This Article focuses on a special problem with performance standards—that their performance criteria are often so subjective as to deny regulated persons a clear idea of what is required. It begins with a discussion of specification and performance standards in American regulatory history. It further discusses attempts by Congress and others to, therefore, require that performance criteria be “objective.” The Article then sets out a case study of how congressional attempts to require “objective” performance criteria have fared. It examines in depth whether one agency, the Occupational Safety and Health Administration (OSHA), has complied with that special requirement and finds that many standards that OSHA has touted as performance standards fail to meet it. This Article also notes how, in rulemaking, OSHA has often styled many of its standards as “performance” standards that would give employers “flexibility” in compliance. It notes, however, that once enforcement begins, promises of such flexibility are often forgotten. The Article then raises some broader jurisprudential issues related to open textured performance standards. It concludes with the suggestion that in some regulatory situations, notwithstanding the arc of regulatory scholarship, specification standards may be more appropriate.
Specification or design standards specify “exactly how the regulated entity must act.” They are also known as “means” or “prescriptive” standards. By contrast, when a regulation is a performance standard, it “identifies its objective but does not prescribe the means for or the specific obligations of the employer to comply with the objective.” Stated simply, specification or design standards dictate means, while performance standards dictate ends. Consider the protection of employees from excessive noise. A specification standard might require that an employer place fiberglass panels fifty millimeters thick—the specification criterion—around machines emitting specific sound levels fiberglass panels fifty millimeters thick—the specification criterion. A performance standard might require that the noise level reaching an employee’s ears not exceed ninety decibels—the performance criterion.
In the early years of the Republic, few people thought of regulations as imposing either performance or specification requirements. Regulations were just that—rules by the government that required or restricted conduct. But most, if not all, early regulations contained significant specification elements.
An early example of the federal government imposing specification standards (although not called such) came in the Steamboat Inspection Act of 1838. It required “that the safety valves on steam engines be opened whenever the steamboat was not underway in order to keep down the steam pressure in the boiler” and that “iron rods or chains shall be employed and used in the navigating of all steamboats, instead of wheel or tiller ropes.” Other provisions had both performance and specification features. Section 8 of the Steamboat Inspection Act required a minimum number of lifeboats based on vessel tonnage, and § 9 required “a suction-hose and fire engine and hose suitable to be worked on said boat in case of fire.” These provisions exemplified what OSHA would observe over a century later—that there is a “continuum between performance standards . . . and specification standards.” A steamboat inspection statute passed in 1852 imposed what today would be called a pure performance requirement: it required tests for “high-pressure” iron or steel steam boilers “[s]ubjecting them to a hydrostatic pressure” of no more than “one hundred and sixty-five pounds to the square inch” and exceeding “the working power allowed, in the ratio of one hundred and sixty-five to one hundred and ten.”
The differences between specification and performance requirements can be better understood by surveying how standards addressing the same problem evolved over time. Consider the evolution of standards regulating the safe packaging of shellfish. The 1906 Pure Food and Drug Act first required that packaging prevent adulteration of food: its § 7 stated that food “shall be deemed to be adulterated . . . [i]f any substance has been mixed and packed with it so as to reduce or lower or injuriously affect its quality or strength.” The Department of Agriculture then issued regulations for the purity of water in which shellfish were shipped, the cleanliness of shipping containers, the salt content of the water, and the prevention of pollution from melting ice. Later, during the New Deal era, regulations added explicit specification requirements. For example, in the case of shellfish packaging and shipping, the regulations imposed these requirements:
(d) Blanching tanks shall not be located in picking room. . . . If picking and packing rooms are in separate buildings such buildings shall not be more than 100 yards apart . . . .
(e) The tops of picking and packing tables and the interior of washtanks, flumes, blanching tanks, brine tanks, and all utensils . . . shall be of metal other than lead or of other smooth, hard, nonporous material that can be readily cleaned.
The modern standard, which first appeared in 1977 and remains today, uses a performance-based approach. The standard for canned shellfish became subsumed into a more general standard for food processing plant construction and design, one that directed producers to “[p]rovide sufficient space for such placement of equipment and storage of materials as is necessary for the maintenance of sanitary operations and the production of safe food.” Mandates for specific materials and maximum distances between facilities were replaced with standards that echoed the statutory goals of minimizing contamination or adulteration. The modern standards used words such as “suitable” or “adequately”—the subjectivity of which poses the notice and related problems that this Article discusses below.
With the rise of the administrative state in the 1960s and 1970s, American businesses began to complain that they were being deluged with excessively detailed and rigid federal specification regulations. Specification regulations were seen as an element of a “regulatory system [that] has become an instructional manual. It tells us and bureaucrats exactly what to do and how to do it. Detailed rule after detailed rule addresses every eventuality, or at least every situation lawmakers and bureaucrats can think of.” Another drawback of specification standards is that they “tend to become obsolete quickly and thus are a potential road-block in the way of the growth of new technology.” Indeed, some have suggested that such standards can reflect a “strategy to perpetuate older technologies.”
They can also be more difficult to draft and slower to adopt, for their text must not only be specific but must intelligently and intelligibly cover all cases within their stated scope. The task of drafting a proposed standard with such detail can be formidable, and the evaluation of ensuing public comments can delay the standard’s completion.
Yet, once promulgated, specification standards can be easier to comply with “[i]n principle . . . [and] easier to enforce than performance standards.” For example, in the case of the mounting height of a fire extinguisher, all that might be needed is a tape measure. As former Justice Stephen Breyer observed, “The regulator often will have to choose between ‘design’ standards, which are readily enforceable, and ‘performance’ standards, which encourage the development of new technology. The need for enforcement will bias his choice in the former direction.”
Indeed, as the administrative state and corresponding regulations expanded, some industries preferred prescription standards as a way of discouraging new entrants. Many standard-setting organizations whose standards were adopted in state laws wanted prescription, rather than performance standards, as a way to both protect the organization’s members from new entrants and to delay obsolescence of the organizations’ manufacturing equipment. This was the case with electric equipment codes adopted by states which used design standards have the same impact. Further, in the past, many states barred corporations from owning pharmacies rather than putting in “performance” standards—like employing trained pharmacists and meeting standards of conduct irrespective of the owner.
Performance standards were often suggested as a cure for the rigidity problems posed by specification standards. They were called more “flexible” and “more cost-effective” and were said to “encourage the development of new technology.” Over the last forty years, scholars have discussed and promoted the superiority of performance standards. The Office of Management and Budget (OMB) issued circulars explicitly advising federal regulatory agencies to use performance standards in preference to specification standards.
Although the preference for performance standards has pretty much prevailed among regulatory scholars, not all agencies adhere to a strictly performance-based model. At times, specification standards can be more efficient or effective. Sometimes hybrid regulations (part specification and part performance) will work best. For example, in 1999, the Nuclear Regulatory Commission (NRC) produced a white paper highlighting a desire to move to performance standards. Given that the issues often involved nuclear safety concerns, the NRC required power plants to develop “[d]efense-in-depth—the use of multiple layers of protection, especially through system redundancy, to guard against or mitigate a reactor accident.” To accomplish this, the NRC required a “complement of prescriptive requirements and performance measures.”
Sometimes agencies use specification standards and performance standards to address different problems or may set them out as alternatives. The Civil Aeronautics Board (CAB) and later its successor, the Federal Aviation Administration (FAA), adopted early regulations that had both specification and performance elements (a kind of hybrid model), depending on the type of aircraft or engine addressed. A 1938 CAB regulation imposed a specification requirement to ensure extra safety protections: in requiring redundant components in aircraft engine-ignition systems, the CAB stated that an engine “of more than 100 horsepower shall be equipped with a dual ignition system having at least two spark plugs per cylinder.” In 1941, the CAB amended the regulation to add another specification (“two separate electrical circuits having separate sources of electrical energy”) but, crucially, added a performance-oriented alternative to all the specification requirements: “[A]n ignition system which will function with equal reliability in flight.” That alternative exists today. By contrast, the standard for the reliability of engine accessories began in 1938 as a performance standard (“reduce to a minimum the chances of failure”) and is still one today, even though it has been expanded to reflect more engine accessory types and the greater complexity of engines.
Performance standards can also have drawbacks. A performance standard shifts the burden of decisionmaking to the regulated entity, who will have the responsibility to ascertain what safety activity will satisfy the government inspectors’ understanding of the standards requirements. “Enforceability is a weakness of management-based [performance-based] regulation. . . . Because businesses are the ultimate enforcers of an effective and innovative management-based regulation, such regulation should not be used when it promotes goals that are adverse to the businesses involved.” Moreover, entities “may lack the expertise and resources to translate performance criteria into suitable engineering designs, especially at small firms that can ill-afford to hire outside expertise.” This is true no matter how objective and specific the performance criterion is. An employer required to reduce noise to ninety decibels may have no idea how to do so and may have to hire a noise abatement company or consultant. A performance standard with a vague or subjective compliance criterion can pose even more severe problems for regulated persons, problems which are central to this Article. “Thus, when OSHA changed its fire safety rule dictating the exact height for mounting fire extinguishers and substituted a performance standard stating that the extinguishers must be ‘accessible,’ some in the industry complained that the burden of compliance became more difficult.” If a performance standard uses a subjective rather than an objective performance criterion, compliance might be difficult for both employers and OSHA to verify. And adjudicators may interpret performance standards with unclear performance criteria in ways not intended by the drafting agency. With performance standards, the regulated often find there is no safe harbor.
An interesting effort was made some years ago to address in law the subjectivity problem posed by some performance standards. One writer distinguished between “primitive” and “precision” performance standards, defining “[p]recision standards” as “those that contain both a scientifically developed means of measurement and a scientifically known and accepted level of performance.” An offered example was the Environmental Protection Agency’s (EPA’s) maximum pollutant discharge levels for an industry. The writer defined “primitive” performance standards, on the other hand, as either “more general,” such as one that prohibits land uses that produce an “objectionable” level of emissions, or not based on levels that are “scientifically substantiated.” Primitive standards, the writer observed, “do not afford the best protection against possible arbitrary enforcement by local governments.” The writer’s distinction between precision and primitive performance standards closely resembles the distinction discussed in Part III below between those performance standards that have and those that lack the “objective” performance criteria required by § 6(b)(5) of the Occupational Safety and Health Act of 1970 (OSH Act).
Sometimes, the dilemma posed by the choice between performance and specification standards can be misstated in a crucially important way. For example, OSHA has offered this summary of what it perceives to be the dilemma posed by the choice between performance and specification standards: “Although enforceability is enhanced by specification standards, they may be unduly restrictive; on the other hand, a performance standard may be too broad to be meaningful.” As Part III argues below, OSHA’s assertion that “a performance standard may be too broad to be meaningful” is inconsistent with the way that Congress intended performance standards be written under the OSH Act. This Article argues there that a standard with the “objective” performance criterion required by OSH Act § 6(b)(5) can never be “too broad to be meaningful.”
The gravamen of this Article is that one size does not fit all. In issues of safety regulation, particularly when regulating negative externalities such as safety and pollution, design, command-and-control, or hybrid regulations may be superior to performance-based regulations. “Command and control regulation can aid in the worthy goal of increasing amusement park safety. . . . [For example,] [s]erious injuries resulting from construction problems can be reduced to minor mishaps with the use of command and control regulations.” In the workplace-related context, EPA “promulgated regulations under [the Toxic Substance Control Act] to fill a gap in the asbestos regulations of the [OSH Act]. These regulations ensure the protection of government employees who work with asbestos and asbestos-containing materials.” Asbestos regulations require “local education agencies to identify friable and nonfriable asbestos-containing material (ACM) in public and private elementary and secondary schools by visually inspecting school buildings for such materials, sampling such materials if they are not assumed to be ACM, and having samples analyzed by appropriate techniques referred to in this rule.” Follow-on regulations set forth the precise manner for asbestos testing and abatement in schools and selected other workplaces. Any assessment of OSHA standard-setting must recognize that there is a place still for specification standards in the regulatory mix.
Congress and the courts have not taken consistent approaches to the use and interpretation of performance standards. As is shown by the following discussions of statutory requirements for performance standards under the National Traffic and Motor Vehicle Safety Act, the Civil Service Reform Act of 1978, and environmental statutes administered by the EPA, Congress and the courts have taken inconsistent approaches to requirements that performance standards use “objective” performance criteria.
In 1966, Congress passed the National Traffic and Motor Vehicle Safety Act, which required that motor vehicle safety standards “be stated in objective terms.” The House report on the underlying bill stated: “In order to [ensure] that the question of whether there is compliance with the standard can be answered by objective measurement and without recourse to any subjective determination, every standard must be stated in objective terms.”
The Sixth Circuit agreed. It held that the test standard was not stated in “objective terms” because a compliant test standard must, quoting the House report, not make “recourse to any subjective determination,” that is, “the subjective opinions of human beings.” “[O]bjectivity requires that each essential element of compliance be made by specified measuring instruments . . . .” This holding appears consistent with the plain meaning of “objective.”
The Civil Service Reform Act of 1978 requires that performance standards for federal employees, “to the maximum extent feasible, permit the accurate evaluation of job performance on the basis of objective criteria.” Upon reviewing such cases, however, the Federal Circuit has held that “the legislative language does not suggest any necessary requirement for numerical measurement, and it is not at all difficult to think of many positions in which such strictly quantitative criteria would be unrevealing, bizarre, or counter-productive.” It stated that the provision required only that “a standard should be sufficiently precise and specific as to invoke a general consensus as to its meaning and content.” The Federal Circuit has since then reiterated that position.
The Federal Circuit’s reasoning is questionable. The “legislative language” uses the word “objective,” and yet the Court did not inquire into whether “objective” has a plain meaning or what it is. As shown in Part III.A below, the dictionary definition of the word and case law indicate that the plain meaning of “objective” is much narrower and more demanding than the Federal Circuit’s case law posits. Instead of recasting the word “objective,” the Court should have instead stated that for the personnel characteristics at issue, the test it articulated came as close as “feasible” (a statutory exemption from the “objectivity” requirement) to what could be stated.
Although Congress has required that performance standards have “objective” performance criteria in several statutes, that does not appear to be true with respect to environmental laws administered by the EPA. Several environmental laws require that the EPA adopts what it calls “a standard of performance,” but none applicable to private parties require that a “standard of performance” use “objective” performance criteria. For example, the Clean Air Act defines “standard of performance” for new stationary sources as a standard that “reflects the degree of emission limitation achievable through the application of the best system of emission reduction.” A similar definition lacking the word “objective” can be found in the Clean Water Act.
The remainder of this Article will be an in-depth study of OSHA’s attempt to create, implement, and enforce performance standards. As we discuss below, the word “objective” disqualifies nearly all the standards that OSHA has called “performance standards” from deserving the name, for nearly all use performance criteria that are subjective. And the OSH Act’s legislative history, to which we now turn specifically, indicates that Congress used the word “objective” to avoid the danger of subjectivity—“the danger of letting each person . . . decide [for] himself” the best way to achieve safety. This Article will conclude with recommendations that OSHA can use to create enforceable standards.
The OSH Act authorized the Labor Department’s OSHA to adopt “occupational safety [and] health standard[s].” OSH Act § 6(b)(5)’s last sentence requires that “Whenever practicable, the standard promulgated shall be expressed in terms of objective criteria and of the performance desired.”
The key word is “objective.” “Objective” means “perceived without distortion by personal feeling, prejudices, or interpretations.” As the Second Circuit has held, “Objective data . . . are independent of what is personal or private in our apprehension and feelings, that use facts without distortion by personal feelings or prejudices and that are publicly or intersubjectively observable or verifiable, especially by scientific methods.” An objective proposition is “susceptible of exact knowledge and correct statement” and “can be discovered and substantiated by external testing.” OSHA understands that this is what “objective” means, for it has repeatedly adopted standards permitting employers to not comply with certain requirements if they have “objective data” showing a lack of hazard; those standards define “objective data” so as to require numerically-expressed information.
Having said this, we must recognize that “objective data” can mean many things. It can mean scientific studies, such as OSHA’s determination that a chemical agent is a “Category I” carcinogen. It can also mean, in certain circumstances, judgments based on anecdotal evidence, for instance OSHA’s ruling that dermal exposure to benzene was carcinogenic. Indeed, we know that, at times, agencies must promulgate regulations based on probabilities, such as OSHA and EPA decisions on possible carcinogenicity of new chemical agents.
It is blackletter law that the Administrative Procedure Act (APA) does not create an independent obligation on agencies to conduct or commission their own research beyond material submitted to the record. The agencies’ responsibility is to make reasonable judgments based on the evidence at hand.
Consider FCC v. Prometheus Radio Project, which affirmed an agency action loosening media ownership rules and overturned a Third Circuit decision that stated, “Even just focusing on the evidence with regard to ownership by racial minorities, however, the FCC’s analysis is so insubstantial that it would receive a failing grade in any introductory statistics class.”
In response, the Supreme Court found that “in assessing the effects on minority and female ownership, the FCC did not have perfect empirical or statistical data. Far from it. But that is not unusual in day-to-day agency decisionmaking within the Executive Branch.” Thus, under the APA at least, standards grounded in science that are based on a probability analysis can still be objective.
It may be that the congressional definition of objectivity is a specialized definition that differs from the definitions used in academia and popular conversation. Put simply, does the plain meaning of “objective” still mean a neutral analysis, or is the very concept of a neutral objectivity now an essentially contested concept? While the definition of objective may be contested today, it is highly unlikely that the 1970 Congress had much appreciation for “progressive” notions of truth and objectivity.
The OSH Act’s legislative history contains nothing to indicate that Congress did not use “objective” in § 6(b)(5) in its ordinary sense and thus, nothing that would justify a court in construing it otherwise. On the contrary, the legislative history indicates that Congress used the word “objective” in its plain sense and did so to avoid the danger of subjectivity—“the danger of letting each person . . . decide himself” the best way to achieve safety.
The OSH Act’s bounded legislative history volume contains no passages discussing what is now the last sentence of the Act’s § 6(b)(5). The bills that were the subject of the Senate and House committee reports had language identical to that provision, but the reports did not discuss the sentence or state why it came to be in the bills.
The congressional hearings, however, are illuminating. In 1968, 1969, and 1970, House and Senate committees held hearings on early bills, all of which lacked language addressing the subject of performance standards. The first mention of a preference for performance standards occurred during House hearings in March 1968 when the Manufacturing Chemists’ Association (Association) submitted written testimony containing a statement that used two of the key terms—“performance” and “practicable”—that came to be in § 6(b)(5). The Association stated: “We urge that the bill provide that safety and health standards should be in terms of performance requirements to the maximum extent practicable.” The word “objective” was not in the statement. The record of the 1968 Senate hearings contains the same statement by the same organization and similar statements by other organizations, often in prepared statements but also orally. Similar statements and testimony were made before the House committee during the 1969 hearings. A preference for performance standards thus appears to have been important to the business community.
The hearings also shed light on the likely reason why the OSH Act’s drafters inserted the phrase “objective criteria” into the resulting committee bills and what they thought “objective” signified. The House hearings first alluded to the meaning of “objective” when witnesses spoke of objective criteria as numbers read off a meter or test instrument. A representative of the American Society of Safety Engineers testified that a performance standard would state, “[T]he exposure to certain types of toxic material, whether gas fumes, or what, must be maintained below a certain point.” A prominent representative of the labor movement during his testimony characterized “objective data” as data read off a meter.
The record of the Senate hearings indicates why the phrase “objective criteria” was likely used to draft what is now the last sentence of § 6(b)(5) of the OSH Act. The president of the U.S. Chamber of Commerce had stated both orally and in prepared submissions that, “Where feasible, all standards should be written in terms of performance requirements rather than specifications.” On behalf of Senator Harrison Williams of New Jersey (co-author of the OSH Act), a committee lawyer, in his express questioning of the witness, first asked the Chamber representative (a witness from DuPont) to clarify what the Chamber of Commerce president had meant by that. The committee lawyer then asked a question that assumed that a performance standard’s compliance criterion would necessarily be subjective: “If you write standards in terms of performance are you then running into the danger of letting each person who builds a stairway or each person who does whatever the standard is applicable decide himself what is the best way to achieve safe performance?” The witness answered that performance standards “circumscribe” plant managers “more than you think” because “[they get] down to some type of specification.”
This exchange—which posited a “danger” from subjective performance criteria—together with the portrayals by witnesses of objective data as data expressed numerically and the plain meaning of the word “objective” indicate the likely reason why the Senate committee inserted “objective criteria” into its bill: To ensure that any duty imposed under the rubric of “performance standard” would be clear enough to prevent the danger of uncertainty and debate over what degree of safety is required.
After the hearings ended, the versions of the bills reported to the House and Senate by their respective committees both contained the language now in the last sentence of OSH Act § 6(b)(5). In the House, the committee bill was at one point replaced by the Steiger-Sikes Substitute, which lacked a performance standard provision. However, the conference committee bill, and, thus, the final act, retained the performance standard language in the Senate bill.
The fourth (and last) sentence of OSH Act § 6(b)(5), by its terms, applies to both health and safety standards. Dictum in the D.C. Circuit decision in the Auto Workers case, however, states that “§ 6(b)(5) does not govern occupational safety standards.” The dictum’s phrasing was unfortunate, for the Court’s reasoning and conclusion pertained to only § 6(b)(5)’s second and third sentences, not its fourth sentence. Moreover, the text of § 6(b)(5) makes clear that its fourth sentence does apply to all standards.
Section 6(b)(5) has no scope provision. Its opening sentence, however, states that it applies to standards regulating “toxic materials or harmful physical agents,” that is, to health standards. A question may, therefore, be raised whether this opening sentence controls all of § 6(b)(5) and thus restricts the scope of the fourth, last sentence of the provision to health standards. Section 6(b)(5) states:
There is strong textual evidence in § 6(b)(5) that the D.C. Circuit’s dictum was incorrect. The third sentence uses the phrase “the highest degree of health and safety protection.” It also requires OSHA to consider “the feasibility of the standards”; that requirement would be superfluous if the third sentence applied only to health standards, for the first sentence already requires that health standards impose duties “to the extent feasible.”
The D.C. Circuit’s broad dictum is also contradicted by the fourth sentence’s legislative history, set out in Part III.B above, which indicates that it was intended to apply to all standards. For example, during the Senate hearings, a safety expert responding to committee counsel’s questions about performance standards referred to them for “stairways” and “guardrails,” classic examples of safety rather than health standards. That witness’s prepared statement, like those of others, had urged that, “Where feasible, all standards should be written in terms of performance requirements rather than specifications.” There is also no conceivable policy reason why performance standards should be preferred for health but not safety matters or why the performance criterion of a health standard should be “objective” but not that of a safety standard.
A review of the precise question before the D.C. Circuit in the Auto Workers case also shows that neither the Court’s holding nor reasoning applies to § 6(b)(5)’s fourth sentence. The union had argued that, in fashioning the Lockout Standard, a safety standard, OSHA was not permitted to apply the stricter cost-benefit concept of feasibility implicit in OSH Act § 3(8). Instead, the union argued that OSHA was required to apply the less onerous feasibility test enunciated by the Supreme Court in the Cotton Dust case for standards governing “toxic materials or harmful physical agents,” the phrase used in § 6(b)(5)’s first sentence. The union, however, argued that “the rest” of § 6(b)(5) applied to the lockout rulemaking. Although the Court rejected the argument, it understood the argument to pertain to only the second and third sentences of § 6(b)(5): It stated that “it is reasonable to conclude that the two sentences do not reach beyond toxic materials and ‘harmful physical agents,’” and at page 1,316 it referred to the “second and third sentences of § 6(b)(5).” To dispose of the union’s arguments, it was therefore unnecessary for the Court to have broadly stated that “§ 6(b)(5) does not govern occupational safety standards.”
Another reason why the Auto Workers dictum cannot be considered definitive is that part of its reasoning was in error. The Court stated, with respect to § 6(b)(5)’s second and third sentences, that “[c]oncern for ‘scientific data’ and ‘experiments’ makes complete sense for regulation of carcinogens but sounds out of place when the hazard is [a safety hazard such as] . . . a spinning saw blade.” Not only are OSHA’s safety standards often built on scientific and experimentation data (as the data assembled to support OSHA’s requirements for safety nets, guardrails, and walking and working surfaces shows), but safety criteria for devices to stop spinning table saw blades in less than five milliseconds after skin contact were developed on the basis of scientific data and experimentation.
The phrase “toxic materials or harmful physical agents” in § 6(b)(5)’s first sentence was not intended to limit its fourth sentence, and the reason for inserting the phrase in the first sentence reveals this intention. The Senate committee bill would have required OSHA to adopt both health and safety standards that “most adequately and feasibly assures . . . that no employee will suffer any impairment of health or functional capacity, or diminished life expectancy.” Senator Dominick objected that it would require “a utopia free from any hazards” and “close every business in this nation.” He, therefore, introduced an amendment to delete § 6(b)(5)’s first sentence and change the second and third sentences, but not the fourth. Later, after consulting the Senate committee’s chairman (Senator Williams of New Jersey) and ranking minority member (Senator Javits of New York), Senator Dominick proposed an agreed substitute amendment. The amendment inserted the phrase “dealing with toxic materials or harmful physical agents” into the first sentence and slightly revised the second sentence. The reason he gave for the substitute amendment had nothing to do with the third or fourth sentences. It was solely to limit the employer’s duty to assure no employee harm over a working lifetime to exposure to “toxic materials or physical agents” and to require only “steps as are feasible and practical.” The slight revision of the second sentence (from “such standards” to “standards under this subsection”) seems aimed at retaining its application to all standards if the word “subsection” is understood as referring to paragraph (b) of § 6; otherwise, the inserted phrase “under this subsection” would be superfluous. And that is exactly how the Senate report had already characterized the last two sentences of § 6(b)(5)—that is, as applying to standards adopted under “section 6(b).”
Moreover, the overbroad statements in Auto Workers rested on that court’s view that, because § 6(b)(5) is ambiguous, it would defer under Chevron to what was perceived as OSHA’s litigation position that § 6(b)(5) applies only to health standards. But the Court was likely misled on this point; at least with respect to § 6(b)(5)’s last sentence and its applicability to safety standards, the Court appears to have deferred to a misunderstood agency position. Even after the Auto Workers decision, OSHA explicitly applied that sentence in a safety rulemaking and discussed the decision in detail. Thus, the preamble to the Permit-Required Confined Spaces Standard, which OSHA called a safety standard, states that “it is OSHA’s policy, as set out in section 6(b)(5) of the OSH Act to state safety and health standards in terms of performance desired wherever possible.” Even in the rulemaking for the Lockout Standard, OSHA applied the last sentence of § 6(b)(5). And inasmuch as Chevron may be sharply limited, ignored, or, as urged by some Justices, overruled, the applicability of the last sentence might well be revisited de novo in the future.
A post-decision order by the Fifth Circuit’s decision in National Grain also contains isolated expressions suggesting that all of § 6(b)(5) applies only to health standards. But a reading of the decision and the order together indicates otherwise. The case presented the same question as Auto Workers, though it involved OSHA’s grain dust standard. As in Auto Workers, the issue involved the meaning of the first, not the fourth sentence. The opinion quoted only the first and third sentences of the provision, said nothing about the fourth sentence, did not address whether all of § 6(b)(5) applied only to health standards, and was careful to use such limiting phrases as “section 6(b)(5)’s feasibility mandate,” “the first sentence of section 6(b)(5),” or “[t]he feasibility requirement contained in section 6(b)(5)’s first sentence.” In dictum in an order on a petition for rehearing, however, the Court characterized OSHA as arguing that the original panel opinion “incorrectly holds that section 6(b)(5) . . . applies to safety standards” and stated that it agreed that “section 6(b)(5) does not apply” to safety standards. This broader usage in an order rather than an opinion should not be seen as a retreat from the more careful characterizations in the panel’s opinion but as a short-hand reference to the feasibility requirement in § 6(b)(5)’s first sentence.
There are also expressions by the Supreme Court in the Benzene Case that appear on first glance to state that all of § 6(b)(5) applies only to health standards. For example, the Court stated, “Where toxic materials or harmful physical agents are concerned, a standard must also comply with § 6(b)(5) . . . .” But the Court expressly stated that it did not hold that all of § 6(b)(5) applied only to health standards. When it quoted § 6(b)(5), the Court not only omitted its fourth sentence, but then immediately appended a footnote explaining that “[t]here is no need for us to decide” whether “[t]he second and third sentences of this section . . . may apply to all health and safety standards.” The dissenting opinion expressly characterized “[t]he remainder of” § 6(b)(5) as “applicable to all safety and health standards.” In sum, the Benzene Case left open the applicability of the fourth sentence to safety standards.
This Article, therefore, treats the fourth sentence of § 6(b)(5) as applicable to safety standards.
Soon after OSHA began enforcing the OSH Act, a problem came to light: The statute had required OSHA to adopt en masse scores of “national consensus standard[s]”—standards adopted under consensus-reaching procedures by private organizations such as the American National Standards Institute and the National Fire Protection Association. OSH Act § 6(a) states that OSHA “shall” adopt “any national consensus standard.” After OSHA adopted and began to enforce those standards, the business community began complaining publicly and to congressional committees that the national consensus standards were, among other things, “unnecessarily specific” and thus imposed unjustifiable compliance costs. The vast majority of the standards at which complaints were leveled were specification standards, not performance standards. It was soon found that the standards had rarely been drafted to be legally enforceable, were unduly restrictive, were often ambiguous, and were sometimes directed at matters other than employee safety or health. Such complaints were reflected in President Ronald Reagan’s view that “‘OSHA’ is a four-letter word that’s giving businessmen fits.”
Unnecessarily specific specification standards were a major impetus for the drive to rid OSHA of what some called “Mickey Mouse” regulations and helped fuel a “Stop OSHA” campaign. One iconic example of such a “Mickey Mouse” regulation is the proposed May 1978 OSHA Field Sanitation Regulation for agricultural workers, on the basis of which Republicans claimed (not necessarily with foundation) that cowboys had to carry a porta-potty to roundup.
OSHA’s Response. OSHA responded to the complaints by starting a process to review its standards. Among the points on which OSHA invited public comment was whether it should use performance or specification standards:
Specification or design versus level of performance provisions. Comment has been generated concerning the performance versus the design or specification type of standard. Many have alleged that OSHA’s present safety standards are too design-oriented, and that the design requirements are not always necessary for employee protection. Additionally, it is said that design standards tend to become obsolete quickly and thus are a potential road-block in the way of the growth of new technology. On the other hand, some have stated that performance standards are too general or vague and do not afford sufficient guidance to employers as to what they must do in order to achieve compliance with the standard.
The 1977 Presidential Task Force Report. At about the same time in 1976, President Gerald Ford appointed a task force to study the problem and recommend solutions. The task force “attempt[ed] to develop a model approach to safety standards” to, inter alia, “alleviat[e] the difficulties derived from the complex, detailed, but narrowly prescribed, current standards.”
Although the presidential task force devoted much discussion to the merits and drawbacks of both performance and specification standards, it recommended that OSHA adopt the performance standard approach. One of its principal observations was this: “The key to using a performance standard . . . is to design the standard so that compliance with it can be objectively measured. Only in that way can employers and employees know what the obligations are before an accident occurs.” The task force urged that standards be written so that the employer’s “compliance . . . is objectively measurable” and noted that, “[i]f properly phrased, employees can readily determine whether their employer is complying.”
OSHA’s responses to the presidential task force report discussed the recommendation that OSHA prefer performance standards when regulating machine safety, but nowhere in its combined nine pages of discussion did it use the word “objective” or mention the recommendation that performance criteria be “objective.”
The 1980 Report of the United States Regulatory Council. In 1978, President Jimmy Carter established the United States Regulatory Council “to improve coordination of Federal regulatory activities and encourage more effective management of the regulatory process.” In 1980, the Council authored a report praising performance standards as “permit[ting] more freedom of action to regulated concerns, reducing compliance costs and providing more freedom to discover new and more efficient compliance technologies.” Unfortunately, the report erroneously described performance standards, stating that they “involve regulating according to general performance criteria . . . .” The error is in the use of the word “general.” The report then stated that OSHA “has begun a sweeping program to systematically replace its existing design-specific occupational safety standards with performance standards.”
The 1987 Administrative Conference of the United States Recommendation. In 1987, the Administrative Conference of the United States (ACUS) issued its Recommendation 87-10. Among its recommendations were that “OSHA should generally use performance standards (i.e., standards that prescribe the regulatory result to be achieved) whenever they will provide equivalent protection as that provided by design standards (i.e., standards that prescribe a specific technology or precise procedure for compliance).” ACUS then added that, “In deciding which type of standard to employ, OSHA also should consider whether the standard can be readily understood and monitored and whether it may lower industry compliance costs.” As we observe in Part VI below, however, the question whether OSHA standards “can be readily understood” should not arise if OSHA followed the requirement of OSH Act § 6(b)(5) that performance standards use “objective” criteria.
Circulars by the Office of Management and Budget. The OMB under President George W. Bush issued Circular A-4 to state “best practices” for regulatory agencies. It explicitly endorsed performance standards, stating:
Performance standards . . . are generally superior to engineering or design standards because performance standards give the regulated parties the flexibility to achieve regulatory objectives in the most cost-effective way. In general, you should take into account both the cost savings to the regulated parties of the greater flexibility and the costs of assuring compliance through monitoring or some other means.
A 2016 circular again preferred performance standards.
This Article now discusses how OSHA has neither followed the requirement of § 6(b)(5) that performance standards use “objective” criteria nor heeded the admonition of the presidential task force that, unless standards use objective measures of performance, employers will not know what their obligations are. As discussed below, the importance of the word “objective” has been overlooked by OSHA, adjudicators, and legal commentators. This Section of this Article describes several commonly cited OSHA standards that OSHA explicitly touted in their text or preambles as being “performance” standards that would give employers “flexibility” in compliance but that, because they lack objective performance criteria, do not meet § 6(b)(5).
OSHA’s Laboratory Standard was explicitly adopted as a health standard. OSHA also stated that it was a performance standard. The difficulty with OSHA’s statement is that, as discussed below and contrary to OSH Act § 6(b)(5), the standard’s “performance” provisions prescribe no objective measure by which an employer can know if said provisions have been met. Two examples are as follows:
The central provision of the Laboratory Standard is the requirement for a written chemical hygiene plan, which OSHA stated used a “performance approach.” For this reason, OSHA declined to prescribe a rule for determining whether a chemical hygiene plan “would be required for each individual laboratory in establishments with many separate laboratory operations or whether a single, facility-specific plan would suffice.” OSHA stated that this was one of the matters in which the standard “generally leaves the particular details to the employer’s discretion.”
Paragraph (e)(3)(iii) of the Laboratory Standard requires that the chemical hygiene plan include “[a] requirement that fume hoods and other protective equipment are functioning properly and specific measures that shall be taken to ensure proper and adequate performance of such equipment . . . .” OSHA stated that it thereby adopted “the performance approach which allows the employer to determine the appropriate face velocities.” But, the standard sets out no method, formula, or algorithm by which the employer can determine what those “measures” are to be or whether they are “adequate.”
OSHA explained in the preamble that the matter is too complex for legal prescription—a judgment that § 6(b)(5) explicitly permits OSHA to make. But the questions may then be raised whether OSHA should have adopted a specification standard instead and, if not, whether OSHA may freely cite an employer if it were to disagree with the employer’s exercise of “discretion.” In the latter case, OSHA should be permitted to cite an employer for failing to state “specific measures” for hood maintenance but not be permitted to cite an employer for prescribing the wrong measures. Otherwise, OSHA’s assertions that the standard is a “performance” standard that gives the employer “discretion” would be illusory.
Some health standards have requirements that OSHA has characterized, sometimes in the standard itself, as “performance” requirements. Instead of setting out objective performance criteria for employers to meet, however, the standards require that employers take “reasonable” steps. For example:
OSHA’s asbestos standard requires in paragraph (d)(3) that post-initial monitoring of a workplace for airborne asbestos levels “be of such frequency and pattern as to represent with reasonable accuracy the levels of exposure of the employees.” OSHA characterized the provision as requiring the employer to sample “based on performance criteria.”
OSHA’s cotton dust standard requires in paragraph (e)(4) “that measurements of the effectiveness of mechanical ventilation equipment be made at reasonable intervals.” OSHA characterized the provision as “performance-oriented.” The previous version of the provision required that equipment effectiveness “be checked every six months and within five days after a production change.” In revising the standard, OSHA quoted with approval an industry comment that “no ‘hard and fast’ rule could be established for the variety of circumstances found in the industry” and concluded that “it was more appropriate to leave the exact frequency of such checks to the professional judgment of the plant engineer . . . .”
Inasmuch as “reasonable” is not an objective criterion, such provisions do not meet the requirement for performance standards in OSH Act § 6(b)(5).
The history of the Lockout Standard supplies a good example of how OSHA promises that a “performance” standard would afford employers “flexibility” in compliance. The Lockout Standard seeks to prevent employees from being injured by the unexpected startup of a machine while servicing it.
The Lockout Standard states in its opening paragraph that it “establishes minimum performance requirements.” As shown below, many statements in the preambles to the standard’s proposed and final versions characterize it as a “performance” standard that would afford employers “flexibility.” Although some of these preamble statements characterized particular provisions as performance standards, many so characterized the standard as a whole.
Thus, the preamble to the standard’s proposed version stated that it “does not contain specifications which must be followed in all circumstances, but, rather, provides flexibility for each employer to develop a program and procedure which meets the needs of the particular workplace.” “The employer would be given considerable flexibility in developing a control program, and such a program would be evaluated by OSHA compliance officers to determine whether it meets all the criteria in this standard.”
The preamble to the final standard repeated the theme: “The standard is written in performance-oriented language, providing considerable flexibility for employers to tailor their energy control programs and procedures to their particular circumstances and working conditions.” “The advantage of writing this OSHA standard in performance language is to allow flexibility of compliance for all systems in which hazardous energy is or may be present.” OSHA even devoted an entire preamble section to the question of whether OSHA “[s]hould . . . state the requirements of this final standard in performance language,” calling the performance language a “[m]ajor” issue. The section discusses the pros and cons of performance language, cites the last section of OSH Act § 6(b)(5), and ends with the statement, “OSHA has decided to retain the performance language in this final standard.” Shortly after the standard was adopted, OSHA issued a compliance directive stating that “[t]he standard incorporates performance requirements which allow employers flexibility in developing lockout/tagout programs suitable for their particular facilities.”
Despite all the above statements, it is difficult to find any “performance” requirement that meets OSH Act § 6(b)(5) in the Lockout Standard.
Take the standard’s training requirements. Section 1910.147(c)(7)(i)(A) of OSHA regulations states: “Each authorized employee shall receive training in the recognition of applicable hazardous energy sources, the type and magnitude of the energy available in the workplace, and the methods and means necessary for energy isolation and control.” The preamble touted this particular provision as “performance oriented, thereby providing the employer with considerable flexibility in how the training should be conducted.” But because the provision uses words that call for subjective judgments, such as “training,” “recognition,” “hazardous,” “magnitude,” “necessary,” and “control,” it does not meet § 6(b)(5).
The training provision of the Lockout Standard also provides employers with no “flexibility” worthy of the name.
It is true that § 1910.147(c)(7)(i)(A) does not tie the employer’s hands in choosing, for example, how the training is to be conducted. In that sense, it provides employers initial flexibility in compliance. But that flexibility is ephemeral. The training provision has so many subjective terms that a compliance officer could criticize almost any aspect of training—for example, that a hazard was not described clearly enough to permit employee “recognition of applicable hazardous energy sources,” that the “magnitude” of the available energy was not sufficiently described, or that employees were not trained in some method that the compliance officer thought “necessary” for lockout. The promised flexibility can thus evaporate on an employer’s first meeting with an OSHA compliance officer.
The employer would then be faced with the choice of paying attorneys to defend the criticized program or meeting OSHA’s demands that the training be altered and that a penalty be paid. Worse, if the employer were to litigate, the administrative law judges and commissioners of the independent Occupational Safety and Health Review Commission (OSHRC or Commission) might well, as they have before, treat all the terms of the standard as raising factual questions that they would then decide de novo. That would effectively transfer the decisionmaking power from the employer to the adjudicator.
Worst of all, OSHA’s attorneys might well argue that the very words that create the promised flexibility are ambiguous and that the court must therefore defer to OSHA’s liability-imposing interpretation. The practical effects would be to turn a so-called “performance” standard with “flexibility” into an ad hoc specification standard with none, and to further transfer the decisionmaking power from the adjudicator to the prosecutor.
None of that would be possible if a standard met § 6(b)(5). Such a standard would confer flexibility because it would protect the employer in their choice of compliance methods. An employer who met a standard’s objective performance criterion (such as ninety decibels or 100 parts per million (ppm)) would know that they could not be cited for a violation. The employer can thus resist pressure from OSHA to do things differently. But if a standard does not meet § 6(b)(5) because its criteria are subjective, no employer can have such confidence. OSHA’s promises that such standards grant employers “flexibility” are, therefore, illusory.
Another example of a standard that OSHA has said is a “performance” standard providing “flexibility” to employers is the Process Safety Management Standard (PSM Standard). The PSM Standard was adopted to prevent “catastrophic” events in chemical plants.
Characterizations of the PSM Standard as a Whole as a “Performance” Standard. The preambles to the PSM Standard’s proposed and final versions described it as having been “written in general, performance-oriented terms” that would give “considerable flexibility” to “employers to use . . . methods of compliance . . . appropriate to the working conditions covered by the standard.” A non-mandatory appendix to the PSM Standard states that certain compliance methods “are not the only means of achieving the performance goals in the standard.” Soon after the PSM Standard was adopted, OSHA denied a stay of its effective date on the ground that it gives employers “broad latitude to design and implement those systems and practices which are calculated to produce a safe workplace.” At least one OSHA interpretation letter stated that “employers have flexibility in complying” with PSM requirements. The above statements were not made to characterize any particular provision within the PSM Standard but to characterize the PSM Standard as a whole.
Some statements characterized particular provisions of the PSM Standard as setting out “performance” requirements. None of these provisions meet OSH Act § 6(b)(5)’s requirement for “objective” performance criteria, however. These include the following:
The Choice of Methodology for Process Hazards Analysis (29 C.F.R. § 1910.119(e)(2)). The PSM Standard requires employers to analyze chemical processes to determine what hazards they pose and in what circumstances. OSHA stated that it was “proposing a performance-oriented requirement with respect to the process hazard analysis so that the employer will have the flexibility to choose the type of analysis that will best address a particular process.” The preamble to the final version repeated the characterization. Although the promised flexibility would presumably have inhered in the proposed provision’s list of alternative analytical methodologies from which the employer could choose, the methodologies use general terms to elicit subjective judgments and are thus not objective. The final version of the paragraph was even more subjective, for OSHA added the requirement that the methodology chosen be “appropriate.” “There is scarcely a word more descriptive of unbridled subjective discretion than ‘appropriate.’ It has no objective meaning . . . .” It is “one of the most wonderful weasel words in the dictionary.” The final provision, therefore, did not meet OSH Act § 6(b)(5).
The Frequency of Inspections and Tests (29 C.F.R. § 1910.119(j)(4)(iii)). How often should a pipe in a chemical plant be inspected for corrosion? In proposing that the frequency of equipment inspections and tests be “consistent with applicable codes and standards,” OSHA stated that “This is a performance-oriented requirement to provide the employer with the flexibility to choose the frequency which will provide the best assurance of equipment integrity.” The preamble to the final rule repeated that the proposed provision “was a performance-oriented requirement.” The final version, however, added a duty to be consistent with “good engineering practices.” The final provision is another example of a so-called “performance” standard that fails to use an objective performance criterion; instead, it uses the criterion “good,” which courts have called “highly subjective.” Ordinarily, this provision too exemplifies a so-called “performance” standard that does not meet § 6(b)(5). However, we must also recognize that performance standard terms that appear to be insufficiently precise can be clarified if read against the context of industry custom or industry standards.
Safe Work Practices (29 C.F.R. § 1910.119(f)(4)). The final version of the PSM Standard contains a provision for which there was no proposed version, paragraph (f)(4). It requires employers to implement procedures “to provide for the control of hazards during operations.” OSHA characterized the provision as embodying a “performance oriented approach.” Inasmuch as the provision boils down to a requirement to “control . . . hazards,” and inasmuch as whether a “hazard” exists and has been “controlled” can both be very much in the eye of the beholder, this so-called “performance” standard does not impose the “objective” criterion required by OSH Act § 6(b)(5).
The preamble to the proposed version of OSHA’s Confined Spaces Standard contains a statement so strange as to call into question whether all OSHA rulemakers know what a performance standard is. That preamble stated that the proposed standard “uses performance-oriented language, except in a few cases where OSHA would set the permissible levels for certain substances in the workplace atmosphere.” This is a not-uncommon misconception. The statement is strange because specifying permissible air levels does not remove a standard from the category of performance standard. On the contrary, a numerical criterion is the classic hallmark of a performance criterion. The remark gives the impression that the author of the preamble did not know the difference between performance and specification standards.
As noted above, the presidential task force report emphasized the importance of performance standards having “objective” performance criteria. Unfortunately, it stands alone. No adjudicative decision has recognized the significance of the word “objective” in § 6(b)(5). Even the two appellate courts that took brief notice of the word “objective” misunderstood it. No law review article discusses it. As a result, adjudicators who have discussed “performance standards” under the OSH Act have commonly misunderstood the term “performance” to refer to standards that, without a limiting construction, would be unconstitutionally vague.
Of the ten or so appellate court cases that have taken note of § 6(b)(5), none discuss the significance of the word “objective” at any length. One court seemed in a footnote to erroneously equate “objective” in § 6(b)(5) with a requirement for reasonableness. One judge, concurring specially in a different case, remarked that a requirement for a medical examination triggered by employee exposure to a “concentration” of asbestos “does not meet the requirement for a feasible standard with objective criteria.” It is not clear in what sense that judge was using the word “objective” or whether it was apt at all, as non-zero appears to be an objective criterion.
Even decisions discussing so-called “performance” standards have failed to notice the word “objective” in § 6(b)(5). As a result, they erroneously characterize performance standards under the OSH Act as standards so vague as to be considered unconstitutionally vague without a limiting gloss.
For example, in C&W Facility Services, the employer was accused of violating OSHA’s personal protective equipment standard in 29 C.F.R. § 1910.132(a), a standard notorious for its amorphousness. The standard essentially states that personal protective equipment must be used when “necessary by reason of hazards . . . capable of causing injury or impairment.” Earlier cases had held that the fair notice required by the Fifth Amendment’s Due Process Clause required OSHA to show either that the sought protective measure is industry custom or that the employer had actual knowledge that a hazard requires the use of some other or additional personal protective equipment. The earlier cases did not use the term “performance standard” and had no need to. The court in C&W Facility Services, however, for an unexplained reason, thought it necessary to characterize the standard as a “performance” standard. It then erroneously stated that “performance standards can create problems of fair notice.” The court neither mentioned § 6(b)(5) nor used the word “objective” in the sense used by the section.
The Court’s equating of performance standards with vague standards was unfortunate, for performance standards are not necessarily vague. They can be and if they are, they can simply be called vague, and their “performance” nature ignored as beside the point. Or performance standards can have objective compliance criteria, in which case no vagueness problem would arise. OSHA standards adopted in compliance with § 6(b)(5)—that is, those with objective performance criteria (such as ninety decibels or fifty ppm)— can never by their required nature be unconstitutionally vague. The court in C&W Facility should, therefore, have omitted the term “performance” from its discussion. Instead, its discussion contributed to the general misunderstanding of what a performance standard is.
The same problems arose earlier in Echo Powerline and Sanderson Farms, where the Fifth Circuit discussed performance standards at greater length than most courts. In Echo Powerline, the court erroneously stated that one “hallmark” of a performance standard is that it “is so general as to require definition by reference to industry standards for the regulation to be reasonable.” Much the same was stated in Sanderson Farms, which cited a number of Review Commission decisions. Neither opinion took notice of the word “objective” or even of § 6(b)(5). As a result, they overlooked that performance standards adopted in compliance with § 6(b)(5)—that is, those with “objective” performance criteria—can never be “so general as to require definition by reference to industry standards.”
The Eighth Circuit’s recent opinion in Jacobs Field Services further exemplifies the confusion endemic to this subject. Section 1910.335(a)(1)(i) of OSHA regulations requires that employees working where there are “potential electrical hazards” use protective equipment that is “appropriate” for the body parts protected and “for the work to be performed.” The Eighth Circuit called this “a broadly worded performance standard” for which OSHA must prove that the need for particular protective equipment is “objectively foreseeable,” as to which industry practice will “most often . . . establish the standard of conduct.” The court was surely right in holding the standard so vague as to require the extra measures of proof it mentioned but its use of the term “performance standard” added nothing to the discussion. Had the court noticed that the standard lacked the “objective” performance criterion required by § 6(b)(5), it might have further refrained from calling the provision a performance standard.
The OSHRC is an administrative agency that independently adjudicates OSHA citations. Although the Commission’s experience should have endowed it with more familiarity with the OSH Act, the situation is no better than in the federal courts.
Recent Commission discussions of performance standards seem to equate them with unconstitutionally vague regulations. In Thomas Industrial Coatings, the Commission flatly stated, “Because performance standards . . . do not identify specific obligations, they are interpreted in light of what is reasonable.” Interpreting standards in light of what is reasonable is a gloss that the Commission uses to alleviate problems of unconstitutional vagueness in standards. The Commission overlooked that if a standard’s performance criterion is “objective” as required by § 6(b)(5), it would not pose a vagueness problem and would not require a “reasonable person” gloss.
In several cases, the Commission described a performance standard but, in doing so, confused the words “performance” and “objective.” In one, it stated, without citing or quoting § 6(b)(5), that “[a]s a performance standard, [a provision] states the objective to be achieved . . . [,] not the means for achieving it.” To have been faithful to § 6(b)(5), it should have written that the standard stated the “performance” to be achieved and treated “objective” not as a noun but as an adjective describing a requirement for a performance criterion. The Commission’s usage reflected a lack of nuanced knowledge of § 6(b)(5) and the role of “objective” in it. In other cases, the Commission described performance standards inaccurately. In Thomas Industrial Coatings and again in Cleveland Wrecking, the Commission erroneously stated that performance standards “require an employer to identify the hazards peculiar to its own workplace.” This is disappointing.
In sum, adjudicative tribunals have not yet focused on the purpose of the word “objective” in § 6(b)(5). As a result, the term “performance standard” has been widely misunderstood and mischaracterized.
Once OSHA starts to enforce a standard that it called a “performance” standard during the rulemaking but that lacks the objective performance criterion required by OSH Act § 6(b)(5), OSHA enforcement officials and lawyers break OSHA’s promises of “flexibility.”
The problem is that no known OSHA document instructs enforcement officials on how to give concrete meaning to the promise of “flexibility” or how to refrain from issuing citations on the basis of any promised flexibility. For example, no known OSHA document instructs enforcement officials to refrain from issuing a citation alleging a violation of such a standard if the employer’s judgment was “reasonable,” even if OSHA thinks it wrong. As a result, OSHA officials freely substitute their own judgment for the employer’s and treat the words of the standard as raising factual questions to be resolved de novo. Thus, they grant the employer no flexibility at all.
OSHA’s lawyers equally disregard OSHA’s promises of “flexibility.” They do so in several ways.
Turning Flexibility into Ambiguity to Trigger Deference. A standard that lacks an objective performance criterion can grant “flexibility” to an employer only if the standard’s words are themselves flexible. The PSM Standard’s provision on methodologies for process hazards analysis, which uses the subjective word “appropriate,” is a good example of a provision touted as a “performance” standard that, therefore, grants “flexibility.” But OSHA’s lawyers often argue that the very words that make for the touted flexibility are ambiguous and should thus trigger judicial deference to OSHA’s duty-imposing “interpretation.” This undermines the standard’s stated intent, which is to provide flexibility to the employer, and substitutes an inflexible and ad hoc criterion—whatever OSHA “reasonably” wants. This way of breaking faith with employers is especially pernicious.
Though decided on another ground, a recent Eleventh Circuit case exemplifies this problem. The preamble to OSHA’s emergency response standard characterized it as a “performance” standard: “This final rule is written so that employees engaged in hazardous waste operations and related emergency response operations . . . would be protected by general, performance oriented standards.” This meant, OSHA stated, that employers would have “flexibility” in compliance: “[T]he performance nature of this final rule, of and by itself, allows for flexibility by . . . providers of emergency response to provide as much safety as possible using varying methods consonant with the conditions in each state.” The case specifically involved the definition of “emergency response” in § 1910.120(a)(3), as to which OSHA’s own compliance directive twice rejected any “formula” for whether an “emergency” response is required, and instead set out “factors” to “consider[].” It states that the standard is “a performance-oriented standard, which allows employers the flexibility to develop a safety and health program suitable for their particular facility or operations.”
OSHA’s attorneys nevertheless argued that “incidental,” “potential,” “immediate,” and “uncontrolled” were ambiguous and, therefore, OSHA’s “reasonable” interpretations of them were entitled to deference in judging the employer’s decision that a certain release did not warrant an “emergency” response. The employer responded that such deference would deprive employers of the flexibility that it was the standard’s clear original intent that they were to have. As noted above, the case was decided on another ground.
Treating the Standard as Raising Merely a Factual Issue to Be Resolved De Novo. OSHA’s lawyers often treat the very words that made for the touted flexibility as merely requiring factual inquiries to be resolved de novo by the Commission, without leeway for employer judgment. Their approach amounts to saying, “Yes, the employer has flexibility—if he’s right.” The approach is seductive enough to have attracted the Commission and at least two appellate courts.
An example is the Albemarle case, which involved the PSM Standard, touted by OSHA as a “performance” standard conferring “flexibility” on the employer. Albemarle’s written operating procedures lacked instructions about an operation called “block-and-bleed.” This was alleged to violate § 1910.119(f)(1), which requires employers to “provide clear instructions for safely conducting activities involved in each covered process.” Instead of including written instructions, Albemarle trained employees on the technique. Before the Commission, Albemarle argued that § 1910.119(f)(1) was a “performance” standard that permitted employers to make judgment calls in choosing the level of detail in operating procedures. It argued that, in writing them, employers must balance the need for information against the dangers of prolixity. It cited a chemical industry manual stating the dilemma and argued that its judgment call—to convey information about “block-and-bleed” in training—was reasonable. The Commission essentially ignored the argument and announced an “include everything” interpretation of § 1910.119(f)(1). Albemarle then made essentially the same argument to the Fifth Circuit but, in an opinion that never used the word “performance,” had the same result.
What Performance Standard? Another way that OSHA avoids the implications of its statements about “performance” and “flexibility” is to argue that, while other provisions of a standard may be performance-oriented, the cited provision is not. An example is Otis Elevator, where OSHA alleged a violation of a provision of the Lockout Standard requiring that host and outside employers “inform each other of their respective lockout or tagout procedures.” The reader will recall that OSHA touted the standard as a whole as a performance standard conferring flexibility on the employer. The employer in the case, therefore, concluded that the provision did not apply to this particular instance because there was no possibility of its employees and those of the host employer interacting and thereby creating misunderstandings.
Although an administrative law judge agreed with the employer, the Commission reversed, holding that this particular provision was a “specification standard,” that is, written so as to leave no room for employer judgment. The D.C. Circuit agreed, stating that “[t]he regulation’s use of mandatory directives like ‘shall’ and ‘whenever’ defy the optionality in operation that Otis Elevator favors.” Otis had cited a statement in OSHA’s compliance directive that described the information exchange process as “performance-oriented.” The Court dismissed this: “[T]he reference to ‘performance-oriented’ applies to which energy control procedures to use . . . [A]t no point does the Directive suggest that employers are free to choose whether they will comply with the information exchange provision at all.” The court seemed unaware that OSHA had, in preambles, characterized the Lockout Standard as a whole as performance oriented.
Ignoring the Issue. OSHA’s lawyers often just ignore the argument, hoping that overworked courts, which often take their cues from what is important from an agency’s brief, will do so too. An example is the Wynnewood Refining case. There, Wynnewood argued that “[i]n keeping with the performance-oriented nature of the rule, the PSM standard gives discretion to employers to determine how best to achieve its purposes.” OSHA’s brief ignored the argument, and so did the Tenth Circuit.
As a result of this lack of attention to § 6(b)(5)’s words, OSHA has been permitted to adopt and enforce so-called “performance standards” that do not comply with the section’s requirement that “objective” performance criteria be used. The adverse consequences have been numerous.
First, OSHA’s conduct undermines the integrity of the rulemaking process because it misuses key documents in that process—the text of and preamble to a proposed regulation. It is only the proposed version and its preamble that notify the public of the agency’s intentions, which allows people to determine whether they should participate in the rulemaking and on which comments to focus. OSHA’s touting there of the words “performance” and “flexible” can mislead rulemaking participants into believing that the final standard will be more forgiving than it is and, thus, induce employers to not oppose such proposed standards or to oppose them less vigorously. And according to some case law, private persons who are so lulled, and thus fail to submit an objection during the rulemaking, may not raise that objection in any pre-enforcement challenge to the validity of the final regulation.
A preamble containing an “affirmative mischaracterization of [a proposed rule’s] import and impact” violates the requirement of the APA in 5 U.S.C. § 553(b)(3) that the agency disclose “either the terms or substance of the proposed rule or a description of the subjects and issues involved.” “Agency notice must be sufficient to fairly apprise interested parties of the issues involved, so that they may present responsive data or argument related thereto.” Such undermining of the rulemaking process should, in a democratic society, be considered important, for an opportunity to comment is a substitute—a poor substitute—for legislation by elected representatives of the people. “An administrative agency . . . is not ordinarily a representative body . . . . Its deliberations are not carried on in public and its members are not subject to direct political controls as are legislators.”
Second, the use of subjective performance criteria in a final standard deprives employers of the notice and certainty that performance standards compliant with § 6(b)(5) would afford. As the presidential task force noted, a standard with an objective performance criterion permits the employer to know for certain whether it has complied far in advance of any OSHA inspection. Concomitantly, no compliance officer could dispute that the employer has complied. Even if an OSHA compliance officer were unhappy with the way that the employer complied (for example, with the way noise exposures were brought down to a permissible level), the objective nature of the performance criterion means that OSHA could not credibly allege a violation or attempt to dictate how the employer must comply. By using subjective compliance criteria, however, OSHA is able to deprive employers of both guidance and certainty.
Third, the use of subjective performance criteria makes it possible for OSHA enforcement officials to freely cite employers for almost anything they dislike. Employers soon learn from their attorneys that such subjective language makes illusory statements by OSHA in the text of or preamble to a standard that its “performance nature” will provide employers “flexibility” in compliance. This will eventually result in a sense by employers of betrayal—a sense that is corrosive to public trust in government.
Fourth, the subjectivity of the supposed performance criteria can lead OSHA enforcement officials to believe that, by the way they write a citation, they can dictate or limit the means of compliance—that is, bar use of a certain disfavored compliance method or fault the employer for not using a favored one. Issuing either sort of citation would further destroy the flexibility promised during the rulemaking and effectively turn a so-called “performance” standard into what is, in reality, a specification standard and an ad hoc one at that.
While we argue supra that the OSH Act § 6(b) objectivity requirement creates a higher bar than other government agencies for clarity in their performance regulations, the problem of “objectivity” and subjectivity in OSHA performance standards raises some significant jurisprudential questions regarding basic principles of administrative law—specifically the concern for “fair notice.” As discussed in this Article, the OSH Act’s textual requirement of “objectivity” for performance regulations ought not (if read correctly) allow for excessive subjectivity (or flexibility) by the regulator. Such subjectivity by the regulator raises issues regarding the regulated party’s knowledge that they are committing an infraction—what is often called “fair warning” or “fair notice.” The basic principle is that you should not be punished for breaching a requirement or requirements that you did not know about. There is some sense that the fair notice (or fair warning) concept is “based on constitutional requirements of fair notice.”
A regulation that does not meet “objective” criteria is likely to create uncertainty as to its meaning, which can lead to a lack of “fair notice.” Depending on the facts, this uncertainty can result in situations where the defendant does not know specifically what is required of them. This can lead to unfair surprise. Unfair surprise suggests something more than an incorrect “reasonable” interpretation by the employer. It suggests that the employer could not have reasonably known what standard was required of them. There is likely a range of interpretations that could be deemed reasonable—some more favorable to the employer, and some are perhaps closer to what OSHA had in mind when it promulgated its performance standard. Justice Felix Frankfurter’s comment on the interpretation of statutes is relevant here:
Anything that is written may present a problem of meaning, and that is the essence of the business of judges in construing legislation. The problem derives from the very nature of words. They are symbols of meaning. But unlike mathematical symbols, the phrasing of a document, especially a complicated enactment, seldom attains more than appropriate precision.
The classic fair warning case is General Electric Co. v. EPA (GE). The fair warning argument in GE related to a case of regulatory confusion—a situation where differing regulatory standards might apply. In GE, the company developed a process for customers to dispose of electric equipment, most often computers that contained polychlorinated biphenyls (PCBs). There was conflicting regulatory guidance from different EPA offices regarding the proper method of enforcement, and GE asserted EPA’s method of enforcement was “arbitrary, capricious, and otherwise impermissible interpretation of its regulations.” The D.C. Circuit found that a regulation must give the regulated party “fair warning of the conduct it prohibits.” Further, the regulated party must be able to identify with ascertainable certainty the standards to which the party is expected to conform, and those standards must be “reasonably comprehensible to people of good faith.” The Court’s concern was that the “interpretation is so far from a reasonable person’s understanding of the regulations that they could not have fairly informed GE of the agency’s perspective.”
The fair warning argument in GE is basically a claim of “regulatory confusion”—a situation where there are competing regulatory standards that might apply; GE chose what the agency considered the “wrong” standard, but the Court found that when the meaning of an agency’s interpretations of regulations cannot be reasonably understood by persons to whom those regulations apply, they cannot be found to be in violation. And competing standards can create such a confusion.
The GE understanding of “regulatory confusion” is one point on a continuum of claims of lack of fair warning. Other points on the continuum have been expressed in case law. As example, in Christopher v. SmithKline Beecham Corp., the agency’s interpretation conflicted with long accepted industry practices, seemingly uncontested by the Department of Labor (DOL), so enforcement was an unfair surprise. Usually, the DOL’s interpretation of the ambiguous provision would be entitled to deference, but the Court held that no such deference was warranted because the DOL’s 2009 interpretation directly conflicted with an industry-wide practice that had been in place, uncontested, since the 1950s. Therefore, to permit DOL’s interpretation to stand would be to “impose potentially massive liability on respondent for conduct that occurred well before that interpretation was announced . . . [and] it would result in precisely the kind of ‘unfair surprise’ against which our cases have long warned.”
By one view, if the reasonable, intelligent judgment of the employer is always a safe harbor, this will give those employers not particularly interested in safety two bites at the apple. The employer will argue that their first stab at compliance was not unreasonable (first bite) and that only when told by an OSHA inspector how he must comply, and failing to do so (second bite), is he liable.
At the same time, we must consider the extent to which a regulated entity has a responsibility to seek to ascertain what the agency’s regulation was intended to accomplish. Consider a hypothetical situation where the DOL interpreted its Wage and Hour overtime regulations to require an employer to pay employees overtime whose abattoir job, by law, required them to suit up in protective gear and the employer claimed the standard is loose enough to allow for a “reasonable interpretation” that overtime need not be paid. Does the employer have some responsibility to be aware of the interpretation the agency might choose and plan accordingly? Or should they ask the agency? Should the employer have a duty to flesh out what the agency believes this statute requires? Or if they can situate their interpretation in the range of reasonableness, do they have a “safe harbor,” even if they know their interpretation may not be the one the agency intended?
Much depends on whether you envision the regulation to have been drafted to guide the “good” man who “finds his reasons for conduct, whether inside the law or outside of it, in the vaguer sanctions of conscience” or to constrain the “bad” man. Justice Oliver Wendell Holmes suggests that the bad man is motivated by “what the . . . courts are likely to do in fact.” If you substitute agency for court, the question is how the agency reasonably interprets the standard.
In Diamond Roofing, the Court found that “a regulation cannot be construed to mean what an agency intended but did not adequately express,” adding that it is the Secretary of Labor’s “responsibility to state with ascertainable certainty what is meant by the standards he has promulgated.” Thus, if the employer’s interpretation is reasonable, they would have a safe harbor. The safe harbor approach, taken to its logical conclusion, actually gives the bad man an advantage. With a “safe harbor,” the bad man does not have to ask what interpretation the agency has chosen. They need only ask, “Is my interpretation a reasonable one?”—whether or not it is the interpretation that the agency would choose. This gives them, one would think, two bites at the apple.
In another case, the Fifth Circuit found that the standard must be “reasonably comprehensible to [people] of good faith.” In Satellite Broadcasting Co. v. FCC, the D.C. Circuit found that if two interpretations are reasonable, the agency could not enforce either. In other cases, they have urged deference to the agency interpretation. As one can see, how much certainty is needed to reach “reasonable certainty” and prevent regulatory confusion is an open question for the courts.
And what if there is regulatory confusion or lack of fair notice? How should a court proceed? One approach is that of United States v. Chrysler Corp., where Chief Judge Edwards noted that a manufacturer cannot be held to have violated a standard “if it had no notice of what [National Highway Traffic Safety Administration] now says is required under the standard.” Alternatively, courts have upheld the agency charge but have regularly rejected penalties when agency interpretations (either because they are new or a choice between interpretative options) are such that the regulated entity cannot be expected to have understood what was required of them. This is illustrated in Rollins v. EPA, where the court found both a competing standard problem and declined to impose a penalty. While it found Rollins’s interpretation more plausible than EPA’s, it decided over a strong dissent that EPA’s “rather more strained” interpretation would be upheld as correct based on deference. The Court described the latter reading as “rather more strained.” Despite this holding, though, the Court refused to impose any penalty.
Former White House Counsel Don McGahn put the fair notice problem starkly:
The first step is preserving individual liberty in the face of the burgeoning federal Leviathan is to insist on fair notice. The government has an obligation to clearly inform parties of the rules that will apply to them. Anyone who is engaged with the regulatory state in the last several decades knows that agencies have often taken precisely the opposite approach. Far too often agencies issue vague regulations or in some cases, no regulations at all. Then they interpret those vague regulations through byzantine interpretive rules, guidance documents, or so-called “Dear Colleague” letters. To administrative law experts, these are known by the Orwellian term ‘subregulatory actions.’ But whatever you call them, they are illegitimate.
In an earlier age Roscoe Pound made a similar point stressing that
the law cannot say to a business man, well, you guess; you employ a lawyer by the year to give you the best guess that he can, and then as the result of litigation we will tell you five years afterwards whether your guess as to the conduct of your business was the correct one or not.
The problem of “fair notice” is a perennial trope conservatives raise to critique the administrative state. In an earlier generation, the Regulation Fair Warning Act underscored this theme. The now-rescinded Trump Executive Order 13892 reflected a more recent iteration of this concern. It stated in relevant part that an agency “may apply only standards of conduct that have been publicly stated in a manner that would not cause unfair surprise,” with “[u]nfair surprise” defined as “a lack of reasonable certainty or fair warning of what a legal standard . . . requires.” While one may argue over the interpretation of “unfair surprise,” “reasonable certainty,” and “fair warning,” the approach serves as a significant and needed check on regulatory overreach.
Administrative law is undergoing a reconceptualization. One aspect of this rethinking is the emergence of a new paradigm demanding clear congressional delegation to support legislative rulemaking. This Article provides an opportunity to rethink another central issue in administrative law—the best approach to performance standards. The Article argues that the OSH Act requirement for “objective” performance criteria in rulemaking has often not been followed in practice and has too often resulted in charges based on an ambiguous statutory text. The ambiguity (what OSHA calls “flexibility”) may mean that the OSHA performance standards that purport to provide objective criteria in fact fail to provide fair warning as to what the agency requires and that the resultant “subjectivity” results in unfair surprise for the regulated community. It is time for all parties involved with OSHA rulemaking—employers, trade associations, unions, and the agency staff, as well as other stakeholders—to up their game.
With the federal courts’ increased watchful interest in “fair warning,” OSHA “rulemakers would do well to proceed in the immediate future with even greater caution than in the past in insuring that their rules give fair notice of what is expected of the regulated and fair procedures for disputing alleged violations” by being as precise as possible. By doing so, they can reduce the range of interpretations that would be considered reasonable (though they will have, of course, reduced some of the regulation’s “flexibility”). Further, after drafting a performance regulation, OSHA should consider providing an extensive list of examples of activities they would consider meeting the standard and those they would not. The greater the number of examples, the less likely an employer would not know what the agency expects. We recognize that this introduces specification elements into a performance standard, but that is not always inherently bad.
We do not believe, as some do, that these examples need be sent out for some type of pre-promulgation notice. While agency interpretative rules and statements of policy should be “handled with care,” there is still great value in allowing agencies flexibility in their understanding of the statutes and regulations for which they are responsible. It is transparency (or the lack thereof) that makes the difference.
At the same time, OSHA might well create a system whereby employers can seek guidance as to whether their interpretation of a performance standard would meet OSHA’s understanding. This would be similar to a request for an IRS private letter ruling. Such capacity would assist both OSHA and the regulated party in avoiding regulatory confusion.
Before developing a future standard, OSHA should analyze whether it can in fact provide “objective” criteria required by § 6(b)(5). Examples of such standards include the provision of OSHA’s noise standard, which uses numerical limits in decibels, and its air contaminants standard, which uses numerical limits in such units as parts per million molecules of air. And if OSHA finds it impracticable to state an objective performance criterion, it should so state and adopt a specification standard instead.
With respect to current standards—such as OSHA’s Lockout, Process Safety Management, and Laboratory standards—OSHA should publicly instruct its compliance personnel that if a standard or its preamble characterized the standard as a “performance” standard providing “flexibility” in compliance, but uses subjective compliance criteria, employer judgments must be respected so long as they are reasonable. By reasonable, we mean the regulated party could have really understood the performance standard to include OSHA’s definition of the required context.
For their part, all stakeholders—employers, trade appreciations, unions, and worker associations—should analyze more clearly any statements in the preamble to a proposed standard that the standard would be a “performance” standard providing “flexibility.” They should instead determine whether a proposed requirement meets OSH Act § 6(b)(5)’s requirement for an “objective” performance criterion—such as a requirement of ninety decibels. If it does not, they should file comments in the rulemaking record that the proposed standard violates the last sentence of § 6(b)(5), that it would be, in fact, “a subjective” standard. And, in fact, any promise of “flexibility” would be illusory. All stakeholders should demand that all standards touted to be performance standards have the objective performance criteria required by § 6(b)(5).
Employers and trade associations should also go further. They should demand that, as § 6(b)(5) expressly requires, all standards by default be performance standards with objective performance criteria unless OSHA finds that such criteria are not “practicable.” Employees and their unions might well find it in their interest to join employers in a demand that OSHA comply with § 6(b)(5) and thus use only objective performance criteria unless that exception is met. That way, all parties would have “fair meaning” as to what is expected of them. If objective performance criteria cannot be set down, then all should demand that specification standards be set.
If OSHA cannot write either kind of standard, then employers and their trade associations must demand that either no standard be set or that OSHA state explicitly that the standard is so phrased as to require the employer only to make a reasonable judgment as to what the standard requires and not a correct one, and therefore that citations will not be issued if the employer’s judgment is reasonable.
As for the Commission, it should stop equating unconstitutionally vague standards with performance standards. Some performance criteria are unconstitutionally vague; some are not. Those that meet OSH Act § 6(b)(5)’s requirement for an “objective” performance criterion can never be unconstitutionally vague. As discussed above, the word “objective” means “expressing or involving the use of facts without distortion by personal feelings or prejudices,” which, as understood by the drafters of the OSH Act and as a practical matter, requires numerically expressible criteria. The Commission and the courts should also stop characterizing a standard as a “performance” standard unless it somehow advances the analysis. If a standard is or would be unconstitutionally vague, it does not advance the analysis to also call it a performance standard.
Second, if a standard is characterized in a proposed standard or its preamble as a “performance” standard that affords “flexibility,” but does not prescribe an objective performance criterion, the Commission and the courts should consider holding that the standard is a sham performance standard which raises significant “fair notice” concerns (whether based on the Due Process clause or the APA). Further, it would be making requirements that the regulated party could not really foresee. Inasmuch as this holding would provide the promised flexibility, it would vindicate the original intent of the standard, prevent the undermining of the rulemaking process by OSHA’s misleading statements, and provide fair notice to all parties. It is time to make the promise of “objective” standards real by making such standards actually objective, and as this study hopes to show if this promise cannot be met, then OSHA should revert to specification standards. Remember, one size does not fit all.
Professor Breger thanks his sterling research assistants, Leon Scotorow (2023) and Emily Bushman (2024) for insights and perseverance. He also thanks Harvey Reiter for helpful comments.
Just as this Article was substantially completed, Arthur Sapper died suddenly and unexpectedly. He was a terrific lawyer and a man of great personal integrity. This Article is dedicated to him. Lacking his wise discernment for the final review, all errors should be laid to me, Professor Marshall Breger. May his memory be for a blessing.
