Essays on collusion and the organization structure

This dissertation consists of three essays in microeconomics and organization theory. These essays analyze collusion problems in the organization using a stochastically repeated game when an explicit contract can not be enforceable. The first essay studies the problem of collusion in organization where the agents have an incentive to collude in a long-term relationship. Contrary to the previous literature which relies on the assumption of the enforceability of collusion side contract, we explicitly investigate the mechanism that ensures compliance to the collusion agreement. By using reputation as an implicit enforcement mechanism, we first analyze a possibility of self-enforcing collusion which requires ex-post incentive constraints to be satisfied. Defining several types of collusion, we characterize the relationship between different collusions and discuss the factors which affect the structure of collusion. The second essay analyzes the effect of collusion possibility on the structure of organization by developing a dynamic model with one principal and two agents where the collusion between the two agents is self-enforced through reputation. The stage game played by the agents has a prisoner\u27s dilemma structure, so that collusion is a dominated action. In a long-term relationship, however, the agents can collude as long as future benefits are sufficient. When monitoring technology is imperfect and the cost of replacing the agents is small, costly rotation of the agents (reducing future benefits of collusion) and payment of a wage higher than the market-clearing wage are found to be optimal for the principal. The optimal structure of organization in terms of wages and rotating schemes is also discussed. The third essay studies the perfect equilibrium in stochastically repeated game which was developed in the previous two essays to model the collusion game between the agents. We provide an example where players\u27 ex-post incentives may be different from ex-ante incentives. Defining simple stochastic strategy profile which is an extension of simple strategy profile by Abreu (1988), we show that discounted stochastic games can be completely analyzed in terms of simple stochastic strategy profiles. This implies that we don\u27t need to consider more general strategy profiles which may involve an infinity of punishments and complex history-dependent prescriptions