Performance evaluation (“rating”) systems not only provide information to users but also motivate the rated worker. This article solves for the optimal (effort-maximizing) rating within the standard career concerns framework. We prove that this rating is a linear function of past observations. The rating, however, is not a Markov process, but rather the sum of two Markov processes. We show how it combines information of different types and vintages. An increase in effort may adversely affect some (but not all) future ratings.

Expert advice often extends beyond a simple recommendation, including information about alternative options. To explore the role of this referential advice, we enrich the expert’s informational advantage in a canonical model of communication with hard information. We show that when constructed just right, referential advice dissuades the decision maker from choosing options other than the recommendation, thereby making the recommendation itself more persuasive. We identify an equilibrium in which, with probability 1, the expert is strictly better off providing referential advice than she is in any equilibrium in which she provides a recommendation alone.