Ranking contingent monitoring systems

This paper seeks to provide a ranking of information systems in a setting of contingent monitoring. Control strategies that make the acquisition of additional information conditional on observing certain outcomes largely elude the existing ranking criteria. We show that this happens because contingent monitoring involves more than the classical trade-off between risk sharing and incentives; it also requires a balancing of incentives and downside risk. We then develop a refinement of the most common information system orderings that conveys this feature. This allows us to reinterpret and generalize some of the literature's key results concerning, for instance, auditing policies with independent or with correlated signals and monitoring systems where the precision of an added signal is endogenous.