translates (on average) into an approximately 6% reduction in aggregate BOD discharges. For the TSS total discharges examined in 2B, the average discharge ratio declines approximately 0.018 in the year following a fine. Given the overall mean discharge ratio, this translates (on average) into an approximately 6% reduction in the aggregate TSS discharges.

Most importantly, results presented in Tables 2A and 2B are extremely similar to those found in the peer-reviewed study Shimshack and Ward (2008). Specifications for the key dependent and the key explanatory variables are identical here and in that study, so results are comparable. Shimshack and Ward (2008) found a general deterrence coefficient for BOD of -0.023. The simplified models presented here yield general deterrence coefficients for BOD of approximately -0.022. After modest rounding, all models translate (on average) into an approximately 6% reduction in BOD aggregate discharges. Similarly, Shimshack and Ward (2008) found a general deterrence coefficient for TSS of -0.024. The simplified models presented here yield general deterrence coefficients for TSS of approximately -0.018. Published results translate (on average) into an approximately 8% reduction in TSS aggregate discharges, while the simplified model results translate (on average) into an approximately 6% reduction in TSS aggregate discharges.

Results from applying the simplified models for measuring general deterrence to the BOD non-compliance status in the pulp and paper dataset are presented in Table 3A. All of the BOD models show a significant general deterrence effect of lagged enforcement actions. The coefficients on ‘fines 1-12 months ago on another plant in the state’ are negative and strongly statistically significant. Most importantly, the results are all nearly identical to those presented in Shimshack and Ward (2005). Specifications for the key dependent and the key explanatory variables are extremely similar here and in that study, so results are comparable. The relevant coefficient in Shimshack and Ward (2005) was -0.509. Here, coefficients vary between -0.487 and -0.514. All models translate (on average) into an approximately 60-65 percent reduction in the statewide probability of a BOD violation in the year following a fine.

Results from applying the simplified models for measuring general deterrence to the TSS non-compliance status in the pulp and paper dataset are presented in Table 3B. None of the TSS models show a general deterrence effect of lagged enforcement actions. The coefficients on ‘fines 1-12 months ago on another plant in the state’ are never close to being statistically meaningful. These results, however, are entirely consistent with the combined analyses of Shimshack and Ward (2005) and Shimshack and Ward (2008) published in the literature. Those studies, and especially the later paper, indicated that TSS compliance is typically indirectly determined by BOD compliance. Thus, the finding of no direct deterrence impact of fines on TSS discharges is consistent with peer- reviewed research.

underlying decision-making process has importantly changed, on average, for the regulated facilities. See the associated white paper for a more detailed discussion.

16