Working Papers 2025
Working paper 1-2025
Kfir Eliaz and Ran Spiegler
Monopolistic Data Dumping
Abstract
A monopolist curates a database of current and historical observations for users who want to learn some parameter. Nowcasters ("forecasters") wish to learn its current (long-run) value. The monopolist chooses the size of each data type, facing constant marginal storage cost, and a menu of contracts, consisting of a fee and access level to each data type. The optimal menu offers full access to historical data, but discriminates access to current data: either full to all consumers or full to nowcasters and none to forecasters. Relative to social optimum, there is too much (little) historical (current) data, and sometimes too much total data.