Optimal Collusion-Proof Contracts with Reciprocity and Correlation

This paper studies optimal collusion-proof contracts in a dynamic model where (ⅰ) agents are willing to collude with each other by reciprocal behavior and (ⅱ) valuations for public goods are positively correlated. As a benchmark case where collusion is enforceable with side-contracts between agents, strong correlation makes collusion deterrence more costly and entails partially-pooled output schedules. In the dynamic structure, however, the degree of correlation may reduce the continuation gain from on-going collusion since the possibility of collusion diminishes. Accordingly, whether optimal outputs are partial-pooling or not depends on the conflicting interaction between the intertemporal incentive and the per-period incentives under self-enforcing collusion. As a result, the government can screen agents' valuations even with strong correlation by taking advantage of the conflicting incentives. In addition, strong correlation makes the convergence to the worst outcomes slow by mitigating the collusion incentives over time.