ElectronicMarketplaces

A of Introduction
Strategic Analysis
Buyers often face substantial search costs in
Electronic orderto obtain informationabout the prices and
productofferings of sellers in a market.These
Marketplaces costs introduce inefficiencies into market-
intermediatedtransactionsand detractfromthe
abilityof markets to provide an optimalalloca-
By: J. Yannis Bakos tion of productiveresources. Interorganizational
Graduate School of Management information systems can create "electronic
University of California, Irvine marketplaces"by serving as intermediariesbe-
Irvine, California 92717 tween the buyers and the sellers in a vertical
market,in the process reducingthe cost buyers
must incur to acquire informationabout seller
prices and productofferings.Thisarticlefocuses
on the role of buyer search costs and certain
other economic factors that determinethe prac-
tical significance and the strategic dymanics of
electronicmarketplacesin commodityand differ-
Abstract entiated markets.
Informationsystems can serve as intermediaries Simple microeconomicmodels typicallyassume
between the buyers and the sellers in a vertical that buyers can costlessly acquire full informa-
market, thus creating an "electronic market- tion about the prices and product offerings of
place. "A majorimpactof these electronicmarket sellers in a market.For example, this assump-
systems is that they typicallyreduce the search tion underliesthe central resultof convergence
costs buyers must pay to obtain information to a unique competitiveprice-takingequilbrium
about the prices and productofferingsavailable in competitivemarkets.Incertainoligopolisticset-
in the market.Economictheorysuggests thatthis tings, it leads to the resultthat sellers willnot be
reductionin search costs plays a majorrole in able to keep prices above marginalcosts and
determiningthe implicationsof these systems for thus willrealizezero profits,knownas Bertrand's
marketefficiencyand competitivebehavior.This paradox.There is littledisagreement that cost-
articledrawson economic models of search and less, perfect informationabout marketprices is
examines how prices, seller profits, and buyer an unrealisticsimplification.Yet the actual im-
welfareare affected by reducingsearch costs in plicationsof deviationsfromthis assumptionfor
commodity and differentiated markets. This marketprices and for the conduct and welfare
reductionresults in direct efficiency gains from of firms and consumers have been the subject
reduced intermediationcosts and in indirectbut of much debate. In real-worldmarkets,the evi-
possibly larger gains in allocational efficiency dent variabilityof prices and the emphasis on
frombetter-informedbuyers. Because electronic disseminationof informationthroughadvertising
marketsystems generallyreduce buyers'search and other media suggest that the cost and the
costs, they ultimatelyincrease the efficiency of availabilityof price and productinformationare
interorganizationaltransactions,in the process importantdeterminantsof economic behavior.
affectingthe marketpowerof buyersand sellers.
The economic characteristics of electronic The goal in this article is to demonstrate how
marketplaces,in additionto theirabilityto reduce economic theory can provide insights into the
search costs, create numerous possibilities for benefitsof electronicmarketplaces,the strategic
the strategic use of these systems. options of the parties involved, and their likely
strategicconduct.Althoughthese questionshave
Keywords: Interorganizationalsystems, elec- already been addressed in the information
tronic markets, strategic informa- systems literature(e.g., Cash and Konsynski,
tion systems 1985), the contributionof this article lies in pro-
vidinga supplementarylineof reasoningground-
ACMCategories: K.1, K.4.3, K.6.0 ed in economic theory; it emphasizes the

MIS Quarterly/September1991 295

 ElectronicMarketplaces

underlyingreasoning and the resultingstrategic Role of electronic marketplace

implications.The interested readercan findthe
model specifications that extend existing As discussed in latersections of this article,the
economic theoryand the complete mathematical cost that buyers incur in order to acquire infor-
formulationsin Bakos (1987), chapters 6 and 7. mationabout prices and productcharacteristics
enables sellers to extract monopolisticrents in
The nextsection reviewsthe concept and the role otherwisecompetitivemarketsand creates inef-
of electronic marketplaces.This is followed by ficiencies inthe allocationof economic resources
a discussion of the implicationsof search costs and the distributionof economic wealth. Elec-
in commoditymarketsand shows how in price- tronic marketplaces require major fixed in-
competitivemarkets,even a small cost of search vestments in system development, but, once in
on the buyers' partmay enable sellers to main- place, they promiseto reducethe marginalcosts
tain prices substantially above their marginal of interorganizationalcoordinationand handle
cost; inthis scenario,the introductionof a market largervolumes of markettransactionsin a more
system providingprice informationcan dramati- timely fashion before they reach their point of
cally reduce seller profits and increase buyer saturation.Ifthey are able to fulfillthis promise,
welfare. Next is a discussion of search costs in they are likely to proliferatebecause of these
differentiatedmarkets.The heterogeneityof pro- superiorcost/performancecharacteristics. Fur-
duct offerings and consumer tastes in such a thermore,some authors have argued that infor-
market allows sellers to exploit buyer search mationtechnologywillfavorcoordinationthrough
costs and enjoy increased profits.An interesting market-based rather than hierarchical gover-
outcome in this setting is that an electronic nance mechanisms (Malone, et al., 1987),
marketplaceimprovesallocationalefficiency by resulting in an increased role for electronic
enabling buyers to locate sellers who better marketplaces.
match their needs; these efficiency gains are
comparable in size to the reduction in search
costs. Finally,the last section discusses the im- Real-worldelectronicmarketsystems typicallyof-
plicationsof the economic characteristicsof elec- fer valuablefeaturesbeyondthe postingof prices
tronic marketsystems for the strategic conduct and productcharacteristics.Forexample, airline
of buyers, sellers, and intermediaries. reservationsystems allowticketingand billingin
additionto their market-relatedfunctionality.In
Electronic Marketplaces this article the focus is primarilyon the role of
electronic marketplaces in bringingtogether a
We are witnessingan increaseinthe numberand supplier and a customer; it is recognized,
functionalityof informationsystems that cross however, that once this relationshiphas been
organizationalboundaries,such as systems link- established,throughan electronicmarketsystem
ing one or more firmsto their customers and/or or by any other means, interorganizationalinfor-
suppliers.The terminterorganizationalinforma- mationsystems can playan importantrolein sup-
tionsystems (IOS),firstintroducedby Barrettand portingthe resulting bilateralrelationship.
Konsynski(1982), is now widelyused to charac-
terize these systems. An electronicmarketplace
(orelectronicmarketsystem) is an interorganiza- Economic characteristics of
tional informationsystem that allows the par-
ticipating buyers and sellers to exchange electronic market systems
informationabout prices and productofferings. In order to use economic theory to understand
The firmoperatingthe system is referredto as the strategic implicationsof electronic market-
the intermediary, which may be a market places, we must focus on their most salient
participant-a buyeror a seller, an independent characteristics, especially the ones that
thirdparty,or a multi-firmconsortium.The use distinguishthem from other types of capital in-
of the term "electronicmarketplace"in this arti- vestments. Inthat context, five characteristicsof
cle has a narrower,system-orientedfocus com- electronic marketsystems can explain, froman
pared with Malone,et al.'s (1987) more generai economic perspective,theirstrategicpotentialas
use of the term "electronicmarket"in referring well as theirimpacton the structureand efficien-
to the correspondinggovernance mechanism. cy of markets.

296 MIS Quarterly/September1991

 ElectronicMarketplaces

1. An electronic marketsystem can reduce thermore, technological and organizational

customers' costs of obtaining informa- resources and expertise acquired during the
tion about the prices and product offer- development and operationof one system may
ings of alternative suppliers as well as be transferableto other systems, resulting in
suppliers' costs of communicatinginfor- economies of scope.
mation about their prices and product
characteristics to additional customers. 5. Potential participants in electronic
marketplacesface substantial uncertain-
Thiscost reductionis likelyto affectthe monopoly ty regarding the actual benefits of join-
powerof the suppliersor the monopsonypower ing such a system. Occasionally this
of the customers in a vertical market that is uncertainty remains even after an
moved "online" by the introductionof an inter- organization joins the system.
organizationalinformationsystem. It will also This uncertaintycan affectthe strategicbehavior
have implicationsforthe efficiencyof thatmarket
in terms of the search costs experienced by of buyers, sellers, and potentialintermediaries,
buyers and their ability to locate appropriate by inducing them to adopt a "wait and see"
sellers. strategywhere they delay introducingor joining
a system in the hope thatthey willlearnfromthe
2. The benefits realized by individual par- experiences of other organizations.
ticipants in an electronic marketplacein- Competitivemoves and technological develop-
crease as more organizations join the ments can affectthe strategicsignificanceof the
system. above factors. In some markets, for example,
This property,known in economics as network competition among systems, new technology,
externalities(Katzand Shapiro,1985),can affect and even governmentregulationare increasingly
the dynamicsof the introductionand adoptionof limitingthe ability of intermediariesto exploit
electronic marketsystems, e.g., by favoringthe switching costs. Furthermore,while the above
first intermediaryintroducingsuch a system. economic characteristicsdeterminethe strategic
potentialof electronicmarketplaces,they are not
3. Electronic marketplaces can impose exclusive to these systems. The reduction in
significant switching costs on their par- buyersearch costs is the single attributethat is
ticipants. most specificto electronicmarketplaces;in other
Electronicmarketsystems may requiresizable aspects, especially in terms of the last two
characteristicsabove, electronicmarketsystems
investmentsfromtheirparticipantsin hardware,
may not be very differentfrom other types of
software,employee training,and organizational
transformations.Such investmentsmay become capital investments, including investments in
worthlessshould the organizationdecide to join strategic informationsystems.
a differentsystem or to revert to the previous For instance, better transportationsystems or
mode of operation. Competing intermediaries advertisingmedia can decrease the search cost
may need to compensate potentialsystem par- of buyers, just like electronic marketplacesdo.
ticipants for their switching costs or invent Competingstandards in the personal computer
technology that minimizesthese costs in order and workstationmarkets offer network exter-
to lure them away from rivalsystems. nalities, similar to electronic marketplaces,
because the wide adoptionof a standardresults
4. Electronic marketplaces typically re- in increasedavailability of hardwareand software
quire large capital investments and of- productsas well as large numbersof othercom-
fer substantial economies of scale and
patible users. Adoptionof any computingplat-
scope. form, not only electronic marketplaces,creates
An intermediaryusuallymust incurlarge system switchingcosts due to the cost of convertingsoft-
developmentand maintenancecosts (bothfixed ware and retrainingusers. Similarly,several in-
and variable);it willthen face relativelysmall in- dustries requirelarge capitalinvestmentsas the
cremental costs for each additionaltransaction price of entry, and most new technologies are
untilthe capacity of the system is approached, characterizedby uncertaintyabout their actual
resultingin substantialeconomies of scale. Fur- benefits. Electronicmarketplaces,however,are

MIS Quarterly/September1991 297

ElectronicMarketplaces

unique in that they typically exhibit all five of these stimulated economists to ask how firms and in-
characteristics. dividuals behave, and how markets function,
when buyers and sellers do not have perfect in-
Informationsystems researchers face the task of
formation about the consequences of their ac-
extending economic theory in the appropriate tions. Inthe case of search costs, it can be shown
areas, drawing the proper conclusions, and bring- that sequentially rational rules (i.e., rules that
ing the relevant results into the information describe a strategy that rational economic actors
systems field (Bakos and Kemerer, 1990). In this would want to follow to the end) will be based on
spirit, the following two sections focus on the abili- a reservation price, which is set so that the ex-
ty of electronic marketplaces to reduce the cost
pected gain from searching once more equals the
buyers face in obtaining price and product infor- cost of the search (Rothschild, 1974).
mation, and they illustratehow economic analysis
can be applied to study the forces driving the Several authors have pointed out that the actual
economic and strategic implications of this tech- distribution of seller prices may be unknown at
nology. The discussion is subsequently expand- the beginning, or it may change over time. For
ed to include the strategic impacts of three other example, when there is only a small number of
characteristics of electronic marketplaces iden- sellers, finding the price of one seller may change
tified above: large fixed costs, network exter- the expected price distribution for the remaining
nalities, and switching costs. sellers. Search models can be extended to cope
with these contingencies by allowing for the
Commodity Markets dynamic updating of buyer beliefs (Rothschild,
1974). Another concern is that in some real
Products in commodity markets are essentially markets it may be unrealistic to expect buyers'
identical across all sellers, as is often the case,
prior beliefs to reflect accurately the actual
for example, in the markets for agricultural grain distribution of seller prices, in which case rules
products or gold bullion. Buyers will typically based strictly on a reservation price are not
choose the seller with the lowest total cost, which robust and can lead a buyer with inaccurate ex-
will usually include the price paid to the seller plus
pectations to a grossly suboptimal outcome.
any search, transportation, and other similar Gastwirth(1971) proposed combining the two into
costs. This section examines how buyers behave
hybrid rules that are both robust and preserve the
when they search for a seller in a commodity reservation price property: buyers will search up
market. Because of the simple structure of a com- to a certain maximum number of times or until
modity market, most search models in the litera- the best price discovered is less than their reser-
ture can be used to study the impact of electronic vation price (which gets dynamically updated as
marketplaces by examining their outcome when new information is gathered about seller prices).
the marginal cost of search is substantially re-
duced. The key insight offered is that under vir-
tually any assumptions about the nature of the
Impact of search costs in
market and the search process, electronic commodity markets
marketplaces are likely to destabilize profitable
monopolistic outcomes, thereby reducing seller Most search rules proposed in the literature, in-
profits and increasing buyer welfare. cluding the ones reviewed above, have the follow-
ing properties: (1) seller prices decrease as the
cost of search decreases; (2) the amount of
Rules of search in commodity search increases as the cost of search
decreases; (3) the amount of search increases
markets as seller prices become more dispersed; and
The perfectly competitive or monopolistic (4) ceteris parabus, as seller prices become
markets of simple microeconomic theory assume more dispersed, buyers' total costs decrease
that buyers are fully informed about the prices (buyers need to search more when there is a wide
of all sellers. This leads to a single market price variety of seller prices, but they are compensated
and offers buyers no incentive to compare the by better deals.)
prices of different sellers. Stigler's (1961) seminal The search costs faced by buyers allow the
work titled "The Economics of Information" sellers to maintain prices at equilibrium substan-

298 MIS Quarterly/September 1991

 ElectronicMarketplaces

tiallyabove their marginalcost; it can be shown price charged by sellers gradually decreases
that, under certain assumptions about buyer fromthe monopolypriceto the competitiveprice.
behavior, at equilibriumall sellers charge the We have seen that in almost any setting, as the
same price, and it is the one that would have cost of buyersearch decreases, it becomes more
been charged by a monopolist.The first half of difficultforsellers to sustain high prices. Reduc-
this result was first proved by Diamond(1971)
and the second halfof Salop and Stiglitz(1977). ing the cost of obtaininginformationaboutseller
Whathappens in this case is that if search costs prices in a commoditymarketmay thus under-
mine a monopolistic outcome. Under certain
are highenough, each seller willslightlyincrease
his or her price knowing that the unfortunate assumptions, for example, prices are mono-
polisticforhighenoughsearch costs and become
buyers that visit will prefer to pay the slight indeterminate(no equilibrium) when search costs
premiumratherthan embarkintoanotherexpen- fall below a certainthreshold(Bakos, 1987). As
sive search. The resultis thatallsellers gradually
raise theirprices to reachthe monopolisticlevel. long as the cost of search is still significant,
sellers are still likelyto enjoy some excess pro-
At this point no seller has an incentiveto lower
fits, althoughsmallerthan underthe monopolistic
the price;because of highsearch costs he or she
will not be able to attract enough additional equilibrium.Ifthe search costs decrease enough,
all profitsenjoyed by the sellers eventuallyare
demand.
competed away. Ineithercase, the resultis a net
These models of buyersearch are limitedto the welfaregain fromimprovedsearch efficiency(as
extent that they offer a theory only about the fewer resources are expended in non-productive
buyers'side of the market.The variability of seller search activities) and from the reduction or
pricesis exogenous and usuallydisappearswhen elimination of inefficiencies that characterize
one triesto close these models by allowingprofit- monopolisticmarkets(such as pricingmarginal
maximizing behavior on the sellers' side; at buyers out of the market).Finally,lower prices
equilibrium,all sellers charge a single price. result in a transferof wealth fromthe sellers to
Buyersthen do not have to search, undermining the buyers.Electronicmarketsystems are social-
the motivationfor the development of these lydesirablewhentheirnet welfaregains outweigh
theories. On the other hand, these models can their development and operatingcosts.
demonstratehow even modest search costs can
lead to prices substantiallyhigherthan marginal
costs, even in commoditymarketsand when the
Role of electronicmarketplaces
sellers behave competitively(i.e., withoutany col- A commodity product bought from different
lusion). Morerecent workin this area suggests sellers can differonly in its price. We have seen
that the key in avoidingthe above problemslies that sellers can still realize substantial profits,
in introducingasymmetries in the buyer and however,as longas comparisonshoppingis cost-
seller sides of the market (i.e., allowing for ly fortheircustomers.As computerand telecom-
heterogeneous buyers and sellers). municationstechnology in the formof electronic
For example, in the real world not every buyer marketsystems makethe distribution of informa-
has access to an electronicmarketsystem. Salop tion moreefficient,the opportunitiesforfat, easy
and Stiglitz(1982)have studieda settingwithtwo profitswill shrink. Commoditymarkets may be
destablized by price wars that wipe out any ex-
types of buyerswho face differentsearch costs. cess profitsenjoyedby the sellers. As articulated
This is similar to a scenario in which certain
in the followingquote by John Phelan Jr., at the
buyershave access to an electronicmarketplace time chairmanof the New YorkStock Exchange,
that provides informationabout seller prices at
a low cost, while buyers withoutaccess to the "Technologyand communicationbringefficien-
electronicmarketplaceface highersearch costs. cy. Money is made in inefficiency" (Hansell,
Theiranalysissuggests a mixedpriceequilibrium 1989, p. 92). Sellers may thus attemptto delay
in whichcertainsellers charge highpricesto take the arrivalof electronic marketsystems, or they
advantage of buyers with high search costs. may try to controltheir development.
Buyerswithlowsearch costs induce the entryof A case in point is the European market for
low-pricedsellers. As the proportionof buyers government and blue chip bond issues. At the
with low search costs increases, the average Eurobondmarket's1987 annualmeetingin Oslo,

MIS Quarterly/September 1991 299

ElectronicMarketplaces

Salomon Brothersled the big Americantrading among majorbonddealersthatthey need to con-

firmsintosuccessfully opposing proposalsfor a trol electronic trading,especially because they
computerizedreal-timeprice-quotation system for providemuch of the data on which such trading
the trading of Eurobonds (Economist, 1987). is based. Itmay be difficultto achieve this goal,
Such a system would have made marketinfor- however, without provoking government
mation available instantlyto all subscribing in- regulation.
vestors and traders, regardless of their size or
location, and could have markedthe beginning
of the end of big trading profits.1 Differentiated Markets
Similarly,inthe domestic marketforU.S. govern- Certainmarkets,likegovernmentbonds or most
ment fixed income securities (whichhas a daily agriculturaland mineralproducts, deal in com-
turnoverof over $100 billion),big bond dealers modityproducts,yet the majorityof marketsare
had a virtualmonopolyin real-timebond prices characterized by differentiated products. Dif-
in the mid-1980s.Thus, they enjoyed an edge in ferentiatedmarketsare commonplace because
tradingand were able to profitby charginga price buyer preferences are heterogeneous for most
differentialto their customers. Despite the fact types of products. Furthermore,these markets
that a General AccountingOffice study in 1987 are attractive to sellers because they offer
called for better public access to bond prices, substantial profit opportunities. Differentiated
these dealers were understandablyopposed to marketsinvolvea varietyof productofferingsand
an open tradingsystem that mightthreatentheir consequentlythe search problembecomes more
monopolyprofits.Improvementsin computerand complex;buyers need to consider boththe price
telecommunication technology,however,allowed of a particularseller and the characteristicsof the
financialdata vendors to gain access to timely correspondingproductoffering.Inthis section we
bond price data in the late 1980s. As a result, focus on describinga modelof search in differen-
these vendors, includingReuters HoldingsPLC, tiated marketsappropriateto study the implica-
Telerate, and Quotron, made inroads in that tions of electronic marketplaces.
marketby postingbondpricedata and eventually
by matching smaller buyers and sellers. In an Modelsof productdifferentiation
ironictwist,the opposingbonddealers eventually
were forced to join these electronic systems Economistshave been interestedin differentiated
themselves to take advantage of their superior markets since the late 1920s. A number of
efficiency, in the process paying millions of models have been proposed for their study,
dollars in fees (Hermanand Power, 1990). which generally fall in two categories:
Inagreement withthe economic argumentspre-
sented, the government bond business turned * Spatialdifferentiation models trace theirorigin
froma gold mineto a minefield injusttwo years; to Hotelling's(1929) formalizationof spatial
several dealers are losing money, and the rest competition.Inthat class of models, product
enjoy razor-thinprofit margins (Herman and attributesare treatedas choices of locationin
Power, 1990). Although the electronic market an n-dimensional space. Consumer prefer-
systems were notthe only structuralchange (for ences determine the location of their ideal
example,the entryof manymoreparticipants,in- productor productmix, and some type of dis-
cludingseveralJapanese houses, has made that tance metricin the productattributespace is
marketmore competitive),these systems seem used as a proxy for consumers' utilityloss
to have played a significantrole in that process. when they are not able to purchasetheirideal
As discussed in more detail later in this article, productor productmix.
there is gradually a consensus being formed * The brandsubstitutionformalization of product
differentiationis oftenattributedto Chamberlin
(1933).Inthis class of models,consumersmay
'Morerecently,otherdevelopmentshave combinedto purchase several brands according to a
severelydamage profitsin the Eurobondmarket,while brand's prices, desirability,and the abilityof
actions taken by the dealers to protectthese profits
caught the attentionof the regulatoryauthorities;the any given pairof brandsto substituteforeach
profitabilityof this markethas suffered as a result. other.

300 MIS Quarterly/September1991

 ElectronicMarketplaces

The advantage of brand substitution models is sellers (Bakos, 1987). Each buyer searches un-
that they allow multiple brands to compete for the til a product close enough to his or her
attention of any particular consumer; in spatial preferences is located. As the search decreases,
models consumers consider only the products buyers become more demanding and keep sear-
closest to their ideal product vector and typical- ching untilthey find a product closer to their ideal
ly buy the one most desirable. Spatial models on preference. If the search cost becomes zero,
the other hand focus more clearly on product buyers look at all product offerings and purchase
charcteristics and on the dynamic aspects of the one best serving their needs, resulting in a
competition through the reformulationof existing socially optimal allocation. If there is a large
brands or through the introduction of new ones number of products, seller profits will be low but
(Perloff and Salop, 1985; Schmalensee, 1978). still not zero because product differentiation
Their typical assumption that consumers pur- prevents all-out price competition. If a higher
chase only one product at a time is acceptable degree of product differentiationcan be sustained
in most real markets. e.g., through advertising, buyers bcome less will-
ing to purchase an offering significantly different
Modeling search costs in from their ideal product (i.e., their "transporta-
differentiatedmarkets tion" cost increases). In that case, sellers will en-
The economic literatureon search costs suggests joy an increase in profits, which may partially or
that heterogeneity of consumers and product of- completely offset the decrease caused by lower
search costs.
ferings is central to the ability of sellers to exploit
buyer search costs and thus extract monopolistic The ability of electronic marketplaces to improve
rents (Reinganum, 1979; Shilony, 1976). It is thus
market efficiency is particularly evident when
appropriate to extend standard models of dif-
ferentiated markets to take account of search high search costs threaten the very existence of
a market. In differentiated markets, a sufficient-
costs.
ly large search cost will force buyers to stay out
One good candidate is the "unit circle" or "city of the market even if they are offered a zero price;
around the lake" model of spatial differentiation they would find the expected cost to locate an
(Salop, 1979) in which seller offerings and buyer acceptable seller too high, even if they know that
preferences are located along a unit circle; such a seller is guaranteed to exist. It follows,
buyers face a "transportation"2 cost when they then, that high search costs can lead to a market
buy from a seller whose offering is not identical breakdown. This is in contrast to the case of free
to their preference. Bakos (1987) modelled the information, where some buyers may stay out of
impact of search costs by requiring each buyer the market because of high transportation costs,
to incur a certain cost in order to be informed but sellers always service their "local" markets.
about the location (i.e., the product attributes)and An important implication of this analysis is that
the price offered by some seller; he or she must electronic market systems providing product and
then decide whether to purchase one of the pro- price information may generate substantial
ducts already identified, keep searching, or give allocational efficiencies by enabling customers
up. to locate suppliers that better match their needs.
At the unique symmetric equilibrium each seller
charges a prices that depends on the search and As search costs decrease, so do price premiums
transportation costs; the higher these costs, the and seller profit margins. The best strategy for
higher the prices and profits enjoyed by the a buyer in such a market is to determine a "price
and inconvenience" threshold and keep search-
2 The term transportation cost originates from the spatial in- ing until a satisfactory product is located. Con-
terpretation of the unit circle model, where the buyer must
sequently, customers with lower search costs
travel to the seller's location or have the product shipped. the become more demanding and are willingto make
larger the distance between buyer and seller, the higher will
fewer compromises concerning their ideal pro-
be the transportation cost. In the context of the model duct. Buyers are better off in two ways: first, they
presented here, the transportation cost formalizes the fact that enjoy lower prices because of the increased com-
the more a product offering differs from a buyer's ideal petition among sellers; second, and potentially
preference, the less desirable it become for that buyer. more important, they enjoy allocational efficien-

MIS Quarterly/September 1991 301

ElectronicMarketplaces

cies from being better informed about the they have been increasinglyprominentin recent
available products,thus makingpurchases that years. The two dominantsystems were devel-
better suit their needs. oped by AmericanAirlines(SABRE)and United
Airlines(Apollo).3The precedingdiscussion sug-
gests that these systems should be expected to
Role of electronic marketplaces promote price competition,reduce the airlines'
It was argued that an electronic marketsystem market power, and result in more demanding
in a differentiatedmarketis likelyto promoteprice customerswho are less willingto compromiseon
competition and reduce the market power of theirpreferredproduct.Airlineshave experienced
sellers. It may thus create a net welfare gain by all of these impacts (Smith, 1987).
lowering the search cost of buyers and also We saw thatin a differentiated market,sellers can
enablingthemto locate productsbettermatching take advantageof buyers'search costs. Although
their needs. If the search cost becomes low no formaldistinctionwas made between the role
enough, buyers willlook at all productofferings of priceand productinformation,it is actuallythe
and purchase the one best serving their needs, priceinformation thatis mostdamagingto sellers'
resultingin a sociallyoptimalallocation.Itfollows profits.Inthat context, sellers have an incentive
thatsuch a system willbe sociallydesirablewhen to manipulateelectronic marketplaces in order
these welfare gains outweigh its development to increase the cost of obtainingprice informa-
cost. tion.Whenthis is combinedwithreadilyavailable
information on product characteristics, it
It is interestingto note that when informational
inefficiencies prevail,a large numberof sellers discourages buyers from searching for price
deals and results in higherprofitsforthe sellers.
does not need to resultin a competitiveand effi- Inthe airlineindustry,forexample,Americanand
cient market. Under certain circumstances the United attempted to bias CRS screen displays
marketmay become more monopolisticas the to discouragepricecomparisons;in a similartac-
number of sellers increases! This can happen,
tic, most airlines now offer a wide gamut of
for example, if a buyer can acquire information thousandsof active fares and promotionsto con-
about a product only by purchasing it. In this fuse these comparisons.Anotherapproachis to
case, buyers often buy fromthe firstseller they increase the degree of product differentiation,
visit. As a result, individualsellers do not have whichwas the majorobjectivein the introduction
a strong incentiveto lowertheir prices because of frequentflierprograms.Finally,sellers and in-
they would attractfew buyers, and even these termediariesoperatingelectronicmarketsystems
buyers may interpretthe lowerprice as a signal can appropriate some of the buyers' gains
of poor quality. As the number of sellers in-
throughuser fees. Forexample, airlinesare im-
creases, it becomes even moredifficultforbuyers
to locate any specific discounter; This further posing transactionfees beyond a certainlevel of
CRS utilization, both to generate additional
decreases the incentives of individualsellers to revenues and to discourage searching for the
cut theirprices and thus results in a more mono-
lowest-pricedfares (Dahl, 1991).
polistic market.This behavioris likelyin certain
marketswithlittleor no advertisingand no cheap Increased competition,fueled by the deregula-
way to assess qualityex ante (or even ex post), tion of the airlineindustryand the computerized
e.g., professionalmarketsfor legal and medical reservationsystems, was adverselyaffectingthe
services. In a setting of this type, electronic profitabilityof flightoperationsfor most airlines
marketsystems could disseminateproductinfor- in the early 1980s. The airlines that pioneered
mation (e.g., through a ratingservice that pro- reservation systems, however, were enjoying
motes the sharingof buyerexperiences withthe high profits from system-related revenues. In
product)as well as providepricecutterswiththe 1980-1982, AmericanAirlinescommanded a 40
means to reach a largerfractionof the buyers; percent gross marginon its SABRE revenues,
the monopolisticnatureof these marketscould while flightoperations yielded as littleas 5 per-
be underminedas a result. cent. TWAmade moremoneyin the same period
Airlinecomputerizedreservationsystems (CRS)
fit well the model of electronic marketsystems 3 Unitedspun offApollointoa subsidiarynamedCovia,which
in differentiatedmarketsdiscussed earlier,and is now owned jointlyby several other airlines.

302 MIS Quarterly/September1991

 ElectronicMarketplaces

on PARS (its reservation system, which eventual- by lowering buyer search costs and thus reduc-
ly was merged into WorldSpan in a joint venture ing sellers' profits and market power. This may
with Northwest and Delta) than from its airline. create a problem for sellers in markets where
As a result, airlines with established reservation technological developments make a market
systems enjoyed stock price premiums as recent- system feasible, or even imminent. In the long
ly as mid-1987 (Smith, 1987). As these systems run it may be impossible to avoid some loss of
have started competing among themselves, market power, especially when this power is
however, operating margins have been falling, based on exploiting high buyer search costs.
and current SABRE profit margins have declined System revenues may compensate for this
to the low teens. Air travelers are the only group decline in the short-run, but, as several systems
made unambiguously better off as a result of are introduced, their profits are likely to be com-
these systems. peted away. In SABRE's case, for example, even
customer lock-in through legally binding con-
Finally, it may be interesting to note that the im- tracts did not prevent the eventual loss of market
plications of search costs for differentiated
share; competitors would offer to foot the travel
markets, discussed above, are consistent with
the traditional transaction cost analysis of ap- agencies' legal bills to induce them to switch to
their systems (Smith, 1987). Ifthe sellers collude
propriate governance structures. It was sug- to prevent an electronic market system, they may
gested, for example, that market inefficiency induce buyers to introduce such a system them-
(deviation from a Pareto optimal outcome) in-
creases when the cost of obtaining information selves, or they may encourage a third-party in-
about a specific product offering is high, which termediary to enter the picture.
is typically the case for complex products. Ifthis The potential profits of running an electronic
inefficiency has resulted in hierarchical gover- marketplace can actually put sellers in a
nance structures, an electronic marketplace may prisoner's dilemma situation. We saw that sellers
promote market-based governance mechanisms, as a group have no incentive to introduce a
suggesting that electronic marketplaces may system offering price and product information,
favor markets over hierarchies, in accordance which makes them all worse off. Yet each of them
with Malone, et al.'s (1987) prediction. individually might benefit from introducing such
a system because the revenues that can be
Strategic Conduct in gained by charging buyers directly and indirect-
Electronic Marketplaces ly for the system's services would probably
outweigh the lost profitsof an individualseller due
The preceding sections discussed how in- to increased competition. This may explain why
terorganizational informationsystems can create sellers often precipitate the introduction of elec-
"electronic marketplaces" by serving as in- tronic market systems, in spite of the potential
termediaries between the buyers and the sellers of these systems to reduce their market power.
in a vertical market, reducing the buyers' search
costs in the process. Also discussed were the im- The best strategy for sellers may be to control
plications of this reduction in search costs and the type of system that is eventually introduced.
the potential of electronic marketplaces to affect If they orchestrate the introduction of systems
the efficiency of interorganizational transactions that emphasize product rather than price infor-
and the market power of buyers and sellers. We mation, buyers will use these systems to locate
now turn to the strategic potential of these the most appropriate product in the market.
systems and inquire whether their economic Sellers would be able to maintain their profits and
characteristics are likely to favor early movers in addition appropriate some of the buyers'
and large intermediaries. benefits through user charges. Such systems
may be able to support a certain degree of im-
plicit collusion and thus maintain a relatively
Marketpower and the introductionof cooperative outcome: other sellers could "get the
electronic marketplaces message" and stay away from systems that em-
phasize price information. Established systems
It was argued in the previous two sections that can raise the barriers of entry for third-party in-
electronic marketplaces usually favor the buyers formation providers and thus delay or avoid the

MIS Quarterly/September 1991 303

ElectronicMarketplaces

competition from systems offered by non-industry This effort faces several obstacles, however,
participants. because the Electronic Joint Venture group is
likely to encounter conflicts of interest with cur-
Naturally, buyers have the opposite incentives rent or potential members. For example, Citibank
and would like to create an electronic market- owns Quotron, and any attempts to recruit Mer-
place that facilitates comparisons among sellers' rillLynch could be stymied by the latter's interest
prices as well as product offerings. The func- in Bloomberg Financial Markets; both Quotron
tionality of an electronic market system is and Bloomberg would be likely competitors with
therefore likely to depend on whether it is in- an EJV system. Furthermore, there is fragmen-
troduced by a buyer or a seller. In a market that tation among the sellers in the bond market
is more concentrated on the seller side, like most because a group of bond brokers is planning to
consumer and several industrialmarkets, a single introduce its own proprietary system; this will
buyer may be too small to introduce a market result in competition between six or more
system, may not be sufficiencly interested, or systems serving the bond market.
may lack the clout to induce seller participation.
Sellers, on the other hand, have higher stakes
in such markets, leading to stronger incentives Early mover advantage
to introduce a market system. This provides The focus thus far has been on the impact of in-
another explanation for the observation that most troducing an electronic market system on market
electronic market systems are pioneered by prices and profits. As pointed out earlier,
sellers, although, theoretically, such systems are however, electronic marketplaces may be able
most beneficial to the buyers. Buyers may form to generate substantial efficiencies, which can be
coalitions, however, and introduce jointly owned translated into profits for the system operators.
systems, or a third party organization with the This is likely to create interest among potential
necessary technological expertise may enter the intermediaries, thus introducing competition to
market. the market for electronic intermediation. A well-
known result of microeconomic theory is that
It is more difficult for sellers to control the out- firms in competitive markets earn no profits in the
come in commodity markets. In the case of the
long run, except for a fair return on their capital
domestic government bond market, big bond assets. Excess profits are possible, however,
dealers refrained from introducing an electronic under such deviations as informationalinefficien-
market system, which allowed financial data ven-
cies, the ability of some firms to create and ex-
dors to make inroads in that market, as discussed
ploit barriers to entry, or the technological
earlier. In order to reassert its position, Merril
sophistication that allows other firms to enjoy pro-
Lynch, which has a minority interest in Bloom- fits from innovation even if they lack any signifi-
berg Financial Markets, has transformed this cant market power. In this context, three
system into an electronic marketplace by mak- characteristics of electronic marketplaces iden-
ing it available to other dealers (Clemons, 1991). tified earlier may allow intermediaries that move
Bloomberg has become the premier source of early in offering electronic market systems to
real-time data for trading in corporate bonds and maintain a competitive advantage: large re-
also gets price feeds for government bonds
quirements for capital investment resulting in
through the primary dealers. Meanwhile, other substantial economies of scale and scope,
big bond dealers formed the Electronic Joint Ven- switching costs imposed on their participants,
ture (EJV) group, which came to be known as the and network externalities.
"Gang of Six" and currently includes Salomon,
Goldman Sachs, Morgan Stanley Group, CS First
Capital Investment
Boston, Citicorp, Shearson Lehman Hutton, and
possibly others. It has been reported that group The large investments in sunk and fixed costs
members are ready to commit up to $100 million that are required to develop electronic market
to launch their own electronic marketplace, and systems are likely to play an important strategic
they plan to provide research, analysis, and role. These costs can raise barriers to entry
related news in an attempt to differentiate (Spence, 1979) and can offer advantage to early
themselves from the other competing systems movers (Fundeberg and Tirole, 1983; 1985). In-
(Herman and Power, 1990). cumbent firms may overinvest in the beginning

304 MIS Quarterly/September 1991

 ElectronicMarketplaces

and keep investingsufficientlyto discouragenew build a locked-in customer base that can be
entrants(Bernheim,1984).Traditionally, the final subsequently exploited.
outcome in such competitionbased on strategic It is often assumed in the informationsystems
investment is determined by the underlying literature that interorganizationalinformation
economies of scale, which typically favor
established entrants(Coursey,et al., 1984), and systems create substantial switching costs
because of sunk investments in hardware,soft-
by the uncertaintyabout technological and de- ware, user training,and organizationalchanges,
mand characteristics(Arvan,1986). Inthe case as well as non-technologybarrierssuch as trust
of electronicmarketplaces,networkexternalities in organizationalpartnersor long-termcontracts
(discussed later)are likelyto be just as signifi- (e.g., Bakos and Treacy,1986; McFarlan,1984).
cant, if notthe dominantfactor.Uncertaintyabout The larger the switching costs, the fiercer the
the actual capacityof the marketmay induce in-
termediaries to offer more systems than the competitionto recruituncommittedusers and the
marketcan support;in that case consolidation largerthe proportionof system benefits appro-
and a war of attritionare the likelyoutcomes. priated by the intermediariesin the long run.
On the other hand, the abilityof switchingcosts
The dominance of American's SABRE and
to create early mover advantage in electronic
Covia's Apolloin the marketfor airlinereserva-
tion systems illustrates the significance of marketplacesis mitigatedby technologicalpro-
gress whentechnologyfacilitatingaccess to com-
strategic investment. These systems are pro- petitivesystems becomes availableor when the
fitable,yet theirunderlyingtechnologycould cer-
changeover to a new generation of the techno-
tainly be duplicated by a sophisticated
logy offers an opportunityto switch to a new in-
intermediary.Itis also unlikelythata new entrant termediary.Similarly,the arrivalof new potential
could legallybe denied access to the needed in-
users can spark competitionamong intermedi-
formation,and, subject to contract-termlimita- aries to recruitthese users to their systems; if
tions recently established by the U.S.
intermediaries cannotprice-discriminate between
government, it takes only 30 days for a travel new and existingusers, theirabilityto profitfrom
agent to switchto a new reservationsystem (Hop-
switchingcosts is significantlyreduced. Inview
per, 1990). The economics of the large com- of mountingevidence that system participants
munication networks and massive transaction
are often able to overcome highswitchingcosts,
processing centers requiredto support a CRS an interestingtopicforfutureresearchinthis area
results in start-upcosts of hundredsof millions
is the indentificationof the switching costs im-
of dollars for a new entrantin that market.The
saturation of the existing market (virtuallyall posed by different types of electronic market
travelagents subscribe to a CRS), the relatively systems on theirparticipantsand the integration
of these resultswiththe economictheoryof swit-
low marginal costs enjoyed by the existing
ching costs.
players, and the possibility of a retaliatory
response create enough uncertaintyabout the
benefits of entryto deter new systems and limit
the aspirationsof marginalplayers. Network Externalities
Electronicmarketsystems with large installed
bases create more value for their participants,
Switching Costs
who are providedwitha widerselection of poten-
Firmsconnected to an electronicmarketsystem tial buyers and sellers. These "networkexter-
mayface substantialtechnologicaland organiza- nalities"create an early moveradvantage (Katz
tional costs if they decide to switch to an alter- and Shapiro,1985) because early movers enjoy
nativesystem; such switchingcosts can play an the opportunityto builda larger installed base.
importantstrategicrole.The studyof the strategic Unless the technology evolves in a way that
implicationsof switchingcosts has emerged as subsequentlypenalizes earlymoversor removes
an active field in economics withthe workof Far- access barriersbetween differentsystems, late
rell(1987), Farrelland Shapiro(1987),Klemperer movers are at a disadvantage.
(1986; 1987a; 1987b),and von Weizsaker(1984).
Analyticmodelsof switchingcosts usuallypredict Networkexternalitiescan interactwithswitching
aggressive behaviorof early movers, who tryto costs to reinforceeach other, increasingthe ag-

MIS Quarterly/September 1991 305

ElectronicMarketplaces

gressiveness of intermediariesin the earlystages Citibank,for example, was a leader in introduc-

of introductionand the advantage of successful ing automatedtellermachines(ATMs)in the New
earlymovers. Networkexternalitiesmake estab- Yorkmarket.Its extensive ATMnetworkallowed
lished systems more attractive to new users, it to lower its operatingcosts and played an im-
reducingthe need for intermediariesto compete portantrole in the revivalof Citibank'sIndividual
for these users on a price basis. This is signifi- BankingUnit, which enjoyed an increase in its
cant because competitionfor newcomers is one retailmarketshare from4 percent in 1977 to 13
of the majorchecks inthe abilityof intermediaries percent in 1988. However,the strategic impact
to exploit switching costs. of this ATMnetworkdid not last long. Citibank's
commandingtechnological leadership induced
the other major consumer banks to cooperate
and offer a competing networkof compatible
Sustainability of Early Mover Advantage ATMs.This seems to have left Citibankwithout
Experiencesuggests that any advantagegained a durableadvantagefromits ATMnetworkas its
by movingearly in offeringan electronicmarket competitorswere able to matchits installedbase
system will most likelycreate only a windowof of ATMs(Glaser, 1988).
opportunity. Advantage based purely on Inthe eastern Pennsylvaniamarket,GirardBank
technologicalsophisticationis difficultto sustain started installingATMmachines in 1975, and it
because of the open and rapidlyevolvingnature
of the technology. When interorganizationalin- soon introducedits proprietaryGeorge ATMnet-
formationtechnology becomes commonplace, work, expecting to capitalize on its traditional
profitsare competed away and the technology strength in the retail market. PhiladelphiaNa-
becomes a "strategicnecessity" (Clemons and tionalBank(PNB)rapidlyresponded to Girard's
Kimbrough,1986). Sustainableadvantagebased GeorgeATMnetworkwithMAC,its ownATMnet-
on informationtechnology typically requires work; although PNB was not able to match
leveraginguniqueresourcesthatcannotbe easily George'sATMbase on its own, it made participa-
tion in MACavailableto otherbankssince MAC's
replicatedor leapfroggedby potentialcompetitors initiallaunch. In 1982 Girardwas acquired by
(Clemonsand Row, 1987), or continuousinnova-
tion that will keep the system a moving target MellonBankof Pittsburgh;George was merged
with Mellon's Cashstream and evolved into a
beyond the reach of its competitors.Inthe case shared ATMnetworkas well (Clemons, 1990).
of electronic marketplaces,specific industryex-
pertise, a locked-incustomer base, or the ability Possiblyas a resultof the rapidresponses in this
to deal with certain organizationaland system market,these ATMnetworksgenerally did not
complexitiesare promisingareas to lookforsus- provideany competitiveadvantage to their par-
tainable advantage. ticipants,as confirmedin a study by Bankerand
Kauffman(1988); the systems had become a
Early movers may be stranded with a bad "strategic necessity."
technologyand offera system thatsoon becomes
obsolete, diminishing the advantage of their Inthe airlinereservationsmarket,CRS operators
strategic investment.Technologicaland market have lost theirabilityto impose severe switching
developments, such as fallingprocessing costs costs and have seen the benefit of networkex-
and the increasingavailabityof software-defined ternalitiesdwindlebecause of governmentregu-
networks (SDNs) and the integrated services lations limiting contract terms and requiring
digital network(ISDN),may reduce the cost of airlinesto equallyprovidetheirflightinformation
the infrastructurerequiredto offer an electronic to all intermediaries.Malone,et al. (1989) have
marketplace.Earlymovers may also findto their arguedthatthis abilityto benefitfroman uneven
chagrinthattechnologyhas been developedthat playing field tilted to the intermediary'sadvan-
reduces the switchingcosts they workedhardat tage will eventually diminish in all electronic
imposingand diminishesthe roleof the standards marketplaces; intermediaries who attempt to
they have established. Finally,a successful ear- keep theircustomers at a disadvantagewillsee
ly system enjoying network externalities may their marketshare dwindle. Responding to this
leave no alternativeto otherindustryparticipants trend,and in an attemptto sustain its competitive
but to forma coalitionand offer a credible com- advantagethroughcontinuousinnovation,Amer-
peting system. ican Airlinesis shiftingits focus fromCRS opera-

306 MIS Quarterly/September 1991

 ElectronicMarketplaces

tion to superior utilization of the information in sunk costs and specialized expertise; big in-
provided by the SABRE system (Hopper, 1990). termediaries can leverage this investment over
a larger number of system participants; (2)
In the financial industry, Barclays de Zoete Wedd
system development, which is often character-
(BZW) has leveraged its unique ability to provide ized by a steep learning curve that allows the
coverage for the widest range of securities in the
London market. It enjoys a sustainable advan- development of subsequent systems at a smaller
cost; (3) economies of scope, especially in
tage from its TRADE system for automatic order
execution (Clemons and Weber, 1990a). This ad- development expertise, the sharing of operational
facilities, and data collection (where data col-
vantage is likely to persist as long as BZW's lected during system operation becomes a valu-
coverage advantage can be maintained, a re- able asset); and (4) network externalities as the
markable achievement in view of the fact that the
number of participants in an electronic market-
deregulation of London's financial markets,
known as the Big Bang, has resulted in a drastic place increases and the market becomes more
decrease of revenues for London brokerage firms successful, providing more benefits (e.g., liquidi-
ty) to its individual participants.
(Clemons and Weber, 1990b). Similarly, Bloom-
berg Financial Markets and the EJV group hope The typical strategy to secure economies of scale
to control real-time prices for bond issues trad- and scope in intermediating electronic
ed outside the exchanges by virtue of their posi- marketplaces is to achieve dominant market
tion as the major dealers in several issues; if they share in an industryor provide intermediationser-
succeed in defending this control against regu- vices across a number of industries. As traditional
latory challenges, their systems could maintain technology with low entry costs is replaced by
a sustainable advantage compensating them for systems based on informationtechnology, which
the reduction in their trading margins. has large fixed costs, the new economics of in-
termediation may favor firms with access to a
Clemons (1990) proposes system topology as an
wealth of resources; this can leverage their know-
important factor affecting the ability of electronic how in different industries and defray their
marketplaces to generate profits for their opera-
development expenses among several systems.
tors; he argues that the point of customer con- Firms with related organizational and techno-
tact is a critical resource in determining the
logical expertise are attempting to build on their
appropriability of system benefits and the sus- customer bases, establish themselves as infor-
tainability of these profits. Intermediaries are mation utilities, and dominate the provision of in-
more likely to appropriate system benefits, and
termediation services. IBM, General Electric
sustain this appropriation, in systems where they
control the market transactions and the access (through its GEISCO subsidiary), and General
Motors (through its EDS subsidiary) are emerg-
to customers (such as the airline reservation
ing as major players among the firms position-
systems); these benefits will be less sustainable
in systems where the suppliers or the customers ing to compete in this arena.
retain control over individual transactions and
These information intermediaries are likely to
provide themselves with the link to the in-
enter individual markets in partnerships with in-
termediary (as is the case with inter-bank ATM
networks). dustry participants that can provide industry-
specific expertise. Conversely, buyers and sellers
in a number of industries are discovering that
they cannot develop these systems competitive-
Role of informationintermediaries ly without the help of a partner with systems ex-
Economies of scale and scope in the develop- pertise. Even well-established, sophisticated
ment of electronic marketplaces and in the pro- players like the providers of airline computer
vision of electronic intermediation services may reservation systems are succumbing to this
become an important element of the competitive trend. As maintenance costs increase and
game. Four areas where such economies may systems require major upgrades to incorporate
arise include the following (Bakos, 1991): developments in communications and user inter-
(1) building and managing systems of substan- face technology, airlines find that they need part-
tial size and functionalityrelying on complex com- ners. For example, Texas Air has agreed to sell
munication networks, requiring large investments to EDS 50 percent of its System One (the fourth-

MIS Quarterly/September 1991 307

ElectronicMarketplaces

largest CRS) for over $250 millionas partof an attempt has been made to address their most
agreement that would outsource to EDS the importantstrategic implications.
operation of most of Continental's computer Withthe notable exception of the economics of
systems. AT&Tis negotiatingwithTWA,North-
west, and Delta for a piece of third-largest system development, the applicabilityof these
economic characteristicsto electronic market-
WorldSpan.Even AMR(the parentof American
places has not been empiricallydocumented in
Airlines)has indicateda willingnessto sell up to the informationsystems literature.Althoughwe
50 percent of SABRE,the largest CRS, at $15
assumed the validity of these characteristics
millionfor each 1 percent, placing on it a total
based on existing anecdotal evidence, a formal
value of $1.5 billion(Business Week, 1990).
verification of their relevance for electronic
A substantialpartof the economies of scale and marketplaces and other types of information
scope enjoyed by large intermediariescomes systems is an important direction for future
fromthe computerand communicationsnetworks research.
underlyingall electronicmarketplaces.Ifaccess Electronicmarketplacesare a fact of lifeand are
to the necessary infrastructurebecomes avail-
becoming more prevalentevery day. Economic
able, players with expertise in particularin-
dustriesmay introducemarketsystems, resulting theorysupportsthe commonargumentthatthese
in a proliferationof such systems. The govern- systems hold great promise for improvingin-
terorganizationalcoordinationin marketsettings.
ment-sponsored infrastructureof the Minitel These economic efficiencies can create poten-
system in France illustratesthis possibility,hav- tial opportunitiesfor informationintermediaries;
ing sparkedthousandsof intermediation services
yet whenthe technologybecomes commonplace,
rangingfrommatchingbuyersand sellers of vin- profitswillbe competed away for intermediaries
tage wines to brokeringindustrialparts. who have not achieved some formof sustainable
Similarlyin the U.S. market,intermediarieslike advantage. The underlyingeconomies of scale
AT&T,MCI,SPRINT,the regional phone com- mayenable certainfirmsto leveragetheirsystem
panies (RBOCs),and otherthird-party value add- development expertise, installed networks,and
ed networkprovidersare increasinglyoffering customer bases in orderto become information
access to their networkinfrastructuresthrough utilitiesand possibly dominate the provisionof
expanded telecommunicationsservices, such as intermediationservices.
high-capacityswitched digital services, ISDN, Electronicmarketsystems are likelyeventually
and completecustomizednetworkmanagement. to become a strategic necessity and part of an
As networkservices become commodityofferings
with relativelylow fixed charges, they are likely industry's infrastructure;it seems that neither
size nor being the first mover will guarantee a
to cease being a majorfactor in the provisionof sustainable advantage or the appropriationof a
electronic marketsystems. The advantage cur- favorableshare of system payoffs.Clemons and
rentlyenjoyed by big intermediariesis thus like- Row's(1987)viewthatsustainableadvantagere-
ly to dissipate, except possibly in the offeringof quiresthe controlof unique resources has been
verycomplexlarge-scalesystems spanningwide supported by a number of case studies of in-
geographical areas. terorganizationalsystems, such as the ones by
Clemons(1990)and Clemonsand Weber(1990a;
1990b).Intermediarieswho do not achieve some
Conclusion formof sustainable advantage can tryto exploit
profitoportunitiesforas longas they last and may
This articlefocuses on the potentialof electronic attemptto controlthe transitionto a more com-
marketplacesto reduce buyer search costs as petitiveenvironmentand bias the finaloutcome
their salient characteristic. It identifies certain in their favor.
other characteristicsas relevant to a strategic
analysis of these systems, namely networkex-
ternalities,technological uncertainty,switching References
costs, and economies of scale and scope. While Arvan,L. "SunkCapacityCosts, Long-RunFixed
a detailedtheoreticaltreatmentof these charac- Costs, and EntryDeterrenceunderComplete
teristics lies outside the scope of this article,an and IncompleteInformation,"Rand Journal

308 MIS Quarterly/September1991

 Electronic
Marketplaces

of Economics (14:1), Spring 1986, pp. formationSystems, San Diego,CA,December

105-121. 1986, pp. 181-194.
Bakos, J.Y. "InterorganizationalInformation Clemons, E.K. and Row, M. "StructuralDif-
Systems: StrategicOpportunities forCompeti- ferences AmongFirms:A PotentialSource of
tion and Cooperation," Ph.D dissertation, CompetitiveAdvantage in the Applicationof
Sloan School of Management,Massachusetts Information Technology,"Proceedingsof the
Institute of Technology, Cambridge, MA, EighthInternationalConference on Informa-
November 1987. tionSystems, Pittsburgh,PA,December1987,
Bakos, J.Y. "InformationLinks and Electronic pp. 1-9.
Marketplaces:The Role of Interorganizational Clemons, E.K. and Weber, B.W. "Barclaysde
InformationSystems in Vertical Markets," Zoete Wedd's TRADE:Evaluatingthe Com-
Journal of MIS, forthcoming,Fall 1991. petitive Impact of a Strategic Information
Bakos, J.Y. and Kemerer,C.F. "RecentApplica- System," Proceedings of the 23rd HawaiiIn-
tions of Economic Theory in Information ternationalConferenceon System Sciences,
Systems Research,"workingpaper,Graduate Kailua-Kona,HI,January 1990a.
School of Management,Universityof Califor- Clemons, E.K.and Weber, B.W. "London'sBig
nia, Irvine,California,October 1990. Forth- Bang: A Case Study of Information
coming in Decision Support Systems. Technology, Competitive Impact, and
Bakos, J.Y. and Treacy, M.E. "Information Organizational Change,"Journalof MIS(6:4),
Technology and Corporate Strategy: A Spring 1990b, pp. 41-60.
Research Perspective,"MISQuarterly(10:2), Coursey, D., Isaac, R.M., Luke, M, and Smith,
June 1986, pp. 107-119. V. "MarketContestabilityin the Presence of
Banker,R. and Kauffman,R. "StrategicContribu- Sunk (Entry) Costs," Rand Journal of
tions of InformationTechnology:An Empirical Economics (15), 1984, pp. 69-84.
Studyof ATMNetworks,"Proceedings of the Dahl, J. "Agents Rankle Airlines With Fare-
NinthInternational Conferenceon Information Checking Programs," Wall Street Journal,
Systems, Minneapolis,Minnesota,December May 20, 1991, p. B1.
1988, pp. 141-150. Diamond,P.A. "A Modelof Price Adjustment,"
Barrett,S., and Konsynski,B. "Inter-Organization Journal of Economic Theory (3), 1971, pp.
InformationSharingSystems." MISQuarter- 156-168.
ly, Special Issue, 1982, pp. 93-105. Economist. "No Computers for Euro-bonds,"
Bernheim, B.D. "Strategic Deterrence of Se- May 30, 1987, p. 14.
quential Entryinto an Industry,"Rand Jour- Farrell,J. "Competitionwith Lock-in,"Working
nal of Economics (15), 1984, pp. 1-11. Paper 8722, Department of Economics,
Business Week. "Wanted:Co-Pilotsfor Reser- Universityof California,Berkeley,CA,January
vation Systems," April9, 1990, pp. 78-79. 1987.
Cash, J.I. and Konsynski,B. "IS RedrawsCom- Farrell,J. and Shapiro, C. "DynamicCompeti-
petitive Boundaries," Harvard Business tion with Lock-in," Discussion Paper 121,
Review(63:2),March-April 1985, pp. 134-142. WoodrowWilsonSchool, PrincetonUniversi-
Chamberlin,E.H. The Theory of Monopolistic ty, Princeton,NJ, February1987.
Competition,HarvardUniversityPress, Cam- Fundeberg,D. and Tirole,J. "Capitalas a Com-
bridge, MA, 1933. mitment:StrategicInvestmentto DeterMobili-
Clemons, E.K. "MAC-Philadelphia National ty," Journal of Economic Theory (31:2),
Bank'sStrategicVenturein SharedATMNet- December 1983.
works,"Journalof MIS(6:5), Summer 1990. Fundeberg, D. And Tirole,J. "Preemptionand
Clemons, E.K. "Evaluation of Strategic In- Rent Equalizationin the Adoption of New
vestments in Information Technology,"Com- Technology," Review of Economic Studies
municationsof the ACM(34:1),January1991, (52), 1985, pp. 383-401.
pp. 22-36. Gastwirth,J.L. "On ProbabilisticModelsof Con-
Clemons, E.K.and Kimbrough,S. "Information sumer Search for Information,"unpublished
Systems, Telecommunications,and their Ef- paper, 1971.
fects on IndustrialOrganization,"Proceedings Glaser, P.F. "UsingTechnologyforCompetitive
of the Seventh InternationalConferenceon In- Advantage:The ATMExperienceat Citicorp,"

MIS Quarterly/September1991 309

ElectronicMarketplaces

in Managing Innovation: Cases from the Ser- Salop, S. "Monopolistic Competition With Out-
vices Industries, B.R. Guile and J.B. Quinn side Goods," Bell Journal of Economics (10),
(eds.), National Academy Press, Washington, 1979, pp. 141-156.
DC, 1988, pp. 108-114. Salop, S. and Stiglitz, J.E. "Bargains and Ripoffs:
Hansell, S. "The Wild, Wired World of Electronic A Model of Monopolistically Competitive Price
Exchanges," InstitutionalInvestor, September Dispersion," Review of Economic Studies
1989, p. 92. (44), October 1977, pp. 493-510.
Herman, T. and Power, W. "Big Bond Dealers Salop, S. and Stiglitz, J.E. "The Theory of Sales:
Plan to Provide Price, Data Service," Wall A Simple Model of Equilibrium Price Disper-
Street Journal, March 23, 1990, pp. C1, C9. sion with Identical Agents," American
Hopper, M.D. "Rattling SABRE-New Ways to Economic Review, December 1982, pp.
Compete on Information," Harvard Business 1121-1130.
Review (68:3), May-June 1990, pp. 118-125. Schmalensee, R. "Entry Deterrence in the
Hotelling, H. "Stability in Competition," Ready-to-Eat Breakfast Cereal Industry," Bell
Economic Journal (39), 1929, pp. 41-57. Journal of Economics (9), 1978, pp. 305-327.
Katz, M.L. and Shapiro, C. "Network Exter- Shilony, Y. "Mixed Pricing in Oligopoly," Jour-
nalities, Competition and Compatibility," nal of Economic Theory (14), 1977, pp.
American Economic Review (75), Spring 1985, 373-388.
pp. 70-83. Smith, R. "AMR's Outlook is Clouded as Com-
Klemperer, P.D. "Markets with Consumer petition Erodes Profit of Airline Unit's SABRE
Switching Costs," Ph.D. dissertaton, Stanford System," Wall Street Journal, May 28, 1987,
University, Stanford, CA, September 1986. p. C5.
Klemperer, P.D. "The Competitiveness of Spence, A.M. "Investment Strategy and Growth
Markets with Switching Costs," Rand Journal in a New Market," Bell Journal of Economics
of Economics (18:1), Spring 1987a, pp. (10), 1979, pp. 1-19.
138-150. Stigler, G. "The Economics of Information,"Jour-
Klemperer, P.D. "Markets with Consumer nal of Political Economy (69), June 1961, pp.
Switching Costs," Quarterly Journal of 213-225.
Economics, May 1987b, pp. 375-394. von Weizsaker, C.D. "The Costs of Substitution,"
Malone, T.W. Yates, J., and Benjamin, R.I. Econometrica (52), 1984, pp. 1085-1116.
"Electronic Markets and Electronic Hierar-
chies: Effects of Information Technology and
Market Structure and Corporate Strategies,"
About the Author
Communications of the ACM (30:6), June J. Yannis Bakos is assistant professor of
1987, pp. 484-497. management at the Graduate School of Manage-
Malone, T.W., Benjamin, R.I., and Yates, J. "The ment, University of California, Irvine. He receiv-
Logic of Electronic Markets," Harvard ed his B.S. and M.S. degrees in electrical
Business Review (67:3), May-June 1989, p. engineering and computer science from MITand
166. holds an M.S. in management and a Ph.D. from
McFarlan, F.W. "Information Technology MIT's Sloan School of Management. He focuses
Changes the Way You Compete," Harvard his teaching and research on opportunities for the
Business Review (62:3), May-June 1984, p. strategic use of information technology. He has
98. special interest in the application of industrial
Perloff, J.M. and Salop, S.C. "Equilibrium with organization theory to the study of interorganiza-
Product Differentiation," Review of Economic tional information systems and their competitive
Studies (52), 1985, pp. 107-120. implications. Professor Bakos also tracks and
Reinganum, J.F. "A Simple Model of Equilibrium forecasts developments in end-user computing
Price Disperson," Journal of Political and corporate informationsystems architectures,
Economy (87), 1979, pp. 851-858. especially in view of advances in communications
Rothschild, M. "Searching for the Lowest Price technology. He lectures primarilyin the areas of
When the Distribution Price is Unknown," corporate telecommunication networks, informa-
Journal of Political Economy (82), 1974, pp. tion systems outsourcing, and the use of infor-
689-711. mation technology for competitive advantage.

310 MIS Quarterly/September 1991