Voluntary environmental management initiatives: Smoke signals or smoke screens?
By
Michael Wayne Toffel
B.A. (Lehigh University) 1990
M.B.A. (Yale University) 1996
M.E.S. (Yale University) 1996
M.S. (University of California, Berkeley) 2003
A dissertation submitted in partial satisfaction of the
requirements for the degree of
Doctor of Philosophy
in
Business Administration
in the
GRADUATE DIVISION
of the
UNIVERSITY OF CALIFORNIA, BERKELEY
Committee in charge:
Professor David I. Levine, Chair
Professor David J. Vogel
Professor Howard A. Shelanski
Fall 2005
Voluntary environmental management initiatives: Smoke signals or smoke screens?
Copyright 2005
by
Michael Wayne Toffel
Except Chapter 4, “Strategic Management of Product Recovery”, which is Copyright ©
2004, by The Regents of the University of California. Reprinted from the California
Management Review, Vol 46, No. 2. By permission of The Regents.
Abstract
Voluntary environmental management initiatives: Smoke signals or smoke screens?
By
Michael Wayne Toffel
Doctor of Philosophy in Business Administration
University of California, Berkeley
Professor David I. Levine, Chair
This dissertation examines several approaches companies are pursuing to address
environmental externalities associated with their operations. I conduct empirical
evaluations of an international management standard and a voluntary government
program, and develop theory to guide manufacturers in managing their end-of-life
products.
I begin by focusing on the ISO 14001 Environmental Management Systems
Standard, an environmental management standard that requires adopters to seek
independent verification of their conformity. I find that this standard attracts facilities that
are emit more pounds of toxic emissions prior to adoption, but were improving their
environmental performance and legal compliance faster than non-adopters. Companies
reduced their toxic emissions as they were upgrading their management practices to meet
the ISO 14001 standard, a trend that continued in the first few years after certification.
However, because their emissions increased in hazardousness (per pound), I find no
1
evidence that adopters reduced the overall health risk they pose on their communities. I
also found no evidence that adopters subsequently improved regulatory compliance.
In a second study, co-authored with Jodi L. Short, we conduct one of the first
empirical evaluations of a government initiative that encourages companies to “selfpolice” their regulatory compliance and self-disclose their violations.
We find that
facilities were more likely to self-disclose environmental violations to the United States
Environmental Protection Agency’s Audit Policy if they were recently inspected or
subjected to an enforcement action, were narrowly targeted for heightened scrutiny by the
regulator, and were more prominent in their community. We find no evidence that
statutory protections or community pressure encourage facilities to self-police.
In a third study, I focus on how manufacturers are responding to regulatory and
market pressures that are increasingly holding them responsible for their products once
customers no longer want them.
Leveraging transaction cost economics, dynamic
capabilities, and resource dependence theories, I find that product recovery technologies,
uncertainty in reverse logistics, manufacturing-related capabilities, and the uniqueness of
recovered assets are influencing manufacturers’ decisions to vertically integrating into
end-of-life product recovery, form consortia, or outsource end-of-life product recovery
activities.
2
To Erin
i
Contents
List of Tables .................................................................................................................... iv
1. Introduction................................................................................................................ 1
2. Resolving Information Asymmetries In Markets: The Role Of Certified
Management Programs ............................................................................................. 6
2.1 Voluntary Management Programs...................................................................... 10
2.2 The ISO 14001 Environmental Management System Standard......................... 19
2.2.1 A Brief Overview...................................................................................... 19
2.2.2 Prior Evaluations of ISO 14001................................................................ 23
2.3 Theory................................................................................................................. 27
2.3.1 Why Adopters May Already Be Superior Performers: Signaling ............ 27
2.3.2 Why These Programs May Not Legitimately Distinguish Adopters: Free
Riding........................................................................................................ 31
2.3.3 The Potential of Voluntary Management Programs to Improve
Performance .............................................................................................. 33
2.4 Sample and Measures ......................................................................................... 35
2.4.1 Sample....................................................................................................... 36
2.4.2 Measures ................................................................................................... 37
2.5 Methods & results............................................................................................... 44
2.5.1 Selection Analysis..................................................................................... 44
2.5.2 Selection Results....................................................................................... 45
2.5.3 Treatment Analysis ................................................................................... 47
2.5.4 Treatment Analysis Results ...................................................................... 57
2.6 Discussion and Conclusions ............................................................................... 64
2.6.1 Is Independent Verification Inadequate? .................................................. 68
2.6.2 Is a Process-Orientation Insufficient to Improve Performance?............... 73
2.6.3 Implications............................................................................................... 76
2.6.4 Limitations and Future Research .............................................................. 79
Appendix to Chapter 2............................................................................................... 96
3. Turning Themselves In: Why Companies Disclose Regulatory Violations...... 100
3.1 Literature Review ............................................................................................ 104
3.2 The US EPA Audit Policy................................................................................ 110
3.3 Who Turns Themselves In?.............................................................................. 114
3.3.1 The Regulatory Environment.................................................................. 114
3.3.2 Community Pressure............................................................................... 120
3.3.3 Organizational Characteristics ................................................................ 121
3.4 Methods ............................................................................................................ 122
3.4.1 Sample..................................................................................................... 122
3.4.2 Measures ................................................................................................. 124
3.5 Empirical Models and Results.......................................................................... 133
ii
3.6 Discussion......................................................................................................... 137
3.7 Future Research and Conclusions .................................................................... 142
4. Strategic Management of Product Recovery ...................................................... 152
4.1 Motives for Voluntary Product Recovery ........................................................ 155
4.1.1 Reducing Production Costs..................................................................... 155
4.1.2 Promoting an Image of Environmentally Responsibility........................ 156
4.1.3 Meeting Customer Demands................................................................... 157
4.1.4 Protecting Aftermarkets .......................................................................... 158
4.1.5 Preempting Regulation............................................................................ 159
4.2 Product Manufacturers’ Strategic Choice ........................................................ 161
4.3 The Role of Recovery Technologies and Supply Uncertainty ......................... 163
4.3.1 Product Recovery Investments ............................................................... 165
4.3.2 Uncertainty.............................................................................................. 166
4.4 Leveraging Manufacturing-Associated Capabilities to Product Recovery ...... 170
4.4.1 Manufacturing, Service, and Repair Capabilities ................................... 173
4.4.2 Acquiring Tacit Disassembly Know-How.............................................. 174
4.4.3 Feeding Back Recovery Know-How to Designers ................................. 175
4.4.4 Environmental Reputation Capabilities .................................................. 177
4.5 Unique Assets and Avoiding Supplier and Buyer Dependence ....................... 178
4.6 Managerial Implications: Crafting a Strategy .................................................. 182
4.6.1 Limitations and Further Research........................................................... 184
5. Conclusions............................................................................................................. 189
References...................................................................................................................... 192
iii
List of Tables
Table 2.1
Prior Evaluations of Voluntary Management Programs................................84
Table 2.2
Summary Statistics ........................................................................................85
Table 2.3
Correlations....................................................................................................86
Table 2.4
Selection Results: Performance Levels .........................................................87
Table 2.5
Selection Results: Performance Trends ........................................................88
Table 2.6
Adoption Model to Generate Propensity Scores............................................89
Table 2.7
Balancing Covariate & Outcome Levels: Comparing Means........................90
Table 2.8
Balancing Covariate & Outcome Levels: Standardized Bias ........................91
Table 2.9
Balancing Pre-Adoption Trends ...................................................................92
Table 2.10 Difference-in-differences Estimates: Compliance and Emissions.................93
Table 2.11 Difference-in-differences Estimates by Propensity Score Quantiles ............94
Table 2.12 Difference-in-differences Estimates, Early versus Late Adopters.................95
Table 2A.1 Difference-in-differences Estimates, Including Time-Varying
Covariates ......................................................................................................96
Table 2A.2 Adoption Model Including Production Index to Generate Alternative
Propensity Scores ..........................................................................................97
Table 2A.3 Matched Samples Based on Alternative Propensity Scores Also
Balance Adoption Covariates & Pre-Adoption Outcome Levels ..................98
Table 2A.4 Difference-in-differences Estimates, Matched Samples Based on
Alternative Propensity Scores........................................................................99
iv
Table 3.1
Facilities disclosing violations to the Audit Policy .....................................145
Table 3.2
Internal audit statutory protections: Privilege and immunity ......................146
Table 3.3
Federal Circuit Court ideology ....................................................................147
Table 3.4
Variable definitions......................................................................................148
Table 3.5
Summary statistics .......................................................................................149
Table 3.6
Correlations..................................................................................................150
Table 3.7
Who participates in the Audit Policy? .........................................................151
Table 4.1
Drivers of Product Recovery Strategy .........................................................188
v
Acknowledgments
I gratefully acknowledge the assistance of my dissertation committee, Professors
David I. Levine, Howard Shelanski, and David Vogel, who helped make my dissertation
empirically robust, grounded in theory, and relevant to business and public policy. I also
appreciate the encouragement and guidance Professors Paul Gertler, Arpad Horvath,
Kellie McElhanie, and Christine Rosen provided.
Each of the dissertation chapters
benefited from thoughtful feedback provided by my fellow students in the Business and
Public Policy group. I am particularly indebted to Jason Snyder and Geoff Edwards, who
were especially generous in sharing their time and insights.
Chapter 2, Resolving Information Asymmetries In Markets: The Role of Certified
Management Programs, benefited from comments by Magali Delmas, Geoff Edwards,
Paul Gertler, Andrew King, Ted London, Jorge Rivera, and Jason Snyder. Research
support was provided by the Center for Responsible Business at the Haas School of
Business and a Sasakawa Fellowship from the Institute of Management, Innovation &
Organization.
Chapter 3, Turning Themselves In: Why Companies Disclose Regulatory
Violations is co-authored with Jodi L. Short, Department of Sociology, University of
California, Berkeley. I owe Jodi a great deal of thanks for bringing this subject to my
attention and teaching me a great deal about legal and sociology theory. This chapter
benefited from insightful comments by Professors Neil Fligstein, Robert Kagan, and
Jason Snyder. I also gratefully acknowledge research funding from the Center for
vi
Responsible Business and the Institute of Business and Economic Research at the Haas
School of Business. Ara Abrahamian provided excellent research assistance.
Chapter 4, Strategic Management of Product Recovery, benefited from the helpful
comments provided by Erin Deemer, Geoff Edwards, V. Daniel R. Guide, Jr., Eric
Masanet, Christine Rosen, and Luk N. van Wassenhove. This chapter was published in
the California Management Review (2004, Vol. 46, No. 2) in its Special Issue on ClosedLoop Supply Chains. Copyright © 2004, by The Regents of the University of California.
Reprinted from the California Management Review, by permission of The Regents.
Above all, I am most indebted to my wife, Erin Deemer, who provided continual
love and support throughout my doctoral student experience, and Elijah Toffel, who
joined us along the way and provides me with joy and inspiration every day.
vii
Chapter 1.
1. Introduction
Throughout the 20th Century, societies in industrialized—and some developing—
nations have passed increasingly complex laws and regulations to force companies to
internalize the costs their operations impose on the natural environment. By the late
1980s, many policymakers in the United Stated and Western Europe were convinced that
“first generation” command-and-control environmental policies—such as requiring “Best
Available [pollution] Control Technologies”—were too inflexible and costly. By then,
the momentum had shifted toward a hybrid approach, where regulators were to play a
more limited role of establishing pollution thresholds, while companies could leverage
market forces to meet these limits more cheaply. Indeed, there are many success stories
of these second generation environmental policies, including the Toxic Release Inventory
program that requires companies in the United States to publicly report their emissions of
toxic chemicals, and the United States Acid Rain Program that established a cap-andtrade scheme for sulfur dioxide emissions. The recently ratified Kyoto Protocol to the
United Nations Framework Convention on Climate Change suggests that policymakers
and industry continue to embrace this hybrid approach as a cost-effective pollution
mitigation strategy.
The past decade has seen dramatic growth in “corporate self-regulation”, the third
generation of environmental policy where companies assume a much larger role in
regulating their own behavior. Industry proponents argue that self-regulation is a more
1
efficient and effective way to achieve regulatory goals, and that voluntary, private
compliance initiatives should largely replace what they see as a cumbersome bureaucratic
regulatory system. A substantial and growing body of academic literature also extols the
virtues of a more cooperative regulatory system.
There are two types of corporate self-regulation initiatives. The first type includes
a burgeoning number of voluntary programs initiated by regulators, industry associations,
and non-governmental organizations that attempt to encourage firms to reduce their
pollution well below regulatory thresholds. Classic examples of these “beyond
compliance” programs include the US Environmental Protection Agency’s 33/50
Program, the chemical industry association’s Responsible Care, and the Coalition for
Environmentally Responsible Economies’ CERES Principles. Remarkably, most of these
voluntary programs, and nearly all that have been empirically evaluated, require no thirdparty verification. Perhaps not surprisingly then, evaluations of these programs have
found little evidence that they have attracted better-than-average participants or that they
lead to improved environmental outcomes (King & Lenox 2000; Welch, Mazur &
Bretschneider 2000; Lenox & Nash 2003; Rivera & de Leon 2004), prompting charges
that they are nothing more than industry “greenwashing” (Eden 1996). Several scholars
that have conducted these evaluations have noted that these programs’ lack of
independent verification seriously undermines their credibility (King & Lenox 2000;
Rivera & de Leon 2004). In Chapter 2 of this dissertation, I evaluate the ISO 14001
Environmental Management System Standard, one of the most widely adopted voluntary
programs that require independent verification and for which performance data are
2
available. I evaluate this program in terms of two of its ultimate objectives: improving
regulatory compliance and reducing pollution. I investigate whether ISO 14001 attracts
manufacturing facilities that already exhibit better-than-average performance, and
whether facilities that adopt this program subsequently improve their performance. To
empirically evaluate the latter issue, I conduct a difference-in-differences analysis using a
quasi-control group based on propensity score matching.
“Self-policing” programs constitute the second type of self-regulation. These
government programs provide incentives to encourage companies to self-disclose their
legal violations, shifting the burden of monitoring regulatory compliance from the
government to the private sector. Examples include the US Department of Agriculture’s
Hazard Analysis and Critical Control Point program and the US Federal Aviation
Administration’s Voluntary Disclosure Reporting Program.1 Despite private-sector
enthusiasm and academic models that demonstrate the potential cost-effectiveness of selfpolicing, little is known about the actual efficacy of self-policing programs, perhaps due
to the difficulty of observing firms’ internal monitoring and policing decisions. In
1
The Hazard Analysis and Critical Control Point program encourages meat and poultry
processing plants to monitor their processes for contamination, report any violations, and
implement corrective actions. The Voluntary Disclosure Reporting Program waives
penalties for qualified air carriers that voluntarily report and correct their own regulatory
noncompliance.
3
Chapter 3, Jodi L. Short2 and I conduct one of the first empirical examinations of a “selfpolicing” initiative. We investigate a variety of factors that might lead companies to
participate in the Audit Policy, a US Environmental Protection Agency program that
waives punitive penalties for companies that discover, promptly self-disclose, and correct
their violations of environmental regulations. We examine how enforcement activities,
statutory protections, community pressure, and organizational characteristics influence
organizations’ decision to self-police.
Despite the recent popularity of self-regulation, governments are continuing to
expand the scope of market-based regulation. Besides mitigating climate change, one of
the most active areas of new environmental regulation is in extending producer
responsibility to internalize costs associated with product disposal. Such regulations are
designed to achieve three goals: (1) to reduce household waste disposal costs typically
borne by municipalities; (2) to internalize disposal costs in prices to provide incentives
for producers to reduce these costs, such as by using less hazardous materials; and (3) to
encourage producers to design their products to facilitate the reuse and recycling of their
materials. The extended producer responsibility laws and regulations that have emerged
over the past decade, primarily in Europe and East Asia, have targeted a diverse array of
products, from packaging to consumer electronics to automobiles. These laws have
provided manufacturers with widely varying levels of flexibility to respond: some require
2
Department of Sociology, University of California, Berkeley
4
firms to join a single nationwide consortium, while others allow them to decide whether
to conduct their own product recovery, form partnerships, or outsource these tasks. In
Chapter 4, I develop theory to guide manufacturers in how to respond to these new
regulations and, in unregulated markets, to market pressures to manage their end-of-life
products. I employ transaction cost economics, dynamic capabilities, and resource
dependence theories to identify factors that influence manufacturers’ decision to
vertically integrate into end-of-life product recovery, form partnerships, or outsource
these tasks.
5
Chapter 2.
2. Resolving Information Asymmetries In Markets:
The Role Of Certified Management Programs
A considerable literature has focused on the problems associated with asymmetric
information in markets. Many types of mechanisms have been evolved to attempt to
overcome such problems, including branding strategies and mandatory information
disclosure regulations. However, relatively little is known about what features enable
some of these mechanisms to be effective in overcoming problems associated with
information asymmetries. This chapter examines one approach whose popularity has
exploded in the past decade: voluntary management programs. These programs call on
participants to adopt various management practices, procedures, and frameworks, but
vary dramatically in their verification requirements. Some require no oversight, while
others require periodic inspections by independent, accredited auditors. While prior
studies have found that voluntary programs with weak verification requirements appear
ineffective, I conduct one of the first evaluations of a program with a strong verification
requirement. I find that a voluntary management program that requires periodic
independent verification to assure conformity with the program’s requirements appears to
be somewhat more effective in addressing problems associated with information
asymmetries.
I focus on information asymmetries associated with companies’ internal
management practices. Why do business customers care about their suppliers’
6
management practices governing quality, financial, environmental, or labor issues? In
part, to manage risk. A supplier’s production practices can harm its customers because it
can damage its customers’ end-product quality, brand image, and corporate reputation.
Inconsistent production processes can reduce the quality of buyers’ final goods, which
can lead to higher costs and damaged reputations.3 In addition, suppliers who manage
regulatory compliance poorly may be more likely to be shut down by regulatory
inspectors, which can impose costly business interruption on their customers (MedinaRoss, 2002). Furthermore, customers are increasingly being held accountable for their
suppliers’ management practices, especially labor and environmental practices viewed as
inappropriate by end-consumers.4 Beyond risk management, some companies are
3
For example, the tires produced by replacement workers during a labor strike at a
Firestone plant were of inferior quality, which subsequently led to a recall of millions of
Ford Explorers and a great deal of negative publicity for both companies (Krueger &
Mas, 2004).
4
While some activists and media exposés target suppliers are located in industrialized
countries, such as those charged with inhumanely treating farm animals supplied to
McDonald’s, Wendy’s, Burger King, and KFC (Becker, 2003; McNeil, 2004), such
charges are more typically levied toward suppliers in less developed countries where
labor and environmental regulations are often lax (as is regulatory enforcement). In those
locales, charges of abusive labor practices among suppliers of Nike, KathyLee, P.
7
interested in selecting suppliers with superior environmental practices to reduce their
overall life-cycle environmental impacts (Chouinard & Brown, 1997; NEETF, 2001;
Walton, Handfield, & Melnyk, 1998; Wycherley, 1999).
While important to some customers and other stakeholders, suppliers’ production
practices are typically difficult for stakeholders to observe directly. While regulations
have arisen to address such information asymmetries in a few circumstances that directly
affect public welfare such as hospital safety (Cutler, Huckman, & Landrum, 2004;
Mukamel & Mushlin, 1998) and restaurant hygiene (Jin & Leslie, 2003), in most cases
firms are left to develop their own mechanisms to address these information asymmetries.
Hundreds of thousands of firms have adopted a burgeoning number of voluntary
management programs. Because voluntary management programs do not include
performance requirements and few require independent verification that participants
have fully implemented them, critics charge that voluntary management programs are
simply marketing gimmicks. Prior studies of voluntary management programs have
found little evidence that better-than-average performers adopt them or that participation
Diddy’s, and many other large clothing retailers (Abrams, 1999; Barnet & Cavanagh,
1994; Collier & Strasburg, 2002; Greenhouse, 1999, 2003; San Francisco Chronicle,
1999; Strom, 1996) and “inhumane working conditions” within suppliers of Dell,
Hewlett-Packard, and IBM (CAFOD, 2004) have subjected these companies to unwanted
negative publicity.
8
improves performance (King & Lenox, 2000; Lenox & Nash, 2003; Naimon, Shastri, &
Sten, 1997; Rivera & de Leon, 2004). These findings have largely been attributed to the
evaluated programs’ lacking verification mechanisms to ensure conformance.
Stronger monitoring might better enable these mechanisms to differentiate
adopters as organizations with superior management practices. In this chatper, I empirical
evaluate one of the few voluntary management programs that require periodic
verification by an authorized third party. I examine whether such differentiation emerges
because the program attracts participants whose ex ante performance is better than
average (a positive selection effect) and whether participation improves performance (a
treatment effect). I focus on the ISO 14001 Environmental Management System
Standard, an international environmental management standard that has been adopted by
tens of thousands of companies around the world. This standard was developed to help
companies assure regulatory compliance and reduce their environmental impacts.
I find that ISO 14001 has attracted “dirtier” facilities that were improving their
compliance and environmental performance faster than non-adopters. After developing a
quasi-control group via propensity score matching, my difference-in-differences analysis
provides evidence that companies reduced their toxic emissions as they were upgrading
the comprehensiveness of their environmental management system to meet the
requirements of the ISO 14001 standard, a trend that continued in the first few years after
certification. However, I also found evidence that these emissions were increasing in
hazardousness (per pound), suggesting adopters were concentrating their pollution.
Combining these effects, I find no evidence that adopters reduced the overall health risk
9
they pose on their communities. I also found no evidence that adopters subsequently
improved their regulatory compliance. These findings suggest that a voluntary
management program that requires third-party verification can be a useful mechanism to
differentiate organizations as having superior management practices, both though a
selection effect and via the preparations required to adopt the program.
The chapter proceeds with a more detailed introduction to voluntary management
programs and a review of prior evaluations. I then describe the ISO 14001 standard and
how its adoption might be a credible signal of superior compliance and environmental
performance, and why it might not be. I combine my field research observing companies
that implemented the ISO 14001 standard with a discussion of organizational learning
and incentives to describe how operating in accordance with the ISO 14001 standard may
improve performance. Next, I describe the sample, measures, and empirical methods, and
present the results. I conclude this chapter with a discussion of the results, their
implications for firms and policymakers, and suggestions for future research.
2.1 Voluntary Management Programs
Vertical integration and contracting were proposed decades ago as alternative
solutions to address the information asymmetry problem where a buyer cannot observe
key processes of the seller. Buying products from suppliers whose production processes
are costly to monitor, yet critical to the performance of the final product, induces risks of
shirking (Alchian & Demsetz, 1972; Williamson, 1981). When supplier practices are of
significant importance to buyers but are difficult to observe, vertical integration and
contracting offer alternative solutions (Alchian & Demsetz, 1972; Williamson, 1985).
10
While research has identified some circumstances where firms respond to these
circumstances by vertically integrating (Anderson & Schmittlein, 1984; Anderson &
Weitz, 1986; Teece, 1978), monitoring is just one of numerous factors that determines the
boundaries of the firm. As for contracting, buyers can seldom require suppliers to provide
insurance against losses caused by their poor performance because of difficulties in
verifiably assigning blame and quantifying damages.5 Without explicit contracts, the
potential self-enforcement incentive of suppliers is often thwarted in markets that eschew
long-term relationships and where reputation effects are limited.6 Alternative solutions
include companies either actively monitoring their own suppliers or relying on voluntary
management programs as a signal of superior management practices and performance.
5
Verifiably assigning blame for end-product quality problems to a particular supplier
remains fraught with controversy, and is sometimes self-defeating for buyers if endcustomers perceive attempts to blame suppliers as avoiding responsibility. Quantifying
damages associated with some of the most significant resulting losses, such as eroded
brand value and potential future sales, lacks well-established methods and remains a
controversial task. Furthermore, many suppliers are unlikely to afford compensating
buyers for such losses, since many are relatively small compared to their buyers.
6
For example, many buyers in the apparel industry (the brands) annually change nearly
half of their suppliers, and they share little information with each other about their
experiences with particular suppliers.
11
While some firms, such as Nike, Gap and Levi’s have dedicated significant resources to
develop elaborate supplier monitoring systems, few companies are pursuing this costly
strategy.7
Instead, I focus on voluntary management programs, which have been initiated by
industry associations, non-governmental organizations (NGOs), and international
agencies and adopted by hundreds of thousands of organizations to reduce information
asymmetries in business-to-business transactions.8 These programs share a common
7
A recent World Bank study that examined buyers' efforts to monitor the social and
environmental aspects of their own suppliers found that this “has resulted in some
improvements, some of them substantial, [but] possibly has reached its limits….[These
buyers] generally acknowledged that the costs of monitoring are becoming increasingly
high and in the long run are not sustainable [especially] as some buyers have started to
look beyond the first tier of suppliers, opening up the possibility of monitoring many
more suppliers. Accordingly, buyers are looking to solutions that would allow cost
sharing.... Moreover, a number of buyers would like to avoid duplication as much as
suppliers” (Jørgensen, Pruzan-Jørgensen, Jungk, & Cramer, 2003: 16-20). These findings
suggest individual monitoring will give way to the use of more standardized approaches
like the voluntary programs discussed in this paper.
8
Examples of industry-initiated programs that address environmental management
practices include Responsible Care (chemical industry), Sustainable Forestry Initiative
12
focus on production processes rather than end-results, and seek to differentiate adopters
as possessing superior management practices related to quality, environmental, or labor
and human rights. Consequently, buyers may view adopters as being less likely to deliver
poor quality goods or have their labor or environmental practices be the subject of media
exposés that can damage buyers’ reputations.
At the same time, voluntary management programs also represent an approach
governments are pursuing to overcome widely recognized concerns of the traditional
command-and-control regulatory approach (e.g., fragmentation, inflexibility, complexity,
(forest products), Sustainable Slopes (skiing), and ECOmmodation Rating (hospitality).
NGO programs include the Social Accountability 8000 labor/human rights standard and
the Forest Management Certification and Chain of Custody Certification environmental
standards. Supranational organizations have also launched voluntary programs aimed at
improving environmental and labor management, including the European Union’s EcoManagement and Audit Scheme (EMAS) and the United Nations Global Compact
National. International process standards include the ISO 9000 Quality Management
System Standard, the ISO 14001 Environmental Management System standard, and the
OHSAS 18001 Occupational Health and Safety Management Systems Specification.
International standards are the most popular form of voluntary management programs,
with over half a million facilities across 161 countries having adopted ISO 9000 (ISO,
2002).
13
high administrative costs, and high compliance costs) (Stewart, 2001). Seeking to
promote a more cooperative, cost-effective method to achieve regulatory objectives,
some government agencies have launched voluntary management programs that
encourage companies to monitor their own compliance and to improve their performance
beyond minimum compliance thresholds.9 The logic of government voluntary programs
is to publicly praise and reward firms that either have strong compliance programs or that
are striving to perform beyond compliance thresholds. “Weak firms, seeking the
incentives that agencies are offering, will emulate their…stronger competitors. Instead of
just punishing the bad, agencies will be able to nurture the good” (Coglianese & Nash,
2001). Some companies have adopted voluntary programs to signal to regulators and
others concerned with their regulatory compliance that they are taking this matter
seriously. Often, participation is driven by the hope of earning fewer inspections,
streamlined permitting, or other manifestations of regulatory goodwill.
9
Examples of government-initiated voluntary management programs that encourage
companies to go “beyond compliance” include the United States Environmental
Protection Agency’s (US EPA) Green Lights and Waste Wise programs that encourage
participants to bolster energy and material efficiency and the US Occupational Health and
Safety Administration’s Voluntary Protection Programs. Other government voluntary
management programs encourage companies to self-police their regulatory compliance
(see Chapter 3).
14
Besides compliance, many are adopting voluntary environmental management
programs “to demonstrate to the public, regulators, and environmental advocacy groups
that participating firms are committed to environmental protection and are taking
responsibility for reducing their impacts” (Nash & Ehrenfeld, 1997: 495).
Uncertainty about whether voluntary management programs actually distinguish
adopters has left many companies and regulators unsure about the extent to which these
programs can provide a reliable substitute for their direct efforts to monitor production
processes. For a program to legitimately differentiate participants from non-participants,
at least one of the following must occur: the program must disproportionately attract
participants with superior ex ante performance (a selection effect), or organizations that
participate in the program must improve their performance faster than non-participants (a
treatment effect).
Robust research has only begun to evaluate whether voluntary management
programs distinguish adopters (Melnyk, Sroufe, Calantone, & Montabon, 2002). Most
prior studies focused on the most common type of voluntary management programs:
those that rely only on an “honor system” to ensure conformance.10 These evaluations
10
The chemical industry’s Responsible Care program, which attracted worse-than-
average performers and its participants performed worse than non-participants (King &
Lenox, 2000; Lenox & Nash, 2003), uses “moral pressure” as its “primary means of
ensuring compliance” (Gunningham, 1995: 69). Even the chemical industry association
15
found no evidence that those with superior performance are adopting these programs (see
Table 2.1). In fact, there is far more evidence that those with inferior performance are
more likely to participate (King & Lenox, 2000; Lenox & Nash, 2003; Naimon et al.,
1997; Rivera & de Leon, 2004). Furthermore, the subset of these studies that evaluated
whether participation in voluntary management programs was associated with
performance improvement found that participation led to worse performance (King &
Lenox, 2000; Naimon et al., 1997). Many have attributed these findings to these
programs’ weak or non-existent verification and enforcement mechanisms (e.g., King
& Lenox, 2000; Rivera & de Leon, 2004).11
admits Responsible Care participants have only been required to make “good faith efforts
to implement [its] program elements” (American Chemistry Council, 2000). Similar
concerns have been expressed about weak enforcement mechanisms in codes of conduct
that address labor and human rights issues (O'Rourke, 2003).
11
For example, King & Lenox (2000: 698) concluded, “effective industry self-regulation
is difficult to maintain without explicit sanctions.” Gunningham (1995: 68) argues that
voluntary programs face a “‘credibility obstacle’ [that] will be insurmountable unless
mechanisms are put in place to give the scheme teeth, and to allow for government and
third-party oversight.” Rivera & de Leon (2004) attribute adverse selection in the ski
industry’s Sustainable Slopes Program “like other typical industry-sponsored voluntary
programs, it does not involve specific environmental standards, lacks third-party
16
Lenox & Nash (2003) investigated the possibility that the threat of expulsion from
a trade association might be a sufficient enforcement mechanism to attract better-thanaverage participants to adopt an industry-initiated voluntary program. Their study yielded
mixed results: one such program attracted better-than-average participants, while another
attracted participants whose ex ante performance was indistinguishable from nonparticipants. Thus, the threat of trade association expulsion is an unreliable enforcement
mechanism to ensure that voluntary management programs distinguish participants’
organizational practices based on superior ex ante performance.
The ISO 9000 Quality Management System standard is one of the few voluntary
management programs with an explicit verification mechanism, as certification to this
international standard requires periodic conformity assessments by third-party
organizations. ISO 9000 has been subjected to a few robust evaluations. Most of these
studies measure performance using financial indicators, which are several steps removed
from the quality-assurance objectives of the standard, and have yielded mixed results.12
oversight, and does not have sanctions for poor performance,” which enabled poor
performers to free-ride on the program’s reputational benefits. Lenox & Nash (2003: 343)
“speculate that only when self-regulatory programs have explicit sanctions for
malfeasance may they avoid adverse selection problems.”
12
While a few studies found that ISO 9000 adoption led to higher production levels,
returns on assets (ROA), and stock prices (Corbett, Montes, Kirsch, & Alvarez-Gil, 2002;
17
The studies that examined the standard’s influence on performance indicators more
directly related its quality-assurance objectives have also yielded mixed results. While
King & Lenox (2001) found that ISO 9000 adoption among US manufacturers led to
lower emissions and waste, Terlaak (2002) examined a similar sample over a longer time
period and found no evidence of these effects. Relying on survey data, Sun (1999) found
that ISO 9000 certified firms reported greater improvements in reducing product defects,
rework costs, warranty costs, and customer complaints than non-certified firms, but the
study did not distinguish whether this resulted from a selection effect or treatment effect.
In this chapter, I evaluate the ISO 14001 Environmental Management System
Standard, which, like the ISO 9000 Quality Management System Standard, requires
periodic verification of conformance by authorized third parties. I focus on ISO 14001
because this standard offers a unique combination. Not only is it widely adopted, which
indicates that many are placing their faith in its signaling or improvement potential, but
data are available to assess its two ultimate objectives of distinguishing adopters based on
their superior regulatory compliance and environmental performance.
Easton & Jarrell, 1998; Terlaak, 2002), other research found no evidence that adoption
led to improvements in a variety of financial indicators (Häversjö, 2000; Heras, Dick, &
Casadesús, 2002; Terziovski, Samson, & Dow, 1997).
18
2.2. The ISO 14001 Environmental Management System
Standard
2.2.1 A Brief Overview
The ISO 14001 Environmental Management System Standard is an international
management standard that provides a framework for conducting environmental
management activities. Established in 1996 by the Geneva-based International
Organization for Standardization, ISO 14001 provides specifications for developing a
comprehensive environmental management system (EMS). The standard requires
organizations to: develop an environmental policy with a commitment to continuous
improvement; identify all of its environmental aspects and then prioritize them based on
the significance of their environmental impacts; establish environmental objectives and
targets; develop work procedures to control environmental aspects; ensure employee
competence by conducting a documented training program; demonstrate a commitment to
comply with environmental laws and regulations; conduct self-assessment audits; and
periodically review the management system (ISO, 1996). The requirements of ISO 14001
are largely based on the “best management practices” of the multinational corporations
such as IBM that participated in creating the standard. ISO 14001 was designed to be
sufficiently flexible so that any type of organization could adopt the standard. Over
66,000 organizations across 113 countries have adopted ISO 14001, including more than
3,500 in the US (ISO, 2004).
19
The standard was created under the premise that organizations that create or
strengthen their EMS in accordance with ISO 14001 will benefit by reducing their
operating costs and environmental impact, enhancing their corporate image, experiencing
fewer and less severe accidents and regulatory violations, and satisfying a requirement of
a growing number of buyers. Like all voluntary management programs, ISO 14001
contains no performance requirements. The nearest the standard comes to discussing
performance are its requiring an organization’s environmental policy to include: (1) a
commitment to comply with relevant environmental legislation and regulations;13 and (2)
a commitment to continual improvement. However, the latter refers to a commitment to
continually improve its environmental management system, not its environmental
performance.
13
Specifically, ISO 14001:1996 requires the environmental policy to include a
“commitment to comply with relevant environmental legislation and regulations, and
with other requirements to which the organization subscribes” (Section 4.2) and that an
adopter must “establish and maintain a procedure to identify and have access to” its legal
requirements (Section 4.3.2).
20
Once an organization believes its EMS conforms to the ISO 14001 standard, it
becomes “certified”14 after a third-party auditor conducts an on-site assessment to verify
that all aspects of the standard have been fully implemented.15 Independent third-party
certification is meant to bring “rigor and discipline” to the process (NAPA, 2001:12).
These third-party auditors must meet ISO auditing requirements,16 and ISO authorizes
one organization in each country to ensure the credibility of the auditing process. In the
14
Unlike elsewhere around the world, this process is officially referred to as
“registration” in the United State rather than “certification.” However, I employ the term
certification because this term is commonly used even in the United States.
15
Technically, ISO 14001 allows a second alternative: organizations may self-declare
their adherence to the standard. In practice, few organizations that claim adherence to the
ISO 14001 standard choose that alternative because certification is significantly more
credible. Since there is no way of knowing the extent to which those who self-declare are
actually adhering to the standard, in this article only organizations that have received
third-party certification are considered to have adopted ISO 14001.
16
The two “Guidelines for Environmental Auditing” standards are “Audit Procedures and
Auditing of Environmental Management Systems” (ISO 14011), and “Qualification
Criteria for Environmental Auditors” (ISO 14012). In 2002, these were replaced by “ISO
19011:2002 Guidelines on Quality and/or Environmental Management Systems
Auditing.”
21
United States, the American National Standards Institute (ANSI) and the Registrar
Accreditation Board (RAB) have operated the ANSI-RAB National Accreditation
Program17 to accredit certifiers and provide training to ensure that audits were performed
consistently and competently in the United States.18 The role of certifiers is to confirm
that the organization has implemented a “management system that should achieve
continuing compliance with regulatory requirements applicable to the environmental
aspects and associated impacts of its activities, products and services” (IAF, 2003: 10).
Rather that verifying regulatory compliance, certifiers confirm that the adopter has
implemented a management system that is fully capable of achieving compliance. As for
17
In January 2005, the ANSI-RAB National Accreditation Program was replaced by the
ANSI-ASQ National Accreditation Board (ANAB), an organization formed by the
American National Standards Institute (ANSI) and the American Society for Quality
(ASQ).
18
Technically, auditors are not required to be accredited to issue ISO 14001
certifications, although in practice the vast majority do. I reviewed a list of 975
organizations across all industries that had been certified to ISO 14001 by mid-2000, and
all of their auditors were accredited. 94% were accredited in the United States, and the
remainder in the United Kingdom, Canada, the Netherlands, or Australia/New Zealand.
In the quality realm of ISO 9001, one survey found that about 90 percent of certifications
were from accredited auditors (Dusharme, 2004).
22
environmental performance, certifiers assess that the organization’s objectives, targets,
and procedures are consistent with its commitments to continual improvement and
prevention of pollution. To maintain certification, adopters must demonstrate that their
management systems continue to uphold the standard in annual surveillance assessments
and full re-assessments every three years (IAF, 2003).
Despite the standard’s absence of performance requirements, many facilities
certified to ISO 14001 want government regulatory agencies to recognize and reward
their proactive approach (NAPA, 2001). Such efforts are often based on the premise that
operating according to the ISO 14001 standard indicates particularly robust management
of regulatory compliance issues as well as credible commitments to reducing
environmental impacts and minimizing the likelihood and environmental consequences
of workplace accidents.
2.2.2 Prior Evaluations of ISO 14001
A plethora of anecdotal evidence recounting how certified companies use less
energy, generate less waste, experience fewer accidents and spills, and have improved
regulatory compliance (Burglund, 1999; Chin & Pun, 1999; Fielding, 1999; Hideaki,
1999; Toffel, 2000) has led ISO 14001 to gain a favorable reputation in some quarters.
ISO 14001 has been referred to as “a tool for demonstrating environmental responsibility
in the global marketplace” (MacLean, 2004b: 13), “a proxy of improved environmental
performance” (NEETF, 2001: 26), and a mechanism “for achieving improvements in
23
environmental performance and for supporting the trade prospects of ‘clean’ firms”
(Smith & Feldman, 2003: 13).19 However, others are unimpressed. Environmental
regulators and activists have focused on the standard’s lack of environmental
performance requirements as a significant impediment to claims that certification
necessarily implies a significant and praiseworthy achievement (Courville, 2003: 288;
Yiridoe, Clark, Marett, Gordon, & Duinker, 2003: 450).
Despite widespread anecdotal evidence, little systematic research has rigorously
evaluated the performance implications associated with ISO 14001 certification (Delmas,
2004; Melnyk et al., 2002; Redinger & Levine, 1998; Rondinelli & Vastag, 2000). Most
prior studies that have examined the effects of ISO 14001 have either (1) focused only on
adopters,20 or (2) compared adopters to non-adopters either before or after adoption
19
The generalizability of anecdotal evidence may be undermined by concerns of
selection bias: adopters who have not achieved much improvement are far more likely to
remain silent about their experience with ISO 14001.
20
Several studies have found that many ISO 14001-certified companies believe that
adopting the standard led them to establish more stringent environmental targets, provide
safer working conditions, improve their regulatory compliance, increase recycling,
improve efficiency in the use of materials and energy, and reduce waste (e.g., Babakri,
Bennett, Rao, & Franchetti, 2004; Berthelot, McGraw, Coulmont, & Morrill, 2003;
24
(Schylander & Zobel, 2003). For example, a recent study found that 75 percent of the 260
facilities certified to ISO 14001 subsequently experienced emissions reductions,
including 53 percent in the year following certification (Szymanski & Tiwari, 2004).
Unfortunately, focusing only on adopters prevents comparing their results to those of
non-certified firms, who may be experiencing similar or even greater performance
improvement. Furthermore, most of these studies do not examine whether the certified
groups’ performance trends changed after they adopted ISO 14001. Comparing adopters
to non-adopters only after certification (without considering selection effects) precludes
parsing out whether any observed differences ex post either already existed prior to
certification or whether operating in accordance with the ISO 14001 standard actually led
to these differences.
Comparative research suggests the enthusiastic perceptions of certified
organizations may be overstated. Several studies that compared adopters to non-adopters
found no evidence of systematic differences in their environmental practices (Hillary,
1997; Hillary & Thorsen, 1999) or regulatory compliance (Dahlström, Howes, Leinster,
& Skea, 2003). Matthews (2001) found no evidence that preparing for certification
reduced toxic emissions. On the other hand, one survey found that companies in Mexico
that have adopted an ISO 14001-style EMS reported better regulatory compliance than
Hamschmidt & Dyllick, 2001; Koenigsberger, 2004; Melnyk et al., 2002; Raines, 2002;
Szymanski & Tiwari, 2004; Welch, Mori, & Aoyagi-Usui, 2002; Yiridoe et al., 2003).
25
others (Dasgupta, Hettige, & Wheeler, 2000), but did not distinguish whether this
difference existed prior to adoption or as a result of adoption.
The one published study that compared the environmental performance of
adopters to non-adopters over time concluded that ISO 14001 certified facilities reduced
their emissions of toxic chemicals more than non-certified facilities (Potoski & Prakash,
Forthcoming). However, that study compared average emissions during 1996-1997 to
2000-2001, but coded facilities as “adopters” if they adopted anytime by 2001. Given that
(1) there has been an overall decline in many facilities’ toxic chemicals emissions since
the early 1990s, and (2) every year since 1996 has seen a growing number of facilities
obtain ISO 14001 certification, the majority of adopters in the sample were probably
certified toward the end of the sample period, when emissions of all facilities were lower.
As such, one cannot clearly identify whether emissions changes during this period were
the result of certification or a selection effect. In addition, like the recent ISO 14001
evaluation by Szymanski & Tiwari (2004), Potoski & Prakash (Forthcoming) do not
control for production volumes, which are directly related to emissions. Consequently,
the study’s finding that ISO 14001 adopters had reduced total emissions more than nonadopters could be attributed to decreased production rather than improved environmental
performance. Finally, their approach relies on compliance with environmental regulations
as an instrument for adoption under the assumption that a facility’s compliance record
does not otherwise directly affect emissions. This assumption is debatable since others
have shown that compliance violations directly lead to behavioral changes that reduced
26
pollution and improved worker safety (Earnhart, 2004; Gray & Scholz, 1993; Kniesner &
Leeth, 2004).
My analysis improves upon these prior studies by comparing adopters to nonadopters over time in a manner that clearly distinguishes selection effects from treatment
effects. I control for production changes when evaluating emissions, and I use a much
broader sample than previous researchers. Finally, I use propensity score matching to
develop quasi-control groups, which avoids often controversial assumptions required for
instrumental variable approaches.
2.3. Theory
In this section, I describe how a voluntary management program with a verification
mechanism might distinguish adopters as possessing superior, difficult-to-observe
management practices. I describe why those with superior practices may or may not be
more likely to select into the program. I also explain how the program could elicit
performance improvement in two areas of central concern to the ISO 14001 standard:
regulatory compliance and environmental performance.
2.3.1 Why Adopters May Already Be Superior Performers: Signaling
The comprehensiveness of a firm’s environmental management practices, its
compliance history, and its environmental performance are largely unobservable to
27
outsiders.21 Nonetheless, a growing number of companies are including environmental
management and regulatory compliance as key criteria in selecting suppliers, and some
21
As discussed earlier, regulations that require information disclosure provide
exceptional cases that could resolve such information asymmetries if the information
disclosed is readily interpretable. Although two information disclosure regulations relate
to the performance measures used in the current study, the information is released with a
several year lag and is quite difficult to interpret. For example, many manufacturing
facilities in the US are required to report their emissions of toxic chemicals to the to US
EPA’s Toxic Release Inventory, although these data are made publicly available after a
two year delay. In addition, substantial differences in the hazardousness of these
substances (Toffel & Marshall, 2004) and the paucity of facility-level production data
make assessing facilities’ environmental performance quite challenging. Websites
designed by non-governmental organizations to inform communities about the emissions
and associated health-hazards of nearby facilities (e.g., The Right-To-Know Network at
www.rtknet.org and Environmental Defense’s Scorecard at www.scorecard.org) do not
normalize the results by production values. As such, larger facilities inevitably appear to
have worse environmental performance than smaller facilities, even if the former emit
less pollution per unit produced. Furthermore, even when governments disclose
compliance data, such as the US EPA’s Enforcement & Compliance History Online
(ECHO) database (www.epa.gov/echo) and the Texas Compliance Histories database
28
are using ISO 14001 certification as a proxy for superior environmental management and
regulatory compliance.
Several studies have found that suppliers that are more geographically remote
from their buyers—and whose production processes are thus particularly costly for
buyers to observe—are more likely to adopt ISO 14001 (Christmann & Taylor, 2001;
King, Lenox, & Terlaak, Forthcoming), suggesting adoption can substitute for direct
monitoring by signaling superior management practices. Whether this signal is credible
depends on whether certification actually differentiates facilities with superior
compliance and environmental performance.
Spence (1973) described how signaling can resolve such information asymmetry
problems. In his classic signaling model, employers seek to hire highly productive
employees, but they could only observe jobs applicants’ educational attainment. The key
insight of this model is that for educational attainment to be a credible signal for ability,
the cost of signaling (pursuing education) must be cheaper for those with “high ability”
than those with “low ability”. For ISO 14001 adoption to provide a credible signal of
unobservable environmental performance or compliance, the difference in the cost of
(www.tceq.state.tx.us/nav/cec), such data are seldom provided in a manner that can easily
be used by firms to compare potential suppliers because these databases cannot be
searched or sorted by industry or products.
29
signaling must be sufficiently large to make it worthwhile (e.g., profitable) to facilities
with superior EMS’s but not to others.
These assumptions that are required to support a signaling story are plausible in
the case of ISO 14001. The most relevant costs to consider are the internal “soft” costs
such as preparing documentation and conducting training, since these represent the
largest proportion of total costs required to implement ISO 14001; fees associated with
hiring a third-party certifier and actual registration are relatively small (Kolk, 2000).22
The very fact that the ISO 14001 standard was designed based on best management
practices implies that organizations with comprehensive EMS’s should incur lower
adjustment costs to become certified.23 In contrast, if poor regulatory compliance and
22
Third-party certifiers charge $5,000 to $20,000 per facility (Dahlström et al., 2003;
Prakash, 2000: 116), depending on facility size and complexity of operations. Such costs
are even lower in countries such as Macau, Singapore and Japan that subsidize these fees
(Macau Productivity and Technology Transfer Centre, 2001; Singapore Ministry of the
Environment, 2002; Welch et al., 2002).
23
Similarly, empirical research has shown that EMS adoption costs are lower
among organizations with more management system experience (e.g., those that had
already implemented Total Quality Management and Just-in-Time inventory systems)
and those that have already implemented pollution prevention practices (Darnall &
Edwards, 2004).
30
environmental performance are due to the absence of a systematic EMS, then such firms
seeking to adopt ISO 14001 would have to invest significantly more to build an ISO
14001-compliant EMS from scratch. This supports the plausibility of the fundamental
assumption underlying a signaling story: that adoption is cheaper and only worthwhile for
those with superior environmental management practices. Given that these management
practices are correlated with environmental performance (King et al., Forthcoming),
companies with better regulatory compliance and/or environmental performance may be
more likely to adopt ISO 14001.
2.3.2 Why These Programs May Not Legitimately Distinguish Adopters: Free
Riding
There are also reasons to doubt that organizations with superior environmental
management practices face lower adoption costs. The standard would fail to provide a
credible signal of superior management capabilities if the costs required to adopt ISO
14001 were similar across facilities, regardless of the caliber of their ex ante
environmental management capabilities. Several plausible scenarios could elicit this
result.
First, suppose ISO 14001 were not based on prevailing “best practices” and
instead required entirely new activities for which superior ex ante environmental
management capabilities offered no complementarities. In this case, all facilities would
face similarly high adjustment costs, regardless of the quality or comprehensiveness of
these ex ante capabilities.
31
Two other scenarios may be more of a concern. If the standard actually requires
very little of adopters, all adopters could “simply implement a ‘paper EMS’…without
any real change in performance” (Andrews, Hutson, & Edwards, 2003). Alternatively, the
certifiers who are supposed to ensure that participants have fully adopted all aspects of
the standard may not be robust, as alleged by several critiques (O'Rourke, 2001, 2002;
Walgenbach, 2001). In either of these two cases, all facilities would face low adjustment
costs, regardless of the caliber of their ex ante environmental management capabilities.
If all facilities face the same costs of adoption regardless of their ex ante
management practices, free riding is likely to occur. Because many remain confident that
adopters exhibit better performance than non-adopters, organizations with little desire to
actually improve performance have an incentive to adopt and free ride on the reputation
gains associated with participation. This leads to a complete “unraveling” of the value of
the signal. For this reason, several voluntary environmental programs that lacked
independent verification requirements failed to disproportionately attract those with
superior environmental performance (King & Lenox, 2000; Rivera & de Leon, 2004). In
the context of ISO 14001, a recent survey found that organizations that decided to adopt
ISO 14001 were less likely to already have an EMS (Corbett & Russo, 2001), which
suggests that eventual adopters initially had inferior environmental management
practices. These arguments suggest that companies with inferior environmental
management practices may be just as likely to adopt ISO 14001.
32
2.3.3 The Potential of Voluntary Management Programs to Improve Performance
Adopting ISO 14001 and maintaining certification can facilitate organizational
learning and can create new incentives to assure legal compliance. As explained below,
each of these can lead organizations to improve their environmental performance and
compliance record.
Developing a sophisticated EMS to meet the ISO 14001 standard facilitates
organizational learning through the development of new routines and skills, new sources
of innovation, and new knowledge networks. ISO 14001 requires organizations to exhibit
a commitment to legal compliance, which is typically accomplished by developing and
reviewing a list of all relevant laws and regulations and then ensuring that all required
measures are incorporated in policies, procedures, and training programs (Toffel, 2000).
In addition, the standard requires organizations to develop a comprehensive inventory of
all the ways in which their activities may impact the environment (its “environmental
aspects”), to rank their significance in terms of potential environmental impacts, and to
develop procedures to control aspects with highly significant environmental impacts.
These requirements are often the first time organizations pursue such a comprehensive
approach and explicitly prioritize their management attention based on the significance of
their environmental impacts. As such, this process may enable organizations to better
target their management efforts to improve their environmental performance.
Internal auditing. ISO 14001 also requires an adopter to conduct and document
periodic audits to ensure that staff follow the organization’s environmental procedures,
improvement projects are on schedule to meet their objectives and targets, and training
33
schedules are being met. Finally, the standard’s requirement of periodic management
reviews of the EMS encourages companies to reconsider their priorities when their
activities or processes change.
Training. In addition, the standard requires organizations to train all employees
whose work may create significant impacts on the environment. This training often
enables employees to better identify pollution prevention opportunities and empowers
them to offer recommendations (Darnall, Gallagher, Andrews, & Amaral, 2000;
Godshall, 2000; Ochsner, 2000; Rondinelli & Vastag, 2000; Toffel, 2000; Wells &
Galbraith, 1999). Such training is typically provided to over half the facility’s employees,
with some companies training over 95% of their employees (Corbett & Luca, 2002).
Furthermore, cross-functional teams, commonly used to implement ISO 14001, can foster
systems-thinking and shared objectives to more efficiently transfer information and tacit
know-how within organizations (Kogut & Zander, 1992), including new ideas to prevent
waste and pollution across various production process stages (King, 1995, 1999; Russo &
Fouts, 1997).
Organizations can also improve their operational performance by leveraging
knowledge within their networks of rivals, buyers, and suppliers (Baum, Calabrese, &
Silverman, 2000; Dyer & Nobeoka, 2000). Organizations implementing ISO 14001 often
tap several new sources of technical and organizational knowledge, such as external
consultants hired to help implement ISO 14001 and certifiers (Prakash, 2000) who can
provide ideas to improve compliance and environmental performance. In addition, ISO1
14001 certification can put the organization on the “radar screen” of NGOs looking for
34
companies with whom to collaborate, facilitating preferential access to these NGOs’
expertise (Rondinelli & London, 2003).
Finally, adopting ISO 14001 may impose new consequences for regulatory noncompliance, which can motivate adopters to increase the efforts to assure they remain in
compliance. ISO 14001 requires organizations to include a commitment to compliance in
their environmental policy and requires documentation of compliance self-assessments
and third-party assessments. Because certifiers may inspect regulatory compliance
records, there is a heightened chance that non-compliance issues will be exposed to third
parties. In addition, non-compliance issues that remain unresolved can jeopardize a
facility’s ability to maintain its ISO 14001 certification status. In addition, many
companies are concerned about legal liability they may accrue by following the ISO
14001 standard’s requirements of documenting the results of their internal compliance
audits, including non-compliance issues and corrective measures taken (Delmas, 2000,
2002). The potential for legal liability increases potential risks associated with noncompliance, and can in turn increase incentives to assure compliance. Indeed, field
research found that ISO 14001 “helped to ‘indoctrinate’ employees so that the goal of
compliance became unquestioned” (Nash, Ehrenfeld, MacDonagh-Dumler, & Thorens,
2000: 10).
2.4. Sample and Measures
To implement our evaluation strategy, I assembled a comprehensive dataset for
the 1991-2003 period using data obtained from several publicly-available government
databases, several commercial databases, journal articles, and reports. I obtained
35
additional data by filing numerous Freedom of Information Act Requests of the United
States Environmental Protection Agency (US EPA). Specific data sources are described
in the measures section below.
2.4.1 Sample
My sample includes manufacturing facilities within the United States that have
reported emissions of toxic chemicals to the US EPA Toxic Release Inventory (TRI)
program. This program includes facilities that manufacture, import, process, or use any of
the listed substances in amounts greater than threshold quantities (typically 10,000 or
25,000 pounds) and have at least 10 full-time employees (US EPA, 1999a). I focus on the
five industries with the most ISO 14001 adopters as of 2001: chemicals, fabricated metal
products, industrial machinery and equipment, electrical and electronic equipment, and
transportation equipment (SIC Codes 28 and 34-37) (McGraw Hill, 2001). Facility
details were obtained from the US EPA’s TRI website and the US EPA’s Risk-Screening
Environmental Indicators 2.1 model.24 While 3,553 ISO 14001 certifications had been
issued in the US by the end of 2003, the International Organization for Standardization
reports that 2620 of these had been issued by 2002, the end of my sample period (ISO,
2004). However, the names and addresses of only 1106 adopters were listed in one of the
24
US EPA provides TRI data at http://www.epa.gov/tri and the Risk-Screening
Environmental Indicators model at http://www.epa.gov/opptintr/rsei/
36
most comprehensive databases available (QSU, 2002b).25 After supplementing this
source with data from state websites and finding that some certifications that
encompassed multiple facilities, I identified 858 TRI-reporting facilities that had adopted
ISO 14001 by 2002 (see Table 2.2).
2.4.2 Measures
Regulatory compliance. Because facilities in my sample are located in many
different states, I assess their compliance with federal environmental regulations. I focus
on the regulations associated with the US Resource Conservation and Recovery Act
(RCRA), which addresses the management of hazardous waste, because the industries in
my sample have been subjected to far more environmental inspections and enforcement
actions based on RCRA than any other federal environmental statute (US EPA, 1995b).
RCRA primarily imposes training and recordkeeping requirements on manufacturing
facilities that generate hazardous waste, including stipulations governing the storage and
labeling of hazardous wastes, training personnel in waste management and emergency
response procedures, and keeping detailed records (US EPA, 2003b). These procedural
requirements make RCRA a particularly relevant statute for evaluating the effects of ISO
14001 adoption, since both emphasize documentation, training, and procedures. I
measure regulatory compliance at the facility level as the annual number of RCRA
25
While the International Organization for Standardization reports annual country total
certifications, it does not collect or report the identities of adopters.
37
violations. I also gathered data on serious non-compliance events regarding any federal
environmental regulations that led to an EPA enforcement action against the facility.
Because very few facilities had more than one per year, I coded this as a dummy variable.
Data for both of these measures were obtained via Freedom of Information Act requests
of US EPA.
Environmental performance. I measure environmental performance using several
metrics based on toxic emissions data reported annually to the US EPA TRI program
from 1991 to 2002, the latest year for which data are available.26 A rare source of
uniformly reported, legally mandated facility-level environmental performance data, the
26
Growing concerns about the accuracy of the TRI dataset (e.g., Environmental
Integrity Project & Galveston-Houston Association for Smog Prevention, 2004; US EPA,
1998; US EPA, 2004: 14) suggests measurement error is possible. However,
measurement error only seriously biases my results if TRI data accuracy is correlated
with ISO 14001 adoption such that certified facilities come to systematically report
higher or lower emissions. An extensive literature review and discussions with TRI
program managers at US EPA (Sockabasin, 2004) and the Indiana Department of
Environmental Management (Teliha, 2004) revealed no evidence that adopters are
reporting more accurately. Supposing adopters became more accurate reporters, these
interviewees were agnostic about if this might lead them to systematically increase or
decrease their reported releases.
38
TRI dataset has been widely employed in the management literature (Gerde & Logsdon,
2001; Toffel & Marshall, 2004).27 My first measure, pounds of emissions, mimics how
the US EPA and the media rank the “dirtiness” of TRI reporters by first aggregating the
total pounds of toxic chemicals each facility released to air each year.28
My second metric estimates the health hazard posed by these emissions. To
accommodate the enormous variation in the toxicity of TRI chemicals, I employ
chemical-specific toxicity weights pertaining to inhalation exposure from the US EPA’s
Risk-Screening Environmental Indicators model (Toffel & Marshall, 2004; US EPA,
2002). To calculate the facility’s annual health hazard, I multiplied the pounds of each
chemical emitted to air by its inhalation toxicity factor, and took the sum of these
products. I also calculated each facility’s annual average emissions hazardousness by
dividing its health hazard by pounds of emissions.
ISO 14001 certification. I ascertained the identity of ISO 14001 adopters and
their certification year from the ISO 14001 Registered Company Directory North
27
Because the list of chemicals and reporting thresholds for some have changed
over time, I include only the “core group” of chemicals that have been required every
year since 1991 and whose reporting threshold remained constant.
28
For the empirical models, I added 100 before taking the log of pounds, health risk, and
emissions hazardousness to avoid overemphasizing differences between small amounts.
39
America (QSU, 2002b) and supplemented this with data from various state environmental
regulator websites.
Adoption determinants. To identify additional potential adoption determinants, I
surveyed the literature on the adoption of ISO 14001 and other voluntary environmental
programs. The comprehensiveness of an EMS should influence facilities’ likelihood of
adopting ISO 14001. In the documents facilities submit under the EPA’s TRI program,
facilities report the sources of pollution prevention or pollution control ideas they
implemented over the past year. A few of these sources imply that the facility has very
likely implemented a formal environmental management system: internal pollution
prevention audits, participative team management, or employee recommendations under
a formal company program (King et al., Forthcoming). I created a dummy variable
Environmental Management System coded one in years when a facility’s TRI
submissions included any of these sources.
Adoption may be driven by a management team’s fondness for the highly
structured and systematic “management systems” approach, or by the complementary
capabilities acquired when implementing a management system to meet the ISO 9000
Quality Management System Standard (Nash et al., 2000). I account for such influences
on the ISO 14001 adoption decision by incorporating a facility’s prior decision to adopt
ISO 9000. ISO 9000 was coded 1 in years when a facility had adopted this standard using
data from ISO 9000 Registered Company Directory North America (QSU, 2002a).
Adopting
bundles
of
proactive
environmental
practices
may
yield
complementarities (MacDonald, 2001; Robèrt, 2000; Rowland & Sheldon, 1999). For
40
example, adopting ISO 14001 and other voluntary programs may reinforce a company’s
environmental leadership strategy (Bansal & Hunter, 2003) since diverse stakeholders
may differ in their views about which programs credibly demonstrate environmental
commitment. To address this potential influence, I control for whether the facility has
adopted a government-initiated voluntary environmental program, the US EPA 33/50
program. 33/50 participants, a dummy variable, pledged voluntary reductions in a small
set of priority pollutants by 1995. A list of 33/50 participants was obtained from US EPA.
Empirical research has found that companies are more likely to adopt
environmental management practices and ISO 14001 when they face more institutional
pressure from a host of stakeholders including legislators and regulators, local
communities, environmental activist organizations, and buyers.29 As such, I included
several variables that measure institutional pressures from these stakeholders. Regulatory
29
Prior literature describes a wide range of stakeholders can influence companies’
decision to adopt environmental management practices (Delmas & Toffel, 2004)
including legislators and regulators (Carraro, Katsoulacos, & Xepapadeas, 1996; Delmas,
2002; Majumdar & Marcus, 2001; Rugman & Verbeke, 1998), local communities
(Florida & Davison, 2001; Henriques & Sadorsky, 1996; Maxwell, Lyon, & Hackett,
2000; Raines, 2002; Vidovic & Khanna, 2003), environmental activist organizations
(Baron, 2003; Lawrence & Morell, 1995), and buyers (Christmann & Taylor, 2001;
Henriques & Sadorsky, 1996; Khanna & Anton, 2002).
41
pressures were measured using averages of one and two year lagged RCRA violations
and EPA enforcement actions related to serious violations of any federal environmental
regulation.
King et al. (Forthcoming) argue that facilities are more likely to adopt ISO 14001
if they release effluent with toxic chemicals to publicly-owned treatment facilities
(POTWs)—government-owned wastewater treatment facilities that process effluent
received via pipes or sewers—because the standard might help them respond to pressure
from POTWs to make effluent flows more predictable and less toxic. As such, I include a
dummy variable to denote years when a facility had any releases of toxic chemicals to a
POTW.
To account for important differences in regulatory pressure and compliance costs
between the states and EPA 10 Regions, I used an indicator of state environmental policy
comprehensiveness index that measures the extent to which each state has implemented
environmental policies to address 50 pollution, waste, and land use issues (Hall & Kerr,
1991) used by others for similar purposes (Welch, Mazur, & Bretschneider, 2000). In
addition, I employed a measure of states’ relative compliance costs from Levinson
(1996).
I also controlled for the potential influence of community socio-economic status
by using two community demographic characteristics, income and education, that are
widely considered to be correlated with the desire for more stringent environmental
protection (Elliott, Sheldon, & Regens, 1997; Klineberg, McKeever, & Rothenbach,
1998). Median household income for 1999 and the percentage attended college among
42
those 25 years and older were obtained for every zip code using 2000 Decennial Survey
data from the US Census Bureau. Averages were calculated for all zip codes within a
five-mile ring around each facility.
Production and facility size. Facility production volumes are strongly linked to
emissions, but are difficult to acquire. As a proxy for production, I create a production
index based on two variables: (1) facility employment, obtained from Dun & Bradstreet
for a single year for each facility in my sample, which I refer to as that facility’s baseline
year, and (2) annual facility production ratios, which is the ratio of production in the
current year to the production in the prior year, obtained from the TRI dataset.30 I
calculate production index in three steps: (1) I set production index equal to facility
employment for the baseline year; (2) for each year following the baseline year, I
calculate the production index by multiplying the prior year’s production index by the
current year’s production ratio; (3) for each year preceding the baseline year, I calculate
the production index by dividing the subsequent year’s production index by the
subsequent year’s production ratio.31 This enables me to control for production in the
models that include emissions levels.
30
In the models, I employ log production index. To avoid overemphasizing differences
between small amounts, I added 100 before taking the log.
31
For example, suppose production ratios during 1998-2002 are 1.2, 1.3, 0.8, and 2,
respectively and facility employment is 100 in 2000. First, I set production index equal to
43
Summary statistics and correlations are provided in Tables 2.2 are 2.3.
2.5. Methods & results
2.5.1 Selection Analysis
The selection analysis seeks to discern whether the performance of adopters
differed from the performance of non-adopters in the years prior to adoption. I create a
dummy variable coded one only for adopters in the year they become certified, and I
employ a probit model. The key variables to detect a selection effect is lagged
performance for each of the five performance metrics.32 In each specification, I include
industry dummies (2-digit SIC Code), year dummies, and EPA Region dummies. The
year dummies control for events in a given year that might impact the emissions of all
facilities, such as changes in federal regulations, federal enforcement priorities, or the
introduction of new technologies. The EPA Region dummies control for variations across
EPA regional enforcement strategies. I also include log production index to account for
100 in 2000. Second, I calculate production index for subsequent years as follows: for
2001, 100 × 0.8 = 80; then for 2002: 80 × 2 = 160. Third, I calculate production index for
years prior to the baseline year: for 1999, 100 ÷ 1.3 = 76.9; then for 1998, 76.9 ÷ 1.2 =
64.1.
32
For each performance metric, I use the average of the 1- and 2-year lags because this
prevents many observations from dropping out of the sample, which contains many
missing values.
44
facility size and changes in production. I include 1991-2002 data, but drop adopters after
their certification year to avoid confounding this selection analysis with any potential
impacts of certification (treatment effects).
I also evaluated whether the two groups’ performance trends differed during the
pre-adoption period. For pollution (pounds) and health hazard, I estimated the following
model individually for each facility i:
yit = αi + β1i t + β2i log(PIit) + εit
where t represents the year of the observation, PIit is the production index, and εit is an
error term assumed to be distributed randomly across facilities. β1i estimates the temporal
trend for each facility. I exclude adopters from the sample starting two years before their
certification year to ensure the estimates are not confounded by efforts associated with
preparing to adopt ISO 14001. For the compliance and hazardousness variables, I omit
production index (PIit). After generating β1i estimates for each facility and for each
performance measure, I conducted t-tests to determine whether the temporal trends
differed between the eventual adopters and the non-adopters. To reduce the influence of
outliers, I omit facilities whose slope values fall outside the 1st to 99th percentiles of the
distribution.
2.5.2 Selection Results
Tables 2.4 and 2.5 present the results of the selection analysis. In the probit
analysis, I report clustered standard errors by facility to account for non-independence
among observations from the same facility. Controlling for facility size, there is no
evidence that the number of RCRA violations or enforcement actions encouraged
45
adoption (Table 2.4 Column 1). However, non-adopters were experiencing a growing
number of enforcement actions whereas the eventual adopters were not (p<0.01, Table
2.5 Row 1). While both groups were improving their compliance with RCRA, eventual
adopters were improving faster, but the difference is not statistically significant (p=0.25,
Table 2.5 Row 2).
Turning to environmental performance, Columns 2 and 3 of Table 2.4 illustrate
that facilities that emitted more pounds of toxic chemicals were more likely to adopt
(p<0.01) even after controlling for facility size (p<0.05). However, comparing trends,
pounds of toxic chemicals emissions were decreasing among eventual adopters, while
they were increasing among non-adopters, and the difference is statistically significant
(p<0.01, Table 2.5 Row 3).
Facilities that emitted more dangerous chemicals (more hazardous per pound)
were more likely to adopt (p<0.01, Table 2.4 Column 4). Indeed, both groups emissions
were increasing in hazardousness prior to adoption, with eventual adopters’ increasing
faster (p<0.01, Table 2.5 Row 4).
Adopters’ emissions were posing greater health hazard (harm-weighted
emissions) to their communities than non-adopters (p<0.01, Table 2.4 Column 5),
although there is no evidence of any distinction once size is controlled for (Table 2.4
Column 6). Furthermore, while the health hazard from the emissions of both groups was
increasing during the pre-adoption period, those from non-adopters were increasing faster
than eventual adopters (p=0.10).
46
In some ways, ISO 14001 adopters are “dirtier” firms when they become
certified: not only were they emitting emit more pounds of toxic chemicals, but these
chemicals were more hazardous than those emitted by non-adopters. However, during
this pre-certification period, adopters were improving their performance more rapidly
than non-adopters across several dimensions: (1) enforcement actions; (2) RCRA
violations; (3) pounds of emissions, and (4) health hazards they were imposing on their
communities.33
2.5.3 Treatment Analysis
To evaluate whether the adoption of ISO 14001 influences regulatory compliance
or emissions performance, I use a difference-in-differences approach, which is widely
employed in the program evaluation literature. This approach uses the control group’s
performance during the post-treatment period as the counterfactual for how the treatment
group would have performed if it had not received the treatment. I estimate the following
equation:
yit = αi + β1τ Di τit + β2τ τit + β3t λt + β4log(PIit) + εit
33
Only along one dimension, the average hazardousness per pound of toxic chemical
emitted, were eventual adopters worse and worsening, compared to the non-adopters.
However, this metric is not particularly important in its own right from an environmental
or health perspective.
47
yit represents each of the five outcome variables. Di is an adoption group dummy coded 1
(in every year) for facilities that adopted ISO 14001 sometime during the sample period.
τit is a set of certification year lead and lag dummies (4-years-until-certification through
4-years-since-certification) where non-adopters are coded according to the certification
year of the adopter to which they were matched. λt and PIit represent year dummies and
the production index. The production index is omitted when estimating compliance or
hazardousness, since there is no theoretical reason to expect annual production changes to
influence these outcomes. To estimate these models, I employ a conditional fixed effects
logit model for enforcement actions, a conditional fixed effects negative binomial model
for RCRA violations,34 and ordinary least squares (OLS) models with plant fixed effects
for the three emissions variables.
34
Three tests suggest the negative binomial is more appropriate than the Poisson model
to analyze the violations variable for the subsample of facilities subject to RCRA. First,
the ratio of the variance to the mean of 6.0 ( σ̂ 2 =2.95, X =0.49) suggests overdispersion.
Second, after running the Poisson model, the extreme significance of the goodness-of-fit
chi squared (χ2=180893, p<0.00) indicates the Poisson regression model is inappropriate.
Third, I ran a negative binomial model and conducted likelihood-ratio tests that the
overdispersion parameter alpha equals zero. The high chi squared value (χ2 >10000,
p<0.001) implies that the probability of observing these data conditional on alpha being
zero is virtually nil.
48
Unbiased estimates from a difference-in-differences approach requires two
assumptions about the similarity between the treatment and control groups: (1) that both
groups would respond similarly to treatment (homogeneous response), and (2) that the
control group’s performance trend serves as the counterfactual of the treatment group
(i.e., if the adopters had not adopted, their performance would have mimicked the control
group’s). The selection analysis conducted above already showed that adopters’ and nonadopters’ differed in some performance outcomes during the pre-adoption period. The
two groups are likely to differ in other observable ways as well. For example, larger
facilities may have more resources to devote to the adoption of ISO 14001 and to
pollution control technologies that affect environmental performance.
To eliminate this source of potential bias, I used propensity score matching to
identify a quasi-control group of non-adopters whose adoption covariates and pre-period
outcomes are statistically indistinguishable from those of the treatment group.
Combining a difference-in-differences approach with propensity score matching is
particularly appropriate in evaluating programs in an environment of important
underlying time trends (Athey & Imbens, 2002). This is exactly the case in my empirical
context since, on average, facilities reduced their pounds of TRI emissions throughout the
1990s.35
35
For example, the US EPA announced the 2003 TRI data by noting that “the amount of
toxic chemicals released into the environment by reporting facilities continues to decline,
49
Propensity score matching. Matching is a widely used approach to construct a
quasi-control group based on similar characteristics as the treatment group (Heckman,
Ichimura, & Todd, 1998). Matching on the propensity score—the probability of receiving
the treatment conditional on covariates—is as valid as matching on a series of individual
covariates individual covariates
(Rosenbaum & Rubin, 1983). The identifying
assumption is that the assignment to the treatment group is associated only with
observable “pre-period” variables, and that all remaining variation across the groups is
random.36
Propensity score matching methods, when used to evaluate job training programs,
have performed well in replicating the results of randomized experiments under three
conditions: (1) the same data sources are used for participants and non-participants; (2)
an extensive set of covariates are employed the program-participation model that is used
to estimate propensity scores; and (3) participants are matched to non-participants in the
same local labor market (Smith & Todd, 2005). In addition, substantial bias can result if:
(4) controls are included whose propensity scores are off the support of the participants’
propensity scores; (5) the distributions of the participants and non-participants’
with total reductions of 42 percent since 1998 and a six percent decrease from 2002 to
2003” (US Environmental Protection Agency, 2005).
36
This assumption is often referred to as the “ignorable treatment assignment” or
“selection on observables.”
50
propensity scores differ; or (6) unobservable factors influence both participation and
outcomes (Heckman, Ichimura, & Todd, 1997)
I address these six potential sources of bias as follows. First, I use the identical
data sources for all facilities (participants and non-participants). Second, I gather an
extensive set of adoption covariates based on a comprehensive literature review. Third, I
ensure that participants and non-participants operate within the same markets by
matching participants to non-participants within the same industry. I address the fourth
and fifth concerns by implementing nearest neighbor matching with a “caliper”
restriction, which prevents matches of nearest neighbors beyond a fixed threshold. The
sixth concern addresses selection on unobservables. In the context of ISO 14001, it is
possible that facilities with an “environmentalist” culture (which I do not observe in my
data) may have managers who are both more likely to insist upon strong environmental
performance and be more prone to adopt ISO 14001. To the extent that such
unobservable factors are fixed over time during the sample period, I address this concern
by employing a differences-in-differences approach that includes facility fixed effects.
Facility fixed effects also addresses this source of bias for unobservable influences that
are highly correlated with the observable factors I include when estimating propensity
scores.37
37
Instrumental variable models control for the possibility that such unobservable
differences may vary over time. This approach has been employed in some program
51
I implement propensity score matching in three steps. First, I estimate propensity
scores by estimating a probit model for adoption status during 1996-2002, omitting
adopters after their adoption year. This is a more comprehensive version of the probit
used in selection analysis that incorporates adoption determinants in addition to the
lagged outcome variables. After surveying the literature on the adoption of ISO 14001
and other voluntary environmental programs, I gathered data on a comprehensive set of
variable that constitute likely adoption determinants. The primary objective of this
process is to identify a specification that balances the pre-program covariates between the
evaluations that examined the performance implications of voluntary management
programs (e.g., Khanna & Damon, 1999; Rivera, 2002; Rivera & de Leon, 2004; Welch
et al., 2000). However, instrumental variable models such as the Heckman two-step
process require a valid instrument to avoid identification entirely relying on the
assumption that the error terms of the selection and performance equations are distributed
jointly normal. Little & Rubin (1987: 230) note that an exclusion restriction is required
“for the [Heckman] method to work in practice,” a conclusion bolstered more recently by
results of Monte-Carlo simulations (Puhani, 2000). For the current study, the instrumental
variable approach would require identifying a variable that is correlated with the decision
to adopt ISO 14001 and has no independent influence on subsequent performance. Valid
instruments are notoriously difficult to identify, and I could not identify a credible
instrument that would enable me to use this alternative method.
52
treatment and comparison groups conditional on the propensity score (Dehejia, 2005).
However, to increase the likelihood that the matched control group’s pre-period
performance would be similar to the adoption group’s (Barbera & Lyona, 1996; Eichler
& Lechner, 2002), I also included lagged outcome variables. Because propensity score
methods are more reliable for estimating treatment effects when more than 1 year of pretreatment outcome data are available (Dehejia & Wahba, 1999), I incorporate two years
of lagged outcome data in the probit specification. The probit results of the full adoption
model are provided in Table 2.6.
In addition to the probit coefficients, I report the change in the probability in
adoption given an infinitesimal change in the continuous independent variables or
discrete changes in the dummy variable—when all independent variables are at their
mean values. I report standard errors clustered by facility to account for nonindependence among observations from the same facility. Briefly, adoption is more likely
in facilities that provided evidence of an environmental management system, were EPA
33/50 participants, adopters of the ISO 9000 Quality Management System standard,
released toxic chemicals in their wastewater to POTWs, were located in states with
higher compliance costs, and controlling for all other covariates, released more pounds of
toxic emissions.
53
Second, I generated predicted values from the probit model and implemented
nearest-neighbor matching without replacement.38 To construct the matched sample, for
each adopter, I identified the ten non-adopters with the most similar propensity scores in
the year the adopter became certified. I refer to this as the “10 overall-nearest-neighbors
matched sample.”
The nearest-neighbor matching method offers the advantage of selecting controls
for every treatment observation. However, by requiring a set number of matches for each
adopter, the nearest neighbor approach can result in poor matches for treatment
observations that have few similar control observations (Becker & Ichino, 2002).
Therefore, I also restricted matches to the subset of the nearest neighbors that were
sufficiently “near” the adopter by imposing a caliper limit of 0.005.39 This restriction
incorporates fewer non-participants when fewer close matches are available. This process
yielded a matched sample of 753 adopters and 5830 non-adopters, for an average of 7.7
matches per adopter.
38
The matching process was implemented in Stata 8.2 using PSMATCH2 (Leuven &
Sianesi, 2003).
39
Selecting the particular caliper distance involves a tradeoff: smaller distances improve
the quality of matches but risks having no matches for some treatment observations
(Becker & Ichino, 2002).
54
Third, I assessed the similarity of the non-adopters and adopters in the matched
control groups in several ways. Sianesi (2004) suggests comparing the pseudo-R2 value
from the initial model used to generate propensity scores (the entire sample) to the
pseudo-R2 value from re-estimating the model on the matched sample. Since the pseudoR2 indicates how well the regressors explain the participation probability, the matching
process should results in a substantial decline in the pseudo-R2 value. In my case, the
pseudo-R2 value is substantially reduced from 0.12 (entire sample) to 0.05 (matched
sample). I also compared the group means during the pre-adoption period. T-test results
confirmed that the differences between group means for most covariates and lagged
performance levels were indistinguishable at conventional significance levels (Table 2.7
Columns 4-6), in sharp contrast to the many significant differences between the adopters
and all non-adopters in the entire sample (Columns 1-3). A third way to assess the extent
to which the matching process reduces observable differences is to compare match the
standardized bias for each covariate before matching to the standardized bias after
matching (Rosenbaum & Rubin, 1985). Standardized bias is calculated as the difference
between group means divided by the square root of the average variance across the two
groups.40 The standardization allows comparisons between covariates, as well as before
40
Standardized Bias = 100 ∗
(X1 − X 0 )
0.5 ∗ (V1 (X ) + V0 (X ))
55
and after matching. The median (mean) standardized bias across all covariates and lagged
outcome variables before matching is 3% (10%), whereas this declines to 2% (2%) in the
matched sample (see Table 2.8).
Finally, I assessed whether the outcome trends of the groups were similar in the
years before the match year.41 I ran the same simple temporal trend models that I
employed in the selection analysis to distinguish whether the two groups in the matched
samples had similar trends prior to adoption. To avoid confounding the analysis with any
adoption effects, I exclude observations from adopters or non-adopters starting one year
before the match year. After estimating each facility’s pre-match trend for all five
outcomes, I compared two groups trends using t-tests. I find that before the match year,
the trends between the matched non-adopters and matched adopters were statistically
indistinguishable across all five outcomes (see Table 2.9, Columns 1-3). This bolsters the
plausibility of the identifying assumption of the difference-in-differences method: that the
matched control group’s performance serves as a valid counterfactual for the treatment
where X1 (V1) represents the mean (variance) in the treatment group and X 0 (V0) is the
analogue for the control group. The standardized difference after matching uses the same
approach but employs the means and variances of the matched sample.
41
For an adopter, the “match year” refers to its certification year. For a non-adopter, the
“match year” refers to the certification year of adopter to which the non-adopter was
matched.
56
group’s.42 Overall, the matching process yielded a quasi-control group that is quite
similar to the adoption group.
2.5.4 Treatment Analysis Results
“Overall matched” rows in Table 2.10 present the results of the difference-indifference models described above. Because it often takes a year or two of preparatory
measures to upgrade a facility’s environmental management practices to develop a
sufficiently comprehensive EMS to meet ISO 14001 requirements, I looked for evidence
of a preparatory treatment effect. I examined the difference-in-differences estimates
pertaining to one or two years before certification and the certification year itself. In the
model examining pounds of emissions (Row E), the statistically significant coefficients
(p<0.05) suggest that adopters reduced their emissions by 25% to 32% as they prepared
for certification.43 I find no evidence of preparatory treatment effects for any of the other
four outcomes, since in each of those models, neither of the two interaction terms had
coefficients that were statistically significant (nor did the sum of these coefficients differ
from zero at conventional significance levels).
42
Additional statistical tests to assess whether the two groups’ pre-period performance
trends differed are include in the treatment model described below.
43
Recall that I earlier confirmed in Row 3 of Table 9 that the two groups’ trends in the
years leading up to this period were indistinguishable; further evidence is provided by the
insignificant coefficients on the interaction terms in Row A of Table 10
57
I also looked for evidence of a longer-term treatment effect by examining the
coefficients on the interaction terms subsequent to certification. Again, the most
interesting results are in the model addressing pounds of emissions (Row E). Adopters
reduced their pounds of emissions by 21% and 29% in the first two years after
certification, and the overall difference between the groups in the 4 years after
certification is almost significant at conventional levels (p=0.13). There is stronger
evidence that the emissions of adopters grew more hazardous per pound (an F-test
suggests we can reject at p=0.06 the null that the sum of the five post-certification
estimates is zero), perhaps suggesting they focused on concentrated the mass of their
emissions. Combining these outcomes (pounds of emissions weighted by harm) into the
health risk metric, I find that these two trends essentially cancel each other out: there is
no evidence that adopters differed from non-adopters in the health risk they imposed on
their communities (an F-test that the sum of these five coefficients differed from zero
yielded p=0.66).
Similarly, adopters and non-adopters do not significantly differ in the number of
enforcement actions or RCRA violations for which they were cited after the event year.
There is, however, some evidence that adopters are cited with more enforcement actions
(an F-test of the sum of the five-post certification estimates yielded p=0.16). Interpreting
this result requires caution for two reasons. First, the ISO 14001 process may have
brought violations to the attention of management who may have subsequently informed
regulators, or regulatory inspectors may have been provided easier access to documented
violations (and presumably to their corresponding corrective actions). Second, because
58
enforcement actions often result from multi-year investigations, finding more
enforcement actions several years after adoption may, instead of indicating worsening
compliance, could suggest that facilities knew about these investigations and adopted ISO
14001 in response, trying to mitigate subsequent penalties or to rebuild their image.
Clearly, more research is needed to investigate these alternative explanations.
Selection out of the sample. Selection out of the sample could bias the results of
the treatment analysis. In the matched sample used in the treatment analysis, 11%
(81/753) of the adopters stopped reporting TRI data (and thus dropped out of the sample)
before 2002, the end of the sample period. In contrast, 32% of non-adopters (1887 of
5830) dropped out.
If worse emissions trends or violations trends led to the non-
adopters’ inferior survival rate, then their exiting the sample would have meant that the
remaining non-adopters’ outcomes would have yielded underestimates of what the nonadopters’ annual average emissions would have been had none dropped out. This, in turn,
would have resulted in biased underestimates of the treatment effect because the
adopters’ outcomes should have been compared to a control group whose outcomes were
worse than those used in the analysis. To detect evidence of this bias, I examined whether
non-survivors’ outcome trends were indeed worse than survivors’. I developed several
measures to test this. First, I created a dummy variable coded 1 if the facility did not
survive in 2002, and 0 if it did. I determined each facility’s survival year as the latest year
that it reported any TRI emissions or production ratios, had any RCRA inspections or
violations, had any enforcement actions, or was certified to ISO 14001. Second, I
calculated each facility’s trend of pounds of TRI emissions during the final 4 years for
59
which the facility reported TRI data during the sample period. I employed a ratio similar
to percent change but that is robust to outliers as follows. I divided (a) the difference
between the average of 2- and 3-years lags and the average of the final and 1-year lags by
(b) the average during this 4-year period. To further reduce the influence of outliers, I
omitted values outside the 5th to 95 percentiles. Third, I calculated each facility’s 4-year
RCRA violation trend over this same period. I measured violation trends as the average
number of violations in the facility’s last two years minus the average number of
violations in the two preceding years.44 Fourth, I created two variables to control for
outcome levels: pounds of TRI emissions and number of RCRA violations, each
averaged over 2-years and 3-years before the facility’s final year.
I ran a probit
regression of the dummy non-survival variable on the outcome trends and levels
variables, and included a series of dummy variables to control for industries (2-digit SIC
Codes) and EPA Regions. I re-ran this probit using three different subsamples: (1) all
non-adopters within the matched sample; (2) only the non-adopters that survived at least
two years after the match year; and (3) the entire matched sample (including adopters).
Each of these three regressions yielded the same substantive result: in no case were
either coefficient on the trend variables positive and statistically significant. As such,
these results provide no evidence that emissions trends or violation trends were
44
Because the majority of facilities had zero violations during many or all of their final 4
years, a percent change metric would have resulted in an abundance of extreme values.
60
associated with facility death, and thus provide no evidence that selection out of the
sample resulted in biased estimates of treatment effects.
Robustness tests. I conduct several tests to assess the robustness of the treatment
effect results. First, I tested the robustness of my estimates to an alternative matching
approach. I created an alternative matched sample by conditioning matches on industry.
For each adopter, I identified the ten non-adopters in the same industry (2-digit SIC
Code) with the most similar propensity scores in the year the adopter became certified,
using propensity scores generated. I refer to this matched sample as “10 nearestneighbors-within-industry.” This process yielded a matched sample with 749 adopters
and 5691 non-adopters, for an average of 7.6 matches per adopter. Before the match year,
the adopters and non-adopters in this matched sample appear quite similar in terms of
covariates and lagged performance levels (Table 2.7 Columns 7-9) and outcome trends of
(Table 2.9 Columns 4-6). The mean and median standardized bias of 2% and 1%,
respectively, are quite low, similar to the original matched sample (Table 2.8 Column 3).
The difference-in-difference results for the five outcome variables using this matched
sample are presented in the “within industry” rows in Table 2.10. As with the main
results, I find evidence that adopters were releasing 18% to 23% less emissions than nonadopters in the two years leading to certification and the certification year itself (p<0.10;
see Row F). Again, I find no evidence that adopters differed from non-adopters in any of
the other four outcomes during this period. Subsequent to adoption, I continue to find
evidence that the chemicals the adopters were emitting were more hazardousness (per
pound) than those emitted by non-adopters (an F-test of the null that the five post61
certification estimates sum to zero is rejected at p=0.08). As in the main analysis,
adopters subsequently reduced their pounds of emissions by 15% to 24% in the first two
after certification (Row E), although the overall difference between adopters and nonadopters in the years after certification is no longer statistically significant.
Next, I used an alternative specification for the treatment analysis by including all
time-varying adoption covariates that were used to estimate propensity scores:
Environmental Management System, ISO 9000, and any releases of toxic chemicals to a
POTW. These are dummy variables lagged one year. I run these regressions on a sub-
sample of data beginning in 1996 (the first adoption year) because these covariates are
coded in my dataset starting that year. The results of the difference-in-differences
estimates of this alternative specification are presented in Table 2A.1. The emissions
results from the main analysis were unchanged: I find evidence that adopters reduced
emissions as they approached certification (p=0.01) and in the first few years subsequent
to certification (an F-test of the null the sum of the coefficients is zero is rejected p=0.10).
In this specification, however, I find no evidence that adopters increased the
hazardousness of their emissions. I continue to find some evidence that adopters were
subsequently cited with more enforcement actions (an F-test of the sum of the
coefficients equals zero can be rejected at p=0.14), but no evidence of preparatory or
durable treatment effects for any of the other outcome variables.
Dehejia (2005) emphasizes the importance of examining the sensitivity of
estimated treatment effects to slight changes in the propensity score specification. As
such, I employed an alternative propensity score specification where I include the
62
average of 1 and 2 year lagged production index in the full adoption model to generate
alternative propensity scores. The production index has many missing values, which
causes many adopters to drop out of the sample and reduces statistical power.
Nonetheless, higher production growth is a plausible determinant of adoption. The results
of this probit model are provided in Table 2A.2. I used the predicted values from this
model to generate alternative propensity scores, and then repeated the two nearestneighbor matching techniques described above (overall 10 nearest, and 10 nearest-withinindustry). The similarity adopters and non-adopters within these alternative matched
samples was comparable to the original two matched samples in terms of adoption
covariates and outcome levels (Table 2A.3). The difference-in-difference models yield
similar results as those that proved robust in the main analysis: adopters emit fewer
pounds of toxic emissions in the years leading up to certification (Rows E and F of Table
2A.4), while subsequent to certification their emissions become more hazardous per
pound (Rows G and H of Table 2A.4).
Extensions. The above results provide no evidence of an average treatment effect
on the treated. However, treatment effects could vary by adoption propensity. For
example, perhaps only facilities with high propensity scores exhibit a positive treatment
effect. To investigate this, I repeated the treatment analysis on two sub-samples: facilities
whose propensity scores were among the lower half of adopters’ propensity scores, and
those in the upper half. The results are presented in Table 2.11. One key insight from
these regressions is that the adopters who significantly reduced their emissions
(pounds)—both in the two years leading up to adoption and subsequent to adoption—are
63
concentrated in the subset of matched facilities with the higher propensity scores; there is
no evidence that adopters in the subset of matched facilities with lower propensity scores
performed any differently from non-adopters.
Treatment effects could also vary between early and late adopters. Others have
found that early adopters of other management programs (e.g., quality management
programs, financial reporting standards, and long-term incentive plans) are more likely to
adopt to bolster efficiency, whereas later adopters are more likely to adopt to acquire
legitimacy (Mezias, 1990; Westphal, Gulati, & Shortell, 1997; Westphal & Zajac, 1994).
Jiang & Bansal (2003) suggest a similar distinction may exist between early and late
adopters of ISO 14001. To test this, I ran the treatment effects model on two subsamples: (1) “early adopters” who became certified during 1996-1999, and their matches,
and (2) “late adopters” who became certified during 2000-2002, and their matches. Table
2.12 presents the results. One key insight from these regressions is that the late adopters,
but not the early adopters, significantly reduced their emissions (pounds) in the two years
leading up to adoption and subsequent to adoption. This conflicts with prior theory about
early versus late adopters, and elicits the need for more research to understand why these
outcomes apparently differ between early and late adopters.
2.6. Discussion and Conclusions
This study evaluated whether adhering to a voluntary management program that
requires periodic third-party verification serves as a credible signal of superior
environmental management practices. Before certification, eventual adopters were
improving their compliance faster than non-adopters, both in terms of enforcement
64
actions and RCRA violations. I examined facilities’ annual pounds of toxic emissions
because the media and EPA often use this metric to rank the “dirtiest” companies. Prior
to adoption, while eventual adopters emitted more pounds of toxic chemicals than nonadopters, they were reducing these emissions while non-adopters were not (the difference
in trends is highly statistically significant). After weighting these emissions by their
potential to harm human health, I found that prior to adoption, the health harm posed by
non-adopters was increasing significantly faster than that posed by eventual adopters.
Despite these important differences at the time of adoption, when comparing
adopters to a matched control group, I found some evidence that adopters’ performance
continued to outpace non-adopters in terms of reducing the quantity of toxic emissions.
However, by looking beyond the simplistic metric of total pounds of toxic emissions, I
found that adopters’ emissions were increasing in hazardousness and as a result the
overall health risk they pose on their communities was unchanged. I also found no
evidence that compliance improved as facilities upgraded their environmental
management systems to meet the ISO 14001 standard, or subsequently operated
according to the standard. This latter finding is consistent with the results of a recent
study in the United Kingdom that found that facilities that adopted ISO 14001 (or the
similar
European
Eco-Management
Audit
Scheme,
EMAS)
had
statistically
indistinguishable numbers of incidents, complaints, and non-compliance events as those
without an externally verified environmental management system (Dahlström et al.,
2003).
65
Taken as a whole, these results demonstrate that a voluntary management
program with a robust enforcement mechanism can distinguish organizations based on
their difficult-to-observe management practices, albeit through only some mechanisms
and along only some performance metrics. In this study, I found evidence that this
differentiation occurs prior to adoption in two ways: (1) through a positive selection
effect; and (2) in preparations upgrading management practices to meet the standard. I
also found some evidence that adopters subsequently reduce their pollution intensity by
reducing their pounds of emissions, but this comes at the price of their emissions
becoming more hazardous per pound. As a whole, these findings represent a dramatic
departure from prior studies that found no evidence that superior performers
disproportionately adopted voluntary management programs that lacked or had only weak
enforcement mechanisms.
Several stakeholders can benefit knowing whether voluntary management
programs such as ISO 14001 distinguish participants because a disproportionate number
of adopters already possess superior ex ante performance (a positive selection effect), or
because such programs elicit improvement (a treatment effect). For example, consider a
marketing department seeking preferential access to buyers who use program
participation as a screening device. Such a department might favor adopting programs
whose credibility rides on a positive selection effect because adoption would immediately
distinguish the firm. On the other hand, managers of quality, or environment, health and
safety who are interested in improving their organization’s management practices and
performance may prefer adopting programs with a demonstrated treatment effect. Given
66
the current findings of a positive selection effect and a preparatory treatment effect, ISO
14001 appears more suitable for marketing purposes than functional managers seeking to
use the program to drive improvements.
This distinction between selection and treatment effects is also important to
buyers and regulators contemplating when to consider participants as superior to nonparticipants. If participants’ superior performance derives from a treatment effect, buyers
and regulators would need to allow a time lag to pass after adoption to enable participants
to attain superior performance. Since ISO 14001 was found to distinguish participants via
a selection effect, no such delay is needed because adopters were already exhibiting
superior ex ante performance.
Why might adopting ISO 14001 fail to improve regulatory compliance or reduce
health risks associated with toxic emissions? Perhaps third-party certifiers are not
assessing adopters stringently enough to ensure their compliance assurance processes are
sufficiently robust or their environmental prioritization schemes are appropriate. More
fundamentally, ISO 14001 might not be well designed to elicit performance
improvement.45 Each of these considerations is discussed below.
45
In addition, ISO 14001 might elicit environmental and compliance benefits other than
those assessed in this study, a point discussed below as a recommended avenue for future
research. Another possibility is that not enough time has passed for adoption benefits to
become apparent. While the time required to implement or to improve an environmental
67
2.6.1 Is Independent Verification Inadequate?
The verification mechanism of third-party assessment is a key distinction between
the ISO 14001 standard and the voluntary management programs others have evaluated.
However, the competence and comprehensiveness of some certifiers have been criticized
(O'Rourke, 2001, 2002), and my field research revealed substantial variation among even
senior certifiers’ environmental knowledge and assessment stringency. MacLean (2004b:
13) notes that many companies are trying to “game the system” by choosing assessors
with the most lenient interpretations of ISO 14001 requirements, a problem he describes
as “brewing for years.” Furthermore, he argues that even ISO 14001 adopters need to
improve their “core risk analysis process that examines past, present and future risks
rigorously” and develop “a robust set of metrics” (MacLean, 2004a).
management system to meet the ISO 14001 standard may range from a few months to a
year or two, tangible improvements in compliance or environmental performance may
take several years to materialize. For example, improvements in compliance may only be
discernable after a facility is subjected to several regulatory inspections, but regulators
inspect many facilities only once every few years. In addition, even if adoption leads
facilities to ideas and projects that will reduce their emissions, implementation may be
delayed to coincide with their next major plant renovation or production change. Others
have observed that environmental benefits emerge years after ISO 14001 adoption
(Babakri et al., 2004: 636).
68
Because ISO 14001 already requires these steps, this charge suggests cases where
certifiers have not ensured that adopters have comprehensively prioritized their
environmental aspects. As a result, these adopters may not have established procedures
required by ISO 14001 that could help them target their management resources toward
opportunities with the greatest potential for performance improvement. MacLean’s call
for better metrics implies that some ISO 14001 adopters have not taken seriously this
fundamental step required to meet the standards’ continuous improvement requirement.
As a result, uneven rigor among certifiers may be allowing poor performers to become
certified, and may not be ensuring that adopters can fully benefit from meeting all
elements of ISO 14001. The National Association for Public Administration (NAPA),
which conducted a comprehensive review of third-party certifiers and the processes
meant to ensure their consistency and rigor, has also raised such concerns. NAPA (2001:
17) concluded that the responsible organizations “have largely refrained from developing
specific guidance on how auditors should assess continual improvement and prevention
of pollution.”
Third-party auditors for ISO 14001 are similar to external financial auditors in
some ways. The purpose of both types of audits is to verify management claims of
conformity to standards, they both use sampling techniques rather than comprehensively
reviewing all issues, and after they are satisfied they both issue their external “seal of
approval”. In the United States, both ISO 14001 and financial auditors face restrictions
69
meant to ensure that their organizations and personnel are independent of the
organization being assessed.46 Both ISO 14001 and financial auditors are vulnerable to
46
For ISO 14001, auditor independence is required by the ISO/IEC Guide 66:1999
“General Requirements for Bodies Operating Assessment and Certification/registration of
Environmental Management Systems.” This states that certifiers shall “…ensure that the
subcontracted body or person…is not involved…with the design, implementation, or
maintenance of an EMS or related management system in such a way that impartiality
could be compromised…”(§4.1.3b). The International Accreditation Forum’s guidance
document to ISO/IEC Guide 66 clarifies that an auditor must be fully independent of the
organization being assessed by proscribing consulting (“participation in an active manner
in the development of an EMS to be assessed…[or] giving specific advice towards the
development
and
implementation
of
management
systems
for
eventual
certification/registration…” (§4.1.21) to organizations auditors subsequently assess. The
guidance document also states that individuals “should not be employed to conduct an
audit…if they have been involved in any consultancy…towards the organization in
question, or any company related to that organization, within the last two years”
(§4.1.29). The Sarbanes-Oxley Act stopped financial auditing companies from
conducting some, but not all, non-audit related professional services including services
related to the accounting records of the audit client, internal audit outsourcing, and the
design and implementation of information technology systems (Riesenberg, 2002).
70
criticism that, because they are hired by the organizations whose practices they are
supposed to assess, their desire to get re-hired in subsequent years creates a serious
conflict of interest that might encourage them to reduce their level of scrutiny.
However, significant differences between the governance systems surrounding
financial auditing and ISO 14001 auditing suggests the former is more robust. For
example, financial auditing firms are liable for the consequences of inadequate audits,
they are monitored by a public agency (the Securities and Exchange Commission), and
have well established qualification criteria including years of formal education and a
standardized qualifying exam. In contrast, ISO 14001 auditors are not liable, are not
subject to oversight by any public agency, and the ISO guidelines governing
qualifications of environmental auditors have been critiques as allowing “potentially wide
variations in the environmental experience of EMS auditors” (NAPA, 2001: 5).
Furthermore, while financial audits typically result in a detailed publicly available
financial report, the ISO 14001 standard does not require any environmental performance
data to be public disclosed. In addition, while Boards of Directors typically select
financial auditors to temper conflicts of interest between the Boards’ desire for robust
audits and management’s desire for swift approval, such measures are rarely taken in
managing relationships with ISO 14001 auditors. These differences suggest that system
governing the ISO 14001 certification process is less robust than that which governs
financial auditing, which itself has been widely attacked as inadequate. They also point
the way toward several possibilities to improve the robustness of ISO 14001’s
71
verification mechanisms, including instituting standardized training requirements and
qualification exams.
Bolstering the verification mechanism may increase ISO 14001 certification costs.
Such cost increases might yield two benefits. First, it might increase the gap in costs
between those with strong and weak environmental management capabilities. This could
better ensure that adoption is profitable only for those with stronger environmental
management capabilities, and therefore deter those wishing to “symbolically adopt” and
free ride on the standard’s reputational benefits. Second, “raising the bar” would go
further to ensure that adopters have indeed implemented a robust EMS, and could thus
better ensure that adopters continuously improve their performance.
Increasing transparency surrounding the certification process is another way to
bolster ISO 14001’s verification mechanisms. The International Organization for
Standardization could require auditors to make public the names of all facilities that hire
third-party auditors, audit dates, and the findings of these audits.47 The bad publicity that
could result from unsuccessful attempts to become certified could deter marginal
facilities from seeking certification, or at least encourage them to delay until they have
improved their EMS. Others have suggested that the International Organization for
Standardization should publish a list of facilities that have lost their certification status
47
Currently, auditors are not even required to publish a comprehensive list of clients they
have certified.
72
(NAPA, 2001). These transparency measures could encourage greater differences (i.e.,
promote separation) between adopters’ and non-adopters’ environmental management
processes and performance.
2.6.2 Is a Process-Orientation Insufficient to Improve Performance?
Even if the independent assessment were perfectly able to ensure that adopters
fully implement all aspects of a voluntary management program, it could be that the
strictly procedural approach of voluntary management program may simply be
insufficient to drive improvement in compliance or health risks. ISO 14001’s emphasis
on documentation has been the subject of much criticism, and it is possible that many
facilities implementing such procedural initiatives are distracted from other tasks that
might elicit better results. For example,
In actual practice, ISO 14001 and EMAS implementation teams start with,
and often get mired in, the paperwork. The standard requirement to ‘go
through the process’ can make it quite difficult to focus less on the details
and develop an EMS with a strategic environmental direction….ISO
14001 and EMAS...may create the illusion to executive management that
all is well because the process is in place; management’s attention may
shift from improving performance goals to completing a procedure and
getting the box checked (MacLean, 2004b: 13).48
48
In addition, a leading certifier’s Chief Executive Officer recently referred to
conformance-based environmental management systems such as ISO 14001 as a
“fundamentally flawed” approach to improve performance (MacLean, 2004b). Many
environmental NGOs have also had substantial doubts about ISO 14001, with some
claiming that ISO 14001 amounts to “greenwashing” where adopters claim to conduct
73
Responding to persistent skepticism about the Responsible Care program, and in
the face of evidence that participants have inferior ex ante performance and have
improved their performance less than non-participants (King & Lenox, 2000; Lenox &
Nash, 2003), the US chemical industry recently announced a complete overhaul of this
program. The Responsible Care program had been far more prescriptive than the ISO
14001 standard, calling for the adoption of nearly 100 management practices based on
seven “Codes of Management Practices” (pollution prevention, process safety,
distribution, employee health and safety, product stewardship, security, and community
awareness and emergency response). The program’s fundamental weakness was that it
allowed—and in the US actually required—any company that is a member of the
national chemical industry association to “join” Responsible Care, regardless of whether
they implemented any of its management codes. Unfortunately, the industry’s complete
superior
environmental
management
practices
while
making
few
substantive
improvements. ECOLOGIA (2000) has expressed concern that “Certification to ISO
14001 can be achieved without improvement in environmental performance or
investment in pollution abatement technologies” The World Wildlife Fund for Nature
(WWF) claims that the International Standards Organization “has insufficient safeguards
in place to prevent unscrupulous companies from using certification to ISO 14001 as a
‘quasi-ecolabel’ which would lead to consumer confusion” that the standard is
performance-based (WWF, 1996).
74
revamping of the program—which is “intended to improve public perceptions of the
industry, and member companies' perceptions of Responsible Care itself” (Chemical
Week, 2002: 33)—may be discarding its most promising elements along with its very
weak participation requirements. While the industry now requires third-party certification
by 2007, the Responsible Care program has abandoned its nearly 100 prescriptive
management practices governing environmental, health, safety, and community issues in
favor of requiring independent certification to a far more general management system
approach (akin to ISO 14001). Given my study failed to find evidence that this approach
leads to improved performance, this shift to less prescription seems risky. Indeed,
combining prescriptive management practices with mandatory periodic third-party
certification may contain the right combination for voluntary management programs to
improve participants’ performance. Indeed, this is the model that some recent industryspecific voluntary management programs have adopted, including the Forest Stewardship
Council’s forest certification program, and merits rigorous independent evaluation.
As an alternative to more prescriptive industry-specific management practices,
voluntary management programs can also stimulate performance improvement by
requiring such improvements as a condition for ongoing participation. The few
government-initiated voluntary programs that have actually elicited performance
improvement are those require improvements, such as the US EPA’s 33/50 and
75
Indonesia’s PROPER-PROKASIH programs (Blackman, Afsah, & Ratunanda, 2004;
Khanna & Damon, 1999).49
2.6.3 Implications
Firms. This study has important implications for the hundreds of thousands of
firms that are relying on voluntary management programs to signal superior management
practices to interested buyers, regulators, and local communities. While prior research has
shown that the vast majority of voluntary management programs that lack verification
mechanisms do not favorably differentiate participants from non-participants, this study
has shown that third-party verification distinguishes adopters via a positive selection
effect and through the preparations required to initially meet the standard’s requirements.
This is an important finding for several types of managers, including those engaged in
marketing and procurement, outsourcing strategy, or responsible for ensuring that
subsidiaries and suppliers meet regulatory requirements and continually improve their
environmental, safety, or quality management practices and performance.
The evidence I found that ISO 14001 distinguishes adopters as organizations with
superior environmental performance may encourage firms concerned about their
49
The US EPA launched its voluntary 33/50 Program in 1991 to encourage companies to
pledge that they would reduce their emissions of 17 high-priority toxic chemicals by 33%
by 1992 and by 50% by 1995. Controlling for selection effects, participants actually
reduced their emissions faster than non-participants (Khanna & Damon, 1999).
76
suppliers’ environmental management practices and performance to use ISO 14001 to
screen suppliers, a practice some firms have already begun implementing (Fielding,
2000; Sissell, 2000; Strachan, Sinclair, & Lal, 2003). Failing to find evidence that
adopters subsequently reduced their environmental impact or improved their regulatory
compliance, firms waiting for evidence that adoption typically leads to such
improvements will unfortunately remain unsatisfied.
Policymakers. Many policymakers are using or considering using voluntary
management programs to improve the efficiency of achieving environmental, labor, and
financial regulatory objectives. Because “priority schemes for [regulatory] inspections
are very unsophisticated” (Wasserman, 1987: 20), if voluntary management programs
credibly indicated superior or improving compliance, regulators could redeploy their
scarce resources from participants to other facilities that are more likely to be in noncompliance. Confidence that ISO 14001 effectively assures regulatory compliance has
been undermined worldwide by mass media stories about compliance problems at ISO
14001 certified firms.50
The absence of a demonstrable link between ISO 14001
certification and either superior or more rapid improvement in compliance or
50
Examples of such media reports include “dioxin compliance problems at Ebara Corp.,
a Japanese facility certified in 1997...the largest Brazilian pollution incident in 25 years at
the certified Petroleo Brasileiro S.A. facility...[and], two Taiwanese ISO-certified
facilities were involved in a hazardous waste dumping scandal” (MacLean, 2004b: 13).
77
environmental performance has resulted in few environmental agencies officially shifting
their enforcement scrutiny away from certified firms (Hillary & Thorsen, 1999; US EPA,
2003a). My finding some evidence that adopters had better ex ante compliance records
suggests regulators should seriously consider using ISO 14001 as an indicator of superior
compliance.
A primary distinction between ISO 14001 and other voluntary management
programs that were associated with adverse selection is that ISO 14001 requires periodic
independent certification. As such, policymakers may wish to reserve endorsing only
voluntary management programs that feature, at a minimum, a robust verification
requirement. However, requiring participation in a robustly monitored voluntary
management program and imposing minimum performance thresholds appears a prudent
combination when regulators consider which companies to de-prioritize. For example,
facilities must have both an EMS subjected to periodic independent assessment (e.g., ISO
14001) and a history of sustained regulatory compliance before they are eligible to join
the US EPA’s National Environmental Performance Track voluntary program.51
51
Regulators in the Netherlands, Denmark, and some US states (e.g., Maine, North
Carolina, Oregon, Texas and Wisconsin) are providing streamlined and more flexible
permitting processes and exemptions from some reporting requirements to participants in
some voluntary management programs; most offer such benefits only to those
78
To gain confidence that voluntary management programs will actually elicit
performance improvement, regulators should consider favoring programs that go beyond
procedural mandates by stipulating performance improvement requirements such as the
US EPA’s 33/50 program. This study confirms the wisdom of this integrated approach.
2.6.4 Limitations and Future Research
It is important to note several limitations of the data and methods employed in this
study. TRI data are self-reported to US EPA by facilities managers and are externally
verified only occasionally by some state environmental agencies and by US EPA, and are
very often based on estimates rather than monitoring. Toxic chemical emissions included
in TRI data represents just one of any number of environmental performance metrics that
could be used to evaluate a voluntary management program whose objective includes
reducing environmental impacts. For example, participants may be more aggressively
improving energy efficiency to reduce greenhouse gas emissions, increasing their use of
recycled materials, or enhancing the recyclability of their own products. They might also
improve training and operational processes and bolster their regulatory knowledge.52This,
participants who also possess good compliance records (Hillary & Thorsen, 1999: 6;
Industrial Economics, 2003; Tjiong, 2002; TNRCC, 2003).
52
For example, UK Environmental Agency inspectors found that ISO 14001 adopters
had significantly better regulatory knowledge, plant maintenance, management and
79
in turn, can expedite responding to regulators’ requests for information and
simultaneously reduce compliance costs and enhance regulatory relations (Gupta &
Piero, 2003). 53
As for methods, while the treatment effects analysis controlled for differences in
several observable factors associated with adoption, there remains the possibility that
unobserved factors that changed during the sample period may have influenced both the
decision to adoption and performance, potentially introducing bias I was unable to
control for. Future research that employs a suitable instrument can address this concern.
This study contributes to the nascent literature examining the effectiveness of
self-regulation programs and enhances the economics and management literature on
training, and operational processes than facilities without an externally verified
environmental management system (Dahlström et al., 2003).
53
Gupta & Piero (2003) describes a telling anecdote: an EPA inspector arrived at a ISO
14001 certified facility and announced it would take three weeks to investigate the source
of groundwater contamination because this required reviewing many regulatory
documents the facility had prepared. However, “The ease with which the company
provided the requested information resulted from requirements achieved through EMS
documentation and recordkeeping, allowing the inspector to be in and out of the facility
in less than four hours…The company felt this experience alone paid for the system”
(Gupta & Piero, 2003: 19).
80
assuring supplier performance. The efficacy of the growing number of initiatives that
seek to substitute for individual monitoring and government regulation will remain in
doubt until robust research demonstrates that they either attract superior performers or
elicit performance improvement (or both). Future researchers should consider using
alternative performance measures and settings other than the United States, and should
strive to identify the mechanisms that lead to selection effects and treatment effects.
Choosing the US as the setting for empirical studies that evaluate the effects of voluntary
management programs on regulatory compliance may significantly underestimate the
effectiveness of such programs in other countries. The relatively stringent enforcement
regime in the US creates comparatively strong incentives for compliance, adopting ISO
14001 in many developing countries may provide the first opportunity to learn about all
applicable laws and regulations. In addition, the limited impact ISO 14001 may have on
reducing toxic emissions in the US may underestimate its effects in other countries
because negative publicity regarding TRI releases in general, especially in the early years
during the late 1980s through mid-1990s—coupled with a growing awareness of costsaving pollution prevention opportunities during this period—may have already driven
eventual ISO 14001 adopters to exhaust most profitable avenues of reducing their TRI
emissions. Future research could examine the impact of voluntary management programs
on environmental performance metrics akin to TRI in other empirical contexts.54 In
54
For example, Australia, Belgium, Canada, Denmark, Finland, Ireland, Italy, Japan,
81
particular, one might expect ISO 14001 adoption to elicit greater improvement in
compliance and environmental performance in countries with less sophisticated
regulatory enforcement regimes and less media pressure to reduce emissions. Dasgupta et
al.’s (2000) finding that Mexican manufacturers who implemented an ISO 14001-style
EMS reported better compliance suggests promising results in such domains.
While most evaluations of voluntary management programs that lacked thirdparty verification requirements found that these programs attract worse-than-average
performers, a few others programs besides ISO 14001 (as shown in this chapter) have
attracted superior performers. More research is needed to understand what key features
differentiate these programs. In addition, more research is needed to understand how
voluntary management programs can improve adopters’ performance, since this (rather
than a selection effect) is the primary outcome that regulators and functional managers
(e.g., quality or environmental managers) seek from voluntary management programs.
Korea, Mexico, the Netherlands, Norway, the Slovak Republic, Switzerland, and the
United Kingdom have begun operating Pollution Release and Transfer Registers akin to
TRI that require facilities to publicly report their emissions (OECD, 2001). Over half of
all ISO 14001 certified organizations are located in these countries (27680 of the 49462,
based on national totals obtained ISO [, 2003 #641]), which presents a formidable
opportunity for researchers to conduct performance evaluations in these countries.
82
Most studies that have evaluated the extent to which voluntary programs are
achieving their ultimate objectives have focused on environmental programs; many
research opportunities exist in other domains. For example, more research is needed to
evaluate whether voluntary programs governing labor and quality management are
distinguishing their participants. Lamenting the absence of rigorous performance
evaluations of labor codes and standards, O'Rourke (2003: 10) notes “It is hard to
determine how much improvement firm-led codes of conduct and monitoring programs
have achieved. Little research exists on the impacts of codes and self-monitoring on
actual workplace conditions.” Do companies that adopt voluntary labor standards have
fewer instances of abusive labor practices than non-adopters, and are they subjected to
fewer media exposés of their labor practices?
Do participants in voluntary quality
programs actually produce goods with fewer defects? If so, are such distinctions based on
a selection effect, a treatment effect, or both?
Codes of conduct, industry-initiatives, government voluntary programs, and
international standards that govern environmental management, occupational health and
safety, human rights, quality management, and other management processes continue to
proliferate. The need is greater than ever to discover which programs are validly
differentiating their adopters, and which program features are critical to ensure their
credibility. Absent such knowledge, many of the millions of hours and dollars spent
implementing these tools may be wasted. The methods presented this chapter could be
adopted to evaluate these questions
83
84
TABLE 2.2
Summary Statistics
Variable
FacilityMean
year
observations
Became ISO 14001 certified this year [note 1]
RCRA violations, sum of 1 & 2 years ago
Enforcement actions, sum of 1 & 2 years ago
Pounds of toxic chemicals emitted, average of 1 & 2
years ago (log)
Emission hazardousness per pound, average of 1 & 2
years ago (log) [Note 2]
Health risk imposed, average of 1 & 2 years ago (log)
[Note 3]
Production index, average of 1 & 2 years ago (log) [Note
4]
Evidence of environmental management system
EPA 33/50 participant
ISO 9000 certified by this year
Any waste transfers to Publicly Owned Treatment
Works (POTW) 1 year ago
Compliance cost per state (state)
State environmental policy comprehensiveness (state)
Percent with at least 4-years college in 2000 (Census
Tract)
Per capita income in 1999 (log) (Census Tract)
Number of facilities
Total TRI reporting facilities
Not ISO 14001 certified in 1996-2002
ISO 14001 certified in 1996-2002
Became certified in 1996
Became certified in 1997
Became certified in 1998
Became certified in 1999
Became certified in 2000
Became certified in 2001
Became certified in 2002
SD
Min
Max
Number coded 1
for dummy
variables
127551
127551
127551
127551
0.01
0.57
0.03
7.39
0.08
1.99
0.17
3.11
858
0
46
0
2
4.61
17.50
127551
4.89
0.88
4.61
17.66
127551
8.27
4.50
4.61
24.11
42355
5.65
0.81
4.64
10.28
127551
127551
127551
127551
0.11
0.22
0.02
0.18
0.31
0.41
0.15
0.38
126454
126468
122112
-0.07
23.58
0.19
0.13
7.42
0.14
-0.57
5
0.00
0.27
38
1.00
122112
9.80
0.38
6.89
12.27
SIC 28
SIC 34
SIC 35
SIC 36
SIC 37
5,464
5,344
120
4
13
12
13
22
29
27
5,537
5,388
149
1
2
17
9
37
28
55
2,179
2,083
96
1
3
18
14
21
18
21
2,597
2,396
201
7
23
24
39
54
24
30
2,677
2,385
292
2
6
36
34
55
70
89
13900
27429
2742
22982
Total
18,454
17,596
858
15
47
107
109
189
169
222
Number of facilities: 18454. Sample includes data for 1996-2002.
Note 1: Became ISO 14001 certified this year is coded 1 at most once per facility.
Note 2: Emission hazardousness per pound is defined as Health risk imposed divided by Pounds of toxic chemicals
emitted.
Note 3: Health risk imposed is defined as the sum across all toxic chemicals, where the pounds of each chemical
released to air is multiplied by the chemical-specific toxicity weights pertaining to inhalation exposure from the US
EPA’s Risk-Screening Environmental Indicators model
Note 4: Production index is calculated as the facility’s employment in a base year, which in other years is adjusted by
the facility’s annual production ratios reported to the US EPA Toxic Release Inventory program
85
86
87
TABLE 2.5
Selection Results: Performance Trends
Average temporal trends during pre-adoption period
(1)
(2)
Facilities that will Facilities that will
not adopt
adopt
Compliance (1) Any EPA enforcement actions
Outcomes
(2) RCRA violations
Emissions (3) Pounds of emissions
Outcomes
(4) Hazardousness of emissions
(health risk per pound)
(5) Health risk created
0.0022
[0.0001]
-0.021
[0.001]
0.028
[0.012]
0.027
[0.003]
0.184
[0.021]
0.0004
[0.0001]
-0.028
[0.007]
-0.092
[0.042]
0.072
[0.014]
0.054
[0.074]
(3)
Difference
between groups
-0.0017
[0.0003]***
-0.007
[0.006]
-0.120
[0.045]***
0.044
[0.013]***
-0.130
[0.080]*
Notes: Columns presents the average estimated time trend (slope) for each outcome variable for facilities
that never adopt ISO 14001 during the sample period; Column 2 presents this for facilities that will adopt
ISO 14001 during the sample period. Column 3 presents the difference in trends between the two groups,
which were calculated for each outcome after omitting those facilities whose slopes fell outside the 1st to
99th percentiles to reduce the influence of outliers. Column 3 also presents the results of t-tests to determine
whether the difference between the group means was statistically significant. * p<0.10; ** p<0.05; ***
p<0.01. The models for pounds of emissions and health risk include the production index to control for
changes in production levels; thus these trends measure changes in pollution intensity. Brackets contain
standard errors. The sample period includes 1991-2001, but adopters are dropped from the sample the year
before their certification year to avoid any effects of certification. Sample sizes for this analysis were 17273
non-adopters and 832 eventual adopters for enforcement actions; 10905 and 645 for RCRA violations; 5351
and 429 for pounds of emissions: 5354 and 426 for hazardousness of emissions; 5348 and 432 for health
risk.
88
TABLE 2.6
Adoption Model to Generate Propensity Scores
Dependent variable: Became ISO 14001 certified this year (dummy)
Probit coef dF/dx
Regulatory
deterrence
Number of RCRA violations in prior 2 years
Number of enforcement actions in prior 2 years
Related
programs
Evidence of environmental management system, 1 year ago
EPA 33/50 participant
ISO 9000 certified as of last year
Regulatory
complexity
Any waste transfers to POTW 1 year ago
Compliance cost per state (state)
State environmental policy comprehensiveness (state)
Community
pressure
Percent with at least 4-years college in 2000 (Census Tract)
Per capita income in 1999 (log) (Census Tract)
Environmental Pounds emitted, average of prior 2 years (log)
performance
Emission hazardousness per pound, average of prior 2 years
(log)
Health risk imposed, average of prior 2 years (log)
Observations
Facilities
Adopters
Pseudo R2
0.002
[0.006]
0.025
[0.069]
0.172
[0.038]***
0.221
[0.030]***
0.283
[0.061]***
0.273
[0.034]***
0.736
[0.149]***
-0.004
[0.003]
0.247
[0.170]
-0.004
[0.066]
0.024
[0.008]***
-0.009
[0.020]
0.009
[0.005]
107,904
15,635
790
0.12
0.0000
0.0003
0.0021
0.0027
0.0043
0.0036
0.0076
0.0000
0.0026
0.0000
0.0002
-0.0001
0.0001
Notes: This table reports probit coefficients and marginal effects. dF/dx is the change in the probability of
adoption for an infinitesimal change in each independent variable evaluated at the mean all variables.
Robust standard errors clustered by facility in brackets. * p<0.10; ** p<0.05; *** p<0.01. All specifications
include dummies for years, industries (2-digit SIC Code), and the 10 EPA Regions. Sample includes 19962002; adopters are excluded from the sample after their certification year.
89
90
TABLE 2.8
Balancing Covariate & Outcome Levels: Standardized Bias
(1)
Entire
Sample
(1996)
RCRA violations, sum of 1 & 2 years ago
Enforcement actions, sum of 1 & 2 years ago
Pounds emitted, average of 1 & 2 years ago (log)
Emission hazardousness per pound, average of 1 & 2
years ago (log)
Health risk, average of 1 & 2 years ago (log)
Evidence of environmental management system
EPA 33/50 participant
ISO 9000 certified by this year
Any waste transfers to POTW 1 year ago
Compliance cost per state (state)
Environmental policy comprehensiveness (state)
Percent with at least 4-years college in 2000 (Tract)
Per capita income in 1999 (log) (Tract)
SIC 28 - Chemicals and allied products
SIC 34 - Fabricated metal products
SIC 35 - Industrial machinery and equipment
SIC 36 - Electrical and electronic equipment
SIC 37 - Transportation equipment
EPA Region 1
EPA Region 2
EPA Region 3
EPA Region 4
EPA Region 5
EPA Region 6
EPA Region 7
EPA Region 8
EPA Region 9
EPA Region 10
Mean standardized bias (all above variables)
Median standardized bias (all above variables)
10%
0%
48%
-1%
35%
24%
44%
9%
44%
24%
-6%
7%
14%
-39%
-33%
0%
24%
49%
-4%
-8%
-8%
3%
26%
-14%
0%
-8%
-15%
-7%
10%
3%
(2)
(3)
Matched Sample:
Matched Sample:
10 Overall-Nearest- 10 Nearest-NeighborsWithin-Industry
Neighbors
(Match Year)
(Match Year)
3%
3%
0%
0%
11%
9%
4%
4%
9%
3%
12%
4%
10%
8%
-3%
0%
6%
-5%
-8%
0%
2%
13%
0%
-4%
-4%
3%
6%
-7%
4%
-8%
-4%
0%
2%
2%
7%
5%
15%
9%
10%
8%
0%
0%
6%
-8%
-10%
-3%
2%
16%
-4%
-4%
-8%
3%
6%
-7%
0%
-9%
-4%
0%
2%
1%
Note: Following Rosenbaum & Rubin (1985), standardized bias before matching is calculated for a given
covariate X as the difference of the sample means in the full treated and non-treated sub-samples as a
percentage of the square root of the average of the sample variances in the full treated and non-treated
groups. Thus:
(X1 − X 0 )
Standardized Bias = 100 ∗
0.5 ∗ (V1 (X ) + V0 (X ))
where X1 (V1) represents the mean (variance) in the treatment group and X 0 (V0) is the analogue for the
control group. The standardized difference after matching uses the same approach but employs the means
and variances of the matched sample. The standardization allows comparisons between variables X and, for
a given X, comparisons before and after matching.
91
92
93
94
95
Appendix to Chapter 2
96
TABLE 2A.2
Adoption Model Including Production Index to Generate Alternative Propensity
Scores
Dependent variable: Became ISO 14001 certified this year (dummy)
Probit coef dF/dx
Regulatory
deterrence
Related
programs
Number of RCRA violations in prior 2 years
Number of enforcement actions in prior 2
years
Evidence of environmental management
system, 1 year ago
EPA 33/50 participant
ISO 9000 certified as of last year
Regulatory
complexity
Any waste transfers to POTW 1 year ago
Compliance cost per state (state)
Community
pressure
Environmental
performance
State environmental policy
comprehensiveness (state)
Percent with at least 4-years college in
2000 (Census Tract)
Per capita income in 1999 (log)
(Census Tract)
Pounds emitted, average of prior 2 years
(log)
Emission hazardousness per pound, average
of prior 2 years (log)
Health risk imposed, average of prior 2 years
(log)
Production index, average of prior 2 years
(log)
Observations
Facilities
Adopters
Pseudo R2
-0.001
0.0000
[0.008]
0.022
0.0003
[0.089]
0.163
0.0029
[0.048]***
0.134
0.0022
[0.045]***
0.176
0.0034
[0.078]**
0.250
0.0044
[0.044]***
0.861
0.0133
[0.207]***
-0.010
-0.0002
[0.005]*
0.526
0.0081
[0.239]**
-0.162
-0.0025
[0.091]*
0.004
0.0001
[0.011]
0.001
0.0000
[0.024]
-0.005
-0.0001
[0.007]
0.189
0.0029
[0.024]***
39,187
8,665
445
0.14
Notes: This table reports probit coefficients and marginal effects. dF/dx is the change in the probability of
adoption for an infinitesimal change in each independent variable evaluated at the mean all variables.
Robust standard errors clustered by facility in brackets. * p<0.10; ** p<0.05; *** p<0.01. All specifications
include dummies for years, industries (2-digit SIC Code), and the 10 EPA Regions. Sample includes 19962002; adopters are excluded from the sample after their certification year.
97
TABLE 2A.3
Matched Samples Based on Alternative Propensity Scores
Also Balance Adoption Covariates & Pre-Adoption Outcome Levels
(1)
(2)
(3)
(4)
(5)
(6)
Matched Sample Based on
Matched Sample Based on
Alternative Propensity Scores: Alternative Propensity Scores:
10-Overall-Nearest-Neighbor 10-Nearest-Neighbor-WithinMatched Sample
Industry
RCRA violations, sum of 1 & 2 years ago
Enforcement actions, sum of 1 & 2 years ago
Pounds emitted, average of 1 & 2 years ago (log)
Emission hazardousness per pound, average of 1 &
2 years ago (log)
Health risk, average of 1 & 2 years ago (log)
Production index, average of 1 & 2 years ago (log)
Evidence of environmental management system
EPA 33/50 participant
ISO 9000 certified by this year
Any waste transfers to POTW 1 year ago
Compliance cost per state (state)
Environmental policy comprehensiveness (state)
Percent with at least 4-years college in 2000
(Tract)
Per capita income in 1999 (log) (Tract)
SIC 28 - Chemicals and allied products
SIC 34 - Fabricated metal products
SIC 35 - Industrial machinery and equipment
SIC 36 - Electrical and electronic equipment
SIC 37 - Transportation equipment
EPA Region 1
EPA Region 2
EPA Region 3
EPA Region 4
EPA Region 5
EPA Region 6
EPA Region 7
EPA Region 8
EPA Region 9
EPA Region 10
Non- Adopters Significance Non- Adopters Significance
adopters
of
adopters
of
difference
difference
N=2652 N=391
N=5691 N=749
0.99
1.07
0.80
0.86
0.05
0.06
0.04
0.04
9.74
10.14
**
8.70
9.00
**
5.04
11.20
6.01
0.25
0.38
0.06
0.45
-0.05
22.74
5.04
11.63
6.28
0.26
0.45
0.06
0.52
-0.05
22.68
0.19
9.82
0.22
0.20
0.14
0.22
0.22
0.06
0.07
0.07
0.18
0.38
0.11
0.07
0.01
0.02
0.03
0.20
9.85
0.17
0.17
0.13
0.25
0.28
0.07
0.05
0.06
0.18
0.40
0.10
0.08
0.02
0.02
0.03
*
***
***
**
**
4.91
9.68
5.86
0.19
0.33
0.05
0.35
-0.05
23.26
4.94
10.04
6.33
0.21
0.40
0.07
0.40
-0.04
23.25
0.20
9.83
0.18
0.22
0.14
0.24
0.23
0.07
0.07
0.08
0.18
0.36
0.09
0.05
0.02
0.07
0.02
0.20
9.85
0.15
0.18
0.13
0.25
0.30
0.06
0.06
0.06
0.19
0.39
0.07
0.05
0.01
0.06
0.02
*
***
**
***
*
**
**
***
Note: Matched non-adopters (adopters) in the match year (certification year) are presented in Columns 1
and 4 (Columns 2 and 5). Columns 3 and 6 display t-test results that the group means are equal: * p<0.10;
** p<0.05; *** p<0.01 Sample sizes are above, except for production index in the 10-Nearest-NeighborWithin-Industry matched sample: 2509 non-adopters (Column 4) and 412 adopters (Columns 5).
98
99
Chapter 3.
3. Turning Themselves In: Why Companies Disclose
Regulatory Violations§
The pitched political battles over regulation in the 1970s and 1980s, from
deregulation to Reagan’s vow to get government “off the backs” of industry, have given
way in recent years to a new wave of voluntary self-regulation programs based on a more
cooperative approach between government and industry.
Regulatory agencies are
embracing regulatory models that see firms as active participants in their own
governance. And from the industry side, talk is increasingly about companies regulating
themselves rather than trying to avoid regulation altogether. Industry proponents argue
that self-regulation is a more efficient and effective way to achieve regulatory goals, and
that voluntary, private compliance initiatives should largely replace what they see as a
cumbersome, bureaucratic and outdated “command-and-control” regulatory system (Orts
1995, Murray 1999). They are supported by a substantial and growing body of academic
literature touting the virtues of a more cooperative regulatory system (Bardach & Kagan
§
This chapter is co-authored with Jodi L. Short, Department of Sociology, University of
California, Berkeley.
100
1982; Scholz 1984; Ayres & Braithwaite 1992; Gunningham & Grabowsky 1998). More
importantly, this cooperative approach has influenced the practices of regulatory
agencies, resulting in the proliferation of voluntary self-regulation programs that engage
firms as partners in regulatory activities from achieving “beyond compliance” results to
policing their own noncompliance.
A regulatory system that relies increasingly on corporate self-regulation
ultimately can be effective only if organizations are willing to admit and correct their
failures as well as tout their successes. To this end, several regulatory agencies have
developed “self-policing” programs that provide incentives to encourage companies to
self-disclose their legal violations, shifting the burden of monitoring regulatory
compliance from the government to the private sector. For example, through its Hazard
Analysis and Critical Control Point program, the US Department of Agriculture recently
reduced the number of onsite inspectors at slaughterhouses and “shifted much of the
responsibility for safety to the plants, requiring them to identify vulnerable points in their
production lines and build in steps to kill germs” (Peterson & Drew 2003:A1).55
55
In addition, the US Department of Justice Corporate Leniency Program encourages
companies to disclose illegal anti-competitive activity by offering amnesty, an approach
many other nations have since replicated (Medinger 2003). The US Department of
Defense established a self-disclosure program to reduce fraud among government
contractors by offering limited liability, maximum confidentiality allowed by law, and
101
These types of initiatives carry promise as well as pitfalls. On the one hand, the
incentives of self-policing programs have encouraged many companies to report and
correct problems that regulators never would have discovered, suggesting the possibility
for real improvements in compliance. On the other hand, without any evidence that they
improve compliance, such programs may give industry an unprecedented and
unwarranted level of control over its own regulation.
As one commentator notes,
“shifting the responsibility for cleanup oversight to the very actors that created the
problem raises public fears of the ‘fox guarding the henhouse.’” (Cox 2004:28). Such
programs risk undermining environmental compliance by publicly praising participants
who may be hiding egregious violations behind their self-disclosure of relatively minor
infractions (Pfaff & Sanchirico 2004). In addition, by providing the regulated community
with broad discretion to determine the scope of regulatory enforcement and to define the
meaning and content of a violation, self-policing programs may subtly alter what it means
to comply and even how regulators define success for the agency.
Until now, the debate over corporate self-regulation has been waged largely in
terms of policy and ideology. We are skeptical of the competing claims this debate has
produced: namely, that corporate self-regulation is mere “greenwashing” or, on the other
other benefits to firms that self-disclose procurement violations (Fleder 1999). Similarly,
the Securities and Exchange Commission encourages self-disclosure by informally
offering prosecutorial leniency (Duggin 2003).
102
hand, that it can supplant the role of government in overseeing industrial activities. In
this chapter, we argue that cooperative strategies and market-based solutions may
effectively complement, but cannot substitute for, more traditional approaches to
regulatory enforcement.
Among the first empirical studies to address self-policing behavior, this article has
two objectives. First, we seek to understand what influences organizations to police their
own operations and “turn themselves in” by self-disclosing their regulatory compliance
infractions. Second, we evaluate the extent to which self-policing actually promotes
regulatory compliance and benefits self-policing organizations. We use longitudinal
cross-sectional data on voluntary disclosures under the United States Environmental
Protection Agency (US EPA) Audit Policy, which provides rich data on how firms
actually behave when they know they have violated the law.
We find that despite the rhetoric of cooperation surrounding self-policing
programs, they work best when coupled with more traditional, coercive regulatory
measures such as inspections and enforcement actions. Facilities were more likely to
self-disclose violations if they were recently inspected or subjected to an enforcement
action, were narrowly targeted for heightened scrutiny by a US EPA initiative, and were
more prominent in their community as indicated by having more employees or revenues.
The chapter proceeds as follows. In the next section, we review the literature on
compliance and self-policing. In Section 3.2, we describe the US EPA Audit Policy, the
empirical setting of our research. In Section 3.3, we hypothesize how various institutional
pressures, organizational characteristics, and legal institutions may influence facilities’
103
decisions whether to self-police. In Section 3.4, we describe our sample, and measures.
We describe our empirical methods and present our results in Section 3.5. We discuss our
results in Section 3.6, and present opportunities for future research and our conclusions in
Section 3.7.
3.1. Literature Review
There is a small but growing literature on corporate self-regulation, consisting
primarily of studies that either evaluate “beyond compliance” initiatives or model selfpolicing behavior.
In the arena of environmental protection, for example, both
government and industry have established programs that recognize and reward firms for
environmental performance and management practices that go above and beyond what
the law requires.56 Evaluations of these “beyond compliance” programs, however, have
found little to support the political enthusiasm for them. Few of these programs have
56
Examples include government partnership programs such as the United States
Environmental Protection Agency’s (US EPA) Greenlights and 33/50 programs and the
United States Department of Energy’s Climate Challenge Program, negotiated
agreements between regulators and industry such as Germany’s Global Warming
Prevention program and the Netherlands’ Declaration on the Implementation of
Environmental Policy, and fully private-sector initiatives such as the chemical industry’s
Responsible Care, the ski industry’s Sustainable Slopes, and the Hotel Green Leaf EcoRating Program.
104
attracted better-than-average participants, and there is little evidence that they improve
performance (King & Lenox 2000; Welch, Mazur & Bretschneider 2000; Lenox & Nash
2003; Rivera & de Leon 2004), prompting some to charge that they are nothing more
than industry “greenwashing” (Eden 1996).57
Much less is known about self-policing programs, largely due to the difficulty of
observing firms’ internal monitoring and policing decisions. The extant literature focuses
on economic models of self-policing behavior, touting it as a way to reduce government
monitoring and enforcement costs (Kaplow & Shavell 1994), optimize levels of selfauditing by firms (Pfaff & Sanchirico 2000), and reduce firms’ costs of avoiding
detection (Innes 2001). Our data on actual firm self-disclosures will provide a valuable
empirical dimension to this literature. To develop hypotheses about why firms turn
themselves in when they have broken the law, we look to the related literature on why
they comply in the first place.
A significant body of research examines what factors influence compliance with
legal obligations. The most common conceptualization of compliance behavior comes
out of deterrence theory, an economic model in which firms are rational, “amoral
calculators” (Kagan & Scholz 1984) that will comply with legal directives only to the
extent that the costs of expected penalties exceed the benefits of non-compliance.
According to deterrence theory, firms’ compliance behavior is influenced both by
57
For exceptions, see Khanna & Damon (1999) and Toffel (2005).
105
specific deterrence -- “the fear engendered by the prior experience of being inspected,
warned or penalized themselves” (Thornton, Gunningham & Kagan 2005:263) – and by
general deterrence, or “hearing about legal sanctions against others” (Thornton,
Gunningham & Kagan 2005:263, Gibbs 1986).
The deterrent effect of potential
sanctions is often viewed as a function of both their likelihood and severity (e.g.,
Friedman 1975).
This rational choice perspective dominates the theoretical literature about selfpolicing, which is largely comprised of economic models that seek to determine optimal
outcomes (e.g., Pfaff and Sanchirico 2000; Innes 2001) as well as legal and policy
arguments supporting or denouncing self-regulation based on normative concerns (e.g.,
Kesan 2000; Murray 1999; Geltman & Mathews 1997; Goldsmith & King 1997; Grayson
& Landgraf 1997; Hunt & Wilkins 1992). The central premise of this literature is that
firms will self-disclose only when it is in their economic self-interest to do so, based on a
strict cost-benefit model of firm decision-making that assesses whether the costs of selfreporting are less than the expected costs of attempting to hide a violation.
While the deterrence-based approach to compliance captures important dynamics
involved in self-regulation and continues to dominate academic literature and regulatory
practice, a significant body of research suggests that compliance is more normativelybased. Several recent studies have found that individuals and organizations alike comply
with law not out of fear, but out of a sense of duty or a desire to do the right thing (May
2004; Gunningham et al. 2004). Vandenbergh (2003), for instance, cites evidence that
compliance with environmental law is motivated by a duty to follow the law and to avoid
106
endangering human health. In addition, several studies have shown a strong link between
legitimacy and compliance, with people more willing to follow the directives of an
authority they see as legitimate (Tyler 1990), and organizations seeking legitimacy
through public displays of legal compliance (Meyer and Rowan 1977).
A growing body of socio-legal research suggests that compliance behavior is best
explained by a complex combination of deterrence instruments as well as social and
moral considerations that shape cost-benefit calculations about the risks of noncompliance. Kagan, Gunningham & Thornton (2003), for instance, argue that firms must
comply not only with the formal legal requirements of their regulatory license, but also
with the “social license” granted by their local communities -- all within the context of
their particular economic constraints. Thus, they show how many firms “overcomply”
with environmental regulations to maintain a good reputation within their local
community.
This body of work suggests that the neo-classical economic basis of the
deterrence model overlooks potentially important differences in how firms perceive and
assess their options to maximize profits as well as distinct, normatively-based
motivations for compliance. We look to new institutional theories of organizations to
understand what kinds of factors might influence compliance decision-making. New
institutional theories of organizations emphasize the symbiotic relationship of
organizations and their environments. From this perspective, law and regulation provide
not only a coercive set of incentives, but a normative framework within which firms
measure their reputation and legitimacy, and a cognitive framework that constitutes what
107
is possible and desirable within a given environment (DiMaggio & Powell 1991,
Edelman & Suchman 1997). In this conception, firm-level cost-benefit calculations are
embedded in the complex interaction of economic demands with broader factors like law,
culture, and norms, and we draw on these insights to develop our hypotheses.
New institutional research suggests that when internal corporate compliance
mechanisms are sufficiently institutionalized, they generate their own commitments and
justifications that can induce compliance despite decreased regulatory enforcement
(Dobbin & Sutton 1998, Edelman et al. 1999). For instance, Dobbin & Sutton (1998)
show how managers responsible for implementing Equal Opportunity Employment
(EEO) regulatory mandates developed independent commitments to procedures that
started out as symbolic gestures, and continued to advocate for the maintenance and
expansion of these procedures on efficiency grounds even as enforcement of EEO law
became less stringent during the Reagan administration. More broadly, Meidinger (1987)
suggests that the contemporary regulatory process, with its broad array of stakeholders
and its structure of ongoing relationships, often produces its own set of “citizenship”
values that leads to regulatory cooperation that cannot be explained exclusively by
economic self-interest.
In fact, some scholars have argued that sufficiently institutionalized compliance
measures would allow industry to regulate itself without the threat of government
sanctions (King & Lenox 2000; Gunningham 1995; Rees 1994).
Rees (1994), for
instance, argues that unsupervised industry self-regulation can work if it comprises a
well-defined “industrial morality,” enforced through peer pressure strong enough to
108
institutionalize a sense of communal responsibility among firms. He shows how this type
of “communitarian regulation” has produced significant safety improvement in the
nuclear power industry since the accident at Three Mile Island. It is not clear, however,
how generalizable his insights are outside this small, highly specialized and
catastrophically dangerous industry that was fighting for its survival. As noted above,
studies of other voluntary “beyond compliance” initiatives have found much more limited
success (King & Lenox 2000; Welch, Mazur & Bretschneider 2000; Lenox & Nash 2003;
Rivera & de Leon 2004). And many researchers suggest that voluntary self-regulation
may only be effective when supplemented by third party oversight and the threat of
sanctions (King & Lenox 2000; Lenox & Nash 2003; Rivera & de Leon 2004). Our
analysis examines how coercive enforcement mechanisms like sanctions and inspections
interact with benefits and incentives to produce desired regulatory outcomes.
Our research expands on the prior literature in three important ways. First, we
apply insights from the compliance/deterrence literature in a novel setting to predict not
whether firms will comply with law, but whether they will come clean when they have
failed to comply. Our data provide a unique window on how firms behave when they
know they have violated the law, thus allowing us to observe behavior that has thus far
largely been studied in the compliance literature only theoretically (May 2004; Thornton,
Gunningham & Kagan 2005) or experimentally (Paternoster & Simpson 1996). Second,
while we rely on the rational-choice perspective of deterrence theory to develop many of
our hypotheses about self-reporting behavior, we attempt to integrate it with a much
broader array of legal, political and cultural factors than recognized in previous studies of
109
self-policing. We draw on institutional theory to develop a more nuanced approach that
acknowledges managers’ cost-benefit calculations are embedded in and influenced by the
broader institutional environment.
Third, by showing how traditional deterrence
strategies as well as broader institutional factors influence self-policing practices, we
hope to invigorate debate in the self-regulation literature, which tends to ignore the
influence of both government and social context.
3.2. The US EPA Audit Policy
the US EPA’s “Incentives for Self-Policing: Discovery, Correction and
Prevention of Violations” (Audit Policy), launched in 1995, provides the empirical
setting for our research. This program provides incentives for companies to identify,
voluntarily report, and correct environmental violations. In exchange, US EPA promises
to reduce or waive penalties that would be owed otherwise and provides a loose
assurance that it will not refer voluntarily reported cases to the US Department of Justice
(US DOJ) for criminal prosecution.
The main objective of the Audit Policy is to
encourage facilities to implement “systematic, objective, and periodic” environmental
auditing and to develop “documented, systematic procedure[s] or practice[s] which
reflects the regulated entity’s due diligence in preventing, detecting, and correcting
violations” (Federal Register 1995:66708).58 US EPA waives 75-100% of the gravity-
58
Facilities must promptly disclose the violation to US EPA, correct the violation, and
take steps to prevent future violations. The Audit Policy does not apply to violations that
110
based (punative) penalties59 associated with self-disclosed violations, depending on
whether they meet all of the Audit Policy’s requirements.60 In addition, US EPA assures
“resulted in serious actual harm or which may have presented an imminent and
substantial endangerment to public health or the environment” (Federal Register 1995:
66709), or to violations that are similar to others the facility experienced within the past
several years. Disclosures under the Audit Policy cannot result from any regulatory or
permit requirements, enforcement actions, employee whistleblowers, or third party
discovery. .
59
The US EPA (2004) notes, “In general, civil penalties that EPA assesses are comprised
of two elements: the economic benefit component and the gravity-based component. The
economic benefit component reflects the economic gain derived from a violator’s illegal
competitive advantage. Gravity-based penalties are that portion of the penalty over and
above the economic benefit. They reflect the egregiousness of the violator’s behavior and
constitute the punitive portion of the penalty.”
60
For self-reporters who meet all of the Audit Policy’s conditions, US EPA waives 100%
of gravity-based penalties. When violations are discovered by means other than
environmental audits or due diligence efforts but all other conditions are met, 75% of
gravity-based penalties are waived. However, US EPA retains full discretion under the
Audit Policy to recover any economic benefit the self-reporter gained as a result of
noncompliance “to preserve a ‘level playing field’ in which violators do not gain a
111
self-disclosers that the agency will not routinely request or use environmental audit
reports as a part of routine inspections or to initiate civil or criminal investigations;61 nor
will it refer self-reported violations to US DOJ for criminal prosecution except in rare
circumstances.62
competitive advantage over regulated entities that do comply” (Federal Register
1995:66712). US EPA may also waive these penalties if it views the economic benefit to
be insignificant.
61
However, US EPA reserves the right to seek such reports if it has independent reason
to believe that a violation has occurred, and federal law provides no audit privilege for
their protection. Many commentators see these guidelines as insufficiently protective of
confidentiality.
62
US EPA will not refer self-reported violations for criminal prosecution so long as they
do not involve a prevalent management philosophy or practice that concealed or
condoned the violations or high-level official involvement in the violations. This aspect
of the policy has been particularly controversial, because US EPA has wide discretion to
determine whether a reported violation qualifies for this kind of relief. In addition, the
criminal environmental enforcement arm of the US DOJ often has a different view on
how cases should be treated and is not bound by US EPA’s Audit Policy. Self-reporters
thus place themselves at some degree of risk that they may face criminal charges from
US DOJ.
112
The Audit Policy has been widely used since its adoption. According to a dataset
we constructed based on US EPA databases and documents (described below), nearly
3500 facilities have disclosed violations during 1997 – 2003, and many of these facilities
simultaneously disclosed involved multiple violations. Even though many self-reporters
incurred significant costs to remedy the violations they voluntarily disclosed, the majority
of participants held favorable views of the program (Federal Register 1999). US EPA
also touts the program as a success: “Discovery and correction of violations under the
policy have removed pollutants from the air and water, reduced health and environmental
risks and improved public information on potential environmental hazards” and ensured
safe management of PCBs and other hazardous wastes (Federal Register 1999:26745).
The Audit Policy explicitly excludes from its purview violations that result in
“serious actual harm or substantial health risk.” Moreover, because this standard is
ambiguous, firms may interpret it more broadly than it is meant to apply and thus avoid
voluntarily reporting compliance violations involving emissions, effluent, or solid waste
that results in little actual harm or heath risk. As a result, voluntarily reported violations
tend to be less serious than EPA-discovered violations (Pfaff & Sanchirico 2004). While
it is important to acknowledge the Audit Policy’s limitations, it would be wrong to infer
that the kinds of record-keeping and reporting violations typically disclosed under the
Audit Policy are trivial.
Emissions violations under all the major environmental
permitting statutes can be discovered and prosecuted only if the regulated community
takes seriously its paperwork obligations.
Thus, while lawful recordkeeping and
reporting are not themselves sufficient, they represent a necessary precondition to
113
environmental compliance and to successful self-policing. As such, in its evaluation of
the Audit Policy, US EPA noted “[The] discovery and correction of violations under the
policy have removed pollutants from the air and water, reduced health and environmental
risks and improved public information on potential environmental hazards” (Federal
Register 1999:26745).
3.3. Who Turns Themselves In?
In this section, we describe several factors that encourage organizations to turn
themselves in. We hypothesize that facilities will self-disclose compliance violations
when they expect to incur more total costs for hiding the violation than for disclosing it,
and argue that these costs include not only monetary penalties but also damage to the
firm’s reputation and relationships with local communities and regulators. We posit that
such economic cost-benefit calculations are deeply influenced by institutional pressures
exerted by regulators and local communities on the facility, the facility’s sensitivity to
these pressures, and the broader legal environment including the presence of legal
protections for self-reporters (Edelman 1990:1406).
3.3.1. The Regulatory Environment
We first examine various legal and regulatory aspects that comprise the
environment in which regulated facilities decide whether or not to turn themselves in. A
facility’s regulatory environment includes not only applicable regulations, but
enforcement activities undertaken by regulators and legal protections and incentives
designed to encourage self-reporting.
We examine how government wields these
regulatory tools in the context of voluntary programs and hypothesize how this influences
114
regulated entities’ decision whether to self-police.
Enforcement Activities: Specific Deterrence
The level of regulatory enforcement is a crucial component of an organization’s
regulatory environment, informing a facility’s expectations about the likelihood of
getting caught out of compliance and creating a framework of legal experience within
which firms ascertain the risks and benefits of self-reporting. Regulatory enforcement
policies typically include both specific and general deterrence strategies.63 Numerous
studies have shown that specific deterrence measures such as regulatory inspections
improve compliance at targeted firms (Gunningham, Thornton & Kagan 2005, Helland
1998, Magat & Viscusi 1990), because frequent inspections increase the likelihood that
regulators will discover and penalize violations (Dimento 1989). In addition, since
regulators often target inspections toward facilities they believe are more likely to have
violations, facilities facing more inspections (regardless of whether violations are
discovered) may be more likely to self-report to bolster regulators’ confidence of their
willingness to comply. Consequently, facilities subjected to more frequent inspections
will be more likely to self-disclose violations to avoid the costs of detection and to
63
As discussed earlier, “specific deterrence” is the deterrence effect of enforcement
actions against a particular facility, whereas “general deterrence” is the deterrence effect
of a facility’s knowledge about enforcement actions against a category to which a facility
belongs, such as an industry or geographic region.
115
generate goodwill with suspicious regulators.
Inspections that uncover violations may have an even greater compliance impact.
Because regulatory agencies are known to target worse violators with their limited
inspection resources (US EPA 1992, Helland 1998), facilities found in violation can
expect to be targeted for more frequent inspections in the near future (Helland 1998,
Harrington 1988). Furthermore, inspections that uncover multiple violations suggest that
the firm has a poor relationship with regulators – both because of their apparent
unwillingness to comply and because high violation rates can result from dismayed
inspectors legalistically interpreting regulations to maximize the number of violations
(Aoki & Coiffi 2000). Such firms may be particularly eager to use self-reports to show
their good faith willingness to comply in an attempt to mitigate their heightened scrutiny
(Helland 1998). Therefore, controlling for inspection rate, we expect organizations with
more violations discovered by inspectors are more likely to self-report regulatory
compliance violations.
Growing empirical evidence suggests that penalties and enforcement actions also
improve facilities’ regulatory compliance (Gunningham, Thornton & Kagan 2005; Gray
& Shadbegian 2005; Mendelhoff & Gray 2005; Gray & Scholz 1991; Aoki & Coiffi
2000). Enforcement actions are administrative or judicial proceedings that subject firms
to fines, penalties and various forms of injunctive relief, and they represent more serious
compliance problems than merely cited violations. Firms with poor compliance records
may tend to go the extra mile to demonstrate compliance for several reasons. First, their
experience may make them more sensitive to the costs of non-compliance, especially
116
since they may face escalating consequences for future violations as “repeat offenders.”
For this reason, they may wish to put themselves back into the good graces of the
regulator (Scholz 1984, Pfaff & Sanchirico 2000, Helland 1998).
In addition to
considerations emanating from a facility’s cost function, “enforcement actions serve to
focus attention on organizational patterns of behavior that may be out of line with
organizational beliefs and accepted social norms” (Gray & Scholz 1993:200). As such,
we predict that facilities with recent enforcement actions will be particularly motivated to
self-disclose violations.
Enforcement Activities: General Deterrence
Beyond their own individual experience, facilities are also influenced by
enforcement activities that affect other organizations in the broader regulatory
community. For example, the overall stringency of an inspection regime can influence
companies’ expectations that regulators will detect their violations (Cohen 1987; Cohen
2000; Epple & Visscher 1984). In addition, high profile enforcement actions against
other firms have motivated some companies to review their compliance programs, and to
modify their equipment, monitoring practices, and employee training (Thornton,
Gunningham & Kagan 2005).
Some regulators have attempted to leverage general deterrence by launching
targeted enforcement initiatives that single out particular industries for added scrutiny
(Epple & Visscher 1984, Cohen 1987, Anderson & Talley 1995, Ross 1982). Such
efforts attempt to increase facilities’ expectations that their violations will be discovered,
thereby motivating greater compliance. For example, US EPA has launched enforcement
117
initiatives to encourage compliance and self-auditing within sectors such as steel minimills and chemical manufacturers. In addition, US EPA releases a list of “National
Priority” sectors where it will target enforcement resources.
Such campaigns are
designed to encourage compliance by increasing these targeted facilities’ perceived
likelihood of getting caught.64 Similarly, we expect that this heightened expectation of
getting caught will encourage organizations facing general deterrence initiatives to selfdisclose violations.
Statutory Protections
State law is an integral part of the regulatory environment facing facilities as they
decide whether to report compliance infractions, with some states providing much greater
protection for corporate disclosures than others. To encourage self-reporting, states have
developed two types of legal protection for voluntary disclosers: (1) audit privilege laws
that prevent state regulatory agencies and private parties from obtaining any documents
produced in connection with an internal environmental audit or using them in court
64
For example, when encouraging facilities to review their prior regulatory reports and
self-disclose any errors or emissions, US EPA has warned that facilities that fail to do so
“will be targeted for potential enforcement inspections,” which “could result in an
enforcement action.” (US EPA’s “Show Cause Letter Regarding EPCRA Section 312
Sector Agreement” and “Asphalt letter”, both obtained via a Freedom of Information Act
Request )
118
against a voluntary discloser; and (2) immunity statutes that shield self-reporters from
prosecution for violations they voluntarily report.
States have taken a variety of
approaches, with some providing one or both of these protections and others providing
none.
Many scholars strongly endorse the adoption of audit privilege laws, stressing the
need for protection against the risks of disclosure, including potential criminal or state
civil liability as well as bad publicity and exposure to citizen suits (Kesan 2000; Murray
1999; Geltman & Mathews 1997; Goldsmith & King 1997; Grayson & Landgraf 1997;
Hunt & Wilkins 1992). Such enthusiasm for audit privilege is typically based on the
argument that companies will not self-disclose violations without such protection for
their audit materials. For example, Hunt & Wilkins (1992:366) note: “Unless current law
and existing policies are modified to broaden confidentiality privileges, … powerful
disincentives to self-examination will remain.” A large survey of manufacturing facilities
in the US conducted in 1998 found that nearly a third of these facilities that were not
conducting internal audits attributed this to a concern that a regulatory agency might
attempt to obtain an audit report and use this information for enforcement actions
(Morandi 1998). A majority of such facilities located in states without immunity or
privilege laws claimed they would begin conducting internal audits if their state passed
such laws (Morandi 1998).
These theoretical arguments and claims by company
representatives have been subjected to little empirical evaluation. We remedy this by
examining the extent to which state-level statutory audit privilege encourages
organizations to self-report compliance infractions.
119
Immunity provides a different way of protecting voluntary disclosers, preserving
state regulators’ access to all relevant information about a violation, but preventing them
from prosecuting a company for voluntarily disclosed violations. In many ways, state
immunity statutes simply mimic the protection that most federal voluntary programs
already provide. However, this protection can be important because facilities often face
overlapping state and federal regulatory obligations, and without immunity, information
disclosed under a federal voluntary program could later be used against them by state
regulators. We examine the extent to which state-level statutory immunity that protects
organizations from prosecution for voluntarily disclosed compliance violations influences
organization’s decision to self-report violations.
3.3.2. Community Pressure
Institutional pressures from community groups—which include local citizens and
environmental groups—represent another important element of an organization’s legal
environment. Several studies have found that company decisions to adopt environmental
management practices are influenced by the desire to improve or maintain relations with
their communities. The desire to improve community relations has influenced firms to
adopt environmental plans (Henriques & Sadorsky 1996), pollution prevention activities
and environmental management systems (Florida & Davison 2001), community advisory
panels (Lynn, Busenberg, Cohen, & Chess 2000), and the ISO 14001 Environmental
Management System Standard (Raines 2002).
Local community and activist group
demands, backed by the threat of adverse publicity, citizen suits, or reports to local
government regulators, can also encourage facilities to improve their environmental
120
performance, creating a kind of “social license” with which firms must comply in
addition to their legal and regulatory obligations (Thornton, Kagan, & Gunningham
2003). This body of research suggests that community pressure induces companies to go
above and beyond what the law requires to maintain their community ties. Because selfreporting violations can be viewed as another manifestation of exhibiting responsible
corporate citizenship, we predict that organizations that face more community pressure
are more likely to self-report regulatory compliance violations.
3.3.3. Organizational Characteristics
Organizations do not respond uniformly to pressures in their legal environment.
Neoinstitutional studies have shown that the normative pressures to comply with
regulatory requirements are stronger on firms that are more prominent or visible within
their legal environment, either because of their size or because of their ties to the public
(Edelman 1990). High profile firms are more sensitive to their legal environment for
several reasons.
Firms with high public visibility “receive more attention from
regulators, the media, and the public, and they are therefore held to higher standards of
institutional compliance than smaller organizations” (Ingram & Simons 1995:1468).
This extra attention makes them more vulnerable to normative pressures (Goodstein
1994; Ingram & Simons 1995) because these firms must “maintain their social
121
legitimacy” (Goodstein 1994:376) and preserve their reputation (May 2004).65 Greater
public scrutiny may also make it more difficult for such firms to hide violations or atone
for them if discovered (Scott & Meyer 1983). Finally, these firms are more likely to have
a “culture of formal rules” and “rational legal authority” (Edelman 1990:1415), with
many already subject to strict reporting obligations; thus, they may be more likely to have
in place a “reporting culture” that would facilitate environmental self-regulation.
Consequently, we hypothesize that organizations with greater public visibility will be
more likely to self-disclose regulatory violations.
3.4. Methods
3.4.1. Sample
Our sampling approach attempts to surmount a major limitation of much selfregulation empirical research. Because the homogeneity of interests among similar firms
fosters the bonds that facilitate effective self-monitoring (Rees 1994), many empirical
studies of self-regulation have focused on a single industry or, in some cases, a handful of
firms (e.g., Rees 1994, King & Lenox 2000, Gunningham, Kagan & Thornton 2003,
65
In addition, there are fixed costs associated with becoming familiar with regulatory
requirements. Therefore, larger firms and firms with multiple facilities within a common
regulatory regime can leverage economies of scale in compliance-oriented tasks. For
example, once a firm develops an audit protocol to ensure compliance, this protocol can
often be leveraged across many of its facilities with only minor revisions.
122
Rivera & de Leon 2004, Welch, Mazur & Bretschneider 2000). Our sample spans a wide
variety of industries, which should enable our insights about the dynamics of industry
self-regulation to be more generalizable. In addition, our approach enables us to examine
whether the determinants and effects of self-policing differ across industries.
Our sampling frame includes facilities that are required to submit data to the US
EPA’s Toxic Release Inventory (TRI) program. This includes manufacturing and other
facilities engaged in pollution-intensive industries66 with 10 or more employees that
manufacture, process, or use significant amounts of toxic chemicals (typically above
10,000 pounds). Our primary sample consists of TRI facilities that are subject to both the
US Resource Conservation and Recovery Act (RCRA) and the US Clean Air Act
(CAA).67 Facilities are subject to RCRA regulations if they generate, manage, store, or
66
Specifically, the TRI program applies to the following facilities engaged in
manufacturing, most metal and coal mining, most electrical utilities, hazardous waste
treatment and disposal facilities, chemical wholesalers, petroleum terminals and bulk
stations, solvent recovery service providers, and all federal facilities (US EPA 2002a).
67
In any particular year, an operating facility may release pollutants below TRI reporting
thresholds and not be subjected to any inspections. In such years, we recoded missing
values from these facilities to zero if we had evidence the facility was active both prior to
and subsequent to that “quiet” year. In making this determination, we considered any of
the following activity: (a) submitted any data to the TRI program; (b) had an inspection
123
treat hazardous waste, and are subject to CAA provisions if they emit air pollutants
beyond regulatory thresholds. We focus on these federal regulations because they are
arguably the most broadly applicable to our sample of facilities.
3.4.2. Measures
We measured voluntary disclosure as a dummy variable, coded 1 for a facility in
a year when it disclosed a compliance violation in conjunction with the US EPA Audit
Policy. We constructed the most comprehensive dataset possible of Audit Policy selfdisclosures by drawing on data from three sources: the US EPA Integrated Compliance
Information System (ICIS) database, the (hardcopy) US EPA Audit Policy Docket, and
lists of facilities that disclosed under the Audit Policy in response to the Compliance
Incentive Programs, discussed below.68 We present total number of facilities selfdisclosing violations to the Audit Policy for each year of our sample in Table 3.1.
or violation recorded in the US EPA’s Resource Conservation and Recovery Act
Information (RCRAInfo) database or Aerometric Information Retrieval System/AIRS
Facility Subsystem (AIRS/AFS) database; or (c) had an enforcement action or voluntary
disclosure recorded in the US EPA Integrated Compliance Information System (ICIS)
database.
68
Discussions with US EPA revealed that both the ICIS database and the Docket were
incomplete, which led us to create our dataset based these sources as well as participant
lists for Compliance Incentive Programs.
124
We measure the specific deterrence effect of inspections and inspector-discovered
violations (Cohen 2000) using data from US EPA’s Resource Conservation and
Recovery Act Information (RCRAInfo) database and Aerometric Information Retrieval
System/AIRS Facility Subsystem (AIRS/AFS) database.69
We measure whether a
facility had an enforcement action based on data obtained from US EPA’s ICIS database.
In our empirical models for the selection analysis, we lagged each of these variables one
year.
We considered two types of general deterrence. First, we considered the facilities
and sectors that were targeted by US EPA Compliance Incentive Programs that
encouraged them to review their compliance status and consider self-disclosing violations
via the Audit Policy. US EPA typically announces Compliance Incentive Programs via
its Enforcement Alert newsletter, the Federal Register, and its website. Facilities may
also learn about these programs through trade associations. We gathered data about
Compliance Incentive Programs via a Freedom of Information Act Request of the US
EPA.70 The second form of general deterrence we considered are US EPA National
69
To reduce the potential influence of outliers, for each of the variable we recoded values
above the 99th percentile to the 99th percentile value.
70
Compliance Incentive Programs that affected our sample include the National Iron &
Steel Mini-mills Program, National Industrial Organic Chemicals Program, National
Nitrate Compounds Program, Region 1 Chemical Industry Program, and Region 5 Iron &
125
Priority Sectors. US EPA announces its two-year priorities in Memoranda of Agreement,
which we obtained from the agency’s website.71 Because US EPA typically announces
its National Priorities the year before they take effect, we considered facilities to be
targeted by National Priorities for three years: the announcement year and the two years
they were in effect. Because some National Priorities are implemented through
Compliance Incentive Programs, we created three dummy variables to reduce
multicollinearity: (1) National Priority Sector and Compliance Incentive Program
Target; (2) National Priority Sector only; and (3) Compliance Incentive Program
Target only.
Steel Mini-mill Program. US EPA Region 1 includes Connecticut, Maine, Massachusetts,
New Hampshire, Rhode Island, and Vermont, and Region 5 includes Illinois, Indiana,
Michigan, Minnesota, Ohio, and Wisconsin.
71
National Priority sectors in our sample include chemical preparation (1998-9), coal-
fired power plants (1996-9), industrial organic chemicals (1996-9), iron and basic steel
products (1996-9), metal electroplating and coating (2000-3), mining (1996-7), petroleum
refining (1996-2003), plastic materials and synthetics (1996-7), primary nonferrous
metals (1996-9), printers (1996-7), and pulp mills (1996-9).
http://www.epa.gov/compliance/data/planning/shortterm.html (last updated March 17,
2005)
126
We created three dummy variables to reflect whether a facility was located in a
state that provided audit privilege only, immunity only, or both audit privilege and
immunity in a given year.72 We constructed these variables using data from Morandi
(1998), US EPA’s Audit Policy website,73 and a private web service run by the Auditing
Roundtable.74
We resolved any inconsistencies by referring to the actual statutory
language in LEXIS-NEXIS state statutory databases. Our coding of this variable is
presented in Table 3.2
While others have used qualitative methods to assess the impact of actual
community pressure on facilities’ environmental behavior (e.g., Gunningham, Thornton
& Kagan 2005), we believe companies may self-disclose violations based not only on the
actual threats they face from their community, but on the community’s potential to
organize and pose such threats. As such, we conceptualize community pressure as a
function of its potential ability to detect violations and exert political influence. Firms in
72
As described earlier, state audit privilege laws prevent state regulatory agencies and
private parties from obtaining any documents produced in connection with an internal
environmental audit or using them in court against a voluntary discloser. States immunity
statutes prevent self-reporters from being prosecuted for violations they voluntarily
report.
73
http://www.epa.gov/region5/orc/audits/audit_apil.htm (last updated April 6, 2004).
74
http://www.auditing-roundtable.org
127
more densely populated areas are likely to be subjected to greater pressure to strictly
comply with environmental laws (Kagan et al. 2004) because there is a greater chance
that a community member might observe evidence of compliance infractions such as
spills into surface waters or releases of black smoke.75 Population density was calculated
as the average number of residents per square mile in the facility’s Census Tract based on
the US Census Bureau’s 2000 Decennial Census. We capture a community’s potential to
apply political pressure by considering income and voter turnout. Communities with
higher household income are expected to be more connected to politicians, and thus
represent a greater threat to facilities. We calculated log household income within each
facility’s Census Tract using data from the 2000 Decennial Census. We employ voter
turnout as a proxy for a community’s level of political awareness and participation
(Hamilton 1993; 1999). We measured voter turnout as the proportion of residents aged
18 and over in the facility’s county who voted for a Presidential candidate in the 2000
75
Densely populated areas also contain numerous business and employment options,
which suggests that citizens and local officials will tend to be less beholden to a corporate
polluter than those in smaller locales where a single industrial employer may be a critical
source of jobs.
128
general election.76 County population data were obtained from the 2000 Decennial
Census, and voting data were obtained from Lublin & Voss (2001) for all states except
Alaska.
We measure a facility’s public visibility using three organizational characteristics:
(1) facility employment; (2) facility revenues (Edelman 1992); and (3) whether the
facility is a member of a publicly-owned company. As facility-level employment is not
publicly available, we estimate this using the log of the nationwide average revenues per
establishment within each 4-digit SIC Code using data from the 1997 Economic Census,
the latest year available. We obtained data on facility revenues and ownership status
from Dun & Bradstreet.77
76
To reduce the potential influence of outliers, we recoded values above the 99th
percentile to the 99th percentile value. The county is the smallest geographic unit for
which we could locate voting data across the United States.
77
Funding constraints prevented us from obtaining data on revenues and ownership status
for all facilities in the sample. We obtained data for all facilities that self-disclosed in our
sample. For the selection analysis, we also obtained data on a random sample of the
remaining facilities. For the treatment analysis, we obtained data for all facilities that
were in the same industry (4-digit SIC.Code) and state as the self-disclosing
organizations.
129
To measure pollution, we created two variables based on emissions data from the
US EPA’s Toxic Release Inventory program.78 We summed each facility’s annual
emissions of toxic chemicals released to all media (air, water, land, and underground
injection), and took the log of this sum after adding 100.79 Using the sum of these
emissions mimics how the US EPA and the media rank the “dirtiness” of facilities. We
also estimate the health risk posed by TRI emissions released to air by incorporating
Chronic Human Health Indicator (CHHI) weighting factors pertaining to inhalation
exposure from the US EPA’s Risk-Screening Environmental Indicators (RSEI) model
(Toffel & Marshall 2004; US EPA 2002b).80 To calculate the facility’s annual health
risk score, we multiplied the pounds of each toxic chemical emitted to air by its CHHI
weighting factor, summed these products, and took the log of this sum after adding 100.
78
US
EPA
Toxic
Release
Inventory
data
is
available
at
http://www.epa.gov/tri/tridata/index.htm
79
Because the list of TRI chemicals changed during our sample period, we included only
“core chemicals”: those that were continually required and whose reporting thresholds
remained constant. We added 100 before taking the log to reduce the influence of large
disparities in log points between differences in very small emissions values.
80
The US EPA Risk-Screening Environmental Indicators (RSEI) model is described at
http://www.epa.gov/oppt/rsei/
130
Emissions are strongly associated with facility production levels, but facility
production levels are not publicly available for the industries in our sample. As a proxy
for production, we create a production index based on two variables: (1) average facility
employment in 1997 for each 4-digit SIC Code (as described above), and (2) annual
facility production ratios, which is the ratio of its production in the current year to its
production in the prior year. After obtaining production ratios from the TRI dataset,81 we
calculate the production index in three steps. First, we set production index equal to
facility employment for 1997. Second, in an iterative fashion for each year after 1997, we
calculate the production index by multiplying the prior year’s production index by the
current year’s production ratio. Third, iteratively for each year before 1997, we calculate
the production index by dividing the subsequent year’s production index by the
subsequent year’s production ratio.82
81
In the models, I employ log production index. To avoid overemphasizing differences
between small amounts, I added 100 before taking the log.
82
For example, suppose production ratios during 1995-1999 are 1.2, 1.3, 0.8, and 2,
respectively and facility employment is 100 in 1997. First, we set production index equal
to 100 in 1997. Second, we calculate production index for subsequent years as follows:
for 1998, 100 × 0.8 = 80; then for 1999: 80 × 2 = 160. Third, we calculate production
index for years prior to 1997 as follows: for 1996, 100 ÷ 1.3 = 76.9; then for 1995, 76.9 ÷
1.2 = 64.1.
131
To control for potential unobserved differences between industries, we create
dummy variables for each 2-digit SIC Code to reflect the facility’s industry. We also
control for the possibility that the composition of the federal judiciary might affect the
self-reporting decisions of companies located in different appellate jurisdictions.83 To
measure Federal Circuit Court ideology during the early 1990s, we obtained data about
every Court of Appeals judge who served during 1990-1994 (the last year for which data
are available) from Zuk, Barrow & Gryski (1996). We calculated the proportion of
judges on each Circuit in each year that were nominated by a Democratic President, since
presidential party is widely viewed to be highly correlated with a judge’s political
ideology (Humphries & Songer 1999; Tate & Handberg 1991; Spence & Murray 1999;
Schultz & Petterson 1992). We then calculated the average for each Circuit over 1990-
83
Numerous studies have shown that judges’ political ideology affects their decision-
making on a wide range of issues (Sunsten, Schkade & Ellman 2004, Cross 2003),
including environmental issues (Malmsheimer & Floyd 2004, Revesz 2001). Studies
have also shown that firm-level decisions are responsive to the prevailing ideology of the
circuit (Guthrie and Roth 1999). While circuit ideology seems likely to effect selfreporting behavior, we lack a robust theory of how that influence would be felt, and a
meaningful measure of its influence is beyond the scope of this paper. Consequently, we
include it here as a control variable.
132
1994. Our coding of this variable is presented in Table 3.3. 84
Variable definitions, descriptive statistics and correlations are provided in Tables
3.4 - 3.6.
3.5. Empirical Models and Results
To identify which factors influence facilities to self-disclose violations, we model
self-disclosure as a dichotomous decision made by each facility in each year. We employ
a pooled probit model with dummies to control for differences between industries (2-digit
SIC Code), years, and EPA Regions.85
Table 3.7 presents the probit results of the participation analysis, with robust
standard errors clustered by facility. Because we could obtain revenues and ownership
data for only on a small subset of our sample, we employ two alternative specifications.
Model 1 excludes those variables to avail a much larger sample size; Model 2 includes
84
While we could have used just the most recent value pertaining to each Circuit Court
(1994), we felt that using the average of the prior 5 years (1990-1994) better reflects the
overall impression of the court and reduces the influence of any aberrations that may
occur in any single year.
85
A recent report by the US General Accounting Office noted substantial variation across
EPA Regions in terms of inspection coverage, enforcement staff, the number and type of
enforcement actions taken, and criteria used to determine penalty assessments, and the
size of penalties assessed (US GAO 2000).
133
them. With few exceptions, which we describe below, the statistical significance of
determinants was robust to these different specifications.
The statistically significant positive coefficients on inspections and enforcement
actions support our hypothesis that specific deterrence measures encourage selfdisclosure. The results of Models 1 and 2, respectively, suggest that an additional RCRA
inspection increases the probability of self-disclosure the next year by 16% to 26%,86 and
that an additional CAA inspection increases this probability by 7% to 14%. Being
subject to at least one enforcement action—a much rarer event—had a much greater
influence on disclosure, as it increased the likelihood of self-disclosing the next year by
70% to 164%.87
We found no evidence that the number of violations had any influence on the
decision to self-disclose a violation the subsequent year (controlling for inspections). The
86
The marginal effects, evaluated at the mean value of all variables, of an additional
RCRA inspection range from 0.0007 (Model 1) to 0.0039 (Model 2). We evaluate these
marginal effects in the context of the probability of disclosure evaluated at the mean
value of all variables, which is 0.0045 in Model 1 and 0.01476 in Model 2. As such, our
results indicate that an additional RCRA inspection increases the probability of disclosure
in the subsequent year by 16% (Model 1) to 26% (Model 2).
87
The marginal effect of this dummy variable, evaluated at the mean values of all
variables, ranged from 0.0104 (Model 2) to 0.0074 (Model 1).
134
two violations variables were also not jointly significantly different from zero at
conventional significance levels.
As for general deterrence mechanisms, facilities targeted by a US EPA
Compliance Incentive Program were significantly more likely to self-disclose violations.
A facility targeted by both a Compliance Incentive Program and a National Priority
Sector was 1.1 (Model 2) to 2.2 times (Model 1) more likely to self-disclose a violation
that year than a facility targeted by neither program. However, a facility targeted by just a
Compliance Incentive Program (and not a National Priority Sector) was 10 to 17 times
more likely to self-disclose a violation that year than a facility targeted by neither
program.88 We found no evidence that a facility targeted only as a National Priority
Sector—and not simultaneously targeted by a Compliance Incentive Program—was any
more likely to self-disclose than facilities that were not targeted by either program.
We found little evidence that state-level statutes providing immunity or privilege
encouraged facilities to self-disclose violations.89 We also found no evidence that any of
88
The marginal effect divided by the probability of disclosure evaluated at mean of all
variables is 0.1540 ÷ 0.0148 = 10.4 for Model 2, and 0.0758 ÷ 0.0045 = 16.8 for Model 1.
89
The coefficients on these three variables were also not jointly significantly different
from zero at conventional significance levels, nor were their sum. Because our statutory
variables were measured at the state-level, we re-ran our models clustering the standard
errors by state. The statistical significance of these coefficients was unchanged.
135
our three measures of community pressure (population density, average per capita
income, and voter turnout) had any affect on facilities’ decision to self-disclose
violations.90
We found support for our hypothesis that more prominent facilities were more
likely to self-disclose violations. A one-unit increase from the mean of log facilityindustry employment is associated with a 36% increase in the likelihood of selfdisclosing, while a one-unit increase from the mean of log facility revenues is associated
with a 13% increase in the likelihood of self-reporting.91 While the coefficient on the
public-owned dummy is positive as predicted, it is not statistically significant. As such,
we find no evidence that a facility’ parent company being publicly-owned (versus
privately-owned) had any influence on self-disclosing.
90
The coefficients on these three variables were also not jointly significantly different
from zero at conventional significance levels, nor were the sum.
91
Because facility employment was measured at the 4-digit SIC Code level, we re-ran
our models clustering standard errors by 4-digit SIC Codes. The statistical significance of
this coefficient was unchanged. The marginal effects of log facility employment,
evaluated at the mean values of all variables, ranged from 0.0016 (Model 1) to 0.0074
(Model 2). As such, a one-unit increase from this variable’s mean value is associated with
36% (Model 1) to 37% (Model 2) increase in the likelihood of self-disclosing. The
marginal effect of log facility revenues is 0.0019.
136
3.6. Discussion
Our findings suggest that even as voluntary industry self-regulation programs
proliferate, government still has an important role to play. We have shown that violators
are more likely to self-report when they are subject to regulatory pressure, including
being inspected, punished and targeted by focused compliance initiatives. In fact, selfreporting is not deterred even by ostensibly hostile relations with regulators. Firms that
recently experienced enforcement actions, which involve significant legal costs and often
result in penalties and injunctive relief, are much more likely to self-disclose than those
with fewer compliance problems. In addition, we find little evidence that state-level legal
protections that seek to incentivize self-reporting by tying local regulators’ hands
encourage firms to self-report. Together, these findings support a regulatory policy that
recognizes the ongoing importance of state regulation and regulators to the success of
public-private regulatory partnerships. Contrary to what many have suggested, we find
no support for the argument that the state should continue to withdraw from the
regulatory arena, leaving firms to police themselves without supervision.
Our finding that more inspections encourage self-disclosure complements related
research showing that more frequent inspections improve compliance (Magat & Viscusi
1990; Laplante & Rilstone 1996, Gunningham, Thornton & Kagan 2005; Gray &
Shadbegian 2005; Kuperan & Sutinen 1998; Winter & May 2001; Braithwaite & Makkai
137
1991).92 Building on this prior work, our results suggest that such coercive, deterrent
regulatory techniques may continue to be necessary even in a more cooperative
regulatory environment.
Because regulators consider a facility’s “motivation” and
“willingness to comply” when prioritizing their enforcement activities (US EPA 1992),
facilities that have faced more inspections recently may be particularly keen to convey a
“pro-compliance” image to convince regulators of their motivation and willingness to
comply. One way facilities may attempt to do this is by self-disclosing their violations
and agreeing to bolster their internal audit processes. Given these results, it remains clear
that the Audit Policy supplements, but cannot replace, regulatory inspections. What the
US EPA noted in 1990 apparently still holds in the era of voluntary self-policing:
“Inspections remain the backbone of agency compliance monitoring programs….Even
with widespread requirements for self-monitoring, inspections play a major role in
assuring quality and lending credibility to self-monitoring programs” (Wasserman 1990).
Our general deterrence results similarly suggest the ongoing importance of
regulatory oversight to the success of self-policing.
Self-reporting was more likely
among facilities targeted by US EPA Compliance Incentive Programs, which are often
announced directly to target firms through letters or trade associations and typically offer
technical compliance assistance along with the incentives of the Audit Policy. On the
92
For example, Magat & Viscusi (1990) found that pulp and paper plants in the United
States halved their non-compliance rates in the quarter following an inspection.
138
other hand, we find no evidence that facilities targeted by industry-wide US EPA
National Priorities were any more likely to voluntarily self-disclose violations than those
in other industries. A number of factors may explain this apparent disparity. For
example, facilities in National Priority sectors might not be aware that they are a target,
since they are not notified via letters from US EPA or trade associations, as typically
occurs with Compliance Incentive Programs. Even if they are aware that they are within a
National Priority sector, such facilities may believe that scrutiny of a broadly defined
industry does not significantly increase the chances of having their violations detected,
whereas Compliance Incentive Programs often target fewer than 100 facilities--and in
some cases as few as 20. In any event, our results suggest an interesting convergence of
compliance/deterrence strategies that has yet to be developed in the literature: general
deterrence is more effective the more targeted or “specific” it is.
Contrary to prior research that showed that more inspector-discovered violations
led facilities to improve their compliance (Gray & Scholz 1993; Helland 1998), we find
no evidence that violation frequency increases the likelihood of self-disclosures. The
mere presence of inspectors at a facility, or the threat of their arrival through targeted
compliance initiatives, apparently encourages self-reporting regardless of what they find
once they get there. This discrepancy may result from differences between compliance
and self-policing behavior: firms previously cited by inspectors for violations only stand
to gain from cleaning up their act and complying with regulations; however, it is less
clear whether disclosing additional violations to regulators will engender goodwill.
139
Reinforcing our argument that regulators remain relevant within a regime of selfregulation, we found no evidence to suggest that state regulatory authorities should forfeit
their own parallel enforcement activities. Facilities shielded by state-level statutory audit
privilege, immunity, or even both, were no more likely to self-report violations than
facilities in states with neither form of protection. Our findings confirm the results of a
1998 survey reporting that facilities self-disclosed violations and conducted internal
audits at the same rate in states that did and did not provide immunity or audit privilege
protection (Morandi 1998). These results demand a thoughtful re-examination of the
many economic and policy arguments in support of secrecy for audit materials and
broader protections for corporate polluters. US EPA and environmental groups have long
resisted the enactment of audit privilege laws on the grounds that they deprive the public
of access to information that is crucial to health and safety and make discovery of and
prosecution for unreported violations much more difficult (Bedford 1996, Woodall 1997).
As the US EPA said in announcing the Audit Policy, "privilege, by definition, invites
secrecy, instead of the openness needed to build public trust in industry's ability to selfpolice" (US EPA 1995a:66710). Access to internal facility data is especially important in
a regulatory system that has come increasingly to rely on information disclosure as its
own instrument of compliance. Rechtshaffen (2004), for instance, discusses the success
that US EPA has had in improving Clean Water Act compliance by publicly disclosing
information about facility performance. Restricting access to such information through
an evidentiary privilege deprives regulators and the public of a potentially valuable
140
compliance tool without providing any countervailing benefit in the form of increased
self-reporting.
Contrary to several existing studies and to our own expectations, our study
yielded little evidence that potential community pressure influenced company decisions
to participate in self-policing activities. Our “non-result” may be due to the inadequacy
of our measures or to more complex relationships than those we modeled. For example,
the influence of well-organized communities may work through regulators, with
communities applying pressure on regulatory enforcement agencies, who respond by
increasing inspection frequency and the likelihood of pursuing enforcement actions
against facilities. Our model cannot distinguish the impact of such community pressure
from the regulatory pressure it produces. Selection bias may also mute the impact of our
community pressure variables if companies choose where to locate their facilities based
(at least in part) on expectations about their ability to meet community demands. Finally,
self-policing may exhibit a different dynamic than compliance in this arena as well.
Perhaps firms under pressure from their communities are conflicted about whether selfdisclosing will enhance or detract from their legitimacy and corporate citizenship
reputation. Self-disclosing offers a double-edged sword. On the one hand, if stakeholders
focus a facility’s disclosing the violation, the facility’s citizenship reputation may be
enhanced because they decided not to hide the violation.
On the other hand, if
stakeholders focus on the fact that the facility had a violation to disclose, this could dispel
the impression that the facility was in full compliance. It is possible that our non-result
represents these two opposing factors, effectively canceling each other out.
141
We found that more prominent facilities—whether measured by more employees
or revenues—are more likely to self-disclose violations. This may result because larger
organizations have a greater need to preserve their reputation (May 2004)—especially
since compliance reputation may spillover between facilities of the same firm and
because they believe their heightened public scrutiny reduces their ability to hide their
violations (Scott & Meyer 1983).93 Together, greater costs (to their legitimacy) of failing
to report combined with a heightened probability of having violations detected amount to
a compelling rationale for larger facilities to self-disclose. More generally, our finding is
consistent with other empirical studies that have found that larger organizations are more
likely to acquiesce to institutional pressures (e.g., Goodstein 1994; Ingram and Simons
1995).
3.7. Future Research and Conclusions
Our results suggest several avenues of future research. More empirical research is
needed to better understand the influence of communities on companies’ self-regulation
behavior. Future research would benefit by leveraging qualitative research that describes
how communities influence facilities’ environmental behavior (e.g., Gunningham,
Thornton & Kagan 2005) to develop large-scale quantitative measures of actual
community pressure to improve upon our measures of potential community pressure.
Future research evaluating the influence of community pressures on companies’ behavior
93
Teasing these factors apart represents an opportunity for future research
142
should also examine the extent to which companies select facility sites based on their
expectations about community pressures and their ability to meet community demands.
A better understanding of this process could effectively address the endogeneity concerns
we raised in discussing our results.
While we have identified a number of factors that influence self-reporting, there
are undoubtedly more explanations for this complex behavior. Future research might
attend to other factors such as the composition of Boards of Directors or the prior
employment of key managers. It would also be useful to learn whether institutionalized
compliance mechanisms like Environmental Management Systems encourage selfreporting.
Such research could compare and contrast the dynamics between self-
reporting and compliance.
Self-regulation and self-policing have been touted as a new paradigm of
regulation that trades outmoded “command-and-control” strategies for industry-directed,
market-based solutions. While it is hard to deny that there are benefits to fostering more
cooperative relationships between the regulators and the regulated, our research counsels
caution in the face of arguments that coercive regulatory strategies are ineffective or
obsolete and that government should cede to corporations the unfettered authority to
regulate themselves.
Offered the option of self-policing under the Audit Policy,
companies were apparently willing to come clean only under the threat that they might be
caught instead.
Even as corporations are given an expanding role in their own
governance, our study shows that the success of “voluntary” self-policing depends on the
continued involvement of regulators with coercive powers.
143
Future research should focus on whether regulators and communities can rely
upon indicators of self-policing, such as disclosing violations to the Audit Policy, to
conclude that such facilities are more likely to be in compliance with their legal
obligations. If so, an important subsequent question is whether self-policing enables
facilities to reduce their pollution intensity.
144
TABLE 3.1
Facilities Disclosing Violations To The Audit Policy
Source:
1997
1998
1999
2000
2001
2002
2003
Total during 1997-2003
(1)
All facilities
(2)
(3)
(4)
a
All facilities TRI Reporters Entire sample b
EPA Reports
457
950
990
Our dataset
153
251
412
773
603
777
520
1,754
614
3489
Our dataset
62
103
222
412
296
187
171
1453
Our dataset
43
61
100
200
161
104
79
748
Data in column 1 were obtained from various US EPA reports and newsletters that provided updates on
participation in the Audit Policy. The other columns reflect facility-level data we obtained from three
sources via Freedom of Information Act requests: the US EPA Integrated Compliance Information System
(ICIS) database, the (hardcopy) US EPA Audit Policy Docket, and lists of facilities that disclosed under the
Audit Policy in response to the Compliance Incentive Programs. According to discussions with US EPA,
the disparities between the database we constructed from US EPA databases and documents and US EPA’s
own aggregate figures are likely due to several factors, including: (1) their reports typically refer to fiscal
years rather than calendar years; (2) US EPA does not necessarily enter facility-level data into their
databases when a corporation simultaneously discloses tens or hundreds of violations across multiple
facilities; and (3) data coding errors or omissions.
a
This column refers to facilities that report data to the US EPA Toxic Release Inventory (TRI) program.
b
This column refers to the sample used in our study: facilities that report data to the US EPA Toxic
Release Inventory (TRI) program and are subject to hazardous waste regulations pursuant to the Resource
Conservation and Recovery Act (RCRA) and air regulations pursuant to the Clean Air Act (CAA).
145
146
TABLE 3.3
Federal Circuit Court Ideology
Proportion of judges on each Circuit that were nominated
by a Democratic President, average during 1990-1994
US Circuit Court
1
2
3
4
5
6
7
8
9
10
11
DC
12.50%
38.89%
26.67%
37.50%
36.36%
35.29%
16.67%
30.77%
46.67%
41.67%
46.15%
42.86%
Source: Based on data from Zuk, Barrow & Gryski (1996)
147
TABLE 3.4
Variable Definitions
Variable
Definition
Self-disclosure via the Audit Policy
Dummy coded 1 in a year when a facility discloses a violation
under the US EPA Audit Policy
Number of inspections per year pertaining to the US Resource
Conservation and Recovery Act
Number of inspections per year pertaining to the US Clean Air Act
Number of violations per year pertaining to the US Resource
Conservation and Recovery Act
Number of violations per year pertaining to the US Clean Air Act
Dummy coded 1 in a year when the US EPA brought an
enforcement action against a facility
Dummy coded 1 in a year when the facility was among those
targeted by a US EPA Compliance Incentive Program
Dummy coded 1 in a year when the facility’s industry was named
as US EPA National Priority sector
Dummy coded 1 in years when a facility’s state provides statutory
audit privilege
Dummy coded 1 in years when a facility’s state provides statutory
immunity
Population density during 2000 of the facility’s 2000 Census Tract
Per capita income in 1999 of the facility’s 2000 Census Tract
Percent of population who voted for US President in the 2000
elections in the facility’s county
Average employees per facility in 1997 in the facility’s 4-digit SIC
Code
Facility revenues in 2004
Dummy coded 1 if the facility was owned by a publicly-traded
company in 2004
Percent of US Circuit Court’s judges in 1990-94 appointed by
Democratic presidents in the facility’s US Circuit
In 1997: Average employees per facility in 1997 in the facility’s 4digit SIC Code. In other years: adjusted by annual production
ratios, which is a facility’s production level divided by its
production in the prior year
Pounds of toxic chemicals released to the environment
Index of risk to human health associated with facilities’ releases of
toxic chemicals to air
RCRA inspections
CAA inspections
RCRA violations
CAA violations
Any enforcement actions
Compliance Incentive Program target
National Priority sector
Audit privilege
Immunity
Population density
Per capita income in 1999
Voter turnout
Facility employees
Facility revenues
Publicly owned
Circuit Court ideology
Production index
TRI emissions
Health hazard score
148
TABLE 3.5
Summary Statistics
Variable
Facility-.
Mean
year
observations
Self-disclosure via the Audit Policy (this year)
RCRA inspections
CAA inspections
RCRA violations
CAA violations
Any enforcement actions
CIP target and National Priority sector (dummy)
CIP target but not National Priority sector (dummy)
National Priority sector but not CIP target (dummy)
State provides audit privilege but not immunity (dummy)
State provides immunity but not audit privilege (dummy)
State provides audit privilege and immunity (dummy)
Log population density (plus 10) (census tract)
Log per capita income in 1999 (plus 1) (census tract)
Percent who voted for 2000 president (county)
Log facility employees in 1997 (plus 10) (SIC4)
Log facility revenues in 2004
Publicly owned in 2004 (dummy)
Proportion of US Circuit Court’s judges in 1990-94
appointed by Democratic presidents (US Circuit)
Log RCRA penalties (plus 1)
Log CAA penalties (plus 1)
Log pounds of TRI emissions
Log health hazard score of TRI air emissions
SD
89285
78261
89285
78261
89285
89285
89285
89285
89285
89285
89285
89285
89285
89285
89285
89285
11281
16332
89285
0.01
0.45
1.18
0.42
0.05
0.03
0.01
0.02
0.14
0.14
0.05
0.30
2.30
9.72
0.52
4.40
17.31
0.55
0.33
0.09
0.87
1.61
1.33
0.22
0.16
0.07
0.13
0.35
0.35
0.21
0.46
0.00
0.87
0.08
0.77
2.01
0.50
0.10
78261
89285
78261
78261
0.08
0.33
5.63
7.84
0.92
1.78
4.99
7.03
Min
Max
(For dummies,
number coded 1)
703
0
0
0
0
0
4
6
7
3
1
480
1580
12758
12924
4082
27137
2.30
2.31
0
12.05
0.05
0.79
2.80
7.78
10.37 23.96
9046
0.13
0.47
0
0
0
0
16.30
17.50
19.86
27.00
RCRA= Resource Conservation and Recovery Act; CAA = Clean Air Act; CIP = Compliance Incentive
Program; TRI = Toxic Release Inventory.
149
150
TABLE 3.7
Who Participates in the Audit Policy?
Dependent variable: Probability of self-disclosure
Specific
deterrence
RCRA evaluations 1 year ago
CAA inspections 1 year ago
RCRA violations 1 year ago
CAA violations 1 year ago
Any enforcement actions 1 year ago
(dummy)
General
CIP target & National Priority sector
deterrence
(dummy)
CIP target but not National Priority
sector (dummy)
National Priority sector but not CIP
target (dummy)
State provides audit privilege but not
Statutory
provisions
immunity (dummy)
State provides immunity but not audit
privilege (dummy)
State provides audit privilege and
immunity (dummy)
Community Log population density (census tract)
pressure
Log per capita income (census tract)
Voter turnout (county)
Prominence Log facility employees (SIC4)
Log facility revenues
Publicly owned (dummy)
Observations (facility-years)
Facilities
Probability of disclosure, at mean of all variables
(1)
(2)
Probit
dF/dX
Probit
dF/dX
coefficient
coefficient
0.05
0.0007
0.10
0.0039
[0.02]**
[0.04]*
0.03
0.0003
0.06
0.0021
[0.01]*
[0.02]**
0.01
0.0001
-0.04
-0.0016
[0.01]
[0.03]
0.05
0.0007
-0.11
-0.0038
[0.06]
[0.15]
0.0074♦
0.0104♦
0.36
0.22
[0.06]**
[0.13]+
0.0050♦
0.0330♦
0.27
0.51
[0.09]**
[0.22]*
1.22
0.0758♦
1.23
0.1540♦
[0.08]**
[0.21]**
-0.09
-0.0011♦
-0.24
-0.0077♦
[0.06]
[0.16]
0.06
0.14
0.0008♦
0.0058♦
[0.07]
[0.15]
0.04
0.0005♦
0.40
0.0219♦
[0.10]
[0.21]+
0.05
0.0007♦
0.04
0.0014♦
[0.04]
[0.11]
-79.59
-1.0447
-322.12
-12.0306
[136.13]
[472.72]
-0.02
-0.0003
-0.02
-0.0008
[0.02]
[0.04]
0.28
0.0037
1.20
0.0448
[0.28]
[0.65]+
0.12
0.0016
0.15
0.0055
[0.03]**
[0.06]*
0.05
0.0019
[0.02]*
0.05
0.0020♦
[0.08]
89285
6849
13726
1021
0.0045
0.0148
dF/dx is the change in the probability of adoption, evaluated at the mean all variables, based on an
infinitesimal change in each independent variable or a discrete change in each dummy variable (which are
denoted by ♦). Brackets contain robust standard errors clustered by facility. ** p<0.01; * p<0.05; +
p<0.10. All specifications include dummy variables to control for industry (2-digit SIC Codes), EPA
Region, Year, and the proportion of US Circuit Court judges appointed by a Democratic President. RCRA
= Resource Conservation and Recovery Act; CAA = Clean Air Act; CIP = Compliance Incentive Program;
TRI = Toxic Release Inventory
151
Chapter 4.
4. Strategic Management of Product Recovery§
A growing concern to durable product manufacturers is how to manage the
products they manufacture once they have reached their EOL (end of life). In part, this
attention is motivated by legislation enacted by a growing number of countries across
Europe and East Asia that imposes greater responsibilities on manufacturers for
managing their EOL products and by related bills that have been introduced in nearly half
of the 50 state legislatures in the United States.94 These “product take-back” laws are
intended to give manufacturers incentives to implement design changes that reduce the
environmental burden of their products at EOL, while also removing a growing waste
management cost from municipal governments. Take-back regulations have targeted
waste packaging, batteries, automobiles, and a variety of electrical and electronic
equipment (including appliances, computers, lighting, and medical equipment) (Toffel,
2003). Instead of simply banning these products from landfills and incinerators, takeback laws encourage manufacturers to refurbish, remanufacture, and recycle products.
For example, product designers can select assembly mechanisms that can be easily
§
Copyright © 2004, by The Regents of the University of California. Reprinted from the
California Management Review, Vol 46, No. 2. By permission of The Regents.
94
For a description of various forms of take-back regulations, see Toffel (2003) and
Californians Against Waste (2003).
152
disassembled, that use fewer and less hazardous materials, and are more readily
recyclable. Designers can also reduce the cost of assessing the quality of components
harvested from EOL products by implanting data logs that reveal how intensively they
have been used (Klausner, Grimm, & Hendrickson, 1998), and they can facilitate material
identification during product recovery by requiring the material composition of all
components to be clearly labeled.
EOL product recovery consists of several sequential activities: collecting EOL
products (reverse logistics); determining the potential for the product’s reuse,
disassembling the product, and segregating valuable components from scrap (collectively
referred to as primary recycling) (National Safety Council, 1999; White, Masanet, Rosen,
& Beckman, 2003); remanufacturing components; recycling materials; and disposing the
residual as municipal solid waste or hazardous waste (National Safety Council, 1999). In
many industries, independent firms have long recovered EOL products to refurbish or
remanufacture to supply aftermarkets. Many industries are highly fragmented, and most
firms engaged in remanufacturing are small, independent, and privately owned (Hauser &
Lund, 2003; Thierry, Salomon, Van Nunen, & Van Wassenhove, 1995).95 However, in
95
For example, the remanufacturing industry in the United States has been estimated to
be composed of 73,000 firms with nearly a half million employees. Of these,
approximately 50,000 remanufacture automotive parts, 13,000 remanufacture electrical
apparatus, and 6,500 remanufacture toner cartridges (Lund, 1996)
153
industries where original equipment manufacturers (OEMs) are also remanufacturers,
they are often the largest both in terms of sales and employment (Hauser & Lund, 2003).
While acknowledging that non-OEMs dominate remanufacturing markets in many
industries, this article focuses on OEM decisions of whether and how to engage in the
first five stages of EOL product recovery (the latter two are seldom performed by
OEMs). A growing number of OEMs are facing this decision due to legislative mandates
as well as a host of market and non-market factors. Voluntary take-back programs have
been initiated by manufacturers of carpets (Fishbein, 2000; Lave et al., 1998; Louwers,
Kip, Peters, Souren, & Flapper, 1999), batteries (McMichael & Hendrickson, 1998),
automotive parts (Hammond, Amezquita, & Bras, 1998; Hormozi, 1997; Thierry et al.,
1995), packaging (Rondinelli, Berry, & Vastag, 1997), tires (Ferrer & Whybark, 2003;
Guide & van Wassenhove, 2003), and various electronic products (including wired and
cellular telephones, power tools, photocopiers, and computers) (Ayres, Ferrer, & Van
Leynseele, 1997; Baumgartner, 1996; Davis, Wilt, Dillon, & Fishbein, 1997; de Ron &
Penev, 1995; Goggin, Reay, & Browne, 2000; Guide & van Wassenhove, 2001; Guide &
van Wassenhove, 2003; Kerr & Ryan, 2001; Klausner & Hendrickson, 2000; Low,
Williams, & Dixon, 1996; Thierry et al., 1995; White et al., 2003). A recent
remanufacturing industry report noted that OEMs “are becoming increasingly aware of
the profit opportunities afforded by remanufacturing. In addition to the profit potential,
remanufacturing provides feedback on product failure modes and durability, and it
permits the firms to maintain brand reputation” (Hauser & Lund, 2003).
154
4.1 Motives for Voluntary Product Recovery
Even before the emergence of take-back laws, some firms were already engaging
in voluntary product recovery. Their motives have included: reducing their production
costs, enhancing their brand image, meeting changing customer expectations, and
protecting their aftermarkets. Since the arrival of take-back legislation, firms are also
motivated to prevent their scope from broadening and to preempt additional legislation.
4.1.1 Reducing Production Costs
Some companies have discovered that components and materials from EOL
durable products can often be refurbished to substitute for virgin parts to be used as
spares or in remanufacturing. For example, Xerox Corporation saves hundreds of million
of dollars a year by disassembling its EOL photocopiers and then cleaning, sorting, and
repairing components and recycling residual materials. Mercedes-Benz accepts and
disassembles EOL Mercedes vehicles to harvest and sell spare parts to both consumers
and commercial customers at a significant discount compared to virgin spare parts.96 In
1999, Ford Motor Company began buying salvage yards in the U.S., Canada, the United
Kingdom, and Germany to dismantle EOL vehicles to provide a source of spare parts that
were cheaper than virgin parts (Blumberg, 2002; Hoffman, 2000; Taylor, 2001).97
96
Mercedes-Benz ATC GmbH, “Mercedes-Benz Ersatzteile, Gebrauchtteile und
Gebrauchtwagen im ATC,” <www.mbatc.de/index.htm>, 2003, accessed June 17, 2003.
97
Ford is apparently divesting from this business (Gibbs, 2002; Recycling Today, 2002).
155
4.1.2 Promoting an Image of Environmentally Responsibility
Companies have also enacted product recovery programs to enhance the
environmental image of their brand. A recent survey evaluated how various management
practices related to labor, philanthropy, local communities, and the environment
influenced consumers’ intention to invest in, work for, or use a company’s products and
services (King & Mackinnon, 2002). Increasing the use of recyclable materials and
becoming an industry leader in developing environmentally sustainable business
practices were perceived as having the greatest positive influence. Implementing a takeback program encompasses both of these activities.
For example, after consumers began referring to Kodak’s single-use cameras as
“disposables” or “throwaways” and the media reported environmental groups’ concerns
of their wastefulness, Kodak and FujiFilm launched a take-back program that recycles
more than 90% of these cameras and reversed the product’s poor environmental image
(Charlton & Vanhorn, 1990; Fishbein, 1998; Yardley, 1989).98 Hewlett-Packard has
received positive media coverage for investing in a recycling infrastructure for EOL
computing equipment (Gaither, 2003; Markoff, 2003). IBM Europe and Xerox have
reported that their product recovery activities have strengthened their brand image
98
Also, see Kodak, “A Tale of Environmental Stewardship: The Single-Use Camera,”
<www.kodak.com/US/en/corp/environment/performance/recycling/suc.shtml>,
updated September 14, 2001, accessed May 15, 2002;
156
last
(Guide, Jayaraman, Srivastava, & Benton, 2000). Such gains are based on legitimate
improvements in environmental performance. For example, on a life cycle basis,
remanufacturing photocopiers consumes 20-70% less materials, water, and energy and
generates 35-50% less waste than conventional manufacturing (Kerr & Ryan, 2001).
4.1.3 Meeting Customer Demands
Customer expectations are driving some OEMs to become increasingly involved
in product recovery in some industries. In the computer industry, the growing trend of
OEMs to lease rather than sell equipment to business customers has created the need to
retrieve equipment when leases expire (National Safety Council, 1999). In addition,
customers are increasingly expecting OEMs who supply “fleets” of computers to
business customers to remove the outdated computers as they install the new ones
(National Safety Council, 1999). Dell collects EOL PCs from their commercial customers
in the U.S. as a service associated with the sale of new equipment. Furthermore, some
companies are recovering EOL products to meet growing customer demand for products
with recycled-content. For example, “Some auto manufacturers are setting ambitious
recycled-content goals for auto parts. The large nylon producers are important suppliers
(automotive products account for about one-third of nylon’s end uses) and can provide
nylon made from recycled carpet for use in auto parts” (Fishbein, 2000). In addition,
some local governments are considering mandating their agencies to purchase only from
firms that recycle or remanufacture their own products (Majumder & Groenevelt, 2001).
157
4.1.4 Protecting Aftermarkets
Aftermarkets refer to the market for parts and accessories to maintain or enhance
a previous purchase, and they are often quite lucrative for OEMs (Klein, 1996). While
independent remanufacturers can attract new buyers into a market by providing “like-new
products at prices that typically range from 45% to 65% of comparable new products”
(Hauser & Lund, 2003) they can also pose a threat to this market that many OEMs highly
value. OEMs may recover their EOL products to deter independent firms from
remanufacturing and selling them, thus preventing potential losses of both market share
and brand image. Hewlett-Packard asks its customers to return their used laser toner
cartridges using the replacement cartridge box that is marked to provide free shipping.
Ford’s and Mercedes’s recent interest in their EOL vehicles can also be viewed as a
strategy to preclude independent competitors from accessing their branded spare parts.
Lexmark offers a “prebate” discount to customers who agree to return their Lexmark
printer cartridges to Lexmark for remanufacturing (Streitfeld, 2003). The terms of sale
prohibit customers of prebate cartridges from selling them to other companies who would
refill, reuse, or remanufacture them. Lexmark says this program is designed to protect its
brand image, claiming its brand is sullied when customers blame its printers for providing
poor print quality when they use cartridges refilled by other companies. According to
Lexmark, this program has boosted their cartridge return rates (Majumder & Groenevelt,
2001). Lexmark has installed security chips in its prebate cartridges and printers that
158
disable printing if these cartridges were refilled by other firms, though this and similar
programs are being challenged in court and in legislatures.99
4.1.5 Preempting Regulation
Some firms have sought to reduce the pressure for new or expanded legislation by
improving their own performance or by attempting to have their trade association impose
more stringent requirements on its entire membership. In the environmental arena,
perhaps the most successful example is the Responsible Care program developed by the
chemical industry to reduce pressure for additional environmental regulation following
several major chemical plant accidents in the mid-1980s.100 Several voluntary take-back
99
After another company sold technology to overcome this technological obstacle,
Lexmark sued and won a preliminary injunction for its claim that its intellectual property
had been violated. The lawsuit remains pendin. (Guy, 2003). Early drafts of California’s
electronics take-back bill called for regulations “prohibiting the use of an electronic or a
mechanical device that prevents, impedes, or limits the reuse, remanufacture, or recycling
of a hazardous electronic device.” See California State Senate, State Bill 20 amended in
Senate, July 29, 2003.
100
The Responsible Care program requires chemical associations in each country to
“develop credible processes to verify that member companies are meeting Responsible
Care expectations” (International Council of Chemical Associations, 2002). Responsible
Care “stands out as the leading (and most influential) example of regulation by industry
159
programs have followed this approach. For example, facing draft take-back regulations,
major manufacturers of power tools that are sold in Germany agreed to voluntarily take
back their EOL products from customers at no charge (Klausner & Hendrickson, 2000).
Similarly, seeking to deter take-back regulations on large appliances, Frigidaire began
working with independent appliance recyclers to determine what design changes it could
make to reduce recyclers’ disassembly costs and bolster recycling rates (Potter, 1996).
More recently, the U.S. rechargeable battery industry responded to a growing number of
landfill bans by states and municipalities, take-back laws by several states, and the threat
of more legislation by establishing the Rechargeable Battery Recycling Corporation
trade association on the U.S. environmental scene today” (Rees, 1997). There are many
examples outside the environmental arena of industry self-regulation intended to forestall
government regulation. For instance, various media trade associations have responded to
calls for government censorship with self-imposed regulations: the National Association
of Broadcasters developed a Code of Conduct, the Motion Picture Association of
America developed a list of “Don’ts and Be Carefuls” in the 1920s and a voluntary rating
system in the late 1960s, and the Record Industry Association of America developed the
Parental Advisory program in the 1980s to include warning labels on records with
explicit lyrics.
160
(RBRC).101 This industry-funded organization takes back and recycles rechargeable
batteries at no cost to consumers. RBRC notes that over 95% of the portable rechargeable
battery power industry across North America is involved in its battery-recycling program.
4.2 Product Manufacturers’ Strategic Choice
Regardless of whether manufacturers choose to engage in product recovery to
reduce production costs, meet customer demands, protect aftermarkets, enhance brand
image, or preempt regulation, they face a strategic choice. Should they contract with
recyclers, establish joint ventures with recyclers, form consortia with competitors,
vertically integrate into product recovery, or simply promote the recycling market
(Toffel, 2003)?
Guide and Van Wassenhove describe several advantages of employing advance
deposit fees (ADFs), credits toward future purchases, cash payments, and leasing (Guide
& van Wassenhove, 2001). Recovery firms that specify a sliding scale of prices they pay
for products with different residual quality levels facilitates product sorting and may even
increase the average quality of recovered products. In addition, they argue that these
101
Florida, Minnesota, and New Jersey have implemented regulations requiring
manufacturers to take-back and manage the disposal of the rechargeable batteries they
produce, while rechargeable battery manufacturers in Rhode Island and Vermont must
ensure that a collection, transportation, and processing system is established (Fishbein,
1998). Also, see U.S. Environmental Protection Agency (2001).
161
mechanisms can decrease recovery firms’ product inventories, reduce disposal costs, and
increase equipment utilization. Where product manufacturers are better positioned than
other recovery companies to employ these tools (e.g., ADFs, discounts on future
purchases, and leasing), they can gain competitive advantages in conducting product
recovery.
Majumder and Groenevelt (2001) describe several advantages manufacturers
possess in retrieving their products from customers, including their ability to provide
trade-in rebates on new equipment and offering prebates. In fact, Dell, Xerox, HewlettPackard, Compaq, and several large appliance and automobile manufacturers provide
trade-in rebates (Corbett & Savaskan, 2003). Lexmark, as mentioned earlier, offers
prebates on its toner cartridges. Dell offers to recycle a customer’s old printer for free
upon their purchase of a new one. Fleischmann (2003) describes several ways the
efficiency of reverse logistics systems can be influenced by whether the OEM or another
party manages the process. For example, he argues that OEMs possess several advantages
in predicting the quality and timing of EOL product flows because they can monitor
equipment usage by using real-time electronic sensors and can forecast return flows
through end-of-lease returns.
Guide et al. describe several reasons why manufacturers may choose to acquire
EOL products from third parties, including buffering themselves against supply
fluctuations to facilitate production planning and improve asset utilization (Guide et al.,
2000). On the other hand, they note that obtaining EOL products directly from customers
can provide manufacturers with better control over EOL product condition and quality.
162
Collecting directly from customers avoids intermediaries who may cherry pick the most
valuable items and supply only the lower quality ones. Savaskan, Bhattacharya, & van
Wassenhove (2004) compare alternative collection methods for manufacturers who
incorporate components from their EOL products into their new products. Their model
shows that, compared to establishing their own reverse logistics network or engaging
other third-parties, manufacturers that provide incentives to retailers to collect their EOL
products will achieve higher collection rates and will encourage retailers to reduce their
prices, thereby increasing sales. As such, higher profitability is predicted for
manufacturers who collect EOL products through their retailer networks instead of
collecting them themselves or contracting with other companies to do so. Ferrer and
Whybark (2001) describe several tradeoffs between a manufacturer’s choice between
conducting remanufacturing within its manufacturing plants (or in its stand-alone
facilities) or whether to outsource remanufacturing to third parties. They focus on
economies of scale, transportation costs, and coordination needs.
4.3 The Role of Recovery Technologies and Supply
Uncertainty
Transaction Cost Economics (TCE) theory predicts the circumstances when firms
govern a particular transaction using the market, a “hierarchical” form (vertical
integration), or a “hybrid” form (joint venture, partnership, or alliance). According to
TCE, this decision depends upon transaction costs—including the costs associated with
identifying transaction partners, negotiating and drafting agreements, monitoring the
163
exchange, and enforcing its terms—and transaction hazards (Williamson, 1985, 1998,
2000). The key hazard of transacting via markets (characterized as spot exchanges
between two unaffiliated organizations) emerges when one party must invest in
transaction-specific assets to conduct the transaction efficiently. Because such
investments lose value when applied to other transactions, whoever makes this
investment becomes dependent upon the other party.102 TCE anticipates the latter would
seek to leverage its position by renegotiating or threatening to “hold-up” the party who
made the investment.
According to TCE, as the need to employ transaction specific assets increases,
“Simple market exchange thus gives way to credible contracting (to include penalties for
premature termination, information disclosure and verification mechanisms, specialized
dispute settlement mechanisms, and the like). Unified ownership (vertical integration) is
predicted as bilateral dependency hazards successively build up” (Williamson, 2002). To
mitigate hold-up risks, hybrids and vertical integration are preferred because they are
better able to “ensure the continued supply of…inputs necessary to keep the specialized
asset fully employed” (Teece, 1984). Hundreds of studies have provided empirical
validation of TCE’s predicting the circumstances under which transactions are most
102
Transaction-specific assets have been defined as investments in physical, temporal,
human, and site assets that lose value when applied to their second best use (Klein,
Crawford, & Alchian, 1978; Williamson, 1983).
164
efficiently governed using vertical integration, markets, or hybrids (Boerner & Macher,
2001; Shelanski & Klein, 1995). As such, TCE can shed insight on the circumstances
where product recovery transactions are more efficient by having OEMs vertically
integrate, rely on third-parties, or develop hybrids such as consortia, alliances, or joint
ventures.
4.3.1 Product Recovery Investments
Various types of equipment and training can bolster the productivity of EOL
product recovery by reducing the cost of assessing, disassembling, or identifying valuable
components in EOL products. When a technology improves the productivity of
recovering only one particular product, its value may decline significantly should it no
longer be applied to that product. For example, consider a specialized machine designed
to disassemble a particular EOL product, where the machine’s value would depreciate if
applied to any other transaction. According to TCE, when productive EOL product
recovery requires transaction-specific investments, hold-up risks are better mitigated by
joint ventures or vertical integration than by relying on markets. An example of a large
investment that features high transaction specificity is Signature Analysis technology,
which facilitates the comparison of noise, heat, or vibration produced by a
165
remanufactured Xerox machine to the company’s new product quality standards.103 The
unified governance by Xerox Corporation in this case is aligned with TCE predictions.
On the other hand, some EOL product recovery investments are not transaction
specific, such as those that improve disassembly productivity across a range of products.
For example, SpectraCode’s Polymer Identification System technology improves
disassembly productivity by quickly identifying plastic polymers used in electronic
products (The Economist, 2001). Because it can be applied to a wide variety of EOL
products, investing in this technology is not transaction-specific and thus neither the
developer nor any individual buyer is subject to hold-up risk. Indeed, it was developed to
facilitate the recycling of both EOL electronics and automotive products.104 In
accordance with TCE predictions, a wide variety of independent primary recyclers and
OEMs are purchasing this technology from an independent developer (Licking, 1998).
4.3.2 Uncertainty
TCE posits that hold-up risks accompanying transactions featuring asset
specificity are exacerbated as transaction uncertainties increase (Williamson, 1985).
103
Xerox Corporation, “Xerox Equipment Remanufacture and Parts Reuse,”
<www2.xerox.com/go/xrx/about_xerox/about_xerox_detail.jsp?view=editorial&id=2714
6>, undated, accessed February 18, 2002.
104
Spectracode,
“Applications:
Plastic
<www.spectracode.com/practical5.html>, accessed September 16, 2003.
166
Recycling,”
Thus, exchanges that feature both transaction-specific assets and high uncertainty are
especially likely to be governed by hybrid mechanisms or by single firms via vertical
integration.105 Reverse supply chains associated with product recovery are subject to
much more uncertainty than forward supply chains for at least seven reasons: “(1) the
uncertain timing and quantity of returns, (2) the need to balance demands with returns,
(3) the need to disassemble the returned products, (4) the uncertainty in materials
recovered from returned items, (5) the requirement for a reverse logistics network, (6) the
complication of material matching restrictions, and (7) the problems of stochastic
[random] routings for materials for repair and remanufacturing operations and highly
variable processing times” (Guide, 2000).106 Managers are significantly challenged to
accurately predict and control the supply of many EOL products—a key success factor
for profitable product recovery (Thierry et al., 1995). A survey of production-planningand-control practices in remanufacturing indicated that a primary cause of late deliveries
of customer orders was a lack of available components from EOL products (Guide et al.,
2003).
Remanufacturers may be able to reduce the high variation in the quality of EOL
products they receive by offering financial incentives to those who return products at a
105
For example, Carter and Ellram (1998) argue that greater environmental uncertainty
will lead to increased vertical coordination among suppliers and buyers.
106
Also, see Guide et al. (2003 ) and Guide and Van Wassenhove (2001).
167
specified quality level (Guide & van Wassenhove, 2001). Companies that provide a
schedule of prices across various quantities and qualities of end-of-use or EOL goods
include ReCellular (cell phones) and Dell (computer equipment) (Guide & van
Wassenhove, 2001).107 To facilitate identifying the residual quality of components in
EOL products, Robert Bosch GmbH has installed electronic data logs in their power tools
to record their usage history, and similar data logs are being developed for other products
such as large household appliances (white goods) (Klausner et al., 1998; Simon, Bee,
Moore, Pu, & Xie, 2001).
Even if these tactics achieve their objective of reducing quality variation or the
cost of detecting the residual quality of EOL components, firms seeking to rely on
recovered products as a key ingredient to manufacturing still face greater uncertainty
107
Dell’s Asset Recovery Services provides a “Used Equipment Purchase Price”
schedule
on
a
monthly
basis.
See
Dell,
“Welcome
to
Tradeups,”
<www.dell.tradeups.com>, accessed September 18, 2003; and Dell, “Asset Recovery
Frequently
Asked
Questions,”
<www.dell.com/us/en/gen/services/asset_004_assetrecovery.htm>, accessed September
18, 2003. Industry grading schemes have also been developed to categorize scrap timber
products sold for remanufacturing. See Western Wood Products Association, “WWPA
Online
Technical
Guide:
Western
Lumber
Grades
and
Quality
<www.wwpa.org/techguide/grades.htm>, 1997, accessed September 18, 2003.
168
Control,”
surrounding the timing and number of recovered products than those relying on virgin
materials and components. Indeed, remanufacturing firms report maintaining high
inventories of EOL products to buffer against uncontrollable fluctuations in their supply
(Guide, 2000; Nasr, Hughson, Varel, & Bauer, 1998). As Guide (2000) recently noted,
“Forecasting models designed to predict the availability of returns are needed to reduce
some of the uncertainty.” Indeed, a recent survey of executives of remanufacturing firms
indicated that they believe one of the greatest threats to their industry is a lack of EOL
products (Guide, 2000).108
Consequently, OEMs that rely upon components or materials harvested,
remanufactured, or recycled from recovered EOL products often face many new sources
of uncertainty. In addition, as independent product recovery companies have little control
over the EOL product return rate, they would be hard pressed to supply exact numbers of
recovered components at precise times with high penalties for late deliveries, as typically
required by manufacturers employing just-in-time (JIT) systems or lean manufacturing
principles. Such concerns are exacerbated when OEMs face difficulties determining
whether a supplier’s claim that it cannot meet a scheduled delivery of recovered
components is a hold-up attempt or the result of legitimate fluctuations in the availability
108
Several quantitative models have been developed to estimate ideal buffer levels for
remanufacturers that face uncertain timing of supplies along with heterogeneous
composition. For example, see Guide and Srivastava (1997).
169
of EOL products. In such cases, manufacturers may have to accumulate stocks or engage
with additional suppliers to buffer against shortfalls from the supplier, adding costs to the
transaction. The additional costs and risks associated with buying EOL product
components from independent firms may convince OEMs to govern the transaction by
developing joint ventures or vertically integrating to gain better access to information and
thus mitigate hold-up risks. This would be consistent with TCE predictions empirically
validated in other domains.
4.4 Leveraging Manufacturing-Associated Capabilities to
Product Recovery
While the TCE analysis provides insight at the transaction level, it is less useful in
explaining why manufacturers in the same industry choose different strategies. The
Resource-Based View of the firm, Dynamic Capabilities, and Core Competencies are
related management theories that provide insight on this question (for simplicity, these
theories will collectively be referred to as the RBV). These theories assert that firms
possess unique sets of resources, capabilities, and competencies—similar terminology for
related concepts—and that these form the foundation of competitive advantage (Barney,
1991; Foss & Knudsen, 1996; Grant, 1991; Mahoney, 2001; Teece, Pisano, & Shuen,
1997; Wernerfelt, 1984). Resources have been defined as “inputs into the production
process” including “items of capital equipment, skills of individual employees, patents,
brand names, [and] finance,”(Grant, 1991) as well as customer loyalty and production
experience acquired from learning-by-doing (Wernerfelt, 1984). Resources that lead to
170
competitive advantage are those that are rare, valuable, difficult to imitate, nonsubstitutable, and costly to transfer across firms (Barney, 1991; Wernerfelt, 1984).
Capabilities, which refer to “the capacity for a team of resources to perform some task or
activity,” can also provide competitive advantage when firms leverage them into new
opportunities (Grant, 1991). Core competencies, defined as “the collective learning in the
organization, especially how to coordinate diverse production skills and integrate
multiple streams of technologies” to significantly enhance a product’s value, are often
difficult for competitors to imitate (Prahalad & Hamel, 1990). Similarly, “the set of
activities that a firm can organize and coordinate better than other firms” have been
termed “distinctive competencies” (Dosi & Teece, 1998).109
According to these theories, transferring tacit knowledge and leveraging related
competencies are easier within firms than between them. Firms create organizing
principles such as coding schemes, values, and common languages that enable them to
outperform markets at sharing and transferring the information and know-how possessed
by individuals and groups within their organization (Kogut & Zander, 1992). As such,
firms internalize activities when tacit knowledge or related competencies are important
109
This notion has been parameterized as the relative strength of ten functional areas:
general management, financial management, marketing/selling, market research, product
research and development, engineering, production, distribution, legal affairs, and
personnel (Snow & Hrebiniak, 1980).
171
drivers of competitive advantage (Conner, 1991; Dierickx & Cool, 1989; Prahalad &
Hamel, 1990; Teece et al., 1997). For example, vertical integration occurs when
competitive advantage requires “inputs that cannot be purchased, such as learning-bydoing and organizational culture” because these are “on average, likely to be more
specific to the firm than purchasable inputs and hence have the potential to be the more
significant rent-generators” (Conner, 1991). Because companies achieve competitive
advantage by leveraging their core competencies into new activities (Prahalad & Hamel,
1990), vertical integration is more likely when their required competencies are
“something about which the firm already has some degree of relevant knowledge”
(Winter, 1988).110 Vertical integration may depend upon “how good a firm is currently at
doing something, how good it is at learning specific capabilities, and the value of these
capabilities as platforms into new markets” (Kogut & Zander, 1992). On the other hand, a
lack of relatedness reduces the likelihood of integration (Conner, 1991; Conner &
Prahalad, 1996). Empirical studies have validated these claims, finding that vertical
integration is much more likely when the integrated activities require similar
technological knowledge, particularly when this knowledge is “partly tacit and teambased and therefore takes significant time to acquire” (Argyres, 1996).
The RBV can offer insight to explain some of the diversity of product recovery
strategies among product manufacturers. The RBV assumes that companies possess
110
Also, see Kogut and Zander (1992) and Guldbrandsen and Haugland (2000).
172
heterogeneous bundles of capabilities and predicts that firms pursue opportunities where
they can leverage their capabilities to secure competitive advantage. Specifically, the
RBV suggests that an OEM’s decision to engage in voluntarily product recovery depends
upon the extent to which it can leverage its existing capabilities. Analysis based on the
RBV thus requires understanding the capabilities involved in product recovery and how
closely these align with capabilities OEMs already possess.
4.4.1 Manufacturing, Service, and Repair Capabilities
To efficiently disassemble EOL products and accurately distinguish reusable,
repairable, recyclable, and non-recyclable components and materials, specialized skills
are often required. As Ferrer and Whybark (2001) note, “A judgment must be made as to
whether an investment in disassembly is warranted. . . . The capability of correctly
making the determination is a key factor in success. Processing bad cores [EOL products]
means that the disassembly investment is not offset by the recovery of enough good parts,
while discarding potentially valuable cores is a waste. This is an area in which
considerable skill is required and experience useful.” Required skills include careful
disassembly to prevent damaging potentially valuable components, quality inspection to
estimate the intensity of prior usage, and sufficient familiarity with materials to
accurately identify often unlabeled materials such as plastic polymers.
Through manufacturing and repair experience, product manufacturers may
acquire material selection, assembly, and quality inspection skills that can be leveraged to
perform disassembly tasks. Since firms are more likely to vertically integrate into
activities that require knowledge similar to that which they already possess, companies
173
with extensive manufacturing, service, and repair experience may be more likely to
vertically integrate into EOL product recovery. This may explain why IBM and HewlettPackard—companies with world-class capabilities in these areas—have become deeply
involved in product recovery, while companies such as Gateway that possess limited
manufacturing capabilities have not.
4.4.2 Acquiring Tacit Disassembly Know-How
Disassembly requires tacit knowledge that is seldom communicated by codified
OEM specifications. Indeed, even among independent recyclers who possessed OEM
specifications, two-thirds still have to reverse engineer the product to understand how to
disassemble it efficiently (Nasr et al., 1998). Reverse engineering is expensive and time
consuming, averaging $37,000 and 23 days per product (Nasr et al., 1998). Since firms
that design, engineer, and manufacture products acquire tacit knowledge of how their
products are assembled, they—unlike independent firms conducting product recovery—
have the opportunity to leverage this knowledge into tacit disassembly knowledge. For
example, “Ford maintains its Experimental Dismantling Center in Germany [which]
dismantles vehicles to benchmark design practices and materials use against materials
recovery capability” (US EPA, 1999b). This suggests that firms that design, engineer, and
manufacture products are apt to possess a cost advantage over other firms in
disassembling EOL products, and this gap widens when economies of scale enable more
EOL product models to be processed in the same facility. In addition, after studying the
product recovery operations of Océ, a Dutch photocopier firm, Krikke, van Harten, and
Schuur (1999) suggest that dismantling, preparation, and reassembly processes situated in
174
the same location can facilitate the dissemination of tacit knowledge that can boost
productivity. In addition, locating remanufacturing within manufacturing facilities may
facilitate the transfer of tacit knowledge between manufacturing and remanufacturing
operations (Ferrer & Whybark, 2001).
4.4.3 Feeding Back Recovery Know-How to Designers
Engaging in product recovery often generates knowledge about EOL products
such as the relative durability of their components and the ease of unfastening assemblies.
In the hands of product designers, this knowledge can lead to design modifications that
facilitate EOL product disassembly, assessment, and recycling and reduce the amount of
non-recyclable residual. Products designed to facilitate disassembly have more
predictable material recovery rates, faster disassembly times, and generate less waste
(Guide, 2000). For example, IBM has used its Asset Recovery Center to evaluate the
effectiveness of its design initiatives meant to facilitate its EOL products’ disassembly
and plastic resin identification (Dunnett et al., 1999). While even disassembly facilities
owned by equipment manufacturers face difficulties “getting this information to the
designers . . . [it] is even more difficult to get designers to listen to feedback and advice
from independent remanufacturing companies, despite the valuable experience they
have” (Ferrer & Whybark, 2001).
This factor alone suggests manufacturers possess advantages over other firms in
reducing the overall cost of recovering their EOL products. BMW operates a Recycling
and Disassembly Center in Germany to conduct “detailed disassembly analyses [to] take
a close look at the amount of time and the tools required for disassembling end-of-life
175
vehicles. This information is then used as a basis for determining whether the vehicle
construction and materials are suitable for recycling. Recommendations are made
regarding recycling-optimized design and the eco-efficient, i.e., ecological and
economical recovery of end-of-life vehicles . . . [The Center] consistently exchanges
information with the BMW research and innovation center” (BMW, 2001). In Japan,
electronics manufacturers have formed two consortia to collect, recover, and recycle their
EOL products pursuant to legislative requirements. However, “each manufacturer holds
at least one treatment plant so that it can compile and communicate information from the
downstream to the upstream, accumulate knowledge and recycling technology, and grasp
the actual cost for recovery and environmentally sound treatment. Exchange of
information between recycling plants and product design department has been taking
place by way of periodical meetings among the personnel involved, seminars, via intranet
and designers’ visits to recovery plants” (Tojo, 2001).
Cross-functional management and continuous improvement capabilities can
enable organizations to leverage EOL product knowledge into design improvements. This
suggests that manufacturers with these capabilities may be better than others at
facilitating this internal knowledge transfer. Total quality management (TQM) programs
typically require “collecting relevant information from all phases of an organization’s
operations” and “quality assurance and improvement efforts [that] include manufacturing
[and] supporting functions which impact operations” (Curkovic, Melnyk, Handfield, &
Calantone, 2000). Firms with TQM programs often possess capabilities that enable them
to implement improvement activities based on recommendations from a wide variety of
176
sources and to manage cross-functional activities (Hart, 1995), both of which can
increase the likelihood that knowledge accumulated while conducting product recovery
activities is transferred quickly and accurately to product designers. In addition, firms
with more comprehensive environmental management programs often possess
capabilities in cross-functional management, stakeholder integration and higher-order
learning processes (Russo & Fouts, 1997; Sharma & Vredenburg, 1998). Many
companies with deeply engrained quality cultures, such as Shell Chemicals and DuPont,
are expanding their TQM programs and quality management systems to include
environmental issues (Ahmed, 2001; Ehrenfeld, 1994; Karapetrovic & Willborn, 1998;
Klassen & McLaughlin, 1993; Toffel, 2000; Wilkinson & Dale;, 1999). As such,
manufacturers with well embedded quality and environmental programs may be more
likely to possess capabilities that enable them to leverage the knowledge accumulated
during product recovery into design improvements. This can lead to lower costs of
managing products throughout all of their life cycle stages.
4.4.4 Environmental Reputation Capabilities
Proactive environmental management can also foster a strong reputation among
customers for environmental leadership. This can provide competitive advantage when
selling to customers who value their suppliers’ environmental performance. The same
drivers that encourage a company to pursue an environmental leadership stance may also
encourage it to actively engage in product recovery. For example, the corporate
environmental sustainability strategy of Hewlett-Packard, a company with a reputation
for environmental leadership, includes “Developing product end-of-life solutions, such as
177
recycling technologies and infrastructures, across high tech industries to create reliable
streams of recycled materials.”111 Interface, the largest commercial carpet manufacturer
in the world, has adopted the ambitious goal of becoming “the world’s first
environmentally restorative company” and views its voluntary product take-back
program as instrumental to its success.112
Launching a voluntary product take-back program can also enhance a firm’s
environmental reputation. Kodak, as mentioned earlier, initiated its take-back program to
overcome the wasteful image associated with its single-use cameras and has largely
removed their environmental stigma.
4.5 Unique Assets and Avoiding Supplier and Buyer
Dependence
Resource Dependence theory provides a third lens to examine product
manufacturers’ decisions of how to manage their EOL products. According to this theory,
firms manage transactions and define their organizational boundaries to avoid depending
upon other organizations for critical resources (Pfeffer & Salancik, 1978). This is a
111
Hewlett-Packard,
“Our
Planet,
Our
Promise,”
Palo
Alto,
CA,
2001,
<www.hp.com/hpinfo/community/environment/pdf/commit.pdf>
112
Interface, Inc., “Closing the Loop,” <www.interfacesustainability.com/closing.html>,
accessed November 19, 2003.
178
particularly important perspective for OEMs considering whether to vertically integrate
into product recovery. As discussed earlier, EOL products may also become critical
resources to product manufacturers’ business models if other firms can refurbish or
remanufacture their products and threaten their customer base. Not only could this erode
the OEM’s market share, but its brand reputation could be sullied if its products, when
refurbished or remanufactured by another firm, do not deliver the performance expected
of its brand—since its brand label typically remains affixed to the product.
The extent to which a buyer is dependent upon a supplier depends upon three
factors: the importance of the resource to the buyer, the extent to which the buyer can
access alternative sources, and the degree to which the supplier has discretionary control
over the resource (Pfeffer & Salancik, 1978). Therefore, a buyer is most dependent upon
a supplier when an individual supplier has complete discretion over resources the buyer
views as crucial to its cost or differentiation advantage and for which the buyer has no
ready substitute (Cool & Henderson, 1998; Medcof, 2001). On the other hand, a supplier
is most dependent upon a buyer when “a buyer represents a large share of a seller’s
revenues and if this buyer cannot be easily replaced” (Cool & Henderson, 1998). From
this perspective, organizational strategies are designed to enable organizations “to
minimize their dependence and increase the dependence of others on them” (Dunford,
1987) and to increase the predictability and stability of relationships with organizations
upon which they depend (Pfeffer & Salancik, 1978).
One strategic option for dependent organizations is to expand their boundaries
through vertical or horizontal integration to absorb the resources upon which they depend
179
(Dunford, 1987). Alternatively, organizations can establish cooperative relations using
“cooptive ties” with organizations that control their critical resources by developing
cartels, alliances, joint ventures, or common members of their boards of directors (Pfeffer
& Salancik, 1978). To mitigate their dependence on suppliers, buyers can develop
buffering strategies such as investing in reserve inventories that provide some protection
against temporary supplier instabilities (Thompson, 1967). Suppliers can mitigate
resource dependence by diversifying their customer base.
Suppose a manufacturer that employs a unique polymer for its product casing can
produce additional units at less cost by harvesting and recycling this material from its
EOL products than by purchasing virgin polymer. Further suppose the manufacturer
values the recovered components more than the next highest bidder. Should other
companies seek to recover and disassemble these products, they risk becoming dependent
upon the manufacturer. If only a few companies recovered this product such that supply
is concentrated, the manufacturer risks becoming dependent upon these suppliers. The
importance of this dependency rises as the gap increases between the cost of using
recovered versus virgin resources (Klein & Leffler, 1981).
The circumstances described above predict that resource dependence will emerge
when the product manufacturer seeks to recover manufacturer-specific components or
materials from its EOL products, especially when precise deliveries are required as in JIT
systems that minimize buffer inventories. In such cases, Resource Dependence theory
predicts that these manufacturers will seek to increase the relationship’s predictability
and stability by employing cooptive ties or by integrating into product recovery. The key
180
concern in relying on other organizations is that the OEM may become dependent upon a
provider of what may become a key resource. This is a parallel concern of TCE
cautioning against the use of markets for transaction-specific assets due to hold-up
concerns. Indeed, Resource Dependence advocates boundary-spanning approaches,
which may include the use of hybrids or vertical integration as encouraged by TCE.
In fact, many manufacturers that reuse manufacturer-specific components from
their EOL products have avoided supplier dependence by developing schemes that
encourage EOL products to be returned directly to them. For example, Hewlett-Packard
offers postage paid labels to encourage customers to return their toner cartridges. Kodak
pays film processors to return their “single use” cameras directly to Kodak. Xerox has
long relied on a leasing strategy that ensures that its used photocopiers are returned to the
company. Interface has recently begun leasing flooring tiles to commercial customers,
with the company reclaiming worn tiles to recycle them into new flooring.
On the other hand, dependence concerns are largely absent in industries where
manufacturers do not gain particular advantages by recovering their own EOL products.
For example, EOL computers are partially disassembled to harvest components with
resale values on secondary markets (e.g., disk drives, memory chips) and are then
crushed and recycled to reclaim metals and plastics. In general, few of these components
or materials are manufacturer-specific, and thus resource dependence is generally not a
concern between computer manufacturers and third-party companies that recover EOL
computers. As such, it is not surprising that third-party companies conduct a great deal of
EOL computer recovery.
181
4.6 Managerial Implications: Crafting a Strategy
This article has described several factors that motivate manufacturers to engage in
voluntary product recovery: reducing production costs, enhancing brand image, meeting
customer demands, protecting aftermarkets, and preempting regulations. Managers
contemplating product recovery strategies should consider which of these drivers
currently apply to their company and industry. This will likely require discussions with
managers from a variety of functions, since knowledge about production costs, brand
reputations, customer expectations, and legislative agendas is typically diffused across an
organization. These discussions should also explore which additional drivers are on the
horizon. Understanding existing drivers and anticipating additional ones informs the
strategic decision of how to proceed. For example, if the firm is seeking to address
legislative pressures, this industry-wide concern might be best addressed through a
response coordinated by an industry association, as the rechargeable battery example
illustrates. However, if customer expectations are the main driver for exploring product
recovery strategies, then attempting to work with competitors might undermine an
opportunity to gain competitive advantage by providing product recovery services that
best meets customer needs.
The three management theories that provide insight on the potential benefits and
drawbacks of various product recovery strategies are summarized in Table 4.1. The
transaction cost analysis suggests that when efficient EOL recovery requires investments
specific to a particular product or its material, independent product recovery firms and
OEMs face hold-up risks. Consequently, recovery of such products is likely to involve
182
more active engagement of its manufacturer, such as through joint ventures or vertical
integration. Similarly, manufacturers must consider whether minimizing product recovery
costs entails sharing proprietary information, as this could deter relying on independent
third parties or developing a consortium with competitors. In addition, the high
environmental uncertainty surrounding some types of product recovery complicates
contractual relations among a manufacturer seeking to buy its EOL components from an
independent product recovery firm. Though some contractual solutions such as
price/quantity schedules have emerged to deal with this issue, manufacturers using JIT
delivery face greater hold-up risk are more likely to use hybrids or vertical integration
than rely on independent product recovery firms.
The capabilities analysis suggests that by leveraging tacit knowledge and
proprietary information acquired during product design, engineering, and production,
product manufacturers are likely to possess competitive advantages compared to
independent product recovery firms in several stages of EOL product recovery.
Furthermore, manufacturers with quality cultures may possess key capabilities that
enable them to more rapidly and accurately provide feedback to designers and engineers
about the knowledge accumulated during EOL product recovery, which can reduce future
product recovery costs.
Finally, the resource dependence analysis suggests that manufacturers and
independent product recovery firms seek to avoid being dependent on each other for
unique resources. One implication is that when a manufacturer seeks to use its particular
EOL products or materials as a source of parts for spares and remanufacturing, the
183
manufacturer will attempt to avoid becoming dependent on an independent product
recovery firm by developing more direct recovery channels. In addition, some EOL
products risk being disposed of inappropriately, such as being exported to developing
nations that feature unsafe working conditions and limited pollution controls. Recent
reports by activists and the media have demonstrated that OEMs are being held publicly
accountable for goods with their brand name (Puckett et al., 2002; San Jose Mercury
News, 2002; Schoenberger, 2002a, b), irrespective of the fact that disposal decisions are
made by customers (and not the OEM). In these cases, OEMs should seek to develop
durable cooperative relations with third-party firms or competitors or else vertically
integrate to exert greater control over the recovery of their EOL products
4.6.1 Limitations and Further Research
That some OEMs are more heavily engaged in product recovery than this analysis
suggests may be due to the novelty of this activity. The U.S. Environmental Protection
Agency recently reported that the “infrastructure for materials recovery is not sufficient
in the U.S. to deliver quantity and quality of recovered inputs demanded by progressive
implementation of recycled content requirements” (US EPA 1999b). This may explain
why some OEMs are deeply engaged in product recovery in circumstances other than
those described above.
While the focus here has been on the collection and primary recycling stages of
product recovery, Fleischmann et al. argue that there are actually several classes of
product recovery networks and that each should be evaluated separately because “reusable item networks, remanufacturing networks, and recycling networks appear each to
184
have their own typical characteristics” (Fleischmann, Krikke, Dekker, & Flapper, 2000).
As such, each product recovery stage may warrant individualized analysis.
While operations management research on product recovery is clearly a growing
field, research on this subject from other perspectives is less common.113For example,
several antitrust and economic questions arise from the prebate concept that prevents
other companies besides the OEM from refurbishing its products. Is this a legitimate
business practice or is it designed to monopolize aftermarkets? Are recycling rates likely
to be higher or lower if prebates are permitted or outlawed, either by court decisions or
legislation? Could manufacturers use prebates to preclude other companies from
113
Product recovery research is most advanced in the operations management
perspective, as indicated by several conferences, special journal issues, and books
focusing on reverse logistics and closed-loop supply chains. For example, conferences on
closed-loop supply chains were held at Carnegie Mellon University in 2001 and INSEAD
in 2002. Six European universities participate in RevLog, a working group focusing on
reverse logistics. Production and Operations Management published a special issue [10/2,
2001] on environmental management and operations that focused on manufacturing and
eco-logistics. An edited book has recently been published on this issue as well (Guide &
van Wassenhove, 2003).
185
refurbishing their EOL durable goods, thus artificially curtailing equipment durability to
force consumers to purchase newer models?114
In addition, product take-back regulations have evoked concerns about
international treaty obligations. Little research has investigated the implications of takeback regulations on international treaties, including whether they violate obligations
imposed by the General Agreement on Tariffs and Trade (1994) and the World Trade
Organization’s Agreement on Technical Barriers to Trade (1994) — as alleged by the
American Electronics Association (Scigliano, 1999) — and whether such regulations are
consistent with the Basel Convention on the Control of Transboundary Movements of
Hazardous Wastes and their Disposal (1992) and the pending Basel Ban (1995).
Finally, only tentative steps have been taken to identify which EOL products
make better candidates for remanufacturing or recycling. For example, Klausner and
Hendrickson suggest that remanufacturing is particularly well suited for EOL products
that include components characterized by long technology cycles and low technological
obsolescence, and when ex ante uncertainty regarding usage intensity results in “overengineering for certain user groups in order to meet the needs of other user groups”
(Klausner & Hendrickson, 2000). Rose (2000) has developed a model that forecasts
whether EOL electronic products should be remanufactured, disassembled and then
recycled, recycled whole (without prior disassembly), or disposed of. According to her
114
For more on this question, see Borenstein, MacKie-Mason, and Netz (2000).
186
model, this decision is based on technical product characteristics including product
durability, rate of technological obsolescence, extent of product complexity, the duration
of a design cycle, and the reason for redesigns.
187
TABLE 4.1
Drivers of Product Recovery Strategy
Theoretical
Perspective
OEMs should consider vertical integration or hybrids (e.g.,
joint ventures, alliances) instead of relying on independent
companies when...
Transaction
Cost
Economics Analysis
Product recovery requires investments that are specific to an
OEM’s products or materials, especially when EOL product
recovery rates are highly uncertain.
Capabilities Analysis
OEMs can leverage tacit knowledge and proprietary
information acquired during product design, engineering, and
production to their product recovery activities (since tacit
knowledge cannot be easily transferred between independent
organizations).
OEMs possess a culture of cross-functional management and
continuous improvement that facilitates rapid transfer of tacit
knowledge from their product recovery operations to their
designers and engineers.
Resource
Dependence
Analysis
OEMs risk becoming dependent upon independent product
recovery firms for rare or unique components or materials
recovered from EOL products.
OEMs can develop durable cooperative relations with thirdparty firms or competitors to exert some control over the
recovery of their EOL products.
View
188
Chapter 5.
5. Conclusions
This dissertation has examined two forms of corporate self-regulation, the most
recent form of environmental policy that has companies assuming a much larger role in
regulating their own behavior. I find mixed evidence that partially supports industry’s
enthusiasm for this approach, and partially supports skeptics concerned that selfregulation constitutes “the foxes guarding the hen house.”
My evaluation of the ISO 14001 Environmental Management Systems Standard
(Chapter 2) provides some evidence in support of self-regulation. ISO 14001 is an
entirely private-sector initiative that, unusual for voluntary environmental programs,
requires periodic verification by independent, third-party monitors. First, I found that
this standard has attracted facilities that were improving their environmental performance
and legal compliance faster than non-adopters. In addition, I found that companies that
adopted this standard reduced their toxic emissions in the two years before adoption,
while they upgraded their management practices to meet the standard, and their superior
performance continued in the first few years after certification. These findings represent a
substantial departure from prior evaluations of voluntary environmental management
programs that do not require any verification, which found little evidence that they have
attracted better-than-average participants or that they lead to improved environmental
outcomes.
189
Further evidence to support self-regulation is provided by Jodi L. Short and my
evaluation of United States Environmental Protection Agency’s Audit Policy (Chapter 3).
We found that larger facilities and those with worse compliance records were more likely
to self-disclose compliance violations. From a public policy perspective, violations by
larger facilities can often have more serious consequences to the environment and to
public health, and so having these facilities disclose violations—along with their promise
to promptly correct them—can prevent serious harm to the environment and public
health. Regulators should also be pleased to learn that facilities with worse compliance
records are more likely to self-disclose violations, as this apparently signals their
newfound dedication to compliance.
While the above results suggest that self-regulation can complement government
regulation, my results cast doubt on the claim that self-regulation can substitute for
government regulation.
In Chapter 2, while I found that ISO 14001 adopters
subsequently reduced the quantity of toxic chemical emissions, I also found that their
emissions were increasing in hazardousness. The net result of these two trends is that
adopters’ subsequent emissions imposed health risks on their communities that were
indistinguishable from those imposed by non-adopters. I also found no evidence that ISO
14001 adopters subsequently improved their regulatory compliance, either in terms of
enforcement actions or violations of the federal hazardous waste regulations pursuant to
the Resource Conservation and Recovery Act (RCRA).
The US EPA Audit Policy evaluation in Chapter 3 suggests further skepticism is
warranted about the claim that self-regulation can provide a viable substitute for
190
government regulation. In that study, Jodi L. Short and I found that facilities were more
likely to self-disclose environmental violations if they were recently inspected or were
narrowly targeted for heightened scrutiny by regulators. Thus private-sector engagement
in “self-policing” seems directly related to pressure exerted by regulators. In the United
States, the recent nationalization of airport passenger and baggage inspectors followed
several federal privatization initiatives of inspections such as those assuring the safety of
meat and poultry within processing plants. These apparently conflicting trends provide
several opportunities to evaluate the circumstances under which self-policing can serve as
a useful complement to or a reliable substitute for traditional government policing.
Finally, in my study of end-of-life product recovery strategies, I identified several factors
that are leading manufacturers to voluntarily assume more responsibility for recovering
their end-of-life products, including reducing production costs, enhancing brand image,
meeting changing customer expectations, protecting their aftermarkets, and preempting
legislation. In addition, I identify key criteria— product recovery technologies,
uncertainty in reverse logistics, manufacturing-related capabilities, and the uniqueness of
recovered assets —that are influencing manufacturers’ decisions to vertically integrate
into end-of-life product recovery activities, form consortia, or outsource these tasks. The
next few years will see take-back regulations enter force across the European Union
while companies in the United States will very likely continue to recover their end-of-life
products on a purely voluntary basis. This natural experiment may enable researchers to
better evaluate the environmental outcomes and financial costs that accompany each
approach.
191
References
Abrams, F. 1999. Behind the designer labels, a world of exploitation and abuse. The
Independent (London), (Sept. 24): 1.
Ahmed, N. U. 2001. Incorporating Environmental Concerns into TQM. Production and
Inventory Management Journal, 42(1): 25-30.
Alchian, A. A. & Demsetz, H. 1972. Production, information costs, and economic
organization. The American Economic Review, 62(5): 777-795.
American
Chemistry
Council.
2000.
Obligation
of
Members.
www.americanchemistry.com/rc.nsf/unid/lsgs-4eesnm?opendocument, accessed
July 11, 2002.
Anderson, E. & Schmittlein, D. C. 1984. Integration of the sales force: An empirical
examination. RAND Journal of Economics, 15(3): 385-395.
Anderson, E. & Weitz, B. A. 1986. Make-or-buy decisions: Vertical integration and
marketing productivity. Sloan Management Review, Spring: 3-19.
Anderson, Eric E. & Wayne K.Talley . 1995. The oil spill size of tanker and barge
accidents: Determinants and policy implications, 71 Land Economics 216-228.
Andrews, R. N. L., Hutson, A., & Edwards, D., Jr. 2003. Environmental Management
Under Pressure: How Do Mandates Affect Performance? Paper presented at the
Leveraging the Private Sector: Management-Based Strategies for Improving
Environmental Performance, Harvard University, John F. Kennedy School of
Government, Cambridge, MA.
Aoki, Kazumasu & John W. Coiffi . 2000. Poles apart: Industrial waste management
regulation and enforcement in the United States and Japan, in Robert A. Kagan &
Lee Axelrod, eds., Regulatory Encounters. Berkeley, CA: University of California
Press.
Argyres, N. 1996. Evidence on the Role of Firm Capabilities in Vertical Integration
Decisions. Strategic Management Journal, 17(2): 129-150.
Athey, S. & Imbens, G. W. 2002. Identification and Inference in Nonlinear Difference-inDifferences Models. Cambridge, MA: National Bureau of Economic Research,
Technical Working Paper 280.
Ayres, Ian, and John Braithwaite . 1992. Responsive Regulation: Transcending the
Deregulation Debate. New York: Oxford University Press.
192
Ayres, R., Ferrer, G., & Van Leynseele, T. 1997. Eco-efficiency, Asset Recovery and
Remanufacturing. European Management Journal, 15(5): 557-574.
Babakri, K. A., Bennett, R. A., Rao, S., & Franchetti, M. 2004. Recycling performance of
firms before and after adoption of the ISO 14001 standard. Journal of Cleaner
Production, 12(6): 633-637.
Bansal, P. & Hunter, T. 2003. Strategic explanations for the early adoption of ISO 14001.
Journal of Business Ethics, 46(3): 289-299.
Barbera, B. M. & Lyona, J. D. 1996. Detecting abnormal operating performance: The
empirical power and specification of test statistics. Journal of Financial
Economics, 41(3): 359-399.
Bardach, E., & Kagan, R. A. 1982. Going by the Book: The Problem of Regulatory
Unreasonableness. Philadelphia: Temple University Press.
Barnet, R. J. & Cavanagh, J. 1994. Just undo it: Nike's exploited workers. The New York
Times, (Feb. 13): C11.
Barney, J. 1991. Firm Resources and Sustained Competitive Advantage. Journal of
Management, 17(1): 99-120.
Baron, D. P. 2003. Private politics. Journal of Economics & Management Strategy,
12(1): 31-66.
Baum, J. A. C., Calabrese, T., & Silverman, B. S. 2000. Don't go it alone: Alliance
network composition and startups' performance in Canadian biotechnology.
Strategic Management Journal, 21(3): 267-294.
Baumgartner, H. 1996. Reproducing Copying Machines. Mechanical Engineering,
118(5): 53.
Becker, E. 2003. Rights group for animals drops lawsuit against KFC. New York Times,
(Sept. 2): A19.
Becker, S. O. & Ichino, A. 2002. Estimation of average treatment effects based on
propensity scores. The Stata Journal, 2(4): 358-377.
Bedford, Christopher. 1996. Dirty Secrets: The Corporations' Campaign for an
Environmental Audit Privilege. Takoma Park, MD: Environmental Action
Foundation. http://www.mapcruzin.com/scruztri/docs/r2.htm
Berthelot, S., McGraw, E., Coulmont, M., & Morrill, J. 2003. ISO 14000: Added value
for Canadian business? Environmental Quality Management, 13(2): 47-57.
193
Blackman, A., Afsah, S., & Ratunanda, D. 2004. How Do Public Disclosure Pollution
Control Programs Work? Evidence from Indonesia. Human Ecology Review,
11(3).
Blumberg, G. P. 2002. Junkyards Discard an Image, and the Scary Dogs, Too. New York
Times, (Oct. 23): G9.
BMW. 2001. Sustainability Value Report: 2001/2002. Munich: Bayerische Motoren
Werke Aktiengesellschaft (BMW Ag).
Boerner, C. S. & Macher, J. T. 2001. Transaction Cost Economics: An Assessment of
Empirical Research in the Social Sciences. Unpublished manuscript.
http://www.msb.georgetown.edu/faculty/jtm4/Papers/eTCE.pdf.
Borenstein, S., MacKie-Mason, J., & Netz, J. S. 2000. Exercising Market Power in
Proprietary Aftermarkets. Journal of Economics and Management Strategy, 9(2):
157 - 188.
Braithwaite, J., & T. Makkai . 1991. Testing an Expected Utility Model of Corporate
Deterrence, Law & Society Review 25: 7–40.
Burglund, M. 1999. Does an environmental management system really improve the
environment? Pollution Control News 19(7): 5.
CAFOD. 2004. Clean Up Your Computer: Working conditions in the electronics sector.
London: Catholic Agency for Overseas Development.
Californians Against Waste. 2003. E-Waste Legislation in Other
http://svtc.igc.org/cleancc/usinit/index.html, accessed May 13, 2003.
States.
Carraro, C., Katsoulacos, Y., & Xepapadeas, A. (Eds.). 1996. Environmental Policy and
Market Structure. Boston: Kluwer Academic Publishers.
Carter, C. R. & Ellram, L. M. 1998. Reverse logistics: A review of the literature and
framework for future investigation. Journal of Business Logistics, 19(1): 85-102.
Charlton, A. & Vanhorn, D. 1990. Throwaway goods, Uncle Sam join plastics on
environmental `hit list'. The Star-Ledger, (Apr 22).
Chemical Week. 2002. Details of a Revamped Responsible Care Take Shape. (Nov. 20):
33.
Chin, K. S. & Pun, K.-F. 1999. Factors influencing ISO 14000 implementation in the
printed circuit board manufacturing industry in Hong Kong. Journal of
Environmental Planning & Management, 42(1): 123-134.
194
Chouinard, Y. & Brown, M. 1997. Going organic: Converting Patagonia's cotton product
line. Journal of Industrial Ecology, 1(1): 117-130.
Christmann, P. & Taylor, G. 2001. Globalization and the environment: determinants of
firm self-regulation in China. Journal of International Business Studies, 32(3):
439-458.
Coglianese, C. & Nash, J. 2001. Bolstering private-sector environmental management.
Issues in Science and Technology, 17(3): 69-74.
Cohen, M. A. 1987. Optimal Enforcement Strategy to Prevent Oil Spills: An Application
of Principal-Agent Model with Moral Hazard, Journal of Law and Economics, 30:
23-51.
Cohen, M. A. 2000. Empirical Research on the Deterrence Effect of Environmental
Monitoring and Enforcement, Environmental Law Reporter, 30: 10245-10252.
Collier, R. & Strasburg, J. 2002. Clothiers fold on sweatshop lawsuit. San Francisco
Chronicle, (Sept. 27).
Conner, K. R. & Prahalad, C. K. 1996. A Resource-Based Theory of the Firm:
Knowledge versus Opportunism. Organization Science, 7(5): 477-501.
Conner, K. R. 1991. A Historical Comparison of Resource-Based Theory and Five
Schools of Thought Within Industrial Organization Economics: Do We Have a
New Theory of the Firm? Journal of Management, 17(1): 121-154.
Cool, K. & Henderson, J. 1998. Power and Firm Profitability in Supply Chains: French
Manufacturing Industry in 1993. Strategic Management Journal, 19: 909-926.
Corbett, C. J. & Luca, A. 2002. Global Survey on ISO 9000 and ISO 14000: Summary of
Findings. Unpublished manuscript, The Anderson School, University of
California at Los Angeles.
Corbett, C. J. & Russo, M. V. 2001. ISO 14001: irrelevant or invaluable? ISO
Management Systems, (Dec.): 23-29.
Corbett, C. J. & Savaskan, R. C. 2003. Contracting and Coordination in Closed-Loop
Supply Chains. In J. V. Daniel R. Guide, & L. N. van Wassenhove (Eds.),
Business Aspects of Closed-Loop Supply Chains: Exploring the Issues: 93-113.
Pittsburgh: Carnegie Mellon University Press.
Corbett, C. J., Montes, M. J., Kirsch, D. A., & Alvarez-Gil, M. J. 2002. Does ISO 9000
certification pay? ISO Management Systems, July-Aug: 31-40.
195
Courville, S. 2003. Social accountability audits: Challenging or defending democratic
governance? Law & Policy, 25(3): 269-298.
Cox, J. 2004. The Case Against a Judicially Created, Common-Law Self-Audit or SelfEvaluation Privilege Applicable to Environmental Cases, Fordham
Environmental Law Journal, 51: 1-30.
Cross, F. B. 2003. Decision Making in the US Circuit Courts of Appeals, California Law
Review, 91(6): 1457-1515.
Curkovic, S., Melnyk, S. A., Handfield, R. B., & Calantone, R. 2000. Investigating the
Linkage Between Total Quality Management and Environmentally Responsible
Manufacturing. IEEE Transactions on Engineering Management, 47(4): 444 464.
Cutler, D. M., Huckman, R. S., & Landrum, M. B. 2004. The Role of Information in
Medical Markets: An Analysis of Publicly Reported Outcomes in Cardiac
Surgery. NBER Working Paper No. w10489.
Dahlström, K., Howes, C., Leinster, P., & Skea, J. 2003. Environmental management
systems and company performance: Assessing the case for extending risk-based
regulation. European Environment, 13: 187–203.
Darnall, N. & Edwards, D. J. 2004. Predicting the cost of environmental management
system adoption. Paper presented at the Academy of Management, New Orleans.
Darnall, N., Gallagher, D. R., Andrews, R. N. L., & Amaral, D. 2000. Environmental
management systems: Opportunities for improved environmental and business
strategy? Environmental Quality Management, 9(3): 1-10.
Dasgupta, S., Hettige, H., & Wheeler, D. 2000. What improves environmental
compliance? Evidence from Mexican industry. Journal of Environmental
Economics and Management, 39: 39-66.
Davis, G. A., Wilt, C. A., Dillon, P. S., & Fishbein, B. K. 1997. Extended Product
Responsibility: A New Principle For Product-Oriented Pollution Prevention.
Knoxville, TN: Center for Clean Products and Clean Technologies.
de Ron, A. & Penev, K. D. 1995. Disassembly and Recycling of Electronic Consumer
Products: An Overview. Technovation, 15(6): 363-374.
Dehejia, R. H. & Wahba, S. 1999. Causal effects in nonexperimental studies:
Reevaluating the evaluation of training programs. Journal of the American
Statistical Association, 94(448): 1053-1062.
196
Dehejia, R. H. & Wahba, S. 2002. Propensity score-matching methods for nonexperimental causal studies. The Review of Economics and Statistics, 84(1): 151161.
Dehejia, R. H. 2005. Practical propensity score matching: a reply to Smith and Todd.
Journal of Econometrics, 125(1-2): 355-364.
Delmas, M. & Toffel, M. W. 2004. Stakeholders and environmental management
practices: An institutional framework. Business Strategy and the Environment,
13(4): 209-222.
Delmas, M. 2004. The take-off period in the diffusion of international standards: The
case of ISO 14001. Unpublished manuscript, Donald Bren School of
Environmental Science & Management.
Delmas, M. A. 2000. Barriers and incentives to the adoption of ISO 14001 by firms in the
United States. Duke Environmental Law & Policy Forum, 11(1): 1-38.
Delmas, M. A. 2002. The diffusion of environmental management standards in Europe
and the United States: An institutional perspective. Policy Sciences, 35: 91-119.
Delmas, Magali & Michael W. Toffel. 2005. Institutional Pressure and Environmental
Management, University of California Center for Responsible Business Working
Paper, Berkeley, CA.
Dierickx, I. & Cool, K. 1989. Asset Stock Accumulation and Sustainability of
Competitive Advantage. Management Science, 35: 1504-1511.
DiMaggio, Paul J. & Walter W. Powell . 1991. Introduction, in Walter W. Powell & Paul
J. DiMaggio, eds., The New Institutionalism in Organizational Analysis. Chicago:
University of Chicago Press.
Dimento, Joseph F. 1989. Can Social Science Explain Organizational Noncompliance
with Environmental Law? Journal of Social Issues, 45(1): 109-132.
Dobbin, Frank, & Sutton, John R. 1998. The Strength of a Weak State: The Rights
Revolution and the Rise of the Human Resources Management Divisions,
American Journal of Sociology, 104: 441-476.
Dosi, G. & Teece, D. 1998. Organizational competencies and the boundaries of the firm.
In R. Arena, & C. Longhi (Eds.), Markets and Organization: 281-301. New York:
Springer-Verlag.
197
Duggin, Sarah Helene . 2003. Internal Corporate Investigations: Legal Ethics,
Professionalism and the Employee Interview, 2003 Columbia Business Law
Review: 859-964.
Dunford, R. 1987. The Suppression of Technology As a Strategy for Controlling
Resource Dependence. Administrative Science Quarterly, 32(4): 512-525.
Dunnett, M., Grenchus, E., Keene, R., Yehle, L., Jacques, M., Karlsson, M., Kirby, J. R.,
& Pitts, D. 1999. Evaluation of IBM end of life products: measuring DFE
effectiveness. Paper presented at the Proceedings of the 1999 IEEE International
Symposium on Electronics and the Environment, Danvers, MA.
Dusharme, D. 2004. ISO 9001 Survey. Quality Digest, July: 24-30.
Dyer, J. H. & Nobeoka, K. 2000. Creating and managing a high-performance knowledgesharing network: The Toyota case. Strategic Management Journal, 21(3): 345367.
Earnhart, D. 2004. Regulatory factors shaping environmental performance at publiclyowned treatment plants. Journal of Environmental Economics and Management,
48(1): 655-681.
Easton, G. S. & Jarrell, S. L. 1998. The effects of Total Quality Management on
corporate performance: An empirical investigation. Journal of Business Ethics,
71(2): 253-307.
ECOLOGIA. 2000. ISO 14000: A Factsheet for NGOs. ECOLOGIA (ECOlogists Linked
for
Organizing
Grassroots
Initiatives
and
Action).
www.ecologia.org/ems/iso14000/resources/factsheets/index.html, accessed Apr.
14, 2003.
Edelman, L. B. & Suchman, M. C. 1997. The Legal Environments of Organizations,
Annual Review of Sociology, 23: 479-515.
Edelman, L. B. 1990. Legal Environments and Organizational Governance: The
Expansion of Due Process in the Workplace, American Journal of Sociology, 95:
1401-1440.
Edelman, L. B. 1992. Legal Ambiguity and Symbolic Structures: Organizational
Mediation of Civil Rights Law, American Journal of Sociology, 97: 1531-1576.
Edelman, L. B., Uggen, C. & Erlanger, H. S.1999. The Endogeneity of Legal Regulation:
Grievance Procedures as Rational Myth, 105(2) American Journal of Sociology
406-54.
198
Eden, Sally . 1996. Environmental Issues and Business: Implications of a Changing
Agenda. Chichester, England: John Wiley & Sons.
Ehrenfeld, J. R. 1994. Industrial Ecology: A Strategic Framework for Product Policy and
Other Sustainable Practices. Paper presented at the Green Goods: The Second
International Conference and Workshop on Product Oriented Policy, Stockholm.
Eichler, M. & Lechner, M. 2002. An evaluation of public employment programmes in the
East German state of Sachsen-Anhalt. Labour Economics, 9: 143–186.
Elliott, E., Sheldon, B. J., & Regens, J. L. 1997. Political and economic determinants of
individuals' support for environmental spending. Journal of Environmental
Management, 51: 15-27.
Environmental Integrity Project & Galveston-Houston Association for Smog Prevention.
2004. Who's Counting? The Systematic Underreporting of Toxic Air Emissions.
Washington, DC: Environmental Integrity Project.
Epple, Dennis, & Visscher, Michael. 1984. Environmental Pollution: Modeling
Occurrence, Detection and Deterrence, Journal of Law and Economics, 27: 29-60.
Federal Register. 1995. Incentives for Self-Policing: Discovery, Disclosure, and
Prevention of Violations, Federal Register, 60: 66706-66712.
Federal Register . 1999. Evaluation of ‘Incentives for Self-Policing: Discovery,
Disclosure, and Prevention of Violations’ Policy Statement, Proposed Revisions
for Public Comment, 64 Federal Register 26745-26756.
Ferrer, G. & Whybark, D. C. 2001. Material Planning for a Remanufacturing Facility.
Production & Operations Management, 10(2): 112-124.
Ferrer, G. & Whybark, D. C. 2003. The Economics of Recycling. In J. V. Daniel R.
Guide, & L. N. van Wassenhove (Eds.), Business Aspects of Closed-Loop Supply
Chains: Exploring the Issues: 317-354. Pittsburgh: Carnegie Mellon University
Press.
Fielding, S. 1999. Going for the green: ISO 14001 delivers profits. Industrial
Management, (March-April): 31.
Fielding, S. 2000. ISO 14001 delivers profits. Automotive Engineering International,
(July): 140.
Fishbein, B. 1998. EPR: What Does It Mean? Where Is It Headed? P2: Pollution
Prevention Review, 8(4): 43-55.
199
Fishbein, B. K. 2000. Carpet Take-back: EPR American style. Environmental Quality
Management, 10(1): 25-36.
Fleder, John R. 1999. A Voluntary Disclosure Program for FDA – The Time Has Come,
54 Food and Drug Law Journal 389-399.
Fleischmann, M. 2003. Reverse Logistics Network Structures and Design. In J. V. Daniel
R. Guide, & L. N. van Wassenhove (Eds.), Business Aspects of Closed-Loop
Supply Chains: Exploring the Issues. Pittsburgh: Carnegie Mellon University
Press.
Fleischmann, M., Krikke, H. R., Dekker, R., & Flapper, S. D. P. 2000. A Characterisation
of Logistics Networks for Product Recovery. Omega, 28(6): 653-666.
Florida, R. & Davison, D. 2001. Gaining from green management: Environmental
management systems inside and outside the factory. California Management
Review, 43(3): 64-84.
Florida, R., & D. Davison . 2001. Gaining from green management: Environmental
management systems inside and outside the factory, 43(3) California
Management Review 64-84.
Foss, N. & Knudsen, C. (Eds.). 1996. Towards a Competence Theory of the Firm. New
York: Routledge.
Friedman, Lawrence (1975) The Legal System: A Social Science Perspective. New York:
Russell Sage Foundation.
Gaither, C. 2003. Giving PCs the Boot - Responsibly while industry has boosted
recycling, most computers are simply tossed. The Boston Globe, (April 22): F1.
Geltman, Elizabeth Glass, & Carey Ann Mathews . 1997. Environmental Law and
Business in the 21st Century: Environmental Democracy, Iowa Journal of
Corporate Law, 22: 395-410.
Gerde, V. W. & Logsdon, J. M. 2001. Measuring environmental performance: use of the
toxics release inventory (TRI) and other US environmental databases. Business
Strategy and the Environment, 10(5): 269-285.
Gibbs, G. 2002. Ford seeks to offload Kwik Fit. The Guardian, (Jan. 16).
Gibbs, Jack (1986) Deterrence Theory and Research,in G. Melton, ed., The Law as a
Behavioral Instrument. Lincoln, Nebraska: University of Nebraska Press.
200
Godshall, L. E. 2000. ISO 14001: A Case Study in Certification at Bayer Pharmaceuticals
in Berkeley, California. Paper presented at the 2nd POSTI meeting, Strasbourg,
France.
Goggin, K., Reay, E., & Browne, J. 2000. Modelling End-of-Life Product Recovery
Chains: A Case Study. Production Planning and Control, 11(2): 187-196.
Goldsmith, Michael, & King, Chad W. 1997. Policing Corporate Crime: The Dilemma of
Internal Compliance Programs, Vanderbilt Law Review, 50: 1-47.
Goodstein, Jerry D. 1994. Institutional Pressures and Strategic Responsiveness: Employer
Involvement in Work-Family Issues, Academy of Management Journal, 37(2):
350-382.
Grant, R. M. 1991. The Resource-Based Theory of Competitive Advantage: Implications
for Strategy Formulation. California Management Review, 33(3): 114-135.
Gray, W. B. & Scholz, J. T. 1993. Does regulatory enforcement work? A panel analysis
of OSHA enforcement. Law & Society Review, 27(1): 177-214.
Gray, Wayne, & Scholz, John T. 1991. Analyzing the Equity and Efficiency of OSHA
Enforcement, Law & Policy, 13: 185-214.
Gray, Wayne, & Deily, Mary E. 1996. Compliance and Enforcement: Air Pollution
Regulation in the US Steel Industry, Journal of Environmental Economics and
Management, 31: 96-111.
Gray, Wayne, & Shadbegian, Ronald. 2005. When and Why Do Plants Comply? Paper
mills in the 1980s, Law & Policy, 27(2): 238-261.
Grayson, E. Lynn, & Landgraf, Christina M. 1997. EPA’s Audit Policy and State Audit
Privilege Laws: Moving Beyond Command and Control? Environmental Law
Reporter, 27: 10243.
Greenhouse, S. 1999. 4 companies gain accord in labor suit. New York Times, (Aug. 10):
A-10.
Greenhouse, S. 2003. A hip-hop star's fashion line is tagged with a sweatshop label. New
York Times, (Oct. 28): B1.
Guide, V. D. R. J. & van Wassenhove, L. N. 2003. Business Aspects of Closed-Loop
Supply Chains. In V. D. R. Guide, Jr., & L. N. van Wassenhove (Eds.), Business
Aspects of Closed-Loop Supply Chains: Exploring the Issues. Pittsburgh:
Carnegie Mellon University Press.
201
Guide, V. D. R., Jayaraman, V., Srivastava, R., & Benton, W. C. 2000. Supply-Chain
Management for Recoverable Manufacturing Systems. Interfaces, 30(3): 125-142.
Guide, V. D. R., Jr. & Shrivastava, R. 1997. Buffering from material recovery uncertainty
in a recoverable manufacturing environment. The Journal of the Operational
Research Society, 48(5): 519-529.
Guide, V. D. R., Jr. & van Wassenhove, L. N. 2001. Managing Product Returns for
Remanufacturing. Production and Operations Management, 10(2): 142-155.
Guide, V. D. R., Jr. 2000. Production Planning and Control for Remanufacturing:
Industry Practice and Research Needs. Journal of Operations Management, 18(4):
467-483.
Guide, V. D. R., Jr., Jayaraman, V., & Linton, J. D. 2003. Building Contingency Planning
for Closed-Loop Supply Chains with Product Recovery. Journal of Operations
Management, 21(3): 259-279.
Guldbrandsen, B. & Haugland, S. A. 2000. Explaining Vertical Integration: Transaction
Cost Economics and Competence Considerations. Paper presented at the
International Society for New Institutional Economics (ISNIE) Annual
Conference, Tübingen, Germany.
Gunningham, N. 1995. Environment, self-regulation, and the chemical industry:
Assessing Responsible Care. Law & Policy, 17(1): 57-109.
Gunningham, Neil, & Grabosky, Peter. 1998. Smart Regulation: Designing
Environmental Policy. Oxford: Clarendon Press.
Gunningham, Neil, Thornton, Dorothy & Kagan, Robert A. 2004. Motivating
Management: Corporate Compliance in Environmental Protection, in JSP/Center
for the Study of Law and Society Faculty Working Papers. Berkeley, CA.
Gunningham, Neil, Thornton, Dorothy & Kagan, Robert A. 2005. Motivating
Management: Corporate Compliance in Environmental Protection, Law & Policy
27(2): 289-316.
Gunningham, Neil, Kagan, Robert A. & Thornton, Dorothy. 2003. Shades of Green:
Regulation, Business, Enforcement. Palo Alto: Stanford University Press.
Gupta, M. & Piero, T. 2003. Environmental management is good business. Industrial
Management, 45(5): 14-19.
202
Guthrie, Doug, & Louise Marie Roth . 1999. The State, Courts, and Maternity Policies in
U.S. Organizations: Specifying Institutional Mechanisms, American Sociological
Review, 64: 41-63.
Guy, S. 2003. Lexmark lawsuit worries cartridge remanufacturers. Chicago Sun-Times,
(May 21): 70.
Hall, B. & Kerr, M. L. 1991. 1991-1992 Green Index: A State-By-State Guide to the
Nation's Environmental Health. Washington DC: Island Press.
Hamilton, James T. 1993. Politics and Social Costs: Estimating the Impact of Collective
Action on Hazardous Waste Facilities, RAND Journal of Economics 24(1): 101125.
Hamilton, James T. 1999. Exercising Property Rights to Pollute: Do Cancer Risks and
Politics Affect Plant Emission Reductions? Journal of Risk and Uncertainty
18(2): 105-124.
Hammond, R., Amezquita, T., & Bras, B. A. 1998. Issues in the Automotive Parts
Remanufacturing Industry: Discussion of Results from Surveys Performed among
Remanufacturers. International Journal of Engineering Design and Automation –
Special Issue on Environmentally Conscious Design and Manufacturing, 4(1): 2746.
Hamschmidt, J. & Dyllick, T. 2001. ISO 14001: Profitable? Yes! But is it eco-effective?
Greener Management International, 34: 43-53.
Harrington, Winston. 1988. Enforcement Leverage when Penalties are Restricted,
Journal of Public Economics 37: 29-53.
Hart, S. 1995. A Natural Resource Based View of the Firm. Academy Of Management
Review, 20: 986-1014.
Hauser, W. & Lund, R. T. 2003. Remanufacturing: An American Resource. Boston:
Boston University.
Häversjö, T. 2000. The financial effects of ISO 9000 registration for Danish companies.
Managerial Auditing Journal, 15(1): 47-52.
Heckman, J. J., Ichimura, H. & Todd, P. 1997. Matching as an Econometric Evaluation
Estimator: Evidence from Evaluating a Job Training Programme, The Review of
Economic Studies 64(4): 605-654.
Heckman, J. J., Ichimura, H., & Todd, P. 1998. Matching as an econometric evaluation
estimator. The Review of Economic Studies, 65(2): 261-294.
203
Helland, Eric. 1998. The Enforcement of Pollution Control Laws: Inspections,
Violations, and Self-Reporting, The Review of Economics and Statistics, 80(1):
141-153.
Henriques, I. & Sadorsky, P. 1996. The determinants of an environmentally responsive
firm: an empirical approach. Journal of Environmental Economics and
Management, (May): 381-395.
Heras, I., Dick, G. P. M., & Casadesús, M. 2002. ISO 9000 registration's impact on sales
and profitability: A longitudinal analysis of performance before and after
accreditation. International Journal of Quality and Reliability Management,
19(6): 774-791.
Hideaki, A. 1999. Environmental Standard Bearers. Look Japan.
Hillary, R. & Thorsen, N. 1999. Regulatory and self-regulatory measures as routes to
promote cleaner production. Journal of Cleaner Production, 7(1): 1-11.
Hillary, R. 1997. The Eco-Management and Audit Scheme: Analysis of the regulation,
implementation and support. Imperial College of Science, University of London,
London.
Hoffman, B. 2000. Ford Finding Treasure in Trash. Automotive News, (Jan. 3): 16.
Hormozi, A. M. 1997. Parts Remanufacturing in the Automotive Industry. Production
and Inventory Management Journal, 38(1): 26-31.
Humphries, Martha A., & Songer, Donald R. 1999. Law and Politics in Judicial
Oversight of Federal Administrative Agencies, Journal of Politics, 61(1): 207220.
Hunt, Terrell E., & Wilkins, Timothy A. 1992. Environmental Audits and Enforcement
Policy, 16 The Harvard Environmental Law Review 365-427.
IAF. 2003. IAF Guidance on the Application of ISO/IEC Guide 66: General
Requirements for Bodies Operating Assessment and Certification/registration of
Environmental Management Systems (EMS) (Issue 3, IAF GD 6:2003).
Cherrybrook, Australia: International Accreditation Forum, Inc.
Industrial Economics. 2003. State Program Status And Trends. Paper presented at the
National Environmental Performance Track State And Regional Conference,
Denver.
204
Ingram, Paul, & Simons, Tal. 1995. Institutional and Resource Dependence Determinants
of Responsiveness to Work-Family Issues, Academy of Management Journal,
38(5): 1466-1482.
Innes, Robert. 2001. Violator Avoidance Activities and Self-Reporting in Optimal Law
Enforcement, Journal of Law, Economics & Organizations, 17: 239-256.
International Council of Chemical Associations. 2002. Responsible What? Brussels.
ISO. 1996. ISO 14001: Environmental Management Systems -- Specification with
Guidance for Use. Geneva: International Organization for Standardization (ISO).
ISO. 2002. The ISO Survey of ISO 9000 and ISO 14000 Certificates: Eleventh cycle: up
to and including 31 December 2001. Geneva: International Organization for
Standardization (ISO).
ISO. 2004. The ISO Survey of ISO 9000 and ISO 14000 Certificates: up to and including
31 December 2003. Geneva: International Organization for Standardization (ISO).
Jiang, R. J. & Bansal, P. 2003. Seeing the need for ISO 14001. Journal of Management
Studies, 40(4): 1047-1067.
Jin, G. Z. & Leslie, P. 2003. The effect of information on product quality: Evidence from
restaurant hygiene grade cards. Quarterly Journal of Economics, 118(2): 409-451.
Jørgensen, H. B., Pruzan-Jørgensen, P. M., Jungk, M., & Cramer, A. 2003. Strengthening
Implementation of Corporate Social Responsibility in Global Supply Chains.
Washington DC: World Bank.
Kagan, Robert A., & John T. Scholz (1984) The Criminology of the Corporation and
Regulatory Enforcement Styles, in K. Hawkins & J.M. Thomas, eds., Enforcing
Regulation. Boston: Kluwer Academic Pub.
Kagan, Robert. A., Gunningham, Neil & Thornton, Dorothy. 2003. Explaining Corporate
Environmental Performance: How Does Regulation Matter? Law & Society
Review 37(1): 51-90.
Kaplow, Louis, & Shavell, Steven. 1994. Optimal Law Enforcement with Self-Reporting
of Behavior, Journal of Political Economy 102: 583-606.
Karapetrovic, S. & Willborn, W. 1998. Integration of Quality and Environmental
Management Systems. The TQM Magazine, 10(3): 204-213.
205
Kerr, W. & Ryan, C. 2001. Eco-efficiency gains from remanufacturing: A case study of
photocopier manufacturing at Fuji Xerox Australia. Journal of Cleaner
Production, 9: 75-81.
Kesan, Jay P. 2000. Innovations in Environmental Policy: Encouraging Firms to Police
Themselves: Strategic Prescriptions to Promote Corporate Self-Auditing,
University of Illinois Law Review 155-184.
Khanna, M. & Anton, W. Q. 2002. Corporate environmental management: Regulatory
and market-based pressures. Land Economics, 78(4).
Khanna, M. & Damon, L. 1999. EPA's voluntary 33/50 program: Impact on toxic
releases and economic performance of firms. Journal of Environmental
Economics and Management, 37: 1-25.
King, A. & Lenox, M. 2000. Industry self-regulation without sanctions: The chemical
industry's responsible care program. Academy of Management Journal, 43(4):
698-716.
King, A. & Lenox, M. 2001. Lean and green? An empirical examination of the
relationship between lean production and environmental performance. Production
and Operations Management, 10(3): 244-256.
King, A. 1995. Innovation from differentiation: Pollution control departments and
innovation in the printed circuit industry. IEEE Transactions on Engineering
Management, 42(3): 270-277.
King, A. 1999. Retrieving and transferring embodied data: Implications for the
management of interdependence within organizations. Management Science,
45(7): 918-935.
King, Andrew A., & Lenox, Michael J. 2000. Industry Self-Regulation Without
Sanctions: The Chemical Industry's Responsible Care Program, Academy of
Management Journal, 43: 698-716.
King, Andrew A., & Lenox, Michael. J., & Terlaak, Anne. Forthcoming. Strategic use of
decentralized institutions: Exploring certification with the ISO 14001
management standard. Academy of Management Journal.
King, D. S. & Mackinnon, A. 2002. Who Cares? Community Perceptions in the
Marketing of Corporate Citizenship. In J. Andriof, S. Waddock, B. Husted, & S.
S. Rahman (Eds.), Unfolding Stakeholder Thinking: 233-250. London: Greenleaf
Publishing.
206
Klassen, R. & McLaughlin, C. 1993. TQM and Environmental Excellence in
Manufacturing. Industrial Management and Data Systems, 93(6): 14-22.
Klausner, M. & Hendrickson, C. T. 2000. Reverse-Logistics Strategy for Product Takeback. Interfaces, 30(3): 156-165.
Klausner, M., Grimm, W., & Hendrickson, C. 1998. Reuse of Electric Motors of
Consumer Products: Design and Analysis of a Usage Data Log. Journal of
Industrial Ecology, 2(2): 89-102.
Klein, B. & Leffler, K. B. 1981. The Role of Market Forces in Assuring Contractual
Performance. Journal of Political Economy, 89(4): 615-641.
Klein, B. 1996. Market Power in Aftermarkets. Managerial and Decision Economics,
17(2): 143-164.
Klein, B., Crawford, R. G., & Alchian, A. A. 1978. Vertical Integration, Appropriable
Rents, and the Competitive Contracting Process. Journal of Law and Economics,
21(October): 297-326.
Klein, Jill, & Dawar, Hiraj. 2004. Corporate Social Responsibility and Consumers’
Attributions and Brand Evaluations in a Product-Harm Crisis, International
Journal of Research in Marketing, 21: 203-217.
Klineberg, S. L., McKeever, M., & Rothenbach, B. 1998. Demographic predictors of
environmental concern: It does make a difference how it's measured. Social
Science Quarterly, 79(4): 734-753.
Kniesner, T. J. & Leeth, J. D. 2004. Data mining mining data: MSHA enforcement
efforts, underground coal mine safety, and new health policy implications.
Journal of Risk and Uncertainty, 29(2): 83-111.
Koenigsberger, M. 2004. Environmental Compliance/Performance and ISO 14001 - a 3M
Case Study. Paper presented at the Auditing Roundtable 2004 Winter Meeting,
Pheonix, AZ.
Kogut, B. & Zander, U. 1992. Knowledge of the Firm, Combinative Capabilities, and the
Replication of Technology. Organization Science, 3(3): 383-397.
Kolk, A. 2000. The Economics of Environmental Management. New York: Financial
Times.
Krikke, H. R., van Harten, A., & Schuur, P. C. 1999. Business case Océ: Reverse logistic
network re-design for copiers. OR Spektrum, 21(3): 381-409.
207
Krueger, A. B. & Mas, A. 2004. Strikes, scabs and tread separations: Labor strife and the
production of defective Bridgestone/Firestone tires. Journal of Political Economy,
112(2): 253-289.
Kuperan, K., & Sutinen, Jon G. 1998. Blue Water Crime: Deterrence, Legitimacy, and
Compliance in Fisheries, Law & Society Review, 32: 309–330.
Laplante, Benoît, & Rilstone, Paul. 1996. Environmental Inspections and Emissions of
the Pulp and Paper Industry in Quebec, Journal of Environmental Economics and
Management, 31(1): 19-36
Lave, L., Conway-Schempf, N., Harvey, J., Hart, D., Bee, T., & MacCracken, C. 1998.
Recycling Post-Consumer Nylon Carpet: A Case Study of the Economics and
Engineering Issues Associated with Recycling Post-Consumer Goods. Journal of
Industrial Ecology, 2(1): 117-126.
Lawrence, A. T. & Morell, D. 1995. Leading-edge environmental management:
Motivation, opportunity, resources and processes. In D. Collins, & M. Starik
(Eds.), Special Research Volume of Research in Corporate Social Performance
and Policy, Sustaining the Natural Environment: Empirical Studies on the
Interface Between Nature and Organizations: 99-126. Greenwich, CT: JAI Press.
Lenox, Michael J. & Nash, Jennifer. 2003. Industry self-regulation and adverse selection:
A comparison across four trade association programs. Business Strategy and the
Environment, 12: 343-356.
Leuven, E. & Sianesi, B. 2003. PSMATCH2: Stata module to perform full Mahalanobis
and propensity score matching, common support graphing, and covariate
imbalance
testing,
Version
1.2.3.
http://ideas.repec.org/c/boc/bocode/s432001.html.
Levinson, A. 1996. Environmental regulations and manufacturers’ location choices:
Evidence from the Census of Manufactures. Journal of Public Economics, 62: 529.
Licking, E. 1998. Drawing a bead on recyclables. Business Week, (Dec. 7): 61.
Little, R. J. A. & Rubin, D. B. 1987. Statistical Analysis with Missing Data. New York:
John Wiley & Sons.
Louwers, D., Kip, B. J., Peters, E., Souren, F., & Flapper, S. D. P. 1999. A Facility
Location Allocation Model for Reusing Carpet Materials. Computers and
Industrial Engineering, 36(40): 1-15.
208
Low, M. K., Williams, D., & Dixon, C. 1996. Choice of End-of-Life Management
Strategy: A Case Study in Alternative Telephone Concepts. Paper presented at the
1996 IEEE International Symposium on Electronics and the Environment, Dallas.
Lublin, David, & D. Stephen Voss . 2001. Federal Elections Project. American
University, Washington, DC and the University of Kentucky.
http://www.american.edu/academic.depts/spa/ccps/fepindex.html
Lund, R. T. 1996. The Remanufacturing Industry: Hidden Giant. Boston: Boston
University.
Lynn, Frances M., Busenberg, George, Cohen, Nevin & Chess, Caron . 2000. Chemical
Industry's Community Advisory Panels: What Has Been Their Impact?
Environmental Science and Technology, 34(10): 1881-1886.
Macau Productivity and Technology Transfer Centre. 2001. Subsidy & Loan Programs:
Incentive Subsidy for ISO 9000 / ISO 14000 Certification Fees.
www.cpttm.org.mo/quality/program/program_e.htm, accessed July 11, 2002.
MacDonald, J. P. 2001. Giving the Ship a Compass: Sustainability Management Systems
Using The Natural Step Framework and ISO 14001. Dalhousie University,
Halifax, Nova Scotia.
MacLean, R. 2004a. Environmental management systems - Part 2: Getting the most from
your EMS. Environmental Protection Magazine, 15(3).
MacLean, R. 2004b. Environmental management systems: Do they provide real business
value? Environmental Protection Magazine, 15(2): 12-14.
Magat, Wesley A., & Viscusi, W. Kip. 1990. Effectiveness of the EPA’s Regulatory
Enforcement: The Case of Industrial Effluent Standards, Journal of Law and
Economics 33: 331-360.
Mahoney, J. T. 2001. A Resource-Based Theory of Sustainable Rents. Journal of
Management, 27(6): 651-660.
Majumdar, S. K. & Marcus, A. A. 2001. Rules versus discretion: The productivity
consequences of flexible regulation. Academy of Management Journal, 44(1):
170-179.
Majumder, P. & Groenevelt, H. 2001. Competition in Remanufacturing. Production &
Operations Management, 10(2): 125-141.
209
Malmsheimer, Robert W., & Floyd, Donald W. 2004. US Courts of Appeals judges'
review of Federal Natural Resource Agencies' Decisions, Society and Natural
Resources, 17(6): 533-546.
Markoff, J. 2003. 2 PC makers given credit and blame in recycling. New York Times,
(June 27): C3.
Matthews, D. H. 2001. Assessment and Design of Industrial Environmental Management
Systems. PhD Dissertation, Carnegie Mellon University, Pittsburgh.
Maxwell, J. W., Lyon, T. P., & Hackett, S. C. 2000. Self-regulation and social welfare:
The political economy of corporate environmentalism. Journal of Law &
Economics, 43(2): 583-619.
May, Peter J. 2004. Compliance Motivations: Affirmative and Negative Bases, Law &
Society Review, 38: 41-68.
McGraw Hill. 2001. Market Report - ISO 14001 Registered Company Directory North
America. Vol 2, No 1. Washington DC.
McMichael, F. C. & Hendrickson, C. T. 1998. Recycling Batteries. IEEE Spectrum,
35(2): 35-42.
McNeil, D. G., Jr. 2004. At last, a company takes PETA seriously. New York Times, (July
25): 4-4.
Medcof, J. W. 2001. Resource-Based Strategy and Managerial Power in Networks of
Internationally Dispersed Technology Units. Strategic Management Journal, 22:
999-1012.
Medina-Ross, V. 2002. Environmental Management in the Mexican Chemical Sector:
Drivers, Obstacles and Implications. Unpublished manuscript, Norwich, UK.
Medinger, Jason D. 2003. Antitrust Leniency Programs: A Call for Increased
Harmonization as Proliferating Programs Undermine Deterrence, Emory Law
Journal, 52: 1439-1481.
Meidinger, Errol. (1987) Regulatory Culture: A Theoretical Outline, 9(4) Law & Policy
355-386.
Melnyk, S. A., Sroufe, R. P., Calantone, R. L., & Montabon, F. L. 2002. Assessing the
effectiveness of US voluntary programmes: An empirical study. International
Journal of Production Research, 40(8): 1853-1878.
210
Mendelhoff, John, & Gray, Wayne. 2005. Inside OSHA’s Black Box: What is the Link
between Inspections, Citations, and Reductions in Different Injury Type? Law &
Policy, 27(2): 219-237.
Meyer, John W., & Rowan, Brian. 1977. Institutionalized Organizations: Formal
Structure as Myth and Ceremony, American Journal of Sociology, 83: 340-363.
Mezias, S. J. 1990. An institutional model of organizational practice: financial reporting
at the Fortune 200. Administrative Science Quarterly, 35: 431-457.
Morandi, Larry . 1998. State Environmental Audit Laws and Policies: An Evaluation.
Denver: National Conference of State Legislatures.
Mukamel, D. B. & Mushlin, A. I. 1998. Quality of care information makes a difference:
An analysis of market share and price changes after publication of the New York
State Cardiage Surgery Mortality Reports. Medical Care, 36(7): 945-954.
Murray, Paula C. 1999. Inching Toward Environmental Regulatory Reform -- ISO
14000: Much Ado About Nothing or a Reinvention Tool? American Business Law
Journal, 37: 35-71.
Naimon, J., Shastri, K., & Sten, M. 1997. Do environmental management programs
improve environmental performance trends? A study of Standard & Poors 500
companies. Environmental Quality Management, 7(1): 81-90.
NAPA. 2001. Third-Party Auditing of Environmental Management Systems: U.S.
Registration Practices for ISO 14001. Washington, DC: National Academy of
Public Administration (NAPA).
Nash, J. & Ehrenfeld, J. 1997. Codes of environmental management practice: Assessing
their potential as a tool for change. Annual Review of Energy and the
Environment, 22: 487-535.
Nash, J., Ehrenfeld, J., MacDonagh-Dumler, J., & Thorens, P. 2000. ISO 14001 and
EPA'S Region 1's Startrack Program: Assessing their Potential as Tools in
Environmental Protection. Washington DC: National Academy of Public
Administration.
Nasr, N., Hughson, C., Varel, E., & Bauer, R. 1998. State-of-the-art Assessment of
Remanufacturing Technology. Rochester, NY: National Center for Resource
Recovery and Remanufacturing, Rochester Institute of Technology.
National Safety Council. 1999. Electronic Product Recovery and Recycling Baseline
Report: Recycling of Selected Electronic Products in the United States.
Washington, DC: National Safety Council.
211
NEETF. 2001. Going Green... Upstream: The Promise of Supplier Environmental
Management. Washington DC: The National Environmental Education &
Training Foundation.
Ochsner, M. 2000. Case study: Risk prioritization and ISO 14001 at Acushnet Rubber
Company. Environmental Quality Management, 9(4): 45-52.
OECD. 2001. OECD Series on Pollutant Release and Transfer Registers, Number 4: Why
Pollutant Release and Transfer Registers (PRTRs) Differ: A Review of National
Programmes (Report ENV/JM/MONO. 2001. 16). Paris: Organisation for
Economic Cooperation and Development (OECD).
O'Rourke, D. 2001. Sweatshops 101: Lessons in Monitoring Apparel Production Around
the World. Dollars & Sense, (Sept./Oct.).
O'Rourke, D. 2002. Monitoring the monitors: A critique of corporate third-party labor
monitoring. In R. Jenkins, R. Pearson, & G. Seyfang. (Eds.), Corporate
responsibility and labour rights: Codes of conduct in the global economy.
London: Earthscan.
O'Rourke, D. 2003. Outsourcing regulation: Analyzing nongovernmental systems of
labor standards and monitoring. Policy Studies Journal, 31(1): 1-29.
Orts, Eric W. 1995. Reflexive Environmental Law, Northwestern University Law Review,
89: 1227-1340.
Paternoster, Raymond & Simpson, Sally. 1996. Sanction Threats and Appeals to
Morality: Testing a Rational Choice Model or Corporate Crime, Law & Society
Review, 30(3): 549-584.
Peloza, John . 2005. Corporate Social Responsibility as Reputation Insurance, Paper
presented at the 2nd Annual Corporate Social Performance Conference, Haas
School of Business, University of California, Berkeley, April.
Peterson, Melody, & Drew, Christopher. 2003. The Slaughterhouse Gamble: The Risk of
Self-Policing; New Safety Rules Fail to Stop Tainted Meat, New York Times (10
Oct.): A-1.
Pfaff, Alexander S.P., & Sanchirico, Chris W. 2000. Environmental Self-Auditing:
Setting the Proper Incentives for Discovery and Correction of Environmental
Harm, Journal of Law, Economics & Organizations, 16: 189-208.
Pfaff, Alexander S.P., & Sanchirico, Chris W. 2004. Big Field, Small Potatoes: An
Empirical Assessment of EPA's Self-Audit Policy, Journal of Policy Analysis and
Management, 23(3): 415-432.
212
Pfeffer, J. & Salancik, G. R. 1978. The External Control of Organizations. New York:
Harper & Row.
Potoski, M. & Prakash, A. Forthcoming. Covenants with Weak Swords: ISO 14001 and
Facilities' Environmental Performance. Journal of Policy Analysis and
Management.
Potter, S. 1996. Transforming Frigidaire. co-design, 5&6: 81-83.
Prahalad, C. K. & Hamel, G. 1990. The Core Competence of the Corporation. Harvard
Business Review, 68(3): 79-91.
Prakash, A. 2000. Greening the Firm: The Politics of Corporate Environmentalism. New
York: Cambridge University Press.
Puckett, J., Byster, L., Westervelt, S., Gutierrez, R., Davis, S., Hussain, A., & Dutta, M.
2002. Exporting Harm: The High Tech Trashing of Asia. Seattle & San Jose:
Basel Action Network and Silicon Valley Toxics Coalition.
Puhani, P. A. 2000. The Heckman correction for sample selection and its critique.
Journal of Economic Surveys, 14(1): 53-68.
QSU. 2002a. ISO 9000 Registered Company Directory North America. Fairfax, VA:
QSU Publishing.
QSU. 2002b. ISO 14001 Registered Company Directory North America, Volume 2,
Number 2. Fairfax, VA: QSU Publishing Company.
Raines, S. S. 2002. Implementing ISO 14001--An international survey assessing the
benefits of certification, Corporate Environmental Strategy, 9(4): 418-426.
Rechtschaffen, C. 2004. Enforcing the Clean Water Act in the Twenty-First Century:
Harnessing the Power of the Public Spotlight, Alabama Law Review, 55: 775-814.
Recycling Today. 2002. Fugère Family Acquires GreenLeaf Canada. (Dec. 10).
Redinger, C. F. & Levine, S. P. 1998. Development and evaluation of the Michigan
Occupational Health and Safety Management System Assessment Instrument: A
universal OHSMS performance measurement tool. American Industrial Hygiene
Association Journal, 59(8): 572-581.
Rees, J. 1994. Hostages of Each Other: The Transformation of Nuclear Safety Since
Three Mile Island. Chicago: University of Chicago Press.
213
Rees, J. 1997. The development of communitarian regulation in the chemical industry.
Law & Policy, 19: 477-528.
Revesz, R.L. 2001. Federalism and Environmental Regulation: A Public Choice
Analysis, 115(2) Harvard Law Review, 553-641.
Riesenberg, T. L. 2002. The Non-Audit Service Restrictions of the Sarbanes-Oxley Act.
Daily Tax Report (185): J-1.
Rivera, Jorge. 2002. Assessing a voluntary environmental initiative in the developing
world: The Costa Rican Certification for Sustainable Tourism. Policy Sciences,
35: 333-360.
Rivera, Jorge, & de Leon, Peter. 2004. Is Greener Whiter? The Sustainable Slopes
Program and the Voluntary Environmental Performance of Western Ski Areas,
Policy Studies Journal, 32(3): 417-437.
Robèrt, K.-H. 2000. Tools and concepts for sustainable development, how do they relate
to a general framework for sustainable development, and to each other? Journal
of Cleaner Production, 8(3): 243-254.
Rondinelli, D. & Vastag, G. 2000. Panacea, common sense, or just a label? The value of
ISO 14001 environmental management systems. European Management Journal,
18(5): 499-510.
Rondinelli, D. A. & London, T. 2003. How corporations and environmental groups
cooperate: Assessing cross-sector alliances and collaborations. Academy of
Management Executive, 17(1): 61-76.
Rondinelli, D. A., Berry, M. A., & Vastag, G. 1997. Strategic Programming for
Environmental Management: Sonoco's Take-back Policy. Business Horizons,
40(3): 23-32.
Rose, C. M. 2000. Design For Environment: A Method for Formulating Product End-ofLife Strategies., Stanford University, Unpublished PhD Dissertation.
Rosenbaum, P. R. & Rubin, D. B. 1983. The central role of the propensity score in
observational studies for causal effects. Biometrika, 70(1): 41-55.
Ross, H. Laurence (1982) Interrupted Time Series Studies of Deterrence of Drinking and
Driving, in J. Hagan, ed., Deterrence Reconsidered: Methodological Innovations.
Beverly Hills: Sage Publications.
214
Rowland, E. & Sheldon, C. 1999. The Natural Step and ISO 14001: Guidance on the
Integration of a Framework for Sustainable Development into Environmental
Management Systems. London: British Standards Institution.
Rugman, A. M. & Verbeke, A. 1998. Corporate strategies and environmental regulations:
An organizing framework. Strategic Management Journal, 19(4): 363-375.
Russo, M. V. & Fouts, P. A. 1997. A Resource-based Perspective on Corporate
Environmental Performance and Profitability. Academy of Management Journal,
40: 534-559.
San Francisco Chronicle. 1999. Ending Saipan Sweatshops (Editorial). (Aug. 17): A18.
San Jose Mercury News. 2002. High tech should take the high road on e-waste
(Editorial). (Nov. 27).
Savaskan, R. C., Bhattacharya, S., & van Wassenhove, L. N. 2004. Closed-Loop Supply
Chain Models with Product Remanufacturing. Management Science, 50(2): 239252.
Schoenberger, K. 2002a. Recycling solutions for personal computers are limited. San
Jose Mercury News, (Nov. 26).
Schoenberger, K. 2002b. Where computers go to die: Poor cities in China become
dumping ground for e-waste. San Jose Mercury News, (Nov. 23).
Scholz, John T. (1984) Cooperation, Deterrence, and the Ecology of Regulatory
Enforcement, Law & Society Review, 18(2): 179-224.
Schultz, Vicki, & Stephen Petterson . 1992. Race, Gender, Work and Choice: An
Empirical Study of the Lack of Interest Defense in Title VII Cases Challenging
Job Segregation, 59 University of Chicago Law Review, 1073-1181.
Schylander, E. & Zobel, T. 2003. Environmental effects of environmental management
systems: Evaluation of the evidence. Paper presented at the Proceedings of
Evaluation of Sustainability European Conferences (EASY-ECO 2), May 15-17,
Vienna.
Scigliano, E. 1999. Europe May Ban Today's Toxic-Laden Computers, If the US and
WTO Don't Block it. Seattle Weekly, (Nov. 18).
Scott, W. Richard, & John W. Meyer (1983) The Organization of Societal Sectors, in
J.W. Meyer & W.R. Scott, eds., Organizational Environments: Ritual and
Rationality. Beverly Hills, CA: Sage.
215
Sharma, S. & Vredenburg, H. 1998. Proactive Corporate Environmental Strategy and the
Development of Competitively Valuable Organizational Capabilities. Strategic
Management Journal, 19: 729-753.
Shelanski, H. A. & Klein, P. G. 1995. Empirical Research in Transaction Cost Economics
- a Review and Assessment. Journal of Law Economics & Organization, 11(2):
335-361.
Shover, Neil, John Lynxwiler, Stephen Groce, & Donald Clelland (1984) Regional
Variation in Regulatory Law Enforcement: The Surface Mining Control and
Reclamation Act of 1977, in K. Hawkins & J. Thomas. eds., Enforcing
Regulation. Boston, MA: Kluwer Nijhoff.
Sianesi, B. 2004. An Evaluation of the Active Labour Market Programmes in Sweden.
The Review of Economics and Statistics, 86(1): 133-155.
Simon, M., Bee, G., Moore, P., Pu, J.-S., & Xie, C. 2001. Modelling of the life cycle of
products with data acquisition features. Computers in Industry (1534): 1-12.
Singapore Ministry of the Environment. 2002. Green Business: ISO 14001 in Singapore.
www.env.gov.sg/info/business/iso14000.htm, accessed July 7, 2002.
Sissell, K. 2000. Autos and electronics drive certification. Chemical Week, 162(14): 4243.
Smith, G. & Feldman, D. 2003. Company Codes of Conduct and International
Standards: An Analytical Comparison - Part I of II. Washington, DC: World
Bank.
Smith, J. A. & Todd, P. E. 2001. Reconciling conflicting evidence on the performance of
propensity-score matching methods. American Economic Review, 91(2): 112-118.
Smith, J. A. & Todd, P. E. 2005. Does matching overcome LaLonde's critique of
nonexperimental estimators? Journal of Econometrics, 125(1-2): 305-353.
Snow, C. C. & Hrebiniak, L. G. 1980. Strategy, Distinctive Competence, and
Organizational Performance. Administrative Science Quarterly, 25(2): 317-336.
Sockabasin, N. (TRI Coordinator, U.S. EPA Region 9). 2004. Personal conversation on
Oct. 6.
Spence, David B., & Murray, Paula. 1999. The Law, Economics, and Politics of Federal
Preemption Jurisprudence: A Quantitative Analysis, California Law Review, 87:
1125-1206
216
Spence, M. 1973. Job market signaling. Quarterly Journal of Economics, 87(3): 355-374.
Stewart, R. B. 2001. A new generation of environmental regulation? Capital University
Law Review, 29: 21-182.
Strachan, P. A., Sinclair, I. M., & Lal, D. 2003. Managing ISO 14001 implementation in
the United Kingdom Continental Shelf (UKCS). Corporate Social Responsibility
and Environmental Management, 10(1): 50-63.
Streitfeld, D. 2003. Media Copyright Law Put to Unexpected Uses; Companies are using
legislation meant to restrain Web piracy to try to shut down rivals. Los Angeles
Times, (February 23): 3-1.
Strom, S. 1996. A sweetheart becomes suspect; Looking behind those Kathie Lee labels.
New York Times, (June 27).
Sun, H. 1999. Diffusion and contribution of total quality management: An empirical
study in Norway. Total Quality Management, 10(6): 901-914.
Sunstein Cass R., Schkade, David & Ellman, Lisa M. 2004. Ideological Voting on
Federal Courts of Appeals: A Preliminary Investigation, Virginia Law Review,
90(1): 301-354.
Szymanski, M. & Tiwari, P. 2004. ISO 14001 and the reduction of toxic emissions.
Policy Reform, 7(1): 31–42.
Tate, C. Neal & Handberg, Roger. 1991. Time Binding and Theory Building in Personal
Attribute Models of Supreme Court Voting Behavior, 1916-88, American Journal
of Political Science, 35(2): 460-480.
Taylor, B. 2001. Chain Reaction. Recycling Today, (Sept. 17).
Teece, D. J. 1978. Vertical integration in the U.S. oil industry. In E. J. Mitchell (Ed.),
Vertical Integration in the Oil Industry. Washington DC: American Enterprise
Institute for Public Policy Research.
Teece, D. J. 1984. Economic Analysis and Strategic Management. California
Management Review, 26: 87-110.
Teece, D. J., Pisano, G., & Shuen, A. 1997. Dynamic Capabilities and Strategic
Management. Strategic Management Journal, 18(7): 509-533.
Teliha, K. (Branch Chief, Pollution Prevention, Indiana Department of Environmental
Management). 2004. Personal conversation on Oct. 7.
217
Terlaak, A. K. 2002. Exploring the Adoption Process and Performance Consequences of
Industry Management Standards: The Case of ISO 9000, Unpublished PhD
Dissertation, University of California at Santa Barbara.
Terziovski, M., Samson, D., & Dow, D. 1997. The business value of quality management
systems certification - Evidence from Australia and New Zealand. Journal of
Operations Management, 15(1): 1-18.
The Economist. 2001. Science and technology: In the black. (January 27): 80-81.
Thierry, M., Salomon, M., Van Nunen, J., & Van Wassenhove, L. 1995. Strategic Issues
in Product Recovery Management. California Management Review, 37(2): 114135.
Thompson, J. D. 1967. Organizations in Action. New York: McGraw-Hill.
Thornton, Dorothy, Gunningham, Neil & Kagan, Robert A. 2005. General Deterrence
and Corporate Environmental Behavior, Law & Policy, 27(2): 262-288.
Thornton, Dorothy, Kagan, Robert A. & Gunningham, Neil. 2003. Sources of Corporate
Environmental Performance, California Management Review, 46(1): 127-141.
Tjiong, H. 2002. Institutional Dynamics in Environmental Corporatism. The Impact of
Market and Technological Change on the Dutch Polder Model. Bonn, Germany:
Max Planck Institute for Research on Collective Goods.
TNRCC. 2003. Environmental Management Systems - Incentives. Texas Natural
Resource
Conservation
Commission.
http://www.tnrcc.state.tx.us/exec/sbea/ems/incentives.html, accessed Sept 27,
2004.
Toffel, Michael. 2000. Anticipating Greener Supply Chain Demands: One Singapore
Company's Journey to ISO 14001. In R. Hillary (Ed.), ISO 14001: Case Studies
and Practical Experiences. Sheffield, UK: Greenleaf Publishing.
Toffel, Michael W. 2003. The Growing Strategic Importance of End-of-Life Product
Management. California Management Review, 45(3): 102-129.
Toffel, Michael W. & Marshall, Julian D. 2004. Improving environmental performance
assessment: A comparative analysis of weighting methods used to evaluate
chemical release inventories. Journal of Industrial Ecology, 8(1-2): 143-172.
Tojo, N. 2001. Extended producer responsibility legislation for electrical and electronic
for electronic equipment - approached in Asia. Paper presented at the Discussion
Paper at 3rd Asia Pacific Roundtable for Cleaner Production (APRCP), Manila.
218
Tyler, Tom (1990) Why People Obey the Law. New Haven: Yale University Press.
US EPA. 1986. Study of the Pulp and Paper Industry in Region IV. Washington, DC:
United States Environmental Protection Agency.
US EPA. 1990. Environmental Enforcement: A Citizen's Guide. Washington DC: United
States Environmental Protection Agency, Office Of Enforcement (March),
http://www.epa.gov/region4/air/enforce/citizenf.htm
US EPA. 1992. Identifying the Regulated Community and Establishing Program
Priorities, Chapter 4 in Principles of Enforcement. Washington DC: United States
Environmental
Protection
Agency.
EPA
Number
300F93001,
www.inece.org/enforcementprinciples.html
US EPA. 1995a. Incentives for Self-Policing: Disclosure, Correction, and Prevention of
Violations, Federal Register, 60: 66706.
US EPA. 1995b. Sector Notebook - Profile of the Electronics and Computer Industry
(EPA/310-R-95-002). Washington, DC: United States Environmental Protection
Agency.
US EPA. 1998. 1996 Toxic Release Inventory Data Quality Report. Washington, DC:
United States Environmental Protection Agency.
US EPA. 1999a. EPCRA Section 313 Release Reporting Requirements blue brochure.
Washington, DC: United States Environmental Protection Agency.
US EPA. 1999b. Making the business case for EPR: A Snapshot of leading practices and
tools. Washington, DC: United States Environmental Protection Agency.
US EPA. 2001. Product Stewardship. Washington, DC: US Environmental Protection
Agency. http://www.epa.gov/epr/about/index.html, accessed Feb 18, 2002.
US EPA. 2002. User’s Manual for RSEI Version 2.1 [1988-2000 TRI Data]. Washington,
DC: United States Environmental Protection Agency.
US EPA. 2002a. The Toxics Release Inventory (TRI) and Factors to Consider When
Using TRI Data. Washington DC: United States Environmental Protection
Agency, http://www.epa.gov/tri/2002_tri_brochure.pdf
US EPA. 2002b. User’s Manual for RSEI Version 2.1 [1988-2000 TRI Data].
Washington, DC: United States Environmental Protection Agency.
219
US EPA. 2003a. Guidance on the Use of Environmental Management Systems in
Enforcement Settlements as Injunctive Relief and Supplemental Environmental
Projects. Washington DC: United States Environmental Protection Agency.
US EPA. 2003b. Regulations governing hazardous waste generators, RCRA Orientation
Manual (EPA530-R-02-016). Washington DC: United States Environmental
Protection Agency.
US EPA. 2004. EPA's Method for Calculating Air Toxics Emissions for Reporting
Results Needs Improvement (Report No. 2004-P-00012). Washington DC: United
States Environmental Protection Agency, Office of Inspector General.
US EPA. 2005. 2003 Toxics Release Inventory Shows Continued Decline in Chemical
Releases (Press Release). Washington, DC: United States Environmental
Protection Agency. (May 11).
US GAO. 2000. Environmental Protection: More Consistency Needed Among EPA
Regions In Approach To Enforcement. Washington DC: United States General
Accounting Office (Report GAO/RCED-00-108).
Vandenbergh, Michael. 2003. Beyond Elegance: A Testable Typology of Social Norms
in Corporate Environmental Compliance, Stanford Environmental Law Journal,
22: 55-144.
Vidovic, M. & Khanna, N. 2003. Reducing Toxic Releases through Voluntary Efforts:
Incentives for Participation in the 33/50 Program and Implications for the
Distribution of Pollution. Paper presented at the Western Economic Association
International 78th Annual Conference, Denver, Colorado.
Walgenbach, P. 2001. The production of distrust by means of producing trust.
Organization Studies, 22(4): 693-714.
Walton, S. V., Handfield, R. B., & Melnyk, S. A. 1998. The green supply chain:
Integrating suppliers into environmental management process. International
Journal of Purchasing and Materials Management, 34(2): 2-11.
Wasserman, C. 1987. Environmental compliance and enforcement: Theory, practice, and
the challenge to environmental economists (EPA Report 230-10-87-030). Paper
presented at the Association of Environmental and Resource Economists,
Newark, Delaware.
Wasserman, Cheryl E. 1990. An Overview of Compliance and Enforcement in the United
States: Philosophy, Strategies, and Management Tools, First International
Enforcement Workshop Proceedings, International Network for Environmental
220
Compliance and Enforcement, Utrecht,
http://www.inece.org/1stvol1/1stTOC.htm
The
Netherlands
(May),
Welch, E. W., Mazur, A., & Bretschneider, S. 2000. Voluntary behavior by electric
utilities: levels of adoption and contribution of the climate challenge program to
the reduction of carbon dioxide. Journal of Policy Analysis and Management, 19:
407–425.
Welch, E. W., Mori, Y., & Aoyagi-Usui, M. 2002. Voluntary adoption of ISO 14001 in
japan: Mechanisms, stages and effects. Business Strategy and the Environment,
11: 43–62.
Welch, Eric W., Allan Mazur, & Stuart Bretschneider . 2000. Voluntary Behavior by
Electric Utilities: Levels of Adoption and Contribution of the Climate Challenge
Program to the Reduction of Carbon Dioxide, Journal of Policy Analysis and
Management, 19: 407–425.
Wells, R. P. & Galbraith, D. 1999. Proyecto Guadalajara: Promoting sustainable
development through the adoption of ISO 14001 by small and medium-sized
enterprises. Greener Management International, (Winter): 90.
Wernerfelt, B. 1984. A Resource-based View of the Firm. Strategic Management
Journal, 5: 171-180.
Westphal, J. D. & Zajac, E. J. 1994. Substance and symbolism in CEO's long-term
incentive plans. Administrative Science Quarterly, 39: 367-390.
Westphal, J. D., Gulati, R., & Shortell, S. M. 1997. Customization or conformity? An
institutional and network perspective on the content and consequences of TQM
adoption. Administrative Science Quarterly, 42(2): 366-394.
White, C. D., Masanet, E., Rosen, C. M., & Beckman, S. 2003. Product Recovery With
Some Byte: An Overview of Management Challenges and Environmental
Consequences in Reverse Manufacturing for the Computer Industry. Journal of
Cleaner Production, 11(4): 445-458.
Wilkinson, G. & Dale;, B. G. 1999. Integrated Management Systems: An Examination of
the Concept and Theory. The TQM Magazine, 11(2): 95-104.
Williamson, O. E. 1981. The economics of organization: The transaction-cost approach.
American Journal of Sociology, 87: 548-577.
Williamson, O. E. 1983. Credible Commitments: Using Hostages to Support Exchange.
American Economic Review, 73(4): 519-540.
221
Williamson, O. E. 1985. The Economic Institutions of Capitalism: Firms, Markets,
Relational Contracting. New York: Free Press.
Williamson, O. E. 1998. Transaction Cost Economics: How it Works; Where it is
Headed. De Economist, 146(1): 23-58.
Williamson, O. E. 2000. The New Institutional Economics: Taking Stock, Looking
Ahead. Journal of Economic Literature, 38(3): 595-613.
Williamson, O. E. 2002. The Lens of Contract: Private Ordering. American Economic
Review, 92(2): 438-443.
Winter, S. G. 1988. On Coase, Competence, and the Corporation. Journal of Law,
Economics, and Organization, 4(1): 163-180.
Winter, Soeren C., & May, Peter J. 2001. Motivation for Compliance with Environmental
Regulations, Journal of Policy Analysis & Management, 20: 675–698.
Woodall, Mark. 1997. Secrecy Bills Derailed in States, The Planet 4(1).
www.sierraclub.org/planet/199701/states.asp
WWF. 1996. Forests For Life 1996 Annual Report. Zürich: WWF International.
Wycherley, I. 1999. Greening supply chains: The case of The Body Shop International.
Business Strategy and the Environment, 8(2): 120-127.
Yardley, J. 1989. On Diapers and Other Disposables. Washington Post, (Jun 26): B2.
Yiridoe, E. K., Clark, J. S., Marett, G. E., Gordon, R., & Duinker, P. 2003. ISO 14001
EMS standard registration decisions among Canadian organizations.
Agribusiness, 19(4): 439-457.
Zuk, Gary, Deborah J. Barrow, & Gerard S. Gryski . 1996. A Multi-User Database On
The Attributes Of U.S. Appeals Court Judges, 1801-1994 [Computer file]. 1st
ICPSR version. Gary Zuk, Deborah J. Barrow, & Gerard S. Gryski, Auburn
University [producers], 1996. Ann Arbor, MI: Inter-university Consortium for
Political and Social Research [distributor], 1997.
222