Optimal Design of Revenue-Maximizing Position Auctions With Consumer Search

Abstract

We examine the impact of consumer search from the perspective of revenue-maximizing optimal auction design. In this two-sided market, advertisers are endowed with privately known qualities, and consumers incur heterogeneous search costs when clicking on advertisements displayed by the search engine. Our general approach eliminates ad hoc restrictions and therefore allows us to uncover the salient economic trade-offs. We adopt the optimal mechanism design approach to examine this two-sided private information problem. We show that from the revenue maximization perspective, the search engine necessarily displays the advertisers in the descending order of their qualities. The search engine shall insert unnecessary hassles to induce the revenue-maximizing search behaviors. Thus, each consumer stops searching pre-maturely compared to the socially optimal search behavior. Moreover, the obfuscation level is higher for consumers with higher search costs. In addition, the search engine shall implement probabilistic forced termination upon consumers' clicking on advertisements, and this probability is higher for advertisers with poorer qualities. When each advertiser has a two-dimensional type on consumer learning and matching, we show that even if consumers' search costs become negligible, the allocative inefficiency does not vanish. Moreover, no strategic obfuscation is imposed. When allowing for non-independent qualities, we show that product variety does not constitute a rationale for advertisers' favoritism.