The Potential For Collusion in Ascending Price Acutions
Posted by Erica Myers on April 11, 2008
The auction design team that authored the October RGGI auction design report recently posted an addendum to on Resources for the Future’s website. The report includes reactions to comments and suggestions that have been made by interested parties and presents new findings based on research done subsequent to the submission of the final report.
One of the more interesting findings is that “clock auctions” may be more susceptible to explicit efforts to collude than sealed-bid formats.
For the purposes of the report and the subsequent addendum, a “clock auction” refers to an ascending uniform price auction, where bidders submit quantity bids for the number of allowances they would buy at a given price. In this case, the initial price is the reserve price and if quantity demanded exceeds the number of allowances being sold, a second round of bidding begins and the price goes up incrementally (like a tick of a clock).
A sealed-bid uniform price auction differs in that there is only a single round of bidding where participants submit demand schedules with both prices and quantities. The highest bids are accepted according to the number of allowances that are for sale and the first rejected bid sets the price for all of the allowances sold in the auction.
The RGGI design team ran a series of experiments to test the performance of different auction designs. The subjects were undergraduate students at the University of Virginia earned real money for participating. When subjects were allowed to collude explicitly through instant message chatroom, they were much more successful in keeping the price well below the competitive equilibrium in the clock auction format as opposed to other auction formats.
An analysis of the actual chat between subjects provides some insight for why collusion is more successful in clock auctions. Most of the initial proposals were based on suggesting quantity reductions for low and high users. The focus on quantity reduction in the clock auction sessions is revealed in some of the participants’ comments: “again, bid for fewer permits earlier on so we can get permits cheaper” and “this will go 5X faster and will all make LOTS more money if everyone just cooperates the first time.” One person suggested “so why doesn’t everyone bid exactly the same amount that we ended last round [auction] with, since we keep getting the same clearing price.” This plan permitted participants to obtain the same final permit allocation without the run up in prices that occurred previously. Of course, quantity discussions occurred with the other auction formats too, but the effect of the clock is to take out the price dimension so that bidders only have to reach an agreement in a single dimension, quantity.
The worry with this type of collusion is that it often leads to inefficient outcomes where those who value the allowances the most are not necessarily the ones who get them.
This chatroom set up is very different than the conditions for the RGGI auctions. However there is some empirical evidence to suggest that coordinated demand reductions have been used in ascending auctions in the past.
There are several widely cited cases in which coordinated demand reductions are used to defeat ascending price auctions for broadcast spectrum. Just prior to the 2001 Austrian UMTS auction, Telekom Austria announced that it “… would be satisfied with just two out of the 12 blocks for offer and if the [five] other bidders behaved similarly, it should be possible to get the frequencies on sensible terms … but that it would bid on a third block if one of its rivals did …” The other bidders clearly understood and the bidding stopped after a couple of rounds, with each bidder obtaining 2 blocks (Klemperer, 2004, p. 136). The extreme symmetry of this situation differs from the bidding environment faced by bidders in RGGI auctions, but the success of collusion triggered by a public announcement is disturbing.
This type of coordination may be more or less of a concern depending on the individual characteristics of the good being sold, the auction participants, and the regulatory environment.
The addendum also has a description for a combined vintage auction that is designed to correct the inefficiencies that could be caused by having two separate uniform price auctions for two different allowance vintages. I mentioned this in a previous post.