Recent evidence suggests that corruption in the form of a “tip off” is a significant problem in many regulatory agencies. For example, the New York Department of Transportation conducts surprise inspections of school bus companies to test the road-worthiness of their buses. School-bus inspectors at the department of transportation in New York, however, were recently known to accept bribes from firms in exchange for informing them of an upcoming (surprise) inspection. Firms were then able hide those buses that were faulty in order to avoid being penalized.
In this paper we develop a Principal-Supervisor-Agent model to study the regulator’s optimal inspection regime when corruption can weaken the power of surprise inspections. In our model a welfare maximizing principal (regulator) must hire supervisors in order to inspect firms (agents). In addition to choosing the penalties for the firms and the supervisor’s wage, regulators choose the frequency of inspections and the inspection regime, that is, whether the inspections will be announced or surprise. Given this regulatory environment, the firm and the supervisor then play a simultaneous move game in which the firm chooses its level of concealment, while the supervisor chooses the level of inspection effort. Following Malik (1990), the supervisor’s probability of detecting a firm’s non-compliance is increasing in the supervisor’s inspection effort and decreasing in the firm’s level of concealment activity.[1] Conditional on finding evidence of non-compliance, firms are fined and supervisors receive some fraction of those fines as a reward.
Our key finding is that in an unannounced inspection regime with corrupt inspectors (and fixed penalty and supervisor wage) the regulator can lower the frequency with which inspections are conducted in order to eliminate corruption. Interestingly, we also show that despite the fact that inspections become less frequent overall compliance and welfare may be raised. We discuss these results in light of Lazear (2006) who examines the conditions under which announced inspections are preferred to surprise inspections.[2] In contrast to our paper, his paper does not allow for corruption, or for costly inspections. We show that his results are sensitive to the presence of these two factors.