E-Bay's happy hour: Non rational herding in online auctions

Abstract

We hypothesized that on-line auction bidders would herd behind other bidders even when observed choices did not reveal private information. A model that inserts bidders engaging in this type of non-rational herding into a competitive market shows that, in equilibrium, (some) sellers set low starting-price in order to attract low valuation bidders who in turn bring high valuation ones. The model leads to three predictions, all of which found support in a sample of 8,300 eBay auctions for DVD movies: (1) conditioning on current price, low starting-price auctions are more likely to receive additional bids, (2) a bid of a certain dollar amount is less likely to win a low starting-price auction, and (3) low starting-price auctions are more likely to attain high selling prices. We rule out alternative explanations based on unobserved heterogeneity across items with different starting-prices, on the possibility that bidders may become attached to items they place early bids on, and that snipers decide what auctions to bid on while prices are still low.