To Score or Not to Score? Estimates of a Sponsored Search Auction Model

Using data from "WebsiteX", one of the largest online marketplaces in the world, we estimate a structural model of sponsored search auctions where bidders have heterogeneous click-through curves. Unlike earlier studies, our model accommodates two stylized empirical facts: the advertiser prominence eff ect and the position paradox. Using our estimates, we simulate the e ffects of introducing bid-scoring to the auctions. We fi nd that scoring reduces equilibrium per-click prices, but boosts the number of clicks by sorting prestigious merchants to the top positions. Overall there is only a very modest reduction in total revenues from introducing bid-scoring, despite the intent to reward high-quality merchants with price discounts. Methodologically, this paper also illustrates an application of a novel "approximate Bayesian" estimation method to a structural multi-item auction model