New economic research suggests some winning strategies for buyers and sellers taking part in auctions in which a number of similar items are sold one after another. The analysis by George Deltas and Georgia Kosmopoulou, published in the January 2004 issue of the Economic Journal, indicates that:
· If you are a buyer and are looking for bargains, bid on lots that are sold late in the auction, have relatively short descriptions in the catalogue, are placed in the middle of pages, and have small or no photographs associated with them.
· If you are a seller and you have some control over the placement of your lot in the auction, place it early in the auction if it is minor item (relative to the other items on sale) and late in the auction if it is a major item.
The researchers study two basic questions. When looking to buy in an auction where similar items are sold sequentially, should you compete for those that are put up for sale relatively early or those that are put up for sale relatively late? Similarly, as a seller in a sequential auction, should you place your item early in the auction or late?
Economists have investigated these questions since the mid-1980s, when Professor Ashenfelter of Princeton University discovered that identical cases of wine sold by auction an hour apart from each other differed systematically in price. The lots of wine sold later in the auction tended to sell for lower prices than the lots of wine sold earlier.
This was dubbed an ''anomaly'' because it presented a clear arbitrage opportunity for potential buyers. A number of studies have since observed that some products sold by such sequential auctions tend to experience declining prices, while other products tend to experience increasing prices. Understanding why that''s the case can help sellers better sell their products and buyers better time their purchases.
Most researchers have attributed the patterns to the strategic behaviour of very sophisticated bidders: they show that such strategic behaviour can indeed create predictable price trends. Other researchers have attributed the patterns to subtle differences in the quality of the items sold over an auction, showing that in some of these auctions items sold early are sometimes of better quality than those sold late.
This research takes a radically different tack. The authors investigate whether some of these price trends could arise from the nature of the auction catalogue. They note that if the price trends arise from the way the catalogue is written, then bids submitted by mail would also exhibit systematic trends. This indeed turns out to be the case. Using data from a sale of rare books, the researchers find that the probability that a lot receives no mail-in bid increases for lots listed further back in the catalogue (and sold late in the auction).
Moreover, the variance in the number of bids tends to increase with the print order. Lots listed early in the catalogue receive a more uniform number of bids than lots listed later in the catalogue. The researchers have been able to trace these patterns to bidder (or rather ''reader'') fatigue. In the first pages of the catalogue, the bidders give full attention to all lots. As they go through the catalogue, fatigue sets in. The bidders'' attention gets ''captured'' by certain features of the description of the lot, such as its location on the page and the length of the description. Lots with a longer than average description length, or with a description that is located at the top of a page, get an increasing number of mail-in bids, while the other lots get a progressively smaller number of bids, and frequently get none.
The increased variability of the number of bids carries over to prices. Had mail-in bidders been the only participants in the auction, price variability and expected price would be higher for lots placed at the end of the catalogue. However, the expected price of lots with short descriptions is lower the further down the catalogue these lots are placed. It is only lots with long and average length descriptions that experience price increases.
The impact of description length is quantitatively important. A lot with a 21-line description that is moved from the first half of the auction to the second half would experience a price increase of 23%; a lot with a 9-line description would experience a 2% decline (the median description length is 14 lines).
These effects arising from the print order of the catalogue also influence, though to a lesser extent, the behaviour of floor bidders, that is, bidders who submit bids in person. Thus, actual prices in any auction with a catalogue are in part determined by these effects.
''"Catalogue" vs "Order-of-sale" Effects in Sequential Auctions: Theory and Evidence from a Rare Book Sale'' by George Deltas and Georgia Kosmopoulou is published in the January 2004 issue of the Economic Journal. Deltas is at the University of Illinois; Kosmopoulou is at the University of Oklahoma.
