Revenue Maximization with Minimal Information

Consider a single-item auction with two buyers whose values are independently drawn. Suppose that the seller lacks comprehensive knowledge of the market but can exert some effort to get a more nuanced understanding. We study the potential increase in the revenue that can be achieved by a seller that acquires a minimal amount of information: a single yes/no question that effectively partitions the set of possible type profiles into two parts. We identify revenue-maximizing partitions within several natural families. Additionally, we investigate cases where the seller is committed to a specific allocation function but seeks to optimize revenue through additional information. We show that allocational efficiency and revenue maximization can be achieved simultaneously in this context.

 

(Joint work with Itai Ashlagi, Jacob Leshno, and Sigal Oren)

 

Shahar Dobzinski's homepage: https://sites.google.com/site/dobzin