Price Cycles in Oligopoly: What We Can Learn from Online Advertising Auctions

By: Juan Feng and Michael Zhang


Abstract: 
This talk is in the area of  mechanism design, at
the intersection of microeconomics and algorithmic game theory.

We study dynamic price competition in Internet advertising auctions.
With a dynamic model, we find three possible equilibrium price patterns:
(1) high-low bids: one bidder dominates with a very high bid; (2)
low-low bids: both bidders bid the reserve price; and (3) Edgeworth
price cycles: prices build up and collapse in cycles. We empirically
examine the real world behavior of the advertisers and our analysis
lends support to the price-cycles result. In contrast to the classical
spatial or temporal forms of price dispersion, we propose a type of
"reactive price dispersion" as the result of dynamic price
competition. Finally, we consider the implications for collusion and
price wars.