A multi-criteria agency model with incomplete preference information

We analyze an agency model in which the principal has only partial information on the agent's preferences. Consequently, the principal can predict the agent's reaction to an incentive system only stochastically. We first present the analytical solution for a simple decision problem, in which the agent can choose only between two alternatives. From this simple model, we derive several hypotheses on how the preference information available to the principal influences her net results. These hypotheses are then tested via a simulation model for more complex decision problems and we show that not all results of the simple model hold for the more general case. Most notably, it is not generally the case that more precise information on the agent's preferences improves the principal's situation.