Re-engineering the Dutch Flower Auctions

1047-7047/98/0901/0001$05.00Copyright q 1998, Institute for Operations Researchand the Management Sciences

Information Systems ResearchVol. 9, No. 1, March 1998 1

Reengineering the Dutch Flower Auctions:A Framework for Analyzing Exchange

Organizations

Ajit Kambil • Eric van HeckInformation Systems Department, Stern School of Business, New York University, 44 West 4th Street,

New York, New York [email protected]

Department of Decision and Information Sciences, Rotterdam School of Management, Erasmus University Rotterdam,P.O. Box 1738, 3000 DR Rotterdam, The Netherlands

[email protected]

This paper specifies a generalizable model of exchange processes and develops a process-stakeholder analysis framework to evaluate alternative market designs. This framework

is applied to analyze a number of information technology initiatives in the Dutch flower mar-kets. The Dutch flower auctions are the world’s leading centers for trading cut flowers andpotted plants. We undertake a cross-case analysis and apply our framework to analyse suc-cesses and failures in the introduction of new IT-based trading mechanisms in these markets.Based on our study, we develop a number of testable propositions on: the separation of physi-cal and informational processes in trading, the responses of stakeholders to changes in avail-able information due to IT initiatives, and economic and incentive conditions required foradoption of new trading processes. Finally, our detailed cases illustrate the institutional andincentive constraints, and complexities encountered in the introduction of new electronicmarkets.(Electronic Markets; Transaction Costs; Reengineering; Technology Adoption)

1. IntroductionInformation technology enables new ways of coordi-nating the exchange of goods and services. Improvedcommunications and processing systems, such as theInternet and video conferencing, allow individuals andorganizations to radically re-engineer existing tradingprocesses enabling new forms of electronic commerceand markets.

Markets are real or virtual meeting places wherebuyers, sellers, and intermediaries meet to exchange ortransfer property rights from one party to another.Typically, buyers and sellers confront many differentuncertainties and risks in trading. These include theinability to forecast or plan for the future given the

traders’ bounded rationality and market uncertainty,combined with the potential for opportunism given thedivergent interests of different market participants(Williamson 1975, 1985). Markets and other gover-nance mechanisms provide specific information pro-cessing capabilities to exchange parties and serve togenerate information and trust to reduce the uncer-tainty and risks inherent in the exchange process(Kambil 1992). To reduce uncertainty and risk, marketsas institutions provide specific routines and proce-dures for coordination, and for the information andguarantees that enable the buyers and sellers to un-dertake the exchange.

Prior research on the effects of information technol-

KAMBIL AND VAN HECKReengineering the Dutch Flower Auctions

Information Systems Research2 Vol. 9, No. 1, March 1998

ogy (IT) on exchange organizations and processes typ-ically applied transaction costs and agency theory topredict shifts from hierarchies toward market or otherintermediate forms of organizations (Bakos 1991a,Johnston and Lawrence 1988, Malone et al. 1987,Powell 1990). A central argument of these articles wasthat information technology would improve commu-nications, searches, monitoring, and information-sorting capabilities to reduce transaction costs and al-low purchasers to take advantage of production econ-omies available in markets. A critical drawback of thisanalysis was the definition of markets in abstract eco-nomic terms (i.e., markets coordinate economic activ-ity through a price mechanism) without considerationfor differences in market organization. For example,some different market types include direct search mar-kets where partners directly seek each other out (e.g.,a typical retail fruit and vegetable market), brokeredmarkets where exchange parties employ agents to seekcompatible partners (e.g., private placement offeringsby investment banks), dealer markets where dealers asintermediaries hold, buy, and sell product inventories(e.g., NASDAQ), and auction markets where traderstransact directly through a centralized intermediary(e.g., Sotheby’s art auctions). Each of these mecha-nisms organizes the trading process and related infor-mation processing activities in different ways. Thus,we can expect the impact and role of IT can vary acrosstypes.

More recently, information systems researchers haveexamined the impact of information technology onmarket institutions. Konsynski et al. (1989, 1990, 1992)provided a number of descriptive case studies of elec-tronic markets providing a rich case base for research.Clemons and Weber (1990, 1991) examined the effectsof computerization on the London Stock Exchange,and Lee (1993) provided a technology to support ordermatching of complex products. These studies begin toconsider the importance of different market types andinstitutional histories.

Despite the increasing number of cases on electronicmarkets, a common language and categorization ofkey processes in a trading system is lacking. Such acategorization is a precondition to building a frame-work to systematically evaluate and compare the

structure and benefits of different markets for variousstakeholders. Building on transaction costs andinformation-processing models (Galbraith 1974,Tushman and Nadler 1978), this paper provides botha model of key exchange processes and a process-stakeholder analysis framework for comparing acrossdifferent market structures. Such a framework is im-portant to managers who must select and design newtrading models. It is also important to researchers whowant to develop models of how information technol-ogy affects exchange organizations and create theoriesthat can be generalized across cases. We illustrate andapply this framework to evaluate different electroniccommerce initiatives in the Dutch flower markets. Theframework for analysis developed in this paper canapply to the analysis of other market structures.

2. Exchange: A Process StakeholderAnalysis

Exchange is a critical subject of study in marketing,economics, finance, and information systems. Buildingon these disciplines, information systems researchershave typically adopted a number of outcome dimen-sions to characterize and evaluate markets. For exam-ple, Clemons and Weber (1990) characterize financialmarkets in terms of liquidity, volatility, and transpar-ency. Liquidity is the ability of the market to absorblarge orders without significant price changes, indicat-ing the depth and extent of market participation bybuyers and sellers. Volatility refers to the variance inday-to-day volumes and prices, and transparency re-fers to keeping the widest group of buyers informedon current prices and allowing them access to trading.Highly liquid markets with low volatility and hightransparency provide stable forums for exchange.

While these outcome characterizations of differentmarket institutions are important, they are not easilygeneralizable across settings. For example, liquidity isa key indicator of market quality for financial marketsbut it is an inappropriate measure in the case of Dutchflower auctions. First, the size of any one flower orderis limited, so it is hard to evaluate the impact on priceof large orders. Second, the flowers are sold from theauction to the buyer without further resale among the



buyers. Common outcome characterizations such as li-quidity do not apply to all market exchange relations.Outcome characterizations also do not help managersevaluate the potential impact of a new informationtechnology innovation on the operations of a tradingsystem, or the costs and benefits to differentstakeholders.

A model of key exchange processes is necessary toprovide a reference point for comparing the organi-zation of different market institutions. Traditional pro-cess classifications of exchange include search, bar-gaining and negotiation, monitoring, and enforcementprocesses (Coase 1960, Dahlman 1979).1 However, thislevel of classification does not provide sufficient detailfor the comparison of alternative market structures.What is required is a classification of exchange pro-cesses that is generalizable, focuses attention on salientdimensions of an exchange relation and provides suf-ficiently detailed information for effective classifica-tion of exchange features into independent processcategories. We adapt work by Kambil (1992) to buildsuch a classification scheme grounded in transactionscost and information-processing theories.

2.1. A Model of Exchange ProcessesKambil (1992) identifies ten distinct processes that canoperate in an exchange relation, providing a basis fora finer grained analysis of market structure and costs.This categorization also recognizes that basic tradeprocesses exist within a wider context of support pro-cesses that implement dispute resolution, and othermechanisms to ensure that counterparties to tradeswill meet their obligations. We adopt, refine, andextend his model into five trade processes (searchvaluation, logistics, payments and settlements, authen-tication) and five trade context processes (commun-

1 Coase (1960) states for “a market transaction it is necessary to dis-cover who it is one wishes to deal with, inform people that onewishes to deal and on what terms, conduct negotiations leading upto a bargain, to draw up a contract, to undertake the inspectionneeded to make sure that terms of the contract are being observedand so on.” This specifies a minimum of three sequential activities:search to gather information and determine products and parties totransactions; bargaining and negotiations to determine the terms ofexchange; and the construction and adoption of monitoring, polic-ing, and enforcement mechanism to ensure parties will satisfactorilyperform the exchange.

ications and computing, product representation, legit-imation, influence, and dispute resolution). The basictrade processes are distinct processes required in alltransactions of goods and services. The trade contextprocesses facilitate and enable or reduce the costs of or“frictions” in the basic processes. Figure 1 illustratesthese processes, which are defined and discussed inTables 1 and 2.

The ten process categories described in Tables 1 and2 provide parties to an exchange relation with the in-formation processing, coordination, and influencemechanisms required to mitigate uncertainty andexchange-related risks. While we cannot prove thatour categorizations define all processes related to ex-change, we propose that these categories define the keyprocesses underlying trades. The categories allow sys-tematic identification of processes affected by a specificadministrative or technological innovation and enableus to systematically examine and represent the impactsof information technology across interdependent pro-cesses and stakeholders in an electronic market.

The implementation and operation of the above pro-cesses and systems create transaction costs for differentstakeholders to the exchange. Transaction costs andthe complexity of these processes increase as productsand production become more complex, and as the mar-ket environment becomes more uncertain. Transactioncosts also increase as exchange parties establish morecomplex mechanisms to defuse opportunism risks.Traditional economics postulate that firms will chooseto organize exchange relations so that they achieve thelowest possible combined transaction and productioncosts. This would suggest the trading mechanismswith the lowest sum of transaction costs across all theprocesses will evolve and dominate the market. How-ever, real-market institutions are complex, constitutinga meeting point for stakeholders with both convergentand divergent interests. Innovations by a specificstakeholder may reduce overall transactions costs butcan increase the costs to other parties. This can resultin the failure to adopt a specific technology or admin-istrative innovation in the marketplace. Thus we pro-pose the need for a process-stakeholder analysis to sys-tematically evaluate the effects of technologicalinnovations on specific processes as well as costs and



Figure 1 Exchange Processes

Reprinted by permission from: “Doing Business in the Wired World,” Ajit Kambil, IEEE Computer, May 1997. Q IEEE 1997.

benefits that arise from the innovation on differentstakeholders.

Below we develop and illustrate the utility of aprocess-stakeholder analysis in focusing attention oncritical features of trading mechanisms, by using it tosystematically represent and evaluate the effects ofvarious information technology initiatives in the Dutchflower auctions (DFA).

3. The Dutch Flower AuctionsThe Netherlands is the world’s leading producer anddistributor of cut flowers and potted plants. This in-dustry consists of about 11,000 growers and nearly5,000 buyers. Growers typically are family businesses,while buyers represent both large and small wholesal-ers and retailers. The Dutch dominate the world exportmarkets, with 59% world market share for cut flowersand 48% world market share for potted plants in 1996.The Dutch flower auctions established and owned bygrower cooperatives play a vital role in Holland’s lead-ership of this industry by providing efficient centersfor price discovery, and the exchange of flowers be-tween buyers and sellers. The world’s two largestflower auctions are the Flower Auction Aalsmeer(VBA) and the Flower Auction Holland (BVH) in Hol-land. These auctions trade over 30 million flowers andconduct nearly 60,000 transactions daily, generatingtogether over $2.4 billion U.S. in annual sales of cutflower and potted products from the Netherlands andother producers such as Israel, Kenya, and Zimbabwe.

Annual sales of the seven Dutch flower auctions ex-ceeded $2.9 billion U.S. in 1996.

The flower auctions use the “Dutch” auction methodfor price determination. This method uses a clock, withthe clock hand starting at a high price determined bythe auctioneer and dropping until a buyer stops theclock by pushing a button to bid for a lot at the pricedetermined by the clock hand. This method was in-vented by a Dutch cauliflower grower in the 1870s totrade agricultural products. It reduces the time spentby growers at markets and frees them from the task ofbidding and price determination, thereby allowingthem to focus on production. It also reduces the timebuyers spend in bidding or bargaining for goods.

The flower auctions provide a central location for themeeting of buyers, with suppliers allowing for effi-ciencies in quality control, logistics, and product re-distribution. Flower Auction Aalsmeer (VBA) nearSchiphol airport and the Flower Auction Holland(BVH) near Rotterdam are located close to major trans-portation facilities and have some of the largest com-mercial buildings in the world. Both auctions spreadacross the equivalent of about 100 soccer fields, eachwith the capacity to host 2,000 buyers. At the VBA theflower auctions occur in different auction rooms under13 different clocks. The computerized clock in the auc-tion hall provides buyers with information on the pro-ducer, product, unit of currency, quality, and mini-mum purchase quantity. The flowers are transportedthrough the auction hall and shown below the clock tobuyers. When the buyer stops the clock, the auctioneeruses the intercom to ask the buyer how many units of



Table 1 Description of Basic Trade Processes

Process Description of Basic Trade Processes

1. Search The information gathering and evaluation process undertaken by buyers and sellers to identify opportunities. A tradingopportunity consists of a supplier willing to offer a good and a buyer interested in purchasing the good. Thus searchprocesses identify potential suppliers, goods, and customers willing to purchase the goods. Different marketmechanisms implement varied methods and venues to identify trading opportunities. Different search methods allocatedifferent search costs to buyers, sellers, or intermediaries.

2. Valuation (PriceNegotiation)

The process and methods of negotiating and discovering a purchase or sale prices for a product. A variety of differentprice discovery and bidding processes exist that differentially attributes costs to buyers, sellers and intermediaries. Forexample, various auction methods are specified in Davis and Holt (Davis and Holt 1993) for finding prices. Pricediscovery mechanisms can also be biased to shift surplus from the trade to specific transaction stakeholders.

3. Logistics The processes specifying and coordinating the actual delivery of goods from the seller to buyer. Again, in a trade thedirect costs of delivery may be assumed differently by different players.

4. Payment and Settlements These processes define the terms and methods of payment permitted and ensure the settlement of payments in theexchange. Third parties often provide the infrastructures for exchange.

5. Authentication This is a core set of activities used to verify the quality and features of the product offered, the authenticity of the tradingparties, and monitor conformance to the contract or agreement among parties. Authentication may involve third partieswho provide credit check, notarization, and other services to reduce the uncertainty of buyers and sellers.Authenticating the quality of a product prior to purchase mitigates against adverse selection risks (Akerlof 1970, Kleinet al. 1978). Monitoring processes in a complex transaction work to ensure that the trade is successfully completedand mitigate against moral hazard risks. Third parties to the transaction may play a critical role in monitoringtransactions, ensuring nonrepudiation and completion of agreements.

a lot he or she will purchase. The buyer provides thebid and, when the bid is accepted, the clock is resetand the process is re-started to clear the remaining in-ventory. The auctioneer may introduce a new mini-mum quantity to ensure the entire lot clears. Each clockcan handle nearly 1,000 transactions an hour (approx-imately one transaction in four seconds). On average,15 million flowers are traded daily at the VBA in 30,000transactions.

We investigated the use of information technologyin these auctions over the span of three years, collect-ing data through interviews with auction officials andparticipants and through review of all relevant pub-lished reports and archival material. Next we used theprocess categories identified in the previous section toorganize the information and systematically evaluateor compare the impacts of specific IT innovations andprocess changes on different stakeholders in the DFA.This process-stakeholder analysis was used to identifychanges in benefits and costs of different exchange pro-cesses, and the relative value of these innovations to

different stakeholders. A case method was used be-cause it enables rich data and “reality” to be capturedin greater detail than other methods. The case methodalso permits us to consider a greater number of ex-planatory variables. We then undertook a cross-caseanalysis on multiple IT initiatives in the Dutch flowermarkets to generate propositions that explain the suc-cess or failure of specific innovations and more generalimpacts of new information technologies on markets.While we acknowledge that a single case cannot begeneralized into a theory, a cross-case analysis is a use-ful step in theory construction by identifying patternsthat provide testable propositions (Eisenhardt 1989,Yin 1981).

3.1. An Evaluation of the Traditional DutchFlower Auction

Utilizing the previous process categories, Table 3 sum-marizes the key features, strengths, and weaknesses ofthe traditional Dutch flower auction. This provides abaseline for comparision and evaluation of processchanges enabled by information technology.



Table 2 Description of Trade Context Processes

Process Description of Trade Context Processes

6. Communications andComputing

Communications and computing processes underlie and bind all trading processes. New communications capabilities interms of richer media, faster transfer speeds, improved ease of use and lower infrastructure costs transformcoordination capabilities within and across trade processes. Similarly, improved processing, storage, input-output, andsoftware technologies, further transform the coordination and decision making capabilities of stakeholders in eachprocess. These transformations change the specific transaction costs and opportunity costs perceived by buyers,sellers, and intermediaries.

7. Product Representation Product representation processes determine how the product attributes are specified to the buyer or other third parties.To what extent is there a standardized product and quality specification language? Both the buyers’ and sellers’ costsare reduced by informational standards or well-specified languages to represent the key quality, functional andaesthetic attributes of the product. If the product is well represented, the uncertainty of the buyer is lower. The costsand methods of implementing a uniform product representation scheme vary in different market mechanisms.

8. Legitimation This process is used to validate the trade or exchange agreement. It defines how bids, or agreements are recognized asvalid and binding commitments by exchange parties. In different trading situations, different procedures and rituals areundertaken to legitimate a transaction. Often this is done by announcing the agreement, and participants to theagreement, to a broader audience through a formal process.

9. Influence Structures andProcesses

These processes implement mechanisms to enforce obligations or penalties to reduce opportunism risks. Kambil (1992)defines two distinct influence processes developed from social exchange and transactions cost theories. These are:incentives and sanctions, and credible commitments. Incentives and sanctions include performance rewards forcompleting a contract on time, or penalties for misrepresenting products, or denial of access to an exchange network.A reputational system defines a special class of incentives and sanctions. Actions taken to enhance and create a goodreputation serve to create trust in the party. When a social system establishes a reputation as a valued asset to sellersor buyers, opportunistic behaviour that can sully the reputation becomes more costly.

Investment in credible commitments (Williamson 1985) or asset-specialized investments by both exchange parties createshostages such that the loss of the relationship will create a significant cost for both parties. This serves to signalcommitment to the relationship by both parties and mitigates against opportunism risks. The types of commitments,the importance of reputation, and institutional authority to create incentives and sanctions can serve to reduce risksand the costs associated with other processes.

10. Dispute Resolution All exchanges occur in a broader legal or institutional context which provide various processes for resolving disputesamong parties and structures decision rights in the event of conflict. Firms may elect to directly resolve disputesthrough negotiation, commit to third-party arbitration or court ordering of disputes. The latter is typically the mostexpensive alternative. Dispute-resolution processes adopted in an exchange relation, and the allocation of decisionrights to structure solutions to disputes are crucial to reducing risks for buyers and sellers. Where goods areexchanged, disputes can be resolved by replacement of a defective product, and warranty mechanisms.

The Dutch flower auctions provide buyers and sell-ers with a highly efficient infrastructure to supportexchange. As summarized in Table 3, they provideefficient search, communication, and product repre-sentation capabilities for highly varied products. Theauctions are also very efficient in determining price,enabling approximately 1,000 transactions per clockper hour on a low-value good. The auction methodeliminates problems of haggling and reduces bargain-ing costs that are otherwise associated with the trade

of nonstandard and unusual-value products (Milgromand Roberts 1990). By operating at a high clock speed,encouraging competition among buyers, and setting afixed time by which to complete a transaction (or theclock goes to zero), the auction enables efficient trad-ing. The hub and spoke operation and the logisticswithin the auction hall allow the timely transfer of alarge volume of flowers from suppliers to buyers. Byintermediating all buyer–grower exchanges, the auc-tion also provides efficient quality control, settlement,



Table 3 Features of the Traditional Dutch Flower Auctions

Exchange Process Implementation Strengths Weaknesses

1. Search All trading opportunities are found at theauctions.

Search costs are reduced as thecentralized auction aggregates majortrading opportunities in one location.

The buyers have to come to the auctionhalls, and sellers have to deliverproducts to the auction hall.

2. Valuation The Dutch auction clock method. Very efficient for small lot trading. Alltransactions occur in a fixed time slotleading to 1,000 transactions perhour.

Auction prices for specific productsdecrease during the day. To assurefairness, assignment of auctionsequence for different growers is doneby lottery. Clock speeds and small lotsstretch the cognitive capabilities ofbuyers favoring higher prices forgrowers.

3. Logistics The auction facilities are a central hubproviding logistics support fortransferring products from sellers tobuyers. These costs are shared amongall buyers.

Very efficient transfer of product fromseller to buyers. The centralizedauction trading allows specialization,transportation, and other resources.

Packaging costs are incurred multipletimes—for transport to and from theauction. Multiple handling of flowerscan damage them.

4. Payments andSettlements

The auctions provide systems for ordertracking, payments, and settlementswith efficient one-day settlementperiods.

Shared costs among auction participantsand economies of scale in thesesystems.

5. Authentication The auction house authenticatesparticipants and grades the quality ofthe products. It is also responsible fortracking and ensuring delivery oforders to buyers.

Very efficient for large numbers trading.The auction reduces counter-partyrisks, product, and relatedauthentication costs for buyers andsellers.

Buyers and growers perceive qualitygrades as too broad, artificiallyinflating valuation of products at thelower end of quality rating.

6. Communication andComputing

Simple visual communications ofcompetitors, product, price, and othertrading information on the clock. Asimple computerized system forcommunicating bids.

Efficient and low costs for sharedcommunications infrastructure.

Requires synchronous communicationsfor trading, and colocation of partiesto the limited trading floor. Growersdo not know final demand patterns forproducts.

7. ProductRepresentation

The product is represented by itself, andsimple codes identifying grower and agross level quality grade.

Buyers can directly inspect the productin the auction hall if they want to doso.

Buyers and growers perceive qualitygrades as too broad. The cost ofvisual inspection.

8. Legitimation The auction intermediary is responsiblefor recording bids and legitimatingtransactions.

9. Influence The auctions can exclude growers orbuyers who do not meet variouscriteria.

The auctions, by centralizing andrequiring the product to be deliveredprior to sale, minimizes opportunismrisks.

Auction rules tend to favor the growers.

10. Dispute Resolution The auctions provide for variousarbitration mechanisms and rules forresolving problems. This reducesdispute resolution costs.



and dispute resolution mechanisms that mitigateagainst opportunism risks encountered by buyers andsellers. These features reduce transaction costs togrowers and buyers.

However, as summarized in Table 3, there are dis-advantages of this specific model of trading. First, buy-ers must be physically present at the correct locationand correct time to bid on a specific product. In addi-tion, it is hard to simultaneously search other marketsto estimate prices. Today, large buying organizationsprovide cellular phones to their buyers to coordinatepurchases across auctions. In response, the auctionsare beginning to coordinate trading similar productsat the same times in different auction houses. This willallow the large buyers to find the best price across auc-tions. The auction structure also inhibits growers fromunderstanding the true preferences of buyers, as grow-ers are separated from the sale of their produce by theauction.

Other disadvantages for specific stakeholders areunderstood when we consider the auction’s ownershipand incentive structure. A number of auction rules fa-vor growers, who are the auction owners, over buyers.First, the clocks move at a high speed. Higher speedsreduce the decision time available to buyers. This re-sults in higher bid prices. The auctions have experi-mented with different clock speeds and it is widelyknown that faster clock speeds result in higher prices.However, the auctions are unwilling to fully disclosethe results of their experiments and selection of auctionclock speeds. Second, the auction rules and servicecosts to buyers for processing trades favors trading insmaller lots instead of purchases of large lots. This en-sures that no one party can purchase the entire lotwithout competition. It also increases the cognitivecomplexity and competition confronted by buyers.Buyers must now purchase products and adjust pur-chasing decisions to availability across multiple clockauctions. This imperfect information and low trans-parency about available inventories favors growers, al-though more empirical work needs to be done to dis-cover the magnitude of the growers’ advantage.

3.2. Challenges to the Traditional Dutch AuctionsIn the early 1990s the Dutch flower industry and auc-tions faced a number of challenges. First, rapid growth

in the flower industry created increasing demands forlogistics support, space in the auctions and complaintsabout nearby traffic congestion. As the two major auc-tions were close to their limits in terms of complexity,capacity, and room to expand, increases in capacitywould require changes in auction logistics or invest-ment in new facilities. A second problem confrontedby the industry was the consolidation of buyers intolarger entities. Buyers for supermarkets and large retailstore chains wanted to purchase larger lot sizes andcoordinate purchases across different markets to takeadvantage of the best price. Some large retailers, likeMarks and Spencer, were bypassing the Dutch auc-tions and their commissions to source directly fromgrowers in Spain and other countries. Third, the suc-cess of the Dutch auctions as the world’s leadingflower auctions and entry point into downtream logis-tics and distribution led to an increased flow of foreigncut flowers to the Dutch flower auctions. In 1994, 15%of flowers traded at the largest Dutch Flower auctionswere imported from and re-exported to foreign coun-tries (sometimes back to the country of origin). Thisincreased foreign competition to Dutch growers, andin 1993 the import of foreign rose stems to the twolargest auctions reduced average prices by 40% duringthe winter season. Subsequently, the Dutch grower co-operatives voted to limit foreign access to the majorauctions in 1994. These challenges led to a number ofsuccessful and unsuccessful information technology-enabled transformations of trading in the industry.These initiatives are critically examined below, usinga process-stakeholder analysis, to illustrate major pro-cess changes and benefits or costs of the initiatives todifferent stakeholders.

4. IT-Enabled Dutch AuctionTransformations: A Process-Stakeholder Analysis

Four major information technology-enabled initiativeswere undertaken to transform the Dutch flower auc-tions and respond to the above challenges. Two un-successful initiatives were the BVH’s Holland Vidi-fleur system and the VBA’s Sample-Based Auction(SBA) system. These initiatives promised to simplify



Table 4 Dutch Flower Auctions Initiatives

Traditional DFA Holland Vidifleur System VBA Sample Based Auction Holland Supply Bank Tele Flower Auction System

Start (End) Year 1890 1991 (1991) 1994 (1994) 1993 1993

Product Flowers andPotted Plants

Potted Plants Potted Plants Flowers and PottedPlants

Flowers

Sellers Growers Growers Growers Growers Growers

Intermediary Flower Auctions Flower Auction Holland(BVH)

Flower Auction Aalsmeer(VBA)

Flower AuctionHolland (BVH)

East African Flowers (EAF)

Buyers Wholesalers Wholesalers Wholesalers Wholesalers Wholesalers

Valuation Dutch AuctionClock

Dutch Auction Clock Dutch Auction Clock Negotiated PostedOffer

Dutch Auction Clock

auction hall logistics and add extra transaction capac-ity to the auction. Two successful initiatives were theHolland Supply Bank (HSB) system and the TeleFlower Auction System (TFA). The former system re-sponded to the need of large buyers, and the latter pro-vided a new auction for foreign produce. While theauctions have used information technology exten-sively to support order processing, inventory, and lo-gistics, they primarily supported or automated exist-ing auction processes having impacts typically limitedto the auction intermediary (Van Heck and Groen1994). The IT-enabled initiatives discussed below fun-damentally changed the implementation or conduct ofone or more specific auction processes, and the way inwhich buyers, auctions, and sellers interact. Summaryinformation on the initiatives is given in the Table 4.

4.1. The Vidifleur InitiativeThe BVH auction implemented the Vidifleur initiativein response to the growing capacity and logistics re-quirements of the auction. Vidifleur intended to usevideo auctioning to decouple the price determinationand logistics mechanisms and to allow buyers to tradefrom outside the auction hall. When the product ar-rived at the auction a picture was taken, digitized, andstored in auction computers. These computers trans-ferred the picture for display to a screen in the auctionhall, where buyers could bid for the product based onthe image of the product. Buyers were also able to bidfor and look at the flowers on computer screens in theirprivate auction offices. The computers in the private

office provided a screen-based representation of theclock which was synchronized with the clock in theauction hall.

The Vidifleur experiment initiated by the BVH in-formation technology staff (Spooner and Copeland,1992),2 was implemented with minimal changes to theoriginal auction clock process. Experimenters con-verted one of three auction clocks in a room to havevideo screens for product display around the clock.The auction also moved the real product under theauction clock, providing a second visual display of theproduct. An auction room solely dedicated to this ex-periment was unavailable because of the demand forlimited auction facilities, and altering existing logisticsprocesses was deemed too expensive for theexperiment.

Auctioneers expected this remote video auctioningto provide buyers with better information, as theycould work from the comfort of their own offices andalso access their office computers for purchasing, or-der, sales, and local inventory information. The auc-tioneers also knew that buyers often tended to selectgoods from specific sets of growers rather than inspectthe product in great detail and expand the selection ofproducers. This buyer behaviour suggested thatgrower reputations played a substantial role in shap-ing purchases, and that the visual inspection of flowers

2 Spooner and Copeland (1992) provide a teaching case of IT initia-tives at BVH (previously called Flower Auction Westland) in whichthey discuss the motivation and technology used to implement theVidifleur experiment.



Table 5 Process-Stakeholder Impacts of Vidifleur

Exchange Process Auction Intermediary Buyer

1. Search No change. No change.2. Valuation No change. No change.3. Logistics Expected: Auction hall logistics would be eliminated

allowing for cheaper and more frequent transactionsand new space for clocks.

No change.

Actual: No change during the experiment due topotential disruptions to the current process. Realproduct appeared in the hall synchronized to clock.

4. Payments and Settlements No change. No change.5. Authentication No change in the quality grading process. In a full-scale system, buyers would find it harder to

visually inspect and authenticate quality.6. Communication and Computing Shift to electronic communications. Electronic

representation of the clock as well as the product.Expected: Electronic communications could substitute

for buyer presence in the auction halls.Reality: Electronic communications does not capture

nonprice information, such as other bidders’expressions and reactions to specific bids. Itreduces social interaction opportunities.

7. Product Representation Expected: Electronic flower clock and videorepresentation of flowers would not adversely affectthe sale.

Expected: Buyers would make decisions from theiroffices using computers rather than the auctionhall.

Result: Negative buyer reaction to video quality. Result: Negative reaction. Buyer perceive the videoquality as poor.

8. Legitimation No change. No change.9. Influence No change. No change.

10. Dispute Resolution No change. No change.Net Benefits None. Negative.

prior to purchase was less important to the process.Despite these expectations, buyer reaction to screen-based trading was negative and led to the terminationof the experiment in late 1991. Table 5 summarizes theareas of key process changes, benefit expectations andactual results of the initiative. We focus only on buyersand the auction intermediary, as no major processchanges occur for sellers.

As highlighted in Table 5, buyers cited three mainreasons for not adopting the new system. First, theclock-based trading system provided no new efficien-cies for buyers. Second, the quality of the auction hallvideo display was perceived as poor, and trading fromoutside the auction hall created an informational dis-advantage. In floor-based trading the buyers could ob-serve each other, and the reactions of other major buy-ers (from supermarket chains, etc.) to specific bids.This important nonprice information was incorporated

into the decision making of the buyers. Video-basedauctioning on a computer was a limiting medium, notrich enough to capture this information. Thus buyershad limited advantage in trading from their offices,when there was also a floor-based auction which pro-vided the traders with more information. Indeed, buy-ers who tried to trade from offices complained thatthey could not feel the “tension” in the marketplace.Third, at the back of each auction hall is a coffee shopwhere buyers interact informally and share informa-tion about the market. Again, access to the social in-teraction and information was more difficult throughscreen-based trading.

4.2. The Aalsmeer Sample-Based AuctionIn contrast to the BVH, the Aalsmeer auction began asample-based auction for trading potted plants(Griffioen 1994a, 1994b; Van Heck and Groen 1994). In



this model, growers send a sample of the product tothe auction house along with information on availableinventory. During the auction the sample representsthe entire inventory available to buyers who can bidfor the product and specify product packaging and de-livery requirements. Growers then package the prod-uct as specified and deliver it the next day to the buyerlocation in the auction complex or to other buyer ware-houses. Growers, buyers and auctions used electronicdata interchange to share all the information requiredin this process.3 This trading model reduces the num-ber of times a product is handled, reducing overallpackaging costs and damage.

The different actors—the auctions, growers, andbuyers—expected a number of different benefits. First,by uncoupling logistics and price determination, theauctions and growers expected the number of trans-actions per hour to increase. In reality the number oftransactions per hour decreased, as buyers had to spec-ify terms of delivery. Second, while the auctions ex-pected 45% of the supply of potted plants to be trans-acted in the sample-based auction, only 10% of theproduct was transacted this way. Thus, sample-basedauctions also did not effectively reduce storage re-quirements at the auction. After numerous attempts toincrease the volume of sample-based auctions, theywere discontinued in late 1994. Table 6 summarizesresults of the initiative. Only areas of change are dis-cussed below.

The sample-based auction failed to meet expecta-tions for many reasons. First, the incentives and bene-fits to buyers and growers (in particular) did notchange substantially to encourage their participationin this market. Specifically, growers received no extracompensation for modifying packaging and deliverypractices to suit the customer. Second, the growers per-ceived that they got lower prices in a slower auction.To overcome this disadvantage growers would breakthe same product into different sample lots so that it

3 Beginning in 1992, growers and auctions adopted electronic datainterchange to communicate product, order, and transaction infor-mation to each other. EDI reduces double entry of data into systemsand paperwork, and reduces errors. Growers who use EDI can sub-mit data on shipments until 3:00 p.m. instead of 12:00 p.m. for papertransfer.

would be priced multiple times during the auction,hoping it would lead to higher prices. Third, the auc-tion rules initially did not provide incentives to buyersby supporting transactions on large lots. Instead, theauction maintained rules to favor transactions in smalllots. Thus, an insufficient number of buyers and sellersinitially adopted this new form of trading (Van Heckand Ribbers 1996).

In response to this failure, the auction undertookvarious rule changes beginning in March 1994. First,the auction established a price floor of 70% of averageprice of a flower type in the most recent five days toreduce volatility and downside risks. Second, the rep-resentative lot was auctioned first. This sample lot typ-ically received a higher price than following lots of thesame type and quality. Buyers believed the first sampleto always be of best quality within any quality rating.Third, the lot size was increased to three stapelwagens4

from one. The growers were also forced to increase thenumber of plants offered in any one transaction, so thatthey did not game the system. To meet the needs oflarge buyers, buyers were permitted to buy either oneor three stapelwagens per transaction. To preventgaming, the grower for any type of flower was onlyallowed one auction per category of product for pricedetermination. These rule changes temporarily stabi-lized the market, but the lack of growers and buyersadopting the method led to its demise in late 1994.

4.3. The Holland Supply Bank: ImageRepresentations and Negotiated Trading

To respond to the needs of larger buyers, the BVH andAalsmeer auctions have diversified to create a bro-kered market for flowers and potted plants. In the “be-middelingsbureau” (BB), or Mediation Office, an auc-tion employee acts as an agent for the growers andnegotiates between growers and buyers in a forwardmarket. Prices, product specifications, amount of lots,and delivery specifications are specified in a contractwhich is legitimized and monitored by the mediationoffice. In 1994, about 18% of flowers and potted plantswere traded this way. In 1996, this amount increasedto 22%. The mediation office is useful for the sale of

4 Stapelwagens are a fixed-size cart used to transport flowersthrough the auction complex.



Table 6 Process-Stakeholder Impacts of the Sample Based Auction

Exchange Process Growers Auctions Buyers

1. Search No change. No change. No change.2. Valuation Expected no change.

Result: Classify into smaller lot sizesso same product is sampled andeffectively priced multiple times.

Price is on average 10% less onremainder than sample lot. Thisresults in less revenue.

Expected no change.Auction process is slowed down due

to the need to specify packinginformation and buyer gaming.

Lower commissions on remainder forsample lot.

70% of previous five days’ averageprice as a price floor.

Greater uncertainty about productquality leads to discounting ofnonsample lots.

3. Logistics Increased burden from needing tocustomize packaging to buyerspecifications.

Lower logistics costs. More efficient delivery of the product.

4. Payments and Settlements No change. No change. No change.5. Authentication No change. Minor change ensuring product is

properly delivered.Inability to authenticate quality ahead

of time. Must verify product isproperly received.

6. Communication and Computing Minor change: More use of EDI tosend coordination information.

Minor change: More use of EDI tosend coordination information.


7. Product Representation Enabled to select a sample lot that isof higher quality than theremaining product.

Sample lot substitutes for actualtransaction lot.

Buyers cannot visually evaluate thewhole offer.

Must decide to buy based on thesample lot.

8. Legitimation No change. No change. No change.9. Influence More difficult due to more complex

transaction.No change. No change.

10. Dispute Resolution No change. No change. No change.Net Benefits Negative. Negative. Negative.

large lots to large buyers, like supermarkets, for theoccasions market.

The mediation office, as an honest broker, reducesthe costs of search, communications, and bargainingfor buyers and sellers. It also provides a mechanism tolegitimate the transaction and resolve disputes in theevent the contract is not met. Image databases of prod-uct types and inventory are especially useful for trans-actions in the mediation office. Wholesale buyers findthat the pictures of the product are also very usefulmarketing tools to downstream retailers. Electronicdata interchange is used for communications of ordersand the coordination of settlements, and delivery. Thisresults in fewer errors and greater transactionefficiency.

Table 7 summarizes the major impacts of the Hol-land Supply Bank on different stakeholders.

4.4. The Tele Flower Auction SystemSince 1993, the Dutch flower industry has been con-cerned about the increasing flow of foreign flowerproducts into the Dutch flower auctions and down-stream logistics and distribution systems. These im-ports were having an adverse effect on the margins ofDutch growers and the prices of key commodity prod-ucts like roses and carnations. After a referendum inSeptember 1994, the growers, who are the owners ofthe Dutch auctions, decided to ban foreign grower par-ticipation in the auctions during the summer. Theseefforts to reduce foreign access to the traditional Dutchauctions led buyer organizations and foreign growersto announce the creation of competing auctions. In par-ticular, the East African Flower (EAF) importers asso-ciation responded to this ban by developing the TeleFlower Auction (TFA) (Van Heck et al. 1997).



Table 7 Process-Stakeholder Impacts of the Holland Supply Bank


1. Search Benefit: The HSB permits growers tosearch for buyers.

Benefit: Efficient image databasecombined with broker matchingsystem.

Benefit: Large buyers’ search costsare reduced.

2. Valuation Expected no change.Benefit: Slightly higher prices with

less uncertainty.

Mediation process is less costly thanlogistics-intensive process.

Cost: Slightly higher prices butreduced uncertainty.

3. Logistics Benefit: Efficient product transfer inbulk to buyer.

Benefits: Less logistics in the auctionhall.

Benefits: Direct delivery from sellerwith no repacking costs.

4. Payments and Settlements As specified in the contract. Managed by mediation office. As specified in the contract.5. Authentication No change. Costs: Quality managers evaluate

sample, and assign a qualitygrade. Monitor and ensuredelivery.

Cost: Inability to authenticate qualityahead of time.

6. Communication and Computing Minor change: More use of EDI tosend coordination information.



7. Product Representation Cost: Growers send data and samplelot. Growers perceive qualitygrades as broad and don’t reflecttrue quality.

Cost: Quality managers create adigital image of sample lot topresent to buyers in addition tothe rating.

Cost: Buyers cannot visually evaluatethe offer and perceive qualitygrades are broad. Benefit:However, images are useful fordownstream trade.

8. Legitimation No change. Mediation office. No change.9. Influence No change. No change. No change.

10. Dispute Resolution No change. Mediation office. No change.Net Benefit Positive—higher prices and less

uncertainty.Positive—less logistics requirements. Positive—ability to transact for large

lots into the future.

TFA is a completely electronic auction that enablesbuyers to trade at a distance. Decoupling the logisticsand price discovery processes, the buyers can bid fordifferent flowers via their personal computer (PC)screens (Bos 1995, 1996; Eras 1995; Van Vliet 1994). Us-ing ISDN services of the Dutch PTT, each buyer’s PCis connected from his or her office to a fully comput-erized Dutch auction clock. The PC provides infor-mation on the producer, product, unit of currency,quality, and minimum purchase quantity. For addedconvenience, the PC also provides the buyer with tex-tual information on the available flowers, specific lotsfor sale, and two digital images of each lot. One imagegives an overview of the lot to inspect ripeness of flow-ers, and the other presents details of the flower bud,to evaluate its size and diagnose potential diseases.The buyer can then use the PC’s custom software toflag interesting lots, so that the computer can alert thebuyer when it is time to auction the specific lot. On the

PC screen the buyer sees the Dutch auction clock. Theclock hand starts at a high price and drops until abuyer stops the clock by pushing the space bar at thekeyboard of the PC. The auctioneer then asks the buyervia an open telephone connection how many flowersof the lot he or she will buy. When the buyer providesthe amount, the auctioneer legitimates the transaction.The clock is then reset, and the process begins for thenext lot until the remainder of the product is sold.

The auction logistics are substantially simpler thanthe traditional DFA. The EAF receives and stores flow-ers and plants from growers in a distribution ware-house facility in the nearby town of Amstelveen. Theflowers from the Amstelveen facility are transportedto the wholesaler’s address directly by transporters ofEAF. In an agreement with the VBA, the EAF was alsopermitted to deliver products to the buyer’s rented fa-cilities at the VBA. This way the auction minimizesmost internal logistics operations and is able to deliver



flowers to buyers in a timely way. The major processchanges relative to the traditional Dutch auction aresummarized in Table 8.

Overall, TFA was an immediate success. Since itsimplementation in March 1995, it has grown from auc-tioning the product of 2 main East African growers toinclude products from 40 growers in Africa, Spain, Co-lombia, France, India, and Israel. By the end of theyear, the TFA had nearly 160 buyers (about half arelocated in the Aalsmeer area; the rest are scatteredthroughout the Netherlands). The TFA was estimatedto be the fourth largest flower auction in Holland bythe end of 1996.

Why were buyers willing to accept this new auction?First, it uniquely satisfied the buyers’ needs to inex-pensively purchase foreign flowers and include themin their existing distribution chains. Second, buyers weinterviewed said the flowers they purchased from theTFA often exceeded their quality expectations as de-termined by the quality grades. Thus, the TFA createda reputation for setting high quality standards, en-hancing trust in the auction and minimizing adverseselection. If necessary buyers close to the warehouselocation could physically inspect the product. Buyerswere also impressed with the computer data, the qual-ity of the products, and the speed of delivery after theirpurchases were completed. The buyers also felt theyhad a better overview of the auction, as compared withthe traditional Dutch auction.

5. Lessons Learned from IT-Innovations in the Dutch FlowerAuctions

Based on our research we did a cross-case analysis toidentify factors that influence the successful or unsuc-cessful use of information technology in markets. Weuse this analysis to develop a series of observations onthe market impacts of IT as well as the success andfailure of IT-based initiatives in markets. These obser-vations stated as propositions, and the supportinganalysis and discussion are given below.

Proposition 1. The application of information tech-nologies to trading can enable increased efficiency and sep-aration of informational and physical trading processes. This

in turn will permit more varied forms of trading, customizedto different user requirements.

The four cases illustrate the use of information tech-nology to separate the informational and physical trad-ing processes. In all cases the valuation and logisticprocesses are increasingly de-coupled from one an-other in time and space. For example, the Holland Sup-ply Bank illustrates the implementation of the media-tion bureau to broker trading of large futures contractsbetween large buyers and different growers separatinglogistics and information processes. Similarily, in theTele Flower auction, the buyers are able to bid forproducts from remote locations. The buyers can cus-tomize the system by specifying preferences and re-ceiving alerts when the product they want is beingauctioned.

These observations also lend support to Malone etal. (1987), who note that information technology en-ables personalized markets where software agentswould support trading customized to individuals andfirm preferences. While the new flower auctions arenot as fully advanced, they achieve a better fit betweenthe transaction preferences of larger buyers and theorganization of exchange processes by overcoming thecapacity limits and colocation requirements for floor-based trading. These new trading mechanisms bettersuit the unique requirements of large versus smallbuyers.

The observations across cases illustrate that infor-mation technology substantially reduces coordinationcosts and enhance communications capabilities. Thesechanges allow transacting parties to separate in space(e.g., TFA bidding process) and time (e.g., HSB deliv-ery logistics) the informational and physical compo-nents of trading processes. Indeed, the informationaland physical components within each of the ten ex-change processes defined in this paper could be de-coupled from one another and provided by differentactors without the necessity for colocation. Over timethis will permit the creation of more distributed andcustomized market institutions for the trade of physi-cal goods.

Proposition 2. Information technology-enabled pro-cess innovations change the information available to differ-ent stakeholders. When different stakeholders perceive these



Table 8 Process-Stakeholder Impacts of the Tele Flower


1. Search No change from DFA. Benefit: Efficient databases andonline system for presenting tradeopportunities.

Benefit: Up-to-date online databasereduces search costs for specificproducts. Alerts inform buyers ofopportunities.

2. Valuation No change. Very efficient screen-based auctionclock process.

Benefit: Screen trading enables betteroversight and concentration onbidding.

3. Logistics Benefit: Efficient product transfer toAmstelveen warehouses.

Benefits: No auction hall logistics arenecessary. Less errors due tosimplicity of a decoupled system.Cost: Transport to buyer locations.

Benefits: Faster delivery from auctiondue to decoupling of logistics andprice discovery.

4. Payments and Settlements 24-hour payment and settlementclearing.

Managed by EAF. 24-hour payment and settlementclearing.

5. Authentication No change. Costs: quality managers evaluatesample, and assign a strict qualitygrade. Monitor and ensuredelivery.

Benefit: Auction quality controlreduces risks—may exceed buyerexpectations. Cost: Inability toauthenticate quality ahead of time.

6. Communication and Computing Simplified phone and datacommunications.

Simplified phone and datacommunications.

Text representations and digitalimages enable simplifiedcommunications.

7. Product Representation Cost: Growers perceive qualitygrades as broad and don’t reflecttrue quality.

Cost: Text representation of productin addition to the quality rating.

Cost: Buyers cannot directly evaluatethe offered product. Benefits:Screen-based representationsenable remote trading.

8. Legitimation Benefit: Growers don’t have to be anEAF member, enabling foreignimports.

EAF is primary authority. No change.

9. Influence Growers have little influence onauction policies.

EAF organization defines the marketrules.

Buyers do not have direct influenceon auction rules.

10. Dispute Resolution No change. No change: Arbitration by the auctionreduces costs.

No change.

Net Benefit Positive: for excluded foreigngrowers.

Positive: lower cost auction model. Positive: product access and remotebuying.

changes reduce the relevant information or trust, they willact to either increase the information available or discountprices to account for increased risks.

Market organizations, like other governance sys-tems, are information processing systems that repre-sent a consensus among different competing stake-holders on organizing the interdependent processesidentified in § 2. These processes serve to generate ororganize information, or to reduce uncertainties andmitigate opportunism risks faced by different stake-holders (Kambil 1992). The prior literature on marketsand technology generally presumes that information

technology applications enable the provision of morerelevant information to decision makers (Bakos 1991b,Brynjolfsson et al. 1994, Gurbaxani and Whang 1991,Malone et al. 1987). However, the cases above illustratethat IT can actually reduce the relevant informationavailable to decision makers. The application of infor-mation technology to communication and product rep-resentation processes had major impacts on relevantinformation available for price discovery, authentica-tion, and logistics processes. When information or trustas perceived by specific stakeholders was reduced inthese processes, they or other interested stakeholders



acted to increase the information and trust, or reducedprices to discount for new risks.

For example, in the SBA, the buyers chose to dis-count the prices bid for non-sample lots by nearly 10%because they could no longer authenticate quality byvisual inspection. Similarily, growers in the SBA de-creased lot sizes or reclassified their product so thatthe same product was displayed multiple times. Thegrowers’ objective was to have multiple new biddingcycles for price determination, hoping that multiplebidding competitions for smaller lots among buyerswould generate new demand information leading tohigher prices. Both of these responses reduced the at-tractiveness and efficiency of the sample-based auc-tion, leading to its ultimate demise.

While the Holland Supply Bank and the Tele Flowerauction reduced the ability of real-time authenticationof the product, they provided new benefits to the dif-ferent stakeholders or created trust in new ways. In theformer, parties were able to trade in a way that re-duced uncertainty and generated information aboutfuture availability of demand, prices and goods. In thelatter, to create trust among buyers in the new auctionprocess and establish a reputation for quality, the TeleFlower auction ensured that the product matched orexceeded buyers’ expectations given a specific qualityrating. While we cannot establish that the product wastraded at a lower price than when it was visually in-spected, the tight quality control that assures a productmatches or exceeds its quality ratings leaves it open tobeing discounted to the price level of the quality ratingrather than the price of the flower given perfect qualityinformation.

The above observations are consistent with prior re-search on the governance of exchange relations. First,the need for information seeking to reduce uncertaintyis well known. The generation of trust (the expectationof future nonopportunistic behavior) and preservationof good reputations is also important to reduce trans-action costs or “friction” in exchange (Arrow 1974,Fukuyama 1995). Third, social exchange theorists haveidentified the importance of equity and fairness (Blau1964, Homans 1961). The cases illustrate that reduc-tions in information available due to an IT-enabledprocess discontinuity creates new information seeking

or trust-generation costs. Market designers can re-spond to these requirements by allowing more infor-mation gathering by different stakeholders (buyersand sellers) or create trust and reputations for fairness.Both these choices lead to differential levels and allo-cations of costs across stakeholders. These costs andtheir distribution across stakeholders must be consid-ered by designers of new IT-enabled market mecha-nisms if the new market is to be adopted.

Proposition 3. Market organizations are the meetingpoint for multiple stakeholders: buyers, sellers, and inter-mediaries with conflicting incentives. Given existing orcompeting market alternatives, no new IT-based initiative islikely to succeed if any key stakeholder is worse off after theIT- enabled innovation.

The process-stakeholder analysis framework devel-oped in § 2 of this paper was applied to map processchanges and their impacts on different stakeholders. Inthe two cases of failure, the application of the frame-work clearly identified either the grower or the buyerwas worse off from the innovation. For example, in theSBA the grower incurred new packaging costs and lo-gistics costs. In the Vidifleur auction, the buyers didnot perceive a new benefit from the system. The videoquality was poor, authentication of quality less con-venient, and trading online did not provide all the in-formation available in the auction hall.

In contrast, the Holland Supply Bank and the TeleFlower auctions created new advantages for all stake-holders who participated. In the Holland Supply Bank,buyers were able to order large quantities of productand specify a contract for delivery on a future date,reducing the growers’ uncertainty about future pricelevels. The Tele Flower auctions provided Dutch buy-ers with a source of foreign flowers, foreign growerswith a trading location that could take advantage ofthe Dutch wholesaler’s demand for flowers, anddownstream logistics facilities.

These cases illustrate the need to carefully designand implement electronic markets and to include ad-equate incentives to different stakeholders for partici-pation. All stakeholders must be better off with thenew system rather than the old system, otherwise thenew market will fail and revert back to the prior sys-tem if this alternative remains available. The process-stakeholder analysis framework can be used by marketdesigners to evaluate the consequences of their designs



on processes and stakeholders to fulfill the aboveconditions.

Proposition 4. Biases in trading will depend on theownership of the market and the availability of cost-effectivetrading alternatives. As owners of existing markets may notbenefit from electronic trading, this incentive incompatibil-ity will impede the transition to new electronic markets. Asolution to this problem is to change market ownership orcreate new sources of competition.

Why did the traditional Dutch markets fail to suc-cessfully adopt an electronic trading system? Why didthe leading auctions allow the Tele Flower auction toreframe competition among the auctions? In part, thevarious electronic initiatives of the traditional auctionsdid not fulfill compelling needs among the differentstakeholders. In addition, the IT-based trading initia-tives were embedded in a specific organizational con-text and set of assumptions that limited the initiatives.

Historically, the Dutch grower cooperatives haveowned the auctions and have been the dominant influ-ence on the use of technology in the auctions. The well-organized growers who own the auction intermediaryestablished a consensus on the trading rules and or-ganization that best favored specific grower groups—from clock speeds to auction mechanism.

The technology-enabled initiatives of the existingauctions, with the exception of the Holland SupplyBank, generally assumed the existing valuation modeland auction structure. As auction officials mentionedto us in interviews prior to the Tele Flower initiative,it was unlikely that any other form of trading or pricediscovery mechanism would be approved by the own-ers other than the Dutch auction. The innovations werethus embedded within the context and framework ofthe traditional auction model. Converting this modelto an IT-based representation had the potential for anumber of advantages to sellers. For example, by hav-ing a very fast auction clock, and trading in small lotsin the traditional auction, the sellers found they couldget higher prices. Electronic trading that preservedthese features benefits the sellers but creates more costsfor larger buyers, who are increasingly important inthe market. Other price discovery mechanisms andtrading rules, including larger lots, can be more effi-cient for large buyers.

As buyers become more powerful, and move to by-pass the auctions or source foreign flowers, the existingconsensus auction arrangement is being renegotiated.For example, by not participating in the sample auc-tion, the buyers were able to influence the restructur-ing of the sample-based auctions. Second, the HollandSupply Bank brokered market represents another newarrangement for large buyers. Finally, the Tele Flowerauction was the EAF and the buyers’ response to re-strictive foreign flower trading rules in the traditionalDutch flower auctions. The latter was a departure fromrenegotiating the consensus around a specific auctionto creating a competing structure. This structure wascreated because the traditional auctions were forced torestrict the sale of foreign flowers by their owners.

The emergence of Tele Flower is especially instruc-tive. Despite the advice of the traditional Dutch Auc-tion officials, the owners voted to deny access to for-eign growers. The subsequent creation of the TeleFlower auction resulted in the loss of a key revenuestream to the existing auctions and its owners. Con-forming to the existing auction models limited inno-vation by the traditional Dutch auctions. In contrast,information technology allowed the EAF to coordinateand implement trading processes without colocationand at lower costs than setting up a facility similar tothe traditional Dutch auctions. As buyers and foreigngrowers implement and continue to innovate in de-veloping alternate markets, they will, most likely, in-creasingly reshape the access, price determination andtrading rules to their favor. The emergence of TeleFlower could probably have been avoided if the buyershad an ownership stake and greater influence in thetraditional Dutch auctions. This would have led to lessbiased or restrictive trading practices. We expect thatinformation technology will enable new trading pat-terns that are less biased in favor of Dutch growers.These new trading models (e.g., clearinghouse auc-tions, brokered markets) could better serve differentbuyer segments.

In summary, current owners of the auctions may nothave sufficient incentives to adopt new or more effi-cient auction mechanisms. The capacity of existingmarket structures to adapt to new electronic and non-electronic innovations will be determined by the rela-tive power and capabilities of different parties to the



trade, the current ownership structures, and potentialof competitive entry. In the Dutch auctions, we see anincreasing shift of power toward buyers as newsources of supply become available, buyers becomelarger, and technology enables new low-cost tradingmechanisms leading to competing markets.

6. ConclusionsThis paper makes four key contributions to the litera-ture on information technology and markets. First, weidentify a series of distinct processes that underlie ex-change relations. In contrast to the traditional casestudies of electronic markets, this categorization pro-vides a useful basis for examining how technology in-novations affect specific trading processes. Second, wepropose and illustrate the use of the process-stake-holder analysis framework for comparing differentforms of trading, and evaluate the impacts on differentmarket participants. This is a useful guide for design-ers of new market systems as well as those studyingmarkets to evaluate or explain the successes and fail-ures of IT-based initiatives in new markets. Third, thedevelopment of a detailed case study of market changeand a cross-case analysis of IT initiatives in the Dutchflower markets result in a series of testable proposi-tions. The propositions address the separation ofphysical and informational processes in trading, theresponses of stakeholders to changes in information inmarkets due to IT initiatives, and economic and incen-tive conditions required for adoption of new tradingprocesses. Finally, we propose a more complex andsubtle view of markets than typically analyzed in theMIS literature.

While there are a number of teaching and researchcases on electronic markets, these are either unsuitablefor generating new theories (Kambil and Short 1994)or articles analytically apply transactions cost or net-work externality models to emergent markets (Bakos1991b, Lee 1996). While some of the analysis providesconditions for market adoption, they shed little lighton why different actors failed to adopt new marketforms. In contrast, the rich cases in this study illustratethat markets represent a socially constructed consen-sus that is constantly renegotiated among stakeholderswith different levels of power. Information technology,by enabling new low-cost trading mechanisms and

changing information available to stakeholders, is onefactor that can alter the relative costs and benefits toactors, or the power of actors leading to a renegotiationof the consensus. When the renegotiation fails, the newmarket form is not adopted. As new information tech-nologies make feasible new models of electronic com-merce, our analysis suggests that designers of newmarkets should carefully manage the context in whichthey apply their technologies by constructing suitableincentives and processes to arrive at a consensus.

Finally, the paper provides new tools and directionsfor inquiry for managers and researchers of electroniccommerce. First, the process-stakeholder frameworkallows managers and researchers to systematicallycharacterize process changes and the allocation of costsand benefits from electronic commerce. Second, thecases highlights new questions for research. As the In-ternet evolves to a powerful and reliable infrastructurefor electronic commerce, Dutch auctions become moreimportant as a trading mechanism. However, there islittle published empirical research on the Dutch auctionmechanism and the effects of clock speed, and otherinformation variables on equilibrium prices, buyerstrategies, and the distribution of benefits. New devel-opment tools (Java, etc.) will enable researchers to de-velop simulations and test market designs over the In-ternet to develop a richer understanding of individualbehaviors in electronic markets.5

5 We thank Hank Lucas, Ted Stohr, and the anonymous reviewersof Information Systems Research for their comments and suggestionson previous versions of this manuscript.

ReferencesAkerlof, G. A., “The Market for ‘Lemons’: Qualitative Uncertainty

and the Market Mechanism,” Quarterly J. Economics, 84 (1970),488–500.

Arrow, K. J., Limits of Organization, Norton, New York, 1974.Bakos, J. Y. “Information Links and Electronic Marketplaces: The

Role of Interorganizational Information Systems in VerticalMarket,” J. Management Information Systems, 8, 2 (1991a), 31–52.

——, “A Strategic Analysis of Electronic Marketplaces,” MIS Quar-terly, September (1991b), 107–119.

Blau, P., Exchange and Power in Social Life, John Wiley & Sons, NewYork, 1964.

Bos, J., “De Klok met Afstandsbediening,” Groot Handelsblad, No-vember (1995), 22–23.

——, “Een Jaar TFA: De Innovatieve Noodsprong,” Groot Handels-blad, March (1996), 8–11.



Brynjolfsson, E., T. W. Malone, V. Gurbaxani, and A. Kambil, “DoesInformation Technology Lead to Smaller Firms?”, ManagementSci., 40, 12 (1994), 1628–1644.

Clemons, E. K. and B. Weber, “London’s Big Bang: A Case Study ofInformation Technology, Competitive Impact, and Organiza-tional Change,” J. Management Information Systems, 6 (1990), 41–60.

—— and ——, “Evaluating the Prospects for Alternative SecuritiesMarkets,” in Proc. Twelfth International Conf. on Information Sys-tems, New York, 1991.

Coase, R. H., “The Problem of Social Cost,” J. Law and Economics, 3(1960), 1–44.

Coase, R. H., “The Nature of the Firm,” Economica, 4 (1937), 386–405.Daft, R. L. and R. H. Lengel, “Organizational Information Processing

Requirements, Media Richness and Structural Requirements,”Management Sci., 32, 5 (1986), 554–571.

Dahlman, C. J., “The Problem of Externality,” J. Law and Economics,22, 1 (1979), 141–462.

Davis, D. and C. Holt, Experimental Economics, Princeton UniversityPress, Princeton, NJ, 1993.

Eisenhardt, K. M., “Building Theories from Case Study Research,”Acad. Management Rev., 14, 4 (1989), 532–550.

Eras, P., “Tele Flower Auction Veilt via Telefoon en Beeldscherm,”Telecommagazine, 10, 10 (1995), 40–43.

Fukuyama, F., Trust—The Social Virtues and the Creation of Prosperity,The Free Press, New York, 1995.

Galbraith, J. R., “Organization Design: An Information ProcessingView,” 4, 3 (1974), 28–36.

Griffioen, A., “Goede Start Informatieveilen in Aalsmeer,” Vakbladvoor de Bloemisterij, 2 (1994a), 9.

——, “Informatieveilen Moet de Stagnatie Overwinnen,” Vakbladvoor de Bloemisterij, 7, (1994b), 56–57.

Gurbaxani, V. and S. Whang, “The Impact of Information Systemson Organizations and Markets,” Comm. ACM, 34, 1 (1991), 59–73.

Homans, G., Social Behavior: Its Elementary Forms, Harcourt, Braceand World, New York, 1961.

Johnston, R. and P. R. Lawrence, “Beyond Vertical Integration—TheRise of the Value-Adding Partnership,” Harvard Business Rev.,66, 4 (1998), 94–101.

Kambil, A., “Electronic Integration: Designing Information Technol-ogy Mediated Exchange Relations and Networks,” DoctoralDissertation, Sloan School of Management, Massachusetts In-stitute of Technology, Cambridge, MA, 1992.

—— and J. E. Short, “Electronic Integration and Business NetworkRedesign,” J. Management Information Systems, 10, 4 (1994), 59–83.

Klein, B., R. Crawford, and A. Alchian, “Vertical Integration, Ap-propriable Rents and the Competitive Contracting Process,” J.Law and Economics, 21 (1978), 297–326.

Konsynski, B., “Thinking of Linking: Managerial Perspectives onElectronic Linkages Across Organizations,” in Proceedings of Sci-

entific Research on EDI, R. J. Streng, C. F. Ekering, E. van Heck,and J. F. Schultz (Eds.), Samson Bedrijfs Informatie, Alphen aanden Rijn, Netherlands, 1992.

—— and A. Warbelow, “American Gem Market System,” CaseStudy, 9-190-001, Harvard Business School, Cambridge, MA,1989.

—— and ——, “Cooperating to Compete: Modelling Interorgani-zational Interchange,” Harvard Business School, Cambridge,MA, 1990.

——, ——, and J. Kokuryo, “AucNet”, Case Study, 9-190-001, Har-vard Business School, Cambridge, MA, 1989.

Lee, H. G., “Intelligent Electronic Markets for Commodity Auction:An Integrated Approach of Economic Theory and Social ChoiceTheory,” Ph.D. Dissertation, University of Texas, Austin, TX,1993.

——, “Electronic Brokerage and Electronic Auctions: The Impact ofIT on Market Structures,” in Proc. Hawaii International Conf. onSystem Sciences, Honolulu, Hawaii, 1996.

Malone, T. W., J. Yates, and R. I. Benjamin, “Electronic Markets andElectronic Hierarchies,” Comm. ACM, 30 June (1987), 484–497.

Milgrom, P. and D. J. Roberts, “Bargaining Costs, Influence Costs,and the Organization of Economic Activity,” in J. Alte and K.Shepsle (Eds.), Perspectives on Positive Political Economy, Cam-bridge University Press, Cambridge, England, 1990.

Powell, W. W., “Neither Market nor Hierarchy: Network Forms ofOrganization,” Organizational Behavior, 12 (1990), 295–336.

Spooner, I. and D. Copeland, “Flower Auction Westland: The Cos-mos Project,” Case Study, University of Western Ontario, Can-ada, 1992.

Tushman, M. L. and D. A. Nadler, “Information Processing as anIntegrating Concept in Organization Design,” Acad. Manage-ment Rev., July (1978), 613–624.

Van Heck, E. and T. P. Groen, “Towards an Electronic Flower Auc-tion?,” in Proc. Management Studies and the Agri-business 1994:Management of Agri-chains, Wageningen Agricultural Univer-sity, Wageningen, 1994.

—— and P. M. A. Ribbers, “Economic Effects of Electronic Markets,”Discussion Paper, No. 9669, Center for Economic Research, Til-burg University, The Netherlands, July, 1996.

——, E. Van Damme, J. Kleijnen, and P. Ribbers, “New Entrants andthe Role of Information Technology, The Case of the TeleFlower Auction in the Netherlands,” in Proc. Hawaii Interna-tional Conf. on System Sciences, Vol. 3, IEEE Computer SocietyPress, Los Alamitos, CA, 1997.

Van Vliet, C., “EAF wil Met Importveiling Centrum Bloemenhandelin Nederland Houden,” Vakblad voor de Bloemisterij, 49, (1994),40–41.

Williamson, O. E., Markets and Hierarchies: Analysis and Antitrust Im-plications, The Free Press, New York, 1975.

——, The Economic Institutions of Capitalism: Firms, Markets and Rela-tional Contracting, Free Press, New York, 1985.

Yin, R. K., “The Case Study Crisis: Some Answers,” Admin. Sci. Quar-terly, 26 (1981), 58–61.

Jiro Kokuryo, Associate Editor. This paper was received on April 11, 1995 and has been with the authors 10 months for 1 revision.

Re-engineering the Dutch Flower Auctions

Business

Transcript of Re-engineering the Dutch Flower Auctions