Trust, but verify? Monitoring, inspection costs, and opportunism under limited observability

Is feedback on trustworthiness necessary for the functioning of economic relationships? In many real-world economic environments, such feedback can at best be acquired through costly monitoring, raising questions of how trust and efficiency can be maintained. In the lab, we conduct a modified finite-horizon binary trust game in which we vary the observability of the trustee's actions. In the baseline condition, trustors can perfectly observe their trustee's actions. We compare this to a condition in which that actions are unobservable and three conditions in which they must be actively monitored, at zero, low or high cost, respectively. Counter-intuitively, differences in observability do not lead to significant differences in trust and efficiency, but the levels are supported by very different information structures: while trustors monitored every action under zero observation costs, most of trusting actions were “blind” – trustors did not learn whether their trust was honored or not – under costly monitoring. Even under complete unobservability almost half of the available surplus was realized. There are distributive impacts, however: the gains from trust favor trustees under the more adverse informational conditions. The behavioral patterns are consistent with the fact that trustors’ beliefs about their trustee's conduct are almost invariant across conditions.