An investigation of price dispersion in Internet auctions

Abstract

It can be observed that in consumer-to-consumer on-line auctions, there is a strong degree of price dispersion, even in liquid markets with a large number of bids per auction item. This Thesis research strives to quantify such dispersion and present explanations for the key findings on the nature of price dispersion in on-line auctions of like goods: Although contrary to common sense and public opinion, I found no significant correlation of winning bid price with feedback rating. This is true for low, medium-high and high value goods and for both new and used goods. This observation holds for all ranges of feedback ratings. Of all of the other variables of an auction listing: (shipping costs, amount of opening bid, number of bids, accepted payment types, picture in listing, and the use of reserve prices), only picture in listing and use of reserve prices are correlated with winning bid price. This observation holds for all types of goods examined. However, correcting for these variables does not significantly reduce price dispersion. The majority of the dispersion in winning bid prices of same-good auctions can be explained by the unique ability of the on-line auction process to obtain the reservation prices of buyers in the market, through its use of maximum bid amounts and proxy bidding. An auction winning bid curve for a good can be constructed by aggregating the winning bid prices of all auctions of that good over a period of time. The downward sloping and isoelastic nature of this curve can be explained in part by the downward sloping demand curve for that good, made up by the range of premiums buyers are willing to pay in order to increase their chances of winning the auction, and getting the item sooner. The exact shape of the curve can be further explained and in fact reconstructed by an examination of the statistics of grouping bids in auctions. Arbitrage to take advantage of the dispersion of winning bid prices can be shown to be possible, and in theory it appears to be surprisingly profitable. This suggests that on-line auction markets are only weakly market efficient. Several explanations were explored as to why arbitrage has not been pursued, thereby eliminating the observed dispersion. These explanations include the unseen non-monetary transaction and arbitrage costs, the level of sophistication necessary to capitalize on opportunities, and the persisting immaturity of the consumer-to-consumer on-line auction market.