Professional Documents
Culture Documents
a r t i c l e
i n f o
Article history:
Received 12 May 2008
Received in revised form 24 October 2008
Accepted 27 January 2009
Available online 5 February 2009
Keywords:
B2B e-marketplaces sustainability
Ownership structure
Market competition
a b s t r a c t
B2B e-marketplaces alter the structure of buyerseller trading networks. To study the network-level
structural changes caused by the emergence of e-marketplaces, we develop a multiple-player noncooperative
game, where rational rms select optimal interrm connections and the network is endogenously formed
and evolved. We examine the conditions under which both neutral and biased B2B e-markets will sustain,
when previous buyerseller connections exist. We test our model in both the oligopoly market and the
oligopsony market. Our analysis explains how ownership structures and market competition interact with
each other to affect e-market sustainability. We also identify other critical factors for sustainable e-markets
and their social welfare implications.
2009 Elsevier B.V. All rights reserved.
1. Introduction
B2B e-marketplaces match buyers and sellers with automated
transactions, lower search costs, and increased process effectiveness
and efciency [5,6,18]. The emergence of such e-marketplaces represents one of the major market transformations brought about by the
proliferation of information technologies [14,24]. However, despite
rms' enthusiasm in Internet-based B2B e-marketplaces, the growth of
B2B transactions has fallen short of earlier expectations [10]. Most
notably, the shakeout of B2B e-markets1 during the late 1990s and early
2000s has spurred skepticism about the earlier high expectations about
the role of these e-markets. A large number of B2B e-markets, such as
Chemdex and Adauction, went out of business, while others, including
e-Steel and Covisint, changed their business model from e-market
operators to technology service providers. Nevertheless, there are still
hundreds of B2B e-markets, such as World Wide Retail Exchange and
SciQuest, that have survived and thrived [22]. It is thus intriguing why
some e-marketplaces have survived while others have failed and what
key factors lead to a sustainable e-marketplace.
In this paper, we try to answer the question by focusing on the
structural dimension of e-markets [12], which refers to the overall
pattern of connections between rms. The structural perspective
provides us insights into network stability, when individual rms
strategically establish interrm links. The resulting network and the
rms' relative positions in the network determine their bargaining
Corresponding author. Tel.: +1 704 687 7637 (K. Zhao), +1 217 333 2878 (M. Xia),
+1 217 333 5159 (M.J. Shaw), +1 704 687 7604 (C. Subramaniam).
E-mail addresses: kzhao2@uncc.edu (K. Zhao), mxia@uiuc.edu (M. Xia),
mjshaw@uiuc.edu (M.J. Shaw), csubrama@uncc.edu (C. Subramaniam).
1
In this paper, we use the terms e-marketplaces and e-markets interchangeably.
0167-9236/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.dss.2009.01.005
power and performance. When equilibrium is reached, both participating rms and the market operator are willing to keep their existing
connections. This implies the emergence of a sustainable B2B emarket. Otherwise, the e-market will fall apart and evolve into other
business models or exit the market.
We develop a multiple-player non-cooperative game to simulate
the endogenous formation of public B2B e-marketplaces (many-tomany connections). In the model, we describe the overall network
structure as a graph, in which rational and self-interested rms are
nodes and their business relationships are the edges between
corresponding nodes. Buyers and sellers select individual connections
in the game, and their payoffs are determined by the overall network
structure and their relative positions in the network. We focus on
three major factors that affect the endogenous formation of B2B emarkets (Table 1). The rst factor is the ownership structure of the emarket. An e-market can be either neutral or biased [30]. A neutral emarketplace, such as EC21 and Alibaba, is owned by an independent
intermediary, while a biased one is operated by a group of buyers (e.g.,
World Wide Retailer) or sellers (e.g., iSteelAsia) [30]. For the biased
type, we focus on buyer-biased e-marketplaces and then show that
seller-biased e-marketplaces demonstrate similar properties.2 Since
previous studies suggest that neutral and biased e-markets have
difference properties [30], we also examine whether the ownership
structure will affect e-market sustainability. The second factor is the
type of market competition faced by both participating rms and the
market operator. An e-market can be established in either an oligopoly
market or an oligopsony market. In an oligopoly market, there are
2
Yoo et al. [30] also nd that biased marketplaces, no matter whether buyer-biased
or seller-biased ones, have similar economic properties.
106
Table 1
The endogenous formation of B2B e-marketplaces.
107
more buyers than sellers and buyers face more intensive competition.
Examples include natural resource markets and major distributors,
such as Ingram Micro, who transact with a large number of resellers/
buyers. In the oligopsony market, there are fewer buyers than sellers.
Contrary to the oligopoly market, buyers have a major advantage over
sellers in the oligopsony market. For example, the market in which
major retailers, such as Wal-Mart and Target, deal with a large number
of suppliers can be characterized as an oligopsony market. Will market
competition affect the equilibrium of e-markets, since buyers and
sellers have different bargaining powers in an oligopoly market
compared to an oligopsony market? Prior theoretical work is limited
in this area. The third factor is the impact of preexisting buyerseller
connections prior to the emergence of e-markets, as rms often have
direct exchanges for doing business in the absence of e-marketplaces.
For example, the defense contractor Raytheon had used in-house
developed tools to connect with its own suppliers before it joined
Exostar, an e-market cofounded with other large manufacturers in the
aerospace and defense industry [31]. A B2B e-marketplace can either
bring prior disconnected rms together or substitute previously
existing direct buyerseller connections. To the best of our knowledge,
this is one of the rst papers to investigate the process of how B2B emarkets supersede direct buyerseller networks.
Our analysis yields several insights on the sustainability as well as
social welfare impacts of B2B e-marketplaces. First, it demonstrates the
interactions between market competition and ownership structures of
e-markets. The emergence of a neutral e-market does not change
market competition, and the sustainability conditions are the same in
either the oligopoly market or the oligopsony market. By contrast, the
emergence of a biased e-market changes market competition by leading
to the formation of a buyer or seller consortium. We nd that a biased emarket formed by the more competitive side is less likely to survive than
one formed by the less competitive side. Second, our analysis underscores the importance of considering prior buyerseller connections,
which determine rms' incentives to join the e-market and the
efciency gains provided by e-markets. Third, sustainable e-marketplaces require low e-market connection costs to attract rms'
participation. Furthermore, the impacts of the number of buyers and
sellers on the e-market sustainability are different depending on the
ownership structure. Lastly, we show that social welfare is improved
with the emergence of a neutral e-marketplace, but the change in social
welfare is ambiguous in the case of biased e-marketplaces.
The rest of our paper is organized as follows. We rst review the
related literature in network economics and B2B e-marketplaces. In
Sections 3 and 4, we present our model and apply it to study direct buyer
seller networks and to establish the pre-e-market equilibrium structures
in both the oligopoly market and the oligopsony market. In Sections 5 and
6, we study the formation of a neutral B2B e-marketplace and a buyerbiased e-marketplace, respectively. We then discuss theoretical contributions, policy implications, and future research opportunities.
2. Literature review
3. Model framework
In this section, we introduce our basic model framework and our
assumptions.
3
While the unit demand assumption may seem restrictive, it does represent
industries where sellers are exible specialists [25], who deliver small batches of
goods (considering each batch as an aggregated unit of good) to buyers in each
production cycle. Many industries, such as apparel, electronic components, and
professional services, feature demands of this type [21].
108
all competing buyers pay the same price. Overall, the auction
mechanism and the network structure jointly determine the nal
transactions between buyers and sellers. Due to the connection
constraint for trade, the model is fundamentally different from a
traditional second price auction. The bilateral exchange constrained by
the network connections also exhibits double-sided network effects
[30]. A buyer enjoys positive network effects as more sellers establish
connections with it. Meanwhile, the buyer faces negative network
effects as more buyers establish relationships with its linked seller set.
3.3. Network equilibrium: pairwise stability
We use the concept of pairwise stability to dene network equilibrium [16,17]. Pairwise stability reects the fact that a relationship in
an exchange network must be mutually benecial, since linkage
formation inherently requires that not only must a rm be desirous of
forming a linkage, it should also be attractive to potential partners [1,
pp.317]. Thus, mutual consent is necessary for establishing a connection between a pair of partners (e.g., both Raytheon and its suppliers), but severing a link needs only a unilateral decision from either
party involved (e.g., either Raytheon or its suppliers).
We use the value function, f (G), to denote the expected payoffs of
any rm f in the network G. f ' represents any other rm in the network G. Pairwise stability requires that links in the equilibrium network G satisfy:
For all gff 0 aG; f G z f G gff 0 and f 0 G z f 0 G gff 0 : 1
For all gff 0 gG; if f Gb f G + gff 0 then f 0 G N f 0 G + gff 0 ;
and vice versa:
buyerseller connections but also changes the overall network structure [26]. In this paper, we will not discuss the case in which e-markets
supersede physical markets and the network structures remain the
same before and after the formation of e-markets (see Bakos [7] for an
excellent discussion about this case).
In direct buyerseller networks, the link costs include, but are not
limited to, search costs to locate potential trading partners, negotiation
costs to exchange information and establish business procedures, and
setup costs to link two internal information systems. We distinguish the
link costs between buyers and sellers, as they incur different activities in
general, but assume all buyers have the same cost, and so do the sellers.
That is, it costs a buyer lB(KBi) to establish and maintain direct buyer
seller links, while it costs a seller lS(kSj) to build and keep such connections. kBi/Sj represents the number of link the buyer Bi/Seller Sj has.
l(0) = 0, l(k) N 0, l(k) 0, indicating that rms incur either a linear or a
concave link costs. In the case where rms could benet from economies
of scale, they will face lower marginal cost as the number of links
increases. For example, it may become easier for a rm to establish an
interorganizational system with a new trading partner if it has used
similar systems before with existing business partners.
Specically, the network formation process is modeled in the
following two-stage game:
Stage 1: Buyers and sellers simultaneously determine whether to
maintain a link with one another. A link is established only
when both the buyer and the seller want to connect with
the other party. All can observe the consequent network G.
Stage 2: Firms' private information is revealed; then they compete in
the second price auction restricted by the structure of the
network G.
Under different ranges of link costs, multiple equilibrium network
structures exist. Here, we focus on least link complete (LAC) networks
[21], dened in the next paragraph, and use it as a benchmark case for
all pre-e-market structures. If the pre-e-market structure is a LAC
network and thus already enables efcient good allocation, it is most
difcult for e-markets to emerge since e-markets can only reduce
network connection costs without increasing the efciency of good
exchanges. If pre-e-market structures are not efcient, the formation
of B2B e-markets not only changes network structures, but can also
enhance exchange efciency. A special type of the pre-e-market structure is a null network with previously unconnected buyers and sellers,
when direct buyerseller links are prohibitively expensive due to
causes such as geographical dispersion. For instance, it is difcult for
many small Chinese suppliers to meet global buyers without the help
of global B2B e-markets, such as EC21 or Alibaba. In this case, rms'
pre-e-market payoffs are zero, and B2B e-markets are more likely to
emerge. We focus on the analysis of LAC networks without loss of
generality.5
In the oligopoly market, n goods will be sold when the market is
fully cleared. An exchange network is allocatively complete (AC) if and
only if, for every subset of n buyers, there is a feasible allocation in
which every buyer in that subset can purchase successfully. Therefore,
for any realization of buyers' private valuation, the n buyers who have
the highest valuation in AC networks can all obtain products from the
constrained auction game. Obviously, when all buyers and sellers are
connected with each other, the network is AC. An LAC network is an AC
with a minimal number of links, which we regard as an efcient
network since the connection in the network is costly. If the most
efcient networks can move to a market, it will be easier for the less
efcient ones to change as well.
5
We can extend our analysis to other pairwise stable structures of direct buyerseller networks. Firms' pre-e-market payoffs will be different, and the sustainability
conditions of e-markets will change accordingly. However, the properties of network
transition processes still hold.
109
lS m ng V m 1
n
m + 1
m + 1
1
to lose at least m +
1 with the probability
then both the buyer and seller are willing to add the link. Therefore,
removing any link from LAC networks is not feasible given the
condition in Proposition 1.
Please note that the layouts of LAC networks are not unique.
110
Property 1a. LAC networks are less likely to sustain in the oligopoly
market as the number of buyers increases.
Based on Proposition 1, we nd that if the number of buyers
increases, the link costs have to be smaller for LAC networks to reach
equilibrium. When there are more buyers in the exchange network,
the competition for goods becomes more intensive and rms expect
less marginal gain from any additional network connection. Consequently, rms have less incentive to keep multiple links, and it is more
difcult for LAC networks to reach equilibrium. In other words, as the
buyer side becomes more fragmented, the resulting stable direct
buyerseller networks are less likely to be efcient.
nn + 1
2mm + 1
iaM
2mm + n1 lS m n + 1 =
m n
2m + 1
lS m n + 1.
The number of buyers' links range from 1 to n. For buyer Bi with kBi
links,
h
i
ELAC Bi G
n:m
1
mn
mn
mn
1:m
2:m
E X
+ E X
+ ::: + E X
lB kBi
=
:
m
m+1
m+1
m+1
1
n
n1
1
nn + 1
+
+ ::: +
lB kBi =
lB kBi :
=
m m+1
m+1
m+1
2mm + 1
Social welfare is the sum of rms' individual payoffs in the network. In our paper, we use W to indicate social welfare. The expected
social welfare in the LAC network is:
h
i
h
i
X
X
n
nlS m n + 1
ELAC W = nELAC Sj G +
ELAC Bi G =
lB kBi :
2
iaM
iaM
As we have discussed before, LAC networks are used as a benchmark for all pre-e-market structures. In the following analysis, we use
LAC networks to represent the equilibrium prior to the emergence of
e-markets.
5. The emergence of a neutral B2B e-market
In direct buyerseller networks, rms have to keep multiple links
in order to maintain a strong bargaining position and achieve efcient
good exchanges. With the entry of a neutral e-market, a large number
of buyers and sellers can be connected to the other side with only a
single link to the intermediary (Fig. 2). Buyers and sellers want to link
to a neutral e-market to save on individual link costs.
Neutral e-markets, such as EC21 (http://www.ec21.com/) and
Texileweb (http://www.textileweb.com), match buyers and sellers
and facilitate information exchange, physical deliveries and payment
between them [7]. To be able to facilitate transactions among multiple
participants, a neutral e-market has to invest I, to build the IT infrastructure and maintain the exchange platform. Its prots come from the
entry fees, eB and eS7, charged to buyers and sellers, respectively. For
instance, EC21 charges $450 for its Trade OK members. Buyers and
sellers incur setup costs, sB and sS, when they adopt e-market connections. Setup costs may include costs to post/check information on emarkets and improvement of internal connectivity to meet the requirement of e-marketplaces. If rms join a neutral e-market, their new
link costs are the sum of the e-market connection costs and the entry fee
7
For the unit demand exchange, the entry fee can also be considered as transactionbased.
8
For the incomplete participation cases, we can only study individual cases by
specifying idiosyncratic network structures.
111
= nflS m n + 1 sS g +
lB kBi msB . Comparing those
ISoc
iaM
X
IInter
=
lB kBi mlB 1 N 0.
The intuition behind the enhanced social welfare with the emergence of a neutral e-market is that it reduces the total number of links
required for efcient good exchange. Buyers and sellers voluntarily
switch to e-market links only if doing so can lead to better payoff.
However, buyers' positions in direct buyerseller networks are asymmetric, since some buyers have fewer links and can save less from
switching to e-market connections. In order to motivate all rms to join,
the independent intermediary has to lower entry fees and share part of
the increased social welfare with less-motivated rms. Consequently,
the neutral intermediary will invest less than the socially optimal level.
In this section, we discussed the conditions under which an efcient direct buyerseller network will migrate to a neutral B2B emarket. We identied factors such as double-sided fragmentation and
low e-market connection costs that lead to the sustainability of a
neutral e-market. Our analysis implies that market competition has no
112
nn + 1
2mm + 1
n
m eS
I
m
sB . In
mlB 1 sB .
We can extend our analyses to the oligopsony market, where the
market competition is on the seller side (m b n). In the oligopsony
market, sellers' production costs are assumed to be independently and
identically distributed on [0,1] with the commonly known uniform
distribution. We assume that buyers' reservation value, vi, is independently and identically distributed on a uniform distribution be+ 1 10
tween 1 and mn +
1 . In the oligopsony market, the equilibrium condition
of a buyer biased e-market is shown in Proposition 5.
Proposition 5. (Sustainability of a buyer-biased e-market in the oligopsony market): When I V m2nm ++ 11 nsS + mlB n m + 1 sB ,
a LAC network will evolve to a buyer-biased e-market that is pairwise
stable as well as efcient.
Proof. In the oligopsony market, sellers' expected payoff from the
second trading stage is
h
i
EBiased Sj G
n:n
1 m+1
m+1
n m + 1:n
=
+ ::: +
E X
E X
sS eS
n
n+1
n+1
=
mm + 1
s S eS :
2nn + 1
Similar to the oligopoly market case, we assume that buyers' reservation value is
higher than the market clearing price, where the second price auction determines the
exchange between the entire set of m buyers and n sellers. Thus all m buyers will
participate in the nal good exchange. If some buyers' reservation value is lower than
the market clearing price, then they cannot obtain goods. The situation is then
analogous to the case in which less than m buyers participate in the exchange.
#
those two equations, we nd that I#
Soc IBiased.
The proof of the oligopoly case is similar and is omitted.
During the network transition to a buyer-biased e-market, buyers
want to join the consortium only if their payoff is better than in pre-emarket structures. However, after all buyers participate in the
consortium, sellers have to switch to e-market-centered connections
Table 2
Firms' expected payoffs and expect social welfare in the biased e-market.
Oligopoly market
Buyer's payoff
Seller's payoff
Social welfare
n
2m
n
m sS
I
m
sB
0
n
2
Oligopsony market
1
2
n
m sS
I
m
sB
0
nsS I msB
m
2
nsS I msB
even if their payoffs are worse, as they have no choice. The change of
social welfare can go either way depending on the extent of sellers' total
payoff adjustment. If not all buyers join buyer-biased markets, buyers
cannot fully capture the benet from enhanced bargaining power, since
there is still competition between them. Sellers' connection choice
becomes much more complicated, depending upon the fraction of
buyers who join the biased market.
7. Discussion, managerial implications, and concluding remarks
In this paper, motivated by the e-commerce shakeouts a few years
ago, we studied the conditions for B2B e-markets sustainability. We
developed and solved a multi-player noncooperative game to examine
the emergence and dynamics of B2B markets. Our paper makes several
theoretical contributions. First, we provide a model of the network-level
analysis of e-markets and demonstrate the structural changes brought
by new B2B connections. Second, our analysis reveals the interaction
between e-market ownership structures and market competition.
When a neutral e-market emerges as an intermediary, we expect to
observe similar properties in both the oligopoly market and the
oligopsony market. However, the sustainability of a biased e-market is
different under the two market conditions. For example, a buyer-biased
e-market is less likely to succeed in an oligopoly market than in an
oligopsony one, since the buyer-biased e-market signicantly reduces
the sellers' bargaining power. Thus, it is not surprising that many emarkets formed by small buyers (called as reverse aggregators in [18]),
such as FOB, have failed. Meanwhile, those buyer-biased e-markets
formed in oligopsony markets, such as World Wide Retail Exchange,
have survived the e-commerce shakeouts. Another successful example
is Exostar, an aerospace-defense e-market formed by manufactures, BAE
Systems, Boeing, Lockheed Martin, Raytheon and Rolls-Royce. It has
attracted tens of thousands of suppliers since its inception [31]. When
buyers form an e-market consortium to connect to their sellers, the
buyers can consolidate their bargaining power and the sellers have no
choice but to connect to the e-market. The buyers in the consortium
appropriate the benets from reduction in link costs and also by driving
down the sellers' prots. This leads to an interesting situation where it is
uncertain whether a biased e-market can enhance the overall social
welfare in the exchange network.
The third contribution of our paper is showing the importance of
examining network structures that existed before the emergence of the
e-markets. Those prior e-market structures are critical in determining
the outcome but have often been overlooked in previous theoretical
analysis. In particular, benets created by B2B e-markets are two-folded:
link cost reduction and efciency gains, both of which depend on the
prior e-market structures. Since rms only need one link to connect with
a group of buyers/sellers in the e-market, their savings are greater if they
have more pre-e-market connections. However, rms' heterogeneous
positions in prior exchange networks lead to different incentives in
joining e-markets. The efciency gains brought about by e-markets also
depend on the structure of prior exchange networks. If prior buyer
seller connections are limited, e-markets can bring more trading partners for a rm and create efciency gains in addition to link cost
reduction. Our model focuses on LAC networks, which are the most
challenging for an e-market to emerge, since LAC networks already
enable efcient product allocations and e-market can only offer link cost
saving benets. Our model can be applied to other non-LAC networks
and when doing so, the conditions for e-markets to emerge and sustain
in such networks will be easier to satisfy.
Our work also offers several managerial implications for e-market
initiators and operators. First, our results suggest that it is important to
keep e-market connection costs low enough to attract rms' participation. It explains why it is important that Exostar spares its members
from buying added technologywhether hardware and softwareto get
a job done; technologies are purchased as a variable cost service and are
included in the platform, rather than being downloaded to clients the
113
114
Kexin Zhao is an assistant professor of MIS at the Belk College of Business at the University
of North CarolinaCharlotte. She received her Ph.D. degree from the University of Illinois at
Urbana-Champaign in 2007. Her research interests are economics of information systems,
development and adoption of e-business standards, and game theory in information
management. She has published one book chapter and several peer-reviewed journal
papers. Her teaching interests include introduction of business computing, system analysis
and design, electronic commerce.