Copyright 1986 by the Georgetown Law Journal Association; Henry H. Perritt, Jr
August, 1986

NEGOTIATED RULEMAKING BEFORE FEDERAL AGENCIES: EVALUATION OF
RECOMMENDATIONS BY THE ADMINISTRATIVE CONFERENCE OF THE UNITED STATES
74 GEO1. L.J. 1625

Henry H. Perritt, Jr. [FNa1]

INTRODUCTION AND SUMMARY

  In 1982, the Administrative Conference of the United States (ACUS)   [FN1] issued Recommendation 82-4, [FN2] encouraging the use of negotiated rulemaking procedures under federal agencies. The concept of negotiated rulemaking arose from dissatisfaction with notice and comment and hybrid rulemaking pursuant to the Administrative Procedure Act (APA) [FN3] due to their increasingly adjudicatory and adversarial character. In 1985 ACUS issued further recommendations after several agencies had tried negotiated rulemaking.

  Negotiated rulemaking is a realistic alternative to adversarial administrative procedures. The technique permits affected interests to have greater control over the content of agency rules while ensuring fairness and balanced participation. It also permits agencies to obtain a more accurate perception of the costs and benefits of policy alternatives than agencies can obtain by digesting voluminous records of testimonial and documentary evidence presented in adversarial hearings.

  Between the time ACUS Recommendation 82-4 was issued and the end of 1985, federal agencies used negotiated rulemaking four times. [FN4] The Federal Aviation Administration (FAA) used negotiated rulemaking to develop a new flight and duty time regulation for airline flight crews. The Environmental Protection Agency (EPA) used negotiated rulemaking to develop proposed rules on non-conformance penalties for vehicle emissions and on emergency exemptions from *1628 pesticide regulations. The Occupational Safety and Health Administration (OSHA) encouraged labor, public interest, and industry representatives to negotiate a standard for occupational exposure to benzene. The benzene negotiations did not result in agreement among the parties on a proposed rule, but the other three negotiations did result in at least partial agreement, producing proposed rules based on the negotiations. Three of the rules have been promulgated in final form.

  Other agencies also have shown an interest in negotiated rulemaking, and several other major negotiated rulemaking efforts are under way. To guide the use of negotiated rulemaking, the ACUS published an additional set of recommendations in December, 1985. [FN5]

  The 1982 ACUS recommendations address negotiated rulemaking within the legal framework of the APA and judicially developed administrative law concepts. The 1985 ACUS recommendations take the next step, addressing the dynamics of the negotiation process in the rulemaking setting. Together, the two ACUS recommendations present a framework for planning a rule negotiation. This article summarizes the political and legal developments that led to the use of negotiated rulemaking, briefly describes the four completed experiences with the negotiation process, and explains the justification for the 1985 ACUS recommendations. The article is largely drawn from my report to the ACUS based on an investigation of the four completed rule negotiations. [FN6]

  The article is divided into nine parts. Part II reviews the evolution of the negotiated rulemaking concept, emphasizing the proposal Philip J. Harter presented in 1982 on negotiated rulemaking. [FN7] Part II also presents conceptual models drawn from political science and dispute resolution literature that lead to hypotheses as to the preconditions for successful negotiation. Part III reviews the benzene negotiations. [FN8] Part IV reviews the negotiations of the FAA. Part V *1629 reviews the two negotiations of the EPA. Part VI summarizes other past experiences and future agency plans with respect to negotiated rulemaking. Part VII compares the four negotiations. Part VIII discusses major legal issues affecting future use of negotiated rulemaking. Part IX explains the basis for the 1985 ACUS recommendations.

  It is important to view both the 1982 and the 1985 recommendations of the ACUS as a guide to issues to be considered rather than a formula to be followed. Negotiation is intrinsically a process that cannot be specified entirely in advance. Accordingly, what will 'work' in a particular case depends on a number of factors: substantive issues, perception of the agency's position by affected parties, relationships among the parties, authority of party representatives in the negotiations, negotiating style of the representatives, divergence of views within each constituency represented, and skill of agency personnel and mediators. [FN9] Some of these variables almost certainly will change several times during the negotiations. An agency cannot expect that the pattern followed successfully by another agency, or even by itself on another issue, can be transplanted without modification to another negotiation.

  The article explains that the APA need not be amended to provide for negotiated rulemaking. Amending the APA would risk destroying the flexibility it provides to adapt the negotiation process to the needs of different regulatory situations. Moreover, the four agency experiences show that the Federal Advisory Committee Act (FACA), [FN10] as interpreted by the sponsoring agencies and participants, was not a serious impediment to effective negotiations. The purpose of the FACA is satisfied when a properly balanced rulemaking negotiation is conducted, and the statute should not be interpreted to impose additional requirements that may jeopardize the success of negotiation. Under current judicial and agency interpretation of the FACA, caucuses and other working group meetings may be held in private when necessary to promote an effective exchange of views. Agencies should not be deterred from using negotiated rulemaking by a perception that the negotiating group will be an 'advisory committee' under the FACA. Some uncertainty can be reduced, however, if the General Services Administration amends its regulations to make it clear that meetings of caucuses and subgroups may be closed.

  Perhaps the most important insight to be gained from an assessment of the four completed negotiated rulemaking efforts is that an agency sponsoring a negotiated rulemaking should take part in negotiations. Negotiations are unlikely to succeed unless all parties are continually motivated by a perception that a negotiated rule would be preferable to a rule developed under traditional processes. As an incentive to negotiate an agreement, the agency should help *1630 create realistic expectations of the consequences of not reaching a consensus. Agencies must be mindful throughout the negotiations of the impact that agency conduct and statements have on party expectations. The agency may need to communicate with participants, perhaps with the assistance of a mediator or facilitator, to ensure that each has realistic expectations about the outcome of agency action in the absence of a negotiated agreement.

  Negotiated rulemaking is only one of several alternative dispute resolution techniques that can be used by administrative agencies. Negotiated rulemaking is designed to facilitate resolution of interest disputes. [FN11] Other techniques are more suitable for rights disputes, such as arbitration, fact-finding, and mediation. [FN12]

II. LEGAL AND POLITICAL BACKGROUND

A. HISTORY OF THE NEGOTIATED RULEMAKING IDEA

  Negotiated rulemaking [FN13] is a process for resolving interest disputes, similar in many ways to the legislative process. Negotiated rulemaking emerged as a distinct administrative law concept in the late 1970's in reaction to the unsuitability of notice and comment and hybrid rulemaking [FN14] processes for making quasi-legislative administrative agency decisions.

  New administrative processes arose from increased use of administrative litigation to address problems previously dealt with in markets, through private contractual negotiation, [FN15] or by elected representatives in legislative assemblies. *1631 The increased use of administrative regulation to address these problems arose from a combination of increased government intervention to correct market imperfections [FN16] and the impracticability of legislative assemblies taking on the resulting burden of decisionmaking.

  When rules for societal conduct are made in private markets or in representative assemblies, negotiation is relied upon. [FN17] Negotiation occurs as part of the legislative process in a representative assembly at two levels: first, during the process of electing representatives and second, during the interaction among the representatives in making the compromises necessary to pass laws. At both levels, no single interest or constituency gets all it wants; each must rank its demands and trade off one for another.  [FN18] There is an incentive to find common ground to form a majority to elect a favored candidate, or to pass desired legislation.

  The migration of problem solving to administrative agencies from markets and legislatures made negotiation more difficult. Agencies make rules because the legislature has left unresolved certain disputes among conflicting interests. In exercising rulemaking responsibility, agencies lack many attributes that facilitate accommodation among conflicting interests. For example, the accommodation and compromise that result from the election of representatives is missing, [FN19] which forces agencies to deal with a broad range of atomistic interests [FN20] and prevents them from benefiting from some subordinate aggregation. In addition, there is no assurance that even a multimember agency represents the major conflicting interests on any given issue.

  Clearly, the legislative delegation of details to specialized agencies arose out of necessity. However, because an administrative agency lacks the institutional attributes that make private accommodation an integral part of a legislative process, it must find other ways to strike a balance between competing views in developing a rule that will be reasonably satisfactory to those bound by it. In addition, the agency is obligated by law to pursue statutory objectives regardless of the views of affected interests.

  The conceptual basis of administrative law has shifted in recent years in search of ways to facilitate accommodation of conflicting interests.  [FN21] Original administrative law models emphasized judicial review to keep agencies within statutory bounds and decisionmaking procedures designed to promote the accuracy, rationality, *1632 and reviewability of agency decisions. [FN22] Increasingly, administrative law has served to provide a surrogate political process to ensure the fair representation of a wide range of affected interests in the administrative decisionmaking process. [FN23] The conceptual shift was motivated by a growing recognition that administrative procedures modeled on judicial processes do not fit the needs of administrative rulemaking. The main problem is that the procedures available to agencies largely have remained static. They were designed around the original conceptual models and are ill suited to a surrogate political process.

  A major part of the mismatch between administrative procedure and the decisionmaking requirements of delegated legislative power arose from the failure to distinguish rights disputes from interest disputes. [FN24] Adjudication is designed only to deal with rights disputes. [FN25] Rights disputes involve application of preexisting legal standards or rules of decision to facts determined by the adjudicator. Interest disputes, in contrast, are characterized by the absence of preexisting rules for decision. Resolution of interest disputes requires parties to work out the rules according to an accommodation of their interests. While the political process and private contractual negotiation are well suited to resolving interest disputes, the adjudicatory process is ill suited to resolve these disputes.  [FN26] Adjudicatory procedures nevertheless were superimposed on the rulemaking process. [FN27] As agencies were delegated more responsibility for legislating, the need for a process suited to interest disputes increased.  [FN28]

  Conceptual development of new administrative processes accelerated in the 1970's. Professor Stewart, writing in 1975, considered some of the alternative procedures available to agencies given responsibility for a surrogate political process. [FN29] He concluded that all of the obvious possibilities for application within traditional procedural frameworks were seriously flawed.  [FN30] He observed that agencies themselves are likely to afford certain interests disproportionate influence, and that the courts are ill suited to force an alteration in the balance of interests. [FN31]

  This led Stewart to explore more 'explicitly political' mechanisms for interest representation: [FN32]

    Since, in the absence of authoritative rules of decision, the resolution *1633 of the conflicting claims of a large number of competing interests is essentially a political process, a solution to the problems raised by the transformation of administrative law into a system of interest representation might better be achieved by a more direct and explicitly political scheme for securing the representation of all relevant interests affected by administrative decisionmaking. . . . Policy would result from a process of bargaining among the representatives of affected interests. [FN33]

  Stewart considered two ways to achieve better representation of interests: popular election of agency members, or their appointment for specific terms by private organizations designated by Congress. [FN34] He concluded that neither alternative had any real prospect of adoption in the foreseeable future. [FN35]

  Near the time that Stewart's article appeared, academic and policy currents converged, identifying negotiation as a process that should be examined as an alternative to adjudicatory litigation in making administrative rules. The policy currents were much influenced by John T. Dunlop, Secretary of Labor from 1974-1975and before that head of President Nixon's wage and price controls program. As much as anyone in the United States at the time, he was expert on the process of negotiation as a means of resolving workplace disputes. As director of wage and price controls, he administered an extremely broad delegation of rulemaking authority from Congress, [FN36] which struggled to conform to restrictions traditionally imposed by administrative law.

  Secretary Dunlop set a high priority on finding new ways to address hotly disputed issues within the jurisdiction of these regulatory agencies. He authored a paper identifying eleven problems with the existing approach to regulation and then suggested that 'the parties who will be affected by a set of regulations should be involved to a greater extent in developing those regulations.' [FN37] The paper offered as a useful example the method Dunlop used to implement section 13(c) of the Urban Mass Transportation Act, [FN38] which precluded distribution of certain mass transit subsidy funds until the Secretary of Labor certified that employees would not be affected adversely by the federally funded activities:

    Rather than prepare regulations, the Department brought together union and transit representatives and got them to prepare a three year agreement as to what protection employees should receive as a consequence of the federally funded activities. The Department mediated and provided technical assistance, helping to create the standards to apply to individual cases presented to it. [FN39]

Dunlop also sought to mediate an agreement between the steelworkers union and the steel industry on a health standard for coke oven emissions. [FN40]

  *1634 Dunlop's ideas required more conceptual development before they were practical. Professor Stewart in 1976 commented on the ideas advanced in Dunlop's The Limits of Legal Compulsion:

    I am very skeptical as to whether the collective bargaining type of alternative suggested will really work in very many areas. Most types of governmental regulation definitely do not involve the situation which characterizes collective bargaining--two more or less well-recognized groups with a mutual interest in a negotiated compromise (as well as objectives that conflict in part). Much governmental regulation is designed to protect loosely scattered, disorganized interests held by individuals--such as those of consumers or persons concerned with environmental quality. Normally these interests do not have authoritative spokesmen who are in a position to reach binding compromise agreements with the regulated Industry on matters of policy. The only leverage which such disorganized interests often have (apart from seeking champions and publicity in the legislature) is to attempt to goad the agency into action on their behalf and, failing administrative protection, recourse to the judiciary. The 'legal game-playing' [a phrase used in the Dunlop paper] may be an important and perhaps necessary tool enabling interests to protect themselves, and I am very unclear as to how the collective bargaining model could be applied in such situations. [FN41]

  Dunlop's idea of using negotiation as a regulatory procedure responded to Stewart's exploration of explicitly political processes. It was another means to permit political bargaining not considered explicitly in Stewart's article. The negotiation idea nevertheless presented difficulties, especially concerning the representation process.

  These difficulties were addressed by Philip J. Harter, under sponsorship of the ACUS. Harter was motivated to define the regulatory negotiation idea further because he thought discussions of the negotiation process at the Department of Labor and in congressional hearings [FN42] failed to give sufficient attention to the relationship between negotiation and the APA.

  Harter wrote a law review article [FN43] in which he carried the Dunlop concept further, addressing some of the concerns raised by Stewart's letter. Harter reviewed developments leading to Stewart's conclusion that the interest representation model had supplanted the traditional model of administrative law. [FN44] Harter concentrated on procedural issues, noting that under the hybrid rulemaking *1635 concept, [FN45] agencies had become umpires bound by the record. Under this concept, the rulemaking record was the vehicle of ultimate interest group control over agency decision. [FN46] The record- generating procedures focus on facts, however, while theoretically reserving to the agency the unilateral right to make policy. [FN47] Adjudicatory procedures for developing a record in hybrid rulemaking do not permit direct interest group interaction on policy issues.

  The procedural inadequacy of hybrid rulemaking led Harter to focus on practical alternatives to the fact oriented adjudicatory procedure. He found examples of such alternatives in the Consumer Product Safety Commission 'offeror process,' the National Institute of Building Sciences, innovations by Secretary Dunlop, ad hoc efforts involving dam building and uranium mine siting controversies, sequential negotiations frequently involved in formulating notices of proposed rulemaking safety and health consensus standards developed by private groups, the National Coal Policy Project, site specific environmental negotiation, and negotiated settlements of lawsuits challenging EPA regulatory decisions. [FN48] He sought to synthesize from these examples criteria for processes Stewart presumably would acknowledge as 'explicitly political.' [FN49] These criteria could be used in generalizing the use of explicitly political processes to improve administrative agency decisionmaking.

  Formulation of ACUS Recommendation 82-4 proceeded contemporaneously with Harter's article. The ACUS committee charged with reviewing the proposed recommendation was well balanced, representing a broad range of interests. Agency interest in the subject matter of the recommendation was intensified by the introduction of S. 1601 [FN50] at the time the recommendation was being considered. S. 1601 provided for the establishment of regulatory negotiation committees to allow direct participation of affected interests in federal rulemaking. The bill was circulated to federal agencies for their comment through the Office of Management and Budget (OMB) legislative clearance process. *1636 The necessity of formulating a position on S. 1601 undoubtedly heightened agency interest in the negotiated rulemaking concept.

  Recommendation 82-4 was adopted by the ACUS with little controversy. The recommendation suggests that:

    Agencies should consider using regulatory negotiation, as described in this recommedation, as a means of drafting for agency consideration the text of a proposed regulation. A proposal to establish a regulatory negotiating group could be made either by the agency (for example, in an advance notice of proposed rulemaking) or by the suggestion of any interested person. [FN51]

  The four efforts reviewed in parts III through VI of this article show that negotiated rulemaking is a practical alternative to notice and comment or hybrid rulemaking, and that Recommendation 82-4 is basically sound. ACUS Recommendation 85-5, developed in light of the four efforts, is discussed in part IX.

B. THEORETICAL FOUNDATION

  Implementation of the negotiated rulemaking concept in practice requires consideration of the characteristics of the negotiation process, interest groups, and regulatory programs.

1. Dynamics of Regulatory Negotiation

  Notwithstanding efforts to impose adjudicatory procedures on the formulation of regulatory policy, regulatory program decisions are made in a political environment. [FN52] The statutory framework of any program is as much a function of the political process that shaped it as it is a function of objective economic or scientific processes. In developing and pursuing a regulatory strategy within this statutory framework, an administrator is unlikely to pursue a strategy that he knows will enrage those groups most influential in the Congress, with the press, or with his superiors in the executive branch, regardless of what 'rational' analysis says about the merits of various options.

  Regulatory negotiation is but one mechanism to accomplish political accommodation. Negotiation will succeed only when persons able to use other processes have an incentive to participate in negotiations and to reach negotiated agreement. Incentives operate at several different levels: at the level of the negotiation itself, and at lower levels, where negotiations within constituencies are necessary to produce party positions. The best way to understand incentives to negotiate is first to consider the viewpoint of a hypothetical, monolithic party. Having described the incentives for this hypothetical party, one then can overlay complications that influence real world regulatory negotiations, especially complications inherent in intragroup interest aggregation.

  A useful conceptual structure for understanding incentives to negotiate is the one offered by Professors Fisher and Ury in their book on the negotiation process. [FN53] They explain that the participation of any party to a negotiation will be guided by that party's 'Best Alternative to Negotiated Agreement' (BATNA). If a party's BATNA is superior to what can be obtained in negotiation, the party will not participate. [FN54]

  For potential participants in a regulatory negotiation, BATNAs are determined by perceptions of what the agency will do in the absence of a negotiation.  [FN55] A rational, monolithic party will participate in regulatory negotiation only if it perceives the probable negotiation outcome to be superior to its BATNA, determined by the party's estimate of probable unilateral agency action. Different parties are likely to have different BATNAs because they predict the unilateral agency outcome differently, or because they have different predictions of the cost impact and benefit of agency action.

  The BATNA determined participation incentive is not invariant; it likely will change over time for each party, as the party gets additional information about the agency's intentions. Even more important, other negotiators, neutral mediators or convenors, and the agency itself can influence party BATNAs, and hence party incentives to negotiate and to agree.

  The most appropriate analogy to a regulatory negotiation is not a traditional labor-management negotiation, where BATNAs are determined by each party's assessment of the opponent's ability to inflict injury or to offer rewards. Instead, the appropriate analogy is to civil litigation settlement negotiation in which party predictions of what a nonparty, the judge or jury, will do determine BATNAs. In regulatory negotiations, as in settlement negotiations, the third party decisionmaker can influence party perception of likely outcome in the absence of a negotiated settlement. In other words, the agency or judge changes BATNAs by what she says about her intentions.

  This model suggests that regulatory negotiations are most likely to be successful when the agency (or some other credible source) persuades each potential participant that unilateral agency action has undesirable consequences for that participant. Lower BATNAs mean greater incentives to negotiate a solution.

  This analysis yields the first hypothesis for effective regulatory negotiation: parties will negotiate only if they perceive the outcome of unilateral agency action to be worse for them than what is attainable in the negotiation. The hypothesis has three corollaries. Regulatory negotiations are more likely to be successful if: (1) the parties agree on what the outcome will be in the absence of negotiations; [FN56] (2) the parties disagree on what the outcome will be in the absence of negotiations and are all pessimistic rather than optimistic; and (3) the *1638 agency actively influences party perceptions of BATNAs, emphasizing to each party the undesirable consequences of unilateral agency action in terms relevant to each party. [FN57]

  The preceding analysis of BATNA driven choices about participation has assumed that parties are monolithic: that parties affected by regulatory decisionmaking behave like rational individuals dealing with a single estimate of risk. Real parties do not behave this way. Real parties are represented by individuals, expressing the views of a group of constituents who usually have divergent views. The relations within constituency groups and between constituency groups and individual negotiators complicate the regulatory negotiation dynamics.

  Experienced mediators know that three agreements are necessary to any successful two-party negotiation: (1) an agreement between negotiator A and his constituents; (2) an agreement between negotiator B and her constituents; and (3) an agreement between negotiators A and B. [FN58] Agreements (1) and (2) can be called intraparty agreements. Frequently the most difficult mediation job involves achieving the intraparty agreements rather than achieving the negotiator-negotiator agreement.

  The intraparty problem is more difficult in regulatory negotiations than in labor negotiations, because the parties to regulatory negotiation are likely to be ad hoc groups or coalitions without formal processes worked out for internal decisionmaking. Ordinarily in regulatory negotiation, there is no equivalent of the 'exclusive representation' principle from the law of collective bargaining [FN59] available to bind constituents to the position taken by a group representative. At any point a constituency may disavow its putative representative in the negotiations, take its own position on matters under negotiation, present formal positions to the agency inconsistent with positions taken by its representative, or sue to have the negotiated rule set aside by the courts. Some risk of this occurring probably is inevitable, but regulatory negotiation cannot be successful unless a way can be found to resolve as many of these potential intraparty disagreements through a representation process in the negotiations instead of outside the negotiations. When representatives must make compromises at the bargaining table, the intraparty problem becomes worse.

  It is important for someone involved in the negotiation process, either the representatives themselves, the mediator or convenors, or the agency personnel, to be adroit at diagnosing intraparty problems and working creatively to facilitate intraparty agreement. Someone familiar with the internal structure and decision processes of institutions such as labor unions, corporations, and public interest groups involved in a particular negotiation can be invaluable. Even more valuable would be a mediator who already knows the key decisionmakers within a particular constituency, and thus is trusted to some degree by them.

  *1639 Individual group representatives in regulatory negotiations incur personal risk by participating and greater personal risk by reaching agreement. Any party representative who sits down with an adversary and agrees on a regulation is responsible in some measure for the regulation. If, on the other hand, the same party representative presents positions in the traditional administrative and judicial processes, the representative is not responsible for the outcome to the same degree; the decisionmaker, agency, or judge can be blamed, and the representative can point to the purity of his or her advocacy as proof of aggressive pursuit of constituency objectives.

  Personal risk is a difficult problem to deal with, and it always will remain in the background as a continuing bias against meaningful regulatory negotiation. Optimally, other participants, mediators or convenors, and the agency will recognize the source of the risk of and be alert to opportunities to reduce the risk of participation--compared with the risk of nonparticipation--on an ad hoc basis for each individual representative. As the undesirable consequences of unilateral agency action are communicated to constituents of representatives, the relative risk of participation is reduced. Periodic constituency meetings may be desirable, [FN60] and full reporting to constituents as negotiations proceed is essential. In addition, selecting negotiators with strong power bases can serve to permit freer exploration of compromises.

  Moreover, it is important to shield the give and take among negotiation participants from publicity. Fragmentary or inaccurate reports of positions taken or compromises considered by their representatives may alarm constituents, endangering the minimum level of constituent support necessary to make the participation of any representative meaningful.

  The choice between open and closed meetings among negotiators, however, also bears on the relative power of different kinds of interest representatives. Public interest groups frequently compensate for small financial and staff resources by mobilizing the press to influence public and congressional opinion. It is easier to accomplish this mobilization when the deliberations of the negotiating group are visible to the press. It is harder to do when the meetings not only are closed to the public, but also when the negotiators pledge nondisclosure. Thus the legal requirement of open meetings may result in a subtle enhancement of public interest group power, even when no outsider comes to the meetings or tries to publicize committee activities. More generally, open meetings benefit groups possessing less influence with agency decisionmakers. The decision whether to open negotiation to public scrutiny bears on the question of who gets to be included in formulating agency policy.

2. Characteristics of Interest Groups

  Political decisions in a complex society are made primarily through the interaction of interest groups. [FN61] Legislative action occurs when the balance of political *1640 power favors change. [FN62] The balance of political power generally is determined by the strength and intensity of feeling of groups within society perceiving that they have similar interests on a particular subject. [FN63]

  Interest groups arise to provide economies of scale in the exercise of power, and to permit individual concerns about particular issues to be focused within representative institutions. [FN64] The proposition that policy decisions are made by group interaction raises both macro and micro issues. The macro issues relate to the interaction among groups. The micro issues relate to interaction within groups.

  The macro issues raise questions of group power, issue maturity, and intensity of feeling. One commentatorhas drawn the following general conclusion about issue maturity: 'The objective must have been the subject of sufficient political debate so that the groups interested in it and opposed to it can be identified, their positions and relative strengths evaluated, and potential sources of support have time to develop.' [FN65]

  Issue maturity plays an important role in the development of how intensely different groups feel about a particular issue and how strongly they prefer different alternatives. Issue maturity also ensures that a range of alternatives has been formulated for consideration.

  Intensity of feeling is an important variable in the calculus of public opinion. [FN66] The influence of a particular group or faction, which combined with the influence of other groups or factions determines popular will, is a function of numerosity and intensity of feeling. [FN67] Transaction costs reduce the desire of interest groups to have intense involvement in the full range of political decisions. [FN68] In other words, the cost-benefit ratio for participation in policy formulation is unfavorable when a potential participant is indifferent among outcomes. [FN69]

  Microanalysis of interest groups focuses on the representation, or interest aggregating function, that interest groups perform. [FN70] It also permits consideration of the ways in which interest groups function to intensify member interest in particular issues, to formulate concrete alternatives, and to articulate member positions. [FN71]

  Interest groups have overcome the fragmentation, transaction-cost, and free- rider difficulties impeding group action through 'organizational entrepreneurship.' [FN72] Especially when the issues involved are complicated, group members may defer almost entirely to the decisions of group representatives. This deference can make negotiations involving such representatives more fruitful. Trade unions or trade union federations dealing with complex technical regulatory disputes present an example of this phenomenon. Rank and file members have a low level of awareness of technical issues, and they tend to defer to a handful of staff experts.

  On the other hand, some public interest groups may prefer short term litigation victories to negotiation. Greater publicity associated with a dramatic victory and extreme statements made in litigation tend to facilitate fund raising and other facets of membership support. These groups tend to be less risk averse than business groups, and they therefore may prefer the all- or-nothing characteristic of litigation rather than accommodation. [FN73] In addition to lower risk aversion, public interest groups and trade unions may perceive that adversarial procedures better compensate for scarce technical resources than negotiation. [FN74] Careful selection of test cases in court permits targeting of scarce resources on key legal issues rather than dissipating them on complex factual disputes. [FN75]

  These are only tendencies of public interest groups. Obviously the most effective consumer groups do more than litigate; many participate regularly in the legislative process, which always involves negotiation and accommodation. For example, the consumer groups involved in the negotiated rulemaking efforts addressed in this article participated actively and in good faith. [FN76]

  Because large firms are more risk averse than small firms or organizational entrepreneurs, a negotiated resolution of a regulatory dispute is likely to be more attractive to interest groups dominated by a few large firms than to public interest groups or other groups with more fragmented membership.  [FN77] Negotiation, however, while attractive, may be more difficult because the large constituents have the resources to develop idiosyncratic positions and priorities, making intragroup compromise more difficult than it is in groups in which constituents defer to their representatives.

3. Characteristics of Regulatory Programs

  Interest groups interact with regulatory programs. Indeed, interest groups arose in part because of increased government regulation. [FN78] The nature of the interaction is determined in part by the nature of the regulatory program, because the nature of the program affects the intensity of interest group feeling on *1642 regulatory issues. [FN79]

  Professor James Q. Wilson offers a classification of regulatory programs based on the incidence of costs and benefits likely to influence intensity of group feeling. [FN80] He suggests that programs fall into three categories. Programs in Wilson's first category concentrate their benefits on a small group and distribute their costs over wide sectors of the population. Economic regulation of railroads and airlines, milk prices and taxicabs fall into this category. Programs in Wilson's second category concentrate both benefits and costs on a small group. Regulatory programs relating to labor- management relations fall into this category; labor or management benefits at the expense of the other. Wilson's third category encompasses most recent consumer and environmental protection and health and safety regulation. Here, benefits are diffused over large parts of the population and costs are concentrated on relatively narrow sectors. As Wilson points out, the development of policy is particularly difficult with respect to the third category of regulation, because the number of transactions subject to the regulation is likely to be far greater than in the first or second categories: a few hundred license applications before the Federal Communications Commission (FCC) compared with safety and health practices in nine million workplaces. One can add to Wilson's list a fourth category. This category includes programs whose costs and benefits both are diffused. An example would be automobile emissions device inspection programs.

  Programs in Wilson's second category are better candidates for negotiation than programs in the fourth category, because it is easier to mobilize interest representatives for the bargaining process when the interest groups are few in number and narrow in scope. Moreover, programs with narrow impact are less likely to attract intervention by Congress, the press, the White House, and the public. [FN81] Between the extremes represented by the second and fourth categories, Wilson's first and third categories present intermediate levels of difficulty in organizing interest representatives for regulatory negotiation.

C. CRITERIA FOR NEGOTIATED RULEMAKING

  ACUS Recommendation 82-4 offers criteria to select subjects for negotiated rulemaking. The recommendation was drafted for the conference by Philip J. Harter. Harter's contemporaneous law review article addresses the same criteria in somewhat greater detail than the recommendation. Harter dissected the negotiation process into six distinct facets:

    (1) assembling the negotiators; [FN82]

    (2) the negotiations themselves; [FN83]

    (3) achieving consensus; [FN84]

    (4) the means of reporting consensus to the agency; [FN85]

    *1643 (5) agency action; [FN86] and

    (6) judicial review. [FN87]

Finally, he articulated a set of nine principles that are embodied almost verbatim in ACUS Recommendation 82-4. [FN88]

  Harter recognized that people will negotiate only as long as they believe negotiations will resolve their dispute in a manner more favorable to each of them than other dispute resolution techniques. [FN89] The acceptability of negotiation as a dispute resolution process is determined by the relative power of the interested parties. [FN90] Harter identified four sources of power:  [FN91]

    (A) the ability to use a set of preexisting criteria to structure decisionmaking; [FN92]

    (B) availability of some other formal process for decisionmaking; [FN93]

    (C) the ability to gain access to these alternative decisional processes in a way that a party could use its own resources most effectively; [FN94] and

    (D) the power to delay a decision by any of these means.

  The relative attractiveness of negotiations will be influenced by these aspects of power. [FN95] Harter offered the following criteria to define situations where negotiation would be most effective. [FN96] He stressed that he did not envision mechanical application of the criteria or satisfaction of every criterion. [FN97]

1. Countervailing Power [FN98]

  Each party must have power to affect the decisional outcome. This can flow from the capacity to influence the legislature, the ability to run an effective public relations campaign, substantial litigation resources, or any other way of obtaining an outcome favorable to the party, or from inflicting costs on opponents, in another forum. Negotiation will be effective as a decisional process only if no *1644 one party has power sufficient to overwhelm the others. [FN99] Increased power on one side, however, strengthens incentives for opposing sides to seek a negotiated solution.

2. Limited Number of Parties [FN100]

  It is difficult to negotiate when a large number of people and demands must be accommodated. 'Pure' negotiation as a method for producing a new tax bill, for example, would not work. [FN101] The number of people whose interests are involved would probably be so large so as to preclude effective communication. [FN102]

3. Mature Issues [FN103]

  The issues must be readily apparent and the parties must be ready to decide them. If information is lacking or the parties are still establishing their positions, [FN104] negotiation cannot be utilized because the parties do not know what their positions are or what compromises they are prepared to make. The importance of issue maturity also has been recognized by other students of political behavior. [FN105] Issue maturity is significant even when disputants are individuals. Each disputant must have some time to think about her position on a new question or proposal. Issue maturity is far more important when disputants are groups. Constituency positions must be determined, and this requires aggregation and trading off within the group.  [FN106] The more complex the issue, and the more novel the possible solution, the longer it will take for this process to occur.

4. Inevitability of Decision [FN107]

  There must be pressure for resolution of the matter. [FN108] Effective negotiation requires compromise that involves making concessions. Most people are reluctant *1645 to make concessions until they are forced to do so by the prospect of something unpleasant. Deadline pressure can result from fear that the decision will be taken away from the negotiators. [FN109]

5. Opportunity for Gain [FN110]

  Negotiation must have the potential to produce gain for all parties. Negotiated solution to 'zero sum games' are difficult to achieve. [FN111] Effective mediation helps negotiating parties discover alternative formulations and to perceive the true value of their BATNAs. [FN112] The heightened perception of loss in alternative forums may be great enough to overcome any fear about loss through negotiations. [FN113]

6. Absence of Fundamental Value Conflict [FN114]

  The regulation to be developed cannot involve compromise of deeply held beliefs or values. This would involve costs higher than most parties could tolerate. [FN115] This criterion does not mean that negotiation cannot be used to resolve minor issues involved in controversies in which fundamental values are at stake. This criterion also does not mean that the implementation of regulations developed on a fundamental matter cannot be negotiated.  [FN116]

  Moreover, the fundamental value criterion frequently is misunderstood. Merely because substantial costs are involved does not mean that fundamental values are involved. Instead, fundamental values are those having ideological, rather than merely economic significance, or, perhaps, economic risks of such magnitude that they seriously threaten a party's very survival.

7. Permitting Trade Offs [FN117]

  The negotiation process is an evaluative one. The parties determine what is most important to them and direct their behavior toward accomplishing it. If there is only one issue to resolve, one position to assume, negotiation is unlikely to result in agreement. A situation where there are two or more issues to resolve, however, with the attendant possibilities of gain on one issue offsetting loss on another, is more amenable to negotiation. [FN118]

*1646 8. Research Not Determinative of Outcome [FN119]

  Resolution of a dispute should not depend on research results. The parties may be unwilling to formulate or compromise positions in the face of scientific uncertainty. Research results might produce a clear victory for one interest.  [FN120] Even if a clear win or loss might result from the research findings, however, the parties still might choose to negotiate parallel or peripheral issues. [FN121] The parties might negotiate issues to be researched and the way in which the research should be conducted.

9. Agreement Implementation [FN122]

  Some kind of effective implementation process must be present. Lack of an implementation process would destroy the core prerequisite for negotiation: the parties' belief that their own interests will be furthered by negotiating.

  Harter recognized implicitly two forms of negotiated rulemaking, one in which the agency participates in the negotiations and another in which it does not.  [FN123] The same two variants had been addressed more explicitly in a Harvard Law Review note published about the same time. [FN124]

  The objective of negotiated rulemaking is to reach 'consensus' among the participants as to the content of the proposed rule. Harter characterized the definition of 'consensus' as 'one of the most difficult and complex questions in regulatory negotiation.' [FN125] He concluded that experience was necessary before anyone could develop more concrete ideas on what consensus should entail, pointing out, however, that the existence of a consensus is more a matter of feel than of mathematical calculation. [FN126] The Harter formulation necessarily omitted detailed formulation of the conditions conducive to closure on an agreement. [FN127]

  The Harter guidelines, reflected in ACUS Recommendation 82-4, proved sound in practice application. The following parts of the article explore application of the guidelines in four actual rule negotiations. The article begins with the benzene negotiation, because this negotiation did not produce consensus and is therefore a useful basis for scrutiny to determine what can go wrong, and because *1647 the author personally interviewed most of the major participants in the benzene negotiation. [FN128]

III. THE BENZENE NEGOTIATIONS

A. INTRODUCTION

  In the summer of 1983, the Occupational Safety and Health Administration  (OSHA) began to develop a revised standard for workplace exposure to benzene through negotiated rulemaking. An earlier benzene standard had been invalidated in litigation that ultimately reached the Supreme Court. [FN129] Negotiations proceeded for more than a year, producing agreement in principle. The parties did not report to the OSHA, however, because no agreement could be reached on some details and because the political climate had changed, and there was dissension among the various interests (and others similarly situated). Philip J. Harter and Gerald Cormick facilitated the negotiations.

  Although one reason the negotiations adjourned without agreement was that the negotiators expected the OSHA to issue a standard unilaterally, more than a year elapsed before the OSHA published a proposed rule. [FN130]

B. HISTORY OF THE REGULATION OF BENZENE BY THE OSHA

  Benzene is a clear, colorless, highly flammable liquid with a strong odor. It evaporates rapidly under ordinary atmospheric conditions, giving off vapors nearly three times heavier than air. The petrochemical and petroleum refining industries produce nearly all the benzene used in the United States. Benzene is used as a solvent or a reactant alone or with another liquid. Some 274,000 workers are exposed to benzene in seven major industries: petrochemicals, petroleum refining, coke and coal manufacturing, rubber tire manufacturing, bulk storage terminals, bulk plants, and transportation. [FN131]

  The OSHA adopted the original standard for benzene exposure in 1971. This was a 'national consensus standard,' [FN132] setting a time-weighted-average exposure level (TWA) [FN133] of 10 parts per million (ppm); ceiling concentrations of 25 ppm; and permitted excursions above the ceiling not to exceed 50 ppm for more than ten minutes in any eight hour work period.  [FN134]

  Revision of this standard was triggered by a 1974 report prepared by the National *1648 Institute of Occupational Safety and Health (NIOSH),  [FN135] acknowledging that benzene might cause leukemia but recommending retention of the national consensus standard for the time being. In 1976, the United Rubber Workers (URW) petitioned unsuccessfully to lower the standard on an emergency basis. Later that same year, NIOSH submitted a revised report concluding that benzene could cause leukemia and that no safe level for benzene exposure could be established. [FN136] It recommended that the OSHA establish a new standard prohibiting exposure in excess of 1 ppm. [FN137]

  In response, the OSHA issued voluntary guidelines in January, 1977, recommending that worker exposure to benzene not exceed an eight hour TWA of 1 ppm in any eight hour shift of a forty hour week. Four months later the OSHA promulgated an emergency temporary standard for occupational exposure to benzene [FN138] setting an eight hour TWA limit of 1 ppm with a ceiling level of 5 ppm for any fifteen minute period during an eight hour work period.  [FN139] Challenges to the standard were filed in Courts of Appeals for both the District of Columbia Circuit and the Fifth Circuit. [FN140] The Fifth Circuit issued a temporary restraining order against the OSHA, [FN141] and the emergency standard never went into effect. [FN142]

  One week later, the OSHA published notice of a proposed final benzene standard. [FN143] Hearings on the proposed standard were held from July 19 through August 10, 1977. Ninety-five persons testified. [FN144]

  The final standard of the OSHA, issued in February, 1978, was identical to the emergency standard, limiting exposure to an eight hour TWA of 1 ppm, with a ceiling of 5 ppm for any fifteen minute period in an eight hour work period. It also prohibited eye and skin contact and required measurement of employee exposure, engineering controls, work practices, personal protective clothing and equipment, signs and labels, employee training, medical surveillance, and record keeping.

  Producers and users appealed to the Fifth Circuit seeking preenforcement review of the final standard. [FN145] The Fifth Circuit invalidated the standard because it was based on findings that were unsupported by the administrative record. [FN146] The court concluded that the OSHA had exceeded its authority because it had not been shown that the eight hour TWA exposure limit was 'reasonably necessary or appropriate to provide safe and healthful employment' as required by section 3(8) of the Act. [FN147] The court further concluded that section 6(b)(5) [FN148] did not give the OSHA the unbridled discretion to set standards designed to create absolutely risk free workplaces without regard for cost. [FN149]

  The Supreme Court affirmed, [FN150] with a plurality concluding that the OSHA's rationale for lowering the permissible exposure limit from 10 ppm of benzene to 1 ppm was based on a series of unsupported assumptions indicating that leukemia might result from exposure to 10 ppm of benzene. [FN151] There had been no showing that leukemia was caused by exposure to 10 ppm of benzene and that it would not be caused by exposure to 1 ppm. [FN152] The plurality reasoned that section 3(8) of the Act implies that before the Secretary of Labor can set a permanent health or safety standard, he must first find that the workplace is unsafe in the sense that significant risks are present and can be eliminated or lessened by a change in practices. [FN153] Therefore, the Secretary exceeded his power by avoiding this threshold responsibility when he relied on a special policy for carcinogens that placed the burden of proving the existence of a safe level of exposure to benzene on the industry.  [FN154] The plurality also noted that the Act's legislative history supports the conclusion that Congress was concerned not with absolute safety but with the elimination of significant risk of harm. [FN155]

  Justice Stevens, joined by Chief Justice Burger and Justice Stewart, also concluded that the burden was on the OSHA to show, on the basis of substantial evidence, that it is at least more likely than not that long-term exposure to 10 ppm of benzene presents a significant risk of material health impairment. [FN156] The OSHA, they concluded, did not satisfy this burden.  [FN157]

  Chief Justice Burger, concurring, emphasized that the requirement for the OSHA to 'retrace its steps' with greater care does not displace its ability to make policy judgments and noted that the Act also required the Secretary to determine that the economic effects of the new standard bore a reasonable relationship to the expected benefits and that the OSHA had not done this.  [FN158] Justice Rehnquist, concurring in the judgment, expressed the view that section 6(b)(5) of the Act, relied on by the Secretary in promulgating the benzene standard, was an unconstitutional delegation of legislative power to the executive branch. [FN159]

  Justices Marshall, Brennan, White, and Blackmun dissented. They stressed that the Act required a reviewing court to uphold the Secretary's determination *1650 if supported by 'substantial evidence in the record considered as a whole.' [FN160] Therefore, the Secretary's standard was fully in accord with the statutory mandate of section 6(b)(5) of the Act that standards for toxic materials or harmful physical agents most adequately assure that no employee will suffer material impairment of health or functional capacity. [FN161]

  Following the Supreme Court's decision, the OSHA considered various quantitative risk assessments to determine whether it should try again to adopt a standard below 10 ppm. Of the 274,000 workers exposed to benzene, the OSHA estimates that some 262,000 are exposed to concentrations at or below 1 ppm as an eight hour TWA, while approximately 10,000 are exposed to levels between 1 and 5 ppm. The OSHA estimates that fewer than 1,500 workers are presently exposed to benzene levels averaging 5 ppm or more. [FN162]

  On April 14, 1983, Dr. Sidney M. Wolfe, Director of the Public Citizen Health Research Group, wrote Thorne G. Auchter, Assistant Secretary of Labor for Occupational Safety and Health, requesting that the OSHA issue an emergency temporary standard under section 6(c)(1) of the OSHA statute. Auchter denied the petition, noting that only eight to nine percent of workers actually are exposed to benzene in excess of 1 ppm on an eight hour TWA basis. He committed the agency, however, to issue a new permanent standard on an expedited basis, not later than June, 1984. Specifically, he committed to a timetable, pursuant to which the OSHA would add a benzene standard to its 'regulatory agenda' by June, 1983; submit a proposed standard to the OMB by November, 1983; publish a proposed standard in the Federal Register by December, 1983; hold a fact- finding hearing by February, 1984; and publish the final standard by June, 1984. [FN163]

C. DECISION TO NEGOTIATE

  At about the same time that Auchter and Wolfe were corresponding over the plans of the OSHA for a revised benzene standard, Philip Harter and Gerald Cormick held discussions with the OSHA staff about the potential for regulatory negotiation to facilitate the standard setting activities of the OSHA. As a result of these discussions, Assistant Secretary Auchter decided to explore negotiated rulemaking as a means of resolving the benzene dispute. The OSHA's interest in regulatory negotiation was heightened by a perception that negotiations had been useful in connection with development of the coke over standard. [FN164]

  Preliminary discussions with the parties and actual negotiations proceeded from July, 1983, until October, 1984. Initially the convenor/facilitators  [FN165] determined that the parties were skeptical that negotiations could produce a consensus on OSHA action, but they thought meetings might be helpful. After the meetings began, the participants decided to attempt to narrow their differences *1651 and to reach agreement on a standard under a protocol that required agreement on a 'total package' before anything substantive was reported to the OSHA. Representation of the petroleum industry, one of the key industry stakeholders, proved difficult because of dissension among firms within the industry about the wisdom of negotiating. Nonparticipant trade unions were concerned that options being considered in the negotiation might set an adverse precedent for other health standards of the OSHA. Nevertheless, the participants nearly agreed on a standard before finally adjourning in the fall of 1984.

D. ISSUES IN NEGOTIATION

  The central issue in the benzene negotiations was the Permissible Exposure Limit (PEL). The existing standard was set at 10 ppm, and the effort of the OSHA to set a standard at 1 ppm had been invalidated in the Supreme Court. The steel industry, facing difficulty in meeting the 10 ppm standard and questioning evidence of health hazards below that level, opposed reducing it. The rubber industry had been meeting a 1 ppm level since the late 1970's.

  Between these extremes were the petroleum and chemical industries, which thought they could accept a level of 2 ppm, although they were not persuaded that it was justified on a scientific or medical basis. A 1 ppm standard was a problem for the petroleum and chemical industries because of variability in measurements. Eighty-five to ninety percent of monitoring results would show levels lower than 1 ppm, but measurements above that level also would occur, involving different locations and circumstances hard to predict or evaluate. Thus frequent measurements in a single facility would show noncompliance with a 1 ppm limit at least some of the time. [FN166]

  Labor vigorously opposed anything higher than 1 ppm, and the number '1' became an article of faith within the labor movement. New scientific data that had become available since the promulgation of the 1978 standard reinforced the labor view that a 1 ppm standard was justified based on health risk.

  Industry hoped to deal with its own concerns, while satisfying labor's desire for a standard of 1 ppm, primarily through the concept of averaging: requiring an employer to show compliance with a 1 ppm standard by an average of measurements taken over some time period. Many different ideas were explored with respect to averaging: a standard expressed in terms of 'ppm-hours,' such as a limit of 40 ppm-hours; providing that a single 'exceedence' of the eight hour TWA would not result in a violation as long as the average eight hour TWA for the previous five readings was within the standard; providing that a single 'exceedence' of the eight hour TWA would not result in a violation as long as the average eight hour TWA for the next five readings was within the standard.

  Labor was concerned that a standard that allowed averaging would set a precedent for other health standards. Some labor participants preferred setting a simple PEL of 1 ppm, and dealing with the problems caused by exceeding the standard through enforcement guidelines.

  The impact of a new benzene standard on tort liability also was a concern for industry, particularly petroleum refiners. Many of those familiar with the negotiations *1652 perceived the tort liability issue as more important to some industry participants than the PEL. Gasoline has benzene in it. Benzene causes leukemia and other blood disorders, and may cause tumors. Everyone who drives a car is exposed to low levels of benzene. Employees exposed to benzene in enterprises outside the petroleum industry may sue manufacturers of benzene instead of, or in addition to, their own employers because of workers' compensation limitations on suit, or for other reasons. Thus the class of potential plaintiffs against petroleum industry defendants is extremely large. Asbestos litigation, and litigation over benzene-caused disease in Gulf Coast shipyard industry made industry executives sensitive to the potential for benzene litigation in which their companies might be defendants.

  The relationship between the ability of petroleum companies to defend such lawsuits and a finding by the OSHA of significant health risk at levels of 10 ppm or below is speculative. Section 4(b)(4) of the Occupational Safety and Health Act [FN167] says that the OSHA standards shall not be outcome determinative in tort litigation. Nevertheless, it was widely perceived that a finding of significant health risk could affect adversely the industry's ability to protect itself against large tort liability. Labor was basically uninterested in tort liability, but it was willing to try to develop language satisfactory to the petroleum refiners as long as the language did not jeopardize the ability of the standard to survive judicial review.

  The negotiators were faced with a dilemma, however, in formulating a compromise. Industry wanted to avoid a strong finding of health risk below 10 ppm to mitigate tort liability problems. If the finding was weakened to reduce tort liability problems, however, the chance that a negotiated standard would be invalidated in the courts increased, leaving the parties in the position of having compromised on some issues without the benefits of the total negotiated package.

  The rubber industry had been concerned previously only with that part of a standard that might deal with skin contact. The industry hoped for an exemption for uses of liquids with a content of less than 0.3% benzene by volume but labor was opposed to an outright exemption. The negotiators agreed upon an exemption for employers who could show both that liquids contacting employees' skin contained less than 0.3% benzene by volume, and absorption rates low enough that a 1 ppm level would not be exceeded. [FN168]

  Labor understood that variability of benzene levels presented legitimate problems for standard definition. Labor participants sought to determine actual industry practice regarding temporary excursions and to write those practices into a standard; for example, requiring action and frequent monitoring after a spill or a ventilation system failure. They perceived, however, that industry participants were never willing to be pinned down about what should be required when the standard was exceeded, that is, what action should be taken when the 'action level' was exceeded.

  A separate goal for labor was to use the benzene negotiations to induce the OSHA to accept the proposition that medical surveillance was most needed for *1653 older workers, some of whom had left active service. Industry participants were reluctant to embrace this idea, apparently out of fear that it would stir up tort claims by the older workers. In addition, industry noted that screening of older workers does not permit prophylactic action to the same degree as screening of younger workers.

  In general, the labor participants were unaccustomed to the idea that different parts of a benzene standard could be traded off against each other, based on economic impact. Labor was accustomed to thinking of each piece of a standard--the PEL, engineering controls, monitoring, medical surveillance, action requirements--as independent components, each with its own rationale. Participation by labor in discussing tradeoffs resulted only from the realization that industry viewed the standard as a total package and was concerned with the overall economic and health impacts.

E. BENZENE AS A CANDIDATE FOR NEGOTIATED RULEMAKING

  In retrospect, benzene was both a good and a bad candidate for negotiated rulemaking. Benzene was a good candidate because formulation of a rule required several different subjects to be addressed, creating the potential for tradeoffs; [FN169] because party positions had crystallized during seven years of rulemaking proceedings and litigation reaching the Supreme Court of the United States; [FN170] and because a limited number of parties were involved. [FN171]

  The history of the litigation also made benzene a good candidate for another reason: a standard generally supported by organized labor had been invalidated in the courts because of aggressive opposition by industry, and promulgation of a revised standard had been delayed for seven years. This demonstrated that a measure of acceptability to industry could speed attainment of labor's health and safety objectives. [FN172]

  Benzene was a bad candidate because the years of administrative and court litigation had solidified the positions of the parties and the agency and generated a voluminous record lending support to each of those positions. In addition, benzene was a bad candidate because of the perception that benzene exposure creates significant health risks, including the risk of fatal disease. This increased the likelihood that basic values would be at issue in negotiations. [FN173] Finally, though benzene is a single substance, the issues confronting the different industries that would be affected by a benzene rule were diverse, resulting in a complex, multipolar negotiation and making it difficult for the participants to assimilate all the data relevant to the issues under discussion. [FN174]

  *1654 Some participants thought benzene was a bad candidate because the issues were essentially technical, reinforcing their view that a 'correct' standard could be determined objectively. [FN175] The technical nature of the issues, in the view of these participants, meant that the most useful process for developing a standard would be probing examination of research data--an activity for which negotiations are less well suited than formal adjudicatory hearings. They thought that labor's disparity of resources and technical expertise would be a more prominent handicap in negotiations than in traditional administrative litigation. They also feared that the negotiations would delay the issuance of a standard, despite the commitment of the OSHA to issue a final standard by June, 1984. [FN176]

  Another reason benzene was a bad candidate was the parties' differing perceptions of the implications of the Supreme Court's decision. Labor believed that a 1 ppm standard could pass court review if the OSHA did a better job of marshaling scientific evidence in support of the standard. Taking this view, there was little to be accomplished in negotiations, since the only remaining problem was a technical data analysis. Industry, on the other hand, viewed the Court's decision as a repudiation of attempts by the OSHA to revise the 10 ppm national consensus standard. This perception suggested that negotiations would be concerned with the full range of issues related to benzene exposure in the workplace. [FN177]

  Benzene also was a bad candidate because the statute authorizing a standard makes a finding of 'significant health risk' a prerequisite for agency power to regulate. The importance of this finding to a valid standard had been underscored by the Supreme Court in Industrial Union Department v. American Petroleum Institute, [FN178] and the participants in the benzene negotiation were acutely aware of it. The centrality of the significant health risk finding presented a dilemma that utlimately proved insurmountable. If a negotiated standard was to survive judicial review, [FN179] the OSHA must find that significant health risks resulted from the existing standard of 10 ppm. A finding of health risks at PELs lower than 10 ppm was perceived, however, by the petroleum industry at least, as increasing potential tort liability. Industry thought the OSHA was inclined to read the requirement of the Supreme Court decision as saying, 'Quantify the risk, making it seem as high as possible at the lowest possible exposure levels.' They hoped to frame a risk finding that would recognize a risk at a PEL of 10 ppm, but would not say that a risk existed at the new PEL. The participants, however, were unable to develop language that satisfied both the tort and statutory criteria.

F. INCENTIVES TO PARTICIPATE--IN GENERAL

  Both labor and industry believed that uncertainty could be reduced by negotiations. The traditional adversary process usually results in opposing parties taking extreme positions, and consequently building a record that supports agency *1655 action anywhere in between. One participant described the result as a 'crap shoot.' In contrast, if the parties negotiated an NPRM, or at least narrowed the issues through negotiations, they could make the ultimate standard more predictable. All of the parties perceived themselves as risk averse--industry more than labor. Thus reducing uncertainty was a benefit.

  Virtually all participants lacked confidence in the OSHA; there was considerable concern that the OSHA would 'screw up' a new standard through the regular process, either by promulgating a standard that could not be defended in the courts or by imposing collateral requirements with high costs for relatively little health benefit.

  The participants recognized that an individual employer, even a member of one of the participating trade associations, or a union entity or public interest group, could challenge a negotiated standard. Most of the participants were nevertheless reasonably confident that any such challenger would be hard pressed to convince a court to invalidate a standard that unions and trade associations most directly affected had not only agreed upon but also actively defended. Others, however, thought that a challenge to a negotiated standard might be strengthened by the negotiation process, because of their belief that a court would find patent violations of the FACA, or a violation of the APA prohibitions on ex parte contact, or prohibitions on delegation of governmental authority to private citizens.

G. INCENTIVES FOR THE OSHA TO PARTICIPATE

  The OSHA did not participate in the benzene negotiations directly. It did not take part in the discussions among industry, labor, and public interest representatives. It did, however, pay the convenor/mediators [FN180] and indicated its willingness to use a negotiated standard as the basis for its rulemaking.

  Enthusiasm for the process was higher at policy levels within the agency than at 'working levels.' Many health standards personnel of the OSHA and their legal staffs had major doubts whether benzene was a good candidate for negotiated rulemaking, and further doubts about the way the negotiations were handled, believing that the convenor/mediators consistently were too optimistic about the prospects for agreement. The commitment of the OSHA to the process, low from the beginning, was retarded further by its lack of participation in the negotiations themselves.

  Nevertheless, sharp adversarial conflict during rulemaking and in court challenges make the agency's job more difficult, and in the words of one official of the OSHA, negotiated rulemaking was 'worth a try.'

H. INCENTIVES FOR LABOR TO PARTICIPATE

  Organized labor had two reasons to believe that negotiations would be preferable to the traditional rulemaking process. First, a negotiated standard accompanied by an agreement not to litigate the legality of the standard could save organized labor significant litigation expenses. Second, the perception that the *1656 OSHA was in unfriendly hands increased labor's fear that an agency promulgated standard would be delayed and might not satisfy labor's desires as well as a negotiated standard.

  Labor's incentive to negotiate was reinforced by industry's power to delay. Labor did not believe a 1 ppm standard was vulnerable to judicial attack. It thought a court would likely find the OSHA entitled to opt for greater health protection, working from uncertain scientific evidence, as long as it supported a lower standard with a finding of significant health risks at benzene levels higher than the standard. On the other hand, industry had already delayed promulgation of a lower standard for eight years, and could delay it further by arguments presented to the OMB [FN181] and by litigating aggressively.

  Some labor participants perceived the OMB as a bigger threat than the courts to a standard from the OSHA. To them, industry support for a standard before the OMB was more important than an agreement not to challenge the standard in court. Labor's motivation to negotiate was strengthened by an informal assurance by the OMB that any proposal by the OSHA based on negotiated agreement would be cleared by the OMB within twenty-four hours.

  Labor's incentives to participate were mitigated by a perception that the OSHA had committed itself to issuing a standard within a short time and a belief that this standard would adopt a PEL of 1 ppm. In addition, some labor participants were more comfortable with the traditional hybrid rulemaking process than with negotiation, believing that adversary administrative litigation, especially cross-examination of industry witnesses, is the best way to develop scientific data and a factual record to support an adequately protective standard. They were pessimistic, however, that a standard acceptable to them could be promulgated within the eight months promised by the OSHA, given what labor perceived as the unsympathetic attitude of the OSHA and the OMB.

  Other labor participants, and some nonparticipants allied in interest with the labor participants, were suspicious of the negotiation process. As noted above in the discussion of benzene as a candidate for negotiated rulemaking, some participants feared that labor's resource limitations would be magnified in negotiations, and that negotiations might delay issuance of a standard. Nevertheless, because they thought the OSHA wanted negotiations to occur, they reluctantly agreed to participate.

I. INCENTIVES FOR INDUSTRY TO PARTICIPATE

  Before identifying positive incentives for industry to participate, it is useful to note a change in industry position since the 1977 standard was issued. During the three years between adoption of the emergency standard and invalidation of the final 1 ppm standard by the Supreme Court, a number of firms had reduced benzene levels in their workplaces. Thus the marginal cost for the industry to reach the 1 ppm standard in 1983-1984 was less, in real dollars, that it would have been in 1977-1978.

  *1657 Industry's incentives to participate in negotiations depended on its perception of probable action by the OSHA in the absence of a negotiated standard. After the Supreme Court litigation, it appeared virtually certain that the OSHA would find a way to justify a 1 ppm standard, and might accompany this standard with findings, not only that exposures at the 10 ppm level posed health hazards, but also that a 1 ppm standard presented health hazards. In addition, a variety of ancillary issues would be addressed in a standard, such as health monitoring, averaging, and the action level, that might be influenced in a negotiation. There was a perception, at least among some in the industry, that some alternative needed to be found to adversarial rulemaking, and that regulatory negotiation was worth a try. Some other industry participants were pessimistic that anything useful could come from the negotiation project but elected to participate for defensive reasons. They were afraid a process excluding them might produce an undesirable outcome.

J. DIFFERENCES AMONG THE INDUSTRY PARTICPANTS AND WITHIN INDUSTRY

CONSTITUENCIES

  The possibility of a 1 ppm PEL presented different concerns to the different industries involved. The petroleum, chemical, and rubber industries were able to tolerate a 1 ppm standard better than was the steel industry. On the other hand, the petroleum and chemical industries were more concerned about tort liability than was the steel industry. This concern with tort liability made petroleum and chemical participants more willing to accept a 1 ppm standard in exchange for the absence of an agency finding that benzene exposure created significant health risks below levels of 10 ppm.

  The rubber industry had much narrower concerns than the other industries, limited to that part of the standard than would address dermal contact. The rubber industry expressed little concern about the PEL during the negotiations. Representatives of the different industries did not undercut each other's positions in formal negotiated sessions, but neither did they offer aggressive support for positions other than their own.

  The American Petroleum Institute (API) played the leading role on the industry side. API participation was viewed as essential by labor. API also had more internal difficulties than the other industry groups participating. These facts combined to make API difficult in plenary sessions and in industry caucuses. Some of the other industry participants resented what they perceived as API dominance, and occasionally maneuvered to have the chemical industry participants, rather than API, speak for industry in sessions of the full negotiating group.

  Most of the industry participants in the benzene negotiations worked in the health and safety functions of regulated enterprises or trade associations. Industrial relations officers in the same organizations were anxious about direct dealings between labor and industry in the benzene negotiations. One concern related to the possible content of a negotiated standard--the possibility that it might prejudice positions in future labor-industry negotiations. Another concern related the possible precedent setting effect of industry-level negotiations; both the petroleum and chemical industries had resisted industrywide bargaining *1658 with trade unions, and opposed conduct that might make such industrywide bargaining more likely in the future.

  One of the most serious difficulties experienced during the negotiation involved internal dissension within API. Much of the difficulty was structural; hundreds of petroleum companies would be adversely affected by a regulation requiring reduced benzene exposures, many more than in any of the other industries. Moreover, the large size of the industry leaders meant that large numbers of people within each company felt entitled to be included in discussion of negotiating positions taken by industry representations.

  The petroleum industry participated through a three-tiered committee system. At the highest level was a committee of thirty to forty company vice presidents with responsibility for environmental and health matters. At least one member of this group was intransigent in opposing industry participation in the benzene negotiations. Chief executive officers of petroleum companies were briefed from time to time on the negotiations and were supportive, but they never effectively restrained their vice president who had more direct control. As a result, the petroleum industry representatives were regularly subjected to criticism by others from within their own industry. Fear of adverse public reaction prevented a decision by the petroleum industry to withdraw from the negotiations. Concerns within the industry, however, made meaningful compromises at the bargaining table difficult.

  Despite these internal difficulties, however, the petroleum industry finally supported a compromise benzene standard with considerable unity. This compromise was the basis of the near-agreement in the negotiations.

K. INTRALABOR MOVEMENT DIFFERENCES

  Labor participants perceived intraconstituency differences as less of a problem for labor than for industry. Union health and safety representatives regularly worked together on matters involving the OSHA, and thus were accustomed to the process of compromise in formulating strategic administrative and judicial litigation stategies. Moreover, labor participants thought they had greater authority to speak for their constituents than industry representatives.

  In part, labor unity was due to general agreement that the 1978 standard was the minimum that would be acceptable. It was further enhanced by respect for the lead role the AFL-CIO traditionally had taken on health and safety issues, and by personal respect for the AFL-CIO spokesperson.

  Opposing these unifying forces were some differences among the participating unions and some strong feelings held by unions that did not participate. The United Rubber Workers (URW) had greater trust in its industry counterparts than did the other unions. It was more willing to accept various compromises proposed during the negotiations. The Oil, Chemical & Atomic Workers (OCAW) historically had invested heavily in health and safety issues, although in the opinion of internal union critics had little to show for the investment. Faced with declining membership and failed merger explorations, OCAW was perceived as having an acute need for some tangible evidence of success on the health and safety front. A negotiated benzene rule would be such evidence, on an issue of particular concern to rank-and-file refinery workers. There also were *1659 some differences between those accustomed to the give and take of collective bargaining and those more accustomed to the formal adversary process of traditional rulemaking.

  Labor's limited resources discouraged additional conferences to iron out differences and arrive at a common position outside the full negotiations with industry representatives. As a result, labor differences tended to surface at the table for everyone to see.

  Labor constituency problems arose, not so much within represented unions as within the labor movement as a whole. Most approached the OSHA standard setting from a global perspective rather than an industry-specific perspective. Accordingly, the precedent setting effects of particular elements of a benzene standard were a concern. After agreement on a standard seemed a real possibility, the United Auto Workers (UAW) and Amalgamated Clothing and Textile Workers Union (ACTWU) were quite critical of the process, and this criticism within the labor movement apparently discouraged participating union representatives from taking major risks to reach final agreement.

L. NEGOTIATION PROTOCOLS

  It is important to recall that, when the benzene negotiations began, no one expected an agreement on a total package but, largely due to the enthusiasm of Assistant Secretary Auchter, the effort was transformed into an attempt actually to negotiate a standard. Some resentment that the process had been transformed lingered and was amplified when Auchter resigned while negotiation was under way.

  The parties agreed that none would promote, as products of the negotiation process, anything that was not agreed to by all participants. This ground rule did not, however, preclude consideration of an agreement that would leave certain issues to be decided by the OSHA. In this regard early agreement of the URW and the rubber industry trade association [FN182] was something of an irritant to industry participants.

  The parties also agreed, at least tacitly, that none of the participants would challenge any aspects of a standard that had been agreed to in the negotiation. In this way, both labor and industry could buy themselves a measure of certainty. On the other hand, the union participants understood that the trade associations could not guarantee that none of their corporate members would litigate the legality of an ultimate standard, even if the participants reached agreement. Nevertheless, it was perceived that an agreement would be a powerful psychological motivation for a court to sustain a standard from the OSHA, which is entitled to considerable deference under both the APA and the OSHA Act.

M. THE NEGOTIATION PROCESS

  Perhaps the most salient feature of the benzene negotiation process is that the agency with the statutory responsibility for establishing a standard did not actually *1660 participate in the negotiations. Instead, the OSHA said that it would use any negotiated standard as a basis for its rule, encouraged the parties to work out their differences in good faith, attempted to increase incentives for affected interests to participate by threatening to proceed with its own standard regardless of the pace of negotiations, and kept itself informed, through the mediators, on negotiating progress.

  This kind of agency nonparticipation had been recognized as one of the two basic forms of negotiated rulemaking. [FN183] One advantage when the agency does not participate is that deliberations may be more candid in the absence of the ultimate decisionmaker.

  In addition to this theoretical justification for the low profile of the OSHA, however, there is significant evidence that the OSHA limited its participation because it did not want the FACA [FN184] to apply to the benzene negotiations, and it perceived that if the agency did not participate, the FACA would not apply. [FN185] The OSHA subsequently was encouraged to limit its pa