Grantee Research Project Results
Final Report: Identifying Methods for Improving The Effectiveness of Audit Policies and Laws
EPA Grant Number: R829686Title: Identifying Methods for Improving The Effectiveness of Audit Policies and Laws
Investigators: Stretesky, Paul B.
Institution: Colorado State University
EPA Project Officer: Hahn, Intaek
Project Period: July 27, 2002 through July 27, 2004 (Extended to August 31, 2005)
Project Amount: $139,173
RFA: Corporate Environmental Behavior: Examining the Effectiveness of Government Interventions and Voluntary Initiatives (2001) RFA Text | Recipients Lists
Research Category: Environmental Justice
Objective:
The “Incentives for Self-Policing: Discovery, Disclosure, Correction, and Prevention of Violations” (hereafter referred to as Audit Policy) was adopted by the U.S. Environmental Protection Agency (EPA) in 1995 to persuade companies to improve their environmental performance by disclosing and correcting environmental violations in exchange for the elimination or reduction of gravity-based penalties and the promise not to initiate a civil or criminal investigation of the violation. Studies of the Audit Policy suggest that it does not improve rates of environmental compliance. What has been missing from research on self-policing is a company-level study predicting Audit Policy use. Information about the potential association between traditional enforcement (i.e., inspections and enforcement actions) and Audit Policy use is essential if regulatory agencies are to more effectively and efficiently implement self-policing strategies. The objective of this research project is to determine whether the self-disclosure of environmental violations under the Audit Policy can be predicted by traditional enforcement variables. Specifically, this project examines the extent to which regulatory inspections and enforcement behavior are associated with the odds of self-disclosing a violation under the Audit Policy as opposed to having it discovered through more direct enforcement mechanisms. Determining whether such a relationship exists is important because the regulatory community generally believes that traditional enforcement efforts push firms toward self-policing.
Summary/Accomplishments (Outputs/Outcomes):
To assess the potential association between traditional enforcement mechanisms and Audit Policy use, companies that used the Audit Policy were compared to companies that did not use the Audit Policy. A case-control design was employed to make that comparison. Case-control designs typically are used to identify factors that help differentiate a population of all known cases that experience an event (the event group) from a random sample of cases that do not experience the event (the control group). In this study a case-control approach was useful because many companies violated environmental regulations during the 2-year time period under investigation, but only a handful of these companies reported their violations to EPA using the Audit Policy.
The event group was composed of 551 companies that reported an environmental violation of the Clean Air Act (CAA); Clean Water Act (CWA); Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA); Resource Conservation and Recovery Act (RCRA); Toxic Substances Control Act (TSCA); and Emergency Planning and Community Right to Know Act (EPCRA) under the Audit Policy during the 1999 and 2000 fiscal years (October 1, 1998, through September 30, 2000). The list of companies that used the Audit Policy was obtained from EPA’s Office of Regulatory Enforcement and contained all publicly releasable Audit Policy cases.
The control group was composed of 551 companies, the names of which were obtained by means of a simple random sample of all companies that EPA discovered to have violated a federal environmental law (i.e., CAA, CWA, CERCLA, RCRA, TSCA, or EPCRA) during the same 2-year time period. The sampling frame needed to select the controls was constructed from a list of all civil and administrative cases obtained from EPA’s Integrated Data for Enforcement Analysis (IDEA) system. The IDEA system was developed by EPA in response to a need for integrated data on facilities that were potentially involved in enforcement or compliance actions and is a major data-retrieval mechanism for accessing multimedia enforcement and compliance data. IDEA was accessed through the National Computing Center’s IBM mainframe.
The dependent variable, Audit Policy use, is dichotomous and indicates whether a company self-disclosed or was discovered to have violated an environmental law (i.e., CAA, CWA, CERCLA, RCRA, TSCA, or EPCRA). If a company disclosed a violation to EPA, it was assigned a score of 1. Alternatively, if a company was investigated by EPA and found to have violated an environmental law at any of its facilities, it was assigned a score of 0.
Control variables and variables representing traditional enforcement mechanisms were created using four data sources. First, IDEA was used to create variables that represent inspections and enforcement. Second, the American Business Discs (1998, 1999) were used to obtain financial and organizational information about companies in the sample. Third, the Dun and Bradstreet Gale Industry Reference Handbooks (1998) were used to create variables that reflect the financial state of a particular industry in which a company operates. Finally, EPA staff and publications were consulted to obtain a list of industries and regions that were targeted with Audit Policy letters.
The independent variables of interest indicate the presence of traditional enforcement mechanisms. Traditional enforcement mechanisms include activities by EPA and state regulatory agencies in the form of inspections, the threat of inspections, enforcement, and the threat of enforcement. The company inspections variable represents the average number of inspections at company facilities 2 years prior to the disclosure or discovery of a violation. Company inspections were calculated by summing the number of Reporting for Enforcement and Compliance Assurance Priorities (RECAP) inspections at all company facilities and then dividing by the number of company facilities eligible for inspection. The regional industry inspections variable represents the average number of inspections at similar facilities within the same industry and EPA region 2 years prior to the disclosure of a violation. Regional industry inspections were calculated by summing the number of RECAP inspections of all companies operating in the same EPA region and industry and then dividing by the number of company facilities eligible for inspection. A dichotomous variable audit letter was created and coded 1 for those companies operating in an industry that was targeted by the Audit Policy (and was therefore likely to receive an Audit Policy incentive letter in fiscal years 1999 and 2000). The letter was coded 0 for those companies not operating in such an industry and therefore not likely to receive an Audit Policy incentive letter.
Formal enforcement actions were examined through the use of two variables. The company enforcement actions variable indicates the number of formal enforcement actions by EPA against a company 2 years prior to the discovery or disclosure of an environmental violation. The regional enforcement levels variableindicates the rate of formal enforcement actions in the company’s same industry and region 2 years prior to the discovery or disclosure of environmental violations.
Control variables represent various company characteristics, the economic climate, and case characteristics. Research indicates that larger companies are more likely to participate in voluntary environmental initiatives. This same pattern may hold true with regard to use of the Audit Policy. A company size factor variable, which is the principal component index of company sales and number of company employees, was created to denote company size. Data on company size reflect the year prior to the year that the violation was discovered or self-disclosed. In addition to company size, a dichotomous public ownership variable was created to indicate whether a company was publicly owned (coded 1) or not publicly owned (coded 0).
Company financial strain has often been cited as a source of corporate violations. The American Business Discs rank companies according to their credit rating, in terms of “satisfactory,” “good,” “very good,” and “excellent.” Very few companies had credit ratings in the top or bottom categories, so the credit rating variable is dichotomous. Companies with satisfactory and good credit ratings were given a score of 1, whereas companies in the top two credit ratings were given the score of 0. A variable (percent industry sales) was created to denote company sales as a percentage of total industry sales.
Case characteristics also may be associated with a company’s environmental violation. First, a variable (multiple laws violated) was created to indicate whether a company’s actions resulted in the identification of the violation of more than one environmental law. Second, because past research indicates that Audit Policy cases are likely to be EPCRA reporting violations, a dichotomous variable (EPCRA) was created to distinguish those cases from other environmental cases. EPCRA cases were coded 1, whereas other environmental violations (i.e., CAA, CWA, CERCLA, RCRA, and TSCA) were coded 0. Under EPCRA, companies were more likely to have reporting violations than emissions violations; because these violations were less serious, companies were more likely to self-disclose. For example, the most common EPCRA violations in these data occurred when a company did not complete a form summarizing its toxic releases for EPA.
There is considerable enforcement inconsistency among the 10 EPA regional offices. Even though EPA headquarters influences environmental enforcement by issuing various policies and memorandums to guide enforcement policy, the General Accounting Office (GAO) found that regional differences in philosophies and other factors such as budget constraints resulted in significant variations in enforcement practices across similar cases. For instance, according to the GAO, regional CAA-related facility inspections ranged from a low of 27 percent to a high of 74 percent. To control for potential regional variation in enforcement practices, a dummy variable (EPA region) was used to represent the EPA region in which the company facility is located (Region 1—CT, MA, ME, RI, NH, and VT; Region 2—NJ, NY, PR, and VI; Region 3—DC, DE, MD, PA, VA, and WV; Region 4—AL, FL, GA, KY, MS, NC, SC, and TN; Region 5—IL, IN, MI, MN, OH, and WI; Region 6—AR, LA, NM, OK, and TX; Region 7—IA, KS, MO, and NE; Region 8—CO, MT, ND, SD, UT, and WY; Region 9—AZ, CA, HI, and NV; Region 10—AK, ID, OR, and WA). The reference category is Region 1.
Table 1 reports the mean values for the variables used in this analysis, broken down by Audit Policy use. As indicated in the table, several of the comparisons appear to distinguish between the cases (those companies that used the Audit Policy) and controls (those companies that violated an environmental law but did not use the Audit Policy).
Table 1. Comparison of Companies Whose Violations Were Discovered by EPA Versus Those That Self-Disclosed Using the Audit Policy
Variable |
Discovered (mean) |
Self-disclosed (mean) |
Company inspections |
.37 | .89* |
Regional industry inspections |
.78 | .93* |
Audit letter |
.07 | .31* |
Company enforcement actions |
.11 | .25* |
Regional enforcement levels |
.04 | .03 |
Company size factor |
-.53 | .56* |
Public ownership |
.05 | .19* |
Credit rating |
.49 | .24* |
% industry sales |
3.7 | 13.0* |
Number of laws violated |
1.0 | 1.2* |
EPCRA |
.08 | .58* |
EPA region |
||
Region 1 |
.05 | .05 |
Region 2 |
.04 | .08* |
Region 3 |
.11 | .11 |
Region 4 |
.08 | .10 |
Region 5 |
.05 | .18* |
Region 6 |
.31 | .12* |
Region 7 |
.11 | .08 |
Region 8 |
.09 | .03* |
Region 9 |
.10 | .04* |
Region 10 |
.03 | .05 |
Multiple regions |
.01 | .16* |
|
||
n |
551 | 551 |
*Mean difference is statistically significant (t test); p < .01 |
The numbers of both company inspections and regional industry inspections are higher for companies that used the Audit Policy than for companies that did not use the Audit Policy. Also, companies that were in an industry targeted by an audit letter were more likely to use the Audit Policy than companies operating in industries not targeted by such a letter. Finally, company enforcement actions were more frequent for companies that used the Audit Policy than for companies that did not use the Audit Policy.
In terms of organizational characteristics, both company size and public ownership appear to distinguish the cases from controls. Specifically, larger companies and companies that are publicly owned appear more likely to use the Audit Policy than smaller and nonpublic companies. Credit ratings, on average, also appear to be somewhat better among companies that use the Audit Policy than among companies that do not use the Audit Policy.
Companies that use the Audit Policy are more likely to report violating multiple environmental laws when compared to companies that do not use the Audit Policy. Companies that use the Audit Policy are more likely to report EPCRA violations than other types of violations.
Although the results presented in Table 1 illustrate the potential differences between cases and controls, they are not conclusive regarding the potential impact of traditional enforcement mechanisms on Audit Policy use, as they fail to adjust for controls. In short, traditional enforcement variables that appear to differentiate between companies that use the Audit Policy and companies that do not use the Audit Policy may prove to be irrelevant in multivariate analysis. To simultaneously adjust for competing explanations of Audit Policy use, logistic regression analysis was used. Logistic regression is an appropriate and typical statistical method for analyzing case-control designs.
To control for industry variations in the analysis, fixed-effects logistic regression was used in addition to ordinary logistic regression. The fixed-effects regression models enable the examination of the possibility that the apparent effects of independent variables reflect the effects of a group of variables correlated with the industry in which the company operates. The drawback of using fixed-effects logistic regression is that the procedure relies on variation within the matched sets. Thus, industries with no variation in the dependent variable are not informative and are discarded from the analysis. Consequently, there are a total of 1,033 cases in the fixed-effects models rather than 1,102 cases.
Table 2. Logistic Regression and Fixed-Effects (SIC Code) Logistic Regression of Audit Policy Use
Variable | Odds ratio (95% confidence interval) | |
Model 1: Ordinary LR | Model 2: Fixed-effects LR | |
Company inspections |
1.04 (0.94,1.16) | 1.06 (0.95,1.19) |
Regional industry inspections |
0.82 (0.62,1.07) | 1.22 (0.80,1.86) |
Audit letter |
3.37 (2.02,5.62)*** | 1.78 (0.77,4.13) |
Company enforcement actions |
0.82 (0.64,1.04) | 0.85 (0.63,1.14) |
Regional enforcement levels |
0.70 (0.17,2.83) | 0.31 (0.00,166.8) |
Company size factor |
2.58 (1.97,3.38)*** | 2.30 (1.67,3.17)*** |
Public ownership |
0.91 (0.47,1.77) | 0.96 (0.45,2.04) |
Credit rating |
0.71 (0.46,1.09) | 0.72 (0.46,1.16) |
% industry sales |
1.00 (0.99,1.00) | 1.00 (0.99,1.01) |
Number of laws violated |
1.27 (0.70,2.31) | 1.35 (0.71,2.60) |
EPCRA |
13.64 (8.71,21.35)*** | 12.94 (7.70,21.73)*** |
EPA region (versus Region 1) |
||
Region 2 |
2.11 (0.78,5.69) | 1.50 (0.47,4.82) |
Region 3 |
1.51 (0.60,3.81) | 1.02 (0.34,3.06) |
Region 4 |
0.81 (0.32,2.02) | 0.60 (0.21,1.69) |
Region 5 |
2.95 (1.20,7.26)* | 2.21 (0.81,6.07) |
Region 6 |
0.55 (0.24,1.24) | 0.54 (0.21,1.39) |
Region 7 |
0.57 (0.23,1.42) | 0.47 (0.17,1.34) |
Region 8 |
0.27 (0.09,0.83)* | 0.19 (0.52,0.70)* |
Region 9 |
0.16 (0.05,0.44)*** | 0.15 (0.04,0.50)** |
Region 10 |
1.42 (0.48,4.21) | 1.19 (0.36,3.97) |
Multiple regions |
5.53 (1.63,18.79)** | 2.96 (0.83,10.60) |
Constant |
.44 (0.16,1.18) | ---- |
Pseudo (Cox and Snell) R2 |
.46 | .36 |
-2 log likelihood |
835.95 | 596.62 |
n |
1,102 | 1,033 |
*p < .05; **p < .01; ***p <.001 |
After controlling for all variables thought to influence Audit Policy use, several variables that appeared to be important in the bivariate analysis are shown in Table 2 to be no longer statistically significant. Moreover, the results are relatively consistent between models 1 and 2, indicating that industry variation has little impact on the variables examined in this analysis.
What is immediately clear from Table 2 is that traditional enforcement efforts as reflected in the measurements of the variables company inspections, company enforcement actions, regional inspection levels, regional enforcement levels, and Audit Policy useare not associated with Audit Policy use. This is an important finding as it implies that increasing traditional enforcement efforts does little to improve levels of self-policing. This finding, of course, does not mean that traditional enforcement is useless, because a large number of environmental violations are discovered each year that would likely go undetected should regulatory agencies not exist. Instead, these findings suggest that objective deterrence is not an important motivator of corporate behavior. Thus, little benefit may be derived from expenditures used to improve self-policing by targeted mailings, inspections, and enforcement, and resources may be better directed elsewhere.
Also interesting is the finding that three control variables are important indicators of Audit Policy use. First, company size appears to matter. Larger companies are more likely to use the Audit Policy than smaller companies (which are more likely to have their violations discovered through traditional regulatory processes). This finding is interesting as it implies—at least to some extent—that larger companies are more environmentally responsible in their self-policing efforts than are smaller companies when it comes to finding, correcting, and reporting environmental violations. This may be because large companies have more financial resources than small companies to employ experts (such as compliance officers) who can monitor environmental performance, interpret and stay current with environmental laws, and track shifts in environmental policy and enforcement.
The importance of company size in this analysis suggests that if the Audit Policy is used to “get in good” with regulators and reduce future regulatory costs, then smaller companies are placed at an economic disadvantage (because of regulatory costs) when compared to their larger counterparts. Moreover, although EPA does have a small business initiative within the Audit Policy, it clearly does not eliminate the clear association between Audit Policy use and company size.
Second, case characteristics also appear to distinguish between the cases and controls. Specifically, companies are much more likely to use the Audit Policy to report EPCRA violations than other types of environmental violations. In fact, if the violation is an EPCRA violation it is 13.7 (8.71, 21.35; 95% confidence interval) times more likely to be reported to EPA than to be discovered through traditional regulatory processes. This finding may reflect the fact that regulatory agencies do not aggressively pursue EPCRA violations because they do not consider a company’s failure to complete a form summarizing its toxic releases to be as serious as an emissions violation that may have a direct impact on public health. Companies, however, may see potential fines resulting from discovered EPCRA violations as excessive (the average fine for an EPCRA administrative violation is $13,874). Moreover, companies also may view the disclosure of EPCRA violations as nonthreatening because there is little potential liability associated with disclosing a reporting violation as opposed to an emissions violation, for which the penalty may be more uncertain. The self-policing of an EPCRA violation by a company, then, poses little risk and may generate goodwill on the part of regulators, which may also translate into fewer future inspections and enforcement actions.
Third, one interaction was observed in these data (analysis not shown). The interaction was between the variables company size factor and audit letter. That interaction suggests that targeted letters may have some impact on increasing levels of self-policing. Specifically, smaller companies are more likely to use the Audit Policy when they operate in an industry targeted by audit letters informing them about the Audit Policy than are larger companies targeted by audit letters. This finding appears to suggest that large companies may already be familiar with the Audit Policy and know when to use it. It may be that larger companies are aware of the policy because they are more likely to have human capital in the form of auditing personnel and attorneys who can monitor changes in regulatory practices and policies. Moreover, these larger companies could also have stronger ties to regulatory agencies and therefore be aware of the benefits of the Audit Policy because of those connections. Smaller companies, in contrast, probably are less likely to have permanent employees who monitor environmental compliance and track changes in environmental policy. Thus, a small company’s receipt of an audit letter containing information about a particular Audit Policy sector initiative may be the precipitating event that initiates its use of the Audit Policy.
Journal Articles on this Report : 1 Displayed | Download in RIS Format
Other project views: | All 3 publications | 1 publications in selected types | All 1 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Stretesky PB, Gabriel J. Self-policing and the environment: predicting self-disclosure of Clean Air Act violations under the U.S. Environmental Protection Agency’s Audit Policy. Society and Natural Resources 2005;18(10):871–887. |
R829686 (Final) |
Exit |
Supplemental Keywords:
public policy, decision-making, cost-benefit, voluntary regulation, regulation, social science, pollution prevention, economic, objective deterrence, social, corporate performance, environmental compliance, corporate environmental behavior, incentives, regulations, environmental policy, environmental performance, motivators,, RFA, Scientific Discipline, Sustainable Industry/Business, Corporate Performance, Economics and Business, Social Science, Environmental Law, compliance assistance, corporate environmental policy, enforcement strategy, policy making, corporate compliance, government intervention, environmental compliance determinants, information dissemination, audit policies, enforcement impact, environmental behaviorProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.