Incentives under collusion in a two-agent hidden-action model of a financial enterprize

This study analyzes collusion in an enterprize in which concerns about hedging cannot be ignored. In our two-agent single-task hidden-action model, where all the parties involved have exponential utility functions and the principal owning normally distributed observable and verifiable returns is restricted to o®er linear contracts, agents may exploit all feasible collusion opportunities via enforceable side contracts. Hence in general, an optimal incentive compatible and individually rational contract is not necessarily immune to collusion. We demonstrate that collusion may be ignored when making the agents work with the highest effort profile is profitable for the principal and either of the following holds: (1) mean of the return is only a®ected by the first agent's effort level, whereas variance of that is only affected by the second agent's, (2) mean is increasing and variance is decreasing separately in effort levels of both of them. On the other hand, for situations in which any of these assumptions are violated, numerical examples, showing that collusion may make the principal strictly worse off, are provided. For the justification of linear contracts as was done in the model of Holmstrom and Milgrom (1987) we consider a variant of its generalization given by Sung (1995), into which collusion possibilities are incorporated. In that continuous-time repeated agency problem including collusion, we prove the optimality of linear contracts.