Online Auction Site: Proxy Bid Service with Time Awareness Engine Supat Petchbordee Software and Web Developer (System Integration) No Magic Asia Ltd, Bangkok Thailand Email: [email protected]

Thotsapon Sortrakul, Ph.D. Assumption University, Bangkok Thailand Email: [email protected]

Abstract

1

Introduction

Many Internet auction sites implement English auction. A great deal of late bidding has been observed on Internet auctions such as eBay that employ a second price auction with a fixed deadline. Based on field evidence, late bids are caused both by sophisticated strategic reasoning, and by irrationality and inexperience. Some bidders might use software agents as tools from third party to facilitate their monitoring of late bidding. Some might use the agent service provided by the auction site itself which both have their own advantages and disadvantages. This paper introduces a software agent that employs the proxy bidding as the automated bidding agent combined with the time tracking feature to facilitate the behavior of late bidders. This system will complement the concept of trading that goes between two parties, to support sellers who want to sell their goods at a highest price possible and to support buyers who want to buy goods at an affordable price. The demonstration results in higher success rate in that buyers with the most purchasing power would win the auction and the seller would attain the best price point he could possibly earn.

In this paper, the authors have gathered the information of the evolution of ascending bid Internet auction. The ascending bid auction is the auction for items that is posted by the seller. The procedure is bidders or buyers will offer their willingness to pay for those items. They compete with each other by submitting the prices higher than previous offers to take the lead for the price. So, the price is ascending till no one take over the highest price and the period of that auction is over, meaning that the auction is ended. The winner is the one who offers that highest price. Usually, auction service provider will allow seller to make the reserve price for their posted items. If the prices offered by bidders do not meet the reserve price, the auction will end without a winner. The ascending bid auction can be classified based on the ending rules. One is “hard close”, the period of auction for bidding will be set by the seller in a fixed period of time. When that time is reached, the auction is ended immediately. Another way is, when the ending time of that auction is approaching, if some bids come in, the ending time will be automatically extended for some short period, say 5 minutes. This is called “soft close”, in the last minute, if some bids are entered, the time will be

Proceedings of the Fourth International Conference on eBusiness, November 19-20, 2005, Bangkok, Thailand

9.1

•

extended for 5 minutes more and continue to be extended until no one submits the bid during the last minute of that auction period. The most popular internet auction site such as the well-known site named eBay uses hard close ascending bid auction rule.

•

MARKET SHARE: EBAY CONTROLS THREE-QUATERS OF ALL C2C LISTED AUCTIONS (SEPTEMBER 1999)

• FairMarket, 2% Amazon, 3% Yahoo, 22%

2

The proxy bid that bidders have to submit for their maximum bid budget and allow the site to bid on their behalf. The additional option to allow the proxy bidders specifies the time for their proxy bid to begin based on their experienced perspectives or their strategies. The module that will keep track on timing of the system to alert the proxy bids at the time set by proxy bidders.

Current Online Bidding

eBay, 73%

2.1 Incremental Bidding Figure 1 Market share of eBay The market share of the internet auction site of eBay is more than 70 percent out of total auction sites listed for c2c market. Many bidders have been observed to study their bidding behaviors that can form some common and popular patterns as the strategies. In the major part of this paper, the authors have introduced the integration of some bidding service and bidder behavior to create a new auction service that serve bidders, even if they are new to internet auction, to gain as many advantages as the experienced bidders. The goal of the Proxy Bid Service with Time Awareness Engine (PBSTAE) is to recapture the sniping market by offering the sniping option on the internet auction site that provides the integration of the features for bidders to gain more control on their attended auction period to facilitate bidders. The specific objectives of the system are to provide services such as: • The live auction for bidders who can spend time to monitor their bids and are likely to make a bid on their own manually.

The purpose of bidding in an internet auction is to gain the leading price submitted to seller. This concept can come up with the basic strategy called Incremental bidding or Lowball bidding. Bidders with this strategy will keep track on bids almost all the time during auction period to keep the leading price status. When the bidder observes that there is someone overbidding his or her price, the incremental bidder will resubmit the bid and raises the current bid just one increment more to take over that leading price. Incremental bidder will know in advance how much they will pay for the item when the auction ends and they are the winners because they themselves submit all the bids. And if the opening price or the reserve price is close to the market value of that item, the incremental bidder could bid that early price. If the next bid increment would require exceeding the market value, others may be discouraged from bidding. Therefore, the incremental bidder could win the auction with the opening bid. On the other hand, any other bid can beat the incremental bidder easily because it is just a small amount bidding. The only way for incremental bidder to gain the lead is to

Special Issue of the International Journal of the Computer, the Internet and Management, Vol. 13 No.SP3, November, 2005

9.2

bid again. There is some bidding history in eBay auction which shows that the incremental bidder bids 42 times trying to maintain the lead and finally they lost that auction. Inevitable in the bidding war, if another incremental bidder shows up, both bidders may end up “fighting” over the item in a very long and tense increment struggle. It will take a lot of unnecessary energy to maintain the lead. And it is possible that neither of this dueling pair will win the auction. 2.2 Proxy Bidding The most popular bidding service for almost every auction site nowadays is the automatic bidding feature introduced by eBay called Proxy bidding. Bidders just enter their maximum budget that they expect to pay, then sit back and watch. eBay uses this helpful bidding system to make bidding on auctions more convenient and less timeconsuming for buyers. The procedure of this automatic bidding is when bidders place bids, they enter the maximum amount they would be willing to pay for the item. This maximum amount will be kept confidentially from other bidders and the sellers. The system will compare a bid to those of other bidders. Then the system places bids on your behalf, using only as much of your bid as necessary to maintain the high bid position. The system will bid up to the maximum amount submitted by bidder. If other bidders have higher maximum budgets submitted, you will be outbid. But if no other bidders have higher maximum budgets submitted, you will win that item. The payment is just one increment above the previous bid or second highest maximum value. If the bidder knows their exact maximum budget for the item, they have to bid just only once. Proxy bidder does not have to present at the end of auction or any time in the auction period as long as the

maximum bid is not exceeded. This is extremely effective for those with busy schedules and do not want to give their password to an automatic bidding service from the third party software. The only one way to lose the auction is if there is someone who is willing to pay for the item more than the proxy bidder. If that is the case, no harm is done. If the proxy bidder’s maximum bid is later tied until the close of the auction period, the first bid will break the tie and win as the saying ‘the early bird gets the worm’ implies. But if the proxy is placed early, the bidder has a greater chance of incremental bidder to outbid. Moreover, one who submits a proxy bid, they cannot feel free to bid on other auctions for the same item until the auction has ended because there is a chance that if you go to bid in another site for the same item that may have lower price bidding, you may end up with winning both items. 2.3 Snipe Bidding (Last-minute Bidding) This is not the service provided by any auction site but there is enough statistics to confirm that if bidders use this strategy, they will have more chance to win the auction even at the lower price than it is supposed to. The feature that makes this kind of strategy so distinguished from the other two biddings is that this bid is placed so late in the auction period. In this case, no one will be able to manually respond to it before the auction ends. The most common reaction for someone who has been sniped is to wish for an auction extension like the soft close auction. No one will know that you are coming, so no one will have time to manually respond to your bid, if you take the lead. But if the bid misses the proxy maximum, there will be no second chance for the sniper to bid again as well.

Proceedings of the Fourth International Conference on eBusiness, November 19-20, 2005, Bangkok, Thailand

9.3

3

Investigation for Consideration

3.1 Introduction to Late Bidding These are the existing outstanding bidding techniques that most of bidders use not only in eBay but in almost every other auction site. Each technique has its own advantages and disadvantages. But there are some outstanding features that can be extracted and combined with another feature to come up with a more flexible auction service. After the information from the bidding histories of completed auctions on eBay were gathered, two common facts that were found in the bidding strategies become evident. First, the bids submitted from both the human and the artificial agent are placed in the closing seconds of an auction, a practice called “sniping”. Second, many bidders bid multiple times, which means that they increase a maximum bid submitted or proxy in the bid that they have set earlier in the auction.

the web based service esnipe.com also has the evidence on the late bidding for the auction scheduled to end in precisely ten days. But the very first bid in this ten day auction was entered on the last day at the last sixteen minutes before the end. Table 1 Frequency of Late Bidding in eBay and Amazon

More evident is the summary of the distribution of bids over time. The left columns show the last bids submitted by each bidder in an auction, while the right columns record the very last bid submitted in each auction. The table shows that more than two-thirds of the eBay auction are still active an hour before the end time, as more bids are coming in. And even in the last ten minutes, more than half of the eBay auctions still received bids. So, there should be very good reasons to make that bidders choose to bid late. The information of the research will clarify why it is that bidders would bid late. 3.2 Bidding Late to Avoid Bidding Wars with Incremental Bidders

Figure 2 Timing of Bid

The probability distribution of the timing of bidders’ last bids in a sample of 308 eBay auctions of computers and antique with a total of 1339 bidders. It illustrates that, although an eBay auction usually last one week, many bidders bid in the very last minutes of the auctions. One of the most interesting of these bidding agents which is

Last-minute bidding can be the best response to the incremental bidding because bidding very near to the deadline of the auction would not give the incremental bidder sufficient time to respond to being outbid. By bidding at the last moment, you might win the auction at incremental bidder’s initial, low bid, even though the incremental bidder’s willingness to pay is more than your willingness to pay.

Special Issue of the International Journal of the Computer, the Internet and Management, Vol. 13 No.SP3, November, 2005

9.4

will not be successfully transmitted, in which case the winner only has to pay the minimum bid. (see Ockenfels and Roth, 2001, for a game theoretic model and formal proof that this can constitute equilibrium behavior.) 3.4 Bidding Late to Protect Information in Auctions with Interdependent Values

Figure 3. Late Bidding as Best Response to Incremental Bidding

Figure 3 shows the bid history. It reveals that about five minutes left, bidder adgatto placed a proxy bid of 25,095, and, finding that it is not the winning bid. Then he raised it twice but it is still not the highest due to another higher proxy bid from richardb2. It thus appears that, if richardb2 had bid later than addgatto, and too late for him to respond, he would have saved over two thousand dollars, since adgatto tried to fight to be the leading bid. 3.3 Bidding Late to Avoid Bidding Wars with Like-minded Bidders For example, suppose that you are willing to pay up to $100 for an antique coin, and that there is only one other potential bidder who you believe has also a willingness to pay for about $100 as yourself. If both of you submit your value early, your earnings (calculated as your value minus the final price if you are the winner, and zero if you are the loser) would be close to zero. But if we consider a strategy to bid at the very last minute and not to bid earlier, unless the others bid earlier, we will have positive expected profits even if both bidders follow this strategy and mutually delay their bid until the last minute, because there is a probability that one of the last-minute bids

Suppose you are a dealer of antique coins who can distinguish whether a coin is genuine or worthless and identify an antique coin auctioned in eBay as genuine. Your early bids may provoke others bidder to increase the willingness to pay because the bid might inform bidders who are inexperienced in the antique item to follow expertise. Due to bidding without such a signal from the expert, it would run the risk of losing money to a worthless coin. Such conditional bidding behavior creates incentive for experts to submit the bid for a genuine item very late in order to prevent other bidders from cashing in on their expertise. Bidding late does not leave sufficient time to uninformed bidders to respond to and outbid experts’ bids (see Ockenfels and Roth, 2001, for a game theoretic model). 3.5 Strategic Versus Non-Strategic Hypotheses for Late Bidding Of course, there can also be nonstrategic reasons why bidders bid late, some of which are listed in the following table. Table 2 Hypotheses of The Cause of Late Bidding

Proceedings of the Fourth International Conference on eBusiness, November 19-20, 2005, Bangkok, Thailand

9.5

3.6 Experiments with Sniping The result from the sniping agent JbidWatch worked with experienced eBay participants. The following table shows the number of auctions and the winning percentage separated into three categories of times. Table 3 Sniping History and Success Rate

The first column shows a manual sniping period when bidders have time to monitor the bid by themselves. The second column shows the agent based sniping period with JbidWatch. Both reflect the effectiveness of the late bidding strategy that can gain the winning percentage of more than 60% for manual or automatic agent of the third party. Thus, the bidder lost only when her maximum bid was too low or when another bidder had a higher willingness to pay for the item. 3.7 Danger of Last Minute Bidding There is a warning from eBay user (Axis Mundi) which states that, there is nothing they can do to help bidders who were “locked out” while trying to place a last-minute bid because of eBay’s high traffic during the close of an auction. Another warning about late bidding comes from auctionwatch.com, a rich source of information for users of Internet auction (“There are inherent risks in sniping. If you wait too long to bid, the auction could close before your bid is processed”) and from esnipe.com, an online agent that places late bids on behalf of its users (“… network traffic and eBay response time can

sometimes prevent a bid from being completed successfully. This is the nature of sniping”). By the survey of 368 eBay bidders who successfully bid at least once in the last minute of an auction were sent a questionnaire. 86 percent of all bidders reported that it happened at least once to them that they started to make a bid, but the auction was closed before the bid was received. Another risk of late bidding is about 90 percent of all bidders say that sometimes, even though they planned to bid late, something came up that prevented them from being available at the end of the auction so that they could not submit a bid as planned. 4

Drawback on Current Bidding

Based on the current proxy system, bidders can set their maximum bid for the agent to bid on their behalf during the auction period. Proxy bid will immediately run to outbid the manual bids that come from other bidders. But due to scenarios of drawbacks of submitting bid too early that have been discussed in previous topics, some might have to pay the cost of time to keep on monitoring the bidding transaction till the ending period of the auction to submit the late bid. Or some might take a risk to express their auction password to the third party software to track their attended auction and place the bid instead of doing it themselves even though, there is no guarantee at all that the late bid submitted by bidders or buy the third party agents will successfully proceed to the auction before the auction closes. There are some statements in the research observation that from a bidding strategy perspective, eBay’s proxy bidding agent does not allow the bidders to specify the time they want to place their bid. The proxy bid will deploy itself immediately after that information is specified and this has a significant impact on their expected

Special Issue of the International Journal of the Computer, the Internet and Management, Vol. 13 No.SP3, November, 2005

9.6

payoff. Why we would not offer a sniping option on the internet auction site itself. 4.1 PBSTAE, the Fair Features As this PBSTAE involves the introduction of the new service in the system, the information is gathered from the research paper. So, the system requirements are not from users but from the problem that can be extracted from the research. Functionalities fall into five categories: Outputs: - The main production to user of the system is the item information for seller to buyer, the information has to be correct and complete, direct from what seller posted to what bidder chose to view. Inputs: - The budget price for this system should be the true maximum budget. Processes: - After the valid bid is submitted and updated, system should show the new minimum requirement for bid price. - When the bidder sets the agent for the second time, the new bid amount should replace the previous one. - For every bid placed, system has to perform agent alert to check whether there are some agent that have higher value to bid or not. - System has to alert just the agents that the current time has passed the start time specified to every agent. - If the agent sets a bid in advance and finally the winner is from the agent and their proxy bids are tied at the end of the auction, the system has to choose the first one who set that agent amount to be the only winner. - In case no one bids for the item, if there is some proxy bid set, the highest proxy bid amount is the winner and pay only one increment above the second highest price.

Timing: - System has to continuously check the start time of the agent for every second to alert the agent that the due time to start is met. - When the time of auction is met, the system should not allow bidder to place any bid. Controls: - Every bid submitted has to be more than the current bid by at least one minimum increment set by the item owner. The result of this project would be that the system would help the bidder by providing the proxy bid system to bidder in order to place a bid on the behalf of bidder after they set their maximum bid amount only once in the auction period. Moreover, bidders would gain the advantages of sniping strategy that is described in the previous section. But it is easier because the bidders do not have to monitor it throughout the auction period or wait till nearly the closing time to place a bid. PBSTAE has the additional feature that allows a proxy bid to specify the time for it to trigger the bid price, even at the last second of the auction. 5

Conclusion and Recommendation

For a long time, the internet auction people have been discussing about their strategies and the service of each internet auction site. Most of the strategic bidding talked about is the snipe bidding. It needs bidders to spend time on the internet at the closing time to place their bid in order to snipe for the item. Many bidders have to express their password that is used to login to the auction site to the agent from the third party software to place the bid for them. This is because the service of the automatic bidding that is currently provided in the internet auction site does not reflect the behavior of most bidders.

Proceedings of the Fourth International Conference on eBusiness, November 19-20, 2005, Bangkok, Thailand

9.7

If sniping and sniping agents become even more widespread than they are today, why do we not employ the technique into the site itself and help the user instead of the third party agents. Ultimately, it is going to be in everybody’s best interest if the online auctioneers offer a full portfolio of auction design choices. To begin with, for a non-sealed bid auction, they should offer the seller the choice of using a hard or soft closing. In addition, they should offer a choice of the ascending English, the descending Dutch, and the sealed bid second price Vickrey auction and let the market decide which is the suitable mechanism for a variety of product. The outcomes of auction are not supposed to depend on how fast one can type in a bid, or on the speed of one’s internet connection. 6

[6] Roth, Alvin E., and Axel Ockenfels (2002), “The Timing of Bids in Internet Auctions: Market Design, Bidder Behavior, and Artificial Agents.” Discuss paper on Strategic Interaction -33 [7] Shelly, Gary B., Rosenblatt, Harry J., Cashman, Thomas J. Systems Analysis and Design, Third Edition. International Thonson Publishing Company. [8] Sommerville, Ian. (1997), Software Engineering, Fifth Edition. Addison Wesley Longman Limited. [9] Utku, M. Utku. (2002), “Internet Auction with Artificial Adaptive Agents: Evolution of Late and Multiple Bidding” Society for Computational Economic.

References

[1] Ariely, Dan; Ockenfels, Axel and Roth, Alvin E. (2002) “An Experimental Analysis of Ending Rules in Internet Auctions.” Unpublished manuscript, Harward University. [2] Bapna, R. (2002), “When Snipers Become Predators: Can Mechanism Design Save Online Auctions?” Communications of the ACM, , forthcoming [3] Cahill, Aileen M., Jaworski, Bernard J., Mohammed, Rafi A., Fisher, Robert J. Internet Marketing: building advantage in a networked economy. McGraw-Hill Companies, Inc. [4] Hoffer, Jeffrey A., George, Joey F., Valacich, Joseph S (2002). Modern System Analysis and Design, Third Edition. Prentice-Hall International, Inc., [5] Roth, Alvin E., and Axel Ockenfels, “LastMinute Bidding and the Rules for Ending Second-Price Auction: Evidence from eBay and Amazon Auction on the Internet,” American Economic Review, forthcoming

Special Issue of the International Journal of the Computer, the Internet and Management, Vol. 13 No.SP3, November, 2005

9.8