Felipe Varas

Contracting Timely Delivery with Hard to Verify Quality

I study a continuous-time principal-agent model with hidden effort and imperfect monitoring about project quality. I consider a situation in which the agent can manipulate output so as to reduce the completion time of a project at the expense of quality. In order to deter manipulation the principal must make payments contingent on the subsequent performance of the project. The principal balances the provision of incentives to exert effort and the cost of deferring compensation. Unlike in previous models with pure moral hazard, the agent may never be terminated. Moreover, if the optimal contract requires termination, this must be unpredictable. If this were not the case the agent would manipulate performance when he is close to being fired. I analyze the case in which the principal can commit to a mixed strategy of termination, and also where the principal cannot commit to terminate the agent. In the latter case, the mixed strategy of termination has to be incentive compatible for the principal. In the optimal contract, cash compensation is deferred longer than in the absence of manipulation. Finally, I observe that contracts have longer duration when quality is more difficult to observe.

Dynamic Certification and Reputation for Quality

We study firms incentives to invest and build reputation for quality, when quality can be certified at a cost. We consider two types of equilibria: one in which certification decisions are made based on firms reputation and the second in which they are made based on the time since last certification. We show that reputation-based certification has a very limited effect on incentives to invest in quality, so that in equilibrium the firm invests only its reputation is the lowest. We also show that the firm in this case suffers from an over-certification trap in which the benefits of reputation are dissipated by excessive certification. These problems can be avoided with time-based certification, which can allow first-best investment in quality for sufficiently small certification cost, despite investment being unobservable. We also show that the optimal certification duration results in the firm certifying when its reputation is high.

Random Inspections and Periodic Reviews: Optimal Dynamic Monitoring

This paper studies the design of monitoring/audit policies in dynamic settings. Firm quality is private information and evolves stochastically via a Markov process with transitions depending on the firms unobservable effort. The firm benefits from having a reputation for quality. A principal designs a monitoring policy that allows him to learn the firms quality by conducting costly reviews. Monitoring plays two roles. First, it plays an incentive role, because when the principal discloses the outcome of inspections to the public, it affects the firms reputation. Second, information is directly valuable when the principals payoff is convex in reputation, for example because it allows consumers to make better choices. Our main result is a characterization of the optimal monitoring policy that induces full effort. The policy is surprisingly simple. It is either deterministic, whereby the date of the next inspection is pre-announced, or random, with a constant hazard rate of next inspection. We discuss how the type of optimal monitoring depends on history, the cost of monitoring, the persistence of quality, and the cost of effort. We also consider how the optimal monitoring policy is affected by the presence of exogenous news, showing how the evolution of the hazard of random monitoring depends on whether the absence of news is perceived to be good or bad news about quality.

Cashing-In Credibility

This paper studies a dynamic communication game in which the seller of an asset, whose credibility is unknown to the market, reports changes in the asset value during multiple periods before selling the asset. We characterize the strategy by which the seller exploits his credibility over time to maximize the assets terminal price. Though the asset value is iid across periods, and its distribution is continuous, the equilibrium has a simple Markov structure that depends on a single state variable: the sellers cumulative report. The possibility of misreporting generates reports that are serially dependent, even when the shocks are independent across periods. In equilibrium, prices are relatively insensitive to good news, particularly in the last period. The sellers cumulative report tends to grow over time, while the sellers credibility tends to decrease slowly. The seller is able to maintain inflated prices for as long as necessary. Furthermore, the terminal price explodes as the sellers horizon grows large.