31 ELR 10811 | Environmental Law Reporter | copyright © 2001 | All rights reserved

Regulatory Negotiation and the Legitimacy Benefit

Jody Freeman is Professor of Law at the University of California, Los Angeles. Laura I. Langbein is a Professor in the School of Public Affairs at American University. The authors thank Gary Blasi, Neil Kerwin, and Rick Sander for helpful comments. Marc Luesebrink provided excellent research assistance. The authors gratefully acknowledge research support from the U.S. Environmental Protection Agency. All errors and omissions are ours. This Article first appeared in the following issue of the New York University Environmental Law Journal: 9 N.Y.U. ENVTL. L.J. 60 (2000). It has been reprinted here with only minor changes.

[31 ELR 10811]

Notwithstanding that only a few agencies use regulatory negotiation¹ for a tiny minority of rules, the process attracts a remarkable amount of scholarly attention. While proponents claim that negotiation improves rule quality, reduces transaction costs, and increases legitimacy, critics contend that the process fails to deliver its purported benefits and abrogates an agency's responsibility to execute its delegated functions. Some commentators have gone so far as to argue that consensus-based policymaking of this sort is fundamentally undemocratic.²

To date, these debates have remained largely theoretical. What little empirical evidence exists consists mostly of case studies of particular reg negs, with no comparison to conventional rulemaking. The examples tend to focus, moreover, on narrow questions, such as whether reg neg reduces rulemaking time and cuts costs. Only one empirical study has so far compared conventional and negotiated rules,³ but it focuses on just two aspects of the process—time and litigation rates—and its methodology has been controversial.⁴ Thus, the majority of claims advanced in the debate over negotiated versus conventional rulemaking remain largely untested, and policymakers still lack guidance on whether and when to use one or the other approach.

In this Article, we present a summary of new empirical evidence from Neil Kerwin and Laura Langbein's two-phase study of U.S. Environmental Protection Agency (EPA) negotiated rulemakings.⁵ Their qualitative and quantitative data reveal more about reg neg than any empirical study to date; although not published in a law review article until now, they unquestionably bear upon the ongoing debate among legal scholars over the desirability of negotiating rules. The findings of the studies tend, on balance, to undermine arguments made by the critics of regulatory negotiation and to bolster the claims of proponents. Kerwin and Langbein found that, according to participants in the study, reg neg generates more learning, better quality rules, and higher satisfaction compared to conventional rulemaking.⁶ At the same time, stakeholder influence on the agency remained about the same using either approach.⁷ Based on these researchers' results, we recommend more frequent use of regulatory negotiation, accompanied by further comparative [31 ELR 10812] and empirical study for the purposes of establishing regulatory standards and resolving implementation and compliance issues. This recommendation contradicts the prevailing view that the process is best used sparingly,⁸ and even then, only for narrow questions of implementation.⁹

In our view, empirical studies of negotiated rulemaking that examine cost, time, and litigation rates tell only part of the story and, we believe, not the most important part. The studies summarized here went beyond these limited measures of success and provided a more textured picture of regulatory negotiation. Along virtually every important qualitative dimension, all participants in this study—whether business, environmental, or government—reacted more favorably to their experience with negotiated rules than did participants in conventional rulemaking.¹⁰ Contrary to the critics' expectations, Kerwin and Langbein found that negotiation of rules reduced conflict between the regulator and regulated entities, and it was no less fair to regulated entities than conventional rulemaking.¹¹ The data contradict claims that regulatory negotiation abrogates an agency's responsibility to implement laws written by Congress¹²; indeed, the process may better enable the agency to fulfill that role. The researchers concluded that regulatory negotiation clearly emerges, moreover, as a superior process for generating information, facilitating learning, and building trust.¹³ Most significantly, consensus-based negotiation increased legitimacy, defined as the acceptability ofthe regulation to those involved in its development.¹⁴ This legitimacy benefit, which was observed independently of the types of rules chosen for conventional versus negotiated rulemaking, and independently of differences among the participants, including their affiliation,¹⁵ is no small accomplishment and we argue that, in any event, it is more important than reducing transaction costs.

The study reveals some weaknesses of the reg neg process as well, most notably the disproportionate costs it imposed on smaller groups with comparatively fewer resources.¹⁶ Whether or not to participate in a reg neg proved a more difficult decision for environmental organizations and other similar groups than for larger parties like big business or state government regulators. Smaller, poorer groups also reported suffering from resource deficits as compared to their larger, richer negotiating partners.¹⁷ The evidence of resource disadvantage provides empirical support for a frequent criticism of reg neg, but, importantly, these disparities did not seem to translate into undue influence over outcomes.¹⁸ In light of the numerous benefits revealed by the data, the story on reg neg remains mostly positive especially when compared to conventional rulemaking.

Inevitably, the study leaves some questions unanswered. For example, Kerwin and Langbein do not identify the precise causal pathway by which regulatory negotiation produces the legitimacy benefit. The data fail to isolate the exact connection between increased satisfaction and the particular procedural or substantive variables (those features of the process or the outcome) responsible for it. Participants appeared more satisfied than their conventional rulemaking counterparts, mostly because they assessed the substantive outcome as better than the rule that would have been promulgated otherwise, but also because some aspects of the reg neg process itself work well. Based on the dramatic reports of higher learning, and evidence that face-to-face interaction built trust, we believe it is a fair inference that the unique features of the reg neg process—its collective, interactive, and consensual nature—played a key role, either directly or indirectly, in generating increased satisfaction. We are dubious, moreover, that other processes, such as informal policy dialogues, have the potential to produce the legitimacy benefit associated with regulatory negotiation, as some have suggested.¹⁹ Participant reports of increased satisfaction appeared uniquely linked to the reg neg process itself, even if the exact etiology of the higher satisfaction effect remains unidentified.

At first glance, one might assume that overcoming cognitive dissonance explains the higher satisfaction rates—after all, participants devoted considerable time and resources to regulatory negotiation. But we think that this possibility, even if plausible, is largely irrelevant.²⁰ First, these sophisticated parties did not hesitate to criticize reg neg; despite their overall positive evaluations, fully 95% identified things they disliked.²¹ Second, and more important, however, we simply reiterate the facts: regardless of the reason [31 ELR 10813] reported, participant satisfaction was higher for negotiated than for conventional rules (on which participants also expended effort), and the legitimacy of the outcomes in their eyes was also greater. Cognitive dissonance (or some other psychological mechanism) might help to explain how regulatory negotiation works, but it does nothing to diminish the value of the process, unless the intervening psychological mechanism responsible for the legitimacy benefit has some independent negative status.²² In other words, reports of satisfaction may be "biased," but it made no difference to the legitimacy benefit.²³

One might also suspect that higher satisfaction rates correlate to disproportionate influence over the agency, which would suggest that the purported legitimacy benefit simply disguises capture of the agency by interested parties. Based on the reported data, however, this is unlikely. The results of the study indicated that the agency was equally responsive to stakeholders in both conventional and negotiated rulemaking contexts.²⁴ Although participants did perceive that some parties exerted disproportionate influence in the reg neg process, the types of parties believed to have exerted that influence were fairly evenly distributed. In fact, the parties that were perceived as exerting the most influence in both types of rulemakings were EPA itself and big business groups of all stripes.²⁵ Environmental groups were slightly more likely to be seen as exercising disproportionate influence in reg negs than in conventional rulemaking.²⁶ Moreover, perceptions of disproportionate influence in conventional rulemaking occurred with the same relative frequency as those in reg negs.²⁷ The researchers think it would be inaccurate to suggest that satisfaction depends on, or disguises, undue influence over the agency that is exacerbated by reg neg.

At worst, then, we believe that regulatory negotiation might enable partial capture, but no more so than conventional rulemaking. Further, assuming reg neg does enable capture, there is no evidence that the nature or extent of the capture produced through this consensus-based process is greater or more sinister than the capture that occurs through traditional notice-and-comment rulemaking. Powerful groups, such as industry trade associations or government agencies, may fare better in all decision contexts because of resource, information, and political asymmetries that work in their favor, but there is no reason to believe that regulatory negotiation enhances their advantage. Although in this Article we identify and recommend ways to ameliorate these asymmetries among parties in the reg neg context, we doubt the differences can be eradicated.

As to whether reg neg produces "better rules" in some objective sense, Kerwin and Langbein do not say. They noted only that participants, including EPA, perceived that reg neg produces better rules than would have otherwise been obtained.²⁸ Empirically assessing rule quality is challenging, of course, given the difficulty of measuring the net social value of any regulation, however it is produced.²⁹ Kerwin and Langbein did observe that participants claimed the rules were improved partly because these participants perceived that the outcomes favored their interests. Indeed, the data show that for each additional point that a respondent ascribed to the net benefits of the rule for her organization and to the overall efficiency of the rule, her rating of the overall process increased by 0.3 on an 11-point scale.³⁰ We emphasize, however, that the researchers' analysis showed that these variables alone do not explain satisfaction rates.³¹ The higher satisfaction with outcomes reported by reg neg participants remained, independently of the participants' view of the rule's net economic benefit to them and independently of their estimation of its overall economic efficiency. Further, we note that greater participant satisfaction could well be consistent (or at least not inconsistent) with the interests of the larger public. That is, higher satisfaction with reg neg among a group of participants does not necessarily indicate that the product of reg neg is any worse for the greater public than the outcome of the conventional process.

The reported results of this study are consistent with the possibility that satisfaction does not wholly depend on perceptions of "winning" or exacting gains from other groups. Instead, the data suggest that the legitimacy benefit turns, to a significant extent, on participation in a process, specifically one that presents an opportunity to affect the outcome. This view is bolstered by the work of social psychologists, which indicates that involvement in a process enhances perceptions of legitimacy among participants, independently of whether outcomes ultimately favor these participants.³² [31 ELR 10814] This interpretation is consistent with yet another finding of the study: two-thirds of the participants in negotiated rulemakings, regardless of group affiliation, believed that their effect on the outcome was substantial.³³ In sum, the data here support what numerous other studies have already demonstrated: process matters.

We believe not only that people assess legitimacy on procedural as well as substantive grounds, but also that, when choosing a process, legitimacy ought to concern us at least as much as cost, time, and litigation rates. This is especially so where the difference between alternative processes on these three scores is not great, as with reg neg versus conventional rulemaking. Given the legitimacy benefit, the learning effect, and the absence of evidence of greater capture, we believe that the latest empirical evidence shifts the burden to those who oppose regulatory negotiation to articulate when and why agencies ought not to use it

I. The Scholarly Debate Over Regulatory Negotiation

A. Enthusiasts

Early proponents of regulatory negotiation advocated its use for a variety of reasons, chief among which were improved rule quality and legitimacy.³⁴ Indeed, they defined legitimacy largely in terms of the attitudes of the most directly affected interest groups, using them (rightly or wrongly) as a proxy for the larger public.³⁵ Advocates such as Philip Harter and Lawrence Susskind believed not only that direct participation in rulemaking would produce better quality rules, but that it would also increase the rules acceptability to those most affected by them. Proponents expected that repeated face-to-face interaction would lead to better information production, which in turn would improve rule quality. That is, not only would negotiations allow parties to trade interests in order to reach agreement, it would also enable them to educate each other, pool knowledge, and cooperate in problem solving.³⁶ In addition, sharing responsibility for rule development would foster in the parties a sense of ownership over the outcome, rendering it more acceptable—that is, more legitimate. Greater legitimacy could be particularly valuable at a time of heightened frustration with conventional rulemaking and broad dissatisfaction with government regulation.³⁷

In turn, enthusiasts hoped that greater acceptability would yield other instrumental benefits, including easier implementation (because obstacles to implementation would likely surface and be addressed in the negotiations) and higher rates of compliance (because parties that consent to the rule in advance would be more likely to comply with it).³⁸ Surely the parties would be more likely to implement a rule they helped to develop, and less inclined to sue.³⁹

Early proponents also anticipated that reg neg might reduce the transaction costs associated with conventional rulemaking.⁴⁰ In the experience of many practitioners, traditional notice-and-comment rulemaking under the Administrative Procedure Act (APA),⁴¹ 5 U.S.C. § 553, had grown needlessly time-consuming and unnecessarily adversarial. As it had evolved, the process encouraged parties to marshal an enormous volume of irrelevant evidence, adopt extreme positions, and use information defensively. Responding to comments required considerable agency staff time, slowed the pace of rulemaking, and produced unnecessary conflict. The academics and practitioners who proposed regulatory negotiation hoped that a consensus-based approach would temper the adversarial nature of rulemaking and help to channel resources in a more fruitful way. Among other [31 ELR 10815] things, they thought, the demands of negotiating would force parties to prioritize among their concerns, focus on key issues, moderate their positions, and share information productively.⁴² Direct engagement with parties holding opposing views would prompt the interests on all sides of a regulatory issue to get to the heart of their disagreements faster, which would help to speed the process along.

Still, reduced transaction costs were incidental to the two principal goals of improved quality and legitimacy. Moreover, none of the early proponents claimed that regulatory negotiation represented a panacea for all that ailed the regulatory process. No one argued, for example, that consensus-based decisionmaking might, by itself, overcome the lamentable ossification that beset notice-and-comment rulemaking and that it, therefore, ought to replace traditional rulemaking. Indeed, even the most enthusiastic adherents of consensus-based rulemaking acknowledged the limitations of the process from the start. They argued simply that regulatory negotiation, carefully deployed, could improve both the quality and legitimacy of rules.⁴³ Further, they believed, based on a combination of experience and theory, that the precise nature of the process—stakeholder representation, face-to-face negotiation, consensus-based decisionmaking—was crucial to its success.⁴⁴

B. Skeptics

Regulatory negotiation proved more popular in alternative dispute resolution circles than among administrative law scholars, who attacked it first on theoretical, and later on empirical, grounds. For some, the mere idea of negotiating rules with stakeholders seemed anathema to the traditional concept of the agency as a faithful agent of Congress. Regulatory negotiation invites agency abdication of responsibility, they argued, by shifting the decisionmaking burden to stakeholders who owe no duty to the public or to Congress. The process thus embodies what many administrative law theorists viscerally fear: the last step from a system of arm's-length interest representation, which preserves the agency's hierarchical authority, to one of direct interest group bargaining.⁴⁵ At a time when public choice theory and its unsentimental account of the legislative and administrative process was on the ascendance in law schools, regulatory negotiation seemed to portend its darkest implications.⁴⁶

Critics argued, moreover, that even if a consensus-based approach to rulemaking might meet democratic standards of legitimacy under some circumstances, surely regulatory negotiation would not succeed in practice. First, the process is insufficiently inclusive because only a limited number of parties can participate without negotiations becoming unwieldy.⁴⁷ Moreover, the power to convene a negotiating group carries with it the power to manipulate outcomes. Alone, or in collusion with powerful groups, the agency might rig outcomes in advance through the selection of some stakeholders and the exclusion of others.

In addition, critics anticipated that a consensus approach would favor more powerful, well-financed interests with access to money, information, and technical expertise.⁴⁸ Trade associations and large firms in particular would enjoy significant advantages over smaller parties or parties with fewer resources, such as state governments, environmental or labor groups, or small businesses. This advantage, critics believed, would translate into influence over the outcomes. Moreover, even if agencies could balance negotiating committees with representatives from all sides, no single interest could adequately represent the average voter or consumer and, for this reason alone, the process would fall short of American standards of democratic legitimacy. Indeed, critics suspected that regulatory negotiation would be more likely than conventional rulemaking to undermine the public interest and lead to outcomes of dubious legality.⁴⁹ For some or all of these reasons, critics viewed regulatory negotiation as, at best, a minor reform for use in limited and [31 ELR 10816] tightly controlled circumstances, or, at worst, fundamentally undemocratic.⁵⁰

C. The Proponents Respond

For the most part, defenders of reg neg retorted that negotiated rules were far from secret deals. The Negotiated Rulemaking Act of 1990 (NRA) requires federal agencies to provide notice of regulatory negotiations in the Federal Register,⁵¹ to formally charter reg neg committees,⁵² and to observe the transparency and accountability requirements⁵³ of the Federal Advisory Committee Act.⁵⁴ Any individual or organization that might be "significantly affected" by a proposed rule can apply for membership in a reg neg committee,⁵⁵ and even if the agency rejects their application, they remain free to attend as spectators.⁵⁶ Most significantly, the NRA requires that the agency submit negotiated rules to traditional notice-and-comment rulemaking.⁵⁷

In addition, many public choice scholars argue that agencies have no incentive to shirk their accountability to congressional principals, who control agency budgets, appoint top personnel, and oversee agency authority. Agencies thus have no incentive to be less responsive to congressional preferences in negotiated rulemaking than with conventional rulemaking.⁵⁸

Proponents of reg neg argued that, in view of these safeguards, agencies are equally accountable for negotiated and conventional rules. Moreover, external checks on agency decisionmaking remain undisturbed by reg neg. Providing they meet traditional standing hurdles, any party may seek judicial review of a negotiated rule, and upon review the rule is entitled to no greater deference for having been negotiated. Indeed, Congress specifically declined to provide for a lower standard of review in the NRA.⁵⁹

Finally, in an attempt to show that reg neg was not a dramatic departure from traditional rulemaking, proponents pointed out that informal negotiation with stakeholders has always been an essential part of the rulemaking process. Negotiated rulemaking merely formalizes negotiation and utilizes it earlier in the rulemaking process, when it is likely to be most useful. In this view, agencies actually conform to congressional intent by using processes like reg neg; consultation with the entities that might be harmed by legislation is precisely what Congress intends when it delegates decisionmaking authority to the agency.⁶⁰

II. Empirical Evidence

A. Early Empirical Studies

This debate remained largely theoretical until empirical evidence about regulatory negotiation began to emerge in the latter half of the 1980s and the early 1990s. Early accounts were mostly anecdotal, written by participants in individual reg negs or by scholars studying a few reg negs at a time.⁶¹ These studies largely ignored proponents' theoretical claims that reg neg would improve rule quality and increase legitimacy. Instead, they focused on the more easily measured instrumental benefits of time, cost, and litigation rates. EPA's first commissioned study of its first seven reg negs concluded that negotiated rulemaking produced rules more quickly than conventional rulemaking and was less resource-intensive.⁶² While acknowledging heavy investment in front-end negotiations, EPA calculated that when it negotiated rules, it spent approximately half the time and money it ordinarily would have spent to collect and analyze data and to respond to public comments.⁶³ In addition, the report claimed that negotiating rules increased the likelihood that rulemaking wouldbe completed on time, saving significant staff hours. Still, EPA conceded that the data were inconclusive with respect to whether negotiation resulted in a net savings of resources.⁶⁴

Other evidence emerged to support the idea that regulatory negotiation could lower costs, but it was piecemeal and unreliable. For example, one early trade association study concluded that negotiated rules were at least one-third less [31 ELR 10817] expensive for the industry than conventional notice-and-comment rulemaking, even factoring in the possibility of litigation and remand—but the data were limited.⁶⁵ In an interview conducted by one of us (Freeman) in 1993, the then-Deputy Director of EPA's Regulatory Negotiation Project acknowledged that the data that EPA had collected on reg neg were "inconclusive with respect to whether negotiation resulted in a net savings of resources."⁶⁶ Further, the EPA official conceded that it was difficult to compare litigation rates for conventional and negotiated rules because it was hard to track how many traditional rules get litigated.⁶⁷ As a result, agency officials could not readily compare any meaningful measure of resource use between conventional and negotiated rules.⁶⁸

In 1997, Cary Coglianese published an empirical analysis of 35 negotiated rulemakings, assessing whether the process had achieved the instrumental goals of reducing both rulemaking time and legal challenges to agency rules.⁶⁹ He calculated that negotiated rulemaking provided only minimal time-saving benefits compared with conventional rulemaking (2.8 years or 1,013 days versus 3.0 years or 1,108 days).⁷⁰

Coglianese then measured litigation rates by examining the sample of EPA's negotiated rulemakings and conventional rulemakings conducted during the same period. He calculated that the baseline frequency of litigation of conventional rules was only 26%—a far lower figure than previously thought. He claimed that litigation rates for major rules promulgated by EPA between 1980 and 1991, while higher, was only 35%.⁷¹ By comparison, according to Coglianese's count, negotiated rulemakings completed by EPA had a 50% rate of litigation.⁷² Based on this data, Coglianese concluded that reg neg had not lived up to its promise to save regulatory time and reduce litigation rates, which he claimed were the primary purposes of, and driving forces behind, reg neg's adoption.⁷³ Subsequently, based on these data, Coglianese has argued more broadly against the use of consensus-based processes such as regulatory negotiation, suggesting that other approaches could deliver its purported benefits as well or better. Reg-neg, he argues, is not a necessary precondition to increased compliance, reduced conflict, improved public policy, or greater public participation.⁷⁴

In his article in this volume, Harter challenges the validity of Coglianese's data, arguing that Coglianese's research is "significantly flawed and hence misleading concerning the actual experience with negotiated rulemaking."⁷⁵ Harter points out that Coglianese simply miscalculated the start and end dates of the negotiated rules, and that he improperly included an abandoned negotiated rulemaking as if it had been completed.⁷⁶ Harter argues that reg neg cuts the time consumed by EPA rulemakings by about 32% if the data on reg negs are properly measured and calculated to reflect consistent start and finish dates, and to exclude the abandoned reg negs.⁷⁷ Harter also points out that of the negotiated rules challenged by litigation, none reflected the consensus rule proposed by participants—suggesting that either EPA's departure from consensus or some other aspect of the rule besides the outcomes produced by reg neg provoked the litigation.⁷⁸

Thus, although enthusiasts and skeptics continue to test their theories about reg neg against the emerging empirical evidence, the terms of debate remain narrow; virtually none of the studies to date move beyond the measures of time and litigation rates to address the question of legitimacy and quality.

B. Kerwin and Langbein: Phase I

However, by 1995, Kerwin and Langbein had produced the first part of their detailed two-phase study of EPA's experience with negotiated rulemaking.⁷⁹ Its results shed light on many of the hypotheses advanced in the scholarly literature on regulatory negotiation. Taken together with the results of the second phase,⁸⁰ which explicitly compares negotiated [31 ELR 10818] and conventional rules, the data provide a textured, and largely favorable, picture of reg neg.

In Phase I, Kerwin and Langbein interviewed by phone over 100 participants in negotiated rulemaking, each of whom had been chosen as part of the traditional convening process⁸¹ and who together represented a cross-section of interests from eight negotiated rulemakings undertaken by EPA during the 1980s.⁸² Kerwin and Langbein asked respondents a series of open-ended questions concerning their perceptions of the key features of the process,⁸³ including their decision to participate, how issues were established and conflicts were resolved, what respondents learned and how, and the reasons for, and extent of, satisfaction with the rule.⁸⁴ In addition to posing specific questions on each of these topics, the researchers asked respondents to rank various aspects of the final rule on a 11-point scale, ranging from -5 meaning "the rule could not be worse" to +5 meaning "the rule could not be better."⁸⁵

Some of the Phase I data bear significantly on the ongoing theoretical debate over reg neg by directly contradicting or calling into question a number of the criticisms that appear in the literature. At the same time, much of the data simply report participant impressions about reg neg, which, in the absence of a comparison to conventional rulemaking, remain inconclusive. Still, we present a summary of the Phase I data here because we believe, standing on its own, it furnishes some interesting information about the reg neg process and because legal audiences are unlikely to have encountered it. Some readers might find the participant impressions (as we do) to be inherently interesting or to provide a "feel" for reg neg that academic descriptions of the process cannot capture.

In addition, much of the data highlight dimensions of the reg neg process that have thus far been relatively ignored in the scholarly literature. It appears, for example, that as familiarity and trust built among participants, informal relationships became increasingly important, suggesting that informal negotiations can be crucial to resolving conflicts, something no study has shown. Further, increased interaction seemed to increase commitment to the process: several participants reported that the longer and harder they worked, the more important a successful resolution of the issues became.

Moreover, when disaggregated, the Phase I data shed light on some interesting internal differences among participants (business versus environmental groups, for example) in their reactions to the process. Finally, the Phase I data provided the basis for the explicitly comparative analysis in Phase II; in our view, presenting the findings from Phase I separately helps make the results of Phase II more easily comprehensible.

The Phase I findings indicate that parties viewed regulatory negotiation as a flexible, broadly inclusive, resource-intensive process through which they learned a great deal, particularly from each other, and through which they exerted significant influence on outcomes. Two-thirds of participants considered their contributions to have had a major or moderate impact on the outcome⁸⁶—indeed, participants reported that the opportunity to make an impact on the outcome was one of the aspects of the process they considered most valuable.⁸⁷ Participants perceived negotiated rulemaking as an effective means for developing regulations in virtually every meaningful way.⁸⁸

Participants' overall evaluations appeared to be consistent with these other favorable impressions. The average [31 ELR 10819] overall evaluation was 2.1 on the aforementioned -5 to +5 scale, with 79% giving a positive answer.⁸⁹ Almost 80% of participants reported that the benefits of participation outweighed the costs, although the costs at times were high and disproportionately so for smaller groups.⁹⁰ The participants' overall ratings of the process, and their ratings of their personal experience with reg neg, are strongly and widely positive. Below, in more detail, we present the most salient findings from Phase I, linking them, where appropriate, to the scholarly debate.

1. The Criteria for Selecting Rules Suitable for Reg Neg Varied

Scholars on both sides of the reg neg debate have tended to agree that only certain kinds of rules would make good candidates for development via regulatory negotiation.⁹¹ The literature claims, inter alia, that reg neg will work best when issues are mature and the positions of the parties clear.⁹² In addition, scholars have argued that the issues most suitable for negotiation would allow for bargaining and enable trade offs; most commentators believe that issues involving deeply held values will be poor candidates.⁹³

Some but not all of these factors emerged as relevant criteria in the interviews. When asked why a particular rule was selected for regulatory negotiation, participants offered a variety of reasons, ranging from "the issues were simple or clear" and "there was high level of conflict over issues" to "best way to obtain views of affected interests."⁹⁴ Moreover, the importance of these factors differed across the 12 reg negs, suggesting that suitability for negotiation might be more variable and idiosyncratic than scholars tend to allow.⁹⁵ Even the widely held view about the need to keep participant numbers of participants relatively low (between 15 and 25) may be overstated: one successful reg neg involved 31 parties.⁹⁶ Whether intentional or not, the agency convening process appears flexible enough to accommodate this variation: none of the respondents, including agency officials, perceived that EPA had systematically applied a rigid set of a priori criteria for selecting rules, as the literature suggests it should.

2. Reg Neg Was Broadly Inclusive

The data call into question the validity of the criticism that reg negs involve only highly organized and well-financed interests.⁹⁷ The majority of respondents reported participation by all parties, including small, seemingly ad hoc citizen groups, small businesses, and local government representatives.⁹⁸ These types of participants were not in the majority, but neither were they rare. The data therefore support the proposition that negotiated rulemaking is at least open to groups that complain about exclusion from other governmental processes, even if those groups are imperfect surrogates for "ordinary citizens." When asked whether all the interests that should have been involved in the negotiated rulemaking were involved, 65% of respondents answered that there was full representation.⁹⁹

The literature on reg neg also identifies as a potential problem EPA's unwillingness to commit, up front, to accept the results of negotiations and use them as the basis for the rule. There is no evidence, however, that this factor affected parties' decisions to participate. In fact, no respondent expressed concerns in this regard.¹⁰⁰ This result is relevant only to the decision to participate. Parties did not seem concerned at the outset about whether the agency would be "bound" by the rule. However, it might be the case that parties would evaluate the reg neg less favorably once they completed their work, if they thought the agency had ignored their contributions.¹⁰¹

3. Some Groups Had More Difficulty Making the Decision to Participate Than Others

For most parties, the decision to participate in a reg neg was easy. When asked whether it was easy or difficult to join, 76% of respondents replied that the decision was easy, while only 10% replied that it was difficult.¹⁰² Participation was an easy decision for those with a great deal at stake, those who believed they had adequate resources and expertise, those attracted by an uncommon opportunity to influence policy, those curious about the process, or those who were simply ordered to participate by their boss or membership.¹⁰³ Participation [31 ELR 10820] was difficult for those with limited resources or a lack of familiarity with—or mistrust of—the process. For example, whereas 42% of environmental group respondents said the decision to participate was difficult, only 23% of business respondents had the same response.¹⁰⁴

The literature on reg neg suggests that the agency must be active and engaged in order for the process to succeed.¹⁰⁵ The data suggest that EPA was frequently viewed as encouraging or supporting the respondents' involvement; 65% reported verbal encouragement, 13% reported receiving monetary assistance, and another 13% reported receiving information or data from EPA.¹⁰⁶ At the same time, several respondents (12%) indicated that they had to press the agency to secure membership in the committee.¹⁰⁷

The results demonstrate that the convening process was perceived to be flexible but not without structure, just as advocates have claimed.¹⁰⁸ The agency did not agree to all requests for participation, usually arguing that the interest in question was adequately represented by another party. At the same time, participants often generated suggestions for additional participants to represent interests that ought to be included.¹⁰⁹

4. Ground Rules Were Well Understood and Flexible Enough to Adapt to New Information

The data on procedural rules also support the claim that the reg neg process is flexible. When discussing how ground rules for the negotiation were established, participants pointed to a variety of parties, naming facilitators, participants, and EPA as the most common sources.¹¹⁰ At the same time, it appears that flexibility does not necessarily generate uncertainty; in most cases (93%), participants understood the ground rules.¹¹¹ Most participants noted that the rules did not change significantly during the course of proceedings, with 44% reporting no change and 52% reporting only small changes.¹¹² When ground rules did change, however, respondents reported that they did so on account of changed circumstances (30% of mentions), because of new issues, coalitions or information (18% of mentions), because trust developed within the committee (17% of mentions), or at the insistence of EPA (10% of mentions).¹¹³ Interviews with participants revealed that as familiarity and trust developed among participants in the negotiations, they viewed the ground rules as less important, while informal discussions became more important and prominent.¹¹⁴

Several respondents complained, however, that achieving consensus on the ground rules consumed too much time and that the conflicts arising at this stage carried over into the substantive negotiations. Given the tendency toward increasing informality it appears that a few fundamental principles could be sufficient to move the negotiations along, subject to modification and supplementation as circumstances require. This would preserve the flexible yet structured approach that advocates say best serves reg neg.

5. The Parties Negotiated Issues in a Fluid Way and Resolved Nearly 80% of Disputes Through the Presentation of Objective Data

Determining which issues will be subject to negotiation represents a crucial dimension of the reg neg process. Here again, the process examined in the study appeared to be dynamic but not entirely without structure. When asked who determined the issues, the most frequent response was "the participants" (44%), followed by "the statute the rule will implement" (28%), and "EPA" (24%).¹¹⁵ From the interviews, it appears that negotiating committees usually did not consider issues in a rigid pre-determined way; rather participants reported that discussions "take on a life of their own," and that parties acted on issues as they emerged and when it seemed that consensus was within reach.¹¹⁶ There is no discernible pattern in the substance of the issues that emerged for negotiation, nor is there a pattern for issues that did not.

With regard to how parties resolved conflicts,¹¹⁷ the respondents' reports showed that "nearly 80% of issues were either successfully negotiated or resolved through the presentation of objective data" or analysis.¹¹⁸ This suggests that regulatory negotiation may produce sufficient scientific and technical information to enable parties to participate effectively in rulemaking, contradicting speculation in the literature that the process merely clarifies interests and facilitates bargaining.¹¹⁹

[31 ELR 10821]

6. Conflicts Produced Shifting Alignments of Interests

When asked about how the contending parties aligned themselves when conflicts arose, respondents reported that predictable alignments developed among like-minded interests or organizations. However, business interests were uniquely in conflict with virtually all the other types of participants at some point during the reg negs.¹²⁰ The participants reported that EPA's position varied considerably; they viewed agency representatives as lining up with virtually every other interest at one time or another. In addition to inter-group conflict, participants reported intra-group conflict with some frequency—especially, but not exclusively, among business interests.¹²¹

7. More Interaction Built Relationships and Increased Commitment to a Successful Result

Data from the interviews indicate that reaching agreement was perceived to be more important to some participants than to others. Respondents explained the difference by citing a variety of reasons, including the impact of the rule on one's organization (34%), commitment to the reg neg process (21%), and the desire to eliminate uncertainty (12%).¹²² Of those who noted "commitment to the process" as the reason why reaching agreement was important, several noted that the longer and harder the group worked, and the more they developed working or personal relationships with other participants, the more important a successful result became.¹²³ This finding underscores how group dynamics can compete with the substance of issues as a factor in consensus-building. However, the research also demonstrated that commitment to the process had no detectable independent effect on the respondent's overall evaluation of the reg neg process. This suggests that "commitment may be important for bringing closure to the process, but is neutral with respect to the overall evaluation of the process and outcome."¹²⁴

8. Regulatory Negotiation Was a Powerful Vehicle for Learning

Participants generally reported that they learned a great deal during the course of a negotiated rulemaking. Among the most frequently reported forms of learning were technical or scientific aspects of the rule (21% of mentions), the issues associated with the rule (11%), the positions of parties other than themselves and reasons why these positions were taken (30%), and how negotiation actually works (18%).¹²⁵ Notably, the most frequently cited source of learning was the other participants in the reg neg (45% of mentions), cited nearly four times as often as EPA.¹²⁶ This contradicts the criticism in the scholarly literature that reg neg does not help parties acquire the scientific and technical information they need to be effective in rulemaking.¹²⁷ However, several respondents noted that the volume of information introduced during the reg neg, and its complexity, was very difficult to absorb.¹²⁸

9. Interests Were Variably Situated With Regard to Information, With Smaller Groups Expressing the Most Disadvantage

Respondents' reports about the information that they needed to participate effectively quite closely tracked their answers about learning. The most frequent type of information needed was technical-scientific (33%), followed by information about the positions of others (18%), knowledge of the issues (18%), legal information (10%), and economic-cost information (5%).¹²⁹ Only 6% of respondents reported needing no additional information to participate in reg neg.¹³⁰ Participants relied on a number of sources for information, including themselves (29%), EPA (20%), other participants (17%), and members of their own coalition (14%).¹³¹ The responses strongly suggest that not all of the participants were equally situated in a negotiated rulemaking with regard to information. EPA and large organizations called upon their own resources or those they could control, while participants with fewer resources most frequently relied on other entities for the information on which they based their decisions.¹³² Some participants did report obtaining funds from the Agency for research or consultants.¹³³

In many instances (60% of mentions), respondents replied that necessary information did not become available during the course of the reg neg.¹³⁴ Technical and scientific information was deemed to be most lacking (20% of mentions), followed by information about the positions of others (13% of mentions), and economic or cost information (10%), while the remaining 40% of responses indicated that no essential information was lacking.¹³⁵ This suggests, contrary to the critics' assertions,¹³⁶ that the process does expose [31 ELR 10822] much, albeit not all, of the essential information for informed decisionmaking.

Reg neg participants reported as reasons for the absence of information, when it was missing, that it was known by some but not shared (31%), that it was too expensive to obtain (23%), or that it was simply not available (20%).¹³⁷ However, of those who reported a lack of information, 23% indicated that it ultimately became available at some point in the negotiation process.¹³⁸ From the Phase I data, it appears that the strategic withholding of information—thought to be common in conventional rulemaking—was perceived by some participants (31%) to be an issue in reg neg as well.¹³⁹

When reservations about the quality of information surfaced, they were more likely to come from groups with limited resource.¹⁴⁰ Environmentalists were the least likely to report that they had all of the information they needed; indeed, no environmental representative mentioned this, compared to 70% of business mentions and 36% of EPA mentions.¹⁴¹ Environmentalists were also most likely to report that they needed scientific and technical information: 64% of their mentions referred to this, compared to about 25% for business and 0% for EPA.¹⁴²

Another dimension of information not directly probed in the interviews emerged when several respondents reported having difficulty absorbing and understanding the implications of information offered during the course of the negotiated rulemaking.¹⁴³ Others referred to a number of technical presentations that occurred simply to establish a minimum level of technical competence in the issue under presentation.¹⁴⁴

This matter is significant for a number of reasons. As Kerwin and Langbein reported:

The intense education that surely occurs in reg neg can properly be viewed as an effort to mitigate the information asymmetries that critics of reg neg assume will persist with this technique. Still, observations of certain participants suggest that some participants will enjoy the powerful advantage of access to and control of superior information which, unless offset, will give them disproportionate control of the agenda relative to the control exercised by smaller, less well-informed, interests. There is a good case to be made that the role of information is as or more important in negotiated rulemaking than in its conventional counterpart due to the pressures created by deadlines and other aspects of the negotiation process. Hence, the problem of information asymmetry that figures so prominently in criticisms of governmental decision-making has been partially addressed in negotiated rulemaking but not completely eliminated.¹⁴⁵

10. Parties Used Numerous Techniques to Allocate Issues to Smaller Decisionmaking Units

The literature on regulatory negotiation suggests that multiple levels of decisionmaking will occur in a reg neg,¹⁴⁶ a hypothesis borne out by the data. Participants identified five distinct decisionmaking units that could be crucial to the final outcome. These included the full negotiating committee, issue-based working subgroups, caucuses or coalitions of participants with similar interests, discussions among the membership of a single interest or organizational participant, and completely informal—at times secret—negotiations between contending participants.¹⁴⁷ The relative influence of these units varied considerably, as did various interests' participation in them.¹⁴⁸

A number of general patterns emerge from the data on decisionmaking. When asked how decisions were made, 59% of respondents answered that subgroups to the full committee made decisions, followed by subgroups alone (18%) and full committee alone (16%).¹⁴⁹ This pattern varied across reg negs. For example, in the coke ovens reg neg, 100% of mentions reported that subgroups alone made decisions, whereas in the minor permit modification reg neg, nearly 50% of respondents reported that the full committee alone made the decisions.¹⁵⁰ Thus, in some cases, subgroups did nearly all of the work and the full committee served to ratify their decisions, while in others the full committee clearly acted as the critical decisionmaking unit. Again, this underscores the variability and flexibility of the reg neg process.

On the important question of defining consensus, 52% of respondents answered that consensus meant unanimity, while 36% said that it meant loose consensus—what people could live with, for example.¹⁵¹ A large number of respondents did not recall taking formal votes, and some noted that the group presumed unanimity unless a party objected outright.¹⁵² Definitions of consensus varied significantly across reg negs, with the unanimity principle most prominent in the coke ovens and clean fuels negotiations, and loose consensus most commonly reported by the asbestos-in-schools group.¹⁵³ This rather flexible approach to consensus blunts [31 ELR 10823] some of the sharpest criticism of reg neg¹⁵⁴ by contradicting the idea that unanimity requires full agreement on every issue. Although holdout problems may well arise¹⁵⁵ and intractable disagreements sometimes do result in the abandonment of a reg neg, a flexible approach to consensus allows participants not bent on sabotage to find ways to concur in the final outcome without agreeing to every decision along the way.

So, for example, if the representative of a trade association or an environmental organization wishes to satisfy her home constituency on a particular issue without ending the reg neg, a loose consensus rule allows her to dissent on the issue without withdrawing from the final consensus.¹⁵⁶ Even a strict, rather than a loose, unanimity requirement may provide this flexibility, if participants decline to take formal votes on each individual issue and vote only on the final package agreement. We liken this to judicial decisionmaking, in which one member of the panel drafts an opinion specifically designed to garner the votes of other members, despite differences remaining among them on specific issues.¹⁵⁷ Of course, the issue concerned cannot be the single deal-breaker for the group, in which case, the representative might prefer to exit the committee and litigate the rule.

11. Informal Negotiation Occurred Frequently and Accomplished a Great Deal

The data clearly suggest that informal negotiations were an inevitable and important part of a process that involved multiple interests with complex interrelationships and long sometimes intense, formal and semi-formal sessions.¹⁵⁸ Participants reported a great deal of informal interaction with others outside of the full committee and established subgroups. When asked about the nature of their communications, respondents characterized them as negotiations (43%), strategy sessions (10%), informational sessions (27%), and other (20%).¹⁵⁹ For those characterized as negotiations, Kerwin and Langbein asked participants to identify both the issues and the parties involved. Once again, they cited an enormous number and diversity of issues, and reported many combinations of parties involved in the negotiations, both within and between coalitions and caucuses.¹⁶⁰

The interviewers then asked the respondents what these informal negotiations did and did not accomplish, whether the results of the informal negotiations could be found in the rule, and their estimate of the magnitude of their contribution to the results of the final regulation. The results were dramatic. Of the 82 responses, only 2 stated that the informal negotiations accomplished nothing.¹⁶¹ The most frequently mentioned functions of informal communication included full or partial resolution of an issue in dispute (44% of responses), the ability to determine whether the participant had the necessary support of his coalition or organization (18% of responses), and enhancement of information (35%).¹⁶² More than 9 out of every 10 participants reported that some or all of the results of these informal communications and negotiations could be found in the rule they developed.¹⁶³ Despite these generally high values, there were, again, significant differences reported across reg negs.¹⁶⁴

[31 ELR 10824]

A small number of interviewees (less than 10%) raised concerns about these informal exchanges, and one expressed the belief that EPA and major interests had rigged the formal negotiation before it began through informal agreements; others expressed a milder, but still significant, concern about the secret and exclusive nature of informal contacts.¹⁶⁵

12. Nearly Two-Thirds of Participants Considered Their Contributions to Have Major or Moderate Impact on the Outcome

The literature on regulatory negotiation reflects fierce disagreement over whether the process exacerbates or ameliorates power imbalances. In a series of questions, Kerwin and Langbein sought to explore explicitly the issue of relative power. They asked participants to describe the impact of their participation on the negotiations. Nearly two-thirds considered their contributions to the rule moderate or major, suggesting that they believed that they had considerable influence over the result.¹⁶⁶ The data on types of participants reveal that environmental and small business representatives were among the most likely to report that their participation had minor or no influence on the proposed rule—41% and 33% respectively.¹⁶⁷ Other groups were less likely to report that their parties had little or no influence—8% of general business participants, 4% of state government participants, and 18% of EPA respondents.¹⁶⁸ While this seems to be a large disparity, the same disparity appeared in conventional rulemaking, with respect to the impact of participation on changes between the proposed and final rule.¹⁶⁹

13. Perceptions of Undue Influence Were About Evenly Divided and the Identity of the Group Seen to Enjoy Disproportionate Influence Varied

When asked whether others exerted disproportionate influence over the development of the rule, 27% of participants said yes, another 48% noted disproportionate participation by some interests, but did not equate this with influence, and the remaining 25% saw no evidence of disproportionate influence.¹⁷⁰ The types of participants most likely to be identified as wielding disproportionate influence were business (27%), EPA (26%), environmental groups (17%), and state agencies (11%).¹⁷¹ Given that both EPA and many state agencies or their national organizations involved in the reg negs have environmental responsibilities and frequently report communication with environmental groups, the perceptions of undue influence appeared to be about evenly divided between EPA, business, and environmental interests.¹⁷² The absence of agreement about the source of disproportionate influence suggests that there was no hidden, behind-the-scene player in reg negs. Further, the same disparities arose in conventional rulemaking.¹⁷³

When asked about evidence of the alleged influence, 44% of participants stated that the content of the rule favored the interest in question; the rest cited some aspect of the process that could result in disproportionate influence, such as the strategic position of EPA, disproportionate access to better information, and a more effective (or noisier) approach to negotiation.¹⁷⁴ There were no significant differences across the reg negs with regard to the occurrence of disproportionate influence, but, when participants did perceive it, the identity of the interest that was perceived as enjoying the influence varied. For example, regulated businesses were perceived as having disproportionate influence in the clean fuels reg neg (36% versus none in the other reg negs), while in the asbestos-in-schools reg neg, consumers of compliance benefits, such as school personnel, were so identified as such (36% versus none in other reg negs).¹⁷⁵ Participants'perceptions of disproportionate influence with regard to the reg negs' content also varied.¹⁷⁶

There is no question that some of the factors identified as the sources of disproportionate influence—superior information, greater resources, access to EPA—would exert a similar effect in conventional rulemaking. Indeed, information and resource asymmetries affect every decisionmaking process. As it turns out, smaller groups with fewer resources tended to dislike both negotiated and conventional rulemaking equally.¹⁷⁷ Finally, at least in some cases, there might be good reason for disproportionate influence: greater influence based on superior information that, in turn, leads to a better quality rule should cause less concern than influence based solely on political clout or threats to sue.

14. Participants Continued to Be Active in the Post-Public Comment Period, but Communicated Less With Each Other

Phase I reports a very active post-negotiation period (after committees have disbanded and after public comments on the notice of proposed rulemaking (NPRM) during which participants continue to communicate with the agency and with members of their own organization and coalition, among others.¹⁷⁸ Individuals reported communicating most [31 ELR 10825] often with EPA and others in their own organization. The reasons most frequently given for such communication were: to provide or obtain information about some element of the rule (72% of responses), to try to effect a change in the final rule (20%), or to plan strategy for the post-rulemaking period (9%).¹⁷⁹ Only 25% of the respondents reported that their post-proposal communications resulted in a change in the rule.¹⁸⁰ Ten percent reported actively working to resolve an outstanding issue.¹⁸¹

Post-proposal communication activity did not differ significantly across the reg negs. However, small business participants were the most likely to report no post-proposal communications (50%), while environmental respondents were the least likely (that is, 0% report no post-proposal communications).¹⁸²

In addition to staying informed and keeping others apprised of the status of the rule, some participants attempted to continue to affect the rule (about 40%).¹⁸³ Most filed comments; others continued to confer with coalition members and EPA.¹⁸⁴ The data suggest variation in the level of effort expended for the different reg negs and in the post-proposal effort across different types of respondents, and this, in turn, may reflect resource disparities. For example, small business and environmental respondents were the most likely to report no or minor post-proposal activity, while big business and EPA respondents were the least likely to report such activity.¹⁸⁵

When asked to evaluate their own contribution to any differences between the proposed and final rule, 32% indicated that they had a major or moderate impact on the change, but 53% said they had no effect.¹⁸⁶ When asked about the effects of the difference on their organization, 25% of respondents reported that the change helped them and 58% stated that the change had no effect at all.¹⁸⁷

15. Proposed Rules Based on Reg Neg Attracted Considerable Public Comment

The results of reg neg on public participation are not entirely consistent with the critics' claim that reg neg silences criticism. A random sampling of the comments received in response to the negotiated rules reveals a large number of concerns and requests for change.¹⁸⁸ Many of the comments were critical, indicating that reg neg's consensus-based approach did not fully anticipate all issues and satisfy all the concerns of affected parties. We show below, however, that the volume of public comments on negotiated rules is less than for conventionally written rules.

16. A Significant Maldistribution of Expenditure Existed Between Larger and Smaller Groups

In response to questions about the type and magnitude of resources they spent on the process, participants reported a wide range of costs. For example, they reported staff hours ranging from 40 to 40,000, clerical staff hours ranging from 0 to 10,000, and monetary expenditures for research, use of consultants, legal counsel, and the like from zero to more than $ 1 million.¹⁸⁹ In light of the wide range of expenditures, and the potential relationship between spending and effectiveness in negotiated rulemaking, Kerwin and Langbein reviewed the identity of the big and small spenders. With respect to research and information collection, the states spent the least (a mean of $ 2,260) and environmental interests were also at the low end; big business spent the most (a mean of $ 432,000).¹⁹⁰ Similarly, the environmental [31 ELR 10826] interests and the states spent the least for consultants and legal counsel while EPA and big business spent the most.¹⁹¹ Although none of these differences is statistically significant, in our view the consistency of the disparity makes the results worth reporting.¹⁹²

When asked to estimate the relative costs of participation in the reg neg for their organizations, respondents on average reported devoting 26% of their available resources to the negotiation.¹⁹³ The differences among respondent types were not significant and were not especially large, but environmentalists reported using among the highest percent of resources (50%) and states among the lowest (12%).¹⁹⁴

17. A Majority (78%) of Participants Said That the Benefits Outweighed the Costs of Participation

Kerwin and Langbein concluded their interviews by asking participants to consider what their organizations had gained, if anything, from participation in a reg neg. They also asked participants to address a number of qualitative dimensions of the rule that resulted from the negotiation.

When asked what their organization gained from negotiated rulemaking, 32% reported that they got a better rule, 28% referred to gaining a better understanding of some aspect of the substantive issue or the process of developing rules, and 11% expressed the belief that they had exercised a greater degree of influence in decisionmaking; only 6% stated that they gained nothing from their participation.¹⁹⁵ When asked whether the benefits of participation outweighed the costs, 78% said that they did, 15% said that the costs outweighed the benefits, and 7% responded that the costs and benefits were roughly equal.¹⁹⁶ The interviewers then asked respondents to rank the rule, in a number of facets,on an 11-point scale from -5 to +5, with -5 meaning "the rule could not be worse" and +5 meaning "the rule could not be better." On all of these important dimensions, participants reacted favorably, as shown in Table 1.¹⁹⁷

18. Learning Topped the List of "Likes" Reported by Respondents

When asked in detail what they liked and disliked about the process, respondents' most common response (43%) related to something that the respondent learned about the issue, other interests, or the process itself.¹⁹⁸ Other respondents mentioned the quality of the rule produced (10%), the willingness of others to be flexible or to negotiate (13%), and interactions with other participants (9%).¹⁹⁹ A significant number (8%) noted that they expected what they had learned and the contacts they had made to pay dividends in future situations involving similar issues or in interaction with the government or other interests, in general.²⁰⁰ Only three respondents reported that they liked nothing about the process.²⁰¹ Virtually all respondents (95%) also found aspects of the process that they disliked, with risk and uncertainty being the most frequent complaint (20%).²⁰² Participants also complained about the time commitment required for participation (18%).²⁰³

19. Respondents Based Their Overall Ratings on a Combination of Process and Substance Variables

The data show that rather than basing their evaluations entirely on either the process or the substance of the rule, respondents based their views on a combination of the two.²⁰⁴ In general, affiliation did not have a significant impact on overall evaluations, except for small business respondents, who rated the reg neg process 2.5 points lower on an 11-point scale than did EPA respondents, once the other variables are held constant.²⁰⁵ Although the Phase I data showed that environmental interests gave the process the lowest ratings, overall those ratings are not significantly lower than EPA's ratings, taking into account the effects of other variables.²⁰⁶ That is, the control variables help to explain why environmental groups gave the final rule the lowest ratings. For example, they may have given the final rule a low rating because they rated its efficiency or the net benefits to their organization lower, or because they found it too complex or less clear. However, the control variables do not entirely explain small business' low ratings.²⁰⁷

Participants' beliefs about whether or not participants were missing from the negotiation process does not appear to have affected respondents' overall evaluation of the reg neg process, even when other variables were held constant.²⁰⁸ [31 ELR 10827] Two other features of the process itself proved significant to respondents' overall evaluations. First, when respondents identified many issues and sides in a reg neg, they perceived the process to be messier and gave it a poorer overall evaluation.²⁰⁹ At the same time, when respondents were unable to explicitly identify sides and issues, they also tended to rate the reg neg lower overall.²¹⁰ So, on the one hand, it appears that too many issues complicate negotiations, but too few may lead participants to oversimplify issues and to miss opportunities to compromise.²¹¹

Respondents' evaluations of the substance of the proposed rule also help to explain their overall evaluations. Each additional point that the respondents ascribed to the benefits of the rule for their organization (on an 11-point scale) correlates to an increase of 0.3 in the rating of the overall process, even when other variables, including the perceived efficiency of the rule, are held constant.²¹² Similarly, for each additional point that respondents ascribed to the proposed rule's economic efficiency, their rating of the overall process also increased by 0.3, holding other variables constant.²¹³ Thus, the two substance and process variables explain some, but not all, of the overall evaluation.

20. Summary

Kerwin and Langbein's Phase I study of eight negotiated rulemakings sheds light on many of the hypotheses advanced in the scholarly literature on regulatory negotiation and tends to undermine a number of criticisms of the process. In sum, the Phase I data reflect favorably on reg neg, supporting claims that it is a flexible, inclusive, and information-intensive process that produces, in the eyes of participants, good quality outcomes that participants think will survive judicial review and result in successful implementation. The study also found that reg neg engendered significant learning, which participants valued highly. Perceptions of undue influence, when they arose, implicated almost all of the participants, to some degree or another; the data do not suggest that reg neg facilitates agency capture. Nor is there any indication that the agency abdicated its role.

Advocates of reg neg will no doubt find the Phase I results encouraging. The degree to which participants embraced the process is striking, as are their evaluations of its outcomes. The internal disparities uncovered by the data, moreover, appear not to be solely a product of the reg neg process: information and resource asymmetries likely disable the same groups in all forms of governmental decisionmaking. Still, as noted above, these data tell us nothing about the relative performance of negotiated rulemaking as compared to conventional rulemaking. For this we turn to Phase II.

C. Kerwin and Langbein: Phase II

In the second phase of the study, Kerwin and Langbein sought to test a number of the hypotheses concerning the comparison between reg neg and conventional rulemaking that had emerged, largely without empirical support, in the scholarly literature. Specifically, they examined whether, compared to conventional rulemaking, negotiated rulemaking reduces conflict, is fairer to regulated parties, and abrogates the agency's responsibility to execute its delegated authority. They also investigated whether complex rules are more likely to be settled through reg neg, and whether reg neg is more costly than conventional rulemaking.

The study consists of a sample of six conventional rulemakings conducted by EPA over roughly the same period as the negotiated rules examined in Phase I. Kerwin and Langbein matched the negotiated and conventional rules in terms of a number of criteria: when they were developed, their authorizing statutes, the substantive issues involved, their complexity, the depth of disagreement, and the scope and scale of their impact on the affected parties.²¹⁴ The interviewers asked participants the same series of open-ended questions as in Phase I, and requested that they rank the rule on the same 11-point scale to provide overall evaluations of the process and the final rule.

Notably, the survey did not directly ask participants to compare reg neg to conventional rulemaking, either generally or on a given dimension.²¹⁵ Rather, interviewers asked [31 ELR 10828] respondents a series of questions about a rule—conventional or negotiated—in which they had participated.²¹⁶ Thus, the findings are not based on the respondents' subjective perceptions of differences between conventional and negotiated rulemakings; instead, findings of any differences emerge from the data analysis, based on differences between the two sets of responses to the same interview instrument.²¹⁷

First, Kerwin and Langbein sought to test whether reg neg reduces or increases conflict between the regulator and the regulated entities, compared to conventional rulemaking. As noted earlier, reg neg proponents argue that direct participation in the creation of the rule will reduce conflict and increase satisfaction with the final rule.²¹⁸ In this view, the consensus-based approach enhances the parties' commitment to the final rule and creates a long-term working relationship among different parties, reducing the likelihood of litigation and noncompliance.²¹⁹ Critics argue, by contrast, that reg neg will increase conflict by increasing the amount of information available to each party, thereby introducing additional, potentially divisive issues not raised in conventional rulemaking.²²⁰

To test these expectations, Kerwin and Langbein examined whether parties learn more in reg negs,²²¹ whether they perceive that more parties are excluded from reg negs than from the conventional process,²²² and whether the participants view reg neg as more consensual than conventional rulemaking.²²³ Assuming that greater satisfaction is indicative of less conflict, they also explored whether participants express more satisfaction with substantive outcomes of reg negs as compared to conventional rules.²²⁴ Next, Kerwin and Langbein investigated the variance (or heterogeneity) in satisfaction with negotiated as compared to conventional rules, assuming that less variance, i.e., more homogeneity, provides another indication of less conflict.²²⁵

The study also tests whether reg neg is fairer to regulated parties than conventional rulemaking, as some scholars suggest.²²⁶ In this view, reg neg levels the playing field among participants by forcing information production and sharing; it allows everyone a greater and more direct voice in rule writing.²²⁷ By contrast, critics claim that the information-intensive nature of the process disadvantages smaller, poorer parties relative to conventional rulemaking.²²⁸ To test these claims, Kerwin and Langbein examined participant perceptions of who is left out of reg negs compared to conventional rulemaking, and the extent to which participants learn in the two processes. To gather data on information asymmetries, they studied disparities between conventional and negotiated rules with regard to the amount of information each makes available and the extent to which participants report learning in each.

Kerwin and Langbein also sought to shed light on the dispute over reg neg's compatibility with democratic ideals. As noted earlier, some scholars claim that reg neg abrogates an agency's responsibility for implementing legislation,²²⁹ while others maintain that reg neg does nothing to undermine agency accountability because rules must still comply with the APA.²³⁰ Indeed, some political scientists have gone still further, contending that consultation with stakeholders through negotiations specifically fulfills congressional intent.²³¹ According to this view, Congress delegates the details of legislation to the agency when members are uncertain about how to achieve their goals; the agency merely behaves as Congress intends when, under conditions of uncertainty, it in turn consults the most affected parties about the details of a regulation.²³² Even conventional rulemaking involves considerable negotiation with affected parties, which Congress also intends and approves²³³; by formalizing these negotiations in the form of reg neg, Congress simply enables itself to better monitor agency consultation with stakeholders. Thus, in this view, agencies are always responsive to affected parties because failure to listen prompts congressional wrath.²³⁴

To test these assertions, Kerwin and Langbein hypothesized that if negotiating rules does abrogate an agency's duty to perform its congressionally delegated functions, parties would feel that they exert greater impact in a reg neg than in conventional rulemaking.²³⁵ However, if reg neg is actually consistent with an agency's responsibility (on the theory that agencies always listen to the affected parties as Congress intends), then, they hypothesized, participants should view their involvement in reg neg as no more or less influential than their participation in conventional rulemaking.²³⁶

The fourth hypothesis tested by Kerwin and Langbein grows out of debate over congressional intent: if uncertainty and complexity prompt agencies to shift rulemaking responsibility to regulated entities through negotiated [31 ELR 10829] rulemaking—that is, if the "agencies always listen to affected parties" hypothesis is correct—then negotiated rules are likely to differ from conventional rules in systematic ways. For example, rules ought to be especially good candidates for negotiation when the agency is particularly uncertain about their outcome or when the issue is especially complex, as when they involve numerous stakeholders or sub-is-sues about which affected parties are likely to be more knowledgeable than the agency.²³⁷ The counter-argument is that under certain circumstances, single issues have a settlement equilibrium at the median, which implies that uni-dimensional issues settle more easily than complex ones.²³⁸

To test these expectations, Kerwin and Langbein used several measures to examine whether the kinds of rules selected for reg neg were different from those developed through conventional rulemaking. First, they investigated whether negotiated and conventional rulemaking were applied to different types of issues; next, they explored whether respondents believed that negotiated rules or conventional rules settle all the issues to be decided; finally, they developed a measure of complexity and clarity for each rulemaking, and examined whether these measures differed between the two types of rulemakings.²³⁹

The final hypothesis tested by Kerwin and Langbein concerns the cost differentials between negotiated and conventional rulemaking.²⁴⁰ They reasoned that if negotiated rules are politically or technically more complicated than conventional rules, it follows that they will be more expensive to produce.²⁴¹ Further, if agencies shift much of the work of rule-writing to the affected parties, it follows that the time, monetary, and informational costs of participation by those parties should be higher in negotiated than conventional rulemaking.²⁴² This is one claim about reg neg that has no counterclaim. The key findings of Kerwin and Langbein's study, plus our interpretations, appear below.²⁴³

1. Negotiated Rulemaking Reduces Conflict

a. Satisfaction and Conflict

If conflict breeds dissatisfaction, the results of Phase II are consistent with the contention that reg neg reduces conflict. Kerwin and Langbein found that participants in regulatory negotiation reported higher satisfaction than their conventional rulemaking counterparts²⁴⁴; in addition, the reg neg participants' ratings of the efficiency, cost-effectiveness, and the overall process were significantly higher than ratings for conventional rules.²⁴⁵ In our view, these data suggest that negotiated rulemaking confers a legitimacy benefit. The higher satisfaction result was detected evenwhen a variety of substance and process variables were held constant. That is, the researchers found that participants in negotiated rulemakings expressed greater satisfaction with the final rule than participants in conventional rulemakings, independently of differences between the types of rules chosen for conventional and negotiated rulemaking, independently of divergent views of the economic net benefits of the particular rule, and independently of their affiliation.²⁴⁶

If conflict breeds disagreement, these results continue to accord with the hypothesis that negotiation reduces conflict. The reported standard deviation (a measure of heterogeneity) of the ratings of conventional rules was generally higher than the standard deviation of ratings for negotiated rules, consistent with the hypothesis that negotiation reduces conflict among parties, resulting in more homogeneous, consensual views of the eventual outcome.²⁴⁷

The open-ended interviews also produced evidence of higher satisfaction among participants in negotiated rulemaking. When asked what they liked and disliked about the process, reg neg participants reported significantly more "likes" than their conventional rulemaking counterparts.²⁴⁸ The evidence indirectly reveals somewhat greater satisfaction among negotiated rule participants. Again, if conflict breeds dissatisfaction, then these results are consistent with the view that negotiation engenders less conflict.

b. Learning and Conflict

Overall, Kerwin and Langbein's evidence upholds the expectation that, compared to conventional rulemaking, which is hierarchically directed by EPA, "participants learn more in negotiated processes and … negotiated processes are more horizontal in their sources of information."²⁴⁹ Nearly 20% of conventional rule participants reported learning nothing new, but not a single reg neg participant offered that response.²⁵⁰ Negotiated rulemaking participants responded more often that they gained new technical information, better knowledge of the issue, and new information about the positions of other parties; overall, 62% of negotiated rulemaking participants offered that they had learned these sorts of new information, compared to only 17% of conventional rule participants.²⁵¹

Moreover, researchers found that, when asked what they liked about the process, negotiated rule participants more often mentioned something that they learned; indeed, 42% of these respondents volunteered that they liked learning about something during the process, be it the positions of others (15%), the process itself (10%), the substantive issue (9%), or EPA or information for use in the future (8%).²⁵² By contrast, just 13% of conventional rule participants mentioned learning something as an example of what they liked about the process, and not a single conventional participant [31 ELR 10830] reported "learning about the positions of others" among their "likes."²⁵³

Still, the relationship between greater learning and conflict remains opaque. Recall that theorists disagree over whether more learning and information sharing among the parties engenders more or less conflict. On the one hand, learning can reduce conflict by increasing mutual understanding²⁵⁴; on the other it can exacerbate conflict by multiplying the issues over which parties might disagree.²⁵⁵ By using "satisfaction" (calculated using a number of measures) as a proxy for conflict, Kerwin and Langbein found no positive or negative relationship between learning and conflict in rulemaking.²⁵⁶ Thus, learning appears to offer no clear instrumental value for conflict reduction. Still, learning may have inherent value, and in this dimension, negotiated rulemaking was found to be clearly superior to conventional rulemaking.

c. Consensus Decision Rules and Conflict

Consensus decision rules used in reg neg are thought to engender two different effects: while they raise conflict and cost during the decisionmaking process itself,²⁵⁷ they increase satisfaction once the parties reach agreement. This view presupposes that conventional rulemaking involves no informal consensual decisionmaking, a presupposition contradicted by the study. In fact, researchers found that conventional rulemaking participants reported informal contact with both EPA and other parties. One-quarter of conventional rule participants reported that they engaged in informal negotiations.²⁵⁸

Despite this evidence of informal contact, the data suggest that negotiated rulemaking achieved a higher level of consensus among participants. When asked what constituted a consensus, 90% of reg neg participants responded either "unanimity" or "what we could all live with,"²⁵⁹ both consistent with a consensual process. By contrast, 45% of conventional participants responded "what EPA wanted"; no reg neg respondents defined consensus in this manner.²⁶⁰ "What EPA wanted" does not describe a consensual process.

When the more consensual reg neg process was used, respondents reported greater satisfaction both with the process and with the net benefits of the final rule to their organization.²⁶¹ Moreover, the standard deviation of judgments was smaller under reg neg.²⁶² These results support the theory that relatively more consensual decision rules lead to greater satisfaction with outcomes, greater homogeneity in judgments about those outcomes, and less conflict.

In sum, more consensual processes yielded significantly higher net benefit ratings and possibly more agreement. Kerwin and Langbein asked reg neg participants what constituted consensus in their formal negotiation sessions, expecting that more consensual decision rules would be associated with greater satisfaction, higher ratings of organizational net benefits, and less conflict, i.e., more homogeneity, about those judgments. The results were consistent with these expectations: ratings of the overall process were lowest and the standard deviations were usually highest when the decision rule was "what EPA wanted."²⁶³

Overall, then, the study supports the claim that negotiated rulemaking is more consensual than conventional rulemaking. Further, if litigation measures conflict, then reg neg seems to perform as well (or as poorly) as conventional rulemaking. Litigation rates for both kinds of rules, according to Kerwin and Langbein, were about the same.²⁶⁴

2. Negotiated Rulemaking Was Fairer to Regulated Parties Than Conventional Rulemaking

To test whether reg neg was fairer to regulated parties, Kerwin and Langbein asked respondents whether EPA solicited their participation and whether they believed anyone was left out of the process. They also examined as well how much the parties learned in each process, and whether they experienced resource or information disparities.

The researchers found that negotiated rule participants were significantly more likely to say that EPA encouraged their participation than conventional rule participants (65% versus 33% respectively).²⁶⁵ Although a higher proportion of conventional rulemaking participants reported that a party that should have been represented in the rulemaking was omitted, the difference is not statistically significant.²⁶⁶ Specifically, the researchers report that "a majority of both negotiated and conventional rule participants believed that the parties who should have been involved were involved (66% versus 52% respectively)."²⁶⁷

In addition, as reported above, participants in regulatory negotiations reported significantly more learning than their conventional rulemaking counterparts.²⁶⁸ Indeed, the disparity between the two types of participants in terms of their reports about learning was one of the study's most striking results. At the same time, the resource disadvantage of poorer, smaller groups was no greater in negotiated rulemaking than in conventional rulemaking.²⁶⁹ So, while smaller groups did report suffering from a lack of resources during regulatory negotiation, they reported the same in conventional rulemakings; no disparity existed between the two processes on this score.²⁷⁰

[31 ELR 10831]

Finally, Kerwin and Langbein's data suggest that the agency is equally responsive to the parties in both negotiated and conventional rulemakings.²⁷¹ This result, together with the finding that participants in regulatory negotiations perceived disproportionate influence to be about evenly distributed,²⁷² suggests that reg neg is at least as fair to the parties as conventional rulemaking. Indeed, because participant learning was so much greater in regulatory negotiation, the process may in fact be more fair.

3. Negotiated Rulemaking Did Not Abrogate the Agency's Responsibility to Execute Delegated Authority

Overall, the evidence from Phase II is generally inconsistent with the theoretical but empirically untested claim that EPA has failed to retain its responsibility for writing rules in negotiated settings. Recall that theorists disagree over whether reg neg will increase agency responsiveness.²⁷³ Most scholars assume that EPA retains more authority in conventional rulemaking, and that participants exert commensurately less influence over conventional as opposed to negotiated rules.²⁷⁴ To test this hypothesis, Kerwin and Langbein asked participants about disproportionate influence and about agency responsiveness to the respondent personally, as well as agency responsiveness to the public in general.²⁷⁵

The results suggest that the agency is equally responsive to participants in conventional and negotiated rulemaking, consistent with the hypothesis that the agency listens to the affected parties regardless of the method of rule development.²⁷⁶ Further, when asked what they disliked about the process, less than 10% of both negotiated and conventional participants volunteered "disproportionate influence."²⁷⁷

When asked whether any party had disproportionate influence during rule development, 44% of conventional respondents answered "yes," compared to 48% of reg neg respondents.²⁷⁸ In addition, EPA was as likely to be viewed as having disproportionate influence in negotiated as conventional rules (25% versus 32% respectively).²⁷⁹ It follows that roughly equal proportions of participants in negotiated and conventional rules viewed other participants, and especially EPA, as having disproportionate influence.

Kerwin and Langbein asked those who reported disproportionate influence what about the rule led them to believe that lopsided influence existed. In response, negotiated rulemaking participants were significantly more likely to see excessive influence by one party in the process rather than in the rule itself, as compared to conventional participants (55% versus 13% respectively).²⁸⁰ However, when asked what it was about the process that fostered disproportionate influence, conventional rule participants were twice as likely as negotiated rule participants to point to the central role of EPA (63% versus 30% respectively).²⁸¹ By contrast, negotiated rule participants pointed to other participants who were particularly vocal and active during the negotiation sessions (26% of negotiated rule respondents versus no conventional respondents).²⁸²

When asked about agency responsiveness, negotiated rule participants were significantly more likely than conventional rule participants to view both general participation, and their personal participation, as having a "major" impact on the proposed rule.²⁸³ By contrast, conventional participants were more likely to see "major" differences between the proposed and final rule and to believe that public participation and their own participation had a "moderate" or "major" impact on that change.²⁸⁴ These results conform to the researchers' expectations: negotiated rules are designed so that public participation should have its greatest impact on the proposed rule; conventional rules are structured so that public participation should have its greatest impact on the final rule.

Given these differences in how the two processes are designed, Kerwin and Langbein sought to measure agency responsiveness overall, rather than at the two separate moments of access.²⁸⁵ Although the differences were not statistically significant, the results suggested that conventional participants perceived their public and personal contribution torulemaking to have had slightly more impact than negotiated rule participants perceived their contribution to have had.²⁸⁶ Still, given the absence of statistical significance, we agree with the researchers that it is safer to conclude that the agency is equally responsive to both conventional and negotiated rule participants.

[31 ELR 10832]

4. Complex Rules Were More Likely to Be Settled Through Negotiated Rulemaking

Recall that theorists disagree over whether complex or simple issues are best suited for negotiation.²⁸⁷ The data suggest that negotiated and conventional rules differ in systematic ways, indicating that EPA officials do not select just any rule for negotiation. When asked how the issues for rulemaking were established, reg neg participants reported more often than their counterparts that the participants established at least some of them (44% versus 0%).²⁸⁸ Conventional rulemaking participants more often admitted to being uninformed of the process for establishing issues (17% versus 0%) or offered that regulated entities set the issues (11% to 0%).²⁸⁹ A majority of both groups reported that EPA or the governing legislation established at least some of the issues.²⁹⁰

Kerwin and Langbein found that the types of issues indeed appeared to differ between negotiated and conventional rules.²⁹¹ When asked about the type of issues to be decided, 52% of participants in conventional groups identified issues regarding the standard, including its level, timing, or measurement (compared to 31% of negotiated rule participants),²⁹² while 58% of the negotiating group identified compliance and implementation issues (compared to 39% of participants in the conventional group).²⁹³ More reg neg participants (53%) also cited compliance issues as causing the greatest conflict, compared to 32% of conventional participants.²⁹⁴ Conventional participants more often reported that the rulemaking failed to resolve all of the issues (30% versus 14%), but also more often reported that they encountered no "surprise" issues (74% versus 44%).²⁹⁵

The study found that participants perceived negotiated rules to be more complex, with more issues and more sides per issue than conventional rules.²⁹⁶ Kerwin and Langbein learned in interviews that reg neg participants tended to develop a more detailed view about the issues to be decided than did their conventional counterparts.²⁹⁷ The researchers interpreted this disparity in reported detail as a perception of complexity.²⁹⁸ To measure it they computed a complexity score: the more issues and the more sides to each issue that respondents in a rulemaking could identify, relative to the number of respondents, the more nuanced or complex the rulemaking.²⁹⁹ Using this calculation, the rules ranged in complexity from 1.9 to 5.0, with a mean complexity score of 3.6.³⁰⁰ The mean complexity score for reg negs (4.1) was significantly higher than the score (2.5) for conventional rulemaking.³⁰¹

Reg neg participants also presented a clearer understanding of the issues to be decided than did conventional participants.³⁰² To test clarity, Kerwin and Langbein measured the number of different issues and sides that respondents perceived in the rulemaking.³⁰³ Some respondents could identify very few separate issues and sides, e.g., the level of the standard is the single issue and the sides are business, environmentalists, and EPA, while others detected as many as four different issues, with three sides on some and two on others.³⁰⁴ Kerwin and Langbein's measurement was in units of issue/sides, representing a combination of the two variables, the recognition of which they were measuring; the mentions ranged from 3 to 10 issue/sides, with a mean of 7.9.³⁰⁵ Negotiated rulemaking participants mentioned an average of 8.9 issue/sides, compared to an average of 6 issue/sides mentioned by their conventional counterparts, a statistically significant difference.³⁰⁶

To illustrate the difference between complexity and clarity: if a party identified the compliance standard as the sole issue, but failed to identify a number of sub-issues, they would be classified as having a clear understanding but not a complex one. Similarly, if the party identified two sides (business versus environment) without recognizing distinctions among business participants or within an environmental coalition, they would also be classified as clear but not complex in their understanding. So, while clarity focuses on comprehension of differences, complexity focuses on the perception of numbers/sides.

The differences in complexity might be explained by the higher reported rates of learning by reg neg participants, rather than by differences in the types of rules processed by reg neg versus conventional rulemaking.³⁰⁷ Kerwin and Langbein found that complexity and clarity were both positively and significantly correlated with learning by respondents, but the association between learning and complexity/clarity disappeared when the type of rulemaking was held constant.³⁰⁸ However, when the amount learned was held constant, the association between complexity/clarity and the type of rulemaking remained positive and significant.³⁰⁹ This signifies that the association between learning and complexity/clarity was due to the negotiation process. In other words, the differences in complexity/clarity are not attributable to higher learning but rather to differences between the processes. The researchers conclude that the evidence is consistent with the hypothesis that issues selected for regulatory negotiation are different from and more complicated than those chosen for conventional rulemaking.³¹⁰ [31 ELR 10833] The data associating reg negs with complexity, together with the researchers' claims that more issues settle in reg negs,³¹¹ are consistent with the proposition that issues with more (and more diverse) sub-issues and sides settle more easily than simple issues.³¹²

5. Negotiated Rulemaking Is No More Costly Than Conventional Rulemaking

As in Phase I, Kerwin and Langbein asked respondents a series of questions about cost issues in Phase II. Specifically, they asked about respondents' expenditures, in terms of professional hours and clerical hours, and amounts spent for information collection, legal counsel, and consultants.³¹³ They also asked about the proportion of available resources that this spending represented and the subjective perception of whether the benefits outweighed the costs.³¹⁴

Reg neg participants reported spending significantly more professional staff hours than conventional participants; the magnitude of this difference is over 2,000 staff hours or about 1 person-year.³¹⁵ However, this difference was mostly attributable to the fact that EPA spent significantly more time than other participants on regulatory negotiation than on conventional rulemaking.³¹⁶ Recall that the agency is not a "participant" in conventional rulemaking, so there were no EPA respondents among conventional rule participants.³¹⁷ After removing EPA, there were no significant differences between negotiated and conventional rules participants in money spent for staff hours, clerical hours, or monetary resources for research, information, legal counsel and consultants.³¹⁸ However, reg neg participants did report spending nearly twice as much as conventional participants in terms of resources relative to those available.³¹⁹

Further, in response to questions about what they liked and disliked, 30% of dislikes volunteered by negotiation participants cited some aspect of cost, measured in time, money, or personal aggravation, compared to 9% of conventional rulemaking participants,³²⁰ whose dislikes centered on EPA or the quality of the rule. Still, nearly 78% of all participants, in both negotiated and conventional rulemaking, believed that the benefits of participation equaled or outweighed the costs.³²¹ According to Kerwin and Langbein, even if the costs of participation in reg negs are higher, the ratio of benefits to costs appears no different.³²² On the average, participation in either rulemaking appears to be, ex post, a "rational" decision.³²³

III. Implications

A. Reg Neg's Overall Performance

On balance, the combined results of Phase I and II of the study suggest that reg neg is superior to conventional rulemaking on all of the measures that were considered. Strikingly, the process engendered a significant learning effect, especially compared to conventional rulemaking; participants reported, moreover, that this learning had long-term value not confined to a particular rulemaking. Most significantly, the negotiation of rules appears to enhance the legitimacy of outcomes. Kerwin and Langbein's data indicate that process matters to perceptions of legitimacy.³²⁴ Moreover, as we have seen, reg neg participant reports of higher satisfaction could not be explained by their assessments of the outcome alone. Instead, higher satisfaction seems to arise in part from a combination of process and substance variables. This suggests a link between procedure and satisfaction, which is consistent with the mounting evidence in social psychology that "satisfaction is one of the principal consequences of procedural fairness."³²⁵ This potential for procedure to enhance satisfaction may prove especially salutary precisely when participants do not favor outcomes. As Tom Tyler and E. Allan Lind have suggested, "hedonic glee" over positive outcomes may "obliterate"procedural effects; perceptions of procedural fairness may matter more, however, "when outcomes are negative [and] organizations have the greatest need to render decisions more palatable, to blunt discontent, and to give losers reasons to stay committed to the organization."³²⁶

At a minimum, the data call into question—and sometimes flatly contradict—most of the theoretical criticisms of reg neg that have surfaced in the scholarly literature over the last 20 years. There is no evidence that negotiated rulemaking abrogates an agency's responsibility to implement legislation. Nor does it appear to exacerbate power imbalances or increase the risk of capture. When asked whether any party seemed to have disproportionate influence during the development of the rule, about the same proportion of reg neg and conventional participants said yes.³²⁷ Parties perceived their influence to be about the same for conventional and negotiated rules, undermining the hypothesis that reg neg exacerbates capture.

In addition to the NRA's formal requirement that negotiated rules conform to the demands of notice-and-comment rulemaking, the data indicated that agencies played a powerful role during negotiations. Respondents frequently cited the agency as the source of both procedural rules and the substantive issues to be determined, suggesting that the agency exerted considerable authority during negotiations. Respondents also perceived EPA (along with business) as most likely to exert disproportionate influence,³²⁸ and they [31 ELR 10834] viewed the agency as equally responsive to participants of reg neg and conventional rulemaking.³²⁹ In addition (and although it does so at the risk of undermining stakeholder confidence in the process), the agency abandons negotiated rulemaking and departs from the negotiated consensus sufficiently regularly to suggest that the agency remains independent of stakeholder influence, and is willing to signal to that effect.³³⁰

The data also suggest that negotiated rulemaking is more flexible than conventional rulemaking,³³¹ and that it leaves room for more surprise.³³² In terms of how the parties share information, how issues surface, and how solutions emerge, negotiated rulemaking proves more "horizontal" than conventional rulemaking, which remains comparatively hierarchical.³³³ And finally, the process is knowledge and information rich.³³⁴ Even if these characteristics provided no instrumental value in terms of higher quality or more legitimate rules, their inherent value strikes us as significant.

At the same time, regulatory negotiation falls short of an ideal process. Participants perceive it to be more resource-and information-intensive than conventional rulemaking, and smaller, poorer participants disproportionately bear these costs.³³⁵ These same groups also suffer from information asymmetries.³³⁶ The data revealed a number of inequities that could, if not mitigated, harm smaller interests with fewer resources. Recall, for example, that smaller, poorer groups found the decision to participate in reg neg more difficult.³³⁷ During interviews, Kerwin and Langbein detected a fear among smaller, poorer participants that refusal to participate might seriously damage the group's interests. That is, even if they might have preferred a conventional rulemaking because of the lower time and effort required, these groups felt pressured to join the reg neg. Thus, unless EPA provides supplemental assistance to mitigate the resource drain on these groups, they might feel somewhat coerced into participation.

Unequal access to information continues to be an ongoing problem in regulatory negotiation. Although the intense education and learning that occurs during negotiations can help to mitigate information asymmetries, the disparities may still be substantial enough to give disproportionate agenda-setting power to groups with greater resources. Information asymmetries seem particularly problematic in this context, given its information-intensive nature.

Finally, the data indicate that parties continue to devote considerable time and resources to the rule after the negotiating committee disbands during the post-proposal phase.³³⁸ Indeed, as noted earlier, post-proposal activity generated changes in the proposed rules, most of which were minor, but some of which were substantial.³³⁹ At the post-comment stage, the adversarial patterns thought to characterize conventional rulemaking may emerge among interest groups, and richer, larger interests may find themselves in a superior position, due to their ability to monitor, communicate with, and influence the agency.³⁴⁰ But it is not clear that reg neg exacerbates this disparity.³⁴¹

In interviews, environmental groups and small business consistently registered the least enthusiasm of all when asked about a number of qualitative dimensions of reg neg. One might dismiss or minimize this result because, when Kerwin and Langbein statistically controlled for other variables, environmental groups (and possibly small business) rated their overall experience no lower than did EPA.³⁴² In addition, it turns out that smaller groups (including both environmental and small business interests) dislike negotiated and conventional rulemaking equally. Still, we believe the consistent disparity in enthusiasm bears mention and further study. Finally, reg negs may still be insufficiently inclusive even if participants fail to perceive representation gaps. The committees consistently lack a representative of the general consumer or median voter—unless one counts the agency as playing this role.

B. Explaining the Legitimacy Benefit

Although reg neg participants reported higher satisfaction with the overall process than their conventional counterparts, many of the reasons for that satisfaction remain unclear. The parties might have been more satisfied because of the kinds of issues that tend to arise in regulatory negotiation, because they felt negotiation improves rule quality, or because they valued particular features of the process, such as the opportunity to engage each other face to face. Higher satisfaction might have been due to a factor that Kerwin and Langbein did not test.

To more closely identify the reasons for satisfaction (and dissatisfaction), Kerwin and Langbein used a multiple regression analysis to isolate individual factors by holding others constant. Their results show that, overall, participants preferred reg neg to conventional rulemaking partly because they got a better rule out of the process, and partly because some aspects of the process, but not all, worked well—that is, a combination of substance and process variables.³⁴³ Three substantive variables exerted a positive effect on satisfaction ratings, when other variables were held constant. When respondents rated either the net benefits of the rule to their organization or the "economic efficiency" of the rule to society higher, their overall evaluation of the [31 ELR 10835] rule-writing experience went up.³⁴⁴ In addition, when one of the primary issues to be decided concerned the standard (level, measurement, or timing), participant evaluations of the process were consistently and significantly higher (by 1 point on an 11-point scale) than was detected when the primary issue concerned compliance or implementation.³⁴⁵

Two process variables exerted an impact on overall evaluations, when the other variables were held constant. The perceived complexity or "messiness" of the rule contributed to lower evaluations,³⁴⁶ while clarity (participant understanding of the issues) contributed to higher satisfaction.³⁴⁷ Recall that while complexity measures the number of issue/sides identified per participant in rulemaking, clarity measures the number of different issues or sides that respondents could identify. Specifically, each additional side or issue per respondent (more messiness) reduces the overall evaluation by 2 points on an 11-point scale, while the ability to see more different issues and sides had a consistently positive, but not particularly large, effect.³⁴⁸ However, although perceived complexity contributes to lower evaluations, and even though reg negs are perceived as entailing greater complexity, reg neg participants continued to give higher ratings to the overall process, even after numerous statistical controls, including a control for complexity.³⁴⁹

We do not claim that acceptability to participants renders either the reg neg process or the outcomes of it, in some objective sense, fully or wholly legitimate. It is possible that some form, or forms, of cognitive consistency processes are really at play, and that they explain why participants evaluate reg neg so positively. Were this the case, the legitimacy benefit might be limited to those closely connected to the particular reg neg process itself. Any claim to a legitimacy benefit under these conditions would rely on a less expansive understanding of legitimacy than we frequently encounter in administrative law and political theory.³⁵⁰

Certainly a process cannot be called legitimate if outsiders think insiders achieve gains at their expense—if they capture the agency, for example—or if the results favor only a minority of interests. However, at a minimum, reg neg should fare no worse when assessed on legitimacy grounds than conventional rulemaking, at least according to positive political theory. That is, if politicians get re-elected by faithfully representing "outsiders" and unorganized interests, and if agencies garner support by being generally accountable within the bounds set by their politically elected principals, then neither rulemaking process should be more subject to capture than the other.³⁵¹

Still, we recognize the possibility that outsiders to any process might view as illegitimate an outcome that insiders find satisfying. However, no empirical study we know of has tried to measure this potential insider/outsider split over legitimacy in the context of either conventional rulemaking or regulatory negotiation.³⁵² It is difficult to know what the "general public" or median voter thinks of the regulatory process, given that most complaints about its legitimacy or illegitimacy come from either insiders, like lawyers, regulators, and politicians, or from academics, who represent only a narrow class of outsiders.³⁵³ As noted above, a lack of participation by outsiders does not necessarily signal that dissatisfaction, nor does it indicate that they view the process as illegitimate. We note that nonparticipants in the conventional rulemaking process may choose not to participate out of satisfaction that their concerns have already been considered. So too with nonparticipants in the reg negs, whose interests may be represented by other parties.

Although Kerwin and Langbein did not directly investigate the potential insider/outsider split over legitimacy, no data emerged that might fuel outsider suspicion about reg neg versus conventional rulemaking. For example, they found no evidence of skewed outcomes or results that seemed, on their face, to undermine the legitimacy of the reg neg process. Were the reg neg process exclusive to a few repeat players and disproportionately beneficial to them, we might worry more about the possibility that insider satisfaction could in fact undermine legitimacy rather than further it. However, the data suggest that the reg neg process is broadly inclusive—more so, specifically, than conventional rulemaking. There is no reason to believe that exclusive access and increased influence were the reasons for higher satisfaction rates, which might lead us to infer that "insider" legitimacy was procured at the expense of "outsider" legitimacy.

C. Litigation Rates

Kerwin and Langbein's cursory inquiry into litigation rates suggests that conventional and negotiated rules attract a similar amount of litigation,³⁵⁴ but because they did not reproduce Coglianese's methodology, their results cannot resolve the debate between him and Harter. Still, we feel compelled to comment on the relevance of litigation rates, in light of the data presented here.

Because Coglianese claimed thatsaved time and lower litigation rates were the most prominent of reg neg's purported benefits, he concluded that a failure to outperform conventional rulemaking on these scores amounted to a failure to deliver on reg neg's promise.³⁵⁵ In our view, however, [31 ELR 10836] focusing on these two measures alone proves reductive and potentially misleading. For us, time and litigation rates tell only part of the story and although relevant, they remain secondary to improved rule quality and legitimacy.³⁵⁶ Even if negotiated rules do take longer to promulgate, other benefits might offset the cost of additional time. Interestingly, even though participants reported spending more as a proportion of available resources on reg negs, and despite the fact that many respondents cited cost as one of their dislikes, the overwhelming majority still found that the benefits of participation outweighed the costs.

Even if saving time were a critical goal, the question of how to measure time remains. Coglianese compared conventional and negotiated rules in terms of the amount of time elapsed between an NPRM and a final rule,³⁵⁷ but the NPRM is not necessarily an accurate measure of when a rulemaking begins. In fact, frequently, interested parties begin participating informally in notice-and-comment rulemaking prior to the official publication of the NPRM. Even if the NPRM is a convenient start date (after all, one must choose something to mark the beginning of rulemaking), measuring the time elapsed between the NPRM and promulgation seems less important to an evaluation of whether reg neg saves time than determining whether the implementation of negotiated rules occurs in a more timely or effective manner than conventional rules.

Moreover, comparing litigation rates for negotiated and conventional rules, especially without statistical or other controls to ensure that the rules being compared are truly comparable, involves considerable guesswork about the appropriate baseline. Perhaps parties that challenge negotiated rules would have sued regardless of the process used for rule development because of the issue's importance—in such a case, other potential benefits of negotiated rulemaking might still warrant its use.

Perhaps a greater percentage of negotiated rules lead to lawsuits simply because the rules chosen for negotiation are typically more contentious than other rules or because the agency lacks the necessary expertise to devise a particularly complex rule, and is aware of it. That is, in the absence of negotiated rulemaking, even more litigation might have ensued—which, if true, suggests that reg neg may decrease litigation rates, only rather invisibly, in the absence of statistical or other controls.

At the same time, challenges to negotiated rules might be qualitatively different from challenges to conventional rules in important respects: they may involve fewer issues, or different ones, for example, or settle more easily. Of course, when assessing reg neg's purported benefits, one might not care to know the reason that litigation ensues, only that it does. However, as a measure of reg neg's success or failure, this is too simplistic. Without knowing the contextual details of the litigation, drawing conclusions from litigation rates alone can be misleading.³⁵⁸ It is also worth remembering that a legal challenge offers neither the only, nor necessarily the best, evidence of dissatisfaction with regulation. Regulated entities can resist regulation (whether imposed via conventionally promulgated rules, negotiated rules, or even through informal policy instruments) merely through delay, incomplete compliance, or noncompliance.³⁵⁹

Litigation rates alone thus reveal only a limited amount of information about the promise of reg neg. The reason for a legal challenge seems equally as important as the mere fact of challenge itself. The lawsuit might signal dissatisfaction with the consensus reached and therefore reflect a "failure" of the reg neg process, but it might also signal concerns that have nothing to do with reg neg itself. Indeed, the lawsuit might actually involve a dimension of the promulgated rule produced through conventional notice-and-comment rulemaking, as with the litigation over the equipment leaks rule.³⁶⁰ Alternatively, as Harter suggests, the challenge might be aimed at the agency's departure from the negotiated consensus, signaling dissatisfaction not with reg neg but with subsequent amendments to the rule in which the party had little voice.

In addition, the identity of the challenger matters. It makes a difference, in our view, whether the party challenging the rule participated in the reg neg committee or not. When a former committee member files a lawsuit, the challenge might indeed suggest a weakness in the process. Perhaps the party opposes the rule because it lacked sufficient resources to participate meaningfully in negotiations, or because its views were so extreme that other parties failed to incorporate them. We might infer something different, however, if the challenging party was never included in the negotiating committee in the first place. Perhaps committees should be larger, or more diverse, in order to reduce litigation further, or maybe they should not disband upon reaching consensus, but remain intact until promulgation. The point is, the reasons for the different challenges, and the identity of the challengers, provide useful information about the strengths and weaknesses of a consensus-based approach, and how it might be improved. Numbers alone, especially without statistical controls for measurable differences, cannot distinguish easily curable defects from fatal flaws.

D. Compliance

The compliance implications of consensus-based processes remain a matter of speculation.³⁶¹ No one has yet produced empirical data on the relationship between negotiated rulemaking and compliance, let alone data comparing the compliance implications of negotiated and conventional rules.³⁶² However, social psychology studies show that in [31 ELR 10837] both legal and organizational settings, "fair procedures lead to greater compliance with the rules and decisions with which they are associated."³⁶³ However, the Phase II results introduced interesting new findings into the debate. The data showed reg neg participants to be significantly more likely than conventional rulemaking participants to report the perception that others will be able to comply with the final rule.³⁶⁴ Perceiving that others will comply might induce more compliance among competitors, along the lines of game theoretic models, at least until evidence of defection emerges.³⁶⁵

Moreover, to the extent that compliance failures are at least partly due to technical and information deficits—rather than to mere political resistance—it seems plausible that reports of the learning effect and more horizontal sharing of information might help to improve compliance in the long run.³⁶⁶

Similarly, negotiated rulemaking might facilitate compliance by bringing to the surface some of the contentious issues earlier in the rulemaking process, where they might be solved collectively rather than dictated by the agency. Although speculative, these hypotheses seem to fit better with Kerwin and Langbein's data than do the rather negative expectations about compliance. Higher satisfaction could well translate into better long-term compliance, even if litigation rates remained the same. Consistent with our contention that process matters, we expect it to matter to compliance as well, although in the absence of empirical evidence, we cannot be sure.

In any event, empirical studies of compliance should no longer be so difficult to produce. A number of negotiated rules are now several years old, with some in the advanced stages of implementation. A study of compliance might compare numbers of enforcement actions for negotiated as compared to conventional rules, measured by notices of violation, or penalties, for example.³⁶⁷ It might investigate as well whether compliance methods differ between the two types of rules: perhaps the enforcement of negotiated rules occurs more cooperatively, or informally, than enforcement of conventional rules. Possibly, relationships struck during the negotiated rulemaking make a difference at the compliance stage.³⁶⁸ To date, the effects of how the rule is developed on eventual compliance remain a matter of speculation, even though it is ultimately an empirical issue on which both theory and empirical evidence must be brought to bear.

E. Consensus and Participation

While a search for consensus may not be a necessary precondition to increasing public participation, consensus may be necessary if one seeks public participation in the service of both increased quality and legitimacy.³⁶⁹ That is, consensus may be crucial to generating a certain quality of participation that itself produces better rules and greater satisfaction. As a general matter, of course, broad participation and inclusiveness are laudable goals; all things being equal, greater participation strikes us as more consistent with democratic values than does less participation. Still, the nature of the participation and its impact on perceptions of legitimacy seem at least as relevant as sheer numbers, and probably more so.

The data on public comments, and their relevance to public participation, require equal caution. Kerwin and Langbein's Phase I data showed that negotiated rules attracted significant public comment.³⁷⁰ At the same time, the Phase II results showed that the substance of comments regarding conventional rulemaking was more diverse and more critical.³⁷¹ Reg neg thus appears to reduce the number and range of public comments.

[31 ELR 10838]

This apparent decline in participation may actually signal greater satisfaction with the proposed rule.³⁷² Indeed, a rule that has already been vetted by the interests most knowledgeable about it—and most likely to file comments on it or sue over it—ought to attract fewer and less divergent comments. Indeed, we view the Phase I and II data as support for the proposition that reg neg reduces conflict and increases agreement, but without hampering public participation. At a minimum, however, even if we cannot infer higher satisfaction from arelative reduction in comment, nor can we infer from it that the parties are especially dissatisfied, as some critics imply.³⁷³

F. Group Dynamics

Coglianese recently argued that consensus-based processes increase conflict by encouraging participants to seek "an outcome which is tangibly better than what they would otherwise have received."³⁷⁴ This hypothesis seems superficially plausible. Regardless of the process used for rule development, sophisticated parties always seek an outcome that is tangibly better than that which they would have received under other circumstances. In every case, the parties weigh whether it serves their interests to do something or nothing, file comments or stay silent, comply or litigate, or adopt any other course of action.

Yet the idea that pursuing better results in a consensus-based process uniquely leads to unrealistic expectations that increase conflict is curious. First, as a theoretical matter, other parties could presumably counter-balance unrealistic expectations. A little experience with the reg neg process would do the same. So, at least for repeat players, we anticipate no "unrealistic" expectations. In any event, surely the solution to such a problem ought to focus on altering the misinformed expectation and not on abandoning a consensus-based approach, particularly when it provides a number of valuable benefits.

Moreover, high expectations might have some positive implications as well. They could be associated with a more intense psychological commitment to the process, for example, and a greater willingness to cooperate. Several respondents reported that the longer and harder the group worked, and the more they developed relationships with other participants, the more important success became.³⁷⁵ The relationship between expectations, commitment, and conflict remains speculative, of course, in the absence of data on the psychological effects of consensus-based versus other processes. However, in our view, it would be a mistake to infer too much about these complicated dynamics from litigation rates alone.³⁷⁶

Although other processes might be equally effective at reducing conflict,³⁷⁷ we have no data on this either way. Based on the empirical evidence we do have, however, we doubt that alternative processes, such as informal policy dialogues, are likely to produce conflict resolution as effectively as regulatory negotiation. We suspect that processes in which the agency occupies a more hierarchical relationship to stakeholders, where interaction is less direct, and where stakeholder contributions are more advisory than binding, will produce comparatively lower satisfaction and comparatively greater conflict.

IV. Recommendations

In light of the data from the Kerwin and Langbein study and our analysis of it, we recommend that agencies use regulatory negotiation more frequently and that they rely on it more often for both standard-setting and questions of implementation. We note that EPA did not initiated a single negotiated rulemaking during the Clinton Administration, and we encourage the Bush Administration to adopt a different course. To the extent that this may be due to an underestimation of its benefits, we consider this recent data particularly valuable. We also suggest as well that agencies commit greater resources to regulatory negotiation as a general matter; but in order to ease the disproportionate burden on smaller, poorer participants, we make an appeal for greater financial help and technical support for those participants in particular. If regulatory negotiation shifts some of the costs of rulemaking from the agency to the other participants, policymakers should pay special attention to the implications for these relatively disadvantaged groups, especially given the information-intensive nature of the process.

We recommend as well that agencies maintain negotiating committees through rule promulgation, rather than disbanding them upon reaching consensus. While the agency would still promulgate the rule, we believe that maintaining the committee will deter much of the bilateral post-consensus activity reported by participants, through which they sometimes seek to alter the rule by reverting to the traditional agency focus. Moreover, we suggest that agencies experiment with keeping committees intact, or recalling them periodically, for post-promulgation consultation, perhaps through the implementation period.³⁷⁸ Because some combination of regulatory negotiation's process and substance variables contributes to higher satisfaction, we believe that prolonging the dynamic that characterizes the negotiations may facilitate implementation. Practically, it would also allow the agency to draw on the group's expertise while it puts the rule into operation.³⁷⁹

We suggest further that agencies consider adding a new player to negotiating committees—a professional economist or policy analyst with skills in assessing the median consumer or voters' willingness to pay for environmental public goods. Such a policy expert might suggest options, [31 ELR 10839] like performance standards, or other market solutions, that the negotiating committee might not otherwise consider. The agency itself could provide this expertise or draw on economists or policy analysts from the Congressional Budget Office or the Office of Management and Budget. Whether via an independent expertor not, however, agencies should pay attention to the need to represent and propose policy options on behalf of diffuse interests that the participants in a regulatory negotiation might overlook or ignore.³⁸⁰ Although reg neg participants did not raise this potential representation gap as an issue (recall that participants generally perceived that all of the appropriate parties tended to be included), we flag it as an ongoing challenge.

Finally, we propose a research agenda that includes further empirical study of negotiated rulemaking compared to conventional rulemaking. We recommend, in particular, further studies comparing enforcement of and compliance with negotiated rules, to the enforcement of and compliance with comparable conventional rules, as well as studies comparing the compliance of parties who participate in regulatory negotiation to those who are affected but did not sit at the table. We also recommend conducting an experiment in which the type of rulemaking process is randomly assigned to better estimate the importance of process on outcomes. These studies ought to provide important information about the instrumental effects of consensual processes beyond the limited criterion of litigation rates.

In terms of analytic tools and conceptual frameworks, we recommend an interdisciplinary approach. Although game theory may be useful for understanding the bargaining dimension of regulatory negotiation, scholars should be careful to pay sufficient attention to behavioral economics and cognitive and social psychology as well. Studies from these fields might help to explain the relationship between the particular features of a process and participant assessments of net benefits, satisfaction, and legitimacy. A psychological perspective might also help to explain outcomes. Some data suggest that when friends negotiate, they make more concessions³⁸¹; perhaps repeated interaction with the same participants over time in the context of negotiated rulemaking similarly will increase the participants' willingness to make concessions, which perhaps will facilitate consensus and contribute to satisfaction. Here, too, as with compliance, researchers should look for downstream effects: Kerwin and Langbein's data suggest that repeated, face-to-face interaction appears to increase trust and lead to greater informality during negotiations, but whether this spills over into other fora or future interactions remains unclear.

Even anthropological data on decisionmaking in small groups might help us to better understand the dynamics of a reg neg. For example, anthropological research indicates that it is possible to maintain face-to-face interaction and consensus-based decisionmaking withinrelatively large social units, through devices such as sequential hierarchies and "integrative institutions," which include rituals to maintain cohesion and task groups with specific responsibilities to divide up the work.³⁸² Reg neg committees in fact already deploy similar mechanisms; further study might focus on the relationship between these specific features and participant satisfaction.

V. Conclusion

The empirical research on regulatory negotiation has just begun, and its relevance to larger questions of institutional design could not be greater. The data presented here paint a largely favorable picture of this particular decisionmaking innovation—it outperformed conventional rulemaking on virtually every measure tested and produced greater satisfaction among participants—and there was no evidence that it exacerbated capture or led the agency to abdicate responsibility. Perhaps most importantly, however, the data suggest that process matters. The legitimacy benefit produced by negotiated rulemaking is at least somewhat due to the process itself, not merely to satisfaction with outcomes. We imagine that other, similar, consensus-based processes might exert the same effect as well.

With this in mind, we encourage agencies to adopt negotiated rulemaking more often and for a greater range of regulatory tasks, and urge its detractors to take another look. We also recommend continued empirical and comparative study of its effects. More broadly, however, we invite scholars and policymakers alike to assess alternative decisionmaking processes not merely in terms of measures like cost, time, and litigation rates, but in terms of their potential to enhance legitimacy. These latest data suggest that administrative legitimacy is, at least in part, a matter of procedural design.

[31 ELR 10840]

TABLE 1 RATING OF DIFFERENT ASPECTS OF FINAL NEGOTIATED RULES: PHASE I

Aspect Rated	*3*Aggregate
	Mean *	Median *	%>0
Quality of Supporting Scientific Analysis	1.6	2	68%
Incorporation of Appropriate Technology	2.1	3	77%
Cost-Effectiveness of Rule	1.5	2	70%
Economic Efficiency of Rule	1.5	2	69%
Ability of EPA to Implement the Rule	1.6	3	67%
Ability of EPA to Implement the Rule Equitably	2.2	3	69%
Ability of Respondent to Comply with the Rule	3.4	4	85%
Ability of Rule to Survive Legal Challenge Prior to	3.3	4	89%
Enforcement
Ability of Rule to Survive Legal Challenge Once	3.4	3.5	93%
Enforced
Overall Benefits of Rule to Respondent's Organization	2.0	2.5	78%
Respondent Overall Rating of the Reg Neg Process	2.1	3	79%
Respondent Rating of Personal Experience with Reg Neg	2.8	3	90%

[31 ELR 10841]

TABLE 2

Independent variable	Measurement/expectation
Ease of decision to participate	1-3 scale, where 1 = easy, 2 =
	mixed, and 3 = difficult. The
	expectation is the easier the
	decision to participate, the
	higher the overall rating.
Anyone not there who should have been?	Coded 1 if yes and 0 if no. The
	expectation is that the
	rating will be lower if
	someone is missing.
Competence of convenor	1-3 scale, where 1 = competent,
	2 = mixed, and 3 =
	incompetent. The expectation
	is that the rating will be
	lower if the convenor is
	regarded as incompetent.
Big business	1 if yes, 0 if no
Small business	1 if yes, 0 if no
Business-size unknown	1 if yes, 0 if no
Environmentalist	1 if yes, 0 if no
State/local agency	1 if yes, 0 if no
Other	1 if not otherwise classified,
	0 if classified above
EPA	Reference group; group other
	affiliations are compared to
Everything negotiated?	1-3 scale, where "no, items
	were left out" = 1: "yes,
	most but not all were
	negotiated" = 2; "yes, all
	were negotiated" = 3. The
	rating of the process is
	expected to be higher if
	everything was negotiated.
Any surprise issues?	1-3 scale, where 1 = no, 2 =
	surprise only at post-reg neg
	events, and 3 = surprise
	issue during reg neg. The
	expectation is that surprise
	issues will lower the
	evaluation of the process.
Complexity/messiness of the reg neg	For each reg neg, messiness =
	(# mentions of sides + #
	mentions of issues)/ #
	respondents. This describes
	the reg neg, not the
	respondent. The more sides
	and the more issues, relative
	to the number of respondents
	in each reg neg, the messier
	and the more difficult the
	negotiation, and the lower
	the evaluation.
Clarity of understanding of the reg neg	For each reg neg,
	clarity/understanding = max.
	number of mentions of issues
	+ sides. In some reg negs not
	one participant could
	identify > than 2 sides and 1
	issue, yet it was clear there
	were more; but the respondent
	could not articulate them. In
	others, respondents had
	little difficulty identifying
	subissues and subconflicts.
	The expectation is that
	greater understanding results
	in higher ratings.
Was all info. needed during reg neg	Coded 1 if all the info the
available?	respondent needed during the
	reg neg was available, and 0
	if all the info was not
	available. The rating of the
	process is expected to be
	higher if needed information
	is available.
Any party with disproportionate influence?	Coded on a 3-point scale, where
	0 = no party with
	disproportionate influence, 1
	= party has disproportionate
	participation not influence,
	and 2 = party has
	disproportionate influence.
	The perception of
	disproportionate influence is
	expected to reduce the rating
	of the process.
Benefits of the rule to the respondents'	Rating scale where -5 =
organization	benefits couldn't be less to
	+5 = benefits couldn't be
	more. When the substance of
	the rule is rated higher, the
	evaluation of the process as
	a whole is also expected to
	increase.
Economic efficiency of the rule	Rating scale where -5 =
	efficiency couldn't be worse
	to +5 = efficiency couldn't
	be better. The expectation is
	that when rules are perceived
	to be efficient, the rating
	of the process will be
	higher.

[31 ELR 10843]

TABLE 3 REGRESSION OF RATING OF RULE-MAKING PROCESS OVERALL ON SELECTED VARIABLES: PHASE I

	Regression	Significance level
Independent variable	coefficient	(p
Ease of decision	-.339	.39
Not there?	-.264	.60
Facilitator competence	.008	.99
Big business	1.007	.36
Small business	-2.465	.08
Business-size unknown	.404	.69
Environmentalists	-.628	.50
State/local officials	-.002	.99
Other	.855	.29
Everything negotiated?	.540	.12
Surprise issue?(NEWCOLUMN) .075	.77
Complexity/messiness	-1.700	.03
Clarity/understanding	.340	.08
Info available?	.230	.66
Disproportionate influence	-.188	.56
Benefits of rule to org.	.284	.01
Economic efficiency of rule	.286	.01
Intercept	4.171	.10
R-squared	.51	.0001
Adjusted R-squared	.36
Number observations	78

[31 ELR 10844]

TABLE 4 MEAN AND STANDARD DEVIATION OF RATINGS OF THE RULE AND THE PROCESS, BY TYPE OF RULE (-5 TO +5 SCALE)

	*3*Conventional	*3*Negotiated
							t-stat. prob
	Xn	s	N	Xc	s	N	(Xn-Xc)
Economic efficiency (general)	-.5	3.2	48	1.5	2.6	97	<.0001
Cost effectiveness (general)	-.4	3.1	48	1.5	2.6	97	<.0001
Rule making process (overall)	.3	2.5	48	2.1	2.5	99	<.0001
Benefits of rule to organization	-.1	2.8	47	2.0	2.6	98	<.0001
Costs of rule to organization	-.4	2.4	46	1.1	2.8	94	<.005
(high rating = > low costs)
Net benefits of rule to R's	-.5	4.5	46	3.1	4.4	94	<.0001
organization *
Quality of scientific analysis	-.4	2.5	48	1.6	2.2	96	<.0001
Incorporation of appropriate	.6	2.1	44	2.1	2.0	94	<.0001
technology
Ability of EPA to implement rule	.9	2.7	47	1.6	2.7	97	<.14
Ability of EPA to implement rule	-.02	3.1	47	2.2	2.5	95	<.0001
equitably
Ability of others to comply with	1.3	2.7	46	2.8	1.9	77	<.0008
rule
Own ability to comply	2.2	2.7	41	3.4	1.9	62	<.02
Ability of rule to survive legal	1.9	2.7	43	3.3	2.4	96	<.005
challenge
Ability of rule to survive legal	1.2	2.8	43	3.4	1.7	90	<.0001
challenge after enforcement
Personal experience	.8	2.6	48	2.8	2.2	99	<.0001

Tables 4-12 are copyrighted by the Journal of Public Administration Research and Theory (J-PART), and they are reproduced with permission from J-PART.

TABLE 5 LIKES AND DISLIKES BY TYPE OF RULE

			Conventional		Negotiated
5.1	*2*Number of likes
	*2*Mentions per	1.6		2.1
	*2*respondent
	*2* N	51		101	(p|t|<.011)
5.2	*2*Number of dislikes
	*2*Mentions per	1.6		1.8
	*2*respondent
	*2* N	51		101	(p|t|<.091)
5.3	*6*Percent of open-ended comments positive, mixed, or negative
	*2* Positive	7%		13%
	Mixed		15%		34%
		Criticisms		93%**		87%**
	Negative	/	78%	/	53%	/
	*2*No. of comments	60		70	[p(x2)<.001]
						**[p(x2)<.35]

[31 ELR 10845]

TABLE 6 WHAT WAS LEARNED, BY TYPE OF RULE (PERCENT OF RESPONDENTS IN EACH TYPE OF RULE MAKING)

	*2*Conventional	*2*Negotiated
*6*6.1 What did you learn that was new?
Technical/scientific information	4%		21%
Knowledge of issue	7%		11%
Positions of other parties	6%		30%
How to negotiate	0%		18%
Info. about the rule or the law	10%		1%
Nothing	19%		0%
Information about EPA	23%		5%
Other	30%		14%
N *	69		197		[p(x2)<.0001]
*6*6.2 Likes about the process (learning)
Learned—issue	7%		9%
Learned—others	0%		15%
Learned—EPA	0%	13%	3%	42%
Learned—future use	0%	/	5%	/
Learned—process	6%	/	10%	/
Other responses not related to
learning	87%		58%
N *	83		211		[p(x2)<.0001]
*6*6.3 Source of new information
Business or environmental groups	2%		20%
EPA	55%		13%
Federal Register	12%		0%
Other participants	0%		45%
Others in respondent's coalition	10%		0%
News reports	10%		0%
Other	11%		22%(NEWLINE) N *	52		143		[p(x2)<.0001]
*6*6.4 Was there information you needed for the RM but didn't have?
No, needed no additional info.	47%		54%
Yes	53%		46%
N *	45		92		[p(x2)<.40]

[31 ELR 10846]

TABLE 7 RESPONSES TO SURVEY QUESTIONS ON DISPROPORTIONATE PARTICIPATION, BY TYPE OF RULE (PERCENT OF RESPONDENTS IN EACH TYPE OF RULE MAKING)

		*2*Conventional	*2*Negotiated
7.1	*6*Was anyone not there who should have been?
	No	52%		66%
	Yes	48%		34%
	N	25		96		[p(x2)<.21]
7.2	*6*Was there any party with disproportionate influence during the
	*6*development of the rule?
	No	56%		25%
	Yes, but it was participation,	0%		27%
	not influence
	Yes	44%		48%
	N	39		100		[p(x2)<.0001]
7.3	*6*Who had disproportionate influence?
	Business, big business	22%		27%
	Environmental groups	9%		17%
	EPA	32%		25%
	State/local government	9%		12%
	Other	28%		19%
	N *	22		95		[p(x2)<.85]
7.4	*6*What is it about the rule that leads you to believe there was
	*6*disproportionate influence?
	Nothing in the rule itself; it	13%		55%
	was the process
	Content of the rule itself	87%		45%
	N	15		69		[p(x2)<.005]
7.5	*6*What was it about the process that enabled disproportionate influence?
	Ingrained bias of EPA	19%		7%
	Strategic position of EPA	22%	63%	17%	30%
	More access to EPA	22%	/	6%	/
	More seats at the table	0%		8%
	More $ resources	15%		3%
	More info/experience on the part
	of the party	15%		19%
	Party was noisier/effective
	negotiator	0%		26%
	Other	7%		14%
	N *	27		115

[31 ELR 10847]

TABLE 8 RESPONSES TO SURVEY QUESTIONS ON PARTICIPANTS' IMPACT, BY TYPE OF RULE

		*3*Conventional	*3*Negotiated
*9*PERCENT OF RESPONDENTS IN EACH TYPE OF RULE MAKING
8.1	*8*Contribution of formal negotiation sessions/public participation to the
	*8*proposed rule
	Major	*3*17%	*3*68%
	Moderate	*3*37%	*3*17%
	Minor	*3*26%	*3*14%
	None	*3*20%	*3* 1%
	N	*3*35	*3*93	[p(x2)<.0001]
8.2	*8*Contribution of your participation to the proposed rule
	Major	*3* 6%	*3*22%
	Moderate	*3*27%	*3*45%
	Moderate/minor	*3* 8%	*3*18%
	Minor	*3*21%	*3*14%
	None	*3*38%	*3* 1%
	N	*3*48	*3*97	[p(x2)<.0001]
8.3	*8*Were there major differences between the proposed and final rule?
	No major differences	*3*44%	*3*65%
	Major differences	*3*56%	*3*35%
	N	*3*36	*3*66	[p(x2)<.005]
8.4	*8*What was the contribution of public participation to change between the
	*8*proposed and final rule?
	Major	*3*44%	*3*19%
	Moderate	*3*44%	*3*11%
	Minor	*3*13%	*3*37%
	None	*3* 0%	*3*33%
	N	*3*16	*3*27	[p(x2)<.005]
8.5	*8*What was the contribution of your participation to change between the
	*8*proposed and final rule?
	Major	*3*25%	*3*11%
	Moderate	*3*40%	*3*21%
	Minor	*3*25%	*3*14%
	None	*3*10%	*3*54%
	N	*3*20	*3*28	[p(x2)<.04]
*9*MEAN SCORE OF RESPONDENTS IN EACH TYPE OF RULE MAKING
8.6	*8*Total contribution (impact of contribution to proposed plus final rule,
	*8*measured on 2 - 8 scale, where 2 = major, 8 = none) by type of
	*8*contribution (public or personal)
			Std.			Std.
	Type of Participation	Mean	dev.	N	Mean	dev.	N
	Public	4.2	1.64	12	4.7	1.41	27	p|t|<.30
	Personal	4.8	1.48	20	5.2	1.41	28	p|t|<.38

[31 ELR 10848]

TABLE 9 RESPONSES TO SURVEY QUESTIONS ON TYPE OF ISSUE, BY TYPE OF RULE

		Conventional	Negotiated
*5*PERCENT OF RESPONDENTS IN EACH TYPE OF RULE MAKING
9.1	*4*What were the types of issues to be decided? (open-ended responses)
	Setting the standard—level,
	timing, measurement	52%	31%
	Compliance/implementation issues	39%	58%
	Degree of uniformity (of std.,
	compliance)	3%	12%
	Other	6%	0%
	N *	97	258	[p(x2)<.001]
9.2	*4*What issues engendered the greatest conflict?
	The standard (level, timing,
	measurement)	39%	34%
	Compliance issues (measurement,
	who complies, who enforces,
	evidence, how to comply)	32%	53%
	Degree of uniformity	9%	8%
	Other	20%	5%
	N *	65	182	[p(x2)<.05]
*5*MEAN SCORE OF RESPONDENTS IN EACH TYPE OF RULE MAKING
9.3	*4*Complexity (# issues + # sides) and clarity of rule making (# different
	*4*issues/sides) Complexity
	Mean	2.5	4.1
	Std. dev.	.48	.58
	N	51	101	(p|t|<.0001)
	Clarity
	Mean	6.0	8.9
	Std. dev.	1.6	2.1
	N	51	101	(p|t|<.0001)

[31 ELR 10849]

TABLE 10 ASPECTS OF COSTS, BY TYPE OF RULE

		*3*Conventional	*3*Negotiated
10.1	*8*Resources spent on participation
		X	s	N	X	s	N	t-stat. prob
	Number of
	professional
	staff hours	378	955	45	2,224	5,812	88	p<.04 **
	Number of clerical
	staff hours	97	381	45	385	1,345	82	p<.16
	Amount spent on
	research,
	information
	collection	47,458	146,268	43	81,748	371,463	83	p<.56
	Amount spent on
	legal counsel,
	consultants	6,755	29,302	42	55,990	306,941	90	p<.30
	Amount spent on
	"other"	1,353	4,848	49	2,516	5,184	80	p<.21
	relative to
	available
	resources (%)	14	24	51	26	29	87	p<.03
10.2	*8*Did the value of the benefits of participation exceed the costs?
	Yes	*3*70%	*3*77%
	Equal	*3* 6%	*3* 7%
	No	*3*23%	*3*15%
	N	*3*47	*3*97	[p(x2)<.45]

[31 ELR 10850]

TABLE 11 INDEPENDENT VARIABLES IN REGRESSION OF "RATING OF RULE MAKING PROCESS OVERALL"

Independent variable	Measurement/Expectation
Big business	1 if yes, 0 if no
Small business	1 if yes, 0 if no
Business—size unknown	1 if yes, 0 if no
Environmentalist/public	1 if yes, 0 if no
interest
State/local agency	1 if yes, 0 if no
Supplier of compliance	1 if yes, 0 if no
goods/services
Other	1 if not otherwise classified, 0 if classified
	above
EPA	Reference group; group other affiliations are
	compared to
Everything	1-3 scale, where "no, items were left out" = 1;
negotiated/settled?	"yes, most but not all were negotiated" = 2;
	"yes, all were negotiated" = 3. The rating of
	the process is expected to be higher if
	everything was negotiated.
Complexity/messiness of the	For each rule, messiness = (number of mentions
rule	of sides + number of mentions of issues) +
	number of respondents. This describes the rule,
	not the respondent. The more sides and the more
	issues, relative to the number of respondents
	for each rule, the messier, the more difficult
	the negotiation, the higher the decision costs,
	and the lower the evaluation.
Clarity of understanding of	For each rule, clarity/understanding = maximum
the rule	number of mentions of different issues + sides.
	In some rules not one participant could cite >
	than 2 sides and 1 issue, it was clear there
	were more; but the respondent could not
	articulate them. In others, respondents had
	little difficulty citing subissues and
	subconflicts. The expectation is that greater
	understanding results in higher ratings.
Any party with	Coded on a 3-point scale, where 0 = no party
disproportionate influence?	with disproportionate influence. 1 = party has
	disproportionate participation not influence,
	and 2 = party has disproportionate influence.
	The perception of disproportionate influence is
	expected to reduce the rating of the process.
Net benefits of the rule to	Rating scale where -5 = benefits couldn't be
the respondents'	less to +5 = benefits couldn't be more, less
organization	rating scale where -5 = costs couldn't be worse
	to +5 = costs couldn't be better. When net
	benefits of the rule to the organization are
	rated higher, the evaluation of the process as
	a whole is also expected to increase.
General economic efficiency	Rating scale where -5 = efficiency couldn't be
of the rule	worse to +5 = efficiency couldn't be better.
	The expectation is that when rules are
	perceived to be efficient, the rating of the
	process will be higher.
Type of process dummy	1 if negotiated rule, 0 if conventional
Level of standard issue	1 if standard mentioned as issue to be decided
	(level, measurement, timing); 0 if not.
Compliance issue	1 if compliance mentioned as issue to be
	decided (measurement, who complies/enforces,
	evidence, how to comply); 0 if not.
"Other" issue	1 if uniformity or "other" mentioned as issue
	to be decided; 0 if not.

[31 ELR 10851]

TABLE 12 ROBUST REGRESSION OF RATING OF RULE MAKING PROCESS OVERALL ON SELECTED INDEPENDENT VARIABLES, INCLUDING TYPE OF RULE MAKING (ERRORS CLUSTERED ON INDIVIDUAL RULE MAKING)

		Unstandardized	Robust
Independent Variable		Coefficient	Std. Err.	t-value	p-value
*2*Big business (v. EPA)	1.18	.52	2.27	.04
*2*Small business (v. EPA)	-.44	.80	-.55	.59
*2*General business (v. EPA)	.17	.03	.29	.78
*2*Env./public interest (v. EPA)	-1.22	1.07	-1.14	.27
*2*State/local agency (v. EPA)	.45	.71	.64	.54
Supplier (v. EPA)		.68	.43	1.59	.14
Other (v. EPA)		.47	.49	.96	.36
All settled?		.43	.24	1.79	.097
Complexity		-2.07	.31	-6.61	.000
Clarity		.42	.10	4.23	.001
*2*Disproportionate influence?	-.41	.27	-1.53	.15
Net benefits		.20	.06	2.99	.01
Efficiency		.25	.09	2.70	.018
Reg-neg dummy		2.52	.53	4.74	.000
*2*Level of standard issue	1.02	.53	1.93	.076
Compliance issue		-.11	.34	-.33	.74
Other issue		.73	.40	1.86	.086
Intercept		1.76	1.10	1.60	.13
N	121
N of clusters	14
R2	.58
F-value	4.24
P-value	<.0075

1. We use the terms "regulatory negotiation" (reg neg) and "negotiated rulemaking" interchangeably to refer to the specific process by which agencies formally negotiate rules pursuant to the Negotiated Rulemaking Act of 1990 (NRA), 5 U.S.C. §§ 561-570, available in ELR STAT. ADMIN. PROC. By contrast, some scholars and practitioners use the term "regulatory negotiation" more broadly to refer to virtually any dialogue process that involves a regulation.

2. See, e.g., Susan Rose-Ackerman, American Administrative Law Under Siege: Is Germany a Model?, 107 HARV. L. REV. 1279, 1281 (1994) [hereinafter Rose-Ackerman, Germany] (claiming that regulatory negotiation is "inadequate" if one accepts a basic commitment to the democratic legitimacy of the administrative process).

3. Cary Coglianese, Assessing Consensus: The Promise and Performance of Negotiated Rulemaking, 46 DUKE L.J. 1255 (1997) [hereinafter, Coglianese, Assessing Consensus] (arguing that regulatory negotiation fails to deliver on its promise to save time and reduce litigation rates).

4. See Philip J. Harter, Assessing the Assessors: The Actual Performance of Negotiated Rulemaking, 9 N.Y.U. ENVTL. L.J. 32 (2000) [hereinafter Harter, Actual Performance] (disputing Coglianese's methodology and challenging his conclusions).

5. Cornelius M. Kerwin & Laura I. Langbein, An Evaluation of Negotiated Rulemaking at the Environmental Protection Agency: Phase I (1993) (unpublished report prepared for Administrative Conference of United States, on file with New York University Environmental Law Journal) [hereinafter Phase I]; Cornelius M. Kerwin & Laura I. Langbein, An Evaluation of Negotiated Rulemaking at the Environmental Protection Agency: Phase II, A Comparison of Conventional and Negotiated Rulemaking (1997) (unpublished report prepared for EPA, on file with New York University Environmental Law Journal) [hereinafter Phase II]. Phase I was funded by the now-defunct Administrative Conference of the United States; Phase II was funded by EPA. For an analysis of the Phase II data, focusing exclusively on the comparison between negotiated and conventional rulemaking, see Laura I. Langbein & Cornelius M. Kerwin, Regulatory Negotiation Versus Conventional Rulemaking: Claims, Counterclaims, and Empirical Evidence, 10 J. PUB. ADMIN. RES. & THEORY 599 (2000) [hereinafter Langbein & Kerwin, Claims]. The quantitative claims we make in this Article are supported by the specific statistics reported in Phase I and Phase II, which, although unpublished, are available on request from Laura Langbein (e-mail: langbei@american.edu). In this Article, we excerpt verbatim significant portions of the reports but we fall short of reproducing them in their entirety. We emphasize the findings that we believe are most salient for the debate over reg neg, and focus on data about learning and legitimacy in particular. These studies, together with the published presentation of the comparative Phase II data in Langbein & Kerwin, Claims, supra, contain a number of tables that report and disaggregate the data, and they provide, among other things, a detailed breakdown of the regression results. For the reader's convenience, we reproduce relevant tables. See infra Appendix. These three sources also provide comprehensive information on methodological questions that might interest some readers, including details concerning instrument design, survey methodology, and response rate. In particular, some readers will want to know more about the process by which rules were chosen in Phase II for purposes of comparison. On this latter issue, see Langbein & Kerwin, Claims, supra, at 600-02. We elected not to reproduce a number of tables, regression results, and methodological detail here in favor of a shorter and more narrative summary that might be more comprehensible to the legal audience. We do, however, refer to specific results from these studies. The results from the Phase I and Phase II reports have not been peer-reviewed; those from Claims have been peer-reviewed. The results from Claims are a large subset of the results from Phase II.

6. Phase I, supra note 5, at 33, 42; Langbein & Kerwin, Claims, supra note 5, at 625.

7. Langbein & Kerwin, Claims, supra note 5, at 610.

8. See Rose-Ackerman, Germany, supra note 2, at 1284-85; Jerry L. Mashaw, Improving the Environment of Agency Rulemaking: An Essay on Management, Games, and Accountability, 57 LAW & CONTEMP. PROBS. 185, 253 (1994) (dismissing reg neg as useful in only a minority of cases and arguing that consensual processes seem a formula for stalemate or capture unless selectively applied).

9. See Mashaw, supra note 8, at 253.

10. However, there are relative differences in their enthusiasm. Environmental interests rated the reg neg process lower than other participants. Phase I, supra note 5, at 40-41. Their ratings were positive, but only marginally so. Id. The extent to which resource constraints affected their ratings is not clear. Although the environmental interests gave the lowest ratings to the process, once other variables were taken into account—including variables that reflect perceptions regarding the substantive quality of the final rule—their ratings are not significantly different from those of EPA. Id. The regression results show that only one affiliation variable, that for small business, is significant; these respondents rated the reg neg process 2.5 points lower on the 11-point scale than EPA respondents, once other variables were held constant. Id.

11. Langbein & Kerwin, Claims, supra note 5, at 602-03, 607-08.

12. See id. at 614.

13. Id. at 605-07.

14. Our definition of legitimacy is elaborated in part II.C.

15. Langbein & Kerwin, Claims, supra note 5, at 619, exh. 7.2.

16. Phase I, supra note 5, at 43.

17. Id.

18. Langbein & Kerwin, Claims, supra note 5, at 608.

19. See, e.g., Cary Coglianese, Is Consensus an Appropriate Basis for Regulatory Policy?, in ENVIRONMENTAL CONTRACTS: COMPARATIVE APPROACHES TO REGULATORY INNOVATION IN THE UNITED STATES AND EUROPE 104-05 (Eric Orts & Kurt Deketelaere eds., 2000) [hereinafter Coglianese, Is Consensus Appropriate?] (asserting that consensus is unnecessary because the benefits of consensus can be accrued through other processes, such as informal policy dialogues). In fact, we believe that such dialogues, although potentially useful, will generate different kinds of information than that produced in a reg neg, as well as different levels of commitment, because the parties will address themselves to the agency and not to each other, and because they will (rightly) not view the result as binding. We believe that a consultative process designed to air opinions is fundamentally different from a consensus-based process designed to generate a rule. In addition, a consensus-based process may enable the parties to make sophisticated trades that are often unavailable in policy dialogues.

20. Contra id. (manuscript at 103 n.31) ("It is far from clear that [relatively higher ratings of outcomes by participants in reg negs] are unbiased…. Cognitive dissonance seems likely to explain the favorable outcome ratings …, as these proceedings involve considerably more time and effort on their part.").

21. Phase I, supra note 5, at 37.

22. That is, the mechanism responsible for an outcome should have no impact on our assessment of the outcome, unless the mechanism is itself independently objectionable. Take the example of a parent verbally prompting her child to study. If the verbal prompt in fact induces the child to study, we might not care why it is effective as long as it produces the desired result. On the other hand, if it is effective only because the parent strikes the child if he refuses to study, then it matters why the verbal method is effective.

23. Arguing that such findings are "biased" may be technically correct—a psychological bias toward validating the expenditure of time and effort leads to a favorable assessment—but the results are no less valid.

24. Langbein & Kerwin, Claims, supra note 5, at 614.

25. Id. at 609, exh. 4.3.

26. Id.

27. Id. at 609, exh. 4.2.

28. See Phase I, supra note 5, at 603. There are some data, however, indicating that face-to-face learning produces socially better outcomes and reduces decision costs. See Elinor Ostrom, A Behavioral Approach to the Rational Choice Theory of Collective Action, 92 AM. POL. SCI. REV. 1, 6-7 (1998). The Kerwin and Langbein data show striking reports of learning in the context of face-to-face interaction. Phase I, supra note 5, at 12.

29. Assessing rule quality is problematic even if one uses a single criterion for evaluating quality, such as efficiency. In any event, we do not believe that efficiency alone is the appropriate measure of rule quality; process also matters and, according to the data reported here, process appears to be valued by the participants independently of two other measures of participants' views of the efficiency of the rule.

30. Phase I, supra note 5, at 41-42.

31. The study revealed two significant process variables (in addition to these two substantive variables): the complexity of the reg neg in terms of numbers of issues and sides, and the ability of the participants to identify that complexity. These four variables are equally important in affecting the overall rating. However, "the [17] variables in the regression explain only 36% of the variance in the respondents' rating of the overall reg neg process." Id. at 42.

32. See E. ALLAN LIND & TOM R. TYLER, THE SOCIAL PSYCHOLOGY OF PROCEDURAL JUSTICE (1988) (surveying the social psychology literature on perceptions of procedural justice and concluding that people positively evaluate processes in which they are permitted to participate and in which their views are considered by decisionmakers). Although most of the studies concern adjudication. Lind and Tyler cite additional studies indicating that the results may be extrapolated to other decisionmaking contexts, including policymaking. See id. at 147-72; see also TOM R. TYLER, WHY PEOPLE OBEY THE LAW (1990); Robert Kidder & Craig McEwen, Taxpaying Behavior in Social Context: A Tentative Typology of Tax Compliance and Noncompliance, in 2 TAXPAYER COMPLIANCE 47, 53-54 (Jeffrey A. Roth & John T. Scholz eds., 1989) ("the more involved people are in making rules and consenting to them, the stronger [their] sense of obligation" to abide by them); Cynthia R. Farina, The Consent of the Governed: Against Simple Rules for a Complex World, 72 CHI.-KENT L. REV. 987, 1027-32 (1997) (discussing important role that perceptions of fairness play in determination of administrative legitimacy). Indeed, data suggest that people "differentiate between receiving favorable outcomes, receiving fair outcomes, and having decisions made by a fair process." LIND & TYLER, supra, at 158; see also Bruno S. Frey & Felix Oberholzer-Gee, Fair Siting Procedures: An Empirical Analysis of Their Importance and Characteristics, 15 J. POL'Y ANALYSIS & MGMT. 353 (1996).

33. Phase I, supra note 5, at 24.

34. See Philip J. Harter, Negotiating Regulations: A Cure for Malaise, 71 GEO. L.J. 1, 7, 17 (1982) [hereinafter Harter, Cure] (arguing that traditional rulemaking does not allow for the "careful trade offs necessary for an enlightened regulation" and suffers from a "crisis of legitimacy"); see also Harter, Actual Performance, supra note 4, at 55 (those present at the creation of reg neg "saw direct negotiations among the parties—a form of representational democracy not explicitly recognized in the Administrative Procedure Act—as resulting in rules that are substantively 'better' and more widely accepted"). Harter proposed that assessments of reg neg ought to focus on the content of the rule and the degree of compliance.

35. When scholars first proposed reg neg and argued that it would improve legitimacy, they often thought in terms of the people and organizations closely involved with rulemaking and the regulatory process, not necessarily the public at large. See Lawrence Susskind & Gerard McMahon, The Theory and Practice of Negotiated Rulemaking, 3 YALE J. ON REG. 133, 133 (1985):

Scholars, government officials, and practitioners have expressed concern over the weakness of the federal rulemaking process and the time it often takes to promulgate rules. Given the many instances in which rules have been challenged in court, both the process of rulemaking and the regulations produced seem to have lost legitimacy in the eyes of many regulatees.

36. Evidence that face-to-face communication results in substantial increases in level of cooperation is "consistent, strong and replicable." Ostrom, supra note 28, at 6.

37. Legitimacy can enhance trust in institutions. See THOMAS C. BEIERLE, PUBLIC PARTICIPATION IN ENVIRONMENTAL DECISIONS: AN EVALUATION FRAMEWORK USING SOCIAL GOALS 12-13 (Resources for the Future, Discussion Paper No. 99-06, 1998). See also LIND & TYLER, supra note 32, at 179 (arguing that attitudes toward institutions and authority are strongly affected by procedural justice).

38. See Harter, Cure, supra note 34, at 28-29 (arguing that negotiated rules ought to be easier to implement and enforce in part because of the increased legitimacy negotiated rules will enjoy in the eyes of the participants).

39. Of course, litigation rates may be a poor measure of legitimacy. A party might view a process as legitimate and still decide to litigate the rule for other reasons. With respect to compliance, as of now there is no research comparing compliance with negotiated rules to compliance with "comparable" conventionally written rules. Even with conventional rulemaking, compliance proves difficult to evaluate, partly because EPA does not have a system for tracking implementation following promulgation. One way to measure timely implementation might be to track the steps that regulated companies have taken to implement the rule, by examining such things as investments in new technology, new managerial practices and production processes, or increased resources devoted to environmental compliance.

40. We define transaction costs as the time, money, and aggravation of any rulemaking process.

41. 5 U.S.C. §§ 500-706, available in ELR STAT. ADMIN. PROC.

42. All of these factors—time, cost, and litigation rates—might be related to legitimacy. After all, observers might view an expensive, slow, and cumbersome rulemaking process less favorably than an expeditious one. But we separate these factors and call them transaction costs because we believe that a process can be legitimate—that is, satisfying to participants—even if costly.

43. See Harter, Actual Performance, supra note 4 (arguing that the primary objectives of negotiated rulemaking are better and more widely accepted rules).

44. See generally Jody Freeman, Collaborative Governance in the Administrative State, 45 UCLA L. REV. 1, 33-35 (1997) (describing regulatory negotiation in hopeful terms, as an example of a larger and mostly desirable trend toward "collaborative governance" rather than a minor process reform best used in moderation). Still, even optimistic accounts recognize the limitations of the process. See id. at 35.

45. See, e.g., William Funk, Bargaining Toward the New Millennium: Regulatory Negotiation and the Subversion of the Public Interest, 46 DUKE L.J. 1351, 1356 (1997) [hereinafter Funk, Bargaining] (stating that "the principles, theory, and practice of negotiated rulemaking subtly subvert the basic, underlying concepts of American administrative law"); see also THEODORE LOWI, THE END OF LIBERALISM: IDEOLOGY, POLICY, AND THE CRISIS OF PUBLIC AUTHORITY (1969).

46. Public choice theory treats legislation and regulation as the product of bargaining by rational, self-interested legislators and other elected officials motivated primarily by the desire for reelection. Procedural rules, such as those contained in the APA, merely enable organized groups—allegedly over-represented by legislators—to ensure that the bargains they strike in the legislature are enacted by agencies. There is no such thing as the "public interest," only bargains made by legislators seeking reelection, implemented through self-interested agency brokers overseen by Congress, the president, and the courts. By bringing stakeholders to the table to negotiate regulations, regulatory negotiation seemed to confirm this deal-making image of the regulatory process. See generally JERRY L. MASHAW, GREED, CHAOS, AND GOVERNANCE: USING PUBLIC CHOICE TO IMPROVE PUBLIC LAW 124-30 (1997); Daniel A. Farber, Democracy and Disgust: Reflections on Public Choice, 65 CHI.-KENT L. REV. 161 (1989); Gary S. Becker, A Theory of Competition Among Pressure Groups for Political Influence, 98 Q. J. ECON. 371 (1983); Mathew D. McCubbins et al., Administrative Procedures as Instruments of Political Control, 3 J.L. ECON. & ORG. 243, 247 (1987).

47. The NRA limits the total number of participants to 25, "unless the agency head determines that a greater number of members is necessary to the functioning of the committee or to achieve balanced membership." 5 U.S.C. § 565(b), available in ELR STAT. ADMIN. PROC. The Act lists a "limited number of identifiable interests that will be significantly affected by the rule" as a consideration for an agency administrator when deciding whether to use negotiated rulemaking. Id. § 563(a)(2), available in ELR STAT. ADMIN. PROC.

48. See Susan Rose-Ackerman, Consensus Versus Incentives: A Skeptical Look at Regulatory Negotiation, 43 DUKE L.J. 1206, 1211 (1994) [hereinafter Rose-Ackerman, Consensus] (arguing that not all interested parties are adequately represented in regulatory negotiation and that agreement among only that subset of interests with organized advocates is insufficient).

49. Some scholars backed up their fears with anecdotal empirical evidence. For example, in a detailed account of one early reg neg, Bill Funk argued that the consensus reached and then promulgated as arule was, in fact, illegal. William Funk, When Smoke Gets in Your Eyes: Regulatory Negotiation and the Public Interest—EPA's Woodstove Standards, 18 ENVTL. L. 55 (1987) [hereinafter Funk, Smoke].

50. See Rose-Ackerman, Germany, supra note 2, at 1296 (claiming that even consensual decisions that include environmental groups or citizen bodies "lack legitimacy"). Funk charges that the incentives for rulemaking by consensus "subvert the principles underlying traditional administrative law by elevating the importance of consensus among the parties above the law, the facts, or the public interest." Funk, Bargaining, supra note 45, at 1387.

51. 5 U.S.C. § 564(a) (1994), available in ELR STAT. ADMIN. PROC.

52. Id. § 565(a), available in ELR STAT. ADMIN. PROC.

53. Id. § 566(d), (g), available in ELR STAT. ADMIN. PROC.

54. Id. app. §§ 1-15, available in ELR STAT. ADMIN. PROC.

55. Id. § 564(b), available in ELR STAT. ADMIN. PROC.

56. Congress passed the NRA "to establish a framework for the conduct of negotiated rulemaking [and] to encourage agencies to use the process when it enhances the informal rulemaking process." Id. § 561, available in ELR STAT. ADMIN. PROC. Despite the requirements that accompany negotiated rulemaking, the threshold determination whether to utilize negotiated rulemaking is discretionary for agencies. See id. § 563(a), available in ELR STAT. ADMIN. PROC. ("an agency may establish a negotiated rulemaking committee") (emphasis added).

57. Id. § 563(a)(7), available in ELR STAT. ADMIN. PROC. (requiring that the agency consider whether it would use a consensus rule as the "basis for the rule proposed by the agency for notice and comment").

58. See McCubbins et al., supra note 46, at 249.

59. 5 U.S.C. § 570, available in ELR STAT. ADMIN. PROC. Congress apparently declined alternative proposals for relaxed review. Harter had proposed that negotiated rules "should be sustained to the extent that [they are] within the agency's jurisdiction and actually reflect[] a consensus among the interested parties." Harter, Cure, supra note 34, at 103. Susskind and McMahon had argued that courts ought to review only the sufficiency of the process. See Susskind & McMahon, supra note 35, at 164.

60. See Terry M. Moe, Political Institutions: The Neglected Side of the Story, 6 J.L. ECON. & ORG. 213, 229-30 (1990) (arguing that input from and compromise on behalf of potentially affected parties is necessary for effective governmental regulation).

61. See, e.g., OFFICE OF POLICY, PLANNING, AND EVALUATION, U.S. EPA, ASSESSMENT OF EPA'S NEGOTIATED RULEMAKING ACTIVITIES (1987), reprinted in NEGOTIATED RULEMAKING SOURCEBOOK 23 (David M. Pritzker & Deborah S. Dalton eds., 1995) [hereinafter EPA ASSESSMENT]; Henry H. Perritt Jr., Negotiated Rulemaking Before Federal Agencies: Evaluation of Recommendations by the Administrative Conference of the United States, 74 GEO. L.J. 1625 (1986); Funk, Smoke, supra note 49; Neil Eisner, Regulatory Negotiation: A Real World Experience, 31 FED. B. & NEWS J. 371(1984); Daniel J. Fiorino & Chris Kirtz, Breaking Down Walls: Negotiated Rulemaking at EPA, 29 TEMP. ENVTL. L. & TECH. J. 29 (1985); Ellen Siegler, Regulatory Negotiations: A Practical Perspective, 22 ELR 10647 (Oct. 1992). See also Freeman, supra note 44, at 41-54 (describing one EPA and one Occupational Safety and Health Administration reg neg).

62. See EPA ASSESSMENT, supra note 61, at 29.

63. For example, the report claimed that one negotiated rulemaking resulted in a savings of $ 150,000. Id. at 29. The agency acknowledged, however, that its calculations did not include costs of litigation, were the rule ultimately challenged. Id. at 32.

64. Id. at 29.

65. Freeman, supra note 44, at 24-25 n.63 (interview by Jody Freeman with Kathy Bailey, counsel for the Chemical Manufacturers Association during the equipment leaks regulatory negotiation, Sept. 21, 1994).

66. Id. (interview by Jody Freeman with Deborah Dalton, Deputy Director, EPA Regulatory Negotiations Project, July 25, 1994).

67. Id.

68. Indeed, in an interview, the then-deputy director of EPA's regulatory negotiation project referred to the difficulty of measuring the true benefits of the process: "There's no database for how often we get sued on a rule." Id. Moreover, the industry response to negotiated rules was hard to compare with its response to conventional rules.

69. Coglianese, Assessing Consensus, supra note 3, at 1257-59.

70. Id. at 1282-83. To calculate the time consumed by a negotiated rule, Coglianese counted the number of days between the agency's announcement of its intent to form a negotiating committee and publication of the final rule in the Federal Register. Id. Of all 35 negotiated rulemakings that yielded final rules at the time of his study, he calculated that the shortest (the U.S. Coast Guard's regulations for drawbridges over the Chicago River) took about half a year, and the longest (EPA's rule on farmworker pesticide protection standards) consumed nearly seven years. Id. at 1279. The latter, especially lengthy effort notably resulted in a failure to reach consensus. See id. at 1279 n.111. The average length of time for all negotiated rulemakings counted by Coglianese was 835 days; the median length of time was 651 days. Id. at 1279-80. To compare negotiated with conventional rules, however, Coglianese reevaluated the findings of a prior study of rulemaking at EPA by Cornelius M. Kerwin and Scott R. Furlong, Time and Rulemaking: An Empirical Test of Theory, 2 J. PUB. ADMIN. RES. & THEORY 113 (1992). Id. at 1280-86. To calculate the average time consumed by conventional rules, he adopted Kerwin and Furlong's method, which calculated average rulemaking time for conventional rules by measuring the time elapsed between the date that the rule entered the agency's regulatory development management system, and the date on which it was finalized. See id. at 1282-84.

71. Id. at 1298-1301. Prior to his calculation, many commentators had simply accepted the unsubstantiated claim that approximately 80% of rules promulgated by EPA are subjected to legal challenge. See id. at 1296 n.183.

72. Id. at 1301-02.

73. Id. at 1271.

74. See Coglianese, Is Consensus Appropriate?, supra note 19 (manuscript at 98-105).

75. Harter, Actual Performance, supra note 4, at 32.

76. Id. at 42 (arguing that the farmworker protection rule should not have been included).

77. Id. at 49.

78. According to Harter, "no rule that implements a consensus reached by the committee in which the parties agree not to challenge it has ever been the subject of a substantive judicial review." Id. at 41. This suggests that the failure of EPA to respect the efforts of the participants, and perhaps the substantive change itself, prompted the litigation. Of course, one would have to look closely at the identity of the litigants and the subject of the lawsuit to confirm this. Coglianese's overview of legal challenges to rules promulgated after regulatory negotiation indicates that the final rule often is challenged by both parties to the regulatory negotiation and non-parties. See Coglianese, Assessing Consensus, supra note 3, at 1307.

79. Phase I, supra note 5.(NEWLINE)

80. Phase II, supra note 5.

81. Pursuant to the NRA, the sponsoring agency may use a convenor to assist it in determining whether it ought to establish a negotiated rulemaking committee. 5 U.S.C. § 563, available in ELR STAT. ADMIN. PROC. An agency may establish a committee if it determines that the committee can adequately represent the interests that will be significantly affected by a proposed rule and that it is feasible and appropriate. Id. § 563(a)(2), (3)(A), available in ELR STAT. ADMIN. PROC. The convenor identifies parties who are interested in, and will be significantly affected by, the proposed rule. Id. § 563(b)(1)(A), available in ELR STAT. ADMIN. PROC. If the agency elects to establish a committee, it must give notice in the Federal Register and solicit membership. Id. § 564(a)-(b), available in ELR STAT. ADMIN. PROC. Parties may either independently apply for membership in a reg neg committee or the agency might actively encourage the party to participate. Id. § 564(a)(4), (b), available in ELR STAT. ADMIN. PROC. Alternatively, the agency may deny a party membership if it feels that the party's interests are already adequately represented, or a party may decline to apply. On the multiple purposes of the convening process, and its importance to a successful reg neg, see Philip J. Harter, Fear of Commitment: An Affliction of Adolescents, 46 DUKE L.J. 1389, 1404-08 (1997) [hereinafter Harter, Fear of Commitment].

82. See Phase I, supra note 5, at 2-3. The eight rulemakings were: (1) a new source performance standard for wood-burning stoves, implementing part of Clean Air Act (CAA) § 111, 42 U.S.C. § 7411, ELR STAT. CAA § 111; (2) a rule requiring inspection and abatement of asbestos-containing materials in school buildings under the Asbestos Hazard Emergency Response Act of 1986; (3) a rule implementing the prohibition on underground injection of hazardous wastes found in the Hazardous and Solid Waste Amendments of 1984 to the Resource Conservation and Recovery Act (RCRA); (4) a rule standardizing the manifest (labeling) system used in the transport and in other activities related to hazardous waste; (5) a rule governing minor permit modifications under RCRA; (6) a rule establishing national emission standards for coke oven batteries under the CAA; (7) a rule governing fugitive emission leaks from equipment under CAA §§ 111 and 112, 42 U.S.C. §§ 7411 and 7412, ELR STAT. CAA §§ 111 and 112; and (8) a rule governing the use of reformulated and oxygenated fuels, under the CAA. Phase I, supra note 5, at 2-3. Of the eight rulemakings, six had resulted in proposed or final rules by the time of the study, one had not yet been finalized, and another failed to result in a consensus proposal. Id. at 2.

83. "Open-ended" means that in response to questions, respondents volunteered the reasons; interviewers did not require respondents to conform their responses to preconceived fixed categories. Interviewers did, however, press respondents to clarify their comments, whenever possible. Each response was coded by numbers that summarized the content of the response. In most cases, a single response was coded by more than one number. Another coder double checked the mapping. This process allowed Kerwin and Langbein to count the number of respondents who made similar remarks about each rule. Allowing open-ended responses prevented the researchers from miscasting responses into preconceived but incorrect categories and from losing information by coding "other" when respondents came up with an unanticipated response. All codes were verified by the person who entered the data, who was never the person who coded the data. There were few differences, and those that arose were resolved by discussion between the coder and the interviewer.

84. The actual number of interviews conducted for each negotiated rule was: (1) woodstoves: 7; (2) asbestos in schools: 16; (3) underground injection: 6; (4) hazardous waste manifest: 19; (5) RCRA minor permit modification: 10; (6) coke oven batteries: 11; (7) fugitive emissions: 12; and (8) clean fuels: 20. Id. at 6. For each negotiated rulemaking, a representative of each of the major interests on the negotiating committee, e.g., the agency, businesses, environmental groups, was part of the sample. With respect to the sampling approach, the report states, "we are confident that the general findings and tentative conclusions presented here would not be contradicted or otherwise significantly affected by the conduct of additional interviews." Id.

85. Id. at 33.

86. Id. at 24.

87. Id. at 42.

88. Id.

89. Id. at 36. Although we cannot interpret the full meaning of this number without an explicit comparison to the analogous ratings for conventional rulemaking, we note simply that participant reactions are positive.

90. See id.

91. See Harter, Cure, supra note 34, at 7; Rose-Ackerman, Germany, supra note 2, at 1284-85.

92. See Harter, Cure, supra note 34, at 112 (arguing that reg neg is best used when the issues are well-defined, and a decision is imminent).

93. See Susskind & McMahon, supra note 35, at 139-40.

94. The following reasons were mentioned most frequently; issues were simple/clear (12% of mentions); number of interests affected (9% of mentions); high level of conflict over issues (9% of mentions); presence of a deadline or mandate (9% of mentions); best way to obtain views of affected interests (8% of mentions); avoid litigation (8% of mentions); affected parties known (7% of mentions); parties believed they would do better with reg neg than with conventional rules (7% of mentions); issue was complex/controversial (7% of mentions); and EPA wanted it (6% of mentions). Phase I, supra note 5, at 7.

95. See id. at 8.

96. Id. at 11. The NRA limits committees to 25 members, "unless the agency head determines that a greater number of members is necessary for the functioning of the committee or to achieve balanced membership." 5 U.S.C. § 565(b), available in ELR STAT. ADMIN. PROC.

97. See generally Rose-Ackerman, Consensus, supra note 48.

98. See Phase I, supra note 5, at 9. In only one case (asbestos) was there a claim of exclusion from the reg neg, and in that case participants alleged both under—and over—inclusion.

99. Id. at 10. Correspondingly, 35% of respondents reported that an interest was missing. The three types of interests most frequently described as missing from the negotiations were environmental groups (20% of mentions), small business (15% of mentions), and unions (10% of mentions). The type of environmental groups most often described as missing were local rather than national organizations. When asked why these and other groups were missing, 43% would not speculate. Those who did cited a lack of time or money, resistance by EPA, and no interest on the part of the missing group in a change in the status quo. Id.

100. Id. at 11.

101. Studies by social psychologists prove helpful here. They demonstrate that participation in a process can lead to favorable evaluations even when participants know that their work will not be used in the decision, as long as they feel they have had a voice. See, e.g., LIND & TYLER, supra note 32, at 176-77 (arguing that participation, not necessarily outcome, is a major determinant of satisfaction); Frey & Oberholzer-Gee, supra note 32.

102. Phase I, supra note 5, at 8. The remaining 14% said their decision was mixed. Id.

103. Id. at 9.

104. Id. at 8. Respondents offered various reasons why the decision to participate in the rulemaking was easy or difficult, but the most common reasons were concerns regarding the impact the rule in question would have on their organization (17% of mentions) and a desire to have an effect on the outcome of the rulemaking(15% of mentions). Id. at 9. These reasons varied somewhat depending on the respondent's affiliation. For example, the reasons business respondents most frequently mentioned were the comparatively worse outcome of conventional rulemaking (23% of mentions) and limited resources (18% of mentions), whereas environmentalists were concerned about limited resources (18% of mentions), wanting to have an impact on the outcome (14% of mentions), and concern about what others would do if respondent was not there (14% of mentions). Id. at 9. We note that "resources" were mentioned with equal relative frequency by environmental and business respondents, as were various comments relating to impact on the final rule. In other words, what is important is the lack of differences in the reasons for participation.

105. See, e.g., Freeman, supra note 44, at 31-32.

106. Phase I, supra note 5, at 10.

107. Id.

108. See Harter, Cure, supra note 34, at 67-82 (describing the open-ended convening process).

109. Phase I, supra note 5, at 10.

110. Id. at 12.

111. Id. at 13.

112. Id.

113. Id.

114. Id. Interestingly, this increased trust and familiarity occurred most in the coke oven negotiations of all the rules studied; 69% of coke oven mentions noted this, compared to 23% overall. Id.

115. Id. at 13.

116. Id. at 15.

117. Not all issues in a negotiated rulemaking generate conflict but those that do are the most important to the ultimate success of the process. The issues reported as generating conflict were numerous and diverse; reg neg participants identified 182 contentious issues. Id. at 15.

118. Id. at 15:

There are statistically significant differences among the reg negs, with full resolution of all issues reported most often in the cases of fugitive emissions (100% so reported), asbestos in schools (91% of mentions) and coke ovens (92% of mentions) and least often in hazardous waste injection (33%) and hazardous waste manifest (50%).

119. See, e.g., Rose-Ackerman, Consensus, supra note 48, at 1211 (claiming that "regulatory negotiation does not help parties acquire technical or scientific information. Its main value is in clarifying the interests at stake and helping disparate interests find common ground").

120. Phase I, supra note 5, at 15.

121. Id.

122. Id.

123. Id. at 16.

124. Id.

125. Id. at 12:

New information about the specifics of reg neg, about how EPA works, and about the costs of compliance was also reported (each with about 5% of mentions). Others observed that they learned how complex rulemaking at EPA is, the tactics of the Agency and other interests, and they became more aware of the programmatic and policy initiatives underway at the Agency.

126. Id.

127. See, e.g., Rose-Ackerman, Consensus, supra note 48, at 1211.

128. Phase I, supra note 5, at 12.

129. Id. at 18.

130. Id.

131. Id. at 19.

132. Id. The extent of reliance on others varied considerably across participants. For example, big business reported relying on their own resources (46% of mentions) more frequently than environmentalists (13% of mentions). Id. Environmental interests particularly relied on others for information, but other smaller, similarly situated participants, such as suppliers of compliance services relied less on others for information (35% of mentions). Id.

133. Id. EPA covered some of the expenses of some participants during the reg neg process prior to the notice of proposed rulemaking. Help with defraying expenses is specifically authorized by the NRA if a party is necessary to ensure adequate representation of the member's interest and could not otherwise participate. 5 U.S.C. § 568(c), available in ELR STAT. ADMIN. PROC.

134. Phase I, supra note 5, at 19.

135. Id.

136. See Rose-Ackerman, Consensus, supra note 48, at 1209 (claiming that "negotiation is not effective when technical information is needed to resolve factual disputes").

137. Phase I, supra note 5, at 19.

138. Id.

139. Id. at 20. Kerwin and Langbein put this question to participants in reg neg, but not to conventional rulemaking participants.

140. Id.

141. Id.

142. Id.

143. Id.

144. Id.

145. Id.

146. See, e.g., Susskind & McMahon, supra note 35, at 150-51 (outlining the process whereby interests decide to send representatives to negotiate, who in turn try to win support for the proposed rule from parties with the same interests).

147. Phase I, supra note 5, at 20-21.

148. Id. at 21.

149. Id. Most subgroups included people interested in the same issue regardless of their position on the issue. Id.

150. Id.

151. Id.

152. Id.

153. Id. In the full report, Kerwin and Langbein offered a more detailed interpretation of data on how decisions were made and by which level of decisionmaking unit than we do here. For example, they speculated that the initial organization of the negotiation by the facilitator would be influential in determining the relative importance of the different fora for decisionmaking. They claimed that when subgroups were tasked with consideration of technical or scientific issues and development of proposals, they were likely to be the focus of decisionmaking, with the full committee acting as a ratifying body. Conversely, when the subgroups were caucuses of particular interests, the full committee was likely to be the place where the alternative proposals were discussed and compromises explored. Id. at 22. Moreover, Kerwin and Langbein note:

It is clear from the data that there is an association between the composition of subgroups and how decisions are made. On one hand, when those with interests in the same issue (no matter what their position) make up a subgroup, respondents are unlikely to report that subgroups alone made actual decisions (just 10% so report); by contrast, when those on the same side of an issue make up subgroups, 41% report that subgroups alone made actual decisions…. On the other hand, when subgroups reflect similar interests, 75% report that the formal negotiation sessions had a major impact on the proposed rule; when subgroups reflect similar sides, only 56% report that the formal negotiation sessions were a major contribution to the proposed rule.

Id.

154. See, e.g., Coglianese, Is Consensus Appropriate?, supra note 19 (manuscript at 95).

155. Technically the voting rules in a reg neg do not entitle individual parties to block consensus, only to withdraw from the group, thereby jeopardizing consensus. Exiting entitles parties to exercise their right to sue, creating potential obstacles for any consensus that emerges in their absence. Although they may sue in any event, remaining in the committee drastically reduces this risk because, when they join the negotiating committee, they typically sign an agreement not to challenge any consensus that emerges from a group of which they are a part. For example, in the coke ovens negotiation, the parties signed an agreement that said: "Each party agrees not to challenge the rule in court to the extent the final rule and its preamble have the same substance and effect as the recommended rule and preamble concurred in by the committee." Although these agreements leave enough interpretive room to allow parties to sue, and although they are arguably unenforceable, violating them would likely entail significant reputational costs, particularly for repeat players. In theory, then, at least some players (probably larger repeat players like trade associations and national environmental groups) are likely to be seen as indispensable to the smooth promulgation and implementation of the rule. Other participants, including the agency, may want to keep them in the group for fear that the reg neg will fail without them. Thus, influential groups may effectively exert blocking power, creating a holdout problem. At the same time, repeat players who interact in other fora face high reputational costs if they hold out unreasonably, thus the risk of holdout may be highest for smaller one-time players (mavericks) who, ironically, are less crucial to consensus and might be let go without high risk. On the classic holdout problem see Lewis S. Peterson, Who's Being Greedy? A Theoretical and Empirical Examination of Holdouts and Coercion in Debt Tender and Exchange Offers, 103 YALE L.J. 505 (1993); Mark J. Roe, The Voting Prohibition in Bond Workouts, 97 YALE L.J. 232 (1987) (discussing the holdout problem). Reliance in reg negs on loose consensus reduces the holdout problem, as does the complexity of the issues to be negotiated. A holdout problem is most likely to emerge when issues are unidimensional and one party takes an extreme position. As noted above, most reg negs entail numerous and diverse issues.

156. There are other strategies as well. See, e.g., Letter from Marie Kocoshis, President, Group Against Smog and Pollution, to committee facilitator Philip Harter (Oct. 28, 1992) (on file with New York University Environmental Law Journal) (stating that one of the participant groups in the coke ovens reg neg abstained from the final consensus rule but agreed that the rule was an appropriate national standard and declined to challenge it). The interviews revealed that participants did shuttle back and forth between their constituencies and the reg neg group, and mediated between them. See Phase I, supra note 5, at 22. One of the frequently cited purposes of informal discussion was determining whether the participant had the necessary support of his coalition or organization (18% of responses). Phase I, supra note 5, at 23.

157. On strategic judicial voting, see, e.g., Evan H. Caminker, Sincere and Strategic Voting Norms on Multimember Courts, 97 MICH. L. REV. 2297, 2319 (1999).

158. See Phase I, supra note 5, at 23.

159. Id. at 22.

160. Id. at 22-23.

161. Id. at 23.

162. Id.

163. Id.

164. Id. The percentage of participants reporting that informal negotiations had a major impact on the content of the proposed rule was highest for coke ovens (91%) and fugitive emissions (67%) and lowest for the hazardous waste manifest (0%). Id.

165. Id.

166. Id. at 24. Aggregate responses were as follows: major, 22%; moderate, 45%; moderate to minor, 17%; minor, 14%; and none, 1%. Id.

167. Id.

168. Id.

169. See Laura I. Langbein, Responsive Bureaus, Equity, and Regulatory Negotiation: An Empirical View 20 (Apr. 2000) (unpublished manuscript, on file with author) [hereinafter Langbein, Responsive Bureaus]; see also Langbein & Kerwin, Claims, supra note 5, at 612-14.

170. Phase I, supra note 5, at 24.

171. Id.

172. Id.

173. See Langbein & Kerwin, Claims, supra note 5, at 609, exh. 4.

174. Phase I, supra note 5, at 24.

175. Id. at 25.

176. Id. For example, in the minor permit modification negotiation, no one mentioned that the content of the rule reflected the preferences of a party with disproportionate influence. On the other hand, participants in the woodstove and clean fuels reg negs submitted a high proportion of remarks saying that the content did reflect the preferences of a party with disproportionate influence (86% and 63%, respectively, compared to 45% overall). Id.

177. See Langbein, Responsive Bureaus, supra note 169, at 17-18. Given the resource-intensive nature of reg neg, and the perception that its costs fall disproportionately on smaller groups, we would expect smaller participants to report lower satisfaction with reg neg than with conventional rulemaking, so this result is all the more surprising.

178. Language in a rider on EPA appropriations prohibits the agency from funding "intervenors" in agency adjudicatory or rulemaking proceedings. This has been interpreted by the agency to apply to activities by participants in reg negs after the notice of proposed rulemaking is published. See Phase I, supra note 5, at 28. Hence EPA reg neg committees formally cease to function once the agency proposes the rule because EPA cannot provide financial help to any intervenor after that point. Telephone interview with Chris Kirtz, Consensus and Dispute Resolution Team Leader, Office of Regulatory Management and Information, U.S. EPA (Aug. 18, 2000). However, there appears to be at least some confusion about the precise reasons why committees disband when they do. One facilitator cited concern about ex parte communications as another potential reason why an agency might feel compelled to disband negotiating committees. E-mail from Philip J. Harter (Aug. 20, 2000) (on file with New York University Environmental Law Journal). Still, EPA staff reported that they use informal means to continue to communicate with members of negotiating committees. Phase I, supra note 5, at 28. The NRA itself contemplates maintaining committees up to promulgation of the final rule. 5 U.S.C. § 567, available in ELR STAT. ADMIN. PROC. (committees shall terminate upon "promulgation of the final rule under consideration, unless the committee's charter contains an earlier termination date or the agency, after consulting the committee, or the committee itself specifies an earlier termination date").

179. Phase I, supra note 5, at 27.

180. Id.

181. Id.

182. Id.

183. Id.

184. Id.

185. Id. at 27. There is a distinction between post-proposal "communication" and "activity." Activity suggests a broader category of behavior that requires more effort, including filing a comment or writing a letter. Communication is less "active" and includes efforts such as making phone contact.

186. Id. at 26.

187. Id.

188. Comments per reg neg were as follows: woodstoves (180 items of communication and 95 conversations between EPA and outside parties prior to the start of the reg neg; EPA received 71 comments and had 6 external conversations on the proposed rule); clean fuels (33 communications with or from external parties prior to the start of the reg neg and 322 pieces of correspondence, formal comments, and additional items, including hearing testimony, on the proposed rule); fugitive emissions (1 public comment before the reg neg and 45 comments on the proposed rule); coke ovens (81 comments, pieces of correspondence, or conversations prior to or during the period of the reg neg and 66 comments after the reg neg was completed); hazardous waste injection (55 public comments after a suspended reg neg); asbestos (50 public comments that arrived at various points); and minor permit modifications (8 comments and other forms of communication prior to the reg neg and 58 comments on the proposed rule). Id. at 29-30.

189. Id. at 31-32. Participants in the coke oven reg neg reported using the most professional staff hours, with a mean of 8,100. Small businesses reported the highest use of professional staff hours, with a mean of 10,700 hours, followed by EPA at 9,500 hours; and at the low end, reporting between 600 and 1,000 professional staff hours, were business, environmental groups, states, and suppliers of compliance services. Id. at 32. On clerical hours, again, the coke oven reg neg used the most hours and the woodstoves and minor permit modifications reg negs used the fewest, but the differences were not statistically significant. Environmental interests used relatively fewer clerical staff hours; small business used none; EPA use was among the highest (mean = 1,780), but again the differences were not statistically significant. Id. at 32. The pattern of monetary expenditures was extremely skewed. For research and information collection, the bottom third of respondents spent nothing and the middle third spent no more than $ 250, but two respondents reported spending in excess of $ 1 million and six others ranged from $ 50,000 to $ 500,000. The overall mean was $ 82,000 but the median was 0. For legal counsel, two-thirds of the respondents spent nothing, but two respondents reported spending in excess of $ 2 million. The most common support activity expenditure was for travel, which was also highly skewed, with a mean of $ 2,500 and a median of 0. Id.

190. Id. at 32.

191. Id.

192. Some reg negs are more costly than others. The data show that with respect to research and information collection, the hazardous waste injection reg neg was the most costly (mean = $ 600,300), while the hazardous waste manifest reg neg was the least costly (mean = $ 53—probably because it was not regarded as a highly technical reg neg). Id. at 32. With respect to consultants and legal counsel, the coke ovens reg neg was the least costly (mean = $ 1,400), while hazardous waste injection reg neg was the most costly (mean = $ 402,000). Id. at 32-33. Again, because of the small number of observations in some categories these differences are not statistically significant, but the large differences are striking.

193. Id. at 33.

194. Id. ("There were significant differences among the rules on this dimension, with coke oven negotiations commanding, on average, 55% of the participants' organizational resources and hazardous waste manifest, minor permit modifications, fugitive emissions and woodstoves using far less.").

195. Id.

196. Id.

197. See infra Table 1 in Appendix.

198. Phase I, supra note 5, at 37.

199. Id.

200. Id.

201. Id.

202. Id. The risk most frequently reported referred to the problem of negotiating on behalf of a constituency that might or might not agree with the person's decisions under pressure. Id.

203. Id.

204. Id. at 37-38. Kerwin and Langbein used a regression analysis that incorporated nine aspects of the reg neg process, two aspects of the substantive rule and several dummy variables to capture the affiliation of the respondent. Id. at 38. The variables were issues on which the interviewers had gathered information, including such things as "ease of decision to participate," "complexity/messiness of reg neg," or "any party have disproportionate influence." Id. at 38-40. Each variable had an expected relation with the overall evaluation. For example, where it was easier for the participant to decide to join the reg neg, one would expect their overall evaluation to be higher. Where they viewed the reg neg as messier or more complex, one would expect their evaluation to drop. For more detail on the regression variables, see infra Table 2 in Appendix.

205. Phase I, supra note 5, at 41.

206. Id.

207. For the regression results see infra Table 3 in Appendix.

208. Phase I, supra note 5, at 10 (stating that there were also several references by participants to overinclusiveness; interests who were included that did not need to be present at all, some whose participation was excessive given what they contributed to the proceedings, and others who were invited but simply failed to attend).

209. Id. at 41.

210. Id.

211. Id. Complexity increases the probability of agreement with multidimensional issues, we believe, in part because it allows gainers to compensate losers on particular issues. It is hard for the agency itself to make these trades. Although complexity can also reduce satisfaction, overall, participants prefer the reg neg process. See Langbein & Kerwin, Claims, supra note 5, at 602-05.

212. Phase I, supra note 5, at 41-42.

213. Id. at 42. The data on timeliness and litigation was fairly limited in this study but deserves mention as well. While 40% of the respondents felt it took too long to produce the proposed rule, only 17% stated it took too long to produce the final rule. These results varied significantly across reg negs. In fact, in the asbestos reg neg, 27% of responses said the time elapsed in writing the proposal was too short. In only two reg negs (woodstoves and minor permits) did 100% of responses describe the elapsed time as just right, compared with 53% overall. These results clearly reflect pressures felt during the reg neg and frustration at post-reg neg events, but they may also reflect the amount of personal time spent on the reg neg by many participants. In addition, the Phase I results supported the then-conventional wisdom that reg neg avoids successful legal challenge. The asbestos rule was challenged but the court ruled in EPA's favor; the hazardous waste injection rule was challenged and once again EPA prevailed. The rule containing the results of the clean fuels reg neg was challenged but the issues in question were not among those negotiated by the committee. Consequently, this challenge cannot be considered a valid legal attack on the nature of the rule as a product of reg neg.

214. The six EPA conventional rulemakings analyzed by Kerwin and Langbein were the "Third-Third" rule under RCRA, the Superfund Right-To-Know rule, the lead paint notification rule under the Residential Lead-Based Paint Hazard Reduction Act of 1992, the Lead Smelters rule under the CAA, the stormwater rule under the Clean Water Act, and the Petroleum Refineries rule under the CAA. See Langbein & Kerwin, Claims, supra note 5, app. at 630-31. While the number of conventional rulemakings analyzed and number of participants in conventional rulemaking were fewer than the number of reg neg rulemakings analyzed and reg neg participants involved in Phase I, Kerwin and Langbein reported that the sample size in Phase II was "sufficient in both size and diversity for the types of statistical analyses we needed in order to compare the conventional rules as a whole to the negotiated rulemakings as a whole." Id. at 601. They contend that, taken together, the set of conventional rules is "roughly comparable" to the set of negotiated rules. Id. In addition, in their final multivariate analyses they controlled for any important differences. Id. The researchers asked the same questions of both pools of respondents, "with obvious changes where a given question was linked only to one or the other technique." Id. at 602.

215. Id. at 602.

216. Id.

217. Id.

218. See supra notes 34-37 and accompanying text.

219. See, e.g., Harter, Cure, supra note 34, at 31, 110; cf. John Mendeloff, Regulatory Reform and OSHA Policy, 5 J. POL'Y ANALYSIS & MGMT. 440, 455-56 (1986) (arguing that another consensus-based procedure, collective bargaining, establishes a stable, ongoing relationship that promotes compliance with agreements); Steven Kelman, Adversary and Cooperationist Institutions for Conflict Resolution in Public Policymaking, 11 J. POL'Y ANALYSIS & MGMT. 178, 187 (1992) (describing how mediation defendants are twice as likely to meet the obligations imposed on them as are adjudication defendants).

220. See, e.g., Coglianese, Assessing Consensus, supra note 3, at 1326-27.

221. Langbein & Kerwin, Claims, supra note 5, at 605-08.

222. Id. at 608-11.

223. Id.

224. Id. at 602-05.

225. Id. They also examined litigation data and the content of public comments. See id. at 614-15.

226. See Susskind & McMahon, supra note 35, at 151.

227. In theory, because negotiations produce more information among the parties, compared to conventional rulemaking, they could help reduce asymmetries. See id. at 154.

228. See Siegler, supra note 61, at 10651-52 (highlighting the high resource demands of reg neg, and acknowledging that a "considerable" amount is spent on collecting data).

229. See, e.g., Rose-Ackerman, Consensus, supra note 48, at 1216-17.

230. Freeman, Collaborative Governance, supra note 44, at 37.

231. See, e.g., Susskind & McMahon, supra note 35, at 158 (arguing that in reg neg the rule still goes through notice-and-comment rulemaking, and the agency still has the discretion not to promulgate it); see generally Morris P. Fiorina, Legislative Choice of Regulatory Forms: Legal Process or Administrative Process?, 39 PUB. CHOICE (1982); Mathew D. McCubbins & Talbot Page, The Congressional Foundations of Agency Performance, 51 PUB. CHOICE 173 (1986).

232. See McCubbins et al., supra note 46, at 258.

233. See id. at 257-59.

234. See id. at 248-49.

235. Langbein & Kerwin, Claims, supra note 5, at 611-12.

236. Id.

237. Id. at 615.

238. Id.

239. Id.

240. Id. at 618-20.

241. Id. at 618.

242. Id.

243. We do not provide a description of each rule here, but summaries of the eight negotiated and six conventional rules that make up the two groups Kerwin and Langbein compared can be found in Phase II, supra note 5 (manuscript at 4-6); Langbein & Kerwin, Claims, supra note 5, app. at 629-31.

244. Langbein & Kerwin, Claims, supra note 5, at 603.

245. Id. at 604, exh. 1.

246. Id. at 625.

247. Id. at 604, exh. 1.

248. Id. at 603, exh. 2. The difference was 0.5 mentions per respondent. Reg neg respondents also mentioned more dislikes, but the difference is smaller, and not statistically significant. Id.

249. Id. at 607.

250. Id. at 606, exh. 3.1.

251. Id.

252. Id. at 606, exh. 3.2.

253. Id.

254. See Harter, Cure, supra note 34, at 141.

255. See Coglianese, Assessing Consensus, supra note 3, at 1326-27.

256. See Langbein & Kerwin, Claims, supra note 5, at 607 (stating that there was no correlation between learning and satisfaction with the final rule).

257. See JAMES M. BUCHANAN & GORDON TULLOCK, THE CALCULUS OF CONSENT: LOGICAL FOUNDATIONS OF CONSTITUTIONAL DEMOCRACY 93 (1962) (describing the costs of consensus-based decisions).

258. Phase II, supra note 5, at 17. As expected, Kerwin and Langbein also found that reg neg participants engaged in informal contacts; indeed, nearly one-half of negotiated rule respondents reported participating in informal negotiations along with formal sessions. Id.

259. Id.

260. Id.

261. Langbein & Kerwin, Claims, supra note 5, at 604, exh. 1.

262. Id.

263. Phase II, supra note 5, at 17.

264. Langbein & Kerwin, Claims, supra note 5, at 614-15. More to say about the implications of litigation rates and in particular with respect to Coglianese's study is included below. See infra note 376 and accompanying text.

265. Phase II, supra note 5, at 11.

266. Langbein & Kerwin, Claims, supra note 5, at 609, exh. 4.1.

267. Id. at 608.

268. Id. at 606, exh. 3.1.

269. Id. at 620.

270. Id. at 608.

271. Id. at 614.

272. Id. at 609, exh. 4.3.

273. See, e.g., Freeman, supra note 44, at 37 (arguing that in reg neg the agency is accountable since it must still comply with the requirements of the APA); Rose-Ackerman, Consensus, supra note 48, at 1216-17 (arguing that reg neg gives rise to the possibility of excessive delegation of administrative tasks to private interests).

274. See, e.g., Rose-Ackerman, Consensus, supra note 48, at 1216-17.

275. See Langbein & Kerwin, Claims, supra note 5, at 610-11.

276. Id. at 611-12.

277. Id. at 608.

278. Id. at 609, exh. 4.1. Conventional participants were more likely to respond with "no" than were negotiation participants (56% versus 25%), but negotiated rule participants volunteered that much of this participation was "air time" at the negotiating table but not actual influence. Id. at 609, exh. 4.2. When the results are viewed as a response of "yes, there was a party with disproportionate influence" versus a category combining "no" and "yes, but it is participation not influence," the significant difference disappears. Id. at 610.

279. Id. at 609, exh. 4.3.

280. Id. at 609, exh. 4.4.

281. Id. at 609, exh. 4.5.

282. Id.

283. Id. at 613, exhs. 5.1-5.2.

284. Id. at 613, exhs. 5.4-5.5.

285. Id. at 612. Kerwin and Langbein constructed a measure of overall responsiveness to public participation by adding two other measures: the respondent's perceived contribution of the public's participation in the proposed rule (measured on a 1-4 scale ranging correspondingly from major to none) and the respondent's perceived contribution of the public's participation, to changes between the proposed and final rule (measured on the same scale). Id. Only a minority of respondents answered both questions, but the mean score among negotiated rule respondents was only slightly higher than that of conventional participants. Id. On a scale ranging from 2-8 (where 8 was no public influence and 2 was major public influence), the mean for negotiated rule participants was 4.7 and that for conventional participants was 4.2; the half-point difference was not statistically significant. Id. Similar results emerged when the total impact of respondent's own contribution was measured by adding the respondent's assessment of her personal contribution to the proposed rule with her contribution to the change between proposed and final rules. Id. at 612-14. The results revealed slightly higher means for total personal than for total public contributions (indicating less total personal influence than public influence). See id. at 613, exh. 5.6. For negotiated rule participants, the mean total personal contribution score was higher (indicating less influence) than it was for conventional rule participants but the difference was not significant. Id. at 614.

286. Id. at 614.

287. See Susskind & McMahon, supra note 35, at 139-40 (arguing that a subject of reg neg should have many different sides to allow for integrative bargaining); Rose-Ackerman, Consensus, supra note 48, at 1209 (arguing that for reg neg to be successful, the range of choices for participants must be restricted).

288. Phase II, supra note 5, at 13.

289. Id.

290. Id.

291. Langbein & Kerwin, Claims, supra note 5, at 616.

292. Id. at 617, exh. 6.1

293. Id.

294. Id. at 617, exh. 6.2.

295. Id. at 616.

296. Id. at 617, exh. 6.3

297. Id.

298. Id.

299. Id.

300. Id.

301. Id. at 617, exh. 6.3.

302. Id.

303. Id. at 618.

304. Id. at 616-17.

305. See id. at 618.

306. Id. at 617, exh. 6.3.

307. Id. at 618.

308. Id.

309. Id.

310. Id.

311. Id.

312. See id.

313. See id. at 618-19.

314. See id.

315. See id. at 619, exh. 7.1.

316. Id. at 619-20.

317. Id. at 620.

318. Id.

319. Id.

320. Id.

321. Id.

322. Id.

323. Id.

324. See id. at 626.

325. LIND & TYLER, supra note 32, at 176-77. Lind and Tyler have further stated that "one of the most potent determinants of the procedural fairness of a social decision-making procedure is the extent to which those affected by the decision are allowed to participate in the decision-making process through the exercise of process control or voice." Id. at 176.

326. Id. at 186.

327. See Langbein & Kerwin, Claims, supra note 5, at 609, exh. 4.2.

328. See id. at 609, exh. 4.2.

329. Id. at 614.

330. For example, EPA abandoned a reg neg effort to establish a standard to protect farmworkers from pesticides, although it later promulgated a rule. See Harter, Actual Performance, supra note 4. For a table summarizing the reg negs abandoned by all agencies between 1983 and 1996, see Coglianese, Assessing Consensus, supra note 3, at 1274, tbl. 1. EPA has also departed from the consensus rule in a number of instances. See Harter, Actual Performance, supra note 4.

331. For example, the data showed that the ground rules at times came from many sources and on occasion were reportedly not formalized. See Phase I, supra note 5, at 12.

332. See Langbein & Kerwin, supra note 5, at 616. Although some commentators would view the potential for surprise as a negative, we believe that it might be the sign of an open and flexible process, which we consider potentially helpful in solving regulatory problems and designing implementation.

333. See id. at 607.

334. See id. at 605.

335. See Phase I, supra note 5, at 43.

336. See id.

337. See id. at 8.

338. Id. at 26-27. Although reg neg committees typically disband after achieving consensus, agencies nonetheless frequently ensure that advance copies of final rule language go to members. Id. at 28.

339. Id. at 29.

340. Id. at 28.

341. See Langbein, Responsive Bureaus, supra note 169, at 17-18.

342. See Phase I, supra note 5, at 41.

343. See Langbein & Kerwin, Claims, supra note 5, at 625.

344. See id. at 623, exh. 9.

345. Id. at 624-25. This is true even though EPA appears more often to select rules that entail compliance issues for regulation. Id. at 625.

346. Id. at 623, exh. 9.

347. Id. at 624.

348. Id. at 623, exh. 9.

349. Id. at 625.

350. Some research in social psychology suggests that perceptions of procedural justice enhance perceptions of distributive justice, but only on the part of participants in the decisionmaking process. See Laurens Walker et al., The Relation Between Procedural and Distributive Justice, 65 VA. L. REV. 1401, 1415 (1979) (discussing this result in studies of adversarial and nonadversarial decisionmaking procedures).

351. See Arthur T. Denzau & Michael C. Munger, Legislators and Interest Groups: How Unorganized Interests Get Represented, 80 AM. POL. SCI. REV. 89, 103 (1986); see also McCubbins et al., supra note 46, at 249.

352. Clearly, this would require another study. Testing this might require data from observers who were not involved in the reg neg process but who were aligned with the interests of those who were. One might then compare the reactions of those outsiders with the reactions of similar observers of the traditional notice-and-comment process.

353. Although we suspect that outsiders may "free-ride" to some extent on insider assessments of legitimacy, we believe further research is needed on this point.

354. The rates were 33% and 29%, respectively. Langbein & Kerwin, Claims, supra note 5, at 614. Still, the study did not focus on litigation, and cannot resolve the dispute between Coglianese and Harter.

355. Coglianese, Assessing Consensus, supra note 3, at 1334-36.

356. As noted earlier, early proponents of regulatory negotiation argued that reg neg's primary benefits would be improved quality and legitimacy. See Harter, Cure, supra note 34, at 7, 17.

357. Coglianese, Assessing Consensus, supra note 3, at 1278-86.

358. For a detailed analysis of the circumstances surrounding a variety of challenges to negotiated rules, and the argument that such rules are remarkably resistant to challenge, see Harter, Actual Performance, supra note 4 (discussing the haphazard promulgation of the reformulated gasoline rule, and its surprising subsequent resistance to judicial review).

359. See generally Laura I. Langbein & Cornelius M. Kerwin, Implementation, Negotiation, and Compliance in Environmental and Safety Regulation, 47 J. POL. 854 (1985) [hereinafter Langbein & Kerwin, Implementation] (arguing that cost-benefit analyses that assume 100% compliance in assessing the net benefits of rules are inaccurate).

360. Following the negotiated rulemaking over equipment leaks, for example, parties that had participated in the negotiations challenged a section of the rule that was not part of the negotiations. See Harter, Actual Performance, supra note 4, at 46-47; see also Coglianese, Assessing Consensus, supra note 3, at 1304-05.

361. See, e.g., Coglianese, Is Consensus Appropriate?, supra note 19 (manuscript at 102).

362. See id. (manuscript at 101) ("researchers have yet to study systematically whether consensus-based policies elicit greater compliance").

363. LIND & TYLER, supra note 32, at 187. The data also suggest that "sham" participation might have the most serious compliance implications. Id. ("when the rule-maker had induced those subject to the rule to participate in a procedure that ultimately led to the adoption of an unfair decision rule, the rate of compliance was particularly low").

364. Langbein & Kerwin, Claims, supra note 5, at 604, exh. 1. Some scholars seem to think that compliance with conventional rules is already high, implying, from what we can make out, that concern over relative compliance rates for negotiated versus conventional rules ought not preoccupy us. See, e.g., Rose-Ackerman, Consensus, supra note 48, at 1212. Although we are dubious about reports of high compliance, if true, this provides a strong reason why scholars should focus more on rule quality and legitimacy. Research suggests that flexible enforcement strategies yield higher compliance rates than does rigid rule-bound enforcement. See John T. Scholz, Cooperative Regulatory Enforcement and the Politics of Administrative Effectiveness, 85 AM. POL. SCI. REV. 115, 117-19 (1991); see also John T. Scholz & Mark Lubell, Trust and Taxpaying: Testing the Heuristic Approach to Collective Action, 42 AM. J. POL. SCI. 398 (1998) (discussing heuristics that affect willingness to comply). On factors likely to affect variation in levels and speed of compliance generally, see Laura I. Langbein, Estimating the Impact of Regulatory Program Enforcement: Practical Implications of Positive Political Theory, 18 EVALUATION REV. 543 (1994).

365. See DOUGLAS G. BAIRD ET AL., GAME THEORY AND THE LAW, 159-87 (1994) (discussing reputation and repeated games); see also ROBERT R. AXELROD, THE EVOLUTION OF COOPERATION (1984) (discussing the importance of detection and punishment to cooperative regimes).

366. See JOEL A. MINTZ, ENFORCEMENT AT THE EPA: HIGH STAKES AND HARD CHOICES 103-06 (1995) (proposing methods of enforcement that contemplate noncompliance caused by "organizational failure"); see also Langbein & Kerwin, Implementation, supra note 359, at 860; cf. CORNELIUS M. KERWIN, RULEMAKING: HOW GOVERNMENT AGENCIES WRITE LAW AND MAKE POLICY 100-01 (1999) (suggesting that complex rules are inconsistently interpreted and may therefore lead to noncompliance).

367. Such a study would necessarily entail numerous controls to ascertain that the rules are otherwise comparable, and to control for other factors (e.g., agency budget; ideology of president; ideology of pivotal congressional floor and committee voters; ideology, political participation, and economy of locality where enforcement occurs) that previous research has shown to affect agency enforcement actions. See, e.g., W. Kip Viscusi & James T. Hamilton, Are Risk Regulators Rational? Evidence From Hazardous Waste Cleanup Decisions, 89 AM. ECON. REV. 1010, 1011-12 (1999) (emphasizing the effect that political factors can have on administrative decisionmaking); see also B. Dan Wood & Richard W. Waterman, The Dynamics of Political Control of the Bureaucracy, 85 AM. POL. SCI. REV. 801, 804-05 (1991) (exploring the impact of the Congress and the president on the bureaucracy); B. Dan Wood & Richard W. Waterman, The Dynamics of Political-Bureaucratic Adaptation, 37 AM. J. POL. SCI. 497, 500-01 (1993) (identifying the president, Congress, and the courts as the stimuli for the bureaucratic adaptation process).

368. See IAN AYRES & JOHN BRAITHWAITE, RESPONSIVE REGULATION: TRANSCENDING THE DEREGULATION DEBATE (1992) (recommending a more tailored and graduated approach to enforcement).

369. But see Coglianese, Is Consensus Appropriate?, supra note 19 (manuscript at 104) ("While it is true that a successful policy deliberation depends on individuals being motivated to deliberate, a quest for consensus is by no means necessary in order to achieve that motivation."). Coglianese continues: "The best way for an agency to motivate outside groups to participate in policy deliberations seems to be to set realistic expectations, meet them, and demonstrate on a repeated basis that the deliberations have an impact on—even if they do not control—the agency's final decisions." Id. at 15.

370. See Phase I, supra note 5, at 29-31.

371. See Langbein & Kerwin, Claims, supra note 5, at 604, exh. 2.

372. See id. at 605.

373. See, e.g., Coglianese, Is Consensus Appropriate?, supra note 19 (manuscript at 100) (arguing that while reg neg may reduce conflict, it may also be counterproductive).

374. Id. (manuscript at 100).

375. Phase I, supra note 5, at 16.

376. Contra Coglianese, Assessing Consensus, supra note 3, at 1290 (equating litigation and conflict in the reformulated gasoline case).

377. See Coglianese, Is Consensus Appropriate?, supra note 19 (manuscript at 100) (claiming that other means—such as government officials' informing themselves of the interests of the various parties and seeking to craft a policy that addresses them—would be equally effective at reducing conflict).

378. We are aware that this may require Congress to amend the NRA, which provides that committees "shall terminate upon promulgation of the final rule under consideration" or, under some circumstances, at an earlier date. 5 U.S.C. § 567, available in ELR STAT. ADMIN. PROC.

379. We believe that a group consultation of this kind would produce different benefits than those produced by one-on-one consultation with regulated entities.

380. This recommendation is not inconsistent with the view that agencies ought to seek balanced reg neg committees on which all of the important interests have representation. See Harter, Fear of Commitment, supra note 81, at 1404-06 (identifying the goal of empanelling a committee with sufficient diversity that all of the relevant issues will be raised and discussed).

381. See Jeffrey T. Polzer et al., The Effects of Relationships and Justification in an Interdependent Allocation Task, 2 GROUP DECISION AND NEGOTIATION 135, 145 (1993) (concluding that "friends are willing to forego resource gains … to maintain equality between themselves and their friend").

382. See Gregory A. Johnson, Organizational Structure and Scalar Stress, in THEORY AND EXPLANATION IN ARCHAEOLOGY: THE SOUTHAMPTON CONFERENCE 389 (Colin Renfrew et al. eds., 1982). Commenting on Johnson's research, Fort and Noone have argued that although it concerns the organization of pre-historic groups, his conclusions are "fundamental to any understanding of human social organization; his findings are as applicable to contemporary as to ancient populations, and provide insight into alternative, nonhierarchical organizations for large groups." Timothy L. Fort & James J. Noone, Banded Contracts, Mediating Institutions, and Corporate Governance: A Naturalist Analysis of Contractual Theories of the Firm, 62 LAW & CONTEMP. PROBS. 163, 210 (1999).

* Value is on a scale from -5 (worst rule possible) to +5 (best rule possible).

* Scale is from -10 to +10

* These Ns include multiple mentions.

* These Ns include multiple responses.

* These Ns include multiple mentions.

** Difference of means = 138, p < .49 when EPA/not EPA dummy held constant.

31 ELR 10811 | Environmental Law Reporter | copyright © 2001 | All rights reserved