Giving Feedback to Clients

Client-agent interactions are prevalent in our daily lives. In this research, we study one such interaction through an economic game that captures the essence of clients’ and agents’ dealings with one another: the fact that agents, whose pay is frequently determined by client happiness, have the incentive to maximize client happiness by giving distorted feedback. In this game, a client undertakes a nontrivial task and her performance is disclosed to an agent. The agent must then give the client feedback to inform her of her performance. Upon receiving the agent’s feedback and before knowing her true performance, the client reports her happiness level, which in turn determines the agent’s monetary payoff. In a series of eight studies involving a total of 844 subjects, we systematically vary the way the agent’s cash earnings depend on the client’s happiness, but we always compensate the client based on her true task performance. In Study 1, we make the agent’s cash earnings proportional to the client’s happiness. We find that the agent inflates feedback, and the client reports a higher level of happiness than that reported in a control condition where the agent always provides honest feedback. As a result, the agent ends up making more money (i.e. inflated feedback pays). In Study 2, we show that the client reports the same higher level of happiness even if she reports her happiness and her preferred payment for the agent on two separate measurement scales. Studies 3 and 4 show that neither the agent nor the client behaves altruistically in their reporting. Study 5 shows that the client reports a higher level of happiness because she over-estimates her own task performance and mistakenly believes that the agent’s inflated message is genuine, leading to higher cash earnings for the agent. Study 6 shows that the agent stops inflating feedback when his cash earnings are made proportional to the client’s ex-post happiness (i.e., happiness after the client finds out her true performance), suggesting that the agent is opportunistic. Studies 7 and 8 show that the main findings extend to client-agent interactions where agent feedback is categorical (e.g., good versus bad) and where agent feedback is consequential in that it can directly affect the client’s payoff in a subsequent task. Together, the eight studies provide strong evidence that the agent behaves opportunistically and this strategy works because the client is overconfident of her own performance.