Reproduce “The Economics of Social Data” on OpenResearch

About this paper

Dirk Bergemann, Alessandro Bonatti, Tan Gan

The authors model a data intermediary that pays consumers for signals and sells information to a producer, who can receive identity-linked or anonymized data. Anonymization preserves market-demand information but removes links between signals and people. With homogeneous consumers, the intermediary prefers anonymity over complete intermediation if and only if anonymity raises social surplus; this result applies only to that choice.

Key insights

• Social data can generate positive or negative externalities: other consumers’ signals may help an individual infer her own preferences, but may also help a producer set more targeted prices against her.
• The socially preferred allocation shares information among consumers while withholding it from the producer, but a profit-maximizing intermediary generally sells some producer access because the producer finances data acquisition.
• Anonymization is valuable not because consumers directly demand privacy in the model, but because removing identity links can reduce inefficient price discrimination while lowering the intermediary’s compensation costs.

Summary by alphaXiv.