Project selection with biased advice: An experiment on competitive cheap talk

Abstract

When agents with private information compete for resources from an uninformed decision-maker and are biased towards their own favored projects (e.g., a CEO decides which division manager’s project to fund), they have incentive to strategically communicate about their project’s value. However, possible future interaction can mitigate this problem even without reputational concerns, since an agent who induces acceptance of a low-valued project today consumes firm resources that crowd out better opportunities that may arrive in the future. We study this organizational environment both theoretically and empirically using laboratory experiments. We hypothesize and find that truth telling is easier to support as low-quality projects lose value or become more likely to occur, but harder to support as agent competition grows. We see an interesting behavioral result in which beliefs influence responsiveness to parameter changes. Specifically, as agents grow more pessimistic about the likelihood of truthful reporting by their competitors, they respond more sharply to parameter changes, in line with the model’s predictions.