Mengqi Zhang

Persuasion Through Reviewers: Implementation Observability and Commitment

We study persuasion when a sender communicates through reviewers whose private reporting types map states into signals. The sender garbles information by managing the distribution of these reviewer mappings. When the receiver observes only a coarse statistic of this distribution, the sender can secretly substitute reviewer types within an observational cell. Observability of implementation therefore becomes a commitment problem. This mapping-management perspective structurally characterizes partial commitment and its effect on persuasion. The sender can commit to an information structure only if it lies in the convex hull of the cellwise mappings selected by its own value vector. This characterization turns partial commitment into a geometric admissibility constraint on the sender’s strategies. The constraint impairs persuasion by forcing the sender to choose a suboptimal admissible strategy, or by making the standard-persuasion optimum admissible only at a discounted value. For policy design, the framework indicates which reviewer types should be observable for given objectives, including robustness, rather than treating full transparency as the only benchmark.