Finding Edgeworth Cycles in Online Advertising Auctions

Search engines make a profit by auctioning off advertisement positions through keyword auctions. I examine the strategies taken by the advertisers. In a theoretical model, it is shown that the equilibrium bids should follow a cyclical pattern— “escalating” phases interconnected by “collapsing” phases— similar to a pattern of “Edgeworth Cycles” that was suggested by Edgeworth (1925) in a different context. I empirically test the validity of the theory. With an empirical framework based on maximum likelihood estimation of latent Markov state switching, I show that Edgeworth price cycles exist in this market. I further examine, on the individual bidder level, how strategic these bidders are. My results suggest that some bidders in this market adjust their bids according to Edgeworth predictions; there are also a group of other bidders who are generally less strategic. Finally, I discuss the important implications of finding such cycles.