State adoption of environmental audit initiatives.
The remainder of the article is organized as follows. Section I provides a brief background on federal and state environmental auditing initiatives, and section II presents some evidence on the results of these initiatives. Section III discusses the theoretical framework for this analysis and identifies potential factors that might drive state adoption of environmental audit initiatives; section IV describes the empirical framework and results; and section V concludes.
I. BACKGROUND ON ENVIRONMENTAL AUDITING INITIATIVES
According to a 1995 General Accounting Office (GAO) report, the potential benefits of environmental auditing to regulated entities include the detection of compliance problems before the problems can pose serious liabilities, cost savings through operating efficiencies, and reduced risks from environmental hazards. Proponents of environmental audits also argue that most voluntarily disclosed violations would not be detected by regulators and that encouraging the use of environmental audits can increase the number of violations that are detected and remediated, thus increasing environmental protection without increasing enforcement costs. Moreover, a number of theoretical studies have shown that an audit immunity or self-policing policy can actually reduce enforcement costs without decreasing environmental protection by allowing regulators to redirect their enforcement efforts (see, e.g., Innes 2001; Kaplow and Shavell 1994). However, many environmental interest groups are opposed to audit or self-policing policing policies because they believe that, regardless of the theoretical benefits of such policies, in practice they are likely to have a detrimental effect on the environment because they lower the expected financial penalty associated with violations, and thus facilities have less incentive to comply.
Starting in the 1980s, the EPA began encouraging facilities to voluntarily undertake environmental audits. In 1986, the EPA issued an Environmental Auditing Policy Statement, which recommended the use of environmental auditing and encouraged states and local governments to develop environmental auditing initiatives. The policy statement also explained when the EPA would and would not request environmental audit reports: the EPA would not routinely request audit reports, but reserved the right to request audit documents if necessary "to accomplish a statutory mission, or where the Government deems it to be material to a criminal investigation" (U.S. EPA 1986).
In 1995, the EPA issued the AUDIT POLICY that both revised the 1986 policy statement and provided incentives for facilities to voluntarily disclose and correct violations of environmental regulations. (2) The AUDIT POLICY is a self-policing policy that eliminates or reduces civil penalties for facilities that voluntarily disclose and correct violations of environmental regulations. Although the violations must be discovered during the course of an environmental audit for the facility to receive a full penalty reduction, environmental auditing is not required. Additionally, a violation must meet a number of other criteria to be eligible for a penalty reduction: The violation must be promptly disclosed; the discovery must be independent of any government actions or requirements; any harm caused by the violation must be remedied expeditiously; and the violation must not have occurred at the same facility within the last three years. Finally, the EPA reserves the right to collect a penalty equal to the economic benefit that might have been realized as a result of the self-disclosed violation.
As discussed, although the EPA encourages the use of environmental audits, it does not support statutory audit privileges or immunity. The EPA's position is that privilege laws prevent states from investigating even the most serious environmental violations and interfere with the public's right to know about existing and potential harms to human health and the environment. The EPA also believes that immunity laws prevent states from obtaining penalties that are appropriate to the seriousness of the violation. Moreover, the EPA feels that such laws are unnecessary given the federal audit policy.
However, over the past two decades a number of states have passed either environmental audit privilege or immunity legislation. In general environmental audit privilege legislation protects information gathered during the course of an environmental audit from disclosure and prohibits the use of such information in judicial or administrative proceedings. (3) Environmental audit immunity legislation grants limited immunity from civil fines and penalties to companies that voluntarily disclose and remediate violations discovered during the course of an environmental audit. Although there are some differences across states in immunity and privilege laws, in most instances differences in how the legislation is actually implemented by individual regulators is likely to outweigh any differences in the wording of the legislation. (4)
[FIGURE 1 OMITTED]
As shown in Figure 1, 19 states currently have both environmental audit privilege and immunity legislation. (5) Although Idaho adopted both privilege and immunity in 1995, the legislation had a 1997 sunset date and has not been renewed. (6) Four states have adopted privilege but not immunity, and only two have adopted immunity but not privilege. Finally, 17 states have adopted self-policing policies that provide for reduced fines when environmental violations are voluntarily disclosed. Like the federal audit policy, these self-policing policies authorize the environmental agency to provide penalty relief but not immunity for self-disclosed violations. (7) Fifteen of the 17 states only have a self-policing policy; the other 2 have self-policing policies in addition to some form of environmental audit legislation. Finally, nine states have no environmental audit legislation or self-policing policy. (8)
II. EVIDENCE ON THE EFFECTIVENESS OF THE FEDERAL AUDIT POLICY AND STATE AUDIT INITIATIVES
The EPA consistently publicizes the AUDIT POLICY as one of its successful, innovative approaches to compliance. In the introduction to the EPA's (2003) 2002 Enforcement and Compliance Assurance Report, then Assistant Administrator John Peter Suarez included the 26% increase in companies self-reporting violations as one of the highlights of the year. Although the EPA provides anecdotal evidence of the AUDIT POLICY's use and statistics on disclosures made under the policy, to date it has not provided any empirical evidence on the policy's effectiveness in terms of compliance behavior or environmental performance. (9) State environmental audit initiatives have also been praised as a "key flexible enforcement approach" (U.S. GAO 1998). However, the claims of successful state audit initiatives also appear to be based primarily on the number of facilities that register their intent to audit or disclose violations rather than on an empirical analysis of changes in environmental performance due to the initiative. For example, the Texas Natural Resource Conservation Commission (the state environmental agency) reported that the results of its environmental audit legislation were "promising" based on the number of facilities that filed notices of audit and disclosed violations discovered during audits after its inception (see Weaver et al. 1997). Similarly, Michigan's Department of Environmental Quality (2001) provided a positive assessment of its audit program based on these same measures.
One statistical analysis of audit policy disclosures has been conducted by outside researchers. Pfaff and Sanchirico (2004) examined all cases filed in the Audit Policy Docket from 1994 to 1999 and compare the profile of disclosed violations to all violations in terms of the statutes violated, types of violations, and average fines. The authors find that the typical disclosed violation from these cases is relatively insignificant and question whether the audit policy is as effective as the EPA claims given this result. However, as Pfaff and Sanchirico note, one possible explanation could be that regulated entities are increasing the use of audits (and presumably environmental performance) but are choosing not to disclose any discovered violations. Thus it is important to consider the effect of the audit policy on compliance as well as disclosures. Stafford (2005) examines compliance with hazardous waste regulations before and after the establishment of the AUDIT POLICY. (10) The analysis controls for changes in both enforcement intensity and targeting but does not find any significant evidence that the federal Audit Policy has affected overall compliance. However, the results of the analysis do suggest that both state audit legislation and state self-policing policies decrease violations of hazardous waste regulations. (11)
Another potential measure of environmental performance is a facility's willingness to move "beyond compliance" through participation in voluntary programs. One of the most well-known of these programs is International Organization for Standardization (ISO) 14001, which requires facilities to develop a comprehensive environmental management system that is externally audited. In a recent study, Potoski and Prakash (2004) find that the percentage of facilities that participate in the ISO 14001 program is higher in states with at least one compliance incentive program (i.e., an environmental leadership program, state audit privilege, or state audit immunity) than in states without. Because ISO 14001 requires facilities to establish environmental audit protocols, it is not surprising that facilities in states with audit initiatives might be more likely to enroll in the program.
III. POTENTIAL FACTORS DRIVING STATE ADOPTION OF ENVIRONMENTAL AUDIT INITIATIVES
Over the past decade, a number of studies have examined state adoption of environmental policies (see Bacot and Dawes 1997 for an overview). There are essentially three competing explanations for the variance in state environmental policies: political context, environmental conditions, and institutional capacity. (12) Although each of these explanations has received some empirical support, according to Potoski and Woods (2002) the ability of each to explain state variation in policies depends on the nature of the environmental policy in question. If the policy allocates resources among competing groups, such as the establishment of pollution standards, the political context will be a very important factor in its adoption. If, on the other hand, the policy is focused on implementation, such as an enforcement program, one might expect the most important factor to be the environmental conditions or scope of the environmental problems in the state. Finally, if the policy is focused on collecting and processing information, the most important factor should be institutional capacity. For example, a pollution monitoring policy requires the ability to collect information as well as a demand for the information in the first place. Both of these are likely to be present only in states with strong institutions.
Obviously environmental audit legislation and self-policing programs are concerned with the implementation of environmental programs, and thus one might expect environmental conditions to be a driving factor in their adoption. Additionally, these initiatives could have significant resource allocation effects because audit immunity and self-policing policies can save regulated entities significant civil penalties and audit privilege will decrease potential liability from future lawsuits. Thus one would expect the political context to be an important factor as well. However, because self-policing policies can be implemented by a state environmental agency without legislation and provide less regulatory relief than audit legislation, one might expect the political context to be less important for self-policing. Because information collection and processing is neither a goal nor a by-product of any of the audit initiatives, in general one would not expect institutional capacity to be a significant factor in a state's decision to adopt. Nevertheless, to the extent that audit immunity and self-policing programs may help reduce enforcement costs, one might expect states with low institutional capacities or tight environmental budgets to adopt them as cost-saving measures. Finally, there is one additional dimension that could play an important role in a state's decision to adopt audit legislation or a self-policing policy. Due to the EPA's opposition to audit legislation, states with strong ties to the EPA should be less likely to adopt such legislation, although one might not expect a similar effect for self-policing policies.
Table 1 summarizes the hypothesized importance of these explanatory factors in explaining state adoption of audit legislation and self-policing policies. Next the article will describe the variables used to measure political context, environmental conditions, and institutional capacity and provide the assessment of the expected relationship between each variable and the adoption decisions. These expectations are also summarized in Table 1.
A. Political Context
Political context can be captured by a number of different measures. First, the economic importance of pollution-generating industries in a state will affect the ability of such industries to effectively lobby state legislatures for policies that they find beneficial. Clearly both audit privilege and audit immunity legislation are beneficial for businesses that are subject to environmental regulations, and thus the author expects a positive relationship for privilege and immunity legislation. The author also expects a positive--although quantitatively less significant--relationship for the self-policing policy because although such policies do not guarantee penalty reduction or privilege audit reports, they do provide some regulatory relief. For this analysis, the author measures the overall importance of pollution-generating industries using the percentage of total sales in a state generated by mining and manufacturing firms. (13) Additionally, because Pfaff and Sanchirico (2004) find that the majority of disclosures under the audit policy are violations of the Toxic Release Inventory (TRI) reporting requirements or violations of hazardous waste regulations, the analysis captures the relative strength of these specific interest groups using two variables: the percentage of establishments in a state that are subject to TRI reporting and the percentage of establishments that are large quantity hazardous waste generators (LQGs). (14)
The author also measures political context by including the percentage of voters in the 2000 U.S. presidential race that voted for the Republican candidate, George W. Bush. (15) Given that Bush's position can be considered more pro-business that that of his opponent, this variable should also pick up states with probusiness attitudes, and thus the author expects a positive relationship between this variable and state adoption. Legislatures may also be lobbied by interest groups or individuals for stricter environmental controls. To measure the strength of the environmental lobby in a state, the author uses the number of people in a state out of 1,000 that are members of national conservation groups, such as the Sierra Club. (16) The author expects this variable to have a negative impact on the adoption of privilege and immunity legislation, as environmental groups have been quite vocal in their opposition to such laws. (17) However, because the EPA's audit policy has not been the focus of significant opposition from these groups, the author does not anticipate a significant effect on the adoption of self-policing policies.
B. Environmental Conditions
As discussed earlier, states are expected to adopt environmental policies that are consistent with the scope of the environmental problem in a state. In states with severe environmental problems or high levels of pollution, traditional enforcement programs are more likely to be successful than voluntary or flexible approaches, such as auditing or self-policing. Moreover, audit immunity and self-policing policies generally do not apply to severe or repeat violations. Thus one might expect states with poor environmental conditions to be less likely to adopt audit legislation or self-policing policies. Additionally, in states with higher levels of pollution per capita, one might expect citizens to put pressure on legislators to adopt more strict environmental regulations. To capture environmental conditions in a state, the author includes three pollution measures: pounds per capita of TRI chemicals released to air, surface water, and land. (18) The author expects that the higher the pollution level, the less likely a state will be to adopt any of these initiatives.
C. Institutional Capacity
The primary measure of institutional capacity is Lester's (1994) classification of states based on their commitment to environmental protection and their ability to carry out strong environmental programs. Progressives are states with "a high commitment to environmental protection coupled with strong institutional capabilities," and strugglers are states "with a strong commitment to environmental protection but with limited institutional capacities." Delayers are states with "a strong institutional capacity but with a limited commitment to environmental protection," and regressives are states with a weak institutional capacity "as well as a limited commitment to environmental protection." (19) Given their commitment to environmental protection and the EPA's stated opposition to audit legislation, the author expects progressives to be less likely to adopt audit legislation than delayers and regressives. However, given that audit legislation may result in decreased enforcement costs, the author also expects strugglers and regressives to be more likely to adopt it than delayers. Because the self-policing policy has been touted by the EPA as more protective of the environment than audit legislation, the author does not expect environmental commitment to be a factor in its adoption. However, because it also can result in decreased enforcement costs, the author might expect strugglers and regressives to be more likely to adopt it than progressives and delayers. Note that for this analysis, the second category, delayer, is the omitted category and thus the category to which others are compared.
Another measure of institutional capacity is the financial resources available for environmental programs. (20) Because audit initiatives can decrease enforcement costs, the author expects states with smaller budgets to be more likely to adopt them. An indirect measure of institutional capacity is whether a state has a voluntary pollution prevention program. (21) The primary focus of most voluntary pollution prevention programs is using facility education and guidance to increase environmental performance rather than mandatory requirements. Lyon and Maxwell (2003) suggest that such public voluntary programs (as opposed to negotiated voluntary agreements between regulators and specific industry groups) may be a sign that stricter regulation is politically infeasible. Thus one might expect states with voluntary pollution prevention programs to have weak environmental agencies (relative to the legislature) and thus more likely to adopt audit legislation. However, the author would not anticipate a significant effect on the adoption of self-policing policies.
D. State-Federal Relationship
Measuring the relationship between states and the federal EPA is not easy. One possible measure is the degree to which the political context of the state conforms to the political ideology of the federal administration. For most of the period over which states adopted audit legislation, 1993 to 2002, the Clinton administration was in place. Thus the measure of the percent of voters in a state that voted republican in the 2000 presidential election will also pick up, to some extent, the relationship between the state and federal government. The higher the percentage voting Republican, the more likely the state is to adopt audit legislation. However, the EPA does not oppose states adopting self-policing policies modeled on the federal policy and thus the state-federal relationship should not affect self-policing policies.
One method the EPA has used to persuade states to not adopt audit legislation is by threatening to revoke state-level enforcement authority if a state adopts it. (22) Because states must be authorized for each regulatory program separately, there is no one measure of state authorization. Thus the author includes a variable measuring the percentage of the key hazardous waste rules that each state has been authorized to implement as a proxy for all environmental programs. (23) The author expects states with a higher level of authorization to be less likely to adopt audit legislation, but she does not expect any effect for self-policing.
In addition to granting or revoking state authority to implement federal environmental regulations, the EPA has other means by which it may persuade a state to adopt or not adopt legislation and policies, such as awarding grants-in-aid. Thus the analysis includes a variable indicating whether the state has a performance partnership agreement (PPA) with the EPA. The goal of the performance partnership program is joint federal-state planning and priority setting. States can develop a PPA with the EPA that identifies "the goals and objectives for environmental protection in the state, the strategies that will be employed in meeting them, the roles and responsibilities of the state and EPA in carrying out the strategies, and the measures that will be used to assess progress." (24)
E. Other Controls
The author includes two additional variables in the analysis, population and area, to control for differences in states that may indirectly impact a state's desire and ability to adopt new environmental policies, but do not have a direct impact or fall into any of the categories described above. These variables are also included in many of the other analyses of state environmental policies. However, the effect of such variables on state adoption is not consistent, and therefore the author has no expectations as to the effect of these variables on the adoption of audit initiatives.
IV. EMPIRICAL ANALYSIS AND RESULTS
As shown in Figure 1, 40 states have adopted either audit legislation or a self-policing policy. Note that the first state to adopt legislation was Oregon in 1993. The following year, five additional states added some type of audit or self-policing program. By 1996, 28 states had some type of program, and by 1998 the total had risen to 38. Given the theoretical framework outlined, the author conducted a reduced-form analysis of the factors that affect whether a state has adopted each of the three possible audit initiatives: privilege legislation, immunity legislation, or a self-policing policy. (25)
A. Cross-Section Analysis
The adoption of the three audit initiatives depends on the benefit to each state of doing so. As already discussed, the (unobservable) net benefit of adopting an audit initiative will be a function of the political context of the state (p), the environmental conditions of the state (e), the institutional capacity of the state (i), the state-federal relationship (f), and other factors (o). Let the net benefits be represented as [y*.sub.i]:
[y*.sub.i] = [p.sub.i][[beta].sub.1] + [e.sub.i][[beta].sub.2] + [i.sub.i][[beta].sub.3] + [f.sub.i][[beta].sub.4] + [o.sub.i][[beta].sub.5] + [[epsilon].sub.i].
If the benefits are greater than 0, that is if [y*.sub.i] > 0, the observable binary variable [y.sub.i] takes on the value of 1 and 0 otherwise. Assuming that the errors are normally and independently distributed, the author can estimate this model with a simple probit regression. (26) The results are of these regressions are presented in Table 2.
In the privilege regression, the coefficients on all of the political context variables except for Percent Mining and Manufacturing Sales are significant. However, the coefficient on Percent Conservation Members is positive, the opposite of what was expected. The hypothesis was that states with a high percentage of conservation members would be less likely to adopt privilege legislation due to the strength of the environmental lobby. However, it could be the case that in states with active environmental groups, regulated entities are more concerned that such groups would try to subpoena environmental audit documents for use in civil litigation and thus have a stronger desire for audit privilege legislation. (27)
Although none of the coefficients on the environmental condition variables in the privilege regression are significant at conventional levels, Per Capita Water Releases and Per Capita Land Releases are jointly significant at the 95% level. (28) Thus as expected, higher levels of pollution do decrease the likelihood that audit privilege will be adopted. With respect to the institutional capacity variables, the coefficient on Progressive is significant and positive as expected. Although the coefficients on Struggler and Regressive are not individually significant, they are both negative as expected and they are jointly significant at the 90% level. Also the coefficient on Voluntary Pollution Prevention is positive as expected and significant. (29) Of the two variables measuring the state-federal relationship, Performance Partnership Agreement has the expected negative sign and is significant, indicating that states with strong ties to the federal EPA are less likely to adopt privilege legislation. (30)
In the immunity regression, Percent Voting Republican is the only political context variable that has a significant coefficient, but it is positive as expected. None of the environmental condition or institutional capacity variables has a significant coefficient. However, one of the state-federal relationship variables, Percent Hazardous Waste Program Authorized, does have a significant negative coefficient, as expected. Also, note that in most cases the coefficients on the explanatory variables have the same sign in both the privilege and immunity regressions, which one would expect given that the two types of legislation are often adopted together (although they need not be).
Interestingly, in the self-policing regression only two coefficients are significant, the coefficient on Percent LQGs and the coefficient on Land Area. For both of these variables, the sign of the coefficient is the opposite of the sign of the (significant) coefficient in the privilege regression. As discussed earlier, there is nothing that precludes a state from adopting both a self-policing policy and audit privilege and immunity legislation, although only two states (Oregon and Minnesota) have done so. Thus it appears that to some extent states view audit legislation and self-policing policies as alternatives to each other.
To get a sense of the magnitude of the effect of the revised penalty policy on compliance behavior and test the hypotheses about the relative importance of these variables, the author calculated the probability of adopting environmental audit legislation and self-policing policies for a hypothetical state that has the median values for the explanatory variables. The estimated probability of adoption is 27% for privilege legislation, 36% for immunity legislation, and 17% for self-policing policies. It is interesting to note that the actual percentages of states adopting these programs are 46%, 42%, and 34%, respectively. To better understand the effect the explanatory variables have on the likelihood of adoption, Table 3 reports the change in the probability (in percentage points) of this hypothetical state adopting the legislation or policy that would result from various changes in the explanatory variables. For example, increasing Percent Voting Republican in the hypothetical state by 1 SD from a median of 50.6% to 59.51% increases the likelihood of audit privilege adoption by 76 percentage points to 100% and the likelihood of audit immunity adoption by 40 percentage points to 76%.
For the privilege legislation note that all of the political context variables have a quantitatively large effect on the likelihood of adoption, particularly compared to most of the institutional capacity variables. For both audit immunity and self-policing policies, the political context variables and the institutional capacity variables are closer in terms of quantitative effect, which is consistent with institutional capacity being relatively more important for these two initiatives. The state-federal relationship variables also have a reasonably large effect on the adoption decision for audit privilege and immunity as expected. However, although the author hypothesized that environmental conditions would be very important in the adoption decision for all three initiatives, the quantitative effects of these variables are small relative to the other types of variables.
B. Duration Analysis
Because the author believes that some of the lack of significance in the cross-section probit might be due to the small number of observations, she also analyzes a duration model of time to adoption to see if those results can provide more insight into the factors driving state adoption. Additionally, a duration analysis allows one to examine the temporal dimension of policy adoption that is not captured by the cross-section analysis. The hazard rate, h(t), is the likelihood that a state adopts audit legislation or a self-policing policy at time t, given that the state has not already adopted the program in question. The author used a Weibull proportional hazard model where the hazard rate function is defined as:
h(t,x(t)) = [e.sup.x[beta]][alpha][t.sup.[alpha]-1]
where x is the vector of explanatory variables and [alpha] is a shape parameter to be estimated from the data. (31) This model assumes that the time until adoption is independent across states. One benefit of using the Weibull model is that the log of the time to adoption, T, can be written as:
ln(T) = x[beta]* + [epsilon],
where [beta]* = -[beta]/[alpha] and [epsilon] is independent of x. (32) The [beta]* coefficients then can be interpreted directly as the percentage change in the time to adoption for a one unit change in the explanatory variable. (33)
The results of the duration analysis are presented in Table 4, and the estimated effects of changes in explanatory variables on time to adoption for the hypothetical state are presented in Table 5. (34) For each initiative, the author observes all 50 states for 10 years, 1993 to 2002. (35) States are included in the analysis in each year up to and including the year the program in question is adopted, so the number of observations varies across programs and is reported in Table 4 as number of observations. This model includes both time-invariant and time-varying explanatory variables. Some time-invariant variables are actually constant over the period of analysis such as the Progressive, Struggler, and Regressive classification and Land Area. For others such as Conservation Members and Per Capita Environmental Spending, the author does not have a panel of data and thus uses observations from one point in time. The variables that vary across time are noted with a subscript t. Where endogeneity is a potential concern, the author lagged the variables by one year and have indicated this with the subscript t-1. Note that all of the variables included in the cross-section probit are included as explanatory variables in this analysis with one exception: Percent Voting Republican has been replaced with Percent of Lower House that is Republican, because the author felt that the latter variable would be a more consistent measure than the former, which would depend significantly on the candidates in the race, whether one was an incumbent, and so on. (36)
In a number of cases, the results of the duration analysis support that of the cross-section analysis. For example, the coefficient on Percent LQGs is negative and significant in the privilege regression, indicating that states with higher percentages of large quantity hazardous waste generators adopt audit privilege more quickly. This is consistent with the positive and significant coefficient in the cross-section analysis. Similarly, although the coefficient on Percent LQGs is not significant in the self-policing duration regression, its positive sign is consistent with the negative coefficient in the self-policing cross-section. Additionally, the results for Percent Lower House Republican in the duration regressions are consistent with the results for Percent Voting Republican in the cross-section analyses.
For a quite a few variables, the results of the duration analysis are stronger than the results of the cross-section. For example, in the privilege regression, the coefficient on Percent Mining and Manufacturing Sales is negative and significant, confirming the expectation that states with more significant polluting industries will be quicker to adopt audit privilege legislation. Also note that for both the privilege and immunity regressions, the coefficient on Per Capita Water Releases is positive and significant, which is consistent with expectations that states with higher pollution levels would be less likely to adopt this legislation. It is also interesting to note that the effect of water pollution levels on time to adoption is much more significant than the effect of any of the political context variables, although in the cross-section analysis, political context is more important.
In the self-policing regression the coefficient on Per Capita Air Releases is negative and significant, which is not consistent with expectations. It could be that states with higher pollution levels adopt self-policing policies as a substitute for audit legislation because self-policing policies are marginally more protective of the environment. This explanation is consistent with the negative and significant coefficient on Percent Conservation Members in the self-policing regression. If some type of audit initiative is seen as inevitable, states with a high number of conservation members may be pressured to adopt self-policing policies sooner in an attempt to reduce the need for audit legislation.
In the duration analyses, the only institutional capacity variable that has any significant effect on time to adoption is Voluntary Pollution Prevention Program. The negative and significant coefficient on this variable in the privilege regression is consistent with results from the cross-section analysis. Moreover, the coefficient in the immunity regression is also negative and significant. This is in line with the expectations that a voluntary pollution prevention program may be a sign that the state environmental agency is weak relative to the legislature, and thus the state will be more easily able to adopt audit legislation. The insignificant coefficient on Voluntary Pollution Prevention Program in the self-policing regression is also consistent with the expectation that the weakness of the agency would not affect the adoption of self-policing policies.
In the privilege regression the coefficient on Percent Hazardous Waste Program Authorized is negative and significant, indicating that states that are authorized to implement a larger percentage of the hazardous waste program take less time to adopt audit privilege legislation. The author hypothesized that states with a larger percentage of their program authorized would have a closer relationship with the EPA and thus would be less likely to adopt audit legislation, and the cross-section results for audit immunity appear to confirm this. However, one also might expect that if a state is inclined to adopt audit legislation, states that were early adopters of other environmental programs might also be early adopters of audit legislation.
Finally, note that the coefficients on Performance Partnership Agreement are positive and significant in both the privilege and self-policing regressions. Although the author expected this relationship for audit privilege, she hypothesized that the state-federal relationship would not have a significant effect on the adoption of a self-policing policy modeled on the EPA's program. Note, however, that the effect of performance partnership agreement is quantitatively higher for audit privilege (an increase of almost six years) than for self-policing (an increase of slightly more than one year).
This article examines the factors that affect whether a state has adopted three environmental audit initiatives: audit privilege legislation, audit immunity legislation, or a self-policing policy. Because such initiatives are primarily concerned with the implementation of a state's environmental programs, the author expected environmental conditions to be a driving factor in their adoption. Additionally, because the initiatives could have significant resource allocation effects, the author also anticipated that political context would be an important factor in the adoption decision. Finally, due to the EPA's opposition to audit legislation, the author expected that the state-federal relationship would affect adoption of audit legislation although she did not expect a similar effect for self-policing policies.
In general the hypotheses are supported by both cross-section and duration analyses of state adoption, although there are several interesting exceptions. Political context is a key factor in the adoption decision for both privilege and immunity as states with more regulated entities and states that are highly Republican are more likely to adopt these two initiatives. However, contrary to expectations, the analysis also finds that states with a higher percentage of environmental group members are more likely to adopt audit privilege legislation. One possible explanation for this result could be that in these states regulated entities are very concerned that such groups might seek to use environmental audit documents against them. A second unexpected result is that states with a higher percentage of environmental group members are quicker to adopt self-policing policies. If some type of audit initiative is seen as inevitable, it could be the case that active environmental lobbies pressure state to adopt self-policing policies as an alternative to audit legislation.
Although environmental conditions are less important than expected in the cross-section results, the duration analyses show that such conditions do significantly affect the timing of adoption decisions for privilege and immunity. However, for self-policing higher levels of air pollution cause states to adopt more quickly, a result the author did not expect. It could be that states with higher pollution levels adopt self-policing policies as a substitute for audit legislation because self-policing policies are marginally more protective of the environment. As expected, institutional capacity is generally not a significant factor in the adoption decision. However, states with voluntary pollution prevention programs, one of the measures of institutional capacity, are more likely to adopt audit legislation and adopt it more quickly than states without such programs. This result is consistent Lyon and Maxwell's (2004) theory that voluntary pollution prevention programs are a sign that the state environmental agency is weak relative to the legislature, and thus the state will be more easily able to adopt audit legislation.
Finally, the state-federal relationship does appear to play an important role in state adoption of audit legislation. States with PPAs with the EPA are less likely to adopt privilege legislation, and states that are authorized to implement a higher percentage of the hazardous waste program are less likely to adopt immunity legislation. Interestingly, although states that have PPAs adopt privilege legislation more slowly, states that are authorized to implement a higher percentage of the hazardous waste program adopt privilege legislation more quickly. It could be that if a state is inclined to adopt audit legislation, states that were early adopters of other environmental programs will also be early adopters of audit legislation.
Overall, the results for privilege and immunity legislation are very similar, although the results are generally a little stronger for audit privilege. This is not too surprising given that the two types of legislation are often adopted together, although they need not be. On the other hand, many explanatory factors affect the adoption of audit legislation and self-policing policies in opposite ways. Although there is nothing that precludes a state from adopting both a self-policing policy and audit legislation, only two states have done so. Thus it appears that to some extent states view audit legislation and self-policing policies as alternatives to each other, with self-policing policies seen as more protective of the environment and audit legislation as more beneficial to the regulated community.
Although there is limited empirical evidence on the effect of state audit initiatives on environmental performance, there have been numerous claims that such initiatives can have a significant positive impact. The results of this analysis suggest that when conducting a more formal evaluation of audit initiatives, researchers should take into account those factors that affect both the underlying decision to adopt an initiative and the performance of that initiative such as environmental conditions and institutional capacity. Finally, if state audit legislation is shown to have a positive impact on environmental performance, for example by increasing participation in ISO 14001, these results also suggest that the EPA may want to reconsider its opposition to such legislation as that appears to have a significant effect on states' adoption decisions.
Bacot, A. H., and R. A. Dawes. "State Expenditures and Policy Outcomes in Environmental Program Management." Policy Studies Journal, 25, 1997, 355-70.
Dahl, R. "Audit-Privilege Laws: The Right to Know Nothing." Environmental Health Perspectives, 107, 1999, 510-13.
Frey, B. C., and K. A. Johnson. "Environmental Auditing since EPA's 1986 Audit Policy." Working Paper, 2000; available online at www.epa.gov/region5/orc/articles/env-audits.htm.
Hall, B., and M. L. Kerr. 1991-1992 Green Index: A State-by-State Guide to the Nation's Environmental Health. Washington, D.C.: Island Press, 1991.
Hays, S. P., M. Esler, and C. E. Hays. "Environmental Commitment among the States: Integrating Alternative Approaches to State Environmental Policy." Publius, 26, 1996, 41-58.
Innes, R. "Violator Avoidance Activities and Self-Reporting in Optimal Law Enforcement." Journal of Law, Economics, and Organization, 17, 2001, 239-56.
Jordanger, D. J., and C. R. Graham. "Protecting Privilege, Recognizing the Risks of Criminal Liability, and Reaping the Benefits of EPA's Audit Policy." Virginia Lawyer, April 2001, 12-19.
Kaplow, L., and S. Shavell. "Optimal Law Enforcement with Self-Reporting of Behavior." Journal of Political Economy, 102, 1994, 583-606.
Lee, J. A., and B. C. Frey. "Environmental Audit Immunity Laws: A State by State Comparison." BNA Environment Reporter, June 13, 1997.
LeSage, J. P. "The Theory and Practice of Spatial Econometrics." Unpublished manuscript, 1999.
Lester, J. P. "A New Federalism? Environmental Policy in the States," in Environmental Policy in the 1990s, edited by N. J. Vig and M. E. Kraft, Washington, D.C.: CQ Press, 1994.
Lyon, T. P., and J. W. Maxwell. "Self-Regulation, Taxation, and Public Voluntary Environmental Agreements." Journal of Public Economics, 87, 2003, 1453-86.
Michigan Department of Environmental Quality. "Environmental Audit Privilege and Immunity Five Year Report: 1996-2001." March 31, 2001.
Mishra, B. K., D. P. Newman, and C. H. Stinson. "Environmental Regulations and Incentives for Compliance Audits." Journal of Accounting and Public Policy, 16, 1997, 187-214.
Pfaff, A., and C. W. Sanchirico. "Big Field, Small Potatoes: An Empirical Assessment of EPA's Self-Audit Policy." Journal of Policy Analysis and Management 23, 2004, 415-32.
Potoski, M., and A. Prakash. "The Regulation Dilemma: Cooperation and Conflict in Environmental Governance," Public Administration Review, 64, 2004, 152-63.
Potoski, M., and N. D. Woods. "Dimensions of State Environmental Policies: Air Pollution Regulation in the United States," Policy Studies Journal, 30, 2002, 208-26.
Stafford, S. L. "Does Self-Policing Help the Environment? EPA's Audit Policy and Hazardous Waste Compliance." Vermont Journal of Environmental Law, 2005.
U.S. EPA (Environmental Protection Agency). "Environmental Auditing Policy Statement--Final Policy Statement." Federal Register, 51, 1986, 25004.
______. "Incentives for Self-Policing: Discovery, Disclosure, Correction, and Prevention of Violations--Proposed Revisions and Request for Public Comment." Federal Register, 64, 1999, 26745-756.
______. "Incentives for Self-Policing: Discovery, Disclosure, Correction and Prevention of Violations--Final Policy Statement," Federal Register, 65, 2000, 19617-626.
______. "Fiscal Year 2002 Enforcement and Compliance Assurance Report." EPA 300-N-03-002, 2003.
U.S. GAO (General Accounting Office). "Environmental Auditing: A Useful Tool that Can Improve Environmental Performance and Reduce Costs." GAO/RCED-95-37, 1995.
______. "EPA and States' Efforts to Focus State Enforcement Efforts on Results." GAO/RECD-98-113, 1998.
Weaver, J. M., R. J. Martineau, and M. K. Stagg. "State Environmental Audit Laws Advance Goal of a Cleaner Environment." Natural Resources and Environment, 11, 1997, 6-13.
Wooldridge, J. M. Econometric Analysis of Cross Section and Panel Data. Cambridge, Mass.: MIT Press, 2002.
SARAH L. STAFFORD*
*The author thanks Julio Videras and an anonymous referee for their valuable comments and suggestions and gratefully acknowledges the financial support of Resources for the Future and EPA STAR Grant R831036.
Stafford: The College of William & Mary, P.O. Box 8795, Williamsburg, VA 23187. Phone 1-757-221-1317, Fax 757-221-1175, E-mail firstname.lastname@example.org
EPA: Environmental Protection Agency
GAO: General Accounting Office
ISO: International Organization for Standardization
LQG: Large Quantity Generator
PPA: Performance Partnership Agreement
RCRA: Resource Conservation and Recovery Act
TRI: Toxic Release Inventory
1. A number of public commenters on EPA's audit policy expressed concern that audit reports would be used for compliance witch hunts (U.S. EPA 1986). Legal practitioners have also cautioned regulated entities about conducting environmental audits (see Jordanger and Graham 2001). It should be noted that the audit policy has also been criticized as being too lenient (see, for example, Dahl 1999).
2. The AUDIT POLICY was issued on December 22, 1995 (60 Federal Register 66705), and took effect on January 22, 1996. The EPA issued minor revisions to the policy on April 11, 2000 (65 Federal Register 19617).
3. Without such laws, neither attorney-client privilege nor the self-evaluative privilege protect the factual material disclosed in an environmental audit report. See Frey and Johnson (2000).
4. Lee and Frey (1997) discuss the differences in state legislation.
5. Information on the adoption of audit privilege, audit immunity, and self-policing policies provided by U.S. EPA Region 5, Office of Regional Council (www.epa.gov/region5/orc/audits/audit_apil.htm). Although Mississippi's legislation does not grant environmental audit immunity per se, the legislation provides de facto immunity.
6. Arizona adopted both immunity and privilege, but it never became effective because it was conditioned on an appropriations bill to fund the program that was never passed.
7. Like the AUDIT POLICY, these policies do not necessarily require that the violation be discovered during the course of an environmental audit. Additionally, the agency is not required by law to provide penalty relief.
8. Maine and New York have self-policing policies that apply only to small businesses.
9. In 1999, the EPA completed an audit policy evaluation based on a voluntary survey of companies that disclosed environmental violations under the policy, but the analysis is limited to descriptive statistics about the users of the policy (U.S. EPA 1999).
10. Because one cannot directly observe compliance, the analysis uses data on detected hazardous waste violations and EPA enforcement actions to determine if there has been an underlying change in the behavior of regulated entities.
11. Whether a state has environmental audit privilege, immunity, or a self-policing policy is included in the analysis as three separate explanatory variables.
12. This categorization is due to Potoski and Woods (2002). Hays et al. (1996) and Bacot and Dawes (1997) provide similar categorizations, although both of these articles break political context and institutional capacity into several different subcategories.
13. Data on total revenues and revenues by industry were taken from the 1997 Census of Enterprise Statistics.
14. Because the EPA does not collect data on revenues or employment statistics for TRI reporters or hazardous waste generators, it is not possible to look at the percentage of revenues or employment generated by these groups. Data on total number of establishments were taken from the 1997 Census of Enterprise Statistics. Data on the number of facilities subject to TRI reporting was taken from the 1997 TRI State Fact Sheets. Data on the number of large quantity generators was taken from the 1997 Biennial Reporting System National Report.
15. These data were taken from the 2001 Book of the States compiled by the Council of State Governments.
16. Environmental membership was taken from the 1991-92 Green Index, Hall and Kerr (1991).
17. In fact, environmental groups in five states that adopted privilege laws petitioned the EPA to revoke delegation of federal environmental programs to those states (U.S. GAO 1998).
18. To address the concern that pollution might be endogenous, the analysis uses TRI releases for 1995. Data on TRI releases were downloaded from the EPA's Web site, www.epa.gov/triexplorer.
19. Progressives include California, Massachusetts, and New Jersey; strugglers include Colorado, Maine, and North Carolina; delayers include Georgia, Illinois, and Texas; and regressives include Arizona, Mississippi, and Nebraska.
20. The per capita environmental spending measure is based on 1996 expenditures for all environmental programs and is taken from the Council of State Government's Resource Guide to State Environmental Management, 5th ed.
21. Data on pollution prevention programs was obtained from the National Pollution Prevention Roundtable, Washington, D.C. To address concerns that this variable might be endogenous, the author used data on whether the state had a program in 1997.
22. For example, in 1998 following a petition by the Michigan Environmental Council, the EPA reviewed Michigan's environmental audit law and advised the state that it would revoke Michigan's delegated authority to administer federal environmental programs. In response, Michigan amended its environmental audit legislation. See www.michiganinbrief.org/edition06/text/issues/issue-26.htm for more details. Mishra et al. (1997) provide additional examples.
23. The author tried other proxies, such as whether the state is authorized to implement the National Pollution Discharge Elimination System program but none performed as well as hazardous waste (Resource Conservation and Recovery Act, RCRA) authorization and due to the small numbers of observations, it was not possible to include other measures. The effective date of authorization for RCRA rules was obtained from the EPA's State Authorization Tracking System's Key Rule Checklist. To address concerns that this variable may be endogenous, the author used data on the percentage of key rules for which the state was authorized in 1997.
24. See www.epa.gov/ocirpage/nepps/pp_agreements.htm for more information on PPAs. To address concerns that this variable may be endogenous, the author used data on whether the state had a PPA in 1997, the first year for which data is available.
25. Due to the limited number of observations, it is not possible to estimate the adoption decisions jointly.
26. Of course, because these are state observations, this may not be the case. In particular, one might be concerned that the error terms could be correlated across neighboring states. Thus, the author also conducted a spatial autoregressive error probit model (see LeSage 1999 for an in-depth discussion of this model). The results of the spatial probit analysis are generally similar to the results of the simple probit analysis. Additionally, the coefficients on the spatial errors are not significant, so the author has chosen not to report the results of that analysis here. (Results of the spatial probit analysis are available on request.)
27. The author would like to test this hypothesis more directly by including an interaction between conservation group participation and the strength of regulated entities. Because there are only 50 observations, the model will not converge if this interaction is included in the regression. However, if one replaces Percent TRI Reporters with a Percent Conservation Members* Percent TRI Reporters interaction, the coefficient on the interaction term is positive and significant and the coefficient on Conservation Members alone is no longer significant.
28. Joint significance was tested using a likelihood ratio test.
29. To address concerns that this variable may be endogenous, the author conducted a Rivers-Vuong test and could not reject the null hypothesis that this variable is exogenous. Wooldridge (2002) discusses the use of the Rivers-Vuong test of endogeneity in probit models (see p. 474).
30. To address endogeneity concerns, the author used data on whether the state had a PPA in 1997, the first year for which data is available. However, by 1997 over half of the states had some type of audit program in place, so there could still be a problem with endogeneity. Therefore the author conducted a Rivers-Vuong test and could not reject the null hypothesis that this variable is exogenous.
31. See Wooldridge (2002) for a complete discussion of duration models in general and the Weibull model in particular.
32. In addition, [epsilon] has an extreme value distribution scaled by 1/[alpha].
33. The transformation to ln(T) is only strictly correct if the explanatory variables are constant over time. However, if one assumes that the change in the explanatory variable would hold over the entire period of analysis, the author can use this interpretation.
34. To evaluate the size of the effect of the explanatory variables on the time to adoption, the author multiplies the [beta]* coefficients by the SD of the explanatory variables to get the percentage change in time to adoption. The author can then evaluate the absolute change in years using the estimated constant years to adoption shown in Table 4.
35. Because all but two adoptions take place before 2000 and there was a significant change in economic trends around 2000, the author also estimated the duration model for the period 1993 to 1999. The results for the shorter time period are not qualitatively different from the results for the longer period.
36. Data for the Percent of Lower House that is Republican comes from various editions of the Book of the States compiled by the Council of State Governments. Data are available for even years only and are interpolated for odd years.
TABLE 1 Expected Importance and Effect of Variables on Adoption Decision Self- Audit Privilege Audit Immunity Policing Legislation Legislation Policy Political context High High Medium Percent mining and + + + manufacturing sales Percent TRI reporters + + + Percent LQGs + + + Percent voting Republican + + + Percent conservation members - - None Environmental conditions High High High Per capita air releases - - - Per capita water releases - - - Per capita land releases - - - Institutional capacity Low Medium Medium Progressive - - None Struggler + + + Regressive + + + Environmental budget per - - - capita Voluntary pollution + + None prevention program State-federal relationship High High Medium Percent hazardous waste - - None program authorized Performance partnership - - None agreement TABLE 2 Probit Regression of Factors Affecting Adoption Environmental Audit Privilege Explanatory Variables Legislation Percent mining and manufacturing sales -0.17 (0.15) Percent TRI reporters 24.54* (12.66) Percent LQGs 81.77* (45.92) Percent voting Republican 1.24* (0.67) Percent conservation members 1.97* (1.12) Per capita air releases 0.28 (0.32) Per capita water releases -1.11 (0.71) Per capita land releases -0.38 (0.23) Progressive -5.86* (3.50) Struggler 0.25 (2.16) Regressive 5.82 (4.26) Environmental budget per capita -0.03 (0.03) Voluntary pollution prevention program 15.18* (8.43) Percent hazardous waste program authorized 2.08 (2.93) Performance partnership agreement -9.45* (4.97) Population (millions) -0.21 (0.17) Land area (10 million acres) 3.88* (2.20) Constant -123.49* (67.52) Pseudo-[R.sup.2] 0.69 Environmental Audit Immunity Explanatory Variables Legislation Percent mining and manufacturing sales -0.04 (0.06) Percent TRI reporters 1.71 (3.06) Percent LQGs 5.81 (4.08) Percent voting Republican 0.12* (0.07) Percent conservation members -0.01 (0.15) Per capita air releases -0.02 (0.06) Per capita water releases -0.47 (0.33) Per capita land releases -0.08 (0.05) Progressive 0.26 (0.89) Struggler 0.56 (0.87) Regressive 1.07 (0.78) Environmental budget per capita -0.0003 (0.01) Voluntary pollution prevention program 0.67 (0.65) Percent hazardous waste program authorized -4.22** (1.91) Performance partnership agreement -0.83 (0.66) Population (millions) -0.10 (0.08) Land area (10 million acres) 0.44* (0.25) Constant -6.23* (3.51) Pseudo-[R.sup.2] 0.38 Self-Policing Explanatory Variables Policy Percent mining and manufacturing sales 0.03 (0.07) Percent TRI reporters 0.65 (2.84) Percent LQGs -7.69** (3.69) Percent voting Republican -0.09 (0.07) Percent conservation members 0.20 (0.15) Per capita air releases 0.01 (0.06) Per capita water releases 0.03 (0.28) Per capita land releases 0.01 (0.11) Progressive 0.16 (0.84) Struggler -0.17 (0.82) Regressive 0.76 (0.91) Environmental budget per capita -0.01 (0.01) Voluntary pollution prevention program -0.88 (0.77) Percent hazardous waste program authorized 2.17 (1.51) Performance partnership agreement 0.12 (0.74) Population (millions) 0.06 (0.06) Land area (10 million acres) -0.29* (0.19) Constant 4.24 (3.94) Pseudo-[R.sup.2] 0.33 Notes: SEs in parentheses. *Significant at the 90% level. **Significant at the 95% level. TABLE 3 Effect of Change in Variables on Probability of Adoption Change in Probability of Adoption (in Percentage Points) Privilege Explanatory Variable Median SD Legislation Baseline probability of adoption 27% Increase % mining & mfg sales by 1 25.20 8.54 -25% SD Increase % TRI reporters by 1 SD 0.29 0.17 73%# Increase % LQGs by 1 SD 0.27 0.11 73%# Increase % voting Republican by 1 50.60 8.71 73%# SD Increase % conservation members by 8.45 3.58 73%# 1 SD Increase per capita air releases 5.07 5.66 56% by 1 SD Increase per capita water releases 0.51 1.47 -26% by 1 SD Increase per capita land releases 0.34 7.76 -27%# by 1 SD Progressive 0 0.40 -27% Struggler 0 0.46 9% Regressive 0 0.40 73% Increase environmental budget per 46.05 55.95 -26% capita by 1 SD No voluntary pollution prevention 1 0.49 -27%# program Increase % haz waste program 0.44 0.27 21% authorized by 1 SD No performance partnership 1 0.50 73%# agreement Increase population by 1 SD 3.79 5.85 -24% Increase land area by 1 SD 3.49 5.50 73%# Change in Probability of Adoption (in Percentage Points) Immunity Self-Policing Explanatory Variable Legislation Policy Baseline probability of adoption 36% 17% Increase % mining & mfg sales by 1 -10% 8% SD Increase % TRI reporters by 1 SD 10% 4% Increase % LQGs by 1 SD 25% -20%# Increase % voting Republican by 1 40%# -19% SD Increase % conservation members by -1% 28% 1 SD Increase per capita air releases -4% 2% by 1 SD Increase per capita water releases -17% 2% by 1 SD Increase per capita land releases -16% 2% by 1 SD Progressive 9% 6% Struggler 21% -5% Regressive 41% 29% Increase environmental budget per -1% -17% capita by 1 SD No voluntary pollution prevention -17% 34% program Increase % haz waste program -23%# 21% authorized by 1 SD No performance partnership 32% -4% agreement Increase population by 1 SD -15% 12% Increase land area by 1 SD 70%# -25%# Note: Significant changes in boldface. Note: Significant changes indicated with #. TABLE 4 Results of Hazard Model of Time to Adoption Environmental Environmental Audit Privilege Audit Immunity Explanatory Variables Legislation Legislation Number of observations 342 360 Percent mining and -0.03* (0.02) -0.02 (0.02) manufacturing sales Percent TRI reporters -0.62 (0.80) 0.35 (1.02) Percent LQGs -1.92* (1.02) -1.55 (1.09) Percent Lower House -0.03** (0.01) -0.02** (0.01) Republican[.sub.t] Percent conservation members 0.03 (0.03) 0.03 (0.04) Per capita air -0.01 (0.01) -0.01 (0.01) releases[.sub.t-1] Per capita water 0.88** (0.24) 0.50** (0.23) releases[.sub.t-1] Per capita wand 0.01 (0.01) 0.01 (0.01) releases[.sub.t-1] Progressive 0.27 (0.29) 0.24 (0.40) Struggler -0.13 (0.26) -0.30 (0.31) Regressive 0.06 (0.24) -0.05 (0.28) Environmental budget per -0.0001 (0.002) 0.001 (0.003) capita Voluntary pollution -0.70** (0.17) -0.42** (0.21) prevention program[.sub.t] Percent hazardous waste -0.52* (0.31) -0.06 (0.37) program authorized[.sub.t-1] Performance partnership 1.23** (0.33) 6.43 (103.2) agreement[.sub.t-1] Population 0.03 (0.03) 0.01 (0.03) (millions)[.sub.t] Land area (10 million acres) -0.04 (0.03) -0.03 (0.03) Constant 4.66** 0.77 3.78** (0.82) [chi] 3.53 3.10 p-value of [chi square] for <0.01 <0.01 regression Self-Policing Explanatory Variables Policy Number of observations 396 Percent mining and 0.01 (0.02) manufacturing sales Percent TRI reporters -0.65 (1.37) Percent LQGs 2.63 (1.61) Percent Lower House 0.01 (0.01) Republican[.sub.t] Percent conservation members -0.13** (0.05) Per capita air -0.04* (0.02) releases[.sub.t-1] Per capita water 0.09 (0.19) releases[.sub.t-1] Per capita wand 0.01 (0.03) releases[.sub.t-1] Progressive -0.24 (0.37) Struggler -0.25 (0.43) Regressive -0.49 (0.46) Environmental budget per 0.01 (0.01) capita Voluntary pollution 0.20 (0.26) prevention program[.sub.t] Percent hazardous waste -0.32 (0.50) program authorized[.sub.t-1] Performance partnership 0.50* (0.29) agreement[.sub.t-1] Population -0.04 (0.03) (millions)[.sub.t] Land area (10 million acres) 0.07 (0.08) Constant 2.32** (0.58) [chi] 2.57 p-value of [chi square] for 0.04 regression Notes: SEs in parentheses. *Significant at the 90% level. **Significant at the 95% level. TABLE 5 Effect of Change in Variables on Time to Adoption Change in Time to Adoption (in Years) Privilege Explanatory Variable Median SD Legislation Increase % Mining & mfg sales 25.26 7.98 -1.08# by 1 SD Increase % TRI reporters by 1 0.29 0.16 -0.47 SD Increase % LQGs by 1 SD 0.27 0.11 -1.00# Increase % Lower House 48.00 15.82 -1.86# Republican[.sub.t] by 1 SD Increase % conservation 8.45 3.55 0.54 members by 1 SD Increase per capita air 4.95 6.08 -0.22 releases[.sub.t-1] by 1 SD Increase per capita water 0.43 3.17 12.98# releases[.sub.t-1] by 1 SD Increase per capita land 0.37 7.27 0.24 releases[.sub.t-1] by 1 SD Progressive 0 0.40 1.28 Struggler 0 0.46 -0.63 Regressive 0 0.40 0.28 Increase environmental budget 46.05 55.45 -0.03 per capita by 1 SD No voluntary pollution 1 0.48 3.24# prevention program[.sub.t] Increase % hazardous waste 0.44 0.28 -0.67# program authorized[.sub.t-1] by 1 SD Performance partnership 0 0.48 5.72# agreement[.sub.t-1] Increase population 3.82 5.92 0.71 (millions)[.sub.t] by 1 SD Increase land area (10 3.49 5.45 -0.92 million acres) by 1 SD Change in Time to Adoption (in Years) Immunity Self-Policing Explanatory Variable Legislation Policy Increase % Mining & mfg sales -0.46 0.11 by 1 SD Increase % TRI reporters by 1 0.21 -0.24 SD Increase % LQGs by 1 SD -0.66 0.68 Increase % Lower House -1.46# 0.48 Republican[.sub.t] by 1 SD Increase % conservation 0.40 -1.07# members by 1 SD Increase per capita air -0.20 -0.52# releases[.sub.t-1] by 1 SD Increase per capita water 6.02# 0.63 releases[.sub.t-1] by 1 SD Increase per capita land 0.30 0.18 releases[.sub.t-1] by 1 SD Progressive 0.92 -0.56 Struggler -1.13 -0.57 Regressive -0.20 -1.13 Increase environmental budget 0.27 0.81 per capita by 1 SD No voluntary pollution 1.60# -0.47 prevention program[.sub.t] Increase % hazardous waste -0.07 -0.20 program authorized[.sub.t-1] by 1 SD Performance partnership 24.30 1.15# agreement[.sub.t-1] Increase population 0.31 -0.53 (millions)[.sub.t] by 1 SD Increase land area (10 -0.67 0.83 million acres) by 1 SD Note: Significant changes in boldface. Note: Significant changes indicated with #.
|Printer friendly Cite/link Email Feedback|
|Author:||Stafford, Sarah L.|
|Publication:||Contemporary Economic Policy|
|Date:||Jan 1, 2006|
|Previous Article:||Factors affecting species' risk of extinction: an empirical analysis of ESA and NatureServe listings.|
|Next Article:||Political economy and the efficiency of compensation for takings.|