1

Online Auctions: A Closer Look*

Alok Gupta

Associate Professor

Information and Decision Sciences Department

Carlson School of Management

University of Minnesota

3-365 Carlson School of Management

321 - 19th Avenue South

Minneapolis, MN 55455

alokgupta@acm.org

Ravi Bapna

Assistant Professor

MIS Area

College of Business Administration

214 Hayden Hall

Northeastern University, Boston, MA 02115,

bapna@mediaone.net

(May 2001)

Published in: Handbook of Electronic Commerce in Business and Society, Boca

Raton, FL: CRC Press, 2002.

*First author’s research is supported by NSF CAREER grant IIS-0092780.

2

1. Introduction to Online Auctions

Online auctions represent a model for the way the Internet is shaping the new

economy. In the absence of spatial, temporal and geographic constraints these

mechanisms provide many benefits to both buyers and sellers. They are now an

important component of the portfolio of mercantile processes that are

transforming the economy from traditionally hierarchical to market oriented

structures [see Kauffman and Walden (2001) for an exhaustive review]. A broad

and deep body of economics literature exists that investigates the theoretical

properties of traditional auctions. However, significant differences in the cost

structures, to both buyers and sellers, participating in online auctions, have

resulted in a need to revisit much of the existing theory. This chapter provides a

broad context, derived from an overview of the current research and practice in

this field, and provides insights into this interesting sphere of economic activity.

Online auctions fall under the ambit of web based dynamic pricing

mechanisms. In these mechanisms, consumers become involved in the price-

setting process. Consumers can now experience the thrill of ‘winning’ a product,

potentially at a bargain, as opposed to the typically relatively tedious notion of

‘buying’ it. For sellers these mechanisms are likely to bring access to newer

markets, help clear aging or perishable inventory, and provide experiential and at

times viral marketing capabilities.

Nowhere are these trends as visible as in the hugely popular online

auction site, eBay. Among other things eBay has resulted in dramatically

improving the efficiency of secondary markets that were typically associated with

3

garage sales and flea markets. eBay’s legion of 10 million monthly visitors

provides the necessary critical mass of buyers and sellers to set market prices

for their goods. The more bids that come in, the more competition there is, and

chances are that higher the price. In retrospect, eBay had and continues to have,

both the positive network externality effect of a growing user base as well as the

first mover advantage, necessary and sufficient conditions for success in today’s

economy [Varian (2000)].

The impact is even more dramatic in business-to-business (B2B) markets

where Forrester Research predicts an increase in sales from $19.3 billion last

year to $52.6 billion by 2002. A full suite of dynamic pricing mechanisms is in use

in B2B markets, including standard auctions where there is a single seller and

multiple buyers, reverse auctions where a single buyer receives bids from

multiple sellers and multiple buyer, and multiple seller exchanges that resemble

the bid-ask markets for stocks and commodities. Mollman (2000) presents an

overview of the top performing B2B auctions.

Beginning with the dotcom euphoria of 1999, one can observe the

emergence of a myriad collection of price-setting processes, such as traditional

first-price auctions for single items (e.g. eBay.com), multi-item auctions selling

multiple identical units (e.g. Onsale.com and eBay’s Dutch auction), reverse

auctions for goods and services (e.g. eLance.com and FreeAgent.com), name-

your-price mechanisms (e.g., Priceline.com), quantity discounters (e.g.,

Mercata.com), and methods that used derivative based pricing for consumer

goods (e.g., Iderive.com). Not surprisingly, some continue to flourish (e.g. eBay

4

and Onsale) while others have floundered (Mercata and Iderive). Despite the

innovativeness of these pricing approaches and the initial dotcom buzz

surrounding them, little attention has been paid to the their effectiveness.

Instead, directionless entrepreneurship, at times by fueled by overzealous

venture capitalists, replaced scientific enquiry and rigor when it came to

examining the efficacy and viability of candidate mechanisms.

We contend that significant research is still needed in designing new and

better mechanisms, as well as examining the efficacy of existing ones in the

contexts of the markets they serve. In this chapter, we touch upon issues of

mechanism design, secondary market creation, bidding costs and strategies,

incentive compatibility, bid taker cheating (shilling), simultaneous substitutability,

and associated research methodologies.

Interestingly, the advent of auctions over open Internet Protocol based

networks, such as the Internet, has also facilitated the pursuance of a richer set

of empirically derived methodologies by today’s researchers. Most pre-Internet

based auction research was either purely theoretical in nature [see McAfee and

McMillan (1987), Milgrom (1989), and Myerson (1981) for a thorough review), or

involved laboratory experimentation [see Kagel and Roth (1997) for an excellent

overview]. Empirical research was rare due the lack of meaningful data sets,

which in turn could be attributed to the lack of mainstream appeal of auctions.

Lucking-Reiley (1999) acknowledges the difficulty in obtaining field data for

testing long-standing hypotheses, such as the supposed revenue equivalence

between the basic auction formats. The best data set available prior to the arrival

5

of the web-based auctions covered US Forest Service sales of contracts for

harvesting timber in the Pacific Northwest during 1977 [see Hansen (1985)]. The

earlier lack of empirical and/or realistic experimental test environments is

increasingly disappearing with the technological advancements in online auction

technology. The widespread popularity of online auctions, coupled with the open

computing paradigm upon which Internet applications are built, together present

a golden opportunity for researchers to revisit the various branches of auction

theory in a setting that is more realistic and has higher inductive value.

In the remainder of this chapter we share the insights of these recent

research developments in online auctions.

2. A Review of Major Online Auction Mechanisms

A key factor that makes electronic markets, such as online auctions, interesting is

the potential for achieving higher efficiency. On surface e-markets such as eBay,

with millions of registered users, would appear to be a close approximation to an

economist ’s idealization of a frictionless, efficient market. One thing for certain

is that using information technology has brought this sphere of economic activity

out of the domain of specialists to that of the common man. The success of eBay

notwithstanding, we contend that the frictionless efficient market is still an ideal to

strive for. We review the popular types of online auctions, with the caution that

this is by no means an exhaustive list of current online auctions. Our objective

here is to isolate mechanisms that are interesting, being currently researched

and in which the online environment influences the strategic spaces of the

6

participants. We begin with the classic single item, open, ascending, English

auction.

2.1 Single item Auctions

This asset-exchange mechanism has been extensively studied

theoretically, beginning with the seminal article by Nobel laureate William Vickrey

(1961). More recent coverage can be found in Rothkopf and Harstad (1994).

Researchers studying this auction commonly make use of assumptions, such as

the independent private values (IPV) assumption to derive its equilibrium

characterization. Such an assumption implies that a single indivisible object is to

be sold to one of several bidders. Each bidder is risk-neutral and knows the value

of the object to himself, but does not know the value of the object to other

bidders. It also implies that there is a finite population of bidders, each of which

draws his valuation independently from some given continuous distribution (see

Milgrom, 1989 for a detailed description).

Consider the applicability of this assumption to common single item online

auctions such as the ones conducted on eBay. Note, eBay’s multi-item Dutch

auction are discussed in the next subsection. For most goods being auctioned,

the IPV assumption is robust. However, for collectibles, a popular category on

eBay, it is reasonable to assume that an individuals’ valuation will be dependent

on the valuations of fellow bidders. Presumably, a collector will have the

objective of at least recovering the cost of the item purchased and thus will

implicitly carve her valuation distribution to be dependent on that of other bidders.

7

The presence of winners curse in auctions, namely that bidders win items

only because they pay too much, has long been of interest to researchers. Bajari

and Horta (2000), in an empirical study of eBay auctions, find that a bidder’s

surplus falls by 3.2 percent when the expected number of bidders increases by

one. In section 6, we describe two recent studies that have compared online

auctions with posted price mechanisms selling the same goods (using matching

SKU’s). The evidence for winners curse from those two studies is mixed.

A common assumption related to IPV, especially popular in experimental

studies in the lab, is that the number of bidders in an auction is exogenously

determined. That somehow this population is known a priori and the design of the

auction itself does not influence the number of participants. Several, recent

empirical studies challenge this notion and point to interesting issues regarding

the choice of a reserve price by the seller. On eBay the seller can set an open

minimum bid that serves as the starting price for the auction. Additionally, the

seller also has the option of setting a hidden reserve price below which she is not

obligated to sell. Once the bidding level exceeds the reserve, an indication of

“reserve met” is displayed to the bidders. The questions are: a) Should the seller

use the hidden reserve at all? b) Does setting a low minimum bid attract bidders

to the auction, thereby increasing competition? And c) Is there an optimal mixed

strategy that can be employed by sellers with respect to these two parameters?

Bajari and Hortacsu (2000) find that items with higher book value tend to be sold

using a secret as opposed to posted reserve price with a low minimum bid. They

8

also find that the minimum bid is the most significant determinant of whether a

bidder enters an auction.

Lucking-Reiley (1999b) describes controlled experiments, conducted on

the Internet, to verify a variety of theoretical properties of electronic auctions by

manipulating the reserve prices in these auctions as an experimental treatment

variable. His findings indicate that bidders consider their bid submission to be

costly and that bidder participation is indeed an endogenous decision.

Additionally, the data shows that a zero reserve price provides higher expected

profits than a reserve price greater than or equal to the auctioneer’s salvage

value for the good. In contrast, Reily and Samuleson (1981) showed that for an

optimal, highest clearing price, English auction the reserve price is a function of

the sellers' valuation of the product.

Perhaps, the most cited property of single item auctions is the Theorem of

Revenue Equivalence [Vickrey (1961), Myerson (1981) and Bulow and Roberts

(1989)]. The idea is that under a set of restrictive assumptions (IPV and risk-

neutral bidders), the expected revenue from a variety of auction types, namely:

English, Dutch, first price and second price sealed bid auctions, is equivalent.

Revenue equivalence results are known not to be robust with respect to the

slightest deviation from the restrictive assumptions of the independent private

values model [Myerson (1981)] or bidder risk preferences [Maskin and Riley

(1984)], which are notoriously difficult to observe. Lucking-Reiley (1999a) tests

revenue equivalence through his field experiments auctioning Magic game cards.

He finds that the Dutch auction produces 30 percent higher revenues than the

9

first-price auction format, a contradiction to the theoretical prediction and a

reversal of previous laboratory results that the English and second price formats

produce roughly equivalent revenues.

It will be interesting to see whether similar trends are observable in online

auctions of a more general nature. Future research in this area promises to bring

interesting new insights into these age old topics of interest.

2.2 Multi-unit (item) auctions

The online environment has spawned a variety of auctions that sell

multiple identical units of the same item or good. These range from auctions of

consumer goods, mostly aging hardware and electronics, through sites such as

Onsale.com and Ubid.com, to auctions of fixed-income and equity securities by

Muniauction.com, to OpenIPO.com -- which allows individual and institutional

investors to bid online for shares of an IPO -- giving both types of investors a

level playing field in the IPO market for the first time in the history. A cursory

examination of the above mentioned sites reveals that what was historically a

sealed-bid dominated market, as in the auction of Treasury bonds, now supports

the a wide range of auction mechanisms, namely ascending English, descending

Dutch, and eBay’s so called "Dutch" -- which is really an ascending open uniform

price auction.

Rothkopf and Harstad (1994) point out that single-item results do not

carry over into multiple-item settings and that this has been a vastly neglected

area of auction theory research. Of late there is evidence of research spawning

in multi-item auctions. List and Lucking-Reiley (1999) examine the case when

10

consumers are allowed to bid for more than one item under two different types of

two-unit, two-person sealed bid auctions. When consumers are allowed to bid for

more than one-item in an m-item auction, Vickrey’s original proposition -- full

demand revelation occurs in a sealed-bid auction -- does not hold [Ausubel and

Crampton (1999)]. Instead, the rule has to be modified such that for an m-item

Vickrey auction bidders can submit as many individual unit bids as they like.

Further, the top m bids are declared winners and for the jth unit won by a bidder,

she pays an amount equal to the jth highest of the rejected bids submitted by

others [Groves (1973) and Clarke (1971)]. Hence this revised mechanism offers

discriminating prices in contrast to the original mechanisms’ uniform pricing.

Importantly, this mechanism is incentive compatible. Bidders gain nothing by not

revealing their true valuations, as they never have to pay what they bid.

In the two-item case, List and Lucking-Reiley (1999) indicate that there is

evidence of demand reduction, i.e. lowering of the second bid below the true

valuation, when the uniform pricing rule is applied. This is a cause for concern

and leads to lower allocative efficiency. In the case of real-world B2C online

multi-item auctions consumers are allowed to bid for more than one-item but

these bids cannot be discriminating, i.e. they all have to be of the same amount.

For instance, a given individual can bid for 3 items at $100 each but cannot bid

for 2 items at $110 and 1 item for $80. Whether this constraint is designed to

prevent demand reduction in auctions that sell multiple (far greater than 2 units)

is an open and interesting research question.

11

Bapna, Goes and Gupta (2000, 2001) present an analytical modeling

approach, to analyze multi-item progressive English auctions, also known as

Yankee auctions, such as those conducted by Onsale.com. Their analytical

modeling is subsequently validated by empirical investigation using data

collected by automated agents which track real-world web auctions, adding a

new methodological dimension to auction theory research.

They focus on hitherto undescribed discrete and sequential nature of the

revenue realization process of such auctions, caused by the presence of the bid

increment. In such auctions, say of ten Palm V PDAs, the ten highest bidders win

and the price they pay is equivalent to their highest bids, technically making this

a discriminatory auction. Usually, a very low opening bid such as $1 is set by the

auctioneer as a way to attract web traffic. In addition, all auctions have a bid

increment that defines the minimum step size for bidders. Bids that fall in

between bid increments are automatically rounded down to the nearest step.

This discretization of the process challenges the common auction theory

assumption that individuals' valuations can be drawn from a known, continuous

distribution. The bid increment also helps determine the minimum required bid at

any time during the auction. This is equal to the lowest winning bid plus the bid

increment.

The list of current winning bidders, the bid increment, the minimum

required bid, and the auction closing time are all continuously updated on the

web. Auction durations’ typically range from one-hour express auctions to day-

long regular auctions. Bids are ranked by bid amount and by time within amount.

12

Unlike traditional, single-item English auctions, a new bidder's high bid does not

automatically displace the existing winner from the winner's list. In fact if the

current number of bidders is less than the lot size than the new high bid does not

affect any of the existing winners.

An interesting observation revealed by their data collection process is that

online auctioneers experiment with the design parameters they can control for

such auctions. For instance, they sell the same goods using a $20 bid increment

one day, followed by using a $15 bid increment on another day. There exists a

great opportunity for researchers in determining how to optimally set the control

factors such as the bid increment, lot size and the opening bid that influence the

efficiency of such auctions.

There have been other attempts to compare the efficiency of different

auction mechanisms both theoretically and empirically. The focus has been on

comparing single-item sealed-bid competitive auctions with sealed-bid

discriminatory auctions. In the former mechanism, the highest bidder wins,

however, the price paid is the second highest bid; whereas in the latter the

highest bidder wins with the price being the highest bid. Competitive auctions

were first suggested by Vickrey (1961) in his seminal article; the special property

of this mechanism is that all the bidders have incentive to bid their true valuation.

Plot and Smith (1978) were among the first to design a controlled laboratory

experiment to compare competitive auctions with discriminatory auctions. Actual

bidding data has also been analyzed by Baker (1976). The key results of these

empirical investigations have been inconclusive with respect to sellers' revenue.

13

Harris and Raviv (1981) compare the efficiency and expected revenue of the

uniform price (Vickrey-like) auction mechanism with that of the discriminating

(first price sealed-bid) mechanism when a fixed quantity of divisible goods are to

be sold to many buyers. Their results indicate that the sellers' revenue under a

specific mechanism depend on the risk characteristics of the bidders.

Given the growing theoretical and empirical interest in multi-item auctions,

it will be interesting to see work that examines the relative efficacy of

mechanisms that are available to the consumers today. Does eBay’s so called

“Dutch” mechanism, which has little theoretical basis and unexplored incentive

characteristics, lead to higher clearing prices than say an equivalent Yankee

auction? The extension of the single-item results, such as revenue equivalence,

to the online setting of multi-item auctions is an interesting area of research. This

will help us understand which mechanism should be adopted under what

circumstance. For instance if there is a pre-dominance of risk-averse bidders

who prefer a certain outcome to an uncertain one, than would a descending

Dutch auction yield higher expected revenue? Of course, this analysis is not

trivial even with the most simplistic of assumptions regarding the consumer type.

2.3 Combinational Auctions

If we allow multiple units of non-identical goods to be sold through online

auctions, then we get into the realm of combinational auctions. Such auctioning

schemes are desirable to sell complementary goods that can be “bundled”

together. A good example is the FCC spectrum auction for different regional

licenses, where the value of having, say Boston increases if the bidder can also

14

acquire neighboring New York. These auctions present many interesting

challenges to practitioners, both sellers as well as bidders, as well as

theoreticians. In general, the auctioneer's problem of determining an optimal set

of bids in a combinatorial auction is an NP-Complete problem, a class of

problems that are tough to solve in a reasonable amount of time. Additionally,

there are the issues of the (i) exposure problem: an unsuccessful attempt to

acquire a collection of assets, when combinational bidding is not allowed, may

lead to paying more for some individual assets than they are worth, and (ii)

threshold problem: a bidder on item A and a bidder on item B may not be able to

coordinate to displace a bid on package AB in the presence of diseconomies of

scale.

Several interesting approaches are being proposed to overcome some of

the above mentioned, computational and mechanism design, difficulties in

combinatorial auctions. One such apporach is the iBundle mechanism of Parkes

(1999). The basic idea of the iBundle mechanism is to use software to calculate

the maximal allocation of products among various users who can bid on bundles.

Each bidder can bid for any number of bundles, so a bidder can offer $10 for A,

$20 for A and B. The iBundle software then calculates the combination of

bundles that maximize total transaction value and notifies the bidders of the

provisional winners. The bidders are then able to make higher bids, and the

process repeats until bidders are satisfied. The contribution of iBundle is to use

IT to quickly solve an optimal allocation problem that would be computationally

infeasible for human agents in real time.

15

Jones and Koehler (2001) are developing another interesting

combinatorial auction mechanism. Their approach accepts incompletely specified

bids that provide a framework to guide, rather than dictate, the choice of goods

that satisfy bidder needs. Their incompletely specified combinatorial auction

mechanism is designed to facilitate large and complex problems commonly

relegated to negotiated sales, where the allocation of goods requires solving a

complex combinatorial problem. A representative example is the complex multi-

dimensional process of media buying, specifically, the sale of television

advertising airtime. Allowing a bid in the form of high-level rules relieves the

buyer from the burden of enumerating all possible acceptable bundles.

Further research is needed to harness the enormous computational power

available to us to make combinatorial auctions mainstream for corporations keen

to optimize their complex logistical decisions, such as determining optimal freight

patterns fro moving goods from manufacturing sites to wholesale sites onwards

to retail sites.

2.4 Multi-dimensional Auctions, Reverse Auctions

In certain cases, for example in many procurement situations, it is not

sufficient to conduct auctions where only the price dimension matters. Often

price and quality go hand in hand and a jointly determine the winning bid -- not

necessarily the lowest price bid. Majority of the literature on auction theory has

focused on the analysis of auctions of a well-defined object or contract so that

the price to be paid is the unique strategic dimension with the exception of

Branco (1997), Che (1993) and Thiel (1988). For example, in the auction for a

16

department of defense contract, say for the construction of an aircraft, the

specification of its characteristics is as important as the price. In such multi-

dimensional auctions bidders submit bids with relevant characteristics of the

project, price being just one such characteristic, and the procurement agency

uses a scoring mechanism to select amongst the bids. Branco (1997) and too

some extent Che (1993) recommend a two-stage auction mechanism where in

the first stage the procurer selects one firm and in the second stage she bargains

to re-adjust the level of the quality to be provided.

Multiple dimensions also exist in the reverse auctioning of service

contracts, such as those conducted by Elance.com and Freeeagent.com, where

price is not the only differentiating factor. For instance, a client who posts a

requirement for a web design project could receive bids from all over the world,

ranging from Bangalore, India to Yugoslavia to the US with hourly rates of $24,

$10 and $50 respectively, as we casually observed on Freeagent.com. It would

be naïve to think that the client would necessarily go with the lowest bid in this

case as service quality may be widely varying, and perhaps even difficult to

assess. Snir (2000) studies Internet based spot markets for service contracts. His

analysis confirms the fact that the transaction costs of posting a project, bidding

on a project and evaluating bids are all significant.

3. Consumer Bidding Strategies

The global scope and reach of the online environment makes it feasible

for “armchair” bidders [Kauffman and Walden (2001)], to be active players in

17

auction markets. While much of the theoretical development assumes buyers

and sellers to be rational, profit maximizing individuals, a tenuous assumption to

begin with, the reality of the online markets landscape indicates a wide disparity

in user experience and information levels. Researchers such as Lucking-Reiley

and List (2000) and Bapna, Goes and Gupta (2001) have been quick to capitalize

on the availability of real auction data, obtained through field experiments and

automated-agent based real-time tracking of online auctions respectively, for

examining consumer bidding strategies on the Internet. Among other things they

test the behavior of uninformed bidders, who exhibit different behavior than

theory predicts.

Lucking-Reiley and List (2000) auctioned 4 types of trading cards ranging

from $3 to $70. They examined whether dealers of such cards would bid more

rationally than non-dealer, less experienced, individual card collectors. They find

that dealers exhibit more of the predicted strategic behavior than do nondealers,

for both lower and higher priced cards and that the predicted strategic behavior is

considerably greater when the auctioned sportscards have higher values,

confirming prior theory [Smith and Walker (1993)] that suggest that rationality is

more likely to be exhibited when the stakes are higher.

Bapna, Goes and Gupta (2001) also find support for the notion that

rationality becomes more evident as the expected payoff is higher. Their

empirical investigation of 90 such auctions identified three distinct types of

bidders. They are summarized in Table 1 [Bapna (1999)]:

18

Evaluators • Early one time high bidders who have a clear idea of their

valuation

• Bids are, usually, significantly greater than the minimum

required bid at that time

• Rare in traditional auction settings - high fixed cost of

making a single bid

• Violates the assumption of rational participatory behavior

described earlier

Participators • Derive some utility (incur a time cost) from the process of

participating in the auction itself

• Make a low initial bid equal to the minimum required bid

• Progressively monitor the progress of the auction and

make ascending bids never bidding higher than minimum

required

Opportunists • Bargain hunters

• Place minimum required bids just before the auction

closes.

Table 1. Bidder Classification

In order to compare the performance of these strategies they introduced a

metric based on loss of surplus. This is the difference between an individual's

winning bid and the minimum winning bid. Loss of surplus evaluates the

performance of an individual or a group with respect to the bidder who had the

19

minimum winning bid in a given auction. Bapna, Goes and Gupta (2000)

compared the relative performance of these three groups with respect to loss of

surplus in the auctions tracked on the WWW. They found that the evaluators as a

group fared worst, the participators were best off, and the opportunists lay in

between.

Much like Lucking-Reiley and List (2000) they also find evidence for the

fact that as the stakes get higher, a larger percentage of the population behaves

strategically. This is evident in Figure 1 below where the percentage of rational,

non-evaluator type, bidders is plotted against the dollar values of the auctions

tracked.

% Rational Bidders

-20%

0%

20%

40%

60%

80%

100%

120%

0 500 1000 1500 2000

$ Value of Auction

Figure 1 – Rationality Increases with the Dollar Value of the Auctions

It is easy to see an increasing trend in the percentage of bidders who

behave in a non-evaluator mode as the auction stakes get higher. Evaluators, as

we know were found to be the worst off of the three bidder categories.

20

5. Opportunism and Trust in Online Auctions

Typically, ignorance of what price to post is a reason for negotiating or

holding an auction. Rothkopf and Harstad (1994) provide a behavioral reason for

holding auctions. They assert that one of the critical reasons for the use of

bidding is that the formality of the auction process provides legitimacy, in a way

that other economic means cannot. Contradicting this belief from academia, is

the practical notion that the lnternet makes the likelihood of fraud detection and

punishment low. Thus, one is not surprised that opportunistic behavior in online

auctions, by both sellers and buyers, is constantly in the media. According to

Internet Fraud Watch, online auction fraud has become the number one type of

Internet fraud over the last two years.

Brandabur and Saunders-Watson (2001) discuss this issue in depth. For

instance, they talk about an auction for a painting that was claimed to by the late

Bay Area painter Richard Diebenkorn. The auction soared to $135,805 with the

winning bid coming from Holland, however, the painting turned out to be a fake.

The seller gave the appearance of being a novice, never actually mentioning the

artist by name. In reality, it turned out this seller along with two accomplices

often used more than 20 eBay screen names for the purpose of shilling -- a

strategy of placing phony bids to run up the closing prices of auctions!

Recently researchers have begun to attempt to model important aspects

of trust and reputation in online auctions. Much of this stream relies on analytical

modeling backed up empirical investigation using automated agents to capture

21

data from real-world auctions. This is an important area of research as it

addresses many of the issues raised by the critics of online auctions and helps

formalize the concept and measurement of trust and reputation of sellers.

There is increasing concrete evidence of shilling by bid-takers in online

auctions. Kauffman and Wood (2000) present an analytical model and provide

empirical evidence of ‘questionable bidding behavior (QBB)’ by sellers on eBay.

They operationalize QBB as bidding on an item when the same or a lower bid

could have been made on the exact same item in a concurrent auction ending

before the bid-upon auction. QBB can be considered irrational, since the buyer

has a greater level of utility if she were to bid on another item for the same or

lower cost. Their research highlights the difficulties associated with identifying

opportunistic sellers using QBB in online environments, when many auctions are

going on in parallel. First, non-reputable sellers try to remain anonymous.

Because they are attempting to hide their identity, it is difficult to identify them.

Second, it is difficult to track multiple Internet auction identities and tie them

together. Third, QBB needs to be reviewed over time in multiple auctions. For

instance, if a bidder (using the same name) consistently rates a particular seller

higher and exhibits active bidding behavior (for shilling purposes) while never

winning many of the items, it would be easy to identify suspicious behavior,

however, in practice finding such behavior is difficult. Using intelligent data

gathering agents Kauffman and Wood (2000) track a number of eBay auctions of

coins, and their initial findings suggest that indeed a significant amount of QBB is

evident.

22

Opportunism can also be exhibited by the bidders, in the form of collusion,

through fictitious identities. Wang, Hidvegi and Whinston (2000) bring out the

importance of appropriate mechanism design to counter such undesirable

behavior. They focus on sealed bid auctions and propose an alternative

mechanism to the Generalized Vickrey Auction (GVA), called the Sealed-bid

Multi-Round Auction Protocol (S-MAP). They give instances when the GVA is no

longer incentive compatible. Their examples show that in GVA under collusion,

truth-telling is not longer a dominant strategy and illustrate how a bidder can

reduce her payment by submitting bids under her false names. This limitation is

overcome in S-MAP, however, the mechanism itself induces a higher cognitive

load on the players and it remains to be seen whether it can become

mainstream.

Pavlou and Ba (2000) recognize that trust is an essential component of

online auctions, and that buyers pay a price premium to transact with reputable

sellers, particularly for expensive products. Results showed a significant

correlation between trust and price premiums for all products. Moreover, this

correlation became increasingly more significant for more expensive products.

In another interesting empirical study Dewan and Hsu (2001) examine the

economic value of trust in electronic markets, based on a comparison of prices

across generalist (eBay) and specialty sites (Michael Rogers, Inc.) in the arena of

person-to-person online auctions. Generally, the two types of sites have very

different mechanisms for providing trust in the marketplace – whereas generalist

sites do not inspect the merchandise and rely instead on a reputation reporting

23

system, specialty sites typically take possession of auction items and provide a

variety of value-added services directed at reducing information asymmetry and

other sources of transaction risk. Their empirical findings appear to confirm that

the observed price differences between eBay and the specialist are driven

primarily by the relative effectiveness of trust mechanisms in the two markets.

6. Simultaneous Substitutable Mechanisms: Auctions v. Posted Price

Another promising stream of IS research in the area of dynamic pricing

deals with comparing auctions with posted price mechanisms for the sale of

identical goods. Seidmann and Vakrat (1999) compared online catalog prices

with online auction prices. They obtained data from 473 online auctions, such as

SurplusAuction (www.surplusauction.com) and OnSale.Com (www.onsale.com).

They compared prices received in these auctions with prices from Internet

catalog sellers, such as Egghead (www.egghead.com) and PriceScan.Com

(www.pricescan.com). Their data analysis revealed that consumers expect

greater discounts for more expensive items. In their studies, Seidmann and

Vakrat employed Internet agents as a data collection tool. Using a similar

methodological approach Lee and Mehta (1999) investigated the existence of

winner's curse using theoretical modeling and empirical validation. Their

preliminary results confirm the existence of the winners’ curse in electronic

auction. The amount overbid is especially pronounced for items where potential

information asymmetries exist as a result of the nature of the product, and it is

24

further augmented in cases where the product is relatively new and not much

information regarding it exists in the retail channels

Interesting current research in this area is being carried out both from the

buyers’ and the sellers’ perspective. Barua and Tomak (2000) are studying under

what conditions should buyers use auctions in contrast to posted price

mechanism?

Aron, Croson and Lucking-Reliey (2000) investigate when should auctions

be used by sellers instead of posted prices? They are developing a theoretical

model of markets with uncertain demand, in an attempt to understand what types

of demand uncertainty make it worthwhile for a seller to consider investing in an

auction mechanism in order to gain more price flexibility.

7. Conclusions and Future Trends

The above review indicates that interesting developments are happening in both

the practice and research of online auctions. The current dotcom shakeout not

withstanding, dynamic pricing mechanisms, such as online auctions, will continue

to be an important component in the portfolio of mercantile processes that will be

deployed by businesses to transact with their customers and suppliers, and for

consumers to transact with other consumers.

We call for greater interaction between the practitioners and researchers in this

area. In many cases. practitioners, fueled by over zealous venture capitalists,

would do well to resist carrying out costly field experiments in the name of

innovation. They can enlist the research community to examine the design of the

dynamic pricing mechanisms they propose to adopt in a give market. Is the

25

mechanism suitable for the targeted market? Would an alternative mechanism fit

the bill? Will it achieve the desired liquidity to sustain itself, and will it achieve

higher allocative efficiency than its current counterparts. The large numbers of

failed real-life experiments in dynamic pricing (mercata.com, priceline.com for

groceries!), and the associated loss of social capital, could have been avoided if

interactions between practitioners and researchers were de rigueur.

We look forward to more research that examines the relevant issues in online

auctions without the baggage of the traditional assumptions made in earlier

auction theory. In the absence of the physical constraints of traditional auctions

the behavior of the different economic agents in auctions is heavily influenced by

the (online) context in which they take place. For instance, the presence of

simultaneous substitutable online auctions - which allows an individual shopping

for, say a computer, to simultaneously bid at Onsale.com or Yahoo.com -

impacts the efficiency of not just the isolated auction under consideration but also

the external market in which it takes place. Auction portals like

www.biddersedge.com are specifically designed to make tracking such

simultaneous substitutable online auctions easy for the consumer.

We firmly believe that the emerging practice and research in this area has high

inductive value and will lead to a significant enhancement to the body of

knowledge dealing with dynamic pricing and auctions.

26

References

1. Aron, R., Croson, D., and Lucking-Reliey, D., Auctions Versus Posted

Prices, Workshop on Information Systems and Economics, Brisbane,

2000.

2. Bajari, P., Horta, A., “Winner’s Curse, Reserve Prices and Endogenous

Entry: Empirical Insights from eBay Auctions,” Working Paper, 2000,

Department of Economics, Stanford University.

3. Bapna, R., "Economic and Experimental Analysis and Design of Auction

Based On-line Mercantile Processes," Ph.D. Dissertation 1999,

Department of Operations and Information Management, University of

Connecticut, Storrs.

4. Bapna, R., Goes, P., and Gupta, A., "A Theoretical and Empirical

Investigation of Multi-item On-line Auctions," Information Technology and

Management, Jan 2000a, 1(1), 1-23.

5. Bapna, R., Goes, P., and Gupta, A., "Online Auctions," Forthcoming in the

Communications of the ACM, 2001.

6. Barua, A., and Tomak, K., Workshop on Information Systems and

Economics, Brisbane, 2000.

7. Branco, F., The design of multidimensional auctions, RAND Journal of

Economics, 28, 1, Spring 1997, 63-81.

8. Brandabur S., and Saunders-Watson C., Fraud Online and Off,

AuctionWatch.com article, May 2001.

27

9. Bulow, J., and Roberts, J., “The Simple Economics of Optimal Auctions,”

Journal of Political Economy, 1989, 7, no 5, 1060-1090.

10. Che, Y-K, Design competition through multidimensional auctions, RAND

Journal of Economics, 24, 4, Winter 1993, 668-680.

11. Dewan, S., Hsu, V., Trust in Electronic Markets: Price Discovery in

Generalist Versus Specialty Online Auctions, Working paper – University

of Washington, Seattle, 2001.

12. Hansen, R. G., “Empirical Testing of Auction Theory,” American Economic

Review, 75( 2), May 1985, 156-159 .

13. Jones. J. L., and Koehler, G. J., “Incompletely Specified Combinatorial

Auctions,” University of Michigan working paper, 2001.

14. Kagel, J. H., Roth, A. E., The Handbook of Experimental Economics, ,

editors, Princeton University Press, Fall 1997.

15. Kauffman, R., and Wood, C., Running Up the Bid: Modeling Seller

Opportunism in Internet Auctions, Proceedings of the AMCIS 2000

Conference, Long Beach CA.

16. Kauffman, R., Walden, E., “Economics And Electronic Commerce: Survey

And Research Directions,” Submitted to International Journal of Electronic

Commerce, 2001.

17. Lee, B., Mehta, K., Efficiency Comparison in Electronic Market

Mechanisms: Posted Price Versus Auction Market, WISE Conference

1999, Charlotte, NC.

28

18. Lucking-Reiley, D., “Using Field Experiments to Test Equivalence

Between Auction Formats: Magic on the Internet,” 1999, American

Economic Review. December 1999a, vol. 89, no. 5, pp.1063-1080

19. Lucking-Reiley, D., List, J. A., "Bidding Behavior and Decision Costs in

Field Experiments", Working Paper – Vanderbilt University, Last revised:

March 2000.

20. Lucking-Reiley, D., "Experimental Evidence on the Endogenous Entry of

Bidders in Internet Auctions." Working Paper – Vanderbilt University, Last

revised: Last revised: May 1999b.

21. Maskin, E, and Riley, J. “Optimal Auctions with Risk Averse Buyers.”

Econometrica, November 1984, 52 (6), pp. 1473-1518.

22. McAfee, R. P., and McMillan, J., "Auctions and Bidding," Journal of

Economic Literature, 25 (1987), 699-738.

23. Milgrom, P., "Auctions and Bidding: A Primer," Journal of Economic

Perspectives, 3 (1989), 3-22.

24. Myerson, R. B., "Optimal Auction Design," Mathematics of Operations

Research, 6, (1981), 58-73.

25. Mollman, S., 2000,

http://www.zdnet.com/pccomp/stories/all/0,6605,2431978,00.html.

26. Parkes, D. C. iBundle: An Efficient Ascending Price Bundle Auction.

Proceedings of the 1999 ACM Conference on Electronic Commerce (EC-

99), Denver, CO, ACM Press, New York, NY.

29

27. Pavlou, P. A., Ba, S., Does Online Reputation Matter? – An Empirical

Investigation of Reputation and Trust in Online Auction Markets,

Proceedings of the AMCIS 2000 Conference, Long Beach CA.

28. Riley, J. and Samuelson, W. F., " Optimal Auctions" The American

Economic Review, Vol. 71, No. 3, 1981, 381 - 392.

29. Rothkopf, M. H., and Harstad, R. M., "Modeling Competitive Bidding: A

Critical Essay," Management Science, 40 (3), 1994, 364-384.

30. Rothkoff Michael H., Aleksandar P. and Harstad, R. M. " Computationally

Manageable Combinatorial Auctions", Management Science, Vol. 44., No.

8, pp 1131 –1147.

31. Smith, V L., and Walker J M., “Monetary Rewards and Decision Cost in

Experimental Economics,” Economic Inquiry, 1993, 31, 245-261.

32. Snir E. M, (2000) "Designing Internet Spot Markets for IT Services,"

Working Paper- Wharton School, U Penn.

33. Theil, S. E., Multidimensional Auctions, Economics Letters, 28, 1998, 37-

40.

34. Vakrat, Y. and Seidmann, A. "Can Online Auctions Beat Online

Catalogs?," In P. De and J DeGross (eds.), Proceedings of the 20th

International Conference on Information Systems (ICIS '99), Charlotte,

NC, December, 1999.

35. Varian, H., “Miles And Miles Of Flexible Track,” Forbes ASAP, 10/02/2000,

http://www.forbes.com/asap/2000/1002/071.html

30

36. Vickrey, W., "Counter-speculation, Auctions, and Competitive Sealed

Tenders," Journal of Finance, 41, 1961, 8-37.

37. Wang W., Hidvegi Z., Whinston A. B., Economic Mechansim Design For

Securing Online Auctions, In Proceedings of the 21th International

Conference on Information Systems (ICIS '00), Brisbane, Australia,

December, 2000.