Giving Feedback to Clients

Abstract

We examine a prevalent form of client–agent interaction through a feedback-giving game. In this game, a client undertakes a nontrivial task and is compensated based on her task performance, which is only made known to her when the client–agent interaction ends. Meanwhile, her performance is disclosed to an agent, who must then give the client feedback on her performance. Upon receiving the feedback, the client reports her happiness level, which in turn determines the agent's payoff. In eight studies involving 928 subjects, we vary the way the agent's cash earnings depend on the client's reported happiness. When we make the agent's earnings proportional to the client's reported happiness, the agent inflates his feedback, and the client reports a higher level of happiness (than that reported in a control condition where the agent always provides honest feedback). We show that neither the agent nor the client behaves altruistically in their reporting. The client reports being happier because she overestimates her performance and mistakenly believes that the agent's feedback is genuine. The agent stops inflating his feedback when doing so no longer benefits him. Finally, our main findings are shown to be robust with respect to several factors, including making the agent's feedback consequential in affecting the client's payoff. In summary, we show that the agent behaves opportunistically, and the client's overconfidence of her own performance is what makes this strategy successful. Data, as supplemental material, are available at http://dx.doi.org/10.1287/mnsc.2013.1846 . This paper was accepted by Uri Gneezy, behavioral economics.