The Effectiveness of Voluntary Environmental Programs-A Policy at a Crossroads?

Introduction

In the United States, voluntary environmental programs (VEPs) have become standard in the environmental policy tool kit since the 1990s. This development is at the confluence of several events, including more complex regulations, technical innovation and scientific discoveries, cuts in regulatory budgets (Brouhle, Griffiths, & Wolverton, 2005), and increased concern with the costs imposed by environmental regulations. In 1981, President Reagan issued Executive Order 12291 (amended with Executive Order 12866), which called for regulatory impact analysis to weigh the potential benefits to society of a regulation with its potential costs. The emphasis on cost raised the profile of environmental economics both within and outside the Environmental Protection Agency (EPA) and in the 1990s helped broaden support for economic instruments, such as emissions trading markets, in lieu of traditional command and control regulation that was increasingly considered less efficient and more costly. In parallel, the international business community made a concerted effort to be a proactive participant in debates on environmental protection with the formation of the World Business Council for Sustainable Development at the 1992 Rio Earth Summit. Business leaders furthermore argued that industry-lead environmental change is both viable and necessary (Schmidheiny, 1992), if not financially strategic (Schmidheiny & Zorraquin, 1996). In 1996, the ISO 14001 environmental management system (EMS) standard was issued unleashing a global wave of voluntary certified management systems targeted broadly at improving corporate environmental management.

Today, in 2007, many fundamental questions remain about the effectiveness of VEPs, despite their popularity. It is not clear what types of voluntary programs can yield an incremental improvement in performance compared with a status quo defined by a set of regulations and market conditions, and if so why this should be. Do VEPs achieve environmental and public health results commensurate with the resources invested by firms and government? Most research has focused on the decision to participate in a VEP targeting production process changes and is starting to yield a profile of the type of firm most likely to participate. Far less research has focused on the actual impact on the environment and public health. With the exception of theoretical work (Lyon & Maxwell, 2004), very little is known about the interplay between market and regulatory pressures on adoption and action under a VEP. For example, if regulation is anticipated, does firm resistance increase or decrease after joining the associated VEP? Even less is known about the actual effectiveness of various program design elements intended to motivate participation and action. As a result, expectations of VEPs may not be properly aligned with the decision-making reality at the firm (or in government) to permit appropriate conclusions on their effectiveness. Should we merely concede that VEPs at best "prepare" firms to address and acknowledge an issue and open the door to greenwashing?

This article presents a summary of research to date on VEPs in the United States. Much of the research has been funded through the EPA's Office of Research and Development Science to Achieve Results program, which, since 1997, has helped create a community of scholars with increasingly more sophisticated skills in program (effectiveness) evaluation. After a brief description of VEPs and their standing in the United States, the remainder of this article focuses on research that evaluates U.S. VEPs. I draw from empirical research on non-U.S. VEPs where it complements and ameliorates our understanding of how VEPs in the United States (might) work. It should be emphasized that the empirical research summarized here focuses on VEPs that target pollution abatement via changes in industrial production processes and not changes to product design, as in various eco-labeling schemes. It is important to note at the outset that industrial production process VEPs probably make up a relatively small share of total U.S. or non-U.S. VEPs given the large number of VEPs targeting areas such as energy efficiency, transportation, food and agriculture, and indoor air quality.

Alberini and Segerson (2002) point out that VEPs can be evaluated from several angles: participation, pollution abatement levels, and the impact on polluting firms and competition. I start with effectiveness determinations based on participation. Two basic mechanisms are thought to motivate participation: (i) positive incentives (e.g., cost-sharing and subsidies) and (ii) the threat of legislation (Segerson & Miceli, 1998). Most of the theoretical work has focused on the interplay between the threat of regulation and the firm's desire to avert that threat by entering into a VEP. Alternatively, theorists posit that VEPs can be useful when the threat of regulation is weak (Lyon & Maxwell, 2003). The next section presents this theoretical work.

While firms appear willing to participate in a VEP, empirical evidence on increased pollution abatement levels beyond business as usual for industrial production process VEPs is scarce. I find somewhat surprisingly that none of the VEPs targeted at production processes initiated by the U.S. government that have been studied show evidence of significant pollution abatement above business as usual, even those programs with specified targets. VEPs managed by the private sector fare no better, with the exception of the ISO 14001 standard, which appears to be mildly successful. In light of what appears to be widespread greenwashing, I turn to empirical research on corporate compliance behavior to understand how firms respond to regimes with stricter sanctions. I suggest that conclusions on the overall effectiveness of VEPs as a policy tool be postponed until we have additional analysis covering the complete spectrum of VEPs.1 I conclude with implications for the policymaker and suggestions for further research.

Definition of VEPs

The most common taxonomy of VEPs, alternatively called voluntary environmental agreements (VEAs), is based on which party(s) initiated and designed the programs. The EU Research Network outlined three widely accepted categories: (i) unilateral industry-led; (ii) public voluntary VEPs initiated by government where industry is invited to participate; and (iii) negotiated agreements between industry and government2 VEPs are, in many instances, designed to enhance the efficacy and scope of existing regulations. Alternatively, they are used to address issues, which do not fall within the current regulatory framework. For example, in the United States, a significant number of VEPs address global warming.

Compared with traditional policy tools, such as command-and-control regulation, VEPs are thought to encourage proactive industry, reduce transactions costs, and accelerate achievement of environmental targets because of less legal action and conflict (Brouhle et al., 2005). VEPs, like taxes and cap-and-trade policies, offer industry greater flexibility and can more readily accommodate specific situations. Participants can benefit from technical assistance, information subsidies, and public recognition. Public VEPs often involve a nonbinding Memorandum of Understanding, under which participants commit to environmental goals, but are not subject to a penalty for withdrawal or nonperformance. Unilateral industry VEPs, on the other hand, have used the threat of dismissal as a sanction (Brouhle et al., 2005; Mazurek, 2002).

The U.S. EPA Commitment to VEPs

According to the 2006 report on EPA partnership programs (PPs) by the EPA's Office of Inspector General there are 54 VEPs currently managed by EPA at the federal level.3 Climate change is targeted by 18 of these programs. According to their designers, EPA PPs attempt to use a market-based approach to motivate organizations and individuals to engage in environmental protection. These programs do not specifically require external parties to take action by law and EPA retains primary responsibility for program design. Overall, individual program budgets are small, with median values at $492,500. Most EPA PPs offer technical assistance and information, PR or the opportunity for partnership with regulatory agencies (both federal and state), while a few offer regulatory flexibility. Some EPA PPs leverage the influence of federal purchasing power and other large, institutional buyers to achieve results. Other EPA PPs in partnership with banks and other financial institutions help companies and other organizations finance cost-saving pollution prevention measures. Participants range from industry to state and local government and the nonprofit sector and are often required to help with recruitment and marketing. While data collection continues to be a challenge, the majority of programs (35) require reporting, and the data are used by the EPA to change programs.

EPA Administrator Steve Johnson has publicly stated support for voluntary EMS in 2005. The Position Statement calls for widespread use of EMSs across a range of organizations as a means to achieve improved environmental performance, compliance, and pollution prevention. Specifically, the EPA will encourage organizations that use EMSs to demonstrate accountability for performance outcomes through measurable objectives and targets. Additionally, the Agency will encourage organizations to publicly disclose information on the performance of their EMSs. Finally, the EPA will encourage the use of recognized environmental management frameworks, such as the ISO 14001 Standard. Issued in 1996, this ISO standard describes a set of EMS and practices, including the development of environmental objectives and policies, the provision of training and documentation, delegation of responsibilities, and internal performance audits (Delmas, 2002).4 Clearly, there is strong support in regulatory and industry circles for VEPs targeting production process changes, although proponents still rely primarily on anecdotal evidence, rather than the increasingly critical empirical research.

Why Do Firms Participate in a VEP?

Theory of Regulatory Threats and Incentives

In the absence of a clear signal from markets, the regulator needs to create a situation where firms can gain from VEP participation. The problem is that pollution abatement requires some form of resource investment, and so a firm must perceive a net gain from participation (Alberini & Segerson, 2002). The most widely proposed mechanism is the threat of regulation or taxation that prompts industry to either form its own VEP (Lyon & Maxwell, 2003) or (if an industry VEP fails5) join a public VEP (Alberini & Segerson, 2002; Segerson & Miceli, 1998,1999). A firm is more likely to join a VEP if the associated costs are lower compared with the anticipated cost of compliance with (current or expected) government mandates or other schemes, such as ecotaxes (Segerson & Miceli, 1998). The regulator can increase the appeal of VEP participation by providing sufficient incentives, such as the threat of regulation, financial subsidies, the provision of services, and cost-sharing schemes.

Each of these has implications for the regulator. On the one hand, the regulator needs to generate the funds necessary to extend financial subsidies to firms participating in a VEP. On the other hand, the threat of regulation needs to be credible enough to induce participation. An important benefit of a credible regulatory threat is that it can reduce the need for government subsidies to induce participation (Khanna, 2001). Segerson and Miceli (1998) theorize that a VEP can be reached even when the regulatory threat is weak, however, the level of abatement achieved will be lower than what could be achieved with mandatory standards. Such a minimal amount of voluntary abatement is in the firm's economic self-interest, as long as it serves to preempt regulation (Khanna, 2001). In circumstances where there is low political will to impose mandatory controls and it is not costly (both politically and financially) to provide subsidies, a regulator is more likely to design a VEP with subsidies to attract participants (Lyon & Maxwell, 2003). If, on the other hand, the probability of regulation is high enough, the first best pollution abatement outcome (possible under mandatory standards) can occur with a VEP (Segerson & Miceli, 1999). In this circumstance, a complementary, and perhaps necessary, policy is for the regulator to implicitly promise not to regulate if a VEP can adequately achieve the desired environmental goals. These models suggest that the level of abatement depends on the probability of regulation and a cost/benefit analysis for both the regulator and the regulated.

Alternatively, when the regulation(s) is already in place, the regulator is conceivably in a stronger position and can extend exemptions to VEP participants. This approach, however, requires potentially expensive monitoring systems (Alberini & Segerson, 2002) and an appropriate regulatory setting. Regulatory relief has been a popular approach in Europe, but faces legal challenge in the United States because Congress has not extended this authority to the EPA. Thus, the EPA relies upon legally nonbinding agreements that do not require congressional approval and cannot rely on tough sanctions to punish defecting firms (Delmas & Terlaak, 2001, 2002; Mazurek, 2002).6 Without this bargaining position, the EPA is constrained to offering a more limited set of incentives such as technical assistance, analytical tools, product differentiation, brand enhancement, and public recognition.

An important concern with VEPs is the potential for participating and nonparticipating firms to "free-ride" and gain the benefits from a VEP without investing resources in pollution abatement For example, when a VEP is created to address an industry-wide regulatory threat and only some firms actively participate, the entire industry is thought to benefit from preemption of regulation because regulationrelated costs are avoided (Dawson & Segerson, 2003). A subset of firms will therefore still opt to participate in a VEP despite the free-rider problem. However, it is the nonparticipants who gain more.

Business Strategy

As an alternative to the aforementioned regulator-industry model of VEPs, researchers theorize that market forces can shape firm environmental behavior. For example, firms may be more inclined to participate in a VEP if they perceive a shift in demand and supply toward more green products. Market pressures can arise not only from consumers and suppliers, but also from customers, competitors, trade associations, community groups, and investors (Henriques & Sardorsky, 1996). Participating in VEPs is one manner in which a firm can demonstrate its environmental consciousness to these various audiences. At another level, VEPs help firms to signal reduced production costs via improved environmental management and reduced financial risk through better management of environmental risks and liabilities (Reinhardt, 1999).

Yet another avenue of research is informed by neoinstitutional sociologists who claim that compliance can be achieved through informal mechanisms, including shaming and public exposure (Braithwaite, 1989), and the emergence of new values and norms that guide and alter members' preferences for collective action (Hoffman, 1997). The desire for social legitimacy therefore also influences management choices and practices (Meyer & Rowan, 1977). In the absence of a clear regulatory threat, corporate managers might still perceive a net gain from participation associated with personal satisfaction from being an environmental steward (Alberini & Segerson, 2002). An analogous debate has been between those who suggest that industry self-regulation will work only if explicit sanctions are included to prevent free-riding (Grief, 1997) and those who argue that sanctions are not necessary. In lieu of sanctions, researchers have theorized that the institutional structure of industry self-regulation7 may control behavior through informal means of coercion, the transfer of norms, and the diffusion of best practices (King & Lenox, 2000; Nash & Ehrenfeld, 1997).

Evidence

The majority of research on U.S. VEPs that target production processes has focused on the decision to participate and has explored most of the above-mentioned factors. Most articles use a discrete choice model to evaluate the decision to participate (or not) and evaluate the relative impact of various factors using regression analysis. Factors, such as regulatory pressure, environmental performance, market pressures, competitive pressures, and financial performance that have empirically been shown to influence the decision to participate are listed in Table 1. Table 1 also lists the VEPs where these factors have been found to influence participation and associated paper(s), the majority of which have been published.

What is striking from a perusal of Table 1 is the consistency of influential factors across analyses of several VEPs, including those initiated by the EPA and Department of Energy (DOE) (e.g., 33/50, Green Lights, WasteWi$e, and Climate Challenge) and those managed by industry (e.g., ISO 14001 EMS, Responsible Care [RC], Responsible Distribution Processes, Sustainable Slopes, Sustainable Forestry Initiative, and Encouraging Environmental Excellence). It should be noted that this perceived consistency is also a function of data availability to evaluate environmental performance and collective learning in the research community of which factors appear to influence participation and need to be included in an analysis. Thus, for example, a significant number of analyses have used the Toxic Release Inventory (TRI) database because it is a longitudinal database (starting in 1987) of toxic chemicals released from a broad swath of manufacturing facilities. As a result, 33/50, an EPA-initiated VEP to reduce toxic chemical emissions, has been analyzed by six different author teams (Arora & Cason, 1995, 1996; Gamper-Rabindran, 2006; Khanna & Damon, 1999; Sam & Innes, 2005; Videras & Alberini, 2000) and the TRI is the most common measure of environmental performance.

If we consider which factors appear statistically significant despite differences across studies in sample, sampling period, and econometric analysis, the research to date shows that a combination of market and nonmarket forces influences the decision to participate in a VEP, as has been posited in theoretical work. The evidence, based upon the number of studies finding the same statistically significant results, indicates that VEP participants tend to be large, more polluting firms, which face more regulatory pressure or future environmental liability, that are image and brand conscious in highly competitive industries, are more innovative global market players, and have engaged in one or more voluntary environmental management activities, such as Green Lights or ISO 14001.

More specifically, high levels of pollution (usually measured with TRI data) induce participation in multiple EPA PPs (33/50, Green Lights, and WasteWi$e) and industry VEPs (RC, Sustainable Forestry Initiative, and ISO 14001). Similarly, higher levels of CO2 emissions per unit output are a motivator for participation in the Climate Challenge Program (Karamanos, 1999). From the perspective of EPA as a designer of VEPs, the goal of attracting more highly polluting firms (or industries) appears successful. It bears mentioning that EPA has strategically targeted high polluters, as in the 33/50 program where the first wave of invitations went out to the highest six hundred polluting firms who, in total, generated 66 percent of 33/50 chemicals in 1988 (Khanna, 2007).

What is somewhat surprising is that high-level polluters, who bear a greater cost of pollution abatement, are more likely to participate in an industry-initiated VEP, such as ISO 14001 or RC (Khanna & Anton, 2002; King & Lenox, 2000). This implies that participants join not to signal higher levels of performance already achieved, but rather it would appear that the decision to participate is motivated by the desire to improve performance, at a minimum; create the appearance that such improvements are taking place or in the making; or preempt future regulation that is likely to be more costly to heavy polluters. From a game theoretic standpoint, a firm may be motivated by the desire to build leverage with the regulator and thereby influence future regulatory actions and legislation. The private sector tends to view VEPs as a manner to promote regulatory goodwill and generate positive public opinion (Mazurek, 2002). Scholars have also theorized that these dirty firms have the most to gain from institutionalized improvement strategies and from reducing the adverse risks of increased regulator and activist scrutiny if they continue to operate too close to or fluctuate around regulated pollution thresholds (Lyon & Maxwell, 2002; Russo & Fouts, 1997). In fact, research has shown that facilities with water pollution discharges build up a regulatory buffer by discharging far below permitted levels because of the high variability in discharge levels (Bandyopadhyay & Horowitz, 2006; Gunningham, Kagan, & Thornton, 2003). Participation in a VEP may simply offer a convenient way to "brand" an existing strategy.

The finding that VEP participants tend to come from industries that are final goods producers and are more visible and concerned with maintaining a strong brand or public image indicates that perceived pressures from various external stakeholders are strong motivators (Henriques & Sardorsky, 1996). These perceived pressures may also indicate either an increase in green demand or concern with higher visibility (i.e., marketing) (Alberini & Segerson, 2002). Similarly, it appears that participants generally are subject to greater (global) competition or tend to be foreign-owned and use VEP participation (e.g., ISO 14001 certification) to signal environmental excellence to distant trading partners.

However, a murkier picture emerges on the question of whether participating firms operate in more competitive industries and are more innovative with greater R&D. While in some studies these factors appear to positively influence VEP participation (1996; Arora & Cason, 1996; Decanio & Watkins, 1998; Khanna & Anton, 2002; King & Lenox, 2001a; Videras & Alberini, 2000), other studies, often by the same authors, find no such effect or find a negative impact on participation (Arora & Cason, 1996; Khanna & Anton, 2002; Khanna & Damon, 1999; Videras & Alberini, 2000). It is also not clear from this research whether more profitable firms tend to participate (Arora & Cason, 1995; Decanio & Watkins, 1998; Khanna & Damon, 1999; Videras & Alberini, 2000). It would appear that firms with older equipment and subject to greater investor scrutiny are more likely to participate (Khanna & Anton, 2002; Khanna & Damon, 1999). These attributes are not necessarily mutually exclusive and overall paint a picture of participating firms that attempt to use VEP participation to differentiate themselves in highly competitive consumer products markets where profit margins are tight and investor perception is tied closely to maintaining a good brand image and reputation. In addition, enterprises that are global market players subject to greater supply chain pressures are more likely to participate.

How Effective Are VEPs?

As noted earlier, VEPs can be evaluated from several angles, including pollution abatement levels and the impact on polluting firms and competition. In this section, I focus on research of VEP effectiveness in inducing participants to abate production-related pollution beyond what could be expected in a business-as-usual scenario and the potential to gain competitive advantage from participation. Multiple scholars have noted that VEPs lack monitoring and reporting requirements, which could yield data that can be used as a comparative baseline in effectiveness analyses (Alberini & Segerson, 2002; Lyon & Maxwell, 2002).8 The lack of baseline data, in part, explains the fact that most VEP research to date has focused on the participation question.

Table 2 provides a summary of various VEPs, both public VEPs launched by the EPA and industry-led VEPs that have been evaluated from the environmental performance perspective; some of which have also analyzed participation. Table 2 shows a program's inception date, any precipitating event that influenced program launch, and a brief program description. Changes in corporate behavior are summarized along with the study time frame and study sample, and the unit of analysis (facility, firm, industry) for each research paper. It becomes quickly evident that several programs were studied by different research teams, which puts us in a stronger position to draw general conclusions on program effectiveness, program design, and methodological issues that might impact the research findings.

I start with the 33/50 program, initiated in 1991 in response to a Science Advisory Board (1990) report on risks of exposure to toxic chemicals and improved transparency based upon data coming from the EPA's TRI starting in 1989 (based on 1987 emissions results) (Khanna, 2007). This program was designed to reduce releases and transfers of 17 high-volume chemicals listed in the TRI by 33 percent in 1992 and 50 percent in 1995 relative to 1988 baseline levels. Firms were specifically invited by the EPA to join and engage in pollution prevention of these chemicals across various media according to their own reduction goals and timetables. Exactly 1,294 firms participated over the lifetime of the program. According to two studies (Khanna & Damon, 1999; Sam & Innes, 2005), the 33/50 program was successful in reducing emissions of the 17 program chemicals, whereas two others show a limited or mixed impact of the program (Gamper-Rabindran, 2006; Vidovic & Khanna, 2005). One obvious difference between these studies is that the former two analyses are at the level of the firm, whereas the latter two focus on changes at the facility and industry level. It is less likely that the facility versus firm difference will have an impact because all researchers assumed that changes in emission levels at all facilities owned by a participating firm were associated with program participation. This assumption should yield industry samples composed of similar facilities. The difficulty in creating a finer distinction between true and estimated participating facilities is attributed to the fact that the EPA kept a record of participating 33/50 firms, while the TRI data used to evaluate the program are gathered at the facility level. Alternatively, the length of study period can impact findings, however, this is a less plausible explanation as three out of the four studies capture the entire program period.

A more plausible explanation of the differences in 33/50's program effectiveness evaluations is the confounding impact of regulation and strategic reporting in analyses using TRI data. The toxic chemicals listed in the TRI are often targeted by multiple regulations and if these emissions reducing influences are not accounted for in an analysis, results will likely yield false positives. Researchers have found evidence of strategic behavior by facilities that have to report to the TRI (Bennear, 2006) and potentially very significant impacts on reported emission levels because of changes in regulation, such as the Clean Air Act Maximum Available Control Technology (MACT) standards and other regulations (Bui, 2005; Koehler & Spengler, 2007). In general, omitting key confounding variables in this type of analysis means that the program under study will appear to be more effective than merited. Based on their 33/50 analysis, Vidovic and Khanna (2005) conclude that the decline in observed emissions from 1991-95 was likely the result of an independent trend rather than a direct consequence of 33/50.

Gamper-Rabindran (2006) specifically excludes 33/50 chemicals that are ozone depleting substance (ODS), which are regulated under the Clean Air Act (in accordance with the 1989 Montreal Protocol), and finds that 33/50 participants shifted emissions to media, which were not included in the program (i.e., off-site recycling). In addition, while fabricated metal and paper industry emissions decreased, chemical and primary metals industry emissions increased! This is a direct contrast with Khanna and Damon's earlier analysis of 33/50 emissions reductions by chemical industry program participants, yet they did not exclude the two ODS substances. These two chemicals alone accounted for the greatest percentage reductions between 1990 and 1996 of all 17 chemicals in the 33/50 program (Khanna, 2007). This important omission in earlier analyses of 33/50 leads me to conclude that there is stronger evidence that the program was not effective in reducing toxic chemical emissions beyond the levels to be expected under business as usual.

Four studies analyze various federal initiatives to reduce greenhouse gas emissions (GHG) (Delmas & Montes-Sancho, 2007; Lyon & Kim, 2006; Morgenstern, Pizer, & Shih, 2007; Welch, Mazur, & Bretschneider, 2000). These studies consistently find that the programs, sponsored by the EPA and DOE, had no or had a transient impact on GHG emissions reductions beyond business as usual. These findings are somewhat nuanced by differences in behavior by early versus late participants. It should be noted that many of these VEPs have been terminated several years ago and are not widely considered among the more successful VEPs by practitioners.

Finally, research into the effectiveness of the EPA's Performance Track program finds that because the program does not adequately recognize and reward the best environmental performers and does not motivate many firms to become high performers, change in environmental performance will be minimal. At best, the program can offer a view inside the firms or facility regarding management decision making with respect to environmental performance (Coglianese & Nash, 2006).

Various industry-led VEPs have been studied, including RC, ISO 14001 EMS, and Sustainable Slopes. The majority of research has been on the first two VEPs and raises interesting questions on the effectiveness of VEP design elements, such as the role of sanctions and third-party auditing. The RC program was created in 1989 to achieve two goals: (i) to improve the environmental and safety performance of members of the Chemical Manufacturers Association (CMA); and (ii) to improve the public perception of the industry. Participants were encouraged to adopt the RC principles and to annually submit a self-assessment to the CMA. Failure to demonstrate progress on implementation did not lead to expulsion, although it could provoke greater peer-to-peer scrutiny. However, members were not held to a particular level of performance. All CMA members were required to adopt RC as a condition of membership (King & Lenox, 2000). In all, 70 percent of chemical firms chose not to join the program, many of them smaller chemical manufacturers (Lenox, 2006).

In the first significant empirical analysis of RC, researchers King and Lenox (2000) find that RC membership did not translate into improvement in the rate of environmental performance. In fact, they find that members improved their performance more slowly than nonmembers. Thus, while toxic chemical emissions for the entire chemical industry decreased, emission from RC members did not decrease any faster. They conclude that explicit sanctions may be needed by informed outsiders to avoid this type of opportunistic behavior, or greenwashing if you will.9

In a related analysis, Lenox and Nash (2003) hypothesize that industry associations that incorporate sanctions in the form of expulsion for noncompliance with an industry code will attract less polluting firms. Of the four programs studied, only two employed the threat of expulsion, the Sustainable Forestry Initiative and the Responsible Distribution Process program (see Table 2 for summary). They find mixed support for their hypothesis. Specifically, participants in RC (without threat of expulsion) were more polluting on average than other chemical firms, while participants in the Sustainable Forestry Initiative were less polluting on average than other pulp and paper firms. Since 1996, when this study was conducted, RC and the Responsible Distribution Process programs have introduced third-party verification of implemented program aspects and program elements aligned with the ISO 14001 EMS standard.

There is a considerable amount of research on ISO 14001, including research on participation and effectiveness. As previously noted, the research finds that adopters are largely more R&D intensive facilities that have previously adopted similar quality management schemes (such as ISO 9001 quality management standards). Certified facilities are also under pressure from foreign firms to demonstrate environmental management Finally, more polluting facilities (measured in absolute TRI emissions) tend to join (King & Lenox, 2001b; King, Lenox, & Terlaak, 2005). According to these researchers, certifying facilities are thus not signaling higher levels of environmental performance. However, when pollution is measured not in absolute levels of emissions, but emissions intensity (air emissions per production volume), cleaner facilities were more likely to adopt ISO 14001 (Toffel, 2006). As larger facilities tend to join VEPs, they will also emit at higher levels, which can explain the King and Lenox result, and does not necessarily contradict the Toffel result It may simply be that the larger facilities in Toffel's sample are less emissions intensive.

In contrast, Khanna and Anton (2002) find that when TRI emissions are disaggregated into on-site emissions (released into the air, water, and soil) and off-site transfers (which are managed) per unit sales (based on Compustat data), firms with larger on-site releases per unit sales and firms with lower off-site transfers per unit sales adopt a more sophisticated EMS.10 Air emissions, as utilized by Toffel, fall within the on-site category of the TRI database, and thus, his finding that cleaner facilities certify contradicts the Khanna and Anton paper.11 The point in this discussion is that choice of environmental performance measure affects results and it is wise to reserve judgment on whether cleaner or dirtier facilities pursue a sophisticated EMS and certify to ISO 14001.

Research on whether certification to ISO 14001 leads to improved environmental performance also yields a complex picture. We would expect that certification to ISO 14001 verifies the existence of an EMS and various related management practices that should, in turn, lead to changes in environmental performance.12 Researchers do find that it is the underlying EMS which leads to future improvements and not certification itself (King etal., 2005), yet this result is sensitive to econometric model specification. Another study (Toffel, 2006), which does not specifically control for the existence of an EMS, finds that certification to ISO 14001 itself is associated with improvements, measured in absolute pounds of toxic chemical emissions and emissions intensity per production volume. Similarly, an analysis that does not take account of the existence of an EMS finds that certification reduces both time out of compliance with clean air regulations and TRI emissions, compared with noncertifying facilities, at least for the short 1995-96 and 2000-01 time frames (Potoski & Prakash, 2005b).

Based on a 2003 survey of 3,189 manufacturing facilities, Andrews, Hutson, and Edwards (2006) did a simple comparative analysis between those who had a certifiable EMS and to those without They find that facilities with an EMS were characterized by modest environmental improvements, such as decreases in energy use, increased use of recycled inputs and recycling of waste, decreases in hazardous and nonhazardous waste generation, and decreased incidence of severe leaks or spills (i.e., "low-hanging fruit").13 They find no significant improvements for air and water pollution and material inputs, which might require more costly capital investments. Similarly, an empirical analysis based upon survey results and publicly available databases finds that at automobile assembly facilities, there was virtually no difference between the environmental performance of facilities with and without ISO-certified EMSs (Matthews, 2001). Furthermore, despite the focus on regulatory compliance, certified auto assembly facilities had more enforcement actions and pay high fines. Finally, at auto assembly facilities with an ISO-certified EMS, improvements in environmental performance did not accelerate from past performance. In some cases, facility performance was actually worse.

Despite these mixed results, the research on EMS does suggest the possibility of institutional learning (and level of comfort with VEPs) and the development of corporate norms of environmental management through participation in various VEPs, as proposed by neoinstitutional sociologists (see discussion earlier). The propensity to participate in multiple programs, indicated in Table 1, may represent an institutional commitment to VEPs. In support of this conjecture, researchers have found that facilities that adopt the ISO 9000 total quality management (TQM) standard are more likely to adopt ISO 14001 (King et al., 2005). TQM practices, such as lean production, are found to be associated with reduced toxic chemicals waste generation (King & Lenox, 2001c). Firms who adopt total quality environmental management (TQEM) adopt more sophisticated pollution prevention activities (Anton, Deltas, & Khanna, 2004). In fact a "consensus" EMS has developed across certifying facilities that is aligned with the ISO 14001 standard and focuses on regulatory issues (Matthews, 2001). A further testament to institutional learning is that firms with past experience with pollution prevention engage in more pollution prevention activities going forward (Khanna, Deltas, & Harrington, 2005).

Institutional and organizational aspects of a firm can also impact the corporate response to environmental initiatives, including VEPs. In a recent analysis of the Performance Track program, researchers find that internal organizational dynamics rather than environmental performance distinguish participants from nonparticipants (Coglianese & Nash, 2006). For example, institutional pressures from different stakeholders are channeled to different organizational functions, which influence how they are received by facility managers and ultimately the firm's response (Delmas & Toffel, 2004a,b). Thus, firm legal departments are more involved in addressing nonmarket (e.g., regulatory agencies, local community, media, and environmental NGOs) pressures, while market pressures (e.g., consumers, suppliers, competitors, socially responsible investors, and trade associations) are handled by corporate marketing departments. When the pressure is coming from market constituents, firms are more likely to adopt ISO 14001. Pressure from nonmarket constituents, however, is found to compel firms to adopt governmentinitiated voluntary programs and less so ISO 14001. Based on this and other analyses, researchers suggest that market pressures, rather than nonmarket pressures, applied through the regulator, are generally more effective at initiating beyond compliance behavior (i.e., participation in VEPs) (Delmas & Toffel, 2004a; Khanna & Anton, 2002). Significantly, these market pressures do not appear to motivate higher levels of pollution abatement (beyond regulated levels) according to the majority of VEP research summarized here.

The role of free-riders in VEPs has been explored by some researchers. Comparing early and late joiners to the Climate Challenge Program, Delmas and Montes-Sancho (2007) find that early joiners reduced emissions more than later joiners, who in fact reduced emissions more slowly than nonparticipants. As a result of free-riding by later joiners, the program in the aggregate appears not to have had an impact on CO2 emissions reductions. Similarly, in Denmark, early joiners in the agreements on industrial energy efficiency achieved more profitable gains than late joiners, indicating the possibility of a first-mover advantage (Morgenstern & Pizer, 2007).14 Analysis of ISO 14001 finds that early adopters, who tend to be more pollution intensive, reduce their TRI emissions more than late adopters (Toffel, 2006). This type of effect appears to be at play with various VEPs and should be incorporated in future analyses.

Other research supports the theoretical conclusions of Dawson and Segerson (2003), who propose that an entire industry can benefit economically from a VEP, though free-riders benefit more. Research on RC finds that while the rate of emissions reductions did not increase after RC creation, RC members benefited from improved financial performance, measured in Tobin's Q (market value over value of tangible assets) (Lenox, 2006). However, nonparticipating chemical firms benefit even more in terms of improved Tobin's Q after inception of RC. Lenox concludes that even though RC yields no environmental improvements and suffers from free-riders, the institution continues to exist because participants and nonparticipants reap some degree of financial reward via surprising changes in investor perception of chemical firms' environmental risk. Similarly, the shareholder value of chemical industry firms did not decrease as significantly after an accident at a chemical firm once RC existed, indicating a protective effect of the VEP. Conversely, negative financial spillover increased for the chemical industry after the deadly chemical accident in Bhopal, India (Barnett & King, in press). However, RC is more protective of nonmembers. Thus, when an accident occurred at nonmember firms, RC members were not protected from a negative financial impact. Nonmembers, however, did not suffer as significant a negative financial impact when the accident occurred at an RC member firm.

Discussion

From the summary in Table 2, it appears that most VEPs targeting production process changes that have been studied are not associated with significant improvements in environmental performance. Several federal VEPs that have been analyzed seem to be unsuccessful or lead to short-lived efforts, with minor nuances because of early versus late adopters. Of the industry-led VEPs studied, only ISO 14001 seems to be associated with some environmental improvements, which tend to be focused at end-of-pipe measures rather than pollution prevention or innovation. The authors of several chapters in Morgenstern and Pizer (2007) found that the various international climate change VEPs demonstrate a 5 percent improvement in energy efficiency during the study period, which while not trivial does suggest that VEPs might inherently have some limitations. Similarly, case studies of voluntary approaches to climate change in the UK, Switzerland, and Denmark found that these VEPs did achieve goals and could improve the efficacy of CO2 tax policies, but that the goals were not often very demanding (Baranzini & Thalmann, 2004). However, it cannot be emphasized enough that there are many VEPs, particularly more recent ones or those targeting greener product design, which might conceivably be more effective. The results presented here are indeed troubling from the perspective of environmental protection and present a fundamental question: Did the minimal environmental improvement arise because the participants fail the institution or did the institution fail participants?

Let us start with the institution: Various researchers have suggested that tighter requirements might improve the performance of VEPs, such as sanctions, rewards, public disclosure, or independent monitoring and verification (Rivera, deLeon, & Koerber, 2006). Environmental targets are set by participating firms in several VEPs. However, the analysis of 33/50, Climate Wise and Climate Challenge indicate that this requirement did not motivate significantly higher pollution abatement above the business-as-usual scenario, possibly because participating firms set very low performance targets. Lenox and Nash (2003) find only limited support for their hypothesis that sanctions in the form of expulsion for noncompliance with a voluntary industry code will attract less polluting firms. VEPs with stronger requirements, such as monitoring and public disclosure used in Performance Track or 33/50, are thought to perform better than weak form programs, such as ISO 14001, which does not employ public disclosure but requires monitoring (Potoski & Prakash, 2005a). However, the research summarized here does not support this conjecture.

Attempts to strengthen VEP requirements have had limited success. VEP designers rightly fear that introducing a more stringent process for participant target setting will scare them away and /or not motivate significant environmental improvements. In fact, although EPA raised the level of rewards for Performance Track participants to compensate for higher entry requirements, the net result was reduced participation rates indicating an important trade-off between level of effort required and willingness to participate (Coglianese & Nash, 2006). Considering five VEPs targeted at energy efficiency in the United States, Germany, the UK, Denmark, and Japan with varying levels of sanction (ranging from taxes, further regulation to industry coercion), Morgenstern and Pizer (2007) conclude that increased stringency does not seem to improve environmental performance. For example, the UK and Danish climate change VEPs were designed as an exemption to a significant tax, but this heavy-handed approach nevertheless did not yield significantly better VEP performance.

In contrast, sanctions (e.g., inspections, penalties, and fines) do work in regulated areas to various degrees. Enforcement activity increases compliance and pollution abatement for pulp and paper facilities and at publicly owned treatments works (Dietrich, 2004; Gray & Shadbegian, 2005). Penalties appear to improve environmental performance and compliance behavior (Foulon, Lanoie, & Laplante, 2002), but for some industries (e.g., pulp and paper plants), the impact lasts only one year (Shimshack & Ward, 2005). Other researchers find that penalties are generally more effective at smaller plants with lower financial resources (Thornton, Gunningham, & Kagan, 2005). Not surprisingly, higher costs of compliance and more complex regulations combined with decreased inspection and detection rates are found to increase the likelihood of violations (Stafford, 2006). Finally, the mere threat of legal sanctions implied by promulgation or enforcement of laws and regulations is much more salient than the imminent threat of legal punishment (Thornton et al., 2005). In other words, while the ultimate punishment, penalties, appears to spur compliance behavior, particularly at small facilities, the threat of regulation and associated uncertainty may be a more significant driver of compliance behavior.

The impact of regulation and regulatory uncertainty has been considered in the empirical research on VEPs. Various researchers (Alberini & Segerson, 2002; Davies, Darnall, Mazurek, & McCarthy, 1996; Videras & Alberini, 2000) have suggested that VEPs with a direct link to an existing regulatory framework, such as Waste Wi$e or 33/50, are more successful because there is a clearer connection to cost savings arising from pollution reduction. For 33/50, however, I conclude that it was in fact direct regulation (under the Montreal Protocol) rather than a "linked" VEP that engendered the significant portion of toxic chemical reductions. Others suggest that various proxy variables for potential future regulatory pressures, such as environmental group membership are motivators for voluntary pollution abatement (Karamanos, 1999; Maxwell, Lyon, & Hackett, 2000; Welch et al., 2000), although the evidence is not definitive. My own experience as an industry participant in crafting a packaging waste management VEP in Hungary supports the notion that VEPs preempt regulation or at least weaken and postpone it (Koehler, 2002).

However, it is noteworthy that the regulatory threat as a motivation to participate does not appear to translate into actual improvements in environmental performance by VEP participants. In contrast, in the traditional compliance regimes, regulatory uncertainty appears to motivate pollution abatement, which may lend support to theories that the perceived seriousness of regulatory threats and subsequent enforcement activities guide the decision to invest in pollution abatement As noted by numerous scholars, VEPs are popular when the threat of regulation is low. Thus, the weak institutional framework for VEPs, both in terms of program design and regulatory threats, partially can explain the lack of effectiveness.

Now consider whether it is the participants who have failed the VEP as an institution. Some of the compliance research provides hints that firms tend to sort themselves into good or bad environmental stewards and will do so more likely because of industry fixed effects (e.g., technology), market pressures and peer pressure than regulatory pressure. Research finds that a relatively small number of plants are responsible for most violations of environmental regulations (Horowitz, 2002), and a positive correlation in the probability of violation across facilities owned by the same firm (Stafford, 2006). Pulp and paper plants that are old, large, host a more pollution intensive pulp mill, and violate water pollution or OSHA regulations, tend to violate air regulations (Gray & Shadbegian, 2005). At the other end of the spectrum, firms that are already committed to compliance for normative and reputarional reasons view themselves as responsible corporate citizens who need not fear the social and economic costs that can be triggered by serious violations (Thornton et al., 2005). In fact, day-to-day environmental management practices appear to have a greater influence on variation in levels of environmental performance (in the pulp and paper industry) than differences in regulatory stringency (Gunningham et al., 2003), indicating that firm internal decisions are critical, and may also explain VEP participation decisions within a neoinstitutional framework.

There appear to be some parallels between those firms who do not comply and those who participate in VEPs: large and polluting, see Table 1. It would be useful to know if these are the same firms or if industry membership makes a difference. For example, large polluting intermediate goods firms may be less inclined to comply than large polluting final goods firms, lending support to theories that consumer and customer demand plays a role in VEP participation; although this continues to be debated (Alberini & Segerson, 2002; Gunningham etal., 2003; Salzmann, Steger, Ionescu-Somers, & Baptist, 2006). Alternatively, large polluting firms in final goods industries may be less inclined to invest in compliance, but nevertheless participate in a VEP to gain regulator goodwill and greenwash.15 Perhaps it is the large polluting final goods firms that join a VEP earlier and actively engage in pollution prevention, while large polluting intermediate goods industries join later and do not engage in pollution prevention, effectively negating the efforts of early joiners with their free-riding behavior. There are multiple possibilities that would need to be examined by more comparative industry analysis. It would also be very interesting to tease out whether more or less compliant firms join VEPs as this could help illuminate why VEPs are not as effective as hoped and which type of sanctioning methods might work better, given the research on compliance behavior referenced earlier.

Finally, it may be factors other than the institutional design and regulatory framework of a VEP or the a priori proactive environmental attitude of certain types of firms that influence participant commitments under VEPs. Researchers suggest that beyond-compliance behavior cannot be explained purely in terms of threats and moral obligations, but rather can be better explained as an interplay of societal pressures and economic constraints (Gunningham, Kagan, & Thornton, 2004). Ramus and Montiel (2005) find evidence that coercive, normative, and mimetic institutional pressures determine whether firms make a public commitment or not to environmental actions, whereas economic incentives determine whether they actually implement their commitment.

In other words, the problem may be that while firms are willing to demonstrate a proactive environmental policy by participating in VEPs, it may not be economically advantageous to subsequently invest in pollution abatement We know from research that U.S. facilities tend to face significant marginal abatement costs (McClelland & Horowitz, 1999; Rezek & Campbell, 2007), which would imply that unless forced to do so, most facilities are unlikely to invest in pollution abatement technology. It is striking that of the VEPs studied (e.g., ISO 14001 and climate change programs), the changes that have been made appear minimal, hardly distinguishable from business as usual, and probably low-hanging fruit The ineffectiveness of Performance Track and other VEPs may be because the incentives that can be legally offered by the EPA to participating firms do not appear to offer sufficient economic reward to compensate for potentially significant investments in production processes required to achieve (costly) improved environmental performance. The final goods firms may have joined these VEPs in expectation of raising their market appeal, but low levels of environmental performance may indicate that these pressures are insufficient or market players are informationally challenged and cannot "punish" lackluster performance. On the other hand, research on the voluntary Costa Rican Certification for Sustainable Tourism shows that those participants who achieve superior (presumably more costly) performance are able to command price premiums for hotel rooms (Rivera, 2002) and that proximity to the country's eco-parks, which attract eco-tourists and more government scrutiny, predicts participation (Rivera, 2004).

Theoretical work and empirical research indicate that the national political context can be an important influence on the decision to participate in a VEP (Lyon & Maxwell, 2004; Potoski & Prakash, 2004) and undertake pollution abatement For example, implementation of the European Union's Eco-Audit and Management Scheme (EMAS) is quite variable based upon domestic factors, including standard promotion and dissemination, stakeholder demands, and the extent of regulatory relief offered by national governments (Glachant, Schucht, Bueltman, & Waetzold, 2002; Kolln & Prakash, 2002). Public disclosure of environmental performance appears to have a marked effect on pollution levels in Indonesia and Canada (Foulon et al., 2002; Garcia, Sterner, & Afsah, 2006). In the latter case, researchers find that appearing on a list of out of compliance firms is more effective at reducing water pollution than traditional sanctions, such as fines.16 Kollman and Prakash (2001) suggest that in an adversarial economy (or what Lyon & Maxwell, 2004, call a "pluralist" political setting), industry willingness to engage in a VEP is influenced by whether individuals and citizen groups or NGOs can use litigation as a means to enforce good environmental behavior. In the United States, the reluctance of firms to adopt ISO 14001 arises in part from concerns that evidence gathered by a third-party auditor can be used in a court of law. Similarly, members of the U.S. forestry product industry preferred to certify to the domestic private Sustainable Forest Initiative rather than join the more rigorous NGO-sponsored international Forest Stewardship Council to preserve an element of control in an adversarial context (Sasser, Prakash, Cahore, & Auld, 2006).

In the United States, the design of VEPs is constrained by regulatory requirements. As noted earlier, the EPA is constrained in its ability to offer regulatory relief without congressional authorization, and therefore, may not be able to offer enough economic incentive to spur pollution abatement Furthermore, antitrust laws limit the types of sanctions participating firms can levy against members who fail to comply with the codes of an industry (self-regulatory) VEP. VEP participants cannot coerce other firms to participate or impose internal sanctions that would impact production costs for competitors (Lenox, 2006; Lenox & Nash, 2003). Another concern in the United States is the potentially adverse impact on VEPs of third-party legal complaint, such as actions initiated by environmental groups (Delmas & Terlaak, 2002; Kollman & Prakash, 2001; Potoski & Prakash, 2004).

In combination, these findings can leave the policymaker in a quandary. To what degree is regulator involvement sufficient or necessary for the success of a VEP in the United States? How do institutional inertia, a pluralistic political setting, and a history of adversity impede VEP adoption and performance? If the regulator increases sanctions in an existing VEP within the limited playroom given by Congress, this may launch widespread defection or have no effect at all on environmental performance. If the regulator increases the threat of future regulation, it is not clear whether this will increase industry effort, although this has been suggested in theory and anecdotally In the United States, at best, regulatory threats appear to induce greater levels of VEP participation, although it is notable that early adopters appear to engage in more pollution abatement for reasons that remain to be explored. For participants, the lack of significant improvement may in fact be attributed to the lack of economic incentive or to very conservative target setting and a wait-and-see attitude. For outside observers, including the EPA, consumers and/or third-party auditors, the information asymmetry may be so great as to make it difficult, if not impossible, to determine whether VEP participants are setting ambitious targets and performing beyond business as usual. This phenomenon has been noted in several climate change VEPs included in the Morgenstern and Pizer (2007) volume. It is not helpful in this regard that many EPA VEPs (i.e., PPs) have not established baselines, which would readily lend themselves to effectiveness analysis.

Conclusions and Future Research

As the EPA considers the effectiveness of EPA-sponsored VEPs and whether this type of policy can be depended on to achieve environmental results, particularly in priority areas, the research offers several warnings. Based on what we know today, VEPs in the Untied States that target pollution arising from production processes can attract participants, but do not incrementally yield significant environmental improvement. It may be that on the margin, a VEP can spur higher rates of improvement in production processes in areas where pollutants are already regulated elsewhere, such as the 33/50 program, or regulation is anticipated, e.g., climate change. However, we have yet to research this question.17 In lieu of blind faith in VEPs as an option for changing production processes, policymakers need to understand that these VEPs are used by industry as a public relations tool, intended to appease critics in government and consumer circles alike. Outside observers and critiques appear justified in viewing VEPs as a stop-gap measure, particularly in politically unsavory areas, such as climate change. For significant change in production-related emissions, different policies, including regulation, will be necessary. Indeed, Khanna and Anton (2002) find that first-order practices, such as environmental staffing, audits, and internal policies are attributable to legal and regulatory factors, whereas second-order practices (e.g., total quality management) are driven by market factors.

The research does suggest that expectations of VEPs are not properly aligned with the decision-making reality at the firm (or in government) to permit appropriate conclusions on the effectiveness of VEPs. It would appear that theories of regulatory threats, market pressures, and neoinstitutional factors can explain the decision to participate, yet require more work to explain why participants do not improve environmental performance. The strongest reason, I believe, is that it is economically unfavorable to invest in pollution abatement under a voluntary regime in the current sociopolitical situation in the United States, which is not thoroughly explored in the theoretical and empirical work. In other words, the rewards, whether they be offered by the government or expectations of consumer demand, are not sufficient to compensate for what are likely to be the expensive investments in production processes that could achieve notable change.

According to some researchers, regardless of the lack of significant environmental improvement by participants, participation in a VEP may in itself be a public good (Morgenstern & Pizer, 2007). The more participants, the greater the potential benefit In fact, the existence of a VEP improves the image of both the regulator and the regulated by signaling the willingness of both sides to engage in a more flexible process of environmental protection. However, a blind pursuit of high participation numbers, such as EPA's strategy of first inviting a high-profile group of organizations to participate in a VEP, is an open invitation for late adopters to free-ride and thereby jeopardizes the overall effectiveness of the program (Delmas & Montes-Sancho, 2007), or as noted by Morgenstern and Pizer (2007), lead to short-lived efforts.

It is notable that of the VEPs studied using empirical methods, researchers tend to focus on industrial production processes. A central reason is that the available environmental performance data come from the EPA, which, through a variety of regulations focused primarily on production processes, has required primarily large facilities to collect a vast array of production-related environmental data. Therefore, the main theoretical thrust of environmental economists to explain the motivation to join a VEP has been to explore the risk of regulation of production processes and not of products. In other words, both theoretical and empirical work on VEPs is colored by the prevailing method of environmental protection in the United States: control of industrial production processes. In this setting, a VEP may be doomed for a variety of reasons, including the adversarial relationship between regulator and regulated, costly abatement, high levels of regulatory interference that make it particularly difficult to empirically isolate an incremental impact associated with a VEP, and environmental databases (from the EPA) that essentially track compliance and do not always lend themselves to beyond compliance analysis without very careful sleuthing to evaluate various confounding variables.

As an institution, VEPs should not merely be discarded. Rather, the role of EPA should be more carefully explored in terms of the market failure that is being targeted and the key drivers creating the failure. It is entirely conceivable that the EPA can play an effective and unique role in resolving the failure, not only through the provision of technical information at a below market cost.18 It is in areas where there are clear economic incentives to innovate and invest in pollution reduction that VEPs can help accelerate improvements beyond what can be expected of mandatory emissions levels alone. I would argue that by focusing on products and leveraging market developments, the EPA is in a better position to align economic and environmental performance in its VEP programs rather than by focusing on production-related pollution abatement alone, which will still fundamentally be considered a cost in markets. The fact that final goods producers tend to participate in VEPs is clear indication that this is based upon a rationale more closely aligned with market conditions than simply compliance requirements. Consequently, EPA can help create clear market signals and rewards for environmentally friendly product design, such as with Energy Star.

A growing body of research indicates that VEPs, which are clearly tied to economic gains can achieve environmental results. These VEPs tend to focus on energy-efficient lighting (Horowitz, 2001, 2004),19 energy-efficient electronics (Webber, Brown, & Koomey, 2000), energy-efficient motors,20 green building,21 land conservation easements,22 wildlife conservation income tax check-off,23 transportation demand management,24 and eco-labeling (Teisl, Roe, & Hicks, 2002). This body of research merits careful attention in the debate on VEP effectiveness.

The research on ISO 14001 certification in the United States tentatively reveals small improvements in environmental performance, which may be attributed to scheduled third-party auditing and the desire to attract new business across the supply chain. By encouraging the frequency and quality of third-party auditing, as with ISO 14001, the EPA can leverage a private sector "enforcement" mechanism that appears to be yielding positive environmental results. Government monitoring and government-sponsored transparency in (quasi) voluntary schemes, such as Indonesia's PROPER (Garcia et al., 2006) and Costa Rica's Certification for Sustainable Tourism (Rivera, 2002, 2004) appear to stimulate higher levels of environmental effort.

The literature on VEPs summarized here shows some interesting behavior that would benefit from a closer look in future research. Particularly intriguing, I find, is the potential role of networks, geographic clusters, peer pressure, trade associations, and a precipitating event that creates a de facto group more willing to engage in a VEP. The research on RC begins to illuminate the beneficial impact of a VEP (sponsored by a trade association), at least for the financial well-being of participants (King & Lenox, 2000). EPA experts on VEPs have noted that trade associations can play a positive role by increasing the pressure to participate and assisting with information exchange, and should therefore be leveraged by policymakers. Indeed, multiple articles find that industry association membership is crucial in the formation of and recruitment into VEPs (King & Lenox, 2000; Potoski & Prakash, 2004; Rivera, 2004; Sasser et al., 2006).

We would also benefit from more neoinstitutional research to understand the role of networking and geographic clusters in environmental protection, both compliance and beyond compliance measures. For example, researchers have found that more inspections at a plant, nearby plants, and other plants in the same state are associated with greater compliance (Gray & Shadbegian, 2007). Penalty actions against other mills in the same state for water discharge violations result in reduced discharge levels for all facilities, both those that are already compliant and those which are noncompliant (Shimshack & Ward, 2006). Research on the U.S. hazardous waste management industry finds that local competition increases compliance (Stafford, 2007). With most compliance enforcement happening at the regional and state level, this appears to be creating clusters of high performers dependent on the extent of enforcement action and local peer pressure. VEPs that are state-run may be more effective than those developed and administered at the federal level for the very reason that continuous local networking and peer pressure appear to play an important role in pollution abatement It would therefore be very interesting to understand whether firms fear sanctions from the regulator more than sanctions from other firms. If it is the former, then regulatory threats and VEP sanctions need to be heightened. If it is the latter, then the policymaker should leverage this peer-to-peer oversight mechanism.

Finally, we need more research to understand when a voluntary agreement is most effective or appropriate. This issue has not been addressed properly by the empirical research to date and few of the theoretical models presented and tested are designed to provide an answer. This is, however, the predominant concern for the EPA, which must determine the environmental effectiveness of various policies, including mandatory and voluntary approaches, which has been explained in theory (Lyon & Maxwell, 2003; Stranlund, 1995). One possibility would be cost-effectiveness analysis of the relative advantage of VEPs compared with traditional regulation, which remains to be explored (Alberini & Segerson, 2002). We might expect economies of scale to appear as more firms join a VEP and a reduction in firm compliance costs, because government programs, such as those that provide information on pollution prevention opportunities, can partially substitute for private effort. The benefits can be even greater when the government program provides "nonrival" (i.e., nonreplicate/nonduplicate) efforts to firms. In fact, researchers (perhaps optimistically) theorize that the benefit of government services associated with a VEP can eventually outweigh mandatory programs as more firms are targeted to participate (Wu & Babcock, 1999). Thus, a social welfare analysis of VEPs needs to consider the industry costs incurred by participation and the costs avoided by not having to comply with mandatory requirements in addition to the net costs to the regulator (Stranlund, 1995).

For the EPA to effectively use VEPs, there has to be greater understanding of the strategic linkage between public policy and industry behavior, both compliance and beyond compliance. There may be contradictory tendencies and complementary assets that need to be understood and leveraged. Research has yet to provide more insights into the mechanism by which VEPs might work and the interplay between various motivating and demotivating factors for participants. For example, did VEP creation forestall regulation or inflate frustration at the lack of action at the federal level? Empirical analysis is not necessarily suited to this type of research and more promising avenues of research lie in fields, such as experimental economics and psychology. Finally, the distinction among different types of VEPs by researchers, policymakers, and the public is critical to understanding the mechanisms of corporate response. Failure to make these distinctions will result in unforeseen consequences and invalid expectations (Welch et al., 2000).