Committees of experts in the lab

Theory predicts that committees of experts may take decisions that look good but are bad and that they show a united front to impress evaluators. Although evaluators see through this behavior, committees persist in it only to avoid worse assessments. We investigate this theory in the lab, using treatments with and without reputation concerns and with and without cheap-talk communication with evaluators. We use the chat among committee members to learn about, e.g., their beliefs about the determinants of evaluators' assessments. We find that a committee's desire to come across as well-informed causes it to garble the information on which evaluators can base their assessments. Evaluators see through this behavior, making their assessments less dependent on actual decisions and statements. With or without reputation concerns, for the majority of committees, words speak louder than costly decisions. Evaluators pick this up. Orthogonality tests show that evaluators use observable clues about ability quite efficiently but struggle to infer ability from infrequent statements. The absence of cheap talk as a means to influence assessments hurts decision making and reduces the overall accuracy of assessments. Evidence that united fronts are consciously formed is limited.