Research and Advances
Computing Applications Virtual extension

Cross-Bidding in Simultaneous Online Auctions

  1. Introduction
  2. What is Cross-Bidding
  3. Findings of Our Study
  4. Discussion
  5. Conclusion
  6. References
  7. Authors
  8. Footnotes
  9. Tables

Online auctions have dramatically transformed the way many people trade goods and services on the Internet. This is evident from online auction giant eBay’s 2005 year-end figures: over 180 million registered users, 1.9 billion listed auctions (33% increase over 2004), and $44.3 billion worth of transacted merchandise (30% increase over 2004).6 Online auctions are popular among buyers because they aid in price discovery for hard-to-price products such as used merchandise, enable market-making for geographically dispersed buyers and sellers who cannot otherwise reach each other, and provide buyers with a sense of winning an item instead of purchasing it. For sellers, auctions allow them to move hard-to-sell and perishable products (e.g., vacation rentals) which are often difficult to sell without significant infrastructure, provide the Internet as an alternative and less expensive distribution channel compared to offline means, and extract the most consumer surplus possible in an efficient market.

The popularity of the online auction market and evolving variations in the auction format from single-item auctions to multiple-item simultaneous or sequential auctions have motivated bidders to experiment with different bidding strategies. One such strategy is “cross-bidding,” where bidders move between multiple simultaneous auctions of the same product.1,10 Given that large auction sites such as eBay often list many units of popular items or “hot sellers” from the same or different sellers at any given instant in time, homogeneity in product quality across different auctions especially for new “in-box” merchandise, and the relative lack of a cost barrier on the part of bidders to compare and switch between these alternate listings, cross-bidding is increasingly becoming a prevalent bidding pattern. However, while some previous studies have examined bidding behaviors and strategies within the context of a single-item auction,7,11 very little investigation have focused on simultaneous multi-unit auctions1 or examined the efficacy of newer bidding strategies such as cross-bidding.

The goal of this paper is to provide greater insights into the cross-bidding phenomena and thereby, improve our understanding of this emerging bidding strategy in simultaneous online auctions. More specifically, we seek to answer the following questions: what proportion of online auction bidders engage in cross-bidding, how many simultaneous auctions do cross-bidders participate in, what is the success rate of cross-bidders in winning online auctions, how do the closing prices for cross-bidders differ from those of traditional bidders, and how do cross-bidders differ from traditional bidders?

Answering the above questions is important not only for enhancing our knowledge of bidder behaviors in simultaneous online auctions, but also for the better design of online auctions. The results reported in this paper have implications for both bidders and sellers in simultaneous online auctions as well as for online auction sites managing such auction formats. Further, the findings may motivate auction researchers to delve deeper into the complex behavioral dynamics of simultaneous online auctions.

Back to Top

What is Cross-Bidding

Cross-bidding refers to bidding by a given bidder across simultaneous and competing auctions with similar ending times.10 When a cross-bidder is outbid in one auction, she may either bid again in that auction or move to one of the simultaneously running competing auctions if the competing auction has a lower standing bid than the current auction. The process is repeated when the bidder is outbid again, until all auctions end. The cross-bidder’s goal is to win a single product at the lowest possible price by continually monitoring and moving between multiple auctions. Bidding across multiple auctions diversifies the bidder’s risk of being outbid in a single auction, increases visibility of product supply across different auctions, and potentially lowering the price paid on the winning auction.

In order to avoid winning multiple items from the competing auction listings, a cross-bidder typically bids in only one auction at any given instant in time. Instead of bidding her true valuations (willingness to pay) upfront in one auction, she will incrementally increase the lowest standing bid in a set of competing auctions by moving back and forth between these auctions in an effort to win one auction with the lowest possible closing price. Because cross-bidding involves considerable complexity and overhead in continual monitoring, evaluating, and bidding between different auctions, cross-bidders are expected to be more experienced in online auctions than the traditional bidders (non cross-bidders).

As an example, consider an auction for a popular product such as an iPod music or video player. Assume that there are two currently active simultaneous auctions 1 and 2 listing the same product and ending at approximately the same time. Also assume that there are three bidders A, B, and C, where A and B are traditional bidders (non cross-bidders) bidding solely in auctions 1 and 2 respectively, while bidder C is a cross-bidder who is bidding in both auctions to win one listing at the lowest price.

Table 1 illustrates bidder C’s cross-bidding strategy. At time T, she notices that traditional bidders A and B have placed bids of $1 and $2 in auctions 1 and 2 respectively. Since auction 1 has the lower standing bid of $1, C bids $2 on this auction at time T+1 (if the minimum increment for each auction is $1) to become the highest bidder. However, if A responds by increasing the current standing bid in auction 1 to $3 at time T+2, then C would move to auction 2, with the then lower standing bid of $2, and bid $3 on this auction at time T+3. C continues to pursue this strategy until both auctions end, or until either A or B drop out of the bidding process when the bid prices exceed their private valuations.

Interestingly, if all bidders cross-bid, then theoretically, the price everyone pays for that product becomes identical and cross-bidding will fail.10 However, because not every bidder cross-bid, this market inefficiency translates into transient opportunities for cross-bidders, helping them realize lower closing prices than otherwise.1 Hence, cross-bidding takes advantage of market inefficiencies across the bidder population. Naturally, the cross-bidder, as a utility-maximizing rational agent, attempts to extract the most consumer surplus (the difference in the amount she is willing to pay versus the price actually paid) by moving back and forth between competing auctions.4

Back to Top

Findings of Our Study

To further investigate the five research questions presented earlier, we collected bidding and bidder data from actual auction listings on eBay. We extracted detailed listing information on six models of Apple iPods over a four-month period from October 2006 to January 2007. These six models included the Apple Shuffle (1GB), Nano (2, 4, and 8 GB), and video iPods (30 and 80 GB), which were the popular iPod models at the time of the study. Our choice of iPods was motivated by the large number of iPods listings on eBay at this time, which provided a large sampling frame for this study, and the large number of iPod buyers and sellers, which ruled out potential price manipulation by either buyers or sellers and made the iPod market more efficient. To eliminate any potential variances in auction outcomes due to varying supply and demand conditions for different products in different auctions, we excluded all other products and auction sites from our data sample.

Data was collected using a Java program which searched the titles of active eBay listings using the keyword “iPod” twice per day. The resulting list of auctions was filtered to remove used or refurbished iPods or those bundled with accessories. For the new iPods, the listing number, auction title, auction start date, and auction end date were entered into a database. After the auctions ended, the program downloaded the auction and bid history pages, parsed each page and extracted information relevant to our study such as bidders’ screen names, bid amount, time of bid, feedback score, etc., and stored them in a local database.

Over the four-month course of this study, we captured a total of 23,919 iPod listings on eBay. Among these listings, 21,714 auctions were successful and resulted in the exchange of goods between bidders and sellers, while the remaining auctions did not meet their prespecified reserve prices or failed to attract bidders and were therefore dropped from the sample. Excluding auctions that were not running simultaneously and hence infeasible for cross-bidding, we had a list of 633 simultaneous auctions in which bidders could cross-bid. The frequencies of successful auction listings, broken down by iPod models, are shown in Table 2.

A total of 5,366 distinct bidders placed bids in the 633 iPod auctions identified above. This group consisted of 304 cross-bidders, identified as those who placed bids in two or more simultaneous auctions ending within five minutes of each other. We considered auctions ending within close proximity of one another for two reasons. First, as most bidding activity tends to occur toward the end of an auction, auctions with significantly different ending times reduce a cross-bidder’s ability to move back and forth between auctions, thereby hindering cross-bidding. Second, following recommendations from previous studies,1 we examined auctions ending within 1, 5,10, and 15 minutes of each other and found no substantive differences in bidder feedback scores, prices paid, or other parameters of interest among the three groups, and therefore chose to retain the sample with a five minute differential. Our selection of the five minute differential was further motivated by Roth and Ockenfels’ study,” which observed that “a considerable share of bidders submit their bid in the last five minutes.”

In our sample, cross-bidders constituted only 5.7% (N=304) of the total bidder population, while the remaining 94.3% (N=5,062) were traditional or single-auction bidders. This suggested that cross-bidding is possibly still in its infancy and not yet popular among the bidder population. On average, cross-bidders placed bids in 2.1 simultaneous auctions, suggesting that they were likely moving between two competing auctions. Within each auction, cross-bidders placed an average of 2.4 bids, compared to 2.0 bids for traditional bidders. These descriptive statistics are shown in Table 3.

While 11.4% of the overall bidder population eventually won their targeted iPod auctions, this proportion was significantly greater for cross-bidders (20.4%, or 62 out of 304) than for traditional bidders (10.9%, or 552 out of 5,062) (see Table 3). Of the 62 successful cross-bidders, 85.5% (53 out of 62) paid less than the retail price of their specific iPod model. This retail price, obtained from Apple’s Web site,2 did not change during the four-month course of our study, even though new iPod models were introduced during this time. In contrast, among the successful traditional bidders, only 72.8% (402 out of 552) paid less then the retail price. Further, the winning price among cross-bidders, on average, was a discount of 8.6% from the retail prices of their respective models, compared to a 4.6% discount for traditional bidders. For example, the retail price for an Apple iPod Shuffle was $79, but cross-bidders paid an average of $72.21 and traditional bidders paid $75.36.

The above statistics collectively demonstrate that cross-bidders are more likely to win an auction than traditional bidders, more cross-bidders are likely to extract a consumer surplus (lower-than-retail price) than traditional bidders, and cross-bidders are likely to extract a greater consumer surplus than traditional bidders.

To examine whether cross-bidders are more experienced than traditional bidders in online auctions, we next proceeded to compare their feedback score on eBay at the time of auction closing as a proxy measure of their experience. eBay calculates a bidder’s feedback score as the difference between the number of positive and negative feedbacks received by that bidder for each auction she has won. Though the summation of positive and negative feedbacks may be a more accurate measure of bidder experience, the feedback score reported by eBay is an approximate measure because sellers rarely post negative feedback for winning bidders, fearing negative ratings in return.1 Further note that feedback score represent auctions won by the bidder, rather than auctions participated, which tends to systematically bias bidder experience toward a lower value. However, since such bias occurs for both all bidders (traditional and cross-bidders), and we are interested in a relative comparison of bidders’ experiences, rather than comparing their experiences against an absolute value, the above bias do not pose a threat to internal validity in our analysis. The average feedback rating for all bidders in our sample was 59.2; this rating was higher for traditional bidders (60.9) than for cross-bidders (31.1). Though this finding seems to contradict our expectation, it may be an artifact of the relative recency of simultaneous auctions and the cross-bidding phenomenon, since “old timers” (more experienced bidders) rarely had an opportunity to cross-bid during their initial days and have habitually tended to remain traditional bidders, while the less experienced “new comers” have had more opportunity and hence greater propensity to cross-bid.

Since cross-bidding could be confounded with sniping, both of which tend to occur toward the end of an auction, we also examined sniping activity among both traditional and cross-bidders. A bidder was considered to be a sniper if he/she placed a bid within the last 10 seconds of the auction. Our selection of 10 seconds was motivated by Roth and Ockenfels,11 which noted that 12% (or 29 out of 240) of eBay auctions had bids in the last 10 seconds of the ending time. Snipers constituted only 9.5% of the cross-bidder population, and 3.1% of traditional bidders, providing evidence that very few cross-bidders are also snipers at the same time. Further, there were no substantive differences between snipers and non-snipers in our cross-bidder population in terms of the number of simultaneous auctions they participated in (2.2 versus 2.1), number of bids per auction (2.1 versus 2.4), proportion of bidders paying less than retail (86% versus 85%), and price discount from retail (8.7% versus 8.6%). Sniping cross-bidders were more successful in winning auctions compared to non-sniping cross-bidders (48% versus 18%), presumably due to their sniping strategy, and were also more experienced (feedback score of 52 versus 29). Similar patterns were observed for the traditional bidder population, with no large difference on any metric except bidder success, bidder experience, and discount from retail. Though our findings demonstrated the occurrence of sniping, sniping does not seem to have any confounding effects because the same pattern of sniping effects were observed in both the cross-bidder and the traditional bidder populations. Because snipers constituted a small fraction (10%) of the cross-bidders, sniping occurred both among traditional and cross-bidders, and the outcome of sniping was equivalent among both traditional and cross-bidders, sniping did not confound our results or pose any significant threat to the internal validity of our findings.

Back to Top


This study started with the goal of addressing five research questions: what proportion of online auction bidders cross-bid, how many simultaneous auctions do cross-bidders participate in, what is the success rate of cross-bidders in winning online auctions, how do the closing prices for cross-bidders differ from those of traditional bidders, and how do cross-bidders differ from traditional bidders? We used actual eBay data on iPod auctions to answer these questions. Below, we elaborate our findings on each of these questions.

First, the results of our study indicate that cross-bidders constitute only a small minority (5.7%) of the bidder population. One potential reason for this may be the relative recency of the cross-bidding phenomenon and hence its lack of awareness among much of the bidder population. A second reason may be that cross-bidding imposes significant cognitive overload for the bidder in terms of their time and effort to monitor several auctions simultaneously, decide on which auction to bid, and determine the right amount to bid. Bidders unwilling to or incapable of taking on such a demanding cognitive load may decide against cross-bidding.

Second, we demonstrate that the average cross-bidder participates in just over two simultaneous auctions. Though this figure may vary with auctions involving other products, given the complexity of the cross-bidding strategy, it is not surprising that cross-bidders can only reasonably handle only a small number of simultaneous auctions. The effort required to monitor, evaluate, and bid in multiple auctions tend to increase exponentially with the number of competing auctions, and will likely decrease a cross-bidder’s efficiency in responding to competing bids, limiting the extent to which she can cross-bid. Special cross-bidding software for automated monitoring and bidding can ease this burden and may therefore increase the frequency of cross-bidding. However, more widespread use of such software will also reduce the overall efficacy of the cross-bidding strategy.

Third, in a population consisting of cross-bidders and traditional bidders, cross-bidders are more likely to win auctions than traditional bidders. In our data sample, cross-bidders were almost twice as likely to win iPod auctions (20.4%) as traditional bidders (10.9%). The higher success rate for cross-bidders can be attributed to their multi-auction bidding strategy, which diversifies their risk of being outbid in a single auction.

Fourth, we demonstrate that more winning cross-bidders (85.5%) pay less than retail price for the auctioned product than do winning traditional bidders (72.8%). Further, the winning price among cross-bidders (8.6% discount off the retail price) tends to be lower than that for traditional bidders (4.6% discount). Not only does cross-bidding tend to provide a better bargain due to its consideration of a basket of auctions, in contrast to locking into a single auction in traditional bidding, but cross-bidders are also less likely to be involved in a bidding war, or get caught in the frenzy of emotionally charged and frantic bidding, and outbid their private valuations—a circumstance termed “the winner’s curse” in the auction literature.8,9

Finally, regarding our last research question, we expected cross-bidders to have greater auction experience than traditional bidders, but were unable to confirm the same. In our sample, cross-bidders were experienced, but not more experienced than traditional bidders. More research is required to identify this and other systematic differences between traditional and cross-bidders.

Back to Top


In closing, this article provided new insights into cross-bidding behaviors in simultaneous online auctions using actual data from eBay. We show that cross-bidding is still in its early stages, and that a cross-bidding strategy can provide better auction outcomes than traditional bidding. We quantify the amount of consumer surplus (such as price discounts) generated from cross-bidding, compared to that from traditional bidding. We also warn that this performance advantage for cross-bidders may work only as long as the bidder population consists of both cross-bidders and traditional bidders, and that it may erode as more traditional bidders become cross-bidders. We hope that our research will attract more attention to cross-bidding and to bidding strategies in simultaneous online auctions, and will motivate other researchers to extend our work by probing into the underlying reasons of cross-bidding and/or extend existing theoretical frameworks to explain cross-bidders’ behaviors.

Back to Top

Back to Top

Back to Top

Back to Top


T1 Table 1. Cross-Bidding Example

T2 Table 2. Distribution of iPod auction sample

T3 Table 3. Comparing Traditional and Cross-Bidders

Back to top

    1. Anwar, S., McMillan, R. and Zheng, M. Bidding behavior in competing auctions: Evidence from eBay. European Economic Review50, (2006), 307–322.

    2. Apple. The Apple Store (U.S.).

    3. Apple. Press releases—September 2006.

    4. Ba, S. and Pavlou, P.A. Evidence of the effect of trust building technology in electronic markets: Price premiums and buyer behavior. MIS Quarterly 26, 3, (2002), 243–268.

    5. Bapna, R., Goes, P. and Gupta, A. Replicating online Yankee auctions to analyze auctioneers' and bidders' strategies. Information Systems Research 14, 3, (2003), 244–268.

    6. eBay. eBay Inc. announces fourth quarter and full year 2005 financial results. http://files.shareholdercom/downloads/ebay/84673029x0x40071/85a41a2f-dd13-40dd-9ebd-61454b020800/EBAY_News_2006_1_18_Earnings.pdf.

    7. Gilkeson, J.H. and Reynolds, K. Determinants of Internet auction success and closing price: An exploratory study. Psychology & Marketing 20, 6, (2003), 537–566.

    8. Ku, G., Malhotra, D. and Murnighan, J.K. Towards a competitive arousal model of decision-making: A study of auction fever in live and Internet auctions. Organizational Behavior and Human Decision Processes 96, 2, (2005), 89–103.

    9. Lind, B. and Plott, C.R. The winner's curse: Experiments with buyers and with sellers. The American Economic Review 81, 1, (1991), 335–346.

    10. Peters, M. and Severinov, S. Internet auctions with many traders. Journal of Economic Theory 130. (2006), 220–245.

    11. Roth, A.E. and Ockenfels, A. Last-minute bidding and the rules for ending second-price auctions: Evidence from eBay and Amazon auction on the Internet. The American Economic Review 92, 4, (2002), 1093–1103.


Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More