Store rating is based on customer feedback on a number of attributes such as

customer service, fulfillment, etc. We record store rating along with price information.

Data are collected for multiple items within each category.

Recall proposition 1 says that the pricing strategy of the featured store first order

stochastically dominates the pricing strategy of the non-featured store. This claim has

two implications. First, it implies mean dominance, i.e., the featured store charges a

higher price on average. Second, it implies that one should observe that the featured

store charges a higher price more frequently. Both of these implications will be tested.

Let pf and pnf denote the prices of an item at the featured store and a non-featured

store, respectively. In a reasonably large sample set, we should observe pf > pnf

(confirmation) with a frequency significantly larger than 50%, where pf is the average

price at the featured stores and pnf is the average price at the non-featured stores. The

percentage of confirmation provides partial support for the two implications mentioned

24

Page 27: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

earlier. To see whether the difference between the average prices at the featured stores

and the average prices of the non-featured stores is significant, we perform a t-test.

Category level likelihood ratio test is conducted to examine whether the prices at the

featured stores are larger than the prices at the non-featured stores at a frequency

significantly above 50%. The test procedure of the category level likelihood ratio test

is as follows. Within each category, we count the number of occasions where the price

of a featured store is larger than that of the non-featured stores. Assume the event

that a featured store charges a higher price is a random draw from a binomial process,

then the likelihood of observing that event is:

L = C(n, k)pk(1− p)n−k

where n is the total number of price comparisons between the featured and non-featured

stores, k is the number of occasions where the featured store charges a higher price.

The null hypothesis is that it is equally likely to observe a featured store to charge a

higher or lower price than the non-featured store (p = 0.5), the alternative hypothesis

is that this probability is larger than 0.5 (p = kn). The test statistic is:

r = −2 lnL0

L1

which has a χ2 distribution. Table 3 presents some summary statistics. A snapshot of

the actual data is given in Appendix C. The complete data set is available from the

author upon request.

The percentages of confirmation, t statistics, and category level likelihood ratio test

statistics across the eight product categories reported in Table 3 support our theoretical

prediction that featured store charges higher prices on average. To check whether

25

Page 28: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Table 3: Some descriptive statistics of the data: Price comparisons between the fea-tured and non-featured stores (p-value in parenthesis)

Category PDAs Camcorders Printers Digital�Cameras

DVD�players

Phones Small�Appliances

DVDs

#�of�Products 10 10 10 10 10 10 6 10%�of�Confirmation 80% 80% 80% 90% 90% 70% 100% 90%

t�Statistics4.18�

(0.002)2.98�����

(0.015)6.28�

(0.0002)2.84�

(0.01)2.83�

(0.01)2.82�

(0.013)3.88����

(0.006)4.42�

(0.001)Category�Level�Likelihood�Ratio�Test�Statistics�(χχχχ2)

24.66�(<0.005)

16.56�(<0.005)

7.62�(0.01)

31.99�(<0.005)

16.25�(<0.005)

7.62�(0.01)

26.39�(<0.005)

11.63�(<0.005)

Table 4: Price–Rating Correlations (p-value in parenthesis)

PDAs Camcorders PrintersDigital�

CamerasDVD�

playersPhones

Small�Appliances

DVDs

Correlation�between�Price�and�

Rating

0.1351����(0.20)

0.0089���(0.93)

0.0785�(0.46)

0.0536��(0.62)

-0.0079�(0.94)

-0.1151�(0.28)

-0.0583�����(0.68)

-0.0423�(0.69)

featured stores charge higher prices due to other factors such as store reputation and

customer service level, we calculate the correlation between price and store rating for

each of the eight product categories. The results are reported in Table 4. It is found

that price is uncorrelated with store rating for all categories.

The fact that price is uncorrelated with store rating does not fully justify that feature

leads to higher price. For example, if consumers who prefer convenience to price rate

stores on the basis of convenience, and consumers who prefer price to convenience rate

stores on the basis of price, then store rating is inconsistent. To cope with this problem,

we conduct a survey. All stores in a Yahoo! Shopping product category are listed in the

survey. The subjects are asked to indicate whether they are aware of these stores, and

rate the perceived reputation for those stores that they are aware of. We search the

prices for ten products in this category. There are seven featured stores and eighty-two

26

Page 29: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Table 5: Regression Results

Intercept ∆∆∆∆A ∆∆∆∆RCoefficients 0.181971 -0.120255 -0.024674

t Stat 4.444724 -0.831226 -1.506315

Coefficients 0.126167 − −t Stat 4.808218 − −

non-featured stores that carry these products. For the featured and non-featured stores

that carry the same product, we calculate the difference in prices (∆P ), the difference

in awareness (∆A), and the difference in perceived reputation (∆R). We then regress

∆P on ∆A and ∆R. The intercept measures the effect of being featured. The results

are reported in Table 5. The coefficients of ∆A and ∆R are insignificant. In contrast,

the intercept is positive and significant. Thus, feature is the driving force behind the

price premiums of the featured stores over the non-featured stores.

2.3.2 A featured store at a large Internet mall charges a higher price than

that of a featured store at a small Internet mall

To verify whether featured stores at a large Internet mall charge higher prices than

those at a small Internet mall, we track the prices at a number of Yahoo! Shopping

featured stores and Shopping.com featured stores. Yahoo! Shopping is a large Internet

mall, and Shopping.com is a minor Internet mall, measured by the traffic volume

of their underlying Internet portals. According to PC Data Online, the numbers of

unique visitors to Yahoo! and AltaVista.com (the underlying portal of Shopping.com)

in December 2000 are 65.9 million and 18.5 million, respectively. We choose five pairs

of featured stores at the two Internet malls from the following two product categories:

computer and electronics. Within each category, we check the price of a particular

item at a Yahoo! Shopping featured store and that of its counterpart at Shopping.com.

27

Page 30: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Let pLf and pSf denote the prices of an item at the featured store of a large Internet mall

and a small Internet mall, respectively. In a reasonably large sample set, we should

observe pLf > pSf (confirmation) with a high frequency. The following table presents

some summary statistics. A snapshot of the actual data is given in Appendix C. The

complete data set is available from the author upon request.

Table 6: Some descriptive statistics of the data: Price comparisons between the fea-tured stores at a large Internet mall and a small Internet mall

Category # of Items # of Confirmation % of ConfirmationComputer 39 38 97.4%Electronics 9 8 88.9%

2 48 46 95.8%

2.4 An alternative featured store selection mechanism

It is not a priori clear how the Emall should choose the featured store. The decision to

feature a store may be an ongoing one, i.e., after the Emall announces its fee structure

for joining, every week the featuring decision is made (just as in a supermarket!).

Rather than collecting a fixed fee or a higher percentage fee from one of the stores

and feature it (mechanism 1), the Emall may choose to feature the store that charges

a higher price (mechanism 2). Intuitively, such a featured store selection mechanism

induces the stores to raise their prices to capture featured store shoppers (by being

featured). Without this inducement, the stores’ natural tendency is to compete for the

comparison shoppers by charging a lower price. Thus, mechanism 2 helps to reduce

price competition between online stores, thereby improves the Emall’s revenue. In this

section we investigate the relative merit of the two alternative featured store selection

mechanisms.

28

Page 31: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

2.4.1 The Setting

Assume that both stores are members of the Emall and the Emall picks the high price

store to be the featured store. When store i charges the reservation price, it will be

selected as the featured store by the Emall and get the business from featured store

shoppers (C segment) and its loyal customers. On the other hand, if store i charges a

lower price than store j (i, j = 1, 2, i 6= j), it will get the business from switchers who

do comparison shopping (A and B segments) and its loyal customers.

2.4.2 Results

Under the alternative featured store selection mechanism, the two stores’ equilibrium

strategies are characterized by the following price distribution function:

F (p) =

0, p <[α+C(1−d)]rα+A+B(1−d)

,

[α+A+B(1−d)]p−[α+C(1−d)]r[A+(B−C)(1−d)]p

,[α+C(1−d)]rα+A+B(1−d)

≤ p ≤ r,

1, p ≥ r.

(6)

Lemma 2. The larger the featured store shopper segment (C), the more profitable are

the stores. On average, the featured store charges a higher price regardless of which

featured store selection mechanism is implemented.

Of the two alternative mechanisms to select the featured store, which one is more

advantageous to the Emall? It turns out that the relative merit of the two mechanisms

is closely related to the percentage of shoppers who shop at the Emall and search time

(γ). Define k as the percentage of shoppers who shop at the Emall, k = B+CA+B+C

. When

most shoppers shop at the Emall (k large), it is better off for the Emall to choose the

high price store to be the featured store (mechanism 2). On the other hand, when a

significant percentage of shoppers shop outside the Emall (k small), it is more desirable

29

Page 32: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.2 0.4 0.6 0.8 1

k

Σπ ΣπΣπΣπ

Mechanism 1 dominates Mechanism 2 dominates

Figure 5: Comparing the two featured store selection mechanisms: a market coverageperspective

for the Emall to collect a fee from one of the stores and set that store as the featured

store (mechanism 1). This result can be visualized in Figure 5.

In Figure 5, we have drawn the total profits of the two firms (∑

π) against the per-

centage of shoppers who shop at the Emall (k). When the Emall has high market

coverage (k > 0.78 in the graph), mechanism 2 yields higher total profits than that

from mechanism 1, which implies that the Emall will receive more revenue by choosing

the high price store as the featured store.

Similarly, when shoppers have strong preference toward the featured store due to high

search costs, the Emall is better off to choose the high price store as the featured store

(mechanism 2). By contrast, when featured store preference is relatively weak (search

costs are low), it is more advantageous for the Emall to let one of the stores pay a

fee and be featured (mechanism 1). Recall that search costs depend on search time

30

Page 33: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

0.56 0.61 0.66 0.71 0.76 0.81 0.86 0.91 0.96

γγγγ

Σπ ΣπΣπΣπ

Mechanism 1dominates

Mechanism 2 dominates

Figure 6: Comparing the two featured store selection mechanisms: a search costsperspective

(γ). We illustrate this result in Figure 6. In Figure 6, the total profits of the two

firms (∑

π) are drawn against category specific search time (γ). When search time is

substantial (γ > 0.67 in the graph), mechanism 1 yields higher total profits than that

from scenario 2. Thus, mechanism 1 is more profitable to the Emall when γ is large.

Proposition 4. It is more profitable for the Emall to select the high price store to be

the featured store when the Emall is sufficiently attractive (k large), or shoppers incur

substantial search costs (γ large).

The intuition for proposition 4 is as follows. Under mechanism 2, online stores trade

off the benefit from charging a lower price with that of charging a higher price. While

a lower price attracts comparison shoppers, a higher price improves the store’s chance

of being featured, in which case the store captures featured store shoppers. The larger

the featured store shoppers segment, the larger the benefit of charging a higher price.

31

Page 34: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

When the featured store shoppers segment is large enough (γ and k large enough), the

incentive of the stores to raise price is so high under mechanism 2 that the resulting

revenue is higher than that under mechanism 1.

3 Summary and Conclusions

This research takes the first step toward understanding Internet malls as a new form

of e-commerce. We begin our analysis by providing a specific type of consumer seg-

mentation and examining the incentives of online stores to join Internet malls. In the

base model, we analyze the pricing policies and profit implications for online stores

and Internet malls, taking the segmentation scheme as given. We then allow consumer

segmentation to interact with the pricing policies of online stores and explore how such

interactions moderate the competitive strategies of online stores and Internet malls.

This paper distinguishes itself from previous research in that it focuses on the unique

mode of interactivity18 afforded by search engines and featured stores in Internet malls.

We demonstrate that Internet mall is a profitable business model through the compe-

tition reducing role of the featured store. Next, we show that Internet malls continue

to be profitable when consumer segmentation is endogenous. This result is supported

by a unique mixed strategy equilibrium that survives an intuitive refinement. Finally,

we prescribe how Internet malls and their member stores may fine-tune their strategies

by responding to market characteristics such as consumer search costs and the reach of

an Internet mall. Specifically, an Internet mall should feature stores for free when its

reach is high and consumer search costs are substantial. Conversely, it should charge

fees to featured stores when both its reach and consumer search costs are low. Our

empirical tests provide support to the model predictions. We show that: 1. on average,

18Alba et al. (1997) define interactivity as a continuous construct capturing the quality of two-waycommunication between two parties.

32

Page 35: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

featured stores charge higher prices than non-featured stores when other factors such

as store reputation and customer service are taken into account, 2. featured stores at

a large Internet mall charge higher prices than those at a small Internet mall, 3. as

search costs increase, the price differential between featured and non-featured stores

becomes larger. To managers, our research presents a positive analysis of Internet

malls as a business model. It serves as a preliminary guide to e-tailers and shows when

it is beneficial to join Internet malls and how they should adapt their strategies in

this particular online shopping environment. Our analysis also helps Internet portals

(owners of Internet malls) to evaluate their pricing schemes and shows what steps can

be taken to improve their profitability.

Our model can be extended in several ways. First, we have only analyzed Internet malls

in the context of homogeneous product market. An obvious extension is to consider

differentiated markets. Second, repeat buyers may form their expectations differently.

Specifically, they may incorporate their prior purchase experience into their future

expectations. To allow such expectations, we have to model consumers as Bayesians

who update their beliefs each period. Third, we have assumed that there is a single

featured store for mathematical tractability. An extension in which multiple featured

stores are allowed may be of interest. To model this situation, we may classify featured

stores and non-featured stores into two groups and analyze intra-group and inter-group

competition. We expect that most of the findings in this paper should carry through

in that case.

33

Page 36: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Appendix A

Technical note for the base model

As noted earlier, our base model is built upon the pioneering works by Varian (1980)

and Narasimhan (1988). We briefly recapitulate some key findings in Narasimhan

(1988) that are relevant to our model:

1. There is no Nash equilibrium in pure strategies.

2. The strategy sets of the two stores are convex.

3. Neither can have a mass point in the interior or at the lower bound of the other’s

price support, nor can either store have a mass point at the upper bound of the

other’s price support if that upper bound is a mass point for the other firm.

4. The two stores’ strategy sets are identical when neither store has a mass point.

If store j has a mass point at the upper bound of the price support, store i will

have zero density at that price in equilibrium.

In the base model we have shown that two-part tariff is superior to pure fixed fees

and pure percentage fees. All subsequent analyses are based on two-part tariff. Recall

that the expected profits for store 1 and store 2 are π1 = [α + C(1 − d)]r and π2 =

p[α+A+B(1−d)], respectively. The loyal customers and featured store shoppers will

buy from store 1 as long as its price is no greater than their reservation price, but the

competition for the comparison shoppers puts downward pressure on price. Clearly,

store 1 has more to lose when it moves away from reservation price. Thus, store 1 has

a mass point at the reservation price r. By contrast, store 2 does not. We can state

the equilibrium conditions for p ∈ [p, r) as follows:

p{α+ C(1− d) + (1− F2(p))[A+B(1− d)]} = [α + C(1− d)]r (A–1)

34

Page 37: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

and

p{α+ (1− F1(p))[A+B(1− d)]} = p[α + A+B(1− d)] (A–2)

Solving for F1(p) and F2(p) from equations (A–1) and (A–2) we get

F1(p) =[α + A+B(1− d)]{[α + A+ (B + C)(1− d)]p− [α + C(1− d)]r}

[α + A+ (B + C)(1− d)][A+B(1− d)]p(A–3)

and

F2(p) =[α + A+ (B + C)(1− d)]p− [α + C(1− d)]r

[A+B(1− d)]p(A–4)

From (A–3) and (A–4) we see that F1(p) = F2(p) = 0, F2(r) = 1, and F2(r) = 1−Q,

where

Q =C(1− d)

α + A+ (B + C)(1− d)(A–5)

This indicates that F1(p) has a mass point at r equals to Q.

Technical note for extension 1: segmentation among switchers

endogenized

Before we can write down the probability distribution functions for pmin and p, we need

to resolve the issue of mass point. Recall that the equilibrium distribution function for

store 1 F1(p) as shown in the base model is not a proper distribution function since

F1(r) = 1−Q. Given the mass point Q, we can derive the conditional price distribution

function for store 1 F c1 (p) from the following equilibrium condition:

p{α + [(1−Q)(1− F c1 (p)) +Q][A+B(1− d)]} = p[α + A+B(1− d)] (A–6)

35

Page 38: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

From (A–6) we get

F c1 (p) =

[α + A+ (B + C)(1− d)]p− [α + C(1− d)]r

[A+B(1− d)]p(A–7)

Note that F c1 (p) is a proper distribution function with F c

1 (p) = 0 and F c1 (r) = 1. We

can now express the probability distribution functions for pmin and p as follows:

Prob{min(p1, p2) < p} = (1−Q)[1− (1− F c1 (p))(1− F2(p))] +QF2(p) (A–8)

and

Prob{pf = p} = Qr + (1−Q)F c1 (p) (A–9)

Derivation of pmax and pmin in section 2.2

From (4) and (5) we can calculate pmin and p from the following expressions:

pmin =

∫ r

p

pd{(1−Q)[1− (1− F c1 (p))(1− F2(p))] +QF2(p)} (A–10)

and

p = Qr + (1−Q)

∫ r

p

pdF c1 (p) (A–11)

Substitute F c1 (p), F2(p), and Q into (A–10) and (A–11) we get

pmin = r(C − Cd+ α)[2B2 − 2Bα− 2Bdα+ 2Aα+ 2B2d2 − 4dB2 + 4BA−

2Cα ln(η) + 2Cαµ+ CBµ− CB ln(η)− 2Bα ln(η) + 2Bαµ+ 2A2−

dCAµ+ 2Cdα ln(η)− 2Cdαµ+ CAµ+ d2CBµ− 2α2 ln(η) + 2α2µ−

4dBA− d2CB ln(η) + 2dCB ln(η)− 2dCBµ+ dCA ln(η) + 2dBα ln(η)−

2dBαµ+ 2Aαµ− CA ln(η)− 2Aα ln(η)]/[(B −Bd+A)2η]

(A–12)

36

Page 39: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

and

p = r[B2αµ+B2α ln(η)− C2d3B ln(η) + 2C2αdµ+ 2Bd2AC ln(η) +A2Cdµ−

A2Cd ln(η) + 3B2d2C ln(η)− 3B2d2Cµ+ 4BACdµ− 4BACd ln(η)− 3B2·

Cd ln(η) + 3B2Cdµ+ 2BAC ln(η)− 2BACµ−B2d2αµ+B2d2α ln(η)+

2BdAαµ− 2BAαµ+ 2BAα ln(η)−B2Cµ+B2C ln(η) +A2C ln(η)−

A2Cµ− 2BdAα ln(η)−A2αµ+A2α ln(η) +B2d3Cµ−B2d3C ln(η)−

2Bd2ACµ− 2B2dα ln(η) + 2B2dαµ+ 2rBd2CA− rB2d3C − rA2Cd+

2rBCA− 4rBdCA+ 3rB2d2C + rA2C − 3rB2Cd+ rB2C − 2Cdα2 ln(η)−

2Bdα2 ln(η) + 2Bα2 ln(η) +BC2 ln(η) + C2α ln(η)−BC2µ− 2Bα2µ−

C2αµ+ α3 ln(η)− α3µ+ C2d2A ln(η)−AC2µ+ 2Bdα2µ− 2AC2d ln(η)+

AC2 ln(η)− 6CBdα ln(η)− C2d2Aµ+ 2AC2dµ+ 2Cdα2µ+ 6CBdαµ+

2Cα2 ln(η)− 2Cα2µ− 3CdAα ln(η) + 3CdAαµ+ C2d3Bµ− 2C2αd ln(η)−

3BdC2 ln(η) + 3BdC2µ+ 3CAα ln(η)− 3CAαµ+ 3CBα ln(η)− 3CBαµ+

2Aα2 ln(η)− 2Aα2µ+ 3Bd2C2 ln(η)− 3Bd2C2µ+ 3Cd2Bα ln(η)− 3Cd2Bαµ+

C2d2α ln(η)− C2d2αµ]/[(B −Bd+A)η2]

(A–13)

where η = α + A+ (B + C)(1− d), µ = ln[α + C(1− d)].

Technical note for extension 2: an alternative featured store

selection mechanism

By a similar argument as in the base model, we can show that the price support is

p ∈ [p, r], where

p =[α + C(1− d)]r

α+ A+B(1− d)(A–14)

Note that if the numerator is larger than the denominator, i.e., more customers are

either loyal customers or featured store shoppers then p = r. The profits for store i at

37

Page 40: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

a price p is given by

πi(p) = {α + Fj(p)C(1− d) + (1− Fj(p))[A+B(1− d)]}p (A–15)

where p ∈ [p, r], i 6= j, i, j = 1, 2. Each store chooses its pricing policy to maximize its

expected profit

maxFi

E(πi) =

πi(p)dFi(p) (A–16)

such that

πi ≥ αr∫ r

p

dFi(p) = 1

By assumption, the two stores are symmetric, therefore, their equilibrium price dis-

tributions are identical, i.e., F1(p) = F2(p) = F (p). We can state the equilibrium

condition for p ∈ [p, r] as

[α + C(1− d)]r = {α + F (p)C(1− d) + (1− F (p))[A+B(1− d)]}p (A–17)

From (A–17),

F (p) =[α + A+B(1− d)]p− [α + C(1− d)]r

[A+ (B − C)(1− d)]p(A–18)

Consumer segmentation endogenized

In this case, the expectation of consumers in segment B remain unchanged (they expect

to pay the average minimum price), but consumers in C segment now expect to pay

the average maximum price. The probability distribution functions of pmax and pmin

38

Page 41: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

are given by

Prob{max(p1, p2) < p} = [F (p)]2 (A–19)

and

Prob{min(p1, p2) < p} = [1− (1− F (p))2] (A–20)

From (A–19) and (A–20) we can calculate pmax and pmin as follows:

pmax =

∫ r

p

pd[F (p)]2 (A–21)

and

pmin =

∫ r

p

pd[1− (1− F (p))2] (A–22)

substitute (A–18) into (A–21) and (A–22) we get

pmax =2r[α+ C(1− d)]{α(χ− φ) + C(1− d) +B[(χ− φ− 1)(1− d)]−A(1 + φ− χ)}

[A+ (B − C)(1− d)]2

(A–23)

and

pmin =2r[α+ C(1− d)][α(φ− χ) +A+B(1− d) + C(1− d)(φ− χ− 1)]

[A+ (B − C)(1− d)]2(A–24)

where φ = ln[α + A+B(1− d)], χ = ln[α + C(1− d)].

The disutility for the B and C segments of consumers are dui = pmin + γs and duf =

pmax, respectively. In equilibrium, s∗ solves dui = duf , s∗ = B

B+C, which implies that

consumers whose time cost is s ∈ [s∗, 1] shop at featured store, whereas consumers of

type s ∈ [0, s∗] use the search engine. The optimal sizes of B and C can be derived

accordingly. As in section 2.2, two types of equilibria emerge. The properties of these

equilibria are illustrated in Figures 7, 8, and 9. By proposition 3, we can eliminate the

unstable equilibrium.

39

Page 42: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Figure 7: Mechanism 2: the stable equilibrium

00.20.40.60.8

11.21.41.61.8

2

0.5 0.55 0.6 0.65 0.7 0.75 0.8

γγγγ

p

Pmax

Pmin

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.5 0.55 0.6 0.65 0.7 0.75 0.8γγγγ

π πππ i iii

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.5 0.55 0.6 0.65 0.7 0.75 0.8

γγγγ

Seg

men

t Siz

e

B

C

40

Page 43: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Figure 8: Mechanism 2: the unstable equilibrium

0

0.1

0.2

0.3

0.4

0.5

0.6

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

γγγγ

Seg

men

t Siz

e

B

C

0

0.5

1

1.5

2

2.5

3

3.5

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

γγγγ

p

Pmax

Pmin

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8γγγγ

π πππ i iii

41

Page 44: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45

C

diff

p

an unstable equilibrium

a stable equilibrium

Figure 9: An illustration of the two types of equilibria: Mechanism 2

We now proceed to present an illuminating example and some interesting comparative

statics. As in section 2.2, suppose that the reservation price is $3 (r = 3) for the

consumers, each store enjoys a loyal segment of size 0.1 (α = 0.1), 90% of the rest of

the shoppers go to the Emall (k = 0.9). We vary search time (γ) and compile a sample

output in Figures 7 and 8. Given the parameter values specified above, the most

profitable scheme is for the Emall to collect 7% of revenue from its member stores.

Under this scheme, the equilibrium segment sizes for B and C are 0.641 and 0.079,

respectively. Each store makes a profit of 0.52, the revenue to the Emall is 0.073.

42

Page 45: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Appendix B

Proof of Proposition 1

Proof. It is easy to verify that for every p ∈ (p, r), F1(p) < F2(p), which implies

that F1(p) first order stochastically dominates F2(p). When more switchers shop at

the Emall, that is, non-Emall shoppers (A segment) become Emall shoppers (B and

C segments), the number of comparison shoppers (B segment) and featured store

shoppers (C segment) increases proportionally. Define E(p) as the expected price, we

have

E(p) =

∫ r

p

pdF (p) (B–1)

= [pF (p)]rp −

∫ r

p

F (p)dp

= r −

∫ r

p

F (p)dp

Thus, E(p) increases (decreases) when F (p) decreases (increases). Recall that F1(p) =

[α+A+B(1−d)]{[α+A+(B+C)(1−d)]p−[α+C(1−d)]r}[α+A+(B+C)(1−d)][A+B(1−d)]p

, and p = [α+C(1−d)]rα+A+(B+C)(1−d)

.

Clearly, p is increasing in C, but the F1(p) is decreasing in C. Note that the profits for

store 1 and store 2 are (α+C)r and p(α+A+B), respectively. Therefore, the average

price as the featured store (E(p1)) and the Emall’s profits (proportional to the sum of

the two stores’ profits) increase as more shoppers purchase through the Emall.

Proof of Proposition 2

Proof. To prove that both stores join the Emall in equilibrium, we need to show that

out-out and in-out cannot occur in equilibrium. It’s easy to verify that out-out is not

an equilibrium (see tables 1 and 2). Suppose that store 1 joins the Emall and store

2 stays out, and in-out is an equilibrium. The expected profits of store 1 before and

43

Page 46: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

after joining the Emall are αr and (α + C)r, respectively. The Emall will charge a

fixed fee in the amount of Cr to fully extract the additional benefits accrued to store

1, which leaves store 1 indifferent between joining the Emall or not. By contrast, store

2’s expected profits after store 1 joining the Emall are (α+C)rα+A+B+C

× (α+A+B), which

is strictly greater than αr. Therefore, both stores want to be the one that stays out,

this contradicts the assumption that in-out is an equilibrium.

Under fixed fees, the outside option for both stores when both stores join the Emall is

(α+C)rα+A+B+C

(α + A + B). As shown in table 1, the payoffs to the featured store and

non-featured store are (α + C)r and (α+C)rα+A+B+C

(α + A + B), respectively. Therefore,

the optimal fixed fees to the featured store and non-featured stores are f1 = (α+C)rCα+A+B+C

and f2 = 0. If the non-featured store chooses not to join, the outside option for the

featured store becomes αr, and the Emall will charge a fixed fee of f = Cr in this

situation. Given the Emall’s strategy, the best response for each store is to join the

Emall. Thus, both stores join the Emall under f1 and f2.

Under percentage fees, the outside option for both stores when both stores join the

Emall is αrα+A

(α + A + B). As shown in table 2, the payoffs to the featured store

and non-featured store are [α + C(1 − d1)]r and [α+C(1−d1)]rα+A+(B+C)(1−d1)

[α + A + B(1 − d2)],

respectively. Therefore, the optimal percentage fees to the featured store and non-

featured stores are d1 = Cα+CA−αB(α+A)C

and d2 = CA−αB(α+A)C

. If the non-featured store chooses

not to join, the outside option for the featured store becomes αr, and the Emall will

charge a percentage fee of d = 100% in this situation. Given the Emall’s strategy, the

best response for each store is to join the Emall. Thus, both stores join the Emall under

d1 and d2.

The Emall’s profits under fixed fees and percentage fees are Rf = (α+C)rCα+A+B+C

and Rp =

Crd1r +[α+C(1−d1)]r

α+A+(B+C)(1−d1)Bd2, respectively. The difference between Rp and Rf is given

44

Page 47: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

by

Rp −Rf =r(α+A+B)(2αB2 + α2B +AαB + 2CαB + Cα2 + 2CαA+ CA2)(CA− αB)

(Cα2 + 2CαA+ CA2 + αB2 + CαB)(α+A)(α+A+B + C)(B–2)

Hence, percentage fees dominate fixed fees when CA > αB. ¤

Proof of Lemma 1

Proof. From proposition 2, we know that fixed fees are superior to percentage fees

when CA ≤ αB. In this parameter range, two-part tariff is equivalent to fixed fees

(by setting percentage fees to 0). When CA > αB, it suffices to show that two-

part tariff yields higher revenue to the Emall than percentage fees. Under two-part

tariff, the Emall charges a uniform percentage fee d to both stores and a fixed fee

to the featured store. The outside option for both stores when both stores join the

Emall is αrα+A

(α + A + B). The payoffs to the featured store and non-featured store

are [α + C(1 − d)]r and p[α + A + B(1 − d)], respectively, where p = [α+C(1−d)]rα+A+(B+C)(1−d)

.

Comparing the optimal two-part tariff percentage fee d to the optimal percentage fees

d1 and d2 under pure percentage fees, it is easy to verify that d2 < d < d1. The Emall’s

revenue from the non-featured store can be written as R2 = pBd. Under percentage

fees, Rper2 = [α+C(1−d1)]r

α+A+(B+C)(1−d1)[α + A+ B(1− d2)]. It can be shown that p is decreasing

in d when CA > αB. Since d2 < d < d1, R2 is larger under two-part tariff. The

Emall’s revenue from the featured store (the sum of fixed fee and percentage fee) can

be written as R1 = [α+C(1− d)]r− αrα+A

(α+A+B). Since d < d1, R1 is larger under

two-part tariff. This completes the proof for the lemma.

Proof of Proposition 3

Proof. In Figure 4, the difference between p and pmin (diffp) is drawn against C. The

stable equilibrium is located in the increasing segment of the graph, whereas the un-

stable equilibrium is located in the decreasing segment of the graph. Suppose that

45

Page 48: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

the two stores’ strategy sets are fixed, as γ increases, more shoppers will shop at the

featured store (C increases). Now, since the stable equilibrium is in the increasing

segment of the graph, as C increases, the gap between p and pmin (diffp) widens—the

gain from using the search engine becomes larger, which counterbalances the increase

in C. Thus, the stable equilibrium is a fixed point. By contrast, since the unstable

equilibrium is in the decreasing segment of the graph, as C increases, the gap between

p and pmin (diffp) becomes narrower, which leads to further increase in C. Thus, an

irrelevant disturbance causes the collapse of the unstable equilibrium, in other words,

the unstable equilibrium does not constitute a fixed point.

Proof of Lemma 2

Proof. The first half of the lemma can be proved by observing that in mechanism 1,

the two stores equilibrium profits are π1 = [α+C(1−d)]r and π2 = [α+C(1−d)]rα+A+(B+C)(1−d)

[α+

A + B(1 − d)], respectively. Both are increasing in C. In mechanism 2, each store’s

equilibrium profit is π = [α+ C(1− d)]r, which is increasing in C. The second half of

the proposition is true for mechanism 2 by definition. For mechanism 1, note that F1(p)

stochastically dominates F2(p) in the first order, which implies E(p1) > E(p2).

Proof of Proposition 4

Proof. This proposition is derived from numerical simulations. Details are available

from the author upon request.

Appendix C

46

Page 49: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Table 7: Featured stores charge higher prices on average: A snapshot of the data

Category Item

Palm Vx 299.98 281.72 √Palm m505 449.99 421.06 √

Palm IIIc 299.99 283.09 √Casio Cassiopeia E-

125638 555.69 √

Casio PC-Unite BZX-20

114 90.73 √Compaq iPAQ

H3650525 529.05 ×

HP Jornada 548 449.99 471.1 ×HP Jornada 547 449 419.44 √HP Jornada 720 899.99 876.87 √Sony Clie PEG-

S300329.99 302.2 √

% of Confirmation

80%

4.18 (0.002)

24.66 (0.0001)

PDAs

t Statistic ( ), p-value in parenthesis

Category Level Likelihood Ratio Test Statistic (χχχχ2222), p-value in parenthesis

nfp f nfp p>

f nfp p>

fp

47

Page 50: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Table 8: Featured stores at a large Internet mall charge higher prices than their counterparts in a small Internet mall: A snapshot of the data

Category

Mall Shopping.com price Yahoo! price

Featured Store ABT Electronics circuit city √Item JVC VCR

HRS4800U226 JVC VCR

HRS4800U249.99 √

Item JVC VCR HRS3800U

181 JVC VCR HRS3800U

199.99 √

Item Sony VCR SLVN80 198 Sony VCR SLVN80 199.99 √

Item Panasonic VCR PVV4620

148 Panasonic VCR PVV4620

149.99 √

Item Sony CD Player CDPCX235

189 Sony CD Player CDPCX235

199.99 √

Item Sony CD Player CDPCX53

169 Sony CD Player CDPCX53

149.99 ×Featured Store BeachCamera RitzCamera

ItemCanon ELURA 2

DIGITAL CAMCORDER

1249Canon ELURA 2

DIGITAL CAMCORDER

1599.95 √

Item Canon ES55 8MM CAMCORDER

285 Canon ES55 8MM CAMCORDER

349.95 √

ItemCanon OPTURA PI

DIGITAL CAMCORDER

1149Canon OPTURA PI

DIGITAL CAMCORDER

1499.95 √

ItemPanasonic - PV-

D100 Digital Camcorder

579Panasonic - PV-

D100 Digital Camcorder

799.95 √

ItemPanasonic - PV-

DV400 Digital Camcorder

704Panasonic - PV-

DV400 Digital Camcorder

999.95 √

Item

Panasonic - PV-DV600 Mini-DV Palmcorder with

Niteshot

839

Panasonic - PV-DV600 Mini-DV Palmcorder with

Niteshot

1199.95 √

ItemPanasonic - PV-L550 Palmcorder

Camcorder 327

Panasonic - PV-L550 Palmcorder

Camcorder 399.95 √

ItemPanasonic - PV-L650 Palmcorder

Camcorder 415

Panasonic - PV-L650 Palmcorder

Camcorder 499.95 √

ItemPanasonic - PV-L750 Palmcorder

Camcorder 449

Panasonic - PV-L750 Palmcorder

Camcorder 599.99 √

ItemCanon - Powershot G1 3.3MP Digital

Camera 848

Canon - Powershot G1 3.3MP Digital

Camera 899.99 √

Item Canon - S20 Digital Camera

554 Canon - S20 Digital Camera

699.99 √

Item FUJI FINEPIX 1400 DIGITAL CAMERA

234 FUJI FINEPIX 1400 DIGITAL CAMERA

299.99 √

Item FUJI FINEPIX 2400 DIGITAL CAMERA

338 FUJI FINEPIX 2400 DIGITAL CAMERA

374.99 √

Item FUJI FINEPIX 4700 DIGITAL CAMERA

599 FUJI FINEPIX 4700 DIGITAL CAMERA

799.99 √

Item Kodak DC-215 DIGITAL CAMERA

279 Kodak DC-215 DIGITAL CAMERA

299.99 √

Item Kodak DC-3400 DIGITAL CAMERA

394 Kodak DC-3400 DIGITAL CAMERA

499.99 √# of

Confirmation38 out of

39% of

Comfirmation97.4%

ElectronicsSf

Lf pp >

48

Page 51: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

References

Alba, J., J. Lynch, and B. Weitz et al. “Interactive Home Shopping: Consumer, Re-

tailer, and Manufacturer Incentives to Participate in Electronic Marketplaces”

Journal of Marketing, vol. 61, 38–53, 1997.

Bakos, Yannis “Reducing Buyer Search Costs: Implications for Electronic Market-

places” Management Science, vol. 43, no. 12, 1676–1692, 1997.

Bakos, Yannis and Erik Brynjolfsson “Bundling and Competition on the Internet”Mar-

keting Science, vol. 19, no. 1, 63–82, 2000.

Balasubramanian, Sridhar “Mail versus Mall: A Strategic Analysis of Competition

between Direct Marketers and Conventional Retailers” Marketing Science, vol.

17, no. 3, 1998.

Bradlow, Eric and David Schmittlein “The Little Engines That Could: Modeling the

Performance of World Wide Web Search Engines” Marketing Science, vol. 19, no.

1, 43–62, 2000.

Carrie Johnson et al. “eCommerce Brokers Arrive” The Forrester Report, March, 2001.

Dudey, Marc “Competition by Choice: The Effects of Consumer Search on Firm Lo-

cation Decisions” American Economic Review, vol. 80, no. 5, 1092–1104, 1990.

Fischer, Jeffery and Joseph Harrington, Jr. “Product variety and firm agglomeration”

Rand Journal of Economics, vol. 27, no. 2, 281–309, 1996.

Hauble, Gerald and Valerie Trifts “Consumer Decision Making in Online Shopping En-

vironments: The Effects of Interactive Decision Aids” Marketing Science, vol. 19,

no. 1, 4–21, 2000.

49

Page 52: Competition In An Internet Mall: A Strategic Analysis of A ... · Competition In An Internet Mall: A Strategic Analysis of A New Marketing Venue Chuan He⁄ Washington University

Iyer, Ganesh “Competition in Heterogenous Markets: Theory and Evidence”Working

paper, Washington University in St. Louis, February, 2001.

Iyer, Ganesh and Amit Pazgal “Internet Shopping Agents: Virtual Co-Location And

Competition” Working paper, Washington University in St. Louis, April, 2001.

Lal, Rajiv and Miklos Sarvary “When and How is Internet Likely to Decrease Price

Competition” Marketing Science, vol. 18, no. 4, 485–503, 1999.

Lynch, John and Dan Ariely “Wine Online: Search Costs Affect Competition on Price,

Quality, and Distribution” Marketing Science, vol. 19, no. 1, 83–103, 2000.

Messinger, Paul and Chakravarthi Narasimhan “A Model of Retail Formats Based on

Consumers’ Economizing on Shopping Time” Marketing Science, vol. 16, no. 1,

1–23, 1997.

Narasimhan, Chakravarthi “Competitive Promotional Strategies” Journal of Business,

vol. 61, no. 4, 427–449, 1988.

Novak, Thomas, Donna Hoffman, and Yiu-Fai Yung “Measuring the Customer Expe-

rience in Online Environments: A Structural Modeling Approach” Marketing

Science, vol. 19, no. 1, 22–42, 2000.

Shapiro, Carl and Hal Varian “Information rules: an analysis of the network economy”

Harvard Business School Press, 1998.

Varian, Hal “A Model of Sales” American Economic Review, vol. 70, no. 4, 651–659,

1980.

Wernerfelt, Birger “Selling Formats for Search Goods” Marketing Science, vol. 13, no.

3, 298–309, 1994.

50