Working Paper: NBER ID: w31036
Authors: Shanglyu Deng; Hanming Fang; Qiang Fu; Zenan Wu
Abstract: Two potentially asymmetric players compete for a prize of common value, which is initially unknown, by exerting efforts. A designer has two instruments for contest design. First, she decides whether and how to disclose an informative signal of the prize value to players. Second, she sets the scoring rule of the contest, which varies the relative competitiveness of the players. We show that the optimum depends on the designer’s objective. A bilateral symmetric contest—in which information is symmetrically distributed and the scoring bias is set to offset the initial asymmetry between players—always maximizes the expected total effort. However, the optimal contest may deliberately create bilateral asymmetry—which discloses the signal privately to one player, while favoring the other in terms of the scoring rule—when the designer is concerned about the expected winner’s effort. The two instruments thus exhibit complementarity, in that the optimum can be made asymmetric in both dimensions even if the players are ex ante symmetric. Our results are qualitatively robust to (i) affiliated signals and (ii) endogenous information structure. We show that information favoritism can play a useful role in addressing affirmative action objectives.
Keywords: contest design; information favoritism; scoring bias; effort maximization; incentive provision
JEL Codes: C72; D44; D82
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| designer's choice of information disclosure (D82) | expected total effort (C13) |
| designer's choice of information disclosure (D82) | expected winner's effort (D79) |
| scoring bias (D91) | expected total effort (C13) |
| scoring bias (D91) | expected winner's effort (D79) |
| optimal contest design (D44) | expected total effort (C13) |
| optimal contest design (D44) | expected winner's effort (D79) |
| information favoritism (M51) | affirmative action objectives (J78) |
| designer's objectives (L21) | optimal contest design (D44) |
| information disclosure and scoring bias are complementary (D83) | optimal contest design can be asymmetric (C72) |