Working Papers
Price Recommendations and the Value of Data. (JMP) (link)
Abstract: This paper presents a novel methodology to regulate data collection for e-commerce platforms based on the marginal value of information. I consider a platform that discloses buyer information to sellers, via price recommendations, to influence their prices. I determine the gain in profit from acquiring datasets with marginally more buyers under two business models. First, when a platform charges buyers and sellers, it discloses information efficiently, but undervalues data. In this case, a subsidy scheme restores efficiency. Second, when a platform provides free access to buyers, it discloses information and ranks datasets inefficiently. I develop a classification of datasets depending on if they increase welfare or consumer surplus.
Short video presentation (8min30) : Link
Collusion under incomplete information on the discount factor. (link)
Abstract: The gradual increase in prices at the start of collusion is a recurrent pattern that has been observed in many discovered cartels. When firms have private information regarding their respective discount factors, I show that optimal collusive schemes feature a transition phase during which prices increase gradually. Impatient firms, for which sustainable collusive prices are low, are willing to mimic patient firms to undercut them at a high market price. To elicit firms’ true discount factors, optimal collusive strategies take advantage of the differences in time preferences across types. Patient firms delay the period during which the highest collusive prices are set to force impatient firms to undercut early and reveal their type. In addition, patient firms find it optimal to reach the highest collusive prices by employing a gradual price path. I characterize the Pareto payoff frontier and compute the optimal speed of price increases during the transition phase.
Work in Progress
Price Recommendation and the Value of Data: Competition. (link)
Abstract: I study whether competition between e-commerce platforms reduces the distortion in their incentives to collect data. In a Hotelling model with co-located outside option, I study competitive equilibria between two platforms charging participation fees and using price recommendations. Both platforms disclose strategically their information about buyers’ valuations to sellers thereby influencing their sellers’ pricing decisions. Platforms gain market shares by lowering the average seller prices, hence increasing buyer expected surplus from trade. Results show that increasing the degree of competition decreases the distortions in the platform’s entry fees and incentives to collect data.
The Production of Information.
Abstract: We call the costly process of transforming data into actionable insights the production of information. We develop a two-stage model of a firm that first produces information and then faces a finite decision problem under uncertainty. We apply our model to a monopolist producing information to price discriminate buyers. We characterize the profit-maximizing posterior belief distribution under costly production. The value of spreading posteriors to take different actions equals the Bregman divergence of the marginal cost of production. We exhibit robust distortions with the welfare-maximizing production of information.