Abstract
Many e-commerce platforms use buyers' personal data to intermediate their transactions with sellers. How much value do such intermediaries derive from the data record of each single individual? We characterize this value and find that one of its key components is a novel externality between records, which arises when the intermediary pools some records to withhold the information they contain. Ignoring this can significantly bias the evaluations of data records. Our analysis has several implications about compensating individuals for the use of their data, guiding companies' investments in data acquisition, and more broadly studying the demand side of data markets. Our method combines modern information design with classic duality theory and applies to a large class of principal-agent problems.