Using Grouped Data to Estimate Revenue Heterogeneity in Online Advertising Auctions

Abstract

This paper estimates the heterogeneous impact of advertising networks from the perspective of a publisher who has access to limited information provided by the advertising platform in the form of grouped data over different auctions and users. The models account for the high-dimensional nature of the data and allow for time-varying interactive effects. We estimate models for different countries, and the measured heterogeneity may reflect factors such as local competition or cost effectiveness.