Governance and Performance: Theory-Based Evidence from US Coast Guard Inspections

Given three stylized facts about the US Coast Guard (USCG), namely, soft penalties for safety violations, low incidence of penalties relative to the number of violations, and substantial resources devoted to inspections of vessels, this paper seeks (i) a theoretical lens to view USCG activities and (ii) an empirical assessment of whether those activities improve performance. Harrington’s (1988) model is motivated by these stylized facts about US regulation in general, and provides a solution via targeting of good and poor performers. The model generates hypotheses about optimal regulation in the context of pollution prevention activities of the USCG. An organization-level panel data set consisting of thousands of US flag tank barges is constructed to test those hypotheses. A count model that controls for vessel heterogeneity yields mixed evidence. If USCG inspections are considered exogenous variables (as the theory presumes), they appear to prevent pollution spills. But if inspections are endogenous and respond to previous spills then correcting for endogeneity reverses the earlier result. In addition, violations are found to be good predictors of pollution occurrences, suggesting that inspections are not as effective as they could be. Targeting as in Harrington’s model therefore appears to be incomplete, and the findings suggest that more complete targeting could increase performance. An interesting finding is that stronger penalties could increase performance.