Credit Scoring and Incentives for Loan Officers in a Principal Agent Model

We analyze incentives for loan officers in a model with hidden action, limited liability and truth-telling constraints under the assumption that the principal has private information from an automatic scoring system. First we show that the truth-telling problem reduces the bank’s expected profit whenever the loan officer cannot only conceal bad types, but can also falsely report bad types. Second, we investigate whether the bank should reveal her private information to the agent. We show that this depends on the percentage of good loans in the population and on the signal’s informativeness. Though we had to define different regions for different parameters, we concluded that it might often be favorable to not reveal the signal. This contradicts current practice.