B2B Online auctions a conceptual framework for a sensitivity by zhouwenjuan


									            B2B Online auctions: a conceptual framework for a sensitivity analysis
                                                 Davide Aloini
 Department of Electrical Systems and Automation, University of Pisa, Via Diotisalvi 2 56126 Pisa – Italy, e-mail:
                   davide.aloini@dsea.unipi.it, tel. +39.0502217384, fax. +39.0502217333.


This work is part of a more extended research proposal which aims to give a contribute in e-
auction optimization through a neural network-based sensitivity analysis on performance
determinants in B2B online auctions. A complete sensitivity analysis on auction process is a
challenging purpose for research, both because of large number of attributes and correlations that
may affect final performances and uncertainty of human behavior. In such complex decision
context, Artificial Intelligence (AI) techniques as neural networks present great operative
advantages. This paper in particular focuses on design of a conceptual framework for parameters
identifications to guide neural network (NN) design and empirical analysis.
Keywords: B2B e-auction, sensitivity analysis, conceptual framework.


   Auction process optimization has experienced a great development in recent years, becoming
a common objective of both business and academic research, in their aim to maximize profit
(price reduction or sale margin growth according to the buyer/seller point of view) and minimize
transaction costs of negotiation process. In this sense, past performance information, as also
market-related ones, has a great value for on line auctions design (De Liu & Chen, 2006).
Moreover, fine grained data provided by online transaction of internet auction applications enable
an high variety of new potential analysis and services both by seller than buyer’s side (Ghani,
2005); so ex-ante use of historical information in auction design could be an interesting
research’s field for its optimization.
   Making an extreme brief of auction research literature, we found that in spite of the great
interest on auction theory, due to poor data availability and expensive data collecting process, as
also Lucking-Reiley (1999) highlights, till today only few empirical studies has been developed,
especially in B2B context; however, a drastic growth of contributes number has been since real
online data were available on internet.
   Hence, studies about online auction’s ground and in particular on performance determinant
factors and final price prediction are relatively young. A principal research branch focus on
bidder/seller behavior dynamics (Bajari and Hortacsu, 2003; Bapna et al.,2004); a second field of
research concerns impact of auction features on process performances (Carter et al., 2004;
Kaufmann & Carter, 2004; Millet et al., 2004); several case studies (Handfield, 2002; Emiliani,
2004) have examined factors related to e-auction success or failure. A specific field of research
concerns impact of auction features on final price but only few contributes on auction price
prediction were deployed in B2C and C2C markets by AI community (Wellman, 2002; Wellman,
   In this sense, Emalghraby (2006) suggests that the use of data mining or other techniques to
learn from previous auction events the suitability of various auction formats under different
market settings, could be an interesting direction for the future, especially in B2B auction. Until
to date, in fact, we are unaware of similar empirical analysis or any solution providers that are
developing auction tools which such capabilities.
Expected contribution

   This research aims to give a contribute in e-auction optimization through a neural network-
based sensitivity analysis on performance determinants in B2B online auctions. We think that a
better knowledge of internal and external variables impact on auction process and final price
performance could induce great benefit (not only for profit and saving maximization) both for
buyers, sellers and third parts; and this benefit may be larger and larger for business players (Li
Xuefeng et al., 2006). B2B auction processes, in fact, are quite different from B2C and C2C ones,
for example, in terms of information exchange, suppliers relationship, market and item
knowledge; so we think it needs to be deeply analyzed in order to identify gaps and process
Now, it is clear that value of past information in e-auction design, especially in B2B field, could
be high; not only for benefits which directly impact on final auction performances but also for
new business potentialities and profit opportunities linked to a better reliability in final price
forecasting which will not constitute a central proposal of our research. So main expected
contribution, in our opinion, could be grouped in the following areas:
1. Optimization of auction parameters. A “What if” analysis, enabled by the NN-prediction
    model, could be an useful device to assist auction makers in this phase of negotiation process.
2. Market decision support. Past auction-events observation by an extended sensitivity analysis,
    could provide relevant information about performances and link them to several dimension
    like market trend, suppliers relationship and item class. Online auctions often result in
    switching to new sources of supply (Emiliani & Stec, 2002); so historical auction data can be
    also used to enable price prediction in order to define an acceptance threshold or as a means
    to accurately define a scheme for purchasing process and suppliers evaluation, for properly
    evaluating auction success.

Research design

   E-auctions are complex processes, since no a unique parameter influences final performances.
Literature in this field is populated of contributes which analyze different features, in terms of
auction class and parameters, market condition, characteristic of purchasing; moreover, variables
are often described and organized differently (Walley and Fortin, 2005).
   So the final objective of this study is first to design and develop the NN-based model to
analyze auction outcomes like final price gain as complex function of several, quali-quantitative,
no-linearly correlated input variables and then run the sensitivity analysis to investigate about
process trend and factors impact. Naturally, this is a challenging purpose but neural models
present great operative advantages in such complex and uncertain contexts. Programmed steps of
work are:
1. Design of a conceptual framework for parameters identification. An extended literature
    analysis finalized to the identification of relevant parameters which could have impact on
    final price definition, was conducted. Process inputs and outputs were identified and relevant
    parameters described and classified. Finally the framework was defined.
2. Data collect and test of factors significance on data set. For neural network training a large
    set of field data is necessary. An accurate data-analysis may be needed to filter available data
    in the framework scheme, identify principal dimensions and design pre-processing of model
3. Design, develop and test of NN model. Using a commercial software some prototypical neural
    networks are designed and developed. Then, after preprocessing activities, data collected are
    used to network training and testing.
4. Run the sensitivity analysis. Input are accurately modulated to perform the analysis and test
    more relevant research hypothesis about parameters variation; finally conclusion will be
This paper concern the first step of the above general research program and focuses on design of
conceptual framework for parameters identification to guide NN design and empirical analysis;
starting from literature evidences a general map of main input, output and process variables is
drawn. Variables are then explained in relation to auction context and process impacts.

Conceptual framework

Introduction to framework

First objective of our research project aims to develop a conceptual framework for auction
parameters identification. To have a clear interpretative scheme of a so multifaceted process,
helping in input-output selection, is an essential starting point for an effective and efficient
sensitivity analysis. B2B auction process are complex items which involve different players and
process elements; consequentially, final outcomes could be deeply influenced by numerous
variables. From a first analysis, main auction players which can directly or indirectly influence
trading, are: Buyer/s, Sellers and eventually Third Part Outsourcers; principal elements in auction
process are: Trading goods, Auction dynamic(process), Transactional phase(costs).
In this light, we have re-analyzed the general framework presented by Parente et al. (2004) and
accomplished an extended literature analysis in order to identify and systematize potentially
interesting success factors of online auctions. First of all, “auction success” definition problem
was faced and a set of appropriate outcome performances and related indicators were
hypothesized. Then variables were identified according to six contextual dimensions and finally
linked to four performance drivers (respectively “Input” and “Drivers” in figure 1), how we will
detail in next sections.
                                         <Here Figure 1>
In next sub-sessions, before explaining framework general architecture and in order to better
explicate the model, we will start introducing the interpretative logic and describing four main
drivers which filter impact of the several variables on auction process. After this step, we will
categorize and classify output, process and input variables, clarifying their potential influence on
process and performances and suggesting possible indicators for their measurement.

Performance drivers

Despite of the intrinsic complexity of process, according to our interpretation, influence on final
auction performances are driven by a limited set of attributes associated to the relative value of
exchanged item, process costs and dynamic efficiency of auction mechanism which players
perceive (See figure 2).
                                        <Here Figure 2>
In particular, we identified four principal performance drivers that could filter determinants
impact on auction process. First two drivers are related to buyer and seller perception of product
and market; Pinker et al. (2003) demonstrated that the relative value of auctioned goods sensibly
weights on financial performances. Drivers are:
1. Seller perception of item value. Sellers value perception of auctioned items can deeply weight
   on motivation, bids and suddenly on final price definition; diverse perceptions are due to
   different cost structures, market conditions, available production capacity or selling
2. Buyer perception of item value. Buyer value perception of auctioned goods directly impact on
   reserve price setting and definition of acceptance threshold for process costs; different
   perceptions can be due to several context attributes such as market competition, strategic
   degree of auctioned item, expected revenues and required quality of products.

Third and forth driver, instead, belong to a “process” dimension:
3. Perceived process costs. Despite researchers agree on the lower transaction costs associated
   with auction-negotiation, for example look at searching, monitoring, holding and others
   opportunity costs linked to time spent participating the auction (De Vany, 1987), switching
   costs due to alternate suppliers and additional costs linked to new incumbent inexperience can
   be enormous. So if these costs exceed potential procurement benefits, it could induce a lack of
   interest on auction process and negatively impact on competition and final results.
4. Auction dynamic efficiency. Online auctions are automated processes which respond to a
   specific, customized, set of rules as for example type, timing mechanism, reserve price,
   minimum bid, extension time, lot assembling and sequencing; numerous rules were developed
   in order to govern trading process (especially bidding and trading) and the way which
   parameters are set could influence auction dynamic, length, process costs and finally financial

Framework architecture and elements

The general framework architecture was inherited by Parente’s work, in particular the basic
system theory logic and input-process-output levels organization were maintained to permit an
easier identification of boundaries/interfaces and interpretation of relation between
interdependent variables (See figure 3).
                                           <Here Figure 3>
We have clarified that auction negotiation is only a single part of a more articulated sourcing
process and even if we can look at negotiation phase as an independent step, this is not really.
Experienced bidders, in fact, take into account several factors in their bidding strategies, so a
large set of variable could be linked with the final auction performances and in particular with the
final price gain.
As Parente et al. (2004) suggest, our interpretative model consist of 3 main groups:
    1. Input: Input component contains all principal subjects/objects involved into the auction
        process: buyer, suppliers and eventually third part which could interact in auction, trading
        goods and general market context in which negotiation develops.
    2. Process: Auction process refers to negotiation phase and trading rules which auctioneer
        set for the event. It involves auction attributes which may impact on outcomes.
    3. Output: Output components refer to the outcomes of the auction such as price efficiency
        or other general auction performance indicators by which we try to provide an effective
        success evaluation.

   We will start describing auction outputs in terms of process performances and results. For the
sensitivity analysis, in fact, is really important to define clearly what are principal directions on
which it has to investigate on; so to explain what auction success means and suggest measurable
metrics is a priority for framework organization.
In our opinion auction success evaluation, especially in B2B procurement context, has to embrace
a broad TCO (Total cost of Ownership) philosophy and consider all costs associated with
acquisition, possession, use and disposition of purchased good. So beyond more traditional
performance indicators such as perceptual price gain, other factors should be observed such as
internal process costs (transaction and switching costs) and participation intensity. Here we
describe selected dimensions and propone a set of quantitative indicators for variables measuring.
   First dimension is “price efficiency” which directly analyze savings on final price obtained for
goods auctioned. Purchase price reduction, in fact, is one of the primary benefits of e-auctions,
even if it is also important to consider process cost saving in assessing e-auctions profitability. So
a possible indicator of price efficiency is the COA (Close-of-auction) index which is defined as
the difference between the historical price paid and the lowest bid price in a given bid event
(Emiliani et al. 2002).
   Emiliani et al. (2002), furthermore, argued that some savings would be lost due to various
factors associated with the award decision, data integrity and additional business expenses; in this
sense attainability of the COA price saving has been often questioned. Since this reason, we
selected “process efficiency” as second outcome dimension; it refers to internal process cost
savings (administrative, transaction and switching costs) and increase of productivity due to e-
auction adoption. Reverse auctions, in fact, force buyer to structure the bid prior to the event,
standardize the procurement process and develop a strategy for group of items; for repeated
auctions it will traduce in shorter cycle time and increase of productivity (Carter et al., 2004).
   However, as we observed above, suppliers qualification, certification, pilot production, quality
testing, tooling, personnel training and orientation, regulation compliance, disposing obsolete
inventories, supplier monitoring, and other similar costs (D. Hur et al. 2007) are difficult to
quantify or estimate and this make process performance dimension more hard to explore.
Quantitative indicators we could use to estimate purchasing process efficiency and productivity
are: 1) Total negotiation cycle time reduction (Pre-auction cycle time + Auction cycle time); 2)
Unit cost of a purchase activity reduction; 3) Reduction of man-hours spent in purchasing
operational activities.
   Finally, third dimension for performance evaluation is intensity in “participation” to the
auction event. This dimension try to fit competition intensity and auction interest of players.
   Auction failure, lack of motivation and sellers collusive behavior could induce great and un-
useful resources waste to firms; a good auction design, in fact, is not “one size fits all” but it
should be sensitive to the details of the context. Pre-auction phase accuracy, correct definition of
RFx and capitulates, supplier training, business context analysis can deeply impact on
participation degree.
   Perceptual number of successful auctions concluded, bids received, bid density, number of
activated auto-extension (if exist) can estimate this performance. Moreover, this dimension has
indirect reflections also on financial performance; Massad and Tucker (2000) observed that the
increased presence of bidders in a particular online auction will result in higher sale prices being
obtained, due to increased bidder competition.
Input and process determinants

    Determinants affecting final auction performances (outputs) can be associated both with input
dimension in which we classified exogenous variables –i.e. context-related variables impacting
on auction process– and with process settings which group endogen and controllable process
variables. In Figure 4 we resume principal links between input classes, performance drivers,
auction process and outcomes we have selected for the analysis.
                                            <Here Figure 4>
    In inputs, we identified two main classes of variables: first class is “Market conditions” which
refer to information about trading players, their relationships and market sector; the second
concerns “Product characteristics” which involve general information about exchanged product.
    In “market conditions” group we assess all information linked to buyer, sellers, third part and
their relationships with general market context data (Buying and selling firm characteristics;
Buyer/Supplier relationship; Market context characteristics).
    Traditional buying and selling firm characteristics continue to strongly influence
procurement exchanges. These factors are frequently included in pre-auction phase for supplier
screening and often, also influence winner determination algorithm and post negotiation phases,
especially in multi-attribute competitions. In this group we focus principally on:
    - Players reputation. As Walley and Fortin (2005) affirm reputation is a subjective concept
that resides in the mind of the buyer; so the inability to observe and quantify reputation makes it
difficult to estimate its effect on price, anywhere it exists. This variable include several attributes
like quality and delivery reliability, economic performance, financial stability and others criteria
traditional used for suppliers selection (Ellram, 1990; Min, 1994; Choi, 1996), which contribute
to define a market supplier profile.
    - Company dimension could play an important role in sourcing process because of critical
buying mass and other benefits as volume discount, process cost saving, contractual power which
it could determine; empirical researches showed that lot size and value have a significant weight
on auction players motivation and final price reduction. Quantitative indicators in this field, could
be: total number of firm’s employees and total turnover.
    - Company IT/E-commerce culture. Information technology (IT) confidence and e-commerce
attitude could have a relevant influence on adoption and learning rate of new tools and practices,
and deeply influence participation level in e-auctions. Furthermore, high financial gain are often
related to primary auction events; Carter et al. (2004) affirm that there is no initial relationship
between the level of buyer experience and auction success, but suppliers tend to be unsuccessful
in the initial events in which they participate. Indicators we hypothesized to estimate IT ad e-
commerce attitude are for example: age of e-commerce tool adoption (E-commerce attitude) and
number of reiteration of the specific auction event.
    Buyer/Supplier relationship. As Parente et al. (2004) suggest “among the other
characteristics that are important for auction success, interaction between buyer and seller is the
most critical”. Factors influencing companies relationships are numerous, many were identified
in marketing research (Jaworski and Kohli, 1993; Parente, 1998). We can model relation
characteristic as “distant factor” which draw dimensional, financial and cultural difference
between companies. Referring to variables previous explained in buying and selling firm
characteristics we will analyze two of most significant gaps between firms: Dimensional gap
and IT/E-commerce gap. Principal indicators we could use to estimate differences in dimension
and in IT/E-commerce confidence are ratio index related to firm employees, turnover and
longevity of tool adoption. Other central variables are for example:
- Age of relationship. Long relationship between buyer and seller could prejudice auction
applicability and motivation to competition. Carter et al. (2004) say that reverse auction are
viewed more favorably by buyers than by suppliers; this condition, in same cases, may cause
decrease in the level of trust, collusive behavior and less motivation to competition. The indicator
we hypothesize is the duration of relation.
- Relative contractual power. Contractual power equilibrium of a trading transaction depends on
credibility and effectiveness by which a player may represent a treat for the other one.
Differences in contractual power certainly contribute to influence bidding behavior, motivation to
competition and final auction performances. Key factors for contractual power analysis are
suggested by Grant (1991).
   Finally, in market context characteristic we analyze information related to market sector.
Obviously market information are relevant in good price definition; bidding behavior as also
buyer reserve price depends on buyers and sellers perception and confidence of market
   - Market concentration gives important information on potential motivation to auction
competition. In particular matching supply and demand concentration, it could be possible to
understand better market attitude to auction sourcing and draw supplier motivation to compete in
e-auction. We think that in a common buyer-driven auction event, concentration in supply market
could differently impact on motivation to competition in relation to the concentration on demand
side. Matching Supply and Demand dimensions, we can identify four scenario which could
differently and critically impact on competition. In this sense, also regularity degree of demand
may meaningfully empathize this phenomena. Principal indicators we can use to evaluate market
concentration are: 1) The “C4” index - combined share of the top four firms in a market - is a
measure of the size of firms in relationship to the industry and an indicator of the amount of
competition among them; 2) Number of firms which have 80% of market share; 3) Market share
of top firm.
- Market price volatility is also an interesting variables in market context dimension. An high
price volatility in selected market could drive high irregularity in price performance; this could
induce suppliers to more aggressive/protective behavior impacting on motivation and auction
competition. In literature different techniques for Volatility measure exist (VIX index).
   In “product characteristics” we classify product attributes which could modify auction costs,
dynamic and players motivation. Klein (1997) suggested that specific auction format may be
better for certain types of products; we link this condition to the following set of variables:
   - Product class, buying situation (New Buy, Rebuy, Straight Rebuy) (Faris et al., 1967) and
Strategic Importance (Kraljic, 1983) are correlated with relative value perceived by buyer and
sellers, managerial complexity and business risk perception. This can impact on their competition
motivation, acceptance thresholds and finally on auction results.
   - Product standardization or customization degree could impact on negotiation complexity
and costs;
   - Description complexity impact on negotiation costs and supplier comprehension of
capitulate; so effectiveness, efficiency, dynamic of auction, motivation and final results could be
   - Repeatability of auction event could impact one side on amortization of costs for complex
capitulates during exchange of high specific products and suppliers motivation to e-auction tool
adoption, on the other side it could induce supplier adaptation and opportunistic/collusive
   Process group refers directly to the auction event and all internal process variables that may
impact on final performances, in particular auction mechanism and its characteristics. In literature
a variety of auction formats is available and each format allows personalization through
numerous parameters settings, so thousand different combinations are effectively configurable.
   Many authors have studied how parameters impact on final auction price, adopting different
analysis approach especially in B2C and C2C context (Bajari et al., 2003; Pinker et al., 2003;
Lucking-Reiley et al., 2006). Carter et al. (2004), for example suggest that auction that utilize
multiple lots are more likely to be successful and that the size of auction in terms of monetary
volume is in general significantly greater for “successful” auctions.
   Online auction designers should pay attention to how their project decision influence
participative behavior of consumers over time. Pinker et al (2003) divided the global set of rules
for online auctions in two sub-set referred one to mechanism design second to bid constraints.
   Here, we present a list of principal parameters we will consider in our analysis: Auction type,
Winner algorithm, Minimum bid, Reserve price level, Information disclosure (reserve price, bids,
etc), Total lot value, Lot size, Auction fees, Timing mechanism (Auto-extension or rigid deadline),
Opening bid, Final auction length, Total number of bidder.

Concluding remarks

Present research is addressed to drive the design, development and testing of a neural network for
a performance sensitivity analysis in B2B e-auction process. This paper in particular investigates,
from a conceptual point of view, on factors influencing auction process and their contribution to
the final performances. Nonetheless, this framework represent a first -general and not exhaustive-
tentative to identify and classify, from literature evidences, main determinants to auction
“success”, drawing them into a scheme functionally to next research’s steps. Then neural network
will be designed, input selected; finally, model will be trained, tested and considerations on
determinants will be achieved.
We think that an interaction effect between market, purchasing item, auction parameters and final
auction performance really exist and this may significantly impact on auction design; so this
study could be a first step in order to provide, from a B2B empirical prospective, a better
knowledge of auction process. Principal benefits are linked to a more effective and efficient
market decision support, buying or selling performance improving and process transaction costs
and time reduction. A possible direction in which our research may be improved and developed is
the realization of a Decision Support System to assist B2B (e-auction based) redesign of
negotiation process.


Bajari, P., & Hortacsu, A., 2003. Winner’s curse, reserve prices, and endogenous entry: Empirical insights
    from Ebay auctions. The Rand Journal of Economics, 3(2), 329–355.
Bapna, R., Goes, P., Gupta, A., & Jin, Y., 2004. User heterogeneity and its impact on electronic auction
    market design: An empirical explanation. MIS Quarterly, 28(1).
Carter, C.R., Kaufmann, L., Beall, S., Carter, P.L., Hendrick, T.E., & Peterson, K.J., 2004. Reverse
    auctions-grounded theory from the buyer and supplier perspective. Transportation Research Part E,
    40, 229–254
Choi, T.Y., 1996. An exploration of supplier selection practices across the supply chain. Journal of
    Operations Management, 14(4), 333-343.
De Vany, A., 1987. Institutions for Stochastic Markets. Journal of Institutional and Theoretical
    Economics. 143, 91-103.
Ellram, L.M., 1990. The supplier selection decision in strategic partnerships. Journal of Purchasing &
    Materials Management, 26(4), 8-14.
Elmaghraby, W., 2006. Auctions within e-Sourcing Events, to appear in Production and Operations
Emiliani, M.L., & Stec, D.J., 2002. Realizing savings from reverse online auction. Supply Chain
    Management, 7(1), 12–23.
Emiliani, M.L., 2004. Sourcing in the global aerospace supply chain using online reverse auctions.
    Industrial Marketing Management, 33(1), 65–73.
Faris, C.W., Robinson, P.J., Wind, Y, 1967. Industrial Buying and Creative Marketing. Allyn & Bacon,
Ghani, R., 2005. Price Prediction and Insurance for Online Auctions. Accenture Technology Labs of
Grant, R.M., Balckwell (Ed.) 1991. Contemporary Strategy Analysis: Concepts; Techniques,
    Applications. Oxford,.
Handfield, R.B., Straight, S.L., & Stirling, W.A., 2002. Reverse auctions: how do supply managers really
    feel about them. Inside Supply Management, 13(11), 56–61.
D. Hur, V. A. Mabert, Janet L. Hartley, 2007. Getting the most out of reverse e-auction investment.
    Omega, 35, 403-416.
Jaworski, B.J. and Kohli, A.K., 1993. Market orientation: antecedents and consequences. Journal of
    Marketing, 57, 53-70.
Kaufmann, L., & Carter, C.R., 2004. Deciding on the mode of negotiation: to auction or not to auction
    electronically. The Journal of Supply Chain Management, 40(2), 15–26.
Klein, S., 1997. Introduction to electronic auctions. Electronic Markets, 7, 3-6.
Kraljic, P., 1983. Purchasing must become supply management. Harvard Business Review, 61, 109-117
De Liu & Chen, J., 2006. Designing online auctions with past performance information. Decision Support
    Systems, 42(3), 1307-1320.
Lucking-Reiley, D., Doug, B., Prasad, N., & Reeves, D., 2000. Pennies from eBay: the Determinants of
    Price in Online Auction. Technical report, University of Arizona.
Lucking-Reiley D., 1999. Using field experiments to test equivalence between auction formats: magic on
    the Internet. American Economic Review, December, 89(5), 1063–1080.
Li Xuefeng , Liu Lu and Wu Lihua, Zhang Zhao, 2006. Predicting the final prices of online auction items,
    Expert Systems with Applications, 31, 542–550.
Massad VJ, Tucker JM., 2000. Comparing bidding and pricing between in-person and online auctions. J
    Prod Brand Manage, 9(5), 325-340.
Millet, I., Parente, D., Fizel, J., Venkataraman, R., 2004. Metrics for managing online procurement
    auctions. Interfaces, 34(3), 171–179.
Min, H., 1994. International supplier selection: a multi-attribute utility approach. International Journal of
    Physical Distribution & Logistics Management, 24(5), 24-33.
Parente, D.H., 1998. Across the manufacturing-marketing interface: classification of significant research.
    International Journal of Operations & Production Management, 18(12), 1205-22.
Parente, D.H., Venkataraman, R., Fizel, J. and Millet, I., 2004. A conceptual research framework for
    analyzing online auctions in a B2B environment, Supply Chain Management: An International
    Journal, 9(4), 287-294
Pinker, J., Seidmann, A. and Vakrat, Y., 2003. Managing online auctions: Current Business and Research
    Issues, Management Science, 49(11), 1457-1484.
Walley, M.J.C., & Fortin, D.R., 2005. Behavioral outcomes from online auctions: reserve price, reserve
    disclosure, and initial bidding influences in the decision process. Journal of Business Research, 58,
Wellman, M. P., Reeves, D. M., Lochner, K. M., & Vorobeychik, Y., 2004. Price prediction in a trading
   agent competition. Journal of Artificial Intelligence Research, 21, 19–36.
Wellman, M. P., Reeves, D. M., Lochner, K. M., & Vorobeychik, Y., 2002. Price prediction in a trading
   agent competition. Technical report, University of Michigan.

                             Figure 1: General framework architecture

                                  Figure 2: Performance drivers

                                                   Figure 4: Framework dimensions

          Figure 3: Framework



  Input                         Output

To top