Working Paper: CEPR ID: DP16507
Authors: Stephen Morris; Dirk Bergemann; Tibor Heumann; Constantine Sorokin; Eyal Winter
Abstract: In digital advertising, a publisher selling impressions faces a trade-off in deciding how precisely to match advertisers with viewers. A more precise match generates efficiency gains that the publisher can hope to exploit. A coarser match will generate a thicker market and thus more competition. The publisher can control the precision of the match by controlling the amount of information that advertisers have about viewers. We characterize the optimal trade-off when impressions are sold by auction. The publisher pools premium matches for advertisers (when there will be less competition on average) but gives advertisers full information about lower quality matches.
Keywords: Second Price Auction; Conflation; Targeted Advertising; Impressions; Two-Sided Private Information; Bayesian Persuasion; Information Design
JEL Codes: D44; D47; D83; D84
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| finer match (C78) | efficiency gains (D61) |
| coarser match (C78) | thicker market (D40) |
| thicker market (D40) | increased competition (L13) |
| competition (L13) | information rents for advertisers (R33) |
| control of information disclosure (D82) | efficiency of allocation (D61) |
| control of information disclosure (D82) | revenue generated (H27) |
| optimal information structure (D83) | pooling of high-value bidders (D44) |
| optimal information structure (D83) | revealing information about lower-value bidders (D44) |
| optimal information structure (D83) | competition among high-value bidders (D44) |
| competition among high-value bidders (D44) | seller's revenue (D49) |