Estimating the Effect of Penalties on Regulatory Compliance


This dissertation has two main objectives. First, we investigate the effectiveness of penalties and other enforcement tools on regulatory compliance, and comprehensively address problems that exist in previous regulatory compliance studies. Second, we develop a model that explains why most empirical studies of regulatory compliance yield results that seem to be inconsistent with the theoretical predictions of Harrington’s (1988) seminal article on regulatory compliance. Thus the dissertation comprises two essays. In Essay One, we estimate facility compliance with the Clean Water Act (CWA) by comprehensively addressing the problems that exist in previous studies. The first problem is the failure to take into account undetected violations. To address this problem, we employ Detection Controlled Estimation (DCE) model, developed by Feinstein (1990). The DCE variant that we use is the two-sided expectation simultaneity version. We use this version because we assume that potential violators will react to what the regulator would do, and vice versa. The second problem that we address is in the measurement of regulatory penalties. Previous studies use dummy variables, but using a continuous measure of penalty enables us to differentiate the responses of minor from substantial violators, and avoid measurement error. Finally, we use a richer set of covariates. We include variables that were found to be statistically and economically significant in different previous studies, but which have never been estimated jointly. The results in Essay One indicate that facilities do respond to penalties, but the effect is economically insignificant. We argue that the small effect of penalties in reducing noncompliance comes from the way regulators enforce the regulations: penalties are rarely imposed on detected violators, or if imposed, the amount is usually negligible. The policy implication that arises from our findings is that if regulators want to see a substantial increase in the probability of compliance, it should consider imposing more frequent and severe penalties. The positive effects of more stringent enforcement on compliance rates come from three sources: (1) through specific deterrence effect; (2) through general deterrence effect; and (3) through an increase in the probability of self-reported violations, which allows for more efficient use of inspection budgets. In Essay Two, we extend Harrington’s (1988) theoretical model by (1) introducing an imperfect detection parameter, and (2) relaxing the movement between the groups, as in Friesen (2003). The extended model shows that when detection is imperfect, the zone for the “always-violate” strategy expands. This expansion has two implications. First, when firms are uniformly distributed in cost space, the number of firms that choose the “always-violate” strategy increases. Second, any empirical study that uses major facilities will be more likely to confirm “always-violate” strategy, but fail to confirm the other two strategies discussed in Harrington (1988). We also discuss other possibilities that can contribute to the difference between empirical results and theoretical predictions.

For full article, click the following link:

%d bloggers like this: