Organizations spend considerable resources in monitoring activities. The usefulness of the monitoring function requires limiting the possibility of side-deals between the agent and the monitor. The conditions for such implicit collusion are that the monitor should expect future rents from a continued relationship and that an agent should have the authority to influence these rents and a credible threat to do so. Many instances of collusion between agents do not involve explicit bribes; instead, long-term relationships and dynamic concerns are the real threat to the independence of monitors in general and of auditors in particular. In Discussion Paper No. 1027, Research Affiliate Daron Acemoglu offers a theory of dynamic collusion that fits these accounts, thus allowing for the examination of a number of questions that a static model cannot address.

The paper shows that the credibility of the agent's threat depends on how the principal will interpret a disagreement between the agent and the monitor, which in turn depends on whether the principal expects collusion to take place. First, the extent of collusion will depend on the monopoly power (rents) of the monitors. Second, in contrast to a static model with explicit side-contracts, using the threat of lawsuits combined with a flat fee for the monitor rather than bonuses may be the optimal method of controlling the monitor. Third, the behaviour of the current monitor will depend on the expected behaviour of other monitors and thus a multiplicity of equilibria is possible. Fourth, the number of tasks that the monitor has may influence the possibility of collusion. Last, the model can be used to determine jointly the allocation of authority and the reliability of information flows within the organization.

A Dynamic Model of Collusion
Daron Acemoglu

Discussion Paper No. 1027, September 1994 (IO)